Animals and Song Recordings
Adult males and females of the B6D2F1/J (BxD), C57BL/6J (B6), B6.129S1-Casp3tm1Flv/J (CASP3), and BALB/c strains were purchased from the Jackson Laboratory (Bar Harbor, Maine). All mice were group housed and kept on a 12 hr light/dark cycle. At 5 weeks old, males were socialized by spending at least one night with an adult female. We selected males for behavioral experiments that readily sang in response to female urine. Therefore, the probability that the males in each of the groups in our experiments were from the same litter is very low. For recordings, males were placed into a recording box with fresh bedding, allowed to acclimate overnight for IEG experiments or 15 min for behavior only experiments, then stimulated to sing by presenting 200 µL of female urine directly into the bedding. Sounds were recorded with UltraSoundGate CM16/CMPA ultrasound microphones that feature a flat frequency response from 30–130 kHz and an UltraSoundGate 416–200 recording interface (Avisoft Bioacoustics). The microphones were suspended over the center of the recording box to minimize the differences in sound pressure level reaching the microphone due to varying horizontal orientation. Sounds were digitized at 250 kHz, 8 bits and captured to disk as .WAV files using Avisoft Recorder USG (Avisoft BioAcoustics). Example songs were pitched and slowed down to the human hearing range using Raven 4.0 (Cornell Laboratory of Bioacoustics). For playback studies, a set of recorded songs from the singing males of the IEG study were played at normal pitch, through an Avisoft ultrasound amplifier Model #70101 and speaker Model #60401 (Avisoft Bioacoustics). The same songs were used for all playback sessions, and sessions were recorded to ensure that no USVs were produced by the listening mouse. Mice that responded with USVs were not included in the analysis. All recordings and playbacks were conducted in dark, enclosed isolation chambers. Movement in the dark was recorded with a Speco Technologies VL-62 color infrared camera, the video was saved to tape, then digitized on a Canopus ADVC110 analog-to-digital converter at full resolution, and stored to disk as MPEG video files.
Behavioral Molecular Mapping
Adult male BxD mice were acclimated by placing them in a dark 15″×24″×12″ sound-attenuating recording chamber overnight. The following day, after a period of 3 hrs with little movement and no ultrasonic songs, males were presented with an olfactory or auditory stimulus. Five normal males (Singing & Hearing group) and 5 deafened males (Deaf-Singing group) were stimulated to sing by presentation of 0.1 cc of fresh urine from BALB/c females, pippetted on the bedding through a small covered opening on the top of the sound chamber. Five males were stimulated to explore the home cage without singing by presentation of 0.1 cc of 10% EtOH (Non-Singing group) to control for possible olfaction and movement-induced IEG expression in the singing groups. Olfactory stimuli were presented at 5 min intervals throughout the 30 min recording session to maintain exploratory and vocal behavior. We had one additional male that spontaneously sang without urine stimulation and showed the same IEG pattern as the urine stimulated singing animals (not shown). Five males were stimulated with 30 min of continuous presentation of identical USVs (Hearing Only group) recorded from a normal adult male played through an Avisoft ultrasound amplifier and speaker as described above. All other procedures were as described in the main text.
In Situ Hybridization
Immediately after the 30 min behavioral sessions, animals were sacrificed by decapitation without anesthesia, as the IEG changes can be sensitive to manipulation within 5–10 minutes of handling. Brains were removed, embedded in Tissue-Tek (Sakura Finetek), frozen on dry ice and then stored at −80°C. Coronal 12 µm sections were cut through the entire brain on a cryostat and every other section was mounted on silanated slides in series of 10. Frozen sections were processed for in situ
hybridization with a 35
S radioactively labeled riboprobe made from cDNAs for mouse egr-1 and rat arc, and processed for emulsion autoradiography following a previously described protocol 
. The egr-1 probe was generated from PCR-amplified sequences of the pCMV-Sport6-egr-1 plasmid containing the full-length mouse egr-1 cDNA (3.1 kb) insert from our own library (Pioneer Clone F6). The arc probe was generated from PCR-amplified samples of a 1.5 kb sequence of the rat arc cDNA, prepared according to a previously described protocol 
Adult male mice 77 to 87 days old were deafened by bilateral cochlear removal. Anesthesia was induced with 5% isofluorane in oxygen and maintained by intramuscular injection of ketamine-xylazine (75 mg/kg ketamine; 5 mg/kg xylazine). A retro-aural incision was made and the skin and muscle were retracted to reveal the tympanic bulla. The lateral wall of the bulla was punctured to reveal the cochlea, and a pair of fine forceps was used to remove the tympanic membrane, stapes and parts of the cochlear walls until no cochlear structure was visible. Sham surgery treated animals received all of the same treatments, except the bulla was not punctured and the tympanic membrane was not accessed.
Acoustic waveforms were processed using custom MATLAB programs that we modified from code written by Timothy E. Holy (Washington University) 
and that we called Syllable Identifier, made available upon request. A sonogram was computed from each waveform (256 samples/block, half-overlap), thresholded to eliminate the white noise component of the signal, and frequencies outside 35–125 kHz were truncated. Syllables with duration longer than 10 ms were identified and classified by presence or absence of instantaneous ‘pitch jumps’ separating notes within a syllable. The morphologically simplest note type doesn't contain any pitch jumps (Type A). The next most complex were those containing two notes separated by a single upward or downward pitch jump (Types B & C, respectively). More complex syllables were identified by the series of upward and downward pitch jumps occurring as the fundamental frequency varies between notes of higher and lower pitch (Types D–K). Much rarer syllable types (<1%) were grouped as other.
The following spectral features were calculated from the sonograms of each of the classified syllables types: Standard deviation of pitch distribution, mean frequency, frequency modulation, and spectral purity. Frequency modulation was measured as the frequency variance, or the squared deviation of peak frequencies from the mean peak frequency, averaged over the length of the syllable. Spectral purity was calculated as the instantaneous maximum power at the peak frequency normalized by the instantaneous total power in the spectrum, averaged across the entire syllable; a pure tone would have a spectral purity of 1, and white noise would approach 0. We also calculated starting frequency, ¼ frequency, ¾ frequency, final frequency, minimum frequency. From these measures, the mean value for each spectral feature (FV) was calculated for each recording epoch (single session, weeks or months depending on the experimental design) for all syllable types. For longitudinal data, we took the logarithm of the normalized mean value for each epoch (n) as the spectral feature score (SFS) such that: SFS(n)
log10[FV(n)/FV(1)]. This log ratio allowed us to obtain a relative difference to the pre-treatment conditions across animals, which could differ in their absolute values. Essentially, the SFS gave us a measure of the change in each measured acoustic feature for each animal normalized to their own pre-treatment baseline. The log ratio made it symmetrical around zero. This approach allowed us to easily visualize and compare both decreases and increases in individual acoustic features across scales and across animals.
Digitized videos were coded for periods during which the animal was sitting still or moving throughout the home cage using the behavioral coding software Annotation by SaySoSoft (v1.0, http://www.saysosoft.com/
). The durations of all locomotor and rotational movement recorded were summed to determine the total amount of movement produced by each animal.
Gene Expression Analysis
Photomicrographs were taken from autoradiographs of hybridized sections. Regions of interest (ROIs) were defined in three serial sections for each brain area on inverted images in ImageJ (NIH, Bethesda). The mean pixel value was recorded for each ROI (Mroi), three regions of each glass slide with no brain tissue (Mbkgnd), and control areas with no difference in the background-adjusted mean pixel values across groups (Mctrl). The control areas with no difference were: 1) the ventral striatum for cingulate cortex, motor cortex, somatosensory cortex, and anterodorsal striatum (); 2) the midbrain reticular nucleus for the auditory cortex (). These values were used to calculate the expression score (ES) for each ROI as follows: ESroi
log10[(Mroi−Mbkgnd)/(Mctrl−Mbkgnd)]. The values were log transformed to visualize comparable magnitudes for expression differences above and below silent control levels in the experimental animals.
For retrograde tracing from laryngeal muscles, we used a recombinant strain of Psuedorabies Bartha (PRV-Bartha) expressing enhanced Green Fluorescent Protein (eGFP) under the control of the histomegalovirus immediate early gene promoter 
. Live virus was received from the laboratory of Dr. Lynn Enquist at Princeton University at a titer of 1×109
pfu/mL, aliquoted at 4 µL per tube, then stored at −80°C, and thawed immediately before injection. General anesthesia was induced with 1% isofluorane and maintained by intramuscular injection of ketamine-xylazine (75 mg/kg ketamine; 5 mg/kg xylazine). A midline incision was made from the sternum to the hyoid bone. The portion of the sternohyoid muscle covering the larynx was removed. Five 200 nL injections were made 1 min apart into the cricothyroid laryngeal muscle using a Nanofil microsyringe system with a 34 gauge stainless steel needle (World Precision Instruments, Sarasota, FL). After 5 min, the microinjection pipette was retracted, and the injection was repeated for the cricoarytenoid lateralis muscle. A single break in the fascia was made for each muscle and sealed with TissueMend adhesive (Veterinary Products Laboratories) to prevent spread of virus to other tissues.
For anterograde tracing of cortico-bulbar projections to nucleus ambiguus we injected 7.5% BDA (Biotinylated Dextran Amine, 10000 MW; Sigma) in sterile water into the motor cortex of 6 adult male mice (5 bilateral and 1 unilateral). Following induction of anesthesia, as above, the scalp was retracted and a small craniotomy made over the injection site. Injections of BDA were made through a glass micropipette using the Nanoject II microinjector (Drummond Scientific) at 4 sites 0.550 mm from the brain surface (50–90 nL per site) along a track 1.2 mm lateral and −0.2 to 0.4 mm anterior to Bregma. Then, 6–11 days later, three of these mice were injected with 0.5 µL of 1% CTb (Cholera Toxin Subunit b) in sterile water in the two laryngeal muscles, as described above for PRV-Bartha. Two days after CTb injections, the mice were sacrificed and transcardially perfused as described below.
For chemical lesions of cortex we injected 7.5% ibotenic acid in sterile water bilaterally (220 nL per injection site) into the motor cortex of 6 adult mice, and into the visual cortex bilaterally (220 nL per injection site) of 4 adult mice. The coordinates for the motor cortex were as above for the tracer experiments, and for visual cortex were 3 mm caudal and 1.5 mm lateral to Bregma. 5 adult mice received sham surgeries in which the scalp and skull were opened, as described above, but no injection was made. After recording USVs three weeks after surgery, all mice were injected with PRV-Bartha in the laryngeal muscles, as described above. To quantify the lesions we counted the number of surviving PRV-Bartha-labeled layer V pyramidal cells in M1 in 7 serial sections per hemisphere. Lesion sizes were expressed as a percentage of cells eliminated from a baseline of 102±16 cells (s.e.m.) counted from 7 similarly quantified unlesioned hemispheres. When calculating the spectral feature scores for post-operative songs one lesion case was confirmed to be an outlier on three univariate feature scores using the Dixon Q Test (Q90%
0.625; Bandwidth: Q
0.5852; Range: Q
0.5727; Frequency Variance: Q
0.7244) and the multivariate Mahalanobis Jackknifed Distance using 11 spectral feature scores (Chi Square (97.5%) JMD
4.6819; Outlier: MJD
7.9143). This case was excluded from further analysis.
Immunohistochemistry for tracers
Unless otherwise noted washes of brain sections were 3 times for 5 min in 0.1 M PBS. Animals were given an overdose of pentobarbital sodium and perfused transcardially with 0.9% saline followed by 4% paraformaldehyde in 0.1 M PBS. Brains were removed, post-fixed in 4% paraformaldehyde overnight, and cryoprotected in 30% sucrose in PB until sectioned. 40 µm coronal sections were cut on a cryostat into 0.1 M PBS.
For visualizing BDA, free-floating sections were quenched 30 min in 0.3% H2O2, then reacted 1 hr in ABC solution (VECTASTAIN Elite Kit, Vector Labs). Sections were then washed 3 times for 10 min in PB, and developed for 15 min in 0.05% 3,3′-diaminobenzidine (DAB) solution (Sigma, #D5905) with nickel (DAB Substrate Kit, Vector Labs) to give a black reaction product.
For double labeling with CTb, following BDA detection, free-floating sections were blocked 1 hr in 0.3% PBST with NRS (VECTASTAIN Elite Kit, Vector Labs). Blocked sections were incubated 2 hrs at RT in goat anti-CTb (1
10000 dilution, List Biological Laboratories) followed by 1 hr in rabbit anti-goat biotinylated secondary antibody (VECTASTAIN Elite Kit) and then reacted 30 min in ABC solution. DAB staining was as above, but for 3 min without nickel to give a brown reaction product.
For labeling of eGFP expressed from the PRV-Bartha recombinant vector, free-floating sections were quenched as above, then blocked 30 min in 0.3% PBST with NGS (VECTASTAIN Elite Kit). Blocked sections were reacted for 3.5 hrs at RT in rabbit anti-eGFP (1
1000 dilution, Open Biosystems), followed by 1 hr incubation in goat anti-rabbit biotinylated secondary antibody (VECTASTAIN Elite Kit) or 2 hrs at RT in donkey anti-rabbit secondary antibody conjugated to Alexa-fluor 488 (1
500 dilution, Invitrogen). ABC reaction and DAB staining were the same as for CTb detection, but for 8 min.