Hearing Research 328 (2015) 1e7

Contents lists available at ScienceDirect

Hearing Research journal homepage: www.elsevier.com/locate/heares

Research paper

Auditory cortex directs the input-specific remodeling of thalamus Sultan L. Nelson 1, Lingzhi Kong 1, Xiuping Liu, Jun Yan* Department of Physiology and Pharmacology, Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary, Calgary, AB T2N 4N1, Canada

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 April 2015 Received in revised form 10 June 2015 Accepted 23 June 2015 Available online 2 July 2015

Input-specific remodeling is observed both in the primary auditory cortex (AI) and the ventral division of the medial geniculate body of the thalamus (MGBv) through motivation such as learning. Here, we show the role of AI in the MGBv remodeling induced by the electrical stimulation (ES) of the central division of the inferior colliculus (ICc). For the MGBv neurons with frequency tunings different from those of electrically stimulated ICc neurons, their frequency tunings shifted towards the tunings of the ICc neurons. AI neurons also showed this input-specific remodeling after ES of the ICc (ESICc). Interestingly, the input-specific remodeling of MGBv was eliminated when the AI was inactivated using cortical application of muscimol. For the MGBv neurons tuned to the same frequency as the stimulated ICc neurons, their tunings were kept but their responses were facilitated after the ESICc. In contrast to the input-specific tuning shifts, this facilitation was rarely impacted by the AI inactivation. Thus, we conclude that AI directs the input-specific remodeling of MGBv induced by ESICc. It is suggested that the tuning shift in the MGBv primarily takes place in the AI and is relayed to the MGBv through the corticofugal system while the MGBv mainly highlights the frequency information emphasized in ICc. © 2015 Elsevier B.V. All rights reserved.

Keywords: Auditory cortex Auditory thalamus Corticofugal Mouse Plasticity

1. Introduction Auditory learning and experience lead to frequency-specific plasticity both in cortex and thalamus, which functions as a plausible neural substrate for the learning-induced or experiencedependent changes in animal behaviors (Yan, 2003; Weinberger, 2004; de Villers-Sidani and Merzenich, 2011; Pantev and Herholz, 2011; Suga, 2012). Through the tonotopic loop formed by thalamocortical and corticothalamic projections, this input-specific remodeling in auditory cortex and thalamus can be aligned to ensure the coordinated behavioral consequences of learning (Zhang et al., 1997; Suga and Ma, 2003; Jafari et al., 2007; Suga, 2008; Zhang and Yan, 2008; Xiong et al., 2009). However, the

Abbreviations: AI, primary auditory cortex; BF, best frequency; ES, electrical stimulation; ESICc, electrical stimulation of the central division of the inferior colliculus; F-scan, frequency scan; GABAA, g-amino-butyric acid-A; ICc, central division of the inferior colliculus; MGBv, ventral division of the medial geniculate body of the thalamus; MT, minimum threshold; PSTC, post-stimulus time-cumulative; SD, standard deviation * Corresponding author. Hotchkiss Brain Institute, Department of Physiology and Pharmacology, Cumming School of Medicine, University of Calgary, 3330 Hospital Drive NW, Calgary, AB, T2N 4N1, Canada. E-mail address: [email protected] (J. Yan). 1 Sultan Nelson and Lingzhi Kong contributed equally to this work. http://dx.doi.org/10.1016/j.heares.2015.06.016 0378-5955/© 2015 Elsevier B.V. All rights reserved.

distinctive role of the auditory cortex and thalamus in frequencyspecific plasticity remains unclear. Behavioral studies show an interesting relationship between the auditory cortex and thalamus in terms of learning-induced changes. After auditory fear conditioning, animals demonstrate freezing response specific to the conditioned sound (Iwata and LeDoux, 1988; Stiedl and Spiess, 1997). When the auditory cortex is inactivated, the freezing response is retained but becomes generalized, i.e., not specific to the conditioned sound (Armony et al., 1997). Moreover, inactivation of the auditory cortex also eliminates animals’ ability of sound-frequency discrimination (Riquimaroux et al., 1991; Talwar et al., 2001). These findings suggest that the cortex is important for enabling the learning process specific to the acquired sound information while auditory thalamus is to capture the acquired sound information. Neurophysiological studies demonstrate that essential auditory information to establish conditioning or association primarily stems from the thalamus rather than from the cortex (Weinberger, 1998; Maren and Quirk, 2004; Sigurdsson et al., 2007). In addition, thalamocortical pathway contributes to the frequency-specific plasticity in the cortex (Jafari et al., 2007; Liu et al., 2015). On the other hand, our previous studies show that the frequency-specific plasticity of the auditory cortex can be relayed to the auditory thalamus and midbrain through the corticofugal systems (Zhang et al., 2005; Wu and Yan, 2007; Zhang and Yan, 2008). We,

2

S.L. Nelson et al. / Hearing Research 328 (2015) 1e7

therefore, hypothesize that the auditory thalamus primarily provides the thalamocortical system with acquired auditory information and the auditory cortex is responsible for the reorganization, i.e., the shift in neuronal frequency tunings in both auditory thalamus and cortex. To test this hypothesis, we examined the frequency-tuning change in the ventral division of the medial geniculate body (MGBv) and the primary auditory cortex (AI) following focal electrical stimulation (ES) of the central nucleus of the inferior colliculus (ICc). We found that the ESICc facilitated AI and MGBv neurons with best frequencies (BF) similar to the BFs of the stimulated ICc neurons while ESICc shifted the frequency tunings of AI and MGBv neurons that had BFs different from the ICc neurons. Importantly, when the AI was inactivated by the g-amino-butyric acid-A (GABAA) receptor agonist muscimol, the MGBv tuning shift disappeared while the ESICc-induced MGBv facilitation retained. 2. Materials and methods Thirty-eight female C57 mice, aged 4e7 weeks old and weighing 16.6e22.5 g (Charles River Laboratories), were used in our study. Animal use was in accordance with the Canadian Council on Animal Care, and the protocol (M10029) was approved by the Animal Care Committee at the University of Calgary. 2.1. Animal preparation The mice were anesthetized throughout the experiment with an intraperitoneal injection of a mixture of ketamine (85 mg/kg, i.p., Bimeda-MTC Animal Health Inc., Canada) and xylazine (15 mg/kg, i.p., Bimeda-MTC Animal Health Inc.). Additional doses of ketamine and xylazine (17 mg/kg and 3 mg/kg) were injected to maintain the anesthetic level by monitoring the tail pinch reflex. The surgical procedure was identical to that used in a previous study (Wu and Yan, 2007). Briefly, the mouse's head was firmly fixed using a custom-made head holder, and the skull was exposed. Three openings were made on the left skull above the AI (2.18e3.64 mm posterior to the bregma and 4e4.5 mm left to the midline), the MGBv (3.1 mm posterior to the bregma and 1.9 mm left to the midline) and the ICc (5.02e5.34 mm posterior to the bregma and 0.5e2.0 mm left to the midline), using the stereotaxic coordinates in the mouse brain map (Franklin and Paxinos, 1996). These openings were used for either the placement of the recording/ stimulating electrode or the local application of drugs (saline or muscimol) (Fig. 1A and B). During surgical procedure and electrophysiological recording, animal body temperature was maintained at 37
C using feedback-controlled heating pad. 2.2. Acoustic stimulation Tone bursts (60-ms duration, 5-ms rise and fall times) were generated by an Enhanced Real-time Processor (RP2, TuckereDavis Tech., Gainesville, FL, USA) and then fed to a free-field loudspeaker (ES1, TuckereDavis Tech., Gainesville, FL, USA) via a digital attenuator (PA5, TuckereDavis Tech., Gainesville, FL, USA). The speaker was placed 45
to the right of and 13 cm away from the mouse's right ear. The output amplitude of the loudspeaker was expressed as dB SPL within one decibel accuracy (reference 20 mPa), calibrated at the right ear of the animal using a condenser microphone (Model 2520, Larson-Davis Laboratories, USA) and a microphone preamplifier (Model 2200C, Larson-Davis Laboratories, USA). Frequencies and intensities of tone bursts were varied either manually or automatically using BrainWare software (TuckereDavis Tech., Gainesville, FL, USA).

Fig. 1. A, A schematic of the mouse brain with recording, electrical stimulation and drug injection sites (adapted from Luo et al., 2008). The ascending and descending projections of interest are indicated by solid dark lines. AI, primary auditory cortex; MGBv, ventral division of the medial geniculate body of the thalamus; ICc, central division of the inferior colliculus. B, A table illustrating the detailed procedures in the three protocols. C, Two examples of auditory responses recorded from one neuron in MGBv and one neuron in AI. The arrows represent the BFs of the two neurons.

2.3. Recording auditory responses in the AI, MGBv and ICc Tungsten electrodes (~2 MU impedance, FHC, USA) were used for recording neuronal activities. Action potentials (spikes) from the electrode(s) were fed to and digitized by a preamplifier (RA16, TuckereDavis Tech., Gainesville, FL, USA) that was connected to another RP2 via an optical cable. The BrainWare software, in addition to control tone stimulation, was also used to control the recording (10k times, 0.3e3 kHz bandpass). Action potentials together with tone information were saved in a DAM file for data analysis. Methods of determining the AI and targeting the MGBv/ ICc were identical to those described elsewhere (Jafari et al., 2007; Wu and Yan, 2007; Luo et al., 2008; Liu et al., 2015). A frequencyamplitude scanning with 1 kHz frequency step and 5 dB amplitude was used to sample the excitatory response areas. The lowest amplitude (minimum threshold, MT) and the corresponding frequency at which the neuron showed a response were determined. Then, a frequency scan (F-scan) was used for data sampling. In an Fscan, the frequency varied from 3 to 40 kHz with increments of 1 kHz and the amplitude was set at 20 dB above the MT. The tone frequency in an F-scan was randomly varied using BrainWare software. The interval between stimuli was 500 ms and the identical stimulus was repeated 20 times. Fig. 1C showed the frequency tunings of two neurons using F-scan, one from MGBv and the other from AI. The frequency to which the neuron showed the largest response was defined as the BF of the neuron.

S.L. Nelson et al. / Hearing Research 328 (2015) 1e7

2.4. ICc stimulation As soon as the auditory response properties, i.e., the BF and MT, were determined, the position of the electrode was secured and its connection was switched from the recording system to the output of a constant current isolator (A360, WPI Inc., Sarasata, FL, USA) for the electrical stimulation. This procedure ensured that the locations of the recording and stimulating sites were identical. The negative pulses (monophasic, 0.1 ms, 700 nA constant current) were generated by a stimulator (Grass S88, Natus Neurology, West Warwick, RI) and a constant-current isolator. During the electrical stimulation, the onset of the electrical pulse was synchronized with the onset of the tone at BF and 10 dB above the MT of the ICc neurons. The combined acoustic and electrical stimuli were repeatedly delivered at a rate of 4 Hz for 30 min. 2.5. Cortical application of saline or muscimol solution Once the location of the AI was established, the dura was carefully removed and a polythene-tube (1.5 mm in inner diameter and 1 mm in length) with a small amount of Vaseline at one end was gently placed on the surface of the brain. The Vaseline was used to seal the circumference around the tube and brain surface and to form a well that covered the AI. Muscimol solution (1 mg/ml, dissolved in 0.9% saline; Sigma, St. Louis, MO) or 0.9% saline was dropped into the well with a 1.0 ml syringe. The muscimol solution was maintained until the electrophysiological recording was finished. 2.6. Experimental protocols Three protocols were designed for examining the ESICc effects on MGB frequency tuning, on AI frequency tuning and on MGB frequency tuning with AI inactivation (Fig. 1). The procedures of three protocols were in general the same as described below. When recordings in the ICc and MGBv or AI were stable and neuronal frequency tunings were sampled by F-scan, the connection of electrode in ICc was switched from recording system to electrical stimulation system. Saline or muscimol were dropped to the well on AI. Five minutes later, the neuronal frequency tunings were sampled again in AI or MGBv before (control) and after ESICc. Sampling frequency tunings of AI or MGBv neurons continued every 30 min until any changes in frequency tunings of MGBv or AI neurons recovered by more than 50%. 2.7. Data acquisition and processing Cluster-cutting was applied to sort single-unit action potentials (spikes) from the recorded multiple unit data. Eight parameters of the action potential waveform were measured including peak, valley, spike height, spike width, peak time, valley time, and two user-defined voltages (Yan and Ehret, 2002; Yan et al., 2005). Single-unit responses to F-scan were displayed using post-stimulus time-cumulative (PSTC) histograms with a bin width of 1 ms. The BF was the frequency to which neurons showed the largest response. Changes in the frequency tunings of MGBv and AI neurons in response to the ESICc were expressed by the difference between the BF and spikes before and after the ESICc. Data were expressed as mean ± SD. The t-test (two-tailed) was used to compare the differences between groups of data. A p value of less than 0.05 was considered to be statistically significant. 3. Results In total, 81 neurons were sampled from 38 mice: 38 neurons/23 mice in Protocol 1, 22 neurons/8 mice in Protocol 2 and 21 neurons/

3

7 mice in Protocol 3. All neurons in the AI, MGBv and ICc were sharply tuned and their BFs were easily identified. Their neuronal BFs ranged from 11 kHz to 26 kHz. 3.1. Frequency-specific changes in MGBv by ESICc (protocol 1) ESICc induced noticeable changes in the frequency tunings of MGBv neurons when the AI function remained intact. Three MGBv neurons shown in Fig. 2 exemplify these changes. Neuron A tuned to 13 kHz (control) with a BF identical to that of the electrically stimulated ICc neuron (Fig. 2A). The ESICc did not change the BF of this neuron. It is however clear that the response of this neuron to the 13-kHz tone increased after ESICc (facilitation). Neuron B had a BF of 16 kHz (control) that was 5 kHz higher than the BF of the electrically stimulated ICc neuron (Fig. 2B). The ESICc shifted the BF of this MGBv neuron lower, from 16 kHz to 13 kHz. The shift in BF resulted from the increased response to 13 kHz and the decreased response to 16 kHz (inhibition). Neuron C, on the other hand, had BF lower than that of electrically stimulated ICc neuron (Fig. 2C). Its BF was 13 kHz (control) and the BF of the ICc neuron was 18 kHz. The ESICc shifted the BF of this neuron higher, from 13 kHz to 16 kHz. Similarly to the Neuron B, the BF shift of Neuron C also resulted from the facilitation of the response to the 16 kHz and the inhibition of the response to the 13 kHz. The three examples clearly demonstrated that the ESICceevoked changes in the frequency tunings of MGBv neurons could be expected according to the relationship of frequency tunings between the recorded MGBv and electrically stimulated ICc neurons. We therefore analyzed the changes in the frequency tunings of all samples as the function of the differences in the BFs of the MGBv and neurons. Fig. 3A shows that the BF changes of MGBv neurons were linearly correlated with the BF differences between the ICc and MGBv neurons (n ¼ 38, p < 0.001). The MGBv neurons were then classified into two different types: unmatched neurons and matched neurons. When the BF difference between the MGBv and ICc neurons was more than 1 kHz, the MGBv neuron was defined as physiologically unmatched neuron (filled circles in Fig. 3A). When the BF difference was less than or equal to 1 kHz, the MGBv neuron was defined as physiologically matched neuron (open circles in Fig. 3A). For the unmatched neurons with BFs lower than the ICc neurons, their BFs significantly shifted higher by 2.40 ± 2.10 kHz (n ¼ 15, p < 0.01). For the unmatched neurons with BFs higher than the ICc neurons, their BFs significantly shifted lower by 1.83 ± 2.10 kHz (n ¼ 12, p < 0.01). For the matched neurons, the ESICc did not change their BFs (n ¼ 11, p ¼ 0.724). Corresponding changes were also observed in the responses of the MGBv neurons (Fig. 3B). The auditory responses at the control BFs for the matched MGBv neurons significantly increased by 31.07 ± 15.45% (n ¼ 11, p < 0.001). In contrast to the matched neurons, the auditory responses at the control BFs significantly decreased by 24.17 ± 20.34% (n ¼ 15, p < 0.001) for the unmatched neurons with BFs lower than the ICc neurons and 32.72 ± 28.81% (n ¼ 12, p < 0.01) for the unmatched neurons with BFs higher than the ICc neurons. 3.2. Frequency-specific changes in AI by ESICc (protocol 2) Similar to what was observed in the MGBv, ESICc induced clear and ordered changes in the frequency tunings of AI neurons. Fig. 4 illustrates the changes in BFs and responses of AI neurons as the function of the BF differences between the stimulated ICc and recorded AI neurons. The changes in AI BFs were linearly correlated with the differences in BFs between ICc and AI neurons (n ¼ 22, p < 0.001). As with the MGBv neurons in protocol 1, they were classified into two types, unmatched neurons and matched neurons. The BFs of unmatched AI neurons significantly shifted higher

4

S.L. Nelson et al. / Hearing Research 328 (2015) 1e7

Fig. 2. Three examples (A, B, C) illustrating the effects of the ESICc on the frequency tunings of MGBv neurons (protocol 1). The ESICc did not change the BF but increased the auditory response of the matched MGBv neuron (A1, A2), while the BFs of the unmatched MGBv neurons shifted towards the ICc BFs (B1, B2 and C1, C2). The ESICc caused facilitation (A3, B3, C3) and inhibition (B3, C3) of MGBv auditory responses. The gray bars and the arrows represent the BFs of the stimulated ICc neurons. In the right column of panels, the dark gray areas represent facilitation, whereas the light gray areas represent the inhibition.

Fig. 3. Frequency-specific change in the BF and the auditory responses (spikes) of MGBv neurons after ESICc (protocol 1). Open circles and bars, BF changes of the matched neurons; filled circles and bars, BF changes of the unmatched neurons. A. The BF changes were systematically correlated to the BF differences between the ICc and MGBv neurons. B. The response changes were different between the matched and unmatched MGBv neurons at control BF and 20 dB above the MT. The error bars represent the SD.

Fig. 4. Frequency-specific change in the BF and the auditory responses (spikes) of AI neurons after ESICc (protocol 2). Open triangles and bars indicate the BF changes and the responses of matched neurons. Filled triangles and bars indicate the BF changes and the responses of unmatched neurons. A, The BF changes were systematically correlated to the BF differences between the ICc and AI neurons. B, The response changes were different between the matched and unmatched AI neurons at control BFs and 20 dB above the MT. The error bars represent the SD.

S.L. Nelson et al. / Hearing Research 328 (2015) 1e7

5

by 3.44 ± 1.24 kHz (n ¼ 9, p < 0.001) when the ICc BFs were higher than the AI BFs and significantly shifted lower by 4.50 ± 2.17 kHz (n ¼ 6, p < 0.01) when the ICc BFs were lower than AI BFs (filled triangles in Fig. 4A). For the matched AI neurons (opened triangles in Fig. 4A), their BFs did not change (n ¼ 7, p ¼ 1). The auditory responses of AI neurons to the control BFs were also facilitated for the matched neurons while suppressed for the unmatched neurons (Fig. 4B). On average, the responses significantly decreased by 30.29 ± 16.98% (n ¼ 9, p < 0.01) for the unmatched neurons with BFs lower than the ICc neurons and 52.18 ± 14.24% (n ¼ 6, p < 0.001) for the unmatched neurons with BFs higher than the ICc neurons. In contrast, the responses of the matched neurons significantly increased by 17.08 ± 13.72% (n ¼ 7, p < 0.05). 3.3. ESICc-evoked no MGBv tuning shift following AI inactivation (protocol 3) The findings presented above are in line with previous findings using either focal electrical stimulation or learning paradigm (Xiong et al., 2009). A critical issue raised here was how they contributed to the frequency-specific plasticity. Since the MGBv receives auditory inputs exclusively from the ipsilateral inferior colliculus (Kudo and Niimi, 1980), the protocol 3 allowed us to address this issue. Three examples are presented in Fig. 5. The neuron in Fig. 5A was tuned to 13 kHz, similar to the electrically stimulated ICc neuron. After the ESICc, the BF of this neuron remained the same while its response increased. Interestingly, ESICc induced only small changes in both BFs and responses of the neurons in Fig. 5 B and C; one had BF higher and the other lower than the BFs of the stimulated ICc neurons. In contrast to what we observed in protocol 1 and 2 (gray circles and gray triangles in Fig. 6A), the BF changes of the MGBv neurons in protocol 3 were not correlated with the BF differences between the MGBv and ICc neurons (n ¼ 21, p ¼ 0.875, Fig. 6A). As with the MGBv neurons in protocol 1, they were classified into two types: unmatched neurons and matched neurons. The BFs did not shift for

Fig. 6. Frequency-specific change in the MGBv neurons after cortical application of muscimol (protocol 3). Open circles and bars, BF changes of the matched neurons; filled circles and bars, BF changes of the unmatched neurons. The gray filled circles and triangles in Fig. 6A indicate the BF changes in MGBv (Fig. 3A) and AI (Fig. 4A) neurons without AI inactivation. A. The changes in MGBv BFs were not correlated to the BF differences between ICc and MGBv neurons. B. The response changes were different between the matched and unmatched MGBv neurons at control BF and 20 dB above the MT. The error bars represent the SD.

Fig. 5. Three examples (A, B, C) illustrating the effects of the ESICc on the frequency tunings of MGBv neurons after cortical application of muscimol (protocol 3). The ESICc did not change the BF but increased the auditory response of the matched MGBv neurons (A1, A2). Different from those without AI inactivation (Fig. 2), the BFs of the unmatched MGBv neurons did not shift towards the ICc BFs following ESICc (B1, B2 and C1, C2). The gray bars and the arrows represent the BFs of the stimulated ICc neurons. In the right column of panels, the dark gray areas represent the facilitation.

6

S.L. Nelson et al. / Hearing Research 328 (2015) 1e7

the unmatched neurons with BFs lower than ICc ones (n ¼ 8, p ¼ 0.685), for the unmatched neurons with BFs higher than ICc ones (n ¼ 6, p ¼ 1), and for the matched neurons (n ¼ 7, p ¼ 0.172). Decreased responses at the control BFs were also not observed for the unmatched neurons (Fig. 6B). The responses were not changed for the unmatched neurons with BFs lower than ICc ones (n ¼ 8, p ¼ 0.685 and for the unmatched neurons with BFs higher than ICc ones (n ¼ 6, p ¼ 338). A sharp contrast is that the increased responses were clear for the matched neurons (Fig. 6B). On average, the responses significantly increased by 40.67 ± 16.25% (n ¼ 7, p < 0.01). We then compared the ESICc-induced changes when the AI was treated by saline (protocol 1) with the ESICc-induced changes when the AI was inactivated by muscimol (protocol 3). The BF changes (2.30 ± 1.71 kHz) for the unmatched neurons (n ¼ 27) in protocol 1 was significantly larger than the BF changes (0.50 ± 0.52 kHz) for the unmatched neurons (n ¼ 14) in protocol 3 (p < 0.001). For the responses of the unmatched neurons at control BFs, the decrease of responses (27.97 ± 24.34%) in protocol 1 was also significantly larger than the decrease of responses (3.50 ± 9.36%) in protocol 3 (p < 0.01). What remained the same was the increase in the responses of the matched MGBv neurons. The increase of responses (31.07 ± 15.45%) for the matched neurons (n ¼ 11) in protocol 1 was similar to the increase of responses (40.67 ± 16.25%) for the matched neurons (n ¼ 7) in protocol 3 (p ¼ 0.742). 4. Discussion It is well established that auditory learning induces frequencyspecific plasticity, i.e., neuronal tuning shifts and/or map reorganization, in the cortex and thalamus (Weinberger, 2004; de VillersSidani and Merzenich, 2011; Pantev and Herholz, 2011; Suga, 2012). The thalamic is highlighted for its essential role in sending acquired auditory information to the amygdala and/or signal association during learning (Weinberger, 1998; Maren and Quirk, 2004; Sigurdsson et al., 2007). The auditory cortex receives attention as the highest auditory processing center and for its ability to direct subcortical functional remodeling (Suga et al., 1997; Perrot et al., 2006; Nunez and Malmierca, 2007; Yan and Ehret, 2002; Luo et al., 2008; Suga, 2012; Markovitz et al., 2013 Slater et al., 2013; Mellott et al., 2014). Our data show here that frequency-specific plasticity was induced in both the MGBv and AI following a bottom-up direction from the ICc (Figs. 3 and 4), a protocol simulating the information flow during auditory learning. Since the ICc projects directly to the MGBv but not to the AI (Kudo and Niimi, 1980; Redies and Brandner, 1991; Winer et al., 1999; Moore and Goldberg, 2004), a logical speculation is that the frequencyspecific plasticity, i.e., facilitation of the acquired frequency channel and frequency tuning shifts of neighboring frequency channels, takes place in the MGBv and then is further augmented or consolidated in the AI (Weinberger, 1998; Suga, 2008, 2012; Xiong et al., 2009). Our data suggest this is not entirely true, the MGBv and AI play distinctly different roles in the frequency-specific plasticity. As discussed above, auditory frequency-specific plasticity consists of two clear-cut components. One is the facilitation of the acquired frequency channel (matched neurons), which can simply result from the enhancement of synaptic transmission (Buonomano and Merzenich, 1998; Froemke et al., 2007; Caporale and Dan, 2008). The other is the suppression and centripetal tuning shifts of neighboring frequency channels (unmatched neurons), which is believed to rely on the cross-channel neural connections of both excitation and inhibition, i.e., through local neural network (Suga, 2012; Xiong et al., 2009; Liu et al., 2011, 2015). Neural mechanisms for the occurrences of these two components have often fallen into a simple logic: they are induced or created in the

same local neural circuitry or the same processing center. Our data in this study suggest that the roles of auditory cortex and thalamus are extremely different and support a novel view of the neural mechanisms underlying the highly specific auditory plasticity in the auditory thalamus and cortex. The MGBv appears to be responsible for the facilitation of the motivated frequency channel while the AI appears to be responsible for the suppression and tuning shifts of the neighboring frequency channels (Fig. 6). The suppression and tuning shifts observed in the MGBv likely originate from those created in the AI and forwarded down to MGBv (Fig. 6) through the corticothalamic feedback projections (Zhang and Yan, 2008). This novel view receives supports from a series of previous findings at the systems level. First, the AI has the capacity of enabling subcortical frequency-specific tuning shifts. Studies from different groups demonstrated that focal electrical stimulation of the AI induces centripetal frequency tuning shifts of cortical neurons in the same way as those induced by auditory learning or experience (Chowdhury and Suga, 2000; Talwar and Gerstein, 2001) Second, we have demonstrated that AI neurons show similar tuning shifts following focal electrical stimulation of the MGBv with the same ES protocol in the current study (Jafari et al., 2007), which can be explained by the long-term potentiation and suppression of thalamocortical synaptic transmission (Liu et al., 2015). Third, previous studies using different species of animals have documented the frequency-specific corticofugal modulation; focal stimulation of the AI induces centripetal frequency shifts of neurons in the MGBv, ICc and even subcollicular nucleus (Suga et al., 1997; Nunez and Malmierca, 2007; Yan and Ehret, 2002; Luo et al., 2008; Suga, 2012). Fourth, the frequency-specific plasticity in the auditory thalamus and midbrain can solely inherit the plastic changes in the AI. We have previously demonstrated that the focal electrical stimulation of the cholinergic basal forebrain paired with a tone induces frequency-specific frequency tuning shifts in the AI, MGBv and ICc. Such plasticity in the MGBv and ICc does not occur when the AI is inactivated by muscimol (Zhang et al., 2005; Zhang and Yan, 2008). Fifth, with the same ES protocol used in this study, we also demonstrated that ESMGBv induces the centripetal frequency tuning shifts in the ICc, which is eliminated when the AI is inactivated by muscimol (Wu and Yan, 2007). Last, the centripetal frequency tuning shift can also been induced in the ICc by focal ICc stimulation, which is absent when the AI is inactivated (Zhang and Suga, 2005). The last one is of particular interest because it suggests that the ICc local circuitry cannot create ICc centripetal frequency tuning shift. However, further investigation remains necessary to confirm whether or not the local neural circuitry in the MGBv and ICc bears intrinsic mechanisms for centripetal frequency-specific tuning shift. Note that lack of evidence demonstrating the role of MGBv in the frequency-tuning shift does not rule out the possibility that MGBv has other functions in inputspecific remodeling, i.e. the stimulus-specific adaptation. It has been shown that AI is not necessary for the generation of stimulusspecific adaption in MGBv, but works as a gain modulation system (Antunes and Malmierca, 2011). Back to the behavioral examples discussed in our introduction, the novel view presented here is able to provide a neural basis that underlies the interesting differences in learning-induced behavioral changes with and without cortical deprivation. If the AI is the only neural substrate responsible for sound-guided or frequencyspecific tuning shifts of auditory neurons, i.e., the reorganization of tonotopic maps in the auditory cortex and subcortical nucleus, auditory training or learning cannot improve the frequency discrimination in animals with cortical inactivation (Riquimaroux et al., 1991; Talwar et al., 2001). For the same reason, auditory fear conditioning induces only generalized conditioned response in

S.L. Nelson et al. / Hearing Research 328 (2015) 1e7

the animals with cortical inactivation (Armony et al., 1997). These animals are able to keep the conditioned response because conditioned and unconditioned information are associated in the thalamus/amygdala and the sound information or the associated information are directed from the thalamus to the amygdala for establishing the conditioning (Weinberger, 1998; Maren and Quirk, 2004; Sigurdsson et al., 2007). In sum, our data clearly demonstrated that the AI directs inputspecific remodeling in MGBv. Our data together with previous findings draw a clear outline of the distinct function of auditory cortex and thalamus for input-specific (i.e., learning-induced and experience-dependent) auditory plasticity at the systems level. The central position of the auditory cortex is further consolidated. It is worthy to note that the frequency-specific plasticity in the AI, MGBv and even ICc must continuously be coordinated through the tonotopic loops that consist of the colliculothalamic, thalamocortical, corticothalamic and corticocollicular projections (Xiong et al., 2009). Acknowledgment We are grateful to Dr. Jos Eggermont for his helpful comments. Conflict of Interest: None declared. This work was supported by grants from the Canadian Institutes of Health Research (grant numbers MOP164961, MOP274494); the Natural Sciences and Engineering Research Council of Canada (DG261338-2009); and the Alberta Innovates-Health Solutions and funds from the Campbell McLaurin Chair for Hearing Deficiencies. References Antunes, F.M., Malmierca, M.M., 2011. Effect of auditory cortex deactivation on stimulus-specific adaptation in the medial geniculate body. J. Neurosci. 31 (47), 17306e17316. Armony, J.L., Servan-Schreiber, D., Romanski, L.M., Cohen, J.D., LeDoux, J.E., 1997. Stimulus generalization of fear responses: effects of auditory cortex lesions in a computational model and in rats. Cereb. Cortex 7, 157e165. Buonomano, D.V., Merzenich, M.M., 1998. Cortical plasticity: from synapses to maps. Ann. Rev. Neurosci. 21, 149e186. Caporale, N., Dan, Y., 2008. Spike timing-dependent plasticity: a Hebbian learning rule. Annu. Rev. Neurosci. 31, 25e46. Chowdhury, S.A., Suga, N., 2000. Reorganization of the frequency map of the auditory cortex evoked by cortical electrical stimulation in the big brown bat. J. Neurophysiol. 83, 1856e1863. de Villers-Sidani, E., Merzenich, M.M., 2011. Lifelong plasticity in the rat auditory cortex: basic mechanisms and role of sensory experience. Prog. Brain Res. 191, 119e131. Franklin, K.B.J., Paxinos, G., 1996. Paxinos and Franklin's the Mouse Brain in Stereotaxic Coordinates. Academic, San Diego. Froemke, R.C., Merzenich, M.M., Schreiner, C.E., 2007. A synaptic memory trace for cortical receptive field plasticity. Nature 450, 425e429. Iwata, J., LeDoux, J.E., 1988. Dissociation of associative and nonassociative concomitants of classical fear conditioning in the freely behaving rat. Behav. Neurosci. 102, 66e76. Jafari, M.R., Zhang, Y., Yan, J., 2007. Multiparametric changes in the receptive field of cortical auditory neurons induced by thalamic activation in the mouse. Cereb. Cortex 17, 71e80. Kudo, M., Niimi, K., 1980. Ascending projections of the inferior colliculus in the cat: an autoradiographic study. J. Comp. Neurol. 191, 545e556. Liu, X., Basavaraj, S., Krishnan, R., Yan, J., 2011. Contribution of the thalamocortical system towards sound-specific auditory plasticity. Neurosci. Biobehav. Rev. 35, 2155e2161. Liu, X., Wang, C., Pan, C., Yan, J., 2015. Physiological correspondence dictates cortical long-term potentiation and depression by thalamic induction. Cereb. Cortex 25, 545e553.

7

Luo, F., Wang, Q., Kashani, A., Yan, J., 2008. Corticofugal modulation of initial sound processing in the brain. J. Neurosci. 28, 11615e11621. Maren, S., Quirk, G.J., 2004. Neuronal signalling of fear memory. Nat. Rev. Neurosci. 5, 844e852. Markovitz, C.D., Tang, T.T., Lim, H.H., 2013. Tonotopic and localized pathways from primary auditory cortex to the central nucleus of the inferior colliculus. Front. Neural. Circuits 7, 77. Mellott, J.G., Bickford, M.E., Schofield, B.R., 2014. Descending projections from auditory cortex to excitatory and inhibitory cells in the nucleus of the brachium of the inferior colliculus. Front. Syst. Neurosci. 8, 188. Moore, C., Goldberg, S., 2004. Successfully developing a cardiovascular research program in a community hospital. J. Cardiovasc. Manag. 15, 13e19. Nunez, A., Malmierca, E., 2007. Corticofugal modulation of sensory information. Adv. Anat. Embryol. Cell. Biol. 187, 1e74. Pantev, C., Herholz, S.C., 2011. Plasticity of the human auditory cortex related to musical training. Neurosci. Biobehav. Rev. 35, 2140e2154. not, M., Catenoix, H., Fischer, C., Mauguie re, F., Perrot, X., Ryvlin, P., Isnard, J., Gue Collet, L., 2006. Evidence for corticofugal modulation of peripheral auditory activity in humans. Cereb. Cortex 16, 941e948. Redies, H., Brandner, S., 1991. Functional-organization of the auditory thalamus in the Guinea-Pig. Exp. Brain Res. 86, 384e392. Riquimaroux, H., Gaioni, S.J., Suga, N., 1991. Cortical computational maps control auditory perception. Science 251, 565e568. Slater, B.J., Willis, A.M., Llano, D.A., 2013. Evidence for layer-specific differences in auditory corticocollicular neurons. Neurosci 229, 144e154. Sigurdsson, T., Doyere, V., Cain, C.K., LeDoux, J.E., 2007. Long-term potentiation in the amygdala: a cellular mechanism of fear learning and memory. Neuropharmacol 52, 215e227. Stiedl, O., Spiess, J., 1997. Effect of tone-dependent fear conditioning on heart rate and behavior of C57BL/6N mice. Behav. Neurosci. 111, 703e711. Suga, N., 2008. Role of corticofugal feedback in hearing. J. Comp. Physiol. A Neuroethol. Sens. Neural. Behav. Physiol. 194, 169e183. Suga, N., 2012. Tuning shifts of the auditory system by corticocortical and corticofugal projections and conditioning. Neurosci. Biobehav. Rev. 36, 969e988. Suga, N., Ma, X., 2003. Multiparametric corticofugal modulation and plasticity in the auditory system. Nat. Rev. Neurosci. 4, 783e794. Suga, N., Zhang, Y., Yan, J., 1997. Sharpening of frequency tuning by inhibition in the thalamic auditory nucleus of the mustached bat. J. Neurophysiol. 77, 2098e2114. Talwar, S.K., Gerstein, G.L., 2001. Reorganization in awake rat auditory cortex by local microstimulation and its effect on frequency-discrimination behavior. J. Neurophysiol. 86, 1555e1572. Talwar, S.K., Musial, P.G., Gerstein, G.L., 2001. Role of mammalian auditory cortex in the perception of elementary sound properties. J. Neurophysiol. 85, 2350e2358. Weinberger, N.M., 1998. Physiological memory in primary auditory cortex: characteristics and mechanisms. Neurobiol. Learn. Mem. 70, 226e251. Weinberger, N.M., 2004. Specific long-term memory traces in primary auditory cortex. Nat. Rev. Neurosci. 5, 279e290. Winer, J.A., Sally, S.L., Larue, D.T., Kelly, J.B., 1999. Origins of medial geniculate body projections to physiologically defined zones of rat primary auditory cortex. Hear. Res. 130, 42e61. Wu, Y., Yan, J., 2007. Modulation of the receptive fields of midbrain neurons elicited by thalamic electrical stimulation through corticofugal feedback. J. Neurosci. 27, 10651e10658. Xiong, Y., Zhang, Y.H., Yan, J., 2009. The neurobiology of sound-specific auditory plasticity: a core neural circuit. Neurosci. Biobehav. Rev. 33, 1178e1184. Yan, J., 2003. Development and plasticity of the auditory cortex. Can. J. Neurol. Sci. 30, 189e200. Yan, J., Ehret, G., 2002. Corticofugal modulation of midbrain sound processing in the house mouse. Eur. J. Neurosci. 16, 119e128. Yan, J., Zhang, Y., Ehret, G., 2005. Corticofugal shaping of frequency tuning curves in the central nucleus of the inferior colliculus of mice. J. Neurophysiol. 93, 71e83. Zhang, Y., Hakes, J.J., Bonfield, S.P., Yan, J., 2005. Corticofugal feedback for auditory midbrain plasticity elicited by tones and electrical stimulation of basal forebrain in mice. Eur. J. Neurosci. 22, 871e879. Zhang, Y., Suga, N., 2005. Corticofugal feedback for collicular plasticity evoked by electric stimulation of the inferior colliculus. J. Neurophysiol. 94, 2676e2682. Zhang, Y., Suga, N., Yan, J., 1997. Corticofugal modulation of frequency processing in bat auditory system. Nature 387, 900e903. Zhang, Y., Yan, J., 2008. Corticothalamic feedback for sound-specific plasticity of auditory thalamic neurons elicited by tones paired with basal forebrain stimulation. Cereb. Cortex 18, 1521e1528.