Fos imaging reveals ageing-related changes in hippocampal response to radial maze discrimination...

Fos imaging reveals ageing-related changes in hippocampalresponse to radial maze discrimination testing in mice

Khalid Touzani,� Aline Marighetto and Robert JaffardCNRS, UMR-5106, Laboratory Neurosciences Cognitives, Avenue des Facultes, 33405 Talence Cedex, France

Keywords: cognitive ageing, declarative/ relational memory, immediate early gene, long-term retention, short-term retention

Abstract

A two-stage radial arm maze (RAM) task has been designed recently to demonstrate a specific age-related memory deficit in mice. Ithighlights the contrast between normal and deficient memory expression in a spatial discrimination problem depending on how to-be dis-criminated arms were presented to the animal. This specific deficit has been interpreted as a preferential loss in a relational/declarativeform of memory, thereby implying an underlying hippocampal dysfunction. To test this hypothesis, neuronal activation measured by Fosimmunostaining was compared in aged (21–23 months) and adult (4–6 months) mice trained in the aforementioned task and killed aftera retention session consisting in age-insensitive probe trials, performed 6 days later (6-day RAM). Two comparison conditions wereincluded: (i) repeated locomotor training on a treadmill (TM); (ii) the same RAM training, except for the use of a longer (30 days insteadof six) retention interval (30-day RAM). Although all RAM groups displayed similar levels of performance at the end of the experiment,immediately before the mice were killed, significant between-group differences in brain activation were observed. In adult mice, 6-dayRAM testing was associated with greater septal and hippocampal (CA1, CA3, DG) Fos expression than the TM condition. Lengtheningthe retention interval from 6 days to 30 days resulted in a significant decrease in RAM testing-induced Fos expression in most of the septo-hippocampal regions. With respect to adult mice, aged mice displayed reduced Fos expression (except for DG) and a lack of inter-relationships between levels of Fos produced in each of the SH regions, in the 6-day RAM testing condition. Conversely, there was noeffect of ageing on Fos expression associated with either TM training or 30-day RAM testing. These results are interpreted as reflectingage- (or time-) related alterations in recruiting of brain structures that underlie a relational/declarative form of memory expression.

Introduction

There is a general agreement in both human and animal experimenta-

tion that memory performance declines from early to late adulthood,

and that such age-related cognitive decline is more prominent and

severe in certain tasks (as reviewed by Gallagher & Rapp, 1997).

Neuroimaging studies in aged human subjects have revealed that

abnormal activity (either higher or lower than in younger adults) in

specific brain regions can be obtained depending on the specific task(s)

used (for a review, see D’Esposito, 1999; Grady & Craik, 2000). Given

the limited history of the application of functional neuroimaging

techniques in the study of cognitive decline associated with normal

ageing, continual effort in this direction is needed to further reveal the

complex functional anatomical changes associated with senescence.

Surprisingly, very few studies in animals (i.e. Nagahara & Handa,

1997; Zhang et al., 2000) have tackled this issue by performing

functional imaging experiments. Using mice as subjects, the present

work attempts just this.

Using a specifically designed two-stage behavioural testing proce-

dures in a radial arm maze (RAM), we have previously provided

evidence that both ageing (Marighetto et al., 1999, 2000) and hippo-

campal lesions in adult subjects (Etchamendy et al. in press) produced

the same pattern of spared vs. impaired performance depending only

on the way discrimanda (i.e. arms of the maze) were presented to the

animal (mouse). Namely, successful discrimination between arms of

opposing valence was observed when these arms were presented one at

a time (successive go-no-go discrimination) or all simultaneously (six-

choice discrimination). Yet such demonstrable acquired knowledge

failed to guide the same aged or hippocampal-lesioned mice towards

the positive arms when confronted with an explicit choice between two

of the same arms (two-choice discrimination). In line with similar

findings in rats (Eichenbaum et al., 1988, 1989), this pattern of deficit

was interpreted as reflecting a selective loss of relational memory

which, in aged animals, would stem from hippocampal physiological

dysfunctions that have been described in the ageing (reviewed in

Gallagher & Rapp, 1997).

The present study was designed to further examine this hypothesis

by comparing testing-induced activation of specific brain regions in

adult and aged mice using immunodetection of Fos protein. The

expression of the protein product of the proto-oncogene c-fos is

believed to be indicative of changes in neuronal activity (Morgan &

Curran, 1989; McCabe & Horn, 1994) and has repeatedly been shown

to be induced under learning conditions (e.g. Herdegen & Leah, 1998;

Tischmeyer & Grimm, 1999; Wan et al., 1999; Vann et al., 2000)

although intense Fos labelling might not by itself represent an index of

functionality of the brain regions considered (see Passino et al., 2002).

In the present experiment we tried to circumvent several potential

‘false positive’ age-related differences in testing-induced brain acti-

vity. First, in most of the experiments so far performed (but see Bennett

et al., 2001) one cannot disambiguate brain activation differences

related to performance from differences related to ageing per se. To

circumvent this difficulty, we first took advantage of the fact that

for certain versions of the RAM task aged mice, like hippocampus-

lesioned mice, were able to express the same level of response

European Journal of Neuroscience, Vol. 17, pp. 628–640, 2003 � Federation of European Neuroscience Societies

doi:10.1046/j.1460-9568.2003.02464.x

Correspondence: Dr A. Marighetto, as above.

E-mail: [email protected]�Present address: Columbia University Centre for Neurobiology and Behaviour, 1051

Riverside Drive, New York, NY10032, USA

Received 28 June 2002, revised 28 October 2002, accepted 19 November 2002

accuracy as adult mice. Previous neuroimaging studies (Bontempi

et al., 1999) have shown that, in adult mice, hippocampal metabolism

was increased significantly by RAM testing on a task that was very

similar to our simultaneous six-choice version. Therefore we hypothe-

sized that age-related differences should be observed in the hippo-

campal response to RAM testing even though the task used was this

(‘ageing- and hippocampal lesion- insensitive’) simultaneous six-

choice discrimination version. If so, one should also be able to show

that changing specific testing parameters would alter (e.g. reduce) the

age-related differences in brain activation. Specifically, we reasoned

that if, as expected, memory testing revealed stronger hippocampal

activation in adults with respect to aged mice, such a difference should

be suppressed (or at least reduced) by lengthening the retention

interval up to several weeks. Indeed, it was shown in the study of

Bontempi et al., in adult mice, that increasing the retention interval

from 5 to 25 days resulted in a reduction of retention testing-induced

activation of hippocampal metabolic activity. In other words, in

addition to equating performance among the two groups of aged,

lengthening the retention interval was used here as a mean to assess the

specificity of the putative age-related differences in brain activities

associated with memory testing.

Materials and methods

The cognitive impairment of senescent mice was shown by training the

animals in a three-pair discrimination task in the RAM (i.e. two-choice

discriminations, stage 1). The same mice were subsequently submitted

to a test-session (stage 2) where the manner of presenting the familiar

arms was changed to a simultaneous presentation of the six arms

altogether (six-choice discrimination). Such testing was aimed at

showing that aged mice did acquire significant knowledge about the

relative valence of the arms during stage 1 training although such

acquired knowledge could only be expressed when the presentation of

the arms was modified. The groups of animals submitted to RAM

training were killed for Fos immunochemistry following a retention

session of the six-choice discrimination performed after a period of

training remission (6 days) for half the animals and 30 days long for the

remainder. This second group corresponded to the first comparison

condition. The second comparison condition consisted in repeated

treadmill (TM) training performed in the same environment as RAM

training.

Animals

Mice were naive males of the C57/bL6 Jico inbred strain obtained from

IFFA Credo (Lyon, France). They were either aged mice of 21–

23 months old (n¼ 24) or adult mice of 4–5 months old (n¼ 24).

Upon arrival they were caged in groups, housed in a climatized

vivarium under a 12 : 12 h light : darkness cycle, and maintained with

access to food and water ad libitum. After 4–6 weeks, the animals were

caged singly a fortnight before the introduction of food deprivation,

which was introduced progressively over a number of days with the

animals eventually receiving a fixed amount of laboratory food chow

per day. Their body weight was maintained at a stable level, at

approximately 90% of their free feeding weight.

Mice of both age groups were allocated randomly into three

between-subject conditions (n¼ 8) according to three behavioural

regimes: (i) 6-day RAM; (ii) 30-day RAM; and (iii), 6-day TM. Mice

in conditions (i) and (ii) were both trained on a spatial discrimination

task in the RAM, but the two conditions differed with respect to the

retention interval, which was either 6 or 30 days before the last session.

Similarly, mice in the 6-day TM condition were given daily training on

the treadmill except for an interruption of 6 days before the last day of

training. Upon completion of the last day of training, all mice were

killed for immunohistochemistry. Eight mice were rejected from the

final analyses: four died before the end of the experiment and four

others yielded poor immunostaining because of inadequate perfusion.

The final number of aged mice in the three conditions (6-day RAM,

30-day RAM and 6-day TM) were 8, 6, and 5, respectively. The final

number of adult mice in the three conditions was seven for each.

All experimental manipulations were carried out in accordance with

the European Communities Council Directive of the 24 November

1986 (86/609/EEC).

Apparatus

Radial arm maze

This was a fully automated 8-arm RAM located in a quiet testing room

enriched with distal spatial cues; the dimensions and construction have

been described in detail elsewhere (see Marighetto et al., 1993). A door

was mounted at the entrance to each arm from the central platform and

a pellet dispenser was installed at the end of each arm. Door move-

ments were controlled by a computer program that also tracked the

position of the mouse within the maze continuously via pressure

detectors underneath the central platform and two pairs of photocell

beams installed along each arm. One pair of photocell beams guarded

the entrance of the arm, and another was placed just in front of the food

well. This enabled a real-time control of the accessibility to the arms

according to a predetermined test schedule.

Treadmill

This consisted of a Plexiglas enclosure measuring 30 cm long, 5 cm

wide and 20 cm high. A moving belt powered by a motor was mounted

on the floor of the enclosure.

Radial arm maze training

Shaping

Before discrimination training, the animals were habituated to the

apparatus over a period of 2 days. On each day, the mice were allowed

to move freely individually in the maze until all the food pellets

prebaited in the food well of each of the eight arms were collected.

Discrimination task

Each animal was separately assigned six adjacent arms of which three

served as positive (baited) arms and the remaining three served as

negative (unbaited) arms. The relative locations of these arms were

such that these six arms could be grouped into three pairs of arms with

opposing valence (designated as pairs A, B and C, as illustrated in

Fig. 1). In addition, it was ensured that if the positive arm of pairs A

and B was on the left (or right), then the positive arm of pair C had to be

on the right (or left).

The discrimination task comprised two consecutive stages, which

differed only in terms of the manner in which the arms were presented

to the mice. In Stage 1, only two arms (pairs A, B, or C) were acce-

ssible at any one time (i.e. two-choice discrimination), whereas all six

arms were accessible to the mice during Stage 2 (i.e. six-choice dis-

crimination). Such a design was based on our previous experiments

showing that most of the aged mice were impaired in acquiring the

three pairs (i.e. two-choice discriminations) as they failed to consis-

tently choose the positive arm within each pair. Nevertheless, after 15

sessions of such training, the same aged mice exhibited a preference

for positive arms similar to the one seen in their younger controls, given

that the pair presentation was changed to the simultaneous presentation

of the six arms altogether (i.e. six-choice discrimination).

Fos imaging of cognitive ageing in mice 629

� 2003 Federation of European Neuroscience Societies, European Journal of Neuroscience, 17, 628–640

By equating the length of training at Stage 1 (i.e. 15 days), we also

needed to cater for a possible overtraining effect from developing in

subjects that might have acquired the task before the 15th daily session.

To this end, mice that had attained a predetermined criterion of

performance (on pairs A, B and C) before the 15th daily session were

trained on a second set of three pairs until the completion of the 15-day

period. The reward valence of the six arms was preserved since one

pair (pair [AB]) was formed by combining an arm from pair A with an

arm (of opposing valence) from pair B, and one pair was unchanged

(i.e. pair C). Lastly, a novel pair (pair N) was introduced which was

formed by the two arms that did not feature in pairs A, B or C. This

supplementary procedure was adopted in order to minimize the risk of

a possible over-training effect developing for the original three pairs

(pairs A, B and C). Performance on the variant task is not reported here

(see Marighetto et al., 1999 for a detailed description of performance

of aged and adult mice on such trials).

Stage 1: Concurrent two-choice discriminations. In each trial, the

subject had access to two adjacent arms with opposing valence (either

of pairs A, B and C). A choice was considered to be made when the

subject had reached the food well of an arm; this also triggered the

closure of the door to the alternative arm. The trial was finished as soon

as the subject returned to the central platform. The subject was then

confined to the central platform for 10 s before the next discrimination

trial began: this constituted the intertrial interval. Each daily session

consisted of 20 consecutive trials comprising alternate presentations of

pairs A, B and C according to a pseudo-random sequence. The mice

were trained on this task for a minimum of six sessions and a maximum

of 15 sessions. Within this interval, if they reached a predetermined

criterion of choice accuracy (see below), then training on the second

set of three pairs followed on the next day till the completion of 15

sessions. As explained above such a procedure was adopted to match

the amount of stage 1 training-sessions (i.e. 15 sessions of two-choice

discriminations) between the subjects and, at the same time, avoid

overtraining on the first set of pairs (A, B and C) in quick learners.

Choice accuracy was measured by the percentage of choice of the

positive arm (percent correct). A mouse was considered to reach crite-

rion performance when its overall choice accuracy was at or above

75% over two consecutive sessions, given that performance in each of

the three discrimination problems was at least 66% correct.

Stage 2: Six-choice discriminations. All subjects were transferred to

this stage on the 16th session of training. All the six arms composing

pairs A, B and C were used again here, and the reward contingency

among them remained unchanged but their presentation was modified.

At the beginning of each trial, the doors of the six arms were opened.

An entry into an arm was considered to be made when the subject

reached the food well of the particular arm. As soon as a subject was

back on the central platform, the door of the visited arm was closed but

those of the unexplored arms remained opened. The trial was ended as

soon as each of the three positive arms had been visited. This also

triggered the closure of the doors (of unexplored arms) when the mouse

has returned to the central platform. The mouse was confined there for

10 s before the next trial began. The session comprised five trials.

Accuracy was measured by the percentage of positive arms visited

among the first three visits of each trial.

Retention: six-choice discriminations. Following an interruption of

training of either 6 or 30 days for the 6-day RAM and 30-day RAM

groups, respectively, the mice were submitted to one more session of

six-choice discrimination training. This was exactly similar to the test

session as previously described.

Each mouse was killed 90 min after the beginning of this last train-

ing session.

Treadmill training

Treadmill control animals were paired with 6-day RAM mice and were

submitted to 16 daily sessions of TM training (5 cm/s). The duration of

each session was rendered as close as possible to the mean time spent

in the maze by RAM subjects (30 min on day 1, 20 min on days 2, 3 and

4, and 10 min per day from day 5 to the last day of testing). At the end

of each session, TM animals were given the same amount of food

pellets as that consumed by the RAM groups. After a 6-day rest

interval without training, animals were submitted to another 10-min

session of TM training, 90 min before being killed for immunohis-

tochemistry.

Fos-like immunohistochemistry and cell counting

Ninety minutes after the retention session, mice were deeply anaes-

thetized with sodium thiopental (70 mg/kg i.p.) and perfused through

the ascending aorta within 10 min with 100 mL saline solution (0.9%)

FIG. 1. General design of the experiment.


630 K. Touzani et al.

followed by 200 mL of fixative containing 4% paraformaldehyde and

0.2% picric acid in 0.1 M phosphate buffer (pH 7.4). Brains were

removed and immersed in the same fixative for 2 h at room temperature

and then soaked overnight in a 20% sucrose solution in 0.1 M phos-

phate buffer, pH 7.4, at 4 8C. The brains were then frozen in cooled

methyl-2-butane and frontally sectioned in a freezing microtome at

50mm. One of each two consecutive sections was stained with thionine

and the second was treated for immunodetection of c-Fos protein. The

sections were collected in Veronal buffer pH 7.4 and rinsed twice with

the same buffer before immersion in a rabbit affinity purified poly-

clonal antibody (Oncogene Research Products, Boston, USA) raised

against amino acid residues 4–17 of human c-Fos protein. Sections

were incubated with the primary antibody at room temperature at a

1 : 2000 dilution in Veronal buffer and 0.4% Triton X-100 under

constant agitation. Approximately 24 h after incubation, sections were

rinsed with Veronal buffer and incubated with a biotinylated goat anti-

rabbit IgG (Jackson Immunoresearch Laboratories, PA, USA) 1 : 2000

in Veronal buffer, 0.4% Triton X-100 for 2 h. Sections were then

washed with Veronal buffer and incubated for 60 min with peroxidase-

conjugated streptavidin (Jackson Immunoresearch) 1 : 2000 in Veronal

buffer, 0.4% Triton X-100. The peroxidase activity was revealed

according to the glucose oxidase-nickel-DAB method (Shu et al.,

1988) and the development of reaction product was monitored under a

light microscope. Sections were then placed in sodium acetate buffer,

pH 6 for 5 min to stop the reaction, rinsed several times in distilled

water, mounted on gelatin-coated glass slides, dried overnight, rapidly

dehydrated through a graded series of ethanol, cleared in toluene and

then coverslipped with Eukitt mounting medium. Neurons with nuclei

exhibiting Fos-like immunoreactivity (FLI) were counted unilaterally

in sections taken through the dorsal hippocampus [CA1, CA3 and

dentate gyrus (DG)], the lateral septum (LS), the medial septum (MS)

and the medial prefrontal cortex (mPFC). For each area, the counts

were performed in three sections per mouse in a blind fashion. Cells

exhibiting FLI in the hippocampus and the septum were counted

unilaterally in the entire structure by caption of these brain regions

through a video camera connected to a light microscope. In mPFC,

Fos-immunoreactive nuclei were counted by direct microscopic obser-

vation within a 200mm2 grid per section viewed at 100�magnification

and positioned on the prelimbic subdivision according to the mouse

brain atlas of Franklin & Paxinos (1997).

Statistical analysis

Data were analysed by analyses of variance (ANOVAs). Each ANOVA

always included a between-subject factor, Age, which contrasted the

4–5-month and 21–23-month conditions. Analysis of behavioural data

also included a repeated-measure factor. Analysis of Fos data consisted

in comparisons between: (i) 6-day RAM and TM groups; and (ii), 6-day

RAM and 30-day RAM groups. This analysis included two between-

subject factors (Age and Behavioural condition) as well as a within-

subject factor (Region). Post hoc comparisons and correlation analyses

were performed using Scheffe’s pairwise comparisons and Pearson’s

test, respectively. An a level of 0.05 was required throughout.

Results

Behaviour

Stage 1: concurrent two-choice discriminations

All 14 adult mice acquired the three-pair discrimination task (mean

number of sessions to criterion 10.41� 0.71; minimum 6, maximum

15) but only 23.2% (3 out of 14) of aged mice managed to reach the

criterion performance within less than 15 sessions (respectively 7, 12

and 15 sessions to criterion for those three aged mice; mean number

11.33). Choice accuracy over the first and the last six sessions of this

stage for the adult and aged groups (comprising 6-day- and 30-day

RAM mice) is illustrated in Fig. 2. The performance of adult mice

improved over the course of training and approximated 80% correct in

the last session. Conversely, aged mice maintained a performance that

did not consistently vary from chance level over repeated training. The

performance was submitted to a 2� 6 ANOVA with the between-subject

factor Age and the within-subject factor Session. Within the first six

sessions, the analysis revealed a significant effect of Age (F1,26 ¼ 6.46,

P¼ 0.017) and its interaction with Session (F5,130¼ 4.68, P¼ 0.0006).

Within the last six sessions, there was a significant effect of Age

(F1,26¼ 33.5, P¼ 0.0001) and Session (F5,130¼ 5.6, P¼ 0.0001) as

well as their interaction (F5,130¼ 2.53, P¼ 0.032). These results

supported the conclusion that aged mice were significantly impaired

in acquiring the three-pair discrimination task.

Stage 2: six-choice discriminations

The progression of performance from the last session of training on

pairs A, B and C to the six-choice test is depicted in Fig. 3 (in the

aged group there were two missing values; data from two mice were

lost because of a computer failure). In the adult group, it can be seen

FIG. 2. Mean (�SEM) percentage of correct responses (choice of the positivearm) in the adult (6-day RAMþ 30-day RAM) and aged (6-day RAMþ30-day RAM) groups over the first and last six sessions before the attainmentof criterion performance in stage 1 (concurrent two-choice discrimination).Each mouse was trained until it reached the predetermined criterion ofacquisition on the first set of three pairs (A, B and C). Therefore the amount oftraining sessions required varied between individuals from a minimum of sixto a maximum of 15. This is the reason why the mean performance for eachgroup are presented for the first six and last six sessions of training. P< 0.05;P< 0.01; P< 0.001, vs. chance.



that choice accuracy was decreased slightly in the six-choice session as

compared with the last session of three-pair discriminations, whereas

an opposite tendency was observed in the aged group.

A two-way ANOVA performed on these data with the between-sub-

ject factor Age and the within-subject factor Session (last-Stage 1 vs.

Stage 2) revealed significant effects of Age (F1,24¼ 29.49, P¼ 0.0001)

and its interaction with Session (F1,24¼ 6.8, P¼ 0.015). A simple

main effect analysis showed that there was a significant between-group

difference in the last session of Stage 1 (P¼ 0.001) but not in Stage 2

(P> 0.1). The disappearance of the group difference was mainly

because of a near significant tendency for improved response accuracy

in the aged group (repeated measures: P¼ 0.067) associated with an

opposite, although nonsignificant, tendency for an impairment in the

adult group (P> 0.10). This supports the conclusion that aged mice

were impaired in the two-choice but not six-choice version involving

the same set of items.

Retention of six-choice discriminations

This session followed an intermission of training of either 6 days (6-

day RAM groups) or 30 days (30-day RAM groups). Although 30-day

performance was slightly reduced in comparison with the 6-day

condition in both adult and aged groups (see Fig. 3), the effect of

retention-interval length was far from statistical significance.

A two-way ANOVA was performed on these data with both Age

and Interval (6-day RAM vs. 30-day RAM) as the between groups

factors and Session (last Stage 2 vs. Retention sessions) as within-

subjects factor. Neither the effect of Age (P> 0.1), nor the Inter-

val� Session (F< 1) and Age� Interval–Session interactions (F< 1)

were significant.

Immunohistochemistry

The mean numbers of Fos-positive nuclei in selected brain regions of

the septo-hippocampal system and in mPFC are depicted in Fig. 4, for

each group of mice, i.e. 6-day RAM experimental groups and control

groups (TM and 30-day RAM). Representative photomicrographs of

coronal sections from the adult and aged groups showing Fos-positive

nuclei in the different brain regions are presented in Fig. 5 (6-day

RAM), Fig. 6 (30-day RAM) and Fig. 7 (TM).

Comparisons between RAM and TM

Results summarized in Fig. 4 showed that, in adult mice, the number

of Fos-positive nuclei was greater in the 6-day RAM group than in the

TM group (training) in all the septo-hippocampal and mPFC regions.

In aged mice this RAM vs. TM between-group difference was nil in the

mPFC and was reduced in all the septo-hippocampal areas except for

DG. A three-way ANOVA with Age and Behaviour (RAM vs. TM) as

between-groups factors and Region (6 levels: CA1, CA3, DG, LS, MS

and mPFC) as the within-subjects factor performed on these data

confirmed this description, indicating a highly significant threeway

Age�Behaviour–Region interaction (F5,115¼ 9.11; P< 0.001). This

was mainly because of the absence of Age (F< 1) and of

Age�Region (F< 1) differences, whereas both the effect of Age

(F1,13¼ 10.05; P¼ 0.007) and its interaction with Region

(F5,65¼ 65.86; P< 0.001) were highly significant in RAM (6-day)

animals. These results support the conclusion that ageing affects the

Fos expression induced by RAM but not TM training in the mPFC and

in certain regions of the septo-hippocampal system.

There were also significant effects of Behaviour (F1,25¼ 17.34;

P< 0.001), Age (F1,25¼ 7.05; P¼ 0.014)) and their interaction (F1,25¼7.84; P¼ 0.01), indicating that RAM training globally increased

FLI contents as compared to TM, but that this increase was signifi-

cantly attenuated in aged mice. The effect of Region (F5,115¼ 97.1;

P< 0.001) and its interaction with Behaviour (F5,115¼ 15.69;

P< 0.001) and Age (F5,115¼ 10.61; P¼ 0.002) were also significant,

indicating thereby that the absolute level of Fos production, as well as

its differential between RAM and TM training and between ages varied

among the brain regions under investigation.

Additional Pearson analyses of correlation were performed on

FLI contents measured in each region within each group. Results

are summarized in Table 1. In adult mice, almost all FLI contents

were positively inter-related in the RAM (6-day) group but not

in the TM group. In contrast, no significant correlation was obser-

ved in aged mice of either the RAM (6-day) group or the TM

group. Although these analyses must be considered cautiously given

the small number of subjects in each group, they suggest that, in

addition to reducing Fos production associated with RAM training,

FIG. 3. (Left) Comparisons between the performance (mean�SEM percentage of correct responses) of aged (6-day RAMþ 30-day RAM) and adult (6-dayRAMþ 30-day RAM) mice in the last session of stage 1 and in the testing session of stage 2. (Right) Comparisons between the six-choice discriminationperformance of aged and adult mice in the testing session of stage 2 and in the retention session performed either 6 days (6-day RAM groups) or 30 days (30-dayRAM groups) later. ��P< 0.001 vs. aged; 888P< 0.001 vs. chance; 88P< 0.01 vs. chance; 8P< 0.05 vs. chance.



ageing also affects the coherence of this production across brain

regions.

Finally, correlation analysis revealed positive interrelations between

FLI contents in several regions and RAM performance in adult mice

(LS r¼ 0.84, P¼ 0.02; MS r¼ 0.73, P¼ 0.06; PFC r¼ 0.78, P¼ 0.04;

for CA1, CA3 and DG; all P> 0.14) but not in aged mice (all r< 0.13

and all P> 0.18).

Brain activation in the subgroup of aged mice displaying

no sign of impairment in stage 1

As detailed in the method section, stage 1 training in the two-choice

discrimination design was not strictly identical among the subjects.

This comes directly from our twofold aim to match the amount of

training across subjects and to avoid possible over-training effects in

fast learners. Thus, all mice reaching a predetermined criterion of

performance on the first set of three pairs before the 15th session (i.e.

all adult mice and three aged mice) were submitted to training on a

second set of three pairs, whereas the mice that failed to acquire the

two-choice task (i.e. all aged mice but three) were trained on the same

set of pairs till the fifteenth session. As can be seen in Fig. 4, the

subgroup of aged mice that displayed no sign of deficit in stage 1 (and

therefore underwent the exact same training as the adult group)

nevertheless displayed patterns of RAM-induced Fos production that

were indistinguishable from that seen in the whole set of aged subjects.

FIG. 4. Mean (�SEM) number of Fos-positive nuclei in different brain regions in each group following testing in the six-choice discrimination. The ‘unimpairedaged group’ was made of those three aged mice, which reached the criterion performance in the concurrent two-choice discrimination task (pairs A, B and C)within less than 15 sessions. Therefore, exactly as the younger adult mice, these three aged mice underwent training on the second set of three pairs ([AB], C & N)till the 15th session of two-choice discrimination training. 8P< 0.05; 88P< 0.01 vs. aged; �P< 0.05; ��P< 0.01 vs. 30-day RAM; þP< 0.05; þþP< 0.01;þþþP< 0.001 vs. 6-day RAM.



Therefore, this observation enables us to exclude the slight between-

subjects differences in the way animals completed initial training

with pairs as a potential cause for the age-related differences in

Fos production observed following testing in the six-choice version

of the task.

Comparison with the long-term retention condition

As can be seen in Fig. 4, FLI contents measured in the septal area

and in the hippocampal CA1 and CA3 subfields of the adult RAM

group of mice were reduced following the long (30-day) with respect to

the short (6-day) retention intervals. Conversely, these contents were

very similar among the two retention intervals in both the DG and

mPFC.

In the aged subjects, the effect of the retention interval on FLI

contents was almost nil in both the septal area and in the hippocampal

CA1 subfield, whereas, as observed in adult mice, increasing the

retention interval was associated with a reduction of Fos production

in CA3. Surprisingly, the content of FLI measured in the mPFC in this

aged group was significantly greater for the long- than for the short

retention interval.

These impressions were confirmed by a three-way ANOVA with

Age and Interval as between-groups factors and Region as the

within-subjects factor indicating a significant Age� Interval–Region

triple interaction (F5,120¼ 5.47; P< 0.001). This was mainly because

of the lack of effect of Age (F< 1) and of its interaction with Region

(F< 1) in the 30-day RAM groups, a result in sharp contrast with

observations previously described in the 6-day RAM groups (see

above), wherein both effects were highly significant. This was also

reflected by a significant Age–Interval interaction (F1,26¼ 4.48;

P¼ 0.045). Finally, the analysis also revealed a significant effect

of Region (F5,120¼ 104.74; P< 0.001) and of its interactions with

both Age (F5,120¼ 8.73; P< 0.001) and Interval (F5,120¼ 2.57;

P¼ 0.03), suggesting that the absolute level of FLI, as well as its

differential between adult and aged groups, and between short and long

retention intervals, varied among the anatomical structures investi-

gated.

These data support the conclusion that in the mPFC and certain

regions of the septo-hippocampal system ageing selectively reduces

Fos expression associated with short-term but not long-term retention

of the RAM discrimination.

FIG. 5. Photomicrographs of coronal sections from young (left column) and aged (right column) groups showing Fos-positive nuclei in the hippocampus (A and D),lateral septum (B and E) and medial prefrontal cortex (C and F) in the 6-day RAM groups. Scale bar, 0.4 mm.



Additionally (see in Table 1), Pearson correlation analysis showed

that FLI contents measured in regions of the septo-hippocampal

system in the adult 30-day RAM group were again positively inter-

related. However, there was no correlation between regions of the

septo-hippocampal system and mPFC as previously observed in the 6-

day condition. Furthermore, in aged animals, a pattern of interrelations

similar to that seen in adults was observed in the 30-day RAM group,

whereas no such inter-relations were actually visible in the 6-day

condition. This suggests that ageing alters the coherence of brain

responses for the short- but not the long-term retention of the dis-

crimination task.

Finally, in the adults, there was no significant relationships between

FLI contents in septo-hippocampal and mPFC regions and RAM per-

formance at the 30-day retention interval (all r< 0.24 and all P> 0.14).

In the aged mice, there was a tendency towards inverse relations

between performance and FLI contents in regions of the septo-hippo-

campal system, although most of them were far from statistical

significance (vs. CA1 r¼�0.73, P¼ 0.09; vs. CA3 r¼�0.72,

P¼ 0.11; vs. DG r¼�0.85, P¼ 0.03; vs. LS r¼�0.65, P¼ 0.17).

Discussion

The behavioural data showed that the performance of aged mice was

significantly diminished in comparison with that of the adult groups in

stage 1 discriminations, but not in the stage 2 testing involving the

same set of RAM arms. Thus, despite prolonged training, aged mice

failed to demonstrate a consistent preference for the positive arms in

the two-choice situation, whereas such a preference was actually

expressed in over-chance performance as early as the first session

of six-choice discriminations. The present study therefore confirms our

previous findings in a novel manner by showing contrasting effects of

ageing between the different versions of this position discrimination

task. This notion is discussed in the next section.

Although the different groups of RAM mice performed the six-

choice version of the task at about the same level of accuracy at the end

of the experiment, immediately before they were killed, Fos imaging

revealed that the brain sites investigated were not recruited similarly

within the different conditions of age and retention-interval length.

The significance of such differences in brain response to RAM training

FIG. 6. Photomicrographs of coronal sections from young (left column) and aged (right column) groups showing Fos-positive nuclei in the hippocampus (A and D),lateral septum (B and E) and medial prefrontal cortex (C and F) in the 30-day RAM group. Scale bar, 0.4 mm.



will be discussed subsequent to the behavioural section with respect to

both the effects of the retention interval in the adults and to the ageing

effect.

Selectivity of senescence-related memory impairment

Experiments using variants of the original RAM design (Marighetto

et al., 1999, 2000; Etchamendy et al., 2001) have shown that aged mice

as well as hippocampal-lesioned mice (Etchamendy et al. in press)

display a bias towards positive arms similar to that seen in younger

adult controls, whenever the arms are made accessible to them

successively one by one (in a go-no-go procedure) or all at a time

(in a multiple choice procedure as the one used in the present

experiment). Conversely, whenever the same aged and lesioned mice

are offered a choice between only two of the same adjacent arms,

they significantly differ from their younger controls in that they fail

to choose the positive arm more often than the negative one. Given

that the basic requirements of the task are mainly identical between

the different testing conditions, such a selective age-related deficit

might be viewed as cognitive in nature. Furthermore, it reveals that

different forms of mnemonic expression for the same piece of

past experience can be assessed in the same subjects through a change

in arm presentation. On the basis of the theorization proposed by

Eichenbaum (1992) and Eichenbaum et al. (1994), we speculate that

the two-choice version would require the ability to mentally compare

and contrast the representations associated with each arm in order to

inform an explicit relative judgement between the available choices.

Therefore, two-choice discrimination would critically rely on a com-

plex kind of mnemonic representation, incorporating the relative

relationships among choice arms, which normally enable comparisons

and contrasts between separately experienced items. By contrast, in

versions that are less demanding with regard to comparison processing

(e.g. when only one arm is opened or too many arms are accessible at

any time), position discrimination could also rely on an adapted

response to each arm which in turn is based on stimulus-response

habit strengths (McDonald & White, 1993; Packard & McGaugh,

1996). Within this view, the deficit seen in aged mice is qualitatively

similar to the preferential loss in declarative/explicit memory seen in

human senescence (Craik & Jennings, 1992; Schugens et al., 1997).

FIG. 7. Photomicrographs of coronal sections from young (left column) and aged (right column) groups showing Fos-positive nuclei in the hippocampus (A and D),lateral septum (B and E) and medial prefrontal cortex (C and F) in the 6-day TM group. Scale bar, 0.4 mm.



The present Fos imaging study gives further support to our view in

showing that the hippocampus is likely to contribute to the mediation

of adult performance and disturbances in such hippocampal mediation

occur with ageing. However, the present data can also be viewed as

contradictory with our hypothesis as the pattern of testing-induced

hippocampal activation was similarly altered with respect to the adult

group, in the subgroup of aged mice that showed no sign of an

impairment in learning the three-pairs discrimination task. This latter

aspect of our study will be discussed first.

Consistency of the senescence-related alteration in testing-induced Fos production

As in our previous study (Marighetto et al., 1999), we observed here

that about one third of the aged group could acquire the two-choice

discrimination task in stage 1 as quickly and accurately as the adult

control mice. Nevertheless, even those ‘apparently unimpaired’ aged

mice subsequently failed in recombination trials where relational

memory was assessed specifically (% correct was 45 in the aged

vs. 72 in the adult group). As previously discussed in detail, spared

initial learning of two-choice discriminations in ‘unimpaired’ aged

mice is likely to rely on a different cognitive strategy than in the adults.

One such a strategy would restrain the need for explicit between-arms

comparisons supposedly required in that task thereby supplying to

their (subsequently revealed) relational memory deficit. Whatever the

case, the reliability of the deficit evidenced in our paradigm in the aged

population might explain the present similarity in hippocampal Fos

expression between ‘unimpaired’ and impaired aged mice. Therefore,

not only are the present Fos data with our hypothesis that senescence is

accompanied by significant alteration in hippocampal-dependent rela-

tional processes, but they also support our previous assumption that

such a dysfunction might be less variable among individuals than

classically believed to be (see Marighetto et al., 1999 for further

discussion).

The consistency of patterns of brain activation among the aged

population also enables us to rule out a methodological concern related

to existing between-subjects differences in the completion of stage 1

training. Indeed, the subgroup of ‘unimpaired’ aged mice underwent

the exact same RAM training as the adult group (i.e. training on a

second set of pairs after acquiring the first three pairs), and never-

theless displayed a significant reduction in hippocampal activation.

Therefore, qualitative differences in stage 1 training might not be

considered as the critical factor for explaining the observed between-

age differences in testing-induced Fos production.

Effect of retention-interval length or age onpattern of brain activation

The present immunohistochemical data from the adult groups show

that the short-term retention of the six-choice discrimination task

produces a significant activation of the septo-hippocampal system,

in comparison with locomotor activity training. This observation is in

line with a report on the Fos imaging in different RAM memory tasks

in rats (Vann et al., 2000). Within the hippocampus, the greatest over-

expressions induced by the RAM testing were observed in CA1 and

CA3 subfields (CA1¼CA3>DG). These subfield-specific changes

extend the findings of Hess et al. (1995) who showed that rats

performing a well-learned odour discrimination task had increased

c-fos mRNA expression in the hippocampus, with the greatest effect

being found in CA1. Frontal FLI contents were also increased in our

RAM-trained adult mice. Furthermore, positive inter-relations

between the different regions investigated were observed, although

they were not visible in TM controls. This could indicate the simulta-

neous recruitment of the septo-hippocampal and frontal areas as parts

of the same extended network mediating the short-term retention of the

RAM six-choice discrimination task.

The RAM testing-induced activation in adult brains was reduced in

all sites investigated, except DG and mPFC, when the retention

interval was increased from 6 to 30 days. In short-term retention,

Fos expression induced by testing was also markedly reduced in aged

brains with respect to adult ones, in all sites except the DG. Both these

retention interval- and age-dependent patterns of brain activation were

TABLE 1. Correlation matrix tables for Fos-like immunoreactivity in brain regions for each experimental group

Adult Aged

CA1 CA3 DG LS MS CA1 CA3 DG LS MS

6-day RAMCA1CA3 0.797� 0.525DG 0.933�� 0.793� 0236 0.699LS 0.802� 0.759� 0.843� 0696 0578 0.565MS 0713 0.67 0.68 0721 �0177 0246 0008 0.228mPFC 0.802� 0.839� 0718 0.867� 0.868� 0531 0.948�� 0533 0358 0.084

30-day RAMCA1CA3 0.878�� 0.789DG 0.821� 0.982�� 0.877� 0.807�

LS 0.873� 0.859� 0.759� 0.82� 0.811� 0.947��

MS 0667 0495 0.33 0.841� 0085 �0116 0313 0.362mPFC �0001 �0225 �0313 0191 0343 0251 0027 0359 0.55 0.78

6-day TMCA1CA3 �0.294DG 0633 �0494 0.783LS 0554 0357 0679 �0287 �0.097MS �0246 �0423 0.42 0802 0157 0117 0.767mPFC �0277 �0443 0484 0713 0.975�� 0293 �0381 0.82 0.867

�P< 0.05, ��P< 0.01, ��P< 0.001, Pearson Correlation



observed despite the lack of between-groups difference in the accuracy

in performing the six-choice discrimination task. Therefore, the basic

requirements of the task (in terms of reward, stimulus and response

involved) can be rejected as potential causes for the septo-hippocampal

activation induced by RAM testing at the short (6-day) retention

interval. However, the fact that apparently similar performances can

be sustained by a differential contribution of the same (here septal and

hippocampal) brain regions can be interpreted in at least three different

ways.

A first possible explanation is the absence of causal relationship

between septo-hippocampal activation and six-choice discriminative

performance. Even though such a hypothesis cannot definitely be

rejected, we will defend the position that activation of the hippocam-

pus has meaningful implications for discrimination performance. Our

opinion is based on recent data (Etchamendy et al., 2003) showing that

although hippocampus-lesioned mice performed the six-choice ver-

sion of the task with the same accuracy as their sham controls, both

groups displayed significant qualitative differences since discrimina-

tive behaviour appeared to be based on a cognitive strategy similar to

the one used in go-no-go discriminations in the former but not in the

later group. Supporting the view that six-choice discrimination might

rely on different kinds of cognitive processes mediated by dissociated

memory systems, these observations led us to favour a second possible

interpretation of our set data as follows.

Although quantitatively similar, the performances were qualita-

tively distinct between the two conditions of retention-interval or

age, i.e. relied on different memory systems within which the

septo-hippocampal region is not equally involved. Thus, we already

postulated that existing relational representations of past episodes in

adult control mice would sustain flexible expression of memory in all

various forms of between-arms discriminations. Meanwhile, as a result

of hippocampal dysfunction, position discrimination in aged mice

would mainly rely on stimulus-response habit. The present age-related

reduction of Fos production in CA1 and CA3 associated with short-

term RAM testing condition supports our view. If we now speculate

that, as time passes, the full richness of relational representation

progressively wanes, then this implies that the mnemonic representa-

tions of adults progressively become qualitatively similar to those of

the aged. Hence, after a long retention interval, discriminative

responses of adult mice might rely on simple associations, mainly

like those of aged mice in short-term retention, and the pattern of brain

responses to long-term retention testing would more closely corre-

spond to a procedural or implicit expression of memory. In this view,

first, the relative reduction in testing-induced hippocampal activation

in the long term condition is coherent with human neuroimaging data

showing that significant activation of the hippocampal formation is

specifically associated with conscious recollection of the learning

episode (Schacter et al., 1996; Eldridge et al., 2000). Second, the

absence of any age-related difference in brain responses to RAM

testing at the 30-day retention interval is coherent with the idea that

ageing is associated with a preferential alteration in the system that

mediates declarative memory.

A third alternative interpretation of our results could be that whether

in aged or in younger mice, in short- or in long-term retention, the six-

choice discrimination task always relies on the same cognitive pro-

cesses, these being possibly mediated by the recruitment of different

neural networks. In fact, such a ‘brain reorganization’ hypothesis is

frequently encountered (at present) in the current literature devoted to

cognitive ageing in humans (Hazlett et al., 1998; McIntosh et al.,

1999). Similarly, a gradual reorganization of the neural substrates

underlying long-term memory storage is postulated in a model of

memory consolidation proposed by Squire & Alvarez (1995). Accord-

ing to this model, the hippocampus is only temporarily involved in

the maintenance of acquired information until such a reorganiza-

tion (i.e. the consolidation process) has ended. Such a view might

indeed explain why the hippocampus was not activated significantly by

the long-term retention of the task in adult mice and, as a direct

consequence, why the differential between the two ages disappeared.

However, as explained below, such an interpretation can hardly account

for the total absence of ageing effect in the long-term condition.

Selectivity of senescence-related changes inthe pattern of brain activation

The present Fos imaging study reveals a triple selectivity of the effect

of ageing on brain responses associated with testing in the six-choice

discriminations task: (i) ageing affects responses to RAM but not TM

training; (ii), ageing reduces c-Fos expression in some but not all brain

regions as FLI contents measured in DG were identical in aged and

adult mice, whatever the condition. This age-related decrease in the

number of Fos-positive neurons within the hippocampus is unlikely to

result from neuronal death as Rapp & Gallagher (1996) clearly

demonstrated, by using modern stereological methods (which pro-

vided estimates of total neuron number in a given region in stead of

measures of neurons density), that hippocampal neuron number was

relatively preserved in aged rats; and (iii), finally, ageing alters Fos

responses in terms of both their amplitude and their inter-area relation-

ships, in the condition of short- but not long-term retention of the RAM

task. This latter selectivity is particularly remarkable and the possi-

bility that it was a pure consequence of the retention interval effect

observed in the adults can be excluded for the following reasons. First,

in mPFC, the relative increase in Fos products in the aged mice was nil

after the 6-day retention but strictly normal after the 30-day retention,

whereas in the adults this particular area was activated equally and

significantly in both conditions of RAM testing. Second, in the adult

groups, although the FLI contents observed in most of the septo-

hippocampal regions (areas) were diminished as a result of the increase

in the retention interval length, they were nevertheless inter-related

positively in both RAM conditions, whereas they were not in the TM

group. Conversely, in the aged mice, these Fos measures were mainly

unchanged in terms of their amplitudes between the 6-day RAM and

the 30-day RAM groups, but they were positively interrelated only in

the latter group. In summary, Fos imaging of septo-hippocampal and

frontal responses to the discrimination task indicates that ageing

selectively modifies the brain response to the short-term retention

of the task.

This selectivity enables us to rule out a serious methodological

concern that is intrinsically related to the use of an indirect correlate

of neural activity to image age-related changes in brain function

(D’Esposito, 1999). Namely, if ageing affects the ability of cells to

express the gene c-fos (or the kinetics of such expression) as well as

neural activity, then age-related differences in testing-induced Fos

activation could be misleading. The risk of such possible confusion is

clearly reduced whenever it can be demonstrated that age-related

differences in the activation of a particular brain area are not obser-

ved systematically but in response to certain kinds of stimulation.

Such is precisely the case here, where the mPFC in aged brains was

normally activated in response to long-term but not short-term reten-

tion testing.

Consequently, the present age-related reductions in certain Fos

protein expressions add to a previous report on the age-related increase

in hippocampal c-fos transcription induced by high-frequency stimu-

lation (Lanahan et al., 1997) to suggest possible modifications in the

activation of signalling pathways upstream of c-fos. Our study further

shows that the apparent coherence of the brain response, in particular



within the septo-hippocampal areas, might or might not be normal in

senescent mice depending on the behavioural situation. Taken as a

whole, our Fos data are therefore strongly suggestive of age-related

disturbances in the connective processes that sustain proper recruit-

ment of a set of structures in some, but not all, circumstances. It is

noteworthy that human data relative to age-related changes in brain

imaging of cognition are also frequently interpreted in terms of

changes in effective connectivity in the neural network underlying

the task (Cabeza et al., 1997; Esposito et al., 1999; McIntosh et al.,

1999).

Conclusion

The present neurofunctional correlates of ageing support our hypoth-

esis that a relational/declarative form of memory undergoes prefer-

ential loss during ageing. Even though the whole set of our existing

data from ageing and lesion experiments is coherent with our inter-

pretation, the precise nature of the processes involved remains spec-

ulative. More experiments are clearly needed to interpret more closely

the seemingly complexity of senescence-related changes in brain

activity associated with cognitive performance in our mice. In any

event, the present study shows c-fos protein expression to be a valuable

tool for such neuroimaging approaches. Our study also demonstrates

that, as in humans, it is possible to evidence subtle and selective effects

of ageing on both the brain and behaviour in mice.

Acknowledgements

This work was supported by the C.N.R.S and the Conseil Regional d’Aquitaine.The authors are very grateful to Drs B. K.Yee, J. Micheau and J. L. Guillou fortheir helpful comments on previous versions of the manuscript.

Abbreviations

DG, dentate gyrus; FLI, Fos-like immunoreactivity; LS, lateral septum; mPFC,medial prefrontal cortex; MS, medial septum; RAM, radial arm maze; TM,treadmill.

References

Bennett, P.J., Sekuler, A.B., McIntosh, A.R. & Della-Maggiore, V. (2001) Theeffects of aging on visual memory: evidence for functional reorganization ofcortical networks. Acta Psychologica, 107, 249–273.

Bontempi, B., Laurent-Demir, C., Destrade, C. & Jaffard, R. (1999) Time-dependent reorganization of brain circuitry underlying long-term memorystorage. Nature, 400, 671–675.

Cabeza, R., McIntosh, A.R., Tulving, E., Nyberg, L. & Grady, C.L. (1997) Age-related differences in effective neural connectivity during encoding andrecall. Neuroreport, 8, 3479–3483.

Craik, F.I.M. & Jennings, J.M. (1992) Human memory. In Craik, F.I.M. &Salthouse, T.A. (Eds) Handbook of Aging and Cognition. Hillsdale, Erlbaum,NJ, pp. 51–83.

D’Esposito, M. (1999) New answers to old questions. Curr. Biol., 9, 939–941.Eichenbaum, H. (1992) The hippocampal system and declarative memory in

animals. J. Cognitive Neurosci., 4, 217–231.Eichenbaum, H., Fagan, A., Mathews, P. & Cohen, N.J. (1988) Hippocampal

system dysfunction and odor discrimination learning in rats: Impairment orfacilitation depending on representational demands. Behav. Neurosci., 102,3531–3542.

Eichenbaum, H., Mathews, P. & Cohen, N.J. (1989) Further studies of hippo-campal representation during odor discrimination learning in rats. Behav.Neurosci., 103, 1207–1216.

Eichenbaum, H., Otto, T. & Cohen, N.J. (1994) The hippocampus � What doesit do? Behav. Neural Biol., 57, 2–36.

Eldridge, L.L., Knowlton, B.J., Furmanski, C.S., Bookheimer, S.Y. & Engel,S.A. (2000) Remembering episodes: a selective role for hippocampus duringretrieval. Nat. Neurosci., 3, 1149–1152.

Esposito, G., Kirkby, B.S., Van Horn, J.D., Ellmore, T.M. & Berman, K.F.(1999) Context dependent neural system-specific neurophysiological con-comitants of aging: mapping PET correlates during cognitive activation.Brain, 122, 963–979.

Etchamendy, N., Desmedt, A., Cortes-Torrea, C., Marighetto, A. & Jaffard, R.(2003) Hippocampal lesions and discrimination performance of mice in theradial maze: sparing or impairment depending on the representationaldemand of the task. Hippocampus, in press.

Etchamendy, N., Enderlin, V., Marighetto, A., Vouimba, R.M., Pallet, V.,Jaffard, R. & Higueret, P. (2001) Alleviation of a selective age-relatedrelational memory deficit in mice by pharmacologically induced normal-ization of brain retinoid signaling. J. Neurosci., 21, 6423–6429.

Franklin, K. & Paxinos, G. (1997) The Mouse Brain in Stereotaxic Coordinates.Academic Press, San Diego, CA.

Gallagher, M. & Rapp, P.R. (1997) The use of animal models to study the effectsof aging on cognition. Annu. Rev. Psychol., 48, 339–370.

Grady, C.L. & Craik, F.I.M. (2000) Changes in memory processing with age.Curr. Opinion Neurobiol., 10, 224–231.

Hazlett, E.A., Buchsbaum, M.S., Mohs, R.C., Spiegel-Cohen, J., Wei, T.C.,Azueta, R., Haznedar, M.M., Shihabuddin, L., Luu-Hsia, C. & Harvey, P.D.(1998) Age-related shift in brain region activity during successful memoryperformance. Neurobiol. Aging, 19, 437–445.

Herdegen, T. & Leah, J.D. (1998) Inducible and constitutive transcrip-tion factors in the mammalian nervous system: control of gene expres-sion by Jun, Fos and Krox, CREB/ATF proteins. Brain Res. Rev., 28,379–490.

Hess, U.S., Lynch, G. & Gall, C.M. (1995) Change in c-fos mRNA ex-pression in rat brain during odor discrimination learning: differentialinvolvement of hippocampal subfields CA1 and CA3. J. Neurosci., 15,4786–4795.

Lanahan, A., Lyford, G., Stevenson, G.S., Worley, P.F. & Barnes, C.A. (1997)Selective alteration of long-term potentiation-induced transcriptionalresponse in hippocampus of aged, memory-impaired rats. J. Neurosci., 17,2876–2885.

Marighetto, A., Etchamendy, N., Touzani, K., Cortes-Torrea, C., Yee, B.K.,Rawlins, J.N.P. & Jaffard, R. (1999) Knowing which and knowing what: apotential mouse model for age-related human declarative memory decline?Eur. J. Neurosci., 11, 3312–3322.

Marighetto, A., Micheau, J. & Jaffard, R. (1993) Relationships between testing-induced alterations of hippocampal cholinergic activity and memory perfor-mance on two spatial tasks in mice. Behav. Brain Res., 56, 133–144.

Marighetto, A., Touzani, K., Etchamendy, N., Cortes-Torrea, C., De Nanteuil,G., Guez, D., Jaffard, R. & Morain, P. (2000) Further evidence for adissociation between different forms of mnemonic expressions in a mousemodel of age-related cognitive decline: effects of tacrine and S 17092: a novelprolyl endopeptidase inhibitor. Learn. Mem., 7, 159–169.

McCabe, B.J. & Horn, G. (1994) Learning-related changes in Fos-like immu-noreactivity in the chick forebrain after imprinting. Proc. Natl. Acad. Sci.USA, 91, 11417–11421.

McDonald, R.J. & White, N.M. (1993) A triple dissociation of memorysystems: Hippocampus, amygdala and dorsal striatum. Behav. Neurosci.,107, 3–22.

McIntosh, A.R., Sekuler, A.B., Penpeci, C., Rajah, M.N., Grady, C.L., Sekuler,R. & Bennett, P.J. (1999) Recruitment of unique neural systems to supportvisual memory in normal aging. Curr. Biol., 9, 1275–1278.

Morgan, J.I. & Curran, T. (1989) Stimulus-transcription coupling in neurons:role of cellular immediate-early genes. Trends Neurosci., 12, 459–462.

Nagahara, A.H. & Handa, R.J. (1997) Age-related changes in c-fosmRNA induction after open-field exposure in the rat brain. Neurobiol. Aging,18, 45–55.

Packard, M.G. & McGaugh, J.L. (1996) Inactivation of the hippocampus orcaudate nucleus with lidocaine differentially affects expression of place andresponse learning. Neurobiol. Learn. Mem, 65, 65–72.

Passino, E., Middei, S., Restivo, L., Bertaina-Anglade, V. & Ammassari-Teule,M. (2002) Genetic approach to variability of memory systems: analysisof Place vs. Response Learning and Fos-related expression in hippo-campal and striatal area of C57BL/6 and DBA/2 mice. Hippocampus, 12,63–75.

Rapp, P.R. & Gallagher, M. (1996) Preserved neuron number in the hippo-campus of aged rats with spatial learning deficits. Proc. Natl Acad. Sci. USA,93, 9926–9930.

Schacter, D.L., Alpert, N.M., Savage, C.R., Rauch, S.L. & Albert, M.S.(1996) Conscious recollection and the human hippocampal formation:evidence from positron emission tomography. Proc. Natl Acad. Sci., 93,321–325.



Schugens, M.M., Daum, I., Spindler, M. & Birbaumer, N. (1997) Differentialeffects of aging on explicit and implicit memory. Aging Neuropsychol. Cogn.,4, 33–44.

Shu, S.Y., Ju, G. & Fan, L.Z. (1988) The glucose oxidase-DAB-nickel methodin peroxidase histochemistry of the nervous system. Neurosci. Lett., 85,169–171.

Squire, L.R. & Alvarez, P. (1995) Retrograde amnesia and memory consolida-tion: a neurobiological perspective. Curr. Opin. Neurobiol., 5, 169–177.

Tischmeyer, W. & Grimm, R. (1999) Activation of immediate early genes andmemory formation. Cell. Mol. Life Sci., 55, 564–574.

Vann, S.D., Brown, M.W., Erichsen, J.T. & Aggleton, J.P. (2000) Fos imagingreveals differential patterns of hippocampal and parahippocampal subfieldactivation in rats in response to different spatial memory tests. J. Neurosci.,20, 2711–2718.

Wan, H., Aggleton, J.P. & Brown, M.W. (1999) Different contributions of thehippocampus and perirhinal cortex to recognition memory. J. Neurosci., 19,1142–1148.

Zhang, Y.Q., Ji, Y.P. & Mei, J. (2000) Behavioral training-induced c-Fosexpression in the rat nucleus basalis of Meynert during aging. Brain Res.,879, 156–162.



Fos imaging reveals ageing-related changes in hippocampal response to radial maze discrimination...

Documents

Transcript of Fos imaging reveals ageing-related changes in hippocampal response to radial maze discrimination...