The Journal of Neuroscience, January 1994, 74(l): 187-197 Ca*+ Entry Via AMPA/KA Receptors and Excitotoxicity in Cultured Cerebellar Purkinje Cells James FL Brorson,’ Patricia A. Manzolillo, and Richard J. Miller’ Departments of ‘Neurology and 2Pharmacological and Physiological Sciences, The University of Chicago, Chicago, Illinois 60637 Initial studies of glutamate receptors activated by kainate certain diseaseprocesses (Rothman and Olney, 1987). NMDA (KA) found them to be Ca*+ impermeable. Activation of these receptors are ion channels that are highly Ca*+ permeable, receptors was thought to produce Ca*+ influx into neurons whereasnon-NMDA ionotropic receptors, activated by the ag- only indirectly by Na+-dependent depolarization. However, onists kainate (KA) and oc-amino-3-hydroxy-5-methyl-4-isox- Ca2+ entry via AMPA/KA receptors has now been demon- azole propionic acid (AMPA), have traditionally been thought strated in several neuronal types, including cerebellar Pur- to be Ca2+ impermeable (Ascher and Nowak, 1988; Mayer et kinje cells. We have investigated whether such Ca*+ influx al., 1988). For this reason, non-NMDA receptors have been is sufficient to induce excitotoxicity in cultures of cerebellar thought to causeCa*+ influx only indirectly due to Na+-depen- neurons enriched for Purkinje cells. Agonists at non-NMDA dent depolarization and the subsequentopening of voltage-gated receptors induced Ca2+ influx in the majority of these cells, Ca*+ channels. However, it is now clear that several types of as measured by whole-cell voltage clamp and by fura- [Ca*+li AMPAKA receptors are also directly Ca*+ permeable,and that microfluorimetry. To assess excitotoxicity, neurons were ex- thesereceptors can be important sourcesof Ca*+ influx in some posed to agonists for 20 min and cell survival was evaluated types of neurons and astrocytes. Murphy and Miller (1989) ob- by a fluorescence assay 24 hr later. KA (100 PM) reduced served that the Ca2+ influx induced by KA in cultured striatal neuronal survival relative to controls to 43 f 3% when ap- neurons could not be entirely blocked by antagonistsof voltage- plied in Na+-containing solution and to 45 Z!I 3% in Na+-free gated Ca*+ channels,and suggestedthat KA receptors might be solution. This toxicity was blocked completely by CNQX but directly permeableto Ca2+. Iino et al. (1990) showed that some only slightly by 100 @I Cd 2+ and 50 @I o-(-)-2-amino-5- cultured hippocampal neurons expressedKA-activated recep- phosphonovaleric acid. Both Purkinje neurons and non-Pur- tors which were Ca2+-permeable,giving rise to an inwardly kinje cell types present in the cultures were similarly vul- rectifying current-voltage (Z-V) relationship. Other authors have nerable to toxic KA exposure, but the population marked by also reported evidence for Ca2+-permeableAMPAKA recep- KA-induced Co2+ uptake was selectively diminished by the tors in various types of cultured neurons (Holopainen et al., excitotoxicity. Na+-independent excitotoxicity could also be 1989; Ogura et al., 1990; Gilbertson et al., 1991; Pruss et al., induced by domoate, AMPA, or glutamate. Compared to KA, 1991) and glia (Glaum et al., 1990; Jensen and Chiu, 1991; NMDA was relatively ineffective in inducing cell death. Most Bumashev et al., 1992b). Additionally, the cloning of several of the KA-induced excitotoxicity could be blocked by re- non-NMDA glutamate receptor subunitshas revealed that when moval of extracellular Ca2+ during the KA exposure and for combinations of theseare expressedin oocytes or cell lines, they a 5 min period thereafter. Furthermore, antagonists of the can give rise to glutamate receptors that are either Ca*+-per- Ca*+-activated enzymes nitric oxide synthase and calpain meable or Ca2+-impermeable(Hollmann et al., 1991; reviewed significantly reduced the KA-induced cell death. These re- in Miller, 1991). Finally, we have provided evidence that cul- sults show that non-NMDA receptor activation can cause tured cerebellar Purkinje neurons simultaneously expressboth excitotoxicity in cerebellar Purkinje neurons by mechanisms Ca2+-permeable and Ca2+-impermeable KA-activated receptors not involving Na+ influx, but rather depending on direct Ca2+ (Brorson et al., 1992a). permeation and activation of Ca2+-dependent enzymatic Work from a number of laboratories has implicated NMDA processes. receptors in in vitro models of glutamate excitotoxicity (Gold- [Key words: calcium, kainate, AMPA, glutamate receptors, berg et al., 1987; Rothman and Olney, 1987; Choi, 1988) and excitotoxicity, calpain I, nitric oxide, Purkinje cells] NMDA receptor activation seemsto be an essentialcomponent of neuronal damagein several in vivo models of ischemia and Glutamate receptor activation can increase the intracellular trauma (Faden and Simon, 1988; Park et al., 1988). However, Caz+concentration ([Ca2+],) in neurons, and this is thought to the NMDA receptor apparently does not account for all types contribute to Caz+-mediated excitotoxic neuronal cell death in ofglutamate-mediated neurotoxicity. In certain modelsofglobal ischemia, specific antagonists of non-NMDA receptors alone Received Dec. 15, 1992; revised July 8, 1993; accepted July 13, 1993. (Sheardown et al., 1990) or in combination with NMDA an- This work was supported by Grants DA-02 12 1 and MH-40 165 to R.J.M., and tagonists (Kaku et al., 1991) are neuroprotective. Domoate poi- by a Howard Hughes Medical Institute Physician Research Fellowship to J.R.B. soning following the ingestion of contaminated musselscauses Correspondence should be addressed to Richard J. Miller, Ph.D., Department of Pharmacological and Physiological Sciences, The University of Chicago, 947 a clinical syndrome resulting from the activation of non-NMDA East 58th Street, Chicago, IL 60637. glutamate receptors(Teitelbaum et al., 1990).The dietary toxins Copyright 0 1994 Society for Neuroscience 0270-6474/94/140187-l 1$05.00/O that have been postulated as the causative agentsin lathyrism 188 Brorson et al. - Ca+ Entry Via AMPA/KA Receptors and Excitotoxicity (Spencer and Schaumburg, 1983) and the Guam amyotrophic microfluorimetry, for whole-cell voltage-clamp measurements, and for lateral sclerosis-Parkinsonism-dementia complex can also ac- excitotoxicity assays contained tetrodotoxin (0.5 PM) and bicuculline (20 FM) in order to eliminate any synaptic contributions to the measured tivate non-NMDA receptors (Spencer et al., 1987; Meldrum effects. and Garthwaite, 1990). Thus, the action of glutamate at Ca2+ - Whole-cell patch-clamp measurements of ligand-gated Ca*+ currents permeable AMPAXA receptors might also be expected to play were performed as previously described (Brorson et al., 1992a). Cells an important role in some forms of experimental and clinical were accepted for study if a stable seal formed with a whole-cell resis- excitotoxicity. tance of at least 200 MR. Intracellular solutions contained (in mM) NMDG, 145; and BAPTA, 10; pH to 7.2 with HF (ATP was omitted Several properties of cerebellar Purkinje neurons indicate that to allow rundown of the voltage-gated Ca2+ currents). Na+-containing Ca2+-mediated processes might be particularly important in these extracellular solutions were as above. Na+-free solutions contained (in cells. First, they have among the highest concentrations in the mM) NMDG, 145; CaCl,, 2; HEPES, 10; and glucose, 10; pH to 7.4 brain of several of the Ca*+-binding proteins, including calbin- with HCl (Ma*+ and K+ were omitted to eliminate all non-Ca2+ cur- rents).Cells w&e held at a membrane potential of - 100 mV and agonists din D,,, and parvalbumin (Celio, 1990), which are thought to applied by bath perfusion of the recording chamber at a rate of l-l.5 be involved in Ca2+ buffering. Second, they exhibit high con- ml/min. Figures have been corrected for the perfusion delay of approx- centrations of markers for intracellular Ca2+ stores such as the imately 20 sec. All experiments were performed at room temperature. inositol trisphosphate and ryanodine receptors (Brorson et al., Data were recorded on an IBM PC-compatible system using an Axo- 199 1). They are known to have profuse glutamatergic synaptic patch 1D amplifier (Axon Instruments, Foster City, CA) and also on magnetic tape using a digital data recorder (VR- 1 OA, Instrutech Corp., inputs and to exhibit Ca2+-dependent action potentials (Llinas Elmont, NY). Current traces displayed have been replayed with elec- and Sugimori, 1980). Activation of glutamate receptors in these tronic filtering at 1 Hz. Because there was a slow rundown of receptor- neurons produces large increases in dendritic [Caz+], due to both gated currents in the whole-cell mode, relative current magnitudes evoked Ca2+ influx (Connor and Tseng, 1988; Llano et al., 199 1) and by different agonists or under different conditions were always compared to the average of flanking control responses by paired two-tailed Stu- Ca2+ mobilization from intracellular stores (Llano et al., 199 1). dent’s t tests. Finally, Purkinje cells are known to be extremely vulnerable to Excitotoxicity assays. Cell death and survival were assayed using hypoxic and ischemic neuronal death. They are often selectively fluorescein diacetate and propidium iodide, as described by Abele et al. damaged after asphyxiation or cardiac arrest. However, most (1990). To reduce background staining of cellular debris, prior to agonist studies have found that in adult tissues, Purkinje neurons lack exposure the coverslips of cultured neurons were placed for 24 hr in medium to which 10%