AMPA receptors and stargazin-like transmembrane AMPA receptor-regulatory proteins mediate hippocampal kainate neurotoxicity
- Susumu Tomita*,†,‡,
- R. Keith Byrd*,
- Nathalie Rouach§,¶,
- Camilla Bellone§,
- Angela Venegas*,
- Jessica L. O'Brien§,
- Kwang S. Kim†,
- Olav Olsen*,
- Roger A. Nicoll*,§,‖, and
- David S. Bredt‖,**
- Departments of *Physiology and
- §Cellular and Molecular Pharmacology, University of California, San Francisco, CA 94143;
- **Department of Integrative Biology, Eli Lilly and Company, Indianapolis, IN 46285; and
- †Department of Cellular and Molecular Physiology, Yale University School of Medicine, New Haven, CT 06520
-
Contributed by Roger A. Nicoll, September 20, 2007 (received for review August 20, 2007)
Abstract
Naturally occurring glutamate analogs, such as kainate and domoate, which cause excitotoxic shellfish poisoning, induce nondesensitizing responses at neuronal α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors. In addition to acting on AMPA receptors, kainate and domoate also activate high-affinity kainate-type glutamate receptors. The receptor type that mediates their neurotoxicity remains uncertain. Here, we show that the transmembrane AMPA receptor-associated protein (TARP) γ-2 (or stargazin) and the related TARP γ-8 augment responses to kainate and domoate by making these neurotoxins more potent and more efficacious AMPA receptor agonists. Genetic deletion of hippocampal enriched γ-8 selectively abolishes sustained depolarizations in hippocampus mediated by kainate activation of AMPA receptors. γ-8 knockout mice display typical kainate-induced seizures; however, the associated neuronal cell death in the hippocampus is attenuated in mice lacking γ-8. This work decisively demonstrates that TARP-associated AMPA receptors mediate kainate neurotoxicity and identifies TARPs as targets for modulating neurotoxic properties of AMPA receptors.
The α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-preferring glutamate receptors are selective cation channels and mediate most of the postsynaptic depolarization that induces neuronal firing. Dynamic trafficking of neuronal AMPA receptor proteins participates in the plasticity of synaptic transmission that underlies aspects of learning and memory. AMPA receptor channels comprise heterotetramers of subunits GluR1–4, and each subunit can be alternatively spliced as either a flip or flop form (1, 2). The distinct expression of these AMPA receptor subunits and alternatively spliced isoforms in discrete neuronal populations helps determine differential AMPA receptor function and synaptic plasticity throughout the brain (3–7).
Neuronal AMPA receptors also contain transmembrane AMPA receptor-regulatory protein (TARP) auxiliary subunits (8). The TARP family comprises five isoforms: γ-2 (or stargazin), γ-3, γ-4, γ-7, and γ-8, which are predominantly expressed in adult cerebellum, adult cerebral cortex, early developmental brain, cerebellar Purkinje cells, and adult hippocampus, respectively (9, 10). Stargazer mice, which show absence epilepsy and cerebellar ataxia, have a genetic mutation in stargazin, the prototypical TARP (11). Physiological studies have shown that cerebellar granule cells from stargazer mice selectively lack functional AMPA receptors (12). Biochemical and cell biological studies have shown that stargazin binds to AMPA receptors and traffics the receptors to the neuronal cell surface (13, 14). Interaction of stargazin with the postsynaptic density-95 and related proteins (15) mediates synaptic clustering of AMPA receptors (16, 17). Stargazin also cooperates with postsynaptic density-95 to control synaptic plasticity (18–20). In addition to trafficking AMPA receptors, stargazin modulates the channel gating and receptor pharmacology (21–24).
Glutamate-evoked ion flow through AMPA receptors terminates within milliseconds because of rapid channel deactivation and desensitization. Structural studies show that channel desensitization involves rearrangement and destabilization of the interface between GluR subunits (25). Whereas endogenous glutamate evokes transient responses at AMPA receptors, exogenous kainate and domoate produce partial but nondesensitizing channel openings. Thereby, these conformationally constrained glutamate analogs (Fig. 1 A) produce powerful neuronal depolarizations. In addition to these actions at AMPA receptors, kainate also activates its own class of glutamate receptors. These so-called “kainate receptors” participate in excitatory transmission at a subset of synapses (26, 27). Kainate receptors comprise heterotetramers of GluR5–7 and KA-1 and KA-2, which share significant sequence homology with AMPA receptor GluR1–4 subunits (1, 2). However, TARPs do not interact with high-affinity kainate receptors (28).
TARPs enhance AMPA receptor responses to kainate and domoate. (A) Structures of glutamate, kainate, and domoate. (B and C) Oocytes were injected with GluR1 flip (GluR1i), GluR1 flop (GluR1o), and stargazin (B) or γ-8 (C) cRNAs, and responses to glutamate (Glu), kainate (KA), and domoate (DA) were recorded. Stargazin and γ-8 enhance responses to kainate (500 μM) and domoate (100 μM) relative to those of glutamate (500 μM). (D) Oocytes were injected with GluR subunits alone (20 ng), GluRi isoforms with stargazin (0.1 ng each), or GluRo isoforms (0.2 ng) with stargazin (0.1 ng). Stargazin enhanced responses of all AMPA receptor isoforms to kainate (500 μM) relative to glutamate (500 μM). The data in B, C, and D show means ± SEM (n = 5). ND, not detectable.
Kainate and domoate are produced by seaweed and plankton (29). During an algae bloom, domoate can become concentrated in shellfish, particularly oysters, mussels, and scallops. In 1987, consumption of domoate-tainted mussels affected hundreds of Canadians, who suffered from seizures, coma, memory loss, and even death (30). Pathological examination of those killed showed that domoate caused major neuropathology, especially in the hippocampus (31). By analogy, classic neurotoxicology studies showed that stereotaxic injection of kainate into rodent striatum causes neuronal degeneration but spares axons of passage (32). Subsequently, kainate and domoate have been used extensively to model excitotoxic and neurodegenerative diseases. Whether the neurotoxicity of domoate and kainate reflects drug actions on AMPA receptors or on high-affinity kainate receptors has remained uncertain.
In neurons, kainate evokes larger steady-state AMPA receptor responses than does glutamate (33, 34); however, kainate is relatively weak in activating many heterologously expressed AMPA receptor GluR isoforms (35, 36). We previously showed that stargazin and related TARPs make kainate more active toward AMPA receptor GluR subunits (37). Here, we asked whether TARPs also modulate the efficacy of the environmental neurotoxin domoate. Furthermore, we took advantage of recently developed TARP γ-8 knockout mice (38) to decisively test whether AMPA receptors mediate neuronal loss by kainate-type excitotoxins.
Results
Stargazin and γ-8 dramatically increase the magnitude of glutamate evoked currents from GluR1 cRNA injected oocytes (28), and this system allows studies of TARP effects on AMPA receptor pharmacology. We found that oocytes injected with GluR1 and stargazin showed 10-fold higher responses to kainate or domoate than to glutamate, whereas oocytes injected with GluR1 alone showed greater responses to glutamate than to kainate or domoate (Fig. 1 B). This potentiation by stargazin of AMPA receptor responses to kainate or domoate occurs on both the flip (GluR1i) and flop (GluR1o) alternatively spliced versions of GluR1 (Fig. 1 B). We found that γ-8 also vastly increased responses of GluR1i and GluR1o to both kainate and domoate (Fig. 1 C). We further explored the AMPA subunit specificity for stargazin-mediated potentiation of kainate responses and tested all eight alternatively spliced AMPA receptor subunits. We found that stargazin increased the I KA/I Glu ratio for all isoforms. Interestingly, this enhancement varied from a minimum of ≈2-fold on GluR2i to maximum of >100-fold for GluR3o (Fig. 1 D). By contrast, neither stargazin nor γ-8 enhanced kainate evoked currents from GluR6 [supporting information (SI) Fig. 5].
We next explored the mechanism for TARP-mediated enhancement of kainate actions on AMPA receptors. Previous studies have shown that kainate serves as a partial agonist on recombinant AMPA receptors (39). To quantify maximal AMPA receptor responses to glutamate and kainate, we added cyclothiazide (CTZ), which blocks receptor desensitization. For oocytes expressing GluR1, CTZ enhanced glutamate-evoked currents 67-fold and CTZ enhanced kainate-evoked currents 5.2-fold (Fig. 2 A). In the presence of CTZ, kainate-evoked currents were 7% that of glutamate-evoked currents, and this reflects partial agonism of kainate. In oocytes that coexpressed GluR1 and either stargazin or γ-8, CTZ dramatically enhanced glutamate-evoked currents by 60- or 10-fold, respectively. However, with stargazin or γ-8, CTZ only increased kainate-evoked currents by 2.4- or 1.4-fold, respectively. Importantly, in oocytes expressing GluR1 with stargazin or γ-8, currents evoked by kainate plus CTZ were fully 60–70% those evoked by glutamate plus CTZ. These data indicate that TARPs preferentially enhance AMPA receptor responses to kainate by making kainate a nearly full agonist. Consistent with previous studies, we found that stargazin increased by ≈3-fold the potency of glutamate and kainate on GluR1 (Fig. 2 B and C). Interestingly, γ-8 increased these potencies by ≈10-fold (Fig. 2 B and C).
TARPs enhance kainate potency and efficacy of AMPA receptor. Oocytes were injected with GluR1 flip (GluR1i) alone or together with stargazin (STG) or γ-8 and responses to glutamate or kainate with or without CTZ were recorded. Agonist-evoked currents were normalized by glutamate (500 μM) with CTZ (100 μM) as maximum response. Stargazin and γ-8 enhance responses to kainate (200 μM) relative to those of glutamate (1000 μM). As reflected by responses in presence of CTZ (100 μM), stargazin and γ-8 make kainate a nearly full agonist. Shown are means ± SEM (n = 6). (B and C) Oocytes were injected with GluR1 flip alone or together with stargazin or γ-8, and dose-dependent responses to glutamate plus CTZ (100 μM) or kainate were recorded. Shown are means ± SEM (n = 7).
To determine whether TARPs contribute to pharmacological and toxicological actions of kainate in vivo, we took advantage of recently generated mice lacking the hippocampal TARP γ-8 (38). These mice show dramatic losses of AMPA receptor protein and extrasynaptic AMPA receptors channels but show a modest decrease in synaptic AMPA receptors (38). The AMPA receptor antagonist GYKI53655 was used to evaluate high-affinity kainate receptor function in these mice. In agreement with previous work showing that TARPs do not traffic kainate receptors (28), mossy-fiber-evoked synaptic responses in CA3 pyramidal cells mediated by these GYKI53655-insensitive kainate receptors are unaltered in the hippocampus of γ-8 knockout mice (38). In outside-out patches from CA1 pyramidal cells, kainate induced substantial inward currents are mediated solely by AMPA receptors, because they are blocked by GYKI53655 (Fig. 3 A and B). These responses were abolished in γ-8 knockout mice (Fig. 3 A and B). It has been reported that bath application of low doses of kainate or domoate induces a kainate receptor-mediated inward whole-cell current in CA1 pyramidal cells (40). These responses are not seen in outside-out somatic patches because of either their low density or their exclusion from the somatic membrane. We repeated these experiments by applying domoate (300 nM) in the presence of GYKI53655 (100 μM) (Fig. 3 C and D). This kainate receptor-mediated current was unaltered in γ-8 knockout mice (Fig. 3 C and D). These results demonstrate that hippocampal kainate receptors function normally in γ-8 knockout mice and that kainate activation of AMPA receptors is dramatically reduced. This genetic separation of AMPA and kainate receptor responses affords a robust preparation to determine which receptor type accounts for kainate-mediated cell death.
Kainate-mediated sustained depolarizations of hippocampal neurons require TARP γ-8. (A and B) Kainate-evoked currents (1 mM, 2 s) from somatic outside-out patches were observed in pyramidal cells from wild-type mice (+/+) (n = 7 from 2 mice) but were undetectable in γ-8 knockouts (−/−) (n = 7 from 2 mice). (Scale bar, 50 pA; 2 s.) (C and D) Domoate-evoked whole-cell currents (500 nM, 4 min) in the presence of 100 μM GYKI 53655 were observed in CA1 pyramidal cells from both wild-type mice (n = 7 from 3 mice) and γ-8 knockouts (−/−) (n = 9 from 2 mice) at similar levels. The data show means ± SEM from the indicated number of experiments.
Peripheral administration of kainate to rodents produces a well characterized seizure syndrome associated with pronounced hippocampal neurotoxicity (41). This preferential lesioning of the hippocampus parallels the toxicity observed in postmortem evaluation of patients poisoned by domoic acid (31, 42). We performed i.p. injections of kainate and found that 35 mg/kg reliably produced seizures in wild-type and γ-8 knockout mice. These seizures were characterized by head movement, body tremor, rearing with falling, and tonic–clonic convulsions. The latency and extent of seizures did not vary between wild-type and γ-8 knockout mice (Fig. 4 A). To avoid variability from mice genetic background, we characterized kainate-induced hippocampal cell loss in γ-8 knockout mice compared with γ-8 heterozygous mice. Two experimentalists who were naïve to genotype or drug treatments independently scored cell loss in the CA1 and CA3 regions of hippocampus. Strikingly, hippocampal neurons in γ-8 knockout mice were resistant to kainate-induced cell death (Fig. 4 B and C), indicating the AMPA receptor–TARP γ-8 complex underlies this well studied excitotoxicity.
Kainate-induced hippocampal cell death requires the hippocampal-enriched γ-8. (A) γ-8 knockouts (−/−) (n = 5) show a similar extent of kainate-induced seizure, as do γ-8 heterozygous mice (+/−) (n = 8). The seizure index was quantified as described in Experimental Procedures. (B) Cresyl violet staining of hippocampi from kainate-treated (35 mg/kg) γ-8 heterozygous mice (+/−) (n = 15) showed extensive cell death in the CA3 region (arrow) and some cell loss in the CA1 region (asterisk) compared with untreated mice (+/−) (n = 15). Kainate did not induce cell death in γ-8 knockouts (−/−) (n = 8). (C) Investigators unaware of the mouse genotype took images of the CA1 and CA3 regions of the hippocampus, followed by quantitation of survival cell density by using cresyl violet staining and ImageJ software. Sr, stratum radiatum; Py, pyramidal cell; Dg, dentate gyrus. The data in A and C show means ± SEM from the indicated number of experiments.
Discussion
Our results show that TARPs confer AMPA receptor sensitivity to both kainate and domoate. The structural mechanism for TARP enhancement of kainate and domoate efficacy on AMPA receptors remains uncertain; however, recent studies of AMPA receptor structure and of AMPA receptor–TARP interactions provide important insights. X-ray crystallography of the glutamate-binding module of AMPA receptors shows that agonists induce a profound closure of the clamshell-shaped extracellular receptor domain (39). For a series of full and partial agonists, the extent of glutamate-binding domain closure matches the degree of agonist efficacy (43). Biochemical and electrophysiological analyses indicate that TARPs directly interact with the glutamate-binding module of AMPA receptors (14, 44). Taken together, these data suggest a model in which TARPs enhance the extent of domain closure induced by kainate and domoate and thereby enhance their efficacy. Structural studies of TARP–AMPA receptor complex will be needed to assess this model.
Through actions on neuronal AMPA receptors, kainate induces strong, sustained inward currents in somatic patches from CA1 pyramidal cells. That we found an absence of these responses in patches from γ-8 knockout mice shows that TARP subunits dictate AMPA receptor pharmacology in vivo. By contrast, high-affinity kainate receptors remain intact at mossy fiber hippocampal synapses of γ-8 knockout mice (38), at CA1 pyramidal cells (Fig. 3 C and D), and at granule cells from stargazer knockout mice (28). This retention of kainate receptors agrees with experiments in heterologous cells showing that stargazin does not regulate GluR6 (28). In Caenorhabditis elegans, functional expression of AMPA receptor subunits requires both a stargazin-like TARP and the single-pass transmembrane SOL-1 protein (45, 46). Whether mammalian kainate receptors also contain auxiliary subunits remains uncertain.
Classic studies have shown that peripheral or central administration of kainate or domoate induces hippocampal-derived seizures that spread to other limbic structures (47). The resulting excitotoxicity causes apoptotic cell death, which manifests prominently in the CA3 region of hippocampus (41). We find that γ-8 knockout mice display an unaltered threshold to kainate-induced seizures; however, these mice are resistant to the ensuing hippocampal neuronal loss. That kainate administration induced seizures in the γ-8 knockouts similar to wild-type mice indicates that seizure initiation in these mice may occur outside of the hippocampus or may involve high-affinity kainate receptors. Indeed, previous studies have shown that kainate-induced seizures are attenuated in mice lacking the GluR6 subunit of high-affinity kainate receptors (48). These studies also have shown that GluR6 knockouts show a blunted kainate-mediated induction of glial fibrillary acidic protein in hippocampus. The absence of kainate-induced hippocampal neuronal death in γ-8 knockouts indicates that TARP-associated AMPA receptors are essential for kainate excitotoxicity. Why CA3 neurons are particularly vulnerable to kainate toxicity remains unclear and may reflect the recurrent connectivity of this region.
Because AMPA receptors play a central role in neurodegenerative, epilepsy, and mental diseases, this work has important therapeutic implications. Indeed, AMPA receptor antagonists have shown efficacy in animal models of cerebral ischemia and in seizure reduction in clinical trials (49). By analogy, compounds that antagonize the interaction between TARPs and AMPA receptors should down-regulate strong depolarizing responses and could be useful in excitotoxic processes. In other circumstances, enhancing TARP interactions may be beneficial, because drugs that potentiate AMPA receptor responses are emerging as novel therapeutics for treatment of cognitive and mental diseases (50–53).
Materials and Methods
Electrophysiology by Using Xenopus laevis Oocytes and Hippocampal Slices.
For oocytes, two electrode voltage clamp recordings were performed as described previously (14). Standard techniques were used to prepare and record from transverse acute hippocampal slices (300–400 μm thick) from 12- to 26-day-old wild-type and γ-8 knockout mice (38). Somatic outside-out patch recordings of AMPA receptor-mediated currents were made from CA1 pyramidal cells held at −70 mV. Currents were evoked by local application of 1 mM kainate for 2 s. Somatic whole-cell voltage clamp recordings of kainate receptor-mediated currents were made from CA1 pyramidal cells by using 2–6 ΩM electrodes. Currents were evoked by local application (4 min) of 500 nM domoate in the presence of 100 μM GYKI53655. All data are expressed as means ± SEM. Statistical significance was determined by unpaired Student t test.
Kainate-Induced Seizures and Neurotoxicity.
Mice were injected i.p. with kainic acid (35 mg/kg). Mice were monitored for 2 h after kainic acid injection for seizure onset and severity, as described previously (48). Seizure severity was scored as no response (score of 0), limb pawing (score of 1), bilateral pawing (score of 2), jumping (score of 3), and death (score of 4). Seizure severities for all animals in each group were averaged to yield the seizure index.
Seven days after kainate injection, animals were anesthetized with pentobarbital and transcardially perfused with 4% paraformaldehyde. Brains were dissected, dehydrated, and embedded in paraffin. Sagittal sections (6 μm) were cut and stained with cresyl violet. Two independent experimentalists unaware of drug treatment or mouse genotype took images of the CA1 and CA3 regions of the hippocampus, followed by analysis of cell density by using ImageJ software. All data are expressed as means ± SEM. Statistical significance was determined by unpaired Student t test.
Acknowledgments
This work was supported by grants from the National Institutes of Health (D.S.B. and R.A.N.) and the Esther A. and Joseph Klingenstein Foundation and the Edward Mallinckrodt Jr. Foundation (S.T.). N.R. was supported by the Human Frontier Science Program Organization.
Footnotes
- ‡To whom correspondence may be addressed at Department of Cellular and Molecular Physiology, Yale University School of Medicine, New Haven, CT 06520. E-mail: susumu.tomita{at}yale.edu
- ‖To whom correspondence may be addressed. E-mail: nicoll{at}cmp.ucsf.edu or E-mail: bredt{at}lilly.com
-
Author contributions: S.T., R.A.N., and D.S.B. designed research; S.T., R.K.B., N.R., C.B., A.V., and J.L.O. performed research; S.T., R.K.B., N.R., C.B., A.V., J.L.O., K.S.K., and O.O. analyzed data; and S.T., O.O., R.A.N., and D.S.B. wrote the paper.
-
↵ ¶Present address: Unité 840, Institut National de la Santé et de la Recherche Médicale, College de France, Paris 75005, France.
-
The authors declare no conflict of interest.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0708970104/DC1.
- Abbreviations:
- AMPA,
- α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid;
- CTZ,
- cyclothiazide;
- TARP,
- transmembrane AMPA receptor-associated protein.
- © 2007 by The National Academy of Sciences of the USA
