Low-side and multitone suppression in the base of the gerbil cochlea

Abstract

The cochlea’s mechanical response to sound stimulation is nonlinear, likely due to saturation of the mechanoelectric transduction current that is part of an electromechanical feedback loop. The ability of a second tone or tones to reduce the response to a probe tone is one manifestation of nonlinearity, termed suppression. Using optical coherence tomography to measure motion within the organ of Corti, regional motion variations have been observed. Here, we report on the suppression that occurs within the organ of Corti when a high-sound-level, low-frequency suppressor tone was delivered along with a sweep of discreet single tones. Responses were measured in the base of the gerbil cochlea at two best frequency locations, with two different directions of observation relative to the sensory tissue’s anatomical axes. Suppression extended over a wide frequency range in the outer hair cell region, whereas it was typically limited to the best frequency peak in the reticular lamina region and at the basilar membrane. Aspects of the observed suppression were consistent with the effect of a saturating nonlinearity. Recent measurements have noted the three-dimensional nature of organ of Corti motion. The effects of suppression observed here could be due to a combination of reduced motion amplitude and altered vibration axis.

Significance

The mammalian auditory organ, the cochlea, relies on a nonlinear active process to achieve sensitivity to low-level sounds and sharp frequency selectivity. Recent work using novel interferometric techniques to explore cochlear activity has revealed complex and nonlinear vibration patterns within the cochlea’s sensory tissue. In this study, the motion response to a pure tone was reduced by additional “suppressor” tones. The observed motion reduction was consistent with the effect of a saturating nonlinearity, possibly compounded by alterations in the axis of cellular vibration, and thus underscoring the three-dimensional character of cell-based cochlear mechanical activity.

Introduction

An acoustic stimulus induces a traveling wave in the mammalian auditory organ, the cochlea, which propagates along the cochlear spiral from the base to the apex. Vibration of the sensory tissue, the organ of Corti (OoC) leads to shearing between the apical surface of the OoC, the reticular lamina (RL), and the tectorial membrane, an acellular structure lying atop the OoC. This shearing motion displaces the hair cell stereocilia that protrude from the apical surface of the inner and outer hair cells (OHCs), activating mechanically gated transduction (MET) channels located close to the stereocilias’ tips, converting the displacements into receptor current and potential. Each longitudinal location along the cochlea responds preferentially to a different best frequency (BF) with high frequencies encoded at the base and low frequencies encoded at the apex, a phenomenon known as tonotopic tuning. (BF is the peak frequency at low sound level in the healthy cochlea.) The cochlea relies on an active process, termed the cochlear amplifier, to boost its mechanical response to low-level sounds. Activity extends the cochlea’s dynamic range over nearly six orders of magnitude in pressure and enhances its frequency selectivity at each location along its tonotopic axis, particularly the high-frequency base.

The active process is compressively nonlinear, with that compression likely due to the saturation of the OHC MET current that drives the cochlear amplifier (see, e.g., (1)). Due to this compressive nonlinearity, when another tone, tones, or wideband noise is delivered along with a probe tone, the response to the probe tone is reduced—this is termed suppression. Intracochlear suppression is a component of auditory masking (see, e.g., (2))—a phenomenon familiar to anyone who found they can no longer hear the radio if they turn on the coffee grinder. The sound pressure levels (SPLs) corresponding to speech are in highly nonlinear regions of cochlear mechanics, and cochlear nonlinearity leads to excitation patterns in the cochlea that are key to auditory discrimination (see, e.g., (3)). Suppression and other aspects of masking have been used to great advantage in the development of auditory compression algorithms, such as MP3, where bit rates of the original digitized music (on a compact disc) can be reduced by a factor of 15 in the compressed version of streamed music (4). This works because, due to auditory masking, there are components of the sound signal that are inaudible; armed with this knowledge, the compression algorithms eliminate unnecessary bits (4,5). Cochlear nonlinearity also produces intermodulation distortion products and otoacoustic emissions (DPOAEs) when two or more frequencies are simultaneously presented, and these are used for monitoring cochlear condition, for example, in newborn hearing screening. Understanding the physiology of cochlear nonlinearity is part of the basic science of how the ear processes sound signals. It is also needed for the clinical and commercial development of hearing prosthetics, and auditory devices and algorithms.

Auditory suppression has been well documented in the auditory periphery in basilar membrane (BM) vibration and the adjacent pressure, extracellular and intracellular voltage, and neural responses (see, e.g., (1,6,7,8,9,10)). This paper is concerned with vibration responses in the OoC and builds on previous measurements of BM motion in the cochlear base, the high-frequency region of the cochlea. In the basal turn of the gerbil cochlea, at locations visible through the cochlea’s round window, BFs range from $\sim$20 to 50 kHz. At any longitudinal location the BM motion is tuned and the degree of tuning is higher at lower SPL (11). BM motion scales nonlinearly in the response peak, at frequencies close to the location’s BF, while sub-BF BM motions scale linearly. Two-tone and multitone experiments at the BM found that the addition of a second tone or set of tones to a probe tone suppressed the response at the probe tone only when it was within the BF peak, the nonlinear part of the frequency response curve (see, e.g., (1,6,8)). These measurements used laser interferometry, which measures the motion of the first surface encountered by the interferometer’s laser beam. Optical coherence tomography (OCT), introduced to cochlear mechanics in 2006 (12,13), has advanced the field by allowing simultaneous measurement of vibration at multiple depths within the sensory tissue of the OoC. In particular, OCT-based measurements have revealed that the OoC region containing the OHCs exhibits sub-BF nonlinearity that is not observed on the BM (14,15,16,17). These findings were not surprising in the sense that OHC electromotility, which is thought to be the source of cochlear amplification, is broadband (18). However, sub-BF activity increases sub-BF vibration at low SPL, diminishing the tuning in the OHC region, which seems counterproductive for frequency resolution. The basis for cochlear tuning, and for the separation into sub-BF and BF zones of BM nonlinearity had been explored and explained in cochlear models (several noted in (19)) and these new intra-OoC findings are leading to new and revised models of cochlear mechanics (20,21,22).

Recently, OCT has been used to study suppression of vibration responses within the OoC. Two-tone suppression was studied in the mouse cochlea; based on the suppression patterns found, activity close to the BF peak propagated along the cochlea, but sub-BF activity in the OHC region did not propagate (23). Multitone and single-tone OoC responses were compared in (24), where multitone stimuli were found to suppress sub-BF responses in the OHC region. The sub-BF activity was termed a “nonamplifying nonlinearity,” because the motion of the BM, the substrate of the cochlear traveling wave, was linear sub-BF and thus the sub-BF activity apparently did not affect the traveling wave. In a later paper, the sub-BF activity was renamed as a “nonpropagating amplification,” (20) which is a more descriptive name, especially in light of the findings in (23) noted just above.

The impact and properties of the sub-BF activity in the cochlea are unsettled, and we continue to explore it in the experiments described here. Responses to pure tones were measured in the presence of an intense low-frequency tone, termed a low-side suppressor. The suppressor was 3 kHz and typically 100 dB SPL; an example using a 90 dB SPL suppressor is also shown. The intense low-frequency suppressor tone is expected to cause significant saturation of the MET current, which is expected to suppress active responses both sub-BF and BF. Responses to multitone stimuli were recorded at the same locations just after. With the multitone stimulation, the tone levels were equal in SPL across the frequency. The overall SPL of the multitone stimuli will be $10\log N$ higher than the pressure of the individual tones. N in this study was 35, leading to a $\sim$15 dB increase—thus multitone stimuli of 85 dB have an overall level of approximately 100 dB SPL. (While this provides a rule of thumb, cochlear tuning increases the response to BF peak tones, which then have a relatively large influence on the effective level (24).) Objectives of this study were to observe low-side suppression within the OoC, and to compare the character of low-side suppression with the suppression arising from multitone stimuli.

Measurements were made near the 25 and 45 kHz BF locations in the base of the gerbil cochlea, at multiple quasiradial positions across the OoC, with a Thorlabs Telesto OCT. Fig. 1 A shows a schematic image of the OoC in the gerbil base. Fig. 1, B and C shows B-scans from the two experimental locations, 45 and 25 kHz BF, respectively. The optical (z axis) is vertical and the x axis was chosen to span the BM approximately radially (left and middle panels of Fig. 1, B and C—the far left panels are labeled and color-coded versions of the middle panel). The Thorlabs OCT is capable of volumetric imaging, which enhanced the ability to identify structures and regions of interest (25,26,27). A volumetric image was rotated to view the $yz$ B-scan (right panels of Fig. 1, B and C—the BM is color coded and labeled). This view allows an evaluation of the degree to which the vertical axis is anatomically transverse (perpendicular to the BM) versus anatomically longitudinal (parallel to a local segment of the BM spiral). To access the 25 kHz BF location through the round window, the OCT beam was directed deep into the cochlea. The $yz$ plane view of Fig. 1 C shows that the optical axis for the 25 kHz BF location was predominantly anatomically longitudinal, whereas from the $yz$ plane view in Fig. 1 B, at the 45 kHz BF location, the optical axis was predominantly anatomically transverse. The measured motion of a structure is the component of the structure’s 3D motion that is projected along the optical axis; components of the motion that are perpendicular to the optical axis will not be detected. Our two measurement locations, at 25 and 45 kHz, differ in both BF and in the optical axis direction in the cochlea, and knowing the direction is especially important intra-OoC, where motion has significant longitudinal and transverse components (15,26). The program described in (25) can be used with a volume scan to quantify the longitudinal, radial, and transverse components of the optical axis, and these numerical $l,r,t$ values are noted in the captions.

Figure 1.

(A) Cartoon cross section of the organ of Corti with anatomical axes and selected structures labeled: AZ and PZ, arcuate and pectinate zones of the basilar membrane (BM); TM, tectorial membrane; TC, tunnel of Corti; OT, outer tunnel; OHCs, outer hair cells; HCs, Hensen’s cells; DCs, Deiters cells; RL, reticular lamina. The schematic was created with BioRender. In the cochlear base, the fluid-filled spaces within the OoC including the TC and outer tunnel are especially prominent. (B) Left and middle panels: 2D B-scan taken at the 45 kHz BF location of the gerbil cochlea. The left panel is a labeled version of the middle panel. At this location it is possible to align the OCT’s beam giving a nearly transverse view of the OoC, resulting in a fairly “classic” cross-sectional image of the OoC. Scale bars, 50 μm. The instrument axes are z, along the optical axis, x and y as indicated in the middle panel. The right panel shows the view found when the volume scan was rotated 90°. The dashed vertical line indicates the y position of the $xz$ cross section. $\left(l,r,t\right)$ values were: (−0.38, −0.27, 0.89). (C) Similar to (B) for the 25 kHz BF location, with left and middle panels the standard $xz$ orientation, and the right panel the $yz$ orientation. The optical axis is substantially longitudinal, and the resulting B-scan is not a classic cross section, so labeling is relatively coarse. $\left(l,r,t\right)$ values are: (0.87, –0.064, 0.49).

Results are presented as frequency response tuning curves. DPOAEs were recorded in response to two-tone stimuli and used as a real-time measure of cochlear condition.

Materials and methods

The experimental protocol was approved by the Columbia University IACUC. Eight adult gerbils of either sex weighing $\sim$60–80 g were used in this study, and results from seven are included in the presented data, three from the 45 kHz BF location, four from the 25 kHz BF location. Six of these experiments used a 100 dB SPL suppressor, which provided a robust suppressive effect through a wide frequency range. Two early experiments were done with a 90 dB SPL suppressor, which produced relatively weak suppression, especially at sub-BF frequencies; results from one of these experiments is shown. The gerbils were anesthetized with 40 mg/kg ketamine, 40 mg/kg sodium pentobarbitol, and 0.2 mg/kg buprenorphine. Supplemental doses of pentobarbitol were given as needed to maintain areflexia in response to a light hind toe pinch and a second dose of buprenorphine was administered 6–8 h after the start of the surgery. Animals were also given 0.01 mL of 2% lidocaine at each incision site. The scalp was removed and the skull fixed to a two-axis goniometer with dental cement. Animals were tracheotomized to facilitate normal breathing and most were given supplemental oxygen by placing a tube flowing oxygen a few centimeters from the tracheotomy. The left pinna, most of the cartilaginous ear canal, and tissue covering the auditory bulla was resected. The bulla was gently opened after softening the bone with a phosphoric acid gel (Etch-Royale, Pulpdent, Watertown, MA.) for 5–10 min. Core body temperature was maintained at 38°C with a servo-controlled heating blanket and a rectal thermometer and additional heating was applied to the head with heat lamps and a disposable hand warmer (Hot Hands, Kobayashi Consumer Products, Dalton, GA) placed on the goniometer. Paper wicks were gently placed within the round window niche to maintain a constant fluid level. Imaging and vibrometry were performed with a ThorLabs (Newton, NJ) Telesto III OCT (Lübeck, Germany) system equipped with an LSM03 5$\times$ objective lens. The ThorImage program was used to position the head and for volumetric (3D) imaging. The axial resolution of an OCT depends on the central wavelength and bandwidth of the light source and for the Telesto III was reported by the manufacturer to be $\sim$4 μm. The lateral resolution is determined by the objective lens and was $\sim$10 μm. For vibrometry, the noise floor is determined by reflectance of the surface of interest and the recording duration, and varied from $\sim$0.05 to 1 nm. The initial conversion of the raw line-camera data to time-locked A-scans (M-scans) was performed with custom software written in C++ using the ThorLabs software development kit; subsequent analyses were performed in custom scripts written in MATLAB. Time waveforms were extracted from local maxima in the A-scans (28) and the amplitudes and phases at each frequency were determined by Fourier analysis. For each frequency, the response was deemed significant if its amplitude was three standard deviations above the noise floor measured in ten neighboring frequency bins. The OCT took structural B-scans before and after each set of vibration measurements and we compared the two images to confirm that the preparation was stable, or repeated the measurements if the drift was too severe. Because OCT vibrometry is an interferometric technique, the instrument can only measure vibrations parallel to its light path. In general, the optical axis was not aligned along any of the anatomical (longitudinal, transverse, radial) axes of the cochlea. Frost et al. (25) presented a method to determine the mapping between the cochlea’s anatomical axes and the OCT optical axis and we used this method to find the longitudinal, radial, and transverse components of the optical axis for each set of measurements. These are reported in the figure captions as $\left(l,r,t\right)$ values.

Acoustic stimuli were generated and ear canal pressures were measured using a Tucker Davis Technologies (Alachua, FL) system running with a sampling rate of 130 kHz (130.20833 kS/s). Sound was delivered closed-field using a Radio Shack tweeter and the pressure in the ear canal was measured with a Sokolich (Newport Beach, CA) ultrasonic microphone. The acoustic system was calibrated at the start of each experiment and the calibration was repeated as needed. In each experiment three distinct stimuli were used: pure tones (referred to as a “frequency sweep”), the same pure tones in the presence of a 3 kHz low-side suppressor (referred to as a “sweep $+$ suppressor”), and multitone zwuis complexes (6,29). For the sweep and sweep + suppressor measurements, 32 equally spaced frequencies were presented for 62.9 ms each ($2^{13}$ samples) including 1 ms (130 samples) rise/fall times tapered with a cosine-squared envelope for a total recording time of $\sim$2 s. In the zwuis measurements a tone complex containing $N=35$ sinusoidal components having frequencies of approximately equal spacing (up to a few percent) was presented for $\sim$1 s ($2^{17}$ samples). The frequencies were chosen to have an integer number of cycles in the recording window and to lack harmonics or intermodulation distortion products up to third order. Each sinusoidal component in the tone complex was assigned a random phase, $-\pi <{\varphi }_i<\pi$ so the total pressure amplitude is $\sim \sqrt{N}$, or $\sim 10\log N$ dB SPL, higher than the amplitude at each primary. The SPLs set and noted in the results correspond to the SPL of the individual frequencies. The computer controlling the OCT can process and save a 1 s recording in $\sim$10 s and do the same for a 2 s recording in $\sim$15 s.

$2f_1-f_2$ DPOAEs were measured in response to swept two-tone stimuli, $f_1$ and $f_2$, held in a fixed ratio of $f_2=1.2f_1$. $f_2$ ranged from 1 to 48 kHz in 1 kHz steps when recording at the 25 kHz BF location and 2 to 32 kHz in half-octave steps followed by 34–60 or 64 kHz in 1 kHz steps when recording at the 45 kHz BF location. The primaries were played at 50 and 70 dB SPL for 64 identical repetitions of 15.7 ms (2048 samples) and averaged in the time domain. DPOAE levels were generally robust at 50 and 70 dB SPL and stable through the hours of experimentation. An example from one experiment is in Fig. 2 and DPOAE stability in all experiments is documented in the supporting material.

Figure 2.

Timeline to illustrate stability of the preparation over the duration of an experiment, with several courses of the intense suppressor tone. Rows (A–F) correspond to different times, with each row containing, left to right, BM motion and OHC region motion (multitone stimulus), B-scan with measurement positions indicated, DPOAEs taken at a similar time. The stimuli were multitone complexes presented from 20 to 80 dB SPL for all examples except those in row (B) where stimuli ranged from 40 to 80 dB SPL. Scale bar, 60 μm (A). Experiment no. 1006.

Following an initial set of vibration measurements along an optical axis that included the OHC region, we 1) measured DPOAEs, 2) took a volumetric image for orientation, and then at several positions spanning the BM radially we measured: 3) tuning curves in response to a frequency sweep, 4) tuning curves in response to the same sweep in the presence of the low-side suppressor, and 5) tuning curve responses to a multitone complex. In a few experiments, an additional set of uniaxial vibrations were measured in response to the zwuis tones after the sweep + suppressor runs. A complete set of these five or six measurements took about 1 h, and we repeated the set at a few closely spaced longitudinal positions, $\sim$25 μm apart. Fig. 3 illustrates the stability of the preparation and image that could be achieved, which is needed for comparing responses at approximately the same position with the three different stimulus types. The B-scans were taken immediately before the vibration measurements and the different measurements were typically 15–30 min apart. The red-hued images in Fig. 3, C and I were superpositions of Fig. 3 (A and B) and (G and H), respectively. Shifting of the preparation would result in misaligned superposition, which would be indicated by emergent magenta and yellow regions in the superposition image. In later figures, red-hued B-scans are shown, and document the imaging stability. The panels at the right of Fig. 3 show example A-scans from the two grayscale B-scans at the left. Motion responses were analyzed at the local maxima of the A-scans. Based on an earlier study, analyzing motion at the local maxima avoids interference artifacts (30).

Figure 3.

2D imaging confirmed that the preparation was stable between measurements. (A) B-scan taken before the frequency sweep measurement at the 25 kHz place. (B) B-scan at the same location taken before the sweep + suppressor measurement, recorded approximately 30 min later. (C) A false-color composite image of the two scans in (A) and (B) with the sweep B-scan shown in the magenta channel and the sweep + suppressor B-scan shown in yellow. The resulting color when the two images superimpose accurately is reddish. Inaccurate superposition or fluctuations in the brightness results in regions of yellow and magenta. Although the two image channels were independently normalized, the brightness fluctuations can be especially pronounced in the portions of the image corresponding to the fluid-filled spaces, where the poor reflectivity results in prominent salt-and-pepper noise, which gives rise to an overall hue in the parts of the image outside the OoC tissue. The white vertical lines in (C) show the quasiradial locations, or slices, where vibrometry data were collected. (D–F) The mean A-scans along the three central slices, through the OHC region, with the sweep A-scans plotted in dark red and the sweep + suppressor A-scans plotted in blue. Scale bar, 40 μm (A). Experiment no. 1008, runs 20 and 21. (G–L) Follow the same layout but from an experiment at the 45 kHz place. Scale bar, 58 μm (G). Note that only the odd-numbered slices are marked in (I). Experiment no. 1007, runs 29 and 31.

We employed slightly different radial spanning patterns at the 25 and 45 kHz BF locations. At the 25 kHz BF location, sweep and sweep + suppressor responses were measured across a 120 μm radial span in 20 μm slices (“slice” refers to measurements from one A-scan in the radially spaced set of A-scan measurements). Responses to the zwuis tones were taken across the radial span in 10 μm slices. For the sweeps, the frequencies ranged from 5 to 40 kHz and were played from 45 to 85 dB SPL in 10 dB steps. The zwuis complex had 35 frequencies covering the same bandwidth and was played at 40, 53, 67, and 80 dB SPL. At the 45 kHz BF location, vibrations were measured in a 144 μm radial span with the sweeps and sweeps + suppressors measured in 11 slices with a 14.4 μm spacing and the frequencies ranging from 5 to 60 kHz and played from 55 to 85 dB SPL. The zwuis responses were measured in 12 μm spaced slices and played from 50 to 80 dB SPL in 10 dB steps. While the mechanical responses in the 45 kHz BF location are consistent with healthy cochleae showing sharp tuning and compressive growth, we have not been able to consistently measure responses at SPLs below $\sim$50 dB due to signal/noise constraints.

The 100 dB SPL suppressor is a loud sound that could possibly cause short- or long-term damage to the cochlea. This was not borne out in our findings, and Figs. 2, S1, and S3 document the stable condition of cochlear responses over several hours of experimentation, both with displacement measurements and DPOAEs.

Results

We show results from two BF locations with two types of suppressor, multitone and two-tone. The results figures are frequency response examples, and are arranged as follows. Figs. 4 and 5: multitone suppression at the two BF locations. Fig. 6: two-tone suppression with a 90 dB SPL suppressor (25 kHz BF location). Figs. 7, 8, 9, and 10: two-tone suppression (100 dB SPL suppressor) at 25 and 45 kHz BF locations, respectively. (Multitone comparisons are included in Figs. 7 and 9.) Fig. 11: post mortem results. Fig. 12: grouped data tabulated.

Figure 4.

Responses to tones and multitone complexes at the 25 kHz BF location. (C) B-scans. The dots show positions of measurement at the BM (blue) and OHC region (red). The lower copy of the top B-scan is colored with blue, orange, and green areas indicating BM, OHC, and lateral regions, respectively. (A) BM and (B) OHC region tuning curves and phase differences in response to swept tones presented from 45 to 85 dB SPL. (E and F) Responses of the same points to multitone zwuis stimuli presented from 40 to 80 dB SPL. (D and G) Phase differences between OHC and BM responses with single-tone and multitone stimulation, respectively. $\left(l,r,t\right)=\left(0.87,-0.064,0.49\right)$. Experiment no. 1008, runs 20 and 21.

Figure 5.

Responses to tones and multitone complexes at the 45 kHz BF location. (D) B-scans. The blue, red, and violet dots indicate points on the BM, in the OHC region, and in the RL region, respectively. The lower copy of the top B-scan is colored with blue, orange, yellow, and green areas indicating BM, OHC, pillar cell, and lateral regions, respectively. (A–C) The tuning curve gains measured in response to pure tone stimuli at these three positions. (F–H) Tuning curves measured in response to zwuis tone complexes. (E and I) Phase differences between RL region and BM and OHC region and BM. $\left(l,r,t\right)=\left(-0.19,-0.32,0.93\right)$. Experiment no. 1009, runs 4 and 6.

Figure 6.

Effect of the 90 dB SPL 3 kHz low-side suppressor at the 25 kHz BF location. (A) BM responses with left panel displacement and right panel displacement normalized to ear canal pressure, as gain. Orange-hued lines are unsuppressed, Blues are suppressed. (B) Same as (A) at the OHC region. (C) B-scan showing measurement locations. Scale bar, 80 μm. (l,r,t) = (0.86, −0.40, 0.31). Experiment no. 1000.

Figure 7.

Effects of the 100 dB SPL 3 kHz low-side suppressor at the 25 kHz BF location. Tuning curves, gain and phase, evoked by single tones at the BM (A) and two locations in the OHC region (B) and (C). Responses from the single-tone sweep are plotted in reds and the sweep + suppressor are plotted in blues. The points selected are shown in the composite B-scan in (D). Scale bar, 40 μm. Responses evoked by the multitone stimuli are shown in columns (F–H) with the suppressed single-tone results overlaid. Experiment no. 1006, runs 40 (sweep), 41 (sweep + suppressor), and 44 (zwuis). $\left(l,r,t\right)=\left(0.91,-0.029,0.41\right)$. (E and I) The phase difference, OHC region-BM (unsuppressed single-tone responses in E, multitone in I).

Figure 8.

Families of tuning curves measured at multiple positions at the 25 kHz BF location. (A) Composite B-scan taken before the frequency single-tone sweep and before the sweep + suppressor measurement. The blue markers are the points whose gain curves are plotted below. Scale bar, 40 μm. Numbered vertical lines indicate slice positions. (B) A-scans for the five slices. A-scans in red correspond to the sweep and those in blue correspond to the sweep + suppressor. The dots indicated local maxima selected for analysis and comparison. Columns (C–G) show families of tuning curves evoked by the frequency sweep and sweep + suppressor at different positions along five slices in the quasiradial direction. As in the previous figure, gains evoked by the sweep are plotted in reds and those evoked by the sweep + suppressor are plotted in blues. Experiment no. 1008, runs 20 (single-tone sweep) and 21 (sweep + suppressor). $\left(l,r,t\right)=\left(0.87,-0.064,0.49\right)$.

Figure 9.

Effects of the low-side suppressor at the 45 kHz BF location. Tuning curves, gain and phase, evoked by pure tones at (A) the BM, (B) OHC/Deiters cell junction, within the OHC region, and (C) close to the RL. Responses from the single-tone sweep are plotted in reds and the sweep + suppressor are plotted in blues. (F)–(H) Show the multitone responses measured at nearby locations with the results of the sweep + suppressor overlaid. Scale bar, 30 μm (D). Small blue arrows in the phase plots of (B and G) show a portion of the suppressed curves (blue dotted lines) shifted up a full cycle. (E and I) Show the phase differences, OHC region-BM (thick lines) and RL-BM (thin lines) (unsuppressed single-tone responses in E, multitone responses in I). Experiment no. 1009, runs 17 (sweeps), 18 (suppressor), and 20 (zwuis). $\left(l,r,t\right)=\left(-0.50,-0.47,0.73\right)$.

Figure 10.

Families of tuning curves measured at different positions at the 45 kHz BF location. The layout of this figure is similar to Fig. 8 with (A) showing the composite B-scans taken before the sweep and sweep + suppressor measurements and (B) showing the A-scans taken through 5 slices in the center of the image. Columns (C–G) show families of tuning curves evoked by the sweep and sweep + suppressor at the different positions using the same color scheme as Fig. 8. Scale bar, 60 μm $\left(l,r,t\right)=\left(-0.38,-0.27,0.89\right)$.

Figure 11.

Responses from active and post mortem cochleae at the 25 and 45 kHz places in response to the sweeps and zwuis complexes. The top half of the figure shows the responses measured at the 25 kHz place in response to the sweeps (panels A–C) and multitone complexes (panels D–F) with the BM in blues, the OHC region in reds, and the post mortem responses in gray. The 80 or 85 dB SPL post mortem responses are shown, when lower SPL responses were visible out of the noise, they scaled linearly with these responses. The B-scan in (G) shows the two recording locations. Scale bar, 60 μm. Experiment no. 1008, runs 20 (sweeps), 49 (sweeps post mortem), 23 (zwuis), and 51 (zwuis post mortem). $\left(l,r,t\right)=\left(0.87,-0.064,0.49\right)$. The bottom half of the figure shows the responses measured at the 45 kHz place using the same color scheme, with the addition of the RL region in violets. Panels (H–K) show results of the sweeps and panels (L–O) show the results of the zwuis measurements The B-scan in (P) shows the three recording locations. Experiment no. 1009, runs 24 (sweeps), 45 (sweeps post mortem), 30 (zwuis), and 48 (zwuis post mortem). $\left(l,r,t\right)=\left(-0.45,-0.33,0.82\right)$.

Figure 12.

Pooled data (mean ± standard deviation) from six experiments; three at each tonotopic location. (A and B) Grouped gains at different SPLs at BF and BF/2 for the 25 kHz BF location. (C and D) Same, for the 45 kHz BF location. The two rows in (A) and (B) are BM and OHC region; the three rows in (C) and (D) are BM, OHC region, and RL region, respectively. Suppressed responses are in red symbols, unsuppressed in blue. The number of data sets included in each symbol is noted above it. If the number of data sets represented in a symbol was less than three, the suppressed/unsuppressed pair was greyed over due to sparse data. Unsuppressed and suppressed gain values were compared using a Student's t-test. When the two values were statistically distinct, an asterisk (for p < 0.05) or double asterisk (for p < 0.001) was placed above the comparison symbols.

Multitone versus single tone

Fig. 4 shows single and multitone responses from the $\sim$25 kHz BF location. In the B-scan of Fig. 3 C, the BM is identified as the upper surface, but with this very longitudinal optical axis the anatomy of the OoC is otherwise obscure. The B-scan is repeated in a color-coded image, with the blue, green, and orange regions the BM, lateral, and OHC regions, respectively. As in Fig. 1 C, the two dark areas are loosely identified with the tunnel of Corti and the outer tunnel, which are medial and lateral to the OHCs. The large reflective region between them is the “OHC region,” and with the substantially longitudinal view, each A-scan could course through several rows of OHCs. A clear difference between the responses to the single-tone and multitone stimulus types is in the sub-BF responses of the OHC region (Fig. 1, B and F). In the single-tone responses (Fig. 1 B), the sub-BF gain was almost independent of sound level, with a value close to 100 nm/Pa. In the multitone responses (Fig. 1 F) sub-BF nonlinearity resulted in lower gains at higher SPL: At 40 dB SPL the sub-BF gain (with sparse data points) was $\sim$100 nm/Pa and it fell monotonically as the sound level increased, to a gain of $\sim$30 nm/Pa at 80 dB SPL. Another property of the multitone responses from the OHC region in Fig. 4 F is a relatively steep (compared with single tone) high-frequency fall-off at high SPL. Except for the few points affected by this steep fall-off, for both stimulus types and in all frequency regions, there was greater gain in the OHC region than at the BM. The BF gain at 45 dB SPL with single-tone was slightly higher than the gain at 40 dB SPL with multitone stimulation, which is reasonable when the overall pressure of the multitones is considered. The phases were similar for the two stimulus types, and the phase difference plots in Fig. 3, D and G show leads of OHC region relative to BM at sub-BF frequencies, consistent with what we have reported previously (14,16,17).

Fig. 5 shows single and multitone responses from the $\sim$45 kHz BF location. In the B-scan of Fig. 5 D the OoC anatomy was more discernible than in Fig. 4. In particular, the surface of the RL can be identified, especially with the use of the color-coded B-scan (compare with Fig. 1 for labeling). With this approximately transverse view, the “OHC region” is smaller than with the longitudinal view of Fig. 4 D, and an A-scan would generally just pass through at most a single OHC. The responses at both the BM (Fig. 5, A and F) and RL regions (Fig. 5, C and H) were largely similar with the two different stimulus types and, like Fig. 4, the greatest response difference between the two stimulus types was in the OHC region (Fig. 5, B and G). In the single-tone OHC region responses (Fig. 5 B), the sub-BF gain was almost independent of sound level, at about 40 nm/Pa. In the multitone OHC region responses (Fig. 5 G) sub-BF nonlinearity was apparent and resulted in lower gain at higher SPL. At 50 dB SPL the sub-BF gain was close to 40 nm/Pa and it fell monotonically as the sound level increased, to a gain of $\sim$10 nm/Pa at 80 dB SPL. Close to the BF peak, at the lowest SPLs the OHC region gains were quite similar with the two stimulus types, but at 80 dB SPL the multitone responses show a deep trough that is accompanied by a phase lift (Fig. 5, G and I). Similar but milder behavior is seen in the single-tone OHC region responses, with a shallow trough observed at 85 dB SPL, accompanied by a ripple in the phase (Fig. 5 B). An explanation for the behavior is in the discussion.

An apparent multi/single-tone difference between RL responses in Fig. 5 is the phase relative to BM phase. With single-tone stimulation, the RL phase led the BM phase (thin lines in Fig. 5 E) by increasing amounts as the frequency was reduced from the BF. With multitone stimulation, the RL phase was nearly equal to BM phase throughout the frequency range (thin lines in Fig. 5 I). However, these different behaviors were not consistently observed and might be due to a small shift in the position of measurement, not the different stimulus types—Cho and Puria (31) observed these two phase behaviors at different radial positions along the RL.

Low-side suppressor

The multitone responses in Figs. 4 and 5 form a foundation for the low-side suppressed responses. In the figures to follow we show unsuppressed single-tone responses in reds and the suppressed responses in blues. In Fig. 6 we show single-tone responses at the 25 kHz place, with and without a 90 dB SPL 3 kHz suppressor tone. Fig. 6, A and B are from the BM and OHC regions, respectively, as indicated in Fig. 6 C. The left panels show displacement responses, the right panels show displacement relative to ear canal pressure. At the BM, suppression was observed in the peak region at 55 and 65 dB SPL. In the OHC region substantial suppression was observed in the peak, and relatively weak suppression was apparent sub-BF. The sub-BF OHC region motion remained substantially elevated relative to BM, and the effect of the 90 dB suppressor was not as strong as the effect of high-level multitones observed in Figs. 4 and 5.

To observe a level of suppression more in line with our multitone responses, we chose to use a 100 dB SPL 3 kHz suppressor for the remainder of the study. The suppressor tone was only on for $\sim$2 s during suppressed data runs and, as was noted in Fig. 2, did not appear to damage the cochlea.

The amplitude of the response to the 3 kHz 100 dB suppressor was greater at the lower BF (more apical) location, likely due to the more compliant OoC at that location (32). The amplitude of the suppressor was greater at the OHC region than at the BM at both BF locations. The mean and standard deviation amplitude values are as noted in Table 1. Fig. S2 in supporting material plots the amplitude of the responses to the suppressor and the probe tones together, from one data set.

Table 1.

Amplitude of the 3 kHz 100 dB SPL suppressor tone (mean $\pm$ standard deviation)

	BM	OHC region
25 kHz location	11.9 $\pm 5.7$ nm ($N=116$)	62.0 $\pm 14.0$ nm ($N=101$)
45 kHz location	5.8 $\pm 2.9$ nm ($N=225$)	23.8 $\pm 9.6$ nm ($N=73$)

In Figs. 7 and 9 we include multitone responses along with the low-side-suppressed single-tone responses, to illustrate the similarities and differences between multitone-based suppression and low-side suppression. Fig. 7 shows responses at the 25 kHz BF location. Fig. 7, A–C shows single-tone responses along with the suppressed responses. Measurement positions were at the BM (Fig. 7 A) and within the OHC region (Fig. 7, B and C). The B-scan in Fig. 7 D identifies the positions, Fig. 7, A to C (top to bottom). As described in Fig. 3, the B-scan is composed of B-scans from the unsuppressed and suppressed runs plotted as a composite and the resulting red-hued image evinces that there was little shifting between runs. The low-side suppressor reduced the responses substantially, and the suppressed responses were only observable (out of the noise) at the highest stimulus levels, 75 or 85 dB SPL. At the BM (Fig. 7 A), the 85 dB SPL suppressed response falls below the 85 dB single-tone level in the BF peak, but sub-BF there is no reduction. In the OHC region (Fig. 7, B and C), the low-side-suppressed response is reduced throughout the frequency range, and the 75 and 85 dB SPL responses scaled linearly, so that when displayed as gains they are approximately equal. Even when suppressed, at sub-BF frequencies the OHC region responses were greater than the BM responses. However, near BF the OHC region suppressed responses dropped off steeply and became smaller than the BM responses. The response phases, shown in the lower panels, were nearly unaffected by the low-side suppressor; suppressed phase responses are plotted in dotted blue lines, and lie close to the red lines. For completeness, Fig. 7 E shows the unsuppressed OHC re BM phase, which shows a lead-to-lag that is consistent with previous findings from the 25 kHz BF location (e.g., (17)).

Fig. 7, F–H shows multitone responses along with suppressed single-tone responses. At both the BM and OHC region, the low-side-suppressed responses lie approximately with the 80 dB SPL multitone responses, supporting that the mechanism for suppression is similar for multitone and low-side two-tone suppression. Fig. 7 I shows the phase differences for the multitone responses, OHC region phase—BM phase, for the OHC location in (Fig. 7 G). The multitone and single-tone phase differences in (Fig. 7, E and I) were similar except for a sudden phase increase in phase at $\sim$28 kHz, 80 dB SPL in (Fig. 7 I). This increase was due to a phase reversal at 80 dB SPL in the OHC region, which is visible upon close inspection in (Fig. 7 G).

Fig. 8 shows a more complete set of responses from a different preparation, spanning the OoC radially at the $\sim$25 kHz BF location. The B-scan in Fig. 8 A shows the positions of the reported measurements. Fig. 8 B shows the A-scans at the various slices, in red for run 20 (single tones) and blue for run 21 (suppressed single tones). If there had been no shifting of the cochlea at all between runs the red and blue A-scans would overlie; only a small degree of shifting is apparent. The pixel positions noted in Fig. 8, C–G correspond to the unsuppressed run, and the matching pixels from the suppressed run were close to these but not identical; measurement positions were at local maxima in the A-scans (28,30). We show data from the measured points in which both suppressed and unsuppressed results from approximately the same position were detectable (out of the noise).

In Fig. 8 A the large reflective region observed in slices 2, 3, and 4, pixels $\sim$150–175, corresponds to the OHC region and, as noted above, with the substantially longitudinal optical axis, each of these slices could have passed through several OHCs. Every measured position from within this region displayed suppressible sub-BF responses. The sub-BF gain of the OHC region was typically $\sim$100 nm/Pa unsuppressed, and $\sim$60 nm/Pa suppressed. Positions outside the OHC region did not display sub-BF suppression. This included the BM (upper reflective region) in slices 3, 4, 5, and 6, and the lateral region—deep positions in slices 5 and 6. BF region suppression was observed at all positions, and the suppressed responses scaled linearly—the blue lines showing the suppressed results lie approximately with each other. The unsuppressed BF gain in the OHC region was close to 2000 nm/Pa at 45 dB SPL at all but the deepest OHC region point, slice 2 pixel 172. Suppressed OHC region responses were low-pass, with gain of at most $\sim$60–70 nm/Pa. Close to the BF, at the BM and lateral region, suppressed responses lay nearly with the 85 dB SPL unsuppressed responses. In contrast, close to the BF in the OHC region, suppressed responses dropped more steeply than the 85 dB SPL unsuppressed responses.

Figs. 9 and 10 are correlates of Figs. 7 and 8, from the $\sim$45 kHz BF location. At this location the RL region could be identified, and responses from the BM, OHC region, and RL region are shown. Fig. 9, A–C shows single-tone responses in reds and suppressed single-tone responses in blues. Fig. 9, F–H shows multitone responses in reds and suppressed single-tone responses in blues, for comparison. Fig. 9, A and F correspond to BM, Fig. 9, B and G to OHC region, Fig. 9, C and H to RL region, and the B-scan in Fig. 9 D indicates the positions of measurement. Single-tone suppressed phases are included in Fig. 9, A–C and F–G as dotted blue lines. Fig. 9, E and I are phase difference results (OHC-BM in thick lines, RL-BM in thin lines) from unsuppressed single tone and multitone stimuli, respectively.

In both single and multitone responses the BM region was nonlinear around BF and approximately linear sub-BF (Fig. 9, A and F). The suppressor reduced the BF peak, but at this 45 kHz BF location the suppressed peak remained prominent at all SPLs, and remained compressively nonlinear. In the OHC region the sub-BF single-tone responses (Fig. 9 B) were elevated relative to BM sub-BF responses and nonlinear at the highest SPLs (75 and 85 dB SPL) when unsuppressed; when suppressed, the sub-BF responses dropped and were linearized (reporting from sparse data where two SPL responses were observable). In the OHC region the sub-BF multitone responses (Fig. 9 G) were nonlinear at all SPLs and the suppressed single-tone responses lay close to the 70 dB SPL multitone response. In the OHC region BF peak, the single-tone responses (Fig. 9 B) were nonlinear when unsuppressed, with a steep fall-off that eliminated the BF peak at 75 dB SPL. However, in both suppressed and unsuppressed responses, at 85 dB SPL and frequencies above a steep fall-off, a BF peak emerged. These emergent peaks were coincident with a phase lift (lower panel in Fig. 9 B). A similar emergent BF peak and phase lift appeared in the multitone OHC response (the response at 80 dB SPL, dark red line in Fig. 9 G). In Fig. 9, B and G, a portion of the suppressed phase data (blue dotted line) is also shown shifted up one full cycle, to show that the suppressed data lay with the lifted phase of the unsuppressed high SPL data (small blue arrows indicate the lift). The phase lift is apparent in Fig. 9, E and I, where the OHC-BM phase (thick lines) typically decreased to a value of $\sim$−0.5 cycles as the BF was approached, but at the highest SPL (thick gray lines), 80 dB for multitone, 85 dB for single tone, abruptly shifted to a phase of $\sim$0, beginning at $\sim$35 kHz. The significance of the phase lift and emergent peaks was discussed in Strimbu et al. (17) and will be reviewed in the discussion. Fig. 9, C and H show RL region data. In the single-tone gains of Fig. 9 C, sub-BF responses were linear (reporting from sparse data where 75 dB gain lay with the 85 dB gain). The low-side suppressor reduced the sub-BF gain slightly. In the multitone responses, the sub-BF gain at the RL (Fig. 9 H) was linear and the suppressed single-tone gain lay with the multitone gain. Close to BF in the RL region, the suppressed single-tone gains lay close to the 80 dB SPL multitone gains (Fig. 9 H). The RL region BF peak remained prominent and nonlinear even at high SPL and with suppression (both low-side and multitone), similar to the observation at the BM (Fig. 9, A and F). The RL-BM phase differences are the thin lines in Fig. 9, E and I. The phase differences were slightly negative sub-BF, growing to slightly positive close to BF. The data were more complete for the multitone responses, and these showed a level dependence in phase at frequencies close to BF (Fig. 9 I). We do not consider these SPL-dependent phase variations further here, and report them for completeness. The OHC-BM phase differences decreased to a value of $\sim$−0.5 cycles or less as the BF was approached.

Fig. 10 shows a set of responses spanning the OoC radially at the $\sim$45 kHz BF location in a different preparation. Red curves are unsuppressed single-tone responses, blue are single-tone responses in the presence of the low-side suppressor. At many positions in Fig. 10, at sub-BF frequencies the suppressed responses lay with the unsuppressed responses, and sub-BF scaling was linear. This included positions in the BM region in all slices, a position at the edge of the lateral region (pixel 52 of slice 7), and positions in the RL region in slices 4 (pixels 56 and 61) and 5 (pixel 62). The BF responses at all positions were suppressed, but the BF peak was observed at positions where moderate SPL (65 dB) responses were out of the noise (BM positions slice 8 pixels 32 and 36, RL region slice 4 pixel 61, slice 5 pixel 62). Clear cases of suppressed sub-BF responses in Fig. 10 were pixel 59 of slice 5 and pixel 57 of slice 6. Unsuppressed, these positions showed the elevated sub-BF responses that are characteristic of the OHC region in the gerbil. The suppressed gain in slice 6 pixel 57 was low pass and similar in shape to the 85 dB single-tone response, but with a reduced value. The behavior is like that of the suppressed OHC region responses in Fig. 9 B, although lacking the emergent BF peak that was observed there (and signal/noise constraints are limiting observations). The suppressed response at pixel 59 slice 5 is different—it was not low-pass and the BF peak was still robust at 85 dB SPL—it is similar to the suppressed responses from slice 4 pixel 61 and slice 5 pixel 62, which were positions in the RL region. Slice 5 pixel 59 was close to the border of the OHC body and RL and the qualitative behavior can change over small distances in this border region (17), so the with/without suppression comparison of slice 5 pixel 59 might be influenced by a measurement position shift (from OHC region to RL region).

Post mortem results

Fig. 11 shows post mortem responses from the two BF regions, with Fig. 11, A–G from the 25 kHz and Fig. 11, H–P from the 45 kHz BF locations. Fig. 11, A–C show responses to frequency sweeps at the BM and OHC regions with the in vivo vibrations plotted in colors (BM, blues; OHC region, reds) and post mortem responses plotted in gray. The phases measured after death are plotted with dotted symbols. Fig. 11, D–F shows the responses at roughly the same positions, measured in response to the multitone complexes. The B-scan in Fig. 11 G shows the measurement positions. Post mortem responses (Fig. 11, A–F) can be compared with the suppressed responses at this BF location in Fig. 7. The post mortem responses are similar to suppressed responses in being linear and low-pass in the OHC region, and linear and reduced at the BF, but they are reduced further than they were when suppressed. (For figure clarity, only the highest SPL responses are included from the post mortem responses; lower SPL responses scaled linearly when they were visible.) It is notable that the post mortem OHC region responses were smaller than at the BM. The phase responses are similar to in vivo high SPL BM phase responses. Fig. 11, H–O shows responses measured at the 45 kHz BF location, with Fig. 11, H–K in response to the sweeps and Fig. 11, L–O to multitone stimuli (BM, blues; OHC region, reds; RL region, violets; post mortem, gray) The B-scan in Fig. 11 P shows the measurement positions. Post mortem responses can be compared with the suppressed responses at this BF location in Fig. 9. The post mortem condition rendered all responses linear (as above, only the high SPL responses are shown). Sub-BF responses were reduced in the OHC region but were not changed significantly at the BM or RL. The responses still reached a maximum at BF, although the peak response of the in vivo condition was eliminated. At this BF location, measurements were made with a nearly transverse optical axis, and post mortem the responses at the BM, OHC region, and RL region were all similar in size and shape. The post mortem phases were largely similar to high-SPL in vivo BM phases, but with a smaller phase excursion close to BF.

Grouped results

Fig. 12 plots grouped data of gains at different SPLs at BF and BF/2 for the 25 kHz (Fig. 12, A and B, respectively) and the 45 kHz BF location (Fig. 12, C and D, respectively). The three preparations from each BF location went into the grouped data. The two rows in Fig. 12, A and B are BM and OHC region; the three rows in Fig. 12, C and D are BM, OHC region, and RL region. Unsuppressed single-tone responses are compared with single-tone responses in the presence of the low-side suppressor. Unsuppressed responses are in red symbols, suppressed in blue. The number of data sets included in each symbol is noted above it. If the number of data sets represented in a symbol was less than three, the suppressed/unsuppressed pair was greyed over; the symbols are still useful by providing a gain value, but there were not enough grouped data to reliably use these symbols for a suppressed/unsuppressed comparison. When the number of data sets was three or greater, a statistical comparison was made with a Student's t-test, which tests whether the suppressed and unsuppressed data in a comparison pair were distinct. When the comparison yielded a statistically significant difference, single or double asterisks were placed above the blue and red symbols, depending on the p value (^∗∗p < 0.001, ^∗p < 0.05). Pairs that lack an asterisk were not statistically distinct.

At both the 25 and 45 kHz BF locations, at BF/2 there was no significant suppression at the BM (Fig. 12, B1 and D1), and there was substantial suppression in the OHC region (Fig. 12, B2 and D2). In Fig. 12, B2 and D2, the unsuppressed and suppressed gain values were each nearly independent of stimulus level (symbols are nearly in a horizontal line—responses scaled linearly), but the suppressed gain (blue symbols) was a smaller value than the unsuppressed gain (red symbols). At the 25 kHz BF location, at BF both BM and OHC regions showed substantial suppression (Fig. 12, A1 and A2). At the BM (Fig. 12 A1), the BF suppressed response gains were approximately independent of stimulus level, at $\sim$20 nm/Pa—the effect of the suppressor was to linearize the responses. At the OHC region (Fig. 12 A2), linearization of suppressed BF responses was also apparent, although less robustly than at the BM. At the 45 kHz BF location both the BM and OHC region showed only a small degree of suppression (Fig. 12, C1 and C2)—suppression was only significant at the BM (Fig. 12 C1). This apparent lack of suppression in the gain values was likely due to the persistent or emergent BF peak at the 45 kHz location, and the fact that there was little measurable suppressed data at SPLs below 75 dB SPL. At the RL region at the 45 kHz location, BF data were available down to 55 dB SPL and the 55 and 65 dB SPL responses showed suppression (Fig. 12 C3). At BF/2 (Fig. 12 D3) RL region suppression was not apparent, and gain was approximately equal at 75 and 85 dB SPL (where it could be evaluated). The absence of sub-BF activity at the RL is consistent with most of the individual responses in Figs. 9 and 10.

Discussion

Multitone suppression

We have previously published multitone motion responses in the gerbil OoC from both the 25 and 45 kHz BF locations (14,16,17) and in a previous study we compared single and multitone voltage and displacement results from the $\sim$25 kHz BF location (24). The results of the comparison between multitone and single-tone responses presented here are consistent with the previous results, and were extended to include the 45 kHz BF location. They are more complete in frequency span and were taken more quickly due to the sweep strategy employed, which improved experimental stability. In the present comparison, at the BM and the RL (RL was only identified at the 45 kHz location) there was little qualitative difference between single-tone and multitone responses (Figs. 4 and 5). The substantial difference between multitone and single-tone responses was in the OHC region, especially sub-BF. There, single-tone responses were nearly linear, but elevated compared with BM and RL responses, and multitone responses were highly nonlinear—they were suppressed. Cochlear mechanical nonlinearity, including suppression, is thought to be based in the saturating input:output relationship of the MET channel in the OHC. The substantial suppression of sub-BF responses by a multitone stimulus is likely caused by the relatively large response to the BF-peak tones that are simultaneously present with the sub-BF tones (24).

Two-tone suppression: Expectations from a compressive nonlinearity

As just noted, cochlear nonlinearity is thought to be based in the saturating input:output relationship of the MET channel in the OHC. In this section, we consider the expected output when a two-tone stimulus comprising a single tone plus a large suppressor tone is delivered to a saturating input:output function. In our experimental observations, the effect of the high-SPL low-side suppressor included linearization of the responses. This effect can be understood within the framework of a saturating input:output curve, as in Fig. 13 A (7,33,34). This curve is a single Boltzmann function; the MET channel input:output function measured in vitro has been fitted with single and double Boltzmann functions (35,36). Fig. 13 B shows the output amplitude of a sinewave that is subject to the input:output curve, with the sinewave input amplitude on the x axis. The red curve is the output amplitude of a single sinewave input, the blue curve is the same sinewave’s output amplitude when the input was summed with a second “suppressor” sinewave of amplitude 10. Two rectangular regions are identified in Fig. 13 B. The blue rectangle to the left is a region where the red curve scaled linearly, and with the suppressor, it (blue curve) still scaled linearly but at a lower level—i.e., with a lower gain. The pink rectangle to the right is a region where the red curve scaled compressively and when suppressed (blue curve), scaled linearly at a reduced level. Both behaviors were observed in our results and are highlighted in the data of Fig. 13, C–H, with the left column from the 25 kHz BF location and the right column from the 45 kHz BF location. Each column is from a single run, and these data are shown because some of the suppressed responses were available down to 65 dB SPL—so linearity could be evaluated over the 65–85 dB range of inputs. The blue rectangles identify regions where the unsuppressed responses scaled linearly, and with the suppressor they also scaled linearly, but with reduced gain. The pink rectangles identify regions where the unsuppressed responses scaled nonlinearly and were linearized and reduced in the presence of the suppressor.

Figure 13.

(A) Input:output function of a saturating nonlinearity. (B) Output amplitude of a probe tone of input amplitude as noted on the x axis. Red, probe tone alone; blue, probe tone + second (suppressor) tone of amplitude 30. The blue rectangle overlies probe tone responses that were linear in the absence of the suppressor tone, and also linear in the presence of the suppressor tone, but at a lower gain. The pink rectangle overlies probe tone responses that scaled compressively in the absence of the suppressor, and were linearized with reduced gain in the presence of the suppressor tone. (C–E) Single-tone and single-tone + suppressor responses from the 25 kHz BF location, with reds unsuppressed and blues suppressed responses. The blue and pink rectangles identify the different response characters noted when describing (B). (F–H) Same as (C–E) for the 45 kHz BF location. The blue arrows in (F and H) identify a persistent BF peak in the BM and RL region responses.

In previous studies of low-side suppression, phasic suppression was sometimes explored, where the suppression was greatest over a certain phase of the low-side suppressor (1,8). Phasic suppression was especially clear when the low-frequency suppressor was less than 1000 Hz, and has been studied for very low-frequency suppressors where they are termed bias tones (37). For 3000 Hz suppressors, as used in the present study, suppression was primarily tonic, meaning that it persisted throughout the cycle of the suppressor (1). We did not explore phasic suppression in this study. In this regard, it is notable that the nonlinear analysis of Fig. 13 is agnostic as to frequency; this was shown analytically with a power series expansion (34) and numerically in our results, which were produced by running a two-frequency signal through a nonlinearity in a MATLAB code. (The code for this program is included in the supporting material.).

Two-tone suppression sub-BF

Until the advent of OCT, measurements at the base of the cochlea were limited to the BM, where sub-BF responses were linear and unaffected by suppression (1,6,8). With OCT, sub-BF activity within the OHC region of the OoC was observed, and could be suppressed by multitones (24) and second tones (23). Suppression of sub-BF extracellular voltage responses by an intense low-frequency suppressor has been studied (7), with results similar to the OHC region motion results summarized in Fig. 13—reduction of responses and linearization of nonlinear responses. Extracellular voltage measured close to the BM is representative of OHC current, and the similarity of the suppressive effect on OHC region motion and extracellular voltage supports the expectation that OHC motion nonlinearity is due to electromotile responses to OHC voltage.

Two-tone suppression in the BF peak

The suppressed responses in the OHC region at frequencies close to BF were remarkable at both the 25 and 45 kHz BF locations. At the 25 kHz BF location, low-side suppression produced a steep fall-off (Fig. 8, slice 3 pixels 159 and 166, slice 4 pixel 157). In Fig. 7, G and H, the high SPL (80 dB) multitone responses show a similar steep fall-off. In previous work (26), measurements along two axes were used to predict overall motion, and showed that, with the longitudinal + transverse optical axis used to probe the 25 kHz BF location, the OHC region motion could be nearly perpendicular to the optical axis (see also (15,17)). With that in mind, suppression could modify the overall motion direction, making it even more nearly perpendicular to the viewing axis, and contributing to the observed steep fall-off; the size of the observed motion reflects both the motion size and the direction of motion, both of which are likely affected by suppression.

At the 45 kHz BF location, the viewing angle was reasonably transverse. Close to BF, OHC region motion was nearly half a cycle out of phase with BM motion (Figs. 5, E and I and 9, E and I). The OHC motion that is measured is a sum of internal active OHC motion and the motion of the BM, the transversely moving supporting structure of the OoC. At high SPL and/or with low-side suppression, internal OHC motion is reduced, and the motion measured in the OHC region close to the BF can develop a trough (Fig. 9, B and G), which is likely due to the cancellation of internal OHC and BM motion (17). At frequencies at and above those of the trough, the OHC region phase lifted to join the BM phase (Fig. 9, B, G, E, and I), indicating that BM motion dominated internal OHC motion in the sum. At frequencies above the trough a peak emerged in the OHC region, which is likely due to the dominance of BM motion.

The blue arrows in Fig. 13 H identified a persistent nonlinear BF peak that was observed at the 45 kHz BF location at the BM and RL region even at high SPL and in the presence of the low-side suppressor. The persistent peak was smaller at the BM than in the RL region, but still present (see also Fig. 10 from the same preparation). It was more prominent in the Fig. 9 results from a different preparation and was observed in previous results by us and others at the 45 kHz BF location (17,31). This persistent peak at 45 kHz is in keeping with this location being generally less affected by the suppressor than the 25 kHz BF location. For example, in Fig. 9, B and G the suppressor reduced the OHC region gain to the level of the 70 dB SPL multitone gain. In contrast, at the 25 kHz BF location (Fig. 7, A and G) the suppressed OHC region gain was reduced further, to the level of the 80 dB SPL multitone gain. The smaller effect of suppression at the more basal location has at least two potential causes. Firstly, as tabulated at the beginning of the low-side suppressor results, the amplitude of the response to the 100 dB 3 kHz suppressor at the BM was about twice as large at the 25 kHz than at the 45 kHz BF location. With the assumption that the BM moves transversely, accounting for the relatively longitudinal optical axis for measurements at the 25 kHz BF location would make this ratio even larger. In the OHC region the 25:45 kHz ratio of suppressor displacement was about 2.6. The larger suppressor displacement at the 25 kHz BF would produce greater suppression, as observed. The greater displacement at the more apical location is predictable due to longitudinally reducing stiffness of the BM and OoC, and quantitatively reasonable given the stiffness variation that was measured in the base of the gerbil cochlea (32). Secondly, the 45 kHz BF location is sandwiched between the stapes and the round window. Activity boosts the traveling wave as it travels apically to its BF location (10,23), and the 45 kHz response has little distance to travel before reaching the BF place; thus, there is little accumulated activity to suppress. In a related vein, the post mortem responses at the 45 kHz BF location retained the $\sim$45 kHz maximum; apparently the response to a 45 kHz stimulus is not relying on active boosting to be attained at this location. Contrast the 25 kHz BF location post mortem observations, where responses to 25 kHz stimuli were barely or no longer detectable.

Post mortem responses

At the 45 kHz location, the optical axis was approximately transverse to the BM, and post mortem every position, BM, OHC and RL, moved with a similar frequency response and size, and linearly. The OHC region post mortem changes were the most dramatic, due to a large reduction in sub-BF responses. BM and RL region sub-BF responses did not change significantly. These findings were consistent with those of (31). The motion still reached a maximum at approximately the same BF as in vivo, with a steep fall-off at higher frequencies, but the post mortem response was nearly low pass. These responses suggest that post mortem the OoC is riding on the BM, with no additional internal motion. However, the 25 kHz location responses, made with a more longitudinal optical axis, belie this conclusion. At the 25 kHz location, post mortem responses at the BM and OHC region were reduced and linearized. Linear sub-BF responses at the BM were unchanged, and the BM reached a mild peak about a half octave below the in vivo peak. The OHC region responses were greatly reduced throughout the frequency range. However, the OHC region responses were not like those at the BM—they were smaller and lacked the mild peak. Similar findings at a similar location, also in gerbil, were noted previously (38). This is an interesting finding because it indicates that internal motions are present post mortem—the entire organ is not simply moving with the same motion as the BM. This internal motion might be purely passive, due to the structural constraints, particularly the longitudinally slanting Deiters cells. Alternatively, electromotility might be operational for many minutes post mortem, even if not functioning properly to amplify the traveling wave. Measurements of extracellular voltage showed a reduction of about a factor of 10 at 70 and 90 dB SPL, $\sim$10 min post mortem, so OHCs continue to be electrically stimulated (39). The measurements at the 25 kHz location were made with a substantially longitudinal optical axis, and the post mortem internal motion might be primarily a longitudinal motion.

Lateral region

The lateral region, observed in three points at the 25 kHz BF location in Fig. 8, responded much like the BM, with the peak suppressed by the low-side suppressor, and sub-BF frequencies unaffected by the suppressor. At the 45 kHz BF location the lateral region was not well-explored due to low reflectivity in this region. One point, pixel 52 of slice 7 in Fig. 10, is on the border of the lateral region. It responded similarly to the BM, with suppression affecting the BF peak. These findings support previous work noting that, in the base of the gerbil cochlea, sub-BF activity is localized to the OHC region (15,17).

Conclusions

Two-tone suppression has long been used to study cochlear activity (1,6,7,8,9,10,23) and OCT allows for the exploration of the effect of suppression on internal OoC components. Our findings regarding BM suppression were like those of previous studies, with suppression affecting the BF region but not the sub-BF region. The OHC region results showed reduced gain and linearization both at BF and sub-BF, findings that are consistent with previous low-side suppression measurements of extracellular voltage (7). The RL region and lateral region of the OoC showed suppressive effects that were generally qualitatively similar to the BM, with BF region gain reduction, and little effect of suppression sub-BF. The suppression due to the 100 dB SPL low-side suppressor was quantitatively equivalent to an $\sim$70–80 dB SPL 35-tone multitone stimulus. The effects of the intense suppressor tone were in keeping with the mathematical expectations of a saturating nonlinearity. Aspects of the results were likely due the axis of intra-OoC motion, and its relation to the optical axis that prescribes the axis of motion measurement.

Acknowledgments

This work was supported by the NIH/NIDCD (grant number: R01 DC-15362) and the Emil Capita Foundation.

Author contributions

E.S.O. and C.E.S. designed the experiments. C.E.S. conducted the OCT measurements and performed the initial analysis on the data. C.E.S. and E.S.O. both contributed to further analyses. C.E.S. and E.S.O. wrote the manuscript.

Declaration of interests

The authors declare no competing interests.

Editor: Christoph Schmidt.