Expression mechanisms of long-term potentiation in the hippocampus

J Ph_vsiolog~ (Pm+)

I 1096) 90. 299-303

OElsevier, Paris

Expression mechanisms

of long-term potentiation in the hippocampus

JTR Isaaca, SHR Olie@, GO Hjelmstadb, RA Nicollc3d, RC Malenkaa,c* “Drpar/n~~nt qf Psychiatry. md Molecular

oj’ Cellular hPro,qram in Nertroscience. CDeparttnent of Phmio1og.v. md dDepartmm Phatmacolog?: Utliversitx of Cal@nia. San Francisco. CA 94143-0984. USA

Summary - We have taken a number of different experimental approaches to address whether long-term potentiation (LIP) in hippocampal CA1 pyramidal cells is due primarily to presynaptic or postsynaptic modifications. Examination of miniature EPSCs or EPSCs evoked using minimal stimulation indicate that quanta1 size increasing during LIT The conversion of silent to functional synapses may contribute to the LTP-induced changes in mEPSC frequency and failure rate that previously have been attributed to an increase in the probability if transmitter release. long-term potentiation (LTF’) / pairing-induced potentiation I probability AMPA receptor I silent synapses I miniature synaptic current (mEPSC)

Despite extensive investigation, it remains unclear whether presynaptic and/or postsynaptic modifications are primarily responsible for the expression of NMDA receptor-dependent long-term potentiation (LTP) in the CA1 region of the hippocampus. One approach that has been used by several laboratories involves the technique of minimal stimulation in which the stimulus is reduced to a level so that only a single or a few fibers are activated. In most of these studies (Malinow and Tsien, 1990; Kullmann and Nicoll, 1992; Larkman et al, 1992; Liao et al, 1992; Stevens and Wang, 1994; Bolshakov and Siegelbaum, 1995) LTP was associated with a decrease in the incidence of so-called failures, a result that is classically attributed to an increase in the probability of transmitter release (Pr). Evidence was also presented that an increase in quanta1 size accompanied LTP (Kullmann and Nicoll, 1992; Larkman et al, 1992; Liao et nl, 1992), a result which, in contrast, is consistent with a postsynaptic modification. Recently, experiments examining the coefficient of variation (CV) of AMPAR- and NMDAR-mediated EPSCs has produced evidence for an alternative explanation for the change in failures associated with LTP (Kullmann, 1994): silent synapses that lack functional AMPA receptors may exist and be converted to functional synapses following the induction of LTP. Such a scenario provides a postsynaptic mechanism that can explain almost all of the electrophysiological changes observed during LTP. We attempted to directly detect silent synapses by using

*Correspondence

and reprints

of transmitter

release I NMDA receptor I

minimal stimulation and whole-cell recordings (Isaac et al, 1995) (fig 1). We reduced stimulation intensity until no EPSCs were detected at a membrane potential of -60 mV for at least 100 trials. Then we depolarized the membrane to between +30 and +60 mV and in 17 of 42 cells detected EPSCs that could be completely blocked by the NMDAR antagonist DAPV. On returning to a holding potential of -60 mV there were still no detectable EPSCs confirming that we had isolated responses from synapses that expressed functional NMDARs only. In a further set of experiments we attempted to induce LTP at silent synapses. After reducing the stimulus intensity until no responses were observed for 100 trials at -60 mV, we depolarized cells to -10 mV while continuing stimulation. In approximately 40% of our cells (n = 10) this pairing protocol caused AMPAR EPSCs to appear when we returned the cell to -60 mV. In the other cells no NMDAR EPSCs were detected indicating that no synapses were being activated. This form of pairing-induced LTP was blocked by D-APV (n = 7). These data suggest that silent synapses are converted to a functional form during LTP possibly by the insertion and/or uncovering of AMPA receptors in the postsynaptic membrane. Thus a postsynaptic increase in the number of functional synapses could contribute to the decrease in the failure rate commonly observed during LTP. Another approach that has proved powerful in other systems is the examination of miniature synaptic currents. However, it has been difficult to use this approach when studying LTP in slice preparations because with standard protocols only a very small fraction of the synapses generating miniature currents are modified and express LTI? Recently, by taking advantage of the effects of strontium (St-*‘) on synaptic transmission, we (Oliet ef al, 1996) have developed a technique that allows us to examine the

300

+30mV

-60mV s

-6OmV

20-

8

C1 -6OmV, higher stimulus

Stimulus -6OmV

Number

+30 mV

+30 mV + D-APV

-60 mV (after +30 mV)

-1

L

5ms

5ms

5ms

*r 20 ms

T

5ms

WI 20 ms

20 Ills

Fig 1. Example of an experiment demonstrating the existence of silent synapses (adapted from Isaac rr crl. 1995). A. Graph of an experiment to illustrate the standard protocol. The cell was held at 40 mV and after obtaining small EPSCs, the stimulus intensity was reduced until no EPSCs were detected for 100 consecutive trials. The cell was depolarized to +30 mV. and stimulation now evoked responses that were completely blocked by D-APV (25 pM) indicating that they were NMDAR EPSCs. On returning to -60 mV again no EPSCs were detected as confirmed by the lack of effect of the addition of CNQX (IO pM) applied at the end of the experiment. B. EPSC recorded at the beginning of the experiment (average of 10 traces). C. Examples of eight superimposed traces (Ct) or the average of 100 traces (C?) taken at the indicated times during the experiment.

miniature excitatory postsynaptic currents (mEPSCs) that originate specifically from the subset of synapses that are expressing LTP. When Sr’+ is substituted for Ca’+, stimulation evoked synchronous release of transmitter is reduced but asynchronous release of quanta is markedly and selectively enhanced. This permits detailed analysis of quantal events from the subset of synapses being stimulated. The amplitude distribution of the asynchronous events evoked in Sr’+ was not different from that of mEPSCs recorded from the same cells in tetrodotoxin indicating that the former are indeed quantal. Moreover, the amplitude distribution of asynchronous mEPSCs generated by stimulation of one input was the same as that of the

asynchronous events originating from an independent input onto the same cell. These findings permitted a comparison of the mEPSCs generated from stimulation of a naive input with those originating from synapses that had been potentiated (fig 2). The amplitude distribution of mEPSCs collected from potentiated synapses was shifted to the right when compared with the control quantal responses and the mean mEPSC size was larger (n = 6) (fig 2A, B). In contrast, LTD was associated with a shift to the left in the amplitude distribution and a smaller quantal size (n = 7) (fig 2C, D). LTP and LTD also caused an increase and decrease, respectively, in the frequency of asynchronous mEPSCs and

301

collect mEPSCs from the same set of synapses that had previously expressed LTP. We found that the increase in mEPSC size and frequency that were associated with LTP were completely reversed following the depotentiation. This examination of mEPSCs suggests that LTP is associated with an increase in quanta1 size, a result that is most easily explained by a modification in the number and/or properties of postsynaptic glutamate receptors. The increase in the frequency of these events may be due

all the observed changes were completely blocked by the NMDA receptor antagonist, D-APV. In a final set of experiments, we determined if in the same set of synapses, the increase in quantal size that accompanies LTP could be reversed. Shortly after the induction of LTP, Sr2’ was applied and mEPSCs were collected. We then returned to a normal Ca’+-containing medium and applied prolonged 1 Hz stimulation to obtain maximal depotentiation. Re-application of the Srz’-containing medium then allowed us to again

A

a

B

b

-

1 lOOpA 50 ms

-If-

I

0”

I

0

I

40

0

Time (;?I)

b

I

a

scaled

%----

-k

C

LTP

control

I

c

D

I

1 2 Normalized amplitude

control

LTD

3

scaled

1 _l________ 1 Hz

1.5 r

0.0

11

I

\/ _I4 PA 10ms

‘-Cl.Oa B 2 zo.5 .z m I

I

0

I

I

40 Time (t%~)

2 (3 o.o+ 0

1 2 Normalized amplitude

Fig 2. LTP and LTD are associated with changes in quanta1 size (adapted from Oliet et al, 1996). A. Summary of six whole-cell recordings obtained from guinea-pig neurons where synaptic stimulation was paired with membrane depolarization in one pathway (filled symbols). The sample records are superimposed averages of six successive sweeps taken before (a), after (b) LTP induction. and in the presence of S?+ (c). B. Corresponding cumulative amplitude distributions of the events associated with Sr”+-induced asynchronous release obtained from the paired (solid line) and unpaired (dotted line) pathways. The distributions are statistically different (P < 0.0001). Averages of 100 of these events obtained in a single cell are shown above the plot. C. Summary graph of seven whole-cell recordings obtained from rat neurons where depression of one pathway (tilled symbols) was induced by low frequency stimulation (I Hr. 6 min. black bar). The sample records are superimposed averages of six successive sweeps taken before (a), after (b) LTD induction, and in the presence of Sr*+ (c). D. Corresponding cumulative amplitude distributions of the events associated with Sr’+-induced asynchronous release obtained from the depressed (solid line) and control (dotted line) pathways. The distributions are statistically different (P < 0.0001). Averages of 100 of these events obtained in a single cell are shown above the plot.

302

A

0.4 2 0.3 h 0.2 ‘5 r;s -

5

1z ..4

0

3

A

0.32 V

0.26 V-

: 0.28 V-

zz 20

0

40 60 80 100 Stimulus Number

g 100 2 75 8 50 ; 25 8 0

0.34 V

I 0.30 V

25 msec

0

0.26

0.28

0.30

Stimulus

B

0.32

Intensity

0.34 (V)

Pairing

-

_5-

G

-10 -

.

8,: “;‘.:*&: ’ . ..: t.%i..? cv .r.: 8. .*. . A: : ‘... . l_ .** l

$ -15 E -20 -

l

-30 -

h

-35

. . . lI . l.. . .‘< . .* . . \ . . . .* . :* .* : .. .* .‘“0 ,$I %O. -* . . :- ‘*.;: . .- * .t** -0.. l ‘.f. l *.f , 8 . .* ._ 1 _.* . .* . ‘:. 0. .. . .. l . . . l ‘*. . . 8.. . . .H.0 . . ....’ .* a’*, . s 2. .* . l.: . . l a. :. ? . ‘. * . . .-. . . . .* . .* . . t . . :* . I I I I I I l

l

l

l

l

l

l

I

I

42

48

25 msec

D 24 -

20

-35

Baseline LTP

-30

-25

l

l

l

36

.

.

l

l

-25 -

8

.

l

l

Q

120

-15 -10 -20 Amplitude (PA)

-5

0

l

l

l

-

303

Fig 3. Example of LTP which was monitored with perforated patch recording and single axon stimulation and which was associated with t_

an increase in potency and no decrease in failure rate (all data in this figure are from one cell, adapted from Isaac et al, 1996). A. Single axon stimulation test. AI, individual (0) and mean (0; 25 responses) EPSC amplitudes for each stimulus intensity during stepwise increases in stimulus intensity (solid line). The stimulus intensity that elicited only two events (0.28 V) was likely right at threshold and therefore was activating the axon inconsistently. AZ, success rate as function of stimulus intensity. As, averaged EPSCs (25 responses) for each stimulus intensity. B. Individual response amplitudes during the course of an experiment. Time 0 (not shown) was the time at which a IO CR seal was established. CI (left panel), average of responses (II = 100) during the baseline. CI (right panel), nine superimposed consecutive traces from the baseline. CZ (left panel), average of responses (n = 100) during LTP (IO min after pairing). CZ (right panel), nine superimposed consecutive responses during LTP. Cx. superimposed averages of successes only during baseline (smaller trace) and LTP (larger trace; same epochs as used for averages in Ci and C? ). D. Amplitude histograms (bin width = 0.5 PA) of all baseline data (thin line). and all LTP data from 5 min after the end of pairing (thick line).

to modulation of transmitter release mechanisms or to the conversion of silent synapses to functional ones (Isaac er al, 1995; Liao ef nl, 1995). Another approach for studying the site of expression of LTP is to use minimal stimulation to activate a single fiber and thus maximize the likelihood that responses from only a single synapse are being monitored (Raastad, 1995). Such an approach by other investigators has led to the conclusion that LTP is entirety due to an increase in the probability of transmitter release (Pr) (Stevens and Wang, 1994; Bolshakov and Siegelbaum, 1995). We have taken a similar approach using perforated patch-clamp recordings (Isaac et al, 1996) (fig 3). The likelihood of reliable activation of a single axon and a presumptive single synapse was maximized by using stimulus intensity ramps in which EPSCs exhibited an abrupt threshold and a plateau in amplitude, and failure rate for at least two further increases in stimulus intensity (fig 3 AI-A& In all cells (n = 8) LTP was found to be associated with an increase in the potency (Stevens and Wang, 1994), defined as the mean success amplitude (fig 3B-D). There was also a decrease in failure rate in four out of eight cells. In a final set of experiments we attempted to induce LTP in the presence of high extracellular Ca” (5 mM), which raised Pr to 0.84 f 0.04 (n = 5) (baseline Pr in 2.5 mM Ca” was 0.44 + 0.05, n = 8). Under these conditions we could reliably induce LTP, which was again accompanied by an increase in potency. In agreement with the studies using Si’ to examine mBPSCs, these data are consistent with an increase in quantal size contributing to LTP In addition, an increase in the number of functional synapses could explain the decrease in failures rate that was sometimes observed.

However, in no cell could the changes observed with LTP be explained by an increase in Pr alone.

References Bolshakov VY, Siegelbaum SA (1995) Regulation of hippocampal transmitter release during development and long-term potentiation. Science 269, 173&1734 Isaac JTR, Nicoll RA, Malenka RC (1995) Evidence for silent synapses: Implications for the expression of LTI? Neuron 15,427-434 Isaac JTR, Hielmstad GO. Nicoll RA. Malenka RC (1996) Loneterm potentiation at single liber inputs to hippocampal CA1 piramidal cells. Proc NatI Acud Sci USA 93, 8710-8715 Kullmann DM (1994) Amplitude fluctuations of dual-component EPSCs in hippocampal pyramidal cells: implications for longterm potentiation. Neuron 12, 1111-I120 Kullmann DM, Nicoll RA (1992) Long-term potentiation is associated with increases in quanta1 content and quanta1 amplitude. Nature 357. 240-244 Larkman A. Hannay T, Stratford K, Jack J (1992) Presynaptic release probability influences the locus of long-term potentiation. Nature 360. 70-73 Liao D, Jones A, Malinow R (1992) Direct measurements of quanta1 changes underlying long-term potentiation in CA1 hippocampus. Neuron 9, 1089-1097 Liao D. Hessler NA, Malinow R (1995) Activation of postsynaptitally silent synapses during pairing-induced LTP in CA 1 region of hippocampal slice. Nature 375, 4oo-404 Malinow R, Tsien RW (1990) Presynaptic enhancement shown by whole-cell recording of long-term potentiation in hippocampal slices. Nature 346. 177-180 Oliet SHR. Malenka RC, Nicoll RA (1996) Bidirectional control of quanta1 size by synaptic activity in the hippocampus. Science 271. 1294-1297 Raastad M (199.5) Extracellular activation of unitary excitatory synapses between hippocampal CA3 and CA 1pyramidal cells. Ettr J Neurosci 7, 1882-1888 Stevens CF. Wang Y (1994) Changes in reliability of synaptic function as a mechanism for plasticity. Nature 371, 704707