Specialization of primary auditory cortex processing by sound exposure in the “critical period”

Abstract

Environmental acoustic exposure to a complex tone sequence within the critical period in infant rats resulted in the emergence of large-scale, selective changes that radically altered primary auditory cortex (A1) organization. In the sound exposure-revised A1, responses were segregated into two explicit zones representing spectrally and temporally separated lower and higher frequency tone sequence progressions. Cortical neurons between these two A1 zones were poorly driven by sound stimuli. Stimulus sequence-specific (“combination-selective”) responses emerged in the A1 of exposed rats. These selective representational changes induced in the critical period persisted into adulthood. These results show that the temporal order and pace of early, repetitive postnatal auditory inputs strongly affect the emergent and enduring functional organization of A1.

Auditory experience plays an important role in the development of the primary auditory cortex (A1) (1–10). Exposure of kittens or rat pups to a modulated tonal stimuli during an early postnatal period resulted in an expansion of the representation of the exposed sound frequency in A1 (2, 7). Synchronous electrical stimulation of a cochlear implant in congenitally deaf cats led to larger cortical regions producing middle-latency evoked responses (11). The formation of an orderly, continuous sound frequency (“tonotopic”) representation and a postnatal sharpening of tonal receptive fields in the rat A1 appeared to require exposure to normally variable auditory inputs during the “critical period” (ref. 1, but also see ref. 12 for a difference in cats). However, it remains unclear how the cortical representations of the dynamic aspects of complex sound are specifically shaped by the spectro-temporal patterns of auditory inputs during development. These studies were designed to initiate our attack on this important issue.

In this initial study, we exposed rat pups through a critical period epoch extending from postnatal day (P) 9 (before hearing is first functional, at ≈P12) to P30 (when the critical period is drawing to a close) to a tone sequence with two specific spectro-temporal patterns. The frequency representation of A1 was radically revised by the exposure to this complex acoustic stimulus. Cortical neurons developed a preference for the temporal sequences of sound that the infant rat was exposed to. Induced changes endured into adulthood. These results underscore the specific instructive role of early auditory experience in sculpting the specific circuits of the auditory system to selectively and enduringly represent inputs that are present in the criticalperiod acoustic environment.

Methods

Early Exposure of Rat Pups. All of the experimental procedures applied in this study were approved by the University of California San Francisco animal research committee. Litters of nine to twelve 9-day-old female rats (Sprague–Dawley) were placed with their mothers in a calibrated, sound-shielded test chamber from P9 to P30 in the presence of auditory stimulation consisting of a sequence of tone pips (Fig. 1). This stimulation consisted of two sets of tone sequences with distinct temporal orders: a set of pulsed low-frequency tones presented in the order 2.8, 5.6, and 4 kHz; followed after a brief pause and a larger sound frequency jump by a set of pulsed high-frequency tones presented in the order 15, 21, and 30 kHz. Each tone lasted 30 ms with an intensity of 65 dB sound pressure level (SPL) and delivered by a speaker in free field placed 20 cm above the rat arena. A 12-h light and a 12-h dark cycle were automatically maintained. No significant harmonics (<2%) were recorded in the chamber during stimulus delivery. We noted no abnormal behaviors of either the mother or the pups during exposure. Activities during waking and sleep behaviors of the rats indicated that the sound stimuli were not stressful. The exposure period was chosen to cover the critical periods of the development of the rat auditory cortex (2).

Fig. 1.

Spectrogram of the stimulus used for the auditory exposure. Colors indicate the sound intensity levels. The low-frequency tone sequence consisted of a 2.8 kHz tone at 0 ms, 5.6 kHz at 150 ms, and 4 kHz at 300 ms, which is followed by a train of high-frequency tones with 15 kHz at 500 ms, 21 kHz at 650 ms, and 30 kHz at 800 ms. The duration of each tone was 30 ms.

Electrophysiology. To examine responses in rat A1, rats were anesthetized with pentobarbital sodium (50 mg/kg). Parylenecoated tungsten microelectrodes (1–2 megaohms at 1 kHz) were advanced 500–600 μm below the pial surface as described (2). The neural signal was amplified (×10,000), filtered at 0.3–30 kHz, and acquired with a PC-based recording system (Tucker-Davis, Alachua, FL). Frequency/intensity response areas (tuning curves) were reconstructed in detail by presenting 60 puretone frequencies (1–30 kHz, 25-ms duration, 5-ms ramps) at each of eight sound intensities (0–70 dB SPL in 10-dB increments) to the contralateral ear at a rate of two stimuli per second, by using a calibrated sound-delivering system with a custom-made tube speaker inserted into the ear. The ages of the studied rats ranged from P16 to P150. The multiunit spike responses within a latency window of 7–30 ms were collected for each penetration site. Characteristic frequency (CF) was defined as the frequency at which responses could be evoked at the lowest stimulus intensity. Locations of the penetrations were referenced by using the cortical surface vasculature as landmarks.

Nonresponsive sites with unusually high thresholds or responses not distinguishable from basal activity were considered to be non-A1 sites. Cortical maps were generated by Voronoi tessellation to divide the auditory cortex into polygons representing each recording site with color indicating CF.

Results

Segregation of Adult Primary Auditory Area in Exposed Rats. Auditory cortical organization was examined by recording tone-evoked responses from unit clusters in the middle layers defined in 60–140 microelectrode penetrations introduced into the auditory-responsive cortex in each animal. Detailed maps were obtained from groups of five or more experimental and control rats at each studied postnatal benchmark age (centered at P21, P35, and P100; see Methods). Although naive control rats showed an orderly, continuous A1 tonotopicity (Fig. 2a), A1 in exposed P35 and P100 adult rats was marked by two distinct tone-responsive sectors (Fig. 2b): a zone with neurons responding to low-frequency sounds (CFs predominantly near 2 and 7 kHz), and a second zone of neurons selective for high-frequency sounds (CF >20 kHz). Interestingly, between these two zones, cortical neurons had weakly driven and poorly tuned sound-evoked responses (Fig. 2 b-II and b-III). This relatively unresponsive area corresponded to the region representing middle frequency tones (CFs between ≈9 and 18 kHz) in naive control rats. It sharply segregated low- and high-frequency A1 sectors representing the sound frequencies of the two environmentally delivered tone sequences. In the low-frequency A1 zone, in striking contradistinction to control rats, neurons at many cortical sites were also characterized by double-peaked tuning curves, again mainly centered on ≈2.0 and 7.0 kHz (Fig. 2 b-I). Double-peaked receptive fields were rarely recorded in control rats. Whereas high-frequency responses in experimental animals were single-peaked as a rule, the average bandwidth of tuning curves in the high-frequency A1 zone was significantly broader than in controls, (bandwidth 10 dB above threshold Q₁₀ = 1.75 ± 0.09 octaves in exposed rats vs. 0.92 ± 0.05 octaves in controls at P100; see Fig. 2 a-II and b-II). A similar broadening of tuning curves was recorded in rats that were exposed to a single pure tone stimulus during the critical period (2).

Fig. 2.

(a and b) Representative tonotopic organizations of the auditory cortex from naive (a) and critical period-exposed P100 (b) (adult) rats. The color of each polygon indicates the CF (kHz) for neurons recorded at that site. ×, Cortical sites that were slightly responsive to tones; •, sites that did not meet the criteria for A1 responses (see Methods). In the diagonally hatched areas, receptive fields showed broader-than-normal tuning curves. Typical examples of tonal receptive fields recorded from cortical sites in the two rats are shown, with numbers indicating their locations in the maps. CFs and secondary peaks (in exposed rats) of receptive fields are indicated by dotted lines. Note that in exposed rats, receptive fields were marked by multiple peaks, broader-than-normal bandwidths, and poor tuning.

As a summary, in the distinct lower frequency sector of A1, cortical representations of frequencies just below tone 1 (t1) and just above tone 2 (t2) of sequence 1 (s1) were significantly larger than in controls. Interestingly, there was a significant decrement of the representation for tone 3 (t3) of s1 (Fig. 3). In addition, increased cortical representation of high frequencies (>20 kHz) was found in the higher frequency A1 sector. The best frequency-defined representations of both t1 and t2 of sequence 2 (s2) were smaller than in controls.

Fig. 3.

Summary of the effects of early exposure to tone sequences on the adult A1. Graph shows the percentage difference between exposed and normal rats in the area of cortical representations by using CFs. Bin size for each frequency range is 0.6 octaves. Areas of representation for bands centered at 1.7 kHz and 5.9 kHz were significantly increased in exposed rats, whereas the representations of 3.9- and 9-kHz centered bands were reduced (P < 0.01, t test).

These results manifest a significant instructive role for complex sound input patterns for auditory cortex development. The decreased cortical representation of mid-frequencies in sound-exposed rats plausibly results from competitive interactions between these two complex auditory input sequences. The different representations of tone frequencies in the exposed sound sequence indicate that the temporal order of sound inputs and the temporal interaction of neuronal activity driven by them must play an important role in the functional processing machinery in A1.

Development of the Primary Auditory Cortex in Exposed Rats. The segregation of A1 recorded in early exposed adult rats could be largely attributed to experience-dependent cortical changes during development, which was marked by the slow emergence of the weakly driven or poorly tuned “competitively suppressed response zone” within A1 (Fig. 4). Neurons that were not effectively driven by simple tonal stimuli were recorded within this zone by P23. At P35, the regions of poor tuning between the two tone-responsive zones usually completely segregated the two tone-driven A1 sectors (Fig. 4 a and b). These changes in cortical representation during development were accompanied by the progressive emergence of double-peaked tone-evoked receptive fields in the low-frequency zone of A1 (Fig. 4c). Double-peaked responses were relatively infrequently recorded in the P21 animals, whereas they predominated in the low-frequency zone of P35 animals. All of these changes were as strongly recorded in P100 (adult) as in P35 animals.

Fig. 4.

Tonotopic organization of A1 of exposed rats at different postnatal ages. (a and b) Examples are auditory maps from two different rat litters reared in the critical period (P9–P30) in the presence of these sequences of sound stimuli. Postnatal ages of rats are indicated at the top of each map. (c) Representative tonal receptive fields obtained from A1 at different ages. Dotted lines indicate the positions of peaks within the receptive field. (d) Distribution of CFs (represented by solid dots) and frequencies of secondary peaks (▵) along the tonotopic axis of the auditory cortex at different developmental stages. The line connecting the two most anterior and posterior penetrations (normally exhibiting the lowest and highest CFs, respectively) within A1 was used to define a normalized tonotopic axis. (e) Distribution along the tonotopic axis of the auditory cortex in control naive adult rats (P100).

CFs (two CFs, if there were double peaks) along a normalized tonotopic axis for neurons recorded in all of the penetration sites obtained from rats in different age groups are shown in Fig. 4d. Those plots again show that CFs and secondary tuning curve peaks in A1 were sharply clustered at 2 and 7 kHz as well as at frequencies >20 kHz, indicating a specific and persistent alteration of A1 tonotopicity induced by this early auditory stimulus exposure.

Selectivity for Temporal Patterns of Early Auditory Input. Previous studies have shown that the temporal pattern of sensory inputs is important for cortical development (13–18). To examine whether specific temporal patterns of auditory input also play an instructive role on the development of selective cortical representations of successive dynamic features in sound inputs, we compared cortical responses to forward and reversed versions of the tone sequence patterns used for developmental exposure. The sampled sites were considered sequence selective when the response evoked by exposed sequence is >25% stronger than that evoked by reversed tone sequences. At a typical cortical site preferring lower frequencies (Fig. 5a), neurons exhibited reliable responses to each of the three individual low-frequency tonal stimuli when the tones were presented in the forward sequence; whereas for the reversed sequence, the second tone rarely evoked any response (Fig. 5b). Thus, even while the tonotopic maps indicated that this 4-kHz tone was “underrepresented,” this specific stimulus strongly engaged the cortex when the tones were presented in the sequence order that was applied for the stimulus sequence that rat pups were exposed to. The same results were recorded for high-frequency tone sequences (Fig. 5 c and d). There, forward tone sequences consistently evoked stronger and more reliable responses than reversed sequences, again indicating that cortical neurons had developed selective responses for the temporal pattern of tone sequences used in developmental exposure. Overall, ≈23% of the sampled cortical sites in exposed rats showed a clear selectivity for forward tone sequences (Fig. 5e). This was very different from control rats (P < 0.01, t test), in which only a very small percentage of cortical sites exhibited selectivity for the testing sequences and in which neurons at an equal number of sites preferred reversed sequences. Again, these induced changes were as strongly expressed in adult (P100) as in P35 rats. These results indicate that the temporal patterns of early auditory inputs are important for instructing the development of cortical representation of successive temporal features of sound signals in A1.

Fig. 5.

Temporal selectivity of cortical neurons in exposed and naive control rats. (a and b) An example from the low-frequency zone in A1 of an exposed rat. (a) Raster plot of spiking responses evoked by exposed tone sequences. (b) Responses evoked by a stimulus in which the tone sequences were presented in reversed order. (c and d) An example of recording from the high-frequency zone. (e) Percentage of temporal-sequence selective cortical sites in exposed and naive control rats. The sampled sites were considered sequence selective when the response evoked by exposed sequence is >25% stronger than that evoked by reversed tone sequences.

Discussion

These studies show that exposure to a specific complex acoustic stimulus in the early postnatal critical period results in large-scale remodeling of A1 neuronal response selectivity. The tonotopic organization of A1 is distorted by this complex stimulus exposure, to elaborate responses to specific sound stimuli and to weaken the representations of others; an unresponsive zone emerges between the cortical zones of representation of hypothetically “competitive” low- and high-tone stimulus sequences; and neurons respond more reliably to the elements of these complex input sequences when they are delivered in the order that was applied in the exposure protocol. These large-scale changes all endured into adulthood without significant change, even while the exposure to this sound stimulus was terminated near the end of the critical period.

How did these changes arise? What specific plasticity mechanism(s) accounted for them? How do induced changes account for the emergent “combination selectivity” recorded in these exposed rats? In other plasticity studies in which single or multiple tonal stimuli have been delivered separately and in random order, very different representational changes were induced (1, 2, 7, 19–22).

We hypothesize here that these large-scale cortical representational changes can be attributed to cooperative and competitive interaction among convergent stimulus-driven inputs, which result in the enhancement or reduction of the functional representation of specific sound frequencies. The temporal order and rate of occurrences of stimulus elements appear to be major determinants of the specific dimension of A1 signal processing induced by sound exposure (“experience”) in the critical period.

The first evidence for temporal order/rate effects was expressed by the almost complete lack of responses selective for t3 of s1, whose representation would normally be located approximately midway in A1 between t1 and t2. This was consistent with one or both of these preceding stimuli temporally inhibiting t3-evoked responses, and/or, with t1 and t2 conditioning cortical network dynamics (e.g., suppressing ongoing background activity levels for the temporal epoch of t3 stimulation). As a consequence, the neurons originally tuned to t3 would have strengthened inputs representing t1/t2, and the heterosynaptic depression would hypothetically result in the progressive weakening of t3-related inputs (23). Interestingly, in striking contrast to selective representational expansion in rats exposed to isolated single tones (2, 7), the expanded representation here were not centered at t1 and t2 in s1, but just below t1 and just above t2. Because the frequency of t3 is represented in A1 midway between t1 and t2, the above difference must be attributed to the spatial and temporal interactions, among those inputs, and to an integration of various plasticity effects. Thus, these findings lead to the inescapable conclusion that predictable following (or possibly long-preceding) stimuli, i.e., stimulus context altered the plasticity outcome.

The similarly striking degradation of the representation of the range of frequencies lying between s1 and s2 and especially the extension of this relatively unresponsive zone into the A1 zone of representation of t1 of s2 is equally provocative. The jump in frequency from the last component of s1 to the first component of s2 was nearly two octaves. Even given this wide spectral (and distant A1 territorial) separation, there were, again, very large impacts for A1 plasticity of these stimuli being presented in this particular temporal order and pace. The suppression of this intermediate zone could arise from long-term depression, which is differentially strongly expressed in the neonatal rat brain (24), because this zone was not efficiently driven by either s1 or s2 stimuli during early development. It is also a surprise to see the striking loss of tuned responses to t1 of s2 instead of its over-representation. Given the wide frequency differences and representational distance separations and the perturbation in time interval between s1 and s2, it would be of great interest to determine how these specific plasticity outcomes relate to the specific parameters that apply within, the two sound input sequences.

Taken together, these studies demonstrate that the sequence order and timing of repetitive, complex stimuli strongly impact the specific representational changes induced by critical period plasticity in the auditory processing system. The sounds of words that bear meaning (phonemes) in an infant human's native language generate selective, phoneme-specific responses in A1 (25, 26). The specific changes in auditory cortex processing induced by such stimuli are likely to be highly dependent on the specific dynamics and contexts of the infant child's language-specific sound input exposure.

These studies also argue that any practical or theoretical interpretation of the impacts of experience on the shaping of auditory cortex representations, or any interpretation of how known plasticity mechanisms account for the representation of complex acoustic stimuli, must take both the statistics of acoustic input dynamics and cortical response dynamics into account.