Although the neocortex in awake, adult animals is resistant to the induction of long-term potentiation (LTP), synaptic potentiation may be enhanced by rhythmic patterns of activation that evoke short- term synaptic facilitation effects. The effectiveness of stimulation patterned after the theta (4–12 Hz) EEG rhythm for the induction of LTP of sensorimotor cortex responses to corpus callosum stimu- lation was assessed in vivo by inducing LTP using either high- frequency (300 Hz) trains or paired trains delivered at a 100 ms (10 Hz) interval. High-frequency trains caused a reduction of the early field potential component, reflecting a potentiation of direct layer V activation, and a potentiation of the late component, reflecting enhanced polysynaptic activation in layer V. Paired trains resulted in a much larger potentiation of polysynaptic responses than was observed following 300 Hz trains. To determine if short-term facilitation effects contributed to the enhanced LTP induction by theta-patterned trains, facilitation effects induced by the trains were challenged with NMDA receptor antagonists. NMDA-receptor antagonism reduced responses to single pulses, and also reduced facilitated responses evoked by theta-patterned stimulation. The effectiveness of theta-patterned stimulation for the induction of LTP of layer V polysynaptic responses is therefore likely due to frequency-dependent synaptic facilitation effects that enhance NMDA receptor activation.
Memory consolidation is thought to occur through the strength- ening of synaptic connections and the formation of cortical cell assemblies (Hebb, 1949). Long-term synaptic potentiation (LTP) is a widely studied cellular model of memory formation and is easily induced in hippocampal pathways by trains of stimulation pulses that cause intense postsynaptic depolarization and activation of NMDA glutamate receptors (Malenka and Nicoll, 1999). LTP in the neocortex, which is thought to relate more directly to memory consolidation (Bear and Kirkwood, 1993; Buonomano and Merzenich, 1998; Rioult-Pedotti et al., 2000), is much more difficult to induce. Acute preparations and juvenile animals have typically been used to examine neocortical LTP in vivo (Wilson and Racine, 1983; Crair and Malenka, 1995; Heynen and Bear, 2001), and the reliable induction of LTP in vitro often requires use of GABAA receptor antagonists to enhance postsynaptic excitability (Kirkwood and Bear, 1994; Hess et al., 1996).
In the hippocampal formation, LTP induction is promoted by intense stimulation trains that mimic temporal patterns of activation observed during endogenous forms of population activity. Stimulation patterned after the theta (4–12 Hz) rhythm (Staubli and Lynch, 1987; Huerta and Lisman, 1995; Chapman and Racine, 1997; Chapman et al., 1998a; Perez et al., 1999; Morgan and Teyler, 2001), and sharp-wave bursts that occur during large amplitude irregular EEG activity [(Yun et al., 2002); see Chrobak and Buzsaki (Chrobak and Buzsaki, 1996)] induce LTP effectively. The induction of LTP by theta-patterned stimulation is due in part to frequency-dependent modulation of inhibitory mechanisms. For example, delivering a priming pulse 100–200 ms prior to a short train of pulses promotes LTP because the priming pulse leads to activation of presynaptic GABAB autoreceptors, and a resulting disinhibition during the train (Rose and Dunwiddie, 1986; Mott et al., 1993; Davies and Collingridge, 1996).
Synaptic facilitation effects associated with endogenous rhythmic activities may also promote neocortical synaptic plasticity (Singer, 1993; Steriade, 2001). Gamma (40–100 Hz) activity dominates cortical EEG activity in the awake rat, but thalamocortical interactions during awake immobility and sleep also generate large-amplitude ~9 Hz spindle waves that fall within the theta- (4–12 Hz) and alpha- (8–13 Hz) frequency bands (Semba and Komisaruk, 1984; McCormick and Bal, 1997). Although gamma-frequency stimulation of white matter recruits inhibitory mechanisms that restrict the spread of cortical activation, 10 Hz stimulation enhances activation within horizontal pathways (Contreras and Llinas, 2001). Horizontal layer V pathways are also activated by 10 Hz stimulation during the thalamocortical augmenting response (Castro-Alamancos and Connors, 1996a), and mediate a marked paired-pulse facilitation in vivo [100 ms interval (Chapman et al., 1998b)]. The strong facilitation of cortical responses by theta-frequency activation, and the dominant influence of inhibition at higher frequencies (Jefferys et al., 1996; Galarreta and Hestrin, 1998; Gibson et al., 1999), suggest that theta-frequency activity may contribute preferentially to inducing levels of postsynaptic depolarization required for enhancing synaptic strength. For example, high-frequency white matter stimulation is ineffective at inducing LTP in layer III of visual cortex slices unless bicuculline is applied concurrently, but theta-patterned trains induce LTP without pharmacological reduction of inhibition (Kirkwood and Bear, 1994). Theta-frequency trains also induce LTP effectively in horizontal layer II/III connections in sensori- motor (Rioult-Pedotti et al., 2000) and motor (Hess et al., 1996) cortex slices.
In the present study, we have evaluated the effectiveness of theta-patterned stimulation in the induction of LTP in corpus callosum inputs to layer V of the sensorimotor cortex in the awake, adult rat. LTP in this preparation requires spaced and repeated exposure to intense stimulation trains over a period of days (Trepel and Racine, 1998). Evoked field potential responses contain components generated by direct and polysynaptic activation of layer V, and therefore allow potentiation of poly- synaptic responses mediated by horizontal collaterals to be assessed in the intact brain (Chapman et al., 1998b). The effectiveness of theta-patterned stimulation was assessed by comparing the LTP induced by either high-frequency stimulation or by pairs of short four-pulse trains separated by the 100 ms period of the theta rhythm. Because short-term facilitation effects evoked during theta-patterned stimulation may promote LTP induction by enhancing NMDA receptor activation, the contribution of NMDA receptors to train-evoked responses was also assessed.
Materials and Methods
Male Long–Evans rats (300–400 g) were anesthetized with ketamine and xylazine (90 and 8 mg/kg, respectively) and placed in a stereotaxic frame. Bipolar, Teflon-coated stainless-steel twisted-wire electrodes (125 μm exposed tips) were implanted in the right corpus callosum (A = 2.0 mm, L = 2.0 mm, V = 2.6 mm from pia) and sensorimotor cortex (A = 2.0 mm, L = 4.0 mm, V = 1.8 mm). Tip separation was 0.5 mm for stimulating electrodes and 1.2 mm for recording electrodes. The vertical placements were adjusted to maximize monosynaptic layer V responses, and to minimize current thresholds. A stainless-steel jeweler’s screw in the left posterior parietal bone served as the ground electrode. Electrode leads were connected to gold-plated pins mounted in a connector, and the assembly was embedded in dental cement and anchored to the skull with screws. Animals were housed individually on a 12 h light–dark cycle and handled every 2–3 days during a ≥2 week recovery period. Experimental testing was conducted during the lights-on period.
Stimulation and Recording
Animals were habituated to a 30 × 40 × 30 cm Plexiglas chamber, and recordings were collected while animals were in a quiet, resting state. Biphasic, square-wave constant current stimulation pulses (0.1 ms duration) were delivered to the corpus callosum using a computer-gated linear stimulus isolation unit (A-M Systems, Model 2200) regulated by a computer DAC channel (50 kHz). Both monopolar and differential evoked field potentials were filtered (0.1 Hz to 5 kHz passband), amplified (A-M Systems, Model 1700), and digitized at 20 kHz (16 bit) for storage on computer hard disk (Datawave Tech., Experimenter’s Workbench).
Input/output tests were used to monitor changes in field potential responses. During each test, 10 evoked potentials were recorded and averaged at each of eight test-pulse intensities (100–1000 μA). The inter-pulse interval was 10 s and pulses were delivered in ascending order with respect to intensity. Input/output tests were conducted every 2 days during a 1 week baseline period and a 20 day LTP induction period. Further tests were conducted up to 4 weeks after LTP induction. Mean input/output curves were obtained after normalizing amplitudes to the average response to 1000 μA pulses in baseline tests.
Groups of animals were matched for response amplitude and morphology and LTP was induced using either high-frequency trains (eight-pulse 300 Hz trains; n = 10), or pairs of trains (two four-pulse, 300 Hz trains separated by 100 ms; n = 11). Ten trains, or pairs of trains, were delivered daily for 20 days to induce LTP. Pulse intensity was 800 μA and the inter-train interval was 10 s. Control animals received no trains (n = 8). In addition to mean input/output curves that illustrate changes in responses at all intensities, the development of potentiation effects was also plotted using responses evoked by 600 μA pulses.
NMDA Receptor Antagonism
The effects of NMDA receptor antagonists on evoked responses and synaptic facilitation effects were assessed using several tests conducted before and after injection of either MK-801 (0.1 mg/kg, i.p.; n = 5) or ketamine [30 mg/kg plus 6 mg/kg after 30 min; n = 9 (Salami et al., 2000)]. The onset of drug effects was monitored by delivering single 800 μA pulses every 20 s for 10 min before, and 20 min after, drug injection. Input/output tests were used to assess NMDA-receptor activation in response to single pulses. Responses evoked by trains used to induce LTP were also monitored before and after drug injection (with the exception that theta-patterned trains consisted of a series of six, rather than two, four-pulse trains). The contribution of NMDA receptors to the facilitation evoked by less intense, prolonged theta-frequency stimulation was also assessed using 2 s trains of 800 μA pulses at 10 Hz.
Amplitudes of field potential components were measured at the peaks of the antidromic spike (2 ms), early monosynaptic (6–8 ms), and late polysynaptic (17–18 ms) components of field potentials in superficial, deep, and differential recordings. Mixed design ANOVAs and t-tests were used to analyze changes in peak amplitudes.
Histological analysis with cresyl violet showed that electrode placements were close to targets in the corpus callosum and sensorimotor cortex, with superficial tips of bipolar recording electrodes in layers II to IV and deep tips in layers V and VI. Evoked field potentials (e.g. Fig. 1A) contained an initial deep-negative spike (0.99 ± 0.18 mV peak amplitude), an early superficial-negative component (1.62 ± 0.22 mV), and a late superficial-negative component (1.23 ± 0.19 mV). These com- ponents reflect antidromic activation of deep layer V cells, fast monosynaptic activation in superficial layer V, and later, polysynaptic activation of layer V (Chapman et al., 1998b).
Similar to previous studies (Trepel and Racine, 1998; Froc et al., 2000), changes in evoked responses were slow to develop during the 20-day LTP induction period, and were largest after ~15 days of tetanization. The amplitude of the initial antidromic spike was increased in a several animals in each group, but no significant potentiation of this component was observed [F(21,504) = 1.18, P = 0.26] and changes in the initial spike were not correlated with changes in the amplitudes of subsequent components of field potential responses. High-frequency stimulation caused a large reduction in the amplitude of the early component [42.6 ± 15.9% of baseline; n = 10; F(21,252) = 6.36, P < 0.001; Fig. 1A2,B], but animals that received paired trains showed a smaller, non-significant reduction in the early component [87.7 ± 12.2% of baseline; n = 11; F(21,189) = 1.05, P = 0.40; Fig. 1A1,B].
Animals that received paired trains showed a much larger increase in the amplitude of the late polysynaptic component relative to animals that received 300 Hz trains (Figs 1A1,C). Responses to 600 μA pulses increased to 220.8 ± 57.5% of baseline following paired trains [F(21,336) = 4.02, P < 0.001], and to 115.9 ± 26.1% of baseline following 300 Hz trains [F(21,315) = 1.62, P < 0.05]. Mean input/output curves show the differences between the groups most clearly (Fig. 1C1). Although the potentiation induced by high-frequency trains was restricted to intermediate test-pulse intensities, paired trains caused a potentiation of the late component at all suprathreshold intensities [F(147,2205) = 1.48, P < 0.001].
Following LTP induction by 300 Hz trains, the amplitude of the early component is affected by changes in potentials in both superficial and deep recording sites (Chapman et al., 1998b). Monosynaptic inputs to layer V evoke superficial-negative and deep-positive potentials that are generated by a current sink in upper layer V, but repetitive population spikes are also evoked deep in layer V, and the resulting deep-negative potentials reduce the amplitude of the differentially recorded early component. Reductions in the differential early component following 300 Hz stimulation result largely from increased deep-negative spiking that masks potentiation of monosynaptically activated super- ficial-negative currents (Chapman et al., 1998b). This reduction in the amplitude of the early component is analogous to the reduction in the peak amplitude of dentate gyrus fEPSPs that results from potentiation of the population spike following tetanization of the perforant path.
In contrast to the reduction in the early component observed following 300 Hz stimulation (Fig. 1B1), paired trains did not cause a significant reduction in the early component. Monopolar recordings were therefore examined to investigate the small change in the early component following paired trains. Monopolar recordings showed negative shifts in both superficial and deep recording sites following paired trains, resulting in little net change in the amplitude of the early component (Fig. 2). Paired trains caused a significant increase in the superficial- negative potential [125.4 ± 16.9%; F(147,1323) = 1.31, P < 0.01; Fig. 2B1], and a growth in deep-negative spiking that reduced the deep-positive potential [51.9 ± 38.0%; F(147,1029) = 1.32, P < 0.01; Fig. 2B2]. Increases in the superficial-negative potential resulting from potentiation of monosynaptic inputs to layer V were therefore masked in differential recordings by concurrent increases in repetitive spiking deep in layer V.
Increases in the amplitude of the early superficial-negative potential were less reliable following 300 Hz stimulation, and were observed in only three of nine animals (106.9 ± 18.2%), and the deep-positive potential was markedly reduced by increases in repetitive spiking [24.3 ± 32.1%; F(147,588) = 1.41, P < 0.01; not shown]. The large decrease in the early component following 300 Hz stimulation therefore resulted mainly from increased repetitive firing in deep layer V neurons.
NMDA Receptor Antagonism
Because LTP in this preparation is dependent on NMDA glutamate receptor activation (Trepel and Racine, 1998), the greater LTP of polysynaptic layer V responses following paired trains may have resulted from enhanced NMDA receptor activation. To investigate this idea, a number of tests were con- ducted before and after administration of the NMDA receptor antagonists MK-801 (0.1 mg/kg) or ketamine (30 mg/kg). The onset of the effects of MK-801 was slower than that of ketamine, but similar changes in the early and late components were observed 15–20 min after administration of either drug (Fig. 3A,B). Drug effects did not differ in further tests, and data were therefore combined. In responses to single pulses in input/output tests, NMDA receptor antagonists had no significant effect on the initial antidromic spike [F(1,13) = 3.68, P = 0.08], produced a moderate reduction in the early component [79.5 ± 3.5% of baseline at 600 μA; F(1,13) = 21.06, P < 0.01], and caused a much larger reduction of the late component [15.7 ± 17.7% of baseline; F(1,13) = 11.17, P < 0.01]. The reduction in the late component could be due partly to an effect on monosynaptically activated NMDA receptors, but the size of the reduction strongly suggests that NMDA receptors also contribute significantly to polysynaptic responses (Fig. 3C).
Train-evoked responses were recorded before and after drug administration to assess the degree of NMDA receptor activation evoked by paired trains and high-frequency trains. Responses to 300 Hz trains were reduced at latencies corresponding to both the early [81.9 ± 4.2% of pre-drug amplitude; t(13) = 3.08, P < 0.01] and late [78.4 ± 6.2%; t(13) = 3.23, P < 0.01] field potential components (Fig. 4A). Similar reductions in early and late components were also observed in response to the first of the four-pulse theta-patterned trains [early: 81.3 ± 6.5%; t(13) = 1.98, P = 0.07; late: 83.6 ± 3.3%; t(13) = 4.06, P < 0.01].
Responses to subsequent theta-patterned trains were also reduced by NMDA receptor antagonists (Fig. 4B). In pre-drug tests, the early component evoked by the second train was reduced to 70.4 ± 15.3% of the first response [t(13) = 4.11, P < 0.01], and the late component was strongly facilitated to 142.9 ± 7.4% of the first response [t(13) = 4.24, P < 0.01; Fig. 4C1,D]. The early component continued to decrease during subsequent trains as spiking activity was increased, and the facilitation of the late component also declined gradually (Fig. 4B,D). NMDA receptor antagonists reduced the early and late field potential components evoked by theta-patterned trains. The early component response to the second train was reduced to 46.0 ± 15.4% of pre-drug levels [F(1,13) = 7.36, P < 0.05] and the late component was reduced to 90.7 ± 5.8% of pre-drug levels [F(2,26) = 9.78, P < 0.01; Fig. 4C,D]. Responses to subsequent trains were also reduced, but the sixth response did not differ from pre-drug recordings.
To determine if NMDA receptor activation is maintained during theta-frequency stimulation with single pulses, 2 s trains of pulses at 10 Hz were also delivered. The early component was reduced throughout the train in pre-drug tests [to 44.5 ± 18.8% of baseline; F(44,616) = 4.15, P < 0.01], and NMDA receptor antagonism did not reduce the early component amplitude during the train [F(1,14) = 0.02, P = 0.90; Fig. 5A,B1]. In con- trast, the late component was strongly facilitated during the first several pulses in pre-drug tests [306.7 ± 45.0% of baseline; F(44,616) = 31.34, P < 0.01], and the facilitation was maintained at the end of the train (163.3 ± 28.3%). NMDA receptor antagonists reduced the late component throughout the train, and responses were at 78.8 ± 13.0% of pre-drug levels at the peak of the facilitation effect [F(1,14) = 16.91, P < 0.01; Fig. 5A,B2].
These experiments demonstrate that paired trains delivered at an interval corresponding to the theta rhythm are highly effective at inducing LTP in the sensorimotor cortex of the awake rat. Similar to previous findings, daily tetanization was required to induce LTP, and potentiation effects were maintained over days to weeks (Trepel and Racine, 1998; Chapman et al., 1998b; Froc et al., 2000). Paired trains and high-frequency trains, however, caused different patterns of changes in the early and late field potential components. High-frequency trains caused a large reduction in the early component, and a moderate increase in the late component, while paired trains caused smaller changes in the early component, and a much larger potentiation of the late polysynaptic component. Increases in repetitive spiking in deep layer V make the potentiation of monosynaptic inputs to superficial layer V difficult to evaluate (Chapman et al., 1998b), but monopolar recordings suggest that differences between groups in changes in the early component resulted from greater potentiation of deep-negative spiking following high-frequency stimulation, and greater potentiation of monosynaptic inputs following paired trains (Fig. 2). The greater LTP of the late, polysynaptic component following paired trains was marked (Fig. 1C1), and indicates that trains delivered at the period of the theta rhythm preferentially enhance LTP of polysynaptic responses mediated by horizontal layer V collaterals.
The facilitation of train-evoked polysynaptic responses by theta-patterned stimulation, and the reduction of these responses by NMDA receptor antagonists (Fig. 4), suggests that the effectiveness of theta-patterned trains for inducing LTP of polysynaptic responses is due to the facilitation of activity in horizontal layer V collaterals that enhances NMDA glutamate receptor activation (Chapman et al., 1998b; Trepel and Racine, 1998). Polysynaptic responses were also facilitated during 10 Hz stimulation with single pulses (Fig. 5), and similar facilitation effects have been observed in the sensorimotor cortex during theta-frequency stimulation in other preparations. Activation in layer V is facilitated during the thalamocortical augmenting response (Castro-Alamancos and Connors, 1996a), and during white matter (Wu et al., 1999; Contreras and Llinas, 2001) and unitary (Reyes and Sakmann, 1999) stimulation in cortical slices. The sensorimotor cortex is therefore highly responsive to theta-frequency activation, and we have shown here that theta-patterned tetanization greatly enhances the potentiation of polysynaptic layer V responses.
LTP of the Early, Monosynaptic Component
The LTP effects observed following high-frequency stimulation were similar to those observed previously using 300 Hz trains (Trepel and Racine, 1998), and involved a reduction in the early field potential component (Fig. 1A,B). The early component is affected both by monosynaptic inputs to superficial layer V and by repetitive spiking in deep layer V neurons. Reductions in the early component following LTP result from increased spiking that masks smaller increases in the superficial-negative potential generated by monosynaptic inputs (Chapman et al., 1998b). Large decreases in the early component were observed following high-frequency stimulation because increases in repetitive spiking were greater than the potentiation of monosynaptic inputs.
A more balanced potentiation of monosynaptic inputs and repetitive spiking was observed following paired trains. Potentiation of the superficial-negative potential was matched by growth in repetitive spiking, resulting in little net change in the early component (Fig. 2). Thus, although potentiation of mono- synaptic inputs to layer V are difficult to assess directly because of concurrent changes in cell firing, the smaller change in the early component following paired trains versus 300 Hz trains appears to reflect potentiation of monosynaptic inputs paired with a more moderate increase in spiking.
LTP of the Late, Polysynaptic Component
While LTP of the late field potential component induced by high-frequency trains was similar to that observed in previous studies, paired trains caused a much larger potentiation of polysynaptic layer V responses. Although potentiation follow- ing 300 Hz trains was restricted to intermediate test-pulse intensities, paired trains induced LTP at all supra-threshold intensities (Fig. 1C1). The LTP induced by paired trains could have resulted from strengthening of synapses within horizontal layer V collaterals, but is also likely to be partly due to potentiation of monosynaptic inputs to layer V neurons, and a resulting increase in firing of horizontal collaterals (Bindman et al., 1988; Baranyi et al., 1991; Kitagawa et al., 1997). The greater potentiation of the early negative component following paired trains was not marked, however, suggesting that the greater potentiation of polysynaptic responses following paired trains resulted mainly from greater synaptic potentiation in horizontal pathways (Aroniadou and Keller, 1993; Jagodzinski and Hess, 2001) rather than from increased monosynaptic drive. Axonal and/or dendritic growth in horizontal projections could contribute to the increase in the strength and/or spatial extent of polysynaptic layer V activation (Klintsova and Greenough, 1999; Teskey et al., 2002). This hypothesis is consistent with the gradual induction of LTP in this preparation, but future studies will be required to directly measure LTP within synapses of horizontal collaterals, and to characterize possible alterations in axonal and dendritic morphology in layer V.
LTP Induction by Theta-patterned Stimulation
The greater LTP of polysynaptic responses induced by paired trains was likely mediated by synaptic facilitation effects in horizontal collaterals that caused enhanced NMDA receptor activation. Similarly, frequency facilitation of dentate gyrus responses during 10–20 Hz perforant path stimulation causes sufficient levels of polysynaptic activation in the CA3 region for the induction of (non-NMDA-dependent) LTP of mossy fiber synapses onto CA3 neurons (Yeckel and Berger, 1998). This poly- synaptic potentiation effect is due both to the responsiveness of the dentate gyrus to theta-frequency stimulation, and to the convergence of dentate gyrus efferents within the CA3 region (Yeckel and Berger, 1998). In the present study, high-frequency trains were more intense than each of the four-pulse paired trains, but the late component was strongly facilitated during stimulation with paired trains (Fig. 4B). NMDA receptor antagonists reduced the facilitated polysynaptic responses evoked by paired trains (Fig. 4D), suggesting that paired trains recruit NMDA receptor activation in responses mediated by horizontal collaterals. Similarly, short-term facilitation effects also contribute to LTP induction by enhancing postsynaptic depolarization in the hippocampus (Rose and Dunwiddie, 1986; Davies and Collingridge, 1996; Yeckel and Berger, 1998), somatosensory cortex (Castro-Alamancos and Connors, 1996b) and entorhinal cortex (Chapman and Racine, 1997). Unlike monosynaptic pathways, however, facilitation of sensorimotor cortex responses causes widespread activation of an extensive network of horizontal layer V collaterals (Landry et al., 1980; Aroniadou and Keller, 1993). The facilitation of the late component indicates that there is stronger activation across a larger cortical region during the second train. The spatial extent of cortex that is activated by paired trains is therefore likely to be much greater relative to high-frequency trains that do not induce a similar facilitation. This may result in increased cooperativity among horizontal collaterals in activating synaptic targets distal to the stimulation site, and the induction of potentiation across a much more widespread cortical region.
The effectiveness of paired trains for LTP induction is likely a result of the responsiveness of the sensorimotor cortex to afferent inputs in the theta-frequency range. Facilitation of polysynaptic responses was maintained during a 2 s train of pulses at 10 Hz, indicating that theta-frequency stimulation causes prolonged facilitation of layer V responses (Fig. 5). This facilitation effect is similar to the thalamocortical augmenting response in which low-frequency (~5–10 Hz) ventrolateral thalamic stimulation results in a strong enhancement of layer V responses (Castro-Alamancos and Connors, 1996a) that is partly mediated by NMDA receptor activation (Addae and Stone, 1987). Extensive spread of activation within superficial and deep layers has also been observed using a voltage-sensitive dye during 10 Hz white matter stimulation in sensorimotor cortex slices (Contreras and Llinas, 2001). Further, spindle waves, which are one of the most striking forms of spontaneous cortical population activity, occur at frequency of ~9 Hz. Spontaneous spindle activity is observed during both sleep and waking, and reflects thalamocortical interactions that dramatically enhance rhythmic cortical activation (Semba and Komisaruk, 1984; McCormick and Bal, 1997; Steriade, 2001).
Frequency-dependent alterations of inhibitory mechanisms also likely contributed to LTP induction. As in the hippocampal formation (Mott et al., 1993; Davies and Collingridge, 1996), paired-pulse facilitation in the cortex is mediated partly by activation of presynaptic GABAB autoreceptors that reduce inhibitory synaptic transmission (Deisz and Prince, 1989; Metherate and Ashe, 1994). Further, long-duration inhibitory postsynaptic potentials evoked during the augmenting response in layer V cause the activation of the hyperpolarization-activated inward cationic current (Ih) which strongly enhances cortical excitability (Castro-Alamancos and Connors, 1996c). Both of these mechanisms could enhance postsynaptic depolarization and promote NMDA receptor activation during paired trains. At higher frequencies of stimulation, however, inhibitory mechanisms restrict the spread of cortical activation along horizontal pathways. Gamma-frequency stimulation induces cortical activation that has a much more restricted spatial extent (Contreras and Llinas, 2001) that reflects the activation of inhibitory circuitry (Galarreta and Hestrin, 1998; Gibson et al., 1999).
The strong activation of horizontal cortical connections by theta-frequency activity, and the present finding that intense theta-frequency stimulation of the corpus callosum results in a large enhancement of polysynaptic layer V responses, suggests that endogenous low-frequency population activity may con- tribute to learning-related synaptic plasticity in the sensorimotor cortex. Plasticity in long-range inter-columnar connections is thought to contribute to normal learning and memory (Hebb, 1949), and facilitation effects induced during synchronous theta-frequency activity may play an important role in generating levels of postsynaptic depolarization required for long-term changes of synaptic strength in these connections.
This research was funded by grants from the Natural Sciences and Engineering Research Council of Canada, Fonds pour la Formation de Chercheurs et l’Aide à la Recherche, and the Canadian Foundation for Innovation.
Address correspondence to C.A. Chapman, Center for Studies in Behavioral Neurobiology, Department of Psychology, Concordia University, Montréal, Québec, Canada H3G 1M8. Email: email@example.com.