Neuroprotection and Neurotoxicity Studies in Rats with AHN649

Home , Enadoline

0022-3565/99/2911-0399$03.00/0 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS Vol. 291, No. 1 Copyright © 1999 by The American Society for Pharmacology and Experimental Therapeutics Printed in U.S.A. JPET 291:399–408, 1999

Neuroprotection (Focal Ischemia) and Neurotoxicity (Electroencephalographic) Studies in Rats with AHN649, a 3-Amino Analog of Dextromethorphan and Low-Affinity 1 N-Methyl-D-Aspartate Antagonist

FRANK C. TORTELLA, PAUL BRITTON, ANTHONY WILLIAMS, XI CHUNG MAY LU, and AMY H. NEWMAN Department of Neuropharmacology and Molecular Biology, Walter Reed Army Institute of Research, Washington, D.C. (F.C.T., P.B., A.W., X.C.M.L.); and Drug Development Group, National Institute for Drug Abuse Addiction Research Center, Baltimore, Maryland (A.H.N.) Accepted for publication June 16, 1999 This paper is available online at http://www.jpet.org

ABSTRACT AHN649, an analog of dextromethorphan (DM) and a relatively Although impaired neurological function scores improved in all selective low-affinity N-methyl-D-aspartate antagonist, was groups by 24 to 72 h, the most dramatic improvement was evaluated for neuroprotective effects using the rat intraluminal associated with AHN649 treatments. In a rat electroencepha- filament model of temporary middle cerebral artery occlusion. lographic model of brain function, separate neurotoxicity ex- Rats were subjected to2hoffocal ischemia followed by 72 h periments revealed that acute i.v. doses of DM caused seizures ϭ ϭ of reperfusion. In vehicle-treated rats, middle cerebral artery (ED50 19 mg/kg) and death (LD50 27 mg/kg). In contrast, occlusion resulted in neurological deficits and severe infarction AHN649 failed to induce seizure activity at doses up to 100 Ϯ 3 ϭ measuring 232 25 mm , representing approximately 25% mg/kg (LD50 79 mg/kg). Collectively, AHN649 is described as contralateral hemispheric infarction. Post-treatment with a potent, efficacious neuroprotective agent devoid of serious AHN649 (0.156–20 mg/kg i.v.) or DM (0.156–10 mg/kg i.v.) central nervous system neurotoxicity and possessing potential significantly reduced cortical infarct volume by 40 to 60% com- therapeutic value as antistroke treatment. Furthermore, the fea- pared with vehicle-control treatments. AHN649 neuroprotec- sibility of targeting low-affinity N-methyl-D-aspartate-site li- ϭ tion was linear and dose dependent (ED50 0.80 mg/kg), gands as postinjury therapy for ischemic brain injury has been ϭ whereas DM neuroprotection (ED50 1.25 mg/kg) was nonlin- confirmed. ear and less effective at the higher doses (2.5–10 mg/kg).

Potent anticonvulsant and neuroprotective actions associ- at the molecular, receptor, and/or ion channel level continue ated with high-affinity blockers of glutamate receptors, espe- to be explored for therapeutic implications (Szatkowski and cially those coupled to the noncompetitive N-methyl-D-aspar- Attwell, 1994; Muir and Lees, 1995; Whetsell, 1996). tate (NMDA)/phencyclidine (PCP) recognition site of the Consistent with excitatory amino acid mechanisms under- channel complex, have been firmly established (Rogawski, lying clinical episodes of severe pathological depolarization 1992; Muir and Lees, 1995). However, increasing awareness or impaired metabolism (i.e., cellular ischemia), low-affinity of the potential neurotoxicity associated with these ligands, NMDA- and/or PCP-site ligands have been synthesized and including neuropathology, motor dysfunction, seizures, and developed with the intent of providing therapeutic efficacy possible hallucinogenic and/or psychotomimetic actions, may with limited toxicity (Rogawski, 1992, 1993; Cregan et al., limit their development as clinical drugs. Because hyperac- 1997). In principle, a relatively selective low-affinity NMDA/ tivity of glutamate synapses remains one of the most signif- PCP ligand would 1) retain significant anticonvulsant or icant pathological consequences of cerebral ischemia, mech- neuroprotective properties by virtue of use-dependent/non- anisms targeting attenuation of glutamatergic excitotoxicity competitive and/or voltage-dependent mechanisms without 2) the unwanted side effects associated with potent activa- Received for publication May 3, 1999. tion of the high-affinity NMDA/PCP receptor complex. PCP 1 The views of the authors do not purport to reflect the position of the analogs have been synthesized that fit this receptor affinity Department of the Army or the Department of Defense (para 4-3, AR 360-5). Preliminary versions of these data were published in abstract form in Phar- and pharmacodynamic profile (Rogawski, 1992), and non- macologist [(1997) 39:84] and Neurosci Abstr [(1997) 23:1946]. PCP compounds possessing micromolar channel-blocking af-

ABBREVIATIONS: NMDA, N-methyl-D-aspartate; MCAo, middle cerebral artery occlusion; EEG, electroencephalogram; NS, neurological scores; AHN649, 3-amino-17-methyl morphinan; DM, dextromethorphan; PCP, phencyclidine; SWS, slow-wave sleep; TTC, 2,3,5-triphenyltetrazolium chloride; TI, therapeutic index. 399 400 Tortella et al. Vol. 291 finity at the noncompetitive NMDA/PCP-binding site have Materials and Methods also been demonstrated as safe and effective anticonvulsant and neuroprotective agents in animal studies (Rogawski, Animal Preparation. For MCAo experiments, 77 male Sprague- Dawley rats (270–330 g; Charles River Labs, Raleigh, VA) were 1993; Palmer et al., 1995; Cregan et al., 1997). implanted with a chronic indwelling jugular vein catheter and pre- We recently described the anticonvulsant and neuroprotec- pared for temporary focal ischemia using the filament method of tive properties of an analog of dextromethorphan (DM), MCAo and reperfusion (Britton et al., 1997). Anesthesia was induced namely AHN649 (3-amino-17-methyl morphinan), which is a with 5% halothane and maintained with 2% halothane delivered in relatively selective ligand for low-affinity NMDA/PCP bind- oxygen. Body temperature was maintained normothermic (37 Ϯ 1°C) ϭ ␮ ing sites (Ki 9.9 M; Newman et al., 1996). In rat models of throughout all surgical procedures by means of a homeothermic experimental epilepsy, including maximal electroshock-, flu- heating system (Harvard Apparatus, South Natick, MA). Briefly, the rothyl-, and NMDA-induced convulsions, AHN649 is a potent right external carotid artery was exposed, and its branches were anticonvulsant devoid of neurobehavioral toxicity (Tortella et coagulated. A 3-0 uncoated monofilament nylon suture with rounded tip was introduced into the internal carotid artery via an arteriotomy al., 1994). In subsequent studies, AHN649 has been shown to in the external carotid artery and advanced (approximately 22 mm prevent glutamate-induced cell death and the associated from the carotid bifurcation) until it lodged in the proximal region of changes in excitotoxic intracellular calcium dynamics in neu- the anterior cerebral artery, thus occluding the origin of the MCA. ronal cultures (Tortella et al., 1995) and to possess in vivo The endovascular suture remained in place for 2 h and then was neuroprotective efficacy similar to those of other NMDA an- retracted to allow reperfusion. After surgery, animals were placed in tagonists in a rat model of global forebrain ischemia (Tortella recovery cages with ambient temperature maintained at 22°C. Dur- et al., 1997a). ing the 2-h ischemia period and the initial 6-h postischemia period, The primary clinical target for an anti-ischemia neuropro- 75-W warming lamps were also positioned directly over the tops of tective drug is the treatment of stroke in humans. Regardless each cage to maintain body temperature normothermic throughout the experiment. [The facilities in which the animals were maintained of etiology, focal disturbance of blood flow and subsequent are fully accredited by the American Association for the Accredita- disruption of neuronal metabolism leading to cell death char- tion of Laboratory Animal Care (AAALAC). In conducting the re- acterize this ischemic insult. Because most clinical strokes search described in this report, the investigators adhered to the are acute focal ischemic episodes, animal models mimicking Guide for the Care and Use of Laboratory Animals, as promulgated focal ischemia to produce brain infarction, such as that oc- by the Committee on the Care and Use of Laboratory Animals of the curring after occlusion of the middle cerebral artery occlusion Institute of Laboratory Resources, National Research Council.] (MCAo), appear to have significant relevance (Hunter et al., For EEG experiments, 55 additional drug-naive male Sprague- 1995). The MCAo filament procedure was first introduced by Dawley rats (250–300 g) were anesthetized with ketamine HCl and xylazine (70 mg/kg and 6 mg/kg i.m., respectively) and surgically Koizumi et al. (1986) and Longa et al. (1989) as an extracra- prepared with a chronic indwelling jugular vein catheter and epi- nial “noninvasive” model of experimental stroke in rats. In dural stainless steel screw electrodes (0-80 ϫ 1/8 inches; Small Parts general, permanent MCAo produces a focal ischemic injury Inc., Miami Lakes, FL.) chronically implanted and affixed to the marked by a core area of hemispheric infarction localized skull with the use of dental acrylate cement (Tortella et al., 1997c). within the immediate borders of the MCA territory, namely The four recording electrodes were implanted bilaterally over the the striatum and overlying parietal/temporal cortical tissues. right and left frontal (3 mm anterior, 2 mm lateral to bregma) and A penumbral area of pathologically compromised neuronal parietal (3 mm posterior, 2 mm lateral to bregma) cortices. A fifth tissue surrounds the core infarction and is located more reference electrode was implanted over the occipital cortex. After distant from the MCA. The intraluminal filament model us- surgery, all animals were allowed 3 to 5 days for recovery before drug testing. During this time, they were provided food and water ad ing a 2-h transient occlusion reproduces this injury, yet per- libitum and allowed to acclimate to the test environment by being mits reflow of the MCA territory. This more closely models individually housed in clear Plexiglas recording chambers (30 ϫ 30 ϫ the clinical event of ischemic stroke and reflow that may 49 cm) under a 12-h light/dark cycle (lights on at 6:00 PM). occur in more than 80% of patients experiencing acute em- MCAo Experiments. Before MCAo surgery, the animals were bolic occlusion of the MCA trunk (Ringelstein et al., 1992). In randomly divided into three treatment groups: vehicle (n ϭ 14, addition, the incorporation of experimental reperfusion is distilled deionized water), AHN649 (0.156–20 mg/kg, n ϭ 7/dose), or critical because although primate studies have shown that DM (0.156–10 mg/kg, n ϭ 7/dose). Beginning at 30 min after occlu- cerebral reperfusion may be beneficial to neuronal survival sion and again at 1, 2, 4, 6, 24, and 48 h after injury, each rat received (Crowell et al., 1970), it has also been suggested that resto- an i.v. injection of the appropriate treatment. Before each injection, the body temperature was recorded, and a neurological examination ration of blood flow may actually worsen outcome by further was performed (see below). Importantly, so as not to compromise the promoting brain edema (Kuroiwa et al., 1988) and blood- results of the functional neurological examination at 2 h, it was brain barrier disruption (Yang et al., 1994) and physiologi- performed immediately before reperfusion and the 2-h injection. As cally promoting cellular mechanisms of delayed neuronal such, the actual 2-h injection was administered immediately after injury (Hallenbeck and Dutka, 1990; Halsey et al., 1991). recovery from the reperfusion surgical procedure (approximately Therefore, by experimentally mimicking this injury process, 125–130 min after injury). Food and water were provided ad libitum, one ensures a pathophysiological complexity consistent with and the animals were individually housed in clear Plexiglas cham- ϫ ϫ the observed clinical disorder. bers (30 30 49 cm) under a 12-h light/dark cycle (lights on at 6 In the present study, we examined AHN649 (in comparison PM). After 70 h of reperfusion/recovery, the rats were euthanized by decapitation, and their brains were removed for quantification of with DM) in a rat model of focal cerebral ischemia and infarction. reperfusion and conducted a comprehensive dose-effect anal- Infarct Analysis. From each rat brain, seven coronal sections (2 ysis of functional neurotoxicity using spontaneous cortical mm thick) were taken from the region beginning 1 mm from the electroencephalographic (EEG) activity in unanesthetized frontal pole and ending just rostral to the corticocerebellar junction. rats. Analysis of ischemic cerebral damage, including total core infarct 1999 AHN649 Neuroprotection and EEG Studies in Rats 401 volume and hemispheric infarct size, was achieved using the 2,3,5- and recorded on the EEG polygraph records as a correlate to their triphenyltetrazolium chloride (TTC) staining method and computer- respective changes in EEG activity. assisted image analysis as described in detail elsewhere (Britton et Data Analysis. Data are presented as the mean Ϯ S.E. For the al., 1997). Briefly, the posterior surface of each TTC-stained fore- MCAo infarct volume and neuroprotection results, statistical analy- brain section (and the anterior surface of the first section) was sis were made using an ANOVA and post hoc Dunnett’s test for digitally imaged (Loats Associates, Westminster, MD). Infarction multiple comparisons. Regression analysis and potency comparisons and hemispheric areas (mm2) were quantified to provide subsequent were performed on the percentage of neuroprotection dose-response determinations of the various infarct characteristics, including: 1) data. For EEG results, the Student’s t test for paired data was used 3 total volume (mm ) of core injury, 2) the percentage of volume of where appropriate. ED50 (the dose required to produce a response in cerebral tissue representing infarction (percentage of hemispheric 50% of the animals tested per group) and LD50 (the dose required to infarction), and 3) hemispheric swelling. Total core infarct volume, produce death in 50% of the animals tested per group) values were where core injury was defined as brain tissue completely lacking derived from quantal dose-response data. For the neuroprotection

TTC staining, was calculated by sequential integration of the respec- ED50, the criteria for identifying a positive responder were defined tive surface areas. Similarly, ipsilateral and contralateral hemi- empirically as a drug-treated MCAo animal whose measured infarct spheric volumes were calculated to account for the contributing effect volume was less than 140 mm3 or the mean infarct volume minus 1 of brain swelling to infarct volume (Barone et al., 1993) where hemi- S.D. for the vehicle control group (or 232 mm3 Ϫ 92 mm3, respec- spheric swelling representing tissue edema was expressed as the tively, in the present study). For the seizure ED50, an animal was percentage of increase in the size of the ipsilateral (occluded) hemi- considered a responder if its postdrug EEG contained ictal activity, sphere compared with the contralateral (uninjured) hemisphere regardless of the pattern or severity. All statistical tests were per- (Britton et al., 1997). formed using the Pharmacological Calculations Computer Programs Neurological Examination. A neurological examination was described by Tallarida and Murray (1986). performed on each rat immediately before MCAo surgery and again Compounds. AHN649 was synthesized at the Walter Reed Army before each injection using a modification of the procedure described Institute of Research (Division of Experimental Therapeutics) as by Bederson et al. (1986). Briefly, neurological scores (NS) were previously described (Newman et al., 1992), and DM was purchased derived using a 10-point sliding scale. Each animal was examined for from Research Biochemicals Inc. (Natick, MA). All compounds were reduced resistance to lateral push (score, 4), open field circling (score, dissolved in distilled deionized water immediately before testing and 3), and shoulder adduction (score, 2) or contralateral forelimb flexion administered as a 1 ml/kg volume. All injections were administered (score, 1) when held by the tail and suspended approximately 0.5 m as a 1-min i.v. infusion without handling or disturbing normal ani- above the floor. Rats extending both forelimbs toward the floor and mal behavior. not showing any other signs of neurological impairment were scored 0. Using this procedure, maximal neurological severity was mea- Results sured as NS of 10. In the present study, all vehicle-treated rats subjected to MCAo exhibited a neurological score of 10 when exam- In control vehicle-treated rats,2hofMCAo followed by ined 2 h after ischemia or immediately before reperfusion (see Fig. 6). 70 h of reperfusion resulted in core infarction of the ipsilat- EEG Experiments. The recording chambers described above eral temporal/parietal cortex and underlying striatal tissue were equipped with custom-designed multichannel mercury swivel (Fig. 1). Significant forebrain infarction was measured from commutators (Dragonfly Inc., Silver Spring, MD). On the morning of approximately 3 to 13 mm from the frontal pole (Fig. 2). Total the EEG experiment, the rats were connected to the swivel system by infarct volume in control rats was 232 Ϯ 25 mm3 (Fig. 3), flexible shielded cables providing a noise-free connection from the representing approximately 25% contralateral hemispheric unrestrained rat to a Grass model 7D polygraph while permitting infarction (Fig. 5B). MCAo resulted in significant hemi- freedom of movement by the animals during all phases of the experiment. Control EEG recordings were collected such that drug testing spheric edema representing an approximately 8% increase in was initiated only after each subject exhibited normal behavioral cerebral volume compared with the contralateral, uninjured and EEG slow-wave sleep (SWS) patterns. All EEG experiments hemisphere (Fig. 5A). All vehicle-treated rats subjected to routinely began between 9:00 and 11:00 AM and were considered MCAo exhibited a neurological score of 10 when examined complete on the reemergence of SWS. Using this protocol, each 2 h after ischemia (see Fig. 6). Although neurological function animal served as its own control. Throughout these experiments, all was severely impaired (NS ϭ 10 Ϯ 0), by 72 h postinjury, a rats were drug naive and used only once. After the onset of normal significant degree of spontaneous recovery was measured in SWS rhythms and behavior, the experiment was initiated by admin- the control animals (NS ϭ 3.8 Ϯ 0.4). However, none of the istering an i.v. injection of the control vehicle (1 ml/kg distilled injured, vehicle-treated animals completely recovered neuro- deionized water). With the reemergence of EEG SWS after the con- ϭ ϭ logical function (NS 0), with at least contralateral forelimb trol injections, the animals (n 5/dose) were injected i.v. with either ϭ AHN649 (10–80 mg/kg) or DM (12.5–31.25 mg/kg). For the duration flexion and shoulder adduction (NS 3) still evident in all of the EEG experiments, each animal was continuously observed by rats examined at 72 h after injury. two of the authors for signs of abnormal behavioral activity, includ- Regardless of the treatment group, all MCAo animals lost ing sedation; ataxia; enhanced locomotion; stereotypic activity in- approximately 15 to 20% b.wt. over the 72-h recovery period, cluding head-weaving, grooming, preening, and scratching; and and there were no significant differences in body weight loss signs of clonic (convulsant) muscle activity. The behavioral state of between groups. Similar to our earlier findings (Britton and each animal was defined subjectively by the clinical appearance, at Tortella, 1997), in the vehicle-control animals MCAo caused any time before the onset to SWS, of the following behaviors: 1) a transient, mild hyperthermia (38.3 Ϯ 0.2 and 37.6 Ϯ 0.3°C sedation, the absence of overt behaviors in an otherwise awake state at 2 and 4 h postocclusion, respectively) that returned to and distinguished from normal quiet awake behavior by the appear- normal (36.9 Ϯ 0.4°C preocclusion) by 4 to 6 h postocclusion ance of drug-induced, synchronized EEG slowing; 2) ataxia, un- Ϯ steady or impaired gate during walking/locomotion; 3) enhanced (36.9 0.3°C). Interestingly, at 48 to 72 h after injury, locomotion, running or increased sniffing and searching behavior; 4) control animals exhibited a mild and unexpected hypother- clonus, a rapid, repetitive movement or twitching of a muscle or mia (35.9 Ϯ 0.2 and 35.4 Ϯ 0.5°C, respectively). Only the muscle group; or 5) sleep, normal sleep posture, eyes closed, and EEG highest doses of AHN649 (10 and 20 mg/kg) significantly SWS patterns. The behavioral responses for each animal were noted blocked the MCAo-induced hyperthermia. At all other doses 402 Tortella et al. Vol. 291

Fig. 1. Computer-scanned images of TTC-stained coronal forebrain sections from representative vehicle or DM- (10 mg/kg) or AHN649- (0.625 mg/kg) treated rats subjected to2hoftransient MCAo. Histopathology was assessed 72 h after injury as described in the text. Top image in each column represents the posterior surface of the first forebrain section, which corresponds to the forebrain profile 3 mm from the frontal pole as seen in Fig. 2.

and timepoints, temperature measurements from AHN649- (Fig. 5A). As expected based on the infarct volume data, both or DM-treated animals were not significantly different from AHN649 and DM significantly reduced the overall percent- the corresponding control, vehicle-treated animals. age of hemispheric infarction (Fig. 5B). However, again, the AHN649 or DM post-treatment significantly reduced isch- effect of DM was significant at lower doses and nonlinear. emic infarction (Figs. 1 and 2). Both drugs were equipotent The results shown in Table 1 summarize the neuroprotec- (potency ratio ϭ 0.643, P ϭ NS; Table 1). However, AHN649 tion and lethality data obtained for each compound. Although was more efficacious than DM, decreasing infarct volume to the potency (i.e., ED50 value) is similar for both drugs, the 100 Ϯ 33 mm3 (Fig. 3) and resulting in a maximal neuropro- large CLs associated with DM reflect the nonlinearity of its tection of 57 Ϯ 14% at the highest dose tested (Fig. 4). At both effect to decrease cerebral infarction. A comparison of the the intermediate (0.625 mg/kg) and high (10 mg/kg) doses of respective LD50 values derived for AHN649 and DM from DM, the maximal decrease in infarct volume measured was normal, uninjured animals reveal a 2.9-fold difference be- 137 Ϯ 31 mm3 (Fig. 3), resulting in a maximal neuroprotec- tween the two drug groups (79 versus 27 mg/kg, respective- tion of only 41 Ϯ 13% (Fig. 4). Importantly, only the neuro- ly). Critically, the calculated therapeutic indexes (TI) for both protective effects of AHN649 were positively correlated with drugs indicate a wide margin of safety, and comparisons of dose (r ϭ 0.90, P Ͻ .026; Fig. 3). By comparison, the neuro- their respective TIs indicate that AHN649 is approximately protection dose-response effect of DM was significantly non- 4.5 times safer than DM. linear (r ϭ 0.55, P Ͻ .23; Fig. 3). Although both compounds Compared with vehicle-treated controls, neurological func- produced a trend toward reducing the amount of injury- tion significantly improved after AHN649 (Fig. 6A) or DM induced cerebral edema, these effects were not significant (Fig. 6B) treatment. Improvements in neurological function 1999 AHN649 Neuroprotection and EEG Studies in Rats 403

Fig. 2. Forebrain profiles of MCAo-induced cerebral damage in AHN649- (top) or DM- (bottom) treated rats at 72 h after injury. Each graph represents area (mm2) measurements of the anterior surface of the first forebrain section (1 mm from frontal pole) and the corresponding posterior sections of each subsequent forebrain surface (3–15 mm from the frontal pole). Data represent mean Ϯ S.E. Group sizes are 14 (vehicle) or 7 (all other dose groups).

TABLE 1

Neuroprotection ED50 and lethality LD50 values (95% CLs) and calculated TI

a b Neuroprotection ED50 Lethality LD50 Drug (95% CLs) (95% CLs) TI LD50/ED50

mg/kg AHN649 0.80 (0.15–4.07) 79 (62–100) 99 DM 1.25 (0.11–13.49) 27 (24–31) 22 a Dose producing neuroprotection (as defined in the text) in 50% of the rats subjected to MCAo. b Dose producing lethality in 50% of normal, drug-naı¨ve uninjured rats (derived from the EEG experiments).

injury with continued improvement measured at 48 h (not shown) and 72 h. At the highest doses tested, the NS for AHN649 (20 mg/kg) at 72 h was 0.7 Ϯ 0.4 compared with an NS of 3.8 Ϯ 0.4 for the corresponding controls. In contrast, an optimal improvement in DM-induced NS (2.0 Ϯ 0.6) was measured after the intermediate dose of DM (0.625 mg/kg), Fig. 3. Effect of vehicle, AHN649, or DM on infarct volume measured 72 h with a slight worsening in the NS at higher doses. Of the after injury. Data represent the mean Ϯ S.E. *P Ͻ .05, **P Ͻ .01, ϭ AHN649-treated rats, 20% completely recovered neurological significantly different from vehicle control, AHN649 (F5,41 3.44), and ϭ ϭ DM (F4,24 2.67). function (NS 0), whereas an additional 57% recovered to an NS of 1. In contrast, although neurological function recovered were time and dose dependent for both drugs. When evalu- to an NS of 1 in 52% of all DM-treated rats, no rats in this ated at comparable times after MCAo, drug-induced neuro- treatment group recovered complete neurological function logical improvements were first seen beginning 24 h after (similar to nontreated injured rats). 404 Tortella et al. Vol. 291

Fig. 4. Percentage of neuroprotection produced by ANH649 or DM. Data are normalized from Fig. 3 and represented as the mean Ϯ S.E. *P Ͻ .05, **P Ͻ .01 (statistical analysis derived from data in Fig. 3).

Results of the EEG and behavior studies are summarized in Table 2 and Figs. 7 to 9. AHN649 (Fig. 7A) produced moderate EEG slowing and sedative behavior but no evidence of EEG seizure activity. DM also induced similar EEG slowing and sedation in animals not exhibiting convulsant activity (Fig. 7B). However, in contrast to AHN649, injections of DM produced mild EEG seizure activity at doses as low as 12.5 mg/kg (not shown) and intense ictal activity character- ized by polyspike and spike-wave complexes consistently observed at higher doses (Fig. 8). Typically, the DM-induced EEG seizure activity emerged within minutes after injection and was accompanied by clonic movements followed by pos- tictal depression. From these results, the calculated safety index (ratio of the neuroprotection ED50 to the seizure ED50) was 15.2 for DM and more than 100 for AHN649. Impor- tantly, as shown in Fig. 9, lethality induced by either compound: 1) occurred within minutes of the injection and 2) was Fig. 5. Percentage of hemispheric edema (A) and percentage of hemi- not necessarily associated with brain seizures. spheric infarction (B) after vehicle, AHN649, or DM treatment in tran- Both AHN649 and DM dose dependently delayed the onset sient MCAo injured rats. Calculated as described in the text. Data rep- to behavioral and EEG SWS. Normal SWS latency after the resent the mean Ϯ S.E. *P Ͻ .05, **P Ͻ .01, significantly different from ϭ vehicle control, ANOVA/Dunnett’s. Edema: ANHN649, F5,41 2.70; DM, control i.v. vehicle injections ranged from 7 to 23 min across ϭ ϭ ϭ F4,34 1.25; hemispheric infarction: AHN649, F5,41 3.54; DM, F4,34 both treatment groups. In AHN649-treated rats, a significant 2.66. increase in SWS latency was not obtained until the 20 mg/kg dose (35.6 Ϯ 6.2 min; p Ͻ .05, paired t test) and was maximal at the 60 mg/kg dose (107 Ϯ 21.2 min). In DM-treated rats, effect of AHN649 was associated with significant improve- the latency to SWS was markedly increased (47.4 Ϯ 9.3 min; ments measured in postinjury neurological function and ap- P Ͻ .01, paired t test) even at the lowest dose tested (12.5 pears to be unrelated to any adverse cardiovascular or tem- mg/kg) and continued to increase to a maximum of 153 min perature effects. Furthermore, EEG studies in normal rats measured in the only animal surviving the highest dose failed to reveal signs of neurotoxic side effects. tested (31.25 mg/kg). An important feature of the injury model used in this study is its pharmacological sensitivity to neuroprotective agents. Discussion For example, in similar preclinical models of transient MCAo-induced cerebral injury, it has been demonstrated The present results describe a comprehensive, dose-re- that some (but not all) antioxidant treatments can reduce sponse characterization of the neuroprotective and neuro- infarction by 40% (Clemens and Panetta, 1994). Neuropro- toxic profiles of the DM analog AHN649. Similar to earlier tection treatments targeting upstream mechanisms (Lu et findings derived from a rat model of global forebrain injury al., 1998) or specific AMPA receptor mechanisms (Kawasaki- (Tortella et al., 1997a), we demonstrate here the ability of Yatsugi et al., 1998) of excitotoxic glutamatergic activity AHN649, administered i.v. as postinjury therapy, to signifi- have also been described to reduce cerebral damage caused cantly decrease brain infarction caused by experimental, fo- by transient MCAo. Not surprisingly, the high-affinity non- cal cerebral ischemia and reperfusion. The neuroprotective competitive NMDA antagonist MK801 also significantly re- 1999 AHN649 Neuroprotection and EEG Studies in Rats 405

Fig. 6. Effect of AHN649 (A) or DM (B) on neurological recovery (NS scores) after 72 h of transient MCAo. Data represent the mean Ϯ S.E. of four to seven rats per group. *P Ͻ .05, **P Ͻ .01, significantly different from corresponding vehicle control group at each time point: 2 h: ϭ ϭ ϭ AHN649, F5,33 1.14; DM, F4,31 3.26; 24 h: AHN649, F5,33 1.40; DM, ϭ ϭ ϭ F4,31 2.29; and 72 h: AHN649, F5,33 7.35; DM, F4,31 1.78.

TABLE 2

Sedation ED50 and EEG seizure ED50 values and 95% CLs

b Fig. 7. Representative effect of high doses of AHN649 (A) and DM (B) on ED50 (95% CLs) a spontaneous cortical EEG activity in noninjured, drug-naı¨ve rats. Verti- Drug n ␮ Sedation EEG Seizures cal calibration bar represents 100 V.

mg/kg AHN649 35 25 (13–45) No effect both compounds exhibited similar neuroprotection potencies, DM 25 30 (13–66) 19 (14–28) the efficacy of DM to decrease cerebral infarction (maximum a Total number of rats tested per drug group. of 41%) was more limited and likely compromised by neuro- b Dose producing sedation or EEG seizure activity in 50% of normal, drug-naı¨ve toxic actions resulting from brain seizures, a problem also uninjured rats. reported with MK801 (Tortella and Hill, 1996). By comparison, the neuroprotective efficacy of AHN649 does not appear duces infarction in rats or cats by approximately 30% (Dezsi to be compromised by behavioral (Tortella et al., 1994) or et al., 1992; Margaill et al., 1996), whereas the low-affinity seizurogenic (present study) neurotoxic side effects. Although noncompetitive NMDA antagonist ARL 15896AR [(S)-␣-phe- AHN649 produced mild sedation and related EEG slowing in nyl-2-pyridine-ethanamine dihydrochloride] has been shown normal rats, there was no EEG or behavioral evidence of to reduce infarct volume by approximately 23% (Cregan et ataxia, marked sedation, or severe EEG slowing. Critically, al., 1997). By comparison, our results describe a neuropro- there was no evidence of EEG seizures or behavioral convul- tective efficacy with AHN649 approaching a 60% reduction in sant activity, even at the highest (and lethal) doses tested infarct volume at dose levels correlating well with significant (i.e., 80 mg/kg). This indicates that the protective index for improvements in neurological recovery and devoid of EEG or AHN649, calculated as the ratio of the seizure ED50 dose Ͼ behavioral neurotoxic side effects. ( 80 mg/kg) to the neuroprotective ED50 dose (0.8 mg/kg), is Collectively, using both the MCAo and EEG models, exper- greater than 100. By comparison, the protective index for imental conditions were established permitting a direct, DM, although still considered relatively safe, is only 15.2. quantitative comparison of the neuroprotective, neurotoxic, Based on an identical rat EEG drug assessment model, the and lethal potencies between AHN649 and DM. Although seizure ED50 for MK801 was 0.13 mg/kg (Tortella et al., 406 Tortella et al. Vol. 291

determined to be approximately 3 times more lethal than AHN649 after systemic i.v. injections. Therefore, using the

respective LD50 and neuroprotection ED50 data for each compound, the respective TIs (see Table 1) indicate that the 3-amino substitution of DM yields a potential therapeutic drug with a margin of safety at least 5 times greater than its parent molecule. It is highly unlikely that the neuroprotective effect of AHN649 or DM in cerebral ischemia can be attributable to indirect effects on key physiological parameters such as blood pressure, heart rate, blood pH or gases, or brain/body temperature. In previous studies (Britton et al., 1997; Tortella et al., 1997b), it has been shown that doses of DM or AHN649 in the range of those used in the present study had no effect on these physiological variables in normal rats. Furthermore, in the present study, careful monitoring of rectal “core” body temperature, which has been reported in similar ischemia studies to closely reflect brain temperature (Xue et al., 1992; Zhang et al., 1994), failed to reveal a drug-induced hypothermia associated with the decrease in brain infarction. It is also unlikely that the neuroprotective effects of AHN649 were due to general barbiturate-like sedative effect on brain metabolism. Profound cortical EEG slowing may reflect a functional state of depressed cerebral metabolic Fig. 8. Representative effect of DM to induce cortical EEG seizure activ- activity. By decreasing the metabolic demands of neuronal ity in noninjured, drug-naı¨ve rats. Vertical calibration bar represents 100 tissue during periods of severe brain ischemia, such as one ␮V. might achieve during a drug-induced coma, preservation of neuronal tissue may be improved. However, in the case of AHN649, it was determined that even at doses several times greater than those required for neuroprotection, there was no EEG evidence of severe cortical slowing. This contrasts with EEG results described elsewhere for high-affinity noncompetitive NMDA antagonists such as MK801 or PCP in which marked cortical EEG slowing can be seen at neuroprotective doses (Marquis et al., 1989; Tortella and Hill, 1997). Another important finding of this study was the absence of brain neurotoxicity seen with AHN649 as measured by EEG seizures. It has been reported for MK801 and other high- affinity PCP and/or ␴ ligands that relatively high i.v. doses in rats may be associated with EEG brain seizures and, in some cases, the emergence of proconvulsant properties leading to clinical signs of behavioral convulsant activity (Marquis et al., 1989; Echevarria et al., 1990; Tortella and Hill, 1997). At the cellular level, it is possible that the seizurogenic actions Fig. 9. Representative lethal effects of DM or AHN649 without EEG of MK801 and related compounds result from actions leading seizure activity in noninjured, drug-naı¨ve rats. Vertical calibration bar to an increase in cerebral glucose utilization or the induction represents 100 ␮V. Note the change in the slower chart speed from 30 of subcortical vacuolization (Nehls et al., 1990; Hargreaves et mm/s to 30 mm/min. al., 1993). Importantly, concentrations of AHN649 up to 200 ␮M are not neurotoxic in cultured rat neurons (Tortella et al., 1996). Although we have not established the neuroprotective 1995), and light microscopic assessments of cortical tissue potency of MK801 in our MCAo model, other studies have derived from normal or ischemic rat forebrain have not reported similar quantitative neuroprotective effects of shown evidence of AHN649 neuropathology (Tortella et al., MK801 to reduce focal ischemic infarction at a dose of 1 1997a). mg/kg (Dezsi et al., 1992; Margaill et al., 1996). Conse- Although the neuroprotection mechanism of action of quently, the protective ratio for MK801 would be projected to AHN649 remains unresolved, it likely involves antagonism of be less than 1. Therefore, using experimental stroke and the NMDA/calcium channel ionophore and blockade of exci- EEG neurotoxicity as outcome measures, AHN649 is approx- totoxic intracellular calcium signaling (Tortella et al., 1995). imately 7 times safer than DM and 100 times safer than Also, as reviewed above and in earlier reports (Newman, MK801. 1996; Newman et al., 1996) AHN649 and related compounds Interestingly, although DM produces seizures, seizure ac- are relatively selective ligands at low-affinity NMDA-binding tivity per se appeared to be unrelated to the lethal properties sites and the pharmacological profile of AHN649 is consistent of this drug or of AHN649 (Fig. 10). DM was, however, with that of an NMDA antagonist (Tortella et al., 1994, 1999 AHN649 Neuroprotection and EEG Studies in Rats 407

1995). Furthermore, its lack of serious neurobehavioral side (1997) [(S)-␣-Phenyl-2-pyridine-ethanamine dihydrochloride], a low-affinity uncompetitive N-methyl-D-aspartic acid antagonist, is effective in rodent models of effects as determined by EEG analysis (present study), ro- global and focal ischemia. J Pharmacol Exp Ther 239:1412–1424. torod performance, and overt behavior (Tortella et al., 1994) Crowell RM, Olsson Y, Klatzo I and Ommaya A (1970) Temporary occlusion of the and morphology (Tortella et al., 1995, 1997a) is predictive of middle cerebral artery in the monkey: Clinical and pathological observations. Stroke 1:439–448. a low-affinity ligand with rapid on-off kinetics (Rogawski, Couture S and Debonnel G (1998) Modulation of the neuronal response to N-methyl- 1993). Other possible mechanisms, including interactions D-aspartate by selective sigma2 ligands. Synapse 29:62–71. ␴ DeCoster MA, Klette KL, Knight ES and Tortella FC (1995) Sigma receptor- with 1 receptors or blockade of neuronal sodium channels, mediated neuroprotection against glutamate toxicity in primary rat neuronal ␴ ϭ cultures. Brain Res 671:45–53. also warrant consideration. AHN649 is selective for 1 (Ki ␮ ␴ Ͼ ␮ Dezsi L, Greenberg JH, Hamar J, Sladky J, Karp A and Reivich M (1992) Acute 0.98 M) versus 2 ( 10 M) binding sites (Newman et al., improvement in the histological outcome by MK-801 following focal cerebral isch- 1996), and we have recently determined using NovaScreen emic and reperfusion in the cat independent of blood flow changes. J Cereb Blood receptor competition analysis that AHN649 (1 ␮M) produces Flow Metab 12:390–399. Echevarria E, Robles L and Tortella FC (1990) Differential profiles of putative 78% inhibition at neuronal sodium channels (F.C.T. and “dextromethorphan sigma” ligands in experimental seizure models. Fed Proc A.H.N., unpublished data). Consistent with these mecha- 4:A311. ␴ Halsey JH, Conger KA, Garcia JH and Sarvary E (1991) The contribution of reoxy- nisms, ligands can modulate NMDA channel function and genation to ischemic brain damage. J Cereb Blood Flow Metab 11:994–1000. produce neuroprotection (DeCoster et al., 1995; Klette et al., Hallenbeck JM and Dutka AJ (1990) Background review and current concepts of reperfusion injury. Arch Neurol 47:1245–1254. 1995; Couture and Debonnel, 1998), and several sodium Hargreaves RJ, Rigby M, Smith D and Hill RG (1993) Lack of effect of L-687,414 channel antagonists are neuroprotective (Taylor and Mel- [(ϩ)-cis-4-methyl-HA-966, and NMDA receptor antagonist acting at the glycine drum, 1995). site, on cerebral glucose metabolism and cortical neuronal morphology. Br J Pharmacol 110:36–42. The possible induction of PCP-like psychotomimetic reac- Hunter AJ, Green AR and Cross AJ (1995) Animal models of acute ischaemic stroke: tions continues to raise serious concerns about the potential Can they predict clinically successful neuroprotective drugs? Trends Pharmacol Sci 16:123–128. clinical value of first-generation high-affinity NMDA antag- Kawasaki-Yatsugi S, Ichiki C, Yatsugi S, Sasamata SS and Yamaguchi T (1998) onists such as MK801. Because DM is rapidly metabolized to YM90K, an AMPA receptor antagonist, protects against ischemic damage caused the potent PCP-like drug dextrorphan, the psychotomimetic by permanent and transient middle cerebral artery occlusion in rats. Naunyn- Schmiedeberg’s Arch Pharmacol 358:586–591. risk potential could be considerable. As such, our singular Klette KL, DeCoster MA, Moreton JE and Tortella FC (1995) Role of calcium in objective has been to synthesize novel analogs of DM that sigma mediated neuroprotection against glutamate toxicity in primary rat neuronal cultures. Brain Res 704:31–41. would not metabolize to dextrorphan or would do so at a Koizumi JY, Nakazawa TO and Ooneda G (1986) Experimental studies of ischemic slower rate, and thereby exhibit a neuropharmacology dis- brain edema. I: A new experimental model of cerebral embolism in rats in which recirculation can be introduced in the ischemic area. Jpn J Stroke 8:1–8. tinct from PCP-like drugs. Experimentally, AHN649 appears Kuroiwa T, Shibutani M and Okeda R (1988) Blood-brain disruption and exacerba- to closely fit this profile because it is devoid of PCP-like tion of ischemic brain edema after restoration of blood flow in experimental focal behavior, EEG, or receptor-binding activity, yet retains many cerebral ischemia. Acta Neuropathol 76:62–70. Longa EZ, Weinstein PR, Carlson S and Cummings R (1989) Reversible middle of the clinical actions of DM. Other analogs of DM have been cerebral artery occlusion without craniectomy in rats. Stroke 20:84–91. synthesized that also exhibit a similar pharmacological pro- Lu X-CM, Tang Z, Liu W, Lin O and Slusher BS (1998) N-Acetyl-aspartyl-glutamate (NAAG) protects against focal cerebral ischemia in a rat model of MCAo. Neurosci file and also have been shown to be neuroprotective, at least Abstr 24:581. in in vitro models of neuronal injury (Tortella et al., 1995). Margaill I, Parmentier S, Callebert J, Allix M, Boulu RG and Plotkine M (1996) Short therapeutic window for MK-801 in transient focal cerebral ischemia in These novel analogs of DM remain to be studied in experi- normotensive rats. J Cereb Blood Flow Metab 16:107–113. mental in vivo brain injury models. Marquis KL, Paquette NC, Gussio RP and Moreton JE (1989) Comparative electro- In conclusion, comprehensive dose-response studies have encephalographic and behavioral effects of phencyclidine, (ϩ)-SKF-10,047 and MK-801 in rats. J Pharmacol Exp Ther 251:1104–1112. confirmed potent neuroprotective actions of AHN649 in a Muir KW and Lees KR (1995) Clinical experience with excitatory amino acid antag- clinically relevant experimental model of focal cerebral isch- onist drugs. Stroke 26:503–513. Nehls DG, Park CK, MacCormack AG and McCulloch J (1990) The effects of N- emia or stroke. In addition, the results obtained from our methyl-D-aspartate receptor blockade with MK-801 upon the relationship between EEG studies indicate that AHN649 is safer and relatively cerebral blood flow and glucose utilization. Brain Res 511:271–279. Newman AH (1996) Dextromethorphan, carbetapentane, and analogs as anticonvul- devoid of the neurotoxicity normally associated with other sant and neuroprotectant agents, in Studies Medicinal Chemistry (Vol 1; NMDA antagonists like DM or MK801. The improved phar- Choudhary MI ed) pp 331–353, Harwood Academic, Switzerland. macological profile has been proposed to confer added safety Newman AH, Bevan K, Bowery N and Tortella FC (1992) Synthesis and evaluation of 3-substituted-17-methylmorphinan analogs as potential anticonvulsant agents. directly attributable to lower-affinity binding at the NMDA J Med Chem 35:4135–4142. receptor complex while retaining neuroprotection efficacy, Newman AH, Shah JH, Izenwasser S, Heller B, Mattson MV and Tortella FC (1996) Highly selective sigma1 ligands based on dextromethorphan. Med Chem Res further supporting the feasibility of targeting low-affinity 6:102–117. NMDA site ligands as a postinjury therapy for injuries to the Palmer GC, Cregan EF, Borrelli AR and Willett F (1995) Neuroprotective properties brain. Critical studies aimed at determining the postinjury of the uncompetitive NMDA receptor antagonist remacemide hydrochloride. Ann NY Acad Sci 765:236–247. therapeutic window for AHN649, which has been reported to Ringelstein EB, Biniek R, Weiller C, Ammeling B, Nolte PN and Thron A (1992) Type be relatively short (i.e., less than 30 min) for MK801 (Mar- and extent of hemispheric brain infarctions and clinical outcome in early and delayed middle cerebral artery recanalization. Neurology 42:289–298. gaill et al., 1996), and therapeutic efficacy in longer recovery Rogawski MA (1992) The NMDA receptor, NMDA antagonists and epilepsy therapy: models of MCAo injury are under way. A status report. Drugs 44:279–292. Rogawski MA (1993) Therapeutic potential of excitatory amino acid antagonists: Channel blockers and 2,3-benzodiazepines. Trends Pharmacol Sci 14:325–331. References Szatkowski M and Attwell D (1994) Triggering and execution of neuronal death in Bederson JB, Pitts LH, Tsuh M, Nishmura MC, Davis RL and Bartkowski H (1986) brain ischemia: Two phases of glutamate release by different mechanisms. Trends Rat middle cerebral artery occlusion: Evaluation of the model and development of Neurosci 17:359–365. a neurological examination. Stroke 17:472–476. Tallarida JJ and Murray RB (1987) Manual of Pharmacologic Calculations with Britton P, Lu May X-C, Laskosky MS and Tortella FC (1997) Dextromethorphan Computer Programs. Springer-Verlag, New York. ϩ protects against cerebral injury following transient, but not permanent, focal Taylor CP and Meldrum BS (1995) Na channels as targets for neuroprotective ischemia in rats. Life Sci 60:1729–1740. drugs. Trends Pharmacol Sci 16:309–316. Clemens JA and Panetta JA (1994) Neuroprotection by antioxidants in models of Tortella FC and Hill RG (1996) EEG seizure activity and behavioral neurotoxicity global and focal ischemia. Ann NY Acad Sci 738:250–256. produced by (ϩ)-MK801, but not the glycine site antagonist L-687,414, in the rat. Cregan EF, Peeling J, Corbett D, Buchan AM, Saunders J, Auer RN, Gao M, Neuropharmacology 35:441–448. McCarthy DJ, Eisman MS, Campell TM, Murray RJ, Stagnitto ML and Palmer GC Tortella FC, Klette KL, DeCoster MA, Davis BJ and Newman AH (1995) Dextro- 408 Tortella et al. Vol. 291

methorphan analogs are neuroprotective in vitro and block glutamate-induced Xue D, Huang ZG, Smith KE and Buchan AM (1992) Immediate or delayed mild excitotoxic calcium signals in neurons. Neurosci Lett 198:79–82. hypothermia prevents focal cerebral infarction. Brain Res 587:66–72. Tortella FC, Lu X-M, Britton P, Laskosky M, Rose J, Robles L, DeCoster M and Yang GY and Betz L (1994) Reperfusion-induced injury to the blood-brain barrier Newman AH (1997a) AHN-649, a novel analog of dextromethorphan, is neuropro- after middle cerebral artery occlusion in rats. Stroke 25:1658–1665. tective and improves recovery following global forebrain injury in rats. Pharmacol Zhang RL, Chopp M, Zarow GJ, Graham SH and Weinstein PR (1994) Postischemic Rev Commun 9:157–163. (1 hour) hypothermia significantly reduces ischemic cell damage in rats subjected Tortella FC, Robles L, Witkin JM and Newman AH (1994) Novel anticonvulsant to 2 hours of middle cerebral artery occlusion. Stroke 24:1235–1240. analogs of dextromethorphan: Improved efficacy, potency, duration and side-effect profile. J Pharmacol Exp Ther 268:727–733. Tortella FC, Rose J, Robles L, Moreton JE, Hughes J and Hunter JC (1997b) EEG Send reprint requests to: Dr. Frank C. Tortella, Department of Neurophar- Spectral analysis of the neuroprotective kappa opioids enadoline and PD117302. macology and Molecular Biology, Division of Neurosciences, Walter Reed Army J Pharmacol Exp Ther 282:286–293. Institute of Research, Washington, DC 20307-5100. E-mail: frank.tortella@ Whetsell WO (1996) Current concepts of excitotoxicity. J Neuropathol Exp Neurol na.amedd.army.mil 55:1–13.