Cortical Map Reorganization Enabled by Nucleus Basalis Activity

See allHide authors and affiliations

Science  13 Mar 1998:
Vol. 279, Issue 5357, pp. 1714-1718
DOI: 10.1126/science.279.5357.1714


Little is known about the mechanisms that allow the cortex to selectively improve the neural representations of behaviorally important stimuli while ignoring irrelevant stimuli. Diffuse neuromodulatory systems may facilitate cortical plasticity by acting as teachers to mark important stimuli. This study demonstrates that episodic electrical stimulation of the nucleus basalis, paired with an auditory stimulus, results in a massive progressive reorganization of the primary auditory cortex in the adult rat. Receptive field sizes can be narrowed, broadened, or left unaltered depending on specific parameters of the acoustic stimulus paired with nucleus basalis activation. This differential plasticity parallels the receptive field remodeling that results from different types of behavioral training. This result suggests that input characteristics may be able to drive appropriate alterations of receptive fields independently of explicit knowledge of the task. These findings also suggest that the basal forebrain plays an active instructional role in representational plasticity.

The mammalian cerebral cortex is a highly sophisticated self-organizing system (1). The statistics of sensory inputs from the external world are not sufficient to guide cortical self-organization, because the behavioral importance of inputs is not strongly correlated with their frequency of occurrence. The behavioral value of stimuli has been shown to regulate learning in experiments conducted over more than a century (2). Recently, behavioral relevance has been shown to directly modulate representational plasticity in cortical learning models (3,4). The cholinergic nucleus basalis (NB) has been implicated in this modulation of learning and memory. The NB is uniquely positioned to provide the cortex with information about the behavioral importance of particular stimuli, because it receives inputs from limbic and paralimbic structures and sends projections to the entire cortex (5). NB neurons are activated as a function of the behavioral significance of stimuli (6). Several forms of learning and memory are impaired by cholinergic antagonists and by NB lesions (7). Even the highly robust cortical map reorganization that follows peripheral denervation is blocked by NB lesions.

Many studies using acute preparations have shown that electrical stimulation of the NB (8) or local administration of acetylcholine (ACh) (9) can modulate stimulus-evoked single-unit responses. The variability across studies in the direction, magnitude, and duration of the modulation has made it difficult to relate these effects to long-term cortical map plasticity (10).

To clarify the role of the NB in representational plasticity, we investigated the consequences of long-term pairing of tones with episodic NB stimulation. A stimulating electrode was implanted in the right NB of 21 adult rats. After recovery, animals were placed in a sound attenuation chamber and a pure tone was paired with brief trial-by-trial epochs of NB stimulation during daily sessions (11). The tone paired with NB stimulation occurred randomly every 8 to 40 s. Pairing was repeated 300 to 500 times per day for 20 to 25 days. The rats were unanesthetized and unrestrained throughout this procedure.

Twenty-four hours after the last session, each animal was anesthetized and a detailed map of the primary auditory cortex (A1) was generated from 70 to 110 microelectrode penetrations (12). During this cortical mapping phase, experimenters were blind to the tone frequency that had been paired with NB stimulation. The frequency-intensity response characteristics of sampled neurons were documented in every penetration by presentation of 45 pure tone frequencies at 15 sound intensities. Tuning curves were defined by a blind experienced observer (13).

Figure 1A illustrates the organization of A1 in a representative naı̈ve rat. The color of each polygon denotes each penetration's best frequency (BF), which is the frequency that evoked a neuronal response at the lowest stimulus intensity. The frequency representation is complete and regular in control rats. Each frequency is represented by a band of neurons that extends roughly dorsoventrally across A1. The 9-kHz isofrequency band, for example, is shaded light blue in Fig. 1A, and penetrations with a BF within a third of an octave of 9 kHz are hatched with white. Figure 1B shows the tips of the tuning curves recorded in every penetration. The tip of the “V” marks the BF; the width of the V denotes the range of frequencies to which the neurons at the site responded at 10 dB above threshold. In naı̈ve rats, BFs were evenly distributed across the entire hearing range of the rat, in accordance with the well-known tonotopic organization of A1 (14).

Figure 1

(A, C, E, andG) Representative maps of A1 that show the effects of pairing 9-kHz tones with electrical stimulation of the NB. (A) Representative map from an experimentally naı̈ve rat demonstrating the normal orderly progression of BFs recorded in the rat A1. Each polygon represents one electrode penetration. The color of each polygon indicates the BF in kilohertz. The polygons (Voronoi tessellations) were generated so that every point on the cortical surface was assumed to have the characteristics of the closest sampled penetration. Hatched polygons designate sites with BFs within one-third of an octave of 9 kHz, illustrating a typical isofrequency band. Penetrations that were either not responsive to tones (O) or did not meet the criteria of A1 responses (X) were used to determine the borders of A1. (C) Map of A1 after pairing a 250-ms 9-kHz tone with NB stimulation. (E) Map of A1 after pairing a train of six 9-kHz tones with NB stimulation. (G) Map of A1 after pairing both 9- and 19-kHz tones with NB stimulation. The expansion of the 9-kHz isofrequency band is shown in (C), (E), and (G). Scale bar, 200 μm. (B, D, F, andH) Distribution of tuning curve tips at every A1 penetration from each map, which indicate the BF, threshold, and receptive field width 10 dB above the threshold for neurons recorded at each penetration. Threshold as a function of frequency (in kilohertz) matches previously defined behavioral thresholds. Solid vertical lines mark the frequency paired with NB stimulation. Dotted vertical lines mark frequencies presented as often as, but not paired with, stimulation.

Pairing a specific tonal stimulus with NB stimulation resulted in remodeling of cortical area A1 in all 21 experimental rats. In the representative example shown in Fig. 1C, a 50-dB 9-kHz tone was paired with NB stimulation approximately 300 times per day over a period of 20 days. This treatment produced a clear expansion of the region of the cortex that represented frequencies near 9 kHz (Fig. 1C). Figure 1D illustrates the clustering of tuning curve BFs near the frequency that was paired with NB stimulation. After pairing, neurons from 20 of the penetrations into the conditioned map shown in Fig. 1C had BFs within a third of an octave of 9 kHz, compared to only 6 kHz in the equivalently sampled control map. The increase in 9-kHz representation resulted in a clear decrease in the area of A1 that responded to lower frequencies. In the control map, 22 penetrations had BFs less than 5 kHz, compared to only 4 penetrations in the conditioned map. It should be noted that the decrease in low frequency responses is not a consistent finding. In other examples, the low-frequency responses appeared unaltered and the representation of higher frequencies was decreased.

Because the tone paired with NB stimulation was well above the threshold, it was important to examine not only the shifts in the tuning curve tips but also the responses of cortical neurons to tones at the conditioned intensity. During pairing, many of the neurons with BFs different from 9 kHz were excited by the auditory stimulus because most rat A1 tuning curves broaden as intensity is increased. In the naı̈ve map, less than 25% of neurons within A1 responded to 9 kHz presented at 50 dB. By contrast, almost 50% of the conditioned cortex responded to the same stimulus.

Figure 2 summarizes the magnitude of representational changes that resulted from pairing one frequency with NB stimulation in 10 animals. Figure 2A represents data from seven naı̈ve controls and illustrates the average percent of the surface of A1 that responded to tones at any combination of frequency and intensity. Figure 2, C and D, shows the percent change relative to controls after pairing NB stimulation with 4-, 9-, or 19-kHz tones, respectively. In each case, the cortical area representing the paired stimulus nearly doubled. These results indicate that the responses of tens of thousands or hundreds of thousands of A1 neurons can be altered by pairing tones with NB stimulation in a passively stimulated animal.

Figure 2

(A) Percent of the surface of A1 that responds to pure tones at each combination of tone frequency and intensity. The average of seven experimentally naı̈ve animals is shown. (B through D) Percent change in the percent of the primary auditory cortex responding to tones after 1 month of 4-, 9-, or 19-kHz tones paired with NB stimulation (n = 4, 4, and 2, respectively). Each group showed a significant increase over controls in the percent responding to the conditioned frequency at 50 dB above the minimum threshold (P < 0.005, two-tailed t test). The percent of A1 responding is the sum of the areas of all of the Voronoi tessellations that responded to the particular frequency and intensity combination of interest, divided by the total area of A1. The function is highly reproducible across naı̈ve controls with an average standard error across frequencies of less than 3%. Tessellation was chosen to derive area measurements from discretely sampled points by assuming that each location on the cortical surface had the characteristics of the closest sampled penetrtion.

In four animals, NB stimulation was paired with a train of six 9-kHz tone pips (25 ms) presented at 15 Hz to test the effects of increasing temporal structure in the auditory stimulus (Fig. 1, E and F). Conditioning with this stimulus unexpectedly resulted in even greater cortical reorganization than conditioning with a 250-ms tone (P < 0.01, Fig. 3C). In the example shown, the 9-kHz isofrequency band was increased from roughly 250 μm wide in a naı̈ve A1 to more than 1 mm wide. After pairing, over 85% of A1 responded to 9 kHz at 50 dB. Additionally, 50% of A1 penetrations had best frequencies within one-third of an octave of 9 kHz, compared to less than 15% in the control animals. The extent of cortical map reorganization generated by NB activation is substantially larger than the reorganization that is typically observed after several months of operant training (15-17).

Figure 3

(A) Percent of the surface of A1 that responds to pure tones of any combination of tone frequency and intensity. The average of seven experimentally naı̈ve animals is shown. (B) Percent change in the percent of A1 responding after 1 week of pairing 9-kHz tone pip trains (15 Hz) with NB stimulation. There was a significant increase in the response to 9 kHz at 50 dB above the minimum threshold as compared to controls (t test; n = 2, P < 0.05). (C) Percent change in the percent of A1 responding after 1 month of pairing 9-kHz tone pip trains (15 Hz) with NB stimulation. There was a significant increase in the response to 9 kHz at 50 dB above the minimum threshold as compared to controls (n = 4,P < 0.00001). (D) Distribution of receptive field width BW10 for every A1 penetration for each of the four classes of experiments. Pairing one frequency with NB stimulation did not significantly effect the BW10 distribution relative to naı̈ve animals, whereas pairing two frequencies (4 and 14, or 9 and 19 kHz) or a 15-Hz train of stimuli caused receptive field width to be decreased and increased, respectively. The same effect is present in the distributions of BW20 to BW40. The dashed vertical line marks the mean of each distribution. Single units were sorted from the multi-unit data derived from the four naı̈ve animals (15 units) and from four train-conditioned animals (33 units). The mean BW10 for single units was also increased by 15-Hz train conditioning (0.91 versus 1.38 octaves, P < 0.005). This widening of tuning curves adds with the BF shifts to generate the large increase in the percent of A1 responding after train conditioning.

The six short tones presented at 15 Hz evoked less than 30% more spikes than did a single tone, because most rat A1 neurons do not follow onsets presented faster than 12 to 14 Hz (18). It seems unlikely that the larger reorganization evoked with stimulus trains is simply due to an increased cortical response to the stimuli.

Two animals were mapped after only 1 week of conditioning with the 15-Hz stimulus to examine the rate of cortical remodeling evoked by NB activation. The 9-kHz representation was increased by 18% after 1 week of training. This reorganization was nearly halfway to the 44% increase that was recorded after a month of conditioning, indicating that the cortical remodeling generated by NB stimulation was progressive in nature (Fig. 3B).

To probe the competitive processes underlying cortical reorganization, five rats were conditioned with two different randomly interleaved tones that were more than an octave apart. Two distinct classes of reorganizations resulted. The tuning curve tips were either shifted toward a point between the two conditioned frequencies, so that both were within the receptive field at 50 dB (n = 3), or shifted toward only one of the two conditioned frequencies (n = 2, Fig. 1, G and H). The two classes of results may be the consequence of subtle variations in A1 before NB pairing, which can have large effects when competitive processes are involved.

To document the fact that NB activation is required for the cortical reorganizations observed in this study, during four of our experiments two additional frequencies were delivered on identical presentation schedules as the paired tones but were not paired with NB stimulation. These stimuli, which never occurred within 8 s of NB stimulation, did not measurably affect cortical responses or representations (19).

Microdialysis experiments have shown that electrical stimulation of the NB results in ACh release in the cortex (20). Additionally, both the short-term plasticity and the electroencephalogram (EEG) desynchronization evoked by NB stimulation are blocked by atropine (21). Thus, the cortical plasticity demonstrated in this study likely involves the release of cortical ACh paired with tones. To test for the necessity of ACh release in our model, a 19-kHz tone was paired with electrical stimulation of the NB in animals with highly specific lesions of the cholinergic NB neurons (22). No significant increase in the 19-kHz representation was observed in lesioned animals. Even though ACh release is clearly important for NB function, it may be too simplistic to focus exclusively on ACh because only one-third of NB projection neurons are cholinergic (23). One-third use γ-amino butyric acid and the remaining third are uncharacterized. Future work is needed to elucidate the function of concurrent release of these transmitters in cortical plasticity.

The nature of the auditory stimuli paired with NB activation had a profound effect on the selectivity of cortical responses (Fig. 3D). Sharpness of tuning was quantified as the width of the tuning curve 10 dB above the threshold (BW10). When a 250-ms tone was used as the conditioning stimulus, the average BW10 was not significantly different from the average BW10 of control rats (0.93 versus 1.02 octaves). Conditioning with a temporally modulated stimulus (a train of six short tones of the same frequency) resulted in a mean cortical response that was less selective than in controls (1.46 octaves, P < 0.0001). Conditioning with two tones engaging different spatial locations on the input array (the cochlea) resulted in cortical responses that were more selective than in controls (0.70 octaves, P < 0.0001). Thus, our model results in receptive fields that are narrowed, broadened, or unaltered depending on specific parameters of the acoustic stimulus paired with NB stimulation.

Similar increases and decreases in receptive field sizes have been recorded in the somatosensory and auditory cortices of New World monkeys that have been trained at tactile or auditory discrimination, detection, or time-order judgment tasks (4). A pure tone discrimination task or a task involving a stimulus that moved across several fingers decreased receptive field diameters by approximately 40% (15, 16). In contrast, a task that required detection of differences in the amplitude modulation rate of tactile stimuli delivered to a constant skin surface increased receptive field diameters by more than 50% (17).

The mechanisms responsible for remodeling receptive fields in a manner that is appropriate for the particular task that an animal practices are not well defined. One possibility is that top-down instruction from a higher cortical field with explicit knowledge of the goals of the operant task directs cortical plasticity. The fact that our simple model, without any behavioral task, can generate the same receptive field effects as are induced by extended periods of operant training suggests that the characteristics of the stimuli paired with subcortical neuromodulatory input are sufficient to determine the direction of receptive field alterations (24).

Adult cortical plasticity appears to be responsible for improvements in a variety of behavioral skills, maintenance of precise sensory representations, compensation for damage to sensory systems, and functional recovery from central nervous system damage (4). Our results suggest that activation of the NB is sufficient to guide both large-scale cortical reorganization and receptive field reorganization to generate representations that are stable and adapted to an individual's environment by labeling which stimuli are behaviorally important.


Stay Connected to Science

Navigate This Article