Address correspondence and reprint requests to Katsutoshi Furukawa, Laboratory of Neurosciences, Gerontology Research Center, 5600 Nathan Shock Drive, Baltimore, MD 21224, USA. E-mail: firstname.lastname@example.org
Intracellular calcium ions regulate the structure and functions of cytoskeletal proteins. On the other hand, recent studies have shown that the cytoskeleton, and actin filaments in particular, can modulate calcium influx through plasma membrane ligand- and voltage-gated channels. We now report that calcium release from inositol trisphosphate (IP3) and ryanodine-sensitive endoplasmic reticulum (ER) stores is modulated by polymerization and depolymerization of actin filaments in cultured hippocampal neurons. Depolymerization of actin filaments with cytochalasin D attenuates calcium release induced by carbamylcholine (CCh; a muscarinic agonist for IP3 pathway), caffeine (a ryanodine receptor agonist) and thapsigargin (an inhibitor of the ER calcium- ATPase) in both the presence and absence of extracellular calcium. Conversely, the actin polymerizing agent jasplakinolide potentiates calcium release induced by CCh, caffeine and thapsigargin. Cytochalasin D attenuated, while jasplakinolide augmented, thapsigargin-induced JNK activation and neuronal cell death. Our data show that the actin cytoskeleton regulates ER calcium release, suggesting roles for actin in the various physiological and pathological processes that involve calcium release.
Calcium release from the endoplasmic reticulum (ER) occurs in neurons in response to a variety of signals including neurotransmitters and neurotrophic factors (Rizzuto 2001). Two different types of ER calcium channels mediate calcium release, inositol trisphosphate (IP3) receptors and ryanodine receptors. IP3 is produced in response to activation of cell surface receptors coupled to a GTP-binding protein which activates phospholipase C, an enzyme that cleaves membrane phosphatidlyinositol-4,5-bisphosphate and thereby liberates IP3 into the cytoplasm. Ryanodine receptors are activated by calcium itself, and can also be activated by caffeine. Following calcium release, ER calcium stores are refilled via the activity of the ER calcium-ATPase. ER calcium release regulates a variety of processes in neurons including neurite outgrowth (Kocsis et al. 1994; Furukawa et al. 1998), synaptic plasticity (Wang et al. 1996; Szinyei et al. 1999) and cell survival or death (Yao et al. 1999; Glazner et al. 2001). In addition, excessive calcium release from the ER may contribute to neuronal dysfunction and death in several different pathological conditions including ischemic stroke and Alzheimer's disease (Paschen and Doutheil 1999; Mattson et al. 2000a).
Calcium plays important roles in regulating dynamic functions of the cell cytoskeleton; the polymerization of actin filaments and microtubules is sensitive to calcium. In neurons, modulation of actin filaments by calcium plays roles in the regulation of growth cone behaviors and synaptic plasticity (Mattson 1999; Matus 2000). Interestingly, recent findings suggest that the cytoskeleton can regulate calcium influx through plasma membrane channels in neurons, suggesting complex interactions between the cytoskeleton and systems that regulate calcium homeostasis. For example, actin depolymerization attenuates, whereas actin polymerization potentiates, calcium influx through voltage-dependent calcium channels and N-methyl-d-aspartate receptor channels in cultured neurons (Johnson and Byerly 1993; Rosenmund and Westbrook 1993; Furukawa et al. 1995, 1997). Although actin filaments are often intimately associated with ER membranes (LeBeux and Willemot 1975), the possibility that actin regulates ER calcium dynamics in neurons has not been investigated. In the present study, we present data that suggest roles for actin filaments in facilitating calcium release from IP3- and ryanodine-sensitive ER stores in hippocampal neurons.
Materials and methods
Cell cultures and experimental treatments
Primary hippocampal cell cultures were established from embryonic rats (day 18 of gestation) as detailed elsewhere (Mattson et al. 1995). Cells were plated onto polyethyleneimine-coated plastic culture dishes or glass coverslips at a density of 70–120/mm2, and were maintained in neurobasal medium containing B-27 supplements, 2 mm l-glutamine, 0.001% gentamicin sulfate and 1 mm HEPES (pH 7.2), in an atmosphere of 6% CO2 and 94% room air. Experiments were performed on cells that had been in culture for 7–10 days. Using these culture conditions, approximately 80–90% of cells were neurons and the remaining cells were astrocytes. Fura-2 and jasplakinolide were purchased from Molecular Probes (Eugene, OR, USA), and all other reagents were purchased from Sigma Chemical Co. (St Louis, MO, USA).
Cell staining with fluorescent phalloidin
Following experimental treatments (vehicle, cytochalasin D or jasplakinolide), cultured hippocampal cells were fixed for 30 min in 4% paraformaldehyde dissolved in phosphate-buffered saline (PBS). Cell membranes were permeabilized by incubating 5 min in a solution of 0.1% Triton X-100 in PBS. Neurons were then incubated for 20 min in PBS containing 5 units/mL fluorescein-phalloidin (Molecular Probes). Fluorescence images were acquired using a Zeiss 510 CSLM microscope. Levels of cellular phalloidin fluorescence were quantified by image analysis using the software provided by the manufacture of the confocal microscope. All the images were taken using identical microscope settings, and average pixel intensity of each image was quantified.
Analyses of neuronal survival
Neuronal survival was quantified by counting viable neurons in the same microscope fields immediately before experimental treatment and at 24 h after treatment as described previously (Mattson et al. 1995). Neurons with intact neurites of uniform diameter and a soma with a smooth and round appearance were considered viable, whereas neurons with fragmented neurites and a vacuolated or swollen soma were considered non-viable. Survival values were expressed as percentages of the initial number of neurons present before experimental treatment. Statistical comparisons were made using analysis of variance and Scheffe's post-hoc tests for pairwise comparisons.
Quantification of intracellular calcium concentrations
Fluorescence ratio imaging of the Ca2+ indicator dye fura-2 was used to quantify [Ca2+]i in neuronal cell bodies using methods detailed previously (Mattson et al. 1993, 1995). Cells were incubated for 30–40 min in the presence of 2 µm of the acetoxymethyl ester form of the [Ca2+]i indicator dye fura-2 and were then washed twice with fresh medium and allowed to incubate at least 40 min before imaging. Immediately before imaging, the normal culture medium was replaced with Hanks' balanced saline solution (Gibco) containing 10 mm HEPES buffer and 10 mm glucose. Cells were imaged using Zeiss Attofluor system with a 40x oil objective. The ratio of the fluorescence emission using two different excitation wave lengths (334 and 380 mm) was used to determine [Ca2+]i as described previously (Grynkiewicz et al. 1985). The system was calibrated using solutions containing either no Ca2+ or a saturating level of Ca2+ (1 mm) according to the following formula: [Ca2+]i = Kd[(R − Rmin)/(Rmax − R)](Fo/Fs). Values represent the average [Ca2+]i in the neuronal cell body. Experimental treatments were added to the bathing medium by dilution from 100–500× stocks. Calcium response to each agonist was also quantified in the [Ca2+]e-free solution (0 mm CaCl2 and 0.5 mm EGTA) after preincubation in the [Ca2+]e-free solution for 30 min.
C-Jun N-terminal kinase (JNK) assay
JNK assay was performed using the SAPK/JNK Assay Kit (Cell Signaling Technology; Beverly, MA, USA). Cell extracts (250 µL) were incubated overnight with 2 µg of c-jun (1–89) fusion protein. After extensive washing, the kinase reaction was performed in the presence of 100 µg cold ATP. Phosphorylation of c-Jun at Ser63 was measured by western blot using phospho-c-Jun (Ser63) antibody. Detected band densities were quantified by photodensitometry.
Evidence that cytochalasin D and jasplakinolide depolymerize and polymerize actin, respectively
Cytochalasins are a group of fungal metabolites that selectively disrupt actin filaments (Cooper 1987). In cell cultures of embryonic rat hippocampus, cytochalasin D (100 nm) caused depolymerization of actin as observed by confocal laser scanning microscope analysis of neurons stained with fluorescent-labeled phalloidin which is a reagent can bind selectively to polymerized actin (Cooper 1987). After treatment with cytochalasin D (100 nm) for 3 h, phalloidin-staining in neurons was reduced (Fig. 1). The mean fluorescence intensity of fluorescein-labeled phalloidin in cytochalasin d-treated cells was 52 ± 4.5% (n = 25) of untreated cells, demonstrating its predicted disruptiveeffecton microfilaments (Fig. 1). In addition, jasplakinonlide, a cyclic peptide which promotes actin polymerization (Bubb et al. 1994), was examined to determine whether this compound indeed polymerizes actin. Treatment with jasplakinolide (10 µm) for 3 h enhanced phalloidin-staining in cultured hippocampal neurons (Fig. 1). The mean fluorescence of fluorescein-labeled phalloidin in jasplakinolide-treated cells was 165 ± 14% (n = 26) of untreated cells, demonstrating its polymerizing effect on microfilaments (Fig. 1).
IP3-mediated calcium release is attenuated by actin depolymerization and enhanced by actin polymerization
The application of carbamylcholine (CCh, 50 µm) to cultured rat hippocampal neurons induced a transient increase in [Ca2+]i (the peak value was 198 ± 14 nm, n = 35). Pre-treatment of neurons with cytochalasin D (100 nm) for 3 h significantly attenuated the [Ca2+]i response to CCh (the peak value was 151 ± 17 nm, n = 29; p < 0.01). Pre-treatment of rat hippocampal neurons with jasplakinolide (10 µm) for 3 h significantly potentiated the [Ca2+]i response to CCh (the peak value was 284 ± 31 nm, n = 32) (Fig. 2a). To confirm that the attenuation of [Ca2+]i response to CCh is indeed mediated by actin depolymerization, we employed latrunculin A (Molecular Probes), another actin depolymerizing agent. After treatment of cultured neurons with 2.5 µm latrunculin A, the [Ca2+]i response to CCh was also significantly attenuated (the peak value was 162 ± 27 nm, n = 20). When [Ca2+]i response was recorded in the extracellular calcium-free condition, cytochalasin D also attenuated the [Ca2+]i response to CCh, whereas jasplakinolide augmented the [Ca2+]i response (Figs 2b and c). These results confirm that cytochalasin D and jasplakinolide can modulate the [Ca2+]i release from intracellular calcium stores in the absence of extracellular calcium. Collectively, these results suggest that actin polymerization enhances, whereas actin depolymerization inhibits calcium release from IP3-sensitive ER stores.
Calcium release from caffeine-sensitive stores is modulated by actin dynamics
Caffeine stimulates Ca2+ release from ryanodine-sensitive ER Ca2+ stores (Verkhratsky and Shmigol 1996). A transient [Ca2+]i increase was observed in hippocampal neurons in response to application of 10 mm caffeine (peak value was 227 ± 18 nm, n = 36). Pre-treatment with cytochalasin D (100 nm) for 3 h significantly attenuated the caffeine-induced [Ca2+]i increase (peak value was 151 ± 21 nm, n = 33) (Fig. 3a). After treatment of cultured neurons with 2.5 µm latrunculin A, the [Ca2+]i response to caffeine was also significantly attenuated (the peak value was 142 ± 23 nm, n = 22). On the other hand, jasplakinolide significantly potentiated the [Ca2+]i response to caffeine (peak value was 286 ± 19, n = 32). When cells were incubated in buffer without added CaCl2 and with 0.5 mm EGTA, cytochalasin D also attenuated the [Ca2+]i response to CCh, whereas jasplakinolide potentiated the [Ca2+]i response by caffeine (Figs 3b and c). These results suggest that calcium release from caffeine-sensitive ER stores is inhibited by actin depolymerization and potentiated by actin polymerization.
Actin filaments modulate ER calcium release induced by calcium-ATPase inhibition
We employed thapsigargin, a selective inhibitor of the ER Ca2+-ATPase (Treiman et al. 1998), to examine the effects of actin polymerization state on depletion of ER calcium stores. The application of thapsigargin (1 µm) elicited a sustained increase in [Ca2+]i (the peak value was 249 ± 24 nm, n = 38). Pre-treatment of neurons with cytochalasin D (100 nm) significantly attenuated the [Ca2+]i increase induced by thapsigargin (the peak value was 167 ± 23 nm, n = 33) (Fig. 4a). After treatment of cultured neurons with 2.5 µm latrunculin A, the [Ca2+]i response to thapsigargin was also significantly attenuated (the peak value was 153 ± 15 nm, n = 25). Neurons pre-treated with jasplakinolide (1 µm) for 3 h exhibited a significantly enhanced [Ca2+]i response to thapsigargin (the peak value was 314 ± 29 nm, n-29). Cytochalasin D also attenuated the [Ca2+]i response to thapsigargin, whereas jasplakinolide augmented the [Ca2+]i response in the calcium-free medium with 0.5 mm EGTA (Figs 4b and c).
Actin polymerization modifies ER calcium release induced neuronal death
Excessive calcium release from the ER may play roles in the death of neurons that occurs in disorders such as Alzheimer's disease and stroke (Paschen and Doutheil 1999; Mattson et al. 2000a). In order to examine how actin polymerization states modulate neuronal death induced by Ca2+ release from ER, we examined effects of cytochalasin D and jasplakinolide on thapsigargin-induced neuronal death. After pre-treatment with cytochalasin D (100 nm) or jasplakinolide (1 µm) for 3 h, cells were treated with thapsigargin (100 nm) in the continued presence of cytochalasin D or jasplakinolide for 24 h. The numbers of surviving neurons were counted before and after the treatment. Neither cytochalasin D nor jasplakinolide affected neuronal death by itself, whereas thapsigargin killed more than 70% of the neurons. Cytochalasin D significantly protected neurons against thapsigargin toxicity, whereas jasplakinolide exacerbated thapsigargin-induced neuronal death (Fig. 5a,b).
In order to determine how actin polymerization and depolymerization can modulate signaling pathways of thapsigargin-induced cell death, we analyzed effects of cytochalasin D and jasplakinolide on JNK activation by thapsigargin (Fig. 5c). Previous studies have implicated JNK in calcium-mediated neuronal cell death (Camandola et al. 2000). After pre-treatment with cytochalasin D (100 nm) or jasplakinolide (1 µm) for 3 h, cells were treated with thapsigargin (1 µm) in the continued presence of cytochalasin D or jasplakinolide for 15 min. As expected, treatment with thapsigargin potentiated JNK activity compared with vehicle treatment in every group (control, cytochalasin D, and jasplakinolide). Cytochalasin D significantly attenuated JNK activation by thapsigargin, whereas jasplakinolide significantly augmented JNK activity. These results suggest that neuronal cell death by thapsigargin could be mediated by JNK activation, and that JNK may be modulated by actin cytoskeletal dynamics.
The present findings suggest a novel role for actin filaments in regulating calcium release from IP3-sensitive and ryanodine-sensitive calcium stores in neurons. Actin polymerization enhances calcium release, whereas actin depolymerization decreases calcium release. Previous studies have provided evidence that actin filaments can modify calcium influx through plasma membrane voltage-dependent calcium channels in several cell types including snail ganglion neurons (Johnson and Byerly 1993), a vascular smooth muscle cell line (Nakamura et al. 2000), and guinea-pig cardiomyocytes (Rueckschloss and Isenberg 2001). The relationships between functions of IP3 and ryanodine receptors and actin filaments have not previously been studied in mammalian neurons. However, Baumann (2001) reported that depolymerization of actin filaments causes redistribution of ryanodine receptors through the entire ER in honeybee photoreceptor cells.
Interactions between various plasma membrane ion channel proteins and the cytoskeletal proteins have been implicated in mediating spatial localization of the channel proteins and in the regulation of their functions. For example, chloride channel proteins have been shown to interact with an actin-containing protein complex that may control the localization of the channels in the membrane (Suginta et al. 2001). An interaction of the actin-binding protein alpha-actinin-2 with cardiac potassium channels was demonstrated, and shown to affect both channel localization and current density (Maruoka et al. 2000). Galli and DeFelice (1994) reported that the microtubule-disrupting agent colchicine alters the kinetics of L-type Ca2+ channels in intact cardiac cells by increasing their probability of being in the closed state, and by decreasing the inactivation time constant. Changes in the distribution and/or channel activities of IP3 receptors and ryanodine receptors may therefore underlie the changes in calcium release that we observed in hippocampal neurons treated with agents that cause depolymerization or polymerization of actin filaments.
Many different signaling pathways affect both cytoskeletal dynamics and cellular calcium homeostasis including growth factors, neurotransmitters and cell adhesion molecules (Anand-Apte and Zetter 1997; Cary et al. 1999). Our findings suggest the possibility of reciprocal functional relationships between calcium and the cytoskeleton in regulating a diverse array of cellular processes. For example, calcium influx into growth cones can induce rapid polymerization of actin and filopodial extension (Lau et al. 1999). Because actin polymerization enhances calcium release from ER stores, this might provide a mechanism for local amplification of the calcium signal and hence increased growth cone motility. On the other hand, greater or more prolonged elevations of intracellular calcium levels can cause depolymerization of actin. Interactions between calcium and the actin cytoskeleton system might also regulate the structural and functional changes underlying different types of synaptic plasticity, such as long-term potentiation (Matus 2000). We previously provided evidence that such calcium-induced actin depolymerization results in decrease calcium influx through plasma membrane voltage-dependent calcium channels and N-methyl-d-aspartate channels as a result of enhance rundown of the channels (Furukawa et al. 1997). Similar processes may mediate the effects of actin dynamics on ER calcium release channels, although the specific molecular mechanisms remain to be determined.
In addition to its important roles in regulating various physiological processes in neurons, calcium release from ER stores is implicated in the death of neurons that occurs in disorders such as Alzheimer's disease (Mattson et al. 2000a) and stroke (Paschen and Doutheil 1999). For example, inhibitors of ER calcium release can protect neurons against excitotoxic and ischemic insults (Guo et al. 1997; Mattson et al. 2000b). We found that hippocampal neuronsin which actin filaments were disrupted with cytochalasin D were more resistant to death induced by thapsigargin, whereas neurons exposed to the actin-polymerizing agent jasplakinolide were more vulnerable. Previous studies have shown that cytochalasin D can protect neurons against excitotoxicity, and it was proposed that the mechanism involved decreased calcium influx through plasma membrane voltage-dependent channels and ionotropic glutamate receptor channels (Furukawa et al. 1997). Our data suggest that reduced calcium release through ER IP3 receptors and/or ryanodine receptors might also contribute to the neuroprotective effect of actin depolymerization. Collectively, the available data suggest that actin dynamics play important roles in regulating neuronal calcium homeostasis in both physiological and pathological settings.