Nicotine induces dendritic spine remodeling in cultured hippocampal neurons

Authors


Abstract

Cholinergic neurons in the CNS are involved in synaptic plasticity and cognition. Both muscarinic and nicotinic acetylcholine receptors (nAChRs) influence plasticity and cognitive function. The mechanism underlying nAChR-induced plasticity, however, has remained elusive. Here, we demonstrate morphological changes in dendritic spines following activation of α4β2* nAChRs, which are expressed on glutamatergic pre-synaptic termini of cultured hippocampal neurons. Exposure of the neurons to nicotine resulted in a lateral enlargement of spine heads. This was abolished by dihydro-β-erythroidine, an antagonist of α4β2* nAChRs, but not by α-bungarotoxin, an antagonist of α7 nAChRs. Tetanus toxin or a mixture of 2-amino-5-phosphonovaleric acid and 6-cyano-7-nitroquinoxaline-2,3-dione, antagonists of NMDA- and AMPA-type glutamate receptors, blocked the nicotine-induced spine remodeling. In addition, nicotine exerted full spine-enlarging response in the post-synaptic neuron whose β2 nAChR expression was knocked down. Finally, pre-treatment with nicotine enhanced the Ca2+-response of the neurons to glutamate. These data suggest that nicotine influences the activity of glutamatergic neurotransmission through the activation of pre-synaptic α4β2 nAChRs, resulting in the modulation of spinal architecture and responsiveness. The present findings may represent one of the cellular mechanisms underlying cholinergic tuning of brain function.

image

Activation of nicotinic acetylcholine receptors (nAChRs) in brain influences plasticity and cognition. Here, activation of α4β2* nAChRs, which are expressed on glutamatergic presynaptic termini, results in the enlargement of dendritic spines through the modulation of the glutamatergic neurotransmission. The remodeled spinal architecture might be responsible for the change in responsiveness of neural circuitry, leading to cholinergic tuning of brain function.

Abbreviations used
α-BgT

α-bungarotoxin

APV

2-amino-5-phosphonovaleric acid

BES

N,N-Bis(2-hydroxyethyl)-2-aminoethanesulfonic acid

CNQX

6-cyano-7-nitroquinoxaline-2,3-dione

DG

dentate gyrus

DhβE

dihydro-β-erythroidine

DIV

days in vitro

DMSO

dimethyl sulfoxide

GI

glycine-induced

LTP

long-term potentiation

nAChRs

nicotinic acetylcholine receptors

PBS

phosphate buffered saline

PFR

picrotoxin/forskolin/rolipram

SIM

structured illumination microscopy

TetTx

tetanus toxin

Cholinergic neurons in the CNS are involved in plasticity and cognition. For example, denervation of cholinergic fibers in the brain results in severe impairment of learning and memory (Toledano and Alvarez 2004). Genetic disruption of the two classes of acetylcholine receptors, muscarinic and nicotinic, also impairs cognition, learning, and memory (Champtiaux and Changeux 2004; Wess 2004). In contrast, the enhancement of cholinergic transmission with acetylcholine esterase inhibitors ameliorates the cognitive deficits observed in Alzheimer's disease patients (Rogers et al. 1998). Nicotinic receptor agonists have also been shown to improve cognitive function in Alzheimer's disease patients (Levin and Rezvani 2000) and in animal models (Levin et al. 2006).

Nicotinic acetylcholine receptors (nAChRs) are ligand-gated cation channels and are expressed on both pre-synaptic termini and post-synaptic dendritic spines of hippocampal pyramidal neurons (Zarei et al. 1999; Xu et al. 2006). Activation of pre-synaptic nAChRs increases glutamate release from pre-synaptic termini (McGehee et al. 1995; Gray et al. 1996; Rousseau et al. 2005). Post-synaptic nAChRs modulate synaptic transmission by altering the membrane permeability of calcium ions (Frazier et al. 1998). The activation of nAChRs also modulates long-term potentiation (LTP) (Fujii et al. 2000; Matsuyama and Matsumoto 2003). The mechanism underlying nAChR-induced plasticity of hippocampal neurons, however, has remained elusive.

The rewiring of neural circuitry may underlie the plasticity of the brain. One of the key features of the rewiring of neural circuits may be the structural modification of synapses. Spines, the small protrusions on neuronal dendrites that constitute the excitatory post-synaptic components, undergo morphological alterations in response to synaptic activity. For example, LTP is correlated with the formation of new spines (Engert and Bonhoeffer 1999; Nagerl et al. 2004) and the enlargement of existing spines (Lang et al. 2004; Matsuzaki et al. 2004). In contrast, long-term depression is correlated with the shrinkage or pruning of spines (Nagerl et al. 2004; Zhou et al. 2004). Recently, a relationship between the activity of nAChRs and spine density has been reported (Ballesteros-Yanez et al. 2010; Lozada et al. 2012). However, more acute change in spine morphology has yet to be studied.

To understand how nAChRs are involved in neuronal plasticity, we investigated dendritic spine morphology following activation of nAChRs in cultured hippocampal neurons. The present findings may represent one of the cellular mechanisms underlying cholinergic tuning of brain function.

Materials and methods

Reagents

Nicotine, α-bungarotoxin (α-BgT), dihydro-β-erythroidine (DhβE), rolipram (Sigma-Aldrich, St. Louis, MO, USA), bicuculline (Enzo Life Science, Farmingdale, NY, USA), and forskolin (Calbiochem, San Diego, CA, USA) were dissolved in dimethyl sulfoxide (DMSO), while tetanus toxin (TetTx) (Calbiochem), 2-amino-5-phosphonovaleric acid (APV), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), strychnine, and picrotoxin (Sigma-Aldrich) were dissolved in water as stock solutions. Antibodies were purchased from Synaptic Systems, Goettingen, Germany (vGLUT1) and Millipore Corporation, Billerica, MA, USA (α4, α7, β2 nAChRs, and PSD-95).

Neuronal culture and transfection

Neuronal culture and transfection were performed as previously described (Okamura et al. 2004). All animal experiments were conducted in accordance with the guidelines of the Animal Care and Use Committee of Osaka University and the Takeda Experimental Animal Care and Use Committee. Neurons dissected from E18 Sprague–Dawley rat hippocampi (Charles River, Yokohama, Japan) were plated at a density of 1.8 × 104 cells/cm2 in poly-l-lysine-coated glass-bottomed dishes (Matsunami, Osaka, Japan) and cultured in Neurobasal Medium supplemented with B-27 (Life Technologies, Carlsbad, CA, USA). On 6–7 days in vitro (DIV), venus cDNA (3 μg) diluted with 45 μL of CaCl2 0.25 M and 45 μL of 2 × N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid (BES)-buffered saline (NaCl 280 mM, Na2HPO4 1.5 mM, and BES 50 mM, pH 7.1) was applied to the dish followed by incubation for 45–60 min.

Immunostaining, immunoblotting, confocal microscopy, and structured illumination microscopy

Immunostaining was performed as previously described (Tanaka et al. 2000). Briefly, 21 DIV neurons were fixed with methanol, and reacted sequentially with primary and secondary antibodies in blocking solution. The antibodies used were guinea pig anti-vGLUT1 (1 : 200), rabbit anti-α4 nAChR (1 : 160), rabbit anti-β2 nAChR (1 : 100), and mouse anti-PSD-95 (1 : 200). For immunoblotting, rabbit antibodies against α4, α7, and β2 nAChR were diluted 1 : 2000. Confocal images were scanned with LSM510META (Carl Zeiss, Oberkochen, Germany) as previously described (Yasuda et al. 2007). Structured illumination microscopy (SIM) images were taken using ELYRA S.1 equipped with αPlan-Apochromat 100×/NA 1.46 (Carl Zeiss). Parameters were set as follows: Laser power, 2–10%; EMCCD-Gain, < 100; Frame size, 1904 × 1900 pixels; Rotations (Grating), 3; Averaging, 1; SIM processing mode, automatic; and Channel alignment, affine.

Time-lapse imaging

Neurons transfected with venus cDNA were cultured until 19–22 DIV. After the addition of HEPES-NaOH 10 mM (pH 7.4), the culture dish was mounted in a chamber at 37°C for live imaging. Fluorescent images were obtained with an Eclipse TE100 microscope (Nikon, Tokyo, Japan) equipped with an objective (PlanApo 60x, NA 1.40), a filter set (MBE34931), and a charge-coupled device camera (DS-Qi1Mc). NIS-Elements software (Nikon) was used to control mechanical shutters. Shutter time and sensitivity of the camera were adjusted so that the fluorescent intensity of the spines did not saturate. Optical section images of 17 focal planes 0.2 μm apart were acquired by the use of a z-motor. The acquired images were subjected to deconvolution using AutoQuant Adaptive (3D) Blind algorithm (Media Cybernetics, Rockville, MD, USA), in which the underlying point spread function was iteratively (10 times) reconstructed from the collected 3D data set (http://www.mediacy.com/index.aspx?page=Auto_Algorithm) and which is built in NIS-Elements AR.

The neurons that fit the morphological criteria for pyramidal cells (Benson et al. 1994) were subjected to analyses. Bipolar neurons with spindle-shaped cell bodies (putative interneurons) or unipolar neurons with very high spine density were excluded from analyses. The accuracy in ruling out interneurons was confirmed by post-experimental staining with anti-GAD65 antibody (Developmental Studies Hybridoma Bank, Iowa City, IA, USA). Secondary dendrites originating from the primary apical dendrite were subjected to imaging. The spines on 10–20 μm-long dendritic segments between 20 and 50 μm from the proximal origin were analyzed. About three to six protrusions including spines were found on the 10 μm-long segments. Data were collected from all the spines that were fully visualized by scanning through the whole z-series of focal planes.

DiOlistic labeling of neurons

Male Sprague–Dawley rats (3 weeks of age, Charles River) were injected intraperitoneally with nicotine hydrogen tartrate salt 0.5 mg/kg (Sigma-Aldrich), waiting 2 h before perfusion with 1.5% paraformaldehyde. Coronal 300 μm-thick brain slices were prepared with a vibratome (Leica, Wetzlar, Germany) and subjected to delivery with DiI-coated ϕ1.0 μm gold particles (2 mg DiI/50 mg gold) at 100 PSI helium pressure using a Helios Gene Gun (Bio-Rad Laboratories, Hercules, CA, USA) as previously described (Gan et al. 2000). After overnight incubation in phosphate buffered saline (PBS), the labeled neurons in the slices were scanned with LSM510META (Carl Zeiss, PlanApo 63×, NA 1.40) (512 × 512 pixels, scan speed 8, repeat ×4, zoom ×2, 16 slices at 0.2-μm intervals) and the z-stack images were measured by using ImageJ software (NIH, Bethesda, MD, USA, scaling: 0.14 μm/pixel).

Chemical LTP

Neurons were briefly washed with 2 mL of glycine-induced (GI) wash (HEPES-NaOH 5 mM, NaCl 125 mM, KCl 2.5 mM, CaCl2 2 mM, and glucose 33 mM, pH 7.4) and then stimulated with 1 mL of GI solution (glycine 0.2 mM, bicuculline 0.02 mM, and strychnine 0.003 mM added to GI wash) or picrotoxin/forskolin/rolipram (PFR) solution (picrotoxin 0.05 mM, forskolin 0.05 mM, and rolipram 0.0001 mM added to GI wash) for 15 min. The stimulated neurons were further cultured in the conditioned medium supplemented with 50 μM APV.

RNAi

We prepared 4 shRNAs (HuSH pRS plasmid vectors, OriGene Technologies, Rockville, MD, USA) and 3 siRNAs (Silencer Selected siRNAs, Life Technologies) targeting β2 nAChR mRNA, as well as their scrambled controls. The sequences are described in Supporting Information. To select the most efficient shRNA or siRNA sequence for knocking down β2 nAChR, we introduced an shRNA-expressing vector (1 μg) or siRNA (10 pmol) into cultured hippocampal neurons using nucleofection (Amaxa Nucleofector; Lonza, Basel, Switzerland) before plating. Venus cDNA (1 μg) was cotransfected to monitor the transfection efficiency. The neurons were lysed 7 days after transfection, and cDNAs were reverse-transcribed using TaqMan Gene Expression Cells-to-CT Kit (Life Technologies). Endogenous expressions of β2 nAChR and β-actin were quantified using TaqMan Gene Expression Assay (Life Technologies), and the ratio of β2 nAChR to β-actin was calculated (Fig. S1). We selected the β2 nAChR shRNA construct (ID: TR710304A) as the most efficient RNAi (65.2% knockdown). For time-lapse imaging, venus cDNA and an shRNA-expressing vector (ratio 3 : 7, total 3 μg) were cotransfected as described above.

Ca2+ imaging

Neurons (1.8 × 104 cells/cm2 in a poly-d-lysine-coated 96-well plate, 21–29 DIV) were treated with/without nicotine for 60 min, and then loaded with Ca2+-indicator (2.5 μg/mL Fluo4-AM, 0.005% Pluronic F127, and 25% quenching buffer [Calcium Kit II; Dojindo, Kumamoto, Japan] in Hanks’ balanced salt solution with/without nicotine) for 30 min. Fluorescence signals in response to 10 μM glutamate were collected under no spontaneous Ca2+ oscillation with 1 μM tetrodotoxin at 1-s intervals using a FLIPR Tetra High Throughput Cellular Screening System (Molecular Devices, Sunnyvale, CA, USA). Calcium level was expressed as a function of [Ca2+]/Kd = [(– Fmin)/(Fmax – F)], where Fmax and Fmin were determined by 5 min-plateau fluorescence values for 10 μM ionomycin and 20 mM EGTA, respectively.

Statistics

Results are presented as the mean and standard error. Statistical calculations were performed with SAS System Version 8.2 (SAS Institute, Cary, NC, USA). Equality of variance was assessed by F-test for two groups or Bartlett's test for three or more groups. Statistical significance between two groups was calculated by Student's t (equal variance) or Welch's (unequal variance) test. Statistical significance among three or more groups was calculated by Dunnett's (equal variance) or Steel's (unequal variance) test. Statistical significance in the distribution of spines was calculated by chi-square test. Bonferroni corrections were used when statistical calculations were repeated.

Results

α4β2* nAChR is localized at glutamatergic termini in cultured hippocampal neurons

Neurons cultured from embryonic hippocampi are mainly composed of glutamatergic pyramidal neurons with ~ 6% GABAergic interneurons (Benson et al. 1994) and lack cholinergic neurons, whose cell bodies typically reside outside of the hippocampus (Schafer et al. 1998). These cultures, however, display a strong expression of nAChRs (Fig. 1a). On immunostaining, the α4 and β2 subunits were colocalized with vGLUT1 as markers for the glutamatergic pr-esynaptic terminus (Fig. 1b). To discriminate whether the nAChRs are localized to pre- or post-synaptic compartments, we utilized SIM. In the super-resolution SIM images, large α4/β2-puncta (Fig. 1c, arrows) were completely colocalized with vGLUT1, whereas smaller puncta, which probably represent non-synaptic intracellular pools, were distributed evenly throughout the cell. On the other hand, almost all the large α4/β2-puncta showed close apposition with PSD-95-labeled post-synaptic puncta (Fig. 1c). These data suggest that α4β2* nAChRs are localized to glutamatergic pre-synaptic termini in cultured hippocampal neurons.

Figure 1.

Pre-synaptic expression of α4β2* nicotinic acetylcholine receptors (nAChRs) in hippocampal neurons. (a) Extracts of cultured hippocampal neurons were subjected to immunoblotting. Arrowheads indicate the corresponding nAChR subunits. (b) Immunostaining for α4/β2 nAChR (green) double-labeled with vGLUT1 (red). Arrows indicate the colocalization of synaptic puncta. (c) Z-stacked structured illumination microscopy (SIM) images. Green, α4/β2 nAChR; red, vGLUT1/PSD-95. Magnified images of synaptic puncta indicated by arrows are shown below. Scale bars: (b) 20 μm; (c) 2 μm (top); and 0.5 μm (bottom).

Nicotine induces lateral enlargement of spine heads

We utilized these neurons to investigate the morphological alterations in dendritic spines following the activation of nAChRs. Neurons were transfected with the cDNA for fluorescent venus protein (Nagai et al. 2002) and subjected to time-lapse imaging at 19–22 DIV (Fig. 2). At this stage, 49.3% of protrusions on dendritic surfaces displayed a cotyloid appearance as we have described previously (Okamura et al. 2004). The spine head was closely apposed with a glutamatergic terminus labeled by vGLUT1 (Fig. 2c), indicating that the spine receives an excitatory input.

Figure 2.

Measurements of spine dimensions. (a) Venus-transfected pyramidal neuron. Rectangle shows the area magnified in panel (b). (c) vGLUT1 puncta (red) were apposed with spine heads. (d) Z-series of optical sections of a dendritic segment. (e–h) Z-series of raw (e), deconvolved (f), and thresholded (g) images of a single spine. (h) The spine length (blue line) and the full lateral dimension (red line) of each spine were determined by going through all the focal planes. The blue line was drawn from the origin of the spine neck to the farthest point of the spine top. The red line was drawn approximately perpendicular to the blue line, so that it captures the maximal lateral dimension. Scale bars: (a) 20 μm; (b) 5 μm; (c) 2 μm; (d) 3 μm, (e–h) 1 μm.

To measure the size of spines, we took serial optical section images at 0.2-μm focus intervals (z-series) (Fig. 2d). The acquired images were deconvolved based on the 3-dimensional data set (adaptive point spread function) (Fig. 2e and f), and were thresholded at half maximal intensity to determine the boundaries of the cells (Fig. 2g). The maximum diameter of each spine was chosen among the whole z-series of the optical sections (Fig. 2h).

The quantification of spine morphology revealed that 0.5 μM nicotine did not affect spine length or density, but induced an enlargement of spine width (Fig. 3a–c). The change became significant 90-120 min after nicotine application (Fig. 3c). Significantly more spines were “enlarged” in nicotine-treated neurons as compared to control neurons, when spines were categorized into three groups (Fig. 3e). We set the criterion level (20%) based on two SD values of spontaneous fluctuation in spine width.

Figure 3.

Nicotine induces lateral spine enlargement. Venus-transfected neurons were subjected to time-lapse imaging. (a) Z-stacked images of a dendritic segment before and after nicotine treatment. (b) Length and width of each spine before and after nicotine treatment are plotted. (c) Percent changes in spine length and width compared to the original sizes (D0), and spine density. Rectangles, control [dimethyl sulfoxide (DMSO), n = 62, 3 neurons]; triangles, nicotine (n = 61, 3 neurons). (d) Percent changes in spine width when neurons were treated for 90 min with control (DMSO, n = 134, 4 neurons), nicotine (0.5 μM, n = 166, 5 neurons), dihydro-β-erythroidine (DhβE) (1 μM, n = 141, 4 neurons), or α-bungarotoxin (α-BgT) (0.1 μM, n = 154, 4 neurons). (e) The spines were categorized by the change in width: enlarged (red), no change (green), or shrunken (blue). (f) Percent changes in spine width when neurons were pre-treated with DhβE (1 μM, n = 113, 3 neurons) or α-BgT (0.1 μM, n = 107, 3 neurons) prior to nicotine treatment (nicotine, n = 73, 3 neurons). The pairs of neuronal cultures for comparison were chosen from identical preparations from a single pregnant rat. The number of preparations used for each experiment was 2 (a–c), 3 (d and e), and 1 (f). (g) Spine widths of CA1 and dentate gyrus (DG) neurons were measured by DiOlistic labeling (n = 220, four animals for each treatment group). **< 0.01 by paired Student's t-test. < 0.05 by Welch's test with Bonferroni corrections. < 0.05, ‡‡< 0.01 by Steel's test. §§§< 0.001 by chi-square test. ¶¶¶< 0.01 by Welch's test.

The activation of α4β2* or α7 nAChRs induces LTP (Fujii et al. 2000; Matsuyama and Matsumoto 2003). However, it has also been proposed that the inactivation of α7 nAChRs facilitates LTP (Fujii et al. 2000). This paradox has been ascribed to the desensitization of nAChRs (Quick and Lester 2002). Taking the desensitization-prone properties of nAChRs into account, we examined whether the blockade of nAChRs would affect spine morphology. Exposure to DhβE, an antagonist of α4β2* nAChRs, or α-BgT, an antagonist of α7 nAChRs, did not produce any change in spine morphology, suggesting that the nicotine-induced spine enlargement did not result from the desensitization of these receptors (Fig. 3d and e).

When neurons were pre-treated with DhβE, the subsequent application of nicotine did not induce a significant change in spine width. In contrast, nicotine induced a full response in the presence of α-BgT (Fig. 3f). These data suggest that the agonistic activity of nicotine on α4β2* nAChRs is responsible for the increases in spine width.

To examine the in vivo effects of nicotine, we performed DiOlistic labeling of hippocampal neurons in rats. The spine width of neurons in CA1 and dentate gyrus (DG) was slightly larger in nicotine-treated (2 h) rats than in control rats (Fig. 3g).

Glutamate mediates the nicotine-induced spine remodeling

The morphological alterations of spines in response to nicotine exposure are reminiscent of our previous observation, in which glutamate receptor activation results in an enlargement of spine head width (Okamura et al. 2004). Here, the response of spines to glutamate was confirmed by adopting chemical LTP protocols: GI and PFR-induced LTP (Musleh et al. 1997; Otmakhov et al. 2004). During GI or PFR induction, spine heads shrank and the subsequent recovery resulted in a lateral enlargement of spine heads (Fig. 4a and b).

Figure 4.

Glutamate mediates nicotine-induced spine remodeling. (a and b) Neurons were treated with glycine-induced (GI) (n = 79, 3 neurons) or picrotoxin/forskolin/rolipram (PFR) (n = 69, 3 neurons) for 15 min followed by recovery for an additional 75 min. (c and d) Neurons were treated with nicotine (90 min, n = 112, 3 neurons) with existence of tetanus toxin (TetTx) (10 nM, n = 103, 3 neurons) or 2-amino-5-phosphonovaleric acid (APV) (50 μM) + 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) (100 μM) (n = 109, 3 neurons). (e) β2 nicotinic acetylcholine receptor (nAChR) mRNA levels after knockdown (n = 6). (f and g) Neurons transfected with the shRNA-expressing vector together with the venus cDNA were treated with nicotine (90 min) (scrambled shRNA, n = 57, 3 neurons; β2 nAChR shRNA, n= 65, 3 neurons). (h) Glutamate-evoked Ca2+-response with (red, n = 23) or without (black, n = 24) nicotine pretreatment. Control, n = 6. (f) Area under the curve of (i) (0–120 s). The pairs of neuronal cultures for comparison were chosen from identical preparations from a single pregnant rat (a–i). **< 0.01 by paired Student's t-test. < 0.05, ††< 0.01 by Steel's test. ‡‡< 0.01 by two-way anova. §§< 0.01 by Student's t-test. Scale bar: 2 μm.

To gain insight into the involvement of glutamatergic neurotransmission in nicotine-induced spine enlargement, we tried to inhibit neurotransmitter release (Schiavo et al. 1992) or membrane insertion of new AMPA receptors (Lu et al. 2001) with TetTx. Pre-incubation with TetTx resulted in a significant blockade of the nicotine-induced spine remodeling. Furthermore, in the presence of glutamate receptor antagonists, APV and CNQX, nicotine failed to induce spine head enlargement (Fig. 4c and d). These data suggest that nicotine influences glutamatergic neurotransmission, which may lead to the subsequent spine remodeling.

Although the immunolocalization of α4β2* nAChRs seems to be more prominent in pre-synaptic termini than post-synaptic spines in the present experimental system (Fig. 1), the site of nicotine action remains an open question. To address this issue, we knocked down the β2 nAChR subunit in the post-synaptic neurons by transfecting the β2 nAChR shRNA vector together with the venus cDNA (Fig. 4e, Fig. S1). In these neurons, nicotine treatment resulted in full enlargement of spines to the same extent as normal neurons (Fig. 4f and g). The data suggest that the contribution of post-synaptic α4β2* nAChRs on spine enlargement is not dominant, although their involvement cannot be fully denied.

Finally, to investigate whether this spine remodeling is associated with synaptic physiology, we performed Ca2+-imaging of the neurons. Pre-treatment with nicotine resulted in a significant increase in Ca2+-response induced by glutamate (Fig. 4h and i). These data suggest that nicotine-induced spine remodeling is associated with a change in neuronal physiological responsiveness.

Discussion

nAChRs are cation channels that induce rapid cellular responses (Albuquerque et al. 2009). Exposure to nicotine, however, produces a persistent craving for nicotine after long periods of abstinence (Dani and De Biasi 2001). The mechanism how the signal triggered by the rapid gating of nAChR is converted to a persistent change in brain function has remained elusive. The nicotine-induced spine remodeling demonstrated in this study may represent one of the early phase mechanisms underlying such a conversion. Although the present study focused on the semi-acute effect of nicotine within 2 h, more dramatic alterations such as spine density (Engert and Bonhoeffer 1999; Nagerl et al. 2004) might be correlated to a more persistent type of plasticity. The relationship between the activity of nAChR and spine density has been recently shown; a region specific decrease in spine density is observed in β2 subunit deleted mice (Ballesteros-Yanez et al. 2010; Lozada et al. 2012). Notably, Lozada and colleagues have reported an acute (1 h) change in spine density in young mouse (P7) but not in older mouse (P15). The age of the neurons utilized in the present experiments (19–22 DIV) might correspond to in vivo neurons of about 3-week-old mouse, hence losing the vigorous ability to change in spine density (Fig. 3c).

The hippocampal culture adopted in this study is mainly composed of glutamatergic and GABAergic neurons (Benson et al. 1994). nAChRs are expressed both in glutamatergic and GABAergic neurons (McGehee et al. 1995; Gray et al. 1996; Alkondon et al. 1997; Rousseau et al. 2005). Our data suggest that nicotine modulates glutamatergic neurotransmission, which in turn induces spine remodeling (Fig. 4). Therefore, there are at least two possible pathways responsible for this mechanism. The simplest pathway involves the direct modulation of glutamatergic synapses on pyramidal neurons. The highly specific localization of α4β2* nAChRs on the glutamatergic pre-synaptic terminus (Fig. 1c) argues for this pathway.

Another pathway involves GABAergic interneurons. The activation of pre-synaptic α4β2* nAChRs in GABAergic neurons stimulates their firing activity (Alkondon et al. 1997). An enhanced release of GABA following the activation of α4β2* nAChRs should suppress the release of glutamate. In this assumption, the spine should be shrunken after nicotine administration, but this was not the case in our study. Furthermore, the blockade of GABA receptors with bicuculline or picrotoxin resulted in an enlargement of spines as evidenced in the GI and PFR experiments (Fig. 4a and b). Therefore, the enhanced release of GABA may not be involved in the nicotine-induced spine remodeling.

It is also possible that the desensitization of nAChRs results in the inhibition of GABA release (Alkondon et al. 2000), which in turn results in a disinhibition of glutamate secretion. However, desensitization would not appear particularly involved in spine remodeling (Fig. 3d and e). Taken together, the indirect pathway through GABAergic neurotransmission is less likely to be involved in spine remodeling in the present setting.

Among the CNS-type nAChRs, α4β2* and α7 nAChRs are predominantly expressed in the hippocampus, and high expression of α4β2* nAChRs is observed in hippocampal pyramidal neurons (Sudweeks and Yakel 2000). However, the pre- and post-synaptic distribution of α4β2* nAChRs has long been controversial (Albuquerque et al. 2009). Recently, heterologous α4β2* nAChRs have been shown to be localized to the pre-synaptic membrane as well as to somatodendritic compartments, whereas α7 nAChRs are restricted to dendrites (Xu et al. 2006; Cheng et al. 2009). In addition, α4β2* nAChRs are reported to induce the release of excitatory amino acids from synaptosomes (Rousseau et al. 2005). We also confirmed the pre-synaptic localization of α4β2* nAChRs by using SIM (Fig. 1). Furthermore, nicotine exerted full spine-enlarging response in the post-synaptic neuron whose β2 nAChR expression was knocked down (Fig. 4e–g). Collectively, it seems possible that the activation of α4β2* nAChRs on pre-synaptic terminal membranes stimulates the release of glutamate, which in turn induces an enlargement of post-synaptic spines.

Acknowledgments

We thank Tatsuya Tanaka and Yosuke Yamazaki for technical assistance, Masanori P. Takahashi and Fumitaka Kimura for discussions, Atsushi Miyawaki for venus cDNA, and Osaka University Center for Medical Research and Education for experimental facilities. This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (K.Y., H.T.), and the Takeda Science Foundation (H.T.). A.O. and H.U. are employees of Takeda Pharmaceutical Company Limited. The authors declare no competing financial interests.

Ancillary