Central Nervous System Delivery of the Antipsychotic Olanzapine Induces Hepatic Insulin Resistance

Male Sprague-Dawley rats (250–275 g), purchased from Charles River Laboratories (Wilmington, MA), were acclimated to our facilities for 7 days before undergoing surgical catheterizations of the left jugular vein and right carotid artery, as previously described (21). A set of rats also received intracerebroventricular cannulae targeting the third ventricle (from Bregma-anterior-posterior: −2.2 mm; dorsal-ventral: −7.5 mm from sagital sinus surface; medial-lateral: 0.0 mm). The jugular catheter served for infusions and the carotid catheter for blood sampling. Rats were allowed to recover for at least 6–10 days after surgery before the clamp studies. The correct placement of the intracerebroventricular cannulae was verified by the induction of a drinking response to 30 ng angiotensin II intracerebroventricularly performed 4 days after the surgery. All animal procedures were approved by the institutional animal care and use committee of the University of Cincinnati.

The dose of intravenous infusion of OLZ (OLZ-IV group) was selected according to two criteria: 1) the extent of occupancy of the D2 receptors (80–90%) achieved with a single dose (14,22) and 2) it was experimentally determined to cause only moderate sedation (up to grade 2 according to the Sedation Rating Scale in Salamone et al. [ 23]). In addition, since the plasma half-life of OLZ in rats is eightfold shorter than humans (22,24), we used a primed-continuous infusion in order to study its effect on steady-steady condition and minimize pharmacokinetic fluctuations (24). A stock solution of OLZ (provided by Eli Lilly, Indianapolis, IN, and the National Institute of Mental Health Chemical Synthesis and Drug Supply Program) was prepared in HCl (0.1 N). Working solutions of OLZ were prepared daily by dissolving the stock solution in saline (0.9%) or artificial cerebrospinal fluid (Harvard Apparatus, Holliston, MA) for intravenous (3 ml) or intracerebroventricular (15 μl) infusions, respectively. Final solutions were adjusted to pH 5.5–6.0 with sodium hydroxide (0.5 N). Control animals received saline (0.9%) or artificial cerebrospinal fluid after pH adjustment to 5.5–6.0 as well.

Sixty minutes before the drug infusion (t = −60), a prime-continuous infusion of [³H-3]-glucose (Perkin-Elmer NEN, Boston, MA) was initiated and maintained throughout the study (prime 5.2 μCi/min for 4 min, then 0.52 μCi/min). At 0 min, a prime-continuous infusion of OLZ or vehicle was started and maintained throughout the study. A total of 4.5 mg/kg was delivered intravenously (prime 375 μg/kg/min for 4 min, then 25 μg/kg/min) and 330 μg/rat was delivered intracerebroventricularly (prime 22 μg/min for 5 min, then 1.9 μg/min). At +60 min, after collecting the last blood sample for the basal period, all rats received 1) a prime-continuous infusion of human insulin (prime 44.3 mU/kg for 1 min, then 2.5 mU/kg/min) and 2) a variable infusion of 25% glucose solution to maintain euglycemia. Blood samples for glucose concentration and specific activity were taken every 10 min during the basal period and the last hour of the clamp period. Larger blood samples were taken at 0, 30, 60, 120, 150, and 180 min and stored for subsequent analysis. To prevent anemia, erythrocytes from the blood samples were washed, resuspended in saline solution, and reinfused. The rate of glucose infusion was adjusted as needed in order to clamp the plasma glucose concentration at 130–140 mg/dl. Plasma radioactivity from [³H-3]-glucose was determined after deproteinization with Ba(OH)₂ and ZnSO₄ and subsequent evaporation to remove tritiated water, as described (25).

Since OLZ administration induced a transient increase in the plasma glucose levels during the basal period, we applied the non–steady-state Steele equation (26) to determine the rate of glucose appearance (R_a) and disappearance (R_d):

where F equals tracer infusion rate (dpm/min/kg), p equals correction factor (0.65), and V equals volume of distribution of glucose (0.25 l/kg). Glu1 and Glu2 equal glucose concentration (mg/ml) at time t1 and t2 (min). SA1 and SA2 (dpm/mg) equal glucose-specific activities at times t1 and t2. The mean values of R_a and R_d during basal (60–120 min) and clamp (140–180 min) periods were used for comparison between groups. The endogenous glucose production (EGP) rate was calculated as the difference between R_d and the glucose infusion rate (GIR).

Plasma glucose concentrations were measured by the glucose oxidase method using a GM7 Micro-Stat Analyzer (ANALOX Instruments, London, U.K.). Insulin levels in plasma were determined by radioimmunoassay. Plasma free fatty acid (FFA) concentration was determined using a colorimetric kit (Wako Chemicals).

Liver and hypothalamus were harvested immediately after the completion of the hyperinsulinemic clamps. Total RNA was isolated by using a TRIzol reagent (Invitrogen) and reverse transcribed into cDNA by using the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen) according to the manufacturer's directions. Quantitative real-time PCR was carried out on the iCycler (Bio-Rad). Briefly, each 20 μl SYBR green reaction consisted of 1–5 μl cDNA (2∼10 ng/μl), 10 μl iQ SYBR green Supermix (Bio-Rad), 1 μl of 4 μmol/l forward and reverse primers and 3–7 μl nuclease-free water. PCR was performed with the following cycling conditions: 40 cycles of 10 s at 95°C and 30–60 s at 56–62°C. Specificity of the PCR product was confirmed by examination of dissociation reaction plots. A distinct single peak indicated that a single DNA sequence was amplified during PCR. Primers were as follows: α-tubulin: forward TACCCTCGCATCCACTTCCCT, reverse CGCTTGGTCTTGATGGTGGCA; glucose-6-phosphatase (G6Pase): forward CTTGTCAGGCATTGCTGTGGC, reverse CAGGAGGTCCACCCCTAGCC; PEPCK: forward ATTCTGCACCCCTGCCAGC, reverse AACACCCCATGCTGCCAGC; AgRP: forward GCAGAAGGCAGAAGCTTTGGC, reverse CCCAAGCAGGACTCGTGCAG; NPY: forward TACTCCGCTCTGCGACACTACATC, reverse CACATGGAAGGGTCTTCAAGCC; proopiomelanocortin: forward AGTGCCAGGACCTCACCACG, reverse AAGCGGTCCCAGCGGAAG; prepro-melanocyte-stimulating hormone (MCH): forward TTCCCAG CTGAGAATGGAGTTCAG, reverse CCCAGCATACACCTGAGCATGTC; and prepro-orexin: forward AGAAGACGTGTTCCTGCCGT, reverse CGTGGTT ACCGTTGGCCTGAA.

Hypothalami were homogenized in freshly prepared ice-cold buffer (150 mmol/l NaCl, 50 mmol/l Tris pH 8.0, 0.1% SDS, 0.5% deoxycholate, and 1% Triton X-100, radioimmunoprecipitation assay buffer modified) with one tablet of protease inhibitor cocktail (catalog no. 04693124001; Roche) and 100 μl of phosphatase inhibitor cocktail Set II (catalog no. 524625; Calbiochem) per 10 ml of buffer. The samples were spun down for 10 min at 14,000 rpm and the supernatant sonicated in a fresh tube four times for 5 s. The insoluble material was removed by centrifugation (14,000 rpm) for 20 min at 4°C. Aliquots of the resulting supernatants containing 400 μg total proteins were diluted in Laemmli sample buffer and boiled for 5 min before electrophoresis separation. For immunoblot analysis, 40 μg of protein from denatured extracts was separated by SDS-PAGE (150 min at 110 V constant) using a miniature-slab gel apparatus (Bio-Rad). Electrotransfer of proteins from the gel to polyvinylidene fluoride membranes (Bio-Rad) was performed (60 min at 100 V constant at 4°C) and blotted with specific antibodies, anti–phospho-AMPK and anti–AMPK (catalog no. 2535 and no. C2532, respectively; Cell Signaling Technology) and anti–β-tubulin (catalog no. MAB380; Millipore). Band intensities were quantified by optical densitometry of developed radiographs (Quantity One 1-D Analysis Software; Bio-Rad).

All data are presented as means ± SE. Plasma glucose, insulin, and FFA levels were analyzed by two-way ANOVA, followed by the Duncan post hoc test for multiple comparisons when appropriate. Student t test for independent samples was performed to the others parameters described here. The statistical calculations were performed using Statistica 6.1 software (StatSoft, Tulsa, OK).

Both OLZ and vehicle groups had similar body weight (vehicle: 317.0 ± 6.3 g vs. OLZ: 316.1 ± 6.0 g), food intake (vehicle: 123.6 ± 4.9 Kcal vs. OLZ: 114.8 ± 5.9 Kcal), and adiposity (epididymal pads, vehicle: 2.165 ± 0.18 g vs. OLZ: 2.198 ± 0.08 g). Intravenous infusion of OLZ transiently increased plasma glucose levels during the basal period compared with rats receiving vehicle [two-way repeated-measures ANOVA group effect, OLZ-IV vs. vehicle: F_(1,224) = 6.0, P = 0.02; group × time interaction effect: F_(14,224) = 5.7, P < 0.0001] (Fig. 1,A). Although the peak in plasma glucose levels was observed 30 min after the start of OLZ administration, there were no differences in the plasma insulin levels during basal or clamp periods between OLZ-IV– and vehicle-treated groups [two-way repeated-measures ANOVA group effect, OLZ-IV vs. vehicle: F_(1,70) = 0.61, P > 0.05; group × time interaction effect: F_(5,70) = 0.42, P > 0.05] (Fig. 1,B). In addition, OLZ increased plasma FFAs, although at the end of the hyperinsulinemic period there was no difference between vehicle- and OLZ-IV–treated groups [two-way repeated-measures ANOVA group effect, OLZ-IV vs. vehicle: F_(1,80) = 15.84, P < 0.01; group × time interaction effect: F_(5,80) = 2.06, P = 0.07] (Fig. 1 C).

During the basal period, the rate of glucose appearance (R_a) was higher (t test = 2.40, P = 0.02) in OLZ-infused animals compared with vehicle (Fig. 2,A). During the clamp period, the infusion rate of exogenous glucose (GIR) required to maintain euglycemia was significantly lower (t test = 4.33, P < 0.001) in the OLZ group (Fig. 2,B), indicating that OLZ decreased insulin sensitivity. Tracer analysis during the clamp period showed that the insulin resistance in OLZ-IV rats was the result of a reduced glucose disposal rate (R_d) (t test = 3.10, P < 0.01) and an increased EGP (t test = 3.4, P < 0.01) compared with the vehicle group (Fig. 2 C and D).

Consistent with the increased EGP, the mRNA levels of hepatic G6Pase (t test = 3.55, P < 0.01) and PEPCK (t test = 2.62, P = 0.01) were higher in the OLZ-IV group compared with control (Fig. 2 E and F).

Since these studies were conducted in male rats, in a separate experiment in mice, we tested whether the acute effect of OLZ on glucose and FFA levels is sex specific. Intraperitoneal injection of OLZ cause significant hyperglycemia and increased FFA levels in both male and females mice (supplemental Fig. 1 in the online appendix, available at http://diabetes.diabetesjournals.org/cgi/content/full/db10-0449/DC1).

Similar to the OLZ-IV studies, intracerebroventricular infusion of OLZ (OLZ-ICV) transiently increased plasma glucose levels during the basal period compared with rats receiving ICV vehicle [two-way repeated-measures ANOVA group effect, OLZ-ICV vs. vehicle: F_(1,224) = 12.23, P < 0.01; group × time interaction effect: F_(14,224) = 7.57, P < 0.0001] (Fig. 3,A). Both groups had similar body weight (vehicle: 329.3 ± 4.4 g vs. OLZ: 325.0 ± 4.5 g), food intake (vehicle: 123.7 ± 2.8 Kcal vs. OLZ: 132.3 ± 4.4 Kcal), and adiposity (vehicle: 2.30 ± 0.11 g vs. OLZ: 2.248 ± 0.09 g). Like in the intravenous studies, OLZ-ICV increased plasma glucose levels in the basal period, with peak values observed 30 min after the bolus of OLZ. Plasma insulin levels during basal or clamp periods were not changed significantly between OLZ-ICV and vehicle groups [two-way repeated-measures ANOVA group effect, OLZ-ICV vs. vehicle: F_(1,50) = 1.25, P = 0.28; group × time interaction effect: F_(5,50) = 0.54, P = 0.74] (Fig. 3,B). Additionally, there were no significant differences in the plasma FFAs between the groups during the basal or hyperinsulinemic periods [two-way repeated-measures ANOVA group effect, OLZ-ICV vs. vehicle: F_(1,60) = 3.51, P = 0.08; group × time interaction effect: F_(5,60) = 1.10, P = 0.37] (Fig. 3 C).

During the basal period, the OLZ-ICV group had a higher R_a (t test = 2.79, P = 0.01) compared with the vehicle group (Fig. 4,A). During the clamp period, the GIR was decreased in the OLZ-ICV group compared with the vehicle group (t test = 4.24, P < 0.001) (Fig. 4,B). The decreased insulin sensitivity of the OLZ-ICV group was accounted for by a higher EGP (t test = 3.30, P < 0.01) (Fig. 4,D) compared with the vehicle group. However, unlike the OLZ-IV studies, the rate of glucose disposal (R_d) of the OLZ-ICV group was not changed compared with control (t test = 0.85, P = 0.40) (Fig. 4 C).

Moreover, intracerbroventricular infusion of OLZ significantly increased gene expression of G6Pase (t test = 3.18, P < 0.01) and tended to increase PEPCK expression (t test = 1.75, P = 0.09) (Fig. 4 E and F).

The phosphorylation of AMPK in hypothalamus of OLZ-IV rats was significantly increased (t test = 2.59, P = 0.01) (Fig. 5,A and B), indicating that peripheral acute administration of OLZ activates hypothalamic AMPK signaling. Despite this activation, we did not detect any change in the levels of NPY or AgRP mRNAs (supplemental fig. 2). However, OLZ-ICV significantly increased expression of hypothalamic NPY (t test = 2.88, P = 0.01) and AgRP (t test = 3.27, P < 0.01) mRNA levels compared with vehicle rats without changing the mRNA levels of orexin, MCH, and proopiomelanocortin (Fig. 5 C).

In the present study, we show that acute infusion of OLZ either intravenously or intracerbroventricularly induces hepatic insulin resistance. These metabolic alterations are accompanied by hypothalamic activation of AMPK and increased expression of NPY and AgRP. These findings are consistent with the notion that insulin resistance associated with the administration of OLZ could, at least in part, be secondary to its effects on the central nervous system.

Treatment with atypical antipsychotics increases the risk of weight gain, diabetes, and other metabolic disorders in comparison to first-generation antipsychotic drugs (1,–3,27). Recent studies (28,–30) have also shown that some atypical antipsychotics affect glucose homeostasis before changes in body weight are observed, suggesting that insulin resistance could occur before or independent of any change in adiposity. We showed here that in lean, male rats an acute intravenous infusion of OLZ induced a transient increase of plasma glucose levels (Fig. 1,A) in the absence of a compensatory insulin response (Fig. 1 B). In several studies, increases in plasma glucose levels have been observed after acute administration of several different atypical antipsychotics in mice. These effects were blocked by a glanglionic blocker, an α2 adrenergic receptor antagonist, or a glucocorticoid receptor antagonist, suggesting a central activation of the sympathetic system (31,–33).

Although in humans OLZ-induced obesity and diabetes are not sex specific (34), early studies have preferentially used female rodents because of an apparent stronger sensitivity to OLZ in causing obesity (6,10). Recent findings and our data show that male rats, and mice of both genders, are similarly at risk for OLZ-induced alterations in body weight and glucose homeostasis (supplemental Fig. 1) (8,35).

OLZ can impair glucose-stimulated insulin secretion after chronic or acute peripheral administration, as shown in hyperglycemic clamp experiments (9,12). We did not detect a change in insulin levels during the basal period of OLZ infusion, despite the fact that intravenous OLZ increased glycemia. The lack of an increase in insulin secretion in response to increased plasma glucose during the OLZ infusion might be related to sympathetic inhibition of insulin release from pancreatic β-cells, as previously suggested (32).

Under hyperinsulinemic-euglycemic clamp conditions, which allowed us to override potential differences in insulin secretion, we found lower rates of exogenous glucose infusion and glucose utilization (GIR and R_d) and higher EGP in OLZ-IV–infused rats (Fig. 2,B–D). These data support the hypothesis that acutely administered OLZ induces insulin resistance. Our observations agree with previous studies showing impairment of insulin action on hepatic glucose production and peripheral glucose utilization after a subcutaneous bolus injection of OLZ (11,–13). In addition, here we showed that OLZ-induced insulin resistance was associated with higher gene expression levels of G6Pase and PEPCK, indicating that OLZ impairs the ability of insulin to suppress the expression of rate-limiting enzymes for hepatic glucose output (Fig. 2 E and F).

To test the hypothesis that OLZ impairs insulin sensitivity through CNS action, we infused OLZ intracerebroventricularly before and during hyperinsulinemic-euglycemic clamps. Indeed, intracerebroventricular OLZ infusion increased the rate of glucose appearance (R_a) during the basal period, impaired the ability of insulin to suppress EGP (Fig. 4,A and D), and decrease the expression of gluconeogenic enzymes in liver (Fig. 4 E and F).

Interestingly, like OLZ-IV, OLZ-ICV markedly impaired hepatic insulin action but did not increase plasma FFAs or impair glucose disposal. Under basal and low insulin levels, OLZ-IV infusion increased plasma FFA levels. However, during hyperinsulinemia, the elevated FFA levels gradually return to control levels (Fig. 1 C). This finding agrees with previous reports showing no differences in FFA levels after the hyperinsulinemic-euglycemic clamp condition in rats (11). Nonetheless, in our OLZ-IV studies, we observe a time-dependent and transient increase in FFA levels. Similarly, we found that acute intraperitoneal administration of OLZ in male and female mice increases plasma FFA levels, an effect that was not blocked by pretreatment with a β3-blocker (supplemental Fig. 1D and E). The effect of OLZ on FFA levels has been reported in studies with more prolonged treatment. Healthy men receiving daily OLZ treatment for 8 days display an impaired ability of hyperinsulinemia to reduce plasma FFA levels and increase R_d, in the absence of changes in body weight and adiposity (36). Thus, the reported lack of effect on FFA plasma levels (11) could be due to differences in time of sampling and/or OLZ route of administration.

It is known that elevated plasma FFA levels impair the ability of insulin to suppress hepatic glucose production and stimulate glucose uptake by skeletal muscle (37). However, we observed that only OLZ-IV administration increased plasma FFAs (Fig. 1,C) and impaired insulin-stimulated R_d (Fig. 2,C), while OLZ effects on hepatic glucose production occurred regardless of the route of administration and the presence of increased plasma FFA levels (Fig. 2,D and Fig. 4 D). These data suggest that increased FFA levels do not explain the deleterious effects of OLZ on glucose production.

Our data suggest that the effect of OLZ on FFA and glucose utilization may be mediated by peripheral receptors. In support of this notion, recent studies have shown that OLZ reduced basal and isoproterenol-stimulated release of FFAs in rat adipocytes, as well as glucose uptake (38,39).

Hepatic insulin resistance is a common feature of obesity and type 2 diabetes (37,40) and may result not only from alterations of insulin signaling in the liver but also from impaired insulin action in the brain. Neuronal circuitry responsive to insulin play an important role in modulating hepatic gluconeogenesis in response to physiologic elevations of plasma insulin (41). In ARC, insulin receptor activation inhibits NPY/AgRP neurons and stimulates proopiomelanocortin neurons in order to control energy homeostasis (42). Furthermore, insulin action on arcuate neurons suppresses hepatic glucose production by a mechanism dependent on phosphatidylinositol kinase–ATP-sensitive K+ channel activation (43).

Intracerebroventricular administration of OLZ increased hypothalamic mRNA levels of NPY and AgRP (Fig. 5 C) in rats killed after 2 h of hyperinsulinemic-euglycemic clamps. This result suggests a potential mechanism by which CNS OLZ reduced hepatic insulin sensitivity. Insulin inhibits arcuate NPY and AgRP gene expression and inhibits AMPK phosphorylation/activity (44), whereas inhibition of insulin action in the hypothalamic arcuate nucleus increases the expression of NPY and AgRP (45). Conversely, atypical antipsychotics, including clozapine and OLZ, increase phosphorylation/activity of AMPK (46). In our intravenous studies, OLZ increases hypothalamic AMPK phosphorylation, suggesting that this atypical antipsychotic might counteract the action of insulin on NPY/AgRP neurons by preventing the insulin-mediated inhibition of AMPK. However, unlike intracerebroventricular OLZ, we did not detect increased expression of NPY/AgRP in the OLZ-IV studies. Several studies with peripheral injection of OLZ have failed to find changes in NPY/AgRP levels (6,47), although several lines of evidence show that peripheral OLZ acutely activates ARC neurons (48). We speculate that our finding of increased NPY/AgRP in ICV-OLZ and not in OLZ-IV can be explained by a greater time of neuronal exposure to OLZ in the intracerebroventricular group. Since OLZ can transiently activate AMPK in hypothalamic explants (46), our results in OLZ-IV rats are consistent with an acute activation of AMPK that would later result in increased NPY/AgRP expression. Taken together, our data support the hypothesis that OLZ counteracts the inhibitory action of insulin on hypothalamic AMPK in NPY/AgRP neurons. Intracerebroventricular administration of NPY impairs insulin's ability to suppress hepatic glucose production (49,50). In addition, selective ablation of the insulin receptor in AgRP-positive neurons leads to hepatic insulin resistance (20). Taken together, these data support the notion that atypical antipsychotics alter hepatic insulin sensitivity by impairing insulin's action on NPY/AgRP neurons.

In conclusion, we have shown that acute administration of OLZ induces insulin resistance in the absence of changes in food intake and body composition. In agreement with previous studies (9,13), this effect is primarily due to hepatic insulin resistance. These effects are accompanied by the activation of hypothalamic AMPK. Moreover, we showed that CNS delivery of OLZ induces hepatic insulin resistance and activates hypothalamic NPY/AgRP neurons. These data are consistent with the notion that the deleterious effects of OLZ on hepatic insulin action are mediated at least in part by impaired hypothalamic control of hepatic glucose metabolism.

This work was supported in part by a grant from Eli Lilly and Company; P.J.F.M. is recipient of a fellowship from CNPq-200863/2007-0 Brazil and the American Heart Association (0825677D). No other potential conflicts of interest relevant to this article were reported.

P.J.F.M. performed experiments and wrote the manuscript. M.H. performed surgeries and clamp studies. S.O. supervised experiments and reviewed/edited the manuscript.

We thank Kay Ellis in Dr. David D'Alessio laboratory, Department of Medicine, University of Cincinnati, for performing the plasma insulin assays. We thank Jack Magrisso, Department of Medicine, University of Cincinnati, for outstanding technical support.