Network stability through homeostatic scaling of excitatory and inhibitory synapses following inactivity in CA3 of rat organotypic hippocampal slice cultures Lucy E. Buckby, 1 Thomas P. Jensen, 1 Paul J.E. Smith, and Ruth M. Empson * School of Biological Sciences, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK Received 4 October 2005; revised 6 December 2005; accepted 11 January 2006 Available online 24 February 2006 Homeostatic plasticity is a phenomenon whereby synaptic strength is scaled in the context of the activity that the network receives. Here, we have analysed excitatory and inhibitory synapses in a model of homeostatic plasticity where rat organotypic hippocampal slice cultures were deprived of excitatory synaptic input by the NMDA and AMPA/KA glutamate receptor antagonists, AP5 and CNQX. We show that chronic excitatory synapse deprivation generates an excitable CA3 network where enhanced amplitude and frequency of spontaneous excitatory post-synaptic potentials were associated with increased glutamate receptor subunit expression and increased number and size of synapsin 1 and VGLUT1 positive puncta. Intact spontaneous inhibitory post-synaptic potentials coincided with persis- tent expression of the GABA-A receptor alpha subunit and GAD65 and an enhancement of parvalbumin-positive puncta. In this model of homeostatic plasticity, scaling up of synaptic excitation and mainte- nance of fast synaptic inhibition promote an excitable, but stable, CA3 network. D 2006 Elsevier Inc. All rights reserved. Keywords: Hippocampus; Synapsin 1; Glutamate; VGLUT1; Synapsin 1; GABA; GABA-A receptor; NR1; GluR1; Homeostatic plasticity; Homeo- static scaling; EPSP; IPSP Introduction Over the past decade, there has been a transformation in our understanding of activity-dependent synaptic plasticity especially regarding the development and refinement of synaptic connections. Throughout life, axonal arbours and synaptic connections are continuously reshaped, refined and reconfigured by activity- dependent mechanisms (Goodman and Shatz, 1993; Munno and Syed, 2003). These activity-dependent mechanisms can be cate- gorised into Hebbian plasticity including long-term potentiation and depression (LTP and LTD) and homeostatic plasticity (Turrigiano et al., 1998; Davis and Bezprozvanny, 2001; Burrone and Murthy, 2003). Hebbian and homeostatic plasticity are complementary yet contrasting. Whilst Hebbian plasticity can act to destabilise neuronal networks in either a positive or negative direction, homeostatic plasticity promotes network stabilisation by returning synaptic strength and neuronal firing to within a set range (Turrigiano, 1999; Turrigiano and Nelson, 2004; Burrone and Murthy, 2003). The concept of physiological homeostasis was originally championed by Cannon in 1932 (Cannon, 1939). However, the idea that neuronal networks could undergo the same form of adaptation was only recently suggested (Davis and Goodman, 1998; Turrigiano et al., 1998). This may be because up until recently, the idea seemed contradictory, how can homeostatic regulation of neuronal activity also permit the changes in synaptic strength required by Hebbian plasticity and learning ? Rather, the view is that instead of returning neuronal activity to a fixed level, homeostatic plasticity acts like a Fnet_ ensuring that neuronal networks do not reach the extremes of their activity. This is particularly important for highly epileptogenic structures such as the hippocampus and cortex. The classical way to model homeostatic plasticity is to subject neurones to a period of long-term inactivity or overactivity. Previous studies examining homeostatic plasticity in a variety of neuronal preparations have reported increases in neuronal excitation entirely consistent with the network acting to correct the inactivity by upregulating synaptic excitation. This has been measured as an increase in the amplitude and frequency of spontaneous miniature excitatory post-synaptic currents (mEPSCs) indicating direct changes in the properties of the synapse, increases in glutamate receptor expression (O’Brien et al., 1998; Turrigiano et al., 1998; Watt et al., 2000; Bacci et al., 2001; Burrone et al., 2002; Kilman et al., 2002; Galvan et al., 2003) and increases in synapse size, including the pre-synaptic active zone and number of vesicles (Murthy et al., 2001; De Gois et al., 2005). Although homeostatic plasticity has been examined in dispersed primary cortical and hippocampal cultures, there are fewer examples 1044-7431/$ - see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.mcn.2006.01.009 * Corresponding author. Fax: +44 1784 434326. E-mail address: [email protected](R.M. Empson). 1 These authors contributed equally to this work. Available online on ScienceDirect (www.sciencedirect.com). www.elsevier.com/locate/ymcne Mol. Cell. Neurosci. 31 (2006) 805 – 816
12
Embed
Network stability through homeostatic scaling of excitatory and inhibitory synapses following inactivity in CA3 of rat organotypic hippocampal slice cultures
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/ymcne
Mol. Cell. Neurosci. 31 (2006) 805 – 816
Network stability through homeostatic scaling of excitatory and
inhibitory synapses following inactivity in CA3 of rat organotypic
hippocampal slice cultures
Lucy E. Buckby,1 Thomas P. Jensen,1 Paul J.E. Smith, and Ruth M. Empson*
School of Biological Sciences, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK
Received 4 October 2005; revised 6 December 2005; accepted 11 January 2006
Available online 24 February 2006
Homeostatic plasticity is a phenomenon whereby synaptic strength is
scaled in the context of the activity that the network receives. Here, we
have analysed excitatory and inhibitory synapses in a model of
homeostatic plasticity where rat organotypic hippocampal slice
cultures were deprived of excitatory synaptic input by the NMDA
and AMPA/KA glutamate receptor antagonists, AP5 and CNQX. We
show that chronic excitatory synapse deprivation generates an
excitable CA3 network where enhanced amplitude and frequency of
spontaneous excitatory post-synaptic potentials were associated with
increased glutamate receptor subunit expression and increased
number and size of synapsin 1 and VGLUT1 positive puncta. Intact
spontaneous inhibitory post-synaptic potentials coincided with persis-
tent expression of the GABA-A receptor alpha subunit and GAD65
and an enhancement of parvalbumin-positive puncta. In this model of
homeostatic plasticity, scaling up of synaptic excitation and mainte-
nance of fast synaptic inhibition promote an excitable, but stable, CA3
applied in CA3c/d, stimulus–response curves for fEPSP peak
amplitudes (Fig. 1B) and duration (Fig. 1C) from AP5/CNQX-
treated organotypic hippocampal slice cultures were shifted to the
left of those from control slice cultures, indicating an increase in the
excitability of the CA3 network, although as seen in Fig. 1B, a
graded response could still be observed from the more excitable,
treated slice cultures. The fEPSP peak amplitude stimulus–response
curve of AP5/CNQX-treated slice cultures (Fig. 1B) had a steeper
gradient between 0 and 10 V, 8.2 mV/V compared to 3.8 mV/V for
control slice cultures; a higher normalised fEPSP peak amplitude at
10 V, 81.5 T 3.9% of maximum (n = 14) compared to 38.1 T 4.8% for
control slice cultures (P < 0.001, t test, n = 10, Fig. 1B) and a lower
50% stimulation value (the stimulus strength required to evoke the
half-maximal fEPSP peak amplitude) 5.3 V, a third of the
stimulation required in control slice cultures, 15 V. The mean
duration of the evoked fEPSPs in AP5/CNQX-treated slice cultures
was also increased at all stimulus strengths (P < 0.001, t test, n =
10). As seen in Figs. 1A and C, the duration of fEPSPs evoked in
AP5/CNQX-treated slice cultures was increased compared with
control responses, from 27 T 4 ms and 42 T 10 ms (n = 10), at 10 V
and 40 V, respectively, in control slice cultures, to 178 T 31 ms and
178 T 21 ms (n = 14), respectively, in AP5/CNQX-treated slice
cultures. We also examined the fEPSPs in control slice cultures for
their mossy fibre component using the metabotropic glutamate
receptor type II agonist DCGIV (5 AM). Application of DCGIV
reduced the peak amplitude of evoked fEPSPs by 66.9 T 4.1% (P <
0.0001, n = 16, t test), indicating that the evoked CA3 fEPSPs were
approximately 65% mossy fibre-derived.
Despite the clear increase in the amplitude and duration of the
evoked fEPSP responses in the AP5/CNQX-treated slice cultures,
we never observed any spontaneous epileptiform activity. All
indications from these electrophysiology experiments indicated
that despite chronic incubation with glutamate receptor antago-
nists, the responses from the AP5/CNQX-treated slice cultures
were excitable but stable. Further parallel experiments showed
that there was no significant loss or gain of neurones or glia, or
any changes in slice culture thickness in the AP5/CNQX-treated
slice cultures compared with control. Estimations of the area of
living (calcein-stained) regions in both CA3 and CA1 revealed
no significant differences between control and AP5/CNQX-
treated slice cultures at 2, 4 and 6 DIV (P = 0.23, ANOVA, n = 4
slice cultures for each time point and condition), consistent with
no detectable changes in slice culture thickness (P = 0.1, ANOVA,
n = 3 slice cultures for each time point and condition). Neither
was there any change in the intensity or area of dead (propidium
iodide-stained) neurones at all ages (P = 0.99, P = 0.85,
respectively, ANOVA, n = 4 slice cultures for each time point
and condition) in slice cultures from AP5/CNQX-treated and
control conditions, nor in the expression of the glial cell marker
protein GFAP within the CA3 cell layer of control and AP5/
CNQX-treated slice cultures (P = 0.37, t test, n = 13). All these
results confirmed that the observed network excitability in the
AP5/CNQX-treated slice cultures was unlikely to be a result of
the loss or gain of neurones or glial cells. We also tested the
reversibility of the alterations in excitability within our chronic
AP5 and CNQX model. If we treated slice cultures for only 1–2
days with AP5 and CNQX (instead of 7 days), we observed a
similar and also reversible enhancement to the excitability, as
described above. In contrast, after 7 days treatment with AP5 and
CNQX, the alterations to the excitability were not easily
reversed.
The amplitude and frequency of spontaneous EPSPs were
increased in CA3 pyramidal neurones from AP5/CNQX-treated
organotypic hippocampal slice cultures compared with controls
Whole cell recordings of spontaneous excitatory post-synaptic
potentials (spEPSPs) from pyramidal cells held at hyperpolarised
levels (Fig. 2A, to similar levels across control and AP5/CNQX-
treated cells) revealed a large increase in the amplitude and
frequency of the spEPSPs from AP5/CNQX-treated cells (n = 7)
compared with control (n = 9), across all spEPSPs within the
recorded population, as shown in Fig. 2B (P < 0.0001, ANOVA,
n = 16). We also detected an alteration in the mean frequency of
spEPSPs, as shown in Fig. 2C, where mean values changed from
3.2 T 1.0 spEPSP per second in cells from control slice cultures to
11.8 T 1.9 spEPSPs per second in cells from AP5/CNQX-treated
slice cultures (P < 0.001, t test, n = 16). None of the cells from
either the control slice cultures or the AP5/CNQX-treated slice
cultures demonstrated epileptiform behaviour in the form of
spontaneous bursts of action potentials or synchronous and
spontaneous depolarising shifts, even when depolarised. However,
we did observe a slightly depolarised average resting membrane
potential in the cells from the AP5/CNQX-treated slice cultures.
The mean resting potential in cells from control slice cultures was
�66 T 1 mV compared with�62 T 1 mV in cells from AP5/CNQX-
treated slice cultures (P < 0.05, t test, n = 16). Added to this, the
cells from the AP5/CNQX-treated slice cultures also demonstrated
a slightly enhanced input resistance; mean values were 305 T 24
MV in control cells compared with 368 T 19 MV in cells from AP5/
CNQX-treated slice cultures (P < 0.05, t test, n = 16).
Our electrophysiological recordings revealed excitable but
stable cells and slice cultures following chronic AP5/CNQX
treatment. In order to further explain our results thus far, we
conducted parallel Western blotting experiments to identify any
changes in the expression of excitatory synaptic proteins at both
the pre- and post-synaptic level.
Enhanced expression of the excitatory post-synaptic receptors
GluR1 and NR1 in AP5/CNQX-treated organotypic hippocampal
slice cultures compared with controls in the absence of any change
in expression of post-synaptic protein markers PSD95 and SAP102
As shown in Fig. 3A, we identified an approximately 50%
increase in the expression of the AMPA and NMDA receptor
subunits GluR1 and NR1 in the AP5/CNQX-treated slice cultures.
Mean NR1 expression increased to 136.4 T 13.1% of the control
value (P < 0.05, t test, n = 3) and mean GluR1 expression increased
to 140 T 14.4% of the control value (P < 0.05, t test, n = 3).
Western blot analysis of the post-synaptic proteins SAP102 and
PSD95, however, showed no alteration in their expression in slice
cultures from control or AP5/CNQX-treated slice cultures, as
shown in Fig. 3B. Mean values of PSD95 and SAP102 expression
from control and AP5/CNQX-treated slice cultures remained
similar at 95 T 12% and 94 T 11% of control values, respectively
(P = 0.47, t test, n = 8 and P = 0.57, t test, n = 4, respectively).
Excitatory synapse pre-synaptic marker protein expression was
also enhanced in AP5/CNQX-treated organotypic hippocampal
slice cultures compared with controls
Homeostatic scaling of synaptic strength in a variety of
models involves pre- and/or post-synaptic changes at excitatory
Fig. 2. The amplitude and frequency of spontaneous EPSPs is increased in pyramidal neurones from AP5/CNQX-treated organotypic hippocampal slice
cultures. Panel A shows representative traces from a whole cell recording from a control neurone (upper trace) and a neurone from an AP5/CNQX-treated slice
culture (lower trace) both held slight hyperpolarised from their resting membrane potential by the injection of a small amount of hyperpolarising current.
Spontaneous EPSPs (spEPSPs) are seen as small depolarising deflections (that were sensitive to the AMPA/KA receptor antagonist CNQX and the NMDA
receptor antagonist, AP5, data not shown) and are less frequent and smaller in the cell from the control slice culture. In panel B, the distribution of the number
and amplitude of spEPSPs collected over similar time epochs from several control (white bars, n = 9) and AP5/CNQX-treated (black bars, n = 7) cells shows an
enhancement of the number of both small and large events in the AP5/CNQX-treated cells. This is confirmed in panel C where the average number of EPSPs
detected per second during the same epochs used to generate the data in panel B is increased in AP5/CNQX-treated cells (black bars), compared with the
controls (white bars). **Denotes significance at P < 0.001.
synapses. Although our Western blot experiments identified an
enhancement in post-synaptic glutamate receptor expression,
the enhanced excitability seen with our electrophysiological
recordings could also be a result of enhanced neurotransmitter
release. We therefore chose to analyse the levels of a protein
expressed exclusively at excitatory synapse terminals, the
vesicular glutamate transporter protein VGLUT1 (Takamori et
al., 2000) and synapsin1, a vesicular protein present in all
pre-synapses. As shown in Fig. 4A, we detected an elevation
of VGLUT1 in AP5/CNQX-treated slice cultures compared
with control, where the mean expression increased to 133 T7.6% of the control value (P = 0.01, t test, n = 5). We could
also detect a similar level of enhancement of VGLUT1
expression in slice cultures (n = 4) treated with AP5/CNQX
for 48 h. To compliment this finding, we also detected an
increase in the number of VGLUT1 expressing puncta in
stratum radiatum of CA3 in AP5/CNQX-treated slice cultures,
as shown in Fig. 4B. The mean number of VGLUT1-labelled
puncta increased from 621 T 12.1 in control slice cultures to
673 T 16.2 (P < 0.05, t test, n = 5). Consistent with the
upregulation of VGLUT1 puncta, we also identified an
increase in the number of synapsin1 expressing puncta in
the CA3 cell body layer of AP5/CNQX-treated slice cultures.
Synapsin1 labelling appeared as larger puncta than seen with
VGLUT1 staining and although we detected fewer particles in
each field of view, the detected changes were greater. Mean
values for the number of synapsin1 puncta detected increased
from 141 T 8 in the control slice cultures to 273 T 12
particles in AP5/CNQX-treated slice cultures (P < 0.0001, t
test, n = 12). We also noted a significant change in the size
distribution of synapsin1-labelled particles in AP5/CNQX-
treated slice cultures where we detected more, larger synap-
Fig. 3. AP5/CNQX treatment of organotypic hippocampal slice cultures increases the expression of post-synaptic NMDA and AMPA glutamate receptor
subunits without changing the expression of post-synaptic density marker proteins. Panels A and B show Western blot-based comparison of the expression of
post-synaptic marker proteins in AP5/CNQX-treated and control organotypic hippocampal slice cultures. Slice cultures were probed with primary antibodies
raised against NR1, an NMDA receptor subunit (¨120 kDa), n = 4, GluR1 an AMPA receptor subunit (¨90 kDa), n = 3, in panel A and to SAP102 (102
kDa), n = 4, and PSD95 (95 kDa), n = 4, shown in panel B. All lanes were equally loaded with 15 Ag protein and the position of the standard protein markers
is shown, 116 and 66 kDa. Values representing protein expression for the AP5/CNQX-treated slice cultures (black bars) are obtained from the optical density
of the band compared as a percentage to the 100% control level (white bars) on the same blot. Bars represent mean expression T SEM for the different
proteins. *Denotes significance at P < 0.05. Representative examples of the protein expression from the same Western blots are shown below each bar graph.
by the action of GABA at GABA-A receptors, as evidenced by
their sensitivity to bicuculline (data not shown). We noted that
although cells from control slice cultures showed robust inhibition,
as expected, shown in Fig. 5A, in cells from AP5/CNQX-treated
slice cultures, there was an increase in the mean frequency of
spIPSPs, Fig. 5C. Mean values changed from 2.7 T 0.6 spIPSPs persecond to 5.2 T 0.8 spIPSPs per second (P < 0.05, t test, n = 5) in
control versus AP5/CNQX-treated cells. However, there was no
significant change in the amplitude distribution of the spIPSPs
recorded from control and AP5/CNQX-treated cells, Fig. 5B (P =
0.98, ANOVA, n = 5), although more, larger spIPSP events can be
Fig. 4. AP5/CNQX treatment of organotypic hippocampal slice cultures increases the expression of the pre-synaptic vesicular glutamate transporter protein
VGLUT1. Panel A shows a Western blot-based comparison of the expression of VGLUT1 in AP5/CNQX-treated and control organotypic hippocampal
slice cultures. All lanes were equally loaded with 15 Ag protein and the position of the standard protein markers is shown, 66 and 45 kDa. Values
representing VGLUT1 expression for the AP5/CNQX-treated slice cultures (black bars) are obtained from the optical density of the band compared as a
percentage of the 100% control level (white bars) on the same blot. In panel A, bars represent mean expression T SEM for VGLUT1 under the two
different conditions. *Denotes significance at P < 0.05. A representative example of VGLUT1 expression from the same Western blot is shown below the
bar graph. In panel B, fluorescent immunohistochemical detection of VGLUT1 shows small punctate particles in stratum radiatum of hippocamal CA3 in
images obtained from control (left panel) and AP5/CNQX-treated slice cultures (right panel). Using thresholding techniques, we quantified the number of
VGLUT1-stained particles in each field of view (as described in Experimental methods) and found the number of VGLUT1 particles to be significantly
enhanced in AP5/CNQX-treated slice cultures (black bars) compared with control (white bars). *Denotes significance at P < 0.05.
In order to confirm the presence of fast synaptic inhibition in
AP5/CNQX-treated slice cultures, we also analysed the expression
of proteins exclusively expressed at GABA-ergic synapses. As
shown in Fig. 6A, there was little change in the presumably post-
synaptic expression levels of the alpha 1 subunit of the GABA-A
receptor or in the pre-synaptic expression of the GABA synthetic
enzyme glutamic acid decarboxylase, GAD65, present in GABA-
ergic terminals, in slice cultures from control or AP5/CNQX-treated
slice cultures. Mean values for GABA-A alpha 1 receptor expression
were unchanged at 93.5 T 4.7% of the 100 T 11.7% of the control
values (P = 0.55, t test, n = 4), whilst mean values for GAD65
expression were not significantly changed at 77 T 9.5% of the 100 T14.7% control values (P = 0.24, t test, n = 7), although there was a
tendency towards a decrease as can be seen in Fig. 6A. In light of the
tendency towards a change in the expression of GAD65, we
Fig. 5. Intact synaptic inhibition in AP5/CNQX-treated CA3, seen as frequent spontaneous IPSPs in pyramidal neurones helps stabilise the excitability of
the network. Panel A shows representative traces from a whole cell recording from a control neurone (upper trace) and a neurone from an AP5/CNQX-
treated slice culture (lower trace) both held slightly depolarised from their resting membrane potential by the injection of a small amount of depolarising
current. Spontaneous IPSPs (spIPSPs) are seen as small hyperpolarising deflections (that were sensitive to bicuculline, data not shown) and are less
frequent, though of a similar size, in the cell from the control slice culture. Panel B shows the distribution of the number and amplitude of spIPSPs
collected over similar time epochs from several control (white bars) and AP5/CNQX-treated (black bars) cells where a larger number of larger events can be
seen in the AP5/CNQX-treated cells compared with control. In panel C, we show that the average number of spIPSPs detected per second during the same
epochs used to generate the data in panel B is increased in AP5/CNQX-treated cells (black bars), compared with the controls (white bars). In panel D, the
effects of abolishing fast GABA-A receptor-mediated events are shown by the population fEPSPs recorded from the cell body layer of CA3 in response to
minimal (1–2 V) electrical stimulation to the mossy fibres. In the example from a control slice culture, when fast synaptic GABA-A receptor-mediated
inhibition is abolished, the fEPSP is lengthened and displays 4 clear afterdischarges, as denoted by small downward triangles. In contrast, in the example
from an AP5/CNQX-treated slice culture, the number of afterdischarges is increased to 10, lower trace, indicating the enhanced recurrent excitation within
conducted immunohistochemistry but could not detect any alter-
ations in the number or distribution of GAD65-stained puncta, in the
CA3 cell body layer, between control and AP5/CNQX-treated slice
cultures. Mean values were 263 T 21 in control slice cultures
compared with 248 T 27 in AP5/CNQX-treated slice cultures (P =
0.67, t test, n = 6). Nor could we detect any change in the amplitude
distribution of GAD65 positive puncta (P = 0.95, ANOVA, n = 6). In
contrast, immunohistochemistry for parvalbumin (PV), a classical
marker protein of a subtype of inhibitory terminals (Freund and
Buzsaki, 1996), showed an enhancement of the number of PV
positive puncta, see Fig. 6B. The mean number of PV puncta in a
field of view increased from 94 T 9 in control slice cultures to 136 T14 in AP5/CNQX-treated slice cultures (P < 0.05, t test, n = 9).
Discussion
The present study set out to examine the consequences of
homeostatic plasticity for both synaptic excitation and inhibition
following a period of chronic inactivity within hippocampal
CA3.
Our key findings are that inactivity scales up excitatory synapse
strength and number at both pre- and post-synaptic sites. In
addition, the strength of fast GABA-A receptor-mediated synaptic
inhibition stabilised the excitable, hippocampal CA3 network. Our
findings highlight the intrinsic ability of this network to rebalance
excitation and inhibition during alterations in excitatory synapse
strength.
Fig. 6. Changes in pre- and post-synaptic markers for GABA-ergic inhibitory synapses in AP5/CNQX-treated organotypic hippocampal slice cultures
compared with control. AWestern blot-based comparison of the expression of two markers for GABA-ergic inhibitory synapses shows little difference in their
expression in AP5/CNQX-treated and control organotypic hippocampal slice cultures. Slice cultures were probed with primary antibodies raised against the
GABA-A receptor alpha 1 subunit and to GAD65. All lanes were equally loaded with 15 Ag protein and the position of the standard protein markers is shown,
66 and 45 kDa. Values representing protein expression for the AP5/CNQX-treated slice cultures (black bars) are compared to the 100% control level (white
bars) on the same blot. Bars represent mean expression T SEM for the different proteins. Representative examples of the protein expression from the same
Western blots are shown below each graph. Panel B shows fluorescent immunohistochemical detection of parvalbumin (PV)-positive terminals within the cell
body layer of CA3 in control (left panel) and AP5/CNQX-treated slice cultures (right panel). Using thresholding techniques, we quantified the number of PV-
stained particles in each field of view (as described in Experimental methods) and found the number of PV particles to be significantly enhanced in AP5/
CNQX-treated slice cultures (black bars) compared with control (white bars). *Denotes significance at P < 0.05.