Transgenic expression of human acetylcholinesteraseinduces progressive cognitive deterioration in mice

Rachel Beeri*t, Christian Andres*t , Efrat Lev-Lehmant, Rina Timberg t,Tamir Hubermant, Moshe Shani* and Hermona Soreqt

tDepartment of Biological Chemistry, The Hebrew University of Jerusalem, 91904 Israel.*Institute of Animal Science, ARO, The Volcani Center, Bet Dagan, 50250 Israel.

Background: Cognitive deterioration is a characteristicsymptom of Alzheimer's disease. This deterioration isnotably associated with structural changes and subsequentcell death which occur, primarily, in acetylcholine-pro-ducing neurons, progressively damaging cholinergic neu-rotransmission. We have reported previously that excessacetylcholinesterase (AChE) alters structural features ofneuromuscular junctions in transgenic Xenopus tadpoles.However, the potential of cholinergic imbalance to induceprogressive decline of memory and learning in mammalshas not been explored.Results: To approach the molecular mechanisms under-lying the progressive memory deficiencies associated withimpaired cholinergic neurotransmission, we created trans-genic mice that express human AChE in brain neurons.With enzyme levels up to two-fold higher than in con-trol mice, transgenic mice displayed an age-independentresistance to the hypothermic effects of the AChE inhi-bitor, paraoxon. In addition to this improved scavenging

capacity for anti-AChEs, however, these transgenic micealso resisted muscarinic, nicotinic and serotonergic ago-nists, indicating that secondary pharmacological changeshad occurred. The transgenic mice also developed pro-gressive learning and memory impairments, althoughtheir locomotor activities and open-field behaviourremained similar to those of matched control mice. Bysix months of age, transgenic mice lost their ability torespond to training in a spatial learning water maze test,whereas they performed normally in this test at the age offour weeks. This animal model is therefore suitable forinvestigating the transcriptional changes associated withcognitive deterioration and for testing drugs that mayattenuate progressive damage.Conclusion: We conclude that upsetting cholinergicbalance may by itself cause progressive memory decline inmammals, suggesting that congenital and/or acquiredchanges in this vulnerable balance may contribute to thephysiopathology of Alzheimer's disease.

Current Biology 1995, 5:1063-1071

Background

Progressive deterioration of memory and learning is acharacteristic manifestation of Alzheimer's disease [1].The molecular mechanisms that underlie this phenome-non are still obscure, as the vast majority (> 90 %) ofAlzheimer's disease cases originate sporadically - pre-sumably as a result of multigenic contributions combinedwith environmental causes [2]. There is therefore a needto search for potential reasons to explain the progressivecognitive decline in Alzheimer's disease.

At the neuropathological level, the neuritic plaques andneurofibrillary tangles which are defining features of thisdisease are particularly concentrated in brain regionswhere cholinergic circuits operate [3]. Moreover, theprogression of disease symptoms is primarily associatedwith structural changes in cholinergic synapses [4], theloss of particular acetylcholine (ACh) receptor subtypes[5], the death of ACh-producing neurons [6] and theconsequent damage to cholinergic neurotransmission [7].These factors result in a relative excess of the ACh-hydrolyzing enzyme, acetylcholinesterase (AChE). Thisspatio-temporal correlation led to the cholinergic theory

of Alzheimer's disease [7] and promoted the developmentof cholinergic therapies [8].

The first of the cholinergic drugs to be approved forclinical use is tacrine (tetrahydroaminoacridine, THA,Cognex®), an AChE inhibitor which temporarily attenu-ates the worsening of certain clinical symptoms in patientsat the early stages of Alzheimer's disease [8]. The conceptbehind the use of tacrine is to restore the cholinergic bal-ance, at least for a while, by elevating ACh levels and aug-menting the function of the remaining ACh receptors [9].That tacrine administration has some therapeutic valueemphasizes the importance of balanced cholinergic neu-rotransmission for satisfactory steady-state cognitive func-tion. However, whether imbalanced cholinergic neuro-transmission can, by itself, contribute to the progressivedecline in Alzheimer's disease has not been addressed.

To examine the consequences of synaptic cholinergicimbalance in vivo, we have recently expressed humanAChE in transiently transgenic embryos of Xenopus laevisfrogs [10]. The recombinant human enzyme accumulatedin neuromuscular junctions and altered their structural fea-tures [11]. Moreover, transgenic expression of the mouse

© Current Biology 1995, Vol 5 No 9

*The first two authors contributed equally to this research. Correspondence to: Hermona Soreq. E-mail address: soreq@vms.huji.ac.il

1063

1064 Current Biology 1995, Vol 5 No 9

muscle nicotinic ACh receptor induced similar structuralchanges at Xenopus neuromuscular junctions, enlargingtheir post-synaptic length and deepening their synapticcleft [11]. These alterations resembled those reportedfor cholinergic brain synapses in patients at the early stagesof Alzheimer's disease [4], suggesting that these changescould also be caused by imbalanced cholinergic neuro-transmission. This, in turn, initiated our interest in creat-ing-transgenic mice with congenital AChE overexpressionand examining cognitive function in such animals.

The construction of our transgenic mice was conceptu-ally different from other recent models designed torecreate the neuropathological features of Alzheimer'sdisease. Transgenic mice expressing normal or mutantforms of the 3-amyloid protein died prematurely [12] ordeveloped amyloid plaques rich in 13-amyloid peptides[13]. This shed new light on the histopathologicalchanges that are characteristic of Alzheimer's disease,particularly in carriers of 13-amyloid mutations, but noton the relevant mental deficiencies. The elimination ofseveral key brain proteins by homologous recombination(gene knockout) caused no histopathological changes,but damaged spatial learning [14-16] or avoidance learn-ing [17]. However, neither the -amyloid transgenicmice nor any of these knockout mice were reported todisplay the neurochemical changes particular to Alz-heimer's disease, one of which is cholinergic imbalance.Furthermore, none of the reported cognitive changesappeared to involve progressive deterioration.

In contrast with the above studies, our efforts weredevoted to creating a subtle change in the normal bal-ance of cholinergic neurotransmission within the mam-malian brain, mimicking conditions that might exist inpatients with no hereditary tendency for Alzheimer's dis-ease. We further limited this change to those cell types inwhich it naturally takes place, and studied its neuro-chemical, physiopathological and cognitive consequencesin an age-dependent manner. Our findings demonstratethat AChE-overexpressing mice with inheritable cholin-ergic imbalance undergo selective progressive decline intheir capacity for spatial learning and memory, andindicate the usefulness of these mice for exploring thedetailed molecular mechanisms that underlie cognitivedeterioration in Alzheimer's disease.

Both promoters were shown previously to direct humanAChE accumulation in Xenopus neuromuscular junctions[10,11], although the CMV promoter was approximatelyten-fold more efficient in its capacity to direct productionof the AChE protein in Xenopus [10].

We carried out parallel experiments with both constructsto produce transgenic mice. In the first experiment, wemicroinjected CMV-ACHE DNA into 70 mouse eggs,but obtained only one viable founder mouse carrying theCMV-ACHE transgene. However, the transgenic DNAwas not expressed in the progeny of that mouse. In thesecond experiment, 40 microinjections of Hp-ACHEDNA yielded three viable pedigrees with 2, 10 and 15copies of the transgene, respectively. Of these, no enzymeproduction was observed in the two pedigrees with highercopy numbers of the transgene, but AChE was expressedin the pedigree with two copies of the Hp-ACHE DNA.

The rate of success in obtaining viable AChE-transgenicmice was considerably lower than we have observed pre-viously with other microinjected DNAs [19,20], suggest-ing that AChE overexpression was compatible with lifeonly at low levels. Seven generations of transgenic micewith unmodified Hp-ACHE DNA were thereafter raisedfrom the successful pedigree, and all of them presentedgrossly normal development, activity and behaviour.

Transgenic Hp-ACHE expression is limited to neurons of thecentral nervous systemReverse transcription and quantitative polymerase chainreaction (PCR) amplification [20] revealed both humanand mouse ACHE mRNA transcripts in dissected brainregions of the apparently homozygous transgenic micefrom the fifth and subsequent generations (Fig. 1). HostACHE mRNA levels and alternative splicing patternsremained apparently unchanged [21]. Bone marrow andadrenal glands expressed only mouse ACHE mRNAs,

Results

Low-level AChE overexpression is compatible with lifeTo produce transgenic mice overexpressing AChE, weused two DNA constructs that included the ACHE cod-ing sequence for the brain and muscle form of humanAChE [10]. In the first construct, this gene was precededby the potent ubiquitous promoter from cytomegalovirus(CMV-ACHE construct) [18]. In the second construct,this gene was preceded by a 596 base-pair fragment fromthe authentic promoter of the human ACHE gene,followed by its first intron (Hp-ACHE construct) [10].

Fig. 1. Identification of human ACHE mRNA. RNA was extractedand amplified from transgenic (+) and control (-) mouse cortex(C), brainstem and cerebellum (BC), basal forebrain (BF) andbone marrow (BM). Species-specific PCR primers were designedfor the indicated positions within the open reading frame (dottedunderline) on the schemes above (see Materials and methods).Resultant PCR products (276 and 786 bp, respectively) wereelectrophoresed on agarose gels. Note that expression of humanACHE mRNA was limited to the central nervous system.

Cognitive damage in AChE transgenic mice Beeri et al.

with apparently unmodified concentrations of the princi-pal alternative products. Although the Hp-ACHE pro-moter includes several MyoD-binding elements [10], thetransgene was not expressed in muscle.

To locate human Hp-ACHE mRNA transcripts in spe-cific mouse brain cell types, we performed in situ hybri-dization experiments. Labelling was observed in essentiallysimilar brain neurons to those observed in other mammals[22,23]; these included cell bodies in the basal forebrain,brainstem and cortex of the transgenic mice. Cholinocep-tive hippocampal neurons were intensively labelled, espe-cially in the CA1-CA2 region (Fig. 2). Thus, expressionof the Hp-ACHE transgene was apparently confined toneurons of the host central nervous system (CNS) thatnormally express the mouse ACHE gene.

Normally processed transgenic AChE accumulates incholinergic brain regionsAChE from the brains of control and transgenic micedisplayed similar sedimentation profiles in sucrose gradientcentrifugation, demonstrating unmodified assembly intomultimeric enzyme forms (Fig. 3). Up to 50 % of theactive enzyme in basal forebrain, but none in bone mar-row, interacted with monoclonal antibodies specific tohuman AChE (Fig. 3 and Table 1) [24]. Catalytic activitymeasurements in tissue homogenates revealed a 100 %increase over control levels in the detergent-extractableamphiphilic AChE fraction from basal forebrain, and morelimited increments in cortex, brainstem, cerebellum andspinal cord extracts (Table 1). There were no age-depen-dent changes in this pattern and no differences in thecell-type composition of the bone marrows of transgenicand control mice. Gel electrophoresis followed by chemi-cal staining for enzyme activity [24] revealed similar migra-tion patterns for AChE fromthe brains of transgenic andcontrol mice, indicating comparable glycosylation patterns.

Fig. 3. Multimeric assembly of transgenic human AChE activity.Acetylthiocholine hydrolysis levels were determined for basalforebrain homogenates from transgenic and control mice aftersucrose gradient centrifugation prior to (clear areas), or following(blue area), adhesion to human-selective anti-AChE monoclonalantibodies. Vertical arrows denote sedimentation of an internalmarker, bovine catalase (11.4 S).

Intensified cytochemical staining of AChE activity wasobserved in brain sections from transgenic mice in all ofthe areas that stain for AChE activity in normal mice[25]. Staining was particularly intense in the neo-striatumand pallidum (Fig. 4a,b) as well as in the hippocampus(Fig. 4c,d); the latter two areas are associated with learn-ing and memory [26,27]. Moreover, depositions of theelectron-dense product of AChE-cytochemical stainingwere more conspicuous within synapse-forming den-drites in the anterior hypothalamus from transgenic micethan in control brain sections (Fig. 4e,f and Table 2).However, synapses interacting with these stained den-drites were of similar lengths in transgenic mice and incontrols (Table 2), demonstrating that, unlike the situa-tion in Xenopus laevis neuromuscular junctions [11,24],AChE overexpression in the mouse brain did not modifythe post-synaptic length of (at least part of) the cholin-ergic synapses. In addition, there was no indication of

Fig. 2. Expression of transgenic Hp-ACHE mRNA in mouse brainneurons. Whole-mount in situ hybridizations were performed onfixed brain sections from (a) transgenic and (b) control mice.CA1 hippocampal neurons (bracketed) were intensely labelled inthe transgenic, but not the control, section.

RESEARCH PAPER 1065

1066 Current Biology 1995, Vol 5 No 9

neurofibrillary tangles or amyloid plaques in the analyzedbrain sections from mice up to six months of age.

Transgenic AChE selectively alters hypothermic drugresponsesTo investigate the physiological effects of overexpressingtransgenic AChE, and because cholinergic synapses in theanterior hypothalamus (where we noted AChE overex-pression) are known to be involved in thermoregulation

[28], we examined the responses of transgenic mice todrugs causing hypothermia. The potent AChE inhibitordiethyl p-nitrophenyl phosphate (paraoxon) is the toxicmetabolite of the agricultural insecticide parathion; bothyoung (4-5 weeks) and adult (5-7 months) transgenicmice were considerably more resistant to paraoxon-induced hypothermia than non-transgenic controls (Fig.5a,b). The transgenic mice were almost totally resistant toa low paraoxon dose (0.25 mg kgl, Fig. 5a). With higherdoses, they displayed limited reduction in body tempera-ture and shorter duration of response than controls (Fig.5b). Most importantly, transgenic mice exposed to a1 mg kg -1 dose of paraoxon retained apparently normalphysical activity, whereas control mice experienced myo-clonia and muscle stiffening, symptoms characteristic ofcholinergic over-stimulation.

Fig. 4. Neuronal localization of AChE. (a-d) Cells that expressAChE. Wholemount cytochemical staining of AChE activity wasperformed on fixed sections of (a,c) control and (b,d) transgenicbrains. Neuronal cell bodies in (a,b) caudate-putamen (CP),globus pallidus (GP) and (c,d) hippocampal regions CA2 andCA3 were stained more intensely in transgenic than in controlbrain. Scale bars: (a,b)= 1mm; (c,d) 100 pLm. (e,f) Dendriticaccumulation of AChE. Cytochemical staining of AChE activityand electron microscopy were performed on fixed brain sectionsincluding axo-dendritic terminals from the anterior hypothala-mus. Excess electron-dense reaction products appeared as darkgrains within dendrites (centre of field) in (f) transgenic brain ascompared with low-density labelling in (e) control brain. Scalebar = 0.5 pIm.

In addition to their improved capacity for scavengingparaoxon, the transgenic mice also displayed resistance tothe hypothermic effects of the muscarinic agonist oxo-tremorine (Fig. 5c), to the less potent effect of nicotine(Fig. 5d), and to the serotonergic agonist 8-hydroxy-2-(di-n-propylamino) tetralin (8-OH-DPAT) (Fig. 5e), butnot to the a 2-adrenergic agonist clonidine (Fig. 5f).However, the thermal response to cold exposure remainedunchanged in both transgenic and cntrol mice: after 60minutes of exposure to 5 C, the body temperature of allmice had decreased to 35.5 C, indicating that no changeshad occurred in their peripherally induced control overbody temperature (data not shown).

Transgenic AChE induces an age-dependent decline inspatial learning capacityTo examine their cognitive functioning, transgenic andcontrol mice were subjected to several behavioural tests.At the ages of one, two-three and five-seven months,transgenic mice retained normal locomotor and explor-ative behaviour, covering the same space and distance inan open field as sex-matched groups of non-transgeniccontrol mice. The open field anxiety of the mice also

Cognitive damage in AChE transgenic mice Beeri et al.

remained similar, as evaluated in the frequency of defeca-tion incidents and grooming behaviour (Table 3).

Conspicuous behavioural differences between the trans-genic and control mice were observed, however, in twoversions of the Morris water maze [29]. In the hiddenplatform version of this test, mice are expected to uselong-distance cues to orient themselves, to find a plat-form submerged under opaque water and to escape aswimming task [29]. The performance of the transgenicmice was apparently normal in this test at the young ageof four weeks - they needed a similar number of train-ing sessions as compared to control mice in order toreach the platform within a significantly shorter time(defined as the escape latency) than untrained mice(Fig. 6a and Table 3). At the age of two or three months,transgenic mice needed eight more training sessions than

control mice to improve their performance; the variationbetween animals also increased, and their shortest escapelatency was significantly longer than that of control miceof the same age (Fig. 6b and Table 3). At the age of sixmonths, naive transgenic mice, but not control mice,became totally incapable of improving their performance

60 90 120 0

Time (min)

30 60 90 120 150 180

o Control* Transgenic

Fig. 5. Transgenic mice display reduced hypothermic responsesto cholinergic and serotonergic, but not adrenergic, agonists.Core body temperature was measured in control and transgenicmice, five-seven months old, at the noted times, after intraperi-toneal injections of the marked doses of (a,b) paraoxon, (c)oxotremorine, (d) nicotine, (e) tetralin (8-OH-DPAT) or (f) cloni-dine. Presented are average data for (a,f) five or (c-e) four maleand female mice, or (b) one representative pair out of three tested.Note different time and temperature scales. Temperatures oftransgenic mice at all time points after 10 min from injectionwere statistically different from those of controls for oxotremorine,nicotine and 8-OH-DPAT (Student's t test, p < 0.05).

Fig. 6. Progressive impairment in spatial learning and memory ofHp-ACHE transgenic mice. Mice were tested in a water mazewith a hidden platform in a fixed location. Averages of dailymean escape latencies are presented for (a) four-weeks-oldtransgenic (n = 10) and control (n = 5) mice, (b) two-months-oldtransgenic (n = 10) and control (n = 10) mice and (c) five-seven-months-old transgenic (n = 9) and control (n = 11) mice. Starsmark statistically significant reductions of latencies, as comparedwith the first day values in each section (one-way ANOVA,followed by Neuman Keuls test, p < 0.02).

-30 0 30 60 90 120 0 90 210 330 450

(c)OxTreoie01 gk~ d ioie(0gk,

C

aE5)

o

-0

0so

38i

36

34

32

30 IV0 60 120 180 -20 0 20 40 60 80

(f) Clonidine 0.5 mg kg')(e) 8-OH-DPAT (1 mg kg ')

371

36

35,

A

0 30

RESEARCH PAPER 1067

WC Oxotremori ne (O. 1 mg kg-') (d) Nicotine O mg kg-')

�yg

30

i�s.v

RA2/IJ I·rDe. I 'r

1068 Current Biology 1995, Vol 5 No 9

in this test through training, such that their escapelatency remained the same as that of untrained mice(Fig. 6c and Table 3).

This pattern of deterioration was not caused by locomo-tor deficiencies, as was clear when the two groups of micewere compared in the open field test (see above). It could,therefore, reflect changes in memory. Alternatively, or inaddition, changes in the acquisition of learning could beinvolved, which might be affected by motivational fac-tors. It should be noted that the progressive behaviouraldecline of these transgenic mice was more pronouncedthan that observed in mice with targetted mutationsin the glutamate receptor 1 [14], the kinase Fyn [15] orcalcium calmoduline kinase II [16], suggesting a moresubstantial perturbation than in any of these cases.

A different, age-independent defect was observed in thevisible platform version of the Morris water maze, inwhich mice are trained to escape the swimming task byusing short-distance cues to climb on a visible platformdecorated by a flag and a paper cone [29]. Interestingly,the transgenic mice could not improve their perfor-mance in this test through training, regardless of theirage (Table 3), perhaps due to early-onset difficulties inshort-distance vision or because of abnormal avoidancebehaviour. In either case, the early occurrence of thisdefect at the age of four weeks, when the hidden plat-form test was still correctly performed by the transgenicmice, demonstrates that the defects in the visual and thehidden platform test were unrelated.

Discussion

Using the authentic promoter from the human ACHEgene in conjunction with the AChE-coding sequence,we created transgenic mice expressing human AChE in

CNS neurons. The levels of overexpression were appar-ently sufficiently low to be compatible with life. Thetransgenic enzyme was processed normally and was detec-ted preferentially in brain areas that normally expressAChE, where it affected the responses of the transgenicanimals to drugs causing hypothermia. Most impor-tantly, there was a rapid decline in the spatial learningcapacities of these transgenic animals, initiated in earlyadulthood and becoming progressively more severe, sup-porting the notion that subtle cholinergic imbalance cancause progressive learning deficiencies in mammals.

Our failure to obtain viable mice with high levels ofAChE overexpression may imply that both the Hp-ACHE promoter, with its limited efficacy, and the lowcopy numbers of the transgene were important in allow-ing survival of the ACHE-transgenic mice. The expres-sion pattern of the transgene excluded peripheral organswhere the ACHE gene is normally expressed [30-32]; inparticular, there was no expression in muscle, althoughMyoD-binding elements are included in the Hp-ACHEpromoter. The restriction of expression to CNS neuronsmay occur because the promoter is incomplete, missingnecessary enhancer elements. Alternatively, silencing ele-ments in the promoter, similar to those reported inupstream sequences of the choline acetyltransferase gene[33], could prevent expression of this transgene in itsnormal target tissues. Finally, the insertion site within thehost genome may affect expression. Additional studieswill be required to distinguish between these possibilities.

The limited neuronal expression and low copy numberof the transgene ensured that it could only cause a subtlecholinergic imbalance. The transgenic protein was pro-cessed and assembled normally, presumably due to theextremely high homology (95 %) between the humanand the mouse enzymes [30-32]. It therefore accumu-lated in all of the relevant sites, with its highest excess

Cognitive damage in AChE transgenic mice Beeri et al.

levels being found in the basal forebrain - the regionmost vulnerable to neuronal loss in Alzheimer's disease[1]. Interestingly, cholinergic synapses within the brainsof the transgenic mice retained their normal length,unlike neuromuscular junctions from AChE-overexpress-ing Xenopus tadpoles. However, the Xenopus synapses alsoaccumulated up to ten-fold more AChE [11,24]; incontrast, enzyme excess was limited to two-fold in themouse synapses analyzed, perhaps reflecting an upperlimit compatible with survival in mammals. The questionof whether cholinergic imbalance can cause progressivestructural changes to other particular types of cholinergicsynapses in the mammalian brain must await furtherexamination.

The resistance of transgenic mice to the induction ofhypothermia by different agents demonstrated that thefunctioning of cholinergic synapses was modified by theoverexpression of human AChE. Resistance to para-oxon was to be expected, as the amount of its targetprotein, AChE, was significantly increased in the brainsof these mice. However, the resistance to muscarinic,nicotinic and serotonergic agonists reflected secondarychanges affecting both cholinergic and serotonergicsynapses, or perhaps only cholinergic ones - a recentreport demonstrates that 8-OH-DPAT affects nicotinicreceptors directly [34]. These changes were caused, eitherdirectly or indirectly, by the transgenic enzyme. How-ever, no general alterations occurred in the network ofneurons controlling thermoregulation, as the transgenicmice retained normal responses to noradrenergic agentsand to cold exposure. This suggests a loss of specificsynapses and/or receptor desensitization as possible cau-ses, and indicates the use of cholinergic and serotonergicagonists for the early diagnosis of imbalanced cholinergicneurotransmission in human patients.

From their young adulthood onward, and also at a timewhen their spatial learning performance with a hiddenplatform was indistinguishable from that of controls, ourtransgenic mice failed to respond to short-distance visi-ble cues. This could reflect a particular avoidance behav-iour or could be due to the inability of these mice tofocus their eyesight on a nearby target. The latter pos-sibility is reminiscent of the differential response ofAlzheimer's disease patients to tropicamide [35]. It ispossible that autonomic control over pupil constrictionand eyesight-focusing muscles requires precisely balancedcholinergic neurotransmission. It will be intriguing toexamine whether Alzheimer's disease patients have moredifficulties in focusing on nearby targets than unaffected,similarly-aged patients.

Of particular interest are the gradual spatial memorydeficits displayed by the AChE-overexpressing transgenicmice in the hidden platform water maze, as opposed totheir normal open field behaviour. The onset of theseimpairments in early adulthood suggests that the cholin-ergic balance essential for spatial memory can be sus-tained in younger transgenic mice, perhaps by a feedback

mechanism that adjusts other key elements in cholinergicsynapses. This intricate process resembles the increase inxo-bungarotoxin binding levels in Xenopus embryos thatoverexpress AChE, and the reciprocal AChE accumula-tion in neuromuscular junctions of tadpoles expressingthe mouse nicotinic ACh receptor [11]. Further to theseobservations, we now learn that, in mammals, this adjust-ment of the cholinergic balance can only be maintainedfor a limited time.

Several mechanisms of various levels have been impli-cated in the control of CNS cholinergic balance. VariousACh receptor genes have been shown to share commonpromoter elements [36,37], thus ensuring co-regulationby trans-acting factors. Furthermore, a single cis-actingpromoter controls the production of both the vesicularACh transporter and choline acetyltransferase [38,39]. Agradual failure to maintain either or both of these mech-anisms, or others, may be involved in the progressivechanges observed in our mice. The appearance of mem-ory perturbation in the early stages of Alzheimer's dis-ease may similarly signal the onset of failure to sustaincholinergic balance at the precise levels required formemory. The increase in brain AChE activity understress [40] and the heterogeneous genomic origin ofAlzheimer's disease in the vast majority of patients [2]may indicate that failure to maintain cholinergic balancecan be due to variable causes.

The possibility exists that the overexpression of AChEmay have effects that relate to the potential non-catalyticfunction(s) of this intriguing protein. Although the phe-notypic differences observed in the transgenic mice aremost probably correlated with acetylcholine hydrolysis,further experiments should be performed to explore anycorrelation with the potential functions of AChE in cell-to-cell interactions [41,42]. Indeed, evidence accumu-lated from patients suggests that AChE per se is closelylinked to Alzheimer's disease: in patients, the brain ratiobetween AChE monomers and tetramers is significantlymodified [43]; in the adrenal gland, selective AChEdepletion has been observed [44]; and in the cerebro-spinal fluid, a catalytically active AChE form with ananomalous isoelectric point has been detected [45]. Theseobservations indicate that a functional excess of AChEmight, independent of the catalytic site, be subsequentlyexerting new and possibly dangerous roles.

In either case, the gradual impairment of learning andmemory due to AChE overexpression is consistent withthe temporary partial improvement seen when anti-AChE drugs are administered to Alzheimer's diseasepatients [8,9]. Moreover, preliminary differential PCRdisplays [46,47] of mRNA from dissected brain regionsof our transgenic mice have revealed age-dependenttranscriptional changes, which await characterization.This approach may unravel the target genes being affec-ted by the transgene. Testing the effects of drug treat-ments on the progressive transcriptional changes thatoccur in the brain of these mice might thereafter lead to

RESEARCH PAPER 1069

1070 Current Biology 1995, Vol 5 No 9

the development of novel strategies for prophylactictreatment of neurodegenerative disorders associated withcholinergic imbalance.

Finally, although AChE-overexpressing mice provide amodel for behavioural, physiological and molecularaspects of cholinergic imbalance in mammals, they alsodemonstrate a dissociation between this imbalance andthe P-amyloid deposits that occur in Alzheimer's disease.The possibility remains that cholinergic neurons are par-ticularly sensitive to the neurotoxicity attributed to theP-amyloid peptides [12,13], in which case 3-amyloidtransgenic mice should develop cholinergic deficiencieswith time. Alternatively, or in addition, cholinergicimbalance Ipay lead to abnormal -amyloid expression,in which case the AChE transgenic mice should develop13-amyloid deposits with time. A third possibility is thatthe linkage between the pathophysiological and thecognitive manifestations in Alzheimer's disease is uniqueto Homo sapiens.

Conclusions

Expression of human AChE in CNS neurons of trans-genic mice creates impairments in spatial learning andmemory which appear shortly after early adulthood andbecome progressively more severe. From early adult-hood, these mice also show a marked, yet stable, decreasein performance in a visible water maze test. The open-field behaviour of the transgenic animals remains nor-mal, although their responses to hypothermia-inducingdrugs are decreased. These findings suggest that subtlealterations in cholinergic balance may cause physiologi-cally observable changes and contribute to memorydeterioration in at least some patients with sporadicAlzheimer's disease.

Materials and methods

Production of transgenic miceDNA constructs were injected into fertilized eggs of FZB/Nmice as described [19,20]. PCR amplification and Southernblot hybridization of tail DNA samples verified integration ofvariable copy numbers of the transgene into the genomes offounder mice and their progeny.

RNA extraction and PCR amplificationThese were performed as detailed [20], except that the anneal-ing temperature was 69 C. Species-specific PCR primers weredesigned for human ACHE mRNA at nucleotides 1522(+) and1797(-) in exons 3 and 4 [24], and for mouse at nucleotides375(+) and 1160(-) in exons 2 and 3 [21]. Resultant PCRproducts (276 and 786 bp, respectively) were electrophoresedon agarose gels.

In situ hybridizationWholemount in situ hybridizations were performed at 65 C on50 pIm-thick, 4 % paraformaldehyde/1 % glutaraldehyde-fixedsections of transgenic and control brains with digoxigenin-labelled complementary human ACHE cRNA. Detection was

with alkaline-phosphatase-conjugated anti-digoxigenin antibody(Boehringer/Mannheim).

AChE activity measurementsAcetylthiocholine hydrolysis levels were determined, followingsucrose gradient centrifugation of tissue homogenates, prior toor following adhesion to human-selective anti-AChE mono-clonal antibodies, essentially as detailed [24,32].

AChE cytochemical stainingWhole-mount cytochemical staining of AChE activity wasperformed on 50 lim-thick, paraformaldehyde-glutaraldehyde-fixed sections of brains as detailed [24], except that incubationwas for 2 h. Electron microscopy was thereafter performed on80 nm sections of paraformaldehyde-glutaraldehyde-fixed brain[24]. Control experiments with no acetylthiocholine verifiedthat these electron-dense deposits were indeed reactionproducts of in situ AChE catalysis.

Hypothermic response measurementsCore body temperature was measured rectally in mice, using athermocoupled element of 1 mm diameter and 10 mm length,at the noted times after intraperitoneal injections of paraoxon,oxotremorine, nicotine, tetralin (8-OH-DPAT) or clonidine.

Morris water maze testsMice were tested in a 60 x 60 x 15 cm water maze filled with1.5 mg 1-1 powdered milk, with a submerged hidden platformin a fixed location 1 cm below water level. Four trials of up to2 min were performed per day for four days, each initiated ran-domly in a different comer of the maze. For the visual platformtest, the water level was lowered by 2 cm, and a 10 cm highstriped flag and a 7 cm high black cone were positioned on topof the now visible platform. Mean escape latencies per day(four sessions) were calculated for transgenic and sex- and age-matched controls, and statistically significant reductions oflatencies were searched for as compared with the first day val-ues in each section (one-way ANOVA, followed by NeumanKeuls test).

Open field testMice were placed in one corner of a 60 x 60 cm Plexiglass boxwith 30 cm high walls and the floor divided into 144 squares of5 x 5 cm each. Using these squares, the motion path of theanimals was manually traced for 5 min, which enabled mea-surements of walking distance and explored area. Groomingbehaviour and incidence of defecations were also noted.

Acknowledgements: We thank T. Bartfai (Stockholm), J. Crawley(Washington, DC), A. Ungerer (Strasbourg) and H. Zakut (TelAviv) for helpful discussions and B. Norgaard-Pedersen (Copen-hagen) for antibodies. This work was supported by USARMRDCgrant 17-94-C-4031 and the Israel Academy of Sciences andHumanities (to H.S.). C.A. was a recipient of an INSERM, Francefellowship, and an INSERM-NCRD exchange fellowship withthe Israel Ministry of Science and Arts. The behavioural experi-ments were performed in the Smith Foundation Institute for Psy-chobiology at the Life Sciences Institute in the Hebrew University.

References

1. Katzman R: Alzheimer's disease. N EnglJ Med 1986, 314:962-973.2. Ashall F, Goate AM: Role of the -amyloid precursor protein in

Alzheimer's disease. Trends Biochem Sci 1994, 19:42-46.3. Yankner BA, Mesulam MM: Seminars in medicine of the Beth Israel

Hospital, Boston. Beta-amyloid and the pathogenesis of Alzheimer'sdisease. N Engl J Med 1991, 325:1849-1857.

Cognitive damage in A~~hE transgenic mice Been et al. RESEARCH PAPER 1071

4. DeKosky ST, Scheff SW: Synapse loss in frontal cortex biopsies inAlzheimer's disease: correlation with cognitive severity. Ann Neu-rol 1990, 27:457-464.

5. Mash DC, Flynn DD, Potter LT: Loss of M2 muscarinic receptors inthe cerebral cortex in Alzheimer's disease and experimental cholin-ergic denervation. Science 1985, 228:1115-1117.

6. Davies P, Maloney Al: Selective loss of central cholinergic neuronsin Alzheimer's disease. Lancet 1976, 2:1403.

7. Coyle T, Price DL, DeLong MR: Alzheimer's disease: a disorder ofcortical cholinergic innervation. Science 1983, 219:1186-1189.

8. Davis RE, Emmerling MR, Jaen JC, Moos WH, Spiegel K: Therapeu-tic intervention in dementia. Clin Rev Neurobiol 1993, 7:41-83.

9. Knapp MI, Knopman DS, Solomon PR, Pendlebury WW, Davis CS,Gracon SI: A 30-week randomized controlled trial of high-dosetacrine in patients with Alzheimer's disease. J Am Med Assoc 1994,271:985-991.

10. Ben Aziz-Aloya R, Seidman S, Timberg R, Sternfeld M, Zakut H,Soreq H: Expression of a human acetylcholinesterase promoter-reporter construct in developing neuromuscular junctions of Xeno-pus embryos. Proc Natl Acad Sci USA 1993, 90:2471-2475.

11. Shapira M, Seidman S, Sternfeld M, Timberg R, Kaufer D, Patrick i,et al.: Transgenic engineering of neuromuscular junctions in Xeno-pus laevis embryos transiently overexpressing key cholinergic pro-teins. Proc Natl Acad Sci USA 1994, 91:9072-9076.

12. LaFerla FM, Tinkle BT, Bieberich CJ, Haudenschield CC, Jay G: TheAlzheimer's ADl peptide induces neurodegeneration and apoptoticcell death in transgenic mice. Nature Genet 1995, 9:21-31.

13. Games D, Adams D, Alessandrini R, Barbour R, Berthelette P, Black-well C, et al.: Alzheimer-type neuropathology in transgenic miceoverexpressing V717F I8-amyloid precursor protein. Nature 1995,373:523-527.

14. Conquet F, Bashir ZI, Davies CH, Daniel H, Ferraguti F, Bordi F, etal.: Motor deficit and impairment of synaptic plasticity in micelacking mGluR1. Nature 1994, 372:237-243.

15. Grant SGN, O'Dell T, Karl KA, Stein PL, Soriano P, Kandel ER:Impaired long-term potentiation, spatial learning, and hippocampaldevelopment in fyn mutant mice. Science 1992, 258:1903-1909.

16. Silva Al, Paylor R, Wehner JM, Tonegawa S: Impaired spatial learn-ing in -calcium-calmodulin kinase II mutant mice. Science 1992,257:206-211.

17. Picciotto MR, Zoli M, Lena C, Bessis A, Lallemand Y, LeNovere N, etal.: Abnormal avoidance learning in mice lacking functional high-affinity nicotine receptor in the brain. Nature 1995, 374:65-67.

18. Schmitt EV, Christoph G, Zeller R, Leder P: The cytomegalovirusenhancer: a pan-active control element in transgenic mice. MolCell Biol 1990, 10:4406-4411.

19. Shani M: Tissue-specific expression of rat myosin light-chain 2 genein transgenic mice. Nature 1985, 314:283-286.

20. Beeri R, Gnatt A, Lapidot-Lifson Y, Ginzberg D, Shani M, Soreq H,Zakut H: Testicular amplification and impaired transmission ofhuman butyrylcholinesterase cDNA in transgenic mice. Hum Reprod1994, 9:284-292.

21. Rachinsky TL, Camp S, Li Y, EkstrBm TJ, Newton M, Taylor P: Mol-ecular cloning of mouse acetylcholinesterase: tissue distribution ofalternatively spliced mRNA species. Neuron 1990, 5:317-327.

22. Landwehrmeyer B, Probst A, Palacios JM, Mengod G: Expression ofacetylcholinesterase messenger RNA in human brain: an in situhybridization study. Neuroscience 1993, 57:615-634.

23. Hammond P, Rao R, Koenigsberger C, Brimijoin S: Regional varia-tion in expression of acetylcholinesterase mRNA in adult rat brainanalyzed by in situ hybridization. Proc Natl Acad Sci USA 1994,91:10933-10937.

24. Seidman S, Sternfeld M, Ben Aziz-Aloya R, Timberg R, Kaufer-Nachum D, Soreq H: Synaptic and epidermal accumulations ofhuman acetylcholinesterase are encoded by alternative 3'-terminalexons. Mol Cell Biol 1995, 14:459-473.

25. Kitt CA, H6hmann C, Coyle iT, Price DL: Cholinergic innervation ofmouse brain structures. J Comp Neurol 1994, 341:117-129.

26. Yu , Thomson R, Huestis PW, Bjelajac VM, Crinella FM: Learningability in young rats with single and double lesions to the 'general

learning system'. Physiol Behav 1989, 45:133-144.27. Eichenbaum H, Steward C, Morris RGM: Hippocampal representa-

tion in place learning. J Neurosci 1990, 10:3531-3542.28. Simpson CV, Ruwe WD, Myers RD: Prostaglandins and hypothala-

mic neurotransmitter receptors involved in hypothermia: a criticalevaluation. Neurosci Behav Rev 1994, 18:1-20.

29. Morris RGM, Garrud P, Rawlins JNP, O'Keefe J: Place navigationimpaired in rats with hippocampal lesions. Nature 1981, 297:681-682.

30. Massoulie J, Pezzementi L, Bon S, Krejci E, Valette FM: Molecularand cellular biology of cholinesterases. Prog Neurobiol 1993, 41:31-91.

31. Taylor P, Radic Z: The cholinesterases: from genes to proteins.Annu Rev Pharmacol Toxicol 1994, 34:281-320.

32. Soreq H, Zakut H: Human Cholinesterases and Anticholinesterases.San Diego: Academic Press, 1993.

33. Li YP, Baskin F, Davis R, Hersch L: Cholinergic neuron-specificexpression of the human choline acetyltransferase gene is con-trolled by silencer elements. i Neurochem 1993, 61:748-751.

34. Garcia-Colunga J, Miledi R: Effects of serotonergic agents on neu-ronal nicotinic acetylcholine receptors. Proc Natl Acad Sci USA1995, 92:2919-2923.

35. Scinto LFM, Daffner KR, Dressier D, Ransil B1, Rentz D, Weintraub Set al.: A potential noninvasive neurobiological test for Alzheimer'sdisease. Science 1994, 266:1051-1054.

36. Jia HT, Tsay Hi, Schmitt J: Analysis of binding and activating func-tions of the chick muscle acetylcholine receptor -y-subunit upstreamsequence. Cell Mol Neurobiol 1992, 12:241-258.

37. Laufer R, Changeux JP: Activity-dependent regulation of geneexpression in muscle and neuronal cells. Mol Neurobiol 1989,3:1-53.

38. Erickson JD, Varoqui H, Schafer MKH, Modi W, Diebler MF, WeiheE, et al.: Functional identification of a vesicular acetylcholine trans-porter and its expression from a 'cholinergic' gene locus. BiolChem 1994, 269:21929-21932.

39. Bejanin S, Cervini R, Mallet J, Berrard S: A unique gene organizationfor two cholinergic markers, choline acetyltransferase and a puta-tive vesicular transporter of acetylcholine. J Biol Chem 1994, 269:21944-21947.

40. Tsakiris S, Kontopoulos AN: Time changes in Na+, K+-ATPase,Mg2+-ATPase, and acetylcholinesterase activities in the rat cere-brum and cerebellum caused by stress. Pharmacol Biochem Behav1993, 44:339-342.

41. Layer PG, Weikert T, Alber R: Cholinesterases regulate neuritegrowth of chick nerve cells in vitro by means of a non-enzymaticmechanism. Cell Tissue Res 1993, 273:219-226.

42. Small DH, Reed G, Whitefield B, Nurcombe V: Cholinergic regula-tion of neurite outgrowth from isolated chick sympathetic neuronsin culture. J Neurosci 1995, 15:144-151.

43. Arendt T, Bruckner MK, Lange M, Bigl V: Changes in acetyl-cholinesterase and butyrylcholinesterase in Alzheimer's diseaseresemble embryonic development - a study of molecular forms.Neurochem Int 1992, 21:381-396.

44. Appleyard ME, McDonald B: Reduced adrenal gland acetyl-cholinesterase activity in Alzheimer's disease. Lancet 1991, 338:1085-1086.

45. Navaratnam DS, Priddle JD, McDonald B, Esiri MM, Robinson JR,Smith AD: Anomalous molecular form of acetylcholinesterase incerebrospinal fluid in histologically diagnosed Alzheimer's disease.Lancet 1991, 337:447-450.

46. Liang P, Pardee AB: Differential display of eukaryotic messengerRNA by means of the polymerase chain reaction. Science 1992,257:967-971.

47. Lev-Lehman E, El-Tamer A, Yaron A, Grifman M, Ginzberg D, HaninI, et al.: Cholinotoxic effects on acetylcholinesterase gene expres-sion are associated with brain-region specific alterations in GC-richtranscripts. Brain Res 1994, 661:75-82.

Received: 11 May 1995; revised: 28 lune 1995.Accepted: 28 June 1995.

Cognitive damage in AChE transgenic mice Beeri et al. RESEARCH PAPER 1071

Erratum

Involvement of MAP kinase in insulin signallingrevealed by non-invasive imaging of luciferase

gene expression in single living cells

Guy A. Rutter, Michael R.H. White and Jeremy M. Tavare

Current Biology 1995, 5:890-899

In this paper, which appeared in the 1 August 1995 issueof Current Biology, part of Figure lb was inadvertentlyomitted. The correct version of the figure is reprinted onthe right.

Fig. 1. Regulation of AP-1 activity by insulin in single living cells.(a) CHO.T cells were microinjected with plasmid pCol.Luc(0.4 mg ml-') and then incubated in the absence or presence of100 nM insulin as shown. The image represents a 15 min datacollection period. (b) Cells were microinjected with pCol.Lucand simultaneously incubated in serum-free Ham's F12 mediumsupplemented with 1 mM luciferin and 100 nM insulin. Eachimage represents a 30 min data collection period. Note the divi-sion of the arrowed cell and the motility of its daughter cell. (c)The number of photons detected from each of the cells shown in(b) was quantified using the area analysis program (Argus-50 soft-ware). The plot shows the induction of luciferase as a function oftime in 11 individual cells.

© Current Biology 1995, Vol 5 No 91072

Corrigenda

Targeting the mouse genome: a compendium of knockouts

E.P. Brandon, R.L. Idzerda and G.S. McKnight

Current Biology 1995, 5:625-634; 758-765; 873-881

The following omissions and errors in the entries up tothe end of 1994 have come to light so far.

Omissions

CD23Only one paper was referenced when, in fact, there havebeen three independent knockouts of CD23. The miss-ing "initial reports" are:

Stief A, Texido G, Sansig GC, Eibel H, Legros G, Vanderputten H: Micedeficient in CD23 reveal its modulatory role in IgE production but norole in T and B cell development. J Immunol 1994, 152:3378-3390.Yu P, Koscovilbois M, Richards M, Kohler G, Lamers MC: Negative feed-back regulation of IgE synthesis by murine CD23. Nature 1994,369:753-756.

To allow for all three papers, the "phenotype" for CD23should be changed to: Defects in IgE regulation andIgE-mediated signalling.

CD28A "follow up" paper that was missed is:

Green JM, Noel PJ, Sperling Al, Walunas TL, Gray GS, Bluestone JA,Thompson CB: Absence of B7-dependent responses in CD28-deficientmice. Immunity 1994, 1:501-508.

CD40This did not appear in the compendium. The "initialreport" is:Kawabe T, Naka T, Yoshida' K, Tanaka T, Fujiwara H, Suematsu Set al.: The immune responses in CD40-deficient mice: Impairedimunoglobulin class switching and germinal center formation. Immunity1994,1:167-178.

The "phenotype" is: Defects in thymus-dependent hu-moral immunity.

CD40LOnly one paper was referenced when, in fact, there havebeen two independent knockouts of CD40L. The miss-ing "initial report" is:Xu JC, Foy TM, Laman JD, Elliott EA, Dunn JJ, Waldschmidt TJ, ElsemoreJ, Noelle RJ, Flavell RA: Mice deficient for the CD40 ligand. Immunity1994, 1:423-431.

The "phenotype" of the pair of papers should be:Defects in thymus-dependent humoral immunity".

Ik (Ikaros gene products)This did not appear in the compendium. The "initialreport" is:Georgopoulos K, Bigby M, Wang J-H, Molnar A, Wu P, Winandy S,Sharpe A: The Ikaros gene is required for the development of alllymphoid lineages. Cell 1994, 79:143-156.

The "phenotype" is: Neonatal lethality; reduced size;lymphocytes and lymphoid progenitors absent.

Interferon a/1 receptorThis did not appear in the compendium. The "initialreport" is:Muller U, Steinhoff U, Reis LF, Hemmi S, Pavlovic J, Zinkernagel RMet al.: Functional role of type I and II interferons in anti-viral defense.Science 1994, 264:1918-1921.

The "phenotype" is: Anti-viral defence impaired. (seeCorrections).

Invariant chainA "follow up" paper that was missed is:Elliot EA, Drake JR, Amigorena S, Elsemore J, Webster P, Mellman I,Flavell RA: The invariant chain is required for intracellular transport andfunction of major histocompatibility complex class II molecules. I ExpMed 1994, 179:681-694.

L-SelectinThis did not appear in the compendium. The "initialreport" is:Arbones ML, Ord DC, Ley K, Ratech H, Maynard-Curry C, Otten GC,Capon DJ, Tedder TF: Lymphocyte homing and leukocyte rolling andmigration are impaired in L-selectin-deficient mice. Immunity 1994,1:247-260.

The "phenotype" is: Defects in lymphocyte homing andleukocyte rolling and migration.

Corrections

Interferon y receptorThe "follow up" paper [253] primarily reports the initialreport of a knock-out of the interferon o/3 receptor (seeOmissions).

Oxidase cytochrome bThis entry should have been listed as: Cytochrome b,phagocyte-specific oxidase

Int-2 (FGF3)This entry might have been better listed as: Fibroblastgrowth factor 3 (FGF3) (int-2)

© Current Biology 1995, Vol 5 No 9 1073