SEARCH

SEARCH BY CITATION

Keywords:

  • hippocampus;
  • long-term depression;
  • long-term potentiation;
  • N-methyl-d-asparate;
  • phosphatidylinositol-3-kinase

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Insulin and its receptor are both present in the central nervous system and are implicated in neuronal survival and hippocampal synaptic plasticity. Here we show that insulin activates phosphatidylinositol 3-kinase (PI3K) and protein kinase B (PKB), and results in an induction of long-term depression (LTD) in hippocampal CA1 neurones. Evaluation of the frequency–response curve of synaptic plasticity revealed that insulin induced LTD at 0.033 Hz and LTP at 10 Hz, whereas in the absence of insulin, 1 Hz induced LTD and 100 Hz induced LTP. LTD induction in the presence of insulin required low frequency synaptic stimulation (0.033 Hz) and blockade of GABAergic transmission. The LTD or LTP induced in the presence of insulin was N-methyl-d-aspartate (NMDA) receptor specific as it could be inhibited by α-amino-5-phosphonopentanoic acid (APV), a specific NMDA receptor antagonist. LTD induction was also facilitated by lowering the extracellular Mg2+ concentration, indicating an involvement of NMDA receptors. Inhibition of PI3K signalling or discontinuing synaptic stimulation also prevented this LTD. These results show that insulin modulates activity-dependent synaptic plasticity, which requires activation of NMDA receptors and the PI3K pathway. The results obtained provide a mechanistic link between insulin and synaptic plasticity, and explain how insulin functions as a neuromodulator.

Abbreviations used
AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

APV

dl-2-amino-5-phophono-valeric acid

BCA

bicinchoninic acid

DHPG

3,5,-dihydroxyphenylglycine

ECL

enhanced chemoluminescence

EPSP

excitatory post-synaptic potential

ERK

extracellular regulated kinase

IRS

insulin receptor substrate

LTD

long-term depression

LTP

long-term potentiation

mGluR

metabotropic glutamate receptor

mTOR

mammalian target of rapamycin

NMDA

N-methyl-d-asparate

OD

optical density

PBS

phosphate-buffered saline

PDK-1

phosphoinositide-dependent kinase 1

PI3K

phosphatidylinositol-3-kinase

PKC

protein kinase C

PP

protein phosphatase

RMP

resting membrane potential

SDS

sodium dodecyl sulfate

SGK

serum and glucocorticoid regulated kinase

Long-lasting changes in synaptic strength underlie experience-dependent plasticity, which has been suggested to form the basis of learning and memory (Malenka and Nicoll 1999). Two opposite forms of activity-dependent synaptic modifications have been identified: long-term potentiation (LTP) and long-term depression (LTD). In many brain areas, including the hippocampus and neocortex, the direction and degree of the synaptic change are functions of conditioning frequency, the level of post-synaptic depolarization and the change in post-synaptic Ca2+ level. LTD is obtained following low frequency stimulation (1 Hz), low levels of post-synaptic depolarization and a small rise in the intracellular Ca2+ level, whereas LTP is produced by high frequency stimulation (100 Hz), a stronger depolarization and a large rise in the intracellular Ca2+ level (Dunwiddie and Lynch 1978; Artola et al. 1990; Dudek and Bear 1992; Ngezahayo et al. 2000). The best characterized forms of LTD and LTP require Ca2+ influx through the NMDA receptor (Malenka and Nicoll 1999). This Ca2+ influx functions as a second messenger that modulates downstream signalling cascades involved in synaptic strength. These cascades include calmodulin, CaMkII, calcineurin, protein phosphatase 1 (PP1), PP2A, PP2B, protein kinase C (PKC), extracellular regulated kinase 1/2 (ERK1/2) and phosphatidylinositol-3-kinase (PI3K) (Soderling and Derkach 2000; Sweatt 2004).

Insulin has long been considered to be a peripheral hormone incapable of crossing the blood–brain barrier. However, the presence of insulin and its receptor has now been established in the brain (Havrankova et al. 1978; Adamo et al. 1989; Marks et al. 1990; Unger et al. 1991; Gerozissis and Kyriaki 2003). Although widely distributed, the insulin receptor is concentrated in specific brain regions, including the dendritic fields of hippocampal neurones and the olfactory bulbs, and the adrenergic terminals in the hypothalamus (Zhao and Alkon 2001), whereas insulin itself is particularly abundant in the hypothalamus and olfactory bulb (Schulingkamp et al. 2000). Brain insulin plays a role in the regulation of food intake and body weight (Schwartz et al. 1999; Hillebrand et al. 2002), and it may act as a neuromodulator, influencing the release and re-uptake of neurotransmitters (Sauter et al. 1983), neuronal survival (Yamaguchi et al. 2001; Zheng et al. 2002) and, probably, also learning and memory (Zhao et al. 1999; Zhao and Alkon 2001). Insulin binding to the insulin receptor leads to intracellular recruitment of insulin receptor substrates (IRS) and activation of phosphatidylinositol-3-kinase (PI3K) signalling (White 1997; Taha and Klip 1999; van der Heide et al. 2003). PI3K has many downstream targets that include survival-promoting kinases such as protein kinase B (PKB), and serum and glucocorticoid-regulated kinase (SGK) (Vanhaesebroeck and Alessi 2000). Besides PI3K signalling, direct recruitment of the adapter protein SHC to the insulin receptor leads to activation of the ERK1/2 pathway (White 1997, 1998). Insulin-induced activation of ERK1/2 has been suggested to underlie changes in synaptic plasticity and hippocampus-dependent learning and memory (Wu et al. 1999; Zhao et al. 1999; Izquierdo et al. 2000; Sweatt 2001). A role for PI3K is also emerging, as PI3K signalling has been implicated in the induction and maintenance of LTP (Kelly and Lynch 2000; Sanna et al. 2002) and specific forms of LTD. Application of 3,5-dihydroxyphenylglycine (DHPG), which selectively activates type I metabotropic glutamate receptors (mGluR), results in a phosphorylation of phosphoinositide-dependent kinase 1 (PDK-1), PKB and mammalian target of rapamycin (mTOR), and induces mGluR-dependent LTD. Specific kinase inhibitors of PI3K or mTOR prevent the DHPG-induced LTD (Hou and Klann 2004). mTOR is suggested to regulate immediate protein synthesis that is required for DHPG-induced LTD. Since mTOR has also been implicated in the regulation of LTP (Tang et al. 2002), it is very likely that it is involved in multiple forms of synaptic plasticity.

Interestingly, insulin application to hippocampal neurones results in a depression of excitatory synaptic transmission due to increased AMPA receptor endocytosis (Man et al. 2000). However, the intracellular signalling cascades involved in this insulin-induced LTD remain unclear. We investigated how insulin mediates LTD induction in the CA1 area of the hippocampus, and the underlying mechanism, by studying the role of the NMDA receptor and PI3K signalling. Our results show that LTD in the presence of insulin depends on low frequency synaptic activity (0.033 Hz) and NMDA receptor activation. In addition, we show that this form of LTD is indeed PI3K-dependent. Stimulation at different frequencies revealed that insulin shifts the frequency response curve of synaptic plasticity in an NMDA receptor-dependent manner. This suggests that insulin or insulin-like compounds function as neuromodulators of activity-dependent synaptic plasticity.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Preparation of hippocampal slices

Hippocampal slices (450 µm) were prepared from 2-week-old male Wistar rats (70–80 g) after isoflurane anaesthesia and decapitation, as described previously (Kamal et al. 2003). The slices were allowed to recover for at least 1 h at room temperature in artificial cerebrospinal fluid (ACSF) saturated with 95% O2 and 5% CO2. The composition of the artificial CSF was (in mm): NaCl, 124; KCl, 3.3; KH2PO4, 1.2; MgSO4, 1.3; CaCl2, 2.5; NaHCO3, 20; and glucose, 10.

Immunoblotting

Hippocampal slices were prepared and treated as described above. Six to seven slices were incubated with or without 500 nm insulin in ACSF saturated with 95% O2/5% CO2 at 30°C. After 15 min, slices were homogenized with ice-cold lysis buffer (pH 7.4) containing: 50 mm Tris, 1 mm EDTA, 1 mm EGTA, 1 mm phenylmethylsulfonyl fluoride, 100 mm sodium fluoride and 1 mm sodium vanadate on ice. After homogenization, Triton X-100 [0.5% (v/v)] was added to the lysates and incubated for 10 min on ice. Total protein in the supernatant fluid was determined by the bicinchoninic acid (BCA) method (Pierce, Rockford, IL, USA). Concentrated sodium dodecyl sulfate (SDS) sample buffer containing 66 mm Tris/HCl pH 6.8, 3% (w/v) SDS, 5% (v/v) glycerol, 0.001% (w/v) bromophenol blue and 2% (v/v) β-mercaptoethanol was added to the samples before they were heated for 10 min at 100°C. Equal amounts of sample protein (15–20 µg) were separated by electrophoresis on 11% SDS polyacrylamide gels. After electrophoresis, protein was transferred to polyvinylidene difluoride membranes (Amersham Biosciences Europe Gmbh, Roosendaal, the Netherlands) using a semi-dry Bio-Rad (Hercules, CA, USA) blotting apparatus, according to manufacturer's instructions. Protein transfer and blotting efficiency were checked with Coomassie stain (50% methanol, 10% acetic acid and 0.1% Coomassie Brilliant Blue R 250). After staining with Coomassie, blots were de-stained and washed with phosphate-buffered saline (PBS).

Protein detection

For PKB and phospho-PKB detection, blots were blocked in 4% non-fat milk powder, 0.05% Tween-20 in PBS for 1 h. Subsequently, blots were incubated with primary antibodies overnight at 4°C in a heat-sealable plastic bag. Total PKB and phospho-PKB (Ser473) polyclonal antibodies (New England Biolabs, Beverly, USA) were diluted in PBS containing 3% (w/v) non-fat milk powder and 0.05% (v/v) Tween-20. After incubation with primary antibodies, blots were washed extensively with wash buffer for 40 min [PBS containing 0.5% (w/v) non-fat milk powder and 0.05% (v/v) Tween-20]. Blots were incubated with horseradish peroxidase-coupled goat-anti-rabbit (Sigma, St Louis, MO, USA) diluted in wash buffer for 1 h at room temperature (RT). After extensive washing, immunoreactivity was visualized using the enhanced chemiluminescence detection kit (ECL; Boehringer Mannheim Gmbh, Mannhein, Germany) and hyperfilm (Amersham). To determine the linear range of hyperfilm ECL, serial dilutions of sample protein were transferred to polyvinylidene difluoride membranes, probed with an appropriate antibody, reacted with ECL reagent and exposed to film. The optical density (O.D.) versus dilution factor was plotted and was linear. Thus, quantification was performed using bands that were not saturated and had O.D. values within this range.

Each experiment was performed in duplicate and repeated four times. Results shown are representative examples of experiments performed.

Electrophysiological recordings

Slices were transferred to a submerged recording chamber where they were perfused at a rate of 1–2 mL/min with ACSF saturated with 95% O2 and 5% CO2 at 30°C. To prevent epileptiform activity during blockade of GABAergic inhibition, a surgical cut was made between CA3 and CA1. For intracellular recording of excitatory post-synaptic potentials (EPSPs), cells were impaled with sharp glass microelectrodes of 60–80 MΩ filled with potassium acetate (2 m, pH 7.5) and potentials were recorded using an Axoclamp-2B amplifier (Axon Instruments, Foster City, CA, USA) in bridge mode. The resting membrane potential (RMP) was determined when the membrane potential had stabilized after impalement of the cell (within 3 min). Schaffer collateral-commissural fibres were stimulated using a bipolar stainless steel stimulating electrode (0.1 mm diameter) placed in the stratum radiatum. At the beginning of each recording, a stimulus–response curve was obtained by using five stimulus intensities (I1–I5) ranging from stimulus intensities evoking threshold responses (I1) to intensities eliciting maximal responses (I5). The stimulus intensity that evoked half-maximum EPSPs was used throughout the experiment (test frequency 0.033 Hz). Only cells with stable synaptic responses, bridge-balance, RMP and input resistance were used for analysis. The data were normalized to the averaged value of the initial slope of the EPSP obtained during the 15 min period prior to the application of the conditioning stimulus.

In all experiments, the conditioning stimulation was applied in the presence of 10 µm bicuculline (Tocris Cookson, Bristol, UK) and 500 nm insulin, except when stated otherwise. dl-2-amino-5-phosphono-valeric acid (APV) (Sigma) or LY294002 (New England Biolabs) was added 15 min before the conditioning. To test whether a conditioning stimulation had a significant effect on the slope of the EPSP, a Wilcoxon's matched-pair test was used in which the baseline responses were compared with responses at 30 min. To compare the effect of different experimental protocols, we first performed an overall analysis of variance and subsequently, a Student's t-test for independent samples; all averages are listed as mean ± SEM.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Insulin activates the PI3K–PKB pathway in hippocampal slices

To investigate insulin signalling in the regulation of synaptic plasticity we first assessed whether insulin activated the PI3K pathway.

Treatment of coronal hippocampal slices with insulin for 15 min resulted in PKB phosphorylation on Thr473. Although ECL is a semi-quantitative technique, we quantified the western blots to estimate the increase in PKB phosphorylation induced by insulin. In the presence of 500 nm insulin, PKB phosphorylation increased more than 2.5-fold compared with control in the experiment shown in Fig. 1. In three additional, independent experiments insulin induced an increase in PKB phosphorylation at Thr473 ranging from 2.0- to 2.6-fold compared with the unstimulated control. On average, PKB phosphorylation was increased 2.4 ± 0.2 times (p = 0.002, Student's t-test, n = 4). Total PKB levels remained unaltered. Thr473 is phosphorylated by an upstream kinase dependent on PI3K activity. Therefore, insulin-induced PKB phosphorylation on Thr473 can be considered an indirect measure of PI3K activity. For PKB to achieve full catalytic activity, phosphorylation of Thr473 and Ser308 is required. We have previously shown that phosphorylation of PKB at Thr473 by insulin parallels phosphorylation of Ser308 and results in an increase in catalytic activity (van der Heide et al. 2003). Remarkable is the high level of PKB phosphorylation under non-stimulated conditions, which could reflect the role this kinase has in survival and maintenance of cellular integrity.

image

Figure 1. Insulin activates PKB in hippocampal slices. Hippocampal slices were treated for 15 min with 500 nm insulin in ACSF and cell lysates were analysed for PKB phosphorylation using a phospho-specific antibody. (a) Insulin increased phosphorylation of PKB (PKB-P) compared with controls. Blots were stripped and re-probed with an antibody that recognizes PKB independently of its phosphorylation state as a loading control (lower 4 lanes). This experiment was performed in duplicate. We observed the same results in four independent experiments. (b) The average densitometric quantification of the results of four independent experiments. Controls (– Ins) were set at 1 and compared relative to insulin-treated conditions (+ Ins). Insulin treatment caused a 2.4 ± 0.2-fold increase in PKB phosphorylation.

Download figure to PowerPoint

Insulin application induces LTD in CA1 neurones

In the presence of bicuculline, insulin application resulted in a rapid and robust decrease in EPSP to 47.6 ± 8.8% (p < 0.05), which was maximal after 10 min of insulin application and remained stable for the remainder of the experiment (Figs 2a and e). The decrease in EPSPs was also observed when insulin was applied for only 15 min (Fig. 2a inset). The results obtained are similar to those that others have shown (Man et al. 2000), and demonstrate the insulin responsiveness of our experimental preparation.

image

Figure 2. LTD induced in the presence of insulin is NMDA receptor-dependent. (a) EPSPs were recorded in CA1 pyramidal cells after stimulation (0.033 Hz) of Schaffer collateral-commissural fibres. GABAergic inhibition was blocked by the addition of 10 µm bicuculline. Insulin application in the presence of bicuculline (filled circles) resulted in LTD within 15 min and persisted for at least 30 min. When insulin was applied with intact inhibition (no bicuculline present; open circles) no change in the EPSPs was induced. Typical examples of EPSPs, recorded at the indicated time point, are shown as insets in the top of the figure. The inset inside the figure illustrates the effect of a 15 min insulin application (open bar) in the presence of bicuculline. (b) To test whether LTD in the presence of insulin requires activation of NMDA receptor, 500 nm insulin was applied in the presence of 100 µm APV. In the presence of APV, insulin did not induce any change in the EPSPs. (c) The involvement of NMDA receptors in LTD induction in the presence of insulin was further investigated by lowering the extracellular Mg2+ concentration from 1.3 mm to 0.2 mm. In 0.2 mm Mg2+ and 500 nm insulin, the slope of the EPSPs reduced significantly. (d) To determine whether LTD induced in the presence of insulin depended on evoked synaptic activation of NMDA receptors, 500 nm insulin was applied in the absence of afferent fibre stimulation. After stimulation was resumed, the slope of the EPSPs was not changed; only after stimulation at 0.033 Hz was continued, LTD did develop. (e) Bar graph summarizing the results of the experiments shown in (a) to (c). The bar graph represents the values of the last 5 min in the presence of 500 nm insulin in the absence of bicuculline (open bar), presence of bicuculline (black bar), the combined presence of bicuculline and APV (dark grey bar) and low extracellular Mg2+ (light grey bar). *Indicates p < 0.05, compared with baseline.

Download figure to PowerPoint

Subsequently, we investigated the requirement of the NMDA receptor in insulin-mediated LTD induction. In the presence of APV, a specific antagonist of the NMDA receptor, insulin did not affect EPSPs; after 30 min of insulin application, the slope of the EPSPs was 103.4 ± 8.8% (p > 0.05, Figs 2b and e). This shows that NMDA receptor blockade blocks the induction of insulin LTD and suggests that insulin-LTD is NMDA receptor-dependent.

With intact GABAergic inhibition (in the absence of bicuculline), application of 500 nm insulin did not affect the EPSPs (Figs 2a and e; 108.7 ± 8.8%, p > 0.05, 30 min after insulin was added). Bicuculline facilitates NMDA receptor activation by reducing GABAergic inhibition. Therefore, we tested whether insulin could induce LTD when the extracellular Mg2+ concentration was lowered from 1.3 mm to 0.2 mm, which also results in an increased activation of NMDA receptors. Under low extracellular Mg2+ concentrations, insulin significantly reduced the EPSP to 76.1 ± 5.1% (p < 0.05, comparing the effect of insulin with intact GABAergic inhibition and the effect of insulin in low extracellular Mg2+Figs 2a, c and e). The magnitude of the LTD observed under low Mg2+ concentration was not as high as with bicuculline, probably because there are still Mg2+ ions remaining.

To address whether the LTD in the presence of insulin required evoked release of glutamate, insulin was applied in the absence of synaptic stimulation. Insulin application for 15 min in the absence of electrical stimulation revealed that directly after electrical stimulation resumed, the slope of the EPSPs was still at baseline levels (99.7 ± 14.5%, p > 0.05, Figs 2d and e). Thereafter, during normal synaptic stimulation at 0.033 Hz, the EPSPs decreased gradually (26.9 ± 7.5% after 30 min of insulin, p < 0.05, Fig. 2d). These results show that synaptic stimulation is required for insulin-mediated LTD induction.

Insulin-mediated LTD is PI3K-dependent

Since PI3K activation has been implicated in insulin signalling and synaptic plasticity, we investigated whether PI3K activation is required for insulin-mediated LTD induction. LY294002 is a cell-permeable specific inhibitor of PI3K activity, and has been shown to inhibit insulin-induced PKB activation (van der Heide et al. 2003). Pre-treatment of hippocampal slices with LY294002 did not affect baseline EPSPs but after 30 min of insulin incubation, the slope of the EPSPs was reduced to 47.6 ± 8.8% in the absence of the PI3K inhibitor (Figs 2a and e). In the presence of LY294002, the EPSPs were significantly less reduced to 77.4 ± 9.9% (p < 0.05, comparing the amount of depression in the absence and presence of LY294002; Figs 2a and e, 3a and b), The reduced magnitude of the insulin-induced LTD in the presence of the PI3K inhibitor LY294002 provides evidence that PI3K is required for insulin-induced LTD.

image

Figure 3. LTD in the presence of insulin is PI3K-dependent. (a) The involvement of the PI3K pathway was assessed by determining the effect of the specific PI3K inhibitor LY294002 (50 µm) on the LTD induced in the presence of 500 nm insulin. In the presence of LY294002, the magnitude of the LTD was strongly reduced. Typical examples of EPSPs, recorded at the indicated time point, are shown as insets. (b) Summary bar graph representing the values of the last 5 min in the presence of 500 nm insulin. In the combined presence of LY294002 and insulin (filled bar), the slope of the EPSP at the end of the insulin application is no longer significantly different from baseline, whereas in the absence of LY294002 (open bar, Fig. 2), LTD was induced. *Indicates p < 0.05, compared with baseline.

Download figure to PowerPoint

Insulin shifts the frequency response curve of synaptic plasticity

To test whether application of 500 nm insulin has an effect on activity-dependent synaptic plasticity, we compared the frequency–response functions for the induction of LTD and LTP under control conditions and in the presence of insulin. Stimulation at test frequency (0.033 Hz) did not induce any change under control conditions, but it induced LTD in the presence of insulin (Fig. 2a). In the control conditions, slices receiving 1 Hz stimulation showed substantial LTD (65.3 ± 3.1% of baseline, n = 5, p = 0.05). Stimulation at 10 Hz did not induce any significant change (105.1 ± 6.2% of baseline, n = 5, p > 0.05), whereas 50 and 100 Hz stimulations resulted in LTP (156.4 ± 17.8% and 162.4 ± 21.4% of baseline for 50 and 100 Hz stimulation, respectively, n = 4, p < 0.05; Fig. 4). The LTD-LTP crossover point of the frequency–response function for induction of LTD and LTP was between 1 and 10 Hz. In the presence of 500 nm insulin, stimulation at 0.033 Hz resulted in LTD (Figs 2 and 4) and stimulation at 1 Hz did not induce any change (104.4 ± 3.1% of baseline, n = 4, p < 0.05; Fig. 4). All the higher frequencies yielded LTP (184.1 ± 6.2%, 174.3 ± 12.6%, 171.3 ± 18.4% of baseline for 10, 50 and 100 Hz stimulation, respectively, n = 4, p < 0.05; Fig. 4). Therefore, bath application of 500 nm insulin resulted in a leftward shift of the frequency–response function for the induction of LTD and LTP (Fig. 4), the LTD-LTP crossover point now being around 1 Hz.

image

Figure 4. The frequency–response curve for synaptic plasticity is shifted to the left in the presence of insulin. (a) The change in the slope of the EPSP at the end of the insulin application is plotted against the frequency (0.033, 1, 10, 50 and 100 Hz). The baseline level (100%) is indicated by the dotted line. Under control conditions (open circles), synaptic stimulation at 0.033 Hz had no effect on the slope of the EPSP; at 1 Hz, LTD is induced, at 10 Hz the EPSPs are not changed and at 50 or 100 Hz, LTP is induced. In the presence of 500 nm insulin (filled circles), synaptic stimulation at 0.033 Hz induced LTD, stimulation at 1 Hz did not affect the EPSPs and stimulation at 10, 50 and 100 Hz induced LTP. (b) To address the question of whether both the LTD at 0.033 Hz and the LTP at 10 Hz in the presence of insulin are dependent on NMDA receptor activation, experiments were performed in the presence of the NMDA receptor antagonist APV. Both LTD (see also Fig. 2b) and LTP were prevented in the presence of this NMDA receptor antagonist.

Download figure to PowerPoint

To address whether both LTD and LTP in the presence of insulin were NMDA receptor-dependent, we determined whether APV application prevented LTD induction as well as LTP induction. In the presence of APV, stimulation at 0.033 Hz did not induce any change (103.4 ± 8.8, n = 4, p > 0.05; Figs 2 and 4b), whereas LTP induced at 10 Hz in the presence of insulin was also completely blocked (184.1 ± 6.2% in the absence of APV and 98.8 ± 7.5% in the presence of APV, n = 4). The block of LTD and LTP induction in the presence of insulin by APV indicated that both are NMDA receptor-dependent.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Insulin application to hippocampal slices results in the induction of LTD when synapses are activated at 0.033 Hz and GABAergic inhibition is blocked (Man et al. 2000; Huang et al. 2004; Fig. 2). We investigated the mechanism underlying this form of LTD. Our results show that insulin potently activates PI3K signalling (Fig. 1) and mediates LTD induction at the CA1 Schaffer collateral synapses in an activity-, frequency-, NMDA receptor- and PI3K-dependent manner (Figs 2 and 3). Furthermore, insulin-mediated LTD induction was part of an NMDA receptor-dependent shift in the frequency response curve of activity-dependent synaptic plasticity (Fig. 4). Insulin shifts this curve to the left, thereby enabling both LTD and LTP induction at lower frequencies. Therefore, our results suggest that insulin or insulin-like compounds may function as neuromodulators that set the threshold for NMDA receptor-dependent LTD and LTP induction.

The shift in the frequency response curve is frequency- and NMDA receptor-dependent (Fig. 4), and may therefore share its underlying mechanism with insulin-mediated LTD induction. For this reason, insight into insulin-mediated LTD induction is highly relevant for understanding the effect of insulin on synaptic plasticity.

Previous studies have shown that an insulin-induced depression of excitatory synaptic transmission is dependent on intracellular Ca2+ (Man et al. 2000). The mechanism underlying the release or the source of the intracellular Ca2+ required for insulin LTD is, however, not known. Here, we show that that the NMDA receptor is required for insulin-mediated LTD induction (Fig. 2). Possibly, the NMDA receptor provides the rise in intracellular Ca2+ that is required for insulin-mediated LTD induction. APV, a specific antagonist of the NMDA receptor, could effectively block insulin-mediated LTD induction, as did the absence of synaptic stimulation at 0.033 Hz. Insulin-mediated LTD induction required blockade of GABAergic transmission with bicuculline or picrotoxin, but lowering of the extracellular Mg2+ concentration to unblock the NMDA receptor abolished the requirement to block GABAergic inhibition. This underlines the NMDA receptor-dependence of this form of LTD. In vivo GABAergic inhibition may set the spatio-temporal conditions for synaptic plasticity and hippocampal-dependent learning and memory (Paulsen and Moser 1998). Others have reported that insulin-mediated LTD induction is not NMDA receptor-dependent and instead, depends on l-type voltage-activated Ca2+ channels (VACCs) (Huang et al. 2004). If, indeed, only Ca2+ influx through VACCs was required for insulin-mediated LTD induction, APV treatment should have no effect on the induction of insulin-mediated LTD. Our finding that APV treatment blocked the insulin-mediated shift in the frequency–response curve indicates an involvement of the NMDA receptor in this effect. A combination of NMDA receptor activation and VACCs is unfeasible, since APV treatment fully prevents both LTD and LTP induction in the presence of insulin. Thus, in our hands, the effects of insulin on the frequency–response curve appear to be fully dependent on the NMDA receptor.

Insulin-induced PKB activation depends on PI3K activity, and the phospho-specific antibody used in this study is directed against a PKB site that is only phosphorylated by a PI3K-activated kinase (Lizcano and Alessi 2002). Therefore, the insulin-induced increase in PKB phosphorylation reflects an increased PI3K activity(van der Heide et al. 2003).

Previous studies have implicated PI3K activity in synaptic plasticity. Here, we show that LTD induced by 0.033 Hz stimulation in the presence of insulin is PI3K-dependent as the induction of insulin LTD could be prevented by LY294002, a specific PI3K inhibitor. Interestingly, LTP induction in the dentate gyrus of the rat is inhibited by wortmannin, an irreversible PI3K inhibitor (Kelly and Lynch 2000), and exerts an inhibitory effect on KCl-stimulated glutamate release and calcium influx (Kelly and Lynch 2000). Later, it was reported that PI3K activity is not needed for the induction of LTP itself but rather, is required for LTP maintenance in the hippocampal CA1 region (Sanna et al. 2002). Therefore, PI3K activity may be required for the insulin-induced shift in the frequency–response curve of synaptic plasticity, since it appears to be an important mediator of regular LTD and LTP.

A possible explanation of how insulin and PI3K shift the frequency–response curve lays in the potentiation of NMDA receptor currents by protein tyrosine kinases (PTKs)(Wang and Salter 1994). Interestingly, pp60c-src, a PTK, is activated by insulin in a PI3K-dependent manner (Shumay et al. 2002), potentiates NMDA receptor currents through tyrosine phosphorylation (Chen and Leonard 1996) and induces LTP in the CA1 area of the hippocampus (Lu et al. 1998). Consequently, the observed insulin-induced shift in the frequency response curve of synaptic plasticity (Fig. 4) might result from a PI3K-pp60c-src-dependent enhancement of NMDA receptor currents. Remarkably, a similar mechanism has been proposed for the effect leptin has on synaptic transmission in the CA1 field of the hippocampus (Shanley et al. 2001).

Another interesting candidate for interaction with PI3K signalling is PKC, which has been implicated in the DHPG-induced leftward shift in the frequency–response curve of synaptic plasticity (van Dam et al. 2004). The PI3K pathway and the PKC pathway have been shown to interact at multiple levels and can influence each other's activity. For instance, insulin-induced PI3K activity can induce certain PKC isoforms that provide negative feedback on IRS-1, which decreases subsequent PI3K activity (Liu et al. 2001). Activated PKC-zeta can directly bind to, and inactivate PKB activity (Doornbos et al. 1999). Interestingly, a calpain-cleaved form of PKC-zeta named PKM has been implicated in both LTP and LTD (Hrabetova and Sacktor 1996), but whether PKM interacts with PKB is unknown. Possibly, insulin and PI3K recruit PKC to mediate changes in synaptic plasticity.

Our results demonstrate that insulin shifts the frequency–response curve of synaptic plasticity to the left, thereby facilitating LTD and LTP induction at lower frequencies. The NMDA receptor and PI3K signalling are presumably central to these effects, acting as important regulators of synaptic plasticity.

Many growth factors in the brain, in addition to insulin, act on the PI3K pathway. These include insulin-like growth factor I (IGF-I) (Zheng and Quirion 2004), brain-derived neurotrophic factor (BDNF) (Zhu et al. 2002) and nerve growth factor (NGF) (Gerling et al. 2004). Interestingly, these factors have all been implicated in survival and the modulation of synaptic plasticity. IGF-1, insulin's closest relative, has also been implicated in hippocampal neurogenesis (O'Kusky et al. 2000) and learning and memory (Huang et al. 2004). This suggests that insulin and insulin-like factor have a pro-survival as well as a neuromodulatory function by acting on the PI3K pathway, linking survival directly to synaptic plasticity.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We would like to thank Dr. Pierre de Graan for comments on the manuscript. A.A. was supported by the ‘Catharijne Stichting’.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References