Opioid receptor antagonist nalmefene stereospecifically inhibits glutamate release during global cerebral ischemia

BRAIN RESEARCH E LS EV I ER

Brain Research 632 (1993) 346-35(I

Short Communication

Opioid receptor antagonist nalmefene stereospecifically inhibits glutamate release during global cerebral ischemia S.H. Graham ~, H. Shimizu b, A. Newman c, p. Weinstein b, A.I. Faden ~,d,. Departments of " Neurology and b Neurosurgery, Unicersity of California at San Francisco, San Francisco, CA 94121, USA, " Department of Applied Biochemistry, Walter ReedArmy Institute of Research, Washington, DC 2030Z USA, a Departments of Neurology and Pharmacology, Georgetown Uni~,ersity School of Medicine, 3900 Resert~oir Road, NW, Washington, DC 20007, USA (Accepted 5 October 1993)

Abstract

The opioid receptor antagonist nalmefene improves cellular bioenergetics and attenuates the reduction in tissue glutamate levels after global cerebral ischemia/reperfusion. The latter finding suggests that nalmefene might inhibit glutamate release during ischemia. To test this hypothesis, we used microdialysis techniques to examine the effect of nalmefene pretreatment on extracellular excitatory amino acid levels during global cerebral ischemia in rats. Saline, (-)-nalmefene (20, 100 or 500 p.g/kg) or the inactive nalmefene enantiomer (+)-nalmefene (100 /zg/kg) were given 15 min prior to induction of ischemia using a multi-vessel occlusion model. Pretreatment with (-)-nalmefene decreased peak dialysate glutamate in a dose-dependent fashion as compared to saline-treated controls, whereas (+)-nalmefene had no effect. These results suggest that opioid receptors may modulate glutamate release during ischemia and that inhibition of excitatory amino acid release may contribute to the protective actions of opioid receptor antagonists in cerebral ischemia. Key words: Opioid antagonist; Microdialysis; Stroke; Excitatory amino acid

Opioid receptor antagonists have been shown to be effective in reducing injury due to both central nervous system (CNS) trauma and ischemia, but their mechanism of action remains unclear [19]. A potential linkage among endogenous opioids, opioid receptors and excitatory amino acids (EAAs) in the pathophysiology of CNS injury is suggested by several studies. Both Nmethyl-D-aspartate (NMDA) antagonists and opioid antagonists block dynorphin-induced paralysis in the rat [2]. Pretreatment with opioid antagonists stereospecifically attenuates dynorphin-induced reduction of spinal cord glutamate levels associated with reduced histological and behavioral changes [1,2]. Opioid antagonist pretreatment also improves the cellular bioenergetic state following global cerebral ischemia in rats, and reduces the depletion of glutamate from ischemic brain tissue [16]. Collectively, these observations indicate that opioid and excitotoxin mechanisms of sec-

* Corresponding author at Georgetown University. 687-2585.

Fax: (1)

(202)

ondary tissue injury are closely related and suggest that opioid antagonists may decrease the release of glutamate into the extracellular space during ischemia. Accordingly, the present study examined the effect of opioid antagonist pretreatment upon changes in extracellular EAAs monitored by microdialysis during complete global ischemia. Male Sprague-Dawley rats weighing 350 +_ 50 g were initially anesthetized with 3% isoflurane, orotracheally intubated with a plastic 16-gauge angiocatheter and ventilated. Muscle paralysis was achieved by an intraperitoneal injection of pancuronium (1 mg/kg). Anesthesia was maintained with 1-1.5% isoflurane, together with administration of 70% N 2 0 and 30% 0 2 mixture. A femoral artery was cannulated for monitoring of blood pressure and arterial blood gases. During the experimental period, the arterial pCO 2 was adjusted to 35-40 mmHg, p O 2 tO 100-140 mmHg, and the rectal temperature was maintained at 37.5-38.0°C throughout the experiment by a heating lamp. Complete global cerebral ischemia was produced using the method described by Shirane et al. [33]. The omohyoid

0006-8993/93/$06.00 © 1993 Elsevier Science Publishers B.V. All rights reserved SSDI 0 0 0 6 - 8 9 9 3 ( 9 3 ) E 1 3 2 8 - Z

347

S.H. Graham et al. / Brain Research 632 (1993) 346-350

muscles were microsurgically transected through a ventral cervical skin incision and the external carotid and pterygopalatine arteries were occluded with bipolar cauterization. The basilar artery was then occluded as described by Kameyama et al. [25]. Silastic tubing (0.047 inches in diameter, Dow Corning) was loosely looped around each common carotid artery and sternomastoid muscle. Following the initial surgery, rats were placed in a stereotactic frame and a 4 mm microdialysis probe was inserted through a left parietal burr hole at stereotactic coordinates AP 0, ML + 4, angled 5° laterally from the sagittal plane. These coordinates place the probe in the dorsal hippocampus. After 60 min for stabilization, 30 min of temporary ischemia was induced by applying tension to the silastic snares to occlude both common carotids. Reperfusion was accomplished by release of tension on the carotid snares. The microdialysis probe was perfused with artificial CSF at 2 ~ l / m i n . 20 p~l samples were collected every 10 min throughout the experiments starting at 20 min before ischemia. At the conclusion of each experiment, probes were inserted into a control amino acid solution. Data from the experiment were excluded if recovery of glutamate was less than 5% or baseline glutamate levels were greater than 5 /xM. Recovery rates were examined across groups to ensure than any observed differences did not reflect differences in recovery rates. Dialysate was derivatized with orthophthaldialdehyde and t-butylthiol, detected electrochemically and quantified by peak areas by an observer blind to the experimental format [21]. Nalmefene, an exomethylene derivative of naltrexone [22], was used as the opioid antagonist for several reasons. This is a potent antagonist which has proved effective in improving the outcome across many major injury models - traumatic brain injury [34], traumatic spinal cord injury [15], dynorphin-induced injury [1], as well as the same global ischemia model as used in the present experiments [16]. A dextrostereoisomer of nalmefene, which is inactive at opioid receptors [1], was also available as an additional control. Experiments were performed in five groups: saline ( n = 9 ) , (-)-nalmefene (20 /zg/kg, n = 5 ) , ( - ) nalmefene (100 /zg/kg, n = 8 ) , (-)-nalmefene (500 /zg/kg, n = 5) and (+)-nalmefene (100/zg/kg, n = 8). Drug or equal volume saline was given i.v. 15 rain prior to. induction of ischemia by common carotid artery occlusion. Differences across groups with regard to peak and total amino acid values were analyzed by analysis of variance followed by post-hoc testing with the Student-Newman-Keul's test. There were no significant differences in blood pressure, P O 2 , p C O 2 , pH, or rectal temperature among the groups. (-)-Nalmefene inhibited glutamate release during ischemia in a dose-related fashion, causing a

maximum decrease in peak glutamate release (t = 30 min after onset of ischemia) at 100 /~g/kg (Fig. 1A). Dialysate glutamate in (+)-nalmefene-treated (100 /zg/kg) animals was not significantly different from saline-treated rats at any time during or after ischemia, but (-)-nalmefene-treated (100 ~ g / k g ) rats had significantly less microdialysate peak glutamate than either control group (Fig. 1B). Furthermore, the total glutamate in the microdialysate in the 60 min. after onset of ischemia was reduced in (-)-nalmefenetreated (100 Izg/kg) rats (1.36 + 0.28 nmol, P < 0.05) compared to saline controls (2.1 +0.29 nmol); (+)nalmefene (1.83 + 0.29 nmol) did not reduce total microdialysate glutamate compared to saline controls. (-)-Nalmefene-treated rats also had significantly smaller increases in extracellular glycine than saline- or the inactive enantiomer-treated group (Fig. 1C). There were no significant differences in aspartate, 7-aminobutryic acid (GABA) or taurine (Table 1). There were no significant differences in fractional recovery of glutamate from the standard solution among saline (0.21 +0.05), (-)-nalmefene 100 /zg/kg (0.27+_0.03) or ( + )-nalmefene (0.19 _+0.03). Several studies have shown that nalmefene is neuroprotective in models of global CNS ischemia/ reperfusion [16,36]. The inverted U-shaped dose-response curve for nalmefene's effects on glutamate release in the present study is consistent with the typically inverted U-shaped dose-response for opioid antagonists in CNS trauma and ischemia [19]. Glutamate has been implicated as a contributing factor leading to tissue damage after cerebral ischemia [3,6]. Glutamate release during ischemia may occur through at least three mechanisms: (1) synaptic release of glutamate from neurotransmitter stores; (2) reversal of the sodium glutamate co-transporter due to depolarization of the cell and Na influx; (3) non-specific release of glutamate from damaged neurons [10,24,32]. Nalmefene could conceivably decrease glutamate release by any of these mechanisms. However, the observation that nalmefene's effect on glutamate release is stereospecific

Table 1 Peak dialysate amino acid concentrations (ttM) per treatment group Amino acid

Saline

(-) Nalmefene 100 m g / k g

(+ ) Nalmefene 100 m g / k g

Glutamate Aspartate Glycine Taurinc GABA

38.4+4.8 6.0 _+0.7 8.5 _+0.9 40.2_+5.8 9.2+4.2

21.9_+4.0 * 4.7 _+0.8 5.5 _+0.8 * 25.2_+3.8 4.2_+ 1.0

35.0_+5.0 5.8 _+0.6 8.6 _+0.7 41.1_+4.7 7.8+ 1.0

Data expressed as mean+S.E.M. * P < 0.05 different than saline group. All peak concentrations occurred at 30 min after the onset of ischemia.

348

S.H. Graham et al. / Brain Research 632 (1993) 346-350

strongly implicates an opioid receptor mechanism in this response. The present findings are consistent with earlier reports that opioid receptor antagonists limit decreases in tissue levels of excitatory amino acids following global cerebral ischemia [16] and intrathecal dynorphin administration [1]. From these data, we suggest that the neuroprotective actions of opioid receptor antagonists may involve, in part, modulation of amino acid release. Which endogenous opioids and opioid receptors are involved in secondary CNS injury after trauma or ischemia has not been clearly established [19]. Dynorphin has been implicated as a pathophysiological factor in both traumatic brain injury (TBI) and spinal cord injury (SCI), in part through a K opioid receptor mechanism [17,18a,28,35]. With both forms of experimental trauma, dynorphin-like immunoreactivity accumulates at trauma sites in proportion to injury severity [14,28]. In SCI, centrally administered dynorphin A-(1-17), but not the receptor inactive dynorphin A-(2-17), exacerbates the effects of trauma [18a]. Moreover, both dynorphin-induced SCI and dynorphin-exacerbated impact SCI are reversed by the x-selective opioid receptor antagonist norbinaltorphimine [1,17]. In addition, traumatic SCI is antagonized by polycional dynorphin antibodies, but not antisera to the related opioid leucine-enkephalin. However, dynorphin contributes to secondary tissue injury through both opioid receptormediated and non-opioid mechanisms [18a]. Likewise, dynorphin-induced changes in excitatory amino acids appear to involve both opioid receptor-mediated [1,18a] and non-opioid [18,18a] actions; such differences may be related to the experimental model or anatomical site. Although considerable experimental evidence supports a role for endogenous opioids and opioid receptors in the pathophysiology of CNS ischemia in a variety of models, these data are somewhat less consistent than those for CNS trauma [19]. Particularly confusing have been the observations that K agonists such as U50,488H [23], CI-977 [9], or GR89696 [4] may be neuroprotective in certain cerebral ischemia models, and that certain of these agents may inhibit glutamate release in hippocampus [20] or reduce excitatory synaptic potentials at other brain sites [27,30]. Whether opioid or non-opioid actions of these compounds are responsible for their neuroprotective effects has been debated [8]. However, the relatively recent documentation of multiple K isoreceptors may provide a potential explanation for the apparent incompatibilities in the experimental literature [7,31,37]. Thus, whereas dynorphin is active at both K~ and K: receptors, compounds like U50,488H are active at the K~ site but virtually inactive at the K2 site [37]. Of potential relevance in this regard is the observation that the benzomorphan opioid antagonist WIN-44,441-3, which is the most

A ~

4o

.~. 35

~ 3o 2 (.9 25 ~ 2O

.__. 15

p

.u_ 10

~-

0

20pg/kg

Saline

100pg/kg 500~g/kg (-) Nalmefene dose

B

-III-- Saline • (-) Nalmefene 100~g/kg --&-- (+) Nalmefene 100lag/kg

60Ischemla > = ReDerfuslon

~ 50~ o~ 40 e 30E _ 20(9 ¢o

10-

.__. "~ o

0-

G

3'0

Time

6'0

after Onset of Ischemia I --=-- Saline - - e - - (-) Nalmefene lO01~g/mg _ A _ (+) Nalmefene lO0~g/mg

C 15v

L I-

Ischemia

. . . . . . . =.

=!~ Reperfusion I

,~10.

8 (3

5¸

2"

o

--

'

[)

3'0

6'0

T i m e after O n s e t o f I s c h e m i a

Fig. 1. A: effect of increasing doses of ( - ) - n a l m e f e n e upon peak microdialysate glutamate. All peaks occurred in samples the collection of which was completed at 30 min after onset of ischemia, the last sample collected during ischemia. Data are mean_+ S.E.M.. * P < 0.05 different from saline-treated group. B: glutamate microdialysis concentrations for saline-treated animals, ( - ) - n a l m e f e n e 100 ~ g / k g and ( + ) - n a l m e f e n e 100 txg/kg. Each time point is plotted at the completion of the collection time for that sample. Data are mean_+ S.E.M. * P < 0.05 different from saline and ( + ) - n a l m e f e n e treated groups. C: glycine microdialysis concentrations for salinetreated animals, ( - ) - n a l m e f e n e 100 i z g / k g and ( + ) - n a l m e f e n e 100 p,g/kg.

S.H. Graham et aL /Brain Research 632 (1993) 346-350

potent opioid antagonist for the treatment of CNS ischemia [11] or trauma [28], has the highest affinity for the K2b site defined by Rothman et al. [31]. Nalmefene is a relatively non-selective opioid receptor antagonist; however, although most active at opioid receptors, its affinity at K opioid receptors is approximately 12 times that of naloxone [29]. We used this compound for two reasons. First, we previously showed that nalmefene pretreatment limited changes in tissue levels of excitatory amino acids after cerebral ischemia in the same model [16], as well as in dynorphin-induced SCI [1]. Secondly, we had access to the enantiomer, which is inactive at opioid receptors [1] and which served as an important additional control. Although we cannot be sure that nalmefene's protective actions are mediated by K opioid receptors, considerable prior evidence has demonstrated that ~ receptors are not involved in the pathophysiology of CNS ischemia or trauma [19]. Low doses of naloxone ( / z g / k g range) that provide some selectivity for/~ receptors are ineffective in trauma or ischemia [12], whereas high doses that affect non-~ receptors (such as 3 or K) are optimal in terms of behavioral and histological outcome [19]. The present dose-response studies, as well as earlier dose-response studies in SCI [15], are consistent with these earlier reports given the relative potencies between nalmefene and naloxone. Whereas 6 receptor antagonists have not been extensively studied, they have proved ineffective [13]. The observations that structurally different opioid receptor antagonists improve outcome in CNS injury, combined with the stereospecific nature of this response, provides strong evidence that opioid receptors participate in the secondary injury response. The fact that the K-selective antagonist norbinaltorphimine improves outcome after brain [35] or spinal cord injury [17] provides further support for the conclusion that K receptors may be involved. This work was supported in part by an American Academy of Neurology Neuropharmacology Fellowship Award to S.H.G. and by Grant R49/CCR306634-02 from the Centers for Disease Control and Grant RO1/NS27849 from the National Institutes of Health to A.I.F. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control. The principles enumerated in the Guide for the Care and Use of Laboratory Animals, prepared by the Committee on Care and Use of Laboratory Animals of the Institute of Laboratory Resources, National Research Council [DHEW Pub. No. (NIH) 85-23, 2985] were adhered to. [1] Bakshi, R., Newman, A.H. and Faden, A.I., Dynorphin A-(l-17) induces alterations in free fatty acids, excitatory amino acids, and motor function through an opiate receptor-mediated mechanism, J. Neurosci., 10 (1990) 3793-3800. [2] Bakshi, R., Ni, R.X. and Faden, A.I., N-Methyl-o-aspartate (NMDA) and opioid receptors mediate dynorphin-induced spinal cord injury: behavioral and histological studies, Brain Res., 580 (1992) 255-264.

349

[3] Benveniste, H., Drejer, J., Schousboe, A. and Diemer, N.H., Elevation of extracellular concentrations of glutamate and aspartate in the rat hippocampus during transient cerebral ischemia observed by intracerebral microdialysis, J. Neurochem., 43 (1984) 1369-1374. [4] Birch, P.F., Rogers, H., Hayes, A.G., Hayward, N.J., Tyers, M.B., Scopes, D.I.C., Naylor, A. and Judd, D.B., Neuroprotective actions of GR89696, a highly potent and selective K-opioid receptor agonist, Br. J. Pharmacol. 103 (1991) 1819-1823. [5] Choi, D.W. and Viseskul, V., Opioids and non-opioid enantiomers selectively attenuate N-methyl-D-aspartate neurotoxicity on cortical neurons, Eur. Z PharmacoL, 155 (1988) 27-35. [6] Choi, D.W., Ionic dependence of glutamate neurotoxicity. J. Neurosci., 7 (1987) 369-379. [7] Clark, J.A., Liu, L., Price, M., Hersh, B., Edelson, M. and Pasternak, G.W. Kappa opiate receptr multiplicity: evidence for two U50,488-sensitive K1 subtypes and a novel K3 subtype, J. Pharmacol. Exp. Ther. 251 (1989) 461-468. [8] Contreras, P.C., Ragan, 19.M., Bremer, M.E., Lanthorn, T.H., Gray, N.M., Iyengar, S., Jacobson, A.E., Rice, K.C. and deCosta, B.R., Evaluation of U-50,488H analogs for neuroprotective activity in the gerbil, Brain Res., 546 (1991) 79-82. [9] Cordon, J.J., Boxer, P.A., Dominick, M.A. and Marcoux, F.W., CI-977, a novel K opioid receptor agonist, reduces infarct size in a model of focal cerebral ischaemia, Soc. Neurosci. Abstr. 16 (1990) 934. [10] Drejer, J., Benveniste, H., Diemer, N.H., Schousboe, A., Cellular origins of ischemia induced glutamate release from brain tissue in vivo and in vitro, J. Neurochem., 45 (1985) 145-151. [ll] Faden, A.I. and Jacobs, T.P., Opiate antagonist WIN44,441-3 stereospecifically improves neurologic recovery after ischemic spinal injury, Neurology 35 (1985) 1311-1315. [12] Faden, A.I., Jacobs, T.P., Smith, M.T., Zivin, J.A. Naloxone in experimental spinal cord ischemia: dose response studies, Eur. J. PharmacoL, 103 (1984) 115-120. [13] Faden, A.I., Jacobs, T.P. and Zivin, J.A., Comparison of naloxone and a A-selective antagonist in experimental spinal stroke, Life Sci. 33 (Suppl 1) (1983) 707-710. [14] Faden, A.I., Molineaux, C.J., Rosenberger, J.C., Jacobs, T.P. and Cox, B.M., Endogenous opioid immunoreactivity in rat spinal cord following traumatic injury, Ann. Neurol., 17 (1985) 386-390. [15] Faden, A.I., Sacksen, I. and Noble, LJ., Opiate-receptor antagonist nalmefene improves neurological recovery after traumatic spinal cord injury in rats through a central mechanism, J. Pharmacol. Exp. Ther., 245 (1988) 742-748. [16] Faden, A.I., Shirane, R., Chang, L.H., James, T.L., Lemke, M. and Weinstein, P.R., Opiate-receptor antagonist improves metabolic recovery and limits neurochemical alterations associated with reperfusion after global brain ischemia in rats, J. Pharmacol. Exp. Ther., 255 (1990) 451-458. [17] Faden, A.I., Takemori, A.E. and Portoghese, P.S., K-Selective opiate antagonist nor-binaltorphimine improves outcome after traumatic spinal cord injury in rats, CNS Trauma 4 (1987) 227-237. [18] Faden, A.I., Dynorphin increases extracellular levels of excitatory amino acids in the brain through a non-opioid mechanism, J. Neurosci., 12 (1992) 425-429. [18a] Faden, A.I., Opioid and non-opioid mechanisms may contribute to dynorphin's pathophysiological actions in spinal cord injury, Ann. Neurol., 27 (1990) 67-74. [19] Faden, A.I., Role of opioids in central nervous system injury. In A. Herz, H. Akil and E.J. Simon (Eds.), Handbook of Experimental Pharmacology, Springer-Verlag, Berlin, 1993, Chapter 42, pp. 325-341. [20] Gannon, R.L. and Terrian, D.M., U-50,488H inhibits dynorphin

350

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

s.tt. Graham et al. / Brain Research 032 (1993) 346-350 and glutamate release from guinea pig hippocampal mossy fiber terminals, Brain Res., 548 (1991) 242-247. Graham, S.H., Shiraishi. K., Panter, S.S., Simon, R.P. and Faden, A.I., Changes in extracellular amino acid neurotransmitters produced by focal cerebral ischemia, Neurosci. Lett., 11(1 (1990) 124-129. Hahn, E.F., Fishman, J. and Hellman, R.D., Narcotic antagonists.4.carbon-6 derivatives of N-substituted noroxymorphones as narcotic antagonists, J. Med. Chem., 18 (1975) 259-262. Hall, E. and Berry, K.P., A quantitative analysis of the effects of opioid agonists on post-ischemic hippocampal CAI neuronal necrosis, Stroke 1990. Ikeda, M., Nakazawa, T., Abe, K., Kaneko, T. and Yamatsu, K. Extracellular accumulation of glutamate in the hippocampus induced by ischemia is not calcium dependent: in vitro and in vivo evidence, Neurosci. Lett., 96 (1989) 202-206. Kameyama, E., Suzuki, J., Shirane, R. and Ogawa, A., A new model of bilateral hemispheric ischemia in the rat: three vessel occlusion model, Stroke, t6 11985) 489-493. Kusumoto, K, Mackay, K.B. and McCulloch, J., The effect of a K-opioid receptor agonist CI-977 in a rat model of cerebral ischemia, Brain Res., 576 (1992) 147-51. McFadzean, I., Lacey, M.G., Hill, R.G. and Henderson, G., Kappa opioid receptor activation depresses excitatory synaptic input to rat locus coeruleus neurons in vitro, Neuroscience, 20 (1987) 231-239. Mclntosh, T.K., Hayes, R., Dewitt, D., Agura, V. and Faden, A.I., Endogenous opioids may mediate secondary damage after experimental brain injury, Am. J. Physiol., 253 (1987) E565E574. Michel, M.B., Bolger, G. and Weissman, B.A., Binding of a new

opiate antagonist nalmefene to rat brain membranes, Pharmacologist, 26 (1986) 201. [31)] Pinnock, R.D., A highly selective K-opioid receptor agonist, CI-977, reduces excitatory synaptic potentials in the rat locus coeruleus in vitro, Neuroscience, 47 (1992) 87 ~4. [31] Rothman, R.B., Bykov, V., deCosta, B.R., Jacobsen, A.E., Rice, K.C., Brady, L.S., Interaction of endogemms opioid peptides and other drugs with four K opioid binding sites in guinea pig brain, Peptides. 11 11990)311 331. [32] Sanchez-Preito, J., Sihra, T.S. and Nicholls, D.G., Characterization of the excitotoxic release of glutamate from the guinea-pig cerebral cortical synaptosomes, J. Neurochem., 49 (1987) 58-64. [33] Shirane, R., Shimizu, H., Kameyama, M. and Weinstein, P.R., A new method for producing temporary complete cerebral ischemia in rats, J. Cerebr. Blood Flow Metab., 11 (1991)949-956. [34] Vink, R., McIntosh, T.K., Romhanyi, R. and Faden, A.I., Opiate antagonist nalmefene improves intracellular free Mg z ' , bioenergetic state and neurological outcome following traumatic brain injury in rats, J. Neurosci., 10 (1990) 3524-3530. [35] Vink, R., Portoghese, A.I. and Faden, A.I., Kappa-opioid antagonist improves cellular bioenergetics and recovery after traumatic brain injury, Am. J. Physiol., 26l 11991) R1527 R1532. [36] Yum, S.W., Faden, A.I. Comparison of the neuroprotective effects of the N-methylm-aspartate antagonist MK 801 and the opiate-receptor antagonist nalmefene in experimental spinal cord ischemia, Arch. Neurol., 47 (1990) 27%281. [37] Zukin, R.S., Eghbali, M., ()live, D., Unterwald, E.M. and Tempel, A., Characterization and visualization of rat and guinea pig brain K-opiate receptors: evidence for KL and Kz opioid receptors, Proc. Natl. Acad. Sci. USA, 85 (1988) 4061-4065.