INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In various neuronal networks, axonal conduction times obey clear functional constraints. In the avian auditory system, for instance, where accurate timing of afferent signals is essential, the conduction time between the cochlear nucleus and its ipsi- and contralateral brain stem targets is kept constant by lengthening the path of axons ending ipsilaterally (Carr and Konishi 1990). Similarly, in fish electric organs, a synchronized activation of electrocytes is achieved by increasing the conduction velocity of the longer axons and lengthening the path of fibers ending in the closer part of the electric organ (Bennett 1970). In the preceding examples, the function of the system was known. Could an analysis of conduction times yield insights into the role of systems whose function is poorly understood? The present study aimed to test this in the amygdalo-perirhinal network.

Despite the fact that the amygdala and perirhinal (PRH) cortex are reciprocally connected (Krettek and Price 1977b; Russchen 1982), lesion studies have generally emphasized their distinct role. Indeed, PRH lesions were found to interfere with recognition and associative memory (Suzuki 1996; Suzuki and Eichenbaum 2000), whereas amygdala lesions had little impact on performance in such tasks (Parker and Gaffan 1998; Zola-Morgan et al. 1989). Instead, amygdala lesions were found to prevent the acquisition of classically conditioned fear responses (Davis et al. 1994; Kapp et al. 1992; LeDoux 2000), an issue that remains controversial (Cahill et al. 1999).

Yet, other data suggest that the amygdala and PRH cortex are functionally coupled. Indeed, PRH lesions performed after classical fear conditioning interfere with conditioned fear responses (Campeau and Davis 1995; Corodimas and LeDoux 1995; Rosen et al. 1992). In humans, emotionally arousing stories are better recalled than neutral ones, and amygdala lesions abolish this effect (Adolphs et al. 1997; Cahill et al. 1995). Moreover, imaging studies have found a high correlation between recall of emotionally arousing or neutral material and the amount of amygdala activation observed when these stimuli were first presented (Cahill et al. 1996; Canli et al. 2000; Hamann et al. 1999). These and other findings suggest that the amygdala promotes memory-storage processes in brain areas that are involved in declarative memory (Cahill 2000; Cahill and McGaugh 1998; McGaugh 2000), such as the PRH cortex.

How could the amygdala facilitate memory formation in the PRH cortex? Most biological theories of memory posit a Hebbian mechanism (Hebb 1949) that would strengthen synapses linking co-active neurons (Martin et al. 2000). However, experimental studies have revealed that the timing of activity in pre- and postsynaptic neurons is critical: potentiation of cortical synapses occurs when excitatory postsynaptic potentials (EPSPs) narrowly precede postsynaptic firing and depression develops when postsynaptic discharges precede presynaptic inputs (reviewed in Abbott and Nelson 2000). As a result, assuming that the amygdala modulates memory storage by facilitating Hebbian interactions in the PRH cortex confronts us with an interesting problem; the PRH area is a long strip of cortex (12 mm in the cat) and the LA is located at rostral PRH levels. Thus LA axons must travel variable distances to reach different rostrocaudal PRH sites. If the amygdala favors Hebbian associations between coincident but spatially distributed activity patterns in the PRH cortex, amygdala axons should display a uniform range of conduction times despite the varying distances they must travel to reach their targets. The present study was undertaken to test this prediction.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Most amygdalo-PRH efferents stem from the basolateral amygdaloid complex (Krettek and Price 1977a,b), a group of nuclei including the lateral (LA), basolateral, and basomedial nuclei. The LA is notable in this respect because its main extra-amygdaloid target is the PRH cortex (Krettek and Price 1977b). Thus the present study focused on the conduction times of LA neurons.

Surgery

Experiments were conducted in agreement with ethical guidelines of the Canadian Council on Animal Care. Cats (2.5-3.5 kg) were preanesthetized with a mixture of ketamine and xylazine (11 and 2 mg/kg, im) and artificially ventilated with a mixture of ambient air and 2% isoflurane. The end-tidal concentration in CO₂ was kept at 3.7 ± 0.2% and the body temperature was maintained at 37-38°C with a heating pad. The level of anesthesia was assessed by continuously monitoring the electroencephalogram. The bone overlying the amygdala and PRH cortex was removed on both sides and the dura mater opened.

Two types of experiments were performed. The first series of experiments aimed to determine how the antidromic response latency of LA neurons fluctuated as a function of distance to their PRH projection sites. To this end, three concentric stimulating electrodes separated by 4.5 mm in the rostrocaudal axis were inserted in the PRH cortex (Fig. 1A1). Then, an array of four tungsten microelectrodes (2-6 M at 1 kHz; outer diameter of 80 µm) was lowered stereotaxically into the LA nucleus, as shown in Fig. 1A1. To construct the array, small holes were drilled in a circular Teflon block and the electrodes were inserted into them prior to the experiments. This experiment was repeated in both hemispheres of three cats.

View larger version (84K): [in this window] [in a new window]   Fig. 1. Microelectrode configuration and histological verification of recording sites. A: scheme showing the intended position of the recording (dots) and stimulating (concentric circles) electrodes in the first (A1) and second (A2) series of experiments. B: thionin-stained coronal sections illustrating selected histological controls performed after the first (B1-B2) and second (B3-B4) series of experiments. Arrows point to small electrolytic lesions performed to mark selected sites at the end of the experiments. Arrowheads point to the tip of stimulating electrodes. B1: a microelectrode track in the LA. B2: a perirhinal (PRH) stimulation site. B3: a LA stimulation site (arrowhead) and a PRH recording site (arrow). Another PRH recording site can be seen in B4. Abbreviations: BL, basolateral nucleus; BM, basomedial nucleus; CA, caudate nucleus; CE, central nucleus; ENT, entorhinal cortex; EC, external capsule; LA, lateral amygdaloid nucleus; OB, olfactory bulb; rh, rhinal sulcus.

The second series of experiments aimed to determine how the antidromic response latency of PRH neurons fluctuated as a function of distance to their LA and/or perirhinal projection sites. To this end, stimulating electrodes were inserted in the LA and PRH cortex as shown in Fig. 1A2. Then, an array of six tungsten microelectrodes, arranged in the configuration shown in Fig. 1A2, was lowered stereotaxically into the PRH cortex. This experiment was repeated five times.

During both types of experiments, the microelectrode arrays were lowered in steps of 4 µm by a piezoelectric micromanipulator. While moving the microelectrodes, electrical stimuli (0.3 ms, 1 mA, 2 Hz) were delivered in the PRH cortex to increase the likelihood of finding responsive cells. When responsive neurons with a high signal-to-noise ratio (3) were encountered, the stimulus intensity and duration was decreased just above threshold (typically 0.05-0.2 ms pulses of 0.1-0.5 mA) and the nature of the response (ortho- vs. antidromic) was determined. See RESULTS for the criteria used to distinguish antidromic and orthodromic responses. The evoked activity of selected neurons was observed on a digital oscilloscope, digitized, and stored on tape.

Histological identification of recording and stimulating sites

At the end of the experiments, recording sites were marked with electrolytic lesions (0.5 mA for 5 s). Following this, the animals were given an overdose of barbiturate (sodium pentobarbital, 40 mg/kg, iv) and perfused with 500 ml of a cold saline solution (0.9%) followed by 1 l of fixative, containing 2% paraformaldehyde and 1% glutaraldehyde in 0.1 M phosphate-buffered saline (pH 7.4). The brains were later sectioned on a vibrating microtome (at 100 µm) and stained with thionin to verify the position of the recording electrodes. The sections were collected serially to facilitate estimation of the distance separating recording and stimulating sites.

The microelectrode tracks were reconstructed by combining micrometer readings with the histological controls. We also estimated the distance between recorded neurons (Fig. 1, B1, B3, and B4) and stimulation sites (Fig. 1, B2 and B3). In doing so, shrinkage due to fixation was taken into account. The amount of shrinkage was determined by measuring the postfixation separation of tracts left by tungsten rods inserted 10 mm apart prior to fixation, as previously described (Paré et al. 1997).

Estimation of the distance between the stimulating and recording sites

It is difficult to measure the length of the path taken by axons to reach their target because axonal trajectories are variable. For instance, some LA axons take a relatively direct route to the PRH cortex, whereas others take a circuitous path (Smith and Paré 1994). Among the latter, some LA axons first course ventrally (i.e., away from the PRH cortex), then follow the external capsule, and later turn lateroventrally to end in the PRH cortex (Smith and Paré 1994). Thus, recognizing these difficulties, we only estimated the distance between the stimulating electrodes and backfired neurons in the rostrocaudal and lateromedial planes using the Pythagorean theorem.

Analysis

Analyses were performed off-line with the software IGOR (Wavemetrics, OR) and homemade software running on Macintosh microcomputers. Spikes were detected using a window discriminator after filtering (0.3-10 kHz) of the raw waves. Peristimulus histograms of unit discharges were computed.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Database

A total of 1862 neurons were recorded in this study. Histological controls (Fig. 1B) revealed that 484 of these cells were located in the LA nucleus and 508 in the PRH cortex. Examples of histologically confirmed recording sites are shown in Fig. 1 (arrows; LA, Fig. 1B1; PRH, Fig. 1, B3 and B4). Neurons recorded in other amygdala nuclei or cortical fields will not be considered here.

Neurons had to meet at least two of the following criteria to be classified as antidromically responsive: fixed response latency (0.2 ms jitter) (Fig. 2A), collision with spontaneous (Fig. 2B) or orthodromically evoked (Fig. 2C) action potentials, and ability to follow high-frequency stimulation (300 Hz; Fig. 2D).

View larger version (26K): [in this window] [in a new window]   Fig. 2. Criteria used to identify antidromic responses. The symbols below stimulation artifacts indicate the stimulation sites (dots, PRH stimuli; triangles, LA stimuli). A: response of two simultaneously recorded LA neurons to PRH stimuli. The first neuron generated antidromic responses (a); the second was activated orthodromically (o). Note the fixed latency of antidromic spikes compared with orthodromic ones. The neurons shown in B-D were recorded in the PRH cortex. Antidromic spikes (B1) colliding with spontaneous action potentials (B2). Same cell is shown in B1 and B2. C: antidromic spikes (a) evoked by LA stimuli (C1) colliding with orthodromic action potentials (o) elicited by PRH shocks (C2 and C3) when the latter occurred within approximately twice the latency of the antidromic response. See inset for the relative position of the stimulating and recording sites. D: antidromic spikes can follow high-frequency stimulation. Note progressive abolition of the somatodendritic component of the spike from the first to the last action potential. Only the initial segment spike was evoked by the last stimulus. The reduction in spike amplitude, distorted spike shape, and associated increase in latency seen in D probably resulted from a combination of factors such as cumulative Na+ channel inactivation, residual K+ channel activation, as well as synaptic inhibition. These various factors contribute to reduce the likelihood of antidromic invasion and increase the chance of evoking initial segment spikes that do not fully invade the somatodendritic compartment (Lipski 1981).

Applying these criteria to our sample of neurons revealed that 44% of LA cells (or 215 neurons) were backfired from the PRH cortex in the first series of experiments (Fig. 1A1). However, only 2% (or 5 neurons) of these antidromically responsive cells were backfired from two or more PRH sites. In the second series of experiments (Fig. 1A2), 35% (or 179 neurons) of PRH cells were backfired from the LA and 54% (or 274 neurons) were backfired from the PRH cortex. However, we did not encounter PRH cells that were antidromically responsive to both LA and PRH stimulation sites.

Conduction times from the LA nucleus to the PRH cortex

Figure 3A1 illustrates a representative experiment where six of eight microelectrodes aimed to the LA were histologically confirmed to have reached their target (2 in one hemisphere and 4 in the other). The graph of Fig. 3A1 shows how the antidromic response latency of LA cells (y-axis) fluctuated as a function of the distance between the recording and PRH stimulation sites (x-axis). In this experiment, note that irrespective of distance, antidromic response latencies varied widely within most microelectrode tracks (maximal range in one track: 25 ms). However, no clear change in the range of antidromic response latencies was observed as the distance increased.

View larger version (19K): [in this window] [in a new window]   Fig. 3. Conduction times from the LA to the PRH cortex do not vary with distance. A: graph plotting antidromic response latencies (y-axis) as a function of distance between LA cells and PRH stimulation sites (x-axis). A1: results obtained in one experiment. A2: results of all experiments combined. B: example of LA neuron backfired from two widely separated PRH stimulation sites. See inset for the relative position of the stimulating and recording sites. Heading in A1 refers to the direction of the pathway, not the direction of the back-propagating action potentials.

One possible explanation for the latency variations observed within each track would be that as the electrodes moved dorsoventrally in the LA, they sampled neurons located at varying distances from the PRH cortex. If this was true, one would expect antidromic response latencies to decrease gradually as the electrode approached the dorsal level of the PRH cortex and then increase progressively as it moved further ventrally. However, no such relationship was seen. In fact, neurons whose antidromic response latency varied by 10 ms were routinely recorded within 100 µm of each other.

Nevertheless, to address this issue quantitatively, we computed the average latency difference between successively recorded neurons in eight electrode tracks during which 10 or more PRH-projecting cells were recorded. Then, we compared this average to that seen in 500 random distributions of the original latencies. The rationale for this analysis is that if latency variations are caused by changes in the depth of recorded neurons with respect to that of the PRH cortex, then the average latency difference between successively recorded neurons should be smaller than that observed randomly. At odds with this idea, however, the average latency difference was situated within 1.18 SD of that found in randomly generated distributions (1.64 SD is required to reach statistical significance in a one-tailed t-test at P < 0.05). This result suggests that variations in the depth of recorded neurons do not account for latency variations observed within the microelectrode tracks.

As shown in Fig. 3A2, similar results were obtained in all experiments. In this graph, the antidromic response latency of 215 LA neurons recorded in 19 microelectrode tracks is plotted as a function of their distance to PRH stimulating electrodes. A negligible correlation (r = 0.025, n = 220) was found between conduction times and distance. Note that the discrepancy between the number of latencies (n = 220) and LA neurons (n = 215) results from the fact that five LA cells were antidromically invaded from two PRH sites.

Such a LA neuron is shown in Fig. 3B. Its antidromic response latencies to two distant (9 mm) PRH stimulation sites (Fig. 3B inset, S1 and S2) differed by <1 ms. This is particularly striking because this cell (*, inset, Fig. 3B) was much closer to S1 than S2. In fact, had the conduction time to S2 been proportional to the response latency observed with S1, the latency difference between responses to S1 and S2 should have been 33 ms. In the five neurons responsive to two PRH sites, the expected latency difference ranged widely (between 13 and 35 ms) because of large cell-to-cell variations in conduction times. Yet, the observed latency difference averaged 1.2 ± 2.05 ms. Thus, even if each of these five cells showed minimal changes in conduction times with distance to target, there were large between-cell variations in conduction times for similar distances to target. In both respects, therefore, the conduction times of these neurons are consistent with that seen in the rest of our sample.

Antidromic response latency of intrinsic PRH axons

In the preceding experiments, it is conceivable that the lack of relationship between distance and conduction time was due to the artifactual backfiring of LA cells by current spread from PRH stimulation sites. However, this possibility would seem unlikely if, using the same methods, we could provide evidence that a different axonal system shows linear increases in latency with distance.

To test this issue, we examined conduction times within the PRH cortex itself. Indeed, the PRH cortex is endowed with an intrinsic system of longitudinal connections that spans the entire rostrocaudal extent of the PRH cortex (Witter et al. 1986). Moreover, two recent studies have revealed that the latency of responses mediated by these intrinsic PRH connections increases with distance (Biella et al. 2001; Martina et al. 2001).

In support of our findings, the antidromic response latency of PRH neurons to electrical stimuli applied in the PRH cortex increased linearly with the distance between the stimulating and recording sites. The results of a representative experiment are shown in Fig. 4A. In this cat, six microelectrode tracks were confirmed histologically to have coursed in the PRH cortex. Despite significant variations in response latency within each electrode track, a strong correlation (r = 0.81; P < 0.05) was found between conduction times and distance. In further contrast with LA to PRH projections, the latency variations observed within each electrode track seemed to increase with distance. Identical results were obtained when the results of all experiments were combined (Fig. 4A2). In this context, it is interesting to note that the associational longitudinal pathway of the dentate gyrus also shows linear increases in conduction times with distance (Hetherington et al. 1994).

View larger version (22K): [in this window] [in a new window]   Fig. 4. The conduction times of intrinsic PRH axons increase with distance. A: graph plotting antidromic response latencies (y-axis) as a function of distance between LA cells and PRH stimulation sites (x-axis). A1: results obtained in 1 experiment. A2: results of all experiments combined. In A2, the points represent averages of antidromic response latencies (±SE) in all cells recorded during each electrode track. B: graph plotting antidromic response latencies (y-axis) as a function of distance between PRH neurons and LA stimulation sites (x-axis). Headings in A1 and B refer to the direction of the pathway, not the direction of the back-propagating action potentials.

Conduction times from the PRH cortex to the LA

During the aforementioned series of experiments, we also examined the conduction times of PRH neurons projecting to the LA. To this end, one or more stimulating electrodes were inserted in the LA using an oblique approach coursing through the contralateral hemisphere (Fig. 1B3).

As was found in LA neurons projecting to the PRH cortex, antidromic response latencies varied widely within most microelectrode tracks (maximal range in one track: 23 ms), irrespective of the distance separating the stimulating and recording sites (Fig. 4B). Also analogous to amygdalo-PRH projections, these variations were unrelated to the depth of PRH cells. This was determined by computing the average latency difference between successively recorded neurons in six electrode tracks that coursed along the lateral bank of the rhinal sulcus and during which we recorded 10 neurons or more. Then, we compared the average latency difference between successively recorded neurons to that seen in 500 random distributions of the original latencies. On average, the actual average latency difference was situated within 0.2 SD of that found in randomly generated distributions, thus suggesting that the depth of recorded neurons does not account for latency variations observed within the microelectrode tracks.

However, in contrast to amygdalo-PRH projections (Fig. 3A) but consistent with previous findings (Lang and Paré 1997), a low but statistically significant correlation (r = 0.27; P < 0.05) was found between conduction times and the distance separating the LA stimulating and PRH recording sites (Fig. 4B).

The possibility that the varying conduction latencies seen between the LA and PRH cortex resulted from stimulation or recording of different PRH layers seems remote. Although we used the lowest possible stimulation intensities, it is unlikely that we activated axons located in specific PRH layers. Even if we did, the difference in axon length between different layers represents a negligible proportion of the total axonal length, whereas latencies varied several fold. Moreover, a similar range of conduction times was seen in electrode tracks that were parallel or perpendicular to the lamination of the PRH cortex.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study was undertaken to determine whether the conduction times of LA axons are fixed or whether they increase with the distance they must cover to reach their PRH targets. The importance of this issue derives from the hypothesized involvement of the amygdala in promoting memory storage in the PRH cortex and the possibility that Hebbian mechanisms are at play.

Although large variations in conduction times were seen, they were not related to the distance between LA neurons and their PRH projection site. In contrast, a robust correlation was found between conduction times and distance in the intrinsic longitudinal connections of the PRH cortex and a weak one in PRH projections to the LA. In the following account, we will discuss these findings in light of available anatomical and physiological data and consider their significance for amygdalo-PRH interactions.

Conduction times from the LA nucleus to the PRH cortex are not related to distance

Overall, the conduction times of LA axons were independent of distance to target. This finding is intriguing because it implies that axonal conduction velocities and/or trajectories are adjusted to compensate for the varying distances separating LA neurons from their PRH targets.

Several factors suggest that this result was not due to the spurious activation of LA axons by current spread. First, if this was the case, antidromic responses from distant PRH sites should have required higher current intensities. However, a similar range of stimulation intensities was used at all stimulation sites. Second, if current spread was involved, a high proportion of LA neurons antidromically responsive to multiple PRH stimulation sites should have been observed. Yet, only 2% of LA neurons responded to more than one PRH stimulation site. Third, direct neuronal activation by current spread is essentially instantaneous, in sharp contrast with the range of antidromic response latencies observed here. Fourth, using identical methods, we observed that the conduction time of PRH neurons projecting to other rostrocaudal levels of the PRH cortex increased linearly with distance. Thus we feel confident that current spread was not involved and that the conduction times of LA neurons to the PRH cortex are truly independent of distance.

In support of this view, there are numerous examples of central and peripheral axonal systems showing constant conduction times despite varying distances to target (Bennett 1970; Carr and Konishi 1990; Sugihara and Lang 1993). In the olivocerebellar pathway for instance, longer climbing fibers conduct faster than shorter ones (Sugihara and Lang 1993), in keeping with the idea that this pathway serves as a timing circuit for motor control (Lang 2001; Llinás et al. 1974). Similarly, entorhinal sharp potentials, a population phenomenon generated by neurons of the basolateral amygdaloid nucleus, occur synchronously throughout the rostrocaudal extent of the entorhinal cortex (Paré et al. 1995) despite the fact that rostral entorhinal sites are adjacent to the amygdala, whereas caudal ones are 1 cm away in the cat.

Uniform range of conduction times of LA axons allows for a synchronous activation of distributed PRH sites

This study originated from pilot experiments where we observed that electrical stimulation of the LA nucleus elicited a PRH field potential whose latency did not vary with rostrocaudal distance (unpublished observations). However, because multiple circuitous pathways might have given rise to this phenomenon, we opted to study the latency of antidromic responses, an easier phenomenon to interpret. Other studies support the idea that LA inputs favor synchronized neuronal activity among distributed PRH sites. Indeed, multiple simultaneous recordings of LA and PRH neurons at various rostrocaudal levels have revealed that they show phase-locked oscillatory activity under anesthesia (Collins et al. 2001).

It might seem paradoxical to state, on the one hand, that LA inputs can promote synchrony in the PRH cortex, and, on the other, that the conduction times of LA neurons to the PRH cortex vary widely. However, the contradiction is only apparent because, instead of increasing with distance, the range of conduction times is the same irrespective of distance to target. Stated otherwise, it is not the variations in conduction times that promote synchrony, but the fact that the range of conduction times is unrelated to distance.

Consequences for amygdaloperihinal interactions

What could be the significance of this physiological arrangement in LA projections to the PRH cortex? To address this issue, we now turn to the anatomical organization of the PRH cortex and amygdalocortical pathways. In the following paragraphs, the term PRH cortex designates the peri- and postrhinal regions (reviewed in Burwell 2000; Witter et al. 2000), for simplicity.

The PRH cortex relays most neocortical sensory inputs to the entorhino-hippocampal system and constitutes the main return path for hippocampo-entorhinal efferents to the neocortex (Burwell and Amaral 1998a,b; Deacon et al. 1983; Insausti et al. 1987; Room and Groenewegen 1986; Shi and Cassell 1999; Van Hoesen and Pandya 1975; Witter and Groenewegen 1986; Witter et al. 1986). Neocortical inputs are organized topographically with rostral cortical areas focusing on rostral PRH levels and posterior ones mainly targeting caudal parts of the PRH cortex (Deacon et al. 1983; Room and Groenewegen 1986).

Superimposed on these transverse neocortical projections is an intrinsic system of longitudinal connections that span the entire rostrocaudal extent of the PRH cortex (Witter et al. 1986). Recently, it was shown that these intrinsic pathways can distribute neocortical inputs reaching different transverse levels of the PRH cortex in the rostrocaudal axis (Biella et al. 2001; Martina et al. 2001). Surprisingly, these long-range horizontal connections do not engage inhibitory interneurons (Martina et al. 2001). These observations and reports of N-methyl-D-aspartate (NMDA)-dependent long-term potentiation (LTP) in the PRH cortex (Bilkey 1996; Ziakopoulos et al. 1999) have led to the suggestion that the PRH network is biased to favor Hebbian associative interactions between coincident and spatially distributed inputs (Martina et al. 2001).

In light of the present findings, an intriguing possibility would be that the LA facilitates this process. This idea is supported by the excitatory nature of amygdalocortical projections. Indeed, amygdalocortical pathways originate from spiny projection neurons (McDonald 1992a,b) whose axon terminals are enriched in glutamate and form asymmetric (presumably excitatory) synaptic contacts, generally with spiny cortical cells (Paré et al. 1995; Smith and Paré 1994). Considering that cortical inhibitory interneurons generally do not bear spines whereas principal cells do (Freund et al. 1983; Ribak 1978), these data suggest that LA axons mostly form excitatory (glutamatergic) synapses onto excitatory cortical neurons.

How would the depolarizing inputs originating in the LA facilitate plastic events in the PRH cortex? Considering the physiological status of PRH neurons in conscious animals provides insight into this issue. In the waking state, PRH cells have much lower firing rates (Collins and Paré 1999) than neocortical neurons (Steriade 1978; Steriade et al. 1974), suggesting that their membrane potentials are relatively hyperpolarized. Because depolarization beyond a certain threshold is required for NMDA-dependent LTP, we hypothesize that LA inputs push PRH neurons in a range of membrane potential where Hebbian interactions are facilitated.

Another difficulty is to reconcile the uniform conduction times of LA axons with the progressive increase in antidromic response latency of intrinsic PRH axons with distance to target. Indeed, with respect to Hebbian plasticity, what would be the significance of a synchronized LA input to the PRH cortex if synaptic activity continued to propagate within the PRH cortex after the arrival of LA inputs? In this context, it is interesting to note that even though the conduction times of LA axons are uniform irrespective of distance to target, they do show considerable temporal dispersion. Thus we speculate that this scatter is essential to compensate for the conduction delays within the PRH cortex.

These considerations lead us to propose the following scenario for LA-PRH interactions. Emotionally arousing conditions, through still unidentified mechanisms, would increase the excitability of LA projection cells (Paré and Collins 2000). In parallel, the LA nucleus and PRH cortex would receive excitatory inputs from an overlapping set of high-order cortical areas (Deacon et al. 1983; McDonald 1998; Room and Groenewegen 1986). Excitation would propagate within the LA nucleus and PRH cortex via their divergent intrinsic pathways (Krettek and Price 1978; Witter et al. 1986). However, the depolarization produced by LA inputs would push the membrane potential of PRH neurons into a range of values favoring NMDA-dependent LTP. Finally, the temporal scatter of conduction times in the LA to PRH pathway would lengthen the period of depolarization to compensate for conduction delays within intrinsic PRH pathways.

Conclusions

In summary, we hypothesize that the uniform but temporally dispersed conduction times of LA neurons generates short time windows facilitating Hebbian associations between spatially distributed activity patterns in the PRH cortex. However, a long tradition of research indicates that interfering with amygdala activity after encoding decreases recall at a later time (Cahill and McGaugh 1998; McGaugh 2000). Thus we speculate that this mechanism is used by the amygdala to modulate long-term memory storage in emotionally arousing conditions, not only during the initial encoding, but possibly later when memory traces are consolidated.