Reduced PDK4 Expression Associates with Increased Insulin Sensitivity in Postobese Patients


Istituto di Medicina Interna, Università Cattolica S. Cuore, Largo A. Gemelli, 8, 00168 Rome, Italy. E-mail:


Objective: The aim of this study was to verify whether changes in PDK4 mRNA expression in skeletal muscle in formerly obese subjects who underwent malabsorptive bariatric surgery [bilio-pancreatic diversion (BPD)] might be related to insulin sensitivity improvement, and if these possible modifications might correlate with a reduction of the intramyocytic lipid level.

Research Methods and Procedures: Six obese women (body mass index 46.6 ± 8.2 kg/m2) were enrolled in the study. Body composition, euglycemic-hyperinsulinemic clamp and muscle biopsies for skeletal muscle lipid analysis, and semiquantitative reverse transcriptase-polymerase chain reaction were performed before and 3 years after BPD.

Results: The average weight loss observed after surgery was ∼42%. Increased glucose uptake was accompanied by a significant decrease of PDK4 mRNA (R2 = 0.71, p < 0.001). The amounts of intramyocytic triglycerides correlate directly with PDK4 mRNA (R2 = 0.87, p = 0.005) and inversely with glucose uptake values (R2 = 0.75, p < 0.001).

Discussion: Our results support the concept that a reduced tissue availability of fatty acids consequent to a massive lipid malabsorption influences glucose metabolism acting through the regulation of PDH complex. In fact, as shown in animals, a higher level of FFA availability is likely to induce overexpression of PDK4 also in humans.


Insulin resistance (IR)1 is a prominent biological marker of obesity and is defined as an impaired biological response to either exogenous or endogenous insulin (1). A strict correlation between circulating levels of free fatty acids (FFAs) or triglycerides (TGs) and IR has been observed repeatedly (2, 3, 4, 5). Several studies have also reported a correlation between the TG content in skeletal muscle and this syndrome (6, 7, 8, 9). Accordingly to Randle's hypothesis (2), the negative effect of increased tissue lipid content, which likely reflects the plasma lipid concentration on insulin sensitivity, is commonly ascribed to the competition between acetyl-coenzyme A particles (CoAs) deriving from FFA β-oxidation and those deriving from glucose oxidation to entry into the Krebs cycle. The inhibitory effect of fatty acids on glucose metabolism is mediated mainly through changes in the activity of pyruvate dehydrogenase kinase (PDK) isoforms (PDK 1 to 4), which in turn regulate the activity of pyruvate dehydrogenase (PDH) (10), the enzymatic complex that catalyzes the first irreversible step of glucose oxidation (11, 12), thus representing an important component of the regulation of glucose homeostasis. The high mitochondrial acetyl-CoA/CoA and NADH/NAD+ concentration ratios deriving from increased rates of FFA β-oxidation are responsible for the activation of PDK. Skeletal muscle PDK4 isoform expression is selectively increased in insulin-deficient (13, 14) and -resistant states, in both rats (15) and humans (16).

IR is a reversible condition in morbidly obese patients who have undergone malabsorptive bariatric surgery [bilio-pancreatic diversion (BPD)] (17). Interestingly, the degree of weight loss achieved and the normalization of insulin sensitivity seem to be independent factors (17), whereas the improvement of insulin-mediated glucose disposal is likely a result of a lower TG content in muscles and the subsequent decrease of intramuscular FFA availability (9, 18).

The aim of the present study is to assess the skeletal muscle variation of PDK4 mRNA expression in postobese subjects when glucose disposal is reversed to normal. For this purpose, we studied a group of morbidly obese subjects before and after BPD operation. Insulin sensitivity was evaluated by the euglycemic-hyperinsulinemic clamp (EHC) technique before and 3 years after the surgical treatment. Skeletal muscle biopsies were performed to measure the amount of intramyocytic TGs and to extract RNA for gene expression analysis.

Research Methods and Procedures


Six obese subjects (six women, age 32 ± 8 years, body mass index 46.6 ± 8.2 kg/m2) were enrolled in the study. Subjects had maintained a stable body weight for at least 2 months before the beginning of the protocol. The subjects were not on regular medication and did not show known complications of obesity, such as established hypertension, diabetes, or dyslipidemia. All subjects were white. All obese subjects received a basal study, after which they underwent biliopancreatic diversion (BPD), consisting of a partial gastrectomy with a distal Roux-en-Y reconstruction (19). These obese subjects were restudied 3 years after BPD.

Biopsies of the vastus lateralis muscle were performed twice, before and 3 years after BPD. Blood samples for chemical analysis were drawn after an overnight fast. All subjects had given written consent, and the experimental protocols were approved by the Ethical Committee of the Catholic University in Rome.

Body Composition

The day preceding the surgical procedure, body weight was measured to the nearest 0.1 kg by a beam scale and height to the nearest 0.5 cm using a stadiometer (Holatin, Crosswell, Wales, UK). Total body water was determined using 0.19 becquerels of tritiated water in 5 mL of saline solution administered as an intravenous bolus injection (20). Blood samples were drawn before and 3 hours after the injected dose. The disintegrations per minute were counted in duplicate on 0.5 mL of plasma using a Beta-scintillation counter (Model 1600TR; Canberra-Packard, Meriden, CT). Corrections were made (5%) for nonaqueous hydrogen exchange (21), and water density at body temperature was assumed to be 0.99371 kg/L. Total body water (kilograms) was computed as 3H2O dilution space (liters) × 0.95 × 0.99371. The within-person day-by-day coefficient of variation reported for this method is 1.5% (22).

EHC Procedure

Peripheral insulin sensitivity was evaluated by the EHC procedure (23). After inserting a cannula in a dorsal hand vein for sampling arterialized venous blood, and another one in the antecubital fossa of the contralateral arm for infusions, the subjects rested in a supine position for at least 1 hour. To obtain arterialized blood samples, one hand was warmed in a heated air box set at 60 °C. Whole-body glucose uptake (M value) in micromoles per kilogram fat free mass (FFM) per minute was determined during a primed-constant infusion of insulin (at the rate of 7 pmoles/kg per minute). The fasting plasma glucose concentration was maintained throughout the insulin infusion by means of a variable glucose infusion and blood glucose determinations every 5 minutes. Whole-body peripheral glucose use was calculated during the last 40-minute period of the steady-state insulin infusion.

Muscle Biopsies

Muscle biopsies were obtained at 8:00 am under local anesthesia from the vastus lateralis portion of the quadriceps femoris muscle in a different section from the EHC procedure. Tissue samples were immediately placed in liquid nitrogen and stored at −80 °C for further analysis.

Semiquantitative Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)

Frozen muscle biopsies were pulverized in liquid nitrogen, and total cellular RNA was obtained using guanidinium thiocyanate-phenol-chloroform extraction (24). The amount of RNA was quantified spectrophotometrically, and its purity was assessed by the ratio 260:280.

The Perkin Elmer Gene Amp RNA PCR kit was used for the RT-PCR reaction, which was performed in the Gene Amp PCR System 9600 (Perkin Elmer Cetus, San Diego, Ca). After removal of contaminating chromosomal DNA with DNase I treatment, 500 ng was reverse transcribed with 2.5 units of Moloney Murine Leukemia Virus RT (Promega, Madison, WI) at 42 °C for 30 minutes. Two microliters of cDNA products were used in each PCR reaction. PDK was amplified with 1 unit of AmpliTaq Gold DNA polymerase and 1.5 mM MgCl2. A first incubation of 10 minutes at 95 °C, 45 seconds at 60 °C, and 1 minute at 72 °C was followed by 26 cycles of 45 seconds at 95 °C, 45 seconds at 60 °C, and 1 minute at 72 °C. The primers used for amplification of PDK were: primer sense 5′-TGCCAATTTCTCGTCTGTATG-3′ and primer antisense 5′-AAAAACAGATGGAAAACTGAGG-3′. The target cDNA was coamplified with aldolase-A as internal control (25) using the following set of primers: 5′-CGCAGAAGGGGTCCTGGTGA-3′ and 5′-CAGCTCCTTCTTCTGCTCCGGGGT-3. Both primer couples were used at the concentration of 0.06 μM. All oligonucleotides were synthesized by Pharmacia Biotech (Uppsala, Sweden).

The number of cycles and the reaction conditions were chosen so that none of the target cDNAs reached a plateau and so that the two pairs of primers did not compete with each other.

PCR products were electrophoresed on a 2% agarose gel, images on the ethidium bromide-stained gels were acquired by a Cohu CCD camera (Cohu, San Diego, CA), and quantification was performed with Phoretix 1D (Phoretix International LTD, Newcastle on Tyne, UK).

Skeletal Muscle Lipid Analysis

To measure intracellular lipids, a specimen of ∼100 mg was taken and immediately placed into a calcium-free Hanks solution combined with EDTA and bubbled with 95% O2 and 5% CO2. The sample was washed and then immersed in a fresh Hanks solution combined with 50 mg of collagenase type IV and calcium ions and agitated in a Dubnoff water bath maintained at 37 °C until the tissue appeared soft. At this point, the specimen was gently removed, cells were brushed with a blunted spatula, filtered, suspended in phosphate-buffered saline, and centrifuged twice at 50g for 2 minutes. The supernatant was discarded, and the muscle cells were dried under a nitrogen stream. After protein precipitation with 5 to 10 mg of trichloroacetic acid, lipids were extracted twice with 8 volumes of chloroform:methanol (2:1, vol/vol), stirring the solutions at 60 °C for 15 minutes. The combined extracts were dried in a GyroVap apparatus (GV1; Gio DeVita, Rome, Italy) operating at 60 °C, coupled with a vacuum pump and a gas trap (FTS System, Stone Ridge, NY). The dry weight of lipid extracts was obtained by weighing the sample tube before and after drying the extracts. The above extracts were redissolved in chloroform:methanol (2:1, vol/vol) and fractionated into their various components by thin-layer chromatography on standard thin-layer plates (Stratocrom SI AP; Carlo Erba, Milan, Italy) coated with a 0.25-mm-thick layer of silica gel and activated by heating at 120 °C for 20 minutes. The plates were developed into successive solvent system as described by Passi et al. (26). The area of silica gel corresponding to the retardation factor of a triolein and tripalmitin standard mixture was scraped off and extracted with peroxide-free diethyl ether. The TG fraction eluted from the thin-layer chromatography plates was saponified by a treatment with 2 N KOH in methanol and successive acidification to pH 2 to 3 with 2 N HCl. FFAs were obtained and finally separated and measured according to a previous described method (27). Recovery of glycerol, measured enzymatically with a fluorimeter (28) in triplicate from a tripalmitoyl-glycerol standard (Sigma Chemical Co. Ltd., Milan, Italy), was 98 ± 3% (mean ± SD), and that from a tripalmitoyl-glycerol standard added to skeletal muscle cell samples was 95 ± 5%.

Blood Chemistry

Plasma glucose levels were measured by a glucose-oxidase method (Beckman, Fullerton, CA). Serum immunoreactive insulin was assayed by using microparticle enzyme immunoassay (Abbott, Pasadena, CA). Serum FFAs and TGs were measured by enzymatic, colorimetric methods.

Data Analysis

Values are given as means ± SE. The nonparametric Wilcoxon test for paired values was used for comparisons before and after BPD. p < 0.05 was the threshold of significance. The Spearman correlation coefficients were calculated for the estimates of the level of association between two variables.


Table 1 shows the anthropometric variables before and after BPD treatment. The average weight loss observed over a 3-year period was ∼42%, depending mainly on a significant decrease of fat mass (∼ 65%) accompanied by a good preservation of FFM.

Table 1.  Anthropometric characteristics of the subjects
CharacteristicBefore BPDAfter BPDp
  1. Values are expressed as mean ± SD.

Height (cm)169.8 ± 5.8169.8 ± 8
Weight (kg)128.8 ± 26.875.1 ± 7.7<0.005
BMI (kg/m2)44.6 ± 8.226.0 ± 2.1<0.005
FFM (kg)67.8 ± 17.753.6 ± 10.40.07
Fat mass (FM) (kg)61.0 ± 13.921.5 ± 5.05<0.005

Fasting plasma insulin concentration was 130.80 ± 22.62 pM before the surgical procedure and significantly (p < 0.01) decreased down to 55.75 ± 4.68 pM after surgery. EHC showed that whole-body glucose uptake (M), normalized by FFM, significantly increased to 4-fold (13.59 ± 2.42 vs. 48.04 ± 4.73 μmol/kg FFM per minute; p < 0.0001). The analysis of intramuscular TGs showed markedly higher levels in obese subjects before BPD treatment compared with values after surgery (3.9 ± 1.1 vs. 0.9 ± 0.3 mg/100 mg of fresh muscle cells; p < 0.005). The levels of mRNA coding for PDK4, expressed as a ratio with the internal control aldolase-A mRNA [relative amounts (RAs)], significantly decreased from 2.02 ± 0.7 to 0.95 ± 0.4 RA (p = 0.01) (Figure 1). Figure 2 shows the individual data of PDK4 mRNA levels (Figure 2A), M values (Figure 2B), and the content of intramyocytic TGs (Figure 2C) in obese subjects before and 3 years after the bariatric surgery. Comparison between PDK4 mRNA and glucose uptake variations showed a negative linear correlation (y = −0.2319x + 13.842, R2 = 0.7127, p = 0.001) (Figure 3). Changes in PDK4 mRNA expression positively correlate with the changes of TG content in skeletal muscle biopsies (y = 1.5714x + 66.5, R2 = 0.8757, p = 0.005) (Figure 4). The reduction in TG muscle content was related to the increase of glucose tissue uptake (y = −0.1416x +35.112, R2 = 0.7492, p = 0.0006) (Figure 5).

Figure 1.

PDK4 expression before and after BPD in skeletal muscle biopsies. (A) Representative agarose gel showing RT-PCR analysis of PDK4 and aldolase-A mRNA content in muscle sample of one obese subject before and after BPD treatment (M, marker; and C, negative control). (B) Densitometric analysis of the ratio PDK4:aldolase-A mRNA abundance before (dotted bar) and after (striped bar) BPD. Bars represent the mean ± SD of six subjects.

Figure 2.

Individual variations in RAs of PDK4 mRNA (A), M values (B), and intramyocytic TGs (C) in skeletal muscle biopsies before and after BPD.

Figure 3.

Correlation (y = −0.2319x +13, 842, R2 = 0.7127, p < 0.001) between the variation of PDK4 mRNA and insulin sensitivity (M values) in muscle of obese patients before and after BPD treatment.

Figure 4.

Correlation (y = 1.5714x +66.5, R2 = 0.8757, p = 0.005) between the variation of PDK4 mRNA and intramyocytic TGs (milligram per 100 milligrams of fresh tissue) in obese patients before and after BPD treatment.

Figure 5.

Correlation (y = −0.1416x − 35.112, R2 = 0.7492, p = 0.0006) between insulin sensitivity and intramyocytic TGs in muscle of obese patients before and after BPD treatment.


The present study shows for the first time that in obese subjects after malabsorptive bariatric surgery, skeletal muscle PDK4 mRNA levels significantly decrease. According to the current literature (6, 7, 8, 9, 18), the increased glucose disposal rates were associated with a lower level of TGs in the muscle tissue. Interestingly, the reduction in PDK4 expression significantly correlates with the marked decrease of intramuscular TG concentration.

The PDH complex (PDHc) is the major determinant of glucose oxidation, catalyzing the oxidative decarboxylation of pyruvate to acetyl-CoA. PDKs (PDK 1 to 4) are a family of kinases that cover an important role in regulating PDHc activity (10) together with PDH phosphatases (29), respectively catalyzing phosphorylation/dephosphorylation of the PDH complex. Starvation induces a stable increase in PDK activity in skeletal muscle that promotes phosphorylation and inactivation of the PDH complex. Part of this PDK activation is likely the result of increased availability and, consequently, of enhanced oxidation of fatty acids and ketone bodies (30, 31, 32). Moreover, a long-term mechanism of muscle PDK activation in starved states is represented by an increased PDK4 protein expression (13). It is reasonable to hypothesize that in obesity, a condition in which there is an elevated oxidation of lipids, IR could be related, to some extent, to an increased expression of PDK.

In nondiabetic Pima Indians, a population with a high prevalence of type 2 diabetes associated with obesity, the expression of PDK2 and PDK4 mRNA in muscle biopsies correlates positively with fasting plasma insulin, a marker for IR, and negatively with insulin-mediated glucose uptake (16). This result was confirmed in our patients, in whom insulin sensitivity was inversely related to PDK4 mRNA amount in muscle biopsies.

In animal models, PDK isoform 4 is up-regulated by starvation (13, 14, 33, 34). Using immunoblot analysis, increased protein expression of PDK4 has been demonstrated in response to starvation in rat pancreatic islet, whereas expression of PDK1 and PDK2 isoforms was suppressed (35). FFAs, whose oxidation is enhanced during starvation, promote PDK4 expression through the activation of peroxisome proliferator-activated receptor alpha (PPARα) (13). This conclusion is supported by the findings that WY-14, 643, a synthetic PPARα activator, increases PDK4 expression in wild-type mice but not in PPARα-null mice. Likewise, starvation increases the expression of PDK4 in tissues of wild-type mice but not in tissues of PPARα-null mice (36).

Like starvation, diabetes enhances phosphorylation of the PDH complex and, therefore, lowers its activity (13, 33). In experimental animals, this effect is at least in part dependent on increased PDK activity, which in turn is attributable to increased PDK4 expression in heart (33), skeletal muscle (13, 14), liver, kidney, and lactating mammary gland (34).

Starvation and diabetes are conditions marked by high levels of FFAs, which may enhance skeletal muscle PDK4 expression by way of PPARα activation. Similarly, high-fat feeding is responsible for a dramatic increase in PDK4 expression (15, 37). The inactivation of the PDH complex due to its phosphorylation by increased PDK4 activity may explain a defective glucose metabolism and an IR status, which are present in a series of diseases, such as obesity, type 2 diabetes, hyperthyroidism, and hypertension, conditions characterized by abnormality in FFA levels.

In our patients, the expression of PDK4 mRNA reflects the availability of lipid substrates: the higher the amount of the transcript, the higher the concentration of intramuscular TGs. Lack of data related to PDK4 protein measurement represents a limit to our study. In fact, we are aware that both transcriptional and translational mechanisms are important in the regulation of PDK4 expression in many tissues. However, if altered PDK4 expression was responsible for changes in PDK4 activity, then an impaired glucose disposal in obesity should be dependent, at least in part, on diminished complex activity due to increased phosphorylation state. After biliopancreatic diversion, when a rapid decline in intramyocytic TG occurs, PDK4 mRNA decreases and insulin sensitivity increases as shown by higher M values.

In conclusion, our study supports the idea that a specific nutritional state, in this case represented by a massive lipid malabsorption that translates into a reduced tissue availability of fatty acids, influences glucose metabolism acting through the regulation of the PDH complex, which catalyzes the rate-limiting step in glucose oxidation. As shown in animals, a higher flux of FFAs is likely to induce an overexpression of PDK4 in humans also. The inhibition of PDHc activity is a consequence of increased expression of PDK4 that leads to the phosphorylation of this enzyme.

We hypothesize that the high intramyocytic TG content in the muscle tissue of morbidly obese patients could account for stimulation of PDK4 mRNA expression, presumably acting through the activation of PPARα receptor by fatty acids or their metabolites, and that this effect could be reversed after BPD when muscular TG concentration is reduced.


This study was supported by Ministero dell'Intruzione, dell'Universitá e della Ricerca Grant 7020509.


  • 1

    Nonstandard abbreviations: IR, insulin resistance; FFA, free fatty acid; TG, triglyceride; CoA, coenzyme A; PDH, pyruvate dehydrogenase; PDK, pyruvate dehydrogenase kinase; BPD, bilio-pancreatic diversion; EHC, euglycemic-hyperinsulinemic clamp; FFM, fat free mass; RT-PCR, reverse transcriptase-polymerase chain reaction; RA, relative amount; PDHc, pyruvate dehydrogenase complex; PPARα, peroxisome proliferator-activated receptor alpha.