Brain Research 764 Ž1997. 101–108
Research report
The medial supramammillary nucleus, spatial learning and the frequency of
hippocampal theta activity
Wei-Xing Pan, Neil McNaughton
)
Department of Psychology and Centre for Neuroscience, UniÕersity of Otago, Dunedin, New Zealand
Accepted 1 April 1997
Abstract
Previous studies have shown that the presence of hippocampal theta activity Žtheta. is important for learning and that the medial
supramammillary nucleus ŽSuM. is involved in the control of the frequency of theta. In the present experiments, a single-day version Ž20
trials. of the Morris water maze was used to investigate the effects of drug injections into SuM on hippocampal theta frequency and
spatial learning. Two groups of rats received an injection of chlordiazepoxide ŽCDP, 0.5 m l, 40 m grm l. or saline Ž0.5 m l. into SuM 10
min before training in the Morris water maze. Two other groups of rats received an i.p. injection of 5 mgrkg CDP or saline, and two
further groups received short Ž10 min. or long Ž15 min. immersion in cool water Ž228C. before training. The results showed: Ž1. in all
groups theta frequency was an inverse logarithmic function of training time; Ž2. systemic CDP or long cool water exposure decreased
theta frequency to a greater extent Žby 1 Hz., and also impaired learning to a greater extent, than the other treatments; Ž3. that SuM-CDP
produced a modest decrease in theta frequency Ž0.35–0.5 Hz. and a modest impairment of spatial learning. These data suggest that theta
frequency per se may be important for spatial learning and that total abolition of theta is not necessary for dysfunction; and that while a
lesser part of the effect of i.p. CDP on spatial learning appears to be mediated by SuM the greater part appears to involve other nuclei as
well. q 1997 Elsevier Science B.V.
Keywords: Hypothalamus; Hippocampus; Theta; Learning; Water maze; Benzodiazepine
1. Introduction
A large body of evidence indicates that the hippocampal
system plays an important role in memory function, especially spatial memory. Spatial memory in the rat is particularly sensitive to hippocampal damage w17x. Disrupting
brain structures associated with the hippocampal formation, e.g. the septum, the entorhinal-perirhinal cortex or
mammillary region, leads to equivalent deficits w4,18,20x.
The hippocampal formation often shows synchronous
rhythmic firing of large numbers of cells Žtheta activity.
which can give rise to a sinusoidal extracellular rhythm.
Theta activity is likely to be important for hippocampal
function. For example, disrupting the septum, which contains the pacemaker for theta, or other structures which
affect theta, results in deficits in spatial and nonspatial
working memory w1,5,6x. Further, systemic injections of
)
Corresponding author. Department of Psychology, University of
Otago, P.O. Box 56, Dunedin, New Zealand. Fax: q64 Ž3. 479-8335;
E-mail: [email protected]
0006-8993r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved.
PII S 0 0 0 6 - 8 9 9 3 Ž 9 7 . 0 0 4 3 1 - 9
benzodiazepines reduce the frequency of theta w14,19x and
also impair spatial learning in the Morris water maze w13x.
This raises the possibility that theta is sufficiently important for hippocampal function that a change solely in its
frequency is sufficient to produce some dysfunction.
Data from both neuroanatomy and neurophysiology indicate that the supramammillary nucleus ŽSuM. is involved
in septo-hippocampal function. SuM projects to the medial
septum and hippocampus w7,22,23x, and affects hippocampal cell firing and theta rhythm w2,8,9,16x. There is good
evidence that SuM is a site at which the intensity of tonic
afferent input from the midbrain is converted to phasic
firing which controls the frequency of theta activity of
cells in the septum and hippocampus w8,9x. It follows that
SuM may be involved in some of the same learning and
memory tasks as the hippocampal formation. We have
previously shown that injection of chlordiazepoxide ŽCDP.
into the medial SuM decreases the frequency of theta
elicited by stimulation of the reticular formation w12x, and
so we predicted that CDP injected into SuM should reduce
both spatial learning and theta frequency in the Morris
water maze.
102
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
The Morris water maze has been widely used to evaluate spatial learning. But the traditional procedure requires
multiple days of training, which is not ideal for testing
with intracranial injections, since repeated injections can
cause lesions of brain cells and fibres. A single-day version of the Morris water maze has recently been developed
w3,10,11x, and so, in the present experiments, we used this
to investigate the effects on theta rhythm and spatial
memory of infusion of CDP into SuM. Since rats with
electrodes had not previously been tested in this paradigm,
we also tested the effects of systemic CDP on theta and
concurrent spatial learning. In addition, since the rats were
exposed to water for a long time in this single day
paradigm; since we found that theta frequency decreased
markedly during testing; and since body temperature affects theta frequency, we also tested the effects of prior
cooling on theta and spatial learning.
2. Materials and methods
2.1. Animals and surgery
Thirty naive Sprague–Dawley rats Ž250–400 g. were
used for intracranial cannula testing, 11 for systemic testing and 13 for water exposure testing. All rats were
implanted with recording electrodes. Intracranial cannula
rats had, in addition, both cannulae and stimulating electrodes. Electrodes and cannulae were stereotaxically implanted under anaesthesia with sodium pentobarbital Ž60
mgrkg, i.p... The recording electrodes consisted of two
stainless-steel wires, each Teflon coated, with a 70 m m
outside diameter, twisted together with the tips separated
by 2.5 mm vertically, and were placed with the upper tip
in the CA1 region of the dorsal hippocampus Žbregma A-P
3.5, M-L 2.0, D-V 2.7 for the upper tip from the skull..
Stimulating electrodes were similar, but with a tip separation of 0.5 mm vertically, and were implanted anterior to
and in the region of nucleus reticularis pontis oralis Žbregma
A-P 7.5, M-L 1.0, D-V 8.5 from skull.. These were
included for testing reticular elicited theta after the water
maze test, as an active control in the event that there was
no effect on theta in the water maze – but were not in fact
needed. The electrodes were soldered to Amphenol gold
connectors. A stainless-steel skull screw served as ground.
A stainless-steel guide cannula Ž25 gauge. was placed 1
mm above the medial SuM ŽA-P 4.5, M-L 0.8, D-V 7.0
from skull, at an angle of 6 degrees to the vertical.. These
coordinates achieved a midline placement of the tip while
avoiding damage to the sagittal sinus. Silica capillary
tubing ŽVS-140-40, Scientific Glass Engineering, UK, 140
m m external diameter and 40 m m internal diameter., the
tip of which extended 1.5 mm beyond the tip of the guide
cannula, was used as an injection needle to minimize
damage to the small nucleus Žmedial SuM.. The depth of
penetration was controlled by a stainless-steel collar glued
to the capillary. The guide cannula and electrodes were
affixed to the skull using six stainless-steel screws and
dental acrylic. An obturator Žfashioned insect pin cut to be
level with the tip of the guide cannula. was inserted into
the guide cannula to prevent occlusion.
All animals were allowed one week to recover from
surgery prior to any testing.
2.2. EEG recording and drug injection and water exposure
treatment
The electrodes were connected via Amphenol gold connectors to a dual field effect transistor, preamplified ŽGrass
P511K, 1–30 Hz band pass filter. and extracellular field
activity was digitised at 100 Hz for subsequent analysis.
The frequency of theta was determined by measuring the
interpeak interval of the theta waves occurring in successive 1 s periods. All of the EEG within each trial was
measured 1 s at a time, and the average frequency across
seconds for each trial was calculated. Theta was essentially
continuous while the rat was in the pool. Microinfusions
Ž0.5 m l. were made over 4 min via a 10 m l Hamilton
syringe connected to an electrical microdrive. The low rate
of infusion was used to minimise the spread of the drug.
Each 0.5 m l infusion contained either CDP Ž40 m grm l in
saline. or saline. This dose and volume has proved effective in reducing theta frequency in previous experiments
with injections into medial SuM w12x. Higher concentrations cannot be used because of cannula blocking, lower
concentrations would produce negligible effects Žsee Section 3. and higher volumes would have increased the likely
spread of the drug up the cannular track, and so a dose-response curve was not attempted. The glass needle was left
in the guide cannula for 1 min after the injection to
facilitate passive drug diffusion, then the needle was removed and the obturator was reinserted. Intraperitoneal
injections of CDP Ž5 mgrmlrkg. or saline Ž1 mlrkg.
were given to separate groups 10 min before testing. In
water exposure testing, rats were initially divided into
three groups receiving different types of exposure to water
in a bucket before water maze testing. One group received
warm water Ž388C. and the rat was put into the water for
15 s 8 times over a period of 15 min before water maze
training. Two other groups received cool water Ž228C.. For
one group Žlong cooling., each rat was put into the water 8
times over 15 min and for the second Žshort cooling. for 4
times over 10 min.
2.3. Apparatus and procedure
The water maze consisted of a rigid black plastic pool
Ž150 cm in diameter, 35 cm high., and was placed in a
fixed position in the testing room surrounded by various
extramaze visual cues, such as a door, tables, lamps and
instruments. The pool was filled to a 25 cm depth with
258C Ž"18C. water. A black plastic platform Ž15 cm
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
square. was placed 1.5 cm beneath the water surface in the
centre of the southeast quadrant. Note that nothing was
added to the water to make it opaque, but that probe tests
indicate that the animals cannot see the black platform
against the background of the black pool.
A single-day’s testing with a total of 20 training trials
was used for each rat Žexcept that only 12 trials were used
where rats had received prior water exposure so that their
total cooling time was no more than that of rats which
were not pre-cooled.. A single intracranial or systemic
injection was given Žsee Section 2.2.. The EEG cable was
connected and was sealed by Blu-tack ŽBostik Pty. Ltd.,
Australia. for water proofing. Ten minutes after the start of
the injection, the rat was placed in the pool facing the wall
and allowed 60 s to find the escape platform. If it found
the platform within the given time it remained there for 15
s, otherwise it was guided to it and stayed there for 15 s.
Then the next trial started. Note that there was no inter-trial
interval as we wished to minimise the total testing time
after injection and that the time after injection of 10 min
was chosen on the basis of our previous experiments so
that the effect of intracranial CDP would be optimal during
the early trials when we presumed the greatest effect on
learning would be obtained Žbut see Section 3.. After the
last of the 20 Žor 12. training trials, a probe test was given
in which the rat was required to swim in the pool without
the escape platform for 60 s. All rats were released from
the north location on the first trial and then followed a
counterbalanced sequence ŽNSWEENSWWENSSWENNSWE..
The swim path was recorded by a video cassette system
and the hippocampal EEG was recorded. The beginning
and the end of each trial was marked by pressing a button
which controlled concurrently both an LED display visible
in the video frame Žbut not visible to the rat. and distinct
event marks on a second EEG channel digitised concurrently with the hippocampal recording channel. The swim
distance, speed and escape time were analyzed later by a
computer image analysis system.
2.4. Histology
Animals were deeply anaesthetized Žsodium pentobarbital, 60 mg per rat. and perfused transcardially with
saline followed by 10% formalin. The brains were removed and placed in 30% sucrose-formalin for 3 days, and
were frozen sectioned Ž60 m m. coronally. The sections
were stained with thionin. Cannula tip positions were
reconstructed according to Paxinos and Watson w28x. Note
that the medial supramammillary nucleus is referred to by
them simply as the supramammillary nucleus.
2.5. Data analysis
The total swim distance, escape latency, swim speed
and average theta frequency of each trial, were each
103
submitted to analysis of variance with extraction of orthogonal linear, quadratic and cubic polynomial components of
trials. For these analyses, trials were combined into blocks
of two trials. In these analyses the linear component of
trial blocks is identical to a conventional linear regression
slope coefficient and the higher order trends represent
symmetric curves with increasing numbers of inflections.
The percent time spent in the correct quadrant of the pool
Žthe percent data were submitted to angular transformation
to normalise the variance., and number of times passing
through the target area during the probe tests were each
submitted to focused t-test.
3. Results
3.1. Systemic injection
The five rats treated with systemic saline learned the
location of the platform surprisingly quickly Ž; 8 trials..
The six rats treated with CDP also showed steady learning,
but this was slower than the saline controls ŽFig. 1.. Path
length averaged over trials was longer for CDP rats Žgroup,
F s 9.10, df s 1r9, P s 0.015., and this difference was
greatest towards the beginning of training Žgroup = trial
blocks, linear: F s 5.06, df s 1r181, P s 0.026.. Escape
latency data showed a similar pattern. The swimming
speed in the CDP group decreased towards the end of
training Žgroup = trial blocks, linear: F s 14.75, df s
1r181, P - 0.001..
During the 60 s probe trial ŽFig. 2., a bias for the target
quadrant was shown in the saline group Ž t s 4.22, df s 4,
P - 0.01, compared to chance level 25%., but not in the
CDP group Ž t s 0.33, df s 5, P ) 0.05.. There is a significant difference between them Ž t s 2.59, df s 9, P - 0.05..
The target crossings ŽFig. 3. in the CDP group were
significantly fewer than in the saline group Ž t s 4.16,
df s 9, P - 0.01..
Over the course of training, the theta frequency in both
the saline and the CDP group decreased steadily with
trials, and there was about a 1 Hz additional drop in theta
frequency in the CDP group compared to the saline group
Žgroup, F s 15.57, df s 1r9, P s 0.003, Fig. 4. which did
not itself vary significantly with trials. Since the rats of the
saline group took less time to finish the 20 training trials
Žabout 15 min compared to over 20 min in the CDP rats.,
it was possible that a substantial part of the theta frequency
change in the CDP group was a result of their greater time
in the water rather than trial numbers or drug. Fig. 4 shows
that the theta frequency for both the saline and CDP
groups is a nearly perfect logarithmic function of time
Ž r 2 s 0.96, r 2 s 0.97, respectively.. However, while some
part of the overall difference in frequency between the
groups in later trials is due to time, the frequency in the
CDP group is consistently lower than that of the saline
group when time after injection is controlled for ŽFig. 4..
104
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
Fig. 1. Effects of systemic CDP Ž5 mgrkg. or saline, SuM-CDP Ž0.5 m l, 40 m grm l. or SuM-saline, and short or long cool water exposure on swim speed
and the distance taken to locate the submerged platform over training trials Žshown as two-trial blocks.. Left column: systemic CDP. An ANOVA on path
length revealed a significant group difference Ž P - 0.05., and a significant difference in the linear trend across trials between groups Ž P - 0.05.. The
swimming speed in the CDP group also decreased towards the end of training Ž P - 0.001.. Centre column: SuM-CDP. The learning of the SuM-saline
group and SuM-CDP group appeared as good as that of the systemic saline group in the first experiment until the 5th trial block. From the 5th trial block
there appeared to be no further learning in the SuM-CDP group while the SuM-saline group continued to learn as well as had systemic saline. An ANOVA
on distance revealed a significant linear group by trial–block interaction Ž P - 0.001.. There were no significant changes in swimming speed. Right
column: water exposure. An ANOVA on path length revealed a significant separation between short cool-water exposure and long cool-water exposure
groups Ž P - 0.001.. But there was no significant difference between the short-cooling group and the systemic saline group taken from the first experiment
Ž P ) 0.05.. There were no significant changes in swimming speed between the water exposure groups Ž P ) 0.05..
3.2. Intracranial infusions
Histology showed that there were 13 rats in the CDP
group and 9 rats in the saline group with needle tips in or
very close to the supramammillary nucleus as defined by
Paxinos and Watson w28x and in the region where we have
previously obtained reductions in theta frequency. Note
that this is the medial supramammillary nucleus in some
other classifications and was referred to as such above.
There were 8 rats with placements further from SuM in
Fig. 2. As for Fig. 1 but showing percentage time spent in each of the four quadrants and the number of times passing through the previously correct target
area during the 60 s probe trial. The black bar represents the target quadrant. The ), )) or ))) represent P - 0.05, P - 0.01 or P - 0.001 respectively
compared to chance level Ž25%-dotted line.. The left nonlinear scale is percentage time, being the inverse transform of the right-hand scale which
represents the angular transformation used to normalise the error variance. SuM-CDP, unlike systemic CDP, did not impair correct quadrant visiting. Long
cool-water exposure impaired correct quadrant visiting, relative to short cooling, but short cooling was not different from the saline controls of the first
experiment.
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
105
a significant frequency reduction on the first trial block
when the effect of time in the water would be at its
weakest Ž0.35 Hz reduction in the CDP group, t s 2.65,
df s 17, P s 0.017; initial recording of EEG in three rats
were lost because of computer failure.. Thus, Sum-CDP
appears to produce a qualitatively similar but quantitatively weaker and perhaps briefer reduction in theta frequency to that seen with systemic CDP.
Fig. 3. The numbers of times the rats passed through the previously
correct target area during the probe trial. Details as for Fig. 1. The ), ))
or ))) represent P - 0.05, P - 0.01 or P - 0.001. Systemic-CDP,
SuM-CDP and long cooling all decreased target crossings.
sites where we have previously failed to obtain reductions
in theta frequency and these were excluded from analysis
ŽFig. 5..
The effects of CDP in SuM on spatial learning are
illustrated in Figs. 1–3. The learning of the SuM-CDP
group was as good as that of the SuM-saline group until
about the 5th trial. From this point, the learning of the
SuM-CDP group was impaired compared to the SuM-saline
group while the latter appeared to continue learning as
well as was previously found with the systemic saline
group. Path length revealed a significant separation of
groups across trial blocks Žgroups= trial blocks, linear:
F s 16.95, df s 1r357, P - 0.001; cubic: F s 4.55, df s
1r357, P - 0.05.. There was no significant difference in
swimming speed between the two groups.
During the 60 s probe trial, a bias in favour of the target
quadrant compared to chance level was shown in both the
SuM-saline group Ž t s 2.10, df s 8, P - 0.05. and the
SuM-CDP group Ž t s 1.80, df s 10, P s 0.05; two rats
were excluded because they stayed at the same point and
did not swim in the probe trial., and there was no significant difference between them Ž t s 0.21, df s 1r18, P )
0.05.. But there were significantly fewer target crossings
in the SuM-CDP group Ž t s 2.19, df s 18, P - 0.05. which
showed a similar impairment on this measure to the systemic CDP group.
The theta frequency of both the SuM-CDP and SuMsaline groups decreased with trial blocks in the same way
as in the systemic experiment ŽFig. 4.. In the SuM-saline
group, the theta frequency appeared to drop more slowly
over the last four trial blocks as was also seen with
systemic saline. There was an increasing difference in
frequency between the two groups as trial blocks increased
Žtrial block= group, linear: F s 5.73, df s 1r357, P 0.05.. However, this interaction could have resulted from
differences in time in the water rather than an increase in
the effect of the drug per se. Nearly perfect logarithmic
regression functions of frequency against time were obtained in both the saline group Ž r 2 s 0.96. and the CDP
group Ž r 2 s 0.95. and, as can be seen from Fig. 4, the
separation between the regression lines did not increase as
time went by. Post-hoc investigation showed that there was
3.3. Water exposure
The learning of the five rats in the long-cooling group
was clearly impaired. But the learning of the four rats of
the short-cooling group appeared as good as that of the
four rats of the warm water group Žpath length, group:
F s 0.05, df s 1r6, P s 0.83.. The theta frequency in the
short-cooling group was also similar to that of the warm
Fig. 4. Effects of systemic CDP or saline, intracranial CDP or saline, and
water exposure on hippocampal theta frequency over the course of
training. Theta frequency changed as an inverse logarithmic function of
elapsed time Žright-hand panel. and was additionally reduced by systemic
CDP, SuM-CDP and cooling.
106
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
Fig. 5. Placements of cannulae. There were 30 rats, 22 of which were included in the analysis. There were 13 in the SuM-CDP group Žv . and 9 in the
SuM-saline group Ž`.. In all cases they had placements in, or within 500 m m of the border of, SuM. There were 8 rats which were excluded because the
injection site was more than 500 m m from SuM Žm. in positions where we have previously obtained no change in theta frequency.
water group Ž F s 3.40, df s 1r6, P s 0.12.. So we combined these two groups to form a single short-cooling
group and then compared this with the long-cooling group,
and, separately, with the systemic saline control group
ŽFig. 1.. Path length revealed a significant separation
between the short-cooling and long-cooling groups Žgroups:
F s 22.03, df s 1r11, P - 0.001.. But there was no significant group difference between the short-cooling and
saline groups Ž F s 1.19, df s 1r11, P s 0.30.. There was
no significant difference in swimming speed between the
short-cooling and long-cooling groups Ž F s 0.33, df s
1r11, P s 0.57..
During the 60 s probe trial ŽFig. 2., a bias in favour of
the target quadrant was shown in the short-cooling group
Ž t s 5.73, df s 7, P - 0.001. comparing with 25% chance
level, but not in the long-cooling group Ž t s 1.99, df s 4,
P ) 0.05., and there was a significant difference between
them Ž t s 2.63, df s 11, P - 0.05.. There were significantly fewer target crossings in the long-cooling group
Ž t s 4.27, df s 11, P - 0.005, Fig. 3..
The theta frequency of both short-cooling and longcooling groups decreased logarithmically with training time
ŽFig. 4., and there was a significant difference between the
groups Žgroup: F s 9.37, df s 1r11, P s 0.01.. Compared
with the saline group, the long-cooling group showed a
significant reduction of 1.2 Hz Žgroup: F s 23.43, df s
1r8, P s 0.001., and the short-cooling group showed a
significant frequency reduction of 0.5 Hz Žgroup: F s 6.22,
df s 1r11, P s 0.03., However, it should be noted that
the saline group was part of a different experiment and so
may not have been strictly comparable.
4. Discussion
In the present experiments systemic injection of CDP
impaired spatial learning by increasing path length and
escape latency during acquisition. It also decreased visits
to the correct quadrant and decreased target site crossings
in the probe test. All of these changes are consistent with
previous reports w11,13x. The present data also suggested
that systemic CDP decreased swimming speed. This is not
consistent with McNaughton and Morris w13x who found
increased speed with CDP, but they gave only 1 or 4 trials
rather than 20 trials per day. But it is not inconsistent with
McNamara and Skelton w11x. They did not obtain a significant difference in speed between their groups; however,
their data did show a trend to a decrease in speed in the
CDP group. Like us, they gave all their training trials in
one day and it seems likely that the number of trials per
day is interacting with the drug to determine swimming
speed. The various changes in speed do not create a
problem for the conclusion that CDP impairs spatial learning since clear impairments were obtained with path length,
quadrant and target scores Žnone of which would be expected to be affected by changes in speed. and the same
impairment was found in all three studies.
The very fast learning in our saline animals raises the
question of whether the platform was completely invisible.
There are two reasons for believing that it was invisible.
First, the rats showed significant quadrant preference and
high target crossing scores in the probe test. Second, CDP
impaired learning in a similar way and to a very similar
extent to its effect in previous spatial learning studies and
CDP has not been reported to affect simple cue learning.
The most likely source of the unusually fast learning is our
lack of intertrial interval. In order to minimize time after
injection, we allowed rats to start the next trial immediately after the 15 s platform period. This may have enhanced learning by decreasing the load on working memory.
Previous studies have shown that systemic CDP decreases the frequency of theta rhythm elicited by stimulating the reticular formation w14x and of spontaneous theta
during open field behaviour w19x. The present results
demonstrate that systemic CDP also decreases the frequency of spontaneous theta in the water maze and that it
does so to a similar extent Ž5 mgrkg producing about a 1
Hz drop in frequency.. We also found that theta frequency
decreased logarithmically with training time in both the
CDP and the control rats. The most likely reason for this
latter result is that the temperature of the rat’s body in the
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
cool water decreased with training time and that this
cooling then reduced the theta frequency. It has been
shown previously that theta frequency is related to body
temperature w26x. Our water exposure results were consistent with this and in every group in our experiment theta
frequency was a nearly perfect logarithmic function of
elapsed time.
It is possible that our water-exposure had other effects
than cooling the rats since, for example, they received
experience of inescapable water. However, it should be
noted that the bucket in which they were exposed is
spatially unlike the maze; they remained in the bucket for
only 15 s at any one time Žwith most of the cooling being
evaporative. and 8 exposures to warm water had similar
effects to 4 exposures to cold water and was different from
8 exposures to cold water. Cooling, therefore, appears to
be the most likely explanation of the observed effects.
The reduction in frequency produced by systemic CDP
summated with that occurring as a result of water exposure. Our results show, therefore, that systemic CDP impairs spatial learning and decreases theta frequency simultaneously. This is consistent with our original hypothesis
that CDP affects learning by disrupting hippocampal function, but the evidence is, of course, purely correlational.
CDP injection into SuM also both impaired spatial
learning and decreased theta frequency, but it did so to a
much lesser extent and with a somewhat different pattern
of results from systemic CDP. SuM-CDP did not affect
initial learning and also, in the probe test, did not affect
correct quadrant times. However, SuM-CDP did produce a
deficit during the later parts of acquisition and a decrease
in correct target crossings in the probe test which was
similar in extent to systemic CDP. This is consistent with
previous reports that specific hippocampal cell lesions w25x,
fimbria-fornix lesions w24x or perirhinal cortex lesions w27x
may not affect correct quadrant learning even when they
affect target crossings.
It is important to note that the results do not suggest any
differential action of the drug at different times during the
experiment but rather suggest that SuM-CDP is acting
selectively on specific learning processes. Thus, its lack of
action early in acquisition, when control performance is
poor and is probably focused on finding the correct quadrant as opposed to finding the precise position of the
platform, is mirrored by its lack of action on the quadrant
measure in the probe test at the end of acquisition. Likewise, its effect later in acquisition, when control performance is good and is probably focused on the precise
position of the platform, is mirrored by its action on target
crossings in the probe test.
One explanation of our results, then, is that total hippocampal lesions Žas opposed to cell-specific lesions. produce a total loss of spatial learning; that systemic CDP
produces a major impairment in theta control which is
sufficient to produce a large degradation of spatial learning
but which leaves some degree of localization intact; and
107
that SuM-CDP produces a minor impairment in theta
control which is sufficient only to affect the most finegrained and detailed spatial localization. However,
Whishaw et al. w24x demonstrated that rats with fimbriafornix lesions display a dissociation between ‘getting there’
and ‘knowing where’, which suggests that the impairment
produced by hippocampal dysfunction ‘may be in some
process of motoric control, such as path integration, rather
than in learning the location of the platform in relation to
ambient cues’. If this is so, we would have to rephrase our
explanation in terms of a graded sensitivity of aspects of
‘motoric control’ matching the graded sensitivity to aspects of spatial localization which we have just described.
Whatever the true nature of the deficit, the important
finding is that the impairment of spatial learning produced
by SuM-CDP is similar to that produced by subtotal
disruption of the hippocampus proper and the pathways
which pass through it. This suggests that SuM-CDP affected spatial learning in the water maze by disrupting
hippocampal function. In turn, this suggests that theta
frequency may be quite important since SuM-CDP produced its, admittedly small, behavioral effects while decreasing theta frequency by only about 0.35 Hz.
There are several possible reasons why the effect of
SuM-CDP is relatively small compared to systemic CDP.
First, is the fact that the rats were put into the maze 10 min
after the start of the injection. This meant that the peak
drug effect on theta was obtained on the first trial and
decreased thereafter. This timing was chosen by us, on the
basis of prior electrophysiological experiments, in the mistaken expectation that larger effects on behaviour would be
obtained early in acquisition. Second, is the fact that SuM
is not the only nucleus contributing to the control of theta
frequency w12,21x and it is likely that CDP is affecting
several other sites in addition to SuM to produce its
systemic effect. Third, is the possibility that damage to
SuM or its afferents prior to testing could have reduced the
effect of the drug w12x, but this possibility can almost
certainly be discounted as there is no obvious difference in
frequency between the systemic-saline and SuM-saline
groups of the type we have previously observed with
lesions to SuM.
A final issue to consider is the question of why shortcooling Žincluding both true short-cooling and longer treatment with warmer water., which apparently produced an
equivalent reduction in theta frequency to SuM-CDP did
not produce an equivalent impairment in learning. One
possibility is a difference in the batch of rats since the
saline controls which provide the baseline for comparison
were taken from a different experiment. Another more
interesting possibility derives from Miller’s w15x theory
that theta is important in relation to the timing of information passing around hippocampo-cortical loops. In the case
of SuM-CDP, theta frequency would be changed but not
hippocampo-cortical loop timing and so produce impulses
out of phase with the normal cortical pattern. However,
108
W.-X. Pan, N. McNaughtonr Brain Research 764 (1997) 101–108
with moderate cooling it is possible that the effect on the
reduction in theta frequency would be largely offset by
increased conduction time around the hippocampo-cortical
loops so that any residual phase differences would be
functionally negligible. On this view, the effect of more
extensive cooling would be such that the change in theta
frequency and the change in conduction time would be
sufficiently mismatched that dysfunction would occur.
Overall, then, our results show that injections of CDP
into SuM Ža nucleus known to contain cells which fire
rhythmically in phase with theta and continue to do so
even when theta itself is blocked more rostrally w8x. reduce
the frequency of theta to a small extent and impair spatial
learning to a similarly small extent. They also suggest that
any manipulation which reduces theta frequency to a larger
extent Že.g. systemic CDP, cooling. can produce a similarly larger behavioural effect. Theta frequency, therefore,
appears important for the integrity of hippocampal function. In our experiments, impairments appear when the
frequency drops below about 6.5 Hz and so it is not clear
whether it is the change in theta frequency, or whether it
is the loss of high frequency theta, that is crucial. Our
results also suggest that SuM is no more the only nucleus
involved in the behavioural effects of CDP than it is the
only nucleus involved in the control of theta frequency
w12,21x.
Acknowledgements
This research was funded by a grant from Lottery
Health Research ŽNew Zealand.; chlordiazepoxide was
kindly donated by F. Hoffman-La Roche Ltd.
References
w1x M. Ammessari-Teule, C. Maho, S. Sara, Clonidine reverses spatial
learning and reinstates theta frequencies in rats with partial fornix
section, Behav. Brain Res. 45 Ž1991. 1–8.
w2x G. Carre, C. Harley, Population spike facillitation in the dentate
gyrus following glutamate to the lateral supramammillary nucleus,
Brain Res. 568 Ž1991. 307–310.
w3x R. Cody, D. Maris, L. Costa, M. Grady, Single-day testing in the
Morris water maze, Soc. Neurosci. Abstr. 21 Ž1995. 1731.
w4x P. Dutar, M. Bassant, M. Senut, Y. Lamour, The septohippocampal
pathway: structure and function of a central cholinergic system,
Physiol. Rev. 75 Ž1995. 393–427.
w5x B. Givens, D. Olton, Cholinergic and GABAergic modulation of
medial septal area: effect on working memory, Behav. Neurosci. 104
Ž1990. 849–855.
w6x B. Givens, D. Olton, Local modulation of basal forebrain: effects on
working and reference memory, J. Neurosci. 14 Ž1994. 3578–3587.
w7x L. Haglund, L. Swanson, C. Kohler, The projection of the supramammillary nucleus to the hippocampal formation: an immunohistochemical and anterograde transport study with the lectin PHA-L in
the rat, J. Comp. Neurol. 229 Ž1984. 171–185.
w8x I. Kirk, N. McNaughton, Supramammillary cell firing and hippocampal rhythmical slow activity, NeuroReport 2 Ž1991. 723–725.
w9x I. Kirk, N. McNaughton, Mapping the differential effects of procaine
on frequency and amplitude of reticularly elicited hippocampal
rhythmical slow activity, Hippocampus 3 Ž1993. 517–526.
w10x P. Kraemer, R. Brown, S. Baldwin, S. Scheff, Validation of singleday Morris water maze procedure used to assess cognitive deficits
associated with brain damage, Brain Res. Bull. 39 Ž1996. 17–22.
w11x R. McNamara, R. Skelton, Effects of intracranal infusions of chlordiazepoxide on spatial learning in the Morris water maze
I. Neuroanatomical specificity, Behav. Brain Res. 59 Ž1993. 175–
191.
w12x N. McNaughton, B. Logan, K. Panickar, I. Kirk, W. Pan, N. Brown,
A. Heenan, Contribution of synapses in the medial supramammillary
nucleus to the frequency of hippocampal theta rhythm in freely
moving rats, Hippocampus 5 Ž1995. 534–545.
w13x N. McNaughton, R. Morris, Chlordiazepoxide, an anxiolytic benzodiazepine, impairs place navigation in rats, Behav. Brain Res. 24
Ž1987. 39–46.
w14x N. McNaughton, J. Richardson, C. Gore, Reticular elicitation of
hippocampal slow waves: common effects of some anxiolytic drugs,
Neuroscience 19 Ž1986. 899–903.
w15x R. Miller, Cortico-Hippocampal Interplay and the Representation of
Contexts in the Brain, Springer-Verlag, 1991.
w16x S. Mizumori, B. McNaughton, C. Barnes, A comparison of supramammillary and medial septal influences on hippocampal field
potentials and single-unit activity, J. Neurosci. 61 Ž1989. 15–31.
w17x R. Morris, P. Garrud, J. Rawlins, J. O’Keefe, Place navigation
impaired in rats with hippocampal lesions, Nature 297 Ž1982. 681–
683.
w18x A. Nagahara, T. Otto, M. Gallagher, Entorhinal-perirhinal lesions
impair performance of rats on two versions of place learning in the
Morris water maze, Behav. Neurosci. 109 Ž1995. 3–9.
w19x W.-X. Pan and N. McNaughton, The effects of injection of procaine
and chlordiazepoxide into the supramammillary nucleus on the
frequency of hippocampal theta rhythm in free moving rats. Abstract
from the Thirteenth International Australasian Winter Conference on
Brain Research, Int. J. Neurosci. Ž1996. in press.
w20x V. Sziklas, M. Petrides, F. Leri, The effects of lesions to the
mammillary region and the hippocampus on conditional associative
learning by rats, Eur. J. Neurosci. 8 Ž1996. 106–115.
w21x J. Thinschmidt, G. Kinney, B. Kocsis, The supramammillary nucleus: is it necessary for the mediation of hippocampal theta rhythm?,
Neuroscience 67 Ž1995. 301–312.
w22x R. Vertes, Brainstem afferents to the basal forebrain in the rat,
Neuroscience 24 Ž1988. 907–935.
w23x R. Vertes, PHA-L analysis of projections from the supramammillary
nucleus in the rat, J. Comp. Neurol. 326 Ž1992. 595–622.
w24x I. Whishaw, J.-C. Cassel, L. Jarrard, Rats with fimbria-fornix lesions
display a place response in a swimming pool: a dissociation between
getting there and knowing where, J. Neurosci. 15 Ž1995. 5779–5788.
w25x I. Whishaw, L. Jarrard, Similarities vs. differences in place learning
and circadian activity in rats after fimbria-fornix section or ibotenate
removal of hippocampal cells, Hippocampus 5 Ž1995. 595–604.
w26x I. Whishaw, C. Vanderwolf, Hippocampal EEG and behavior: effects of variation in body temperature and relation of EEG to
vibrissae movements, swimming, and shivering, Physiol. Behav. 6
Ž1971. 391–397.
w27x K. Wiig, D. Bilkey, The effects of perirhinal cortical lesions on
spatial reference memory in the rat, Behav. Brain Res. 63 Ž1994.
101–109.
w28x G. Paxinos, C. Watson, The Rat Brain in Stereotaxic Coordinates,
2nd edn., Academic Press, Australia, 1986.