Synaptic activity initiates many adaptive responses in neurons. Here we report a novel form of structural plasticity in dissociated hippocampal cultures and slice preparations. Using a recently developed algorithm for three-dimensional image reconstruction and quantitative measurements of cell organelles, we found that many nuclei from hippocampal neurons are highly infolded and form unequally sized nuclear compartments. Nuclear infoldings are dynamic structures, which can radically transform the geometry of the nucleus in response to neuronal activity. Action potential bursting causing synaptic NMDA receptor activation dramatically increases the number of infolded nuclei via a process that requires the ERK-MAP kinase pathway and new protein synthesis. In contrast, death-signaling pathways triggered by extrasynaptic NMDA receptors cause a rapid loss of nuclear infoldings. Compared with near-spherical nuclei, infolded nuclei have a larger surface and increased nuclear pore complex immunoreactivity. Nuclear calcium signals evoked by cytosolic calcium transients are larger in small nuclear compartments than in the large compartments of the same nucleus; moreover, small compartments are more efficient in temporally resolving calcium signals induced by trains of action potentials in the theta frequency range (5 Hz). Synaptic activity-induced phosphorylation of histone H3 on serine 10 was more robust in neurons with infolded nuclei compared with neurons with near-spherical nuclei, suggesting a functional link between nuclear geometry and transcriptional regulation. The translation of synaptic activity-induced signaling events into changes in nuclear geometry facilitates the relay of calcium signals to the nucleus, may lead to the formation of nuclear signaling microdomains, and could enhance signal-regulated transcription.
Cells undergo functional and morphological alterations following stimulation from their environment. Among the best-characterized signal-regulated intracellular biochemical consequences are changes in gene expression. In neurons, transcriptional responses, triggered by electrical activity and mediated by NMDA receptor-induced calcium signaling pathways, are essential for most long-lasting adaptive responses and, in particular, play a key role in information storage and memory formation (Milner et al., 1998; Bading, 2000; West et al., 2002). However, there is an increasing appreciation of the importance of cell shape changes in modulating function. For example, the dendritic trees of neurons can undergo activity-induced geometry changes that will alter the neurons' ability to receive, process, and transmit information within the network (Muller et al., 2002; Van Aelst and Cline, 2004; Hayashi and Majewska, 2005; Tada and Sheng, 2006). Moreover, a computational analysis suggests that changes in the overall size or shape of a cell allows gradient-controlled signaling pathways to be switched on or off by changing the distance between the source of the signal and its targets (Meyers et al., 2006). In this study, we investigated the possibility that not only the cell surface geometry but also the geometry of cell organelles can undergo signal-induced, functionally relevant alterations. We focused on the nucleus because we reasoned that nuclear geometry changes may have an impact on the amplitude and/or kinetics of nuclear calcium signals following cytosolic calcium transients and thus could modulate the way calcium relays information to the nucleus. We used a recently developed algorithm to reconstruct the three-dimensional (3D) structure of neuronal nuclei from confocal microscopy data and quantified their volumes and surface areas. We found that nuclei from hippocampal neurons are extremely plastic and can adopt either near-spherical or complex, highly infolded structures, which affects the propagation of calcium signals into the nucleus and the activation of transcription-regulating events. NMDA receptor signaling pathways and the ERK-MAP kinases regulate the geometry of the nucleus, with synaptic and extrasynaptic NMDA receptors having opposing effects.
Materials and Methods
Hippocampal cultures and stimulations.
Hippocampal neurons from Sprague Dawley rats were cultured as described (Bading and Greenberg, 1991), except that growth medium was supplemented with B27 (Invitrogen). Cultures were plated onto 12 mm glass coverslips at a density of ∼450 cells (neurons plus glial cells) per mm2. Stimulations were done after a culturing period of 9–10 d during which hippocampal neurons develop a rich network of processes, express functional NMDA-type and AMPA/kainate-type glutamate receptors, and form synaptic contacts (Bading et al., 1995; Arnold et al., 2005). NMDA, d(−)-APV, MK-801, anisomycin and cycloheximide were from Sigma; bicuculline was from MP Biomedicals; actinomycin D was from AppliChem; tetrodotoxin (TTX) was from Tocris Bioscience; PD98059 and 4-aminopyridine were from Calbiochem (Merck); UO126 was from Axxora; DiD was from Biotrend Chemikalien. All experiments in which signal-regulation of nuclear geometry was analyzed were done at 37°C. DNA transfection was done using Lipofectamine 2000 (Invitrogen).
Organotypic hippocampal slice cultures.
Brain slices were prepared from 6-d-old Sprague Dawley rats. The pups were decapitated and the brains were removed and placed for trimming the tissue into a block into ice-cold dissection medium [Hanks' basal salt solution (Invitrogen) supplemented with glucose (Sigma) at a final concentration of 6.5 mg/ml. Brains were then transferred to the Teflon stage of a tissue chopper (H. Saur Laborbedarf, Reutlingen, Germany) and 350-μm-thick slices were cut. Hippocampal slices were micro-dissected from each brain slice in ice-cold dissection medium and transferred onto PTFE-cell culture inserts (Millipore). The slices were kept in a humidified atmosphere (5% CO2/95% air) at 37°C for 2 weeks as described (Stoppini et al., 1991), except that the culturing medium had the following composition: 50 ml of Basal Medium Eagle, 25 ml of Earle's balanced salt solution with Ca2+, Mg2+, Phenol Red (Sigma), 25 ml of horse serum (Invitrogen), 6.5 mg/ml glucose, 1 mm l-glutamine and 0.5% penicillin/streptomycin (Sigma). Half of the medium was replaced every 3 d.
Cultured hippocampal neurons were fixed between day in vitro (DIV) 10 and 14 with PBS containing 2% (w/v) paraformaldehyde and 2% (w/v) sucrose for 10 min at room temperature. Cells were washed 2 times with PBS, followed by incubation with the primary antibodies in PBS containing 0.1% (v/v) Triton X-100 and 2% (w/v) bovine serum albumin (BSA). Antibodies to the following proteins were used: lamin B and lamin A/C (mouse monoclonal antibodies, generous gift from Harald Hermann-Lerdon, German Cancer Research Center, Heidelberg, Germany; the monoclonal antibody to lamin B binds to rat lamin B1 and rat lamin B2); nuclear pore complex proteins recognized by Mab414 (mouse monoclonal antibody; Abcam), the nuclear pore complex protein NUP93 (mouse monoclonal antibody; Abcam), histone H3 phosphorylated at serine 10 (rabbit polyclonal antibody; Millipore), MSK1 phosphorylated at threonine 581 (rabbit polyclonal antibody; Cell Signaling Technology). The following antibody dilutions were used: anti-lamin B and anti-lamin A/C cell culture supernatant: 1:10; Mab414: 1:5000; NUP93: 1:2000; anti-phospho-histone H3: 1:200; anti phospho-MSK1: 1:200. All incubations with primary antibodies were done for 1 h at room temperature. Cells were washed 3 times with PBS and incubated for 15 min at room temperature with PBS containing 0.1% (v/v) Triton X-100, 2% (w/v) BSA and either Alexa 488-labeled secondary antibodies (Invitrogen) or Cy3-labeled secondary antibodies (Jackson ImmunoResearch). For nuclear counterstaining Hoechst 33258 (Serva) was applied in a 1:5000 dilution for 5 min. Cells were washed with PBS and were mounted using a Mowiol-based mounting medium. Experiments in which the cultured neurons were fixed with methanol for 10 min at −20°C generated virtually identical results. Specimens were analyzed using a Leica SP2 confocal microscope (Leica). Different regions of the coverslips were imaged using identical microscope settings and the same number of planes within the projection. To quantify nuclear pore complex (NPC) immunoreactivity, stacks of confocal images spanning the entire nucleus were projected by summation onto a single plane using NIH ImageJ. Regions of interest were defined for near-spherical and infolded nuclei and the mean fluorescence was measured. Mean values of both groups of nuclei were averaged and the percentage differences were plotted as arbitrary values.
For brain section immunohistochemistry, Sprague Dawley rats were killed by transcardiac perfusion under deep anesthesia (pentobarbital). After perfusion with PBS, brains were fixed by perfusion with 4% (w/v) paraformaldehyde in PBS. Brains were removed and postfixed overnight in the same fixative before cryoprotection in 30% (w/v) sucrose in PBS. Coronal sections (40 μm) were cut using a cryostat. Free-floating sections were rinsed with PBS and permeabilized with PBS containing 0.2% (v/v) Triton X-100 and 5% (w/v) BSA for 2 h. Sections were rinsed with PBS and incubated overnight with the monoclonal antibody to lamin B or with Mab414 at room temperature in PBS containing 5% (w/v) BSA and 0.2% (v/v) Triton X-100. Sections were washed with PBS before incubation with Alexa 488-labeled secondary antibodies (dilution 1:1000) for 2 h at room temperature, followed by extensive washing with PBS. Specimens were analyzed using a Leica SP2 confocal microscope.
For immunohistochemistry using organotypic slice cultures, the membranes holding the slices were briefly dipped into 37°C warm PBS to wash away any remaining growth medium before they were fixed with 2% paraformaldehyde (w/v)/2% sucrose (w/v) in PBS for 4 h at room temperature. Slices were subsequently washed with PBS and permeabilized in 0.1% Triton X-100 in PBS-Tween (0.1%) (PBS-T) overnight. Any remaining free aldehyde groups were then quenched with 50 mm glycine for 1 h at room temperature. After washing with PBS-T, slices were blocked with 10% normal goat serum, 2% BSA in PBS-T for 4 h at room temperature before they were incubated with the primary antibodies in 2% normal goat serum, 2% BSA in PBS-T overnight at 4°C. The primary antibodies (mouse monoclonal antibody to lamin A/C and rabbit polyclonal antibody to histone H3 phosphorylated at serine 10) and secondary antibodies used and their dilutions were identical to the ones used for immunocytochemistry of dissociated hippocampal cultures (see above). For nuclear counterstaining, Hoechst 33258 (Serva) was applied in a 1:5000 dilution for 5 min. Slices were washed with PBS and mounted in a Mowiol-based mounting medium. See supplemental material, available at www.jneurosci.org, for details on transmission electron microscopy.
3D image reconstruction and mathematical modeling.
The mathematical framework for the 3D image reconstruction of cell nuclei has been described (Queisser et al., 2008). See supplemental material, available at www.jneurosci.org, for a brief description of the algorithms used and for details of the mathematical modeling of nuclear calcium dynamics in nuclei with different geometries.
Confocal microscopy and calcium imaging.
For calcium imaging cells were loaded with Fluo-4 AM (3.6 μm) (Invitrogen) and nuclear compartments were visualized by concurrent loading with ER-Tracker Blue-White DPX dye (1 μm) (Invitrogen). Fluorescence images were obtained using an inverted Leica SP2 confocal microscope with an HCX PL APO CS 40.0× 1.25 NA OIL UV objective. Cells were mounted in a perfusion chamber (LIS) and imaged at room temperature in a CO2-independent buffered salt-glucose-glycine (SGG) solution containing (in mm) NaCl 140.1, KCl 5.3, MgCl2 1.0, CaCl2 2.0, HEPES 10.0, glycine 1.0, glucose 30.0, and sodium pyruvate 0.5. The image acquisition rate was 1.5 Hz. To calibrate the fluorescence signal (F), Fluo-4 was saturated by adding 50 μm ionomycin (MP Biomedicals) to the perfusion solution to obtain the maximal fluorescence signal (Fmax) and then quenched with MnCl2 to obtain the minimal fluorescence signal (Fmin). [Ca2+] was expressed as a function of the Fluo-4 fluorescence Kd × [(F − Fmin)/(Fmax − F)] (Grynkiewicz et al., 1985). The validity of calcium measurements critically depends on the accuracy of the calibration, particularly when calcium concentrations are being analyzed and compared in different cellular compartments. To avoid artifacts, calibration was done pixel by pixel to relate the fluorescence values to the corresponding values obtained during the calibration. See supplemental material, available at www.jneurosci.org, for details on combined patch-clamp and calcium imaging using X-Rhod-1 and fura-2, and for details on calcium imaging using organotypic hippocampal slice cultures.
Morphological analysis and 3D image reconstruction of hippocampal cell nuclei
To visualize the structure of cell nuclei, we used a recently developed algorithm that allows 3D image reconstruction of cell organelles from optical sections generated using a laser-scanning confocal microscope (Queisser et al., 2008; see also supplemental material, available at www.jneurosci.org). For the microscopy analysis, cultured hippocampal neurons were fixed and subjected to immunocytochemistry using an antibody to lamin B that decorates the inner nuclear envelope (Fig. 1A) (Goldman et al., 2002). Stacks of confocal images were processed using a nonlinear anisotropic diffusion filter (Broser et al., 2004) and a cluster-segmentation process (Otsu-segmentation algorithm) (Otsu, 1979), from which a mathematical surface grid could be created, that represents the nuclear lamina. A total of 102 nuclei were reconstructed. We found that ∼30% of the nuclei of cultured hippocampal neurons are not near-spherical but have a complex infolded shape. Typical examples of 3D reconstructed hippocampal nuclei are shown in Figure 1B. A quantitative analysis of the nuclear geometry in hippocampal tissue of adult rats revealed that ∼4% of all nuclei in the CA1 pyramidal cell layer showed large invaginations of the nuclear envelope. Examples of electron micrographs of nuclei from the CA1 pyramidal cell layer of rat hippocampal slices and, for comparison, from cultured hippocampal neurons are shown in Figure 2. A quantitative analysis using brain tissue from postnatal day 5 (P5) rats revealed a similar percentage of infolded nuclei in the CA1 pyramidal layer of the hippocampus; a typical image of a hippocampal section from a P5 rat that was used for the quantitative analysis is shown in supplemental Figure 1, available at www.jneurosci.org as supplemental material. The difference between the incidence of infoldings in cultured neurons and in vivo may be due to the activity states of these neurons (see below).
During the course of the analysis of P5 brains, we noticed many irregularly shaped nuclei within the subcallosal zone (SCZ), a prominent cell layer between the hippocampus and the corpus callosum. In ∼40–60% of these cells, the nuclei were either highly infolded or even consisted of two or more lobes connected by thin bridges that resemble envelope-limited chromatin sheets (Olins et al., 1998) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). The SCZ as well as the related brain structure, the subventricular zone (SVZ) contain a mixed population of cells including glial cells, neuroblasts, and putative stem cells, some of which are known to have nuclei with deep invaginations (Doetsch et al., 1997; Seri et al., 2006).
To investigate the possibility that only the inner (and not the outer) nuclear envelope invades the cell nucleus to create an intranuclear reticulum [previously termed nucleoplasmic reticulum (Echevarría et al., 2003)], we analyzed hippocampal nuclei by electron microscopy. Electron micrographs obtained from both cultured hippocampal neurons and hippocampal slices of rat brains revealed that all clefts including those that dive deeply into the nucleus are lined by two closely juxtaposed membranes representing the inner and outer nuclear envelope (Fig. 2). These structures have some resemblance to the indentations of the nuclear membranes and grooves observed in several cell types in the primate and rodent hippocampus (Ribak and Seress, 1983; Frotscher et al., 1988; Soriano et al., 1990), onion epidermal cells and tobacco culture cells (Collings et al., 2000), and also to alterations in nuclear architecture found following stress, injury, or aging (Honavar and Lantos, 1987; Lafarga et al., 1992; Borsello et al., 2002; Dorsey et al., 2006). However, they appear morphologically distinct from the tube-like structures found in HeLa cells and various other non-neuronal cell lines (Fricker et al., 1997) and the highly lobulated nuclei found in fibroblasts from patients with Hutchinson-Gilford Progeria Syndrome (Goldman et al., 2004). Our results indicate that the inner nuclear envelope does not form intranuclear cisterns in hippocampal neurons and thus the surface of the 3D images marks the border between the cytosol and the nucleus.
Infolded nuclei have larger surfaces and increased nuclear pore complex immunoreactivity
We next investigated quantitative differences between infolded and near-spherical nuclei. First, we determined the volume and surface of the nuclei. We found that the volume of infolded and near-spherical nuclei was similar; however, infolded nuclei have a 23.8 ± 5.4% larger surface area (Fig. 3A). Electron microscopy revealed that nuclear infoldings contain nuclear pore complexes (NPCs) (Fig. 3Ba,b), raising the possibility that an increased nuclear surface is accompanied by an increased number of NPCs. To investigate this, immunostaining experiments were done using the NPC antibody Mab414 and an antibody to NUP93, which is part of the NPC (Lim and Fahrenkrog, 2006). Examples of projections from confocal stacks showing NPC immunoreactivity in cultured hippocampal neurons obtained with these monoclonal antibodies are shown in Figure 3B, c and d. The punctate nature of the NPC immunoreactivity was evident after processing of Mab414 immunostaining images with a deconvolution algorithm (data not shown). The quantified NPC immunostaining signals reflect the number of NPC expressed in a given nucleus (for details on the quantification of immunostaining experiments see Materials and Methods). By comparing NPC immunoreactivities, we obtained the relative abundance of NPCs in near-spherical versus infolded nuclei. For the two different antibodies, we observed significantly more NPC immunoreactivity in infolded nuclei compared with near-spherical nuclei (Fig. 3Be,f). This indicates that the increase in nuclear surface is accompanied by a similar increase in the number of NPCs, thus keeping the NPC density constant. We also analyzed NPCs in acute rat hippocampal slices. NPC immunostaining experiments with Mab414 identified infolded nuclei in the pyramidal cell layers of CA1 and CA3 (Fig. 3C).
Infoldings generate nuclear signaling microdomains
The differences in the amount of NPCs, nuclear surface area, and distances between cytosolic and nuclear sites observed in infolded versus near-spherical nuclei may be particularly relevant for diffusion in and out of the nucleus. We therefore used mathematical modeling to test the influence of the different aspects of the nuclear geometry on the dynamics of calcium transients in the cell nucleus following increases in the cytosolic calcium concentration. Nuclear calcium signaling was chosen because it controls gene expression mediated by the CREB/CBP transcription factor complex (Bading et al., 1993; Hardingham et al., 1997, 2001a; Chawla et al., 1998) that is important for adaptive processes in the brain including neuronal survival, plasticity, and learning (Milner et al., 1998; Bading, 2000; Lonze and Ginty, 2002; Limbäck-Stokin et al., 2004; Papadia et al., 2005). Nuclear calcium transients are triggered by calcium entry into neurons either through synaptic NMDA receptors or L-type voltage-gated calcium channels (Hardingham et al., 2001a, 2002). Through a still unresolved mechanism, synaptic activity-induced calcium signals travel to the cell soma where they readily invade the nucleus most likely through NPCs (Nakazawa and Murphy, 1999; Hardingham et al., 2001a; Power and Sah, 2002; Raymond and Redman, 2006; Watanabe et al., 2006; Johenning and Holthoff, 2007; Eder and Bading, 2007 Hagenston et al., 2008). In rat hippocampal neurons and amphibian sympathetic neurons, the nuclear compartment border does not sustain a calcium gradient across the nuclear envelope (O'Malley, 1994; Eder and Bading, 2007), although in other cell types the nucleus may be insulated from cytosolic calcium transients (al-Mohanna et al., 1994; Badminton et al., 1996). Mathematical modeling of nuclear calcium dynamics revealed that infoldings divide the nucleus into compartments that can function as nuclear signaling microdomains (Figs. 4 and 5).
To assess the effect of infolding-induced compartmentalization on calcium signaling in the nucleus, mathematical models were developed to run systematic simulations of calcium diffusion between the cytosol and the nucleus through NPCs (for details of the mathematical modeling see supplemental material, available at www.jneurosci.org). The model is based on a diffusion equation, a partial differential equation, which models the propagation of calcium signals into and within the nucleus. The nuclear geometry was generated from confocal microscopy data, processed with the 3D reconstruction tools outlined above and in the supplemental material, available at www.jneurosci.org. As boundary conditions for the diffusion equation, we developed a method to input, experimentally measured cytosolic calcium transients (CCTs) directly into the simulation software UG (Bastian et al., 2000), which allows mathematical modeling on highly unstructured geometries. In brief, mathematical modeling of calcium signaling in the nucleus was done by supplying real, experimentally measured cytosolic calcium signals as input to the diffusion equation and applying it to 3D reconstructions of real nuclear geometries. Nuclear calcium signaling was modeled in infolded nuclei with either differently sized or nearly equally sized compartments (Fig. 4A,C). Parallel to the modeling experiments, we performed confocal calcium imaging experiments with cultured hippocampal neurons containing nuclei with either differently sized or nearly equally sized compartments (Fig. 4B,D).
Simulations showed that compartments, formed by the infolding of the nuclear envelope, exhibit different calcium signals depending on their size. Small nuclear compartments show larger amplitude calcium signals compared with those of large compartments of the same nucleus (Fig. 4A), most likely due to faster rise and decay times of the signal in smaller compartments. Differences in the rise and decay times of calcium signals have previously been reported in small non-nuclear subcellular compartments such as dendrites, spines, and presynaptic terminals (Augustine and Neher, 1992; Regehr and Tank, 1994; Yuste et al., 2000; Helmchen, 2002). In contrast, calcium signals are equivalent in nearly equally sized nuclear compartments of the same nucleus (Fig. 4C). Similar results were observed in calcium imaging experiments. Here too, small compartments of infolded nuclei showed larger amplitude calcium signals than large compartments of the same nucleus (Fig. 4B), whereas in nuclei with nearly equally sized nuclear compartments the calcium signals were similar in the two compartments (Fig. 4D). Raw fluorescence of Fluo-4 in large and small compartments was equal suggesting that compartmental differences in dye concentration and thus calcium buffering did not exist. We quantified the observed effects by allocating cells with infolded nuclei into two groups, those with similar and those with dissimilar nuclear compartment sizes (large compartment size to small compartment size ratios of <2:1 and >2:1, respectively). Relative to the larger compartments, the smaller compartments of neurons with compartment size ratios of >2:1 showed significantly higher calcium levels during bursts of action potential (AP) firing than neurons with nuclear compartment size ratios of <2:1 (Fig. 4E).
These results indicate that the geometry of the nucleus influences the relay of calcium signals to the nucleus. Infoldings of the nuclear envelope separate the nucleus into microdomains that can be seen as functional units of the nucleus, which generate distinct calcium signals. Due to reduced diffusion distances in small microdomains, small compartments generate larger calcium signals, which are more likely to be above threshold for calcium-regulated events than calcium signals in large microdomains. Our modeling data indicated faster rise and decay times of calcium signals in small compartments, which predicts that calcium signals in small compartments more accurately follow an oscillating calcium signal originating in the soma, especially at higher frequencies. Thus, the transmission of somatic calcium signals to small nuclear compartments may help to preserve information contained in the frequency of the signal.
Improved temporal resolution of multiple calcium signals in smaller nuclear compartments
To examine the temporal resolution of an oscillating calcium signal, we compared modeling predictions generated from an artificial somatic calcium signal with experimentally observed data generated from a similar somatic signal. Patch-clamp recordings combined with calcium imaging using X-Rhod-1 (Fig. 5) or the ratiometric calcium indicator, fura-2 (supplemental Fig. 3, available at www.jneurosci.org as supplemental material) were made from neurons with infolded nuclei. Current injection was used to depolarize the membrane to generate a series of action potentials (i.e., a burst) at specified frequencies and burst intervals (for details, see supplemental material, available at www.jneurosci.org). These bursts generated somatic calcium signals, which presumably originated predominantly from the activation of voltage-operated calcium channels in the plasma membrane and traversed the nuclear envelope to enter each nuclear compartment. Recurrent bursting at 5 and 10 Hz was seen to generate a sharp rise in nuclear calcium levels over the 10–15 s of stimulation (Fig. 5A; supplemental Fig. 3A, available at www.jneurosci.org as supplemental material). This nuclear calcium signal also showed a higher frequency component corresponding to the stimulation frequency, which was more visible after high-pass filtering (>1 Hz, Fig. 5C; supplemental Fig. 3C, available at www.jneurosci.org as supplemental material). Unfiltered signals were analyzed with a fast Fourier transform (FFT) to resolve the power of the signal across the frequency spectrum. Recordings from neurons with similarly sized nuclear compartments showed no difference in the power of calcium signals generated at 5 Hz (n = 3 with fura-2) whereas neurons with unequally sized compartments showed either no difference in power (n = 5 with X-Rhod-1; n = 3 with fura-2) or a larger power in the smaller compartment (n = 4 with X-Rhod-1; n = 5 with fura-2) for signals generated at 5 Hz (Fig. 5G,I; supplemental Fig. 3E, available at www.jneurosci.org as supplemental material). Oscillating calcium signals were also larger in smaller compartments than in larger compartments of some nuclei when the stimulation frequency was increased to 10 Hz (n = 3, no difference; n = 2 larger power in smaller compartment; all with X-Rhod-1; Fig. 5I). While these experiments were done at room temperature, similar results were obtained at 32−34°C (data not shown). Similar results were also obtained when identical analysis procedures were applied to simulation data generated from an artificial somatic calcium signal of a similar sampling frequency and kinetic (Fig. 5D,F,H). Calcium oscillations were larger in the small than in the large nuclear compartment with a relative difference in the power of this signal comparable to that observed experimentally in live cells (compare Fig. 5G,H). Thus, both experiment and model show that frequency information in calcium signals is preserved better by smaller nuclear compartments and that small compartments of infolded nuclei better resolve a calcium signal originating in the soma, which oscillates in the 5–10 Hz range.
Activation of synaptic NMDA receptors promote the formation of nuclear infoldings
We next investigated the possibility that nuclear infoldings are dynamic structures regulated by intracellular signaling pathways. Nuclear geometry was assessed by fluorescence microscopy in cultured hippocampal neurons following immunocytochemical labeling of the nuclear lamina using the antibody to lamin B. We found that indeed electrical activity can dramatically change the nuclear geometry. We focused in particular on electrical stimulation paradigms that activate calcium entry through NMDA receptors that play a central role in both the physiology and pathology of neurons (Hardingham and Bading, 2003). We found that stimulation of synaptic NMDA receptors increased the percentage of infolded nuclei. Synaptic NMDA receptors were stimulated by exposing hippocampal neurons to the GABAA receptor antagonist bicuculline. Because the hippocampal network contains ∼11% GABA-ergic interneurons that impose a tonic inhibition onto the network, removal of GABAA receptor activity leads to bursts of AP firing (Hardingham et al., 2001b; Arnold et al., 2005). AP bursting causes calcium entry through synaptic NMDA receptors, generates robust calcium transients in the cytosol and nucleus, activates gene transcription, and promotes neuronal plasticity and survival (Hardingham et al., 2002; Hardingham and Bading, 2003; Arnold et al., 2005; Lee et al., 2005; Papadia et al., 2005). Analysis of the nuclear geometry showed that AP bursting rapidly increased the number of infolded nuclei (Fig. 6A–C). We also used time-lapse confocal imaging to visualize the formation of nuclear infoldings. For those experiments we labeled the membranes of cultured hippocampal neurons with the fluorescent stain, DiD, recorded over time confocal stacks covering the entire nucleus and projected the images for each time point into one plane. Supplemental Figure 4, available at www.jneurosci.org as supplemental material, illustrates an example of a hippocampal neuron that on AP bursting formed nuclear infoldings. The induction of nuclear infoldings following AP bursting was dependent on NMDA receptor activation and was inhibited by the NMDA receptor antagonist, MK-801 (Fig. 6C).
To investigate whether the presence of nuclear infoldings in already ∼30% of hippocampal neurons was due to ongoing electrical activity during the first week of cell culture, we blocked all glutamatergic synaptic transmission in the hippocampal network during this culturing period by including kynurenate, a nonselective glutamate receptor antagonist, and 11 mm MgCl2 in the culture medium. Under these conditions far fewer neurons have nuclear infoldings although infoldings can be rapidly induced in these cells by subsequent induction of AP bursting (Fig. 6B). This indicates that indeed neuronal activity (both basal and evoked activity) is a prime regulator of nuclear geometry.
Rapid loss of nuclear infoldings following stimulation of extrasynaptic NMDA receptors
In contrast to the effects of AP bursting and synaptic NMDA receptors on nuclear geometry, the stimulation of extrasynaptic NMDA receptors using bath application of NMDA led to a rapid loss of nuclear infoldings. Within 15–30 min of NMDA bath application virtually all infolded nuclei were transformed into a near-spherical shape (Fig. 6D). To monitor the behavior of nuclear infoldings in real time we transfected hippocampal neurons with an expression vector for a fusion protein containing human lamin B1 and GFP (lamin B-GFP) (Daigle et al., 2001). High levels of lamin B-GFP expression caused neuronal nuclei to acquire a highly lobulated shape (data not shown), which resembles to some extent the morphology of nuclei found in aged non-neuronal cell types in Caenorhabditis elegans (Haithcock et al., 2005) and in Hutchinson-Gilford Progeria Syndrome, an accelerated aging disease in humans caused by mutations in lamin A (Eriksson et al., 2003; Mounkes et al., 2003; Goldman et al., 2004; Scaffidi and Misteli, 2006). We therefore selected for the analysis neurons that expressed low levels of lamin B-GFP and did not show a lobulated nuclear shape. Exposure of those cells to 50 μm NMDA led to a rapid loss of nuclear infoldings (Fig. 6E).
The striking differences between the effects of AP bursting and bath application of NMDA suggest that synaptic and extrasynaptic NMDA receptors antagonistically control nuclear geometry. However, it is also possible that differences in the amplitude and/or kinetics of the calcium transients evoked by the two stimulation paradigms rather than the location of the NMDA receptor activated explain the result. AP bursting generates intermittent calcium signals, whereas bath application of NMDA triggers an intense sustained increase in calcium levels (see, for example, Hardingham et al., 2001a, 2002; Bengtson et al., 2008). We therefore sought to establish a stimulation paradigm that activates synaptic NMDA receptors and generates sustained calcium transients. This was achieved by the use of 4-aminopyridine (4-AP), a potassium channel blocker that increases the frequency of bicuculline-induced AP bursting in hippocampal networks (Hardingham et al., 2001a, 2002) causing sustained increases in calcium levels very similar in amplitude, duration, and net calcium influx (area under the curve) to those obtained with bath application of NMDA (Fig. 6F,G). However, the effects of bicuculline/4-AP and bath application of NMDA on nuclear geometry were dramatically different: bicuculline/4-AP increased whereas bath application of NMDA decreased the percentage of hippocampal neurons with infolded nuclei (Fig. 6H,I). These results indicate that, indeed, the location of the NMDA receptor activated (synaptic vs extrasynaptic) determines whether nuclear infoldings are being newly formed or lost.
Analysis of glial cells present in hippocampal cultures revealed that ∼12% of GFAP-positive cells have nuclear infoldings. However, the infoldings are generally less prominent than those in hippocampal neurons (supplemental Fig. 5, available at www.jneurosci.org as supplemental material) and the percentage of infolded glial cell nuclei did not change after stimulation of hippocampal cultures with bicuculline (data not shown).
Activity-induced formation of nuclear infoldings requires ERK-MAP kinase signaling and protein synthesis
One of the intracellular signaling pathways activated by synaptic activity and a localized calcium signal in the immediate vicinity of synaptic NMDA receptors is the ERK-MAP kinase cascade (Bading and Greenberg, 1991; Grewal et al., 1999; Hardingham et al., 2001b; Sweatt, 2004; Wiegert et al., 2007). To investigate whether this signaling pathway is important for NMDA receptor-dependent regulation of nuclear geometry, we pharmacologically blocked this pathway using MAP kinase/ERK kinase (MEK) inhibitors PD98059 or UO126. We found that indeed, PD98059 and UO126 blocked the signal-induced formation of infoldings (Fig. 7A). Thus, neuronal activity stimulates the formation of nuclear infoldings through a mechanism that is triggered by calcium entry through synaptic NMDA receptors and requires the ERK-MAP kinase signaling pathway. To investigate the persistence of newly formed infoldings, we first induced AP bursting for 1 h, which was followed by treatment with tetrodotoxin (TTX) to stop all electrical activity (Fig. 7B). Under those conditions, most infoldings disappeared within 1–2 h of blocking activity (Fig. 7C, dashed line). However, if the neurons were induced to fire bursts of APs for prolonged periods of time (i.e., 40 h) causing repeated activation of synaptic NMDA receptors, nuclear infoldings persisted for several hours even after neuronal activity was blocked with TTX (Fig. 7C, solid line). This indicates that activity-induced changes in nuclear shape are initially unstable but become more stable and independent of ongoing electrical activity during repeated stimulations. Since activity-induced, long-lasting structural and functional changes of neurons are often dependent on the synthesis of new proteins, we investigated the requirement of gene transcription and mRNA translation for activity-induced formation of nuclear infoldings. We found that two inhibitors of protein synthesis, anisomycin and cycloheximide, completely blocked the AP bursting-induced formation of new infoldings (Fig. 7D,E). In contrast, actinomycin D, an inhibitor of gene transcription did not prevent the increase in the percentage of infolded nuclei in cultured hippocampal neurons following AP bursting (Fig. 7D), and neither did anisomycin at a low concentration which did not inhibit mRNA translation but activated a stress response and caused phosphorylation of the p38 MAP kinase (data not shown). These results indicate that the translation of preexisting mRNAs is required for activity-induced formation of nuclear infoldings.
Synaptic activity-induced phosphorylation of histone H3 at serine 10 correlates with the degree of nuclear infolding
We next investigated whether the presence of nuclear infoldings correlates with transcription-relevant events. The phosphorylation of histone H3 at serine residue 10 is important for synaptic activity-induced chromatin remodeling and serves as an indicator of gene transcription (Brami-Cherrier et al., 2005, 2007). Histone H3 serine 10 phosphorylation is induced in hippocampal neurons in an activity-dependent manner (Chwang et al., 2007; Chandramohan et al., 2008) (see also Fig. 8A,B); this phosphorylation event is mediated by the mitogen-activated stress-kinases 1 and 2 (MSK1/2) (Soloaga et al., 2003), which are nuclear targets of the ERK-MAP kinase and the p38-MAP kinase cascades (Deak et al., 1998). To unearth a possible correlation between histone H3 phosphorylation at serine 10 and the degree of infolding, we performed immunocytochemical experiments using antibodies that bind to the serine 10 phosphorylated form of histone H3. In the same neurons, we assessed the nuclear geometry using antibodies to lamin A/C. Infolded nuclei were categorized as weakly infolded (only one invagination that spans at least one third of the size of the nucleus), moderately infolded (two invaginations) and highly infolded (three or more invaginations). To categorize the degree of nuclear infoldings anaglyph images were generated from maximum intensity z-stacks using ImageJ. Using 3D glasses, the 3D structure of each nucleus in an image stack was analyzed. This procedure enabled us to categorize the degree of nuclear infolding very accurately without the need for 3D reconstructions of individual nuclei. In control cells, 66% of the nuclei were near-spherical, 26% were weakly infolded, and 8% were moderately infolded. After 1 h of AP bursting, the fraction of near-spherical nuclei decreased to 25%, whereas the fractions of weakly infolded, moderately infolded, and highly infolded nuclei increased to 34%, 25%, and 16%, respectively (Fig. 8C).
Using the antibody to the serine 10-phosphorylated form of histone H3 we observed a robust increase in phospho-H3 immunoreactivity 1 h after induction of AP bursting (Fig. 8A,B). Since the phospho-H3 staining varied considerably within the neuron population, we tested the possibility that the level of signal-induced histone phosphorylation is linked to nuclear geometry. Indeed, we found that after AP bursting, phospho-H3 immunoreactivity correlated with the complexity of nuclear infoldings, i.e., the strongest phospho-H3 signals were obtained in highly infolded nuclei whereas the weakest phospho-H3 signals were obtained in weakly infolded nuclei (Fig. 8D). Even in unstimulated controls, phospho-H3 immunoreactivity was significantly stronger in moderately infolded nuclei compared with near-spherical or lightly infolded nuclei (Fig. 8D). Intranuclear differences between small and large compartments were not detected. Given that the phosphorylation of histone H3 on serine 10 is catalyzed by MSK1/2 (Soloaga et al., 2003), we also investigated a possible correlation between MSK1/2 activation and the degree of infolding. Using an antibody to the activated (i.e., threonine 581 phosphorylated) form of MSK1, we found that AP bursting caused a robust increase in nuclear phospho-MSK1 immunoreactivity (Fig. 8B). Similar to the results obtained for histone H3 phosphorylation, we found that infolded nuclei showed significantly higher nuclear phospho-MSK1 immunoreactivity compared with near-spherical nuclei (Fig. 8E). These results establish a link between nuclear infoldings, the activation of the ERK-MAP kinase target MSK1, and the phosphorylation of histone H3 and suggest that nuclear geometry may be relevant for signal-induced chromatin remodeling and gene expression.
Burst activity induces nuclear infoldings and histone H3 phosphorylation in hippocampal slices
We finally investigated whether the synaptic activity-induced structural changes observed in dissociated hippocampal cultures also take place in an experimental system with at least partially preserved neuronal circuitry. We used organotypic hippocampal slice cultures that were exposed to bicuculline to induce AP bursting. Calcium imaging experiments using Oregon Green 488 BAPTA-1-AM (OGB-1) revealed that bicuculline treatment of the slices gave rise to periodically occurring intracellular calcium transients (Fig. 9A). Within 1 h after induction of burst activity with bicuculline, many nuclei of hippocampal neurons in the CA1 pyramidal cell layer had undergone dramatic changes in their geometry (Fig. 9B,C). The percentage of infolded nuclei increased, in a TTX-dependent manner, from 25% ± 4.5% (category 1), 4% ± 1.8% (category 2), and 1% ± 0.5% (category 3) in the control conditions to 40% ± 2.2% (category 1), 33% ± 2.1% (category 2), and 12% ± 1.9% (category 3) after bicuculline treatment (Fig. 9D). The increase in nuclear infoldings during burst activity was inhibited by MK-801 and UO126; in particular the fractions of nuclei that were in categories 2 and 3 after bicuculline treatment were virtually eliminated by the blockade of NMDA receptors and the ERK-MAP kinase signaling pathway (Fig. 9D). The activity-induced changes in nuclear geometry were paralleled by a robust induction of histone H3 phosphorylation on serine 10 with the strongest phospho-H3 immunoreactivity following bicuculline treatment being observed in infolded nuclei of the categories 2 and 3 (Fig. 9E). Thus, similar to the result obtained using dissociated hippocampal cultures, the nuclei of synaptically activated CA1 pyramidal neurons in a hippocampal slice preparation can undergo dramatic structural changes. The correlation between the formation of infoldings and the phosphorylation of histone H3 suggest a functional link between nuclear geometry and transcriptional regulation.
Structural plasticity of the cell nucleus
The development of a method for 3D image reconstruction of cell organelles from confocal microscopy data revealed a novel and rather surprising feature of NMDA receptor signaling in neurons. Neuronal nuclei are extremely plastic and their geometry is controlled by synaptic activity and NMDA receptor activation, with synaptically and extrasynaptically localized NMDA receptors having antagonistic effects. Activation of extrasynaptic NMDA signaling pathways that direct neurons toward degeneration and cell death (Hardingham et al., 2002) led to a loss of nuclear infoldings. In contrast, synaptic activity and calcium entry through synaptic NMDA receptors promoted the formation of infoldings. Given that chromosomes are not randomly arranged within the nucleus but organized in distinct chromosome territories (Cremer and Cremer, 2001), the dramatic changes in the nuclear architecture following stimulation of synaptic and extrasynaptic NMDA receptors may cause a spatial reorganization of chromosome territories. This could alter chromatin structures and accessibility and may affect gene transcription as well as other nuclear functions including pre-mRNA splicing and DNA repair.
ERK-MAP kinase pathway and nuclear geometry
Although the precise mechanisms through which the activation of synaptic or extrasynaptic NMDA receptors is translated into structural alterations of the cell nucleus remains to be investigated, the ERK-MAP kinase pathway appears to play a crucial role in the formation of new infoldings during synaptic activity. ERK-MAP kinases are part of a multifunctional signaling module that has been implicated in many aspects of neuronal physiology (Grewal et al., 1999; Sweatt, 2004). Their newly discovered and unexpected role in the translation of electrical events into changes in nuclear geometry raises the possibility that at least some of the numerous effects of blocking the ERK-MAP kinase signaling pathway on neuronal functions are the consequence of the blockade of signal-induced formation of nuclear infoldings.
Is there a ‘nucleoplasmic reticulum’ in hippocampal neurons?
Our study answered a longstanding question about the existence of a nuclear calcium store. Intranuclear extensions of the ER have been described previously in several cell types and species (Echevarría et al., 2003). The ER is a massive calcium store and an ER extending into the nucleus could operate as an intranuclear calcium reservoir. Provided that calcium release channels reside on the inner nuclear envelope [as has been suggested previously (Mak and Foskett, 1994; Gerasimenko et al., 1995; Humbert et al., 1996; Santella and Kyozuka, 1997; Marchenko et al., 2005)], such a calcium store could release calcium directly into the nucleus thereby bypassing the cytosol. Our 3D reconstruction and EM results show that the invaginations are lined by both the inner and the outer nuclear envelope; this demonstrates that the invaginated space is filled with cytosol (and not ER lumen) and thus a nucleoplasmic reticulum does not exist in hippocampal neurons. Nevertheless, infoldings enhance nuclear calcium signaling not only because of the larger nuclear surface and the possible increased number of NPCs, which facilitates calcium flux into and out of the nucleus but also because diffusion distances from cytosolic to nuclear locations are smaller. Moreover, in highly infolded nuclei, compartmentalization occurs allowing microdomains and nuclear substructures to be formed that regulate calcium dynamics differently from other parts of the nucleus.
Nuclear geometry, calcium signaling, and gene expression
The compartments formed are often unequal in size with small nuclear compartments being better than large compartments in resolving a 5 Hz calcium signal originating in the soma. The increased resolution of high-frequency calcium oscillations in small compartments may be relevant for signal transmission and efficient activation of target molecules. There are indeed examples for individual molecules or cellular processes such as transcription being activated by calcium oscillations in a frequency-dependent manner. In the case of αCaM kinase II, pulses of calcium signals at 4 Hz produce a more rapid and complete autophosphorylation of αCaM kinase II than pulses at 1 Hz (De Koninck and Schulman, 1998). Moreover, in a Jurkat T cell line, particular calcium oscillations can enhance the signaling efficiency leading to the activation of gene expression (Dolmetsch et al., 1997). It remains to be investigated whether neuronal nuclei contain calcium-regulated molecules capable of decoding calcium oscillations in particular in the theta frequency range (i.e., 4–8 Hz), which can be better resolved by the small than the large nuclear compartments.
Another feature of infolded nuclei that could impact on transcriptional regulation is the—compared with near-spherical nuclei—apparent increase in surface area and NPC immunoreactivity. There is very good evidence suggesting that actively transcribed DNA segments are located near NPCs (Taddei et al., 2006; Akhtar and Gasser, 2007). A possible increase in the number of NPCs in infolded nuclei may enhance the capacity of the nucleus for transcription-coupled anchoring of promoter regions to NPCs. This together with the phosphorylation of histones on serine 10 that preferentially takes place in highly infolded nuclei, may provide a significant advantage for neurons with infolded nuclei (over neurons with near-spherical nuclei) in optimizing steady-state gene expression and executing signal-induced transcriptional responses. Given that synaptically activated neurons form nuclear infoldings, this form of structural plasticity may present an adaptation to a metabolically more active state with an increased demand for gene expression.
This work was supported by the Alexander von Humboldt Foundation (Wolfgang Paul Prize to H.B.), the Deutsche Forschungsgemeinschaft, SFB 488, SFB 636, the European Union (EU) Network of Excellence NeuroNE, the EU Project GRIPANNT, and the Bundesministerium fuer Bildung und Forschung (Project DMSPiN). M.W. was supported by the Boehringer Ingelheim Fonds. We thank Harald Hermann-Lerdon for the antibodies to lamin B and lamin A/C, Jan Ellenberg for the lamin B-GFP construct, Alexei Ponomarenko for his help with power spectrum analysis, Ursula Weiss for carrying out immunoblot analyses, and Dirk Görlich for discussion.
- Correspondence should be addressed to Hilmar Bading, Department of Neurobiology, Interdisciplinary Center for Neurosciences, University of Heidelberg, Im Neuenheimer Feld 364, 69120 Heidelberg, Germany.