Age-associated cholesterol reduction triggers brain insulin resistance by facilitating ligand-independent receptor activation and pathway desensitization

Abstract

In the brain, insulin plays an important role in cognitive processes. During aging, these faculties decline, as does insulin signaling. The mechanism behind this last phenomenon is unclear. In recent studies, we reported that the mild and gradual loss of cholesterol in the synaptic fraction of hippocampal neurons during aging leads to a decrease in synaptic plasticity evoked by glutamate receptor activation and also by receptor tyrosine kinase (RTK) signaling. As insulin and insulin growth factor activity are dependent on tyrosine kinase receptors, we investigated whether the constitutive loss of brain cholesterol is also involved in the decay of insulin function with age. Using long-term depression (LTD) induced by application of insulin to hippocampal slices as a read-out, we found that the decline in insulin function during aging could be monitored as a progressive impairment of insulin-LTD. The application of a cholesterol inclusion complex, which donates cholesterol to the membrane and increases membrane cholesterol levels, rescued the insulin signaling deficit and insulin-LTD. In contrast, extraction of cholesterol from hippocampal neurons of adult mice produced the opposite effect. Furthermore, in vivo inhibition of Cyp46A1, an enzyme involved in brain cholesterol loss with age, improved insulin signaling. Fluorescence resonance energy transfer (FRET) experiments pointed to a change in receptor conformation by reduced membrane cholesterol, favoring ligand-independent autophosphorylation. Together, these results indicate that changes in membrane fluidity of brain cells during aging play a key role in the decay of synaptic plasticity and cognition that occurs at this late stage of life.

Keywords: aging; cholesterol; insulin signaling.

Figure 1

Insulin signaling is impaired in old mice due to PI3K/Akt hyperactivation. (a) Insulin‐LTD is completely abolished in old mice (20–24 months old; n = 7) compared to adult mice (7–12 months old; n = 10). The graphic represents insulin‐LTD as fEPSP slope. Gray box indicates the time of insulin application. Representative analogue traces on the right were collected at the indicated time points. (b) Insulin receptor (IR) basal activity levels detected by Western blot using Phospho‐Tyrosine 1150/1151 antibody after total protein immunoprecipitation. (c) Western blots reflect IGF‐1R Phospho‐Tyrosine 1135/1136 basal levels after total protein immunoprecipitation. Old mice hippocampus shows higher basal activity levels of IR and IGF1‐R compared to adult mice. (d) Akt Phospho‐Serine 473 activating residue detected by Western blot in adult and old mice hippocampus at basal level. (e) Western blot analysis of GSK3β in hippocampal extracts from adult and old mice. Consistent with Akt hyperactivation, the old hippocampus presents high GSK3β inhibitory mark (Phospho‐Serine 9). (f, g) PI3K/Akt inhibition rescues insulin‐LTD in old mice. Graphics showing insulin‐LTD induction in hippocampal slices from old mice incubated with Wortmannin (f; 0.5 μM; n = 7) and Quercetin (g; 20 μM; n = 8). Black box indicates time of insulin application. Gray box indicates time of inhibitors application. Representative analogue traces below were collected at the indicated time points. Numbers in bars reflect number of independent experiments. Data are represented as mean ± SEM. t test for (b, c, d, e), one‐way ANOVA with post hoc Bonferroni's test for (a, f, g). The asterisks p values (*p < 0.05; **p < 0.01; ns = not significant)

Figure 2

Ex vivo cholesterol replenishment in hippocampal slices of old mice restores insulin signaling. (a) Addition of cyclodextrin‐cholesterol (referred as MβCD‐Ch or Cholesterol) solution to old mice hippocampal slices induces a 20% increase in membrane cholesterol. (b, c) Cholesterol addition (as in a) reduces hyperactivated Akt and its downstream target p70S6K. (b) Detection of Akt Phospho‐Serine 473; (c) detection of p70S6K Phospho‐Threonine 389. (d) Cholesterol addition to old hippocampal slices reduces Phospho‐Serine 9 GSK3β inhibitory residue. (e) Addition of the cholesterol solution to hippocampal slices of old mice rescues the impairment of insulin‐LTD. Graphic showing insulin‐LTD induction in old mice slices (n = 7) and old mice slices incubated with MβCD‐Ch (n = 9). Black box indicates time of insulin application. Gray box indicates the time of MβCD‐Ch mix application. Representative analogue traces below were collected at the indicated time points. (f, g) Cholesterol replenishment decreases insulin resistance marks on IRS‐1 protein. (f) IRS‐1, Phospho‐Serine 632; (g) Phospho‐Serine 307. The value inside the bars indicates the number of independent experiments. Data are represented as mean ± SEM. t test for (a, b, d, g), Wilcoxon test for (c, f), one‐way ANOVA with post hoc Bonferroni's test for (e). The asterisks indicate the p values (*p < 0.05; **p < 0.01)

Figure 3

In vivo pharmacological inhibition of cholesterol loss in old mice restores hippocampal insulin sensitivity and insulin‐LTD. (a, b) Oral administration of the Cyp46A1 inhibitor Voriconazole reduces hippocampal PI3K/Akt and downstream effector p70S6K hyperactivation in old mice. (a) Western blot of Phospho‐Serine 473 activating residue on Akt. (b) Western blot of p70S6K Phospho‐Threonine 389. (c, d) Oral treatment with Voriconazole decreases insulin resistance marks in hippocampal samples of old mice. (c) Western blots for IRS‐1 resistance marks Phospho‐Serine 632; (d) Phospho‐Serine 307. (e) Voriconazole treatment reduces the inhibitory phosphorylation on GSK3β. (f) Graphic showing insulin‐LTD induction in old mice slices (n = 7) and old mice slices incubated with Voriconazole (n = 8). Hippocampal slices were incubated with Voriconazole 10 nM for 60 min starting 30 min before insulin stimulus. Black box indicates time of insulin application. Gray box indicates time of Voriconazole application. Representative analogue traces below were collected at the indicated time points. Bar graphs: Numbers inside indicate the number of independent experiments. Data are represented as mean ± SEM. t test for (a, b, c, d, e). One‐way ANOVA with post hoc Bonferroni's test for (F). The asterisks indicate the p values (*p < 0.05)

Figure 4

Cholesterol loss favors in neurons the conformational changes required for insulin growth factor 1 receptor's activation. (a, b, c) FRET experiment. Laser microscopy images of cultured neurons co‐transfected with IGF‐1R‐EYFP (donor fluorophore) and IGF‐1R‐mCherry (acceptor fluorophore), with the fluorophores replacing the cytosolic tails of the entire receptor sequence. Neurons maintained in Neurobasal+B27 (without serum) remained un‐stimulated (a), were stimulated with 4 μM IGF‐1 (b), or were treated with 10 IU/ml Choox before fixation (c). Acceptor signal was bleached, intensity of donor signal was measured, and FRET efficiencies were calculated (see Materials and Methods). Scale bars represent 10 μm. (d) Cholesterol loss enables in neurons a plasma membrane environment sufficient to induce the conformational change on IGF‐1R required for receptor activation, thus increasing FRET efficiency. The graphic represents FRET efficiencies data obtained from 80 images quantified of four independent experiments for each condition. Data are represented as mean ± SEM. One‐way ANOVA with post hoc Bonferroni's test for (a, b, c). The asterisks indicate the p values (*p < 0.05; ***p < 0.001). (e) Summary of how age produces insulin resistance in the mouse hippocampus: cholesterol loss, due in part to Cyp46a1 activation (and also to SCAP reduction, and perhaps others), favors an increase in plasma membrane ordered domains where the IR/IGF‐1R preferentially cluster (membrane yellow segments). This clustering facilitates the conformational changes required for the receptor's autophosphorylation in a ligand‐independent manner (phosphor motifs in IR/IGF‐1R). If this type of receptor change occurs over a prolonged period of time and incrementally (as it is the case with the loss of cholesterol with age), the post‐receptor pathway will also be activated chronically, leading to desensitization (phospho‐motifs in IRS‐1) and loss of insulin synaptic plasticity function (Akt inhibition on GSK3β and AMPAR inhibited internalization)