Electrophysiological Basis of Benzodiazepine-Induced Spatial Learning Dc ficits in the Rat
by
A C C E P T E D
FA C U LTY OF GR A D U A T E S T U D I E S
Robert Keith M cNamara B.Sc., University of Lethbridge, 1989 d e a n M.Sc., University of Victoria, 1990
OAT E .
A D issertation Subm itted in Partial Fulfillment of the Requirem ents for the Degree of
DOCTOR OF PHILOSOPHY in the D epartm ent of Psychology
We accept this thesis as conforming to the required standard
Dr. R. W . Skelton. Supervisor (Departm ent of Psychology)
Dr. M. E. Corcoran. D epartm ent M ember (Departm ent of Psychology)
Dr. B. Gold water. D epartm ent M ember (Departm ent of Psychology)
Dr. D. Paul, O utside M§ii$>er (Department of Biology) C ~ - v “ ■- / —'
Dr. A. G. Phillips, External Examiner (Departm ent of Psychology, U.B.C)
© ROBERT KEITH McNAMARA \<\<\%
U niversity of Victoria
All rights reserved. Thesis m ay not be reproduced in whole or in part, by m im eograph or other m eans, w ithout the perm ission of the author.
Supervisor: Dr. Ronald W. Skelton
ABSTRACT
Benzodiazepine (BZ) drugs, such as diazepam (Valium®) and c h l o r d ia z e p o x i d e (L ib riu m ® ), are w id ely p re sc rib e d for th e ir sedative/anxiolytic properties b u t also im pair m nem onic processes in both hum ans and animals. In the Morris w ater m aze, an aversively m otivated s p a tia l le a rn in g ta sk , BZs im p a ir s p a tia l le a rn in g b u t s p a re retention/perform ance. This spatial learning deficit cannot be attributed to sedation, gross sensorim otor im pairm ents, hypotherm ia, state-d ep en d en t learning, or reductions of escape m otivation (anxiolysis). The follow ing series of experim ents so u g h t to further characterize the neurochem ical, n eu ro an ato m ical, and electro p h y sio lo g ical su b stra te s of B Z -induced im pairm ents of spatial learning. In Experiment I, the role of endogenous BZs in spatial learning was assessed. The BZ receptor antagonists flum azenil (Ro 15-1788) and CGS 8216, as well as the BZ receptor inverse-agonist j8-carboline, en hanced spatial learn in g in an inverted-U d o s e -d e p e n d e n t m an n er, suggesting th at endogenously released BZs im pede optim al learning. In Experiment II, the role of the BZ toi receptor subtype in spatial learning was assessed. CL 218,872, a selective agonist for the BZ o)i receptor subtype, im paired spatial learning in a dose-dependent and flum azenil-reversible m anner, thereby im plicating the wi receptor subtype in BZ-induced amnesia. Together these results suggest that endogenous BZs activity, like BZ drugs, is detrim ental to spatial learning and that specific BZ receptors m ediate this im pairm ent.
Se veral neurochemical systems are im portant for spatial learning in the MWM and arc inflxienced by BZs. The contributions of tw o of these neurochemical systems, the opioids and acetylcholine (ACh), to the spatial learning deficit produced by BZs were assessed. In Experim ent III, a better u nderstanding of the role of opioid system s in spatial learning was sought. M orphine, a p ro to ty p ical opioid, im p aired sp atial learn in g in a dose- d e p e n d e n t a n d naloxone-reversible m anner. H o w ever, m o rp h in e also im paired perform ance an d escape to a visible platform an d its effects on spatial learning could be attenuated by increasing the escape incentive
(colder water). This im pairm ent p attern suggests th at m orphine im pairs spatial learning by reducing escape motivation. Because both BZs and cold w ater im m ersion increase endogenous opioid activity, it seem ed possible th at the com bination of d ru g - an d w ater-induced opioid release m ight m ediate the spatial learning deficit produced by BZs. In Experim ent IV, naloxone, an opioid receptor antagonist, com pletely blocked the s p a ta l learning deficit produced by m orphine but failed, even at a higher dose, to block the sp atial learning deficit p ro d u ced by diazepam . C onversely, flum azenil, a BZ receptor antagonist, completely blocked the spatial learning deficit pro d u ced by d iazepam b u t failed to affect the am nesic effects of m orphine. Together, these findings strongly suggest that the spatial learning deficit pro Juced by BZs is not due to enhanced opioid activity.
There is also biochemical evidt-.ve that BZs interact w ith ACh systems. In Experim ent V, flumazenil attenuated the spatial learning deficit produced by scopolam ine, an ACh (m uscarinic) antagonist, b u t physostigm ine, an acetylcholinesterase inhibitor, failed to attenuate the spatial learning deficit produced by chlordiazepoxide, even at doses that completely reversed the spatial learning deficit produced by scopolamine. Together these results fail to su p p o rt the notion that BZs im pair spatial learning by reducing ACh activity b u t suggest th at scopolamine im pairs spatial learning by enhancing endogenous BZ activity.
Several neuroanatom ical regions possess a high density of BZ recep to rs and are also integral for spatial learning in the MWM. In Experim ent VI, infusions of chlordiazepoxide into the media! septum , but not frontal cortex, nu .eus basalis magnocellularis, am ygdala, hippocam pus, or cerebellum , im p aired spatial learning b u t had little effect on anxiety. Conversely, infusions of chlordiazepoxide into the am ygdala reduced anxiety but h ad little effect on spatial learning. These results suggest that the medial septum m ediates the amnesic effects of BZs and that the am ygdala m ediates the an x io ly tic effects. In E xperim ent VII, in tra s e p ta l in fu sio n s of chlordiazepoxide w ere additionally founu to im pair spatial learning in a d ose-dependent and flum azenil-reversible m anner. How ever, infusions of flum azenil into the m edial septum faJled to block the am nesic effects of system ically adm inistered chlordiazepoxide, suggesting that the am nesic
effects of BZs are n o t m ed iated by the m edial sep tu m exclusively. T e trah y d ro am in o acrid in e, a n acetylcholinesterase in h ib ito r, failed to attenuate the spatial learning deficit produced by intraseptal infusions of chlordiazepoxide, suggesting that the deficit w as not d u e to a disruption of the septohippocam pal ACh projection. Together, these results suggest th at c h lo rd ia z e p o x id e im p airs sp a tia l learn in g by in tera c tin g w ith the septohippocam pal GABAergic projection.
The septohippocam pal GABAergic projection regulates the excitability of hippocam pal afferents (e.g., perforant path). Experim ent VIII assessed the effects of system ically adm inistered BZs o n the induction of long-term potentiation (LTP) in the perforant path. CL 218,872, b u t n ot c h lo rd ia z e p o x id e o r d ia z e p a m , sig n ifica n tly su p p re s se d lo n g -te rm potentiation. H ow ever, all d ru g s im paired spatial learning. These findings suggest that CL 218,872 im pairs spatial learning by suppressing LTP but that BZ-induced spatial learning deficits can occur in the absence of perforant path LTP suppression.
Taken together, the above results suggest th a t endogenous BZ system s, particularly those in the septohippocam pal system , are im portant m odulators of m nem onic processes. These findings are discussed in the co n te x t of u n d e rs ta n d in g in fo rm a tio n sto ra g e p ro ce sse s a n d th e implications for clinical populations.
Examiners:
Dr. R. W. Skelton, Supervisor (D epartm ent of Psychology)
Dr. M. E. Corcoran. D eoartm ent M ember (D epartm ent of Psychology)
:— —
---Dr. B. Gold w ater, Departm ent M ember (Departm ent of Psychology)
Dr. D. Paul, O utside M ember (Department of Biology) v - ^ ^ ^ ^
A B S T R A C T ... ii TABLE OF CONTENTS...v LIST OF TABLES...‘ LIST OF FIGURES... x ACKNOWLEDGEMENT... xv DEDICATION...xvi CHAPTER 1: INTRODUCTION... 1
Benzodiazepine history, use and a b u s e ...1
Benzodiazepine-induced amnesia: H um an studies...2
Benzodiazepine-induced amnesia: Animal s tu d ie s ... 5
Morris w ater m a z e ... 7
Effects of benzodiazepines on water maze acq u isitio n... 12
Benzodiazepine n eurochem islry... 26
Sum m ary of ex p erim en ts... 32
CHAPTER 2: Role of I3Z in place le a rn in g ... 34
EXPERIMENT I: Endozepines and m e m o ry ... 34
METHODS... 38
RESULTS...41
DISCUSSION...55
EXPERIMENT II: BZ receptor specificity...61
METHODS... 64
DISCUSSION...74
CHAPTER 3: Neurochemical in teractions...77
Morris water maze neuropharm acology/neurochem istry... 77
Acetylcholine...77 G lu tam ate...83 Som atostatin... 85 O p io id s... 86 C orticosteroids...86 C atecholam ines... 87 S erotonin ... 89 S u m m ary ... 89
EXPERIMENT III: Opioidergic systems and m e m o ry ... 90
EXPERIMENT I lia ...90 METHODS... 92 RESULTS...93 DISCUSSION...103 EXPERIMENT IH b...108 METHODS... 108 RESULTS... 109 DISCUSSION...114 GENERAL DISCUSSION... 115
EXPERIMENT IV: BZ interaction with opioid s y ste m s ... 117
METHODS... 117
RESULTS... 118
DISCUSSION... 124
EXPERIMENT V: BZ interaction with cholinergic sy ste m s... 131
EXPERIMENT V a ... 133
METHODS... 133
d i s c u s s i o n ...142 EXPERIMENT V b ...143 M ETHODS... 144 RESULTS...144 DISCUSSION...155 GENERAL DISCUSSION...157 CHAPTER 4: N eu ro an ato m y ...159
EXPERIMENT VI: Intracranial infusions of B Z ...161
M ETHODS... 162
RESULTS... 164
DISCUSSION...197
The septo-hippocam pal sy ste m ... 203
A n ato m y ...204
N eurochem istry... 205
C onclusions... 211
EXPERIMENT VII: Intraseptal BZ: neu ro ch em istry... 212
M ETHODS... 213 RESULTS...216 DISCUSSION...243 CHAPTER 5: Electrophysiology...248 Theta rh y th m ... 248 Signal am plification... 250
Long-term p o te n tia tio n ... 251
EXPERIMENT VIII: BZ and *ong-term p o te n tia tio n ... 255
M ETHODS... 257
RESULTS...264
DISCUSSION...274
viii
LIST OF TABLES
TABLE I: Effect of FLU, CGS, /J-CCM and diazepam o n ...54 open field behavior.
TABLE II: Effects of m orphine and naloxone on c o r e ... 104 body temperature.
TABLE Til: Effects of systemic and intracranial injections o f ... 165 chlordiazepoxide on open field behavior.
Figure 1.1: Morris water maze illu stratio n ... 9
Figure 1.2: effect of diazepam on maze acquisition ... 15
Figure 1.3: Effects of diazepam on probe trial perform ance... 17
Figure 1.4: Representative swim paths during the probe t r i a l ... 19
Figure 1.5: Effect of diazepam on core body tem perature during training ... 21
Figure 1.6: Distance and heading error during reversal acq u isitio n ...22
Figure 1.7: Effects of diazepam on reversal p ro b e ...24
Figure 2.1: Effects of flumazenil, CGS 8216.. and /J-CCM o n ...43
m aze acquisition. Figure 2.2: Effects of flumazenil, CGS 8216, and p-CCM on m a z e ...45
acquisition averaged across days. Figure 2.3: Effects of flumazenil, CGS 8216, and /J-CCM on probe t r i a l 48
perform ance. Figure 2.4: Effects of flumazenil, CGS 8216, and /J-CCM on swim s p e e d ...50
Figure 2.5: Effects of combinations of flumazenil, d ia z e p a m ... 52
and /3-CCM on maze acquisition. Figure 2.6: A composite illustration com paring the relationship anxiety ....57
and learning effects of flumazenil, CGS 8216 and /J-CCM. Figure 2.7: Effects of CL 218,872 on activity and thigm otaxia...68
Figure 2.8: Effects of diazepam, CL 218,872, and flumazenil on t h e ...71
distance taken to locate the submerged escape platform. Figure 2.9: Effect of CL 218,872 on probe perform ance...72
Figure 2.10: Effect of CL 218,872 on swim s p e e d ... 73
Figure 3.1: Effects of m orphine and naloxone on maze acq u isitio n ...95
Figure 3.2: Effects of m orphine and naloxone on probe p erfo rm an ce...98
and swim speed. Figure 3.3: Effects of m orphine and naloxone on the v isib le ... 100
platform task. Figure 3.4: Effects of chronic m orphine on maze acq u isitio n ... 102
Figure 3.5: Effects of w ater tem perature on m orphine acquisition r a t e ... I l l Figure 3.6: Interaction between m orphine, naloxone a n d ... 113
water temperature. Figure 3.7: Effects of diazepam, m orphine, naloxone and flu m az en il 120
on m aze acquisition. Figure 3.8: Effects of diazepam, m orphine, naloxone and flu m az en il 123 on probe trial performance. Figure 3.9: Effects of diazepam , m orphine, naloxone and flu m az en il... 126
on swim speed.
Figure 3.10: Effects of diazepam , m orphine, naloxone and flu m az en il 128 on the visible platform task.
Figure 3.11: Effects of chlordiazepoxide + flumazenil and scopolamine .... 136 + flumazenil on lze acquisition.
Figure 3.12: Effects of chlordiazepoxide + flumazenil and scopolamine .... 139 + flumazenil on probe trial performance.
Figure 3.13: Effects of chlordiazepoxide + flumazenil and scopolamine .... 141 + flumazenil on swim speed.
Figure 3.14: Effects of chlordiazepoxide and physostigmine, scopolamine 146 and physostigmine and (C) physostigm ine alone on
acquisition.
Figure 3.15: Interactions between chlordiazepoxide, physostgim ine and . 148 scopolamine on m aze acquisition.
Figure 3.16: Interactions between chlordiazepoxide, physostigm ine and .. 152 scopolamine on probe trial performance.
Figure 3.17: Interactions between chlordiazepoxide, physostigm ine and .. 154 scopolamine and physostigmine alone on sw im speed.
Figure 4.1: Effects of systemic chlordiazepoxide on maze acq u isitio n 168
Figure 4.2: Cannula placements in frontal c o rte x ... 171
Figure 4.3: Effects of intra-cortical chlordiazepoxide on m aze acquisition . 173
Figure 4.4: Cannula placements in nucleus basalis m agnocellularis 176
Figure 4.5: Effects of intra-I\BM chlordiazepoxide on m aze acquisition .... 178
Figure 4.6: Cannula placements in medial s e p tu m ...180
Figure 4.8: Cannula placements in h ip p o cam p u s...185
Figure 4.9: Effects of intra-hippocampal chlordiazepoxide o n ... 187 maze acquisition.
Figure 4.10: Cannula placements in a m y g d a la... 189
Figure 4.11: Effects of intra-am ygdalar chlordiazepoxide o n ... 191 maze acquisition.
Figure 4.12: Cannula placements in cerebellum ... 194
Figure 4.13: Effects of intra-cerebellar chlordiazepoxide o n ... 196 maze acquisition.
Figure 4.14: Schematic of septo-hippocampal a n a to m y ... 207
Figure 4.15: Effects of scopolamine and chlordiazepoxide o n ...218 maze acquisition
Figure 4.16: Cannula placements in medial septum (uose-response) 222
Figure 4.17: Dose-response effects of intra-septal chlordiazepoxide o n 224 maze acquisition.
Figure 4.18: Drug-free reversal acquisition... 227
Figure 4.19: Cannula placements in medial s e p tu m ...230 (flum azenu antagonism ).
Figure 4.20: Effects of systemic flumazenil on in tra-sep tal... 232 chlordiazepoxide.
(chlordiazepox;de antagonism).
Figure 4.22: Effects of intra-septal flumazenil on sy stem ic... 237 chlordiazepoxide.
Figure 4 23: Cannula placements in medial s e p tu m ... 240 (THA antagonism).
Figure 4.24: Effects of systemic THA on intra-septal chlordiazepoxide 242
Figure 5.1: Population spike m easurem ent and rep re se n ta tiv e ...260 evoked potentials.
Figure 5.2: Sample in p u t/o u tp u t curves (pre & post tetan izatio n ) 262
Figure 5.3: Electrode placements in the perforant path and dentate gyrus 266
Figure 5.4: Effects of systemic CL 218,872 on L IP in d u c tio n ...268 and spatial learning.
Figure 5.5: Effects of systemic chlordiazepoxide on LTP in d u c tio n ... 271 and spatial learning.
Figure 5.6: Effects of systemic diazepam on LTP in d u ctio n ...273 spatial learning.
ACKNOWLEDGEMENT
The author w ould like to acknowledge and thank the following contributors:
Dr. D. M. Dean, Hoffmann-La Roche Inc., Canada Dr. C. M ondadori, Ciba-Geigy LTM, Switzerland
Dr. W. J. Fanshawe, Cyanamide Co., U.S.A.
Dr. S. O. Cole Dr. M. E. Corcoran Mr. G. E. de Pape Dr. S. E. File Dr. N. M cNaughton Dr. R. G. M. Morris Dr. R. J. Sutherland Dr. D. Treit Dr. T. Walsh Dr. I. Q. W hishaw
N atural Sciences and Engineering Council of Canada B. C. Health Research Foundation
This thesis is dedicated firstly to my wife for here faith and patience, to my mom for her continued support, and to m y dogs Toben and Leiba for their thoughtful advise. I w ould also like to sincerely thank Ron for his insights, support, and selfless dedication to my endeavors.
CHAPTER 1: INTRODUCTION
Benzodiazepine history, use and abuse
Since their synthesis by Hoffm ann-La Roche, chlordiazepoxide (Librium®;1955) and its m ore potent analog diazepam (Valium®;!959), have come to dom inate the m arket for anxiolytic/sedative drugs and are the most w idely prescribed drugs in the w orld (Blackwell, 1973). In N o rth America d u rin g 1972, m ore than 77 million prescriptions for BZs w ere filled, sixty percent of which were for diazepam (Greenblatt & Shader, 1974). In 1977, 54 m illio n p re sc rip tio n s w ere filled for d iazep am an d 13 m illion for chlordiazepoxide (Skegg & Perry, 1977). In Canada in 1984, approximately 1 of every 10 Canadians reports using BZs at least once per year, and typically 10 percent of these users continue their use for m ore than 1 year (Balter, M anheim er & Mellinger, 1984). BZ d ru g use is higher am ong adults aged 50 years or over an d as th e pro p o rtio n of the aged p o p u latio n in C anada increases, a parallel increase in BZ consum ption is expected (Balter et al., 1984). A recent longitudinal stu d y conducted in C anada revealed that BZ consum ption between 1978 and 1987 increased by 145 percent and that there is a general trend away from slowly metabolized BZs, such as diazepam and chlordiazepoxide, in favor of the rapidly m etabolized BZs, such as lorazepam a n d triazo lam (Busto, Lanctot, Isaac & A drian, 1989). Because these c o m p o u n d s a re so co m m o n ly p re sc rib e d , th e ir effects o n h u m an perform ance, particularly in the elderly, w arrant further characterization.
A lthough m ost BZ drugs are prescribed prim arily for their anxiolytic and sed ativ e /h y p n o tic properties (Rickels, 1978), several side effects appear significant: (1) long-term consum ption of BZ drugs can lead to dependence (Petursson & Lader, 1981), though BZ-associated deaths are rare (Finkle, McCloskey & G oodm an, 1979), (2) BZs are know n to produce "paradoxical" rage attacks, increased incidence of hostile-aggressive feelings an d BZ drugs are often used, in com bination w ith o ther sedatives such as alcohol, to com m it suicide (Zisook & DeVaul, 1977) and (3) BZ drugs im pair cognitive p ro c e s s e s , m o st n o ta b ly le a rn in g a n d m e m o ry (L iste r, 1985).
Hence, in addition their efficacy as anxiolytic sed ative/hypnotics, BZ drugs also have properties that are clearly undesirable for the clinical population.
Benzodiazepine-induced amnesia: H um an studies
BZ drugs were discovered to have am nesic properties soon after their release onto the m arket in the early 1960s (Feldman, 1963; Knight & Burgess, 1968). The am nesic effects of BZs w ere first observed in the clinical setting where intravenous diazepam was used as a prem edicant (Knight & Burgess, 1968). In these settings, it w as discovered th a t p a tie n ts ad m in istered d iazep am p rio r to su rg e ry could recall little a b o u t th e p re su rg ic a l environm ent o r events leading u p to su rg ery (K night & Burgess, 1968). Initially this effect was considered a positive side-effect of diazepam , since it would "erase" the anxiety-provoking presurgical events from the patient's memory. While these early observations w ere often poorly controlled and often confounded by the adm inistration of other anesthetics (e.g., Feldm an, 1963), num erous, well controlled experim ents have subsequently confirm ed that BZs im pair the storage of inform ation (see Cole, i986; C urran, 1986; Ghoneim & M ewaldt, 1990; Lister, 1985; Romney & Angus, 1984 for reviews).
Although the am, lesic effects of BZs may be considered a positive side- effect in presurgical situations, problem s m ay arise clinically because [1] BZs are so widely prescribed an d are typically taken while the patient is engaged in their daily activities, [2] little tolerance develops to the am nesic effects (Ghoneim, M ew aldt, Berie, & H inrichs, 1981; Griffiths, M cLeod, Bigelow, Liebson & Roache, 1984; Petersen & Ghoneim, 1980), [3] subjects m ay not be subjectively aw are of their m em ory im pairm ent, even after doing poorly on m em ory tasks (Hinrichs, M ewaldt, Ghoneim , & Berie, 1982), [4] the am nesia can last up to 14 hours after BZ adm inistration (Ghoneim e t al., 1981), [5] the use of BZs in a psychiatric context (i.e., to com bat phobias) m ay h in d er habituation or the acquisition of appropriate coping skills (H afner & M arks, 1976), and [6] geriatric populations are prim ary consum ers of BZs despite a heightened susceptibility to their am nesic effects (Cook, Flanagan, & James, 1984; N ikaido, Ellin w ood, H eatherly & G upta, 1990; Pom ara, Stanley, Block, Guido, Stanley, G reenblatt, N ew ton, & G ershon, 1984). R egarding the latter issue, a recent stu d y d em o n strated th at 25% of n u rsin g hom e p atien ts
d iag n o sed w ith dem entia of the A lzheim er's type w ere receiving BZ m edication (Beere, A vorn, Soum erai, Everitt, Sherm an & Salem, 1988). Hence, the relative clinical utility of BZ drugs needs to be reevaluated in light of these noted liabilities.
Typically, the influences of BZs on hum an m em ory processes are studied through testing the recall and recognition of visually or auditorily presented item s, usually pictures o r infrequently used English w ords (Ghoneim & M ew aldt, 1990). O ther indexes of m em ory perform ance, including serial position (M ew aldt, Hinrichs & G honeim , 1983) and W echsler M emory or Benton Visual R etention scales (G entil, G orenstein, C am argo & Singer, 1989), have yielded congruent results. Subjects are typically between the ages of 20 and 35 years and the BZ is adm inistered either orally, intravenously, or intram u scu larly in doses ran g in g from 0.1 - 0.3 m g /k g , doses th at are commonly prescribed clinically.
Inform ation storage is considered to be a m ultiple stage process that can occur either im m ediately or require several hours before true consolidation occurs. A popular model, both conceptually and intuitively, was proposed by A tkinson and Shiffrin (1968) and involves a tw o-stages, short-term memory and long-term storage. In this model, inform ation m ust first be registered in sh o rt-term m em ory from w hich it is tra n sferred to long-term storage (encoding). The p attern of m em ory failure observed in hum ans after BZ ad m in istratio n is strik in g in its selectivity for im pairing the transfer of inform ation from short-term m em ory to long-term storage w hile having little effect on short-term m em ory (im m ediate recall) or reten tio n /retriev al (Lister, 1985). In fact, BZs m ay actually produce a significant retrograde facilitation of m em ory (Brown, Brown & Bowes, 1983; Hinrichs, Ghoneim & M ew aldt, 1984). Increasing the dose of the BZ increases the m agnitude and d u ratio n of the m em ory im pedim ent (e.g., G honeim , H inrichs, M ew aldt, 1984) and the tim e course of the amnesic effect can range from 2 m inutes to 14 ho u rs after BZ adm inistration (e.g., Ghoneim, M ewaldt, Berie & Hinrichs, 1981).
In addition to a selectivity of BZ drugs to im pede information storage, the type of inform ation also appears to be selectively affected by BZ drugs. According to the theory of Squire (1987), m em ory can be dichotom ized into declarative memory (or explicit memory; Graf & Schactar, 1985), inform ation that is directly accessible to conscious recollection (i.e., facts, n a n w p l a c e s etc.), and procedural m em ory (or im plicit m em ory; G raf & Schacter, 1985), which is m em ory that is contained w ithin learned skills (e.g., playing darts). BZs im pair only the storage of declarative inform ation w hile p rocedural m em ories are spared (Danion, Zim m erm ann, W illard-Schroeder, G range & Singer, 1989; Fang, Hinrichs & Ghoneim, 1987). For exam ple, diazepam (0.3 m g /k g ) im paired the acquisition of declarative inform ation (free recall of w ord list) b u t spared a procedural, w ord-stem com pletion task. In this task, subjects are presented w ith a series of w ords and later asked to com plete three-letter w ord stem s; these w ord stem s typically elicit those w o rd s previously presented. The selective im pairm ent of declarative m em ory and the sparing of procedural m em ory is also found in patients w ith organic amnesias such as Korsakoff's disease o r patients w ith hippocam pal dam age (Squire, 1987). The sim ilarity betw een the m nem onic deficit of BZ-treated subjects and patients w ith organic am nesia has led to the suggestion that BZ- induced am nesia m ay represent a convenient neurophannacological m odel which may facilitate the understanding the n ature of m em ory and am nesia (Wolkowitz, W eingartner, Thompson, Pickard, Paul & H om m er, 1987).
The failure of BZs to im pede short-term m em ory or retrieval is taken as evidence that the m em ory im pairm ent is not secondary to the sedation that is typically experienced during testing (M ewaldt et al., 1983). A dditional evidence su p p o rts the notion th at BZ-induced sedation and am nesia are in d ep en d en t processes. Firstly, there is a differential tim e course of the amnesic and sedative effects; the am nesic effects persist beyond the sedative effects (Pandit, H eisterkam p & Cohen, 1976; Ghoneim et al., 1981). Secondly, differential tolerance develops to the am nesic an d sedative effects of BZs; tolerance develops to the sedative effects of BZ well before it develops to the am nesic effects (Lucki, Rickels & G eller, 1986). Together th ese findings suggest that the BZ-induced amnesia is not a secondary result of sedation.
In sum, BZ drugs are frequently prescribed for their anxiolytic, sedative, and hypnotic properties an d are used by a large segm ent of the p o p u la tio n , m o st n o tab ly th e eld erly . O ne side-effect of BZ d ru g adm inistration is an im pairm ent of the transfer of inform ation from short term m em ory to long-term storage. More specifically, inform ation that is declarative cannot be encoded, unlike procedural skills which are readily acquired. The am nesic effects of BZs cannot be attributed to their sedative effects. Despite their sidj-effects, the large experimental and clinical literature p ro v id e s an o p p o rtu n ity to identify a congruency betw een the am nesic effects of BZs in hum ans and animals.
Benzodiazepine-induced amnesia: Animal studies
BZs have been found to im pair performance in anim als ranging from m onkeys to pigeons on a w ide variety of learning and m em ory tasks (see Cole, 1986; Dantzer, 1977; Thiebot, 1985 for reviews). Animal m em ory tasks com m only em p lo y ed to stu d y BZ-induced am nesia include avoidance conditioning, discrim ination learning, and spatial learning.
The avoidance conditioning task typically involves placing a rat/m o u se into one com partm ent of a tw o com partm ent box (one cham ber w hite, one cham ber black) seperated by a wall w ith t door. When the rat enters the opposite chamber (typically the black and preferred side), a breif foot-shock is adm inistered. This is the training session. D uring the test session, typically given 24 h later, the rat is returned to the w hite cham ber and the latency to reen ter the black cham ber is the index of acq u isitio n /rete n tio n . G ood a cq u isitio n /re te n tio n is m anifested as a long latency to enter the black cham ber. On avoidance conditioning tasks, pre-training adm inistration of a BZ im pairs acquisition of the avoidance responce b u t post-training-trial ad m in istratio n has little effect on acquisition (Brcekkam p, Le Pichon, & Lloyd, 1984; Essm an, 1973; Jensen, M artinez, Vasquez, & M cGaugh, 1979; Patel, Ciofalo, & Iorio, 1979; Tohyama, Nabeshim a, Ichihara, & Kameyama, 1991). It is difficult to attribute these deficits to amnesia, how ever, since BZs increase pain thresholds (H ouser & Pare, 1973; W uster, Duka & H erz, 1980) and pu n ish ed responses (Theibot, 1985). Hence, the analgesic effects of BZs m ay have reduced the significance of the shock and the m otivation to learn
(e.g., Kesner & Calder, 1980). M oreover, the BZ-induced im pairm ents of avoidance conditioning can be overcome by re-adm inistering the d ru g prior to the test session (state-dependent learning; Patel, Ciofalo & Lorio.. 1979). Hence, the avoidance conditioning task does not appear suitable for assessing the amnesic effects of BZs.
A second task used to assess the am nesic effects of BZs is sim ple discrim ination. In these tasks, the anim al m u st respond (press a lever for rev/ard) w hen a given stim ulus (light; S+) is p resen ted a n d w ith h o ld responding w hen a second stim ulus is offered (S'). The S+ an d S ' can be presented together sim ultaneously o r individually in succession. BZs im pair successive discrim ination acquisition (Cole & M ichaleski, 1986; Fukuda & Iwahara, 1976; Ksir & Slifer, 1982; M cN aughton, 1985; N icholson & W right, 1974; Thom pson, 1974) b ut spare sim ultaneous discrim ination acquisition in rats (Iwasaki, Ezawa, & Iw ahara, 1976) an d m onkeys (Ha&ogawa, Ibuka, & Iw ahara, 1973; Sahgal & Iversen, 1980). The com m on deficit observed on discrim ination tasks is a failure to w ithhold responding to the S ', an effect attributed to the disinhibitory actions of BZs (Cole, 1982, 1983, 1990; Gray, 1983; Tye, Sahgal, & Iversen, 1977). H ence, BZs im p air d iscrim ination acquisition by disinhibiting p u n ish ed /n o n -re w ard e d responses rather than learning and m em ory per se.
In monkeys, lorazepam im paired spatial delayed response perform ance which required the m onkey to retain the spatial location of a sym bol on a com puter screen (Rupniak, Samson, Stevenson, & Iversen, 1990). In rats, the effects of BZs on spatial learning has been exam ined using the appetitively m otivated radial arm maze (Olton & Samuelson, 1976). O n this task, the rat m ust retrieve a piece of food from the end of several (usually 8) arm s/alley s th a t pro ject o u t from a cen tral area. The in d iv id u a l a rm s can be discrim inated only by th eir location relative to am b ien t room cues. A reference m em ory error is recorded if the ra t enters an arm th a t has never been baited and a working m em ory error is recorded if the rat visits the same arm twice d u rin g a given test session. A ssessm ent of BZs on radial m aze learning has yielded inconsistent results. For exam ple, chlordiazepoxide im paired the rats' ability to acquire the location of a single baited arm on an 8
arm radial maze (reference m em ory error; W illner & Birbeck, 1986, see also Shum sky &; Lucki, 1991). O n a cued version of the radial arm m aze (i.e., distinguishable arms), rats treated w ith chlordiazepoxide m ade both working and reference m em ory erro rs (H odges & G reen, 1986). These au th o rs suggested that the errors committed by chlordiazepoxide treated rats may be attributed to response perseveration, rather than a specific m nem onic deficit. In a seperate study, no deficit w as observed w hen the BZ-treated rat was required to retrieve a food pellet from each of eight arm s (Hiraga & Iwasaki, 1984). It should be noted, how ever, th a t the latter task could be solved w ithout using spatial cues. For exam ple, the rat m erely had to continue to turn left upon exiting each arm to acheive a perfect score. In sum, although BZs im p air perform ance on the rad ial arm m aze, this deficit m ay be a ttrib u te d to resp o n se persev eratio n ra th e r than to a tru e nm em onic im p airm en t.
In sum , perform ance across a variety of m em ory tasks is im paired by p retrain in g BZ adm inistration. H ow ever, it is difficult to attrib u te these im pairm ents to a specific disruption of m nem onic processes per se. A task that appears well suited to the assessm ent of BZs on learning and mem ory processes is the Morris w ater maze (Morris, 1981; Morris, 1984).
The M orris w ater maze
The Morris w ater m aze (MWM; Fig. 1.1) consists of a large circular pool filled w ith cool w ater rendered o p aque w ith m ilk pow der. Subm erged som ew here w ithin the pool is a platform onto which th» rat can climb to em erge from the cool w ater and escape from the necessity of swimm ing. In the sta n d a rd use of the MWM, the ra t is placed into the pool at one of several random ly ordered start locations near the wall (e.g., n o rth , south, w est, or east "pole") end sw im s to a subm erged platform m aintained in a fixed position (e.g., center of the northw est quadrant) thro u g h o u t training. Though m any variations have been used, and m any m ore are possible.
The MWM w as originally developed to test rats' abilities to learn, rem em ber, and go to a place in space defined only by its position relative to distal, extram aze cues. Because the rat is generally placed into the pool from
Figure 1.1: Illustration of the Morris w ater maze, (top illustration adapted from "Plasticity in the Neocortex: M echanisms U nderlying Recovery From Early Brain Damage" by Kolb, B., and W hishaw , I. Q., 1989, Progress in N e u r o b io lo g v. 32. p. 242. © 1989 by P ergam on Press. A d a p te d w ith perm ission.
a variety of start locations (usually near the wall), and because there are no cues w ithin the pool to guide the rat to the platform , m ost rats solve this maze by resorting to a spatial (locale) strategy (O'Keefe & Nadel, 1978). O ther possible strategies include cue (taxon) strategies in w hich the rat guides its movements relative to a single cue in the environm ent, o r response (praxis) strategies in which the rat m oves according to a specific sequence of m ovements, though these strategies are less efficient in the MWM and are generally not considered to represent true spatial learning (O'Keefe & N adel, 1978). In this thesis, "spatial learning" will refer to acquisition of know ledge of a subm erged platform location oased on extra-maze cues (a locale strategy). The n ew term "spatial recall" will be used to refer to the dem onstration of the kn o w led g e of a su b m erg ed platfo rm location a cq u ired p rio r to anatomical or pharm acological m anipulations. "Cue learning" w ill refer to acquisition of an escape response to a platform clearly identified by an unambiguous, proximal cue (e.g, the platform itself, visible above the surface of the water).
A g reat strength of the MWM is the availability of procedures for evaluating the strategy being used to locate the platform , and for dissociating pharm acological im pairm ents of m em ory processes from non-m nem onic perform ance deficits. First and forem ost are probe trials (transfer tests) in which the ra t is perm itted to sw im freely a b o u t the pool w ith o u t any platform present. The use of a place strategy is inferred if the ra t spends more tim e in the quadrant that previously contained the platform , or crosses over th e old platfo rm p o sitio n (annulus) m o re o ften th an eq u iv alen t positions in the other three quadrants. These m easures quantify the strength and accuracy of the original learning. A lternative strategies, such as cue (taxon) or response (praxis) strategies, are revealed by sw im paths th at fail to be biased tow ards a particular q u ad ran t or platform location an d w hich either describe large circles at the appropriate distance from the pool wall or which seem to be directed tow ards or aw ay from a single cue (such as the experim enter).
A second valuable procedure is cue learning, which involves the use of a single proxim al cue to identify the location of the platform . U sually, the
platform is sim ply visible above the surface and the rat is required only to learn to sw im to it and chmb on. This task typically requires only one trial to m aster, and is capable of revealing deficits in sensory, m otor, sensorimotor, or m otivational processes. The visible platform can be m aintained in one position, or m oved from trial to trial in order to prevent the use of a spatial strategy (e.g., W hishaw & Mittleman, 1986). This task may best be viewed as a m inim ally acceptable control p rocedure for it m ay be sensitive only to relatively gross deficits. A m ore sensitive (though less frequently used) procedure is the spatial recall test. This consists of subm erged platform or probe trials given to lesioned or drugged rats that learned the 1ocation of the platform earlier, while still intact and undrugged. Spatial recall tests should reveal im pairm ents to alm ost all of the processes req u ired for p roper perform ance of the task (sensory, m otor, m otivation, m em ory retrieval, spatial inform ation processing) except for processes required for learning and for m em ory formation. A fourth valuable procedure requires the rat to learn to d iscrim inate betw een tw o visible platform s, only one of w hich will su p p o rt the w eight of the rat. The correct platform is distinguished either by its position in space (spatial version) or by its visual appearance (non-spatial version). Deficits of the non-spatial task reveal im pairm en ts of sim ple discrim ination learning or sen so ry /p ercep tu al processes, w hereas deficits specific to the spatial version reveal im pairm ents of spatial learning.
In addition to the ability of the MWM to dissociate learning, memory, and perform ance deficits and its paradigm atic flexibility, several components of this task com m end it over other animal m em ory tasks such as avoidance conditioning or m ore traditional m easures of spatial mem ory, such as the T- m aze or the radial-arm m aze. For example: 1) both learning and cognitive perform ance can be assessed sim ultaneously over the course of training, 2) sw im speed can be used to assess m otoric an d m otivational deficits within each learning trial, 3) intram aze cues such as odour trails are obviated, 4) no p re tra in in g is re q u ire d a n d acq u isitio n is q u ite ra p id (5-16 trials to asym ptote), allowing for large num ber of anim als/treatm ents to be assessed in short periods of time, 5) rats can be tested and retested over several days or m an ip u latio n s, 6) shock a d m in istratio n a n d food d ep riv atio n are not necessary, thereby reducing the stress to the anim al, and 7) the level of
m otivation [i.e., w ater tem perature (M cNamara & Skelton, 1991B)] can be m anipulated. A lthough it is possible th at im m ersion into cool w ater m ay cau se p h y sio lo g ic a l e v e n ts w h ich in te ra c t w ith p h a rm a c o lo g ic a l m anipulations in an undefined m anner, the high concordance betw een the effects of pharm acological m a n ip u la tio n s on sp a tia l lea rn in g in the aversively m o tiv ated MWM an d the ap petitively m o tiv ated rad ial arm m aze [see (Levin, 1988) for a review of radial arm m aze pharm acology], suggests that both tasks assess a common substrate of learning an d m em ory and that the cool w ater produces no substantial alteration in physiological responses to drugs.
Effects of BZs on MWM acquisition
The seminal description of the effects of BZs on spatial learning in the MWM was conducted by M cN aughton an d M orris (1987). These authors found that a single dose of chlordiazepoxide (5 m g /k g ) im paired MWM acquisition, increasing the distance to located the platform over the course of three days training. W hen the platform w as rem oved from the pool, control rats show ed a bias for the q u ad ran t th at had contained the platform while rats treated w ith chlordiazepoxide failed to show such a preference. The latter finding is taken as evidence th at the rat d id n o t acquire the spatial location of the platform . In subsequent investigations, it was dem onstrated that diazepam im paired spatial learning in a dose-dependent a n d flumazenil reversible m anner (Arolfa & Brioni, 1991; McNamara & W hishaw , 1991). For exam ple, diazepam im paired a learning-set task th a t req u ired the ra t to acquire a new platform location each day over a series of days (M cNamara & W hishaw, 1991). This learning-set does not require the retention of the prior days training, in fact forgetting the previous platform location w o u ld be a d v a n ta g e o u s. H ence, d ia z e p a m im p a irs a c q u isitio n th a t re q u ire s inform ation to be stored for only a sh o rt tim e (~ 5 m in) w hile sp aring retention processes. It w as further found in this study th at diazepam d id not im p air cue learning, suggesting th a t d iazepam does n ot im p air sim ple associative le a rn in g processes, escape m o tiv atio n , o r p ro d u c e gross sensorim otor deficits. These data suggest th at BZs im pair spatial learning in the MWM by im pairing the acquisition of spatial information.
While the above data suggest that BZs selectively im pair m nemonic processes, it is still possible that BZs affect other aspects of performance that p revent the expression of w hat has been acquired. H ow ever, w e recently d em o n strated (M cN am ara & Skelton, 1991 A) th at ra ts th at learn the platform location prior to receiving an amnesic dose of diazepam have total saving o f the platform location (Fig. 1.2A, B). D uring the probe trial, rats switched to diazepam post-acquisition show a bias for the correct quadrant to a degree equal w ith controls (Fig. 1.3A). Representative swim paths taken d uring the p ro b e trial for each treatm ent g ro u p (Fig. 1.4) show s that the group sw itched to diazepam (Switch) swim in m anner m ore sim ilar to the diazepam gro u p (wide loops) b u t are able to concentrate their time in the correct quadrant. M oreover, the sw itch gro u p show ed proficiency despite reductions of core-body tem p eratu re (Fig. 1.5). W hen the location of the subm erged platform w as m oved to a new location, the switch gro u p show ed an acquistion im pairm ent (Fig. 1.6, Fig. 1.7), as did rats treated w ith diazepam from the begining of training (Diazepam in figures), an d rats treated w ith saline d u rin g initial acquisition a n d given d iazep am d u rin g reversal tra in in g (saline-diazepam : SD in figures). H ow ever, rats treated w ith d iazep am d u rin g in itial acquisition an d given salin e d u rin g reversal (diazepam -saline: DS in figures) train in g acquired the revered platform location at control levels.
D iazepam has sedative properties w hich have tw o consequences: a tte n tio n a l/p e rc e p tu a l im p airm en ts an d m yorelaxation. In the p resen t stu d y , the place-learning deficit cannot be attributed to the m yorelaxation effects of d iazep am for th ree reasons: [1] the D iazepam g ro u p sw am consistently slo w er th a n th e Saline g ro u p , even as the g ro u p reached criterio n levels, [2] th e Sw itch g ro u p d id n o t sh o w im p aired m aze perform ance in spite of a reduction in sw im speed (Fig. 1.2C), and [3] the D iazepam g ro u p did n o t sw im slow er d u rin g the probe trials (Figs. 1.4B, 1.7B). These resu lts su g g est th at the place-learning deficit p ro d u ced by d iazepam can n o t be a ttrib u te d to the m yorelaxation. N either could the deficits have been due to perceptual/attentional factors. In the present study, the Switch group, while receiving diazepam , w as still able to navigate to the h id d e n platform . Because the distance taken to locate the platform d id not
Figure 1.2: Effects of diazepam on (A) the distance taken to locate the escape platform , (B) heading errors, an d (C) sw im speeds o v er the course of training. N ote that the preadm inistration of diazepam resulted in greater distances taken to locate the platform as well as heading errors. Also w hen the Switch group w as sw itched from saline to diazepam on day 7, neither the distance nor heading errors increased substantially, despite reductions of swim speed.
1 5 0 0 - i Diazepam Switch Saline 1000 - 500-Swltch B. a ID O O t _ i — Ui CO c CO o X 60 -i 50 40 30 -Switch 20 10 -(0
E
o
TJ <D 0) Q. (/) 5 (/) 35 -| 30 -*. Switch • * Im 25 -4 5 6 7 8 9 1 0 1 1 1 2 1 3 1-4DAY
Fig ure 1.3: Effects of diazepam on (A) the distance sp en t in the correct quadrant, (B) sw im speed, and (C) post-sw im Tc (post-sw im - pre-swim ) during the first probe trial. Note th at both the Saline and Switch groups, b u t not the D iazepam g ro u p , d em onstrated an 'above chance' bias for the quadrant th at previously contained the escape platform . Also, none of the group's sw im speeds differed d u rin g the probe trial. Finally, only the Saline group's post-swim Tc was reduced. Data expressed as m ean ± S.E.M.. *p<0.01 com pared to chance level (25%) in (A) and to Saline Tc in (C).
-0 .5 & 6 & b is M O N CO D ia z e p a m S a lin e S w itc h
Figure 1.4: Representative swim paths d u rin g the first probe trial. N ote that the Switch group rat, despite having an elongated and circuitous sw im path, still spent the m ajority of tim e in the correct q u ad ran t. The 'F denotes where the rat was rem oved from the pool after the trial w as finished.
19 Saline Sw itch i I D iazepam I
Figure 1.5: Effects of diazepam on (A) pre-sw im Tc (pre-swim - pre-drug) and (B) post-swim Tc (post-swim - pre-swim) d uring initial acquisition. Note: [1] the consistently lower pre-sw im Tc over the course of testing, [2] w hen the Switch group is switched to diazepam on day 7, the pre-sw im Tc is reduced and [3] all three groups show increases in post-sw im Tc over the course of testing (B).
P o st -S w im Tc (D E G C ) P re -S w im Tc (D E G C ) A . Diazepam Saline Switch 1.0 0.5 -0.0 0.5 --1.0 -Switch -1.5 C. 2 1 Switch 0 -1 ■2 1 2 3 4 5 6 7 8 9 10 11 1 2 1 3 1 4
DAY
A. E o <D O c (0 (0 o SD Diazepam Sw itch Saline DS -a 200 B. O UJ Q Oi -i . ai O) c •5 (0 a> X 5 6 7 8 2 3 4 1
DAY
Figure 1.6: A n illustration of (A) the distance taken to locate the subm erged platform and (B) heading errors during reversal acquisition. N ote that those groups receiving diazepam (saline -> diazepam-SD, Diazepam , Switch) have greater distances and heading errors.
Figure 1.7: The percentage distance in the correct q u ad ran t (A), swim speed (B), and post-sw im Tc (pre-swim - post-swim; C) during the final probe trial. N ote th a t the saline treated groups (Saline and DS), b u t not the groups receiving diazepam (Diazepam , SD, Switch), dem onstrate a preference for the correct quadrant, despite com parable sw im speeds and post-swim Tcs. *p<0.01 com pared to chance level (25%).
a> o c to *-< v> Q w E o "U 0) a>
a
co
5 co ■ Diazepam K SD 11 Switch E3 Saline [3 DS l-'jyince_ O 0.1 -(3 0.0 -ID O 0.1 -o h0.2 -E 5 0.3 -(0 <-• tn -o.< -o a. 0.5-increase when diazepam was adm inistered, it is likely that the Switch group continued to use the sam e efficient strategy. Further, the D iazepam group was n o t im paired w hen required to navigate to single, visible platform . These findings suggest that the diazepam -treated ra t can perceive and use distal spatial cues to locate the platform.
The BZ-induced acquisition im pairm ent in the MWM m ay be d u e to h y p o th erm ia. S tudies have d em o n stra ted th a t d iazep am can in d u ce hypotherm ia (present results, Z arrindast & Dibay an, 1989), and hypotherm ia alone can produce anterograde am nesia (Richardson et al., 1983). Hence, the combination of both drug-induced and water-induced hypotherm ia m ight be sufficient to im pair spatial learning. H ow ever, three results argue against this interpretation: [1] the pre-sw im hypotherm ia rem ained consistently low th ro u g h o u t acquisition < ven th o u g h the anim als acquired the platform location (Fig. 1.3A), [2] the body tem perature in the Diazepam group did not decrease d u rin g sw im m ing (post-sw im Tc) m ore than that of controls (Fig. 1.3B), and [3] the Diazepam group was im paired but not hypotherm ic on all three probe trials (Figs. 1.4C, 1.7C). Further, the average change in the body tem perature for the groups receiving diazepam w as -0.35C (±0.1) prior to sw im m ing a n d -0.21 °C (±0.2°C) after sw im m ing. These d ro p s in body tem perature are not as severe as those previously found to induce amnesia, which typically exceed 5°C above or below norm otherm ia (Richardson et al., 1983). Together, these results suggest that hypotherm ia did n ot produce the observed anterograde amnesia.
Previously, the amnesic effects of BZs have been attributed to state- dependent learning (Patel et al., 1979), but such was not the case here. Here, rats adm inistered diazepam d uring both acquisition and retrieval (Diazepam group) were im paired w hereas rats trained u n d er saline an d sw itched to diazepam (Switch group) were not. Furtherm ore, the group w hich displayed the m ost proactive interference (as indicated by perform ance on first day of reversal training) in the reversal phase w as the one th at had been trained u n d er saline and then reversed to diazepam . These results are opposite to w hat would have been predicted by the state-dependent learning hypothesis (Overton, 1974).
Although the diazepam groups eventually learned to swim to the subm ergeo -*?atform in both the initial acquisition an d reversal phases, it w as clear th a t these rats never acquired its spatial location. This was dem onstrated in the probe trials, in w hich the diazepam -treated rats sw am random ly about the pool, failing to concentrate their search in the correct quadrant. Indeed, it appeared that diazepam produced a total anterograde amnesia and the rats adopted alternative strategies to locate the platform . For exam ple, this m ay have included a ’taxon' strategy, such as sw im m ing tow ards or aw ay from a single cue, or a 'praxis' strategy, such as swim m ing in a particular pattern (e.g., sequence of loops). Support for the latter strategy comes from the finding th at diazepam -treated rats typically sw am in a circular pattern until eventually bum ping into the platform . This 'praxis' strategy can be seen in the illustrative sw im paths draw n from the first probe trial (Fig. 1.5). These findings suggest th at diazepam produced a severe and p ersistin g an te ro g ra d e am nesia w hich n ecessitated the a d o p tio n o f a response-based search strategy.
In sum : tary, BZs im pair spatial learning, b u t not spatial recall, in the MWM. This im pairm ent does not appear to be the result of BZ-induced im p a ir m e n ts o f m o to r ic a l e f fic ie n c y , p e r c e p t i o n / a t t e n t i o n , retention/retrieval, or m otivation (anxiolysis). Further, the deficit cannot be accounted for by hypotherm ia or state-d ep en d en t learning. Finally, BZ- treated rats appear to ad o p t non-mnemonic search strategies to com pensate for their im pairm ent. In conclusion, the MWM appears w ell su ited for assessing the amnesic actions of BZs.
N eurochem istrv of Benzodiazepines
In o rd er to understand how BZs im pair spatial learning, it is first im portant to u n d erstan d the neurochem ical m echanism s by w ith BZs exert their effects. There is now little d o u b t th a t BZs exert th e ir actions by enhancing the actions of the inhibitory neu ro tran sm itter y -a m in o b u ty ric acid (GABA). Hence, the follow ing section will review w h at is currently known about GABA as well as the interactions betw een GABA and BZs.
GABAergic cell m orphology and localization
GABA is a four carbon am ino-acid w hich functions as a m ajor inhibitory neurotransm itter in the m am m alian CNS. Only trace am ounts of GABA are present in peripheral nerve tissue (Cooper, Bloom, & Roth, 1986). G lutam ic acid decarboxylase (GAD, an anabolic enzym e of GABA) im m unohistochem istry has revealed th at GABA is d istributed throughout the rat CNS, though there are regional variations (Mugnaini & Oertel, 1985). H igh d en sity (G A D -positive cells rep resen tin g >90% of total neurons) regions include the septum , am ygdala (central nucleus), nucleus accumbens, corpus callosum , some regions of the hippocam pus and dentate gyrus (see below), suprachiasm atic nucleus, and certain laminae of the cerebellum. Low density (GAD-positive cells representing <15% of total neurons) regions include b ed nucleus of the anterior com m isure, m ost of the thalam us, m am m aliary bodies, locus coeruleus, and the reticular form ation. W ithin the h ip p o cam p u s p ro p e r (CA1-CA4), GABAergic n eu ro n s and axonal varicosites are differentially distributed, w ith the greatest density of GAD- positive cells an d term inals located in the stratum lacunosum -m oleculare and stratum radiatum ; m oderate densities are located in the pyram idal and oriens layers (Babb et al., 1988; Berger et al., 1977; C am rani et al., 1986; M ugnaini & Oertel, 1985; Penny et al., 1981). In the dentate gyrus, the greatest density of GAD-positive cells and term inals are located in the hilus (60% of hilar neurons are GAD-positive) and m olecular layer, with a low er density in the stratum granulosum (cell body; M ugnaini & Oertel, 1985).
GABAergic neurons are both local circuit neurons (interneurons) and projection n eurons. G A D -positive cells are n o n-pyram idal in tern e u ro n s (Storm-M athisen, 1972), projection neurons (Seress & Ribak, 1983) as well as neurons th at have both local and distal innervations (Schwerdtfeger & Buhl, 1986). The m o rp h o lo g y o f th e G A B A ergic in te rn e u ro n s is m o re heterogenous th an once thought. In the hippocam pus pro p er, pyram idal basket cells, in verted basket cells, an d horizontal basket cells have been identified in stratu m oriens; stellate cells an d pyram idal-like aspiny cells have been identified in the stratu m rad iatu m (CA1 and CA3; Knowles & Schw artzkroin, 1981; Lacaille et al., 1987; Schw erdtfeger & Buhl, 1986; Seress & Riback, 1985). In the d en tate g y ru s, four d ifferen t GABAergic cell
m orphologies have been identified: pyram idal basket cell, fusiform basket cell, horizontal basket cell, inverted fusiform cell (Seress & Riback, 1983). Despite their m orphological vai lability, these interneurons all have short, locally arborizing axons that form a dense plexus aro u n d the som ata and 'appendages' of principle cells (granule in the dentate gyrus and pyram idal cells in hippocam pus proper).
Three popu latio n s of GABAergic projection n eu ro n s have been identified that are afferent to the hippocam pus. The first is a small band of fibers that originate in the entorhinal cortex (layers II and III), travel via the perforant path (m ed ial/lateral n o t specified) and term inate in th e dentate gyrus (G erm roth et al., 1989). The second b an d of fibers originate in the medial septum and term inate in the CA1, CA3 and dentate gyrus (Chronister & DeFrance, 1979; Kohler et al., 1984). The third m ajor b an d of projection neurons, arising largely from the hilar region, innervates the contralateral hippocam pus (CA1, CA3, dentate gyrus; Seress & Riback, 1983; Seroogy et al., 1983).
The main action of GABAergic intem eurons is to m ediate presynaptic and postsynaptic inhibition, recurrent a n d /o r lateral inhibition, an d feed fo rw a rd in h ib itio n . In m o st b ra in re g io n s, p o sts y n a p tic in h ib itio n predom inates and is m ediated by axodendritic and axosomatic synapses onto principle cells (Alfer & Nicoll, 1982A; Ben-Ari e t al., 1981; F rotscher & Zimmer, 1983; Frotscher, 1989; M uller & M isgeld, 1990; Schw artzkroin, 1986; Seress & Riback, 1983; Soriano & Frotscher, 1989; see Buzsaki, 1984 for a review) and local circuit neurons (Bilkey & G oddard, 1985; Gulyas et al., 1991; Krnjevic et al., 1988). A n exam ple of GABA-mediated recu rren t inhibition can be dem onstrated in the dentate gyrus w hen paired pulses are applied to the perforant path. The first pulse fires granule cells, which trigger recurrent inhibition and depress the response to a second pulse if it is elicited in tem poral proxim ity (<100 m ses; p aired -p u lse d epression). P aire d -p u lse depression is augm ented by GABAergic agonists (muscimol) and blocked by GABAergic antagonists (bicuculline; Albertson & Joy, 1987; Tuff et al., 1983). GABAergic projection n eu ro n s m ed iate in h ib itio n a t d ista l sites. For exam ple, the GABAergic com ponent of the septohippocam pal projection
in n erv ates in h ib ito ry in te rn e u ro n s (F reund & A ntal, 1988), p ro d u cin g 'disinhibition' w hen activated (e.g., Krnjevic et al., 1988).
GABA release
Neurochem ical and pharm acological studies have revealed that GABA release is activated by several different trigger m echanism s. For example, GABA release is activated by excitatory am ino acid (EAA) neurotransm itter glutam ate (Fonnum, 1984). [3HjGABA release from hippocam pal neurons in prim ary cell culture can be stim ulated by glutam ate as well as by N -m ethyl- D-aspartate (NMDA) and kainate (Drejer e t al., 1987; H arris & Miller, 1989). [3H]GABA release induced by glutam ate, NMDA and kainate is blocked by the NMDA receptor antagonists CPP and D-AP5 (Drejer et al., 1987; Harris & Miller, 1989). EA A-induced [3H]GABA release can be achieved by either a C a2+-d.ependent or Ca2+-in d ep en d en t m echanism . For exam ple, NMDA- induced [3H]GABA release is inhibited by the removal of external Ca2+ while the [3H]GABA release evoked by glutam ate and kainate are unaffected by the rem o v al of e x tern a l C a 2+ (H arris & M iller, 1989). M oreover, the d ep o larization of neurons (in vitro) by high external K+ levels elicits the release of [3H]-GABA from pre-synaptic term inals by a Ca2+-d e p e n d e n t m echanism (Fan et al., 1982; Pin e t al., 1988; Srinivasan et al., 1969). Ca2+- in d e p e n d e n t [3H]GABA release has been su g g ested to occur by the depolarization-induced reversal of electrogenic, N a+-coupIed GABA uptake (Nelson & Blaustein, 1982).
The synaptic actions of GABA are term inated by reuptake into terminals and glia cells (Krogsgaard-Larsen, 1980). Newly taken up GABA is stored in at least tw o different pools, one of which includes new ly synthesized GABA (Abe & M atsuda, 1983).
GABA receptors
GABA released from presynaptic terminals binds to stereospecific, high affinity pre- and postsynaptic recognition sites (Enna et al., 1977; Harrison et al., 1988). C om bined evidence from pharm acological an d physiological stu d ies su p p o rt the existence of at least two m ajor subclasses of GABA receptors, term ed GABAa and GABAg. While both receptor subtypes are
rec e p tiv e to GABA a n d b o th m e d ia te n e u ro n a l in h ib itio n , th e pharm acology, physiology, an d anatom ical d istrib u tio n of these tw o receptors is quite d istin ct However, as there is no evidence th at BZs interact with GABAb receptors, they will not be discussed here.
G A BA a receptors. Initial characterization of th e GABAa receptor show ed that it consisted of tw o glycosylated polypeptide subunits, designated a and P (see Stephenson, 1988 a n d Villar et al., 1991 for review s). Subsequently, molecular cloning of GABAa receptor cDNAs has led to the identification of several closely related 042,3 and P1.3 subunit variants as well as tw o addition subclasses, Y2 and 8 (Senke et al., 1991; Olsen & Tobin, 1990; P ritchett et al., 1989; Schofield et al., 1987). Each su b u n it h as a transm em brane dom ain (oligom er), one or m ore of which co n tributes to th e w all of th e Cl" ionophore. Each oligom er contains at least one copy o f an a and P su b u n it and m ay com bine w ith o th er oligom ers (e.g., a8; O lsen & Tobin, 1990). Channel sensitivity to GABA depends on both the com bination of subunits (e.g., 014PY > Pycxi or a2), the m olecular w eight of in d ividual subunits (58- kilodalton [kDa] P > 56 kDa P) as well as the concentration of GABA at the receptor (Yasui et al., "985). In general, how ever, th e recognition site for GABA, as well as picrotoxin (Sigel et al., 1989), requires the presence of a P subunit (Schofield et al., 1987).
The relative distribution of th e subunits differs anatom ically, as determ ined by N orthern blot analysis and in situ hyb rid izatio n to locate subunit mRNAs (Levitan et al., 1988). For example, the highest density of a j mRNA is in the cerebellum, o ^m R N A in the hippocam pus an d 013 m R N A in cortex (Levitan e t al., 1988). W ithin the h ip p o cam p u s, <*2 m R N A predom inates in the dentate gyrus a n d CA3 region w hereas 04 is evenly distributed; 013 is present only at low levels (Wisden et al., 1988). Distribution of a and P mRNAs show considerable overlap, although there is n o t a 100 percent correspondence (Sequier et al., 1988). The distribution of y mRNAs is sim ilar to th at of d j mRNAs w hile 8 mRNAs are m ore a b u n d a n t in the cerebellum an d alm ost nil in the hippocam pus (Shivers et al., 1989). Hence, subunit distribution is heterogeneous.
Physiologically, the GABAa receptor is directly coupled to the Cl" ionophore, a transm em brane ion-conducting channel. A ctivation of the G A B Aa receptor opens the Cl" io n o p h o re, in creasing the m em brane conductance to Cl" and, to a lesser extent, other anions (e.g., Br", F"). D epending on the nu m b er of ionophores an d the concentration of Cl" anions on either side of the m em brane, GABAa receptor activation m ay h y p erpolarize (Cl" influx) or depolarize (Cl" efflux) the neuron (Bormann, 1988; Farrant e t al., 1990; Segal & Barker, 1984). Indeed, the neuronal response to GABAa receptor activation also depends on the location of the receptor. For exam ple, th e io n to p h o retic ap p licatio n of GABA to the som a of h ip p o cam p al p y ra m id a l n eu ro n s elicits h y p e rp o la riz a tio n w hile the application of GABA to basilar or apical d endrites elicits depolarization (Thalm ann et al., 1981). H ow ever, u n d er norm al conditions, the net effect will be hyperpolarization w ith the depolarization being m asked by the more pow erful somatic in p u t (Nicoll & D utar, 1989).
In sum , the GABAa receptor is a hetero-oligomeric protein complex that is coupled to the Cl" ionophore. A ctivation of the GABAa receptor increases m em brane perm iability to Cl", hyperpolarizing the cell. If the neu ro n th at is hyp erp o larized is excitatory, net inhibition ensues. If the neuron is itself inhibitory, then net excitation ensues.
Allosteric m odulation of the GABA^ receptor
BZs ex ert their effects by enhancing the efficacy of GABA at the GABAa receptor, an oligometric protein complex w hich possesses several distinct high affinity receptors. M odulators of the GABAa receptor do not act directly on the GABA receptor but, rather, influence the efficacy of the receptor to o p en the Cl" channel in an allosteric m anner (Costa, Alho, Favaron & M anev, 1989). GABAa receptors are positively m odulated by a n u m b er of different drugs, including BZs, barbiturates and ethanol, and negatively m o d u lated by inverse-agonist d ru g s such as /J-carbolines. For exam ple, BZ agonists enhance the affinity of GABA to its receptor, thereby increasing the frequency th a t GABA op en s the Cl" io n o p h o re (Costa, R o d b ard , & P ert, 1979; M arangos & M artino, 1981; U nnerstall, K uhar, Niehoff, & Palacios, 1981; Vicini, Mienville, & Costa, 1986) and /J-carbolines
