Tissue selectivity of insulin detemir action in vivo
- First Online:
- Cite this article as:
- Hennige, A.M., Sartorius, T., Tschritter, O. et al. Diabetologia (2006) 49: 1274. doi:10.1007/s00125-006-0192-9
Recombinant DNA technology is a useful tool that can be used to create insulin analogues with modified absorption kinetics to improve glycaemic control in patients with type 1 and type 2 diabetes. Among conventional insulin analogues, which are usually created by amino acid exchange, insulin detemir is the first analogue to be acylated with a fatty acid to enable reversible albumin binding. In this study we determined activation of the insulin receptor (IR)-signalling cascade by insulin detemir at the level of IR and IR substrate (Irs) phosphorylation, as well as downstream signalling elements such as phosphatidylinositol 3-kinase and Akt, and performed epidural EEG in vivo.
C57Bl/6 mice were injected i.v. with either insulin detemir or human insulin and Western blot analysis was performed on liver, muscle, hypothalamic and cerebrocortical tissues. Moreover, cerebrocortical activity was detected by EEG in awake mice and cerebral insulin concentrations were measured following human insulin and insulin detemir injection.
The time course and extent of IR phosphorylation in peripheral tissues were similar following insulin detemir treatment compared with human insulin, but insulin signalling in hypothalamic and cerebrocortical tissue determined by tyrosine-phosphorylation of the IR and Irs2 proteins occurred faster and was enhanced due to a higher insulin detemir concentration in the brain. Moreover, epidural EEG in mice displayed increased cortical activity using insulin detemir.
Taken together, these data suggest that insulin detemir has a tissue-selective action, with a relative preference for brain compared with peripheral tissues.
KeywordsBrainCerebral cortexElectroencephalographyHypothalamusInsulin analoguesInsulin detemirInsulin receptor signalling
insulin receptor substrate
The treatment strategies for patients with type 1 and type 2 diabetes have been continually improved over the past decades. One milestone towards optimised glycaemic control are rapid and long-acting insulin analogues that more closely mimic the physiological insulin response to a meal following injection [1, 2] and achieve sufficient basal day-long glycaemic control . The characteristics are based on the fact that modification of the amino acid sequence of the insulin molecule results in distinguished hexamer formation and therefore altered absorption kinetics. Among conventional insulin analogues that are usually created by amino acid exchange, insulin detemir is the first analogue that is acylated with a fatty acid to enable reversible albumin binding [4–7]. Albumin binding is a common principle to delay absorption and results in retention of the insulin molecule in the s.c. depot for a longer period of time . Due to the profile of action of insulin detemir and the fact that the albumin-bound insulin buffers against rapid changes of absorption, insulin detemir action is supposed to be more predictable in terms of the risk for hypoglycaemic episodes compared with other basal insulins [9–11].
The initiation of insulin therapy in type 1 and type 2 diabetes is usually associated with weight gain [12, 13]. Indeed, there is a positive correlation between improved glycaemic control and weight gain, which is clearly undesirable as obesity is present in most patients with type 2 diabetes.
Over the past years data have accumulated supporting the fact that insulin therapy using insulin detemir leads to both improved glycaemic control and weight stability , opening new perspectives for insulin therapy. However, the mechanism(s) underlying the altered weight gain with insulin detemir are unknown.
At the molecular level, insulin binds to its cell surface receptor and stimulates autophosphorylation of the β-subunit, followed by phosphorylation of substrates, including the insulin receptor substrate (Irs) protein family members and activation of the lipid kinase phosphatidylinositol 3-kinase (PI 3-kinase) . In peripheral tissue, activation of this pathway is involved in processes like glucose transport, suppression of hepatic gluconeogenesis, protein synthesis and gene transcription [7, 16]. However, in the brain, intact insulin signalling via the Irs-PI 3-kinase pathway is essential for nutrient homeostasis and appetite regulation as pharmacological inhibition of insulin signalling, especially in the hypothalamus, leads to obesity-induced diabetes [17, 18]. Keeping in mind that insulin detemir therapy is characterised by weight stability or even weight loss, we reasoned that the kinetics of insulin signalling in the brain might be altered by insulin detemir.
In this study we tested this hypothesis and provide in vivo data on the impact of insulin detemir on the insulin-signalling cascade in peripheral and brain tissues. C57Bl/6 mice were injected with either human insulin or the analogue and IR phosphorylation and activation of downstream signalling elements were analysed in muscle and liver and hypothalamic and cerebrocortical tissues. The data suggest that during regular activation of the insulin-signalling cascade in the periphery, insulin detemir displays a predominant activation of the cerebral insulin-signalling cascade.
Materials and methods
Ten-week-old male C57Bl/6 mice were obtained from Charles River WIGA GmbH (Sulzfeld, Germany) and studied after 2 weeks of acclimatisation. All procedures were conducted according to the guidelines of laboratory animal care and were approved by the local governmental commission for animal research. Mice were kept on a normal light/darkness cycle and on a regular chow. Glucose levels were sampled from tail bleeds of ketamine-anaesthetised mice using a Glucometer Elite (Bayer, Elkhart, IN, USA).
In vivo insulin stimulation and Western blot analysis
For in vivo stimulation, a bolus of either 1 U/kg body weight of human insulin (0.6 mmol/l, 100 U/ml) or 2 U/kg body weight of insulin detemir (2.4 mmol/l, 100 U/ml) diluted in phosphate buffer was injected into the inferior vena cava of ketamine-anaesthetised mice after an overnight fast. This dose was used to achieve comparable peripheral insulin-signalling kinetics and glucose-lowering effects. Tissues (liver, muscle, hypothalamus and cerebral cortex) were removed at the indicated time points and homogenised at 4°C as described . Homogenates were allowed to solubilise for 30 min on ice and then clarified by centrifugation at 12,000× g for 20 min. For detection of insulin-stimulated tyrosine phosphorylation, supernatants were immunoprecipitated with antibodies directed against the carboxy terminus of the IR (KKNGRILTLPRSNPS; a gift from R. Lammers, University of Tübingen, Germany) and Irs2 (No. 06-506; Upstate, Charlottesville, VA, USA). Visualisation of immunocomplexes after gel electrophoresis and Western blotting with anti-phosphotyrosine (p-Tyr) antibody (PY20; Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-phospho-Akt (Ser473) and anti-Akt (Upstate) was performed with a non-radioactively enhanced chemiluminescence system.
Assay of PI 3-kinase activity
The same amount of total protein from tissue lysates was immunopurified with anti-p-Tyr antibody and immunocomplexes were absorbed by Protein A-Sepharose for 12 h. Immunoprecipitates were washed three times and pellets were directly incubated with 0.1 mg/ml l-α-phosphatidylinositol (Sigma, Munich, Germany) and 50 μmol/l [γ32P]ATP (Perkin Elmer Life Science, Zaventem, Belgium) at room temperature for 10 min. After addition of 150 μl 1 mol/l HCl, lipids were extracted twice with 450 μl chloroform/methanol (1:1 by volume). Products were separated by thin layer chromatography as described . 32P-labelled phospholipids were detected by autoradiography.
Insulin extracts of total brain
Mice were stimulated i.v. with either human insulin (1 U/kg mouse) or insulin detemir (2 U/kg mouse). At the indicated time points, total brain was homogenised in 1.0 ml acid-ethanol and insulin was extracted overnight at −20°C. The following day, samples were centrifuged and insulin concentration was measured in the supernatant with a competitive RIA kit (Linco Research, St Charles, MO, USA) using a human insulin and insulin detemir standard curve. The data presented for insulin detemir are divided by eight to normalise for the injected dose.
Telemetric EEG analysis was performed using implantable radio transmitters (PhysioTel transmitter TA10EA-F20; Data Sciences International [DSI], Lexington, MA, USA). All surgical procedures were conducted with the mouse under isoflurane (5% induction, 1.5% maintenance) anaesthesia. The mouse was placed in a stereotaxic head holder fitted with a mouse incisor bar. Midline skin incisions (10 mm on the head and 15 mm down the neck) were made and the s.c. tissue was bluntly separated. The body of the telemetry transmitter was implanted s.c. in a pouch made in the loose skin of the back and neck. The two lead wires were tunnelled s.c. through the incision on the scull. Holes (0.7 mm diameter) were drilled in the cranium to implant two screws, around which the tips of the EEG leads were wrapped. For the recording electrode, the screws were placed epidurally 1 mm anterior to the lambda and 1 mm left of the sutura saggitalis, and for the reference electrode they were placed 1 mm anterior to the bregma and 1 mm right to the sutura coronaria. The electrodes and screws were then covered with dental acrylic cement, and the scalp sutured closely around the resulting wound with non-absorbable 5–0 suture material (Ethilon polyamide; Ethicon, Germany). Proper screw placement was confirmed histologically post mortem. The mice were given 1 week of post-surgical recovery before undergoing EEG recordings for further analysis. For in vivo stimulation, human insulin (1 U/kg body weight) or insulin detemir (2 U/kg body weight) was administered i.v. in the tail vein of awake, overnight-fasted mice in a cross-over design 5 days apart. After the injection, the mice were placed separately in their transparent home cages and an EEG was immediately recorded, lasting 30 min. Basal EEG data represent 1 h measurement in fed mice between 09.00 and 10.00 h when insulin was usually applied. Telemetry signals (EEG and motor activity) were processed by a Data-Sciences analog converter (Data Exchange Matrix; DSI) and stored digitally using the Dataquest A.R.T. 3.1 software (DSI). EEG activity was sampled at 250 Hz with a filter cut-off of 40 Hz. A video monitoring system was used in order to record referential EEGs and simultaneous video images showing the behaviour of the mice.
Data analysis for EEG measurements
The acquired EEG data were imported into Somnologica Science software (version 3.3.1; Medcare, Reykjavík, Iceland) and power spectral analysis were calculated. Power spectral estimates provided the basic frequency characteristics of each recorded region. Consecutive 2-s epochs were subjected to a fast Fourier transform routine (Hanning smoothing window, each epoch divided into three 256-point overlapping segment windows with 50% overlap). Epochs of 2 s are to be preferred in intraoperative monitoring because they allow more rapid detection of change by increasing the statistical reliability of the power spectral estimate . EEG power density (the square of the amplitude) was computed for a frequency range of 0.5–100 Hz (delta: 0.5–4 Hz, theta: 4–8 Hz, alpha: 8–12 Hz, beta: 12–30 Hz, and gamma: 30–100 Hz). The value of the EEG power density in each one of the different frequency bands was then examined and compared.
Statistical analyses for Western blot analysis were done by a two-sided unpaired Student’s t-test. Data in figures are expressed as means±SEM. The data from EEG measurements were analysed by a two-side paired Student’s t-test.
Metabolic effects and autophosphorylation of the IR and downstream elements in muscle tissue after stimulation with insulin detemir
To determine whether the insulin analogue insulin detemir activates the insulin-signalling cascade to the same extent as human insulin, we injected either human insulin or insulin detemir into 12-week-old mice. Insulin administration in this study was performed by injection into the inferior vena cava to avoid the effect of altered s.c. absorption kinetics and to focus on direct IR-signalling characteristics. Throughout the 30-min experiment, glucose levels significantly dropped from 9.7±0.3 to 3.5±0.2 mmol/l for human insulin and to 4±0.6 mmol/l for insulin detemir (p<0.01, n=7).
Autophosphorylation of the IR and downstream elements in liver tissue after stimulation with insulin detemir
Phosphorylation of the IR and Irs2 in hypothalamic tissue after stimulation with insulin detemir
Recent studies revealed that dysregulation of IR signalling in the brain is associated with the development of obesity and diabetes. Since IR and Irs2 are expressed in the hypothalamus, we evaluated the IR-signalling characteristics in vivo after insulin injection.
Irs2 is essential for intact insulin signalling in the brain, as partially dysregulated Irs2 signalling is responsible for hyperphagia and obesity . Therefore, we determined the tyrosine phosphorylation state of Irs2 in the hypothalamus of mice. After insulin injection, tissues were removed after 2, 5, 10, 15 and 30 min, and Western blot analysis was performed. These data suggest that the activation levels of Irs2 in mice treated with insulin detemir are significantly elevated compared with human insulin with a shift to an earlier activation of the insulin-signalling cascade for insulin detemir (Fig. 5b).
Phosphorylation of the IR in cortical tissue, EEG and total brain insulin concentrations following insulin detemir treatment
To determine the impact of an altered transport across the blood-brain barrier that might contribute to the brain data obtained, we measured total insulin concentrations in ethanol-acid brain extracts. Following i.v. injection, insulin detemir reached significantly higher concentrations in the brain compared with human insulin (Fig. 6c, *p<0.05, **p<0.01, n=5).
We therefore conclude that treatment of mice with an acute i.v. dose of insulin detemir preferentially activates the cerebral insulin-signalling cascade while there is an equal IR activation in the periphery.
Recombinant DNA technology has been used for almost two decades now to develop insulin analogues with advanced kinetic profiles and distinguished absorption . Recently, the analogue insulin detemir was developed by the attachment of a 14-carbon fatty acid chain to LysB29 resulting in delayed absorption kinetics.
Here we report in vivo data from mice that were treated with either human insulin or the insulin analogue insulin detemir. In order to overcome the effect of altered s.c. absorption kinetics and to focus on direct IR-signalling characteristics, insulin was administrated by i.v. injection. Thereby, plasma concentrations immediately rose and a comparable onset of action and tissue distribution was achieved.
In our in vivo experiments, insulin detemir activates the insulin-signalling cascade in muscle and liver tissues to the same extent as human insulin does. However, in hypothalamic and cerebrocortical tissues, insulin-signalling kinetics determined by tyrosine-phosphorylation of the IR and Irs2 proteins were enhanced and the maximum was shifted to earlier time points, suggesting an altered profile of insulin detemir action in the cerebral nervous system.
As the time course and extent of activation of the insulin-signalling cascade in liver and muscle were comparable, we believe that insulin detemir action in the brain is enhanced due to the attachment of the fatty acid chain. This is reflected by elevated insulin detemir concentrations in the brain.
Although several mechanisms are suggested by which insulin may enter the brain from the blood, including receptor-mediated transport processes, conclusive data are not available . However, it is known that albumin directly penetrates into the cerebrospinal fluid across choroids plexus epithelial cells , and there is evidence that as lipophilicity and membrane permeability of a molecule increases, the concentration in the cerebrospinal fluid rises . Moreover, one could speculate that in the presence of a low albumin concentration such as in the cerebrospinal fluid, the rate of unbound insulin detemir is increased, and therefore an enhanced amount of ‘active’ insulin detemir molecules is available to bind to the IR. In addition, the impact of the fatty acid chain on IR binding in the brain is unknown; however, it could indeed also count towards enhanced IR signalling.
The observation of enhanced insulin signalling in the brain using insulin detemir is of particular interest in the context of former studies in humans using intranasal insulin, which enters the cerebrospinal fluid compartment without affecting circulating insulin levels in the bloodstream. In these healthy subjects the administration of nasal insulin led to a loss of body weight and body fat over an 8-week treatment accompanied by a drop in plasma leptin levels. These data provide a profound basis for a negative feedback signal of insulin in the regulation of body weight and adiposity . Moreover, a recent study in humans compared the effect of human insulin and insulin detemir on hepatic glucose production . While the glucose infusion rate and glucose disposal were similar after s.c. injection, hepatic glucose production within a 16-h euglycaemic clamp was increased using insulin detemir. These data are somehow contradictory to our in vivo data on equal IR signalling in liver tissue; however, the altered observation time (short-term vs long-term) and the route of administration (s.c. vs i.v.) might be responsible for these findings.
On a molecular basis, impressive data from mouse models have accumulated over the past years suggesting that dysregulation of insulin action at the level of the IR and downstream signalling targets in the central nervous system is associated with obesity and diabetes [27, 28]. The precise underlying mechanisms are uncertain and difficult to establish, but it is known that female mice lacking the IR specifically in the brain display increased food intake, and both male and female mice developed diet-sensitive obesity with elevated body fat and plasma leptin levels including mild insulin resistance. Furthermore, experimental desensitisation of the cerebral IR resulted in both a cognitive deficit and abnormalities in cerebral oxidative glucose metabolism , supporting the idea that intact insulin signalling in the brain is a pivotal element for brain function and glucose homeostasis. At the Irs level, Irs2 is potentially important since a lack of Irs2 in the hypothalamus results in increased appetite and body mass, leading to insulin resistance and finally diabetes . Moreover, the loss of Irs2 in the mouse brain is associated with reduced neuronal proliferation during development, as well as accumulation of neurofibrillary tangles containing phosphorylated tau in the hippocampus, favouring intact Irs2 signalling in the hypothalamus as neuroprotective . This is in agreement with a study in humans showing that intranasal administration of insulin leads to an improvement of cognitive function and memory, opening a whole new aspect in insulin therapy .
At the level of PI 3-kinase, it has been demonstrated that the food-intake-lowering effect of insulin in mice is reversed by intracerebroventricular infusion of PI 3-kinase inhibitors , further supporting the idea that the Irs-PI 3-kinase pathway is a strong mediator of appetite and weight control.
Since the impact of an earlier and increased tyrosine phosphorylation of the IR and Irs2 in hypothalamic tissue observed during insulin detemir treatment remains to be further elucidated, it can be suggested that pronounced activation of the insulin-signalling cascade in the brain is sufficient to prevent hyperphagia and weight gain in insulin detemir therapy. If this is the case, effects obtained in this study might be related to the differences in weight development observed in clinical trials comparing insulin detemir with NPH insulin treatment.
Taken together, the current in vivo data suggest that the novel insulin analogue insulin detemir is characterised by an increased rate of brain action vs peripheral action. This characteristic profile of insulin detemir may provide a new concept for insulin therapy in obese people with diabetes.
We wish to thank R. Lammers for providing the anti-IR antibody and S. Waldert for technical assistance. This study was supported by a grant from Novo Nordisk (Bagsvaerd, Denmark) and the Deutsche Forschungsgemeinschaft (KFO 114/1).
Duality of interest
H.-U. Häring and A. Fritsche have received grant support, honoraria for speaking engagements and consulting fees for participating on scientific boards of Novo Nordisk.