The Journal of Neuroscience, August 1995, 15(a): 5820-5830
Long-Term Potentiation Disrupts Auditory Gating in the Rat Hippocampus
Christine L. Miller,i~2 Paula C. Bickford, 1,3 Anne K. Wiser,* and Gregory M. Rosei,2,3 lDepattment of Pharmacology and *Neuroscience Training Program, University of Colorado Health Sciences Center, and 3Veterans Affairs Medical Center, Denver, Colorado 80262
The consequence of long-term potentiation (LTP) of hip- implicated as a mechanismunderlying learning and memory pocampal commissural inputs was investigated in an au- (Swanson et al., 1982; Teyler and Discenna, 1984; Morris, ditory gating paradigm. Auditory evoked potentials (AEPs) 1990). Alternatively, the information processingthat precedes were recorded in the CA3, region of the hippocampus of memory encodingmay require LTP (DeJongeand Racine, 1985; rats anesthetized with chloral hydrate. Two tones were de- Squire, 1987). livered 0.5 set apart; in this paradigm, the second AEP is Auditory information is part of the sensoryspectrum that the diminished compared to the first. Electrical stimulation was hippocampusacts upon (Berger et al., 1976; Rose, 1983; Bran- applied to hippocampal commissural fibers to generate kack et al., 1986; Bickford-Wimer et al., 1990; Edeline et al., field potentials and population spikes which were recorded 1990). This information relayed to the hippocampusfrom sev- at the same site as the AEPs. LTP of the commissural input eral brain regions, including the medial septum (Harrison et al., (initiated by three trains of 250 Hz/l set stimulation) was 1988; Miller and Freedman, 1993) and the auditory responsive associated with changes in the AEPs: on average, the re- region of the entorhinal cortex (Stafekhina and Vinogradova, sponse to the first tone decreased and the response to the 1975; Vayssettes-Courchay and Sessler, 1983; Foster et al., second tone increased, resulting in the disruption of au- 1988). ditory gating. When high-frequency stimulation of the com- In general, auditory processingcan be divided into two cat- missural input failed to result in LTP, no effect on the AEPs egories: (1) that occurring in the classical lemniscal system, was seen. If 3-(2-carboxypiperazin-4-yl)-propyl+-phos- which primarily involves signal frequency, intensity, and timing phonic acid (CPP; 6 mg/kg, i.p.), an antagonist to the NMDA discriminations (Kelly, 1991); and (2) that occurring in extra- subclass of glutamate receptors, was administered prior to lemniscal pathways, which exhibits plasticity to soundsof be- high-frequency stimulation, LTP induction was blocked havioral significance (Weinberger and Diamond, 1987; Wein- and AEPs were not affected. Finally, reversal of LTP, berger et al., 1990). Hippocampal processingis extra-lemniscal achieved by high-frequency stimulation of CA3 input that (Jirsa et al., 1992). Thus, learning changesthe responseof hip- was heterosynaptic to the particular commissural fibers at pocampal neuronsto auditory stimuli which have acquired be- which the LTP was originally generated, caused disrupted havioral significance(Berger et al., 1976; Edeline et al., 1990). auditory gating to return to normal. A model of reciprocal Since thesechanges must be the result of learning-relatedmod- LTP and heterosynaptic depression of commissural and ifications in the selective weighting of inputs to the hippocampal auditory input pathways is proposed to explain these find- neurons,changes in synaptic strengthinduced by artificial means ings. (e.g., LTP stimulation) might also be expected to modify hip- [Key words: long-term potentiation (LTP), plasticity, hip- pocampalprocessing of auditory stimuli. pocampus, CA3, auditory processing, sensory gating, rat, The type of hippocampalauditory processingwe have studied auditory evoked potentials] is shown in Figure 1, which illustrates a mid-latency (N40) The role of the hippocampusin the information processingre- evoked potential that can be recorded in the hippocampusof a quired for learning and memory has been extensively studied. It rat 40 2 10 msec after a 3 kHz, 70 dB tone and which shows is known that the hippocampusserves as a multi-modal asso- diminishedamplitude if an identical tone is presented500 msec ciation cortex, bringing together diverse sensory input (Swan- later. This phenomenonis termedauditory gating, and may relate son, 1983). Hippocampal neurons are responsive to complex to modelsof hippocampalfunction involving the filtering out of spatial cues, a characteristic which is consistentwith a role for irrelevant stimuli (Douglas, 1972; Solomon, 1977). Auditory the hippocampusin spatial learning (O’Keefe and Nadel, 1978; gating is reflective of processingin the central extra-lemniscal Kubie and Ranck, 1983; McNaughton et al., 1983). At the cel- pathway and is not observed in nuclei of the central lemniscal lular level, hippocampallong-term potentiation (LTP) has been pathway (Bickford-Wimer et al., 1990; Bickford et al., 1993). How might LTP interact with auditory gating in the hippo- campus?The impact of LTP has been difficult to determinein Received Dec. 28, 1994; revised Apr. 5, 1995; accepted Apr. 11, 1995. This work was supported by the VA Medical Research Service and USPHS terms of a specific in viva consequenceto an animal (Eichen- Grants MH-38321 and MH-44212. C.L.M. was supported by NIH Grant 1 T32 baum and Otto, 1993). Yet, assumingthat alterations in the HD07408-OlAl and a grant from the Stanley Foundation. strengthsof synaptic input occur in vivo, such changesshould Correspondence should be addressed to Dr. G. M. Rose, Medical Research Service (151), VAMC, 1055 Clermont Street, Denver, CO 80220. affect hippocampalfunction in some manner.In this study, we Copyright 0 1995 Society for Neuroscience 0270-6474/95/l 55820- 11$05.00/O investigated the effect of LTP on auditory gating of the N40 The Journal of Neuroscience, August 1995, 75(E) 5821
Conditioning Test TIC Response Response Ratio
0.21 Gating
200 uv / 50 ms p . *-” 0.72 Gating Lost 2nd tone Figure I. Examples of N40 potentials (marked by &is) recorded in CA3,, elicited by paired auditory stimuli that are separated by 500 msec (marked by arrows). ,The upper set of potentials is representative of baseline auditory gating. A T/C ratio of 0.5 is considered to be the level above which gating is lost (see Materials and Methods). The lower set of potentials illustrates a loss of auditory gating resulting from a decrease in the conditioning response amplitude and an increase in the test response amplitude. potential recorded in the hippocampus of the rat. In common Electrophysiology. The rats were secured in a stereotaxic apparatus, adjustments were made to ensure that the skull was flat, and after re- with LTP, auditory gating is a process that is reflected in extra- moval of small sections of the skull and dura, a recording electrode cellular potentials and can be measured electrophysiologically at (tungsten, coated with non-conducting varnish; 3 to 5 MR resistance) the site of LTP generation. The paradigm involves fewer dimen- was lowered into the hippocampus (from bregma, -3.8 mm AP, 3.6 sions than learning and memory studies and is more closely mm ML). A stimulating electrode (tungsten, coated with nonconducting associated with the substrate upon which LTP acts: excitatory varnish; 0.25 MR resistance) was lowered to one of three different stereotaxic locations to stimulate commissural input to the recording input to hippocampal pyramidal cells and interneurons. electrode site: (1) in the ventral hippocampal commissure (- 1.8 mm AP, 1.0 mm ML, and 2.7-2.9 mm DV; see Swanson, 1992), (2) in the Materials and Methods contralateral CA3 anterior to the plane of the recording electrode (- 1.8 Animals and surgery. Thirty-four adult male Sprague-Dawley rats mm AP, 1.0 mm ML, and 3.4-3.5 mm DV), or (3) in the contralateral (250-350 gm; Harlan Laboratories, Indianapolis, IN) were recorded dur- CA3 homotopic to the recording electrode site (-3.8 mm AP, 3.7 mm ing the course of this study. The animals were anesthetized with chloral ML. and 2.9-3.4 mm DV). Stereotaxic locations of recorded arouus of hydrate (400 mg/kg, i.p.) which had been prepared in a buffered solu- cells was used as an aid to correct placement of the recording ilect;ode. tion of 100 mM NaCl and 50 mu HEPES (final pH 7.2) and stored at The initial landmarks used were the field potential induced by com- 4°C until use. Chloral hydrate was selected as the anesthetic because it missural stimulation and the presence of neurons firing with the com- preserved the auditory response very well, whereas other anesthetics plex spike characteristic of hippocampal pyramidal cells, first encoun- tried (pentobarbital and urethane) did not. To extend the period of stable tered at the depth corresponding to the CA1 layer (usually 1.9-2.2 mm anesthesia, the metabolism of chloral hydrate to trichloroacetic acid was from the dura). Extracellular recordings of action potentials were am- blocked by the administration of pyrazole, an inhibitor of alcohol de- plified by a high-input impedance amplifier and then further amplified hydrogenase (Cooper and Friedman, 1958; Miller et al., 1993). Pyrazole and displayed on the oscilloscope. For the action potential recordings, was injected intraperitoneally at a loading dose of 200 mg/kg, with 30 the bandpass filter was set to limit the response to 300-10,000 Hz. mg/kg supplements at 30 min intervals. Below the CA1 pyramidal layer, the potentials evoked by stimulation The level of anesthesia was carefully monitored throughout the of the commissure were characteristic dendritic negativities until the course of the experiment, and was adjusted with supplemental doses of CA3 pyramidal layer was reached at depths generally > 2.9 mm. Com- chloral hydrate as necessary. The anesthesia level was judged to be plex spikes of CA3 pyramidal cells were always recorded, but more inadequate when there was response to toe pinch and/or presence of the diagnostic was the distinctive antidromic response to commissural stim- cornea1 reflex to touch with dry cotton wisp. A state of excessive an- ulation found in CA3 but not in CA1 (Fig. 2). The electrode was de- esthesia was determined to exist when the cornea1 reflex was no longer termined to be in the pyramidal cell layer of CA3, and not CA3, or present to the application of a wet cotton wisp. Based on observations CA3,, if, while lowering the electrode, (1) the dentate gyms granule from previous work (Miller et al., 1992, 1993), and also observed here, cell layer was not encountered, and (2) pyramidal cells were recorded conditions of either too much or too little anesthesia resulted in variable in a discrete band and not in a more continuous dorsal/ventral band changes in the size of hippocampal field potentials and auditory N40, corresponding to CA2lCA3,. and consistently disrupted auditory gating. However, when a stable and Sufficient negative (cathodal) current of 150 ysec pulse width was appropriate level of anesthesia was maintained these problems were not administered to the stimulating electrode to generate a CA3 population observed (Miller et al., 1993; see Results). spike and/or a field potential, each of at least 1 mV amplitude. The 5822 Miller et al. * LTP and Auditory Gating
4 Table 1. Stimulator currents required to generate a field CA1 potential or population spike of 1 mV in amplitude
Sub- sequent Current Current Stimulator LTP mean range placement status0 (-eSD, bA) N (1.4 VHC’ (+I 81 232 8 40-140 3 C-1 143 + 137 3 50-300 Anterior CA3 fi I 1 mV (+I 73 2 29 11 25-l 15 5 (-) 185 2 90-300 Homotopic CA3 (+) 572 17 5 40-90 C-1 50 2 40-60 clLTP of the population spike. . * Ventral hippocampal commissure. CA3 (see Results). This approach was successful in three of three rats, which were then included in the +LTP data set. Once the post-LTP response had stabilized, the variability of the pop- ulation spike amplitudes, averaged over 5 min intervals of time (similar to the time period of the baseline collection), ranged from 3% to 32%, or 0.6-3-fold the variability seen in the pre-LTP period (paired, within- animal observations). The variability in the field potential amplitudes 3 ranged from 2% to 18%, or 0.6-2.2-fold the variability seen in the pre- LTP period. The auditory evoked potentials were elicited by presentation of tones h (3000 Hz, 70 dB SPL, 10 msec duration) through hollow stereotaxic I 2 mv earbars. The auditory stimulus intensity was selected to fall in the mid- 5 ms dle of a plateau. At intensities above approximately 50 dB the amplitude of the evoked potential shows no significant change with intensity level 60 UV (unpublished data), and below 90 dB the startle mechanism is not a 5 ms 2 confounding factor (Keith et al., 1991). Two tones, separated by 500 msec, were presented every IO sec. Signals were filtered such that only electrical activity with frequencies between 0.1 and 1000 Hz were recorded. Sixteen pairs of tones were presented in each trial and an average waveform was constructed from Figure 2. Spontaneous pyramidal cell complex spikes and potentials the digitized components of the 16 N40 responses. Test/conditioning (2” evoked in the hippocampal pyramidal cell layers by electrical stimula- C) ratios were calculated from the average waveform. The mean T/C tion of commissural fibers recorded in areas CA1 (A) and CA3 (B). I, ratio for a specific period of analysis represents the mean of all the T/ Stimulation artifact; 2, antidromic population spike; 3, orthodromic pop- C ratios calculated for each trial during that period. Historically, a T/C ulation spike; 4, positive-going field potential. Note that the time scale ratio of >0.5 has been associated with psychosis in humans (Adler et for the action potentials is different between A and B, and that the al., 1982; Freedman et al., 1987). Similar T/C ratios are seen following voltage scales are different in all cases. the administration of drugs at doses that are known to have psychoto- mimetic effects in rats (Adler et al., 1986; Miller et al., 1992). A T/C ratio of 0.5 is >2 SD units above the mean value of 0.25-0.35 found for controls in previous studies (Miller et al., 1992, 1993) and is con- range of current required was 25-300 PA (Table 1). In five cases it was sidered to be the value above which gating is lost. The minimum time not possible to elicit a population spike with commissural stimulation separating trials of auditory evoked potentials was 5 min. A minimum prior to LTP even though a large positive field potential was generated. of four trials was collected prior to high-frequency electrical stimulation A series of at least six test pulses was delivered to establish the mean to confirm that a given animal’s auditory gating was stable. Any rat field potential and population spike amplitude prior to high-frequency which failed to show consistent auditory gating was excluded from electricalstimulation. The variability in the populationspike amplitude further study. Viral infections or, as was explained earlier, cyanosis from generally ranged from 5 to 20%, and the variability in the EPSP am- too much anesthesia can lead to lack of gating in our paradigm. In plitude was 3-20%. With the same current used to record the baseline addition, approximately 10% of the rat subjects fail to gate the N40 for data, three trains of 250 Hz, 1 set pulses, each separated by 5 set, were then delivered to the stimulating electrode. This stimulation protocol was developed in an attempt to reliably achieve a significant degree of Table 2. LTP of the population spike or the field potential LTP More physiologically relevant protocols were also successful (e.g., depends on stimulator placement Rose and Dunwiddie, 1986) but less reliable and not as likely to yield as large a degree of potentiation. Potentiation of the electrically evoked LTP of population LTP of field potential was then monitored by administering a trial of at least six test spike0 potential pulses. Potentiation was deemed to have occurred if a two-tailed t test Stimulator (mean % (mean % of a series of population spike or positive field potential amplitudes placement control 2 SD) control 5 SD) revealed a significant increase between those values and the baseline values (p < 0.05). LTP was deemed to have occurred if the potentiation VHC 210 5 170 228 + 92* was observed across all trials during the period of at least 1 hr (Racine Anterior CA3 327 ?z 232 168 t 68 et al., 1983; Linden, 1994) and if successive measures revealed a stable Homotopic CA3 264 +- 138 118 t 16 potentiation during the 1 hr long period (i.e., no significant decrease between successive measurements of the electrically evoked potential). y Values do not include those experiments in which no prior population spike If potentiation was induced that was not stable, a second high-frequency was recorded. train of electrical stimuli was administered in an attempt to achieve LTP *Value is significantly different from the others in this column, p i 0.03. The Journal of Neuroscience, August 1995, 15(a) 5823 unknown reasons; those rats were excluded as well. Following high- Auditory-evoked potentials to paired tones were recorded at frequency electrical stimulation, trials of auditory evoked potentials and the same site and through the same electrode as the electrically electrically evoked potentials were conducted. The minimum time in- terval separating a trial of auditory evoked potentials from a trial of evoked potentials, although not at the same time. The ratio of electrically evoked potentials was 1 min, and the maximum interval the test response to the conditioning response (7°C ratio) was was usually
B. Field Potentials and Population Spikes
u^
e .o 0.75
a E? .a 050 .s a Figure 3. The effect of LTP on au- 8 0.25 ditory gating. A, A paired r test of the Y . T/C ratios before and after LTP re- z BASELINE 1+LTP vealed that the gating loss was statis- tically significant (n = 28; p < 000 Pre-LTP LTP 0.0001). The T/C ratios are graphed as N40 Evoked Potentials means ? SEM. B, An representative experiment. Top, LTP was induced by stimulation of the VHC. Bottom, Four consecutive samples of auditory evoked potentials recorded prior to LTP stimulation (left) and during LTP (right). These data illustrate the transi- tion from test amplitude increase to conditioning amplitude depression which followed successful LTP induc- tion. The elapsed time from the first BASELINE N40 evoked potentials shown to the last was 30 min. The elapsed time from the first +LTP N40 evoked potentials shown to the last was 15 min.
tetani delivered through the contralaterally placed stimulators also, no significant changefrom baselineT/C ratio was observed (Fig. 5). after high-frequency electrical stimulation (+0.07 ? 0.09; p > In no case did LTP stimulation have an immediate effect on 0.30; Table 4, group D). auditory gating. This finding was independent of stimulator As a further control for potential nonspecificeffects of high- placement. From the time of the peak potentiation observed frequency electrical stimulation, three rats received and intra- within the first 10 min following the high-frequency train, the peritonealdose of 6 mg/kg of the competitive NMDA antagonist mean time to onset of the auditory gating loss was 10.5 lr 7 CPP 20 min prior to high-frequency stimulation of the ventral min (range,4-29 min). hippocampal commissure(two animals) or anterior CA3 (one Following LTP, both the N40 conditioning amplitude and test animal). CommissuralLTP is NMDA dependent (Collingridge amplitudewere significantly alteredcompared to baselinevalues et al., 1983; Zalutsky and Nicoll, 1990); therefore, it was pos- (p < 0.05; paired t test): the conditioning amplitude decreased sibleto block LTP with the prior administrationof CPP In CPP- and the testing amplitudeincreased. Furthermore, the effect had treated rats, as expected, LTP was not generated by the high- two phases(Table 3). Initially, the auditory gating was disrupted frequency electrical stimulation; in addition, auditory gating was primarily by a large increasein the test amplitude. These short not perturbed. The mean change was -2.6 ? 2.3% 0, > 0.15) term values for test amplitudewere significantly greater than the in the population spike amplitude, +lO 2 37% 0, > 0.50) in test amplitude values at later time points (p < 0.01). It was the positive field potential amplitude and -0.04 ? 0.12 (p > generally the casethat, within 5-15 min of the initial disruption 0.6) in the T/C ratio (Table 4, group F). These data show that of gating, the conditioning amplitude depressionbecame more the auditory gating loss and the LTP are correlated in that they pronounced, although overall this change was not significantly both require activation of the NMDA receptor in this paradigm. different from the reduction in conditioning amplitudeassociated Thus, the CPP data serves as a control for aspectsof high-fre- with the initial gating loss(p > 0.05). Figure 3B illustratessuch quency stimulationthat are unrelatedto activation of the NMDA a transition in the conditioning and test amplitudes. receptor. Is it a significant finding that rats which did not develop LTP Controlsfor effectsof high-frequency electrical stimulation not of the population spike also did not experience a changein au- related to LTP of the population spike ditory gating? Dividing the 34 memberexperimental population In three rats, high-frequency electrical stimulation failed to po- (Table 4) into two groups, thosethat developed LTP of the pop- tentiate either the field potential (mean change = -30 ? 44%; ulation spike and those that did not, and applying one-way p > 0.35) or the population spike (mean change -3.7 * 5.5%; ANOVA to the change in T/C ratio experienced by the two p > 0.36). In thesecases, the resulting mean changefrom base- groups following high-frequency stimulation, the finding was line T/C ratio was not significant (-0.02 t 0.08, p > 0.40; Table determinedto be significant (F = 17.4; p < 0.0002). In contrast, 4, group E). In four other rats, only the field potential was po- the sameanalysis applied to LTP of the positive field potential tentiated (meanchange = +83 ? 6%; p < 0.001; mean change revealed no relationshipto changesin auditory gating (F = 0.62; in population spike = - 11 2 22%; p > 0.41). For theseanimals p = 0.44). The Journal of Neuroscience, August 1995, 15(8) 5825
0 Population Spike Amplitude A Test/Conditioning Ratio 25 3.0
2.5 20
2.0 15 1.5 10 1.0
5 0.5
0 -i 0.0 0 50 100 150 200 Time (min)
Figure 4. Time course of an experiment in which the first high-frequency train (delivered at the 20 min time point) elicited a potentiation that was not stable (see Materials and Methods). Disruption of auditory gating (mean T/C ratio > 0.5) occurred only with the development ot LTP following the second train (delivered at the 95 min time point). The time course of the potentiation following the second high-frequency train is characteristic of the +LTP experiments. The mean T/C ratios were 0.12 + 0.11 for the baseline period, 0.44 2 0.15 for the period following the first train, and 1.25 t 0.84 for the period following the second train. This finding was replicated in two additional rats.
Contralateral Contralateral A. VHC Stimulation B. Anterior CA3 Stimulation C. Homotopic CA3 Stimulation . J . , . 2 5.6 en
1 ,/86/f% ‘;;
. a-o.o \ 1 0.0, , o.o\ l-2 -0.15 0.84 -0.15 0.94 -0.05 I. Change in T/C ratio (LTP - pre-LTP)
Figure 5. Correlation between LTP of the population spike and loss of auditory gating for three different stimulator placements. The data set does not include experiments in which there was no population spike prior to LTP A, Stimulator location in the ventral hippocampal commissure. B, Stimulator location in the anterior CA3, contralateral to the recording electrode. C, Stimulator location in CA3,, contralateral and homotopic to the recording electrode. In all cases there was a positive correlation between LTP-induced growth of the population spike and an increased T/C ratio, indicating a reduction in gating. However, these correlations were statistically significant only for the anterior CA3 and contralateral/homotopic stimulator placements. 5826 Miller et al. - LTP and Auditory Gating
Table 3. The contribution of conditioning amplitude depression seven rats for varying amountsof time. In one of the failures, and test amplitude increase to the loss of auditory gating seen reversal of the original LTP occurred but LTP developed from with LTP: initial and delayed effects the second site of high-frequency stimulation, and in the other, no reversal of LTP took place. In the successfulreversal exper- Mean change in Elapsed conditioning Mean change in iments, auditory gating was normalized to a mean T/C < 0.5 time amplitude test amplitude (as defined in Materials and Methods) for the period during (min) (% control + SD) (% control + SD) which the population spike amplitudewas lessthan or equal to the original baselinevalue (Fig. 6). The mean time period of 0 -25 + 30 +107 -t57 successfulreversal of LTP was 24 +- 8 min (range, 15-35 min), 5-60h -43 t 20 +78 ?56* which covered at least four trials of auditory gating. In individ- “time since first onset of auditory gating loss. Zero minutes represents only ual rats, the T/C ratios during reversal were compared to an one trial. equivalent number of trials during LTP The minimum mean T/ hThe 5-60 min time point represents the mean, for each experiment, of all the trials for 60 min following the onset of the auditory gating loss, summed for C ratio during LTP was significantly greater than the mean T/C all experiments in which there was LTP of the population spike (N = 21) to ratio during reversal in all five animals(p < 0.05). Furthermore, generate an overall mean. there was no significant difference between the T/C values dur- * Significantly different from 0 time point, p < 0.01, ing reversal and the original baselineT/C values (mean differ- ence, 0.02 t 0.10; range, -0.10 to $0.14; p > 1.0). Within Reversal of LTP and normalization of auditory gating individual rats, the variability in the T/C ratio decreasedfrom Placementof stimulating electrodesin the anterior CA3 and the approximately fourfold during LTP to an average of 1.3 times homotopic CA3 revealed an apparentheterosynaptic depression that of the baselineduring the reversal period. if high-frequency stimulation was applied alternately to both Comparisonof the mean difference between the electrically sitesof stimulation. Furthermore, if the second input yielded a evoked positive field potentials during reversal and the original poor population spike responsewhile the first yielded a robust baselinevalues revealed no significant difference (mean, - 11.7 population spike response,it was possible to generate LTP at * 16.6%; range, -29% to + 10%). Comparisonsfor individual the first input and then reverse it by administering a high-fre- rats indicated no significant change in four rats (p > 0.3) and quency stimulusto the secondinput, without generating LTP of significant reduction in one (p < 0.005). Mean population spike the secondinput. Thus, it was possibleto use this heterosynaptic amplitudesduring reversal were -22 2 37% (range, -58% to stimulation paradigm to first generate LTP, and then return to +33%) compared to pre-LTP baselinevalues. Individual com- the baselinecondition of no potentiated inputs. parisonsrevealed no significant difference for two rats (p > Two stimulating electrodeswere lowered, one to the contra- 0.25) and a significant reduction in the other three (p < 0.05). lateral anterior CA3 and the other to the contralateral CA3 homotopic to the recording electrode.The stimulationsite yield- Discussion ing the more sensitive population spike responsefor a given LTP of commissuralinput to the hippocampusdisrupted audi- current intensity was selectedas the site for initial LTP gener- tory gating in a proportional and reversible manner.This effect ation. Following LTP induction, reversal of the LTP was at- is unlikely to have resulted from an aspect of high-frequency tempted by applying a high-frequency train (250 Hz, up to 400 stimulation unrelated to LTP because(1) when high-frequency PA, 150 psec pulseduration, for 1 set) to the other commissural stimulation failed to generate LTP gating was not altered; (2) stimulator. Complete reversal of LTP was achieved in five of when LTP was purposely prevented by the administrationof an
Table 4. Summary of treatment and results
Group Treatment Result N A High-frequency stimulation LTP of population spike 18 LTP of positive field potential Loss of auditory gating B As above LTP of population spike 2 No LTP of positive field potential Loss of auditory gating C As above LTP of population spike 4 LTP of positive field potential No change in auditory gating D As above No LTP of population spike 4 LTP of positive field potential No change in auditory gating E As above No LTP of population spike 3 No LTP of positive field potential No change in auditory gating F CCP before high-frequency stimulation No LTP of population spike 3 No LTP of positive field potential No change in auditory gating The Journal of Neuroscience, August 1995, 15(8) 5827
Population Spikes
Figure 6. Example of a reversal ex- periment. LTP was generated by stim- ulation of afferents from the contralat- . era1 CA3 homotopic to the recording site, and then reversed by high-fre- quency stimulation of fibers originating 5mV in the anterior CA3. Top, Averages of --I stimulus-evoked potentials. A large in- 10ms i crease in the orthodromic population spike was observed during LTP, but I was returned to the pre-LTP amplitude +LTP BASELINE REVERSAL OF LTP by heterosynaptic stimulation. The an- tidromic population spike remained N40 Evoked Potentials stable throughout all manipulations. Mean T/C Ratio = 0.3 I Mean TIC Ratio = 0.99 Mean TIC Ratio = 0.32 Bottom, Averages of auditory evoked , ,,,,, / ,,,, , , /111/,/111,1~1,1,,,,,,, , , I, I., I, , ,, ,,,, , potentials. Gating of N40 was disrupt- ed following LTP induction, but re- turned to baseline levels following LTP reversal. The N40 evoked potentials are four consecutive traces from each of the resnective time oeriods. The elapsed time from the firs’t BASELINE &LPJ--% N40 evoked potentials shown to the last was 18 min. The elapsed time from q-ydFJbt9 the first -+LTf N40 evoked potentials shown to the last was 15 min. The elapsed time from the first REVERSAL l l pJ--y+ytM., OF LTP N40 evoked potentials shown 1st tone 2nd tone 50 ms to the last was 15 min.
NMDA antagonist, gating was unchanged following high-fre- mentionedin the introductory section,commissural synapses are quency stimulation; and (3) high-frequency stimulation itself not likely to be carrying the auditory signal to the hippocampus. was usedto normalize gating that had previously been perturbed Rather, entorhinal cortex afferents are thought to be the primary by LTI? input pathway for auditory information (Foster et al., 1988). Au- Gating loss was correlated with an LTP-induced increasein ditory signalsmay reach CA3 indirectly via the dentate gyrus, the population spike, but not the positive field potential, recorded or via a direct projection (Yeckel and Berger, 1990; Colbert and in the CA3 pyramidal cell layer. Population spikes are summed Levy, 1992). action potentials of pyramidal cells (Andersen et al., 1971) and One mechanismby which commissuralLTP could decrease therefore, are a reflection of postsynaptic excitation. Extracel- the conditioning amplitude is by causingheterosynaptic depres- lular field potentials are thought to reflect excitatory and inhib- sion of the perforant path synaptic response.Others have ob- itory synaptic currents generatedat postsynaptic siteson pyra- servedlong-term synaptic depressionin the perforant path (Pang midal cells (Schlag, 1973; Leung, 1979; Di et al., 1990). Elec- et al., 1993) and negative heterosynaptic interactions between trical stimulation of the commissureactivates both excitatory commissuraland perforant path inputs in CA3 (Tomasuloand and inhibitory synaptic currents within a few milliseconds of Ramirez, 1993). This is a prediction that can be tested in the each other (Buzsaki and Eidelberg, 1982); the resulting the pos- future by looking for depressionof the population spike evoked itive and negative currents sum, tending to cancel each other by perforant path stimulation following LTP of commissuralfi- out. Due to this interaction between positive and negative cur- bers. Assuming that heterosynaptic depressionof the perforant rents,the peak magnitudeof the extracellularly recordedpositive path is involved in depressionof the conditioning (first tone) field potential is not likely to reflect changesin excitatory input response,why doesthe amplitude of the test (secondtone) re- to pyramidal cells in a gradedmanner. In contrast,the population sponseincrease? spike should provide more of a graded reflection of excitatory Previous work has shown that interneuronalexcitation in the input to the pyramidal cells becausethe faster frequency action 500 msec following the first tone correlates strongly with the potential currents are not affected by the slower frequency field diminishedN40 responseto the secondtone (Miller and Freed- potential currents. This suggests that changes in synaptic man, in press). Current evidence suggeststhat the excitation of strengthwill be more accurately representedby alterationsin the interneurons by the tone does not occur via feedforward acti- population spike rather that in a field potential which is recorded vation by pyramidal cells (Miller and Freedman,in press).Rath- distant from the site of synaptic input. er, the excitation of interneuronsby the tone must originate from Theoretically, LTP of the excitatory synapsescarrying the an external sourcesuch as the perforant path. LTP of commis- tone input should cause the N40 conditioning amplitude to in- sural input to interneuronscould causeheterosynaptic depression creasein size. Rogan et al. (1993) have demonstratedsuch a of monosynapticor disynaptic perforant path input to interneu- relationshipbetween tone evoked potentials and LTP of medial rons, which would diminish the interneuron responsebetween geniculateafferents to the amygdala. In the present work, LTP the tones.This would result in an increasedN40 responseto the of hippocampalcommissural connections resulted in a decrease second tone. While most studies of heterosynaptic depression in N40 conditioning amplitude was observed.However, as was have been carried out in pyramidal cells (for a review, see Lin- 5828 Miller et al. l LTP and Auditory Gating
LTPOF I.= HETEROSYNAPTIC DEPRESSION OF 2. and Abraham, 1992; Kerr and Abraham, 1993), but also that it is possibleto induce depressionof the LTP without getting LTP REVERSAL OF LTP 1. = REVERSAL OF DEPRESSION 2. at the most recently tetanized synapse(Bradler and Barrionuevo, 1990). In addition, there is data demonstratingthat depression Il at an unstimulated synapse(e.g., the perforant path) can be re- commiswral input (electri~~mulation) versed if the LTP causing the heterosynaptic depressionis re- versed by tetanus of yet another heterosynaptic input (Bradler and Barrionuevo, 1990). Although input from the entorhinal cortex may be necessary e l interneuron for the responseof the hippocampusto tones, other facilitatory pyramidal cell A i. inputs are important as well (Harrison et al., 1988; Luntz-Leyb- man et al., 1993; Miller and Freedman, 1993). Heterosynaptic depressionof another facilitatory input could have the sameau- ditory gating effect as the perforant path depressionproposed here. perforant I path Understandingthe significance of the current findings in the (input of tone) context of a behaving animal depends upon the physiological relevance of LTP itself and whether or not long-term synaptic Figure 7. A schematic model of the synaptic contacts likely to be plasticity is related to information processingand/or learningand important to the effect of commissural LTP on auditory gating. Stim- memory. In addition, although the effect of LTP upon auditory ulationof commissuralinput 1 generatesLTP at synapse1 for both the gating is not an artifact of high-frequency electrical stimulation, pyramidal cell and the interneuron. The LTP also causes heterosynaptic there may still be aspectsof the stimulation protocol that are depressionof perforantpath input to theseneurons (2), resultingin the pertinent when assessingthe physiological relevance of the re- depression of N40 conditioning amplitude. The perforant path input refers to either the monosynaptic connection from the entorhinal cortex sults. The electrical stimulation is broad and nonspecific, char- or the disynapticconnection via the dentateand the mossyfiber system. acteristicswhich must certainly differ from the synaptic distri- Heterosynaptic depression of the perforant path input to the interneuron bution of learning-inducedchanges. The hippocampusis exten- prevents the inter-tone excitation of the interneuron that normally in- sively affected by the commissuralLTP generatedin this study, hibits the pyramidalcell synapticpotential evoked by the secondtone; thus, the N40 test potential increases and auditory gating is lost. The and therefore, the cells respondingto the tone input and those reciprocalconnections between the pyramidalcell andthe interneuron experiencing LTP are likely to overlap. In a freely behaving are not shown. To reverse the auditory gating loss, a high-frequency animal, direct parallels with such an experimental outcome are stimulusis appliedto commissuralinput 1’. Reversalof the LTP at likely occur only under a limited set of circumstances.Learning synapse 1, without generating LTP at synapse l’, reverses the hetero- that is specifically related to the tone would, in all probability, synaptic depression of the perforant path inputs to the pyramidal cell and the interneuron.Auditory gatingthen returnsto normal. affect the samecells involved in gating the responseto the tone. Otherwise, situations in which cells involved in learning might fortuitously overlap with those involved in auditory gating den, 1994), long-term heterosynapticdepression has been dem- would seem to be limited to conditions of information “over- onstratedin GABAergic interneuronsin the hippocampus(Ree- load” or involving hippocampal pathological conditions, such ce and Redman, 1992), as well as in other brain regions (Kom- that only a small amount of hippocampaltissue would be avail- bian and Malenka, 1994). In addition, heterosynapticdepression able for processingfunctions. of perforant path input to interneurons has been implicated in If cells involved in auditory gating shouldexperience auditory the responseof the dentate to commissuraltetanus (Tomasulo gating lossresulting from learning processesor from LTP of the and Ramirez, 1993). commissuralinput, there are likely to be some measurablebe- That synapseson different cell types may be involved in the havioral consequencesin the behaving animal. As describedear- changesin the conditioning versustest amplitudesis supported lier, a loss of auditory gating is correlated with the administra- by a clear separationof the timing of the changesin some of tion of psychotomimetic drugs in rats and with psychosis in the experiments (Table 3, Fig. 3B). The initial consequenceof humans,some of the symptomsof which are auditory in nature. LTP induction was to increasetest N40 amplitudewhile, in some However, it is not known how many different regions of the cases,conditioning N40 amplitude remained unchanged (Fig. brain must experiencethe auditory gating loss to causethe au- 3B). This transition is consistentwith our hypothesized loss of ditory-specific features of psychosis. At least two other brain interneuron activation by the perforant path and the resulting regions are known to exhibit auditory gating [the medialseptum lossof inhibitory control of the test N40. Next, the conditioning (Miller and Freedman, 1993) and the brainstem reticular for- amplitude decreasedand the test amplitude decreasedslightly mation (Bickford et al., 1993)]; it is probable that other extra- also (although it did not return to baselinevalues). This transi- lemniscal auditory regions do as well. Here, we have attempted tion fits well with a decreasedperforant path input to pyramidal to specifically disrupt auditory processingin the hippocampus, cells in responseto both the first and secondtones. and the behavioral sequelaeare likely to be more limited than A possiblemodel to explain the effect of commissuralLT& if gating in all of the extra-lemniscal systems was disrupted. and its reversal, upon hippocampalauditory gating is shown in Sakurai (1990) has proposedthat the hippocampusis important Figure 7. In other contexts, all of the interactionsproposed in for initiating motor responseto sounds;if so, then loss of hip- Figure 7 have been reportedby others.From a condition of LTP pocampalauditory gating causedby commissuralLTP might be at one synapse,it has not only been demonstratedthat LTP at a expected to alter the motor responseto repetitive sounds.Alter- heterosynaptic site (e.g., synapse 1’) can introduce depression natively, the auditory gating disruption may interact with the of the LTP at the first synapse(Levy and Steward, 1979; Christie place cell function of the hippocampus(O’Keefe and Nadel, The Journal of Neuroscience, August 1995, 75(8) 5829
1978; Breese et al., 1989; Wilson and McNaughton, 1993), mak- Kombian SB, Malenka RC (1994) Simultaneous LTP of non-NMDA- ing the spatial attribution of sound less rapid or accurate. receptor-mediated responses in the nucleus accumbens. Nature 368: 242-246. References Kubie JL, Ranck JB Jr (1983) Sensory-behavioral correlates in indi- vidual hippocampus neurons in three situations: space and context. Adler LE, Pachtman E, Franks RD, Pecevich M, Waldo MC, Freedman In: Neurobiology of the hippocampus (Seifert W, ed), pp 433-447. R (1982) Neurophysiological evidence for a defect in neuronal New York: Academic. mechanisms involved in sensory gating in schizophrenia. Biol Psy- Leung L (1979) Potentials evoked by alvear tract stimulation in hip- chiatry 17:639-654. pocampal CA1 region of rats. 2. Spatial field analysis. J Neurophysiol Adler LE, Rose G, Freedman R (1986) Neurophysiological studies of 42:1571-1589. sensory gating in rats: effects of amphetamine, phencyclidine and Levy WB, Steward 0 (1979) Synapses as associative memory elements halooeridol. Biol Psvchiatrv 21:787-798. in the hippocampal formation. Brain Res 175:233-245. Andersen P, Bliss TVP Skredk KK (1971) Unit analysis of hippocam- Linden DJ (1994) Long-term synaptic depression in the mammalian pal population spikes. Exp Brain Res 13:208-221. brain. Neuron 12:457-472. Berger TW, Alger B, Thompson RF (1976) Neuronal substrate of clas- Luntz-Leybman V, Bickford PC, Freedman R (1993) Cholinergic gat- sical conditioning in the hippocampus. Science 192:483-485. ing of response to auditory stimuli in rat hippocampus. Brain Res Bickford P, Luntz-Leybman V, Freedman R (1993) Auditory sensory 587:130-136. gating in the rat hippocampus: modulation by brainstem activity. McNaughton BL, Barnes CA, O’Keefe J (1983) The contributions of Brain Res 607:33-38. position, direction, and velocity to single unit activity in the hippo- Bickford-Wimer PC, Nagamoto H, Johnson R, Adler LE, Egan M, Rose campus of freely-moving rats. Exp Brain Res 52:41-49. G, Freedman R (1990) Auditory sensory gating in hippocampal neu- Miller CL, Freedman R (1993) Medial septal neuron activity in relation rons: a model system in the rat. Biol Psychiatry 27: 183-192. to an auditory sensory gating paradigm. Neuroscience 55:373-380. Bradler JE, Barridneuvo G (1990) Heter&ynap& correlates of long- Miller CL, Freedman R (in press) The activity of hippocampal inter- term potentiation induction in hippocampal CA3 neurons. Neurosci- neurons and pyramidal cells during the response of the hippocampus ence 35:265-271. to repeated auditory stimuli. Neuroscience, in press. Brankack J, Buzsaki G (1986) Hippocampal responses evoked by tooth Miller CL, Bickford PC, Luntz-Leybman V, Adler L, Gerhardt GA, pulp and acoustic stimulation: depth profiles and effect of behavior. Freedman R (1992) Phencyclidine and auditory sensory gating in Brain Res 378:303-3 14. the hippocampus of the rat. Neuropharmacology 3 1: 104 l-l 048. Breese CR, Hampson RE, Deadwyler SA (1989) Hippocampal place Miller CL, Hruby VJ, Matsunaga TO, Bickford PC (1993) Alpha-MSH cells: stereotypy and plasticity. J Neurosci 9: 1098-l 111. and MCH are functional antagonists in a CNS auditory gating para- Buzsaki G, Eidelberg E (1982) Direct afferent excitation and long-term digm. Peptides 14:43 1440. potentiation of hippocampal intemeurons. J Neurophysiol 48597607. Morris RGM (1990) Toward a representational hypothesis of the role Christie BR, Abraham WC (1992) NMDA-dependent heterosynaptic _^ long-term depression in the dentate gyrus of anaesthetized rats. Syn- of hippocampal synaptic plasticity in spatial and other forms of learn- apse 10: 1-6. ine. Cold Sorine Harbor Svmo Ouant Biol 55:161-173. Colbert CM, Levy WB (1992) Electrophysiological and pharmacolog- O’Keefe J, Nadel t (1978) The’hippocampus as a cognitive map. Ox- ical characterization of perforant path synapses in CAl: mediation ford: Clarendon. by glutamate receptors. J Neurophysiol 68:1-8. Pang K, Williams MJ, Olton DS (1993) Activation of the medial septal Collingridge GL, Kehl SJ, McLennan H (1983) Excitatory amino acids area attenuates LTP of the lateral perforant path and enhances het- in synaptic transmission in the Schaffer collateral-commissural path- erosynaptic LTD of the medial perforant path in aged rats. Brain Res way of the rat hippocampus. J Physiol (Lond) 334:3346. 632-150-160. Cooper JR, Friedman PJ (1958) The enzymatic oxidation of chloral Racine RJ, Milgram NW, Hafner S (1983) Long-term potentiation phe- hydrate to trichloroacetic acid. Biochem Pharmacol 1:76-82. nomena in the rat limbic forebrain. Brain Res 260:217-231. DeJonge M, Racine RJ (1985) The effects of repeated induction of Reece LJ, Redman SJ (1992) Epsps in hippocampal interneurons ex- long-term potentiation in the dentate gyrus. Brain Res 328:181-185. hibit long-term enhancement. Sot Neurosci Abstr 22:628.28. Di S, Baumgartner C, Barth D (1990) Laminar analysis of extracellular Rogan M, Bordi F, LeDoux JE (1993) Long-term increases in auditory- field potentials in rat vibrissaibarrel cortex. J Neurophysiol 63:832- evoked responses accompany tetanically-induced LTP in the thalamo- 840. amygdala pathway. Sot Neurosci Abstr 23504.9. Douglas RJ (1972) Pavlovian conditioning and the brain. In: Inhibition Rose GM (1983) Physiological and behavioral characteristics of den- and learning (Boakes RA, Halliday MS, eds). New York: Academic. tate granule cell In: Neurobiology of the hippocampus (Seifert W, Edeline JM, Neuenschwander-El Massioui N, Dutrieux G (1990) Dis- ed), pp 449-472. London: Academic. criminative long-term retention of rapidly induced multiunit changes Rose GM, Dunwiddie TV (1986) Induction of hippocampal long-term in the hippocampus, medial geniculate and auditory cortex. Behav potentiation using physiologically patterned stimulation. Neurosci Brain Res 39:145-155. Lett 69:244-248. Eichenbaum H, Otto T (1993) LTP and memory: can we enhance the Rose GM, Branch BJ, Humphreys AG, Moore CI, Diamond DM (1993) connection? Trends Neurosci 16:163-164. The population spike is a valid measure of hippocampal long-term Foster TC, Hampson RE, West MO, Deadwyler SA (1988) Control of potentiation. Sot Neurosci Abstr 23:375.16. Sakurai Y (1990) Hip- sensory activation of granule cells in the fascia dentata by extrinsic pocampal cells have behavioral correlates during the performance of afferents: septal and entorhinal inputs. J Neurosci 8:3869-3878. an auditory working memory task in the rat. Behav Neurosci 104: Freedman R, Adler LE, Gerhardt GA, Waldo M, Baker N, Rose G, 253-263. Drebing C, Nagamoto H, Bickford-Wimer P, Franks R (1987) Neu- Schlag J (1973) Generation of brain evoked potentials. In: Bioelectric robiological studies of sensory gating in schizophrenia. Schizophre- recording techniques (Thompson RE Patterson MM, eds), pp 272- nia Bull 13669-677. 316. New York: Academic. Harrison JB, Buchwald JS, Kaga K, Woolf NJ, Butcher LL (1988) ‘Cat Solomon PR (1977) Role of the hippocampus in blocking and condi- P300’ disappears after septal lesions. Electroencephalogr Clin Neu- tioned inhibition of the rabbit’s nictitating membrane response. J rophysiol 69:55-64. Comp Physiol Psycho1 91:407-4 17. Jirsa R, Pot P Radil T (1992) Hippocampal auditory evoked response Squire L (1987) Memory and brain. Oxford: Oxford UP threshold in the rat: behavioral modulation. Brain Res Bull 28:149-153. Stafekhina VS, Vinogradova OS (1975) Sensory characteristics of the Keith VA. Mansbach RS. Gever MA (1991) Failure of halooeridol to cortical input to the hippocampus: the entorhinal cortex. Zh Vyssh block the effects of phencyclidine and dizocilpine on prepulse inhi- Nerv Deiat 25:119-127. bition of startle. Biol Psychiatry 30:557-566. Steward 0, Tomasulo R, Levy WB (1990) Blockade of inhibition in a Kelly JP (1991) Hearing. In: Principles of neural science (Kandel ER, pathway with dual excitatory and inhibitory action unmasks a capability Schwartz JH, Jesse11TM, eds). Amsterdam: Elsevier. for LTP that is otherwise not exnressed. Brain Res 516:292-300. Kerr DS, Abraham WC (1993) A comparison of associative and non- Swanson LW’ (1983) The hippocampus and the concept of the limbic associative conditioning procedures in induction of LTD in CA1 of system. In: Neurobiology of the hippocampus (Seifert W, ed), pp 3- the hippocampus. Synapse 14:305-313. 19. London: Academic. 5830 Miller et al. - LTP and Auditory Gating
Swanson LW (I 992) Brain maps: structure of the rat brain. Amsterdam: Voneida TJ. Vardaris RM. Fish SE. Reiheld CT (1981) The origin of Elsevier. the hippocampal commissure in the rat. Anat Ret 201:91-3031 Swanson LW, Teyler TJ, Thompson RF (1982) Hippocampal long-term Weinberaer NM. Diamond DM (1987) Phvsiological olasticitv in au- potentiation: mechanisms and implications for memory. Neurosci Res ditory-cortex:‘rapid induction by learning. ProiNeu;obiol 2$:1-55. Prog Bull 20:613-769. Weinberger NM, Ashe JH, Metherate R, McKenna TM, Diamond DM, Teyler TJ, Discenna P (1984) Long-term potentiation as a candidate Bakin J (1990) Retuning auditory cortex bv learning: a ureliminarv mnemonic device. Brain Res Rev 7:15-28. model of receptive fieldplasticity. Concepts Neurosz 1:91-132. d Wilson MA, McNaughton BL (1993) Dynamics of the hippocampal Tomasulo RA, Ramirez JJ (1993) Activity-mediated changes in feed- ensemble code for space. Science 261:1055-1058. forward inhibition in the dentate commissural pathway: relationship Yeckel ME Berger TW (1990) Feedforward excitation of the hiooocamous to epsplspike dissociation in the perforant path. J Neurophysiol 69: by afferents ?rom the entorhinal cortex: redefinition of the role of the 165-173. trisynaotic pathway. Proc Nat1 Acad Sci USA 87:5832-5836. Vayssettes-Courchay C, Sessler FM (1983) Evidence for sensory conver- ZalutskykA,-NicoliRA (1990) Comparison of two forms of long-term gences in rat entorhinal cortex. C R Seances Acad Sci 3 296:877-879. potentiation in single hippocampal neurons. Science 248: 1619-1624.