The effects of cholinergic drugs and cholinergic-rich foetal neural transplants on alcohol-induced deficits in radial maze performance in rats

Behavioural Brain Research, 43 (1991) 7-28 Elsevier BBR 01160

The effects of cholinergic drugs and cholinergic-rich foetal neural transplants on alcohol-induced deficits in radial maze performance in rats H. H o d g e s

1 y.

Allen 3, j . S i n d e n i, S . N . Mitchell ~, T. A r e n d t 4, P . L . L a n t o s 2 a n d J . A . G r a y l

~Department of Psychology (MRC Brain, Behaviour and Psychiatry Research Group) and 2Department of Neuropathology, Institute of Psychiatry, London (U.K.) 3Department of Anatomy and Cell Biology, University of Liverpool, Liverpool (U.K.) and 4Department of Neurochemistry. Paul Flechsig blstitute for Brain Research, Karl Marx University. Leipzig (Germany) (Received 28 August 1990) (Revised version received 30 November 1990) (Accepted I December 1990)

Key words: Forebrain cholinergic projection system; Foetal neural transplant; Chronic alcohol treatment; Radial-arm maze; Working memory; Reference memory; Cholinergic antagonist; Rat

Chronic alcohol (20~ v/v in drinking water for 28 weeks) impaired acquisition of radial maze spatial and associative tasks by increasing both within-trial working and long-term reference memory errors; animals with high (above the median of 100 mg/100 ml) blood alcohol concentrations (BACs) duiing treatment were significantly more impaired than those with BACs below the median. Alcohol-treated rats showed improvements in radial maze performance after treatment with cholinergic agonists (arecoline and nicotine) and disruption with antagonists (scopolamine and mecamylamine) at low doses which did not affect controls. These effects were more pronounced for working than reference memory, and not manifest with the peripherally acting antagonists hexamethonium and N-methylscopolamine. Transplants into cortex and hippocampus of cholinergic-rich basal forebrain (BF) and ventral meseneephalon (VM)foetal neural tissue improved radial maze performance of alcohol-treated rats to control level over a period of 9-12 weeks after grafting. Cholinergic-poor foetal hippocampal (I IC) grafts were without effect. BF and VM, but not HC, grafts showed dense acetylcholinesterase (ACHE) staining, tyrosine-hydroxylase staining was most pronounced in VM sections and dopamine-fl-hydroxylasestaining was minimal in all grafts. Choline aeetyltransferase (CHAT) activity was significantly reduced in cortex and hlppocampus of alcohol-treated rats, except those given cholinergic- rich transplants. Alcohol treatment also significantly reduced AChE-positive cell counts in the nucleus basalis, medial septal and diagonal band brain areas, at the sources of the forebrain cholinergic projection system (FCPS). Cortical levels ofnoradrenaline were significantly reduced in all alcohol-treated rats, regardless of transplant, whereas cortical dopamine content was significantly elevated in all rats receiving transplants, regardless of behavioural effect, but not in alcohol-treated controls. Forebrain serotonin levels were not significantly altered by grafting or alcohol treatment. These results suggest that damage to the FCPS, as shown by reduce d ChAT activity in target areas, and reduced AChE cell counts in projection areas, played an important part in the radial maze deficits displayed by alcohol-treated rats, since these animals were sensitive to cholinergie drug challenge, and cholinergic-rich transplants from two different sites in foetal brain elevated ChAT activity and restored cognitive function. In contrast alcohol- or graft-induced alterations in other transmitter systems did not correlate xvith the pattern ofbehavioural deficit and recovery.

INTRODUCTION

Chronic alcohol consumption is associated with cognitive deficits, including loss of memory, in animals a n d m a n s3"s6. In man neuropathological changes are varied and widespread, ranging from the ventricular enlargement and cortical atrophy of chronic alcoholism, to the gross diencephalic lesions found with Korsakoff's d i s e a s e 13,31"32"s3, so that it is difficult to map particular Correspondence: H. Hodges, Department of Psychology (MRC Brain Behaviour and Psychiatry Research Group), Institute of Psychiatry, Denmark Itill, London SE5 8AF, U.K. 0166-4328/91/S03.50 9 1991 Elsevier Science Publishers B.V.

functional deficits onto specific types of neuronal loss. Thus Mair and McEntee 3~ have argued that noradrenergic depletion arising from loss of locus coeruleus cells31'32 in Korsakoff's disease is responsible for memory deficits, whereas Butters 13 has proposed that they arise from loss of cholinergic basal forebrain projections, through thiamine deficiency superimposed on alcohol-induced diencephalic damage. Alcohol-induced cognitive deficits reliably occur in animals after 3-6 months of treatment 54"56 when there are no gross structural changes to the brain, though there may be subtle alterations (e.g. loss of spines on hippocampal dendrites 47'56 and some regional neuronal 1OSS4'8"55).

However there are reliable biochemical indices of changes in several neuronal systems, notably noradrenergic, cholinergic and serotonergic forebrain projections 3.7:~. Thus in animals it may be possible to investigate the contribution of discrete neuronal systems to alcohol-induced cognitive deficits, by use of selective antagonist and agonist drugs, or by transplantation of specific cell types, without the confounding influence of more widespread damage to structures which integrate several different systems, as occurs in man. This would provide valuable information about those systems which show early vulnerability to the effects of alcohol, and which may play a key role in alcohol-induced deficits. Such information would provide a basis for the development of rational treatment strategies, and early detection of alcohol-induced neuronal dysfunction, by drug challenge. Arendt et al. 3 examined the relationship between behavioural impairment, length of alcohol treatment, and neurochemical changes. In rats treated with 20~0 alcohol in drinking water, as the sole source of fluid, and killed after 4 weeks withdrawal following periods of consumption of 4, 8, 19 and 26 weeks, there was a progressive decline in cortical and hippocampal cholinergic, noradrenergic and serotonergic markers, and, after a transient increase, a less marked decline in dopaminergic indices also. Deficits in radial maze performance, tested in both a spatial ('place') and an associative ('cue') task 24, occurred in parallel with decline in these transmitter systems, being insignificant after 4 weeks, partially reversible after 18 weeks, and longlasting after 26 weeks of treatment. However radial maze memory deficits which occurred after 26 weeks of alcohol treatment were substantially reversed by cholinergic-rich, but not by cholinergic-poor foetal neural transplants into cortex, and/or hippocampus. Improvement was most marked in rats with transplants in both cortex and hippocampus, and in the spatial task, which was the aspect of performance most adversely affected by alcohol. These findings suggested that decline in cholinergic transmission, rather than in the other systems affected by alcohol, was responsible for the behavioural deficits. This would be consistent with previous evidence ~ that in amnesic patients with different diagnoses (Korsakoff's disease, Parkinson's disease or Alzheimer's disease) there was a common loss of cells in the nucleus basalis, despite differences in other types of pathology. It is also consistent with converging evidence from human post mortem studies42, lesion studies in animals 5.Z4,45:6,st,52, and from effects of cholinergic receptor ligands 14.45,46,48,5~ that acetylcholine (ACh) plays a key role in some memory processes, and with evidence that cholinergic dysfunction is

involved in the memory deficits shown by aged 5 or Alzheimer26 populations. Effects of cholinergic-rich foetal neural transplants in alleviating cognitive deficits in alcohol-treated2, lesioned ~5,23, and aged 16 animals have provided a critical test for the cholinergic hypothesis, and despite some null and negative results (see Gray et al. 2~ for a review), such transplants have proved surprisingly efficacious. The present experiments further investigated the contribution of ACh to alcohol-induced memory deficits in two ways: (1) by examining the effects of cholinergic receptor ligands on the radial maze performance of alcohol-treated rats; and (2) by seeking to replicate the findings of Arendt et al. 2 that alcohol-induced radial maze deficits are alleviated by cholinergic-rich transplants, using tissue from two different sites in foetal brain, the basal forebrain, in which primordial cholinergic cells of the nucleus basalis are located, and the ventral mesencephalon, a catecholamine-rich area which also contains cholinergic pontomesencephalic cells. Our previous findings22a3 indicated that rats with ibotenate lesions to the nucleus basalis (NBM) and medial septal/diagonal band (MS-DB) brain regions, at the source of the cortical and hippocampal branches of the forebrain cholinergie projection system (FCPS), showed long-lasting impairment in radial maze performance, and exhibited enhanced sensitivity to effects of cholinergic receptor ligands. Thus lesioned rats showed improvement with agonists (nicotine and arecoline) and impairment with antagonists (scopolamine and mecamylamine) at low doses which did not affect controls 23. These findings are consistent with evidence for sensitivity to effects of cholinergic drugs in marmosets with ibotenate NBM lesions46 or in the rats with colchicine NBM lesions 52. Alcohol treatment and cholinergie lesions both produce decreases in choline acetyltransferase (CHAT) activity in animals 2'3"51, as do chronic alcoholism and Alzheimer's disease in man 37'38,42. Although it is not clear how a decrease in a presynaptic ACh marker translates into behavioural sensitivity to cholinergic drugs, it is apparent that this sensitivity provides a good index of impaired cholinergic function, both in lesioned animals, and in Alzheimer patients. Sahakian e t a l : ~ for instance, found that Alzheimer patients were more responsive than age-matched controls to the effects of nicotine in improving attention. Thus we would expect that if the detrimental effects of alcohol on radial maze performance chiefly involve impaired cholinergic transmission, then alcohol-treated rats should manifest behavioural sensitivity to cholinergic drugs, resembling the response of rats with lesions to the FCPS.

9 The work ofArendt et al. 2"3 showed that cholinergicrich transplants dissected from foetal basal forebrain alleviated the radial maze deficits of alcohol-treated rats. It is not clear, however, whether it was the presence of cholinergic cells, or some other property of the cholinergic-rich but not the cholinergic-poor control transplant, that was responsible for the behavioural improvement, such as trophic factors (e.g. nerve growth factor (NGF) 21,35,a~ or particular proteins (Wets etal., in preparation). Moreover the efficacy of cholinergic-rich tranplants in poorly performing alcohol-treated rats does not rule out a contribution from other neuronal systems to alcohol-induced deficits. For instance Arendt etal. 3 found that the decline in noradrenaline (NA) correlated as strongly with behavioural deficits as the reduction in ACh. NA has also been associated by post mortem 3~ and pharmacological 3~ evidence with cognitive deficits in alcoholic (Korsakof0 and aged populations. We plan eventually to compare the contribution of NA and ACh to alcohol-induced deficits by pharmacological probes, and by examining effects of cholinergic-rich and noradrenergie-rich transplants in alcoholtreated rats. However, the foetal ventral mesencephalon (VM) dissection, which contains catecholaminergic cell bodies of the developing nigra and locus coeruleus, also includes a substantial number of cholinergic cells, so that use of this tissue would not distinguish between effects of cells which express dopamine (DA), NA and ACh. Since the locus coeruleus is very small (1800 ceils in all), it is well nigh impossible to obtain a clean noradrenergic-rich dissection free from adjacent dopaminergie, cholinergic and other cell types. Leaving aside potential effects of DA and other neurons, one possible strategy to dissociate effects of noradrenergic and cholinergic cells is to use the VM dissection treated so as to remove either NA or ACh cells, to show whether behavioural recovery occurs with one or other of these grafts, and, if both are effective, whether additive effects occur with untreated VM grafts; containing both cell types. Use of trypsin to dissociate cells for suspension has been shown to promote survival of cholinergic, but not noradrenergic cells 11't2, and thus may be used to obtain a NA free VM implant, with the additional advantage that cholinergic cells in this region are not associated with N G F receptors 58, and so may be compared with the BF graft in which N G F receptors are present. Selective destruction of ACh cells is at present more problematic, though it may be possible to employ a diphtheria toxin conjugate 27 in the suspension medium. However as a first step, the present experiment used trypsin to dissociate VM tissue, and thus aimed to see whether cholinergic- but not noradrenergic- rich

tissue from two different milieus sustains behaviourai recovery in alcohol-treated rats.

MATERIALS AND METHODS

The experiment consisted of 4 main stages: (1) alcohol treatment for 28weeks, and withdrawal for 4 weeks; (2) pretraining and training in the radial maze for ca. 20weeks; (3)drug testing for 8weeks; and (4) transplantation and post transplant radial maze testing. The time course of these stages is set out in Table I. An#rials Eighty male Sprague-Dawley rats, with a mean weight of 397 g at the start of alcohol consumption, and 651 g at the end, were used. They were housed in pairs, and allowed unrestricted access to laboratory chow during alcohol treatment and a fixed quantity

TABLE I Time course of the experiment IVeeks Procedure 1

Rats in at 300-350 g, housed in pairs, fed ad libitum, handled

2~-4

ALC rats given alcohol, increasing in concentration by 2~/day, to 20~, as sole source of fluid

5-12

ALC rats maintained on 20% alcohol for 8 weeks

13-34

ALC rats given water for 1 h/day, and controls given 20 ml of 8.75% sucrose (10.00-11.00 h). Otherwise ALC rats maintained on 20~ alcohol sweetened with 8.75~ sucrose, and controls on tap water

30

Blood sample taken from all rats for BAC assay (02.00-08.00 h)

34-40

ALC rats withdrawn (alcohol concentration decreased by 2~/day) and remain alcohol-free for 4 weeks

41-49

Rats put on restricted feeding (15-20 g/day) to reduce to, and maintain at ca. 85% of body weight. Radial maze pretraining

50-61

Radial maze acquisition. Rats given 4 trials/week on place and cue tasks, with the same task given in two consecutive trials, one task given in the morning, and the other in the afternoon of the next day

60-66

Radial maze pharmacological experiments

67

Transplant and sham surgery

70-81

Radial maze testing of transplant effects (two consecutive trials/task/week)

84

Rats killed, post mortem studies

10 (15-20 g/day) during behavioural testing to maintain them at a stable weight. Animals were kept on a 14-h light/10-h dark schedule (lights on 07.00-21.00 h).

Alcohol treatment Animals were given increasing concentrations of alcohol in drinking water until a 20~o solution was reached. After 8 weeks with alcohol as the sole source of fluid they were given access to drinking water for 1 h/day, and 20~o alcohol for 23 h, so that alcohol consumption was to some extent voluntary. The alcohol solution was sweetened with sucrose (8.75 mg/ml) to increase palatability, so that a stable consumption, averaging 20 ml/rat/day was maintained for the rest of the 28 weeks treatment period, with an average intake of 4.0-8.5 g/kg/day. Controls were given 20 ml of the 8.75 sucrose solution, and consumed ca. 9 g/day more chow than alcohol-treated rats, so that average calorie intake was very similar (see Table II). After 28 weeks of alcohol treatment rats were gradually withdrawn over 20days using concentrations decreasing in 2~o steps/2 days. There were no overt symptoms of withdrawal. Blood alcohol concentration (BA C) BACs were taken from each rat to provide an estimate of consumption, since individual intake could not be measured accurately in pair-housed animals. Samples were collected in week 24 of treatment in the second half of the dark period (02.00-08.00 h). Animals were restrained and the tip of the tail cut to allow collection of ca. 0.5 ml ofblood, and the cut end of the tail was dipped into aureomycin powder. Twenty microlitres from each sample was immediately shaken in 1.5 ml of 3.5~o perchloric acid, and stored at - 4 0 ~ until assay. This used a standard spectrophotometric enzymatic procedure (Sigma 332-UV kit). Samples from a group of 10 randomly chosen rats h a d been taken on 3 previous occasions (weeks 12, 16 and 20 of treatment) and assayed by the same method, to estimate stability of consumption over time. Behavioral testing A month after withdrawal rats were deprived to 85 of their free-feeding weight and fed a measured amount of food daily (15-20 g) to maintain them at weights which did not vary by more than 5g/week, and behavioural training commenced. Rats were trained to find sucrose (ca. 0.25 ml of a 25~o solution) in 4 arms of a radial maze with 8 arms (62 cm long and I0 cm wide) radiating from a central circular platform (62 cm diameter). Two identical mazes were used for the Jarrard 24 place and cue tasks, with the lower 62 cm and

the upper 112 cm from the ground. The upper maze, with good visibility for room cues (posters, cupboards etc.) was used for the place task, and the lower maze for the cue task, to enable simultaneous testing of two rats, following the procedure described in Hodges et al. 22. During pretraining rats were encouraged with food pellets scattered around the maze to explore all the arms in 5-min periods. Unlike rats trained as juveniles, these large adult animals were extremely reluctant to move around the maze, so that ca. 15 pretraining periods were required before the rats moved freely enough to begin work on the place and cue tasks. For the place task rewards were always in the same 4 arms and for the cue task each arm contained inserts of different textures (e.g. carpet, wire mesh, sandpaper) which were moved to a different arm after each trial, so that the rat had to learn the association between cue and food, regardless of cue position. The rat was placed in the centre of the maze, and allowed to explore freely, and removed either after obtaining all 4 rewards, or after 10 min. Two types of error were scored: reference memory errors when a rat entered a never-rewarded arm for the first time on each trial, and working memory errors when a rewarded arm was revisited, from which food had already been taken on that trial. Non-specific errors of re-entry into non-rewarded arms were noted separately, but were classified as working memory errors for analysis. Thus the maze assessed 4 aspects of memory: short/intermediate within-trial working memory and long-term reference memory for which arms/cues were consistently rewarded across trials, within both a spatial and an associative memory task. Effects of alcohol on acquisition of place and cue tasks were assessed in two consecutive trials (with a 5-min inter-trial interval) on each task, with one task given in the morning, and the other in the afternoon of the next day. Rats (after initial training with 2 trials/task/week, since they moved slowly) were given 4trials/week on each task and were trained for 12weeks (40trials/task) until control performance approached asymptote. After this, effects of cholinergic drugs on performance were investigated, prior to transplantation (see Table I).

Drug lrealmettls Alcohol (ALC) rats were divided into two groups, those above (High ALC) or below (Low ALC) the median BAC of 100 mg/100 ml (mg~o), obtained previously in the treatment period. Within this division they were randomly allocated to 4 groups of 8 or 9 rats, for treatment with cholinergie agonist and antagonist drugs acting on nicotinic (nicotine and mecamylamine) and muscarinic (arecoline and scopolamine) receptors. Two

11 doses of each compound were tested. Each animal received 4 treatments in all, of one agonist and one antagonist, given twice at the same dosage before the place and cue task. Sixteen High ALC rats also received one dose, before the place and cue task, of either the peripheral muscarinic antagonist N-methylscopolamine or the peripheral nicotinic antagonist hexamethonium. Since there were fewer controls, these rats re6eived a total of 14 treatments, including 3 of the 4 centrally acting compounds plus one of the peripheral agents. Controls received two treatments]week with one baseline (BL) day between any treatment, and alcohol rats received one treatment/week. However, to equate injection experience, all rats were given saline injections when not receiving drug on BL days. Tasks, doses and treatments were counterbalanced. Drugs were freshly prepared each day, and tested at the following doses: arecoline (Sigma), 0.5 and 1.0 mg[kg; scopolamine (Sigma), 0.05 and 0.1 mg/kg; nicotine (Sigma), 0.05 and 0.I mg/kg; mecamylamine (donated by Merck, Sharp and Dohme) 1.0 and 2.0 mg/kg, N-methylscopolamine (Sigma), 0.1 mg/kg; hexamethonium (Sigma) 1 mg]kg. All drugs were dissolved in saline, and injected intraperitoneally (i.p.) in a volume of 1 ml[kg, with testing 15 min later, except for mecamylamine, where a 20-min interval was used.

Transplants Following acquisition and drug trials, and 30-31 weeks after withdrawal from alcohol, many rats, particularly those with high BAC's, showed stable impairment in comparison with controls. The 40 poorest performers over the last 6trials (mean BAC = 123 + 19 m g ~ ) were selected by error scores and divided into 4 groups (11 = 10) matched for error rates. There were 3 transplant groups receiving cholinergic-rich tissue from basal forebrain (BF) or VM, or cholinergic-poor hippocampal (HC) control tiss/ae. Non-transplanted alcohol-treated rats (ALC) and controis (CON) received sham surgery to provide additional controls for transplant effects, so that there were 5 groups (BF, VM, HC, ALC and CON) in the posttransplant phase ofthe experiments. Two rats died after surgery, one from the VM and one from the HC transplant groups. The cell suspension transplant procedures followed those described by Arendt et al. 2, Bjorklund et al. 9'11,12 and Sinden et al. 5~. Sections of basal forebrain, hippocampus or ventral mesencephalon were dissected under microscope from foetuses removed by laparotomy from a dam previously overdosed with pentobarbitone. For BF grafts a l-mm 3 piece of tissue was dissected bilaterally from the basal forebrain, most caudally,

beginning at the rostral border of the hypothalamus, and most rostrally, finishing at the rostral tip of the olfactory tubercle. Embryos were at gestational day (E) 15-16 on the day of dissection with E1 being the morning that plugs were detected. For VM grafts a 1.5-2 mm 3 piece of tissue was dissected right at the point of the mesencephalic flexure, on El4, HC control grafts were taken from foetal hippocampi at El8. Embryonic tissue was dissociated by incubation for 20 min at 37 ~ with 0.25~ trypsin (Sigma) in 0.6~o glucose-saline, gently flushed through a polished Pasteur pipette, washed 4 times in this solution, further dissociated by repeated pipetting and resuspended using 10/11 of 0 . 6 ~ glucose-saline for each dissected piece of tissue. Rats were anaesthetized with 3.0 mg/kg ofequithesin, and holes drilled in the skull under stereotaxic control, to allow penetration of the needle of a 10/~1 Hamilton pipette into two neocortical and two dorsal hippocampal sites bilaterally, at the following coordinates based on bregma, with vertical measurements from dura: cortex, AP + 3.2, L + 3.5, V - 3.0 and A P - 1.0, L + 5.0, V - 2.0; hippocampus, AP - 3.3, L + 1.5, V - 3.8 and AP - 4.3, L + 2.2, V - 3.4. At each site 2 ltl of foetal cell suspension was injected over a 2-min period, and the needle was left in situ for a further 2 min. The sham CON and ALC groups were anaesthetized, and holes were drilled in the skull at the same sites as the transplant groups. After surgery au'reomycin powder was sprinkled onto the wound and the scalp sutured. The rats were given a week for recovery on ad libitum food, then food deprivation w a s resumed. Behaviourai testing commenced 2 weeks after surgery, and continued for 12weeks (2 trials/task/week).

Histology and biochemistry When behavioral testing was completed, rats were killed a year after withdrawal from alcohol, and 5 months after transplantation. Three to four rats per group were overdosed with pentobarbitone, transcardially perfused with 10~o formol saline, and brains removed for histology. The remainder were killed by cervical dislocation and the brains were rapidly removed and bisected along the mid-sagittal plane. Strips of tissue about 300/ag in weight were dissected from frontal (FC) and intermediate (dorsolateral) cortex (IC) and dorsal hippocampus (Hipp) near the cannula tracts, so as to include pieces of transplant material (which was generally visible) and adjacent tissue. These were stored at - 70 ~ for later assays. Tissue from one set of half brains was used for ChAT activity, and the other for ,HPLC analysis.

12

Histology H a l f brains were sectioned by cryostat and 2 0 - 3 0 p m sections were thaw mounted onto slides. Adjacent sections were stained with Cresyl violet to reveal Nissl substance and for acetylcholinesterase (ACHE) histochemistry. Additional sections were treated with antibodies to tyrosine hydroxylase (TH) and dopamine-fl-hydroxylase (DBH). For AChE sections were incubated 3 h at 37~ for staining according to the method of El Badawi and Shenk ~8, using acetylthiocholine as the substrate and iso OMPA to inhibit non-specific cholinesterase. TH and DBH (Eugene Tech International) staining was carried out using the biotin streptavidin bridging system.

Cell co,mrs in basal forebrahl Cell counts were made in 14 alcohol-treated rats, 12 of which had not been used in the transplant phase of the experiments, but which had shown acquisition deficits in the radial maze, and two rats which had proceeded through all phases of the experiments, in the alcohol control group (mean BAC = 90.3 + 14 mg~o for all 14 rats), and 9 non-alcohol controls. Animals were transcardially perfused with 10~o formol saline, and brains infiltrated with 30~o sucrose in buffered saline (pH 7.3) after fixation. Sections of 30 ILm were cut in the coronal plane on a freezing microtome. Mounted sections were histochemically reacted for the presence of AChE (see ref. 25) and the total number of AChE-positive cells was counted in every 5th section throughout the basal nucleus complex (i.e., MS, DB and NBM). No correction factor was used, as the nucleii counted were very small in comparison with the thickness ofthe section, so that sectioning would not be likely to lead to an overestimation of cells.

ChA T activity Tissue samples from 5-7 half brains per group were assayed in duplicate for ChAT activity as the rate of formation of [~4C]ACh from [14C]acetylcoenzyme A ([14C]AcCOA) by a method based on that of Fonnum 19. Samples were homogenized in 4 ~ NP-40 (Sigma) using 8.3 pl/mg tissue wet weight. Tissue homogenate (101d) was incubated in duplicates at 37 ~ with 10 pl incubation medium containing 0.75 M NaC1, 135 mM NaH2PO4, pH 7.4, 20 mM choline, 50 mM EDTA, 1 mM AcCOA, 0.4 mM physostigmine sulphate and 10pl of 2pCi [~4C]AcCOA (New England Nuclear, spec. act. 59.3 mCi/mM). Incubation was for 5 min and was stopped with 5 ml ofcold 10 mM NaH2PO 4, pH 7.4. [14C]ACh was extracted with ace-

tonitrile containing 20 mg/ml tetraphenylboron and counted in P P O - P O P O P toluene scintillant.

High performance liquM chromatography (ItPLC) Brain samples were sonicated in 0.1 M perchloric acid (containing 0.1 mM EDTA) and centrifuged at 13000 g for 10 min. Supernatants were removed, refrozen in liquid nitrogen and stored at - 70 ~ The HPLC system consisted of an ACS 351 series plump (HPLC Technology); an on-line degasser (ERC 3510, Erma Inc.), a Chromspher C18 cartridge column (5 pm particle size), a guard column and a saturation precolumn (all from Chrompack U.K. Ltd.). Electrochemical detection was accomplished with a LC-2A detector (BAS Inc.) maintained at + 0 . 7 5 V with respect to an Ag/AgCI reference electrode. Chromatographic separation and electrochemical detection were performed at 10 ~ The mobile phase consisted of a citrate-phosphate buffer containing 1.5 mM octane sulphonic acid, 12~o methanol and 1 mM EDTA, at pH 2.65; all constituents were Analar grade (BDH). A flow rate-of 0.5 ml/min was used. Peaks were displayed, integrated and stored using a Shimadzu C-R3A coupled to an FDD-1A disk drive (Dyson Instruments Ltd).

Statistics Effects of alcohol 'treatment and transplants were assessed b y split plot analyses of variance (BMDP), with groups as the between subjects (S) factor, and weeks of testing, task (place/cue) and error type (reference/working) aswithin S factors. Drug studies used independent dosage groups as a second between S factor. BACs were correlated with the sum of errors over the last 6 acquisition trials using Spearman's rho. Single factor ANOVAs were used for comparison of biochemical measurements (e.g. ChAT activity, levels of NA, DA, 5-HT and their metabolites) in different brain regions. HPLC data were subject to square root transformation to restore homogeneity of variance and covariance, because inclusion of host and transplant material resulted in some extreme scores. For acquisition Dunn's test was used for comparison of control with Low and High ALC groups, and of Low and High ALC groups with each other, as 3 sets of comparisons were made. However in the other experiments, where only two sets of comparisons were made based on specific predictions (either control with experimental groups or baseline with treatment scores within each group), the t-ratio, based on the appropriate error terms in the ANOVAs, was used for comparison of means. Independent t-tests were used to compare basal forebrain cell counts in control and alcohol-treated groups.

13 were 114.6 + 25, 114.0 + 37, 114.8 + 36 and 115.7 + 34 mg% for the 4 samples (i.e. including week 24), the difference between the lowest and highest score in each rat, as a % ofits average BAC across the 4 measures averaged a change of 18.9% and the rank order of BACs did not change over the 4 measures. Table II shows the mean weight and consumption of control and alcohol-treated rats over the 28 weeks of treatment. ALC rats put on weight at a slightly slower rate than controls, but their average weight never fell below 90% ofcontrol level. The mean intake of calories in C O N and ALC groups was almost identical, since controls ate more chow to balance the extra calories from the alcohol itself. However the mean total fluid consumption of ALC rats was only 60-65% of controls, since C O N rats drank more water than ALC rats drank alcohol, and also consumed all of the 20 ml su-

RESULTS

Effects of chronic alcohol treatment on weight and consumption Alcohol-treated rats drank a mean of 20 ml of 20% alcohol; from weeks 8 to 26 this was sweetened with 8.75% sucrose and rats were also allowed access to water for 1 h/day. Controls were given sucrose in this period. With this element of voluntary consumption alcohol intake appeared variable, and blood alcohol concentrations (BACs) taken in all rats in week 24 showed that levels ranged from 23.5 to 167.4 mg%, with a mean of 101.2 mg~o. Some confidence may be attached to this measure, since in a group of 10 rats, BACs taken on 3 previous occasions (weeks 12, 16 and 20) did not show significant variation over time (/73.9 = 0.04, n.s.). In this group mean BACs

T A B L E II

Effects of chronic alcohol treaonent on weight and consumption IVeeks 0

1

4

8

12

16

20

24

28

403 392 103

445 480 93

487 536 91

517 574 90

560 600 93

587 642 91

605 645 94

623 670 93

630 671 94

IVe~ht(g) ALC CON ALC a s ~ o C O N

Mean fluM intake Onl). ALC group given access to water and controls to sucrosefor lh/day from week 8 ALC Alcohol Water

19.0 8.5

19.0 8.9

20.2 8.7

19.8 8.6

21.9 8.0

27.5

27.9

28.9

28.4

29.9

22.3 20.0

27.2 20.0

27.6 20.0

25.5 20.0

26.6 20.0

Total

42.3

47.2

47.6

45.5

46.6

ALC as ~o CON

65

59

61

62

64

20.9 -

24.1 -

21.7 -

Total

CON Water Sucrose

39.4

40.8

39.2

Mean daily consumption~calorie intake T)pe

CON

K/C

ALC

K/C

Chow Sucrose Alcohol

28 g 1.75 g -

75.6 6.9 -

19 g 1.75 g 4.0 g

51.3 6.9 28.4

Total calories

82.5

86.6

14 crose made available for 1 h/day, whereas ALC rats drank only 8 - 9 ml of water in this period. This reduced fluid intake, together with the dehydrating effects of alcohol, may account for the shrivelled appearance of fecal boli from ALC rats. However rate of weight gain in ALC rats was not much affected, no withdrawal symptoms were observed during the withdrawal period, and alcohol treatment did not appear to exert any longterm effects on general health during the year after withdrawal. For instance the incidence of tumours and minor respiratory infections in CON and ALC groups was very similar. Effects of chronic alcohol treatment on radial maze acquisition Training began ca. 12weeks after withdrawal, because of the lengthy pretraining period required, and was continued for a further 12 weeks. As training progressed it became apparent that learning in the alcohol-

treated rats was very variable; some rats showed marked impairment, and others performed nearly as well as controls. Since this might relate to former differences in alcohol consumption, total errors for the last 6 trials/task (weeks 11-12) were correlated with BACs taken 6 months previously, and showed a highly sigificant positive relationship (r = + 0.83). Therefore, in order to examine the relationship between consumption and learning, alcohol-treated animals were divided into two groups, above and below a BAC of 100 rag% which was close to the median. This produced groups of 34 High ( > 1 0 0 m g % : High ALC) and 33 Low (< 100 m g ~ : Low ALC) animals. Fig. 1 shows the acquisition curves for High and Low ALC groups and controls over 12 weeks for place (Fig. la, b) and cue (Fig. lc, d) tasks (first

c:

Cue

Task:

Reference

Memory

6-

a: Place Task: Reference Memory B

L.

9

021,I

A

LOWA

0

r

IIIGII

A

t~

CCN

L.

.t

o/ L,. O

W

J

2SE

LOWA I-'I',GHA

= | J l l ~ l l l l ~ l

2 0

1.1.1

I

2-

1 2 3 4

6

1

11

3

4

5

6

7

8

9

10 11 12

WEEKS

d:

7 8 9

2

2SE

Cue

Task:

Working

Memory

12

WEEKS

cct,i

~

~

~

'

b: Place Task: Working Memory

~

1

j LOWA HIGHA

m

6-

L.

5-

1/1 m mt~

2SE o L_ L,.

ul

4=

ccz~

9

ttlGH A

3

r O L_ LU

LOW A

2 2SE

1

0 0

i

i

i

i

i

i

i

i

i

1

2

3

4

5

6

7

8

9

WEEKS

i

i

i

10 11 12

WEEKS

Fig. I. Mean number ofradial maze place (a, b) and cue (c, d) task reference and working memory errors over 12 weeks of acquisition in control (CON) and alcohol-treated rats with BACs below (LOW A: Low ALC) or above (HIGtt A: Itigh ALC) the median of 100 mg%. Training began I month after withdrawal from alcohol.

15 2 trials/week)..ALC groups were somewhat slower to learn the reference memory tasks (i.e. which arms contained rewards) than controls, and showed very marked impairment in working memory, particularly in the spatial task, where error rates in controls fell rapidly. However it is apparent that the High ALC group was more impaired than the Low, and that this was also most evident in the place task, particularly for working memory. By analysis ofvariance the difference between groups was robust (F2.74 = 45.07, P < 0.0001), and the linear trend of decrease in errors over trials, as learning progressed, was highly significant (F1o74 = 136.08, P < 0.0001). The marked interaction between groups and the linear trend of trials (F2,74 = 13.22, P < 0.0001) demonstrated the different rates of learning in the 3 groups. All rats were slower to learn the place than the cue task (Fi,74 = 13.39, P < 0.001) and made far more reference than working memory e r r o r s ( F i , 7 4 = 50.13, P < 0.0001). However alcohol groups made substantially more working memory errors than controls so that there was an interaction between groups and error type (F2.74 = 2.49; P < 0.001), with the interaction between groups, error type and the linear trend of trials ( F I , 7 4 = 5 . 3 8 , P < 0 . 0 2 5 ) showing the rapid decrease in working memory errors over trials in C O N as opposed to ALC groups. Comparisons of means of the alcohol groups with controls by Dunn's test using the within-cell error term in the ANOVA to calculate the critical difference value for 0.05 and 0.01 probability levels, showed that impairment in reference memory was relatively slight. ALC groups were marginally (P < 0.05) inferior to controls in the cue task only on weeks I0 and 11 (High ALC), and week 6 (Low ALC). In the place task reference memory impairment was more pronounced; High ALC rats differed from controls from week 7 onwards (P < 0.01, apart from weeks 7 and 8), and Low ALC rats differed on weeks 10-12 (P < 0.01). Deficits in working memory in both place and cue tasks were more substantial, though here, too, differences from controls were more marked in the place than the cue task. Working memory errors were significantly ( P < 0.01) increased in the High ALC group from weeks 4 - 5 onwards, and from week 6 in the Low ALC group. However by the end of testing (weeks 11-12) Low ALC animals had improved to the extent that they no longer differed significantly from controls, whilst High ALC rats remained profoundly impaired, particularly in the place task. Comparison of the performance of the Low and High ALC groups showed that they did not differ with respect to reference memory errors. However working memory errors revealed the increasing disparity between the two

groups towards the end of training, as the Low ALC group improved. Thus the difference between the two groups was significant from week 8 onwards in the place task (P < 0.01) where High ALC rats were most impaired, and on weeks 10 and 11 in the cue task.

Effects of cholinergic drags hi alcohol-treated rats N-Methylscopolamhle and hexamethonhtm. Fig. 2 shows that although there were significant differences between control and High ALC r a t s ( F t j 4 = 22.76 and 28.72, P < 0.0005) in the groups tested for response to N-methylscopolamine (0.1 mg/kg; Fig. 2a) and hexamethonium (1.0 mg/kg; Fig. 2b), these compounds had no effect on error rates, and did not interact with task or error type. Thus these peripherally acting cholinergic

a:

Effects

of

N-methylscopolamine

s-Place Task Reference

Cue Task

Working

Reference

Working

"BE

~

NSCOP0.1 mg.kg

~

2 2SE 1

o ~

N.C

IIN

b: Effects Place

s- I Reference

,t

~

0SN

NC

~

~

of H e x a m e t h o n l u m

Task

Cue Task

Working

Reference

Working

ttEX 1.0 mg.kg

o

2

12SE

'o

OSN ~

~

~

OSN ~

~

NI;

Fig. 2. Mean number of reference and working memory errors in control (CON) and alcohol-treated (tligh ALC)groups during baseline (BL) performance and after treatment with N-methylscopolamlne 0.1 mg/kg (a) and hexamethonium 1.0 mg/kg (b).

16 antagonists did not affect memory scores, or alter the preponderance of reference memory and place task errors consistently shown by the animals throughout testing. Arecoline. Fig. 3 shows that arecoline (0.5 and 1.0 mg/kg) substantially reduced errors in both alcohol groups, leaving the low error rate of controls unaffected. Working memory errors were more significantly reduced than reference, and improvements were somewhat more marked in the place task, where baseline error rates were higher, than in the cue task. Arecolineinduced improvement tended also to be more marked in the High than the Low ALC group, as shown most clearly by the larger reduction in reference memory errors below BL (Fig. 3a, b). By analysis ofvariance the difference between groups

was significant (/'2.23 = 19.56, P < 0.0001), indicating impairment in the alcohol groups relative to controls, but the High and Low ALC groups did not differ significantly. More errors were made in the place than the cue task (Fi,23 = 10.50, P < 0.01), and for reference rather than working memory (FI,23 = 97.05, P<0.0001). Effects of arecoline were broadly dose-related, as shown by the linear trend of doses (FI.23 = 14.42, P < 0.01). The effect of arecoline in dose-relatedly reducing errors in the alcohol, but not control groups, was shown by the interaction between groups and the linear trend of doses (/'2.23 = 3.48, P < 0.05), and the more marked effect of arecoline on working, rather than reference memory errors was shown by the interaction of error type and the linear trend of doses (F1.23 = 5.01, P = 0.05). Scopolamhle. Fig. 4 shows that scopolamine (0.05

a: Effects of Arecollne: Place T a s k s,]

Reference

a:

Effects of Scopolamine:

Working

7-

9 BL 4 9 *

[]

AREC0.5 mg,'kg

[]

A R E C 1.0 mg,'kg

Reference

Place

Working

Task

.... I

6

5

3

13L

[]

SCOP 0.05 mg.'kg

[]

SCOP 0.1 rng,'kg

i...

2SE O

"

2

"

o

9.

""

2SE

"

o i... LU

0

CON

LOA

HI A

CEN

LOA

HI A

,

O3q

,

LOA

GROUPS

tn

Reference

CCN

LOA

HIA

GROUPS

b:

b: Effects of Arecoline: C u e T a s k 4.

HIA

Effects of Scopolamine:

Cue

Task

6-

Working

3

Reference

[]

BL

[]

AREC0.5 mg:kg

[]

A R E C 1.0 mg,~,g

...

Working

....

9 BL

5-

4

3

o ,t;,

SGOP o.o5 mg,~g

[]

SCOPo.1 rng,*,g

2SE

2SE o

[]

2-

1-

0

0

CCN

LOA

HI A

COq

LOA

HI A

GROUPS

Fig. 3. Mean number of reference and working memory errors in place (a) and cue (b) tasks in control (CON), low ALC (LO A) and ltigh ALC (HI A) groups during baseline (BL) performance, and after treatment with two doses of arecoline. Difference from BL: *P < 0.05, **P < 0.025, ***P < 0.01.

CCIrl

LO A

ttl A

C~I

LO A

HI A

GROUPS

Fig. 4. Mean number of reference and working memory errors in place (a) and cue (b) tasks in control (CON), low ALC (LO A) and high ALC (I11 A) groups during baseline (BL) performance, and after treatment with two doses of scopolamine. Difference from BL: *P < 0.05, **P < 0.025, ***P < 0.01, ****P < 0.00l.

17

and 0.1 mg/kg) dose-relatedly increased error rates in both alcohol groups, where rates were high, and in controls, where they were low, and that this effect was more marked for working than reference memory errors. However alcohol-treated rats were more sensitive to the effects of scopolamine than controls, showing increases in working memory errors at the low dose which did not significantly impair controls. Moreover working memory was more disrupted by scopolamine in the High ALC group than the low, as shown by the greater increase in errors above BL (Fig. 4a, b). By analysis of variance the difference between control and alcohol groups was substantial (F2.21 = 13.88, P < 0.0001). The marked increase in working, rather than reference errors that occurred with scopolamine is shown by the interaction between error type and the linear trend of doses (Fx,21 = 16.57, P = 0.0005). A further pointer to the potent action of scopolamine on working memory was that it abolished the typical occurrence of more numerous errors of reference, rather than working memory. Place errors were more numerous than cue (Ft,21 = 7.01, P < 0.02), and scopolamine did not interact with task. Nicothze. Fig. 5 shows that nicotine (0.05 and 0.1 mg/kg) dose-relatedly reduced the high error rates of both alcohol groups, particularly for working memory, but did not affect the performance of controls. The High ALC group made more BL errors than the Low, particularly in spatial working memory (Fig. 5a). However nicotine, appeared to be less effective in reducing working memory errors in alcohol4reated rats than arecoline, and the High ALC group was more resistant to the action of nicotine than the low. Thus working memory errors, which fell to control level after treatment with arec01ine, remained significantly above control baselines in High ALC animals following nicotine treatment. By analysis ofvariance the difference between gro/aps was highly significant (Fz.zl = 28.67, P < 0.0001), reflecting differences between control and alcohol groups, and between the two alcohol groups. The effect of nicotine in dose-relatedly reducing errors in alcohol, but not control groups was shown by an interaction between groups and the linear trend of doses (F2,21 = 7.52, P < 0.001). There was no overall difference between tasks, but there were substantially more reference than working memory errors (Fi.2~ = 130.63, P < 0.0001). Despite the apparently more marked effect of nicotine on working than reference memory errors (see Fig. 5a, b), there was no two-way interaction between error type and dose. However dose-related decrease in errors was more clear cut in the place than the cue task, in working rather than reference memory, and in alco~

Effects of Reference

a:

Nicotine:

Place

Task

Working :..

m r 9:-

9

BL

[]

NIC 0.05 mg,l~g

[]

NIC 0.1 mg;kg

3.

2SE

L%I

H ~

O LU

LOA

120'4

HI

A

O~N

IDA

HI A

GROUPS Effects of 5 Reference b:

Nicotine: Cue T a s k

Working 9

BL

[]

NIC 0.05 mg,~g

[]

NIC 0.1 mg,'kg

4

ttl I

--

~

t..

2SE

U,I !

0

9

LOA

HIA

~

LOA

HIA

GROUPS Fig. 5. Mean number of reference and working memory errors in place (a) and cue (b) tasks in control (CON), low ALC (LO A) and high A L e (HI A) groups during baseline (BL) performance, and after treatment with two doses of nicotine. Difference from BL: 9 P < 0.05, **P < 0.025, ***P < 0.0l.

hol rather than control groups, so that the quadruple interaction between task, error type and groups with the linear trend of doses approached significance (F2,21 = 3.09, P = 0.06). Mecano,lamine. Fig. 6 shows that mecamylamine (1.0 and 2.0 mg/kg) marginally increased the low error rates of controls, notably in the place task, whilst it substantially increased the high rates of the alcohol groups, particularly in working memory. BL error rates in Low and High ALC groups were roughly equivalent, and mecamylamine exerted comparable effects in both groups, apart from a much more substantial increase in cue-working errors in the High than the Low group. Analysis of variance indicated that the difference

18

Effects of foetal cell transplants

a: Effects of M e c a m y l a m i n e : Place Task Working

Reference

9

BL

[]

MEC 1.0 rng~,g

[]

MEC 2.0 mg'kg

cq

2SE

=_ LU

~

IDA

HI A

CCN

LOA

HI A

GROUPS b: Effects of M e c a m y l a m i n e : Cue Task s.

Reference

Working ==.~

5.

4.

E c~

dd

3.

LOA

HI A

CEN

LOA

9 eL []

MEC 1 .o mg,'kg

[]

MEC 2.0 mg/kg

2 SE

HI A

GROUPS

Fig. 6. Mean number of reference and working memory errors in place (a) and cue (b) tasks in control (CON), low ALC (LO A) and high ALC (HI A) groups in baseline (BL) performance and after treatment with two doses of mecamylamine. Difference from BL: * P < 0.05, **P < 0.025, ***P< 0.01, ****P< 0.001.

between groups was reliable (F2.2, = 8.73, P < 0.002), and there was also a highly significant linear trend of increase in errors with dose (F~,zl = 32.33, P < 0.0001). Despite the more significant increases in errors in ALC than C O N groups (see Fig. 6a, b) the interaction between groups and the linear trend of doses was not significant. The typical pattern of more reference than working memory errors (Fi.21 = 14.96, P < 0.001) and more place than cue errors (F1.21 = 10.00, P < 0.005) was not modified by mecamylamine. However, since the High ALC group showed a more marked increase in working memory errors in the cue than the place task, whilst the reverse pattern occurred in the Low group and controls, there was a marginal interaction between groups, task and error type (Fz.2~ = 3.49, P < 0.05).

Fig. 7 a - d shows the performance in place and cue tasks of 5 groups of rats (BF and VM with cholinergicrich transplants, H C with cholinergic-poor grafts, and sham-operated C O N and ALC control groups), over 12 weeks of post-transplant testing. In the C O N group reference memory errors were relatively high following the 2 week break for surgery, but fell to preoperative levels after 5 - 6 weeks, whilst working memory errors hovered at one per trial, as in asymptotic performance. The two cholinergic-rich transplant groups (BF and VM) showed clear improvement over the 12 weeks of testing. Thus from week 7 (i.e. week 9 post-transplant) the BF group did not differ from controls on any of the 4 measures. Improvement was slightly more delayed and less marked in the VM group, where error rates fell to C O N level from weeks 8 - 9 (i.e. 10-11 weeks posttransplant). In contrast to the linear trend of improvement in the BF and VM grouPs , the control transplant (HC) and ALC groups showed variable error rates from week to week, with no sign of consistent improvement to C O N levels. Apart from weeks 1-4 for Place-Reference errors, H C and ALC groups differed reliably from controls throughout testing. Working memory errors in particular were substantially ( P < 0 . 0 1 ) increased above control level in each week of testing, except for weeks 8-10 (P < 0.025) in the ALC group. Analysis of variance showed that groups differed significantly (F4,42 -- 12.53, P < 0.0001), and that there was a substantial linear decrease in errors over trials (F~.az = 59.02, P < 0.0001). However the marked interaction between groups and the linear trend of trials (F4.42 = 10.07, P < 0 . 0 0 0 1 ) underlined the different patterns of recovery or non-recovery in the 5 groups. There was also an interaction between error type and groups (F4,42 = 3.30, P < 0.025), occasioned by the more rapid decline in working than reference memory error rates in BF, VM and C O N groups, relative to high error rates in H C and ALC groups. This was also manifest in the interaction between trials, error type and groups (F4,az = 2.93, P < 0.05).

Post mortem studies Histology. Under light microscopy no obvious structural damage was observed in alcohol-treated rats. However cell counts (see Table III) showed that the number of AChE-positive cells was reduced to 7 0 ~ of control level in the MS area (t2~ = 8.51, P < 0.0001), to 79~o in DB (tzl = 5.12, P < 0.0001), and less substantially to 90~o in the N B M (t2, = 2.40, P < 0.025). Grafts had been placed bilaterally in both cortex and hippocampus, and all 8 brains from the 3 transplant groups (2 or 3 per group) contained graft tissue. How-

19

Transplant Effects: Place-Reference

a:

c: Transplant effects: Cue-Reference

5-

5-

BF - ~

VM

~

~C

,I,

3

VM 1-13

O4 O4

ALC

tO L.. L..

I,.,

IJJ

2

9

LLI

2SE 1

0

1

2

3

i

i

*

i

i

l

4

5

6

7

8

9

i

l

I0

CQN

2SE

l

11 12

1

WEEKS

2

4

5

6

8

9

10

11 12

WEEKS

b: Transplant Effects: Place-Working

d: Transplant Effects: Cue-Working

7-

5"

64"

r.

S"

13F

2

VM

"z-

m

,Ik

4'

9

HC O4

"~

w

=

BF

3"

--

~

VM

r

3-

ALC

2-

CCIJ

:L0 6 6

co

i.

2"

o

~

ILl

9

CEN

1 1-

I 2SE

2 SE J

0 0

i

i

i

i

1

2

3

4

D

5

1

~

i

i

i

6

7

8

9

10

~

0

i

11 12

WEEKS

WEEKS

Fig. 7. Mean number of reference and working memory errors in place (a, b) and cue (e, d) tasks in sham-operated control (CON) and alcohol-treated (ALC) rats, and "alcohol-treated animals

receiving transplants of cholinergic-rich basal forebrain (BF) or ventral mesencephalon (VM) foetal tissue, or cholinergic-poor hippocampal (He) tissue.

ever two transplants into cortex (one from the VM and one from the HC group) and one into hippocampus (from the HC group) were not visible on one side of the brain. Hippocampal transplants were larger than cortical, and in several cases distorted the intrinsic architecture of the brain (see Fig. 9a, c). In 3 instances these

hippocampal transplants spread either upwards into cortex from hippocampus, through the cannula tract, or downwards into the dorsal thalamus. Cortical transplants remained bounded by the corpus callosum, but two had grown up through the cannula tract to form a sheet over the surface of the frontal cortex (see Fig. 8c). Fig. 8a, c and d shows examples of AChE staining in HC, BF and VM cortical grafts. AChE-positive cells were abundant in both B F and VM, but not in HC sections, though Nissl staining revealed the presence of many other cell types in HC grafts (Fig. 8b), which were comparable in density to Nissl-stained cells in the cholinergic-rieh grafts. Fig. 9 shows AChE stained sections from hippocampal BF (Fig. 9a), VM (Fig. 9b) and HC (Fig. 9c) grafts, where the contrast between cholinergic-rich and -poor tissue is equally apparent. In BF grafts staining was

TABLE 11I Mean cholinesterase-positive cell count (+ S.D.) hz control and alcoholtreated rats in basal forebrahz

A L C ( n = 14) CON (n = 9)

),IS

DB

NBM

3194_+ 284** 4577 + 498

9655 +_ 731"* 12268 + 1696

9263 + 851' 10341 _+ 1309

Difference from control: *P < 0.025, **P < 0.000l.

20

m

1

--.

-

. .."

~, . . . . . . . . . ,

:..

. ".:7,

,.

9

.

~ ,

.

, ~

.

~

[

:

,',.

9

,':4*:

:"

,

.

',

: "'. ,~,

~

.

9 "

~

l

t."

.

:

"" :.:, "-:" ....

,--'.~"

-

:~

-.

"

,*

. 1 .

-,*

.

"-

:.

.

.-

t"~

3,.-.i,

~.1~"

.

.

.

.

.

~

:,.?,-.-..,-.

,:

','-

+ .'

.-411~

.~ .

"'" ~'.! "'"" ~" "" ",i,~ " "~'" "" 9: . . . . , ; - . - : ' -

.

.

,~

" ~

~~.

.

bQ~:,~.

.

' "k

i

;;

. ~.

.

"

.-v:i~.

""-

'd~7~.,

9

~~ . ' "

, , ~- -

..'~

.

/~' " ,

.

~'.-:'~'

, '.

. . . . . "t,"

, :-

t5

"

"

"~ .... .

, ,

4~

:

....

",

.

. . . .

Jl*

"~

.

. . . .

. "" , ..-,"*" .... ,.

""

. ,: ,,,,~

9 .'.

*

.

'

,~ .

.

"l i

. ": " .~,::', "':'~ -.

~ -

, t

. J

C

i

i-~ ":-. ..i

""

Fig. 8. Coronal cryostat sections of brain from rats in these experiments stained for cholinesterase 5 months after grafting of foetal hippocampal (HC: a), basal forebrain (BF: c), or ventral mesencephalon (VM: d) tissue into the neocortex. Dense staining is visible in the cholinergic-rich BF and VM grafts (c, d) but not the HC graft (a). Fig. 8b shows the cellular composition by Nissl staining of the HC graft shown in Fig. 8a. Scale bars represent 200 Bm.

patchy and unevenly distributed in intense clumps through the transplant (see Fig. 9a). In VM transplants staining was dense and uniform, clearly delineating the transplant boundaries. Within this neuropil ACHEpositive cells were clearly visible (Fig. 9b). Fig. 10 shows that TH-positive cells were more numerous in VM (Fig. 10b) than in BF (Fig. 10a) sections. HC, like BF grafts, showed minimal TH staining. Virtually no DBH-positive cells were found in any type of transplants examined; in particular there was no evidence for the presence of noradrenergic cells in the

sections from VM grafts, consistent with their destruction by trypsin. C h A T activity. Fig. 11 shows ChAT activity in 3 brain regions in the 5 groups of transplant and control rats. In FC activity was significantly reduced in HC and ALC groups (to 79 and 6 9 ~ of control level, respectively), whilst it was significantly increased in the BF, and non-significantly increased in the VM group, relative to controls (to 124 and 113~ of control level, respectively), so that there was a reliable difference between groups (F4.3o = 4.06, P < 0 . 0 1 ) . A similar, but less

21

~-)~".:,-.~."~,r>,

~"~ "" I~.-

.~-~:- ~

~,.-~:~..-.::."

.

~-_ ,r g~.:.''.

:

~-:. "_'

. - - ..': : , C

. . ~ . :... . .:....~:;L;..

"

" . . . . . . "~'

~

;'~I ;

.m'l

41"

! ....

'

; '

J~

"

'" '

.... "

! " -

"

,.-.

" :'':

-~:'

' "

.

"~ :,.~)-~

: ~

~t~ :

-.

..;~,.'-.~.. ~:::,:~ ~--'''.,'~L:-- I

.,, :

.

i..:.iil.:--;,:'i?:i;

."':~..:...:: <~..j.

'.:_iii-

.'::: i

c "."

i

~

-

Fig. 10. Tyrosine hydroxylase-stained sections from grafts sited in hippocampus of foetal basal forebrain (a) or ventral mesencephalon (b) tissue, showing the presence of TH-positive cells, marked with arrows, in the VM, but not the BF graft, where TH staining was very sparse. Scale bars represent 800 pm for a and 200 ym for b.

z.

+

Fig. 9. Cholinesterase-stained sections, as in Fig. 8, from transplants sites in hippocampus of foetal basal forebrain (BF: a), ventral mesencephalon (VM: b), or hippocamp.al (HC: c) tissue, showing reactivity in the BF and VM, but not the HC grafts. Scale bars represent 400 ym.

marked pattern was found in IC and Hipp. Activity in both the HC and ALC groups was reliably reduced (to 87 and 79~o of control level in the HC group and to 80~o and 85~o of control in the ALC group). Activity in the BF group remained above control level (114 and 102~o), and close to control level in the VM group (95 and 90~o of control). The overall difference between groups was significant in IC (F4.3o = 2.82, P < 0.05) and approached significance in Hipp (F4,3o = 2.48,

P = 0.06). Significant differences by the t-ratio with respect to controls are shown in Fig. 11. HPL C analysis of amine and metabolite levels and utilization ratios (UR). (1) NA. Table IVa shows that NA content in FC was marginally reduced in all alcohol groups relative to controls, regardless of transplant. This depletion was significant in the 3 transplant groups at the 0.05 level (BF, t28 = 2.172; VM, t28 = 2.127; HC, t28 = 2.129), but not in the ALC group. In IC, NA depletion was more marked (F4.3o = 6.09, P = 0.001), and concentrations in all alcohol-treated groups were highly significantly (P<0.0005) below control level (smallest t3o = 4.03, P = 0.0005, for the ALC group). In Hipp NA depletion in alcohol-treated groups was not significant.

22 Choline 20-

[--

acetyltransferase

activity 9

TABLE IV

BaALC [] H~ []

BF

[]

VM

2SE 9

o

0

R3

HIPP

Regions Fig. 11. Mean choline acetyltransfcrase activity in sham-operated control (CON) and alcohol-treated rats (ALC), and alcohol-treated rats receiving transplants of foetal basal forebrain (BF), ventral mesencephalon (VM) and hippocampal (HC) tissue, in frontal cortex (FC), intermediate cortex (IC) and hippocampus (IIIPP). Difference from control: *P < 0.05, **P < 0.025, ***P < 0.01.

(2) DA. Table IVb shows that DA content in FC and IC, in contrast to NA, was significantly elevated in all transplant groups, relative to controls (FC, F4,u8 = 3.27, P = 0 . 0 2 5 ; IC, F4,3o=4.19, P < 0 . 0 1 ) , and that this increase was comparable (P < 0.025) in all 3groups in FC whether or not they had shown behavioural recovery (smallest t28 = 2.586 for the VM group). However in IC the VM group was the only one to differ reliably from control level (t3o = 2.904, P < 0.01), and in Hipp only the HC group showed a marginal increase (/3o -- 2.05, P < 0.05) above control level. (3) Dihydroxyphenylacetic acid (DOPAC). In parallel with DA increases, DOPAC levels were also elevated in FC and IC (see Table IVc) in transplant groups. For FC (F4,28 = 3.33, P < 0.025) the effect was reliable in all 3 transplant groups, and most marked in BF rats (smallest tub = 2.48, P < 0.025 for the HC group). In IC (F4.3o = 4.08, P < 0.01) only the VM transplant group (/3o = 3.174, P < 0.005) differed reliably from controls. In hippocampus the increase in DOPAC content was significant in the VM group (/3o = 2.887, P < 0.01) and

Mean levels (+ 2 S.E.M.) hz 3 brain regions of NA , D A . D O PA C. H VA, h, ng/g tissue, and the dopamine utilization ratio (DA- UR) in control and alcohol-treated rats, after transplantation Groups

(a) NA BF VM HC ALC CON 2 S.E.M. (b) DA BF VM HC ALC CON 2 S.E.M. (c) DOPAC BF VM HC ALC CON 2 S.E.M.

Brain regions FC

IC

Hipp

155.3' 159.8* 153.3' 201.1 222.9

89.6~ 89.9 w 91.0 ~ 130.4"** 233.5

193.5 197.0 177.1 196.8 229.8

61.3

954.3** 857.8** 1158.3'* 111.4 24.1

58.6

53.6

33.4 797.5*** 306.7 9.5 100.8

5.5 12.0 23.8* 2.6 2.0

802.6

518.9

18.8

194.2* ** 169.2** 169.6'* 30.4 13.0

I ! .4 204.3*** 67.4 3.4 18.1

2.9 6.8*** 4.2* 2.7 1.4

150.0

140.5

2.6

(d) HVA BF VM tIC ALC CON 2 S.E.M. (e) DA-UR BF VM tIC ALC CON 2 S.E.M.

81.0"** 69.4** 115.7'** 22.4 12.2

58.4

14.5 72.2 39.1 8.4 26.5

4.6 5.2* 7.2*** 3.5 3.4

65.0

2.1

0.61'** 0.39 ~ 0.48*** 1.13 1.69

1.41 0.73 0.80 1.36 1.20

1.37 1.00 3.92 0.53 1.63

0.69

0.69

2.86

Difference from control: *P < 0.05, **P < 0.025, ***P < 0.01, ~P < 0.0001.

in t h e H C g r o u p (/3o = 2.05, p < 0.05).

(4) Homovaniilic acid (HVA). In FC (F4.us = 3.93, P = 0 . 0 1 ) , IC (F4.3o= 3.53, P < 0 . 0 2 ) and Hipp (F4.29 = 3.11, P < 0.03) HVA content was significantly increased in all transplant groups relative to CON and ALC groups, which did not differ (see Table IVd). The most significant increases were in FC (smallest t28 --- 2.542, P < 0.025 for the VM group). In IC no

group differed from control level, whilst in Hipp the increases in both the VM (t29 = 2.137, P < 0.05) and HC (t29 = 3.257, P < 0.01) groups were reliable. (5) Dopamine utilization ratio (DA-UR). URs give a measure of turnover, and thus of neuronal functional activity. From Table IVb, c and d it is apparent that

23 levels of FC DA were increased relative to metabolites in the 3 transplant groups, so that the D A - U R was significantly below control level in FC (F4.28 = 6.22, 9P < 0.001 ; smallest tz8 = 3.349, P < 0.003 for the BF group). In IC and Hipp this disparity was less marked, so that there were no significant decreases below control level (see Table IVe). (6) Serotonin (5-HT), 5-hydroxyindoleacetic acid (5-HIAA) and the 5-HT-UR. Concentrations of 5-HT and 5-HIAA rarely differed from control levels in any of the experimental groups, apart from an increase (t28 = 2.487, P < 0.025) in FC 5-HT concentration in the HC group, attended by an increase in FC (t28 = 2.116, P < 0.05) 5-HIAA. 5-HT-URs were reduced in all alcohol groups in frontal cortex, an effect reliable only for VM (t28 = 2.275, P < 0.05) and ALC (t28 = 2.259, P < 0.05) groups. There were no differences between groups for IC or Hipp URs (see Table Va-c).

TABLE V

Mean levels (+ 2 S.E.M.) of 5-HT and 5-111AA h~ ng[g tissue and the 5-HT utilization ratio (5-HT.UR) in control and alcohol-treated groups after transplantation Groups

(a) 5-HT BF VM HC ALC CON

Brahz regions FC

IC

tlipp

131.2 172.6 298.4** 188.3 140.1

81.6 79.3 122.6 65.9 88.1

181.5 252.2 247.5 170.4 194.8

2 S.E.M.

102.6

65.0

194.8

(b) 5-HIAA BF VM HC ALC CON

117.9 157.8 227.1' 119.8 132.2

95.2 103.2 149.3 90.6 101.2

210.2 264.0 248.3 187.8 240.9

2 S.E.M.

83.1

49.4

22.6

(e) 5-IIT-UR BF VM HC ALC CON

1.09 0.69* 0.75 0.67* 1.90

1.25 1.44 1.46 1.58 1.32

1.24 I. 19 1.22 1.14 1.26

2 S.E.M.

0.37

0.76

0.27

Difference from control: *P < 0.05, **P < 0.025.

DISCUSSION

The results confirm the findings of Arendt et al. 2'3 that 6 months of alcohol ingestion impairs both the acquisition and performance of radial maze tasks in rats. Indeed performance in alcohol-treated groups remained inferior to controls for a full year after withdrawal, suggesting that without graft intervention the deficits were permanent. In the present experiment error rates were not as high as those reported by Arendt et al. 2'3, where alcohol was given as the sole source of fluid, and some animals did not differ from controls by the end of 12 weeks of training. However the degree of voluntary consumption in our schedule (in which rats were permitted access to water for 1 h/day) enabled us to relate blood alcohol levels, as a measure of consumption, to subsequent behaviour. The high correlation between errors and BACs, and the different rates of acquisition of rats below and above the median of 100 m g ~ indicate that cognitive impairment is related to the amount of consumption, in line with the doserelated effects found by Walker and Freund 54. Taken together, Arendt et al.'s 3 findings that duration, and our findings that quantity of consumption affect the degree of cognitive impairment are consistent with Ryback's 49 continuity hypothesis of alcohol-induced deficits. In man the different magnitude of cognitive complexity may mask the presence of cognitive deficits in everyday life, until gross damage has occurred, so that symptoms may not appear to be progressive or continuous ~3. It was noteworthy that these effects on memory were obtained in the absence of gross structural damage, though, in agreement with Arendt et al.'s 3"4 findings, there was biochemical evidence for depletion in some transmitter systems, notably NA, reduced ChAT activity in cortex and hippocampus, and a reduction in AChE-positive cell counts in NBM, DB and MS projection regions of the FCPS. The rostrocaudal gradient of cell loss was comparable to that reported by Arendt et al. 4. Indeed the extent of loss (ca. 3 0 ~ in MS, 2 0 ~ in DB and 10~o in NBM) was very similar to that obtained by Arendt et al. 4 after 12 weeks of alcohol treatment (20~o v/v in drinking water) though a more marked loss might have been expected with 28 weeks of treatment. However animals used for cell counts had relatively low BACs averaging 90 m g ~ , so that our limited cell loss may reflect a low intake of alcohol. Cell count and ChAT activity indices of impaired cholinergie function were somewhat inconsistent, in that cell loss was more marked in MS and DB regions, projecting to hippocampus, than in NBM, projecting to cortex, whereas ChAT activity was more reduced in cortex than hippocampus, again confirming Arendt

24 et al.'s 3"4 findings. In brains of human alcoholics decreases in forebrain ChAT activity, muscarinic receptors, and AChE-positive cells have also been reported 1'37"38, but the indices have not been intercorrelated. Extrapolations from animal to human performance are dangerous, but our findings do suggest that it may be possible, under systematic testing conditions, to detect deleterious effects of chronic alcohol consumption in man relatively early, for instance by use of drug challenge (e.g. response to scopolamine; see below), at a stage of biochemical dysfunction and circumscribed cell loss which might be amenable to drug therapy 2s. The results from our pharmacological probes further support Arendt et al.'s 2'3 contention that damage to the FCPS plays an important role in alcohol-induced radial maze deficits. We have found that both alcohol treatment and lesions to the FCPS produce long-lasting deficits in radial maze performance22'23, though error rates of lesioned rats, particularly for working memory, were approximately twice as high as those of the alcohol-treated animals in the present experiment. Despite differences in baseline performance, the results of the present experiment show that response of alcohol-treated rats to cholinergic drugs was remarkably similar to that found in lesioned animals. Both groups of rats were more sensitive to the disruptive effects of cholinergic antagonists than controls, and showed improvements with low doses ofagonists which did not affect controls. In both groups the most substantial effects occurred in working memory. Since, as in lesioned rats, neither N-methylscopolamine nor hexamethonium altered performance, and the effects of the cholinergic treatments were bi-directional, they appear to involve central processes and not to arise from baseline artefacts. We have argued 22"23 that such a predominant effect ofcholinergic treatments on working memory is consistent with an involvement of ACh in attention rather than short- or long-term storage of information. This would be in agreement with the effects of cholinergic drugs on rapid information processing in normal human 57 and Alzheimers~ subjects, and on working memory or acquisition tasks in animals 6'52. However the present experiments found that reference memory was also affected by cholinergic agonists and antagonists in alcohol-treated rats, albeit to a lesser extent than working memory. Reference memory error rates in alcohol-treated animals were relatively low, suggesting that, unlike lesioned rats, long-term memory was not profoundly disrupted. Moreover the fact that reference memory errors were amenable to acute pharmacological influence suggests that long-term storage and retrieval processes were functional, but that alcohol-treated rats were not using them efficiently. Thus it is possible that

radial maze reference memory is not a 'pure' measure of long-term memory processes, but also makes demands on attention, for instance to register salient spatial/associative cues denoting rewarded/non-rewarded arms. If so the modest effect of cholinergic drugs on reference errors, and the marked effect on working memory errors in alcohol-treated rats would also be consistent with a role for ACh in attention, suggesting that alcohol-treated animals showed a mild attentional deficit which globally disrupted radial maze performance. We are currently using paradigms tailored specifically to attention (e.g. latent inhibition 29 and continuous non-matching to sample 44) to investigate the role of ACh in attention more precisely. We had anticipated that if alcohol-induced cognitive impairment is related to progressive damage to cholinergic systems, then the High ALC group should be more 'lesion-like' in response to cholinergic drugs than the Low; i.e. show more sensitivity to drug probes. This was partly borne out, in that error rates increased more substantially above baselines in the High ALC group than the Low with the antagonists, despite higher baselines. However, if anything, the High group was more resistant to the ameliorating action of the agonists, than the Low, and in general working memory improvements in the ALC groups were not as dramatic as in those of lesioned rats, despi!e lower baselines 22'23. This may reflect the greater age of the alcohol groups at the time of drug testing, and/or their different training history (lesioned rats were prdtrained, ALC groups were trained after 6 months of alcohol treatment). Alternatively it may indicate that systems additional to ACh contribute to the cognitive deficits shown by alcoholtreated animals. Arendt et al. 3 showed that chronic alcohol treatment reduced forebrain content of several transmitters, and that effects on NA were as marked as those on ACh. Our results for NA closely paralleled these findings. Since NA has been implicated in alcohol-induced cognitive deficits by post mortem and pharmacological studies in m a n 3~ it is important to assess the possible contribution of this system to the effects of alcohol, and its interaction with ACh. For instance preliminary findings in our laboratory (McLernon, unpublished data) indicate that radial maze working memory errors of alcohol-treated rats are also reduced by low doses of the ~2 agonist clonidine and ~2 antagonist yohimbine. In the transplant phase of the present experiments two groups of rats showed improvement over 12 weeks of post-transplant testing - - those with basal forebrain and ventral mesencephalon foetal grafts, both of which showed positive AChE staining and elevated levels of cortical and hippocampal ChAT activity relative to con-

25 trol transplant and alcohol-treated groups. Improvement in rats with cholinergic-rich transplants occurred in all 4 aspects of memory, to the extent that from weeks 8-10 (i.e. 10-12 weeks post-transplant) error rates, in general, did not differ from control level. This time course closely parallels the findings of Arendt et al. 2"3, and is consistent with evidence of Bjorklund et al? ~ for the time course of recovery of hippocampal ChAT activity and ACh synthesis following septal grafts. In contrast, error rates remained high, with no signs of consistent improvement, in the hippocampal control transplant and alcohol control groups. Several of the findings, apart from restoration of ChAT activity in the BF and VM groups, suggested that behavioural recovery was related to the presence of cholinergic cells in the two successful transplant groups. Firstly there was no evidence for an association between behavioural recovery and changes in the other transmitters measured. Cortical NA was significantly reduced in all alcohol-treated groups, regardless of type of transplant or level of performance. There were very few DBH-positive cells in any of the grafts, including the catecholamine-rich VM transplant, which would be expected to contain locus coeruleus cells. Thus trypsin processing seems effectively to have prevented survival of these noradrenergic cells, as Bjorklund et al. 11'12 demonstrated. In contrast to NA, DA content in cortex, but not hippocampus, of all transplant groups was substantially increased, whilst it was marginally reduced in IC in the sham-operated ALC group. TH-positive staining was sparse in all but the VM transplants, so that the grafts were not likely to have imported sufficient dopaminergie cells to account for the increase in DA. Moreover the same tissue, with the same distribution of TH-positive ce!ls, was implanted in cortex, which showed massive increases in DA levels, and hippocampus, which did not. Thus elevated DA levels may reflect a response to grafting within the host cortex, where grafts may have triggered sprouting in dopaminergic neurons. If so, it would seem imperative to measure amine levels routinely in transplant studies, to investigate unforeseen effects of grafting. However the occurrence of increased DA in groups which did and did not show behavioural recovery indicates that the changes in DA transmission were not related to the cognitive effects measured in these experiments. By observation grafted animals did not appear to show changes in activity or running speed, and the reduced cortical DA utilization ratios of grafted rats suggest that the excess DA may not have been functional. Alterations in 5-HT, 5-HIAA and the 5-HT utilization ratio, which were minimal and inconsistent, are not likely to account for the behavioural changes found exclusively in the BF

and VM transplant groups. The lack of effect of alcohol treatment on 5-HT in this experiment, apart from a decreased FC utilization ratio, contrasts with the significant reduction found by Arendt et al. 3 following 28 weeks of alcohol consumption as the sole source of fluid. This may reflect differences in the treatment regime. Secondly the behavioural efficacy of cholinergic-rieh tissue from two different regions of foetal brain (basal forebrain and ventral mesencephalon) strengthens the case for the involvement of cholinergic cells in behavioural recovery, since these grafts would be likely to differ with respect to other types of cell. TH staining, for instance, was more abundant in the VM than BF grafts. The evidence of Woolf et al. 58 shows that the N G F receptor is associated with NBM, but not mesencephalic cholinergic cells. If this applies also to foetal grafts, the equipotent effects of BF and VM tissue would tend to rule out a critical role for N G F (supposing that the distribution of this factor is similar to that of its receptor) in behavioural recovery in these experiments. Moreover trophic factors maximally affect early recovery 3s, and our data indicates improvement 8-10 weeks after transplantation, which is more consistent with the time course of the establishment of synaptic connectivity in cholinergie-rich septal grafts 1o. Thus, as suggested also for nigral/striatal grafts Iv, it would seem unlikely that recovery in these experiments can be attributed mainly to trophic factors. However mechanisms of transplant action were not studied in detail in this experiment. Transplants were placed in both cortex and hippocampus, so that it was not possible to relate site to particular aspects of behaviour, nor to measure additive effects between transplants at different sites, as found by Arendt et al. 2. Grafts were variable in size and location, and in some cases produced distortions in host brain, notably in the hippocampus. Their boundaries were sharply delimited, with relatively few fibres penetrating at most 2 m m into host brain. These factors, together with the ectopic nature of the graft, suggest that release of substance from the graft, or non-specific synaptic connections with the host z~ were involved, rather than a precise point-to-point restoration of host circuitry 59. Moreover the efficacy of cholinergic-rich grafts does not preclude the possibility that grafts expressing other neurotransmitters may also be effective in alleviating alcohol-induced deficits. Work is in progress to see whether noradrenergic-rich VM grafts promote behavioural recovery to the same extent as the cholinergic-rich (noradrenergic-poor) VM grafts in the present experiment. In conclusion these results demonstrated that 6 months alcohol ingestion produced prolonged impair-

26 ment in radial maze learning and performance, which was related to degree of consumption, as shown by correlation with BACs, and which may reflect a global attentional deficit. This impairment was alleviated by cholinergic agonists, and by cholinergic-rich foetal cell grafts, and aggravated by cholinergic antagonists, but not affected by peripherally acting agents, suggesting that it was related to damage to the FCPS. Evidence for such damage was provided by significant reductions in cortical and hippocampal ChAT activity in alcoholtreated animals which did not receive cholinergic-rich transplants, and in a reduction in AChE-positive cells throughout the basal forebrain complex. The finding that cholinergic-rich grafts from two different foetal brain regions, which differed in the extent of other types of cell, both promoted behavioural improvement, is consistent with a relationship between functional recovery and the presence ofcells expressing ACh in the graft. Moreover other transmitters measured by HPLC did not correlate with recovery. These experiments therefore supported previous evidence2"3 for the involvement of ACh in alcohol-induced behavioural deficits, but they do not rule out the possibility of a contribution from other neuronal systems, notably NA. Thus further work is needed to assess the part played by other systems in the detrimental effects of alcohol on cognition.

ACKNOWLEDGEMENTS

This work was supported by the Wellcome Trust, the UK Medical Research Council, and R.J. Reynolds Inc. We wish to thank Tim Kershaw, Sanjay Patel and Peter Sowinski, Psychology Department, Institute of Psychiatry, respectively for assistance with histology, neuroanatomy, and care of the animals.

REFERENCES 1 Arendt, T., Bigl, V., Arendt, A. and Tennestedt, A., Loss of neurons in the nucleus basalis of Meynert in Alzheimer's disease, paralysis agitans and Korsakoffs disease, Acta NeuropathoL, 61 (1983) 101-108. 2 Arendt, T., Allen, Y., Sinden, J., Shugens, M.M., Marchbanks, R.M., Lantos, P. and Gray, J.A., Cholinergic-rich brain transplants reverse alcohol-induced memory deficits, Nature, 332 (1988) 448-450. 3 Arendt, T., Allen, Y., Marchbanks, R.M., Schugens, M.M., Sinden, J., Lantos, P.L. and Gray, J.A., Cholinergie system and memory in the rat: effects of chronic ethanol, embryonic basal forebrain transplants, and excitotoxic lesions of the cholinergic basal forebrain projection system, Neuroscience, 33 (1989) 435-462.

4 Arendt, T., Hennig, D., Gray, J.A. and Marchbanks, R.M., Loss of neurons in the rat basal forebrain projection system after prolonged intake ofethanol, Brain Res. Bull., 21 (1988) 563-570. 5 Bartus, R.T., Dean, R.L., Beer, B. and Lippa, A.S., The cholinergie hypothesis of geriatric memory dysfunction, Science, 217 (1982) 408-417. 6 Benninger, R.J., Jhamandras, K., Boegman, R.J. and El Dafrawy, S.R., Effects of scopolamine and unilateral lesions of the basal forebrain on T-maze spatial discrimination and alternation in rats, Pharmacol. Biochem. Behav., 24 (1986) 1353-1360. 7 Beracochea, D., Durkin, T.P. and Jaffard, R., On the involvement of the central eholinergic system in the memory deficits induced by long-term ethanol consumption in mice, PharmacoL Biochem. Behav., 24 (1986) 519-524. 8 Beracochea, D., Lescaudron, L., Tako, A., Verna, A. and Jaffard, R., Build up and release from proactive interference during chronic ethanol consumption in mice: a behavioural and neuroanatomical study, Behav. Brahz Res., 25 (1986) 63-74. 9 Bjorklund, A., Stcnevi, U., Schmidt, R.II., Dunnett, S.B. and Gage, F.H., Intracerebral grafting of neuronal cell suspensions. I. Introduction and general methods ofpreparation,Acta PhysioL Scand., Suppl. 522 (1983) I-7. 10 Bjorklund, A., Gage, F.H., Schmidt, R.II., Stenevi, U. and Dunnett, S.B., Intracerebral grafting ofneuronal cell suspension. VII. Recovery of choline acetyltransferase activity and acetylcholine synthesis in the denervated hippocampus reinnervated by septal suspension implants, Acta PhysioL Scand., Suppl. 522 (1983) 59-66. I 1 Bjorklund, A., Nornes, It. and Dunnett, S.B., Intracerebral and intraspinal implants of locus coeruleus cell suspensions: deleterious effects of trypsin in the suspension medium, Soc. Neurosci. Abstr., 9 (1983) I01. 12 Bjorklund, A., Nornes,tl. and Gage, F.H., Cell suspension grafts of noradrenergic locus coeruleus neurons in rat hippocampus and spinal cord: reinnervation and transmitter turnover, Neuroscience, 18 (I 986) 685,-698. 13 Butters, N., Alcoholic Korsakoffs syndrome: some unresolved issues concerning etiology, neuropathology and cognitive deficits, J. Clin. Exp.. NeuropsychoL, 7 (1985) 181-210. 14 Dunnet, S.B., Comparative effects of cholinergic drugs and lesions of the nucleus basalis or fimbria fornix on delayed matching in rats, Psychopharmacology, 87 (1985) 357-363. 15 Dunnett, S.B., Toniolo, G., Fine, A., Ryan, C.N., Bjorklund, A. and Iversen, S.D. Transplantation of embryonic ventral forebrain neurons into the neocortex of rats with lesions of the nucleus basalis magnocellularis. II. sensorimotor and learning impairments, Neuroscience, 16 (1985) 787-797. 16 Dunnett, S.B., Badman, F., Rogers, D.C., Evenden, J.L. and Iversen, S.D., Cholinergic grafts in the neocortex or hippoeampus of aged rats: reduction of delay dependent deficits in the delayed non-matching to position task, Exp. NeuroL, 102 (1988) 57-64. 17 Dunnett, S.B. and Isaacson, O., Trophic mechanisms are not enough, Trends Ne,lrosci., 12 (1989) 257. 18 El Badawi, A. and Schenk, E.A., tlistochemical method for separate, consecutive and simultaneous demonstration ofacetylcholinesterase and norepinephrine in cryostat sections, J. llistochem. C)'tochem., 15 (1967) 580-588. 19 Fonnum, F., A rapid radiochemical method for the determination of choline acetyltransferase, J. Neurochem., 24 (1975) 407-409. 20 Gray, J.A., Sinden, J.D. and Hodges, tt., Cognitive function, neural degeneration and transplantation. In B. Anderton (Ed.),

27

21

22

23

24

25

26

27

28 29 30

31

32

33

34 35

36

37

38

39

Neurodegenerative Disease: Seminars fir the Neurosciences, VoL 2, Saunders, London, 1990, pp. 133-142. tlefti, F., Nerve growth factor promotes survival of septal cholinergic neurons after fimbrial transections, J. Neurosci., 6 (1986) 2155-2162. Hodges, tt., Thrasher, S. and Gray, J.A., Improved radial maze performance induced by the benzodiazepine antagonist ZK93426 in cholinergic lesioned and alcohol-treated rats, Behav. PharmacoL, 1 (1989)44-54. Hodges, tI., Allen, Y., Sinden, J., Lantos, P. and Gray, J.A., Cholinergic-rich foetal transplants improve cognitive function in lesioned rats, but exacerbate response to cholinergic drugs, Prog. Brahl Res., 82 (1990) 345-358. Jarrard, L.E., Selective hippocampal lesions and behaviour" Implications for current research and theorizing. In K.H. Pribram and R.L. Isaacson (Eds.), The ttippocampus, Vol. 4, Plenum, New York, 1986, pp. 93-126. Karnowsky, N.J. and Roots, L.A., A 'direct colouring' thiocholine method for cholinesterase, J. Histochem. Cytochem., 12 (1964) 219-221. Kopelman, M.D., The cholinergic neurotransmitter system in human memory and dementia: a review, Q.J.E.P., 38A (1986) 535-573. Kudo, Y., Shiosaka, S., Matsuda, M. and Tohiyama, M., An attempt to cause the selective loss of the cholinergic neurons in the basal forebrain of the rat: a new model of Alzheimer's disease, Neurosci. Lett., 102 (1989) 125-130. Lishman, W.A., Alcoholic dementia: a hypothesis, Lancet, May 24 (1986) 1184-1185. Lubow, R.E., Latent Inhibition, Psychol. Bull., 79 (1973) 398-407. Mair, R.G. and glcEntee, WJ., Korsakoffs psychosis: noradrenergie systems and cognitive impairment, Behav. Brain Res., 9 (1983) 1-32. glair, R.G., Warrington, E.K. and Weiskranz, L., Memory disorder in Korsakoff's psychosis. A neuropathological and neuropsychological investigation of two cases, Brain, 102 (1979) 749-783. Mayes, A.R., Meudell, P.R., Mann, D. and Pickering, A., Location of lesions in Korsakoffs syndrome. Neuropsychological and neuroanatomical data on two patients, Cortex, 24 (1988) 367-388. McEntee, W.L and Mair, R.G., Memory impairment in Korsakoffs psychosis: a correlation with brain noradrenergic activity, Science, 202 (1978) 905-907. McEntee, W.J. and Crook, J.H., Age-associated memory impairment: a role for catecholamines, Neurology, 40 (1990) 526-530. Nieto-Sampedro, M., Manthorpe, M., Barbin, G., Varon, S. and Cotman, C.W., Injury-induced neuronotropie activity in adult rat brain: correlation with survival of delayed implants in the wound cavity, J. Neurosci., 3 (1983) 2219-2229. Nilsson, O.G., Shapiro, M.L., Gage, F.II., Olton, D.S. and Bjorklund, A., Spatial learning and memory following fimbriafornix transection and grafting of fetal septal neurons to the hippocampus, Exp. Brain Res., 67 (1987) 195-215. Nordberg, A., Larsson, C., Perdahl, E. and Winblad, B., Changes in cholinergic'activity in hippocampus in chronic alcoholism, Drug Alcohol Depend., 10 (1982) 333-344. Nordberg, A., Larsson, C., Perdahl, E. and Winblad, B., Changes in cholinergic activity in human hippocampus following chronic alcohol abuse, PharmacoL Biochem. Behav., 18 (1983) 397-403. O'Donnell, V.M., Pitts, W.M. and Fann, W.E., Noradrenergic and cholinergic agents in Korsakoffs syndrome, Clin. Neurophannacol., 9 (1986) 65-70.

40 Pallage, V., Toniolo, G., Will, B. and tlefti, F., Long-term effects of nerve growth factor and neural transplants on behaviour of rats with medial septal lesions, Brain lies., 386 (1986) 197-208. 41 Pelham, J.W., glarquis, J.K., Kingelman, K. and Munsat, M.T.L., Prolonged ethanol consumption produces persistent alterations of cholinergie function in rat brain, Alcohol. Clin. Exper. Res., 4 (1980) 282-287. 42 Perry, E.K., Tomlinson, B.E., Blessed, G., Bergman, K., Gibson, P.tl. and Perry, R.H., Correlation of cholinergir abnormalities xHth senile plaques and mental test scores in senile dementia, Br. Med. J., 2 (1978) 1457-1459. 43 Pioro, E.P. and Cuello, A.C., Distribution ofnerve growth factor receptor-like immunoreaetivity in the adult rat central nervous system. Effect ofcolchicine and correlation with the cholinergie system. 1. Forebrain, Neuroscienee, 34 (1990) 57-87. 44 Pontecorvo, M.J., Effects ofproactive interference on rats' continuous nonmatching to sample, Animal Learning and Behav., 11 (1983) 356-355. 45 Ridley, R.M., Baker, t I.F., Drewett, B. and Johnson, J.A., Effects ofibotenie acid lesions of the basal forebrain on serial reversal learning in marmosets, Psychophannaeology, 86 (1985) 438-443. 46 Ridley, R.M., Murray, T.K., Johnson, J.A. and Baker, tt.F., Learning impairment following lesions of the nucleus basalis of Meynert in the marmoset: modification by cholinergic drugs, Bra{n Res., 376 (1986) 108-116. 47 Riley, J.N. and Walker, D.W., Morphological alterations in hippocampus after long-term alcohol consumption in mice, Science, 201 (1978) 646-648. 48 Rupniak, N.M.J., Steventon, M.L, Field, M.J., Jennings, C.A. and Iversen, S.D., Comparison of the effects of 4 cholinomimetics on cognition in primates following disruption by scopolamine or by lists of objects, Psychopharmacology, 99 (1989) 189-195. 49 Ryback, R.J., The continuum and specificity of the effects of alcohol on memory: a review, Q.J. Stud. AlcohoL, 32 (1971) 995-1016. 50 Sahakian, B., Jones, G., Levy, R., Gray, J.A. and Warburton, D., The effects of nicotine on attention, information processing and short term memory in patients with dementia of the Alzheimer type, Br. J. Psychiat., 154 (1989) 797-800. 51 Sinden, J.D., Allen, Y.S., Rawlins, J.N.P. and Gray, J.A., The effects of ibotenic acid lesions of the nucleus basalis and cholinergic-rich neural transplants on win-stay/lose-shift and win-shift/lose-stay performance in the rat, Behav. Brahz Bes., 36 (1990) 229-249. 52 Tilson, H.A., glcLamb, R.L., Shaw, S., Rogers B.C., Pediatikakis, P. and Cook, L., Radial-arm maze deficits produced by colchicine into the area of the nucleus basalis are ameliorated by cholinergic agents, Brain Res., 438 (1988) 83-94. 53 Victor, gl. and Adams, R., The alcoholic dementias. In I.A.M. Fredericks (Ed.), Handbook of Clinical Neurology, Elsevier, Amsterdam, 1985, pp. 335-352. 54 Walker, D.W. and Freund, G., Impairment of timing behaviour after prolonged alcohol consumption in rats, Science, 182 (1973) 597-599. 55 Walker, D.W., Barnes, D.E., Zornetzer, S.V., Hunter, B.E. and Kubano, P., Neuronal loss in hippocampus after long-term alcohol consumption in mice, Science, 209 (1980) 711-712. 56 Walker, D.W., ttunter, B.E. and Wicliffe, C.A., Neuroanatomical and functional deficits subsequent to chronic ethanol administration in animals. Alcohol. Clin. Exp. Res., 5 (1980) 267-282. 57 Warburton, D.M. and Wesnes, K., Drugs as research tools in

28 psychology: cholinergic drugs and information processing,

Neuropsychobiology, 11 (1984) 121-132. 58 Woolf, N.J., Gould, E. and Butcher, L.L., Nerve growth factor is associated with cholinergic neurons of the basal forebrain but not the pontomesencephalon, Neuroscience, 30 (1989) 143-152.

59 Zhou, F.C., Li, Y. and Raisman, G., Embryonic entorhinal transplants project selectively to the deafferented entorhinal zone of adult mouse hippocampi, as demonstrated by the use of Thy-1 allelic immunohistochemistry. Effect oftiming of transplantation in relation to deafferentation. Neuroscience, 32 (1989) 349-362.