Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Excess activation of N-methyl-D-aspartate (NMDA) or -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate receptors produces neuronal death and has been implicated as a major cause of hypoxic-ischemic neuronal injuries (Choi, 1988; Watkins, 1991; Olney, 1994). Consequently, maneuvers antagonizing NMDA and AMPA/kainate neurotoxicity have been developed for the prevention of neuronal death after hypoxic ischemia. Several NMDA receptor antagonists, including MK-801 and dextrorphan, provide substantial protection against ischemic injuries induced by oxygen-glucose deprivation in vitro or occlusion of middle cerebral artery (Simon et al., 1984; Goldberg et al., 1987). However, the therapeutic potential of NMDA receptor antagonists is limited by hypertension (Muir et al., 1997), behavioral effects such as psychosis and hyperlocomotion (Ouagazzal et al., 1993; Ellison, 1995), less protective effects against global ischemic injuries (Ross and Duhaime, 1989; Buchan et al., 1991), and direct neurotoxicity in several areas of the brain (Olney et al., 1991). Recently, low-affinity channel-blocking NMDA receptor antagonists have been developed that reduce the unacceptable side effects of NMDA antagonists (Parsons et al., 1999; Rogawski, 2000). In addition to NMDA receptor antagonists, antagonists [such as 2,3-dihydroxy-6-nitro-7-sulfamoylbenzo(f)quinoxaline] acting on AMPA and kainate receptors mitigate selective neuronal loss after transient global ischemia in the rat (Sheardown et al., 1990) but they are partially neuroprotective against focal ischemic insults (Buchan et al., 1991).

Intracellular signaling pathways that mediate excitotoxicity have been outlined. Excess activation of ionotropic glutamate receptors results in massive influx and accumulation of Ca²⁺ ions as well as Na⁺ and Cl ions. The former is required for rapidly evolving excitotoxicity through Ca²⁺-permeable NMDA and AMPA/kainate glutamate receptors (Choi, 1987). In turn, an accumulation of Ca²⁺ ions produces toxic molecules such as superoxide, hydroxyl radicals, and nitric oxide (Dawson et al., 1991; Lafon-Cazal et al., 1993; Dugan et al., 1995). However, blockers of voltage-gated Ca²⁺ channels, antioxidants, or inhibitors of nitric-oxide synthase were partially neuroprotective against excitotoxicity and hypoxic-ischemic insults (Greenberg et al., 1990; Hall et al., 1990; Kinouchi et al., 1991; Nagafuji et al., 1992). Although calpains and caspases are activated to digest various proteins subsequent to activation of glutamate receptors, selective inhibitors of these proteases slightly alleviate excitotoxicity (Lee et al., 1991; Endres et al., 1997). Thus, efficient and secure strategies that effectively reduce NMDA or AMPA/kainate neurotoxicity need to be developed to counteract hypoxic-ischemic neuronal death.

Prokaryotic organisms such as Bacillus and Streptomyces species were shown to produce various antioxidants and apoptosis-modulating agents as process of stress adaptations (Hochman et al., 1997). Several diphenazine compounds that were isolated from Streptomyces species attenuated glutamate toxicity in the rat N18-RE-105 neuronal cell line (Kim et al., 1997). We have screened approximately 7000 cultural broths of soilborne Streptomyces species to seek out neuroprotective agents that protect against excitotoxicity. Among these, complestatin, a bicyclo hexapeptide, prevented kainate-induced neuronal death at micromolar concentrations. Complestatin (having the structural components 4-hydroxyphenylglycine and 3,5-dichloro-4-hydroxyphenylglycine) was first isolated from the mycelium of Streptomyces lavendulae as an anticomplement agent (Kaneko et al., 1980). The present study was performed to delineate neuroprotective effects and mechanisms of complestatin against excitotoxicity and to investigate whether the neuroprotective effects of complestatin could be extended to ischemic injuries in vitro and in vivo.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Primary Cortical Cell Cultures. Cerebral cortices were removed from the brains of 15-day-old fetal mice in accordance with a protocol approved by our institutional animal care committee. The neocortices were triturated and plated on 24-well plates (with approximately 10⁵ cells/culture well) precoated with 100 µg/ml poly-D-lysine and 4 µg/ml laminine, in Eagle's minimal essential media (Earle's salts, supplied glutamine-free), and supplemented with horse serum (5%), fetal bovine serum (5%), 2 mM glutamine, and 21 mM glucose. Cultures were maintained at 37°C in a humidified atmosphere of 5% CO₂. After 7 days in vitro (DIV 7), the cultures were shifted to the plating media containing 10 µM cytosine arabinoside without fetal serum. Cultures were then fed twice per week.

Induction and Analysis of Excitotoxicity, Oxidative Stress, or Apoptosis. Mixed cortical cell cultures containing neurons and glia (DIV 10-14) were exposed to excitotoxins (NMDA, AMPA, or kainate), oxidative stress (Fe²⁺, H₂O₂, buthionine sulfoximine), or proapoptosis agents (staurosporine or cyclosporine A) in Eagle's minimal essential media supplemented with 21 mM glucose and 26.5 mM bicarbonate. The morphology of the degenerating neurons was observed under a phase contrast microscope over the next 24 h. Neuronal death was analyzed 24 h later by measuring how much lactate dehydrogenase (LDH) had been released into the bathing medium. The percentage of neurons undergoing actual neuronal death was normalized to the mean LDH value that is found after a 24-h exposure to 500 µM NMDA (defined as 100) or a sham control (defined as 0). The levels of neuronal injury by LDH assay were routinely confirmed by counting viable neurons excluding trypan blue.

Purification of Complestatin. Complestatin was isolated from the culture broth of a microorganism identified as Streptomyces species 60910. The fermentation broth (10 liters) of strain 60910 was centrifuged at 6000 rpm to produce both supernatant and mycelial cake. The supernatant was subjected to a column of Diaion HP-20 and washed with 70% methanol. The active fraction was eluted with 70% acetone, concentrated in vacuo, and then chromatographed on a SiO₂ column eluted with butyl acetate/butanol/H₂O/acetic acid (4:4:1:1, v/v). Active fractions were repeatedly precipitated with methanol to derive about 120 mg of pure active compound. The mycelial cake was extracted with 70% acetone and concentrated under reduced pressure. The residue was adjusted to pH 4 and extracted twice with ethyl acetate. The solvent layer was precipitated with MeOH. The purity (>99%) of the compound was finally checked by high-performance liquid chromatography with an ODS column with 45% CH₃CN and 0.04% trifluoroacetic acid as elution solvent. By using UV, high-resolution fast atom bombardment-mass spectroscopy, ¹H NMR, ¹³C NMR, distortionless enhancement by polarization transfer, heteronuclear multiple quantum coherence spectroscopy, and heteronuclear multiple-bond correlation spectroscopy spectra, the chemical structure of the purified compound was verified as complestatin.

Analysis of Calcium Accumulation and Influx. Measurement of intracellular free calcium concentration ([Ca²⁺]_i) was carried out using a Ca²⁺-sensitive indicator, fura-2, as previously reported (Seo et al., 1999). Cortical cell cultures (DIV 12) grown on 35-mm glass-bottom dishes (MatTek, Ashland, MA) were loaded with 5 µM fura-2 acetoxymethyl ester plus 2% Pluronic F-127 for 30 min at room temperature. Cells were washed three times with a HEPES salt solution containing 120 mM NaCl, 5 nM KCl, 2.3 mM CaCl₂, 15 mM glucose, 20 mM HEPES, and 10 mM NaOH, with pH 7.4. The fura-2 fluorescent signals (excitation = 340/380 nm, emission = 510 nm) were acquired with a Nikon (Tokyo, Japan) Diaphot inverted microscope and charge-coupled device camera. Fura-2 ratio images were analyzed using a Quanticell 700 system (Applied Imaging, New Castle, UK).

Patch-Clamp Recordings. Whole-cell voltage-clamp recordings were obtained from cortical neurons (DIV 12-14) by using an Axopatch 100A amplifier (Axon Instruments, Foster City, CA) at room temperature (22-24°C). Patch pipettes (2.5-3.0 M) were pulled on a Narishige (Tokyo, Japan) PP-83 electrode puller from Kimax borosilicate glass capillaries (Garner Glass, Claremont, CA). Ionotropic currents that were elicited from cortical neurons, held at 70 mV, were stored through an analog-to-digital converter (Digidata-1200; Axon Instruments) and connected to a PC by using Axotape software (Axon Instruments). To block synaptic transmissions, 0.5 µM tetrodotoxin was added to the extracellular solution containing the following: 140 mM NaCl, 2 mM KCl, 2 mM CaCl₂, 25 mM D-glucose, 10 mM HEPES, and 0.01 mM glycine (pH of 7.4 titrated with NaOH). The patch pipette solution was composed of the following: 135 mM CsCl, 10 mM HEPES, 1.2 mM MgCl₂, 4 mM ATP-Na₂, 0.5 mM CaCl₂, and 11 mM EGTA (pH of 7.3 with CsOH). Either 10 µM AMPA or 100 µM NMDA dissolved in extracellular solution was perfused onto cortical neurons by using a gravity-driven fast perfusion system, and 10 µM complestatin was pressure-ejected onto the neurons with Picospritzer II (General Valve, Fairfield, NJ).

Deprivation of Oxygen and Glucose. Cortical cell cultures (DIV 16-17) were transferred to an anaerobic chamber containing CO₂ (5%), H₂ (10%), and N₂ (85%). Oxygen and glucose deprivation was initiated by replacing the culture medium with a glucose-free, deoxygenated balanced salt solution containing the following: 143.6 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl₂, 0.8 mM MgSO₄, 1 mM NaH₂PO₄, 26.2 mM NaHCO₃, and 10 mg/l phenol red. To terminate the injury, the oxygen-glucose-deprived cultures were removed from the anaerobic chamber, concentrated glucose (final glucose concentration being 5.5 mM) was added, and the cultures were then placed in the aerobic CO₂ incubator.

Retinal Ischemia. Male rats (Sprague-Dawley, 180-220 g) were anesthetized by chloral hydrate (400 mg/kg i.p.). To induce retinal ischemia, the intraocular pressure was increased to a range of 160 to 180 mm Hg for 90 min by inserting a 301/2-gauge needle into the anterior chamber (Joo et al., 1999). Vehicle solution or 20 to 100 µM complestatin dissolved in 10 µl of saline solution was injected into the eye through a Hamilton syringe inserted into the vitreous chamber 15 min before ischemic injury was induced. Rats were euthanized 24 h after reperfusion. The eyes were removed, fixed in 4% glutaraldehyde, sectioned at 1 µm thickness, and stained with 1% toluidine blue.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Complestatin Blocks NMDA, AMPA, and Kainate Neurotoxicity in Noncompetitive Manner. The purified complestatin is a bicyclo-hexapeptide consisting of aromatic amino acids such as tryptophan, D()-4-hydroxyphenylglycine, and D()-3,5-dichloro-4-hydroxyphenylglycine (Fig. 1A). Exposure of cortical cell cultures to 20 µM NMDA or 40 µM kainate resulted in a rapid swelling of the neuronal cell body within 2 h (Fig. 1, B and C). This swelling was blocked by the inclusion of 3 µM complestatin. Exposure of cortical cell cultures to 20 µM NMDA, 40 µM kainate, or 10 µM AMPA caused 90 to 100% neuronal death over the next day (Fig. 1D). These excitotoxic neuronal deaths were prevented by inclusion of 3 to 10 µM complestatin. Increasing the doses of excitotoxins up to 1 mM did not abrogate the neuroprotective effects of 3 µM complestatin (Fig. 2), suggesting that complestatin prevented excitotoxicity in a noncompetitive manner.

View larger version (36K): [in this window] [in a new window]   Fig. 1.   Complestatin blocks NMDA, AMPA, and kainate neurotoxicity in cortical cell cultures. A, structure of complestatin. B and C, phase contrast photomicrographs of cortical cultures (DIV 12) at 2 h after exposure to 20 µM NMDA, alone (B) or with 3 µM complestatin (C). Scale bar, 30 µm. D, cortical cultures (DIV 12-14) were exposed to 20 µM NMDA, 40 µM kainate, or 10 µM AMPA for 24 h, alone or in the presence of 1 to 10 µM complestatin. Neuronal death was assessed by measuring the LDH efflux into the bathing medium [n = 12 culture wells/condition, scaled to the mean LDH value released after 24-h exposure to 500 µM NMDA (=100)].

View larger version (10K): [in this window] [in a new window]   Fig. 2.   Noncompetitive blockade of excitotoxicity by complestatin. Cortical cultures (DIV 12-14) were exposed continuously to 10 to 1000 µM concentrations of NMDA (A), kainate (B), or AMPA (C), alone () or with 10 µM complestatin (). Neuronal death was analyzed 24 h later (n = 12 culture wells/condition). *, significant difference from the relevant control group (NMDA, AMPA, or kainate alone) at P < 0.05 by using an analysis of variance and Student-Newman-Keuls test.

Complestatin Attenuates neither Oxidative Stress nor Apoptosis. Additional experiments were performed to see whether complestatin would influence the oxidative stress that is known, to some extent, to mediate excitotoxic process (Dykens, 1994; Dugan et al., 1995). Cortical cell cultures exposed to 50 µM Fe²⁺, 100 µM H₂O₂, or 10 mM buthionine sulfoximine underwent 50 to 100% neuronal death (Table 1), which was prevented by addition of antioxidants (data not shown). This free radical-mediated neurotoxicity was not altered in the presence of 3 µM complestatin. Although apoptosis could occur subsequent to the process of excitotoxicity (Ankarcrona et al., 1995), complestatin was not neuroprotective against apoptosis induced in cortical cell cultures exposed to 100 nM staurosporine or 20 µM cyclosporine A (Table 1) (Koh et al., 1995; McDonald et al., 1996). Therefore, complestatin appeared to prevent excitotoxicity independently of oxidative stress and apoptosis.

Complestatin Interferes with Entry and Accumulation of Calcium by Excitotoxins. Because an accumulation of [Ca²⁺]_i is essential for the initiation of excitotoxic neuronal death (Choi, 1987), the possibility that complestatin would influence [Ca²⁺]_i, accumulation in cortical neurons after exposure to excitotoxins was examined. Neuronal [Ca²⁺]_i was gradually increased, 5 min after the exposure of cortical cell cultures to 100 µM NMDA (Fig. 3A). This NMDA-induced accumulation of [Ca²⁺]_i was markedly attenuated with the concurrent inclusion of 10 µM complestatin. Accumulation of [Ca²⁺]_i was also observed in cortical neurons exposed to 10 µM AMPA in the presence of 100 µM cyclothiazide, which is known to inhibit desensitization of AMPA currents and increase Ca²⁺ influx through AMPA receptor (David et al., 1996; Yamada and Turetsky, 1996). The addition of complestatin reduced the [Ca²⁺]_i accumulation after treatment with AMPA and cyclothiazide. The effects of complestatin against [Ca²⁺]_i accumulation appeared to be specific to NMDA and AMPA receptors, because complestatin did not interfere with depolarization-induced [Ca²⁺]_i accumulation, after exposure to 25 mM KCl (Fig. 3A). These results imply that complestatin most likely attenuates the accumulation of [Ca²⁺]_i by interfering with the influx of Ca²⁺ ions after activation of the NMDA and AMPA receptors. This is further supported by the observation that addition of 10 µM complestatin prevented influx of Ca²⁺ ions after 5-min exposure of cortical cell cultures to 100 µM NMDA or 10 µM AMPA plus 100 µM cyclothiazide (Fig. 3B).

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Complestatin prevents NMDA- or AMPA/cyclothiazide-induced accumulation and influx of Ca2+ in cortical neurons. A, analysis of [Ca2+]i by using fura-2 fluorescence signal after exposure of cortical cell cultures (DIV 12) to 100 µM NMDA, 10 µM AMPA plus 100 µM cyclothiazide (AMPA/CTZ), or 25 mM KCl, alone or with 10 µM complestatin (COMP). Neuronal [Ca2+]i was measured 5 min later (for NMDA and AMPA/CTZ) and immediately after exposure (for KCl) (n = 50-60 neurons, randomly chosen from four culture wells for each condition). B, cortical cultures (DIV 12) were exposed to 100 µM NMDA or 10 µM AMPA plus 100 µM cyclothiazide (AMPA/CTZ), alone () or with 10 µM complestatin (). The Ca2+ influx was analyzed 0, 5, or 20 min later by measuring 45Ca2+ entry (n = 4 culture wells/condition). *, significant difference from relevant control group (NMDA, AMPA/CTZ, or KCl alone) at P < 0.05 by using analysis of variance and Student-Newman-Keuls test.

Complestatin Prevents NMDA- and AMPA-Induced Inward Currents. The ability of complestatin to prevent Ca²⁺ ion influx and accumulation after activation of NMDA and AMPA receptors may be related to the direct modulation of the ionotropic glutamate receptors. Complestatin itself did not elicit any current in voltage-clamped cortical neurons. Complestatin almost completely prevented an inward current, produced immediately after application of 100 µM NMDA (Fig. 4A). Complestatin also blocked an inward current induced in cortical neurons after exposure to 10 µM AMPA (Fig. 4B). The blocking effects of complestatin against NMDA- and AMPA-induced responses were rapid and reversible.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Complestatin inhibits NMDA- and AMPA-induced inward currents. A and B, whole cell currents were elicited from cortical neurons at a holding potential of 70 mV after exposure to 100 µM NMDA (A1) and 10 µM AMPA (B1). These neurons then received a pressure-ejected application of 10 µM complestatin for the indicated times. Effects of complestatin against NMDA (A2) and AMPA (B2) currents were averaged (n = 7-15 neurons/condition). Note that complestatin reversibly inhibits the NMDA and AMPA currents. *P < 0.05 by using Student's t test.

Complestatin Protects Neurons from Prolonged Deprivation of Oxygen and Glucose. We next examined whether the neuroprotective effects of complestatin could be extended against deprivation of oxygen and glucose. Cortical neurons deprived of oxygen and glucose underwent necrotic degeneration, evident by a swelling of the neuronal cell body (Fig. 5A). Complestatin as well as a mixture of MK-801 and CNQX attenuated both the (rapidly appearing) swelling of the neuronal cell body, and neuronal death over 24 h after a deprivation of oxygen and glucose for 60 min (Fig. 5, B and C). Interestingly, the neuroprotective effects of MK-801 and CNQX are markedly reduced in cortical cell cultures that are deprived of oxygen and glucose for 120 min. Gwag et al. (1995) reported that such prolonged oxygen-glucose deprivation to neuronal cultures in the presence of AMDA and AMPA/kainate antagonists revealed neuronal apoptosis (Gwag et al., 1995). Moreover, cortical neurons continuously treated with MK-801 and CNQX alone revealed a shrinkage of the cell body, vacuolation, and widespread degeneration within 24 to 36 h (Fig. 5D). In contrast, inclusion of 10 µM complestatin produced little neurotoxicity by itself and protected neurons after 120 min of oxygen and glucose deprivation (Fig. 5, C and D). The protective effects of complestatin were confined to neurons without beneficial effects against degeneration of glial cells deprived of oxygen and glucose (data not shown). These results imply that complestatin prevents the deprivation-induced neuronal death at subtoxic doses.

View larger version (26K): [in this window] [in a new window]   Fig. 5.   Complestatin protects cortical neurons from prolonged deprivation of oxygen and glucose without producing neurotoxicity. A and B, phase contrast photomicrographs of cortical neurons (DIV 17) deprived of oxygen and glucose for 120 min, alone (A) or with 3 µM complestatin (B). Scale bar, 30 µm. C, cortical cell cultures (DIV 17) were deprived of oxygen and glucose for 60 (OGD 60 min) or 120 min (OGD 120 min), alone or in the presence of 10 µM MK-801 plus 50 µM CNQX (MK/CNQX) or 3 µM complestatin (COMP). Neuronal death was analyzed 24 h later by measuring LDH efflux into the bathing medium (n = 8-12 culture wells/condition). *, significant difference from relevant control group (OGD alone) at P < 0.05 by using analysis of variance and Student-Newman-Keuls test. #, significant difference between MK/CNQX and COMP at P < 0.05 by using independent sample's t test. D, cortical cell cultures (DIV 17) were exposed to 10 µM MK-801 plus 50 µM CNQX (MK/CNQX) or 3 µM complestatin (COMP). Cell survival was analyzed 24 or 36 h later by counting viable neurons (n = 16 fields randomly chosen from four culture wells/condition). *, significant difference between MK/CNQX and COMP treatment at P < 0.05 by using Student's t test.

Complestatin Prevents Ischemic Neuronal Death in Vivo. Finally, we examined effects of complestatin against ischemic injuries in the retina that were known to cause an increase in glutamate release and excitotoxic neuronal death (Yoon and Marmor, 1989; Joo et al., 1999). The pressure-induced retinal ischemia produced approximately 50 and 60% neuronal degeneration in the inner nuclear and ganglion cell layers within 24 h, respectively (Fig. 6, A-C). The intravitreal injections of 20 to 100 µM complestatin attenuated the ischemic neuronal degeneration in a dose-dependent manner, up to 90% of the sham-operated control. Complestatin alone did not produce detectable toxicity in the retina (data not shown).

View larger version (64K): [in this window] [in a new window]   Fig. 6.   Complestatin attenuates ischemic neuronal death in retina. A to C, bright field photomicrographs of retinal sections stained with toluidine blue 24 h after a sham operation (A) or ischemia for 90 min, alone (B) or with the intravitreal injections of 100 µM complestatin (C). Arrows indicate degenerating neurons. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bar, 50 µm. D, animals received a sham operation (control) or 90 min of ischemia, alone or with the intravitreal injections of 20 to 100 µM complestatin. Neuronal death in GCL and INL was analyzed 24 h later by counting viable neurons within a 100 × 25-mm square overlying GCL and INL [n = 10 squares randomly chosen from each rat (5 rats/condition)]. *, significant difference from the relevant control (ischemia alone) at P < 0.05 by using an analysis of variance and Student-Newman-Keuls test.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Excitotoxicity is the foremost cause of fulminant brain damage after hypoxic-ischemia and other acute brain injuries. Here, we report a novel neuroprotective action of complestatin against NMDA- and AMPA/kainate-mediated neurotoxicity. Complestatin prevents excitotoxicity by blocking ionic influx through NMDA and AMPA/kainate receptors. Unlike neurotoxicity induced by glutamate antagonists MK-801 and CNQX, prolonged exposure to complestatin produced little neuronal loss. This neuroprotective action of complestatin has been extended to ischemic insults in vitro and in vivo.

Complestatin showed notable features of neuroprotection against excitotoxicity. This peptide isolated from Streptomyces species near completely blocked both NMDA- and AMPA/kainate-induced neuronal death at doses as low as 3 to 10 µM. Although kynurenate and 6,7-dichloro-3-hydroxy-2-quinoxalinecarboxylate have been reported to reduce NMDA responses through the strychnine-insensitive glycine site and AMPA/kainate responses in a competitive manner (Birch et al., 1988, 1989), complestatin is the first peptide that noncompetitively blocks both NMDA and AMPA/kainate neurotoxicity.

Superoxide and hydroxyl radicals are produced after activation of NMDA and AMPA/kainate receptors and have been proposed to account for the mediation of excitotoxicity in cortical, hippocampal, and cerebellar neurons (Lafon-Cazal et al., 1993; Dugan et al., 1995; Ciani et al., 1996; Sengpiel et al., 1998). However, complestatin did not attenuate neurotoxicity induced by pro-oxidants that produce superoxide or hydroxyl radicals. Complestatin should block excitotoxicity, apart from apoptosis, because necrosis is a predominant pattern of excitotoxicity in cortical neurons (Gwag et al., 1997; Tenneti and Lipton, 2000); and complestatin did not attenuate neuronal apoptosis induced by staurosporine or cyclosporine A. We observed that complestatin blocked swelling of neuronal cell body rapidly, evolving within a few minutes after administration of NMDA, AMPA, or kainate, and thus reasoned that complestatin would block intracellular accumulation of Ca²⁺ and Na⁺ after activation of the ionotropic glutamate receptors.

It has been well documented that deregulation of Ca²⁺ homeostasis in the early phase causes delayed cell death by the production of free radicals and the activation of enzymes that degrade lipid, nucleic acid, and proteins (Siesjo et al., 1989; Orrenius and Nicotera, 1994). The influx and accumulation of Ca²⁺ ions constitute an essential component of excitotoxicity resulting from excess activation of Ca²⁺-permeable glutamate receptors (Choi, 1987). Complestatin blocked the influx and accumulation of Ca²⁺ after exposures to NMDA or AMPA plus cyclothiazide without influencing depolarization-induced Ca²⁺ change. Analysis of patch-clamp recordings in cortical neurons showed that complestatin reversibly blocked inward currents through NMDA and AMPA receptors, suggesting that complestatin antagonizes excitatory responses by interfering with the influx of Na⁺ ions as well as Ca²⁺ ions.

It is conceivable that complestatin may act on the regulatory sites of NMDA and AMPA/kainate receptors. However, inclusion of 3 to 10 µM complestatin did not significantly compete with the binding sites of NMDA, 1-(1-phenylcyclohexyl)piperidine (phencyclidine), glycine, polyamine, AMPA, and kainate in the membrane fraction from rat cortical tissues (S. Y. Seo and B. J. Gwag, unpublished data). These results raise an intriguing possibility that complestatin may block excitotoxicity through novel regulatory site(s) in the NMDA and AMPA/kainate receptor complex. Alternatively, complestatin may modulate redox state, phosphorylation, and coupling to PSD-95 and SAP102 of NMDA or AMPA/kainate receptors that influence excitatory synaptic transmission (Wang et al., 1991; Sullivan et al., 1994; Lau et al., 1996; Sattler et al., 1999).

Complestatin possesses unique structural and pharmacological properties for preventing excitotoxicity. Complestatin, a bicyclo-hexapeptide consisting of aromatic amino acids such as hydroxyphenylglycine and tryptophan, blocks NMDA and AMPA/kainate receptors in a noncompetitive and reversible manner. Phenylglycine derivatives have been developed as selective agonists and antagonists of metabotropic glutamate receptors (Watkins and Collingridge, 1994). Among these, (S)-4-carboxy-3-hydroxyphenylglycine, a potent competitive antagonist of metabotropic glutamate receptor 1, showed neuroprotective effects against ischemic injuries through selective attenuation of NMDA neurotoxicity (Buisson and Choi, 1995; Rauca et al., 1998). We found that neither (R)-4-hydroxy-phenylglycine nor tryptophan attenuated excitotoxicity (B. J. Gwag, unpublished data). Although further study is needed to identify those structural components of complestatin that block excitotoxicity, the bicyclic structure containing (R)-4-hydroxy-phenylglycine and 3,5-dichloro-4-hydroxy-phenylglycine may constitute a key component underlying the dual antagonistic action of complestatin against NMDA and AMPA/kainate receptors.

Antagonists of ionotropic glutamate receptors have been developed and are reported to reduce brain injury from hypoxic ischemia. In particular, NMDA antagonists show preferential neuroprotection against focal cerebral ischemia. The therapeutic potential of NMDA antagonists should be compromised due to neurotoxicity that would appear in various areas of the brain after administration of NMDA antagonists (Olney et al., 1991). Moreover, AMPA/kainate receptors mediate delayed neuronal death in the hippocampal formation after transient forebrain ischemia and degeneration of oligodendrocytes and astrocytes that are also vulnerable to ischemic insults (Sheardown et al., 1990; David et al., 1996; McDonald et al., 1998). This implies that NMDA-induced and AMPA/kainate-induced excitotoxic responses should be controlled together for the efficient intervention of ischemic brain damage. This is supported by the present findings that complestatin protected cortical neurons from prolonged deprivation of oxygen and glucose as did MK-801 and CNQX (Kaku et al., 1991; Gwag et al., 1995). In contrast to the NMDA and AMPA/kainate antagonists that produced severe neuronal death, cortical neurons were spared by a continuous exposure to complestatin.

We examined the neuroprotective action of complestatin against ischemic injury in retina in vivo that was shown to produce neuronal death primarily through activation of ionotropic glutamate receptors (Mosinger et al., 1991). The intravitreal administration of 0.65 to 1.3 µg of complestatin markedly reduced ischemic neuronal death in retina. The infarct volume 24 h after occlusion of middle cerebral artery was also reduced in adult rats that received injections of 8 to 10 µg of complestatin in the lateral ventricle at the beginning of occlusion (B. J. Gwag, unpublished data). This suggests that the neuroprotective action of complestatin can be applied to treat hypoxic-ischemic neuronal death. Because approximately 100 to 120 mg of complestatin is purified from the fermentation broth (10 liters) of Streptomyces species 60910, further study will be needed to purify on a large scale or synthesize complestatin and then to verify the therapeutic efficacy of systemically administered complestatin against ischemic injuries in brain.

In conclusion, complestatin securely blocks both NMDA and AMPA/kainate neurotoxicity in a noncompetitive and reversible manner. This unique property of complestatin holds significant potential for efficiently intervening in excitotoxicity, a leading cause of neuronal death in hypoxic ischemia, epilepsy, and chronic neurodegenerative diseases.