Neurobiology of Learning and Memory 77, 1–16 (2002)

doi:10.1006/nlme.2000.3998, available online at http://www.idealibrary.com on

A Comparison of the Effects of Fimbria-Fornix,Hippocampal, or Entorhinal Cortex Lesions on

Spatial Reference and Working Memory in Rats:Short versus Long Postsurgical Recovery Period

Rodrigue Galani,1 Stephanie Obis, Etienne Coutureau,2 Len Jarrard,3 andJean-Christophe Cassel

Laboratoire de Neurosciences Comportementales et Cognitives,Universite Louis Pasteur, UMR 7521 ULP/CNRS, 12 rue Goethe, F-67000 Strasbourg, France

Published online September 10, 2001

Using a radial maze task and different postoperative recovery periods, this experi-ment assessed and compared the reference and working memory performances ofadult Long–Evans male rats subjected to entorhinal cortex, fimbria-fornix, andhippocampus lesions. Sham-operated rats were used as controls. In order to seewhether the duration of the postsurgical recovery period would influence acquisitionof the complex radial maze task, training began 1 month following surgery (Delay1) for half the rats in each group, while for the other half training was started 6.5months following surgery (Delay 2). The results indicated that at both recoveryperiods the entorhinal cortex lesions failed to affect either working or referencememory in the spatial task. Conversely, both fimbria-fornix and hippocampus lesionsimpaired both reference and working memory. While the reference memory deficitwas generally similar in both fimbria-fornix and hippocampal lesion groups, analysisof the results for working memory indicated that at the longer delay rats withfimbria-fornix lesions were still impaired but in animals that had the hippocampusremoved, working memory did not differ from that of controls. These results suggestthat there was some recovery in those rats with hippocampal lesions (e.g., on theworking memory task) but both hippocampal and fimbria-fornix animals were stillimpaired compared to controls when training was delayed 6.5 months followingthe operations. q 2001 Elsevier Science

1 Present address: Brain Research Laboratory, Emory University, 575 Rollins Way, Atlanta, Georgia 30322.2 Present address: School of Psychology, Cardiff University, Tower Building Park Place, Cardiff CF1 3YG

Wales, United Kingdom.3 Present address: Department of Psychology, Washington and Lee University, Lexington, VA 24450.The authors wholeheartedly acknowledge Ms. C. Lazarus for expert technical assistance and Mr. O. Bildstein

and Mr. R. Paul for their excellent work in animal care. L. Jarrard collaborated on this work during his stayat the Universite Louis Pasteur as an invited professor.

Address correspondence and reprint requests to J-C. Cassel, LN2C UMR 7521 ULP/CNRS, 12, rue Goethe,F-67000 Strasbourg, France. Fax: 33 388 358 442. E-mail: jean-christophe.cassel@psycho-ulp.u-strasbg.fr.

1 1074-7427/01 $35.00q 2001 Elsevier Science

All rights reserved.

2 GALANI ET AL.

Key Words: fimbria-fornix; entorhinal cortex; hippocampus; ibotenic acid;N-methyl-D-aspartate; place-learning; radial maze; spatial memory.

INTRODUCTION

There is a growing body of evidence that supports the view that lesions of differentcomponents of the hippocampal formation (i.e., Ammon’s horn, dentate gyrus, subicularcomplex, entorhinal cortex) or interruption of afferent and efferent pathways such as thefimbria-fornix or perforant path result in different behavioral/functional deficits (see Wan,Aggleton, & Brown, 1999; Jarrard, 1993). Certainly, an examination of the literature showsthat there are many discrepancies reported concerning the behavioral effects produced bylesions of the different structures and/or fiber systems of the hippocampal formation.Many of these contradictory results can be attributed to the variety of behavioral tasksand testing procedures that have been used, while others are no doubt a result of differentlesion techniques and surgical procedures. Of special concern in the present research isthe length of the postoperative recovery period and the possibility that there may berecovery of function with different delays.

There is a suggestion in the literature that the length of time between surgery and thebeginning of behavioral testing may be an important variable that accounts for somecontradictory results that have been reported. For example, Whishaw and Jarrard (1995)found that lesions of the fimbria and dorsal fornix (one of the main afferent/efferentpathways of the hippocampus) resulted in a greater effect on spontaneous activity andspatial memory than excitotoxic lesions that were limited to the hippocampus. In theirstudy, it is important to note that the authors began testing their rats approximately 2weeks after surgery. Their results are at some variance with the results reported in earlierexperiments in which we have shown that rats with fimbria-fornix lesions were notdifferent from intact controls when their locomotor activity was tested 3.5 months followingsurgery (Greene et al., 1994; Jeltsch et al., 1994). In a more recent experiment (Casselet al., 1998) testing was started 4.5 months following surgery and we found that hippocam-pal rats were more active in their home cage than rats with fimbria-fornix lesions. In thesame study, hippocampal rats experienced a greater impairment than fimbria-fornix ratsin a radial maze task that assessed working memory; however, in the Morris water mazetest, hippocampal rats and rats with fimbria-fornix lesions showed comparable levels ofperformance. In attempting to account for these results, we hypothetized that rats withfimbria-fornix lesions, unlike those with hippocampal removal, might have partially recov-ered their ability to deal with spatial information during the long delay between surgeryand behavioral testing. This is a possibility since in the experiment by Cassel et al. (1998),neither the cingular bundle nor the ventral route septohippocampal projections weredamaged by the fimbria-fornix lesions. Thus, the remaining septohippocampal fibers mayhave contributed to the recovery by compensating for some of the initial deficits (e.g.,Gage and Bjorklund, 1986). Moreover, we have also pointed out that when the fimbria-fornix pathways are disrupted, other afferent (and efferent) hippocampal fibers, such asthose from (to) the entorhinal cortex, are still present. Interestingly, lesions of the entorhinalcortex or transections of the perforant path that connect the entorhinal cortex to thehippocampus were also reported to disrupt spatial memory (Otto et al., 1997; Skelton and

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 3

McNamara, 1992; but see Galani et al., 1998a) and to increase locomotor activity (e.g.,Galani et al., 1998b; but see Yee et al., 1995). Thus, it does seem possible that thediminished deficit found in rats sustaining fimbria-fornix lesions in the study by Casselet al. (1998) may be due to the fact that the entorhinal cortex, cingular bundle, and ventralfibers continued to provide the hippocampus with useful information.

In the present experiment we have tried to clarify whether a short as opposed to a longpostsurgical delay could differentially affect spatial memory capabilities in rats subjectedto lesions of the hippocampus, the fimbria-fornix, or the entorhinal cortex. Adult malerats sustained electrolytic lesions of the fimbria fornix pathways (group FIFX), ibotenicacid lesions of the hippocampus (group HIPP), or N-methyl-D-aspartate (NMDA) lesionsof the medial region of the entorhinal cortex (group ENT). A group of sham-operatedrats was used as controls (group SHAM). All rats were subsequently tested for workingand reference memory in the eight-arm radial maze. Testing was started 1 month aftersurgery for half of the rats in each group. For the other half, testing was started 6.5 monthsfollowing surgery. The behavioral testing protocols used were those that, in the study byCassel et al. (1998), had shown differences between rats with fimbria-fornix lesions andthose with hippocampal removal. After completion of behavioral testing, all rats weresacrificed for histological examination.

MATERIAL AND METHODS

Subjects

The study used Long–Evans male rats obtained from R. Janvier (France) at the age ofapproximately 3 months (body weight, 380–420 g). All rats arrived at the laboratory atthe same time. They were housed singly on sawdust in transparent Makrolon cages(42 3 26 3 15 cm) with food and water available ad libitum. The colony and experimentroom were maintained on a 12:12 h light–dark cycle (lights on at 07:00) under controlledtemperature (218 6 18C). All procedures involving animals and their care were conductedin conformity with the institutional guidelines that are in compliance with national andinternational laws and policies (Council Directive 87848, October 19th, 1987, Ministerede l’Agriculture et de la Foret, Servic Veterinaire de la Sante et de la Protection Animale;Permission 6212 to J-C.C; NIH Publication N086-23, revised 1985).

Lesion Surgery

All surgeries were performed at the same time (61 week) under aseptic conditions,using equithesin anesthesia (3.3 ml/kg, ip). Lesions of the fimbria-fornix pathways weremade by passing a rectifed current of 1 mA for 40 s through an epoxylite-coated stainlesselectrode (0.15 mm in diameter), uninsulated at the tip (approximately 0.5 mm), whichwas lowered into the brain at five sites according to the coordinates shown in Table 1.The incisor bar of the stereotaxic apparatus was placed 3.0 mm below the interaural line.Compared to our previous study (Cassel et al.; 1998), the body weight of the rats usedin the present experiment was 396 6 2.7 g, approximately 100 g heavier. Therefore, weran pilot studies to check the stereotaxic coordinates. In these pilot studies we found thatlesions made according to our previous coordinates damaged less than half of the axons

4 GALANI ET AL.

TABLE 1

Stereotaxic Coordinates for Electrolysis of the Fimbria-Fornix Pathways, Ibotenate

Injections into the Hippocampus, and NMDA Injections into the Entorhinal Cortex

Lesion A-P M-L D-V

Fimbria-fornix (in mm from lambda) 16.0 0 23.815.7 60.8 24.015.4 61.8 24.3

Hippocampus (in mm from bregma) 25.4 65.0 26.6,b 25.9,b 25.2,b 24.5b

25.4 64.2 24.4,a 23.9a

24.8 64.9 27.2,a 26.4a

24.8 64.3 27.7,b 27.1,b 23.5a

24.0 63.7 22.7c

24.0 62.5 22.7,a 21.8a

23.2 63.0 22.7c

23.2 61.4 23.1,a 22.3a

22.4 61.0 23.5a

Entorhinal cortex (in mm from bregma) 28.3 63.5 27.5, 27.0, 26.028.0 64.0 28.5,27.5

Note. Coordiantes are given according to lambda or bregma (Paxinos & Watson, 1986). A-P, anteroposterior;M-L, mediolateral; D-V, dorsoventral. Amounts of ibotenic acid injected at each site: a0.05 ml, b0.06 ml, c0.08 ml.

in the fimbria-fornix pathways, whereas the coordinates shown in Table 1 permitted usto obtain a lesion extent similar to the one obtained by Cassel et al. (1998). Hippocampallesions were performed using multiple injections of ibotenic acid (Biosearch Technologies,San Rafael, CA) according to the technique described by Jarrard (1989), but with stereo-taxic coordinates adapted to Long–Evans rats (Table 1). The ibotenic acid (concentration:10 mg/ml) was injected into various hippocampal sites with a 5-ml Hamilton syringeequipped with a glass micropipette and held in a Kopf Microinjector. The amount ofibotenic acid injected varied from 0.05 to 0.08 ml depending on the injection site (seeTable 1 for details).

Lesions of the entorhinal cortex were performed using multiple injections of NMDA(Sigma). NMDA was employed since NMDA and IBO are thought to be generally similarin terms of their toxic qualities (see Jarrard & Meldrum, 1993) and since we had previouslyestablished that cells in the entorhinal cortex could be selectively removed using thisexcitotoxin (see Coutureau et al., 1999). NMDA was disolved in PBS (pH 7.4) in orderto obtain a 40 mM solution and was injected into various sites (see Table 1) of theentorhinal cortex. A cannula (external diameter 0.28 mm) connected via a polyethylenetube and a 10-ml Emire microsyringe was driven by an automated syringe pump. Thecannula was lowered into the entorhinal cortex with a mediolateral angle of 158. At eachsite, 0.2 ml of NMDA was injected over a period of 1 min. The cannula was left in placeafter each injection for 1 min. At the time of surgery, there were 21 sham-operated rats(SHAM), 24 rats with entorhinal cortex lesions (EC), 21 with fimbria-fornix lesions(FIFX), and 22 with hippocampal lesions (HIPP).

Sham operations consisted of using the same approach as for the entorhinal cortexlesions in 7 rats, hippocampal lesions in 7 rats, and fimbria-fornix lesions in 7 rats, exceptthat no neurotoxic or electrolytic lesion was performed (needle or electrode lowered into

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 5

the brain, no injection or no current). According to their postsurgical scores in a locomotoractivity test (not reported), rats in each group were then randomly assigned to either ashort (1 month after surgery) or a long (6.5 months after surgery) period of recoverybefore being tested in the eight-arm radial maze.

Eight-Arm Radial Maze

Spatial reference memory and working memory abilities of rats were assessed in theeight-arm radial maze. In the rats with the short period of postoperative recovery, mazetesting was started 1 month after surgery. In the rats with the long period of recovery,maze testing began 6.5 months after surgery. The gray, polyvinylchloride maze was locatedin an experimental room with several obvious extramaze visual cues (pictures on the wall,chair, computer, desk, fan). The octagonal central platform was 40 cm in diameter andthere were eight arms radiating from the platform. The arms were 56 cm long and 10 cmwide. A recessed food well was located 3 cm from the end of each arm. There was a 3-cm-high wall along the edge of the arms, and 30 3 20 cm walls were attached to eacharm entrance. Access to the arms was controlled by transparent Perspex guillotine doors.The maze, elevated 68 cm above the floor level, was illuminated by a 40-W white lightlocated 180 cm above the center of the platform. Photocells connected to a computerwere used to record the choices of the rats (two photocells per arm, one 12 cm from theentrance and the other 10 cm from the end of the arm).

Before training was started, the body weight of each rat was progressively reduced to80% of the free-feeding value (over 10 day). It was maintained at this value throughouttesting. Water was available ad libitum.

Preliminary training. Preliminary training was run over 5 days. On the first 2 daysall arms were closed and each rat was placed on the central platform (several 45-mgpellets on the floor) for a maximum of 10 min or until all the pellets had been eaten. Thethird day all arms were open, one food pellet was placed at the entry of each arm, andeach rat remained on the maze until all pellets had been eaten or for a maximum of 10min. On days 4 and 5, this procedure was repeated with one pellet placed at the end of eacharm. Following preliminary training, the rats were tested using three testing procedures.

Place learning, working memory test. From testing days 1 to 20, each rat was givenan “information” trial followed by four testing trials. For the information trial, one randomlyselected arm was opened and baited with four pellets. The rat was placed on the centralplatform and given access to only the open baited arm. The rat was removed from themaze when the four pellets had been eaten and returned to its home cage for 30 s. Duringthis time, the remaining doors were opened and four pellets were again placed in theinformation arm. The rat was then placed in the center and remained on the maze untilthe correct arm was visited and the pellets were eaten (Trial 1). The experimenter baitedthe same arm again and without delay replaced the rat in the center of the maze for Trial 2.This testing procedure was repeated for the two remaining trials. When the four trialswere completed, the rat was returned to its home cage, the maze was cleaned, and thenext rat was tested. The arm used for the information run was changed from rat to ratand, for a given rat, from day to day Tn the four testing trials subsequent to the informationtrial, any visit to an arm other than the one which was opened during the information run

6 GALANI ET AL.

was recorded as an error. The above protocol was taken as a “simple” matching to sample(place) task measuring spatial working memory capablities.

Place learning, reference memory test. From days 21 to 33, the procedure was similarto that described above except that the same arm was baited across days for a given rat(but the arm differed from rat to rat). Starting on day 21, a single infomation run wasgiven, and this was followed by the rat being released from the end of one of four armswhich were designated as north, south, west, and east. The information arm was arbitrarilydesignated as N-E regardless of its actual spatial location. The four daily test trials weregiven according to a random sequence (e.g., N, W, E, S), and the correct (baited) armwas the same as the baited one on the information run. As before, this arm was rebaitedwith four pellets before each of the four daily trials. An entry in an unbaited arm wasrecorded as an error. This protocol is considered to provide a simple measure of spatialreference memory capabilities.

Standard working memory test. From days 34 to 57 the rats were tested once a dayusing the standard working memory testing procedure: all arms were baited (one pellet/arm) at the begining of each trial and the rat was allowed to move in the maze until theeight arms had been visited. A reentry into an already visited arm during a trial wasrecorded as an error.

Histological Verification

Rats were sacrificed 5 months (short period) or 11 months (long period) after surgery.They were given an overdose of pentobarbital (100 mg/kg, ip) and transcardially perfusedwith 60 ml of saline followed by 60 ml of 0.1 M phosphate-buffered paraformaldehyde(48C, pH 8). After extraction, the brains were postfixed for approximately 4 h in the samefixative and transferred into a 0.1 M phosphate-buffered 20% sucrose solution for 36–40 h.The brains of all HIPP and EC rats, and those of half SHAM rats were cut into 30-mm-thick horizontal sections, while those of all FIFX rats and the remaining SHAM rats werecut into 30-mm-thick coronal sections. All sections were dried on slides at room temperatureand stained with cresyl violet or for AChE using a method similar to that described byKoelle (1954): ethoprapazine (0.3 mM) was used to block nonspecific cholinesterasesand acetylthiocholine iodide (4 mM) was used as the substrate. Cresyl violet staining wasdone as described by Sirkin (1983).

In order to compare the extent of the loss of the hippocampus in the two postoperativedelay conditions, a procedure was followed that permitted quantification of the lesions.Using the horizontal sections of the brains, we measured the area of the damage at 5dorsoventral levels using a graphic table connected to a computer and a camera lucidamounted on an Olympus microscope. The five levels were 23.1, 23.8, 24.7, 26.1,and 27.6 mm from bregma (Paxinos & Watson, 1986). Left and right brain sides wereassessed separately and the distinction was kept for statistical analysis. Each measure wasconverted into a percentage of the surface of the hippocampus in an intact rat at thesame scale.

Statistical Analysis

All data were analyzed with analyses of variance (Winer, 1971). As there was nosignificant difference among the average scores of the rats from the three subgroups of

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 7

sham operations, all rats of these subgroups were considered in a single control group(SHAM). ANOVA considered factors Group, Delay, and Trial or Trial Block for behavioraldata and Delay, Side, and Depth level for the extent of the hippocampal lesions. Two bytwo comparisons employed the Newman–Keuls test (Winer, 1971).

RESULTS

Extent of the Lesions

The location and extent of the lesions are illustrated in Fig. 1. The data from 4 ratswith hippocampal lesions, 5 rats with entorhinal cortex lesions, and 2 rats with fimbria-fornix lesions were discarded from the study because of insufficient or nonselective lesions.In the remaining EC rats, most of the entorhinal cortex was damaged. Typically, thelesions extended from 24.7 to 27.6 mm ventral to bregma. The medial part of theentorhinal cortex was damaged in all rats. The lateral area was only partially lesioned.In 3 of 9 rats (delay 1) and another 4 of 10 rats (delay 2), the lesion also slightly encroachedonto the subiculum, but it was restricted to its most superficial layers.

In FIFX rats, the lesions completely damaged the dorsal fornix and the fimbria, withonly a thin ventrolateral part of the latter being spared in 5 of 10 rats (delay 1) and another7 of 9 (delay 2). As observed in previous studies (e.g., Cassel et al., 1998), these lesionsslightly encroached onto the most dorsal portion of the lateral septum, the overlying partof the corpus callosum and the most anterior pole of the hippocampus. As a result ofthese lesions, we observed a substantial reduction (estimated to be between 90 and 100%,by comparing with previous work; e.g., Cassel et al., 1998) of the hippocampal AChE-positive reaction products (Fig. 2). There was, however, residual staining in the mostventral portion of the hippocampus (not illustrated). There was no obvious differencebetween the rats from delay 1 and those from delay 2.

Typically, the hippocampal lesions consisted of extensive loss of cells throughout thestructure, and no significant damage to structures adjacent to the hippocampus (e.g.,subiculum or entorhinal cortex) could be found. These lesions were very similar to theones described in a previous article (Cassel et al., 1998). There was some sparing in themost posterior part of the dentate gyrus, but only in ventral portions of the hippocampus.Except for 2 rats discarded from the study, there was no noticeable variation in the lesionextent accross the 10 HIPP rats used for statistical analysis. Also, there did not seem tobe a difference between rats from delay 1 and those from delay 2. This was confirmedby the analysis of our morphometric data (Table 2).

ANOVA failed to show a significant effect of the Delay (F(1, 32) 5 .059), but therewas a significant effect of the Side (F(1, 32) 5 7.46, p , .05) and the Depth level (F(4,128) 5 4.98, p , .001). None of the different interactions was significant. The Side effectwas due to lesions which were slightly but significantly larger on the right than the leftside. The Depth level effect was due to lesions which were relatively smallerat 26.1 mm compared to all other levels ( p , .01).

Behavioral Evaluations

Place learning, working memory. In this task the arm to be baited was changed everyday and four consecutive trials were given after a daily information run. Data are shown

8 GALANI ET AL.

B

FIG. 1. Schematic drawing of the smallest (black areas) and largest (black 1 greyish areas) extent ofhippocampal (HIPP), entorhinal (EC), and fimbria-fornix (FIFX) lesions according to plates redrawn from thestereotaxic atlas of Paxinos and Watson (1986). (A) Short postsurgical delay; (B) long postsurgical delay.Coordinates are in millimeters from bregma. Fimbria-fornix lesions are shown on drawings of coronal sectionsthrough the rat brain. Entorhinal cortex and hippocampal lesions are shown on drawings of horizontal sections.

in Fig. 3. ANOVA of the number of errors showed a significant effect of factors Group(F(3, 69) 5 37.75, p , .001) and Trial (F(3, 207) 5 18.64, p , .001), as well as of theGroup 3 Delay (F(3, 69) 5 3.54, p , .05) and Delay 3 Trial (F(3, 207) 5 5.56,p , .01) interactions. The Group effect (see inset in Fig. 3) was due to the overall numberof errors which was significantly higher in FIFX and HIPP rats than in SHAM and ECrats ( p , .001 in each case). Further, FIFX rats made significantly more errors than HIPPrats ( p , .001). There was no significant difference between SHAM and EC rats. TheTrial effect was due to overall values which were significantly smaller on Trials 2, 3, and4 than on Trial 1 ( p , .001, in all cases). Further analysis of the Group 3 Delay interaction

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 9

FIG. 2. Representative example of the acetylcholinesterase-positive staining on a coronal section throughthe dorsal hippocampus of a sham-operated rat (A) and a rat with a fimbria-fornix lesion (B). Both are fromthe second delay. Scale bar, 500 mm.

shows that the performances of SHAM, EC, and FIFX rats were not significantly differentwhen the animals were tested at Delay 2 (after 6.5 months) compared to Delay 1 (after1 month). Interestingly, within the HIPP groups, rats tested at Delay 2 made fewer workingmemory errors than those tested at Delay 1 ( p , .01). Further, the Delay 2 HIPP grouphad recovered enough after the 6.5 months that they did not differ from either SHAM orEC rats tested at Delay 2. The Delay 3 Trial interaction was mainly due to the fact thaton Delay 2, the overall number of errors made on the first trials was significantly higherthan that found on delay 1 ( p , .001).

Place learning, reference memory. The single baited arm was the same over the 12-day testing period. For each of the four daily trials, the rat was released from the end ofa different arm. Data are shown in Fig. 4. ANOVA indicated a significant effect of factorsGroup (F(3, 69) 5 28.81, p , .001) and Trial (F(3, 207) 5 37.80, p , .001). There was

10 GALANI ET AL.

TABLE 2

Extent of Hippocampal Lesions at 5 Dorsoventral Levels Defined According to Bregma

and Expressed as a Percentage of the Surface Covered by the Hippocampus in an

Intact Rat at an Identical Scale.

Level (mm)

Delay Side 23.1 23.8 24.7 26.1 27.6

1 Left 79.5 75.9 71.8 60.6 75.367.4 64.7 62.9 63.4 63.8

Right 87.3 77.5 82.3 66.8 82.762.7 64.1 64.8 64.8 62.3

2 Left 67.9 73.7 74.5 64.3 74.6610.1 65.7 65.3 64.7 63.4

Right 81.6 81.8 86.6 72.6 74.9610.1 62.8 61.5 64.4 66.8

FIG. 3. Average number (1SEM) of errors recorded in the control (SHAM), entorhinal cortex lesion (EC),fimbria-fornix lesion (FIFX), and hippocampal lesion (HIPP) groups in the radial maze task peformed accordingto a place learning testing protocol placing emphasis on working memory (one arm baited). Filled symbols orbars refer to the groups tested at the first delay (from 1 month postsurgery onward) and open symbols or emptybars to those tested at the second delay (from 6.5 months postsurgery onward). Data are the average numberof errors recorded over 20 days on each trial. The inset shows the mean number (1SEM) of errors with alltrials collapsed.

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 11

FIG. 4. Average number (1SEM) of errors recorded in the control (SHAM), entorhinal cortex lesion (EC),fimbria-fornix lesion (FIFX), and hippocampal lesion (HIPP) groups in the radial maze task peformed accordingto a place learning testing protocol placing emphasis on reference memory (always the same arm baited). Filledsymbols or bars refer to the groups tested at the first delay (from 1 month postsurgery onward) and open symbolsor empty bars to those tested at the second delay (from 6.5 months postsurgery onward). Data are the averagenumber of errors recorded over 12 days presented in 3-day blocks. The inset shows the mean number (1SEM)of errors with all trial blocks collapsed.

no significant effect of Delay. Among all possible interactions, only the Group 3 Trialinteraction was significant (F(9, 207) 5 3.53, p , .001). The overall Group effect wasdue to the number of errors, which was significantly higher in FIFX and HIPP rats thanin SHAM and EC rats ( p , .001, in each case). The difference between SHAM and ECrats, and that between FIFX and HIPP rats, was not significant. The Trial effect was dueto an overall number of errors, which was significantly lower on Trials 2, 3, and 4 thanon Trial 1 ( p , .001, in each case) and on Trial 4 than on trials 2 and 3 ( p , .001, ineach case). The difference between Trails 4 and 3 was not significant. Finally, the significantGroup 3 Trial interaction can be interpreted as reflecting a decrease of the number oferrors within a day between Trails 1 and 2 which was more pronounced in FIFX rats thanin the three other groups.

Standard working memory protocol in the radial maze. In the standard workingmemory procedure all arms were baited. The rat was placed in the center of the mazeand permitted to choose arms until all eight had been visited and the pellets eaten. Dataare shown in Fig. 5. The results of the ANOVA indicated that there was a significanteffect for Groups (F(3, 69) 5 18.07, p , .001) and Trials (F(5, 345) 5 44.47, p , .001).There was no significant effect for Delay 1 compared to Delay 2. The only interaction

12 GALANI ET AL.

FIG. 5. Average number (1SEM) of errors recorded in the control (SHAM), entorhinal cortex lesion (EC),fimbria-fornix lesion (FIFX), and hippocampal lesion (HIPP) groups in the radial maze task peformed accordingto a standard working memory testing protocol (all arms baited and open). Filled symbols or bars refer to thegroups tested at the first delay (from 1 month postsurgery onward) and open symbols or empty bars to thosetested at the second delay (from 6.5 months postsurgery onward). Data are the average number of errors recordedover 24 days presented in four-trial blocks. The inset shows the mean number (1SEM) of errors with all trialblocks collapsed.

that was significant was that for Delay and Trials (F(5, 345) 5 2.58, p , .05). The Groupeffect was due to an overall number of errors, which was significantly larger in FIFX andHIPP rats than in SHAM or EC rats ( p , .01), but also FIFX rats made significantlymore errors than HIPP rats ( p , .01). The difference between SHAM and EC rats wasnot significant. Even though the interaction of Groups 3 Delay was not significant (F(3,69) 5 1.00, p . .10), thus preventing further analysis with posthoc tests, inspection ofFig. 5 suggests that HIPP rats that learned 6.5 months after the operations tended to makefewer errors than HIPP rats that learned 1 month following surgery (average number oferrors/trial, 8.00 6 1.5 and 5.5 6 1.0 in HIPP1 and HIPP2 rats, respectively). The Trialeffect was due to an overall number of errors, which decreased significantly over the firstfour Trials ( p , .05), while the significant Delay 3 Trial interaction indicates a decreasein number of errors between Trails 1 and 2 on the first compared to the second delay.

DISCUSSION

The main findings of the present research are that (1) neutotoxic lesions of the EC hadno effect on working or reference memory, (2) electrolytic FIFX lesions and ibotenate

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 13

lesions of the HIPP impaired reference and working memory, (3) rats with FIFX lesionsmade significantly more working memory (not reference memory) errors than HIPP rats,and (4) while there was little recovery in FIFX rats, a weak but significant recovery wasobserved in HIPP rats on the first working memory task.

The primary purpose of the present experiment was to compare, at two postsurgicaldelays, spatial memory capabilities after selective removal of the hippocampus, electrolyticlesions of the fimbria-fornix, or damage to the entorhinal cortex. In a previous experiment(Cassel et al., 1998) in which rats with FIFX or HIPP lesions were tested in the radialmaze from 4.5 months after surgery onward, it was found that HIPP rats were moreimpaired than FIFX rats. This difference was found for working and for reference memoryperformances using the testing protocols employed in the present experiment. Thoughimpaired, the two lesion groups did not differ in the water maze. As the greater impairmentin HIPP rats compared to FIFX rats did not agree with that reported earlier by Whishawand Jarrard (1995), and as the latter authors started to test their rats 15 days after surgery,we hypothetized that rats with FIFX lesions, unlike HIPP rats, had experienced functionalrecovery over time (Cassel et al., 1998).

In the 1998 article, we speculated that this recovery might be explained by supracallosaland ventral septohippocampal pathways that were spared by the lesions, as well as byundamaged connections with caudal structures, such as the subiculum and entorhinalcortex. Such a view is compatible with that of Hampson et al. (1999), who suggested thatfollowing lesions restricted to hippocampal neurons, extrahippocampal structures trainedafter surgery might, in a limited way, mediate some of the processes normally handledby the hippocampus.

The results of the present experiment failed to support the assumption that the entorhinalcortex could have a role in such recovery processes. Despite the fact that the entorhinalcortex lesions primarily involved the medial portion of this area, these lesions had nosignificant effect on spatial memory capabilities, suggesting that cell bodies located inthis cortical region must play a negligible role in processing spatial information underthe test conditions used. This observation, which does not preclude a possible role inspatial memory of fibers coursing through the entorhinal cortex, agrees with findingsreporting a lack of effects on spatial information processing after lesions restricted to cellsin the lateral or medial entorhinal cortex (Cho and Jaffard, 1995; Galani et al., 1998a;Glenn and Mumby, 1998; Holscher and Schmidt, 1994; Jarrard and Hyko, 1994; Pouzetet al., 1999). It also indirectly indicates that neurons in the entorhinal cortex probably donot compensate for the impairment of spatial memory processes found after hippocampallesions, as their role in such processes seems minor. This view is also in line with thereport by Kesner and Giles (1998), who showed that the pre- and parasubiculum, but notthe medial or lateral entorhinal cortex, are important for working memory processesdealing with spatial information. However, when lesions produce extensive damage toboth the medial and lateral parts of the entorhinal cortex as well as the perirhinal cortex(Otto et al., 1997), or when they are performed with techniques that damage cells andaxons (e.g., Jarrard, 1993), spatial learning and memory may be impaired.

In fact, although the present results generally confirm the behavioral effects of hippocam-pal and fimbria-fornix lesions (e.g., Jarrard, 1995; Morris et al., 1982; O’Keefe and Nadel,1978; Olton et al., 1979; Whishaw and Jarrard, 1995), we did not replicate the greaterimpairment of HIPP lesions found by Cassel et al. (1998). Even though the extent and

14 GALANI ET AL.

location of the lesions in the present study were comparable to those in the study byCassel et al. (1998), the working memory impairment in rats with FIFX lesions wasgreater than that found in HIPP rats. The main difference between the present study andthat by Cassel et al. (1998) is that Cassel et al. first tested their rats in a water maze taskassessing working and reference memory before testing them in the radial maze. In thepresent study, the rats were tested only in the radial maze because performances of HIPPand FIFX rats of the 1998 study were different only in the radial maze. Due to earlierexperience in the water maze, it is possible that the differences found by Cassel et al.(1998) reflected a differential response of HIPP and FIFX rats toward transfer phenomenafrom the water maze to the radial maze. Nevertheless the present observations supportthe conclusions that Whishaw and Jarrard (1995) draw from a study in which HIPP andFIFX rats were tested in a water maze for place learning soon after surgery. The authorsconcluded that while both lesions generally have similar effects on spatial memory/learning processes, the extrahippocampal fibers in the fimbria-fornix pathways are probablyinvolved in additional processes that partially differ from those of the hippocampus.

Interruption of the axons passing in the fimbria-fornix/cingular bundle pathways (e.g.,the “fimbria-fornix” lesion) has often been considered the same as “a hippocampus lesion,”the assumption being that both types of lesions would remove the influence of the hippo-campus on behavior. There are now several studies that indicate that the two lesionsshould not be considered equivalent (Cassel et al., 1998; Jarrard et al., 1984; Whishawand Jarrard, 1995).

Furthermore, in the present experiment, whereas the performances of rats with fimbria-fornix lesions were similar at both delays, HIPP rats had weakly but significantly recoveredin the “simple” working memory task. There is evidence in the literature that rats withhippocampal lesions may recover to some extent when exposed to enriched housingconditions following surgery. Galani et al. (1997) reported that rats with hippocampallesions virtually identical to the present ones, and which had been housed for 30 dayspostsurgery in an enriched environment, showed better performance in both a Morriswater maze reference memory task and a Hebb and Williams test than their counterpartskept for the same period in individual cages. Similar findings have been reported in ratswith hippocampal damage produced with other lesion techniques (see Will and Kelche,1992, for a review). Our present results obtained when lesioned rats remained in the homecages for postsurgical periods of different length seem to indicate that following selectiveremoval of the hippocampus some degree of spontaneous recovery is possible even inthe absence of exposure to enriched housing conditions.

The possibility that some of the effects on behavior resulting from hippocampal lesionsmay change over time could have important implications for theories that try to explainthe function(s) of the hippocampus. Often the studies that are used to support the varioustheories not only differ from each other in terms of the type of lesion performed and thedetails of behavioral testing, but also there are often differences in terms of the durationof the interval between surgery and testing. It does seem that for some processes, thelesion-induced effects may depend on the postoperative recovery period. While there isa paucity of research in this area, it could be that investigators may want to be moreconcerned about the time that passes between when the surgery is performed and whentesting begins.

HIPPOCAMPAL DAMAGE AND POSTSURGICAL RECOVERY 15

REFERENCES

Cassel, J-C., Cassel, S., Galani, R., Kelche, C., Will, B., & Jarrard, L. E. (1998). Fimbria-fornix vs selectivehippocampal lesions in rats: Effects on locomotor activity and spatial learning and memory. Neurobiologyof Learning and Memory, 69, 22–45.

Cho, Y. H., & Jaffard, R. (1995). Spatial location learning in mice with ibotenate lesions of the entorhinalcortex or subiculum. Neurobiology of Learning and Memory, 64, 285–290.

Coutureau, E., Galani, R., Gosselin, O., Majchrzak, M., & DiScala, G. (1999). Entorhinal but not hippocampalor subicular lesions disrupt latent inhibition in rats. Neurobiology of Learning and Memory, 72, 143–157.

Gage, F., & Bjorklund, A. (1986). Compensatory collateral sprouting of aminergic systems in the hippocampalformation following partial deafferentation. In R. L. Isacson & K. H. Pribram (Eds.), The hippocampus(Vol. 3, pp. 33–63). New York: Plenum.

Galani, R., Coutureau, E., & Kelche, C. (1998a). Effects of enriched postoperative housing conditions on saptialmemory deficits in rats with selective lesions of either hippocampus, subiculum or entorhinal cortex.Restorative Neurology and Neuroscience, 13, 173–184.

Galani, R., Jarrard, L. E., Will, B. E., & Kelche, C. (1997). Effects of postoperative housing conditions onfunctional recovery in rats with lesions of the hippocampus, subiculum, or entorhinal cortex. Neurobiologyof Learning and Memory, 67, 43–56.

Galani, R., Weiss, I., Cassel, J-C., & Kelche, C. (1998b). Spatial memory, habituation, and reactions to spatialand nonspatial changes in rats with selective lesions of the hippocampus, the entorhinal cortex or thesubiculum. Behavioural Brain Research, 96, 1–12.

Glenn, M. J., & Mumby, D. G. (1998). Place memory is intact in rats with perirhinal cortex lesions. BehavioralNeurosciences, 112, 1353–1365.

Greene, P. L., Cassel, J-C., Kelche, C., Jeltsch, H., Bratt, A. M., & Will, B. E. (1994). Differential behavioraleffects of supracallosal and infracallosal lesions of the septohippocampal pathways: No ameliorative effectsof oxotremorine or pilocarpine on radial-maze performance. Behavioral Neural Biology, 62, 4–14.

Hampson, R. E., Simeral, J. D., & Deadwyler, S. A. (1999). Distribution of spatial and nonspatial informationin dorsal hippocampus. Nature, 402, 610–614.

Holsher, C., & Schmidt, W. J. (1994). Quinolinic acid lesion of the rat entorhinal cortex pars medialis producesselective amnesia in allocentric working memory (WM), but not in egocentric WM. Behavioural BrainResearch, 63, 187–194.

Hunt, M. E., Kesner, R. P., & Evans, R. B. (1994). Memory for spatial location: Functional dissociation ofentorhinal cortex and hippocampus. Psychobiology, 3, 186–194.

Jarrard, L. (1995). What does the hippocampus really do? Behavioural Brain Research, 71, 1–10.

Jarrard, L. E. (1989). On the use of ibotenic acid to lesion selectively different components of the hippocampalformation. Journal of Neuroscience Methods, 29, 251–259.

Jarrard, L. E. (1993). On the role of the hippocampus in learning and memory in the rat. Behavioral NeuralBiology, 60, 9–26.

Jarrard, L. E., & Hyko, A. (1994). Complementary roles for the hippocampus and entorhinal perirhinal cortexin the acquisition of spatial and nonspatial tasks in the rat, Society for Neuroscience Abstracts, 20, 1015.

Jarrard, L. E., Okaichi, H. Steward, O., & Goldschmidt, R. B. (1984). On the role of hippocampal connectionsin the performance of place and cue tasks: Comparisons with damage to hippocampus. Behavioral Neurosci-ence, 6, 946–954.

Jeltsch, H., Cassel, J-C., Neufang, B., Kelche, C., Hertting, G., Jackisch, R., & Will, B. (1994). The effects ofintrahippocampal raphe and/or septal grafts in rats with fimbria-fornix lesions depend on the origin of thegrafted tissue and the behavioural task used. Neuroscience, 1, 19–39.

Kesner, R. P., & Giles, R. (1998) Neural circuit analysis of spatial workin memory: Role of pre- and parasubiculum,medial and lateral entorhinal cortex. Hippocampus, 8, 416–423.

Koelle, G. B. (1954). The histochemical localization of cholinesterases in the central nervous system of the rat.Journal of Comparative Neurology, 100, 211–235.

16 GALANI ET AL.

McDonald, R. J., Murphy, R. A., Guarraci, F. A., Gortler, J. R., White, N. M., & Baker, A. G. (1997). Systematiccomparison of the effects of hippocampal and fimbria-fornix lesions on acquisition of three configuraldiscriminations. Hippocampus, 7, 371–388.

Morris, R. G. M., Garrud, P., Rawlins, J. N. P., & O’Keefe, J. (1982). Place navigation impaired in rats withhippocampal lesions. Nature, 297, 681–683.

Morris, R. G. M., Schenk, F., Tweedie, F., & Jarrard, L. E. (1990). Ibotenate lesions of hippocampus and/orsubiculum: Dissociating components of allocentric spatial learning. European Journal of Neuroscience,2, 1016–1028.

O’Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Clarendon Press.

Olton, D. S., Becker, J. T., & Handelmann, G. E. (1979). Hippocampus, space, and memory. The Behavioraland Brain Sciences, 2, 313–365.

Otto, T., Wolf, D., & Walsh, T. J. (1997). Combined lesions of perirhinal and entorhinal cortex impair rats’performance in two versions of the spatially guided radial-arm maze. Neurobiology of Learning and Memory,68, 21–31.

Paxinos, G., & Watson, C. (1986). The rat brain in stereotaxic coordinates. New York: Academic Press.

Pouzet, B., Welzl, H., Gubler, M. K., Broersen, L., Veenman, C. L., Feldon, J., Rawlins, J. N., & Yee, B. K.(1999). The effects of NMDA-induced retrohippocampal lesions on performance of four spatial memory tasksknown to be sensitive to hippocampal damage in the rat. European Journal of Neuroscience, 11, 123–140.

Sirkin, D. W. (1983). Critical defatting of frozen brain sections for optimal differentiation with the cresyl violetstain. Stain Technology, 58, 121–122.

Skelton, R. W., & McNamara, R. K. (1992). Bilateral knife cuts to the perforant path disrupt spatial learningin the Morris water maze. Hippocampus, 1, 73–80.

Wan, H., Aggleton, J. P., & Brown, M. W. (1999). Different contributions of the hippocampus and perirhinalcortex to recognition memory. Journal of Neuroscience, 19, 1142–1148.

Whishaw, I. Q., & Jarrard, L. E. (1995). Similarities vs. differences in place learning and circadian activity inrats after fimbria-fornix section or ibotenate removal of hippocampal cells. Hippocampus, 5, 595–604.

Will, B., & Kelche, C. (1992). Environmental approaches to recovery of function from brain damage: A reviewof animal studies (1981–1991). In F. D. Rose & D. A. Johnson (Eds.), Recovery from brain damage (pp.79–103). New York: Plenum.

Will, B., Deluzarche, F., & Kelche, C. (1978). Does post-operative environment attenuate or exacerbate symptomswhich follow hippocampal lesions in rats? Behavioural Brain Research, 7, 125–132.

Winer, B. J. (1971). Statistical principles in experimental design. New York: McGraw–Hill.

Yee, B. K., Feldon, J., & Rawlins, J. N. (1995). Potentiation of amphetamine-induced locomotor activity followingNMDA-induced retrohippocampal neuronal loss in the rat. Experimental Brain Research, 106, 356–364.