See discussions, stats, and author profiles for this publication at: [Link]

net/publication/14223142

Regulation of Synaptic Efficacy by Coincidence of Postsynaptic APs and EPSPs

Article in Science · February 1997

DOI: 10.1126/science.275.5297.213 · Source: PubMed

CITATIONS READS
3,853 2,229

4 authors, including:

Henry Markram Joachim H. R. Lübke

Swiss Federal Institute of Technology in Lausanne Forschungszentrum Jülich
362 PUBLICATIONS 45,639 CITATIONS 111 PUBLICATIONS 13,139 CITATIONS

SEE PROFILE SEE PROFILE

Michael Frotscher
Hamburg University
454 PUBLICATIONS 34,895 CITATIONS

SEE PROFILE

All content following this page was uploaded by Joachim H. R. Lübke on 15 May 2014.

The user has requested enhancement of the downloaded file.

 REPORTS
London 298, 227 (1982); P. R. Montague and T. J. J. Neurophysiol. 76, 3460 (1996). for helpful discussions and comments on the manu-
Sejnowski, Learn. Mem. 1, 1 (1994). 29. N. Lasser-Ross, H. Miyakawa, V. Lev-Ram, S. R. script. Supported by NIH grants NS09482 (J.C.M.)
27. G. J. Stuart, H. U. Dodt, B. Sakmann, Pflugers Arch. Young, W. N. Ross, J. Neurosci. Methods 36, 253 and NS11535, MH44754, and MH48432 (D.J.).
423, 511 (1993). (1991).
28. J. C. Magee, R. B. Avery, B. R. Christie, D. Johnston, 30. We thank B. Christie, C. Colbert, and P. R. Montague 17 September 1996; accepted 5 November 1996

Regulation of Synaptic Efficacy by Coincidence was further depolarized by current injection

to produce a burst of APs during the EPSPs,
of Postsynaptic APs and EPSPs then a persistent increase (.20%) was ob-
served in 8 of 11 connections (Fig. 1, C and
Henry Markram,* Joachim Lübke, Michael Frotscher, D; 94 6 23% increase) (8, 9).
Bert Sakmann To establish whether the occurrence of
postsynaptic APs during EPSPs was indeed
Activity-driven modifications in synaptic connections between neurons in the neocortex critical for the induction of the increase in
may occur during development and learning. In dual whole-cell voltage recordings from EPSP amplitude, a number of control ex-
pyramidal neurons, the coincidence of postsynaptic action potentials (APs) and unitary periments were performed. Pairing of indi-
excitatory postsynaptic potentials (EPSPs) was found to induce changes in EPSPs. Their vidual postsynaptic APs with EPSPs and
average amplitudes were differentially up- or down-regulated, depending on the precise without a sustained postsynaptic depolariza-
timing of postsynaptic APs relative to EPSPs. These observations suggest that APs tion (Fig. 2A) induced a persistent increase
propagating back into dendrites serve to modify single active synaptic connections, in EPSP amplitudes (38 6 9%; n 5 21; 20
depending on the pattern of electrical activity in the pre- and postsynaptic neurons. Hz; Fig. 2B) that was not associated with
measurable changes in input resistance, cur-
rent–AP discharge relation, or AP thresh-
old. Neither bursts of postsynaptic APs
Repetitive activation of neuronal circuits tudes decreased rapidly and a sufficiently alone nor high-frequency bursts of presyn-
can induce long-term changes in subse- large postsynaptic depolarization was not aptic APs induced persistent changes in
quent responses generated by synapses in reached (3). When the postsynaptic neuron EPSP amplitudes (Fig. 2B). The increase in
many regions of the brain, and such plastic-
ity of synaptic connections is regarded as a
cellular basis for developmental and learn-
ing-related changes in the central nervous A
system (1, 2). The actual triggers for synap-
tic modifications between two neurons are,
however, unclear (3). Postsynaptic APs are
initiated in the axon and then propagate
back into the dendritic arbor of neocortical
pyramidal neurons (4), evoking an activity-
dependent dendritic Ca21 influx (5) that
could be a signal to induce modifications at
the dendritic synapses that were active
around the time of AP initiation. To test
this hypothesis, we made dual whole-cell
voltage recordings from neighboring, thick,
tufted pyramidal neurons in layer 5 of the
neocortex (Fig. 1A) for which the dendritic
locations of synaptic contacts were known
(6, 7), and we investigated whether the
postsynaptic AP could induce changes in
unitary EPSP amplitudes. Fig. 1. Simultaneous pre- and postsynaptic
When depolarizing current was injected activity in synaptically coupled neurons induc-
only into the cell body of a presynaptic es an increase in EPSPs. (A) Camera lucida
50 µm reconstruction of a bidirectionally coupled pair
neuron to evoke a burst of APs, the result-
of thick-tufted layer-5 pyramidal neurons. Pu-
ing high-frequency train of subthreshold
tative synaptic contacts are marked by green
unitary EPSPs (Fig. 1B) failed to trigger dots (from the black neuron; five contacts; mean distance from soma, 95 mm; range, 73 to 126 mm) and
changes in the average EPSP amplitudes blue dots (from the red neuron; six contacts; mean distance from soma, 95 mm; range, 50 to 283 mm).
(Fig. 1D), possibly because EPSP ampli- Thin dotted lines represent axon collaterals (blue is for the cell drawn in red; green is for the cell drawn
in black). An average of 5.5 contacts are made per connection, and more than 80% of contacts are
H. Markram and B. Sakmann, Max-Planck-Institut für within 200 mm of the soma. (B) Characteristic synaptic response. A presynaptic burst of APs (Pre. APs)
Medizinische Forschung, Abteilung Zellphysiologie,
Jahnstraße 29, D-69120 Heidelberg, Germany.
evoked by a 100-ms current pulse (400 pA, cell body injection) evokes EPSPs in the postsynaptic
J. Lübke and M. Frotscher, Anatomisches Institut der neuron. (C) Mean unitary EPSPs before and after pairing. Averages of 75 EPSPs from the onset and after
Albert-Ludwigs Universität Freiburg, Albertstraße 17, 50 min. (D) Synchronization of pre- and postsynaptic activity. Each dot represents the amplitude of a
D-79104 Freiburg, Germany. single, test, AP-evoked EPSP shown as a percent of the average (of 75 responses, 5 min) control EPSP.
* To whom correspondence should be addressed at the Whole-cell recording was established about 3 min before time 0. After 10 min of recording, bursts of
Department of Neurobiology, Weizmann Institute for Sci- EPSPs were evoked 10 times every 20 s (indicated by the bar labeled EPSPs). Test EPSPs were
ence, Rehovot, 76100, Israel. E-mail: bnmark@weizmann. continuously sampled every 4 s in between these bursts. After 20 min of recording, a burst of postsyn-
[Link] aptic APs was evoked during EPSPs (15 times every 20 s; indicated by bar labeled EPSPs and APs).

SCIENCE z VOL. 275 z 10 JANUARY 1997 213

EPSPs was also dependent on the frequency crease in EPSP amplitude, pairing of sus- the absence of AP initiation) with a high-
of AP-EPSP pairs in the train, with a sharp tained depolarization of the postsynaptic frequency train of EPSPs failed to cause a
onset at 10 Hz (Fig. 2C). Although APs neuron to below the threshold for AP ini- significant change in EPSP amplitudes (Fig.
paired with EPSPs reliably induced an in- tiation or to between 230 and 210 mV (in 2D) (10), which suggests that initiation of
postsynaptic APs and their coincidence
with EPSPs was required to induce persis-
Fig. 2. AP-EPSP coinci-
dence requirement. (A)
tent changes in active synapses. Because
Synchronization of indi- most synaptic contacts were located on
vidual postsynaptic APs dendrites around 100 mm away from the
with EPSPs. Two APs, cell body (Fig. 1A) (7), the coincidence of
separated by 50 ms, the back-propagating AP with active syn-
were evoked in a presyn- apses (11) seems to be a critical step for
aptic neuron (Pre. APs), induction of changes in EPSPs. The persis-
which elicited unitary tent increase in EPSPs caused by pairing of
EPSPs in the postsynap- individual APs and EPSPs was prevented
tic neuron (Post. EPSPs).
when N-methyl-D-aspartate (NMDA) re-
Postsynaptic APs (Post.
APs) were synchronized
ceptors were blocked (12), which suggests
with each EPSP by injec- that the back-propagating AP interacts
tion of current pulses with activated NMDA receptors to trigger
into the postsynaptic the synaptic modification, for example,
neuron (approximately 5 through Ca21 inflow.
ms after the presynaptic The limits of the time window for AP-
AP). APs were evoked EPSP interaction were investigated in uni-
by injection of 5-ms, and bidirectionally connected neurons (13).
1-nA current pulses into When postsynaptic APs preceded EPSPs by
cell bodies (lm). (B) AP-
EPSP coincidence. Three
100 ms (Fig. 3, A and C) or followed the last
stimulation protocols EPSP in a burst by 100 ms (Fig. 3, A and C),
were used. (i) Two, 5, or no effect on EPSP amplitudes was observed,
10 EPSPs and postsyn- which suggests that APs and EPSPs must
aptic APs at 20 Hz, coincide within 100 ms in order to induce
paired 10 times every 4 s changes in EPSPs. To determine the limits of
(n 5 21; solid squares); (ii) 5 or 10 postsynaptic APs alone at 20 Hz (n 5 5; open diamonds); and (iii) 5 the AP-EPSP coincidence window, the ef-
or 10 EPSPs alone at 20 Hz (n 5 5; solid diamonds). Time axes are normalized to the onset of the pairing fect of an AP initiated shortly before or
(arrow at time 0). (C) Frequency dependence. Five EPSPs and postsynaptic APs at 2 (n 5 2), 5 (n 5 5), shortly after the onset of an EPSP was tested
10 (n 5 9), 20 (n 5 11), 30 (n 5 3), and 40 Hz (n 5 4) were synchronized in bursts and repeated 10 times
simultaneously in bidirectionally connected
every 4 s. (D) Sustained postsynaptic depolarization. Two stimulation protocols were used. (i) Circles
represent eight experiments in which bursts of 5 to 10 EPSPs (at 20 to 40 Hz, repeated 10 times every
neurons (Fig. 3B). A burst of APs was trig-
4 or 20 s) were synchronized with sustained depolarization to below the AP threshold (200 to 500 ms; gered at a frequency of 10 Hz in one cell;
250 mV ). (ii) Squares represent six experiments in which similar bursts were synchronized with with a 10-ms delay, an identical burst was
depolarizations to between 230 and 210 mV in neurons loaded with Lidocaine [N-ethylbromide triggered in the other cell. This ensured that
quaternary salt (QX-314, 5 mM; RBI, Bethesda, Maryland)]. each postsynaptic AP in the burst occurred

Fig. 3. Interaction and coincidence intervals. (A) Interaction interval. Bursts of connected pair of neurons. This resulted in the postsynaptic APs occurring
postsynaptic APs were evoked in two bidirectionally coupled cells 100 ms about 10 ms before the onset of the EPSP in cell 1 [210 ms; solid squares in
apart, thus the interval of 1100 ms and 2100 ms between EPSPs and (C)] and about 10 ms after the onset of EPSPs in cell 2 [110 ms; open
postsynaptic APs was investigated in the same slice and at the same time. squares in (C)]. This pattern was evoked in a burst of five such temporally
The EPSPs are represented by the vertically expanded traces between the shifted APs, at 10 Hz, and repeated 10 to 15 times every 4 s. (C) Interaction
time points indicated by the arrows. Vertically expanded EPSPs of cell 1 were and coincidence intervals. Coincidence interval is represented by data from
corrected for the skewed baseline. (B) Coincidence interval. A 10-Hz train of six bidirectionally coupled neurons. The averaged data when EPSPs and
5 APs in cell 1 (only first AP shown) was shifted by 10 ms in cell 2. An AP was postsynaptic APs were 100 ms apart are also represented (1100 ms, open
thus triggered in cell 1 10 ms before an AP in cell 2 in a bidirectionally circles, n 5 6; 2100 ms, closed circles, n 5 4).

214 SCIENCE z VOL. 275 z 10 JANUARY 1997

 REPORTS
either 10 ms before the onset of the EPSP in corded for 30 to 60 min. Amplitudes of EPSPs (aver- 11. The conduction velocity in thin basal dendrites is
age of 75) were measured as the difference between estimated to be roughly 0.2 m/s [estimated from (4)]
one cell and 10 ms after the onset of each onset (5 to 10 ms before) and peak of the EPSP (1 to and the back-propagating AP would thus reach
EPSP in the other cell (Fig. 3, B and C). 2 ms). Pairs in which the control average EPSPs dif- most synapses within 1 ms.
After this stimulation, the EPSP amplitude fered by more than 10% were discarded. The maxi- 12. D(2)-2-Amino-5-phosphonopentanoic acid (50 mM)
mum deviation from the baseline at any time during a was applied to six connections (2).
in the neuron in which the postsynaptic APs 10- to 50-min period after pairing was used to repre- 13. P. K. Stanton and T. J. Sejnowski, Nature 339, 215
occurred 10 ms before the EPSPs was re- sent the change in the EPSP. The time axes of all (1989).
duced, whereas it was increased when the experiments were normalized to the pairing period. 14. W. Singer, Annu. Rev. Physiol. 55, 349 (1993).
postsynaptic AP occurred 10 ms after the 10. B. Gustafsson, H. Wigstrom, W. C. Abraham, Y.-Y. 15. We thank L. Wollmuth, F. Helmchen, G. Borst, D.
Huang, J. Neurosci. 7, 774 (1987); S. R. Kelso, A. H. Feldmeyer, A. Roth, and O. Ohana for their com-
onset of the EPSPs (Fig. 3, B and C), which Ganong, T. H. Brown, Proc. Natl. Acad. Sci. U.S.A. ments on the manuscript and A. Roth for simulations
suggests that postsynaptic AP can up- or 83, 5326 (1986). Block of calcium channels by QX- to estimate AP delays in the dendrite. H.M. was
down-regulate the amplitude of EPSPs, de- 314 could contribute to the lack of effects (M. J. supported by a Minerva Fellowship.
Talbot and R. J. Sayer, J. Neurophysiol. 70, 2120
pending on the precise timing of synapse (1996). 4 October 1996; accepted 22 November 1996
activation.
These observations suggest that the den-
dritic APs are a trigger for modifications of
the functional synaptic connectivity be-
Microtubule Treadmilling in Vivo
tween neocortical pyramidal neurons. The Vladimir I. Rodionov and Gary G. Borisy*
result of synaptic integration, encoded
within the back-propagating AP, is con-
veyed to dendrites, and hence the back- In vivo, cytoplasmic microtubules are nucleated and anchored by their minus ends at the
propagating AP could be regarded as a centrosome and are believed to turn over by a mechanism termed dynamic instability:
“binding signal” for active synaptic con- depolymerization and repolymerization at their plus ends. In cytoplasmic fragments of
tacts. A single synaptic contact could there- fish melanophores, microtubules were shown to detach from their nucleation site and
fore be modified in the context of the depolymerize from their minus ends. Free microtubules moved toward the periphery by
summed activity of all the contacts on the treadmilling— growth at one end and shortening from the opposite end. Frequent release
neuron. The finding that EPSPs were up- or from nucleation sites may be a general property of centrosomes and permit a minus-end
down-regulated, depending on the precise mechanism of microtubule turnover and treadmilling.
timing of EPSPs with respect to APs, fur-
ther suggests that single synapses are mod-
ified according to the temporal relations of
the electrical activity of the neurons in a Microtubules (MTs) are fibrillar intracel- the periphery (dispersion) along a radial
network (14). lular structures that play important roles in array of MTs (5). Remarkably, melano-
multiple cellular activities, including mito- phore fragments retained the ability to ag-
REFERENCES AND NOTES sis, transport, positioning of membrane or- gregate pigment and organize a radial MT
___________________________ ganelles, and determination of cellular array of correct polarity orientation (with
1. D. O. Hebb, The Organization of Behavior ( Wiley, shape. MT arrays within cells are capable of the minus ends at the center) (6) in the
New York, 1949).
2. T. V. P. Bliss and G. L. Collingridge, Nature 361, 31 rapid rearrangement that depends to a large apparent absence of the centrosome (7). To
(1993). extent on MT dynamics—the ability to ex- study MT dynamics, we fluorescently
3. M. J. Friedlander, R. J. Sayer, S. J. Redman, J. Neu- change subunits between the soluble and tagged MTs in melanophores by microin-
rosci. 10, 814 (1990); R. Malinow, Science 252, 722
(1991); V. Y. Bolshakov and S. A. Siegelbaum, ibid.
polymer pools (1). Studies on MT dynamics jection of labeled tubulin, microsurgically
269, 1730 (1995); C. Stricker, A. C. Field, S. J. Red- in vitro demonstrate the existence of two dissected fragments from the parental cells,
man, J. Physiol. 490, 443 (1996). principal mechanisms of subunit exchange and induced pigment aggregation and for-
4. G. Stuart and B. Sakmann, Nature 367, 69 (1994). known as treadmilling (2) and dynamic in- mation of the MT aster (8). Images of la-
5. W. G. Regehr, J. A. Connor, D. W. Tank, ibid. 341,
533 (1989); H. Markram, P. J. Helm, B. Sakmann, stability (3). Treadmilling involves the ad- beled MTs were then sequentially acquired
J. Physiol. 485, 1 (1995); J. Schiller, F. Helmchen, B. dition of subunits to one (plus) end of an in the living fragments at short time inter-
Sakmann, ibid. 487, 583 (1995); F. Helmchen, K. MT and loss of subunits from the opposite vals (3 s) for extended periods (10 min) (9).
Imoto, B. Sakmann, Biophys. J. 70, 1069 (1996).
6. Slice preparation, electrical recordings using Nomar-
(minus) end. Dynamic instability is defined Playback of the image sequences re-
ski optics, and anatomical reconstructions for the by gain and loss of subunits at the same end vealed characteristic patterns of MT behav-
localization of synapses were as described (7). Rats (either plus or minus) of an MT during ior. At any given time, about 80% of the
( Wistar) aged 14 to 16 days were used because
cortical synapses are most plastic at these ages [K.
growth and shortening. In living cells, MTs (n 5 1067) appeared to be static, with
Fox, Neuron 15, 485 (1995)]. Data are presented as where the minus ends of MTs are believed one end at the pigment aggregate and the
the mean 6 SEM. The significance of changes was to be tightly anchored at the centrosome, other at the plasma membrane. The other
determined with independent two-tailed t tests (sig- MTs are thought to exchange subunits by 20% of MTs showed dynamic behavior,
nificance, P , 0.05).
7. H. Markram, J. Lübke, M. Frotscher, A. Roth, B. polymerization and depolymerization at either primarily growing (10.6%), primarily
Sakmann, J. Physiol., in press. their plus ends, thus using the dynamic shortening (9.3%), or moving (;1%) (Fig.
8. H. Markram and M. Tsodyks [Nature 382, 807 instability mechanism (4). 1A). MTs showed only short length excur-
(1996)] showed that the increase caused by pairing
of pre- and postsynaptic activity only applies when
We studied microtubule dynamics in cy- sions at their free ends, as if dynamic insta-
tested by low-frequency (,5 Hz) stimulation. toplasmic fragments of fish melanophores, bility was suppressed. MTs emerged from
9. Passive membrane properties were measured im- which translocate cytoplasmic pigment the pigment mass and grew toward the pe-
mediately after whole-cell measurements. Control granules to the center (aggregation) or to riphery (Fig. 1A, MT1), which indicates
test EPSPs evoked by single presynaptic APs at 0.25
Hz were recorded from 2 to 3 min after whole-cell that the pigment aggregate had the capacity
recording. After a 10- to 12-min control period, the
Laboratory of Molecular Biology, University of Wisconsin,
to nucleate MTs. The aggregate also seemed
pairing protocol (detailed in the figure legends) was to stochastically release MTs, after which
Madison, WI 53706, USA.
initiated, test EPSPs were recorded for 10 to 15 min,
and the passive membrane properties were mea- * To whom correspondence should be addressed. E-mail: they shortened at their proximal ends (Fig.
sured again. Test EPSPs were then continuously re- ggborisy@[Link] 1A, MT2). Thus, the two populations of

SCIENCE z VOL. 275 z 10 JANUARY 1997 215

View publication stats