Role of pyruvate dehydrogenase kinase isoenzyme 4 (PDHK4) in glucose homoeostasis during starvation

Materials

A BAC (bacterial artificial chromosome) clone (no. 19893) containing the mouse PDHK4 gene was purchased from Incyte Genomics (St. Louis, MO, U.S.A.). C57BL/6J black mice for backcrossing were purchased from The Jackson Laboratory (Bar Harbor, ME, U.S.A.). Enzymes and reagents were purchased from Promega (Madison, WI, U.S.A.) and Sigma (St. Louis, MO, U.S.A.). Ultra-sensitive mouse insulin ELISA kit was obtained from Mercodia (Winston Salem, NC, U.S.A.). All radiochemicals were purchased from Amersham Biosciences (Piscataway, NJ, U.S.A.).

Targeting of the mouse PDHK4 gene

To construct the targeting vector (Figure 1A), a PCR-amplified 0.9 kb DNA fragment located upstream of exon 1 of the PDHK4 gene was inserted into the SalI site of a pBluescript II/neo-TK vector [13] that contains the phosphoglycerate kinase promoter driving the neomycin-resistant gene (PGK-neo) and the herpes virus thymidine kinase gene. A 7.5 kb DNA fragment containing a part of exon 2 and extending through intron 7 of the gene was inserted at the XhoI site. The vector was linearized at NotI and used for electroporation into R1 embryonic stem cells (129SvJ origin) by the Transgenic Facility of the Duke Comprehensive Cancer Center (Durham, NC, U.S.A.). The disrupted allele was confirmed by Southern-blot analysis with a radiolabelled PCR-amplified DNA fragment from exon 8 to exon 10. Genotyping of mice was performed by PCR (40 cycles at 95 °C for 30 s, 57 °C for 30 s and 65 °C for 4 min). Primer 1 (5′-gactttcataacacccagtctcc-3′) and primer 2 (5′-cgcttttctggattcatcgactgtggc-3′) were used to amplify a 1.4 kb fragment from the targeted allele. Primers 1 and 3 (5′-ggtgctcgagcctgggtgaagg-3′) produced a 1.2 kb fragment from the wild-type allele. To stabilize the genetic background, heterozygous mice were backcrossed with C57BL/6J wild-type mice for six generations.

Design of experiments

Mice were housed in an AALAC (Association for Assessment and Accreditation of Laboratory Animal Care)-approved pathogen-free barrier facility (12 h light/dark cycles with temperature maintained at 23±2 °C) and fed ad libitum a standard rodent chow diet (Harlan; no. 7071). Studies were conducted with the approval of Institutional Animal Care and Use Committee of Indiana University School of Medicine. Tail blood was collected for glucose determination. To compare blood and tissue metabolite levels of fed and starved mice, groups of male mice [wild-type and PDHK4^−/− mice (homozygous PDHK4 knockout mice)] were either fed a chow diet or deprived of food for up to 48 h. Mice were killed by injection of Nembutal. Liver, heart, kidney, diaphragm and gastrocnemius muscle were freeze-clamped at the temperature of liquid nitrogen.

Measurement of metabolite concentrations in blood and liver

Glucose and lactate were measured in tail blood by an Accu-Chek advantage glucometer (Roche, Indianapolis, IN, U.S.A.) and a Lactate Prolactate test meter (Arkray, Shiga, Japan) respectively. Acetoacetate and β-hydroxybutyrate [14], pyruvate [15] and BCAAs (branched chain amino acids) [16] were assayed by enzymatic methods with deproteinized serum. Serum triacylglycerol and non-esterified fatty acids were determined enzymatically with the INFINITY triacylglycerol reagent (Sigma; 343-25P) and the NEFA (non-esterified fatty acid) Half Micro Assay (Roche) respectively. Glycogen was measured by a colorimetric assay [17]. Metabolites were measured in HClO₄ extracts of liver by enzymatic methods [18].

Glucose and insulin tolerance tests

Overnight-fasted mice were injected intraperitoneally with either a sterile solution of glucose (2 g/kg of body mass) or human recombinant insulin (0.75 unit/kg of body mass; Eli Lilly, Indianapolis, IN, U.S.A.). Tail blood was collected as a function of time for glucose determination.

Measurement of PDC activity

For the determination of actual PDC activity, pulverized tissue was homogenized in 5 vol. (w/v) of extraction buffer [19] containing 30 mM Hepes/KOH (pH 7.5), 0.5 mM thiamin pyrophosphate, 3% (v/v) Triton X-100, 5 mM EDTA, 2% (v/v) bovine serum, 5 mM DTT (dithiothreitol), 10 μM Tos-Phe-CH₂Cl (tosylphenylalanylchloromethane; ‘TPCK’), 10 μg/ml trypsin inhibitor, 1 μM leupeptin, 2 mM DCA (dichloroacetate) and 50 mM KF. The supernatant obtained by centrifugation at 10000 g for 10 min at 4 °C was made 9% (w/v) in PEG [poly(ethylene glycol)] 6000 to precipitate PDC. Pellets produced by centrifugation at 12000 g for 10 min were suspended in a suspension buffer containing 30 mM Hepes/KOH (pH 7.5), 1% Triton X-100, 0.2 mM EDTA, 2% bovine serum, 1 μM leupeptin and 5 mM DTT. For the determination of total PDC activity, tissues were homogenized in 5 vol. of the extraction buffer described above but without DCA and KF. The supernatant (200 μl) obtained by centrifugation of the homogenate at 10000 g for 10 min at 4 °C was added to 100 μl of an activation buffer (suspension buffer containing 25 mM MgCl₂, 1.5 mM CaCl₂ and 1 μg of recombinant PDP1 protein). After incubation for 20 min at 30 °C to dephosphorylate and activate PDC, samples were treated with PEG 6000 (final 9%) to precipitate PDC. An aliquot of an extract of the pellet was used to assay total PDC activity. PDC activity was measured spectrophotometrically in a 96-well plate reader (Spectra Max 190; Molecular Devices, Sunnyvale, CA, U.S.A.) with a coupled assay based on the reaction catalysed by arylamine acetyltransferase [19,20]. One unit of PDC activity corresponds to the acetylation of 1 μmol of p-(p-aminophenylazo)benzenesulphonate per min at 30 °C.

Western-blot analysis

Tissue powders (50 mg) prepared under liquid nitrogen were homogenized with 0.5 ml of extraction buffer (10 mM Tris/HCl, pH 7.5, 1 mM EDTA, 0.5% Triton X-100, 2% bovine serum, 1 μM leupeptin, 10 μM Tos-Phe-CH₂Cl and 10 μg/ml trypsin inhibitor). Protein concentrations of tissue extracts were determined with a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA, U.S.A.) using BSA as the standard. Proteins (50 μg) were separated by SDS/PAGE (12.5% gel) [21] and transferred on to a nitrocellulose membrane (Schleicher and Schuell, Keene, NH, U.S.A.) by the semi-dry electroblotting method. Western-blot analysis was carried out as described previously [22] using an enhanced chemiluminescence immunodetection system (Roche Diagnostics). Immunoblotting was performed with polyclonal rabbit antisera against mouse PDHK4 peptide (HQENRPSLTPVEAT) and mouse PDHK2 peptide (EIQEVNATNANQPIH) custom-produced by Sigma-Genosys (The Woodlands, TX, U.S.A.).

Glucose utilization and fatty acid oxidation by isolated diaphragms

Diaphragms were removed from 48 h starved mice, rinsed in Krebs–Henseleit bicarbonate buffer [23], blotted, weighed and placed in 10 ml Erlenmeyer flasks containing 1.5 ml of Krebs–Henseleit bicarbonate buffer (pH 7.4), 5 mM glucose and 0.2% (w/v) BSA. The flasks were flushed with 95% O₂/5% CO₂, sealed with rubber stoppers, placed in a shaking (60 cycles/min) water bath at 37 °C and pre-incubated for 30 min. Diaphragms were removed from the flasks, blotted and transferred on to new flasks containing 1.5 ml of Krebs–Henseleit bicarbonate buffer supplemented with 5 mM glucose containing 20 μCi/mmol [U-¹⁴C]glucose and 80 μCi/mmol of [5-³H]glucose and 1 m-unit/ml insulin (Novo-Nordisk, Princeton, NJ, U.S.A.) as described by Clark et al. [24]. Flasks were flushed with 95% O₂/5% CO₂, sealed with rubber serum caps fitted with hanging centre wells (Kontes, Vineland, NJ, U.S.A.) and incubated for 1 h with shaking at 37 °C. Reactions were terminated by the injection of 0.25 ml of phenethylamine/methanol (1:1, v/v) into the centre wells and 0.1 ml of 60% (w/v) HClO₄ into the contents of the flask. The rate of glucose oxidation was determined from the production of ¹⁴CO₂; the rate of glycolysis was determined from the difference between the rate of ³H₂O formed [25] and the rate of substrate recycling; the rate of substrate recycling was determined from the difference between the rates of glycogen synthesis from [U-¹⁴C]glucose and [5-³H]glucose [26]. Glycogen was purified [17] to determine the amount of glycogen synthesized from [U-¹⁴C]glucose and [5-³H]glucose incorporation.

To measure fatty acid oxidation, diaphragms obtained from 48 h starved mice were rinsed, weighed and pre-incubated for 30 min in Erlenmeyer flasks as described above. Diaphragms were removed from the flasks, blotted and transferred to new flasks containing 1.5 ml of Krebs–Henseleit bicarbonate buffer (pH 7.4), 5 mM glucose, 1 m-unit/ml insulin, 0.15 mM palmitate containing 500 μCi/mmol [1-¹⁴C]palmitate and 0.2% BSA. ¹⁴CO₂ was collected after 1 h incubation as described above. Acid-soluble radioactive products formed during the incubation were measured in aliquots (100 μl) of the HClO₄-treated medium.

Statistical analysis

The statistical significance of differences between groups was determined with Student's t test. All values are presented as means±S.E.M. for the indicated number of independent samples.

Generation of PDHK4^−/− mice

A null mutation of the PDHK4 gene was generated by exchange of exon 1 and a part of exon 2 with a neomycin gene through homologous recombination (Figure 1A). Genotyping of progeny mice was determined by PCR with DNA prepared from toe clips (Figure 1B). PDHK4 protein could not be detected by Western-blot analysis in skeletal-muscle extracts of 24 h fasted PDHK4^−/− mice (Figure 1C). The amount of PDHK4 protein expressed in skeletal muscle of heterozygotic (PDHK4^+/−) mice was approximately half of that of wild-type mice.

Characteristics of PDHK4^−/− mice

The genotype distribution (56:87:45) of PDHK4^+/+/PDHK4^+/−/PDHK4^−/− progeny of heterozygous mating was close to the 1:2:1 ratio expected for Mendelian inheritance. Body masses at weaning (results not shown) and at 9 weeks were not different between wild-type and PDHK4^−/− mice (24.0±0.3 and 23.6±0.2 g respectively; means±S.E.M.; n=20 in each group). Food consumption measured over 6 days was not different for 8–9-week-old mice (3.6±0.1 and 3.4±0.2 g·day⁻¹·mouse⁻¹ respectively; means±S.E.M.; n=10 in each group). Liver, heart, kidney, epididymal fat pad and brain weights also were not significantly different (results not shown).

Effects of PDHK4 deficiency on blood analyte levels in the fed state

No difference was found in blood glucose levels between fed PDHK4^−/− and wild-type mice. For example, blood glucose levels in wild-type mice were 140±8 mg/dl versus 133±10 mg/dl in PDHK4^−/− mice (means±S.E.M.; n=20 in each group). Levels of blood lactate, non-esterified fatty acids, triacylglycerol, β-hydroxybutyrate, acetoacetate and insulin were likewise not significantly different between PDHK4^−/− and wild-type in the fed state (results not shown).

Effects of PDHK4 deficiency on blood glucose levels after overnight fasting

Blood glucose levels were invariably lower in overnight-fasted PDHK4^−/− mice (45±1 mg/dl) than in overnight-fasted wild-type mice (64±3 mg/dl; means±S.E.M.; n=12 in each group; P<0.0001).

Effect of PDHK4 deficiency on blood analyte levels in the starved state

Blood glucose levels of PDHK4^−/− mice were 30% lower than that of wild-type mice after starvation for 24 h (Figure 2). A rebound back to a higher level of blood glucose occurred between 24 and 48 h of starvation in wild-type mice but not in PDHK4^−/− mice (Figure 2). Blood glucose levels of PDHK4^−/− mice were 58% lower than that of wild-type mice after starvation for 48 h (Table 1).

Serum concentrations of lactate, pyruvate and alanine in PDHK4^−/− mice (Table 1) were also lower in PDHK4^−/− mice, consistent with either a higher rate of oxidation or a reduced rate of production of these compounds in PDHK4^−/− mice. In contrast, higher serum concentrations of non-esterified fatty acids, acetoacetate, β-hydroxybutyrate, BCAAs and triacylglycerol were induced by starvation in PDHK4^−/− mice relative to wild-type mice (Table 1). A small increase in serum concentration of insulin was also observed (Table 1).

Glycogen levels in the livers of PDHK4^−/− and wild-type mice were high and almost identical in the fed state (Figure 2). However, more glycogen was lost from the livers of PDHK4^−/− mice during 24 h of starvation, and analogous to what was observed for blood glucose, less rebound in the amount of liver glycogen occurred between 24 and 48 h of starvation in mice lacking PDHK4 (Figure 2). Skeletal-muscle glycogen levels of PDHK4^−/− mice and wild-type mice were also the same in the fed state (20.3±1.6 versus 20.6±1.5 μmol of glucose/g wet weight; means±S.E.M.; n=6 in each group). Starvation for 48 h reduced skeletal-muscle glycogen levels to the same extent (40%) in both groups of mice (results not shown).

Effect of PDHK4 deficiency on glucose tolerance test and blood insulin levels

PDHK4^−/− mice were slightly but significantly more glucose-tolerant than PDHK4^+/+ mice (Figure 3). Serum insulin concentrations measured during the glucose tolerance test did not differ in the two groups of mice (Figure 3), suggesting that PDHK4^−/− mice may be slightly more sensitive to insulin than PDHK4^+/+ mice. No difference in insulin sensitivity between the two groups of mice was detected, however, by intraperitoneal insulin tolerance test (results not shown).

Effect of PDHK4 deficiency on metabolic intermediates in the liver in the starved state

The livers of starved PDHK4^−/− mice were found to contain 42% less free glucose than livers of starved wild-type mice (Table 2). Concentrations of intermediates of the gluconeogenesis pathway were likewise lower (glucose 6-phosphate, fructose 1,6-bisphosphate, pyruvate, lactate and citrate) in the livers of PDHK4^−/− mice or not different (dihydroxyacetone phosphate, glyceraldehyde 3-phosphate and phosphoenolpyruvate) compared with that of wild-type mice (Table 2). No significant differences were found in concentrations of malate, glutamate and alanine. In agreement with the difference found in the blood, ketone bodies were elevated and the ratio of [β-hydroxybutyrate] to [acetoacetate] was more oxidized in the liver of PDHK4^−/− mice. Acetyl-CoA was lower in the liver of PDHK4^−/− mice. Lactate and pyruvate were both lower in concentration without a significant change in their ratio in the livers of PDHK4^−/− mice.

Effect of PDHK4 deficiency on rates of glucose oxidation and glycolysis by diaphragms from starved mice

Glucose oxidation rate, measured by ¹⁴CO₂ production from [U-¹⁴C]glucose, was 36% greater in diaphragms from PDHK4^−/− mice relative to diaphragms from wild-type mice (Table 3). However, rates of lactate release and net glycolysis, measured by ³H₂O production from [5-³H]glucose, were 37 and 35% less respectively in diaphragms from PDHK4^−/− mice. This was accompanied by a trend for a greater rate of glycogen synthesis in diaphragms from PDHK4^−/− mice that did not reach statistical significance (Table 3).

Effect of PDHK4 deficiency on rates of fatty acid oxidation by the diaphragms from starved mice

Rates of fatty acid oxidation, measured either by ¹⁴CO₂ production or by acid-soluble product formation from [1-¹⁴C]palmitate, or by the sum of the two, were approximately 50% lower in diaphragms obtained from PDHK4^−/− mice than in diaphragms from wild-type mice (Table 4).

Effects of PDHK4 deficiency on PDC activity

Since PDHK4 expression is relatively low in tissues of fed rats [11], lack of PDHK4 was expected to have little or no effect on actual PDC activities (measured as extracted from the tissue without activation by dephosphorylation) and PDC activity states (percentage of total PDC that is active) in tissues of mice in the fed state. Indeed, this was found to be the case for kidney and the gastrocnemius muscle (Table 5). However, a significantly greater actual PDC activity and PDC activity state were found in the diaphragms of fed PDHK4^−/− mice. Trends towards greater actual PDC activities and PDC activity states that did not make statistical significance were also found for the heart and liver. These findings suggest that PDHK4 contributes to the PDC activity state in the diaphragm and possibly the heart and liver in the fed state.

Since there is little data on PDC activity in the literature for the mouse, it is interesting to note that the highest total PDC activities were observed in the heart, whereas the lowest total activities were observed in the liver. The lowest activity states were found in the diaphragm and heart from starved mice, whereas the highest activity states were found in the diaphragm and liver of fed mice.

As expected from many studies with rats [7,9,11,22], starvation markedly decreased actual PDC activities in wild-type mice without significantly affecting total PDC activity, thereby decreasing the activity states of PDC to values typical of those induced in tissues of starved rats (Table 5). Since it is well established in studies with rats that starvation decreases the activity state of PDC and increases PDHK4 expression [11], we expected that the activity states of PDC would be maintained at higher levels during starvation in tissues of mice lacking PDHK4 than in the tissues of wild-type mice. To our surprise, the activity state of PDC was still markedly reduced by starvation in all tissues of PDHK4^−/− mice examined (Table 5). Tissue-specific expression of PDHK1 may contribute to the very low activity state induced in the heart by starvation. Nevertheless, a significantly higher PDC activity state was observed in the heart, muscle, kidney and diaphragm of PDHK4^−/− mice (Table 5), consistent with a role for PDHK4 in inactivation of PDC in these tissues. The differential in activity states between wild-type and PDHK4^−/− mice was the greatest for diaphragm, followed by kidney, skeletal muscle and then heart. No difference in PDC activity states was observed in the livers of wild-type and PDHK4^−/− mice. It is noteworthy that starvation did not induce a significant decrease in PDC activity state in gastrocnemius muscle of PDHK4^−/− mice, whereas it did in wild-type mice, suggesting that much of the decrease in PDC activity induced in skeletal muscle by starvation is caused by increased expression of PDHK4.

Effect of starvation on PDHK2 and PDHK4 expressions

The amount of PDHK4 protein in gastrocnemius muscle of wild-type mice was very low in the fed state, but markedly increased by starvation (Figures 4A and 4B). The amounts of PDHK2 protein were similar in gastrocnemius muscle in the two groups of mice in the fed state (Figure 4C). Starvation had no effect in wild-type mice but, surprisingly, caused a decrease in PDHK2 protein in PDHK4^−/− mice. PDHK4 protein in heart of wild-type mice was very low in the fed state, but markedly increased by starvation (Figures 5A and 5B). PDHK2 protein in heart was greater in PDHK4^−/− mice, perhaps partially compensating for the absence of PDHK4 (Figure 5C). Starvation induced an increase in PDHK2 protein of wild-type mice, but had no effect on the already elevated amount of PDHK2 protein in the PDHK4^−/− mice. The response of the kidney was similar to that of the heart (Figure 6). PDHK4 protein was markedly increased in response to starvation in the wild-type mice (Figures 6A and 6B). PDHK2 protein was greater in the heart of fed PDHK4^−/− mice, but in contrast with the heart, starvation caused an additional increment in PDHK2 protein in the PDHK4^−/− mice as well as the PDHK4^+/+ mice (Figure 6C). The response of the liver to starvation was qualitatively the same as the other tissues (Figure 7). Starvation also increased PDHK4 protein in the liver of wild-type mice (Figures 7A and 7B). PDHK2 protein was similar in amount in the livers of fed PDHK4^+/+ mice and PDHK4^−/− mice (Figure 7C). Starvation increased the amounts of PDHK2 protein in both groups of mice. The amount of PDHK4 protein in the diaphragm of fed wild-type mice was low but detectable by Western-blot analysis (Figure 8A). Starvation also markedly increased the amount of PDHK4 in the diaphragm of wild-type mice (Figure 8B). In contrast with the other tissues, similar amounts of PDHK2 protein were found in the diaphragms of both types of mice and starvation had no effect (Figure 8C).

Why are blood glucose levels lower in PDHK4^−/− mice in the starved state?

PDHK4 deficiency induces effects similar to those induced by the PDHK inhibitor DCA [30]. PDHK4 deficiency and treatment with DCA [31–33] lower blood glucose levels in starved rats but not in fed rats. PDHK4 deficiency and treatment of starved rats with DCA [31,33] lower blood levels of gluconeogenic precursors. Studies with isolated hepatocytes [34] and intact dogs [35] have shown the importance of substrate supply for hepatic gluconeogenesis. Gluconeogenesis normally operates below maximal capacity at the concentrations of the precursors present in the blood [34]. Induction of a lower concentration of precursors as a consequence of higher PDC activity further limits gluconeogenesis by the liver. Consistent with this mechanism, PDHK4 deficiency and DCA [31] cause either no change or a decrease in steady-state concentrations of the intermediates of gluconeogenesis in the liver. Lack of a crossover point in the intermediates of gluconeogenesis suggests a limitation of substrate supply rather than a specific site of inhibition.

Why are gluconeogenic precursors reduced in the blood of PDHK4^−/− mice?

DCA decreases the concentration of lactate, pyruvate and alanine in the blood by stimulation of pyruvate oxidation [31,36–38] and inhibition of glycolysis [39] in peripheral tissues. Decreased pyruvate availability in muscle inhibits transamination of BCAAs by limiting the availability of α-oxoglutarate [40], resulting in an elevation in BCAAs and a reduction in alanine in DCA-treated animals [41]. PDHK4 deficiency likewise resulted in lower than normal levels of alanine and higher than normal levels of BCAAs in the blood. Studies with isolated diaphragms are also consistent with a DCA-like mechanism for PDHK4^−/− mice. Diaphragms from PDHK4^−/− mice produce less lactate, oxidize more glucose, but utilize less glucose by the glycolytic pathway. These are the same effects produced by incubation of skeletal muscle with DCA [24]. As found for DCA, inhibition of glycolysis accounts for most of the decrease in lactate production by diaphragms from PDHK4^−/− mice. Generation of ATP by oxidation of glucose means less has to be generated by anaerobic metabolism of glucose, but like inhibition of glycolysis by DCA [39], the exact mechanism by which greater ATP production by oxidative metabolism results in less ATP production by glycolysis remains uncertain.

Why are the levels of ketone bodies greater in the blood of starved PDHK4^−/− mice?

Starvation induces much greater increases in non-esterified fatty acids and ketone bodies in the blood of PDHK4^−/− mice than in wild-type mice. Whether high ketone bodies levels in starved PDHK4^−/− mice are caused by greater hepatic ketogenesis or less ketone bodies utilization by peripheral tissues of PDHK4^−/− mice is not known. Perhaps both are involved. Elevated non-esterified fatty acids should promote fatty acid oxidation and the reduced supply of three-carbon compounds should limit oxaloacetate availability in the liver and direct acetyl-CoA into ketone body formation. However, acetyl-CoA levels would be expected to be increased by a greater rate of fatty acid oxidation and reduced flux into the tricarboxylic acid cycle. The opposite was found: acetyl-CoA levels are lower in the liver of PDHK4^−/− mice than in the liver of wild-type mice. Furthermore, the ratio of β-hydroxybutyrate to acetoacetate was found lower in the blood of PDHK4^−/− mice than in the blood of wild-type mice. Since a greater rate of ketone body production by the liver would be expected to increase rather than decrease the β-hydroxybutyrate/acetoacetate ratio [31], tissues other than the liver must be involved in increasing the ketone body concentration in starved PDHK4^−/− mice. Therefore inhibition of fatty acid and ketone body oxidation as a consequence of greater PDC activity in peripheral tissues likely contributes to the higher blood levels of fatty acids and ketone bodies. Indeed, diaphragms from PDHK4^−/− mice oxidize palmitate at a lower rate than diaphragms from wild-type mice. A difference in sensitivity of acetoacetate oxidation versus β-hydroxybutyrate oxidation may account for the lower β-hydroxybutyrate/acetoacetate ratio. Why greater PDC activity induces these effects has not been explained in a satisfactory way. Competition between pathways for CoA has been suggested [36] and is currently being investigated as a possible cause.