Download Dissociating Hippocampal Subregions: A Double

HIPPOCAMPUS 11:626 – 636 (2001)
Dissociating Hippocampal Subregions: A Double
Dissociation Between Dentate Gyrus and CA1
Paul E. Gilbert, Raymond P. Kesner,* and Inah Lee
Department of Psychology, University of Utah,
Salt Lake City, Utah 84112
ABSTRACT:
This study presents a double dissociation between the
dentate gyrus (DG) and CA1. Rats with either DG or CA1 lesions were
tested on tasks requiring either spatial or spatial temporal order pattern
separation. To assess spatial pattern separation, rats were trained to
displace an object which covered a baited food-well. The rats were then
allowed to choose between two identical objects: one covered the same
well as the sample phase object (correct choice), and a second object
covered a different unbaited well (incorrect choice). Spatial separations of
15–105 cm were used to separate the correct object from the incorrect
object. To assess spatial temporal order pattern separation, rats were
allowed to visit each arm of a radial eight-arm maze once in a randomly
determined sequence. The rats were then presented with two arms and
were required to choose the arm which occurred earliest in the sequence.
The choice arms varied according to temporal separation (0, 2, 4, or 6) or
the number of arms that occurred between the two choice arms in the
sample phase sequence. On each task, once a preoperative criterion was
reached, each rat was given either a DG, CA1, or control lesion and then
retested. The results demonstrated that DG lesions resulted in a deficit on
the spatial task but not the temporal task. In contrast, CA1 lesions resulted in
a deficit on the temporal task but not the spatial task. Results suggest that the
DG supports spatial pattern separation, whereas CA1 supports temporal
pattern separation. Hippocampus 2001;11:626–636.
©
2001 Wiley-Liss, Inc.
KEY WORDS:
memory
pattern separation; spatial; temporal; hippocampus;
INTRODUCTION
Computational models of hippocampal function and cellular recording
studies suggest that the hippocampus supports pattern separation or orthogonalization of sensory input information (Marr, 1971; McNaughton,
1989; O’Reilly and McClelland, 1994; Rolls, 1996; Shapiro and Olton,
1994; Tanila, 1999). Models of hippocampal function proposed by Rolls
(1996), O’Reilly and McClelland (1994), and Shapiro and Olton (1994)
suggest that pattern separation may be a function associated specifically with
the dentate gyrus (DG). These models propose that pattern separation is
mediated by a competitive inhibitory network at the level of the DG (Rolls
and Treves, 1998) as well as facilitated by sparse connections in the mossyGrant sponsor: NSF; Grant number: BNS 892-1532; Grant sponsor: Human
Frontiers HFSP; Grant number: RG 0110/1998B.
*Correspondence to: Raymond P. Kesner, Department of Psychology, University of Utah, 380 S. 1530 E., Room 502, Salt Lake City, UT 84112.
E-mail: [email protected]
Accepted for publication 24 May 2001
© 2001 WILEY-LISS, INC.
DOI 10.1002/hipo.1077
fiber system that connect DG neurons to CA3 neurons.
The separation of patterns is accomplished due to the low
probability that any two CA3 neurons will receive mossyfiber input synapses from a similar subset of DG cells.
Shapiro and Olton (1994) also suggested that pattern
separation may be facilitated by projections from the entorhinal cortex to CA1 and also across connections between CA3 and CA1. Thus, all three models suggest that
pattern separation may be a function of the dentate gyrus;
however, Shapiro and Olton (1994) also suggested that
CA1 may separate patterns as well.
To examine the role of the hippocampus in spatial
pattern separation, Gilbert et al. (1998) developed a behavioral paradigm based on measuring short-term memory for spatial location information as a function of spatial similarity between two spatial locations. The results
indicated that lesions of the hippocampus decrease efficiency in spatial pattern separation, which resulted in
impairments on trials with increased spatial proximity,
and hence increased spatial similarity among working
memory representations.
Researchers have demonstrated that items which occur
further apart in a temporal sequence are remembered
better than items which are temporally adjacent (Banks,
1987; Estes, 1986; Madsen and Kesner, 1989). This temporal separation effect is assumed to occur because there
is more interference and a greater need to separate temporally proximal events than temporally distant events.
Based on these findings, Chiba et al. (1994) developed an
experiment to test memory for the temporal order of a
sequence of items, when the number of items between
the two choice items in the sequence varied. The results
of this study demonstrated a temporal separation effect
for a sequence of spatial locations on a radial eight-arm
maze; however, following hippocampal lesions, rats were
significantly impaired. The results suggest that the hippocampus may be involved in separating events in time.
Based on the discrepancies among the computational
models, it is clear that the role of hippocampal subregions
in separating patterns of spatial and temporal information is in need of a more detailed behavioral analysis. The
studies conducted by Gilbert et al. (1998) and Chiba et
al. (1994) demonstrated that the hippocampus is involved in pattern separation, but did not address the lo-
____________________________________
DISSOCIATING FUNCTION OF DENTATE GYRUS AND CA1
calization of a pattern separation mechanism to a specific hippocampal subregion. The current study was designed to determine
whether DG or CA1 was preferentially involved in spatial and/or
temporal pattern separation by testing rats with either dorsal DG
or CA1 lesions on the behavioral paradigms developed by Gilbert
et al. (1998) and Chiba et al. (1994) to assess the location of the
mechanism(s) supporting pattern separation.
627
diameter. One hundred seventy-seven food-wells (2.5 cm in diameter and 1.5 cm in depth) were drilled into the surface of the maze
in evenly spaced parallel rows and columns 2 cm apart. A small,
black start-box was placed on top of the maze surface, centered
perpendicular to the rows of food-wells with the posterior edge of
the box placed along the edge of the apparatus.
Spatial Temporal Order Pattern Separation Task
Apparatus
METHODS
Subjects
Twenty-nine male Long-Evans rats were used as subjects. Each
rat was initially food-deprived to 85% of its free-feeding weight
and allowed continuous access to water. Each rat was trained on
only one of the two tasks and received only one type of lesion.
Spatial Pattern Separation Task Apparatus
The test apparatus for the spatial separation task was a cheeseboard maze (Fig. 1). The surface of the apparatus was 119 cm in
FIGURE 1.
Schematic of spatial separation task apparatus and an
example of a sample phase (A) and a 15-cm choice phase (B). The
object in B in the same location as the object in A is the correct choice,
and the other object is the incorrect choice.
A radial eight-arm maze was used as the test apparatus to assess
temporal pattern separation. The maze consisted of an octagonal
central platform 42 cm in diameter. Eight arms radiated from the
central platform like the spokes of a wheel. Each arm was 71 cm
long and 9.5 cm wide. Each arm had clear Plexiglas sides which
rose above the surface of the arm. A small food-well was drilled 1.5
cm deep at the distal end of each arm. A Plexiglas guillotine door
was located at the juncture between the platform and the arm.
Spatial Pattern Separation Task Preoperative
Training Procedure
A delayed-match-to-sample for spatial location task was used to
assess spatial pattern separation. Each animal received 16 trials per
day. Each trial consisted of a sample phase followed by a choice
phase. During the sample phase (Fig. 1A), a randomly positioned
object covered a baited food-well in one of 15 spatial locations
along the centermost row of food-wells perpendicular to the start
box. The rat was placed in the start box with the guillotine door in
the closed position. The door was then opened, and the animal was
allowed to exit the box, displace the object to receive a food reward,
and return to the box. The same food-well was quickly re-baited, a
second identical object was placed to cover the food-well, and a
third identical object was placed in a different location along the
row of food-wells covering a different unbaited food-well. The
object used in the sample phase was never used in the test phase as
either the correct or foil object. Instead, two objects identical to the
original object were randomly assigned to either cover the correct
food-well or the incorrect food-well on the choice phase, thus
eliminating the possibility of using object and/or odor cues to
choose the correct object. On the ensuing choice phase (Fig. 1B),
the animal was allowed to choose between the two objects. The
object which covered the same food-well as the object in the sample
phase was the correct choice, and the second foil object was the
incorrect choice. Five spatial separations, 15 cm, 37.5 cm, 60 cm,
82.5 cm, and 105 cm were randomly used to separate the correct
object from the foil object . Each time a particular spatial separation (15–105 cm) was presented, the distance between the two
objects was held constant; however, the two objects could be in
different positions along the row of wells on different trials. The
position of the correct object relative to the foil object was counterbalanced with regard to left vs. right and closer vs. further with
respect to the animal across all separations. Once an animal established a criterion of 75% correct based on 80 trials across all spatial
separations, the preoperative training period was ended and the
animal was scheduled for surgery.
628
GILBERT ET AL.
Spatial Temporal Order Pattern Separation Task
Preoperative Training Procedure
Each animal was given one daily trial. Each trial consisted of a
sample phase followed by a choice phase. During the sample phase,
the animal was allowed to visit each of the eight arms once in a
randomly predetermined order which varied each day. The sequence of eight arms was presented to the animal by sequentially
opening each door (one at a time) to allow the animal access to the
arm and the food reward at the end of the arm. Once the animal
retrieved the reward and exited the arm, the door was closed and
the next arm in the sequence was presented. This procedure was
followed until all eight arms had been presented. The choice phase
began immediately following the presentation of the last arm in the
sequence. On the choice phase, two arms on the maze were opened
simultaneously and the animal was allowed to choose between the
two arms. The rule to be learned in order to obtain a food reward
was to always enter the arm which occurred earliest in the sequence. Temporal separations of 0, 2, 4, and 6 were randomly
selected for each choice phase and represented the number of arms
that occurred between the two choice arms in the sample phase.
For example, a 0 separation would involve two arms which followed each other in the sequence, whereas a 6 separation would
involve two arms in which six arms occurred between the two arms
in the sequence. Once an animal reached a criterion of 75% correct
across all temporal separations, with the exception of the 0 separation, based on 32 trials, the animal was scheduled for surgery. Eight
sample phases for each of the temporal separations were presented
across the 32-trial block.
Neurotoxin Surgery
Each animal was then randomly assigned to either receive a
bilateral intracranial infusion of colchicine (2.5 mg/ml; total volume of 3.2 ␮l; two injection sites/hemisphere) into the dorsal
dentate gyrus (spatial task n ⫽ 5, temporal task n ⫽ 4), ibotenic
acid (8 mg/ml; total volume of 1.4 ␮l; one injection site/hemisphere) into dorsal CA1 (spatial task n ⫽ 6, temporal task n ⫽ 4),
or the vehicle, phosphate-buffered sodium chloride (PBS), into
either the dorsal dentate gyrus or CA1 (spatial task n ⫽ 5, temporal
task n ⫽ 5). The sample sizes used in the present experiments are
consistent with prior publications from our laboratory which produced highly significant results and minimal variability (Gilbert et
al., 1998; Kesner et al., 1993). Each animal was prepared for surgery as described previously (Gilbert et al., 1998); however, in this
study, lesions were produced by slowly infusing (0.2 ␮l/min) the
neurotoxin or vehicle intracranially via a 10-␮l Hamilton syringe.
The lesion coordinates for the dorsal dentate gyrus group are 2.7
mm posterior to bregma, 2.1 mm lateral to midline, and 4.0 mm
ventral from skull, and 3.7 mm posterior to bregma, 2.3 mm lateral
to midline, and 4.0 mm ventral from skull. The lesion coordinates
for the CA1 lesion are 3.6 mm posterior to bregma, 2.0 mm lateral
to midline, and 3.2 mm ventral from skull.
The rationale for using two different neurotoxins in these experiments is due to the neurotoxic properties of the toxins and the
proximity of the hippocampal subregions. The choice of colchicine
to lesion the dentate gyrus was based on an extensive literature
which describes the use of colchicine to destroy granule cells. Colchicine has been used extensively to destroy dentate gyrus granule
cells but not pyramidal cells due to its selectivity for a particular
isomer of tubulin, which is found on granule-cell somas but not on
the somas of pyramidal neurons (Mundy and Tilson, 1990). In
addition to the present study, multiple studies have demonstrated
that intracranial infusions of colchicine result in significant cell loss
in the dentate gyrus, with little or no cell loss in the CA fields
(Emerich and Walsh, 1989; Mundy and Tilson, 1990; Narny et
al., 1989; Walsh et al., 1986). With a toxin such as ibotenic acid, it
would be very difficult, if not impossible, to lesion only the dentate
gyrus without causing damage to CA3/4 due to its proximity to the
dentate gyrus. Therefore, it would not be feasible to use ibotenic
acid to lesion the dentate gyrus. In addition, it is not reasonable to
use colchicine to lesion CA1 because of the selectivity of the drug
and the high doses that would be required to generate complete
CA1 lesions. Ibotenic acid was chosen to lesion CA1 based on its
excitotoxic properties and affinity for NMDA receptors found on
pyramidal cells.
Spatial Pattern Separation Task Postoperative
Testing
Following a 7–10-day recovery period from surgery, each animal was again tested on the task, for two blocks of 80 trials over a
2-week period, following the same procedure used in the preoperative training trials.
Spatial Temporal Order Pattern Separation Task
Postoperative Testing
Following a 7–10-day recovery period, each animal was again
tested on the task for one block of 32 trials, following the same
procedure used in the preoperative training trials.
Histology
At the conclusion of all testing, each animal was deeply anesthetized with an intraperitoneal injection of 1.5 ml sodium pentobarbital (70 mg/kg) and perfused intracardially with normal saline,
followed by a 10% formalin solution. The brain was removed from
the skull and stored in a 10% formalin/30% sucrose solution in a
refrigerator (4°C) for 72 h to equalize the tissue-shrinkage rates
across brains. A tissue block (bregma ⫺2.0 through ⬃⫺4.0) containing only the dorsal hippocampus was cut perpendicularly from
each brain. The block was then frozen and cut at 24-␮m sections,
and every third section was mounted on a glass slide (the surfaceto-surface distance between collected sections was 72 ␮m), producing 29 collected sections throughout the dorsal hippocampus
for each brain. The sections were stained with cresyl violet and
examined for histological verification of the lesion placement.
Computer-Based Three-Dimensional Volumetry
The damage to cell layers in each of the three hippocampal
subregions (DG, CA1, and CA3) was measured by computerbased three-dimensional (3-D) volumetry. A subset of animals
(n ⫽ 6) from each lesion group and task was randomly selected for
____________________________________
DISSOCIATING FUNCTION OF DENTATE GYRUS AND CA1
3-D volumetric analysis. This procedure has been used by other
researchers to examine hippocampal volume following hippocampal lesions (Moser, et al., 1995; Moser and Moser, 1998). Twentynine sections from the dorsal hippocampus (bregma ⫺2.0 through
⬃⫺4.0) were used to compute volumetry for each rat. See Lee et
al. (1999) for a detailed description of this method. Briefly, the
sections were projected (final magnification ⫻30) onto tracing
paper using a microslide projector (Bausch and Lomb, NY), and
the boundaries of pyramidal cell layers in CA1 and CA3 and granule-cell layers in DG were traced with an extra-fine tip pen. The
tracing procedure was conducted by an experimenter who was
blind to the experimental design. Prior to tracing, the serial sections were subjected to microscopic inspection with high magnification (⫻40) to verify intact pyramidal cells in CA1 and CA3, and
intact granule cells in the dentate gyrus in the lesioned and control
groups. The microscopic examination also afforded the ability to
determine whether a given outline of a cell layer, especially in a
lesioned area, was produced by infiltration of glial cells. The
boundaries of cell layers heavily infiltrated by glial cells were not
traced. For aligning purposes, the boundaries of the overlying cortices and the corpus callosum were also traced throughout serial
sections.
After all the sections were traced and aligned, those traced sections were scanned into a computer as two-dimensional graphic
files (resolution ⫽ 600 dpi). The intralayer zone of each cell layer
was filled with different gray values assigned to different cell layers:
white for CA1, light gray for DG, dark gray for CA3, and black for
background. These color-coded digitized images were then imported into a commercial software package (Vowxin 1.2.2., Voxar
Co., UK) for three-dimensional reconstruction and volumetry.
The software produced three-dimensional images of reconstructed
cell layers that had only intact pyramidal or granule cells. The cell
layer of each subregion could also be reconstructed individually by
adjusting the threshold for rendering.
Volumetry was carried out based on the reconstructed images.
The software calculated the number of voxels used to generate the
three-dimensionally reconstructed structures. The number of voxels composing an intact cell layer of each subregion (e.g., CA1,
CA3, or DG) was compared between each lesion group and the
control group. Therefore, it was possible to calculate the percent
damage to each subregion of the hippocampus following either
DG or CA1 lesions.
RESULTS
Spatial Pattern Separation Task
The data were grouped into blocks of 80 trials for analysis. This
included one set of preoperative criterion trials and four sets of
postoperative trials. For graphical simplicity, each animal’s performance on postoperative blocks 1 and 2 (Fig. 2, POST 1/2) was
averaged, as was their performance on postoperative blocks 3 and 4
(Fig. 2, POST 3/4). As shown in Figure 2A, the CA1 and dentate
gyrus lesioned groups’ preoperative performance matched the pre-
629
operative performance of the control group across all spatial separations. On the first two blocks of postoperative trials, as shown in
Figure 2B, the CA1 lesion group’s performance matched the performance of the control group across all spatial separations,
whereas the DG lesioned group was significantly impaired on postoperative trials across all spatial separations, with the exceptions of
the largest 82.5-cm and 105-cm separations. Similarly, Figure 2C
shows that on the third and fourth postoperative blocks of testing,
the DG lesioned animals were significantly impaired on all spatial
separations, with the exception of the 82.5-cm and 105-cm separations. Thus, DG lesioned animals show no indication of recovery
of function or a transitive deficit. On short and medium separation
trials (15– 60 cm) with an increased overlap of distal cues and
presumably increased spatial similarity among distal cues, DG lesioned animals were impaired. However, on trials with increased
spatial separation (82.5–105 cm), less overlap of distal cues resulted in less spatial similarity; the DG lesioned animals performed
the task as well as controls. A linear trend analysis of the DG
lesioned group’s average postoperative performance across separations revealed a significant linear increase in performance as a function of increased spatial separation (F(1,12) ⫽ 74.63; P ⬍ 0.0001).
Since the CA1 lesions were restricted to dorsal CA1, there is a
concern that there may be sparing in CA1 which explains the lack
of a deficit. However, recent unpublished data from our laboratory
showed that rats with significant dorsal and ventral CA1 lesions
perform this task as well as controls. The results of the present
study are consistent with the hypothesis that the DG may serve to
separate incoming spatial information into patterns or categories
by aiding the storage of one place as separate from another place
into CA3. It is proposed that DG, but not CA1, lesions result in a
decrease in efficiency in pattern separation which may result in an
impairment on trials with increased spatial similarity or interference among spatial working memory representations.
A repeated-measures three-way ANOVA with lesion group
(control, CA1, and DG) as the between-factor and block (PRE,
POST 1, POST 2, POST 3, and POST 4) and separation (15 cm,
37.5 cm, 60 cm, 82.5 cm, and 105 cm) as the within-factors
revealed a significant lesion effect (F(2,13) ⫽ 5.82, P ⬍ 0.05), a
significant block effect (F(4,52) ⫽ 4.07; P ⬍ 0.01), and a significant
separation effect (F(4,52) ⫽ 14.92; P ⬍ 0.0001). Furthermore, the
analysis revealed a significant block ⫻ lesion interaction effect
(F(8,52) ⫽ 2.93; P ⬍ 0.01), and a significant separation ⫻ lesion
interaction effect (F(8,52) ⫽ 3.49; P ⬍ 0.01).
A Newman-Keuls comparison test of the main effect for lesion
showed that the DG group’s performance was significantly different (P ⬍ 0.05) from the control and CA1 lesioned group’s performance. However, the CA1 group’s performance was not significantly different from the control group’s performance. A
Newman-Keuls test of the block ⫻ lesion interaction effect revealed no significant differences between the control, DG lesioned,
or CA1 lesioned groups’ preoperative performance. However, the
analysis showed that the DG lesioned group’s postoperative performance was significantly different (P ⬍ 0.05) from both the
control and CA1 lesioned groups’ postoperative performance. The
postoperative performance of the CA1 lesioned group did not
differ from the postoperative performance of the control group.
630
GILBERT ET AL.
Finally, a Newman-Keuls test of the distance ⫻ lesion interaction
effect showed that the DG animals’ performance on the 105-cm
separation was not significantly different from the other groups’
105-cm performance. However, their performance on short separations was significantly different (P ⬍ 0.05) from both of the
other groups’ performances on all separations.
Spatial Temporal Order Pattern Separation Task
The data were grouped into blocks of 32 trials for analysis. This
included one block of preoperative criterion trials and one block of
postoperative trials. The data shown in Figure 3A indicate that the
DG and CA1 lesioned groups’ preoperative performance matched
the preoperative performance of the control group across all temporal separations. The data shown in Figure 3B indicate the DG
group’s postoperative performance also matched the performance
of the control group on postoperative trials across all temporal
separations. In contrast, the data in Figure 3B indicate that the
CA1 lesion group was significantly impaired on postoperative trials
relative to control and DG lesioned animals. The data suggest that
the mechanism which supports the separation of temporal events
may reside within the CA1 region of the hippocampus.
A repeated-measures three-way ANOVA with lesion group
(control, DG, or CA1) as the between-factor and block (PRE or
POST) and temporal separation (0, 2, 4, or 6) as the within-factors
revealed that there was a significant main effect for block (F(1,10) ⫽
10.66; P ⬍ 0.01) and a significant main effect for temporal separation (F(3,30) ⫽ 19.68; P ⬍ 0.0001). Furthermore, the analysis
revealed a significant block ⫻ lesion interaction (F(2,10) ⫽ 9.40;
P ⬍ 0.01). The significant main effect for temporal separation
demonstrates that the animals’ performance tended to increase as a
function of increased temporal separation between spatial events.
This illustrates the temporal separation effect.
A Newman-Keuls comparison test of the block ⫻ lesion interaction revealed no significant differences between the preoperative
performances of the control, DG lesioned, and CA1 lesioned
groups. On postoperative trials, the analysis revealed no significant
difference between the control and CA1 lesioned groups’ performance. However, the CA1 lesioned group’s postoperative performance was found to be significantly different (P ⬍ 0.05) from the
postoperative performance of both the DG lesioned and control
groups. The results suggest that CA1, but not DG, lesions reduce
efficiency in separating events in time.
Histology
A photomicrograph (⫻20 magnification) of a representative
vehicle-infused control lesion is shown in Figure 4A. Intracranial
infusions of the vehicle did not tend to produce significant damage
to any brain region. Photomicrographs (⫻20 magnification) of
representative dorsal DG and CA1 lesions are shown in Figure
FIGURE 2.
A: Mean percent correct performance as a function of
spatial separation of control group, CA1 lesion group, and dentate
gyrus lesion group on preoperative trials. B, C: Mean percent correct
performance as a function of spatial separation of control group, CA1
lesion group, and dentate gyrus lesion group on two averaged blocks.
For graphical simplicity, each animal’s performance on postoperative
blocks 1 and 2 (B, POST 1/2) was averaged, and so was their performance on postoperative blocks 3 and 4 (C, POST 3/4).
____________________________________
DISSOCIATING FUNCTION OF DENTATE GYRUS AND CA1
631
damage to the entorhinal cortex in the present experiments. In the
present experiments, there was minimal damage to the hippocampus anterior to ⫺2.3 AP or posterior to ⫺4.3 AP. Although it is
possible that some small quantities of neurotoxin spread into the
ventricles, there was little evidence of cell loss in regions adjacent to
the ventricles or global enlargement of the ventricles in either lesion group.
Figure 5A shows a representative photomicrograph (⫻40) of the
upper blade of the dorsal dentate gyrus (approximately 3.6 mm
posterior to bregma) in a vehicle-infused control animal. Figure 5B
shows a representative photomicrograph (⫻40) of a similar section
of dentate gyrus in an animal infused with colchicine. It is clear that
intracranial infusions of colchicine resulted in a significant decrease
FIGURE 3.
A: Mean percent correct performance as a function of
temporal separation of control group, dentate gyrus lesion group, and
CA1 lesion group on preoperative trials. B: Mean percent correct
performance as a function of temporal separation of control group,
dentate gyrus lesion group, and CA1 lesion group on postoperative
trials
4B,C, respectively. All lesions tended to be quite complete within
the targeted subregion of the dorsal hippocampus, with some damage to any other hippocampal subregions. See below, “Three-Dimensional Reconstruction and Volumetry,” and Figures 6 and 7
for a quantitative analysis of the damage to each hippocampal
subregion following DG and CA1 lesions. Prior experiments have
also shown similar lesion selectivity within hippocampus (Emerich
and Walsh, 1989; Mundy and Tilson, 1990; Narny et al., 1989;
Walsh et al., 1986). Other studies have also demonstrated highly
selective DG lesions with minimal CA field damage following
long-term adrenalectomy (Conrad and Roy, 1993). Similar to
prior experiments which did not report entorhinal cortex damage
following colchicine or ibotenic acid infusions into the dorsal hippocampus (Goldschmidt and Stewart, 1980; Jarrard and Meldrum, 1993; Jarrard et al., 1984), there was minimal evidence of
FIGURE 4.
Bilateral photomicrographs (ⴛ20) of a representative control (A), dentate gyrus lesion (B), and CA1 lesion (C).
632
GILBERT ET AL.
FIGURE 5.
Photomicrographs (ⴛ40) of a representative section
of (A) upper blade of dentate gyrus (⬃3.6 mm posterior to bregma) in
a vehicle-infused animal, (B) similar section of dentate gyrus in a
colchicine-infused animal, (C) CA1 (⬃3.6 mm posterior to bregma)
in a vehicle-infused animal, and (D) similar section of CA1 in an
ibotenic-acid infused animal.
in the number and density of neurons within the dentate gyrus and
an increase in glial cells. Damage of a similar magnitude was also
found at 2.8 mm and 4.3 mm posterior to bregma. Figure 5C
shows a representative photomicrograph (⫻40) of a section of
dorsal CA1 (approximately 3.6 mm posterior to bregma) in a control animal. Figure 5D shows a representative photomicrograph
(⫻40) of a similar section of CA1 in an animal infused with ibotenic acid. It is clear that there is a significant decrease in the
number and density of neurons and an increase in glial cells within
CA1 following ibotenic acid infusions. The damage was similar in
magnitude at 2.8 mm and 4.3 mm posterior to bregma.
struction revealed that the damage was most pronounced in the
medial portion of CA1 (arrow in Fig. 6), which reflects the location
of drug injection. A small quantity of ibotenic acid also diffused
ventrally to the granule-cell layer of DG, which caused 18% reduction in volume in DG in the CA1-lesioned group. The upper blade
of the granule-cell layer was the main area affected by this diffusion
from CA1 (compare arrows in 3-D reconstructions in the DG
regions between the control and the CA1-lesion group in Fig. 6).
Furthermore, the cell loss within the CA3 subregion of the dorsal
hippocampus was minimal (11%). This is very important, since
there is no difference in glutamate receptor sensitivity to ibotenic
acid between CA1 and CA3. Therefore, ibotenic acid infusions
into the CA1 subregion produced significant damage to CA1, but
resulted in minimal damage to either DG or CA3.
The colchicine infusions into DG showed marked destruction
(95%) of granule cells in DG compared to controls. However,
colchicine infusions resulted in a relatively small amount damage
in the CA1 pyramidal-cell layer (18%), but CA1 maintained the
identical 3-D morphology compared to controls. This reduction in
volume came mainly from thinning of pyramidal-cell layers in
CA1 after the colchicine injection into DG. The reason for this
Three-Dimensional Reconstruction and
Volumetry
Three-dimensional images of the damage to the cell layers of
each subregion of the dorsal hippocampus following either DG or
CA1 lesions are shown in Figure 6. The volumetric data of the
percent damage to each hippocampal subregion following either
DG or CA1 lesions are shown in Figure 7. Infusions of ibotenic
acid into CA1 produced an 83% reduction in volume of the CA1
pyramidal-cell layer compared to control animals. The 3-D recon-
____________________________________
DISSOCIATING FUNCTION OF DENTATE GYRUS AND CA1
633
FIGURE 6.
Three-dimensional reconstructed images of intact
cell layers in dorsal hippocampal subregions of representative animals
from the control, CA1, and DG lesion groups. Viewing angles for the
reconstructions are indicated at bottom (A ⴝ anterior, P ⴝ posterior,
R ⴝ right, L ⴝ left). CA1 lesions produced damage (83%) in mostly
the medial part (arrow) of CA1 pyramidal-cell layers, leaving CA3
pyramidal-cell layers relatively intact (11% reduction in volume). In
the CA1-lesioned rats, there was some damage (24%) in DG upper
blades (arrow), but not in lower blades compared to intact DG upper
blades (arrow) in controls. Dentate gyrus lesions produced significant
cell loss (95%) in granule-cell layers in DG compared to controls.
Dentate gyrus lesions produced a relatively small reduction (18%) in
the volume of CA1 pyramidal-cell layers compared to controls. The
pyramidal-cell layers in CA3 were not affected by colchicine injections.
thinning of CA1 pyramidal-cell layers with colchicine is still unknown but may likely be due to mechanical damage from the
infusion cannulae. Furthermore, colchicine infusions into DG resulted in no cell loss in the CA3 subregion of the hippocampus.
Therefore, infusions of colchicine into the dorsal dentate gyrus
resulted in significant damage to the DG and minimal or no damage to CA1 and CA3, respectively.
Xavier et al. (1999) conducted a stereological analysis of the
hippocampal subregions after multiple injections (nine injections
per hemisphere) of colchicine into DG. Their volumetric data
showed remarkable resemblance to the present data. Xavier et al.
(1999) reported 87% cell loss in DG compared to 95% damage in
the present experiments. Furthermore, Xavier et al. (1999) reported 21% cell loss in CA1 and 0% loss in CA3, compared to
18% loss in CA1 and 0% loss in CA3 in the present experiments.
The comparisons between the study conducted by Xavier et al.
(1999) and the present study illustrate two important points. First,
multiple injections, as many as 18 bilateral injections of colchicine
into DG, made little difference with respect to the percentage of
damage in volume to the neurons in any hippocampal subregion
compared to the four bilateral injections in the present study.
Second, for the volumetry for principal cell types (e.g., pyramidal
cells in CA1 and CA3 and granule cells in DG) in the hippocampus, the computer-based volumetry with 3-D reconstruction can
generate as reliable a set of volumetric data as stereology can produce.
DISCUSSION
Based on the findings of Gilbert et al. (1998) and Chiba et al.
(1994), it appeared that the hippocampus was involved in mediating spatial and spatial temporal order pattern separation. However,
from these studies, it was not clear whether the subregions of the
634
GILBERT ET AL.
FIGURE 7.
Mean percent damage to each dorsal hippocampal
subregion (CA1, CA3, and DG), following (A) colchicine infusions
into dorsal DG and (B) ibotenic acid infusions into dorsal CA1.
hippocampus function as an ensemble to separate patterns of incoming information, or whether a specific subregion may be responsible for this process, as suggested by computational models
(O’Reilly and McClelland, 1994; Rolls, 1996; Shapiro and Olton,
1994). If efficient separation of patterns is dependent on a functional hippocampal ensemble, then a lesion to any region of the
hippocampus should produce a deficit in both tasks. However, the
current data demonstrated that it is possible to dissociate behaviorally the functions of DG and CA1. On the spatial separation
task, rats with DG lesions were significantly impaired; however,
rats with CA1 lesions matched the performance of controls. In
contrast, on the temporal separation task, rats with CA1 lesions
showed significant impairments, whereas rats with DG lesions
matched the performance of the control group. The lack of a deficit
in the CA1 group on the spatial task indicates that an intact hippocampal system is not necessary for accurate spatial pattern separation. Similarly, the lack of a deficit in the DG group on the
temporal task indicates that an intact hippocampal system is not
necessary for the accurate separation of patterns of temporal information. Furthermore, the data indicate that DG is involved in
separating spatial patterns but not temporal patterns, whereas CA1
is involved in separating temporal patterns but not spatial patterns.
The finding that the DG is involved in spatial pattern separation
offers support for the models of Rolls (1996), O’Reilly and McClelland (1994), and Shapiro and Olton (1994). The finding that
CA1 is involved in temporal pattern separation also supports the
model of Shapiro and Olton (1994). The study represents one of
the first behavioral double dissociations between subregions of the
hippocampus.
The DG lesioned group’s performance on the spatial pattern
separation task increased as a function of increased spatial separation between the correct object and the foil on the test phases. The
DG lesioned group performed at chance on 15-cm and 37.5-cm
separations but showed some improvement on 60-cm separation.
However, the DG lesioned group matched their preoperative performance and the performance of controls on the 82.5-cm and
105-cm separations. The graded nature of the impairment and the
significant linear increase in performance as a function of increased
separation illustrate the deficit in pattern separation. Furthermore,
the data do not support a deficit in working memory, rule learning,
an inability to form internal representations of their environment,
or consolidation, since the DG lesioned animals performed the
task as well a controls when the separation was large. In a study
conducted by Gilbert et al. (1998) using this same task, transfer
tasks were conducted to examine how animals solved this particular task. On the first transfer task, the sample phase was conducted
as described above; however, the choice phase was conducted in the
dark. Rats showed a significant deficit on this task. These data
suggest that even though allothetic and ideothetic information
may be used to solve this task, ideothetic information alone is not
sufficient for accurate task performance. On the second transfer
task, the sample phase was conducted as described above; however,
on the choice phase, the entire maze was shifted either to the left or
the right at the distance of the choice phase separation. Therefore,
on the choice phase, one object was in the same location as the
sample phase object relative to the maze and start box, whereas the
other object was in the same location as the sample phase object
relative to the environmental cues. Rats tended to choose the object
in the same location as the sample phase object relative to the
environmental cues. The data demonstrate that animals were relying on relationships among distal environmental cues to solve the
task. Based on the findings mentioned above, the most parsimonious interpretation of the data is that DG lesions decrease efficiency
in spatial pattern separation.
On the spatial temporal order task, it is assumed that arms
which occurred further apart in a temporal sequence were remembered better than arms which were temporally adjacent because
there was more interference and a greater need to separate temporally proximal events than temporally distant events. Therefore,
the animals’ performance on the task improved as a function of
increased temporal separation between the choice arms, which
illustrates the temporal separation effect. On postoperative trials,
the CA1 lesioned animals showed a similar increase in performance
as a function of temporal separation; however, their accuracy was
decreased relative to their preoperative performance. These data
suggest that CA1 lesions result in decreased efficiency in separating
temporal events in time. Since this task requires spatial memory
____________________________________
DISSOCIATING FUNCTION OF DENTATE GYRUS AND CA1
and the two arms which were used in the choice phase were spatially adjacent, one may question why the DG lesion group did not
display a deficit on this task. Based on the configurations of the
room where testing took place, the distal cues used to differentiate
two adjacent arms of the maze were on average 104 cm apart. Thus,
just as in the spatial pattern separation task, animals with DG
lesions were able to differentiate, with a high degree of accuracy,
between two different locations which were 105 cm apart. Furthermore, probe trials were conducted where the choice phase arms
were separated by one arm spatially and there were no differences
between these trials and the trials where the arms were spatially
adjacent.
The three-dimensional technique applied in the present experiments has proven useful in accurately estimating either the absolute volume of a neural structure such as the lateral geniculate body
of the cat and the dog (Lee et al., 1999) or the relative volume of the
hippocampus after neurotoxic damage (Moser et al., 1995; Moser
and Moser, 1998). The results of the 3-D reconstructions and the
volumetry analysis in the present study revealed that lesions of the
CA1 resulted in 83% cell loss in CA1, 24% cell loss in DG, and
11% cell loss in CA3 on average. Dentate gyrus lesions resulted in
95% cell loss in DG, 18% cell loss in CA1, and 0% cell loss in CA3
on average. Therefore, the lesions resulted in significant cell loss
within the targeted subregion and some cell loss within other hippocampal subregions. These data are comparable to prior experiments that used stereology to analyze cell loss in the hippocampus
following multiple injections (nine/hemisphere) of colchicine
(Xavier et al., 1999). Furthermore, this present technique provides
the ability to generate 3-D reconstructions of the hippocampal
subregions following DG and CA1 lesions, so that patterns of
damage can be visualized.
If the DG is lesioned, it is clear that information can bypass DG
within the hippocampus via perforant path connections to CA3
and CA1. However, if CA1 is the primary output from the hippocampus, how can the information be transmitted out of the
hippocampus once CA1 is destroyed? This is very important, since
other studies have also reported a lack of a deficit in spatial memory
following selective damage to CA1 (Mizumori et al., 1995; Jarrard,
1978; Davis et al., 1988). Since the CA1 lesions in the present
study were restricted to dorsal CA1, it is possible that the sparing in
ventral CA1 could support the transfer of information from CA1
out of the hippocampus. However, recent unpublished data from
our laboratory has shown that rats with significant dorsal and ventral CA1 lesions perform this task as well as controls. Therefore, the
information does not appear to be transmitted out of the hippocampus via ventral CA1. It is suggested that the information is
passed via direct CA3 extrahippocampal connections. CA3 has
direct projections to the medial and lateral septal nuclei (Amaral
and Witter, 1995; Gaykema et al., 1991; Risold and Swanson,
1997). The lateral septum has connections with the medial septum
(Jakab and Leranth, 1995), and, in turn, the medial septum has
projections to the subiculum and eventually entorhinal cortex
(Amaral and Witter, 1995; Jakab and Leranth, 1995). Thus, it is
possible for CA3 output to bypass the CA1 region.
The neurons which comprise CA1 have direct connections with
neurons in the prefrontal cortex (Jay and Witter, 1991; Verwer et
635
al., 1997). Specifically, CA1 neurons have been shown to project to
the medial and lateral prefrontal cortices in the rat (Jay and Witter,
1991; Verwer et al., 1997). Using the temporal pattern separation
task described in the present study, Chiba et al. (1997) showed that
rats with lesions in the medial prefrontal cortex show a deficit
similar to CA1 lesions. Therefore, these data, coupled with the
results of the present experiment, suggest that the medial prefrontal cortex and CA1 may be involved in mediating memory for
temporal order and may support pattern separation for temporal
events.
The results of the present experiment illustrate that the DG and
its mossy-fiber connections support the separation of incoming
patterns of spatial information, whereas the mechanism which supports the separation of temporal events resides within CA1. From
these data, it appears that the hippocampal subregions do not
function as an ensemble to separate patterns of information. Furthermore, the results indicate that it is possible to behaviorally
dissociate the functions of the DG and CA1. Based on these results, it is suggested that sensory information may be processed by
hippocampal neurons by providing sensory information with a
spatial and/or temporal marker. This would ensure that new highly
processed sensory information is organized within the hippocampus and would enhance the possibility of remembering and temporarily storing one place as separate from another place in space
and one event as separate from another event in time.
Acknowledgments
The authors thank Heather Luker, Cinnamon Wuthrich, Kerstin Forsythe, Chris Jensen, and Evan Riddle for their assistance
with data collection, and Robert Schaffer for his capable histological work.
REFERENCES
Amaral DG, Witter MP. 1995. Hippocampal formation. In: Paxinos G,
editor. The rat nervous system, 2nd ed. San Diego: Academic Press. p
443– 493.
Banks WP. 1987. Encoding and processing of symbolic information in
comparative judgments. In: Bower GH, editor. The psychology of
learning and motivation: advances in theory and research. New York:
Academic Press. p 101–159.
Chiba AA, Kesner RP, Reynolds AM. 1994. Memory for spatial location
as a function of temporal lag in rats: role of hippocampus and medial
prefrontal cortex. Behav Neural Biol 61:123–131.
Chiba AA, Kesner RP, Gibson CJ. 1997. Memory for temporal order of
new and familiar spatial location sequences: role of the medial prefrontal cortex. Learn Mem 4:311–317.
Conrad CD, Roy EJ. 1993. Selective loss of hippocampal granule cells
following adrenalectomy: implications for spatial memory. J Neurosci
13:2582–2590.
Davis HP, Colombo PJ, Volpe BT. 1988. Preoperative training effects on
radial maze performance in animals with ischemia or ibotenic acid
hippocampal injury. Soc Neurosci Abstr 14:1228.
Emerich DF, Walsh TJ. 1989. Selective working memory impairments
following intradentate injection of colchicine: attenuation of the be-
636
GILBERT ET AL.
havioral but not the neuropathological effects by gangliosides GM1
and AGF2. Physiol Behav 45:93–101.
Estes WK. 1986. Memory for temporal information. In: Michon JA,
Jackson JL, editors. Time, mind, and behavior. New York: SpringerVerlag. p 151–168.
Gaykema RP, van der Kuil J, Hersh LB, Luiten PG. 1991. Pattern of
direct projections from the hippocampus to the medial septum-diagonal band complex: anterograde tracing with Phaseolus vulgaris leucoagglutinin combined with immunohistochemistry of choline acetyltransferase. Neuroscience 43:349 –360.
Gilbert PE, Kesner RP, DeCoteau WE. 1998. The role of the hippocampus in mediating spatial pattern separation. J Neurosci 18:804 – 810.
Goldschmidt RB, Steward O. 1980. Preferential neurotoxicity of colchicine for granule cells of the dentate gyrus of the adult rat. Proc Natl
Acad Sci USA 77:3047–3051.
Jakab RL, Leranth C. 1995. Septum. In: Paxinos G, editor. The rat nervous system, 2nd ed. San Diego: Academic Press. p 405— 442.
Jarrard LE. 1978. Selective hippocampal lesions: differential effects on
performance by rats on a spatial task with preoperative versus postoperative training. J Comp Physiol Psychol 92:1119 –1127.
Jarrard LE, Meldrum BS. 1993. Selective excitotoxic pathology in the rat
hippocampus. Neuropathol Appl Neurobiol 19:381–389.
Jarrard LE, Okaichi H, Steward O, Goldschmidt RB. 1984. On the role of
hippocampal connections in the performance of place and cue tasks:
comparisons with damage to hippocampus. Behav Neurosci 98:946 –
954.
Jay TM, Witter MP. 1991. Distribution of hippocampal CA1 and subicular efferents in the prefrontal cortex of the rat studied by means of
anterograde transport of Phaseolus vulgaris-leucoagglutinin. J Comp
Neurol 313:574 –586.
Kesner RP, Bolland BL, Dakis M. 1993. Memory for spatial locations,
motor responses, and objects: triple dissociation among the hippocampus, caudate nucleus, and extrastriate visual cortex. Exp Brain Res
93:462– 470.
Lee I, Kim J, Lee C. 1999. Anatomical characteristics and three-dimensional model of the dog dorsal lateral geniculate body. Anat Re 256:
29 –39.
Madsen J, Kesner RP. 1989. Temporal order information in normal subjects and patients with dementia of the Alzheimer’s type. Soc Neurosci
Abstr 15:728.
Marr D. 1971. Simple memory: a theory for archicortex. Philos Trans R
Soc Lond [Biol] 262:23– 81.
McNaughton BL. 1989. Neural mechanisms for spatial computation and
information storage. In: Nadel L ,Cooper LA, Culicover P, Harnish
RM, editors. Neural connection, mental computation. Cambridge,
MA: MIT Press. p 285–350.
Mizumori SJY, Garcia PA, Raja MA, Volpe BT. 1994. Spatial- and locomotion-related neural representation in the rat hippocampus following long-term survival from ischemia. Behav Neurosci
109:1081–1094.
Moser M, Moser EI. 1998 Functional differentiation in the hippocampus.
Hippocampus 8:608 – 619.
Moser M, Moser EI, Forrest E, Andersen P, Morris RGM. 1995. Spatial
learning with a minislab in the dorsal hippocampus. Proc Natl Acad
Sci USA 92:9697–9701.
Mundy WR, Tilson HA. 1990. Neurotoxic effects of colchicine. Neurotoxicology 11:539 –548.
Narny KP, Mundy WR, Tilson HA. 1989. Colchicine-induced alterations of reference memory in rats: role of spatial vs. non-spatial task
components. Behav Brain Res 35:45–53.
O’Reilly RC, McClelland JL. 1994. Hippocampal conjunctive encoding,
storage, and recall: avoiding a trade-off. Hippocampus 4:661– 682.
Risold PY, Swanson LW. 1997. Connections of the rat lateral septal complex. Brain Res Rev 24:115–195.
Rolls ET. 1996. A theory of hippocampal function in memory. Hippocampus 6:601– 620.
Rolls ET, Treves A. 1998. Neural networks and brain function. Oxford:
Oxford University Press.
Shapiro ML, Olton DS. 1994. Hippocampal function and interference.
In: Schacter DL, Tulving E, editors. Memory systems, 1994. London:
MIT Press. p 141–146.
Tanila H. 1999. Hippocampal place cells can develop distinct representations of two visually identical environments. Hippocampus 8:235–
246.
Verwer RWH, Meijer RJ, Van Uum HFM, Witter MP. 1997. Collateral
projections from the rat hippocampal formation to the lateral and
medial prefrontal cortex. Hippocampus 7:397– 402.
Walsh TJ, Schultz DW, Tilson HA, Schmechel DE. 1986. Colchicineinduced granule cell loss in rat hippocampus: selective behavioral and
histological alterations. Brain Res 398:23–36.
Xavier GF, Oliveira-Filho FJB, Santos AMG. 1999. Dentate gyrus-selective colchicine lesion and disruption of performance in spatial tasks:
difficulties in “place strategy” because of a lack of flexibility in the use
of environmental cues? Hippocampus 9:668 – 681.