Longitudinal axis of the hippocampus: Both septal and temporal

advertisement
HIPPOCAMPUS 13:587– 603 (2003)
Longitudinal Axis of the Hippocampus: Both Septal
and Temporal Poles of the Hippocampus Support
Water Maze Spatial Learning Depending on the
Training Protocol
Livia de Hoz,* Jane Knox, and Richard G.M. Morris
Division of Neuroscience, University of Edinburgh,
Edinburgh, Scotland, United Kingdom
ABSTRACT:
It has been suggested previously that 30% sparing of the
hippocampus is enough to support spatial learning of a reference memory
task in a water maze provided the spared tissue is located septally (Moser et
al. 1995, Proc Natl Acad Sci USA 92:9697–9701). Therefore, the temporal
hippocampus may not be involved in spatial memory. Place cells are also
found in this part of the structure, and it has been suggested that these place
cells have larger, less well-tuned place fields than are found in the septal
hippocampus. We tested the possibility that the temporal hippocampus might
be involved in spatial learning when the animals are required to distinguish
between different contexts. Experiment 1 was a replication of the findings
reported by Moser et al., using their protocol (8 trials/day, 6 days) and the
groups with 20 – 40% hippocampus spared septally or temporally (volume
assessed by quantitative volumetric techniques). In experiment 2, rats with
also 20 – 40% sparing of the hippocampus either septally or temporally were
trained in two water maze concurrently (four trials/day/water maze, 8 days).
Rats with 20 – 40% hippocampus spared temporally were able to learn the
two water maze tasks normally, and no difference was observed between rats
with septal and temporal hippocampus spared across different measures of
performance. In experiment 3, rats with 20 – 40% hippocampus spared septally or temporally were trained in one water maze as in experiment 1, but
using a spaced training protocol similar to that of experiment 2 (four trials/
day, 8 days). Rats with temporal hippocampus spared developed a preference
for the training quadrant and acquired levels of performance indistinguishable from those of rats with septal hippocampus spared. The results suggest
that the temporal hippocampus can support the learning of two, but also one,
spatial water maze reference memory task, provided the training protocol is
adequate. Hippocampus 2003;13:587– 603. © 2003 Wiley-Liss, Inc.
KEY WORDS:
memory; septotemporal; dorsal; ventral; lesion
INTRODUCTION
Anatomical studies dating back to Lorente de No (1934) have suggested
that the hippocampus is not completely homogeneous along its longitudinal
Grant sponsor: European Science Foundation; Grant sponsor: MRC Programme; Grant number: G9200370/2.
*Correspondence to: Livia de Hoz, Division of Neuroscience, 1, George
Square, University of Edinburgh, Edinburgh EH8 9JZ, Scotland, UK.
E-mail: livia.dehoz@ed.ac.uk
Accepted for publication 7 May 2002
DOI 10.1002/hipo.10079
©
2003 WILEY-LISS, INC.
(septotemporal) axis (for review of recent findings, see
Amaral and Witter, 1995). Early electrophysiological
(Elul, 1964; Brazier, 1970) and behavioral (Nadel, 1968;
Stevens and Cowey, 1973; Koreli, 1977) studies found
correlates of this anatomical heterogeneity but no clear
functional differentiation along the longitudinal axis.
Moser et al. (1993, 1995) reported that lesions made to
the septal part of the hippocampus (sparing temporal
regions) cause a deficit in a water maze spatial learning
task, whereas those made in the temporal part (sparing
septal regions) left acquisition and probe test performance unaffected. This observation, together with those
reported by other investigators (Moser and Moser,
1998a; Hock and Bunsey, 1998; Bannerman et al., 1999;
Richmond et al., 1999; Passino and Ammassari-Teule,
1999; Alescio-Lautier et al., 2000; Ferbinteanu and MacDonald, 2000, 2001), has reawakened interest in the possibility of functional distinctions along the longitudinal
axis of the hippocampal formation.
In discussing different regions of the hippocampus, we
use the terms “septal” and “temporal” to refer to the
rostralmost and the ventralmost poles of the longitudinal
axis, respectively, because this terminology allows an even
division of this axis into septal and temporal halves. The
terms “dorsal” and “ventral” are sometimes used to refer
to the same areas; the dorsal hippocampus is, however,
more extensive than the ventral. In referring to the work
of others, we use their terminology to describe these studies accurately.
The present work was begun with a view to distinguishing two possible interpretations of the findings reported by Moser and colleagues; it has since broadened to
encompass more general concerns. A functional differentiation hypothesis (Moser and Moser, 1998b) asserts
that, rather than carrying out a single cognitive function,
the hippocampus might support different functions at
different points along its longitudinal axis. The straightforward interpretation of the data reported by Moser et
al. (1993, 1995) is that the septal (dorsal) hippocampus
(posterior in primates) is involved in spatial learning,
whereas the temporal (ventral) hippocampus (anterior in
588
HOZ ET AL.
primates) might mediate nonspatial aspects of hippocampal-dependent learning. Even small minislabs of tissue are thought to be
sufficient for learning. The anatomical support for this hypothesis
is that sensory information projects primarily to the septal hippocampus (with spatial information believed to depend on this),
while projections from the amygdala and hypothalamus—areas
thought of as providing nonspatial information—terminate in the
temporal hippocampus (Amaral and Witter, 1995). Functional
differentiation along the longitudinal axis has also been highlighted in the human literature. For example, the HYPER model of
Lepage et al. (1998) proposes that the rostrocaudal axis maps onto
a functional differentiation between encoding-related and retrieval-related activation in human functional imaging.
An alternative unitary processing framework asserts that the
hippocampus has a single function, or single set of linked functions
(whatever these are), and that task dissociations arising from lesions placed at different points along the longitudinal axis may
reflect subtle effects on memory processing, such as learning rate,
consolidation, representational acuity, or other parameters. It is
important to bear in mind that, despite the differences summarized
by Amaral and Witter (1995), the basic architecture of hippocampal intrinsic circuitry is broadly equivalent at all points along its
longitudinal axis, and it is therefore likely to be carrying out a
common processing algorithm. Moreover, although differences in
the outcome of this common processing in the septal and temporal
hippocampus may occur by virtue of the segregation of the incoming information along this same axis, associational projections
crossing over the septotemporal boundaries are abundant in the
hippocampus, and segregation might not generally occur. According to this, one would expect the temporal hippocampus to be
involved in spatial learning under certain conditions. Consistent
with this hypothesis, place cells are found in both the dorsal and
temporal hippocampus (Jung et al., 1994; Poucet et al., 1994). We
were initially struck by the observation made by Jung et al. (1994)
that place fields in the temporal hippocampus of rats are generally
larger than those in the more dorsally located septal hippocampus.
Although not all studies have found this result (e.g., Poucet et al.,
1994), this finding raises the possibility that a hippocampal mapping system of the kind envisaged by O’Keefe and Nadel (1978)
might use place fields of varying size to work effectively. Large place
fields might, for example, differentiate one context from another,
while smaller fields enable place recognition within a context.
We therefore decided to explore whether the septal hippocampus is sufficient to support water maze spatial learning when the
task involves learning two concurrent reference memory tasks in
two different environments. To this aim, we extended the training
protocol of Moser et al. (1995) to include the concurrent learning
of two water maze tasks and variation in both the number of trials
per day and number of days of training. If the temporal hippocampus in rodents plays no role in spatial learning, varying the training
protocol should have little or no effect. Conversely, if the temporal
hippocampus contributes to spatial learning under certain conditions, but not others, varying the training protocol may reveal the
basis of the differential contribution to spatial learning of cells
along the longitudinal axis. This argument is analogous to that of
Rawlins (1985), who suggested that lesion dissociations related to
systematic manipulation of quantifiable parameters are likely to be
particularly revealing with respect to function.
Three experiments are described, followed by additional data
and a meta-analysis of the outcome in relation to the parameters
varied (number of water mazes, trials, and days of training). Partial
lesions were made sparing the septal or temporal hippocampus. In
addition to a conventional volumetric analysis of lesion size, fluorescent retrograde tracer injections were used to ensure the normal
extrinsic connectivity of the spared pole of hippocampal tissue. We
begin by describing an exact replication of selected groups of the
study reported by Moser et al. (1995).
MATERIALS AND METHODS
Subjects
We used a total of 172 male Lister hooded rats obtained from a
commercial supplier (Charles River, UK) across the series of studies (experiments 1–3). The rats were housed singly in plastic cages
with sawdust bedding and ad libitum food and water. Their care
and maintenance and the experimental procedures complied with
the UK Home Office Regulations conducted under Project License PPL 60/1625.
Apparatus
Training was conducted in a water maze (Morris, 1984), consisting of a circular pool (2.0-m diameter and 60-cm height) of
water (25° ⫾ 1°C; made opaque with latex liquid) located in a
well-lit room with numerous visual cues. The escape platform was
10 cm in diameter with its top surface 1 cm below the water level so
as to be hidden from view at the water surface. The animals’ swimming was monitored by an overhead video camera connected to a
video recorder and an on-line data acquisition system (Watermaze™ see Spooner et al., 1994). This digitizes the path taken
by the animal and computes various spatial parameters (e.g., latency, quadrant-time, time in platform zones). The data acquisition system was located in an adjacent room.
Surgery
Selected animals were given neurotoxic lesions aiming to spare
the septal or temporal 20 – 40% of the hippocampus (DG and CA
fields), while removing the remaining tissue without damage to
structures outside the hippocampal formation. Complete hippocampal lesions, ⬎90%, were also made. Lesions were made with
ibotenic acid (Sigma, UK, or Biotechnology, USA; dissolved in
phosphate-buffered saline [PBS], pH 7.4, at 10 mg/ml) following
the protocol of Jarrard (1989). Rats were anesthetized with tribromoethanol (Avertin, 10 ml/kg) i.p. and placed in a Kopf stereotactic frame. Several injections of ibotenic acid were made at different
rostrocaudal and dorsoventral levels via a 1-␮l SGE syringe securely attached to the frame by a Kopf stereotactic arm. Ibotenic
acid (0.05 ␮l, 0.08 ␮l or 0.1 ␮l per injection) was injected at a rate
of 0.1 ␮l/min, beginning 30 s after the needle was lowered. The
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
589
TABLE 1
Coordinates for Up to 26 Ibotenic Acid Injections Per Rat for Different Types of Lesions
Complete lesion
Septal 70% lesion
Temporal 70% lesion
L
V
␮la
AP
L
V
␮la
AP
L
V
␮la
⫺2.4
⫺3.0
⫾ 1.0
⫾ 3.0
⫾ 1.4
⫺3.0
⫺2.7
⫺2.1
⫺2.9
(0.05)
(0.10)
(0.05)
(0.05)
⫺2.4
⫺3.0
⫾ 1.0
⫾ 3.0
⫾ 1.4
⫺3.0
⫺2.7
⫺2.1
⫺2.9
(0.05)
(0.10)
(0.05)
(0.05)
⫺4.3
⫺4.8
⫾ 4.0
⫾ 3.9
⫾ 5.4
⫺7.0
⫺7.0
⫺5.1
(0.10)
(0.10)
(0.10)
⫺5.0
⫾ 3.0
⫺3.0
(0.05)
⫺4.0
⫾ 3.7
⫾ 2.6
⫺2.7
⫺1.8
⫺2.8
(0.10)
(0.05)
(0.05)
⫺4.0
⫾ 3.7
⫾ 2.6
⫺2.7
⫺1.8
⫺2.8
(0.10)
(0.05)
(0.05)
⫺5.7
⫾ 3.9
⫺4.3
⫾ 4.0
⫺7.0
(0.05)
⫺5.0
⫾ 5.0
⫾ 3.9
⫺4.0
⫺3.1
(0.05)
(0.05)
⫺3.6
⫺7.0
⫺3.6
⫺4.9
(0.05)
(0.05)
(0.05)
(0.05)
⫺4.6
(0.05)
⫺4.9
⫾ 3.9
⫾ 3.9
⫺3.6
(0.05)
⫾ 5.1
(0.05)
(0.10)
(0.08)
(0.08)
(0.10)
⫺5.5
⫺5.9
⫺3.5
⫺7.0
⫺4.5
⫺5.3
⫺3.9
AP
⫾ 5.0
⫺6.15
⫾ 4.3
⫾ 4.6
AP, anteroposterior, L, lateral; V, ventral.
a
␮l of ibotenic acid injected.
needle was removed very slowly 90 s after the injection. A total of
0.6 ␮l or 0.91 ␮l per hemisphere was necessary for the partial and
complete lesions, respectively. The coordinates were modified
from Jarrard (1989) to suit the slightly different brain size of Lister
hooded rats. Sham lesions were done in the same way, but the
injections were substituted by piercing of the dura with a needle,
with the intention of creating comparable neocortical damage.
Table 1 lists coordinates for up to 26 ibotenic acid injections per rat
for the different lesion types.
Behavioral Testing
In each of three experiments, the animals were trained in one or
two concurrent reference memory (RM) task(s) beginning at least
10 days after the lesions. The experiments were conducted in replications consisting of a small number of animals in each lesion
group: septal hippocampus spared (septal), temporal hippocampus
spared (temporal), no hippocampus spared (no-hpc), and sham
lesion (sham). The training was conducted blind with respect to
the group to which the animals were assigned.
The protocol for spatial learning was as follows. The rats were
allowed 120 s to find the escape platform located at the center of
the northeast (NE, one-half of the animals) or southwest (SW)
quadrant for the upstairs water maze, and southeast (SE, half the
animals) and northwest (NW) for the downstairs water maze. In
two concurrent water maze experiments (experiment 2), rats were
trained to one of the four possible combinations of upstairs-downstairs platform positions (SW-NW, NE-SE, SW-SE, or NE-NW).
The layout of the rooms is shown in Figure 1. The animals were
then left on the platform for 30 s. For each rat, the platform
position was constant throughout acquisition, but start positions
(at N, W, S, or E) were varied pseudo-randomly across trials. The
number of trials per day varied with the experiment (see below). A
transfer, or probe, in which the rat swam for 60 s in the absence of
the platform, was conducted at the start of day 5 and on the day
after the end of training. There were several measures of performance: escape latency, path length, swim speed, time spent swimming in each of the four quadrants during the probe test, and time
spent in a 20-cm-radius zone centered around the platform position. These data were analyzed using analysis of variance
(ANOVA), followed by the appropriate further statistical tests.
After the last probe test, all the rats in experiments 2 and 3 were
given four trials with a visible platform. This section of the protocol was used to ensure that any behavioral deficit during probe
testing was not due to a visual impairment that incapacitated the
rat from using the spatial cues located around the pool. The curtains were drawn around the pool to prevent the use of spatial cues;
a modified platform, with vertical black and white stripes and 2 cm
visible above the water was used. The platform position was random as well as the starting position and the time to find the platform was measured.
Retrograde Tracer Injection
Some of the partial lesion and sham rats were given unilateral
retrograde tracer injections after the training. The injection proce-
590
HOZ ET AL.
FIGURE 1.
Diagrammatic representations of upstairs and downstairs water maze room. The
pool is represented by a circle. The position of the two possible platform positions is indicated by
dashed lines inside the pool.
dure is identical to the ibotenic acid lesion. Fast blue (1 ␮l) was
injected in the septal hippocampus (AP:⫺4.3, ML:⫺2.4, DV:
⫺3.6) and diamidino yellow (2 ␮l) in the temporal hippocampus
(AP:⫺5.2, ML:⫺4.8, DV:⫺6.7) or vice versa. Rats were allowed 7
days recovery before perfusion.
Perfusion, Histology, and Analysis of Lesions
Immediately after the end of training, or 7 days after the retrograde tracer injections, the rats were perfused intracardially with
4% formalin after terminal anesthesia with euthetal (1 ml). Their
brains were removed and stored in formalin for 24 h before being
blocked and embedded in egg yolk. The embedding procedure was
as follows: the brain was placed in a cube filled with fresh egg yolk
and, at 5°C, it was incubated in a tray of formalin for 24 h, and
then left in formalin for 24 h. The cube was removed and the block
of hardened egg yolk containing the brain left in formalin for a
further 24 h. Coronal 30-␮m sections through the hippocampus
and other structures (in the case of retrograde tracer injected
brains) were obtained in a cryostat. One in 5 sections was recovered
(or 2 in 5 sections in the case of the retrograde tracer injected
brains). These were mounted on slides, stained with cresyl violet
and coverslipped in DPX. The parallel series cut in the retrograde
tracer injected brains was not stained.
damaged. The cutting point between ventral CA1 and ventral
subiculum was chosen at the point where the cell layer becomes
wider, a feature characteristic of the subicular molecular layer
(Amaral and Witter, 1995). The software then computed the hippocampal area within each section. The sections were spaced 150
␮M apart, yielding up to 38 sections in a sham lesion animal and
less in animals with acceptable partial lesions. The sum of all the
areas containing hippocampus for an individual animal was then
compared with the total mean hippocampal area of the sham animals. The left and right sides of the brain were initially computed
separately and then averaged. The resulting percentage sparing
value is the proportion of hippocampus spared in a lesioned rat.
Strict criteria for acceptance of a lesion were used. The lesion
had to (1) be confined to the hippocampus, (2) leave intact tissue
volumes of 20 – 40% in septal or temporal hippocampus, and (3)
cause minimal sparing (⬍10%) elsewhere in the hippocampus.
The complete lesions were accepted when ⬍10% of the hippocampus was spared. Medial and ventral subicular damage is often observed after ibotenic acid lesions of the hippocampus. A small
number of lesions with minimal subicular damage limited to the
most medial and temporal levels were accepted. No correlation was
found between the amount of subicular damage in the accepted
lesions and behavioral performance.
Measuring Spared Tissue
The relative volume of spared tissue was calculated by measuring
the area of hippocampus (hippocampus proper plus dentate gyrus)
spared in each section of a particular brain according to the following protocol. Each coronal section containing hippocampus was
placed under a Leica photomacroscope, with the image taken by a
videocamera into the computer and opened into NIH Image 1.63.
The area of spared tissue was outlined excluding surrounding fibers
such as the fimbria; the reason to exempt the fimbria is that it
would not be considered in a section were all the hippocampal cells
RESULTS
Lesion Assessment
Of the 172 rats that were given lesions, 26 were excluded after
the histological analysis for failing to reach our strict criteria. A
further 7 animals died after surgery, leaving a total of 139 animals
on which the behavioral analysis is based. Figure 2a is a compilation of representative examples of the different types of lesions.
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
Retrograde tracer injections were used to assess the status of the
extrinsic connectivity of the residual minislabs of hippocampus in
partially lesioned animals. The preservation of extrinsic connectivity in an animal with a septal lesion indicates that the temporal
hippocampal minislab has not been deafferented. It is not the aim
of this study to use retrograde tracers to describe the detailed topography of hippocampal projections, but merely to examine
whether spared sections maintain normal inputs. For this reason,
description of each individual case has been omitted. Injection sites
were consistent between rats and, for Figures 2b and 2c, the most
representative cases have been chosen. We looked for labeling in
the structures that project to the septal and/or temporal hippocampus (i.e., entorhinal cortex, medial septum and diagonal band of
Broca, locus coeruleus [LC], and raphe) and its topographical pattern.
Our observations were consistent across brains and matched
what is known about the topography of the inputs into different
parts of the hippocampus. Figure 2b shows photomicrographs of
medial septum and LC. As expected, spared temporal hippocampus maintains the projection from the lateral part of the medial
septum. The pattern is the opposite in animals with septal hippocampus spared where labeling was now seen in the medial part of
the medial septum (Meibach and Siegel, 1977; Amaral and Kurtz,
1985). The LC is labeled in both cases, which is expected from the
diffuse projection from LC to all levels of hippocampus (Montone
et al., 1988). In the case of the entorhinal cortex (EC), flat maps
were plotted to show the pattern of labeling. In the three cases
shown in Figure 2c, the labeling is in accordance with the topography of the entorhinal projection to the dentate gyrus (DG) described by Dolorfo and Amaral (1998a), i.e., that the lateral and
caudal part of the EC projects to the septal half of the DG, while
the more medially and rostrally located cells in EC project to the
temporal DG.
Finally, in comparing the performance of rats with partial sparing of the hippocampus in different regions it is important to
ensure equivalence of the residual minislabs of spared tissue, and
lesion extent in rats with complete lesions. Figure 2d plots the
amount of hippocampus spared in the different groups used in
these series of experiments.
Experiment 1
Aim, group designation, and protocol. We began by establishing that the results obtained by Moser et al. (1995) are replicable.
The protocol used was identical and the apparatus the same as that
used by them in their 1995 study conducted in Edinburgh. Only a
subset of the groups used in the original study were trained, specifically those with 20 – 40% septal hippocampus spared (septal,
average 34 ⫾ 1, n ⫽ 14), 20 – 40% temporal hippocampus spared
(temporal, average 30 ⫾ 2, n ⫽ 10) and sham lesioned rats (sham,
n ⫽ 19).
The rats were given two sessions per day (morning and afternoon) of four trials each for 6 days (48 trials; Fig. 3). Probe tests
were carried out at the beginning of day 5 (before the normal
training) and on day 7. Some animals were then trained for a
591
further 2 days and received a probe test on day 9 (but these data are
not presented until after experiment 3 below).
Results. All rats swam normally with the usual adult swimming
posture and had no difficulty climbing onto the platform. Overall
analysis of escape latency across training (Fig. 4a) showed a clear
difference between groups (F[2, 40] ⫽ 20.9, P ⬍ 0.001], a decline
in latency across sessions (F[11, 440] ⫽ 59.2, P ⬍ 0.001), but no
group ⫻ session interaction (F ⬍ 1). Post hoc multiple comparisons (Ryan-Einot-Gabriel-Welsch) showed that the septal group
(i.e., rats with septal hippocampus spared) was not significantly
different from the sham group (septal X៮ : 79.6 s on the first session
to 10.5 s on the last session; sham: 69.6 – 6.3 s), whereas the temporal group (temporal spared rats) was significantly different from
both (X៮ : 87.5–35.2 s; P ⬍ 0.05).
By probe test 1 (Fig. 4b), the sham group was above chance in
searching in the correct quadrant, while the septal group was beginning to show a bias. The temporal group remained at chance.
Overall analysis showed an effect of quadrant (F[1.8, 70.8] ⫽ 50.9,
P ⬍ 0.001; degrees of freedom corrected for sphericity according
to Greenhouse-Geisser, SPSS) and a group by quadrant interaction
(F[3.5, 70.8] ⫽ 11.9, P ⬍ 0.001). Analysis percentage time in the
training quadrant only showed a striking difference between
groups (F[2, 40] ⫽ 17.8, P ⬍ 0.001). Post-hoc multiple comparisons (Ryan-Einot-Gabriel-Welsch) indicate that partial lesion
groups (X ⫽ 37.8% for the septal group and 30 for the temporal
group) did not differ from each other, but both differed from the
sham group (P ⬍ 0.001, X ⫽ 55.6%).
In probe test 2 (Fig. 4c), the septal group had clearly differentiated from the temporal group and was searching in a manner
similar to that of the shams. Overall analysis gave an effect of
quadrant (F[2.2, 87.5] ⫽ 42.7, P ⬍ 0.001) and a group by quadrant interaction (F[4.4, 87.5] ⫽ 5.2, P ⬍ 0.005). Analysis of the
training quadrant only showed a group effect (F[2, 40]⫽ 8.1, P ⬍
0.005). Post-hoc comparisons (Ryan-Einot-Gabriel-Welsch)
showed that the temporal group (X៮ ⫽ 30.3%) was significantly
poorer than both the septal group (46.5%; P ⬍ 0.01) and the sham
group (50.7%; P ⬍ 0.01). Swim paths taken by representative
animals during this test are shown in Figure 4d.
These data represent a successful replication of the results found
by Moser et al. (1993 and 1995). Figure 4e presents percentage
time in training quadrant in the same manner as in Figure 3c of
Moser et al. (1995) and the pattern is identical.
Experiment 2
Aim, group designation, and protocol. We then explored a
different reference memory protocol with a view to testing the
different interpretations of this septal vs temporal dissociation as
highlighted in the introduction. This protocol used exactly the
same number of trials/day as in experiment 1 above, but these were
scheduled in two different water mazes trained concurrently. The
groups trained were those with septal 20 – 40% hippocampus
spared (septal, average of 35 ⫾ 1%, n ⫽ 7); with temporal 20 –
40% hippocampus spared (temporal, average of 31 ⫾ 1% spared,
n ⫽ 12); rats in which the entire hippocampus had been lesioned
(no-hpc, n ⫽ 11) and sham-lesioned rats (sham, n ⫽ 14).
592
HOZ ET AL.
FIGURE 2.
(Continued)
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
593
FIGURE 2.
a: Photographs of evenly spaced cresyl violet-stained
coronal sections across the hippocampus of representative septal 34%
spared, sham and temporal 40% hippocampus spared rats. The red
outline encloses the area considered when measuring spared tissue. b:
Photomicrographs of fast blue and diamidino yellow labeling in the
medial septum (LMS, lateral medial septum; MMS, medial medial
septum) and locus coeruleus of representative septal spared, sham and
temporal hippocampus spared rats. The inputs from these areas to the
hippocampus are maintained and display a normal topographical
pattern. See text. Diagrams based on Paxinos and Watson (1998). c:
Flat maps of the entorhinal cortex of representative septal spared,
sham and temporal hippocampus spared rats. Retrograde tracer labeling is plotted such that light and dark gray reflect tracer injections in
the temporal and septal hippocampus, respectively. EC input to the
hippocampus and its topography is maintained after a hippocampal
partial lesion. See text. d: Percentage hippocampus spared across the
different groups in experiments 1–3.
The rats had two sessions per day (morning and afternoon) of
four trials each, with one session in water maze A and one session in
water maze B (the water maze constituting the morning session was
semirandomized; Fig. 3). The two water mazes were located on
different floors in the laboratory, each in a room with distinctive
cues. Rats were trained in this way for an extended period of 8 days
(64 trials) as pilot studies had shown that after 6 days, as in experiment 1, animals with partial lesions had learned little about the
platform location. Transfer tests were carried out at the beginning
of days 5 and 9.
Results. As in experiment 1, all rats swam normally and there were
no obvious qualitative differences in swimming style, use of the platform as refuge etc. In preparing the data analysis, we observed no
significant difference in acquisition and transfer test performance in
the two water mazes; thus, the data from these have been averaged.
594
HOZ ET AL.
FIGURE 3.
Diagrammatic explanation of the training protocol used in experiment 1 (replication of Moser et al. 1995), experiment 2 (2 water maze), and experiment 3 (1 water maze).
Overall analysis of escape latency (Fig. 5a) during training showed a
strong group effect (F[3, 40] ⫽ 11.6, P ⬍ 0.001), a decline in latency
across Sessions (F[7, 280] ⫽ 96.3, P ⬍ 0.001), but no group ⫻ day
interaction (F ⬍ 1). Multiple comparisons (Ryan-Einot-GabrielWelsch) showed that the sham group was significantly different from
all groups (P ⬍ 0.05), that groups septal and temporal group were not
significantly different from each other and that no-hpc were different
from the sham group and the temporal group (P ⬍ 0.05). Mean
escape latency (average of four trials): sham ⫽ 59.7 on first session to
7.1 s on last session; no-hpc ⫽ 85.5–28.1 s; septal ⫽ 76.1–12.2 s; and
temporal ⫽ 73.5–11.7 s.
Probe test 1 (day 5) showed little evidence of learning in any
group except those with sham-lesions (Fig. 5b). However, by probe
test 2 (day 9), groups septal and temporal showed a striking bias
toward searching in the training quadrant that did not differ significantly from that shown by the sham group (Fig. 5c). Overall
analysis of percentage time in each of the four quadrants (Fig. 5c)
shows an effect of quadrant (F[1.8, 73.2] ⫽ 65.1, P ⬍ 0.001) and
a group ⫻ quadrant interaction (F[5.5, 73.2] ⫽ 4.6, P ⫽ 0.001).
Analysis of percentage time spent in the training quadrant only
showed a strong group effect (F[3, 40] ⫽ 6.7, P ⫽ 0.001). Orthogonal analysis showed that the septal group (X៮ ⫽ 44.9%) did not
differ from the temporal group (X៮ ⫽ 41.2%; F ⬍ 1); that these
groups did not differ from the sham group (X៮ ⫽ 47.3%; F[3, 40]
⫽ 1.66), and that partial-lesioned and sham-lesioned groups were
better than the no-hpc group (X៮ ⫽ 29.4%; F[3, 40] ⫽ 16.2, P ⬍
0.001). Do the septal group and temporal differ in any other measure of performance, such as accuracy of swim? Analysis of percentage time in a zone (20-cm radius) centered around the platform
also showed no difference in search accuracy between rats with
septal and temporal hippocampus spared (Fig. 5d). Overall analysis showed an effect of quadrant (F[1.8, 71] ⫽ 56.8, P ⬍ 0.001)
and a group ⫻ quadrant interaction (F[5.3, 71] ⫽ 9.4, P ⬍ 0.001).
Analysis of percentage time spent in the training zone only showed
a group effect (F[3, 40] ⫽ 6.7, P ⬍ 0.001). Multiple post-hoc
comparisons ((Ryan-Einot-Gabriel-Welsch) showed no difference
between the septal group (X៮ ⫽ 14.8%) and the temporal group
(X៮ ⫽ 13.6%). No differences between these two groups were observed in swimming speed, latency of first crossing over platform
position or number of crossings during the test (data not shown).
In the visible platform task, all the animals swam directly toward
the platform and there was no difference between groups (Fig. 5a).
This indicates that none of the groups had a gross visual impairment, which could account for deficits found in the spatial task.
Discussion. These results are surprising and do not bear out the
prediction of either of the two hypotheses that the study was designed
to test. These predictions were (1) lesions sparing the temporal hippocampus only would impair spatial learning, whereas lesions sparing
the septal hippocampus would not, irrespective of the protocol; or (2)
a temporal lesion effect would be seen with training on two concurrent
water mazes, despite its absence in training with one water maze. Our
results show that training on an ostensibly more demanding protocol,
and one shown to be more demanding by the poorer level of performance reached at probe test 1 relative to that in experiment 1, is
actually insensitive to lesions that leave 20– 40% of either septal or
temporal hippocampus intact.
Despite having successfully replicated the relevant part of the
original study by Moser et al. (1995), we now find that not only is
a similar block of tissue sufficient to support the concurrent learning of two water mazes but, surprisingly, that it is irrelevant in
which pole of the hippocampus that tissue is spared.
Experiment 3
Aim, group designation, and protocol. What might be the reason(s) for the unexpected change in performance between experiments 1 and 2? experiment 2 differs from the Moser et al. replication
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
595
FIGURE 4.
Experiment 1: Replication of Moser et al. (1995). a:
Escape latency is averaged across sessions (four trials/session, two
sessions/day). PT1 and PT2, first and second probe tests, respectively.
First (b) and second (c) probe test results plotted as a function of
percentage time in each of the four quadrants (plotted in this order:
adjacent left to training quadrant, training quadrant, shown by black
bar, adjacent right and opposite). d: Example of second probe test
swim paths (percentage: amount of hippocampus spared). e: Second
probe test plotted as in Moser et al. (1995) figure 3.c to facilitate
comparison. percentage time in training quadrant is plotted against
amount of hippocampus spared for each group. Dashed line represents chance performance (25%). Sept, septal hippocampus spared;
temp, temporal hippocampus spared.
of experiment 1 in the use of two water mazes rather than one, but also
in two other ways. First, although the number of trials per day was the
same in both experiments (8), the animals had, of necessity, only four
trials/day in each water maze in experiment 2. Second, because performance was expected to be poorer, the rats were trained for 8 days
rather than 6. As the spacing of trials and extent of training are known
to be important parameters in many types of learning, experiment 3
was conducted using four trials/day in a single water maze and with
training continued for 8 days (32 trials). Probe tests were carried out
on days 5 and 9 (Fig. 3).
Rats with septal 20 – 40% hippocampus spared (septal, 34% ⫾
2 spared, n ⫽ 9) were compared with rats with temporal 20 – 40%
hippocampus (temporal, 30% ⫾ 2 spared, n ⫽ 13). Rats with no
hippocampus spared (no-hpc, n ⫽ 9) and sham lesioned (sham,
n ⫽ 21) groups were used as controls.
Results. All rats displayed normal swimming patterns and
climbed onto the platform in the usual way. Overall analysis of
escape latency across training (Fig. 6a) showed a strong effect of
group (F[3, 48] ⫽ 10.3, P ⬍ 0.001), a decline in latency across
596
HOZ ET AL.
FIGURE 5.
Experiment 2: Two concurrent water mazes. a: Escape latency is averaged across days (four trials/day and water maze).
PT1 and PT2, first and second probe tests, respectively. First (b) and
second (c) probe test results averaged across water mazes plotted as a
function of percentage time in each of the four quadrants (order:
adjacent left to training quadrant, training quadrant, shown by black
bar, adjacent right and opposite). d: Further analysis of probe test 2:
percentage time in zone (20-cm radius) centered around platform
position in training quadrant and compared with equivalent zones in
the other three quadrants (order: adjacent left to training quadrant,
training quadrant, shown by black bar, adjacent right and opposite).
e: Example of second probe test swim paths (percentage: amount of
hippocampus spared). Dashed line represents chance performance
(25%). Vis, visible platform task, average of four trials; Sept, septal
hippocampus spared; temp, temporal hippocampus spared; no-hpc,
no hippocampus spared.
Days (F[7, 336] ⫽ 78.6, P ⬍ 0.001) but no group ⫻ Day interaction (F ⬍ 1). Multiple comparisons (Ryan-Einot-GabrielWelsch) showed no differences in escape latency between groups
septal, temporal and no-hpc (X៮ ⫽ 91.2 s on the first day to 22.6 s
on the last day; X៮ ⫽ 92.6 –21.9 s; X៮ ⫽ 92.4 –22.4 s, respectively).
All three lesioned groups (partial and complete) were significantly
different from shams (X៮ ⫽ 73.1–7.8 s; P ⬍ 0.05).
Analysis of the probe test 1 (Fig. 6b) showed that none of the
lesioned groups was spending more time in the training quadrant
than in any of the other quadrants, while shams were starting to
show a spatial bias. Analysis of the probe test 2 (Fig. 6c) showed an
effect of quadrant (F[2.3, 111] ⫽ 34.1, P ⬍ 0.001) and a group by
quadrant interaction (F[6.9, 111] ⫽ 6.3, P ⬍ 0.001). Analysis of
percentage time in training quadrant only showed an effect of
group (F[3, 48] ⫽ 11.5, P ⬍ 0.001). Orthogonal comparisons
indicated that groups septal (X៮ ⫽ 39 ⫾ 4.1%) and temporal (X៮ ⫽
37.6 ⫾ 4.2%) did not differ from each other (F ⬍ 1). Both partiallesioned groups spent significantly less time in the training quadrant than shams (X៮ ⫽ 54.2%; F[1, 48] ⫽ 16.54, P ⬍ 0.001], while
the mean of the partial-lesioned and sham groups was higher than
that of the no-hpc group (X៮ ⫽ 27.7%; F[1, 48] ⫽ 12.09, P ⬍
0.001). In a separate comparison, both septal and temporal spared
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
597
FIGURE 6.
Experiment 3: One water maze (8 days). a: Escape
latency is averaged across days (four trials/day). PT1 and PT2: first
and second probe tests, respectively. First (b) and second (c) probe test
results plotted as a function of percentage time in each of the four
quadrants (order: adjacent left to training quadrant, training quadrant, shown by black bar, adjacent right and opposite). d: Further
analysis of probe test 2: percentage time in zone (20-cm radius) centered around platform position in training quadrant and compared
with equivalent zones in the other three quadrants (order: adjacent
left to training quadrant, training quadrant, shown by black bar,
adjacent right and opposite). e: Example of second probe test swim
paths (percentage: amount of hippocampus spared). Dashed line represents chance performance (25%). Vis, visible platform task, average
of four trials; Sept, septal hippocampus spared; temp, temporal hippocampus spared; no-hpc, no hippocampus spared.
rats were found to have probe test times that were significantly
above chance (t-test ⫽ 3.4 and 3.00, respectively, P ⬍ 0.05). As in
experiment 2, analysis of percentage time in an area (20-cm radius)
centered around the platform during probe test 2 (Fig. 6d) showed
an effect of quadrant (F[1.8, 89] ⫽ 33.7, P ⬍ 0.001) and a
group ⫻ quadrant interaction (F[5.5, 89] ⫽ 7.6, P ⬍ 0.001).
Analysis of percentage time spent in the training zone only showed
a group effect (F[3, 48] ⫽ 10.3, P ⬍ 0.001). Multiple post-hoc
comparisons (Ryan-Einot-Gabriel-Welsch) showed no difference
between the septal group (X៮ ⫽ 10%) and the temporal group (X៮ ⫽
10.5%). No differences between these two groups were observed in
swimming speed, latency of first crossing over platform position or
number of crossings during the test (data not shown). As in experiment 2, no deficits were observed in the subsequent four trials of
the visible platform task (Fig. 6a).
Discussion. The unexpected difference found between experiment 1 (replication of Moser et al., 1995) and experiment 2 (two
concurrent water mazes) is clarified by experiment 3. A possible
reason why rats with temporal hippocampus spared could learn the
task in experiment 2 is due to the spacing and extent of training in
one water maze, rather than the concurrent training in two water
mazes. For example, one protocol (four trials/day for 8 days; total
number of trials 32; experiment 3) gives rise to paradoxically better
598
HOZ ET AL.
FIGURE 7.
Transfer test (day 9) after additional training the
temporal hippocampus spared group in experiment 1.
probe test performance by the temporal group than the other (8
trials/day protocol for 6 days; total number of trials 48; experiment 1).
resulting graphs. However, the amount of hippocampus spared in
each of the partial lesioned groups is equivalent across experiments.
Rats with no-hippocampus spared were at chance in all experiments and therefore were not considered in this analysis. When
plotted as a function of the number of trials of training, the performance of rats with septal hippocampus spared show a monotonic function, whereas those with temporal hippocampus spared
do not (Fig. 8, left). However, when the data are regrouped and
plotted as a function of number of days of training, the rats with
temporal hippocampus spared show a monotonic function (Fig. 8,
right). In reading Figure 8, it is important to recognize that this is
a meta-analysis and that the downward portions of either set of
data do not reflect worsening performance of individual rats with
training. Also, it is patent that rats with temporal hippocampus
spared are not above chance before day 8, independent of the
training protocol, and that septal and temporal spared rats reach
the same level of performance on this day. The plot of performance
by trials suggests that equivalent levels of performance would have
been reached after 64 trials had septal spared rats been given this
additional training.
Additional Training of Selected Groups and
Meta-analysis
Additional training in experiment 1.
Experiment 3 differs from experiment 1 above in two aspects of the
protocol: the number of days and the number of trials per day. As
some replications of experiment 1 were, in practice, conducted
chronologically after experiment 3, we were able to give a further 2
days of training to some of the temporal hippocampus spared rats
(5 of 10) in experiment 1. This consisted of 8 trials per day in two
sessions of four trials (total of 16 trials) starting immediately after
probe test 2 done on day 7. A further transfer test was then given
one day after this extra training (day 9; Fig. 3). The prior performance of these 5 animals was representative of their group. Figure
7 shows that the 5 rats with temporal hippocampus spared were
significantly above chance in the correct quadrant (51 ⫾ 9.7%; t ⫽
2.68, P ⬍ 0.05).
DISCUSSION
The key new finding of this series of experiments is that rats with
a 20 – 40% minislab of temporal hippocampus can learn an allocentric spatial reference memory task with certain training protocols. Our experiments establish that the finding of poor learning in
such animals by Moser et al. (1993, 1995) can be replicated, but
the implication that has been widely drawn from their results to the
effect that the temporal hippocampus might not be involved in
spatial learning and, therefore, that the septal and temporal hippocampus may fundamentally differ in function could be an overinterpretation. Changing the training protocol to one that increased the number of days of training, rather than the number of
trials, favored learning by lesioned animals with temporal hip-
Meta-analysis.
The results of experiments 1–3, although apparently inconsistent
with each other, point to the different outcomes being a result of
the training protocols used rather than any real disagreement. The
results of all the three experiments were therefore plotted together
for each group across experiments 1–3 (Fig. 8) as percentage time
in training quadrant during the probe tests against, on the x-axis,
either number of trials of training or number of days. Thus, when
plotted against the number of days, the data points correspond to
day 5 (PT 1 in experiments 1–3), day 7 (PT 2 in experiment 1), and
day 9 (PT 2 in experiments 2 and 3). Similarly when plotted
against the number of trials the points correspond to 16 trials (PT
1 in experiments 2 and 3), 32 trials (PT 1 in experiment 1), 48
trials (PT 2 in experiment 1), and 56 (PT 2 in experiments 2 and
3). Detailed statistical analysis of within-subject trends cannot be
done as the numbers of animals per data point vary through the
FIGURE 8.
Meta-analysis. percentage time in training quadrant
averaged across number of trials or number of days before given
transfer test for septal and temporal hippocampus spared rats. Insets:
the same for sham-lesioned animals. Dashed line represents chance
performance (25%).
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
pocampus spared. Moreover, no differences were observed in experiments 2 and 3 between rats with septal hippocampus spared
and those with temporal hippocampus spared across various measures of performance during probe test 2 (day 9).
It has been suggested before that rats with lesions to the hippocampus can learn a spatial water maze task, provided they are
either overtrained (Morris et al., 1990) or carefully shaped (Whishaw and Jarrard, 1996; but see McDonald and Hong, 2000). A
point to clarify about the successful learning of the one and the two
water maze tasks by the present rats with only the temporal hippocampus spared is that it is not merely a consequence of “overtraining.” First, rats with no hippocampus spared remained at
chance in all probe tests indicating that above chance performance
is hippocampus-dependent in these protocols. Second, the performance of rats with only temporal hippocampus spared was better
in experiment 3 (with only 32 training trials) than in experiment 1
(with 48 trials). Third, if the only reason for the above chance
performance of rats with temporal hippocampus spared was due to
overtraining, one might expect rats with septal hippocampus
spared to have been even better after such training if a septotemporal gradient of spatial learning capacity existed. Contrary to this,
in experiment 3, the level of performance of both septal and temporal hippocampus spared rats was identical after 8 days and did
not differ throughout training.
It was been observed previously that very little hippocampus is
enough to support learning of a reference memory task in the water
maze (Moser et al., 1995). We found that rats with as little as 20%
spared temporally or septally could learn a water maze task. These
findings are consistent with the idea of the hippocampus as a distributed memory system. We saw no correlation between the
amount of tissue spared and performance within the 20 – 40%
spared range. With respect to the concurrent learning of two water
mazes, it would not be unreasonable to expect that concurrent
learning of two reference memory tasks in different environments
would require more functional tissue than learning just one. However, the final level of performance reached by the different partial
lesion groups is equivalent in experiments 2 and 3. Curiously, it is
the sham animals that seem to perform somewhat better when
trained in just one water maze (experiment 3).
An interesting finding of this series of experiments is that, although both rats with only septal or temporal hippocampus spared
can learn a reference memory water maze task and reach levels of
performance that are equivalent across a variety of measures, they
are differentially sensitive to the training protocol. The level of
performance reached by rats with septal hippocampus spared
seems to be dependent on the total number of trials of training,
whereas that of rats with temporal hippocampus spared is more
sensitive to number of days of training (Fig. 8). Why might an
increased number of training days be more favorable for rats with
only temporal hippocampus spared than an increased number of
training sessions spaced within the day? How crucial is the length
of a potential consolidation period (i.e., time between two training
sessions)?
We discuss this subtle, rather than major, difference between the
septal and the temporal hippocampus with reference to the two
599
hypotheses proposed in the introduction: the functional differentiation hypothesis and the unitary framework.
The functional differentiation hypothesis (Moser and Moser,
1998b) is rooted in the finding that lesions to the septal hippocampus, but not the temporal, impaired learning of a reference memory task in the water maze (Moser et al., 1993, 1995). Since the
functional differentiation hypothesis was proposed, supporting evidence has been found in some behavioral studies in rodents (Hock
and Bunsey, 1998; Bannerman et al., 1999; Ferbinteanu and McDonald, 2001; but see Richmond et al., 1999), functional magnetic resonance imaging (fMRI) studies in humans (Maguire et al.,
2000), and molecular analysis in rats (Blum et al., 1999).
Our findings that partial lesions to the septal or the temporal
hippocampus do not always result in behavioral differences in the
water maze are at odds with a strict functional differentiation hypothesis. Anatomical support for the hypothesis is drawn mainly
from the pattern of extrinsic connectivity of the hippocampus:
specifically that peri- and postrhinal cortices (considered to convey
spatial information) project to the hippocampus via the EC such
that the septal half of the hippocampus receives a bigger proportion
of these projections than the temporal half (Burwell et al., 1995;
Burwell and Amaral, 1998a,b; Dolorfo and Amaral, 1998a,b),
whereas reciprocal connections with the amygdala and projections
to the hypothalamus (considered to deal with emotional and interoceptive information) arise exclusively from the temporal hippocampus (van Groen and Wyss, 1990; Jay et al., 1989; Köhler et
al., 1985). And yet, despite receiving comparatively few projections from the peri- and postrhinal cortices the temporal pole of the
hippocampus is capable of supporting spatial learning to levels
equivalent of those reached by rats with only septal hippocampus
in certain conditions. Thus, dividing the inputs from peri- and
postrhinal cortices, and those from the amygdala and hypothalamus as being, respectively, spatially and nonspatially (emotional or
interoceptive) relevant might be an oversimplication. Moreover, it
remains controversial as to whether the perirhinal cortex is necessary for spatial memory (for: Wiig and Bilkey, 1994a,b; Nagahara
et al., 1995; Liu and Bilkey, 1998a,b,c; reviewed in Suzuki, 1996;
against: Kolb et al., 1994; Wan et al., 1999; Bussey et al., 1999,
2000). Also, the amygdala (Packard et al., 1994; Packard and
Teather, 1998) and the hypothalamus (Ohl and Fuchs, 1999; Hagan et al., 1998; Oomura et al., 1993) have been found to modulate spatial processing in the hippocampus.
In addition, the implication that it should be possible to find a
double dissociation (i.e., a temporal, but not septal, hippocampusdependent nonspatial task in the absence of spatial learning) lacks
direct experimental evidence so far. Studies investigating this possibility have either failed to demonstrate a nonspatial task dependent on the temporal hippocampus (Hock and Bunsey, 1998;
Ferbinteanu and McDonald, 2000) or found that the same animals
with temporal hippocampus spared and good performance in a
nonspatial task were unimpaired in a spatial reference memory task
in the water maze (Richmond et al., 1999; but see Kjelstrup et al.,
2002). Moreover, spatial and nonspatial functions may not be
incompatible, with the septal part of the hippocampus being involved in nonspatial forms of memory such as social interaction
600
HOZ ET AL.
(File et al., 1998), nonspatial aspects of a delayed matching to
sample (DMS) task (Hampson et al., 1999), and verbal episodic
memory in humans (Fernández et al., 1998).
With respect to the unitary framework, this would predict that
a more demanding task (acquisition of two concurrent water
mazes) would engage the temporal hippocampus such that both
septal and temporal parts of the structure would be required to
learn the task. To our surprise, it seems that either of them is
sufficient to learn such a task with a training protocol that also
turns out to be adequate when learning a single water maze. Although it has been suggested that place fields are bigger in the
temporal hippocampus than in the septal hippocampus (Jung et
al., 1994; but see Poucet et al., 1994), the size of the place field does
not seem to have a critical function in context differentiation as
both septal and temporal partial lesion groups reach equivalent
levels of performance when trained in two concurrent water mazes.
That rats with only temporal hippocampus spared can learn a
spatial task under certain conditions does not directly support the
idea that the hippocampus performs a single function. It does
suggest, however, that the processing capacity in the septal and the
temporal hippocampus is similar enough that, in isolation, these
poles of the hippocampus can yield the same behavioral outcome.
Our studies are by no means the first ones to point out a role for
the temporal hippocampus in spatial memory. Infusion of galanin
into the ventral hippocampus of rats 20 min before training impairs acquisition of a spatial task (Ogren et al., 1996; Schött et al.,
1998). Acquisition of a delayed and a nondelayed spatial cued
radial-arm task is impaired by temporary inactivation of the ventral
CA1/subiculum (Floresco et al., 1997). Infusion of nicotine in the
ventral hippocampus facilitates performance in a working memory
task in a radial-arm maze (Levin et al., 1999). Also, Vann et al.
(2000) find c-fos activation in the temporal, as well as the septal,
hippocampus of rats trained in a spatial radial arm maze. In primates, single units with spatial correlates were found in the anterior
(temporal) hippocampus (Colombo and Gross, 1994; Colombo et
al., 1998). Our findings are therefore consistent with others in
implicating temporal hippocampus in spatial learning.
Differences along the longitudinal axis of the hippocampus do
nonetheless exist and these might explain why the septal and the
temporal hippocampus respond differently to variations in the
training protocol. We have already mentioned some of the differences in the extrinsic circuit, such as quantitative differences in the
density of the peri- and postrhinal inputs. Certain aspects of the
intrinsic circuit also vary quantitatively along the longitudinal axis
despite an otherwise homogeneous intrinsic architecture of the
hippocampus along this axis. Differences in the organization of the
projections result in gradients along this axis (Seress and Pokorny,
1981; Li et al., 1994; Amaral and Witter, 1995). Differences in
synaptic plasticity between the septal and temporal hippocampus
have been reported in the capacity to induce long-term potentiation (LTP) (Papatheodoropoulos and Kostopoulos, 2000) and
long-term depression (LTD) (Izaki et al., 2000). Neuromodulation also differs along the septotemporal axis with neuromodulatory transmitters found in higher concentration in the temporal
hippocampus (Gall et al., 1981; Verney et al., 1985; van Leeuwen
et al., 1985; Caffé et al., 1987; Febvret et al., 1991; Valkna et al.,
1995; Florin et al., 2000).
With the exception of the connections with the amygdala and
hypothalamus, the differences described are quantitative rather
than qualitative. This could explain the quantitative nature of the
behavioral difference observed across the three experiments. However, although one can see the utility of having two different consolidation speeds within the hippocampus (the efficiency of such
system has been highlighted before by Murre (1996) in relation to
hippocampus and neocortex), little is known about the functional
implications of each of the aspects of the circuitry mentioned to
suggest an underlying mechanism for the differential sensitivity to
training protocol observed between rats with septal and temporal
hippocampus spared.
The temporal hippocampus can therefore support spatial learning of a water maze reference memory task under certain training
conditions. The extent to which this generalizes to other, more
demanding, spatial tasks remains unknown. The conditions under
which this part of the hippocampus would be sufficient or even
necessary are not clear and might depend on more complex parameters than an adequate training protocol or a spatial versus nonspatial task dichotomy. While recognizing anatomical differences between the septal and the temporal hippocampus, these differences
are mainly quantitative and their functional implication is far from
clear. The reason it is so difficult to find behavioral correlates of the
septotemporal gradient might reside in one aspect of the hippocampal anatomy that has drawn the attention of anatomists and
network modelers alike: the intrinsic associational circuits. The
longitudinal-associational pathway of CA3 and Schaffer collaterals
extends broadly, although admittedly the distribution is less broad
in the temporal hippocampus (Swanson et al., 1978; Ishizuka et
al., 1990; Tamamaki and Nojyo, 1991; Li et al., 1994). Consequently, information could spread outside individual lamellae and
across septotemporal borders such that the segregation of extrinsic
inputs along the septotemporal axis is lost as information moves
along the transverse axis. For this reason, although differences
along the septotemporal axis suggest that the type of information
and the way it is processed may vary along this axis, it is possible
that a particular level of hippocampus has access to information
reaching other levels along the longitudinal axis by virtue of this
profuse associational system. Episodic memory (Tulving, 1983;
Vargha-Kahdem et al., 1997; Morris and Frey, 1997) is one example in which the need for a system that can integrate different types
of information becomes apparent, for episodic memories are made
of unique combinations of stimuli of a very different nature. It is in
these circumstances that the value of an intrinsic associational system, crossing over the septotemporal boundaries, is made clear, as
a memory only acquires its full significance when the different
components are considered together. Thus, by testing septal or
temporal hippocampal lesioned animals in tasks that rely mainly
on one memory component (sensory-perirhinal, emotional-amygdala), one might find support for the functional differentiation
hypothesis but also be discarding an essential asset of the hippocampus: its capacity to integrate information from different origins.
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
Acknowledgments
The authors thank Andrew Bernard for the animal care and
May-Britt Moser, Edvard Moser, Gordon Arbuthnott, Cali Ingham, and Richard Ribchester for discussions and providing equipment for the anatomical analysis. Livia de Hoz is in debt to David
Amaral (University of California, Davis) and Rebecca Burwell
(Brown University) for advice on the flat map techniques. This
work was supported by the ESF and by MRC Programme grant
G9200370/2 (to R.G.M.M.).
REFERENCES
Alescio-Lautier B, Paban V, Soumireu-Mourat B. 2000. Neuromodulation of memory in the hippocampus by vasopressin. Eur J Pharmacol
405:63–72.
Amaral DG, Kurz J. 1985. An analysis of the origins of the cholinergic and
noncholinergic septal projections to the hippocampal formation of the
rat. J Comp Neurol 240:37–59.
Amaral DG, Witter M. 1995. The hippocampal formation. In: Paxinos G,
editor. The rat nervous system. London: Academic Press. p 443– 493.
Bannerman DM, Yee BK, Good MA, Heupel MJ, Iversen SD, Rawlins
JN. 1999. Double dissociation of function within the hippocampus: a
comparison of dorsal, ventral, and complete hippocampal cytotoxic
lesions. Behav Neurosci 113:1170 –1188.
Blum S, Moore AN, Adams F, Dash PK. 1999. A mitogen-activated
protein kinase cascade in the CA1/CA2 subfield of the dorsal hippocampus is essential for long-term spatial memory. J Neurosci 19:
3535–3544.
Brazier MAB. 1970. Regional activities within the human hippocampus
and hippocampal gyrus. Exp Neurol 26:354 –368.
Burwell RD, Amaral DG. 1998a. Cortical afferents of the perirhinal,
postrhinal, and entorhinal cortices of the rat. J Comp Neurol 398:
179 –205.
Burwell RD, Amaral DG. 1998b. Perirhinal and postrhinal cortices of the
rat: interconnectivity and connections with the entorhinal cortex.
J Comp Neurol 391:293–321.
Burwell RD, Witter MP, Amaral DG. 1995. Perirhinal and postrhinal
cortices of the rat: a review of the neuroanatomical literature and
comparison with findings from the monkey brain [published erratum appears in Hippocampus 1996;6:340]. Hippocampus 5:390 –
408.
Bussey TJ, Muir JL, Aggleton JP. 1999. Functionally dissociating aspects
of event memory: the effects of combined perirhinal and postrhinal
cortex lesions on object and place memory in the rat. Journal of Neuroscience 19:495–502.
Bussey TJ, Duck J, Muir JL, Aggleton JP. 2000. Distinct patterns of
behavioural impairments resulting from fornix transection or neurotoxic lesions of the perirhinal and postrhinal cortices in the rat. Behav
Brain Res 111:187–202.
Caffé AR, van Leeuwen FW, Luiten PG. 1987. Vasopressin cells in the
medial amygdala of the rat project to the lateral septum and ventral
hippocampus. J Comp Neurol 261:237–252.
Colombo M, Gross CG. 1994. Responses of inferior temporal cortex and
hippocampal neurons during delayed matching to sample in monkeys
(Macaca fascicularis). Behav Neurosci 108:443– 455.
Colombo M, Fernandez T, Nakamura K, Gross CG. 1998. Functional
differentiation along the anterior-posterior axis of the hippocampus in
monkeys. J Neurophysiol 80:1002–1005.
601
Dolorfo CL, Amaral DG. 1998a. Entorhinal cortex of the rat: topographic
organization of the cells of origin of the perforant path projection to
the dentate gyrus. J Comp Neurol 398:25– 48.
Dolorfo CL, Amaral DG. 1998b. Entorhinal cortex of the rat: organization of intrinsic connections. J Comp Neurol 398:49 – 82.
Elul R. 1964. Regional differences in the hippocampus of the cat. I.
Specific discharge patterns of the dorsal and ventral hippocampus in
their role in generalized seizures. Electroencephalogr Clin Neurophysiol 16:470 – 488.
Febvret A, Berger B, Gaspar P, Verney C. 1991. Further indication that
distinct dopaminergic subsets project to the rat cerebral cortex: lack of
colocalization with neurotensin in the superficial dopaminergic fields
of the anterior cingulate, motor, retrosplenial and visual cortices. Brain
Res 547:37–52.
Ferbinteanu J, McDonald RJ. 2000. Dorsal and ventral hippocampus—
same or different? Psychobiology 28:314 –324.
Ferbinteanu J, McDonald RJ. 2001. Dorsal/ventral hippocampus, fornix,
and conditioned place preference. Hippocampus 11:187–200.
Fernández G, Weyerts H, Schrader-Bölsche M, Tendolkar I, Smid HG,
Tempelmann C, Hinrichs H, Scheich H, Elger CE, Mangun GR,
Heinze HJ. 1998. Successful verbal encoding into episodic memory
engages the posterior hippocampus: a parametrically analyzed functional magnetic resonance imaging study. J Neurosci 18:1841–
1847.
File S, Kenny P, Ouagazzal A. 1998. Bimodal modulation by nicotine of
anxiety in the social interaction test: role of the dorsal hippocampus.
Behav Neurosci 112:1423–1429.
Floresco SB, Seamans JK, Phillips AG. 1997. Selective roles for hippocampal, prefrontal cortical, and ventral striatal circuits in radial-arm maze
tasks with or without a delay. J Neurosci 17:1880 –1890.
Florin S, Meunier J, Costentin J. 2000. Autoradiographic localization of
[3H]nociceptin binding sites in the rat brain. Brain Res 880:11–16.
Gall C, Brecha N, Karten HJ, Chang KJ. 1981. Localization of enkephalin-like immunoreactivity to identified axonal and neuronal populations of the rat hippocampus. J Comp Neurol 198:335–350.
Hagan MM, Castaneda E, Sumaya IC, Fleming SM, Galloway J, Moss
DE. 1998. The effect of hypothalamic peptide YY on hippocampal
acetylcholine release in vivo: implications for limbic function in bingeeating behavior. Brain Res 805:20 –28.
Hampson RE, Simeral JD, Deadwyler SA. 1999. Distribution of spatial
and nonspatial information in dorsal hippocampus. Nature 402:610 –
614.
Hock BJ Jr, Bunsey MD. 1998. Differential effects of dorsal and ventral
hippocampal lesions. J Neurosci 18:7027–7032.
Izaki Y, Takita M, Nomura M. 2000. Comparative induction of longterm depression between dorsal and ventral hippocampal CA1 in the
anesthetized rat. Neurosci Lett 294:171–174.
Ishizuka N, Weber J, Amaral DG. 1990. Organization of intrahippocampal projections originating from CA3 pyramidal cells in the rat. J
Comp Neurol 295:580 – 623.
Jarrard LE. 1989. On the use of ibotenic acid to lesion selectively different
components of the hippocampal formation. J Neurosci Methods 29:
251–259.
Jay TM, Glowinski J, Thierry AM. 1989. Selectivity of the hippocampal
projection to the prelimbic area of the prefrontal cortex in the rat.
Brain Res 505:337–340.
Jung MW, Wiener SI, McNaughton BL. 1994. Comparison of spatial
firing characteristics of units in dorsal and ventral hippocampus of the
rat. J Neurosci 14:7347–7356.
Kjelstrup KG, Tuvnes FA, Steffenach HA, Murison R, Moser EI, Moser
MB. 2002. Reduced fear expression after lesions of the ventral hippocampus. Proc Natal Acad Sci USA 99:10825–10830.
Kolb B, Buhrmann K, McDonald R, Sutherland RJ. 1994. Dissociation of
the medial prefrontal, posterior parietal, and posterior temporal cortex
602
HOZ ET AL.
for spatial navigation and recognition memory in the rat. Cerebral
Cortex 4:664 – 680.
Koreli A. 1977. Influences of dorsal and ventral hippocampus on hypothalamic self-stimulation. Physiol Behav 19:713–717.
Köhler C, Swanson LW, Haglund L, Wu JY. 1985. The cytoarchitecture,
histochemistry and projections of the tuberomammillary nucleus in
the rat. Neuroscience 16:85–110.
Lepage M, Habib R, Tulving E. 1998. Hippocampal PET activations of
memory encoding and retrieval: the HIPER model. Hippocampus
8:313–322.
Levin ED, Christopher NC, Weaver T, Moore J, Brucato F. 1999. Ventral
hippocampal ibotenic acid lesions block chronic nicotine-induced spatial working memory improvement in rats. Brain Res Cogn Brain Res
7:405– 410.
Li X, Somogyi P, Ylinen A, Buzsaki G. 1994. The hippocampal CA3
network: an in vivo intracellular labelling study. J Comp Neurol 339:
181–208.
Liu P, Bilkey DK. 1998a. Excitotoxic lesions centered on perirhinal cortex
produce delay-dependent deficits in a test of spatial memory. Behav
Neurosci 112:512–524.
Liu P, Bilkey DK. 1998b. Perirhinal cortex contributions to performance
in the Morris water maze. Behav Neurosci 112:304 –315.
Liu P, Bilkey DK. 1998c. Lesions of perirhinal cortex produce spatial
memory deficits in the radial maze. Hippocampus 8:114 –121.
Lorente de No R. 1934. Studies on the structure of the cerebral cortex. III.
Continuation of the study of the ammonic system. J Psychol Neurol
46:113–177.
Maguire EA, Gadian DG, Johnsrude IS, Good CD, Ashburner J,
Frackowiak RS, Frith CD. 2000. Navigation-related structural
change in the hippocampi of taxi drivers. Proc Natl Acad Sci USA
97:4398 – 4403.
McDonald RJ, Hong NS. 2000. Rats with hippocampal damage are impaired on place learning in the water task when overtrained under
constrained conditions. Hippocampus 10:153–161.
Meibach RC, Siegel A. 1977. Efferent connections of the septal area in the
rat: an analysis utilizing retrograde and anterograde transport methods.
Brain Res 119:1–20.
Montone KT, Fass B, Hamill GS. 1988. Serotonergic and nonserotonergic projections from the rat interpeduncular nucleus to the septum,
hippocampal formation and raphe: a combined immunocytochemical
and fluorescent retrograde labelling study of neurons in the apical
subnucleus. Brain Res Bull 20:233–240.
Morris RG, Frey U. 1997. Hippocampal synaptic plasticity: role in spatial
learning or the automatic recording of attended experience? Philos
Trans R Soc Lond Ser B 352:1489 –1503.
Morris RGM. 1984. Developments of a water maze procedure for studying spatial learning in the rat. J Neurosci Methods 11:47– 60.
Morris RGM, Schenk F, Tweedie F, Jarrard LE. 1990. Ibotenate lesions of
hippocampus and/or subiculum: dissociating components of allocentric spatial learning. Eur J Neurosci 2:1016 –1028.
Moser E, Moser M-B, Andersen P. 1993. Spatial learning impairment
parallels the magnitude of dorsal hippocampal lesions, but is hardly
present following ventral lesions. J Neurosci 13:3916 –3925.
Moser MB, Moser EI, Forrest E, Andersen P, Morris RG. 1995. Spatial
learning with a minislab in the dorsal hippocampus. Proc Natl Acad
Sci USA 92:9697–9701.
Moser MB, Moser EI. 1998a. Distributed encoding and retrieval of spatial
memory in the hippocampus. J Neurosci 18:7535–7542.
Moser MB, Moser EI. 1998b. Functional differentiation in the hippocampus. Hippocampus 8:608 – 619.
Murre JM. 1996. TraceLink: a model of amnesia and consolidation of
memory. Hippocampus 6:675– 684.
Nadel L. 1968. Dorsal and ventral hippocampal lesions and behaviour.
Physiol Behav 3:891– 900.
Nagahara AH, Otto T, Gallagher M. 1995. Entorhinal-perirhinal lesions
impair performance of rats on two versions of place learning in the
Morris water maze. Behav Neurosci 109:3–9.
Ogren SO, Kehr J, Schott PA. 1996. Effects of ventral hippocampal galanin on spatial learning and on in vivo acetylcholine release in the rat.
Neuroscience 75:1127–1140.
Ohl F, Fuchs E. 1999. Differential effects of chronic stress on memory
processes in the tree shrew. Brain Res Cogn Brain Res 7:379 –387.
O’Keefe J, Nadel L. 1978. The hippocampus as a cognitive map. Oxford:
Clarendon.
Oomura Y, Sasaki K, Li AJ. 1993. Memory facilitation educed by food
intake. Physiol Behav 54:493– 498.
Packard MG, Teather LA. 1998. Amygdala modulation of multiple memory systems: hippocampus and caudate-putamen. Neurobiol Learn
Mem 69:163–203.
Packard MG, Cahill L, McGaugh JL. 1994. Amygdala modulation of
hippocampal-dependent and caudate nucleus-dependent memory
processes. Proc Natl Acad Sci USA 91:8477– 8481.
Papatheodoropoulos C, Kostopoulos G. 2000. Decreased ability of rat
temporal hippocampal CA1 region to produce long-term potentiation. Neurosci Lett 279:177–180.
Passino E, Ammassari-Teule M. 1999. Visual discrimination in inbred
mice: strain-specific involvement of hippocampal regions. Physiol Behav 67:393–399.
Paxinos G, Watson C. 1998. The rat brain in stereotaxic coordinates. San
Diego, CA: Academic Press.
Poucet B, Thinus-Blanc C, Muller RU. 1994. Place cells in the ventral
hippocampus of rats. NeuroReport 5:2045–2048.
Rawlins JNP. 1985. Associations across time: the hippocampus as a temporary memory store. Behav Brain Sci 8:479 – 528.
Richmond MA, Yee BK, Pouzet B, Veenman L, Rawlins JN, Feldon J,
Bannerman DM. 1999. Dissociating context and space within the
hippocampus: effects of complete, dorsal, and ventral excitotoxic hippocampal lesions on conditioned freezing and spatial learning. Behav
Neurosci 113:1189 –1203.
Schött PA, Bjelke B, Ogren SO. 1998. Distribution and kinetics of galanin infused into the ventral hippocampus of the rat: relationship to
spatial learning. Neuroscience 83:123–136.
Seress L, Pokorny J. 1981. Structure of the granular layer of the rat dentate
gyrus: a light microscopic and Golgi study. J Anat 133:181–195.
Spooner RI, Thomson A, Hall J, Morris RG, Salter SH. 1994. The Atlantis platform: a new design and further developments of Buresova’s
on-demand platform for the water maze. Learn Mem 1:203–211.
Stevens R, Cowey A. 1973. Effects of dorsal and ventral hippocampal
lesions on spontaneous alternation, learned alternation and probability
learning in rats. Brain Res 52:203–224.
Suzuki WA. 1996. The anatomy, physiology and functions of the perirhinal cortex. Curr Opin Neurobiol 6:179 –186.
Swanson LW, Wyss JM, Cowan WM. 1978. An autoradiographic study
of the organization of intrahippocampal association pathways in the
rat. J Comp Neurol 181:681–715.
Tamamaki N, Nojyo Y. 1991. Crossing fiber arrays in the rat hippocampus as demonstrated by three-dimensional reconstruction. J Comp
Neurol 303:435– 442.
Tulving E. 1983. Elements of episodic memory. London: Oxford University Press.
Valkna A, Juréus A, Karelson E, Zilmer M, Bartfai T, Langel U. 1995.
Differential regulation of adenylate cyclase activity in rat ventral and
dorsal hippocampus by rat galanin. Neurosci Lett 187:75–78.
van Groen T, Wyss JM. 1990. Extrinsic projections from area CA1 of the
rat hippocampus: olfactory, cortical, subcortical, and bilateral hippocampal formation projections. J Comp Neurol 302:515–528.
van Leeuwen FW, van Heerikhuize J, van der Meulen G, Wolters P. 1985.
Light microscopic autoradiographic localization of [3H]oxytocin
_________________________
SEPTOTEMPORAL AXIS OF HIPPOCAMPUS AND SPATIAL LEARNING
binding sites in the rat brain, pituitary and mammary gland. Brain Res
359:320 –325.
Vann SD, Brown MW, Erichsen JT, Aggleton JP. 2000. Fos imaging
reveals differential patterns of hippocampal and parahippocampal subfield activation in rats in response to different spatial memory tests.
J Neurosci 20:2711–2718.
Vargha-Khadem F, Gadian DG, Watkins KE, Connelly A, Van Paesschen
W, Mishkin M. 1997. Differential effects of early hippocampal pathology on episodic and semantic memory. Science 277:376 –380.
Verney CB, Baulac M, Berger B, Alvarez C, Vigny A, Helle KB. 1985. Morphological evidence for a dopaminergic terminal field in the hippocampal
formation of young and adult rat. Neuroscience 14:1039 –1052.
603
Wan H, Aggleton JP, Brown MW. 1999. Different contributions of the
hippocampus and perirhinal cortex to recognition memory. J Neurosci
19:1142–1148.
Whishaw IQ, Jarrard LE. 1996. Evidence for extrahippocampal involvement in place learning and hippocampal involvement in path integration. Hippocampus 6:513–524.
Wiig KA, Bilkey DK. 1994a. The effects of perirhinal cortical lesions
on spatial reference memory in the rat. Behav Brain Res 63:101–
109.
Wiig KA, Bilkey DK. 1994b. Perirhinal cortex lesions in rats disrupt
performance in a spatial DNMS task. NeuroReport 5:1405–
1408.
Download