Enhanced insulin sensitivity of gene-targeted mice lacking functional KCNQ1

METHODS

All animal experiments were conducted according to the German law for the welfare of animals and were approved by Regierungspräsidium Tübingen.

Experiments were performed in mice deficient in KCNQ1 (kcnq1^−/−) and their wild-type littermates (kcnq1^+/+) generated as described earlier (10). For the study, 4- to 10-mo-old KCNQ1 knockout mice (kcnq1^−/−; 2–8 males, 2–7 females) and their wild-type littermates (kcnq1^+/+; 3–8 males, 3–9 females) were selected. Prior to the experiment, the age- and sex-matched mice were fed with control diet (cat. no. 1310/1314; Altromin, Heidenau, Germany) and were allowed free access to tap water.

To study the cellular K⁺ uptake, livers of 14- to 20-wk-old male and female mice with a body weight of 22–28 g (kcnq1^−/−: 22–26 g, kcnq1^+/+ 22–28 g) were perfused in situ, essentially as described previously for the rat, in an open nonrecirculating manner (54), with bicarbonate-buffered Krebs-Henseleit saline (in mM: 143.63 Na⁺, 5.9 K⁺, 1.18 Mg²⁺, 1.25 Ca²⁺, 125.76 Cl⁻, 25 HCO₃⁻, 1.23 H₂PO₄⁻, 1.2 SO₄) as perfusion medium plus lactate (2.1 mM) and pyruvate (0.3 mM), gassed with O₂/CO₂ (19:1 vol/vol) at 37°C. The average flow rate was kept constant throughout each perfusion experiment at 4–5 ml·min⁻¹·g liver⁻¹. Where indicated, insulin (100 nM) was added to the Krebs-Henseleit buffer. The effluent K⁺ concentration was registered continuously with K⁺-selective electrodes (Radiometer, Munich, Germany) and calculations were made by planimetry. The intracellular water space in the same perfused livers was calculated from the difference of washout profiles of simultaneously infused [¹⁴C]urea and [³H]inulin (∼500 Bq/ml perfusate), as described previously for the rat liver (54). In brief, both isotopes were infused for 5 min, to allow equilibration of both tracers within the intracellular and extracellular compartments. Tracer infusion was then stopped, and the effluent perfusate was collected for another 5 min, a time period allowing complete washout of both isotopes. The effluent was monitored for the total sum of released [³H] and [¹⁴C] label by dual spectrum scintillation spectrometry. Taking the [³H]-to-[¹⁴C] ratio under steady-state conditions, a putative ¹⁴C-accessible, but not ³H-accessible space was calculated. This space, expressed as intracellular water space (μl/g liver) was presumed to reflect the cellular hydration state. The cellular hydration state was measured under control conditions and 30 min after administration of insulin. A spontaneous decrease of the intracellular water space of 2.1% within 30 min in fed mice was considered. When present, a baseline drift was taken into account.

For the determination of the glucose tolerance, mice were fasted overnight and glucose (3 g/kg body wt) was injected intraperitoneally. A drop of blood was then drawn via the tail vein and read by a glucometer (Accutrend; Roche, Mannheim, Germany). Blood glucose levels were measured before and 15, 30, 45, 60, 75, 90, 120, 150, and 180 min after the injection.

In a further series of experiments, long-acting insulin (Novo Nordisk, Mainz, Germany) was injected intraperitoneally (0.15 U/kg body wt), and plasma glucose concentrations were determined 0, 15, 30, 45, 60, and 90 min postinjection, as described above.

For the determination of fasted blood glucose levels, food was withheld overnight, from 7:00 PM until 9:00 AM the next morning, when blood samples were taken, along with blood samples from nonfasted controls. For determination of the plasma insulin concentrations, ∼80 μl of blood was drawn from overnight-fasted and nonfasted animals into heparinized hematocrit capillary tubes, by retroorbital puncture during light anesthesia with diethylether (Roth, Karlsruhe, Germany). Plasma insulin concentrations were determined using an ELISA, according to the manufacturer's instructions (Crystal Chem). The plasma glucagon concentration was determined in ∼200 μl of blood, using an ELISA according to the manufacturer's instructions (Abcam, Cambridge, UK).

The in vivo tissue glucose uptake during a glucose tolerance test was determined in equal numbers of male and female mice that were fasted for 16 h prior to the experiment. 2-deoxy-d-[1,2-³H]-glucose was mixed with 20% of regular glucose (10 μCi/mouse, 3 g/kg body wt) and injected intraperitoneally. Blood glucose levels were monitored to ensure adequate injection. After 120 min, the mice were killed, and 100 mg of tissues were homogenized in 1 ml of water. Then 800 μl of 7% ice-cold perchloric acid were added to 800 μl of homogenate. The sample was then cleared by centrifugation, and 1 ml of the supernatant was neutralized with 2.2 M KHCO₃. After 30-min incubation, the precipitate was removed by centrifugation and 500 μl of the supernatant were used to determine total ³H radioactivity.

Western blot analysis was performed to explore whether differences in liver GLUT1 protein abundance in kcnq1^+/+ and kcnq1^−/− mice accounted for the differences in glucose uptake between the genotypes. To this end, kcnq1^+/+ and kcnq1^−/− mice (n = 4 mice for each group) were anesthetized with diethylether. The livers were removed and immediately shock-frozen in liquid nitrogen. The liver tissues were then homogenized in lysis buffer containing 50 mM Tris·HCl, pH 7.4, 150 mM NaCl, 1 mM sodium fluoride, 1 mM sodium orthovanadate, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, and a protease inhibitor cocktail (Complete Mini; Roche, Mannheim, Germany). The homogenates were centrifuged at ∼20,000 g at 4°C for 15 min, and the supernatant was removed and used for Western blot analysis. Total protein (100 μg) was separated by SDS-PAGE (10% Tris-glycine), transferred to nitrocellulose membranes, blocked overnight in blocking buffer (5% fat-free milk in PBS containing 0.1% Tween) at 4°C, and incubated for 1 h with a polyclonal anti-GLUT1 antibody (diluted 1:1,000 in blocking buffer; Abcam). After incubation with a horseradish peroxidase-conjugated anti-rabbit secondary antibody (Amersham, Freiburg, Germany), the bands were visualized with enhanced chemiluminescence according to the manufacturer's instructions. Homogenates were also probed with a primary GAPDH (Santa Cruz Biotechnology, Santa Cruz, CA) antibody as a loading control. Densitometric analysis of GLUT1 was performed using Quantity One software (Bio-Rad Laboratories, Hercules, CA).

The extracellular space was determined in an additional series of experiments using a separate group of mice. To this end, FITC-inulin (Sigma, Taufkirchen, Germany) was dissolved at 70 mg/ml in 0.9% NaCl and solubilized by heating. Mice were injected retroorbitally (5 μl/g body wt) under light ether anesthesia. Syringes were weighed before and after injection to determine the exact injected volume. Approximately 8–10 μl of blood were collected from the tail vein 3, 7, 10, 15, 35, 55, and 75 min after injection (38). After centrifugation, 3 μl of plasma was aliquotted, combined with 97 μl of PBS, and loaded onto a 96-well plate. Fluorescence was determined with 485-nm excitation and 535-nm emission on a multilabel counter (Victor 1420; PerkinElmer). The inulin space was calculated, assuming a two-compartment model. According to this model, the rapid initial decay phase represents distribution of the inulin in the intravascular space, followed by a slower decay of the tracer, predominantly reflecting systemic clearance by the kidney. Data were fitted to a two-phase exponential decay curve using nonlinear regression (GraphPad Prism; GraphPad Software, San Diego, CA) according to the equation C(t) = A·e^−k1·t + B·e^−k2·t, where C(t) is the time-dependent concentration and k₁ and k₂ the decay constants. A and B reflect the imaginary initial concentration of the tracer in each respective compartment. Division of the injected dose (counts) by A and B (counts/μl) yielded the distribution volume of the two compartments, respectively. Both values were added to obtain the total distribution volume of inulin.

Data are provided as arithmetic means ± SE, n represents the number of independent experiments. All data were tested for significance using paired or unpaired Student t-test, as applicable. Only results with P < 0.05 were considered statistically significant.

RESULTS

To explore whether lack of KCNQ1 affects the insulin-induced cellular net K⁺ uptake, livers from KCNQ1 knockout mice (kcnq1^−/−) and their wild-type littermates (kcnq1^+/+) were perfused, and the K⁺ concentration in the effluent was continuously recorded. As illustrated in Fig. 1 A and B, the addition of insulin (100 nM) to the perfusate was followed by a significant transient decrease of the effluent K⁺ concentration, pointing to cellular K⁺ uptake. The effect of insulin (100 nM) was significantly more sustained in kcnq1^−/− than in kcnq1^+/+ mice. Accordingly, the amount of K⁺ taken up by hepatocytes following treatment with insulin was significantly larger in kcnq1^−/− than in kcnq1^+/+ mice (Fig. 1B). The stimulation of cellular K⁺ uptake by insulin was expected to be paralleled by cell volume increase. Thus, cell volume was determined by utilizing urea/inulin distribution. As shown in Fig. 1C, insulin (100 nM) indeed increased hepatocyte hydration. The cell swelling tended to be larger in kcnq1^−/− than in kcnq1^+/+ mice without reaching statistical significance. Additional experiments were performed to explore whether lack of KCNQ1 similarly affects K⁺ fluxes following cell swelling. Fig. 1D demonstrates that cell swelling, by a decrease of perfusate osmolarity (removal of 40 mM NaCl), was followed by K⁺ release, which was reflected by a transient increase in the K⁺ concentration in the perfusate. The K⁺ release was significantly larger in kcnq1^+/+ mice (9.4 ± 0.8 μM/g liver, n = 4) than in kcnq1^−/− mice (6.9 ± 0.3 μM/g liver, n = 3).

Further experiments were performed to elucidate whether the lack of KCNQ1 affects glucose metabolism. Under both fasting and nonfasting conditions, the plasma glucose concentration was significantly lower in kcnq1^−/− than in kcnq1^+/+ mice (Fig. 2).

To explore whether the differences in plasma glucose concentration were secondary to enhanced plasma insulin levels in kcnq1^−/− mice, the plasma insulin levels were assayed under fasted vs. nonfasted conditions. Under both conditions, the plasma insulin concentration was significantly lower in kcnq1^−/− than in kcnq1^+/+ mice (Fig. 2). These data indicate that enhanced insulin concentrations did not account for the lower plasma glucose concentrations in kcnq1^−/− mice.

The coincidence of decreased plasma glucose and decreased plasma insulin concentrations points to enhanced insulin sensitivity of cellular glucose uptake. To explore this possibility, insulin was injected intraperitoneally, and plasma glucose concentrations were determined. As illustrated in Fig. 3, the intraperitoneal injection of 0.15 U insulin/kg body wt resulted in a decreased plasma glucose concentration. The hypoglycemia was significantly more sustained in kcnq1^−/− than in kcnq1^+/+ mice, suggesting enhanced insulin sensitivity of cellular glucose uptake in kcnq1^−/− mice.

To further test for insulin sensitivity of cellular glucose uptake, an intraperitoneal glucose tolerance test was performed. Following an intraperitoneal glucose injection (3 g/kg body wt), the peak plasma glucose concentration was significantly lower in kcnq1^−/− than in kcnq1^+/+ mice (Fig. 4).

Next, the inulin space was determined to investigate whether the alterations of extracellular glucose concentration following a glucose load, or following injection of insulin, were partially due to differences in extracellular fluid volume. The inulin space amounted to 4.4 ± 0.6 ml (n = 4) in kcnq1^+/+ mice and to 3.8 ± 0.3 ml (n = 4) in kcnq1^−/− mice; these values were not significantly different.

Additional experiments were performed to define the tissues responsible for altered cellular glucose uptake. To this end, the uptake of radiolabeled ³H-deoxy-glucose into several tissues following intraperitoneal injection was determined. As a result, the ³H-deoxy-glucose uptake into liver, skeletal muscle, kidney and lung tissue was significantly higher in kcnq1^−/− than in kcnq1^+/+ mice (Fig. 5).

Finally, genotypic differences in GLUT1 expression were tested. The difference in hepatic glucose uptake between kcnq1^−/− and kcnq1^+/+ mice could not be explained by differences in GLUT1 expression, as GLUT1 protein levels were similar in kcnq1^−/− (0.60 ± 0.06 GLUT1/GAPDH, n = 4) and in kcnq1^+/+ (0.64 ± 0.08 GLUT1/GAPDH, n = 4) mice (Fig. 6). Furthermore, the differences in glucose metabolism could not be explained by differences in plasma glucagon concentration, because it tended to be higher in kcnq1^−/− (357 ± 19 ng/ml, n = 10) than in kcnq1^+/+ mice (302 ± 29 ng/ml, n = 11). These effects, however, did not reach statistical significance.

DISCUSSION

The observations in the present study indicate that KCNQ1 is a novel molecule affecting peripheral glucose uptake and insulin sensitivity of glucose metabolism. The uptake of ³H-deoxy-glucose into skeletal muscle, liver, kidney, and lung tissue was apparently higher in KCNQ1 knockout (kcnq1^−/−) mice than in their wild-type littermates (kcnq1^+/+). The enhanced insulin sensitivity leads to lower plasma glucose concentrations, which in turn inhibit insulin release and thus result in the decreased plasma insulin concentrations.

The organs particularly sensitive to a lack of KCNQ1 are the liver, skeletal muscle, kidney, and lung tissue. KCNQ1 expression was demonstrated in liver (12, 25, 26), skeletal muscle (15), and several epithelia, including kidney (11, 17, 22, 34, 47, 49, 51, 52). The intrarenal glucose concentration not only includes glucose that has been taken up by cells, but also glucose in the tubular fluid. Intrarenal glucose is therefore enhanced not only by increased cellular uptake, but also by the delayed glucose reabsorption. Usually, kcnq1^−/− mice do not suffer from glucosuria, because the transport capacity is sufficient for complete reabsorption of filtered glucose, despite the decreased electrical driving force. During a glucose tolerance test, however, the tubular transport system may be saturated, and the reduced driving force may thus result in glucosuria (52).

The liver perfusion experiments provide a clue to the link between KCNQ1 and glucose metabolism. Lack of K⁺ channels is expected to impair regulatory cell volume decrease and foster cell swelling (27). In theory, an increase in glucose uptake into the liver could be due to upregulation of the carrier protein or due to enhanced utilization of glucose for glycolysis and glycogen formation. According to our observations, hepatic GLUT1 expression is not different between kcnq1^−/− and kcnq1^+/+ mice. While this observation does not rule out differences in transport activity between the genotypes, cell swelling has previously been shown to stimulate hepatic glycogen formation and cell shrinkage to trigger glycogenolysis (5, 23, 29). The incorporation of glucose into glycogen maintains a driving force for passive glucose entry into the cell and thus indirectly increases cellular glucose uptake. Insulin has been shown to stimulate cellular K⁺ uptake by activation of a Na⁺/H⁺ exchanger, Na⁺-K⁺-2Cl⁻ cotransport, and Na⁺ K⁺/ATPase (19–21). The stimulation of cellular K⁺ uptake was shown to be required for the effect of the hormone on the regulation of metabolism (19–21) and on gene expression (43). Moreover, the K⁺ uptake may confer support for cell survival (7, 8, 16, 46, 48). To the extent that KCNQ1 counteracts the stimulation of cellular K⁺ uptake by insulin, the channel may blunt insulin sensitivity. Conversely, a lack of KCNQ1 would be expected to enhance insulin sensitivity.

At least in theory, KCNQ1 activity may further influence glucose metabolism by modulating intracellular pH. Decreased K⁺ channel activity leads to depolarization, which in turn decreases the electrical driving force for electrogenic bicarbonate exit and thus favors cellular HCO₃⁻ retention and intracellular alkalinisation (1, 28). Intracellular alkalinisation, in turn, stimulates flux through glycolysis and thus enhances glucose utilization (1, 4, 37, 39). As enhanced glucose consumption is paralleled by increased glucose uptake across the cell membrane, the alkalinisation could contribute to the enhanced insulin sensitivity of cellular glucose uptake.

KCNQ1 is important for a variety of crucial functions including cardiac rhythm (6, 33, 33, 41), hearing (10, 30), gastric acid secretion (30, 30, 42), as well as intestinal and renal transport (52). Most importantly, impaired cardiac KCNQ1 function jeopardizes the timely repolarization from cardiac action potential, and thus could lead to life-threatening cardiac arrhythmia. The present observations cannot rule out that indirect effects of KCNQ1 deficiency contribute to altered insulin sensitivity. For instance, the shaker phenotype may influence muscle activity, and thus insulin sensitivity of skeletal muscle.

Besides KCNQ1, other K⁺ channels have been shown to participate in the regulation of glucose metabolism (2, 14, 24, 31, 40). As shown previously, deletion of the ATP-sensitive potassium channels through ablation of the sur1, sur2, and Kir6.2 subunits modulates insulin secretion and glucose metabolism (2, 9, 13, 32, 35, 36). An enhanced insulin effect in Kir6.2-deficient mice has been attributed, in part, to decreased hepatic glucose production due to impaired glucagon secretion (40). However, in the present study, glucagon plasma concentration tended to be rather higher in kcnq1^−/− than in kcnq1^+/+ mice. Thus, impaired glucagon release does not account for the enhanced insulin sensitivity of KCNQ1-deficient mice. Obviously, several K⁺ channels participate in the regulation of glucose metabolism by influencing several distinct mechanisms.

PERSPECTIVE AND SIGNIFICANCE

The present observations indicate that insulin sensitivity is strikingly increased in the KCNQ1 knockout mouse, leading to enhanced cellular glucose uptake, decreased plasma glucose concentration, and reduced insulin release. Thus, KCNQ1 is a completely novel element in tuning of cellular glucose uptake. The significance of the present observations for human pathophysiology is illustrated by the observation that KCNQ1 polymorphisms are associated with the development of type 2 diabetes (50, 56). The observations further underscore the physiological and pathophysiological role of K⁺ transport and cell volume in the metabolic effects of insulin (19–21). Future studies may address the role of KCNQ1 polymorphisms in insulin-induced K⁺ fluxes.

GRANTS

This study was supported by the Deutsche Forschungsgemeinschaft (GK 1302) and Sonderforschungsbereich 575 Experimental Hepatology.