Distinct modulation of voltage-gated and ligand-gated Ca2+ currents by PPAR-γ agonists in cultured hippocampal neurons


  • Tristano Pancani,

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author
  • Jeremiah T. Phelps,

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author
  • James L. Searcy,

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author
  • Michael W. Kilgore,

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author
  • Kuey-Chu Chen,

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author
  • Nada M. Porter,

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author
  • Olivier Thibault

    1. Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose Street, MS 310, Lexington, Kentucky 40536-0298, USA
    Search for more papers by this author

Address correspondence and reprint requests to Olivier Thibault, Department of Molecular and Biomedical Pharmacology, University of Kentucky Medical Center, 800 Rose street, MS310, Lexington, KY 40536-0298, USA. E-mail:othibau@uky.edu


Type 2 diabetes mellitus is a metabolic disorder characterized by hyperglycemia and is especially prevalent in the elderly. Because aging is a risk factor for type 2 diabetes mellitus, and insulin resistance may contribute to the pathogenesis of Alzheimer’s disease (AD), anti-diabetic agents (thiazolidinediones-TZDs) are being studied for the treatment of cognitive decline associated with AD. These agents normalize insulin sensitivity in the periphery and can improve cognition and verbal memory in AD patients. Based on evidence that Ca2+ dysregulation is a pathogenic factor of brain aging/AD, we tested the hypothesis that TZDs could impact Ca2+ signaling/homeostasis in neurons. We assessed the effects of pioglitazone and rosiglitazone (TZDs) on two major sources of Ca2+ influx in primary hippocampal cultured neurons, voltage-gated Ca2+ channel (VGCC) and the NMDA receptor (NMDAR). VGCC- and NMDAR-mediated Ca2+ currents were recorded using patch-clamp techniques, and Ca2+ intracellular levels were monitored with Ca2+ imaging techniques. Rosiglitazone, but not pioglitazone reduced VGCC currents. In contrast, NMDAR-mediated currents were significantly reduced by pioglitazone but not rosiglitazone. These results show that TZDs modulate Ca2+-dependent pathways in the brain and have different inhibitory profiles on two major Ca2+ sources, potentially conferring neuroprotection to an area of the brain that is particularly vulnerable to the effects of aging and/or AD.

Abbreviations used

Alzheimer’s disease


days in vitro




NMDA receptor




peroxisomal proliferator-activated receptor


peroxisome proliferator response element




type 2 diabetes mellitus








voltage-gated Ca2+ channel

Several studies have reported that similar neuropathological mechanisms are present in aging and/or Alzheimer’s disease (AD) and in diabetes mellitus. An increased risk of dementia (Alzheimer’s and vascular type) is seen in people with diabetes (Ott et al. 1999; Schnaider Beeri et al. 2004; Biessels et al. 2006), and aging is a recognized risk factor for both AD and diabetes (Hoyer 2004; Yaffe et al. 2004; Launer 2005; McNay 2005; Reagan 2007). Furthermore, insulin resistance associated with type 2 diabetes mellitus (T2DM) may be an important pathogenic factor in AD (Watson and Craft 2003; Steen et al. 2005; Li and Holscher 2007), and thought to be present in 80% of the AD population (Janson et al. 2004). It has been suggested that cognitive dysfunction in older adults and AD patients might reflect a CNS state of diabetes characterized by reduced insulin signaling and/or resistance, and termed ‘Type 3 diabetes’ (Lannert and Hoyer 1998; Gasparini et al. 2002; Steen et al. 2005; de la Monte et al. 2006). In older T2DM adults, cognitive dysfunction is associated with poor control of diabetes (Munshi et al. 2006), and improving glycemic control leads to improved cognition and memory (Ryan et al. 2006). Therefore, it appears that common underlying mechanisms coexist in T2DM, aging and/or AD.

Peroxisome proliferator-activated receptor (PPAR)-γ is a nuclear receptor that is a target of thiazolidinediones (TZDs) and regulates the expression of genes controlling inflammation, metabolism, cell growth, and differentiation (Berger and Moller 2002; Hauner 2002). PPAR subtypes (α, β, and γ) are present in almost all cells, including adipocytes, fibroblasts, macrophages, microglia, and neurons (Berger and Moller 2002; Hauner 2002; Cimini et al. 2005; Inestrosa et al. 2005). TZDs’ mechanisms of action and biological effects are cell-type specific (Berger et al. 2005) and are also dependent on the structural makeup of the ligand (Nolte et al. 1998; Rangwala and Lazar 2002; Zhang et al. 2007; Gani and Sylte 2008). In adipocytes, TZDs promote lipid accumulation by increasing gene transcription for several fatty acid binding proteins (Schoonjans et al. 1996; Martin et al. 1997; Berger and Moller 2002). TZDs also appear capable of reestablishing insulin sensitivity in fat, muscle, and liver tissues, by decreasing plasma free fatty acids, lipid levels (Berger et al. 2005), and inflammation (Hofmann et al. 1994).

In a preliminary study, TZD treatment with rosiglitazone during early stages of AD has been shown to improve cognition (Watson et al. 2005). Testing the effects of rosiglitazone using the AD assessment scale (ADAS-Cog) in a larger population of AD patients has revealed that patients lacking the APOE4 allele appear selectively protected by the TZD (Risner et al. 2006). Several studies have focused on the mechanisms of action and functional role of TZDs in the brain (Feinstein 2003; Sundararajan et al. 2006; Jiang et al. 2008; Lichtor et al. 2008). In models of AD and neurodegeneration, mechanisms identified include decreased inflammation (Combs et al. 2000; Heneka et al. 2000, 2005; Daynes and Jones 2002; Garcia-Bueno et al. 2005; Hunter et al. 2007; Xing et al. 2008), decreased oxidative stress (Aoun et al. 2003; Garcia-Bueno et al. 2005; Kumar et al. 2009), decreased Aβ load (Combs et al. 2000; Yan et al. 2003; Camacho et al. 2004; Heneka et al. 2005; Jiang et al. 2008), improved mitochondrial function/number (Dello Russo et al. 2003; Feinstein et al. 2005; Fuenzalida et al. 2007; Strum et al. 2007; Hunter et al. 2008), and increased glial glutamate uptake (Romera et al. 2007).

Despite evidence that TZDs are neuroprotective, few studies have considered the impact of PPAR-γ agonists on neuronal Ca2+ homeostasis. This is surprising given the degree of neuronal Ca2+ dysregulation present in aging/AD, and the evidence that in animal models of diabetes, alterations in Ca2+ signaling are seen. In the streptozotocin model of diabetes, increased cytosolic and mitochondrial Ca2+ (Huang et al. 2002, 2005), larger Ca2+ action potentials and Ca2+-dependent afterhyperpolarization, and reduced intracellular Ca2+ release (Huang et al. 2002; Kamal et al. 2003; Kruglikov et al. 2004) are seen. Further, in this model, alterations in the NMDA receptor (NMDAR) complex are thought to mediate deficits in learning and memory (Biessels et al. 1998; Li and Wei 2001; Gardoni et al. 2002). Recent evidence also shows the PPAR-γ agonist pioglitazone attenuates memory impairment in the intracerebral streptozotocin model (Pathan et al. 2006). Here, therefore, we tested the hypothesis that pioglitazone and rosiglitazone could reduce signaling through two aging-sensitive neuronal Ca2+ targets, the voltage-gated Ca2+ channel (VGCC) and NMDAR. Because Ca2+ dysregulation is a hallmark of brain aging that is linked to cognitive decline (Disterhoft et al. 1996; Thibault and Landfield 1996; Moyer et al. 2000; Tombaugh et al. 2005; Murphy et al. 2006), re-establishment of Ca2+ homeostasis with PPAR-γ agonists might represent a mechanism by which these compounds improve cognition with aging/AD.

Materials and methods

Cell cultures

Mixed (neuron/glia) hippocampal cultures were established from pregnant Sprague-Dawley rats as previously described (Porter et al. 1997). Briefly, hippocampi from E18 fetuses were removed and treated with trypsin 0.25% in Hank's balanced salt solution for 10 min. The cell suspension was triturated, diluted with minimum essential medium at a final concentration of 5 × 105 cells/mL, and added to 35 mm culture dishes. For electrophysiological experiments, the 35 mm culture dish contained three plastic coverslips (Corning Inc., Corning, NY, USA). For Ca2+ imaging experiments 35 mm glass bottom culture dishes were used (Mattek Corp., Ashland MA, USA). All dishes were coated with poly-l-lysine, and were maintained in an incubator until used at 13–16 days in vitro (DIV). For PPAR-γ binding and western blot assays, six-well plates (35 mm/well; Corning) coated with poly-l-lysine were used. Approximately 1 × 106 cells were plated in each well and used at 13–16 DIV. Solutions and media were obtained from Invitrogen Corp. (Carlsbad, CA, USA).

PPAR-γ binding assay

Quantitative measurement of PPAR-γ activation was achieved using an ELISA-based assay (TransAM kit; Active Motif, Carlsbad, CA, USA). The assay was performed according to the manufacturer’s protocol. Signals were normalized to vehicle-treated cells [dimethylsulfoxide (DMSO)] and were background corrected. In brief, cells from at least three 35 mm dishes were harvested for each treatment group and processed to obtain a purified nuclear extract. The TransAM kit was supplied with a 96 well plate containing the peroxisome proliferator response element (PPRE) sequence bound at the bottom of each well, the primary antibody against the activated form of PPAR-γ and the secondary antibody (horseradish peroxidase conjugated). The nuclear extract was added to each well to let the activated PPAR-γ bind specifically to the bound oligonucleotide. After washing, the sample was incubated with the primary antibody and then with the secondary antibody. Quantification of the PPRE-bound PPAR-γ colorimetric reaction was obtained using a HTS7000+ plate reader (PerkinElmer, Waltham, MA, USA).

Western blots

The nuclear fraction from at least six 35 mm dishes per condition was isolated as above for the PPAR-γ binding assay. Total protein content was evaluated using a Bradford assay and results were used to load the same amount of protein per lane on a 10% Tris-HCl gel (Bio-Rad, Hercules, CA, USA). After electrophoresis, proteins were transferred to a nitrocellulose membrane and then incubated overnight at 4°C with the primary antibody against the N-terminus fragment of PPAR-γ 1 : 1000 (Santa Cruz Biotechnologies, Santa Cruz, CA, USA). The membrane was then incubated for 1 h with the secondary antibody, horseradish peroxidase-conjugated 1 : 5000 (Santa Cruz), and developed after treatment with ECL plus. The membrane was exposed to a radiographic film (ISCBioExpress, Kaysville, UT, USA), and imaged/quantified on a Gel Logic 2200 Imaging System (Kodak Inc., Rochester, NY, USA). The same procedure was used for β-actin immunostaining. Signal intensities for each lane was normalized to β-actin signal for that lane.

Electrophysiology and Ca2+ imaging

VGCC recording solutions

The extracellular solution was as follows (in mM): 111 NaCl, 5 BaCl2, 5 CsCl, 2 MgCl2, 10 Glucose, 10 HEPES, 20 tetraethylammonium (TEA)-Cl, pH 7.35 with NaOH. Tetrodotoxin (500 nM) was added before recording. Intracellular pipette solution was (in mM): 145 methanesulfonic acid, 10 HEPES, 3 MgCl2, 11 EGTA, 1 CaCl2, 13 TEA-Cl, 14 phosphocreatine Tris-salt, 4 Tris-ATP, 0.3 Tris-GTP, pH 7.3 with CsOH.

NMDAR recording solutions

The extracellular solution was as follows (in mM): 145 NaCl, 2.5 KCl, 10 HEPES, 10 Glucose, 2 CaCl2, 10 TEA-Cl, pH 7.35 with NaOH. Glycine (10 μM), tetrodotoxin (500 nM), and 6-cyano-7-nitroquinoxaline-2,3-dione (10 μM) were added immediately prior to recording. Intracellular pipette solution was (in mM): 145 methanesulfonic acid, 10 HEPES, 11 EGTA, 1 CaCl2, 14 phosphocreatine Tris-salt, 4 Tris-ATP, 0.3 Tris-GTP, pH 7.3 with CsOH.

Ca2+ imaging solutions

Bath solution contained (in mM): 145 NaCl, 2.5 KCl, 10 HEPES, 10 d-glucose, 2 CaCl2, 1 MgCl2, 0.01 glycine, pH 7.3 with NaOH.

Whole-cell recording

Electrodes were made from glass capillary tubes (Drummond Scientific, Broomall, PA, USA) using a P-87 micropipette puller (Sutter Instruments, Novato, CA, USA), coated with polystyrene Q-dope (GC Electronics, Rockford, IL, USA) and fire polished before recording. A coverslip was taken from the incubator, rinsed twice with recording solution and placed in the recording chamber containing 2 mL of the extracellular solution. The recording chamber was then fixed on the stage of an E600FN microscope (Nikon Inc. Melville, NY, USA). An Axopatch-1D amplifier (Molecular Devices, Sunnyvale, CA, USA) and a Digidata 1200 (Molecular Devices) were used in combination with pClamp7 (Molecular Devices) to control membrane voltage and to acquire current records (5–10 KHz). Tip resistance was 3.2 ± 0.7 MΩ. Junction potential was nulled prior to each experiment and pipette capacitance was compensated. Once whole cell configuration was achieved, a 5–10 min run-up period was allowed prior to recording of currents. Data were lowpass filtered at 2–5 kHz. For VGCC recording, an I/V relationship was conducted (from −60 to +30 mV, 150 ms step) to identify the voltage step corresponding to the largest current. Each cell was then held at −70 mV or −40 mV and currents were recorded in response to a step depolarization (150 or 350 ms, respectively) eliciting the largest current (as determined from the I/V). For each cell, leak subtraction was accomplished online using a fractional method (5–8 scaled hyperpolarizing sub-pulses). Series resistance compensation was not routinely performed since we have found that at these current amplitudes, compensation (∼80%) has minimal impact (Porter et al. 1997). For NMDA-mediated current recording, cells were held at −70 mV and the current elicited by exposure to NMDA was monitored (see drug delivery below). All experiments were performed at room temperature (22–24°C). Because cell size contributes to current amplitude, all currents were normalized to cell capacitance (pF) and are reported as current density measures (pA/pF). Membrane capacitance was derived from the integral of the capacitative transient evoked by a 150 ms/10 mV hyperpolarizing pulse from −70 mV (Porter et al. 1997). For each cell, input resistance was determined from the steady current necessary to hyperpolarize the cell by 10 mV.

Ca2+ imaging

Glass pieces from broken 35 mm glass bottom dishes containing hippocampal cells were incubated for 30 min in the dark, in the Ca2+ imaging solution supplemented with 2 μM final Fura-2 AM, 0.085% final DMSO, and 0.015% final Pluronic F127 (Molecular Probes, Invitrogen). Following a 20 min deesterification period in indicator-free imaging solution, glass fragments were transferred to an imaging chamber (glass-bottom 35 mm dish) on the stage of an E600FN microscope (Nikon Inc.). Excitation of the fluorophore (340 and 380 nm excitation) was achieved using a high speed filter changer (Lambda DG4, Sutter instruments), and an Andor iXon EMCCD camera (Andor Technology, Belfast, Ireland) was used to capture emitted light (510 nm). Data acquisition and analysis were performed using Imaging Workbench 5.0 (Indec BioSystems, Santa Clara, CA, USA). Signal intensity from the somatic area of each cell was background subtracted by removing the average signal from an area adjacent the cell imaged but devoid of cellular components. Ratios were then converted to absolute Ca2+ levels using the following equation [Ca2+] = KDβ (R−Rmin)/(Rmax−R) where R is the 340/380 emission ratio, and Rmin and Rmax represent ratios of the lowest and the highest standard Ca2+ concentration imaged (respectively). Calibration of the ratios was accomplished using a series of increasing free Ca2+ concentrations from 0 to 39 μM containing 1 mM Mg2+(Invitrogen). Rmin, Rmax, and KDβ were determined by fitting ratios from the calibration curve with a 4-terms sigmoid function (SigmaPlot, Systat, Chicago, IL, USA), and were 0.29, 5.17, and 2.14 μM, respectively.

Drugs and drug delivery

Thiazolidinedione concentrations and exposure durations were chosen to match conditions where these drugs are protective in culture (Uryu et al. 2002; Dello Russo et al. 2003; Camacho et al. 2004), take into consideration the higher affinity of rosiglitazone (ROSI) for PPAR-γ compared to pioglitazone (PIO) (Willson et al. 1996; Awais et al. 2007), and are clinically relevant (Asano et al. 1999; Brunton et al. 2005; Feinstein et al. 2005). ROSI and T0070907 (TIO) were purchased from Cayman Chemical (Ann Arbor, MI, USA). TIO, a selective/irreversible PPAR-γ antagonist promotes the recruitment of co-repressors and blocks agonist-induced recruitment of co-activators (Lee et al. 2002). PIO was donated by MW Kilgore. These drugs were dissolved in DMSO and stored at −20°C until used as a 1 : 1000 dilution into cell culture medium. Cells were washed twice following exposure (2, 24, 72 h) and TZDs were absent during Ca2+ imaging or electrophysiological experiments. Vehicle-treated cultures were exposed to 0.1% DMSO for similar times. NMDA (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in water and stored at −20°C. All other drugs and salts were purchased from Sigma-Aldrich and were dissolved in HPLC-grade water. NMDAR-mediated Ca2+ currents and transients were recorded in response to NMDA exposure delivered with a SF77A rapid solution exchange system (Warner Instruments Corp., Hamden, CT, USA). This system is composed of a computer-controlled stepper motor which rapidly switches the position of separate glass barrels (700 micron-wide) carrying a constant stream of recording solutions across the cells (∼300 μL/min). Barrels are placed ∼ 75 microns above the cells and control the environment surrounding the cells. For NMDAR-mediated currents, a 500 ms, 300 μM NMDA exposure was used, and for Ca2+ imaging, a 5 s application of the same concentration was used. This NMDA concentration was chosen because it induces maximal currents in hippocampal cultures recorded at this DIV (Brewer et al. 2007).


Cells with membrane resistance < 300 MΩ, holding current > 200 pA (from −70 mV holding potential) or resting Ca2+ levels > 200 nM were removed from the analysis. Effects of treatments on electrophysiologic and imaging variables were assessed with one-way anova (with repeated measure as noted) using Prism (GraphPad software, La Jolla, CA, USA). Tukey’s post-hoc test was used for pairwise comparisons. A one-way anova by rank analysis (Kruskal-Wallis) was used for statistical comparisons on DNA binding fold change analyses, and a t-test was used for analysis of western blots. All data presented in graph form represent means ± SEM.


PPAR-γ DNA binding and PPAR-γ protein expression

In order to assess the involvement of PPAR-γ activation in response to TZD treatment, we monitored DNA binding using nuclear extracts from hippocampal neurons treated 24 h with vehicle (0.1% DMSO), PIO (0.1–10 μM), or ROSI (1 μM). Nuclear extracts were hybridized to an ELISA-based PPAR-γ assay which quantifies PPAR-γ DNA binding to PPREs (Fig. 1a). The significant increase in DNA binding following 24 h treatment with 10 μM PIO (anova, < 0.05) was reduced by the selective PPAR-γ antagonist TIO (1 μM). Treatment of cultures with TIO alone did not reduce PPAR-γ DNA binding below control conditions (anova, > 0.05), indicating little endogenous PPAR-γ activation is present in hippocampal cultures. Surprisingly, at 24 h treatment with 1 μM ROSI, DNA binding was not enhanced while nuclear PPAR-γ protein levels were elevated, as reported by western blot techniques (Fig. 1b). Because ROSI did not enhance DNA binding, we did not test for reversibility with a combination of ROSI and TIO. These results suggest that selective modulation of PPAR-γ via different agonists may affect DNA binding and protein expression differently (see Discussion). Furthermore, results from the TIO experiments indicate that under the conditions tested, the antagonist can significantly reduce DNA binding. In subsequent experiments, we use this antagonist to test the dependence of PIO and ROSI’s effects on VGCCs and NMDARs through PPAR-γ activation.

Figure 1.

 Effects of PIO and ROSI on PPAR-γ DNA binding and protein expression. (a) shows dose-dependent increase in PPAR-γ DNA binding in response to 24 h treatment with increasing concentrations of pioglitazone (PIO) or rosiglitazone (ROSI). T0070907 (TIO) the selective PPAR-γ antagonist prevented binding. Statistical comparisons were made against a vehicle treatment group (0.1% DMSO) (= 3–15/group, *< 0.05, anova). (b) shows a representative western blot (top) and group data responses (bottom) to 24 h treatment with 10 μM PIO or 1 μM ROSI (= 3/group, *< 0.05, t-test).

Rosiglitazone but not pioglitazone reduces VGCC currents

We used patch clamping techniques on hippocampal neurons treated 24 h with PIO or ROSI to record VGCC currents and monitor the effects of these two TZDs. Figure 2 shows representative examples of VGCC current traces (a, c, e), mean VGCC current densities (b, d, f), and I/V relationships (g) measured across several treatment conditions. No main effect of drug on VGCC current density (Fig. 2b) or passive membrane properties (Table 1) were detected at three concentrations of PIO tested (0.1 not shown, 1 and 10 μM; anova, > 0.05). Similar results were seen with 2 h (= 7–12), or 72 h (= 15–17) PIO treatment at 10 μM (data not shown). In contrast, 24 h ROSI treatment significantly reduced VGCC current density measured from −70 mV (Fig. 2d; anova; < 0.05) or −40 mV (data not shown). Interestingly, greater ROSI-mediated VGCC inhibition was noted on currents elicited from −40 mV compared to −70 mV (data not shown, < 0.0005), suggesting high voltage-activated L-type currents might be more sensitive than low voltage-activated Ca2+ currents to the inhibitory effects of ROSI. In vascular myocytes, rapid PPAR-γ-independent inhibitory effects of ROSI on L-VGCC currents have been observed (Knock et al. 1999). A direct effect of ROSI on L-VGCC is not likely the mechanism of inhibition reported here, however, because co-treatment of cultures with ROSI (1 μM) and the PPAR-γ antagonist TIO (1 μM) for 24 h, was able to completely reverse ROSI’s effects on VGCCs (Fig. 2d). Further, ROSI was not present during these experiments. TIO alone for 24 h (= 18), or ROSI for 2 h (= 6, data not shown) had no detectable effect on VGCC current densities (Fig. 2e and f). As seen in the I/V relationship analysis (Fig. 2g), ROSI inhibited peak currents density but did not shift the activation curve and voltage-sensitivity of the currents. Taken together, these data demonstrate that ROSI’s effects on VGCC are dependent on PPAR-γ activation, and also suggest little basal PPAR-γ modulation of these Ca2+ currents.

Figure 2.

 Effects of 24 h treatment with TZDs on VGCC currents. (a), (c) and (e) show representative examples of whole-cell currents elicited upon depolarization from −70 to 0 mV for 150 ms (voltage steps are shown above each trace). (b), (d) and (f) show peak current densities (mean ± SEM) recorded across several treatment conditions. Dashed line represents the peak current density measured in control cells. ROSI 1 μM, but not PIO significantly reduced peak current densities. While TIO did not change peak current density when tested alone, it was able to prevent the inhibition caused by ROSI (d). Asterisks indicates significant difference from the vehicle-treated group (0.1% DMSO-CTRL), (= 10–47 neurons/group; < 0.01; anova). I/V relationship analysis shows ROSI, while inhibiting peak current densities (* 0.05; Tukey’s post-hoc test), did not shift VGCC current activation.

Table 1.  Passive membrane properties
GroupCm (pF)Rm (MΩ)Ra (MΩ)HC (pA)
  1. Values represent means ± SEM on measures of membrane capacitance (Cm), membrane resistance (Rm), access resistance (Ra), and holding current (HC) necessary to hold a cell at −70 mV (= 10–52 cells/group). No significant difference was found in any of these parameters across treatment groups (> 0.05, one-way anova).

CTRL75.0 ± 3.2543.5 ± 24.06.7 ± 0.3−73.0 ± 6.6
PIO 1 μM61.4 ± 3.9504.0 ± 54.17.0 ± 0.7−109.0 ± 20.1
PIO 10 μM82.0 ± 5.6519.0 ± 33.07.1 ± 0.4−92.4 ± 18.6
PIO 10 μM + TIO 1 μM67.7 ± 5.4628.6 ± 58.04.4 ± 0.7−50.6 ± 11.5
ROSI 1 μM78.6 ± 3.3453.3 ± 33.77.3 ± 0.6−89.0 ± 14.7
ROSI 1 μM + TIO 1 μM74.0 ± 5.8546.1 ± 80.05.8 ± 0.6−84.7 ± 21.0
ROSI 10 μM62.6 ± 6.4464.4 ± 22.05.8 ± 0.7−81.5 ± 12.2
TIO 1 μM64.1 ± 4.4661.0 ± 64.96.8 ± 0.5−52.0 ± 8.8

Pioglitazone but not rosiglitazone reduces NMDAR-induced Ca2+ currents

We assessed the effect of PIO and ROSI on NMDAR-mediated Ca2+ currents, another primary source of Ca2+ in hippocampal neurons. Figure 3 shows representative NMDA-induced Ca2+ current traces (a, c, e) and mean current densities (b, d, f) measured across several treatment conditions. Compared to control-treated cells, 24 h PIO treatment significantly reduced NMDAR-induced Ca2+ currents (500 ms, 300 μM NMDA exposure) (Fig. 3a and b; anova; < 0.05). This effect was prevented with co-treatment of cultures using both PIO (10 μM) and TIO (1 μM) for 24 h. TIO alone for 24 h (= 11) did not modify NMDA-induced Ca2+ currents (Fig. 3e and f), suggesting that there is little PPAR-γ tone on hippocampal neurons in these culture conditions. Conversely, 24 h ROSI treatment did not reduce NMDAR-induced currents (Fig. 3d; anova; > 0.05). For this reason, we did not monitor Ca2+ levels during NMDA application following a 24 h ROSI treatment. These data show that NMDARs are a novel target of PIO whose inhibition appears dependent on PPAR-γ activation.

Figure 3.

 Twenty-four hour PIO reduces NMDAR-mediated currents. (a), (c) and (e) show representative current traces in response to a 500 ms, 300 μM NMDA pulse (horizontal bar). Cultures were treated either with DMSO (0.1%), PIO (10 μM), ROSI (1 μM), or TIO (1 μM) alone or in combination with the agonists. (b), (d) and (f) show corresponding mean group current density data (means ± SEM) recorded under these treatment conditions (= 6–34). While PIO significantly reduced NMDAR-mediated current density (*< 0.05; anova), ROSI had no significant effect (c and d). The PIO effect on NMDAR-mediated current was prevented when TIO was added in combination with PIO for 24 h. TIO alone did not affect NMDA-induced Ca2+ currents (f).

Pioglitazone reduces NMDAR-mediated Ca2+ levels

In order to identify whether the effects reported thus far could have an impact on Ca2+ homeostasis during NMDAR-mediated activation, we monitored intracellular Ca2+ levels using the ratiometric Ca2+ indicator Fura-2. Effects of chronic PIO (24 h) treatment on NMDAR-induced intracellular Ca2+ elevations (5 s, 300 μM NMDA exposure) were measured and compared to control conditions. In response to NMDA, robust increases in [Ca2+] (∼ 2.0 μM) were seen in cells treated 24 h with 0.1% DMSO, however, the same NMDA exposure yielded much reduced (∼1.4 μM) [Ca2+] in cells treated 24 h with 10 μM PIO (Fig. 4). While a significant reduction in [Ca2+] was seen during the NMDA exposure (anova; < 0.0001), PIO had no effect on resting Ca2+ concentrations. These results are consistent with the interpretation that PIO’s neuroprotective effects might be mediated, at least in part, by rectification of Ca2+ homeostasis during an insult.

Figure 4.

 Twenty-four hour PIO treatment reduces intracellular NMDA-induced [Ca2+]. (a) shows representative Fura-2-loaded hippocampal neurons used for ratiometric analyses on intracellular Ca2+ concentrations taken before (baseline), and during (peak) NMDA exposure (300 μM, 5 s) in control-treated (CTRL) and PIO-treated cells. (b) shows group data in control-treated (CTRL) and PIO-treated cells (= 32–35) imaged before, during (thick horizontal bar) and after NMDA exposure (*< 0.05; Tukey’s post-hoc test).


Elevated and long-lasting increases in cytoplasmic [Ca2+] are consistent biomarkers of aging that represent one aspect of neuronal Ca2+ dyshomeostasis (Kirischuk and Verkhratsky 1996; Thibault et al. 2001; Murchison et al. 2004; Hemond and Jaffe 2005; Gant et al. 2006). Enhanced Ca2+ transients are also present in AD models (Mattson and Guo 1997; LaFerla 2002; Stutzmann et al. 2007; Cheung et al. 2008) and in cells from AD patients (Ito et al. 1994; Gibson et al. 1996; Hirashima et al. 1996; Etcheberrigaray et al. 1998). Because VGCC and NMDAR blockers show improved cognition in animal models of aging (Deyo et al. 1989; Moyer et al. 1992; Kowalska and Disterhoft 1994), and can reduce risk of dementia in hypertensive or AD patients (Forette et al. 2002; Reisberg et al. 2003; Lipton 2004), we sought to test the hypothesis that TZDs could reduce Ca2+ signaling in neurons. Furthermore, while there has been some reports that PPAR-γ agonists (including ROSI) reduce [Ca2+] during Aβ-mediated insults in cortical neurons (Inestrosa et al. 2005), neither troglitazone nor the endogenous PPAR-γ agonist 15d-PGJ2 appear to reduce [Ca2+] during an NMDA-mediated insult (Uryu et al. 2002; Zhao et al. 2006). We show here that indeed, TZDs can have very different effects, and that the only two clinically-relevant TZDs approved for the treatment of T2DM (ROSI and PIO) reduce Ca2+ flux into hippocampal neurons by targeting two different ion channels. PIO reduced NMDAR-mediated Ca2+ currents and transients but had little to no impact on VGCC currents. Conversely, ROSI reduced VGCC currents, but not NMDAR-mediated currents.

Pioglitazone’s effects on NMDAR-mediated currents and Ca2+ transients provide evidence for new mechanisms associated with its neuroprotective effects. It is well documented that Ca2+-mediated toxicity (excitotoxicity) occurs in response to prolonged activation of NMDARs (Choi et al. 1988; Mattson et al. 1991; Lipton and Rosenberg 1994; Olney 2002) and that NMDAR antagonism, provides neuroprotection in several culture models of aging and neurodegeneration, by reducing Ca2+ levels during an insult (Choi et al. 1988; Levy and Lipton 1990; Zhou and Baudry 2006; Brewer et al. 2007). Use of the low affinity uncompetitive NMDAR antagonist memantine has been associated with a slowing of disease progression in patients with moderate-to-severe forms of AD (Reisberg et al. 2003). Thus, one of PIO’s neuroprotective actions might be mediated, in part, by a reduction in NMDAR-mediated Ca2+ currents and levels, particularly in conditions associated with NMDA over-activation (e.g., neurodegeneration, ischemia, and stroke). Further, because PIO’s effect on Ca2+ currents was reversed by the PPAR-γ antagonist TIO (Fig. 3a and b), the underlying mechanism appears to be transcriptionally mediated.

In contrast to PIO, ROSI specifically reduced Ca2+ signaling through VGCCs. As shown in Fig. 2, this effect was precluded by TIO treatment, indicating that ROSI’s effects on VGCCs also are transcriptionally regulated. These results provide insights into the mechanisms underlying ROSI’s effects on cognition. Functional memory improvement has been seen in response to ROSI treatment in vivo (see introduction), and it is tempting to speculate that this effect might be mediated by reduction of signaling through VGCC. Indeed, use of VGCC blockers (e.g., dihydropyridines) is associated with improved learning and memory in animal models of aging (Deyo et al. 1989; Moyer et al. 1992; Kowalska and Disterhoft 1994). In humans, use of VGCC blockers for the treatment of hypertension has been associated with a significant improvement in cognition (Trompet et al. 2008), and a reduced risk for developing dementia (Forette et al. 2002; Khachaturian et al. 2006). However, because ROSI does not appear to appreciably cross the blood brain barrier (Maeshiba et al. 1997; Pedersen et al. 2006), prior beneficial effects of ROSI on brain function have mostly been attributed to peripheral actions, such as reestablishment of normal glucose (Pedersen and Flynn 2004; Ryan et al. 2006) or insulin levels (Watson et al. 2005; Risner et al. 2006). Still, peripherally administered ROSI has been shown to increase PPAR-γ DNA binding in the brain (Luo et al. 2006), and may more readily gain access under conditions associated with a weakened blood brain barrier such as aging/AD (Gemma et al. 2004; Watson et al. 2005). Irrespective of its pharmacokinetic profile, ROSI’s reduction in VGCC function in the brain may represent a new and beneficial mechanism of action which draws a parallel with the use of L-type VGCC blockers and their association with a decreased incidence of dementia.

Thiazolidinediones have been shown to activate both PPAR-γ-dependent and independent pathways in the brain (Feinstein et al. 2005; Sundararajan et al. 2006; Kapadia et al. 2008). We show here that treatment of hippocampal cultures with PIO increases PPAR-γ binding to PPREs at 24 h, and that this event is significantly reduced by TIO (Fig. 1). Further, the effects reported here on VGCC- and NMDAR-mediated Ca2+ homeostasis occurred in the absence of PIO or ROSI in the recording medium and are sensitive to TIO (Figs 2 and 3). Also, 2 h treatment with either drugs did not affect VGCC recordings. Together, these results are consistent with effects of TZDs occurring through classic nuclear receptor binding. Indeed, numerous reports have identified receptor-dependent beneficial aspects of TZD use, including reduced neuronal Ca2+ levels during Aβ-mediated insults through activation of Wnt signaling (Inestrosa et al. 2005), reduced inflammation in models of neurodegeneration through suppression of proinflammatory genes and proteins (Combs et al. 2000; Daynes and Jones 2002; Heneka et al. 2005), and suppression of inducible nitric oxide synthase (Heneka et al. 2000). Our results emphasize a TZD-mediated reduction in Ca2+ overload which could act complementarily with mechanisms of anti-inflammation or anti-oxidation. In fact, a synergistic pathway between the NMDAR antagonist MK-801 and ROSI has been identified, where drug combination therapy improves neurological outcome when compared to either compound administered alone (Allahtavakoli et al. 2007). Nevertheless, in the brain there has been several reports of non-PPAR-γ-dependent effects of TZDs, including rapid effects on mitochondrial function (Dello Russo et al. 2003; Colca et al. 2004; Feinstein et al. 2005; Hunter et al. 2008) and oxidative stress (Aoun et al. 2003), and further studies are necessary to identify whether these pathways can co-exist.

Our results demonstrate the presence of a ‘double dissociation’ between the effects of PIO and ROSI on VGCCs and NMDARs, albeit through a single receptor. Interestingly, while PIO and ROSI increased PPAR-γ protein levels at 24 h, results from PPRE binding assay did not follow that profile (Fig. 1). Structural differences between PIO and ROSI are likely to mediate the double dissociation seen here, given that separate ligands for the same nuclear receptor can differently alter the conformation of that receptor and mediate unique biological responses (Nolte et al. 1998; Gani and Sylte 2008). In fact, the development of selective estrogen receptor modulators is the basis for major drug discovery efforts in the treatment of breast cancer and osteoporosis (Riggs and Hartmann 2003). Similarly, selective PPAR-γ modulators (Rangwala and Lazar 2002; Zhang et al. 2007) also have been reported to alter the recruitment of co-activators/repressors in different tissues and mediate distinct gene expression profiles (Nolte et al. 1998; Norris et al. 1999; Paige et al. 1999). Under these conditions, selective PPAR-γ modulators appear able to impart distinct physiological responses in breast cancer cells (Thoennes et al. 2000; Allred and Kilgore 2005). Thus we believe the effects of ROSI and PIO on two different ion channels, while dependent on PPAR-γ, as evidenced by their sensitivity to TIO, are likely mediated by subtle structural changes in the agonists, changes in co-activators/repressors recruitment, and/or changes in the duration of the signal (Feige et al. 2005; Gani and Sylte 2008).

The present study demonstrates novel and potentially beneficial actions of two TZDs used clinically for the treatment of T2DM. While their conventional mechanism of action is associated with reestablishing insulin sensitivity in peripheral tissues, we show here that NMDAR and VGCC pathways are novel brain targets of TZDs whose inhibition may account, at least in part, for the therapeutic neurological benefits associated with use of these compounds. Both NMDARs and VGCCs are key participants in learning and memory processes and have also been shown to be altered with aging and/or AD. Therefore, TZDs may have potential applications in conditions associated with impaired learning and memory.


We thank Drs. Hadley and Brewer for their critical reading and valuable input on the manuscript. We thank Dr. Wang for her technical expertise on the DNA binding assay, and Dr. S. Kraner for the modifications she recommended on the assay protocol. This study was supported by grants AG029268, NCRR-P20-RR15592, and a gift from the Neurosciences Education and Research Foundation.