Otoferlin knock-out animals were generated as described in [29
]. We analyzed the vocal behavior from 16 Otof
, deaf) and 15 heterozygous (Otof+/-
, hearing) pups at postnatal age of four or five days (P4-5), eight or nine days (P8-9) and 15 or 16 days (P15-16) while isolated from their mothers. Pups were bred using Otof-/-
females with Otof+/-
males and thus raised by deaf females. All mice were of mixed background (129 ola and C57N). Litters and mothers were kept in standard cages, one litter per cage. For identification, pups were marked with tattoos on the paws one day after birth.
In addition, we analyzed the vocal behavior from 12 knock-out and 16 control males. For the adults, deaf mice (Otof-/-) and heterozygous Otof+/- mice of 129ola/C57N mixed background were raised in groups of 2 to 5 males per cage until they were 8-9 weeks old. Each male was put to a separate cage in which the courtship songs were recorded one day after isolation. For the recordings, their cages were placed in a sound-attenuated Styrofoam box. After three minutes, a female was introduced for three minutes.
All experiments complied with national animal care guidelines and were approved by the University of Göttingen Board for animal welfare and the animal welfareoffice of the state of Lower Saxony (AZ 33.11.42502-04-044/08).
Mouse pups were recorded three times at postnatal age of four or five (P4-5), eight or nine (P8-9) and 15 or 16 days (P15-16). For each recording the cage with the litter to be measured was taken to a bench (room temperature: 21-22 C) and pups were selected randomly, weighed and placed in a soundproofed custom made plastic box (diameter 13.5 cm). An ultrasound microphone (UltraSoundGate CM16) was placed in the lid of the box 12 cm above the bottom and connected to a preamplifier (UltraSoundGate 116) which was connected to a notebook computer. We recorded the pups for 150 seconds using the recording software Avisoft Recorder 3.4 with a sampling frequency of 300 kHz (hardware and software from Avisoft Bioacoustics, Berlin, Germany). In older pups, short-term isolation from mother did not evoke isolation calls in a predictable manner. Therefore we put for the P15-16 pups with the plastic box in ice water (bottom temperature = 5-6 C), and recorded for 180 seconds. We chose such a relatively old age to record the pups because they do not start to hear before day 10-12 [12
To elicit courtship songs from male mice, we isolated males at the age of two months one day before the recording in a macrolon 2 cage (36.5 × 21 × 14 cm). On the recording day we placed the males in their own cage in a sound-attenuated styrofoam box (30 × 43 × 24 cm). After three minutes we introduced a female for three minutes and recorded the vocalizations with the same recording equipment we used for the recording of the pup isolation calls. The ultrasound microphone was placed in the lid of the box, 24 cm above the floor.
We counted the number of calls per recording session with AVISOFT Recorder 3.4. To separate isolation calls from the rest of the recording we used the whistles detection algorithm with following selection criteria: possible changes per step = 4 (4687 Hz), minimal continuity = 8 ms, possible frequency range = 40 to 150 kHz. These criteria were compared with former analysis of pup vocalizations [4
]. In addition, we visually controlled the procedure to ensure that the automated sampling routine selected only calls of mouse pups and no other sounds such as toe clicking. The AVISOFT recorder software stores the selected sounds in separate wave files, and, in addition logs the time of call onset.
From the stored calls, we calculated spectrograms (frequency range: 150 kHz, frequency resolution: 293 Hz, time resolution: 0.21 ms). We submitted the resulting spectrograms to the custom software program LMA 2011 to extract a set of characteristic acoustic parameters. As mice typically concentrate the energy of their USV in a single-frequency band, so-called 'pure tone-like sounds' or 'whistles', we focused on peak frequency of USV, i.e. the loudest frequency of a respective time frame. Just small head movements can lead to strong amplitude fluctuations in USVs. In addition, mice produce often soft sounds in the ultrasonic range. To ensure correct parameter estimations, we visually controlled the estimation and excluded incorrect estimated calls from the analysis. For each call we determined the duration of a call and the duration of amplitude gaps within a call (sound parts whose intensity is below 10% (at the start) and 15% (at the end of call) of the mean maximum amplitude of a call). Furthermore, we determined start, maximum peak frequency, as well as the greatest difference in peak frequency between two consecutive 0.21 ms bins. Typical whistles concentrate their energy to one amplitude peak. Therefore, the peak frequency corresponds to the fundamental frequency, although it is difficult to prove as long as no harmonics can be detected. In addition, we calculated the location of the maximum frequency and the location of peak frequency jump within the call. To describe the call modulation we calculated the slope of a linear trend through the peak frequencies of consecutive 0.21 ms bins. We did the same calculation for the male vocalizations (Table ). In addition to estimating the number of given calls, we estimated the latency to call.
Recording of auditory brainstem response (ABR)
To confirm correct genotyping and exclude hearing impairment in control animals, auditory brainstem responses (ABR) to click stimuli were recorded from 57 out of 59 animals in the analysis. Measurements in heterozygous controls yielded typical ABR waveforms with 5 waves representing synchronous postsynaptic potential generation in the auditory nerve (wave I) and brainstem with a mean threshold of 32 ± 2 dB peak equivalent (Figure ). Otoferlin knockout animals only showed a small early wave component at high sound intensities (mean threshold 100 ± 2 dB), which most likely represents the summating potential, reflecting normal hair cell transduction currents upstream of the synaptic deficit. In summary, ABR recordings confirmed normally hearing in all control animals and were consistent with profound deafness due to abolished hair cell exocytosis in otoferlin knockout animals.
Figure 4 ABR waveforms. Grand averages ± s.e.m. of ABR waveforms in response to 100 dB click stimuli presented at 20 Hz in 15 Otoferlin knockout animals (red) and 14 heterozygous littermates (black) used for pup vocalization studies. Roman numbers denote (more ...)
Animals were anaesthetized intraperitoneally with a combination of ketamine (125 mg kg-1) and xylazine (2.5 mg kg-1) and the heart rate was constantly monitored to control the depth of anaesthesia. The core temperature was maintained constant at 37°C using a rectal temperature-controlled heat blanket (Hugo Sachs Elektronik - Harvard Apparatus GmbH, March-Hugstetten, Germany). For stimulus generation, presentation and data acquisition we used the TDT III Systems (Tucker-Davis-Technologies, Ft Lauderdale, FL) run by custom-written Matlab software (The Mathworks). Clicks of 0.03 ms duration were calibrated using a ¼″ Brüel and Kjaer microphone (D 4039, Brüel & Kjaer GmbH, Bremen, Germany) and were presented at 20 Hz in the free field ipsilaterally using a JBL 2402 speaker (JBL GmbH & Co., Neuhofen, Germany). The difference potential between vertex and mastoid subdermal needles was amplified 20 times and sampled at a rate of 50 kHz for 20 ms, 2 × 2000 times to obtain two mean ABRs for each sound intensity. Hearing threshold was determined with 10 dB precision as the lowest stimulus intensity that evoked a reproducible response waveform in both traces by visual inspection.
We used a two-step cluster analysis (CA, SPSS 19) to establish vocal categories. We used the log-likelihood distance measure to establish different vocal cluster (up to 15 clusters) and the Schwarz-Bayesian information criterion (BIC) to decide which cluster solution showed the best fit. We used the eight acoustic parameters described above to calculate the CA. A higher number of parameters would have provided no advantage, because highly correlating acoustic parameters render it difficult to find appropriate cluster centers. In addition, they will shift the result in the direction of the most highly correlating parameters. The acoustic analysis program provided a set of further parameters. However, these parameters obtained a high correlation with the already chosen parameters (correlation coefficient above 0.7). Therefore, there was no advantage to include theses parameters in the analysis. They only would lead to an increase to correct for multiple testing. To confirm the cluster solution and to estimate the contribution of different acoustic parameters to distinguish between the established call categories we conducted a discriminant function analysis (DFA, SPSS 19) with the same eight acoustic parameters. We used a stepwise DFA. The selection criterion for an acoustic parameter to be entered was p = 0.05 and p = 0.1 to be removed from the analysis. The assignment of the calls was cross-validated by the leaving-one-out method of SPSS 19.
To ensure that the statistic results are not simply an artifact by the chosen number of call categories, we calculated for the male mouse courtship analysis call type solution of higher order and tested them for differences in relation to hearing ability (Additional file 1
To test for differences in structure and number of isolation calls between deaf and normally hearing pups we used a mixed linear model (SPSS 19) with hearing ability, recording age (P5-6, P8-9 and P15-16) as fixed factors, and subject, weight and litter as random factor. To test the courtship vocalization for structural differences regarding hearing ability we used a mixed linear model (SPSS 19), with hearing ability as fixed factors and subject as random factor. We conducted separate tests for all vocal types. To test the courtship vocalization for differences regarding call number, latency to call, rhythm of calling (start/start intervals) and call type usage we used Mann-Whitney-U test (SPSS 19). Where it was necessary we applied a Simes correction to correct for multiple testing. We chose Simes correction because it belongs to the correction methods which minimize the β error.