Temporal constraints in associative synaptic plasticity ...

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by China University of Science and Technology on 06/04/13 For personal use only.

Tempora constraints in associative synaptic plasticity in hippocampus and neo Dominfque Debanne, Daniel E. Shulz, and Ywes Frdgnae

Abstract: We present comparative experimental evidence for the induction of synaptic potentiation and depression in organotypic cultures of hippocampus and in visual cortex in vitro and in vivo. The effects of associative pairings on the efficacy of synaptic transmission are analyzed as a hnction of the temporal delay between presynaptic activity and post-synaptic changes imposed in membrane potentid. Synchron~us association at a low temporal frequency ( ~ 0 . 5Hz) between presynaptic input and postsynaptic deplarization resulted in homosynaptic potentiation of functionally identified postsynaptic potentials in the three types of preparation. Synchronous pairing of afferent activity with hyperpolarization of the postsynaptic cell resulted in homosynaptic depression in visual cortex in vivo and in vitro. An associative form of depression was induced in hippocampus when the test input was followed repeatedly with a fixed-delay postsynaptic depolarization imposed either by intracelldar current injection or synaptically. The latter process might play a significant role in heterosynaptic plasticity in visual cortex in vivo and in vitro, if it is assumed that associative depression still operates in visual cortex a few seconds after the initial surge of calcium in the postsynaptic cell. We conclude that the precise timing between presynaptic activity and polarization changes in postsynaptic membrane potential up- and down-regulates the efficacy of active pathways. Key words : synaptic potentiation, synaptic depression, asynchrony , covariance, supervised learning.

R6um9aC : Nous comparons des donnCes expkrimentales relatives 21 l'induction d'une potentialisation et d'une dkpression synaptiques, dans des cultures organotypiques d'hippocampe et dans le cortex visuel in vitro et in vivo. Nous andysons les effets d'appaf-iements associatifs sur l'efficacitk de la transmission synaptique en fonction du dblai temporel entre l'activitk presynaptique et les variations postsynaptiques imposbes au potentiel de membrane. %'association synchrone, il basse frCquence ( 0.5). The resting membrane potential (-67 mV) or the input resistance of the cell (368 Ma) were unaffected by the pairing procedure (D. Bebanne, A. Baranyi, D. Shulz, and Y. FrCgnac, unpublished).

cells (6 of 22) submitted to alternate stimulation sf two competing pathways. An example is shown in Fig. 3b, and the mean change averaged among d l modifid cells appears to be of a lower amplitude (20-25 96) than hat seen in hippocampal cultured neurons (30 -40%). In summary, it cannot be excluded that associative depression may play a role in heterosynaptic plasticity in visual cortex, but the IS1 delay (values up to 2.5 s in vitro and 4.0 s in vivo) appears much longer &an hat demonstrated to be effective in inducing associative LTD in hippoempus. Further experiments still have to be performed to address specifically such an inte~g~sretatioaa.

Does pstsynaptie hypergolarization and gresynaptic activation induce depression?

0

8.5

5

8.5

TIME (rnin)

In the neocortex The role of postsynaptic hypepslarization in the induction sf homosynaptic depression in the neocortex has been proposed on the basis of distinct but converging experimental strategies. One strategy is based on population analysis and ocular dominance histograms. It is well established that, in 4- to 6-week-old kittens, monocular deprivation induces a shift of the ocular dominance histogram toward the open eye (Hubel and Wiesel 1970). If a similar deprivation period is associated with intracorticd infusion (using an osmotic minipump) of muscimol, a GABAA agowist, electrsphysiologicd assessment of ocular dominance (once the phamaeolsgicd effect of muscimol is gone) shows that most cortical cells are now responding domhmtfy through the closed eye (Reiter and Stryker 1988). A straightforward application of the


Fig. 4. Does synchronous pairing with postsynaptic hyperpolarization induce depression? (a) In rat hid~~)ocampal CAB neurons in orgarnotypic cultures. Synchronous pairings (91-) between a current-induced hyperpolarization (240 ms, -0.5 to -2 .O nA, 80 repetitions) and synaptic activation did not induce significant depression in CAI pyramidal neurons (- 11%, n = 6 , paired b test, p > 0.01) (D. Debanne, unpublished). (b) In kitten visual cortical neurons in vibro. Synchronous pairings (S-) between a current-induced hyperpolarization (50-80 ms, 2 to -7 d,50 repetitions) induced a short-term depression in cortical neurons. Time course of average effects are illustrated for cells that were significantly affected (KS test, p < 0.05 in dl cases) by the S- procedure (n = 12). All values indicated are the mean peak amplitude of the averaged PSPs normalized to control values (100%) f SEM (taken from Fig. 12C in Fr6gnac et al. 1994.b, with permission). The example shown in the inset is taken from a layer 3 area 17 corticd neuron in a 4-week-old fitten. Hntracortical input pathway was paired with -3.0-r~4 pulses for W trials. Following pairing, the paired pathway PSP decreased by -33% ( p < 0.85, KS test), while the test unpaired pathway was unaffected ( + 2 % , p > 0.1, KS test). The resting membrane potential and input resistance remained stable at -65 mV and 65 Ma, respectively, throughout the recording period. (c) In kittf visual cortical neurons in vivo. Homosynaptic depression in vivo of a Complex cell recorded in a 10-week-old kitten. The extinction of an optimally oriented bright bar (OFF response) in the central zone During the control period following pairing the of the WF was associated during pairing (22 trials) with a negative current (-2 d). OFF response was depressed, while the ON response was unchanged. The slope as shown by the superimposition of averaged responses (lower trace) before (thin line) and after (thick line, arrow) pairing seems to be also reduced, but because of the intrinsic variability of in vivo recordings, this effect did not reach statistical significance (D. Debanne, A. Baranyi, D. Shulz, and Y. Fr$gnac, unpublished).

sjw a

c

P

3

5

1

1,

180

2iC &

M

50

P W

+'i"

Li'

5

\

i02

'W,

0

z

'I'INIE (mlw)

b 50 nms

TIME (s)

covariance rule to that experimental situation predicts a longlasting depression of active inputs (driven by the open eye) during the muscimol-induced postsynaptic hyperpolarization (since the treatment imposes a negative change in covariance). The gain of silent synapses driven by the closed eye should remain unchanged because the covariance term for the considered input is nul. In other terms, this form of

depression is associative since it requires presynaptic activity. A more direct demonstration at the cellular level was obtained using our cellular conditioning protocols, in which negatively covaried presynaptic visual stimulation and postsynaptic firing of a cortical neuron induced long-lasting depression of suprathreshold visual responses (Fr6gnac et al. 1988, 1992; Shulz and Frdgnac 1992). However, as for


Can. J. Physiol. Pharmaeal. Val. 73, 1995

positive covaiance pairings, a limitation in the interpretation came from the fact that the postsynaptic membrane potential was not directly controlled in these experiments, and that covariance changes were inferred from measurements of changes in firing rates. We have consequently looked for the effects of contiguous pairing of subthreshold input with somatic hypepolarization, during the murse of intracellular recordings, both in vivo (Baanyi et al. 1991) and in vitro (Pregnac et d 1994b). An example of a cell recorded in vivo, in collaboration with Attila Baranyi, is shown in Fig. 4c, where both the slope and the peak amplitude of the paired response are depressed. The absence of change for the unpaired characteristic of h e stimulus demonstrates the input specificity of the effect. Similar observations were obtained in the visud cortical slice, in collaboration with Michael Friedander and David Smith. In cat and guinea-pig visud cortex in vitro, postsynaptic hypevolarizing current pulses (80 ms, -2 to -7 d) were paired at a low temporal frequency with a test input, and resulted in significant depression of the PSB in a third of the cases (Fregnac et al. 1998, 1994b; Fig. 4b). However, these changes, similar to those observed in vivo, were generally of a shorter duration (10-20 min vs. 18126%min). It is possible that, in dud stimulation protocols, the more permanent form of depression induced in vivo is due not only to contiguous pairing of the postsynaptic hyperpolarization (see Fig. 489)but d s o to heterosynaptic changes resulting from the temporal asynchrony between the increased postsynaptic activity and the stimulation of other hnctiond inputs (see preceding section). However, significant depression was d s o observed when only one pathway was stimul a t d and paired with hyperpolarization (Figs. 4-6 in Fregnac et d. 1994b).

induced (Thiels et d. 1994; Xie et d. 1992). These results are compatible with those observed by us in visual cortex in vivo and in vitro= In contrast, low frequency pairing with hyperpolarizing current (240 ms, -8.5 to -2.0 nA) did not induce homosynaptic LTD in the organotypic culture of hippocampus (Bebanne et al, 1994)). In spite of a reduction of 11 % in the mean size of the BSP and a notable increase in the variance of the response following pairing of test input, no significant effect was found (Fig. 3a; paired 8 test, p > 0.01). It m y be observed that in most cases of LTD previously reported in the hippcampus, the frequency of stimulation of the test synapse was sufficiently high (5 - 10 Hz) to elicit a significant potentiation per se in the absence of current-induced or GABAA-mediatedhyperpolarization (Stanton and Sejnowski 1989; Thiels et d . 19%; Xie et d. 1992). NMBA receptor activation and an elevation of the intracellular calcium concentration, as well as perhaps metabotropic glutamate receptor activation, are required for the induction of hippocampd depression (Thiels et d . 1994; Xie et d. 1992; Yang et al. 1994). Conditioning protocols used in visual cortex and hippocampus differ in other aspects: in visual cortex, rather short duration current pulses were applied contiguously with the test input at the same temporal frequency as that used to establish the response baseline (before and after pairing). It remains to be determined, in the latter case, if NMDA or metabotropic glutamate receptor activation, and increases in intraeellular calcium concentration above the level reached during normal test of synaptic transmission, are also required for the induction of depression.

In the hippocampus The first experimental data in hippocmpus demonstrating the role of the membrane potential hyperpolarization were provided by Stanton and Sejnowski ((1989). They found that the current-induced hyperpolarization of a CAI neuron paired with the tetanization of the afferent fibers at low frequency (5 Hz) induced a 30% depression of excitatory synaptic transmission, whereas the synaptic gain of a control pathway, kept unstimulated during the pairing procedure, was unchanged. Similar observations have since been reported when NMDA receptor mediated EPSPs (after blockade of inhibitory and mn-NMDA excitatory responses) are tetanized at 10 Hz during a constant postsynaptic kyperpolarization of -30 rnV (Xie et d*1992). A more physiologicd way to induce hyperpolarization in the postsynaptic neuron can be obtained by activating in addition to the test pathway (often presumed to be excitatory) an inhibitory input during the conditioning train. Two independent studies recently showed that LTD can be induced when excitatory synaptic inputs are stimulated during postsynaptic GABA-mediakd inhibition (Yang et d- 1994: Thiels et d. 199%). An accepted interpretation is that the role of postsynaptic hyperpolarization is to lower calcium influx to such an extent that depression rather than potentiation is induced (Miyakawa et al. 1992). Indeed when hypepolarizing current is suppressed or when the GABA-mediated hyperpolarization is blocked, a weak but significant potentiation is

Our results demonstrate that the efficacy of particular synaptie pathways to neocoflical and hippcampal neurons can be modulated depending on the degree of synchrony between presynaptic activation and appropriate brief changes in postsynaptic membrane potential- The protocols used in the three types of preparation (hippcampd organotypic cultures, visual cortex in vivo and in vitro) did not require high frequency stimulation of the conditioned pathway, but o d y a low frequency (