- Email: [email protected]
Priming of associative long-term depression in the dentate gyrus by θ frequency synaptic activity
Neuron,
Vol.
9, 79-84,
July,
1992,
Copyright
0
1992
by Cell
Press
Prim ing of Associative Long-Term Depression in the Dentate Gyrus by 8 Frequency Synaptic Activity Brian R. Christie and Wickliffe C. Abraham Department of Psychology and Neuroscience Research Centre University of Otago Dunedin New Zealand
Summary Associative long-term synaptic depression (LTD) was investigated utilizing negatively correlated activity patterns in the medial and lateral perforant path inputs to the dentate gyrus in anesthetized rats. Normally only nonassociative, or heterosynaptic, LTD is elicited in naive pathways. W e report here, however, that associative LTD in the lateral path is readily induced after being “primed” by a brief period of lateral path synaptic activity at a 0 rhythm frequency (5 Hz). Priming of associative LTD lasts at least 2 hr and is not seen following priming activity at non8 frequencies (1 and 15 Hz). N-methyl-o-aspartate receptor activation is critical for establishing the priming effect, but not for the subsequent induction of the associative LTD. These data suggest that 0 rhythm activity in the dentate gyrus may predispose the system to a specific form of synaptic plasticity, associative LTD. Introduction In most adaptive neural network models, learning is simulated by modifications of synaptic weights that occur in response to temporal correlations between pre- and postsynaptic activity (Hebb, 1949; Kohonen, 1984; Hopfield and Tank, 1986; Bear et al., 1987; Sejnowski and Tesauro, 1989; Brown et al., 1990; Hashemzadeh-Gargari et al., 1991). In the hippocampal formation, long-term potentiation (LTP) generally conforms to this learning rule, in that a positive temporal correlation between presynaptic activity and sufficient depolarization to activate glutamate receptors/channels of the N-methyl-o-aspartate (NMDA) subtype results in a persistent enhancement of synaptic efficacy at the active synapses (Bliss and Ltimo, 1973; Harris et al., 1984; Kelso et al., 1986; Wigstrijm et al., 1986). However, the rules governing the reverse effect, long-term depression (LTD) of synaptic efficacy, are poorly understood. Pathways in the dentate gyrus have been shown to sustain nonassociative, or heterosynaptic, LTD when they are inactive during postsynaptic depolarization, as induced by conditioning stimulation to a separate pathway that converges on the same postsynaptic neuronal population (Levy and Steward, 1979, 1983; Abraham and Goddard, 1983; Christie and Abraham, 1992). The appearance of the LTD in the inactive pathway, as well as the LTP in the conditioned pathway, is
prevented by prior administration of NMDA receptor antagonists (Desmond et al., 1991; Christie and Abraham, 1992). Recently, an NMDA-independent associative LTD has been reported to occur in areas CA1 and CA3 of the hippocampus in vitro (Stanton and Sejnowski, 1989; Chattarji et al., 1989; Stanton et al., 1991). In these experiments, one pathway was administered short, highfrequencyconditioningtrainsat200msintervals, while a second test pathway received single pulses interleaved between these stimulus trains. Establishing this negative correlation between activity in the conditioned and test pathways led to the induction of LTD in the test pathway. Since activity in either pathway alone was insufficient to induce LTD, this effect has been termed “associative” LTD. Administration of the NMDA receptor antagonist 2-amino5-phosphonovaleric acid (AP5) did not block the associative LTD, even though it did prevent the appearance of LTP in the conditioned pathway (Stanton and Sejnowski, 1989). The induction of LTD in the test pathway could be prevented, however, by 2-amino-3phosphonopropionate (AP3), which inhibits phosphoinositide turnover stimulated by the activation of metabotropic glutamate receptors. The effect of AP3 was specific to LTD, since it did not impair the induction of LTP in the conditioned pathway (Stanton et al., 1991). The reliability of the associative LTD effect is somewhat controversial, however, and a number of laboratories have failed to produce associative LTD similarto that reported by Stanton and colleagues. Associative LTD has not yet been observed in the dentate gyrus in vivo (Christie and Abraham, 1992), and a number of abstracts have reported a failure to replicate the phenomenon in CA1 in vitro (see Paulsen et al., 1990, Eur. Neurosci., abstract; Kerr and Abraham, 1992, Int. J. Neurosci., abstract). W e present evidence here that prior synaptic activation within the frequency range of hippocampal 0 activity is a critical prerequisite for the appearance of associative LTD in the dentategyrus of anesthetized rats. Such activity also reduces the expression of the nonassociative LTD that is exhibited in naive pathways (Christie and Abraham,, 1992). Results Prior Synaptic Activation Selectively Facilitates Associative LTD Conditioning protocols consisted of two forms: anonassociative paradigm involving trains of pulses to the medial path with no concurrent stimulation of the lateral path, and an associative paradigm whereby single pulses to the lateral path were interleaved between the medial path trains, thus establishing a negative correlation between presynaptic activity in the
Neuron 80
Prime
A
MPP
+
LPPI
+ Prime
B
MPP
Non-Associative I I I I I ““I Associative I’ll’
:
I
,
I 200 ms
C
F
D
E
LPP-NORMAL
LPP-PRIMED
PRIMED
NORMAL
2; 0
NA
20 10 Time Post-Condltlonlng
Priming
Facilitates
Conditioning stimulation was normally delivered to the medial path and consisted of 8 stimulus trains of 2 s duration, spaced 1 min apart. Within each train the stimuli occurred as a burst of 5 pulses at 100 Hz, repeated at 200 ms intervals. In the nonassociative condition (A), the lateral path was kept inactiveduring medial pathconditioning. In the associative condition (B), single pulses to the lateral path were interleaved between the onsets of the medial path bursts. Conditioning stimulation produced equivalent medial path ILTP and lateral path LTD for both associative and nonassociative paradigms in naive pathways. Averages of 5 waveforms before (dashed line) and 30 min after (solid line) associative conditioningareshownforthemedialpath (MPP [Cl) and the lateral path (LPP [D]). Means and standard errors in 5 min epochs are given in (F) (A, associative n = 13; NA, nonassociative n = 11). (E and G) 5 Hz priming stimulation, 10 min prior to conditioning, enhanced LTD following associative stimulation (n = 5) but notfollowing nonassociative stimulation (n = 5). Asterisks indicate those points at which significantly more associative LTD was produced (p < 0.05). Calibration bar, 5 ms, 2.5 mV.
lllll
I I I I I I
LPPI
Figure 1. 5 Hz Synaptic Associative LTD
Illll
-50 1, 30
(mln)
0
10 Time
20
30
Post-Condltlonlng (mln)
lateral path and strong postsynaptic activity as induced by the medial path (Figures IA and IS). The application of either stimulus paradigm resulted in 20%-25% LTP of the medial path field excitatory postsynaptic potential (EPSP) recorded in the dentate hilus and an approximatelyl5% decrease in the lateral path field EPSP, as measured 30 min following conditioning (Figures IC, ID, and 1 F). These data are in accord with previous findings that the associative stimulation paradigm normally elicits only the nonassociative form of LTD in the dentate gyrus (Christie and Abraham, 1992). However, when the lateral path was first briefly stimulated with 8 low frequency (5 Hz) trains of 10 pulses each, the amount of lateral path LTD 30 min following associative conditioning was more than doubled compared with that following normal associative conditioning (-40.0% 5 3.1%; tc16) = 3.8; p < 0.01; Figures IE and IG). The application of the 5 Hz stimulation to the lateral path did not induce any significant changes in the lateral path EPSP (6.3% f 5.0% as measured just prior to associative stimulation, n = 18) and did not affect the degree of subsequent LTP exhibited in the medial path. The preconditioning stimulation effect was spe-
cific for the associative conditioning paradigm, since such stimulation prior to the administration of the nonassociative paradigm (-8.5% + 8.6%, n = 5), or a second application of the 5 Hz stimuli (0.2% +_ 6.4%, n = 2), did not result in significant lateral path LTD. Identical preconditioning stimulation applied to the medial path, at an intensity sufficient to cause postsynaptic cell firing to each stimulus, did not facilitate lateral path associative LTD (11.5% k 7.5%, n = 3).This indicates that the preconditioning, or “priming,” of associative LTD is input specific and not the result of general postsynaptic depolarization or cell firing. Duration and Frequency Dependence of the Priming Effect To establish the time course and consistency of the priming effect, the 5 Hz stimulation was delivered at various time points from 10 min to 2 hr before associative conditioning. Significantly enhanced lateral path LTD was observed at all time intervals (group means ranging from -31% to -44%; Figure 2). The consistency and robustness of the priming effect were shown by the fact that 17 of the 18 animals tested at the various time intervals showed lateral path LTD
Priming of Associative 81
Depression
40-
30-
20 -
NONE
1
Priming
5 Frequency
15 (Hz)
10
50-
0 ‘I
40NONE
10
Prime-Condition Figure 2. Duration
of the Synaptic
30
60
Interval Priming
120
(min)
Effect
Priming stimulation enhanced associative LTD in the lateral path at all intervals examined, from 10 min to 2 hr between 5 Hz priming stimulation and associative conditioning. NONE = nonprimed associative conditioning (n = 10). The four primed groups (IO min, n = 5 [data taken from Figure ICI; 30 min, n = 5; 60 min, n = 4; 120 min, n = 4) all showed significantly greater LTD compared with the normal group, using a one-way analysis of variance (FC4,26) = 7.3, p < 0.001) and post-hoc Student NewmanKeuls test (asterisks; p < 0.05). Data are means and standard errors obtained 30 min after associative conditioning.
NONE Priming Figure 3. Frequency Specificity tation of Associative LTD
greater than 25%. Only 1 of 5 animals in the primed nonassociative condition and only 2 of 13 animals in the nonprimed associativecondition showed a similar degree of LTD. A remarkable feature of the priming effect was that low frequencies other than 5 Hz, but outside the range of hippocampal0 activity, were ineffective in enhancing LTD. The LTD subsequent to 15 Hz priming stimulation (-12.5% +_ 6.6%, n = 5) was not different from the LTD produced in the normal, nonprimed condition (-13%), while 1 Hz stimulation resulted in a modest, but still nonsignificant, increase in LTD (-23.7% + 8.8%, n = 7; Figure 3). No changes in the amount of LTP produced in the medial path were observed either at the different time intervals following the 5 Hz priming, or forthedifferent frequenciesof lateral path priming. This indicates that low frequency priming of heterosynaptic afferents does not affect the induction of homosynaptic LTP.
5
1
Frequency of Synaptic
15 (Hz)
Priming
for the Facili-
Priming stimulation (5 Hz) significantly enhanced associative LTD (asterisks; tlzs, = 25.4; p < 0.05; n = 18) in the lateral path compared with the normal, nonprimed group (NONE, n = IO). Priming stimulation of 1 Hz (n = 7) and 15 Hz (n = 5) did not have a significant influence on LTD (one-way analysis of variance: [F,,39r = 8.4, p < 0.0011 and post-hoc Student Newman-Keuls test: p > 0.05). Medial path LTP was unaffected by priming stimulation delivered to the lateral path at any frequency. Data are presented as mean f standard error percent change, as measured 25-30 min after conditioning. These findings indicate that the degree of medial path LTP and the degree of lateral path LTD are poorly correlated.Thedataforthenonprimedgroupandthe5Hzgroup were taken from Figures IF and 2 (collapsed across all time points), respectively.
NMDA Receptor Involvement in Associative LTD Since the priming stimulation specifically facilitated the LTD arising from negatively correlated activity, this may reflect the induction of associative LTD, which could be mechanistically different from, and additive to, the normally occurring nonassociative LTD. To test this hypothesis, we took advantage of the fact that the two types of LTD have been shown to be
NWKNl a2
MEDIAL
401
30 1
-20 ! -15
'
1 -5
5 Time
20
25
15
(min)
LATERAL
1
lon~c--o -.-. -.- ._._.-.-.-.-.-.-._-._.-._._.-.-.. _._._._._._
04 - ‘-‘-’
l +. '8.
-lO- +-
0 %
h+j--
L --
L 5.
-2o-30 -40 ! -15
--cl--
CPP BEFORE PRIMING
-
CPP AFTER PRIMING
1 -5
5 Time
Figure
4. NMDA
Receptor
Antagonism
15
25
(min) with
CPP
CPP (IO mgikg i.p., Cambridge Research Biochemicals) given 2 hr prior to 5 Hz priming stimulation effectively blocked associative LTD in the lateral path (CPP BEFORE PRIMING, closed circles, n = 4). Significant associative LTD (two-way ANOVA, group main however, when CPP effect, Fci,6)= 10.12, p < 0.01) was observed, was given just after priming stimulation (CPP AFTER PRIMING; open circles, n = 4). Medial path LTP was reduced in both conditions. Note the gradually incrementing LTD when the CPP was given after priming condition. A similar trend is apparent for associative LTD in Figure IC. Data are mean and standard error for 5 min epochs following associative conditioning.
pharmacologically dissociable: nonassociative LTD is prevented by NMDA receptor antagonism (Abraham and Wickens, 1991; Desmond et al., 1991; Christie and Abraham, 1992), but associative LTD is not (Stanton and Sejnowski, 1989; Chattarji et al., 1989). As expected from these prior experiments, when the competitive NMDA receptor antagonist CPP (3-[(+)-2carboxypiperazin-byI]-propyl-I-phosphonic acid; 10 mglkg i.p.) was given after priming stimulation, but before the conditioning stimulation, therewas no LTD in the lateral path following the nonassociative para-
digm (5.5% k 7.6%, n = 3). In contrast, a significant, although reduced, LTD developed following the associative paradigm (-18.9% i 5.8%; tc3) = 3.2; p < 0.05; Figure 4, CPP AFTER PRIMING). As medial path LTP was reduced under both conditions, the LTD did not result from a failure to antagonize postsynaptic NMDA receptors. CPP, given after1 Hz priming stimulation, completely blocked subsequent LTD induction (-2.0% + 8.4%, n = 3), suggesting that following 1 Hz activity only NMDA-dependent, nonassociative LTD is induced. To test whether priming stimulation itself involved NMDA receptor activation, CPP was administered prior to the 5 Hz priming stimulation. Under this condition, the associative paradigm was no longer effective at eliciting LTD (4.6% + 8.0%, n = 4; Figure 4, CPP BEFORE PRIMING): These results indicate that 5 Hz stimulation initiates priming mechanisms that are coupled to NMDA receptor activation, but once the relevant cellular machinery is activated, the subsequent induction of associative LTD is independent of NMDA receptor activation. Taken together, our data support the hypothesis that associative and nonassociative LTD are independent and additive. The 35% LTD seen after 5 Hz priming in drug-free animals appears to consist roughly of 15% nonassociative LTD (i.e., the NMDA-dependent LTD seen in naive pathways) and 20% associative LTD (as induced by theassociative stimulus protocol when CPP follows priming stimulation). Specificity of the Priming Effect for Associative LTD To test whether priming stimulation facilitates synaptic plasticity in general, 5 Hz priming stimulation was delivered to the lateral path prior to the induction of LTP in this pathway. In this situation, priming stimulation significantly reduced (tcsI = 25; p < 0.05), by about 50%, the amount of lateral path LTP generated (normal LTP: 47.8% F 7.7%, n = 6; LTP after priming: 22.6% i 7.7%, n = 5). This effect may reflect an interference with the mechanisms involved in the induction of LTP, or a concurrently elicited associative LTD that masks the true extent of the LTP. In either case, it is apparent that LTP is not facilitated, indicating that the priming activity specifically facilitates associative LTD. The reduction in lateral path LTP is specific to priming of the lateral path, as medial path priming had no effect on lateral path LTP (49.9% f lO.l%, n = 5). Discussion The present results provide evidence that two pharmacologically distinct forms of LTD can be elicited separately, and in combination, within a single monosynaptic pathway. Nonassociative LTD is activity independent and can occur at synapses that are silent during strong postsynaptic depolarization elicited by other inputs. It involves at least the activation of the NMDA subtype of glutamate receptor by these active inputs (Abraham and Wickens, 1991; Desmond et al.,
Priming of Associative 83
Depression
1991; Christie and Abraham, 1992). It is not yet clear how signals travel to the silent synapses to trigger LTD induction there, but it is possible that the spread of strong depolarization across the postsynaptic membrane and the subsequent activation of synaptically located voltage-sensitive calcium channels are sufficient for this purpose (Lisman, 1989; Wickens and Abraham, 1991). In contrast, associative LTD is activity dependent in at least two distinct ways. First, it was produced in the present study only when a negative temporal correlation between presynaptic activity and strong postsynaptic depolarization was established. This temporal relationship seems crucial to the induction of associative LTD in areas CA1 and CA3 of the hippocampus. Pathways that exhibited LTD with the negativelycorrelated stimuli in these regions were found to exhibit associative LTP when the stimuli were administered so as to coincide (Stanton and Sejnowski, 1989; Chattarji et al., 1989). We are currently investigating the necessity of establishing the negative temporal correlation, to ascertain whether the LTD is reliant on its establishment, or requires only presynaptic activity per se. Second, prior synaptic activation at 8 rhythm periodicity primes those synapses for the subsequent induction of associative LTD and may also mask or interfere with the development of LTP and nonassociative LTD. Failure to prime the synapses results in the appearance of only nonassociative, NMDAdependent LTD, even when negatively correlated stimuli are administered. It is important to note that with these stimulation protocols, 5 Hz activity in the lateral path alone is insufficient to induce LTD, even when the stimuli are administered repeatedly. Associative LTD in the dentate gyrus is NMDA independent (as in areas CA1 and CA3), but NMDA receptor activation is necessary for the initial priming step. That the priming effect lasts for at least 2 hr suggests that NMDA receptor activation may be coupled to second messenger pathways (e.g., East and GarthWaite, 1991) necessary for maintainingthe synapses in a state of readiness for subsequent LTD. The recent demonstration that associative LTD in region CA1 may be blocked by inhibiting postsynaptic phosphoinositide turnover mediated by metabotropic glutamate receptor activation (Stanton et al., 1991) may be relevant to this hypothesis. It is not clear, however, whether priming activity is necessary for associative LTD in area CAI, nor whether AP3 will block associative LTD in the dentate gyrus. These questions await further investigation. It is intriguing that the critical frequency for the priming effect lies within the frequency range of the naturally occurring8 rhythm in the hippocampal electroencephalogram. 8 rhythm activity has been found to correlate with exploratory behaviors in rodents and has been suggested to reflect neuronal participation in a variety of cognitive functions including arousal, anxiety, sensory processing, and learning (Berry and Thompson, 1978, 1979; Bland, 1986; McNaughton,
1991). Of particular interest here is the possibility that neural activity in the hippocampus at 0 rhythm frequencies plays an important function in learning mechanisms. Recently Buszhki and Gage (1991) presented a two stage theory of hippocampal synaptic plasticity: a weak labile trace is set up by 8 activity occurring during exploration, and this trace is subsequently converted into a stronger, long-lasting form by patterned activity and postsynaptic depolarization at the termination of exploration. Although this model concentrates on LTP induction, the general similarity between it and the rules for associative LTD induction in the dentate gyrus are striking. Thus, our findings give substance to the model’s concept that 8 activity not only influences the current information being processed in the hippocampus, but also has a longer lasting impact on the magnitude and selectivity of synaptic modifications arising from subsequent patterns of activity. Experimental Procedures Adult male Sprague-Dawley rats (250-500 g) were anesthetized with sodium pentobarbital(65 mg/kg) and placed in a stereotaxic apparatus. Supplemental doses of pentobarbital were given as required to maintain a surgical plane of anesthesia. Rectal temperature was maintained at 37OC + 0.5OC. A 75 pm stainless steel extracellular recording electrode was placed in the dentate hilus 3.5 m m posterior and 2.5 m m lateral to bregma. Two 125 pm monopolar stimulating electrodes were placed separately in the ipsilateral medial perforant path (4.0 m m lateral to lambda) and the lateral perforant path (5.0 m m lateral to lambda). Recordings wereamplified and filtered at 0.1 Hz and 3.0 kHz half-amplitude. Prior to each experiment, lack of cross-facilitation with pairedpulse stimuli was used to ensure adequate separation (>90%) of the two sets of fibers being stimulated. A spatiotemporal summation test (McNaughton and Barnes, 19n; Abraham and Goddard, 1983) was also used to verify that the two stimulated pathways converged onto a common population of granule cells. For acceptance, the population spike in response to the stimulation of the two pathways together was required to be at least twice the amplitude of the sum of the spikes elicited by separate stimulation of the pathways. Test pulses of fixed amplitude and duration (0.1 ms) were delivered alternately to the medial and lateral perforant paths at 0.1 Hz. The initial slope of the recorded EPSPs was measured for each response beginning 15 min before conditioning stimulation and lasting at least 30 min after. EPSP slope values were averaged across 5 min epochs, and all data in the text are given as mean f standard error percent change from baseline, as observed 30 min following conditioning.
Acknowledgments We thank Drs. D. Bilkey, C. Darlington, N. McNaughton, and J. Wickens for critical readings of an earlier version of this manuscript. This research was supported by grants from the New Zealand Health Research Council and the Human Frontiers in Science Program to W. C. A. B. R. C. was supported by a Canadian NSERC postgraduate fellowship. The costs of.publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received
February
19, 1992, revised
April
20, 1992.
References Abraham,
W. C., and Goddard,
G. V. (1983). Asymmetric
relations
m 84
between homosynaptic long-term potentiation and heterosynaptic long-term depression. Nature 30.5, 717-719.
processes underlying learning Sci. USA 86, 9574-9578.
Abraham, W. C., and McNaughton, N. (1983). Differences in synaptic transmission between medial and lateral components of the perforant path. Brain Res. 303, 251-269.
McNaughton, N. (1991). The role of hippocampai RSA: spatial memory versus anxiety. In Memory Mechanisms: A Tribute to G. V. Coddard,W. C. Abraham, M. C. Corballis, and K. G. White, eds. (CITY: New Jersey: Lawrence Erlbaum Associates), pp. 129148.
Abraham, W. C., and Wickens, J. R. (1991). Heterosynaptic term depression is facilitated by blockade of inhibition CA1 of the hippocampus. Brain Res. 546, 336-340.
longin area
Bear, M. F., Cooper, L. N., and Ebner, F. F. (1987). A physiological basis for a theory of synapse modification. Science 237, 42-48. Berry, S. D., and Thompson, R. F. (1978). Prediction rate from the hippocampal electroencephalogram. 1298-1300.
of learning Science 200,
Berry, S. D., and Thompson, R. F. (1979). Medial septal lesions retard classical conditioning of the nictitating membrane response in rabbits. Science 205, 209-211. Bland, 8. H. (1986). The physiology campal formation theta rhythms.
and pharmacology of hippoProg. Neurobiol. 26, 199-218.
Bliss, T. V. P., and Lomo, T. (1973). Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. J. Physiol. 232, 331-356. Brown, T. H., Kairiss, E. W., and Keenan, C. 1. (1990). Hebbian synapses: biophysical mechanisms and algorithms. Annu. Rev. Neurosci. 73, 475-511. Buszaki, G., and Gage, F. H. (1991). Long-term potentiation: does it happen in the normal brain? When and how? In MemoryMechanisms: A Tribute to G. V. Goddard, W. C. Abraham, M. C. Corballis,andK.G. White,eds.(CITY,NewJersey: LawrenceErlbaum Associates), pp. 79-104. Chattarji, S., Stanton, P. K., and Sejnowski, T. J. (1989). Commissural synapses, but not mossy fiber synapses, in the hippocampal field CA3 exhibit associative long-term potentiation and depression. Brain Res. 495, 145-150. Christie, B. R., and Abraham, W. C. (1992). NMDA-dependent heterosynaptic long-term depression in the dentate gyrus of anaesthetized rats. Synapse. 70, 1-6. Desmond, N. L., Colbert, C. M., De, X. Z., and Levy, W. B. (1991). NMDA receptor antagonists block the induction of long-term depression in the hippocampal dentate gyrus of the anesthetized rat. Brain Res. 552, 93-98. East, S. J., and Garthwaite, J. (1991). NMDA receptor activation in rat hippocampus induces cyclic CMP formation through the t-arginine-nitric oxide pathway. Neurosci. Lett. 123, 17-19. Harris, E. W., Ganong,A. H., and Cotman, C. W. (1984). Long-term potentiation in the hippocampus involves activation of N-methyl-o-aspartate receptors. Brain Res. 323, 132-137. Hashemzadeh-Gargari, H., Colbert, C. M., and Levy, W. 5. (1991). Refiningthetemporaldefinitionofanassociationattheneuronal level using long-term potentiation and long-term depression in the dentate gyrus. Neurosci. Lett. 726, 37-40. Hebb, D. 0. (1949). Organization Wiley & Sons).
of Behavior
(New York: John
Hopfield, J. J., and Tank, D. W. (1986). Computing circuits: a model. Science 233, 625-633.
with
neural
Kelso, S. R., Canon& A. H., and Brown, T. H. (1986). Hebbian synapses in hippocampus. Proc. Natl. Acad. Sci. USA 83, 53265330. Kohonen, T. (1984). Self-Organization (Berlin: Springer-Verlag).
and Associative
Levy, W. B., and Steward, 0. (1979). Synapses ory elements in the hippocampal formation. 245.
as associative memBrain Res. 175,233-
Levy, W. B., and Steward, 0. (1983). Temporal contiguity ments for long-term associative potentiation/depression hippocampus. Neuroscience 8, 791-797. Lisman,
J. (1989). A mechanism
for the Hebb
Memory
requirein the
and the anti-Hebb
and memory.
Proc. Natl, Acad.
McNaughton, B. L., and Barnes, C. A. (1977). Physiological identification and analysis ofdentategranule cell responses to stimulation of the medial and lateral perforant pathways in the rat. J, Comp. Neural. 775, 439-454. Sejnowski, T. Jo, and Tesauro, G. (1989). The Hebb rule for synaptic plasticity: algorithms and implementations. In Neural Models of Plasticity, J. H. Byrne and W. 0. Berry, eds. (New York: Academic Press, Inc.), pp. 94-103. Stanton, P. K., and Sejnowski, depression in the hippocampus Nature 339, 215-218.
T. J. (1989). Associative long-term induced by Hebbian covariance.
Stanton, P. K., Chattarji, S., and Sejnowski, T. J. (1991). 2-Amino..3phosphonopropionic acid, an inhibitor of glutamate-stimulated phosphonoinositide turnover, blocks induction of homosynaptic long-term depression, but not potentiation, in rat hippocampus. Neurosci. Lett. 727, 61-66. Wickens, J. R., and Abraham, W. C. (1991). The involvement of L-type calcium channels in heterosynaptic long-term depression in the hippocampus. Neurosci. Lett. 730, 128-132. Wigstrom, H., Custafsson, B., Huang, Y.-Y., and Abraham, W. C. (1986). Hippocampal long-term potentiation is induced by pairing single afferent volleys with intracellularly injected depolarizing current pulses. Acta Physiol. Stand. 726, 317-319.
Vol.
9, 79-84,
July,
1992,
Copyright
0
1992
by Cell
Press
Prim ing of Associative Long-Term Depression in the Dentate Gyrus by 8 Frequency Synaptic Activity Brian R. Christie and Wickliffe C. Abraham Department of Psychology and Neuroscience Research Centre University of Otago Dunedin New Zealand
Summary Associative long-term synaptic depression (LTD) was investigated utilizing negatively correlated activity patterns in the medial and lateral perforant path inputs to the dentate gyrus in anesthetized rats. Normally only nonassociative, or heterosynaptic, LTD is elicited in naive pathways. W e report here, however, that associative LTD in the lateral path is readily induced after being “primed” by a brief period of lateral path synaptic activity at a 0 rhythm frequency (5 Hz). Priming of associative LTD lasts at least 2 hr and is not seen following priming activity at non8 frequencies (1 and 15 Hz). N-methyl-o-aspartate receptor activation is critical for establishing the priming effect, but not for the subsequent induction of the associative LTD. These data suggest that 0 rhythm activity in the dentate gyrus may predispose the system to a specific form of synaptic plasticity, associative LTD. Introduction In most adaptive neural network models, learning is simulated by modifications of synaptic weights that occur in response to temporal correlations between pre- and postsynaptic activity (Hebb, 1949; Kohonen, 1984; Hopfield and Tank, 1986; Bear et al., 1987; Sejnowski and Tesauro, 1989; Brown et al., 1990; Hashemzadeh-Gargari et al., 1991). In the hippocampal formation, long-term potentiation (LTP) generally conforms to this learning rule, in that a positive temporal correlation between presynaptic activity and sufficient depolarization to activate glutamate receptors/channels of the N-methyl-o-aspartate (NMDA) subtype results in a persistent enhancement of synaptic efficacy at the active synapses (Bliss and Ltimo, 1973; Harris et al., 1984; Kelso et al., 1986; Wigstrijm et al., 1986). However, the rules governing the reverse effect, long-term depression (LTD) of synaptic efficacy, are poorly understood. Pathways in the dentate gyrus have been shown to sustain nonassociative, or heterosynaptic, LTD when they are inactive during postsynaptic depolarization, as induced by conditioning stimulation to a separate pathway that converges on the same postsynaptic neuronal population (Levy and Steward, 1979, 1983; Abraham and Goddard, 1983; Christie and Abraham, 1992). The appearance of the LTD in the inactive pathway, as well as the LTP in the conditioned pathway, is
prevented by prior administration of NMDA receptor antagonists (Desmond et al., 1991; Christie and Abraham, 1992). Recently, an NMDA-independent associative LTD has been reported to occur in areas CA1 and CA3 of the hippocampus in vitro (Stanton and Sejnowski, 1989; Chattarji et al., 1989; Stanton et al., 1991). In these experiments, one pathway was administered short, highfrequencyconditioningtrainsat200msintervals, while a second test pathway received single pulses interleaved between these stimulus trains. Establishing this negative correlation between activity in the conditioned and test pathways led to the induction of LTD in the test pathway. Since activity in either pathway alone was insufficient to induce LTD, this effect has been termed “associative” LTD. Administration of the NMDA receptor antagonist 2-amino5-phosphonovaleric acid (AP5) did not block the associative LTD, even though it did prevent the appearance of LTP in the conditioned pathway (Stanton and Sejnowski, 1989). The induction of LTD in the test pathway could be prevented, however, by 2-amino-3phosphonopropionate (AP3), which inhibits phosphoinositide turnover stimulated by the activation of metabotropic glutamate receptors. The effect of AP3 was specific to LTD, since it did not impair the induction of LTP in the conditioned pathway (Stanton et al., 1991). The reliability of the associative LTD effect is somewhat controversial, however, and a number of laboratories have failed to produce associative LTD similarto that reported by Stanton and colleagues. Associative LTD has not yet been observed in the dentate gyrus in vivo (Christie and Abraham, 1992), and a number of abstracts have reported a failure to replicate the phenomenon in CA1 in vitro (see Paulsen et al., 1990, Eur. Neurosci., abstract; Kerr and Abraham, 1992, Int. J. Neurosci., abstract). W e present evidence here that prior synaptic activation within the frequency range of hippocampal 0 activity is a critical prerequisite for the appearance of associative LTD in the dentategyrus of anesthetized rats. Such activity also reduces the expression of the nonassociative LTD that is exhibited in naive pathways (Christie and Abraham,, 1992). Results Prior Synaptic Activation Selectively Facilitates Associative LTD Conditioning protocols consisted of two forms: anonassociative paradigm involving trains of pulses to the medial path with no concurrent stimulation of the lateral path, and an associative paradigm whereby single pulses to the lateral path were interleaved between the medial path trains, thus establishing a negative correlation between presynaptic activity in the
Neuron 80
Prime
A
MPP
+
LPPI
+ Prime
B
MPP
Non-Associative I I I I I ““I Associative I’ll’
:
I
,
I 200 ms
C
F
D
E
LPP-NORMAL
LPP-PRIMED
PRIMED
NORMAL
2; 0
NA
20 10 Time Post-Condltlonlng
Priming
Facilitates
Conditioning stimulation was normally delivered to the medial path and consisted of 8 stimulus trains of 2 s duration, spaced 1 min apart. Within each train the stimuli occurred as a burst of 5 pulses at 100 Hz, repeated at 200 ms intervals. In the nonassociative condition (A), the lateral path was kept inactiveduring medial pathconditioning. In the associative condition (B), single pulses to the lateral path were interleaved between the onsets of the medial path bursts. Conditioning stimulation produced equivalent medial path ILTP and lateral path LTD for both associative and nonassociative paradigms in naive pathways. Averages of 5 waveforms before (dashed line) and 30 min after (solid line) associative conditioningareshownforthemedialpath (MPP [Cl) and the lateral path (LPP [D]). Means and standard errors in 5 min epochs are given in (F) (A, associative n = 13; NA, nonassociative n = 11). (E and G) 5 Hz priming stimulation, 10 min prior to conditioning, enhanced LTD following associative stimulation (n = 5) but notfollowing nonassociative stimulation (n = 5). Asterisks indicate those points at which significantly more associative LTD was produced (p < 0.05). Calibration bar, 5 ms, 2.5 mV.
lllll
I I I I I I
LPPI
Figure 1. 5 Hz Synaptic Associative LTD
Illll
-50 1, 30
(mln)
0
10 Time
20
30
Post-Condltlonlng (mln)
lateral path and strong postsynaptic activity as induced by the medial path (Figures IA and IS). The application of either stimulus paradigm resulted in 20%-25% LTP of the medial path field excitatory postsynaptic potential (EPSP) recorded in the dentate hilus and an approximatelyl5% decrease in the lateral path field EPSP, as measured 30 min following conditioning (Figures IC, ID, and 1 F). These data are in accord with previous findings that the associative stimulation paradigm normally elicits only the nonassociative form of LTD in the dentate gyrus (Christie and Abraham, 1992). However, when the lateral path was first briefly stimulated with 8 low frequency (5 Hz) trains of 10 pulses each, the amount of lateral path LTD 30 min following associative conditioning was more than doubled compared with that following normal associative conditioning (-40.0% 5 3.1%; tc16) = 3.8; p < 0.01; Figures IE and IG). The application of the 5 Hz stimulation to the lateral path did not induce any significant changes in the lateral path EPSP (6.3% f 5.0% as measured just prior to associative stimulation, n = 18) and did not affect the degree of subsequent LTP exhibited in the medial path. The preconditioning stimulation effect was spe-
cific for the associative conditioning paradigm, since such stimulation prior to the administration of the nonassociative paradigm (-8.5% + 8.6%, n = 5), or a second application of the 5 Hz stimuli (0.2% +_ 6.4%, n = 2), did not result in significant lateral path LTD. Identical preconditioning stimulation applied to the medial path, at an intensity sufficient to cause postsynaptic cell firing to each stimulus, did not facilitate lateral path associative LTD (11.5% k 7.5%, n = 3).This indicates that the preconditioning, or “priming,” of associative LTD is input specific and not the result of general postsynaptic depolarization or cell firing. Duration and Frequency Dependence of the Priming Effect To establish the time course and consistency of the priming effect, the 5 Hz stimulation was delivered at various time points from 10 min to 2 hr before associative conditioning. Significantly enhanced lateral path LTD was observed at all time intervals (group means ranging from -31% to -44%; Figure 2). The consistency and robustness of the priming effect were shown by the fact that 17 of the 18 animals tested at the various time intervals showed lateral path LTD
Priming of Associative 81
Depression
40-
30-
20 -
NONE
1
Priming
5 Frequency
15 (Hz)
10
50-
0 ‘I
40NONE
10
Prime-Condition Figure 2. Duration
of the Synaptic
30
60
Interval Priming
120
(min)
Effect
Priming stimulation enhanced associative LTD in the lateral path at all intervals examined, from 10 min to 2 hr between 5 Hz priming stimulation and associative conditioning. NONE = nonprimed associative conditioning (n = 10). The four primed groups (IO min, n = 5 [data taken from Figure ICI; 30 min, n = 5; 60 min, n = 4; 120 min, n = 4) all showed significantly greater LTD compared with the normal group, using a one-way analysis of variance (FC4,26) = 7.3, p < 0.001) and post-hoc Student NewmanKeuls test (asterisks; p < 0.05). Data are means and standard errors obtained 30 min after associative conditioning.
NONE Priming Figure 3. Frequency Specificity tation of Associative LTD
greater than 25%. Only 1 of 5 animals in the primed nonassociative condition and only 2 of 13 animals in the nonprimed associativecondition showed a similar degree of LTD. A remarkable feature of the priming effect was that low frequencies other than 5 Hz, but outside the range of hippocampal0 activity, were ineffective in enhancing LTD. The LTD subsequent to 15 Hz priming stimulation (-12.5% +_ 6.6%, n = 5) was not different from the LTD produced in the normal, nonprimed condition (-13%), while 1 Hz stimulation resulted in a modest, but still nonsignificant, increase in LTD (-23.7% + 8.8%, n = 7; Figure 3). No changes in the amount of LTP produced in the medial path were observed either at the different time intervals following the 5 Hz priming, or forthedifferent frequenciesof lateral path priming. This indicates that low frequency priming of heterosynaptic afferents does not affect the induction of homosynaptic LTP.
5
1
Frequency of Synaptic
15 (Hz)
Priming
for the Facili-
Priming stimulation (5 Hz) significantly enhanced associative LTD (asterisks; tlzs, = 25.4; p < 0.05; n = 18) in the lateral path compared with the normal, nonprimed group (NONE, n = IO). Priming stimulation of 1 Hz (n = 7) and 15 Hz (n = 5) did not have a significant influence on LTD (one-way analysis of variance: [F,,39r = 8.4, p < 0.0011 and post-hoc Student Newman-Keuls test: p > 0.05). Medial path LTP was unaffected by priming stimulation delivered to the lateral path at any frequency. Data are presented as mean f standard error percent change, as measured 25-30 min after conditioning. These findings indicate that the degree of medial path LTP and the degree of lateral path LTD are poorly correlated.Thedataforthenonprimedgroupandthe5Hzgroup were taken from Figures IF and 2 (collapsed across all time points), respectively.
NMDA Receptor Involvement in Associative LTD Since the priming stimulation specifically facilitated the LTD arising from negatively correlated activity, this may reflect the induction of associative LTD, which could be mechanistically different from, and additive to, the normally occurring nonassociative LTD. To test this hypothesis, we took advantage of the fact that the two types of LTD have been shown to be
NWKNl a2
MEDIAL
401
30 1
-20 ! -15
'
1 -5
5 Time
20
25
15
(min)
LATERAL
1
lon~c--o -.-. -.- ._._.-.-.-.-.-.-._-._.-._._.-.-.. _._._._._._
04 - ‘-‘-’
l +. '8.
-lO- +-
0 %
h+j--
L --
L 5.
-2o-30 -40 ! -15
--cl--
CPP BEFORE PRIMING
-
CPP AFTER PRIMING
1 -5
5 Time
Figure
4. NMDA
Receptor
Antagonism
15
25
(min) with
CPP
CPP (IO mgikg i.p., Cambridge Research Biochemicals) given 2 hr prior to 5 Hz priming stimulation effectively blocked associative LTD in the lateral path (CPP BEFORE PRIMING, closed circles, n = 4). Significant associative LTD (two-way ANOVA, group main however, when CPP effect, Fci,6)= 10.12, p < 0.01) was observed, was given just after priming stimulation (CPP AFTER PRIMING; open circles, n = 4). Medial path LTP was reduced in both conditions. Note the gradually incrementing LTD when the CPP was given after priming condition. A similar trend is apparent for associative LTD in Figure IC. Data are mean and standard error for 5 min epochs following associative conditioning.
pharmacologically dissociable: nonassociative LTD is prevented by NMDA receptor antagonism (Abraham and Wickens, 1991; Desmond et al., 1991; Christie and Abraham, 1992), but associative LTD is not (Stanton and Sejnowski, 1989; Chattarji et al., 1989). As expected from these prior experiments, when the competitive NMDA receptor antagonist CPP (3-[(+)-2carboxypiperazin-byI]-propyl-I-phosphonic acid; 10 mglkg i.p.) was given after priming stimulation, but before the conditioning stimulation, therewas no LTD in the lateral path following the nonassociative para-
digm (5.5% k 7.6%, n = 3). In contrast, a significant, although reduced, LTD developed following the associative paradigm (-18.9% i 5.8%; tc3) = 3.2; p < 0.05; Figure 4, CPP AFTER PRIMING). As medial path LTP was reduced under both conditions, the LTD did not result from a failure to antagonize postsynaptic NMDA receptors. CPP, given after1 Hz priming stimulation, completely blocked subsequent LTD induction (-2.0% + 8.4%, n = 3), suggesting that following 1 Hz activity only NMDA-dependent, nonassociative LTD is induced. To test whether priming stimulation itself involved NMDA receptor activation, CPP was administered prior to the 5 Hz priming stimulation. Under this condition, the associative paradigm was no longer effective at eliciting LTD (4.6% + 8.0%, n = 4; Figure 4, CPP BEFORE PRIMING): These results indicate that 5 Hz stimulation initiates priming mechanisms that are coupled to NMDA receptor activation, but once the relevant cellular machinery is activated, the subsequent induction of associative LTD is independent of NMDA receptor activation. Taken together, our data support the hypothesis that associative and nonassociative LTD are independent and additive. The 35% LTD seen after 5 Hz priming in drug-free animals appears to consist roughly of 15% nonassociative LTD (i.e., the NMDA-dependent LTD seen in naive pathways) and 20% associative LTD (as induced by theassociative stimulus protocol when CPP follows priming stimulation). Specificity of the Priming Effect for Associative LTD To test whether priming stimulation facilitates synaptic plasticity in general, 5 Hz priming stimulation was delivered to the lateral path prior to the induction of LTP in this pathway. In this situation, priming stimulation significantly reduced (tcsI = 25; p < 0.05), by about 50%, the amount of lateral path LTP generated (normal LTP: 47.8% F 7.7%, n = 6; LTP after priming: 22.6% i 7.7%, n = 5). This effect may reflect an interference with the mechanisms involved in the induction of LTP, or a concurrently elicited associative LTD that masks the true extent of the LTP. In either case, it is apparent that LTP is not facilitated, indicating that the priming activity specifically facilitates associative LTD. The reduction in lateral path LTP is specific to priming of the lateral path, as medial path priming had no effect on lateral path LTP (49.9% f lO.l%, n = 5). Discussion The present results provide evidence that two pharmacologically distinct forms of LTD can be elicited separately, and in combination, within a single monosynaptic pathway. Nonassociative LTD is activity independent and can occur at synapses that are silent during strong postsynaptic depolarization elicited by other inputs. It involves at least the activation of the NMDA subtype of glutamate receptor by these active inputs (Abraham and Wickens, 1991; Desmond et al.,
Priming of Associative 83
Depression
1991; Christie and Abraham, 1992). It is not yet clear how signals travel to the silent synapses to trigger LTD induction there, but it is possible that the spread of strong depolarization across the postsynaptic membrane and the subsequent activation of synaptically located voltage-sensitive calcium channels are sufficient for this purpose (Lisman, 1989; Wickens and Abraham, 1991). In contrast, associative LTD is activity dependent in at least two distinct ways. First, it was produced in the present study only when a negative temporal correlation between presynaptic activity and strong postsynaptic depolarization was established. This temporal relationship seems crucial to the induction of associative LTD in areas CA1 and CA3 of the hippocampus. Pathways that exhibited LTD with the negativelycorrelated stimuli in these regions were found to exhibit associative LTP when the stimuli were administered so as to coincide (Stanton and Sejnowski, 1989; Chattarji et al., 1989). We are currently investigating the necessity of establishing the negative temporal correlation, to ascertain whether the LTD is reliant on its establishment, or requires only presynaptic activity per se. Second, prior synaptic activation at 8 rhythm periodicity primes those synapses for the subsequent induction of associative LTD and may also mask or interfere with the development of LTP and nonassociative LTD. Failure to prime the synapses results in the appearance of only nonassociative, NMDAdependent LTD, even when negatively correlated stimuli are administered. It is important to note that with these stimulation protocols, 5 Hz activity in the lateral path alone is insufficient to induce LTD, even when the stimuli are administered repeatedly. Associative LTD in the dentate gyrus is NMDA independent (as in areas CA1 and CA3), but NMDA receptor activation is necessary for the initial priming step. That the priming effect lasts for at least 2 hr suggests that NMDA receptor activation may be coupled to second messenger pathways (e.g., East and GarthWaite, 1991) necessary for maintainingthe synapses in a state of readiness for subsequent LTD. The recent demonstration that associative LTD in region CA1 may be blocked by inhibiting postsynaptic phosphoinositide turnover mediated by metabotropic glutamate receptor activation (Stanton et al., 1991) may be relevant to this hypothesis. It is not clear, however, whether priming activity is necessary for associative LTD in area CAI, nor whether AP3 will block associative LTD in the dentate gyrus. These questions await further investigation. It is intriguing that the critical frequency for the priming effect lies within the frequency range of the naturally occurring8 rhythm in the hippocampal electroencephalogram. 8 rhythm activity has been found to correlate with exploratory behaviors in rodents and has been suggested to reflect neuronal participation in a variety of cognitive functions including arousal, anxiety, sensory processing, and learning (Berry and Thompson, 1978, 1979; Bland, 1986; McNaughton,
1991). Of particular interest here is the possibility that neural activity in the hippocampus at 0 rhythm frequencies plays an important function in learning mechanisms. Recently Buszhki and Gage (1991) presented a two stage theory of hippocampal synaptic plasticity: a weak labile trace is set up by 8 activity occurring during exploration, and this trace is subsequently converted into a stronger, long-lasting form by patterned activity and postsynaptic depolarization at the termination of exploration. Although this model concentrates on LTP induction, the general similarity between it and the rules for associative LTD induction in the dentate gyrus are striking. Thus, our findings give substance to the model’s concept that 8 activity not only influences the current information being processed in the hippocampus, but also has a longer lasting impact on the magnitude and selectivity of synaptic modifications arising from subsequent patterns of activity. Experimental Procedures Adult male Sprague-Dawley rats (250-500 g) were anesthetized with sodium pentobarbital(65 mg/kg) and placed in a stereotaxic apparatus. Supplemental doses of pentobarbital were given as required to maintain a surgical plane of anesthesia. Rectal temperature was maintained at 37OC + 0.5OC. A 75 pm stainless steel extracellular recording electrode was placed in the dentate hilus 3.5 m m posterior and 2.5 m m lateral to bregma. Two 125 pm monopolar stimulating electrodes were placed separately in the ipsilateral medial perforant path (4.0 m m lateral to lambda) and the lateral perforant path (5.0 m m lateral to lambda). Recordings wereamplified and filtered at 0.1 Hz and 3.0 kHz half-amplitude. Prior to each experiment, lack of cross-facilitation with pairedpulse stimuli was used to ensure adequate separation (>90%) of the two sets of fibers being stimulated. A spatiotemporal summation test (McNaughton and Barnes, 19n; Abraham and Goddard, 1983) was also used to verify that the two stimulated pathways converged onto a common population of granule cells. For acceptance, the population spike in response to the stimulation of the two pathways together was required to be at least twice the amplitude of the sum of the spikes elicited by separate stimulation of the pathways. Test pulses of fixed amplitude and duration (0.1 ms) were delivered alternately to the medial and lateral perforant paths at 0.1 Hz. The initial slope of the recorded EPSPs was measured for each response beginning 15 min before conditioning stimulation and lasting at least 30 min after. EPSP slope values were averaged across 5 min epochs, and all data in the text are given as mean f standard error percent change from baseline, as observed 30 min following conditioning.
Acknowledgments We thank Drs. D. Bilkey, C. Darlington, N. McNaughton, and J. Wickens for critical readings of an earlier version of this manuscript. This research was supported by grants from the New Zealand Health Research Council and the Human Frontiers in Science Program to W. C. A. B. R. C. was supported by a Canadian NSERC postgraduate fellowship. The costs of.publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received
February
19, 1992, revised
April
20, 1992.
References Abraham,
W. C., and Goddard,
G. V. (1983). Asymmetric
relations
m 84
between homosynaptic long-term potentiation and heterosynaptic long-term depression. Nature 30.5, 717-719.
processes underlying learning Sci. USA 86, 9574-9578.
Abraham, W. C., and McNaughton, N. (1983). Differences in synaptic transmission between medial and lateral components of the perforant path. Brain Res. 303, 251-269.
McNaughton, N. (1991). The role of hippocampai RSA: spatial memory versus anxiety. In Memory Mechanisms: A Tribute to G. V. Coddard,W. C. Abraham, M. C. Corballis, and K. G. White, eds. (CITY: New Jersey: Lawrence Erlbaum Associates), pp. 129148.
Abraham, W. C., and Wickens, J. R. (1991). Heterosynaptic term depression is facilitated by blockade of inhibition CA1 of the hippocampus. Brain Res. 546, 336-340.
longin area
Bear, M. F., Cooper, L. N., and Ebner, F. F. (1987). A physiological basis for a theory of synapse modification. Science 237, 42-48. Berry, S. D., and Thompson, R. F. (1978). Prediction rate from the hippocampal electroencephalogram. 1298-1300.
of learning Science 200,
Berry, S. D., and Thompson, R. F. (1979). Medial septal lesions retard classical conditioning of the nictitating membrane response in rabbits. Science 205, 209-211. Bland, 8. H. (1986). The physiology campal formation theta rhythms.
and pharmacology of hippoProg. Neurobiol. 26, 199-218.
Bliss, T. V. P., and Lomo, T. (1973). Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. J. Physiol. 232, 331-356. Brown, T. H., Kairiss, E. W., and Keenan, C. 1. (1990). Hebbian synapses: biophysical mechanisms and algorithms. Annu. Rev. Neurosci. 73, 475-511. Buszaki, G., and Gage, F. H. (1991). Long-term potentiation: does it happen in the normal brain? When and how? In MemoryMechanisms: A Tribute to G. V. Goddard, W. C. Abraham, M. C. Corballis,andK.G. White,eds.(CITY,NewJersey: LawrenceErlbaum Associates), pp. 79-104. Chattarji, S., Stanton, P. K., and Sejnowski, T. J. (1989). Commissural synapses, but not mossy fiber synapses, in the hippocampal field CA3 exhibit associative long-term potentiation and depression. Brain Res. 495, 145-150. Christie, B. R., and Abraham, W. C. (1992). NMDA-dependent heterosynaptic long-term depression in the dentate gyrus of anaesthetized rats. Synapse. 70, 1-6. Desmond, N. L., Colbert, C. M., De, X. Z., and Levy, W. B. (1991). NMDA receptor antagonists block the induction of long-term depression in the hippocampal dentate gyrus of the anesthetized rat. Brain Res. 552, 93-98. East, S. J., and Garthwaite, J. (1991). NMDA receptor activation in rat hippocampus induces cyclic CMP formation through the t-arginine-nitric oxide pathway. Neurosci. Lett. 123, 17-19. Harris, E. W., Ganong,A. H., and Cotman, C. W. (1984). Long-term potentiation in the hippocampus involves activation of N-methyl-o-aspartate receptors. Brain Res. 323, 132-137. Hashemzadeh-Gargari, H., Colbert, C. M., and Levy, W. 5. (1991). Refiningthetemporaldefinitionofanassociationattheneuronal level using long-term potentiation and long-term depression in the dentate gyrus. Neurosci. Lett. 726, 37-40. Hebb, D. 0. (1949). Organization Wiley & Sons).
of Behavior
(New York: John
Hopfield, J. J., and Tank, D. W. (1986). Computing circuits: a model. Science 233, 625-633.
with
neural
Kelso, S. R., Canon& A. H., and Brown, T. H. (1986). Hebbian synapses in hippocampus. Proc. Natl. Acad. Sci. USA 83, 53265330. Kohonen, T. (1984). Self-Organization (Berlin: Springer-Verlag).
and Associative
Levy, W. B., and Steward, 0. (1979). Synapses ory elements in the hippocampal formation. 245.
as associative memBrain Res. 175,233-
Levy, W. B., and Steward, 0. (1983). Temporal contiguity ments for long-term associative potentiation/depression hippocampus. Neuroscience 8, 791-797. Lisman,
J. (1989). A mechanism
for the Hebb
Memory
requirein the
and the anti-Hebb
and memory.
Proc. Natl, Acad.
McNaughton, B. L., and Barnes, C. A. (1977). Physiological identification and analysis ofdentategranule cell responses to stimulation of the medial and lateral perforant pathways in the rat. J, Comp. Neural. 775, 439-454. Sejnowski, T. Jo, and Tesauro, G. (1989). The Hebb rule for synaptic plasticity: algorithms and implementations. In Neural Models of Plasticity, J. H. Byrne and W. 0. Berry, eds. (New York: Academic Press, Inc.), pp. 94-103. Stanton, P. K., and Sejnowski, depression in the hippocampus Nature 339, 215-218.
T. J. (1989). Associative long-term induced by Hebbian covariance.
Stanton, P. K., Chattarji, S., and Sejnowski, T. J. (1991). 2-Amino..3phosphonopropionic acid, an inhibitor of glutamate-stimulated phosphonoinositide turnover, blocks induction of homosynaptic long-term depression, but not potentiation, in rat hippocampus. Neurosci. Lett. 727, 61-66. Wickens, J. R., and Abraham, W. C. (1991). The involvement of L-type calcium channels in heterosynaptic long-term depression in the hippocampus. Neurosci. Lett. 730, 128-132. Wigstrom, H., Custafsson, B., Huang, Y.-Y., and Abraham, W. C. (1986). Hippocampal long-term potentiation is induced by pairing single afferent volleys with intracellularly injected depolarizing current pulses. Acta Physiol. Stand. 726, 317-319.