Essays

Available online at www.sciencedirect.com R Hearing Research 180 (2003) 39^50 www.elsevier.com/locate/heares A behavioral paradigm to judge acute sodium salicylate-induced sound experience in rats: a new approach for an animal model on tinnitus Lukas Ruttiger à , Jurgen Ciu¡ani, Hans-Peter Zenner, Marlies Knipper « « THRC Tubingen Hearing Research Center, Molecular Neurobiology, Department for Otorhinolaryngology, University of Tubingen, « « Elfriede-Aulhorn-Str. 5, D72076 Tubingen, Germany « Received 21 October 2002; accepted 10 February 2003 Abstract Behavioral conditioning studies on rats have been proven to be a valid animal model for the evaluation of acute and chronic phantom auditory experience (tinnitus). We developed an animal model for short-term, acute induced phantom auditory sensations in rats. Rats were trained in a conditioning chamber to actively access a liquid feeder whenever a constant white noise sound was present. During silence, no reward was given. Fulfilling the demands of animal protection laws for maximal avoidance of pain and fear, punitive paradigms were maximally reduced. After 15^17 learning sessions, all animals performed more accesses to the reward feeder during periods of sound than during periods of silence. Tinnitus was induced by the administration of sodium salicylate (350 mg/kg body weight) given 3 h before testing. The feeder access activity of a rat treated with salicylate was significantly increased during periods of silence, indicating a phantom auditory experience. The presumptive auditory experience was estimated to be comparable to a white noise sound of about 30 dB SPL rms. The activity increase was less pronounced for lower doses of sodium salicylate (150 mg/kg body weight) and was not found in animals trained on a dark^light paradigm, as expected. As the learning sessions of the operant conditioning were performed without pharmacological treatment, unintentional drug effects, for example, on learning and motivation of a rat were minimized in this behavioral paradigm. Furthermore, the behavioral changes reported here were shown to be a specific drug effect evoking a phantom auditory experience of a rat and cannot be explained by unspecific drug effects on motor activity, motivation, learning or hearing loss. The conditional paradigm is discussed in the context of its potential as a model for testing drugs that may have a therapeutic value in tinnitus research. ß 2003 Elsevier Science B.V. All rights reserved. Key words: Animal model; Tinnitus; Operant behavior; Sound experience; Rat; Salicylate 1. Introduction Tinnitus is a phantom auditory experience in the absence of an external or internal, physically quanti¢able, sound. Tinnitus is also known as ear sound or ear ringing (Wilson and Sutton, 1981) and is assumed to originate from damage or reversal suppression of auditory sensory neurons by ototoxic agents (McFadden et al., 1984; Kenmochi and Eggermont, 1997) or loud auditory events (Chermak and Dengerink, 1987; Szczepaniak and MÖller, 1996; Bauer and Brozoski, 2001). The * Corresponding author. need for an animal model in tinnitus research has been stressed and in the last decade two animal models have been introduced to induce a tinnitus-like phantom auditory experience in animals (Jastrebo¡ and Brennan, 1994; Jastrebo¡ and Sasaki, 1994; Penner and Jastrebo¡, 1996; Bauer et al., 1999). Both animal models are based on conditioning techniques and activity observations on the behavior of rats within standard conditioning chambers (Skinner box, Estes and Skinner, 1941). The animal model introduced by Jastrebo¡ et al. (1988a,b) is based on licking suppression of rats induced by electrical foot shock during silence. The authors show that inducing a sound perception in rats before suppression training prevents the animals from 0378-5955 / 03 / $ ^ see front matter ß 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0378-5955(03)00075-3 40 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « learning to connect silence and foot shock. In contrast, animals treated with salicylate after the suppression training obviously experience sound (tinnitus) and resume drinking activity earlier. As a consequence, the behavioral change induced by sodium salicylate (‘salicylate’) was calculated as a result of the integrative differences between the animals treated before and after suppression training. In Jastrebo¡’s animal model, the judgment of an accurate behavioral change depended on continuous daily measures over an extinction time of 5^10 days after the suppression training. This method was shown to be suitable to estimate the intensity (Jastrebo¡ and Brennan, 1994) and the pitch (Jastrebo¡ and Sasaki, 1994) of the assumed phantom auditory experience in the rat and was useful for initial pharmacological studies (Jastrebo¡ et al., 1997). The other model, established by Bauer et al. (1999), focused on the evaluation of chronically induced tinnitus in rats. Bauer et al. chose operant conditioning techniques to train the animals to press a lever bar continuously while sound is being played in the Skinner box. Animals were rewarded in a random interval schedule after 6^30 s. Silent periods were unrewarded and electrical foot shocks were applied only to reinforce learning. Both these inventories maintained a stable lever press response and a suppression of response during a 1 min silent period. To verify the presence of phantom auditory experience in chronic state, lever pressing was recorded over a long time for control animals drinking tap water versus animals drinking a solution containing salicylate. Unfortunately, both conditioning paradigms are based on the deprivation of either food (Bauer et al., 1999) or drinking water (Jastrebo¡ et al., 1988a,b) and unavoidable electrical foot shocks (Jastrebo¡ et al., 1988a,b). Therefore, they are not in line with the novel guidelines formulated by federal animal protection laws in Germany. The maximal avoidance of fear, pain and deprivation forced us to alter the current training procedures substantially towards a paradigm based predominantly on rewards. In addition, we focused on the development of an animal model meeting the requirements for a study on the acute e¡ects of salicylate rather than the long-term e¡ects studied by Bauer et al. (1999) since we were interested in the primary molecular events following a phantom auditory experience. For salicylate, the acute e¡ects on click-evoked action potentials and intensity-dependent summation potentials are maximal after 3 h (Stypulkowski, 1990). Preliminary experiments also indicated that the brainstem auditory evoked potentials of rats (response audiometry) were maximally a¡ected after this time (Ruttiger et al., « 2001). We therefore adopted a time lag of 3 h between salicylate administration and recording to verify the behavioral changes in our animal subjects. 2. Materials and methods 2.1. Animals Nine female albino rats (Wistar, aged 8^20 weeks) were trained and tested on ¢ve consecutive days per week. Animals were held in pairs in standard holding cages (0.28U0.44U0.16 m). Dietary intake was monitored and controlled to maintain a body weight di¡erence of within þ 10% between consecutive weeks and to maintain a balanced water uptake (Verplanck and Hayes, 1953). 2.2. Conditioning chamber Training and testing took place in a commercial conditioning chamber (rat shuttle box, TSE), adapted for the purposes of this study (Fig. 1). The layout was chosen to excite high left^right locomotion activity. The conditioning chamber was housed in a sound-attenuating surrounding box (0.6 mU0.6 mU1.2 m, minimal attenuation 40 dB, 1^40 kHz). All acoustic stimuli were calibrated using a high precision sound level meter (remote microphone BpK 1/2Q Falcon 4191, preampli¢er BpK 2669 C, ampli¢er BpK Nexus 2690). Environmental noise within the conditioning chamber was between 3.5 dB and 30.2 dB SPL rms with a peak at 1 kHz (3.5 dB) and a minimum at 40 kHz (30.2 dB SPL rms) measured with a resolution of 1024 points from 1 to 40 kHz. The conditioning chamber was internally illuminated by remotely switched house lights (Fig. 1A, house lights). Electrical stimuli (0.1^0.5 mA, 100 V, 0.5 s) could be supplied via a shockable £oor ground (Fig. 1, shockable £oor ground). A resting platform with a mechanical sensor was mounted on one side of the cage, covering the shockable £oor and serving as a resting location for the animal (Fig. 1A, resting platform, platform sensor). The cage was separated by a wall into two short hallways (Fig. 1, separating wall). At both ends of the hallways, within a recess (Fig. 1, liquid feeder), small amounts of £uid could be given to the animal (sucrose in water, 3%), gravity-advanced and controlled by £ow resistance- and vibration-muted magnetic shutter valves (Fig. 1A, magnetic shutters, neoLab 3-1167). A typical open time was 0.5 s, resulting in a reward drop of ca. 20 Wl, supplied to the animal via a curved metal drinking cannula (Fig. 1B, metal drinking cannula). Reward drops not taken up by the animal were drained o¡ into a reservoir unreachable by the rat. Photo sensors registered the visits of an animal at the feeder recesses (Fig. 1, photo sensor). All sensors were monitored on a computer screen and a top-mounted USB camera (Fig. 1A, camera) gave pictures of the whole £oor dimensions of the cage interior. L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « 41 Fig. 1. Schematic top (A) and side view (B) of the rat behavior box. Dimensions are 0.3U0.3U0.3 m. Magnetic shutters, photo sensors, house lights, platform sensor, and stimuli were computer-controlled within a temporal accuracy of 0.01 s. A rat could interrupt the photo sensor to trigger a drop of sugar water in rewarded training situations or evoke a punitive electrical impulse stimulus in unrewarded training situations (electrically induced impulse, 0.1^0.5 mA). 2.3. Stimuli Auditory stimuli were generated by means of a digital^analog card (PCI-MIO-16-E, National Instruments) and presented over three broadband speakers mounted vertically in the cage (Fig. 1, central speaker, Fig. 1B, left and right side speaker, Beyerdynamic DT911) separately ampli¢ed and attenuated by computer control. A continuous white noise sound (bandwidth 0.01^50 kHz, 70 dB SPL rms) could be played on the central loudspeaker (Fig. 1, central speaker) switched o¡ and on with a 1000 ms ramp. The Fourier spectrum of the noise sound was measured with a resolution of 1024 frequencies from 1 to 40 kHz. The spectrum had a peak at 6.4 kHz (49 dB SPL), a minimum at 40 kHz (20.2 dB SPL) and a roll-o¡ from peak frequency to 1 kHz (42.9 dB SPL), 20 kHz (40.6 dB SPL) and on to 40 kHz (20.2 dB SPL). In parallel to the white noise sound, a pure tone (cue tone, 8 kHz, 70 dB SPL, 200 ms length, 25 ms ramp, repeated ¢ve times with 300 ms pause) could be presented over loudspeakers mounted directly over the left and right feeder recesses (Fig. 1B, left and right side speaker, liquid feeder). Auditory stimuli were calibrated before testing with the sound level meter, and the remote microphone, placed at a position estimated to be the location of the animal’s head, if it had been resting on the platform. All acoustic stimuli could be monitored by the experimenter over loudspeakers (Yamaha MSP5) and could be visually controlled on an oscilloscope display (Tektronix TDS 210). 2.4. Training and experimental testing Animals were trained on auditory stimuli for 30^60 min/day for 5 days/week. Training session length was adapted to the animal’s activity. Always 15^18 h prior to behavioral testing (experimental session), the drinking water was withdrawn. The conditioned rats were divided into two groups (one animal per cage for either group). Animals from the ¢rst group received an intraperitoneal injection of sodium salicylate (350 mg/kg bw), while animals from the second group received an intraperitoneal injection of an equivalent volume of saline. Animals from either group were tested on the same day in a semi-random order exactly 3 h after injection. During the experimental session electrical stimuli were omitted. Four minutes after the start of the session the sugar water reward was stopped and the behavioral performances were recorded for 12^16 min and subsequently analyzed. Within the next 2^5 days rats received the same training as before the experiment. On the next experimental day animals from the group previously treated with salicylate were injected with saline and vice versa and tested again. 42 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « 2.5. Data analysis Frequencies of feeder access action of a rat were calculated for periods of sound and periods of silence separately (accesses/min) and normalized. Silence activity ratio (SA ratio): SA ratio ¼ nsilence =tsilence ; SA ratiov0 nsound =tsound ð1Þ stitute of Phonetic Sciences, 2001). The Wilcoxon t-statistic is a robust test for the comparison of paired values with no requirements for data distribution and is applicable for sample numbers of ¢ve and more. 2.7. Training procedure Animals were conditioned to discriminate between periods of sound and periods of silence. While sound was played, a rat could access one of the two feeders to draw a sugar water droplet from a metal drinking cannula (Fig. 1B, metal drinking cannula). Initially, rats were lured to the liquid feeders by cue tones given from the side speakers. Once the paradigm was learned, these cue tones could be omitted. Electrical punitive impulses (0.1^0.5 mA, duration 0.5 s) were applied incidentally during the non-rewarded silence periods by means of the shockable £oor ground. With training and experimental session start, the internal house light of ca. 50 lux was switched on remotely (Fig. 1A, house lights). For a control experiment, animals were trained on a light^dark paradigm by additional light sources of ca. 500 lux. For the conditional training we followed a protocol shown in Fig. 2. Training sessions were organized in a modular pattern with increasing level of di⁄culty. In all training blocks, a white noise sound was played (Fig. 2A, constant white noise sound). 2.7.1. First training block (Fig. 2A) Animals were conditioned to access one of the two liquid feeders to draw a single drop of sugar water from the metal drinking cannula. Rats unfamiliar with this task were attracted to the feeders by spontaneously being o¡ered a drop of sugar water (Fig. 2A, drop symbol) on one feeder side. A repetitive cue tone from where n is the number of accesses to a reward feeder during silence and sound (nsilence and nsound , respectively) and t is the time of silence and sound being played (tsilence and tsound , respectively). The di¡erence of silence activity ratios (vSA ratio) was calculated by: vSA ratio ¼ SA ratiosalicylate 3SA ratiosaline ð2Þ where SA ratiosalicylate is the silence activity ratio of an animal tested after salicylate injection and SA ratiosaline is the silence activity ratio of the same individual after saline injection, both calculated from Eq. 1. In an analogous manner as in Eqs. 1 and 2 the LA ratio and vLA ratio of animals trained on a light^dark paradigm was calculated. 2.6. Statistics For each individual rat, the di¡erence of activity ratios (vSA ratio, see Eq. 2) was calculated. Rats performing fewer than three accesses/min in their experimental session were excluded from the experiment data regardless of the treatment since their performance did not fall within the 92% con¢dence interval of the mean training value. The vSA ratios were tested for signi¢cance with the Wilcoxon signed rank t-statistics at a con¢dence level of P 6 0.05 (Table 1, Lowry, 2000; In- Table 1 SA ratios of individual rats after injection of 350 mg/kg bw sodium salicylate vs. saline injection, tested with the Wilcoxon t-statistic Rat SA ratio salicylate 1 2 3 4 5 6 7 8 9 Mean þ S.D. Median Rank sum Wilcoxon t-statistic 0.70 0.52 0.25 0.57 0.74 0.41 0.56 0.73 0.39 0.54 þ 0.17 0.56 saline 0.57 0.31 0.13 0.42 0.10 0.28 0.48 0.14 0.17 0.29 þ 0.17 0.28 0.12 0.21 0.12 0.15 0.64 0.12 0.09 0.59 0.22 0.25 þ 0.21 0.15 0 (3) 45 (+) +4 +6 +3 +5 +9 +2 +1 +8 +7 vSA ratio Rank (vSA ratio) K = 0.01, n = 9 Critical value = 3 t=0 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « 43 the speaker mounted above the feeder side where the sugar water drop was given was simultaneously played (Fig. 2A, speaker and wave front symbol). Voluntary access to a liquid feeder (Fig. 2A, arrow symbol) triggered the same cue tone and sugar water reward sequence. First, animals were rewarded every time they just accessed one of the two feeders (Fig. 1A, arrows) but during the following training sessions, rewards were only given four, three, two and one times consecutively at the same feeder side. The animal then had to switch to the opposite feeder to get further rewards. The signi¢cance of this training was to get a vital locomotion activity that is countable in the sense of characterizing the animal’s behavior. 2.7.2. Second training block (Fig. 2B) Several silent periods of 60 s duration were introduced, 120^180 s after white noise sound had been played (Fig. 2B, unrewarded silence periods). During silence, access to either feeder was non-rewarded (Fig. 2B, unrewarded access). In this training block the animals learned that accessing a feeder in non-reward situations (silence) was a redundant action that could be avoided by attentively listening to the white noise sound. 2.7.3. Third training block (Fig. 2C) The cue tone was played in advance over either the left or the right side speaker (Fig. 2C, speaker and wave front symbol) and a reward was only given at the cage side where the cue tone was playing. The cue tone was played continuously for 30 s and stopped with an animal’s access to the correct feeder side, initiating a reward. The cue tone was faded out later in training block D, when the rat performed su⁄cient voluntary accesses to the feeders (Fig. 2D, cue tone is omitted). As in the second training block (Fig. 2B), the constant white noise sound was alternated by randomly introduced periods of 60 s silence. 2.7.4. Last training block (Fig. 2D) A punitive impulse was introduced during silence (Fig. 2D, electrical impulse). First, an impulse of intensity 0.1 mA was applied every time a rat was misled to access a feeder in the non-rewarded period of silence. In subsequent sessions, the intensity of the impulse could automatically increase (up to 0.3 mA in steps of 0.1 6 Fig. 2. Illustration of the animal operant training behavior and schedule. Periods of constant white noise sound are represented as ripples within the horizontal bars. Periods of silence are indicated as un¢lled areas within the horizontal bars. (A) First training block: a constant white noise sound was presented (noise ripples of horizontal bar). Every access by the animal to a feeder (arrow symbol) was rewarded with a drop of sugar water (drop symbol) and paired with a cue tone (speaker and wave front symbol). Naive rats were attracted to, for example, the left side feeder by being spontaneously o¡ered a drop of sugar water at the left drinking cannula, while a cue tone from the left side speaker was playing (spontaneous reward). (B) Second training block (from session number 3 onwards): the constant white noise sound was arbitrarily interrupted by periods of silence of 60 s duration (blank sections of horizontal bar). During silence, accesses to a feeder were not rewarded (unrewarded access). (C) Third training block (from session number 6 onwards): a cue tone indicated the side to be rewarded (speaker and wave front symbol). Five seconds after a rewarded access (arrow and drop symbol) the cue tone was switched o¡. Accesses to the wrong side and accesses during silence were unrewarded. (D) Fourth training block (from session number 12 onwards): an electrical impulse (£ash symbol) was introduced as a punitive stimulus during the silent periods. The impulse was applied whenever the rat accessed a feeder during unrewarded silent periods. As the training progresses, the cue tone was omitted. 44 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « mA) when a rat performed more than 10 silence accesses within 10 min. Rarely, when the rats were insensitive to the impulse, the intensities were manually increased to 0.5 mA. Depending on the learning capability of a rat, an impulse was given on average every 2^3 min, and occasionally every 0.8 min. When the learning criterion was reached (SA ratio 0.2), an impulse was evoked once every 7^29 min within the conditioning sessions. To achieve a stable access behavior also in the experimental situation, rewards were restricted randomly to every second or third time of access. This could maintain the activity of a rat for 15^20 min after the reward was omitted in experimental recordings. Fig. 4. Time course of access activity (activity, accesses/min). Mean (solid line) and standard deviation (vertical bars) of accesses to the feeders during periods of sound for nine rats over 17 sessions within 23 days of behavioral training on white noise sound. Symbols and letters (A^D) mark a change to the more demanding training level (see Figs. 2 and 3). From day 9 onwards, the activity ranged from four to 20 accesses/min (average 14.6, dotted line) whereas learning progressed continuously (decreasing SA ratio, see Fig. 3). 3. Results 3.1. Training Frequencies of feeder access action of a rat (accesses/ min) were calculated separately for periods of sound and periods of silence. To judge for individual activity di¡erences the silence activity was normalized to the mean activity during sound. This normalization resulted in a calculated SA ratio being robust against daily activity variations. A SA ratio of 1 would indicate no Fig. 3. Time course of the conditioning level expressed as SA ratio. Mean (solid line) and standard deviation (vertical bars) for nine rats over 17 sessions within 23 days of behavioral training on white noise sound. Symbols and letters (A^D) mark a change to the more demanding training level. A: Each access is rewarded; B: no reward during silence; C: a cue tone indicates the rewarded side; D: access during silence evokes a punitive impulse (see also Fig. 2). The dotted lines represent the SA ratio for untrained rats (SA ratio = 1) and the criterion for well-trained rats (SA ratio = 0.2). Once the conditional task was learned (SA ratio below 0.2) the SA ratio remained stable during training sessions. di¡erence between periods of silence and periods of sound, while a SA ratio of 0 would mean complete activity suppression during silence. The SA ratio changed with training. Before learning the access activity was not suppressed during silence, resulting in a similar SA ratio for periods of sound and periods of silence. During the learning phase, rats were trained in the way of becoming active during periods of auditory sensation while resting still during periods of silence (suppression learning). With suppression learning the SA ratio decreased from 1 to 6 0.2. The SA ratios were followed over the time course of learning. Fig. 3 summarizes the SA ratio change (mean of nine rats with standard deviation). Animals were trained in 17 sessions over a period of 23 days. Once the task was learned, the SA ratio was a stable measure of the conditioning state of an animal. Naive rats reached the criterion level of SA ratio 0.2 after 15 conditioning sessions (Fig. 3). Even after 6^8 months without training, four of ¢ve rats retested still performed excellently and easily reached the criterion level. Fig. 4 shows the time course of the reward feeder accesses (activity) during periods of sound for nine rats. The mean activity ranged from four to 20 accesses/min and showed large variations with the switch to the next conditioning paradigm. After seven learning sessions within 9 days, the mean activity stabilized at 14.6 accesses/min. At about the same time the SA ratio decreased constantly, indicating that learning progressed beyond day 10 (Fig. 3). L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « 45 Fig. 5. SA ratio (A) and overall access frequency to the liquid feeders (Activity, B) after injection of saline (open bars) and 350 mg/kg bw of sodium salicylate (closed bars), mean of nine rats. Rats treated with salicylate were more active during periods of silence (A, closed bar, mean 0.54, þ S.D. = 0.17, n = 9) compared to animals treated with saline (A, open bar, mean = 0.29, þ S.D. = 0.17, n = 9), resulting in a higher SA ratio (A, positive vSA ratio, arrow, P 6 0.01). The overall activity was similar for animals treated with salicylate and animals treated with saline (B, Activity, P s 0.05). Error bars (B) indicate the standard deviation. 3.2. Salicylate treatment Experiments were scheduled when a stable access activity and the criterion level (SA ratio 0.2) was reached. The animals were injected with salicylate (350 mg/kg bw) or an equivalent volume of saline and tested 3 h later. The behavioral e¡ect induced by salicylate is encoded in the di¡erence values of the SA ratio after saline and salicylate treatment (Fig. 5A, vSA ratio). Table 1 gives the SA ratio and vSA ratio for individual rats and the test statistics applied. The SA ratio in an experimental session (saline: mean 0.29) could be higher than in a training session (normally 6 0.2). This increase was due to the di¡erent procedure in training and experimental session. In experimental sessions, no rewards and no electrical impulses were given. In any case, this applied equally to both saline- and salicylatetreated animals. The SA ratio of animals treated with salicylate was signi¢cantly higher than the SA ratio for animals treated with saline (Fig. 5A, P 6 0.01). The mean activity was similar for rats treated with saline and salicylate (Fig. 5B, 5^22 accesses/min) and was in the range of the activity found in previous training sessions (Fig. 4, 4^20 accesses/min). Data of rats performing fewer than three accesses/min in a single experimental session were omitted from the analysis data regardless of treatment. The exclusion was necessary because the SA ratio cannot be calculated reliably for access numbers smaller than three accesses/min. In one single instance data had to be excluded from the analysis. To get further insight into the correlation of salicylate treatment with an observed behavioral change, dose dependence was analyzed. Animals were injected with either saline or a dose of 150 mg/kg bw of salicylate. Fig. 6 shows the vSA ratios for single rats after salicylate administration of 150 and 350 mg/kg bw. The median values are shown as horizontal bars for either experimental group. With the higher dose of salicylate the vSA ratios increased, re£ecting the higher silence access activity of the rats. Fig. 6. Dose dependence of the vSA ratio for salicylate dosages of 150 mg/kg bw (circles) and 350 mg/kg bw (triangles). Horizontal bars show the median values for either group. Activity ratios found after salicylate injection were signi¢cantly higher than those found with saline injection (350 mg: **P 6 0.01; 150 mg: *P 6 0.05, n = 9). 46 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « Fig. 7. Estimation of the loudness of the presumptive phantom auditory experience in rats after injection of 150 mg/kg bw and 350 mg/kg bw salicylate. Dots show the average access activity behavior of 11 rats during white noise sound of 10^70 dB SPL and silence. Vertical bars represent one standard deviation of the mean. Rats behave invariantly at sound pressure levels above 50 dB. With sound pressures below 50 dB, the SA ratio decreases gradually. The access activity at sound pressures of 10 dB SPL and more was always higher than the activity during silence. The dashed line shows the regression of the SA ratios below 50 dB SPL. 3.3. Intensity of the phantom auditory experience The intensity of the sound experienced by a rat after salicylate injection was tested with 11 untreated welltrained rats (seven of the nine rats used in the previous experiments and four new rats). The sound protocol involved periodical changes of the sound pressure starting with 70 dB (SPL rms), then changing to 10, 20, 30, 40, 50 or 60 dB and back to 70 dB before a period of silence started. With the lower sound pressure, the animals became gradually less active until, during silence, a minimal access activity was reached (Fig. 7). The psychometric function can also serve as an estimate of the individual behavioral hearing threshold of an animal though this is not a subject of the present study. Our aim was to estimate the average perceived sound pressure of rats treated with salicylate from the psychometric function. This was done by comparing the mean SA ratio of animals treated with salicylate to the corresponding access activity of the mean psychometric function (Fig. 7). The sound pressure levels that evoked a behavioral e¡ect comparable to the salicylate e¡ect were found to be 28 dB and 3 dB for salicylate doses of 350 mg/kg bw and 150 mg/kg bw, respectively. This means that the perceived phantom auditory experience after 350 mg/kg bw salicylate was equivalent to a broadband sound of 28 dB SPL. It should be noted that the dose dependence might be di¡erent when using pigmented rats. Pigmented rats have a higher sensitivity to lower doses of salicylate injections (Jastrebo¡ et al., 1988a,b). Granted that animals did actually experience a phantom sound of about constant intensity after injection of 350 mg/kg bw salicylate, the calculated SA ratio should be larger when the animals were tested on a sound less intense than 70 dB SPL. However, the sound pressure equivalent to the phantom auditory percept after 350 mg/kg bw salicylate should be unchanged. To test this we reduced the sound pressure level of the white band noise that was played during the sound periods. Animal behavior was tested in a sound^silence interval schedule with a reduced sound pressure level of 45 dB SPL. The resulting SA ratios are shown in Fig. 8 for nine rats (circles) and as the mean SA ratio (black horizontal bar). From the mean SA ratio found (0.749) the presumptive auditory experience was calculated to be equivalent to 32 dB SPL. This sound pressure equivalent is close to the calculated intensity percept of 28 dB SPL found with the 70 dB SPL training (Fig. 7). L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « 47 3.4. Dark^light control Rats trained on a light^dark paradigm learned the task in a similar way and as reliably as rats trained on sound and silence: the criterion level (LA ratio 0.2) was reached after 12^15 sessions of about 30 min within 17^20 days. The access activity of animals trained on a dark^light paradigm after saline and salicylate treatment is shown in Fig. 9, expressed as vLA ratio. Salicylate treatment of 350 mg/kg bw did not systematically change the mean activity behavior of rats trained on dark^light (median 0.09). 4. Discussion Using operant conditioning, we were able to demonstrate a signi¢cant behavioral change in animals treated with a dose of 350 mg/kg bw of sodium salicylate and a less pronounced but still signi¢cant behavioral change with a lower salicylate concentration (150 mg/kg bw). It is well known from previous studies that high doses of salicylate or Aspirin0 induce ear ringing (tin- Fig. 9. vLA ratio for four rats (one or two measurements on each rat) conditioned on a light^dark paradigm with constant white noise sound of 60 dB SPL. The mean vLA ratio of approximately zero indicates that these control animals conditioned on a light^dark paradigm did not show behavioral access activity changes after salicylate injection, in contrast to animals trained on a sound^silence paradigm (see Fig. 7, vSA ratio). Fig. 8. Estimation of the loudness of the presumptive phantom auditory experience from rats trained on a 45 dB SPL broadband white noise sound after injection of 350 mg/kg bw salicylate. The dashed line illustrates the regression of the access behavior onto 45 dB SPL (SA ratio 1) and silence (SA ratio 0.095, taken from the regression line in Fig. 7). Open circles show the access activity behavior of individual rats during white noise sound. To illustrate the sound pressure equivalent for the perceived sound intensity, data points were plotted twice: one upon the other (open circles) and horizontally shifted to meet the assumed psychometric function (dots and horizontal thin lines). The horizontal bar gives the mean of the SA ratios (mean 0.749, S.D. 0.192, n = 9). Arrows indicate the calculated mean equivalent sound pressure for the phantom auditory experience (mean 32.5 dB SPL, S.D. 9.76 dB SPL, n = 9). nitus) in humans (Myers and Bernstein, 1965; McFadden and Plattsmier, 1983) and in animal subjects (Penner and Jastrebo¡, 1996; Bauer et al., 1999). The mechanisms of phantom auditory sound perception and the ways of its generation by means of salicylate are still obscure. While some studies reported an upregulation of the neuronal spontaneous activity within the inferior colliculus (IC) with salicylate (Jastrebo¡ and Sasaki, 1986) and a partial up-regulation of the metabolic activity marker [14 C]2-deoxyglucose (2DG) (Sasaki et al., 1980), others reported a down-regulation of 2DG within the IC and other auditory structures (Wallhausser-Franke et al., 1996). Also a loss of the « sharp frequency tuning of single auditory nerve ¢bers has been described after salicylate (Evans and Borerwe, 1982; Stypulkowski, 1990). In in vitro preparations of the cochlea, salicylate is reported to a¡ect the turgor and ionic conductance of outer hair cells (Brownell and Winston, 1989; Lue and Brownell, 1999) and it has been shown that the voltage-driven contractibility as well as the membrane capacitance of isolated outer hair cells was reduced after salicylate (Tunstall et al., 1995). Only recently, salicylate has come into consideration for altering the non-linear membrane capacitance and motor protein-mediated fast motility of mammalian outer hair cells (Oliver et al., 2001). Most likely, 48 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « sensorineural hearing loss in peripheral auditory structures (Boettcher et al., 1989; Chen and Jastrebo¡, 1995; Manabe et al., 1997; Bauer et al., 2000) contributes to the tinnitus-related neuronal activity in higher auditory centers (Kenmochi and Eggermont, 1997; Eggermont and Kenmochi, 1998). To get further insight into the mechanisms and the molecular basis of soundless hearing perception on the one hand and to verify the therapeutic potential of existing tinnitus treatment on the other hand, pharmacological studies are required. An animal model o¡ers the possibility of screening drugs that may have a therapeutic potential in alleviating tinnitus symptoms ¢rst on animals, before extending them to clinical trials. Behavioral models were used to verify the salicylate-induced endogenous sound experience in animals (Jastrebo¡ et al., 1988a,b; Jastrebo¡ and Sasaki, 1994). We nevertheless developed a new strategy by adopting a paradigm that avoids fear induction in the animals. Positive reinforcement of the relevant behavior parameters, locomotion and access activity, was achieved by closely pairing the action (activity) and the reward (sugar water). Introducing mild electrical impulse stimuli as a punitive measure led to fast learning of the task. Though the electrical impulses were meant to be unpleasant, they were far from noxious (0.1^0.5 mA) and since they were easily avoidable by the animal, they were not expected to induce stress or fear. Rats provoked with even many impulses in a short period of time ( s 1/min) were still motivated by the reward to perform the behavioral paradigm. Therefore, rats learned easily to avoid the punishment and to restrict their feeder accesses to the rewarded time periods. 4.1. A new concept for a behavioral model on phantom sound perception The behavioral model we present here allows the strict separation of times of learning and times of experimental treatments. We could achieve this by some novelties within our conditioning paradigm. (1) Animals passed a behavioral training free of drug administration until a de¢ned level of learning was reached. The suppression training to stabilize the behavior took place within 12^15 sessions prior to an experiment (Fig. 3). Drug application was restricted to one experimental session and behavioral testing could take place on a single day. Long-term drug application by injection or oral supply was unnecessary. This schedule enabled us to control the level of learning for each single rat and to avoid side e¡ects of drugs that can a¡ect learning and motivation during the learning phase. (2) Each individual animal contributed twice to the experimental data, on one day as a control animal and on another day as a treated animal. Thus, we were able to calculate individual di¡erences in activity ratios (vSA ratio, Figs. 5 and 6 or vLA ratio, Fig. 9) from the data of only two measurements of about 20 min. The vSA ratio proved to be suitable to detect behavioral changes between control and salicylate treatment with a small number of animals. We were able to reproduce the results in a subgroup of these nine animals and a group of six other animals. (3) By using the vSA ratio to judge whether a rat perceived a phantom sound, the precise level of learning is not critical. Though learning gradually changed the animal’s performance over time (SA ratio, Fig. 3), the di¡erence of SA ratios after saline and salicylate injections (vSA ratio) was a robust estimate for the subjective sound experience in experimental sessions. 4.2. The speci¢city of salicylate in inducing a phantom auditory experience After the application of salicylate at dosages as high as 350 mg/kg bw, non-auditory behavioral e¡ects such as variations in locomotion activity must be considered. For humans it is described that high salicylate doses may induce a non-auditory vertigo sensation (Germann et al., 1996; Jarboe and Hallworth, 1999). In humans (Janssen et al., 2000), as in rats (Kay and Davies, 1993; Ruttiger et al., 2001), hearing thresholds were reported « to be elevated. So how can we be sure that the behavioral changes reported in this study were not due to an unspeci¢c e¡ect on locomotion and learning, or a masking e¡ect, or even a hearing loss' (1) Any unspeci¢c e¡ects of salicylate on locomotion, motivation and learning were ruled out as the overall activity of rats remained una¡ected after salicylate injection (measured as the access frequency). Additionally, animals trained on a dark^light paradigm for reward avoidance did not show a signi¢cantly di¡erent behavior after salicylate or saline treatment. Also the overall activity of rats (measured as the feeder access frequency) remains una¡ected after salicylate injection. Therefore, drug e¡ects on locomotion, motivation and learning cannot explain the behavioral changes. (2) Masking of the external, physical sound by tinnitus may account for the behavioral change we report here. We think this is unlikely since, from our calculations, salicylate produces an auditory phantom experience equivalent to a broadband noise of 28^32 dB SPL (Figs. 7 and 8) regardless of the sound pressure level of the trained and tested sound, be it 70 or 45 dB. However, after training with the 45 dB SPL broadband sound, rats treated with salicylate showed larger behavioral changes than rats trained with the 70 dB SPL broadband sound (SA ratio 0.75 and 0.54, respectively). This may be due to the similarity of the external sound (45 dB SPL) and the phantom sound experience of the L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « 49 rat (equivalent to 32 dB SPL). However, the access activity of a rat after salicylate treatment is still suppressed in periods of silence after the 45 dB SPL training and was suppressed to a larger extent after the 70 dB SPL training. Therefore, auditory masking cannot account for the behavioral changes after salicylate treatment. (3) Salicylate in a dose of 350 mg/kg bw induced a hearing loss of 15^20 dB on click or pure tone stimuli (Ruttiger et al., 2001). After salicylate treatment, rats « should therefore experience less intense sound than after saline. As Fig. 7 shows, the activity was reduced when the sound was muted below 60 dB SPL even in untreated rats. However, if a hearing loss were responsible for the behavioral changes after salicylate treatment, we would have expected a di¡erent result: the animals should become less active during both periods of silence and sound, since the perceived sound intensity would be lower. Fig. 5B shows this is not the case: the mean activity is similar for saline and salicylate treatments. During periods of silence, animals treated with salicylate were more active (Fig. 5A). This indicates that a salicylate-induced hearing loss cannot account for the behavioral changes during silence after salicylate. In conclusion, the SA ratio, the main parameter we used to estimate the salicylate impact on the phantom auditory experience, turned out to be independent of unspeci¢c motor or learning impairment, auditory masking or hearing loss. The new concept presented here is reasonable for the examination of acute induced phantom auditory sensations in rats and therefore o¡ers a new method for experiments towards the understanding of the molecular and physiological genesis of tinnitus in man. To re¢ne this animal model, it will be necessary to conduct further experiments to establish the frequency, bandwidth and intensity properties of the ringing in the ears that an animal is experiencing in our conditioning paradigm. Acknowledgements The authors wish to thank Susanti Hidayat, Mirko Jaumann and Johannes Wendeberg for assistance during the experiments. Joachim Ostwald is acknowledged for supplying information and advice for animal training and conditioning. Special thanks are due to Marcus Muller and Justin Tan for improving the quality of the « manuscript. References Bauer, C.A., Brozoski, T.J., 2001. Assessing tinnitus and prospective tinnitus therapeutics using a psychophysical animal model. J. Assoc. Res. Otolaryngol. 2, 54^64. Bauer, C.A., Brozoski, T.J., Rojas, P., Boley, J., Wyder, M., 1999. Behavioral model of chronic tinnitus in rats. Otolaryngol. Head Neck Surg. 121, 457^462. Bauer, C.A., Brozoski, T.J., Rojas, P., Boley, J., Wyder, M., 2000. E¡ects of chronic salicylate on GABAergic activity in rat inferior colliculus. Hear. Res. 147, 175^182. Boettcher, F.A., Bancroft, B.R., Salvi, R.J., Henderson, D., 1989. E¡ects of sodium salicylate on evoked-response measures of hearing. Hear. Res. 42, 129^141. Brownell, W.E., Winston, J.B., 1989. Slow electrically evoked volume changes in guinea pig outer hair cells. Abstr. Meet. Assoc. Res. Otolaryngol. 138. Chen, G.D., Jastrebo¡, P.J., 1995. Salicylate-induced abnormal activity in the inferior colliculus of rats. Hear. Res. 82, 158^178. Chermak, G.D., Dengerink, J.E., 1987. Characteristics of temporary noise-induced tinnitus in male and female subjects. Scand. Audiol. 16, 67^73. Eggermont, J.J., Kenmochi, M., 1998. Salicylate and quinine selectively increase spontaneous ¢ring rates in secondary auditory cortex. Hear. Res. 117, 149^160. Estes, W.K., Skinner, B.F., 1941. Some quantitative properties of anxiety. J. Exp. Psychol. 29, 390^400. Evans, E.F., Borerwe, T.A., 1982. Ototoxic e¡ects of salicylates on the responses of single cochlear nerve ¢bres and on cochlear potentials. Br. J. Audiol. 16, 101^108. Germann, R., Schindera, I., Kuch, M., Seitz, U., Altmeyer, S., Schindera, F., 1996. Life threatening salicylate poisoning caused by percutaneous absorption in severe ichthyosis vulgaris. Hautarzt 47, 624^627. Institute of Phonetic Sciences, 2001. IFA Services: Statistics, Overview of tests. IFA, http://fonsg3.let.uva.nl/Service/Statistics.html. Janssen, T., Boege, P., Oestreicher, E., Arnold, W., 2000. Tinnitus and 2f1-f2 distortion product otoacoustic emissions following salicylate overdose. J. Acoust. Soc. Am. 107, 1790^1792. Jarboe, J.K., Hallworth, R., 1999. The e¡ect of quinine on outer hair cell shape, compliance and force. Hear. Res. 132, 43^50. Jastrebo¡, P.J., Brennan, J.F., 1994. Evaluating the loudness of phantom auditory perception (tinnitus) in rats. Audiology 33, 202^217. Jastrebo¡, P.J., Sasaki, C.T., 1986. Salicylate-induced changes in spontaneous activity of single units in the inferior colliculus in the guinea pig. J. Acoust. Soc. Am. 80, 1384^1391. Jastrebo¡, P.J., Sasaki, C.T., 1994. An animal model of tinnitus: a decade of development. Am. J. Otol. 15, 19^27. Jastrebo¡, P.J., Brennan, J.F., Sasaki, C.T., 1988a. An animal model for tinnitus. Laryngoscope 98, 280^286. Jastrebo¡, P.J., Issing, W., Brennan, J.F., Sasaki, C.T., 1988b. Pigmentation, anesthesia, behavioral factors, and salicylate uptake. Arch. Otolaryngol. Head Neck Surg. 114, 186^191. Jastrebo¡, P.J., Zhou, S.T., Jastrebo¡, M.M., Kwapisz, U., Gryczynska, U., 1997. Attenuation of salicylate-induced tinnitus by Ginkgo biloba extract in rats. Audiol. Neuro-Otol. 2, 197^212. Kay, I.S., Davies, W.E., 1993. The e¡ect of nimodipine on salicylate ototoxicity in the rat as revealed by the auditory evoked brainstem response. Eur. Arch. Otorhinolaryngol. 250, 51^54. Kenmochi, M., Eggermont, J.J., 1997. Salicylate and quinine a¡ect the central nervous system. Hear. Res. 113, 110^116. Lowry, R., 2000. VassarStats: Statistical Computation Web Site. Lowry@vassar.edu. Lue, A.J.-C., Brownell, W.E., 1999. Salicylate induced changes in outer hair cell lateral wall sti¡ness. Hear. Res. 135, 163^168. Manabe, Y., Yoshida, S., Saito, H., Oka, H., 1997. E¡ects of lidocaine on salicylate-induced discharge of neurons in the inferior colliculus of the guinea pig. Hear. Res. 103, 192^198. McFadden, D., Plattsmier, H., 1983. Aspirin can potentiate the temporary hearing loss induced by intense sounds. Hear. Res. 9, 295^ 316. 50 L. Ruttiger et al. / Hearing Research 180 (2003) 39^50 « pathway ^ a model for examining tinnitus. Brain Res. 194, 511^ 516. Stypulkowski, P.H., 1990. Mechanisms of salicylate ototoxicity. Hear. Res. 46, 113^146. Szczepaniak, W.S., MÖller, A.R., 1996. E¡ects of (3)-baclofen, clonazepam, and diazepam on tone exposure-induced hyperexcitability of the inferior colliculus in the rat: possible therapeutic implications for pharmacological management of tinnitus and hyperacusis. Hear. Res. 97, 46^53. Tunstall, M., Gale, J., Ashmore, J., 1995. Action of salicylate on membrane capacitance of outer hair cells from the guinea-pig cochlea. J. Physiol. 485, 739^752. Verplanck, W.S., Hayes, J.R., 1953. Eating and drinking as a function of maintenance schedule. J. Comp. Physiol. Psychol. 46, 327^ 333. Wallhausser-Franke, E., Braun, S., Langner, G., 1996. Salicylate al« ters 2-DG uptake in the auditory system: A model for tinnitus' NeuroReport 7, 1585^1588. Wilson, J.P., Sutton, G.J., 1981. Acoustical correlates of tonal tinnitus. In: CIBA Foundation Symposium 85, Tinnitus. Pitman, London, pp. 82^100. McFadden, D., Plattsmier, H.S., Pasanen, E.G., 1984. Aspirin-induced hearing loss as a model of sensorineural hearing loss. Hear. Res. 16, 251^260. Myers, E.N., Bernstein, J.M., 1965. Salicylate ototoxicity. Arch. Otolaryngol. 82, 483^493. Oliver, D., He, D., Klocker, N., Ludwig, J., Schulte, U., Waldegger, S., Ruppersberg, J., Dallos, P., Fakler, B., 2001. Intracellular anions as the voltage sensor of prestin, the outer hair cell motor protein. Science 292, 2340^2343. Penner, M.J., Jastrebo¡, P.J., 1996. Tinnitus: psychophysical observations in humans and an animal model. In: van de Water, T.R., Popper, A.N., Fax, R.R. (Eds.), Clinical Aspects of Hearing. Springer, New York, pp. 258^304. Ruttiger, L., Limberger, A., Muller, M., Zenner, H.-P., Knipper, M., « « 2001. Sodium salicylate induced changes of auditory brainstem responses in rat: search for a peripheral or central tinnitus correlate. In: Elsner, E., Kreutzberg, G.W. (Eds.), Gottingen Neuro« biology Report, 4th Meeting of the German Neuroscience Society, Gottingen, Germany. Thieme, Stuttgart, p. 438. « Sasaki, C.T., Kauer, J.S., Babitz, L., 1980. Di¡erential [14 C]2-deoxyglucose uptake after dea¡erentiation of the mammalian auditory