Asymmetric mechanotransduction by hair cells of the zebrafish lateral line

Summary

In the lateral line system, water motion is detected by neuromast organs, fundamental units that are arrayed on a fish’s surface. Each neuromast contains hair cells, specialized mechanoreceptors that convert mechanical stimuli, in the form of water movement, into electrical signals. The orientation of hair cells’ mechanosensitive structures ensures that the opening of mechanically-gated channels is maximal when deflected in a single direction. In each neuromast organ, hair cells have two opposing orientations, enabling bi-directional detection of water movement. Interestingly, Tmc2b and Tmc2a proteins, which constitute the mechanotransduction channels in neuromasts, distribute asymmetrically so that Tmc2a is expressed in hair cells of only one orientation. Here, using both in vivo recording of extracellular potentials and calcium imaging of neuromasts, we demonstrate that hair cells of one orientation have larger mechanosensitive responses. The associated afferent neuron processes that innervate neuromast hair cells faithfully preserve this functional difference. Moreover, Emx2, a transcription factor required for the formation of hair cells with opposing orientations, is necessary to establish this functional asymmetry within neuromasts. Remarkably, loss of Tmc2a does not impact hair cell orientation but abolishes the functional asymmetry as measured by recording extracellular potentials and calcium imaging. Overall, our work indicates that oppositely oriented hair cells within a neuromast employ different proteins to alter mechanotransduction to sense the direction of water motion.

eTOC Blurb

Kindig et al. demonstrate that the zebrafish posterior lateral line mechanosensory cells that detect opposing directions of fluid flow have asymmetric mechanotransduction. Cells that detect anterior flow have larger mechanosensitive responses due to Tmc2a expression, indicating that the mechanotransduction channel can tune sensory system function.

Introduction

Aquatic vertebrates must establish an internal representation of the surrounding aquatic environment for survival. To detect flows in the aquatic environment the mechanosensory lateral line system of fishes mediates a range of behaviors, including orienting in a stream of water (rheotaxis)^1–3 and predator or prey detection.^4,5 The fundamental units of the lateral line system are neuromast organs. In the larval zebrafish, these organs are distributed along the body’s surface, and each neuromast contains approximately 16 sensory hair cells that are dispersed among supporting cells. For mechanotransduction, the sensory cells rely on apical structures called hair bundles. Hair bundles are ensembles of actin-based protrusions termed stereocilia that are arranged in rows of graded heights; proximal to the tallest stereocilia is a single microtubule-based kinocilium.⁶ Deflection of the hair bundle towards the kinocilium opens force-gated mechanoelectrical transduction (MET) channels.^7–9 MET channel opening causes an influx of cations that depolarizes the cell. Conversely, deflection of the hair bundle in the opposite direction, away from the kinocilium, closes MET channels. When hair cells are positively stimulated and depolarize, voltage-gated calcium channels are opened at the presynapse, allowing for calcium influx and the release of neurotransmitter, activating afferent neurons.

Within the well-studied zebrafish posterior lateral line (pLL; Figure 1A), each neuromast has hair cells oriented in opposition along an anterior-posterior or a dorsal-ventral axis of mirror symmetry.^10,11 The axis of mirror symmetry of pairs of hair cells derived from a precursor cell is defined by the position of the kinocilia of the two cells relative to their stereocilia. Each pair of these hair cells have their kinocilia oriented toward each other, giving rise to an axis of mirror symmetry.^12–15 For example, in anterior-posterior neuromasts, one cell will form with its kinocilium on the posterior side of the hair bundle, and depolarize when deflected posteriorly (A-to-P sensitive, head-to-tail flow). The other cell will form with its kinocilium on the anterior side of the hair bundle and depolarize when deflected anteriorly (P-to-A sensitive, tail-to-head flow). As these pairs form, afferent neurons selectively innervate hair cells based on orientation.¹⁶ The presence of hair cells with opposing orientations within a neuromast establishes the neuromast’s axis of best physiological sensitivity.⁶ Differences in orientation define the two hair cell populations within a neuromast, but the extent of molecular and functional differences between these two populations of hair cells remains less clear.

Figure 1. Overview of the configurations for recording microphonic potentials.

(A) Diagram of the zebrafish lateral line. The posterior lateral line (pLL) neuromasts L1-L4 used in this study are indicated. Afferent neurons are marked with blue or red. (B) Lateral-view illustration of a neuromast with the stimulus pipette and recording pipette in their correct z-axis position. + and – demonstrate directions of push and pull stimuli, respectively. Hair bundle stereocilia are yellow or blue apical tufts adjacent to kinocilia (black) (C) Recordings were taken from the dorsal, ventral, and posterior side of the neuromast in Configuration I or on the anterior, ventral, and posterior side after the fish is rotated 180º in Configuration II. Blue hair cells are P-to-A sensitive, yellow hair cells are A-to-P sensitive. Kinocilia are gray dots. (D) DIC image at 100✕ magnification of a pLL neuromast viewed top down, with pipettes in proper recording positions. Top panel: plane for stimulus pipette, aligned with the tip of the kinocilia. Bottom panel: plane for recording pipette, near the surface of the skin. Scale bar = 5 μm.

One important molecular feature that discriminates neuromast hair cells with opposing orientations is the composition of the MET channel. In mammals, transmembrane channel-like (TMC) 1 and TMC2 are thought to form the MET channel of hair cells.^17,18 In mice, mature auditory hair cells use TMC1 whereas vestibular hair cells use TMC1 and TMC2.^19,20 Zebrafish have a single tmc1 gene and two tmc2 orthologs, tmc2a and tmc2b.^21–23 In zebrafish, auditory responses are diminished when just Tmc2a is absent and abolished when Tmc1, Tmc2a, and Tmc2b are not present ²⁴; by contrast, in the zebrafish lateral line, hair cells rely just on Tmc2a and Tmc2b for mechanotransduction.²³ Importantly, in the lateral line, differences in Tmc channel composition have been observed in hair cells with opposing orientations.²³ Recently, we demonstrated that in pLL anterior-posterior sensitive neuromasts, all hair cells are dependent on Tmc2b, but a subset that are sensitive to anterior flow (P-to-A sensitive) depend on both Tmc2a and Tmc2b.²³ In addition to Tmcs, in the pLL, Emx2 is expressed in hair cells that are sensitive to posterior flow (A-to-P sensitive), but not those sensitive to anterior flow. Emx2 expression during development functions to reverse hair cell orientation in A-to-P sensitive cells to create an axis of mirror symmetry.²⁵

Here, we demonstrate that the mechanosensitive responses of neuromast hair cells with opposing orientations are functionally asymmetric. This asymmetry arises from differences in Tmc2 and Emx2 expression between hair cells that have opposing orientations. Overall, this functional difference provides a mechanism that may enable fish to more precisely identify the direction of waterflows in their natural environments.

Results

Electrode placement impacts microphonic responses

Neuromasts are composed of two populations of hair cells that are oriented in opposition,180º relative to one another, and respond best to stimuli along a single axis of best physiological sensitivity (Figure 1B-C). Hair cells in the posterior lateral line (pLL; Figure 1A) have a differential dependence on Tmc channel proteins that is stereotyped based on their orientation.²³ Therefore, we aimed to determine whether there are differences in mechanotransduction linked to orientation in the pLL of wild-type zebrafish. Previous work demonstrated that microphonic potential recordings are a powerful readout of mechanotransduction. Microphonic recordings from neuromasts have been used to measure the flow of cations that occurs adjacent to populations of like-oriented hair cells during MET channel opening caused by deflection of hair bundles toward the kinocilia (Figure 1B).^23,26–28 We sought to improve the reliability of microphonic potential recordings to be able to quantify whether there are differences in mechanotransduction based on hair cell orientation.

For our analyses, we examined posteriorly (A-to-P) and anteriorly (P-to-A) sensitive cells of neuromasts of the pLL (L1-L4; Figure 1A-C, A-to-P sensitive cells in yellow, P-to-A sensitive cells in blue; Figure 1B-C). We used a 50-Hz stimulus that alternates in the positive (push) and negative (pull) direction along the A-P axis of the fish and recorded the magnitude of potentials recorded extracellularly from each population of hair cells.

During our initial experiments, we noticed that the position of the recording electrode impacted our measurements. We hypothesized that the proximity of the recording electrode to one hair cell type may bias the recording towards that type. For example, although each neuromast has an equal number of A-to-P and P-to-A sensitive cells (examples: Figure S1), each side of the neuromast (anterior versus posterior) is not populated by an equal proportion of each type of hair cell. To test whether electrode position is a confounding factor in our measurements, we compared recordings taken from two stimulation configurations. In Configuration I, we placed the stimulus pipette on the anterior side of the neuromast and the recording pipette at the posterior side of the organ. In Configuration II, we rotated the fish 180º and made recordings on the anterior side with the stimulus pipette on the posterior side of the neuromast (Figure 1C). In both stimulation configurations, we placed our recording electrode in dorsal (D) and ventral (V) positions, as well as in anterior (A) or posterior (P) positions (Figure 1C).

Using this approach, we discovered a bias when the recording electrode was placed in the anterior or posterior positions. We found a bias towards A-to-P sensitive cells when the recording pipette was placed on the anterior side of the neuromast, and a bias towards P-to-A sensitive cells when the recording pipette was placed on the posterior side (Figure 2). Specifically, the mean microphonic magnitude ± SEM (for these and all subsequent measurements) for A-to-P sensitive cells was higher when the recording was made on the anterior side of the neuromast (14.95 ± 0.73 μV) compared to the posterior side of the neuromast (7.32 ± 0.31 μV, n = 8 neuromasts, P = 0.0078 Wilcoxon test). Similarly, the mean microphonic magnitude for P-to-A sensitive cells was lower when the recording was made on the anterior side of the neuromast (8.87 ± 0.49 μV) compared to the posterior side of the neuromast (15.46 ± 1.08 μV, n = 8 neuromasts, P = 0.0234, Wilcoxon test). In contrast, when the recording pipette was placed on the dorsal or ventral side of the neuromast the recordings were not significantly different for hair cells of the same orientation (Figure 2B).

Figure 2. Recording pipette position affects microphonic potential magnitude.

(A) DIC image of the neuromast on which subsequent illustrations in c are based. (B) Mean potential magnitude of A-to-P sensitive (yellow) and P-to-A sensitive (blue) hair cells from each of the six recording possibilities (n = 8 neuromasts). Along the x-axis, the first letter indicates the origin of the stimulus—i.e. A- means the stimulus pipette is on anterior side and a positive deflection pushes hair cells toward the posterior. The second letter indicates recording pipette position. ****P < 0.0001, one-way ANOVA, all groups. ***P = 0.0003, one-way ANOVA, all A-to-P sensitive groups. *P = 0.0236, one-way ANOVA, all P-to-A sensitive groups. **P = 0.0078, Wilcoxon test, P-A vs A-P, A-to-P sensitive. *P = 0.0234, Wilcoxon test, A-P vs. P-A, P-to-A sensitive. (C) Illustration of the 6 recording possibilities and corresponding traces from the sample neuromast shown in A. Note the bottom row: in P-A the pipette is closer to A-to-P sensitive cells (left), while in A-P the pipette is closer to P-to-A sensitive cells (right). These positions bias magnitudes towards one orientation even though both hair cell types are present in equal number. Scale bar = 10 μm. See also Figure S1,S2.

Our results strongly suggest that placement of the recording electrode closer to hair cells of one orientation (A or P positions) biases the recording. To test this hypothesis, we determined the average distance from the electrode to the nearest hair bundle of each orientation (Figure S2). We found that the electrode was indeed closer on average to an A-to-P sensitive cell when placed on the anterior side, whereas it was generally closer to a P-to-A sensitive cell when on the posterior side (Figure S2). In contrast, the mean distance to either orientation was similar for the dorsal and ventral recording positions (Figure S2). Consequently, all subsequent recordings were made from the dorsal and ventral side of the neuromast. We consider these neutral or unbiased positions on each side of the neuromast. Having an unbiased recording position in the zebrafish preparation is an essential requirement to test whether there are differences in mechanotransduction in hair cells of different orientations.

Microphonic responses are larger in P-to-A sensitive hair cells

To explicitly test the hypothesis that hair cells with opposing orientations have differences in the properties of mechanotransduction, we measured stimulus-evoked microphonic potentials at neutral recording positions (Figure 3). When comparing recordings from these positions, we observed a difference in magnitude between A-to-P and P-to-A sensitive cells. To ensure that this difference was not simply due to an unequal number of hair cells, microphonic potentials were divided by the number of A-to-P or P- to-A sensitive cells in that neuromast to determine the mean response per hair cell (Figure 3A,B). Using this approach, we found within the same neuromast, during a 50-Hz stimulus, P-to-A sensitive cells had a significantly larger mean microphonic magnitude (1.35 ± 0.02 μV per cell) compared to A-to-P sensitive cells (1.12 ± 0.01 μV per cell) (Figure 3B, n = 21 neuromasts, *P = 0.0233, paired t-test).

Figure 3. The microphonic potential magnitude is larger in P-to-A sensitive cells.

(A) Representative microphonic traces from an A-P oriented neuromast during a 50-Hz sinusoidal stimulus, which excites hair cells of opposing orientations sequentially. Top and bottom traces correspond to Configuration I and II respectively (See Figure 1C). P and A above the trace indicates the deflection of the neuromast is toward that side of the animal. (B) During sinusoidal stimuli the mean microphonic response of A-to-P sensitive (1.123 ± 0.014 μV/HC) is smaller than P-to-A sensitive (1.353 ± 0.024 μV/HC) cells (n = 21 neuromasts, *P = 0.0233, paired t-test). (C) Sample microphonic traces from an A-P oriented neuromast during steps of sustained deflection. Top and bottom traces correspond to Configuration I and II respectively (See Figure 1C). (D) During step stimuli the mean microphonic response of A-to-P sensitive (1.131 ± 0.025 μV/HC) is smaller than P-to-A sensitive (1.436 ± 0.032 μV/HC) cells (n = 9 neuromasts, *P = 0.0123, paired t-test).

To determine if the difference in response magnitude was affected by the duration of stimulation, we performed similar experiments but with longer, sustained 200-ms stimuli, instead of 50-Hz sinusoidal stimuli (Figure 3C). When using a step stimulus we found that P-to-A sensitive cells had a significantly larger mean microphonic magnitude (1.44 ± 0.03 μV per cell) compared to A-to-P sensitive cells (1.13 ± 0.02 μV per cell) (Figure 3D, n = 9 neuromasts, *P = 0.0123, paired t-test). These findings indicate that the ensemble of hair cells that are sensitive to anterior water flow has larger microphonic potentials.

Evoked calcium signals are larger in P-to-A sensitive hair cells

Our electrophysiology recordings revealed that for a comparable stimulus, P-to-A sensitive cells had larger mechanosensitive responses when compared to A-to-P sensitive cells. To verify this finding, we performed calcium imaging to provide an additional readout of the process of mechanotransduction. Previous work established a hair cell-specific, membrane-localized GCaMP6s transgenic line²⁹ (Figure 4A). This transgenic line can be used to visualize evoked calcium signals in individual hair bundles of pLL neuromasts and provide a readout of mechanotransduction. Therefore, we compared the GCaMP6s signals in A-to-P sensitive and P-to-A sensitive hair bundles. The stimuli were delivered in a similar manner as our microphonic potential recordings, using a 500-ms sustained positive deflection (push) as the stimulus.

Figure 4. Mechanosensitive calcium signals are larger in hair bundles of P-to-A sensitive cells.

(A) Schematic of calcium imaging setup. Left: a side-view of a neuromast expressing membrane localized GCaMP6s (GCaMP6s-CAAX; yellow/blue). The z-depth used for imaging is indicated. Right: a top-down view the imaging plane showing A-to-P sensitive (yellow) and P-to-A sensitive (blue) hair bundles. Pipettes indicate the direction and position of the fluid jet during the stimulus. (B) Representative example of evoked-mechanosensitive calcium signals in hair bundles of an A-P pLL neuromast during a 500-ms A-to-P (C) or a 500-ms P-to-A (D) directed stimulus. Spatial patterns of GCaMP6s signals during stimulation (C,D) are colorized according to the ∆F heat map and superimposed onto a baseline GCaMP6s images (B). ROIs in E were used to generate the ∆F/F GCaMP6s traces from individual A-to-P sensitive (F) and P-to-A sensitive (G) hair bundles. (H) Traces show the average GCaMP6s response per neuromast in A-to-P sensitive versus P-to-A sensitive hair-bundle populations (n = 9 neuromasts). Dot plot in I shows that the average GCaMP6s increase for A-to-P sensitive (63.29 ± 5.99) is less than P-to-A sensitive (108.0 % ± 15.79) hair bundles for each neuromast (*P = 0.0175, unpaired t-test). Scale bar = 5 μm. See also Figure S3.

Using this approach, we were able to identify A-to-P sensitive and P-to-A sensitive cells based on stimulus-evoked GCaMP6s signals in individual hair bundles (Figure 4B-G). By examining the magnitude of the calcium signals, we observed that overall, the P- to-A sensitive GCaMP6s signals were larger compared to A-to-P sensitive GCaMP6s signals. We found that during a 500-ms saturating stimulus, the average magnitude of GCaMP6s signals of P-to-A sensitive cells (108.0 ± 15.79 % per neuromast) were significantly larger compared to A-to-P sensitive cells (63.29 ± 5.99 % per neuromast) (Figure 4H-I, n = 9 neuromasts, P = 0.0175, unpaired t-test). We also examined GCaMP6s signals during a 500-ms deflection across three stimulus intensities (see Methods). At all three stimulus intensities the average magnitude of GCaMP6s signals of P-to-A sensitive cells were significantly larger compared to A-to-P sensitive cells (Figure S3). Overall, our calcium imaging results indicate that hair cells that are sensitive to anterior flow have larger mechanosensitive calcium signals.

Emx2 is linked to the magnitude of mechanosensitive calcium signals

The transcription factor Emx2 plays a pivotal role in determining hair cell orientation in both the zebrafish pLL and mouse vestibular hair cells.²⁵ In zebrafish A-P neuromasts in the pLL, A-to-P sensitive cells express Emx2, whereas P-to-A sensitive cells do not (Figure 5A-C). Furthermore, in emx2 loss-of-function (lof) zebrafish mutants, all hair bundles are sensitive to anterior flow, and, when Emx2 expression is driven in all lateral-line hair cells (emx2 gain-of-function; gof), the majority of cells are sensitive to posterior flow (Figure 5D,G). Our functional analyses found that in wild-type animals, the mechanosensitive responses of P-to-A sensitive cells were larger compared to A-to-P sensitive cells. We hypothesized that in addition to determining hair cell orientation, the presence or absence of Emx2 expression could also impact the mechanosensitive properties of lateral-line hair cells. As a transcription factor, Emx2 could alter Tmc expression, localization, or function by another mechanism in an orientation-specific manner. Therefore, we examined whether the wild-type response properties of P-to-A sensitive and A-to-P sensitive cells are intact in emx2 lof or emx2 gof hair cells, respectively.

Figure 5. The mechanosensitive calcium responses are larger in emx2 loss of function (lof) hair cells.

(A) Top-down view of a wild-type, A-P-oriented pLL neuromast. Phalloidin is used to visualize hair bundle orientation. In the same neuromast, Emx2 is present only in A-to-P sensitive cells (yellow asterisks) (B). Myo7a is present in all hair cells. (C) Schematic linking Emx2 and hair bundle orientation in A-P pLL neuromasts. (D,G) Schematic of hair bundle orientation in emx2 lof (D) and gof (G) mutants. (E-F, H-I) Representative examples of evoked-mechanosensitive calcium signals from neuromasts in emx2 lof or gof mutants where all the hair bundles are P-to-A or A-to-P sensitive, respectively. Spatial patterns of GCaMP6s signals during a 500-ms A to P and P to A stimulation (E,H) are colorized according to the ∆F heat map and superimposed onto a baseline GCaMP6s image. In F and I, ∆F/F GCaMP6s traces from individual hair bundles in E and H are plotted. (J-K) Averaged traces reveal an increase in GCaMP6s signal only during the P to A stimulus in emx2 lof mutants (n = 7 neuromasts) and during the A to P stimulus in emx2 gof mutants (n = 7 neuromasts). (L) The magnitude of the GCaMP6s signals of P-to-A sensitive cells in emx2 lof mutants (87.63 % ± 5.03) is larger than the A-to-P sensitive cells in emx2 gof mutants (39.00 ± 2.57) (****P< 0.0001, unpaired t-test). Scale bars = 5 μm.

To examine the mechanosensitive response properties of emx2 lof and gof mutant hair cells, we measured stimulus-evoked GCaMP6s signals in hair bundles (Figure 5). In emx2 lof hair bundles, we only observed GCaMP6s signal increases in response in the P to A direction (Figure 5E-F). Similarly, in the majority of emx2 gof hair bundles, we only observed GCaMP6s signal increases in response in the A to P direction (Figure 5H-I). This is consistent with the hair bundle orientation phenotype in emx2 lof and gof mutants, where the majority of hair bundles are P-to-A or A-to-P sensitive, respectively. Interestingly, we also observed that the magnitude of the GCaMP6s signals were larger in emx2 lof hair bundles compared to emx2 gof hair bundles (Example: Figure 5E-F,H-I). We averaged the magnitude of the P-to-A sensitive and A-to-P sensitive GCaMP6s signals per neuromast and found that P-to-A sensitive cells in emx2 lof mutants (87.63 ± 5.03 % per neuromast) had larger magnitudes compared to the A-to-P sensitive cells in emx2 gof mutants (39.00 ± 2.57 % per neuromast) (Figure 5J-L, n = 7 neuromasts, P < 0.0001, unpaired t-test). Overall, our calcium imaging results indicate that the larger mechanosensitive responses observed in P-to-A sensitive cells in the wild-type zebrafish are preserved in emx2 lof mutants where all hair cells have P-to-A sensitivity. Furthermore, our results suggest that in addition to directing hair cell orientation, Emx2 expression also functions to impact the mechanosensitive properties in groups of hair cells with opposing orientations within a neuromast.

Loss of Tmc2a abolishes the asymmetry in mechanotransduction

To further understand the molecular origins of the asymmetric mechanosensitive responses in groups of oppositely facing hair cells in the neuromasts of the pLL, we tested the hypothesis that differences in Tmc2a expression give rise to functional differences in A-to-P sensitive versus P-to-A sensitive cells. Previously, we have shown that in neuromasts oriented along the A-P axis of the fish, mechanotransduction is absent in tmc2b ^cwr2 mutants, except in a subset of P-to-A sensitive cells. In tmc2b ^cwr2 tmc2a ^cwr3 double mutants, loss of Tmc2a eliminates all mechanosensitive responses, even in the remaining subset of P-to-A sensitive cells.²³ In the pLL, many P-to-A sensitive cells depend on both Tmc2a and Tmc2b, as opposed to A-to-P sensitive cells that rely primarily on Tmc2b (Figure 6A).²³ To test if differences in Tmc2a expression could account for the elevated response magnitude in P-to-A sensitive cells, we measured mechanotransduction using microphonic potentials and calcium imaging in tmc2a ^cwr3 mutants.²⁴

Figure 6. Microphonic potentials and mechanosensitive calcium responses show a loss of functional asymmetry in hair cells of a tmc2a mutant.

(A) Schematic outlining the relationship between hair cell orientation and Tmc2 use in A-P neuromasts of the pLL. (B) Example microphonic traces from tmc2a heterozygotes and tmc2a homozygous siblings with equal numbers of P-to-A and A-to-P sensitive cells per neuromast (stimulus was from the posterior). (C) Mean magnitude of microphonic potentials for A-to-P (0.9 ± 0.06 μV/HC) and P-to-A sensitive (0.87 ± 0.07 μV/HC) cells is not significantly different in tmc2a^cwr3 mutants (n = 12 neuromasts, P = 0.634, paired t-test; sibling control is shown in Figure. S4). (D-F) The increase of the GCaMP6s signals in P-to-A sensitive hair bundles in tmc2a^cwr3/+ siblings (tmc2a^cwr3/+or tmc2a^+/+, 106.40 % ± 9.21) is significantly larger than A-to-P sensitive hair bundles (60.19 % ± 7.38) (n = 9 neuromasts, P = 0.0035, 2-way ANOVA). In contrast, the increase of the GCaMP6s signals of P-to-A sensitive hair bundles in tmc2a^cwr3 mutants (68.90 % ± 10.72) is not significantly larger than A-to-P sensitive hair bundles (70.03 % ± 6.18) (n = 9 neuromasts, P > 0.9999, 2-way ANOVA). In F, the GCaMP6s data shown in D-E is plotted to compare the average GCaMP6s increase for each neuromast in tmc2a^cwr3/+ siblings (D) and tmc2a^cwr3 mutants (E). See also Figure S4 and Figure S6.

We first recorded microphonic potentials and found that for sibling controls (tmc2a^+/−) the mean magnitude per hair cell was significantly different between A-to-P (0.83 ± 0.06 μV) and P-to-A sensitive (1.09 ± 0.08 μV) cells (Figure S4, n = 7 neuromasts, P = 0.0038, paired t-test). Importantly, we found that the mean magnitude per hair cell was not significantly different in A-to-P sensitive (0.90 ± 0.06 μV) and P-to-A sensitive (0.87 ± 0.07 μV) cells of tmc2a^cwr3 mutants (Figures 6B,C, S4, n = 12 neuromasts, P = 0.634, paired t-test). To verify this result, we also examined GCaMP6s responses in the hair bundles of tmc2a^cwr3 mutants. We found that similar to wild-type animals, in tmc2a siblings the magnitude of GCaMP6s responses in P-to-A sensitive cells (106.40 ± 9.21 % per neuromast) was significantly larger than in A-to-P sensitive cells (60.19 ± 7.38 % per neuromast) (Figure 6D,F, n = 9 neuromasts, P = 0.0035, two-way ANOVA). In contrast, in tmc2a^cwr3 mutants, the magnitude of GCaMP6s signals in P-to-A sensitive (68.90 ± 10.72 % per neuromast) and A-to-P sensitive (70.03 ± 6.18 % per neuromast) cells were not significantly different (Figure 6E,F, n = 9 neuromasts, P > 0.9999, two-way ANOVA).

Together our GCaMP6s and microphonic measurements of mechanosensitive function in pLL hair cells indicate that loss of Tmc2a results in a loss of mechanosensitive asymmetry between A-to-P sensitive and P-to-A sensitive cells. Further, these data indicate that the presence of Tmc2a in P-to-A sensitive cells augments mechanosensitive responses.

Asymmetric responses are preserved in afferent neuron processes

Our calcium-imaging and electrophysiological measurements demonstrated that the presence of Tmc2a can augment the mechanosensitive responses of P-to-A sensitive cells. This enhancement could allow the zebrafish pLL to be more sensitive to anterior-versus posterior-directed fluid flow. However, to preserve this increased sensitivity to anterior-directed fluid flow, this information must be faithfully transmitted to the downstream neural circuitry. In the pLL each neuromast is innervated by afferent neurons that selectively innervate hair cells based on orientation ³⁰ (Figure 7A). Therefore, in wild-type animals, we examined the response properties of the pLL afferents during anterior or posterior-directed flow.

Figure 7. Differences in functional asymmetry are preserved in pLL afferent processes.

(A) Schematic of a side-view of a wild-type A-P pLL neuromast. Hair cells are selectively innervated by afferent neurons based on orientation. The z-depth for imaging is indicated. (B) Top-down image of GCaMP6s in the afferent process beneath a neuromast. (C-F) Representative example of evoked calcium signals in afferent process beneath during a 500-ms A to P (C) or P to A (E) directed stimulus. Spatial patterns of GCaMP6s signals during stimulation (C,E) are colorized according to the ∆F heat map and superimposed onto a baseline GCaMP6s image (B). In B, the colored ROIs indicate calcium signals linked to stimulus direction (A to P, yellow; P to A, blue) and were used to generate the ∆F/F GCaMP6s traces in D and F. (G) Traces show the average afferent response is larger when P-to-A sensitive cells are stimulated compared to when A-to-P sensitive are stimulated (n = 12 neuromasts). (H) The dot plot shows that the average GCaMP6s increase in the afferent process is larger when P-to-A (87.49 % ± 7.18) versus A-to-P (57.40 ± 4.07) sensitive cells are stimulated (**P = 0.0014, unpaired t-test). Scale bar = 5 μm. See also Figure S5.

Prior to our functional analysis, we examined whether there were any obvious differences in the synaptic architecture between A-to-P sensitive and P-to-A sensitive cells. For this analysis, we performed immunohistochemistry to label the hair cell presynaptic densities, or ribbons, using an antibody against the main component, Ribeye, while also labeling the postsynaptic densities using a pan-Maguk antibody (Figure S5A-B). Careful examination of synaptic architecture in A-to-P sensitive and P-to-A sensitive cells revealed no significant difference in the number of complete synapses per cell type. In addition, we observed no difference in the size of the presynaptic or postsynaptic densities between in A-to-P sensitive and P-to-A sensitive cells (Figure S5C-E, n = 8 neuromasts, no significance).

After verifying that the synaptic architecture was largely similar in P-to-A and A-to-P sensitive cells, we examined the functional properties of pLL afferents. To examine response properties of these neurons, we used a transgenic line that expresses GCaMP6s in the afferent neurons of the lateral line system.²⁹ Using this line, we could reliably detect evoked GCaMP6s signals in the afferent process beneath pLL hair cells during an anterior or posterior-directed 500-ms stimulus (Figure 7C-F). We found that the average increase of GCaMP6s signals in the afferent process was significantly larger when we stimulated P-to-A sensitive cells (87.49 ± 7.18 % per neuromast) compared to when we stimulated A-to-P sensitive cells (57.40 ± 4.07 % per neuromast) (Figure 7G-H, n = 12 neuromasts, P = 0.0014, unpaired t-test). Overall, our calcium imaging results indicate that the enhanced response of P-to-A sensitive cells is transmitted to the downstream afferent neuron processes. Further, these data indicate the pLL system is more responsive to anteriorly directed fluid flow originating from a stimulus pipette and similar stimuli found in nature.

Discussion

In the zebrafish pLL, hair cells with A-P orientations within a neuromast reliably detect and transmit information about water flow along the A-P axis of the fish. Although they appear morphologically similar, hair cells with opposing orientations depend on different Tmc proteins for mechanotransduction. Based on this molecular difference, we hypothesized that within neuromasts, groups of hair cells that face opposite directions have different mechanosensitive properties. This hypothesis was counter to the general notion that opposing hair cells have similar mechanosensitive properties. Overall, we demonstrate using microphonic potentials (Figure 3) and calcium imaging (Figure 4) that P-to-A sensitive cells have larger mechanosensitive responses compared to A-to-P sensitive cells, indicating a functional asymmetry based on hair cell orientation (Figure 3).

Next, we tested the hypothesis that this functional asymmetry was dependent on the presence of Tmc2a protein, which has only been observed in P-to-A sensitive cells. Our results indicate that functionally eliminating Tmc2a from the hair cells of neuromasts in the pLL results in symmetric rather than asymmetric mechanotransduction in neuromast organs (Figure 6). This result strongly suggests that differences in MET channel proteins give rise to different mechanosensitive properties. Here, cells that express both Tmc2b and Tmc2a (P-to-A sensitive) have enhanced responses compared to cells that express only Tmc2b (A-to-P sensitive). Enhanced responses could be due to differences in ion permeability. For example, work in mammalian hair cells has shown that Tmc proteins that give rise to channels with increased calcium permeability also have larger mechanosensitive currents.^17,18,31,32 This idea is supported by our results, as we show that the calcium signals in hair bundles expressing both Tmc2b and Tmc2a are larger compared to those expressing just Tmc2b (Figure 6). Molecular mechanisms that may permit P-to-A sensitive hair cells to have enhanced responses are Tmc2a and Tmc2b homomerization or Tmc2a and Tmc2b heteromerization within mechanotransduction channels (Figure S6). In the future, measurements of single-channel properties in the presence or absence of Tmc2a and Tmc2b may give further insight into the effect their direct or indirect association has on mechanotransduction.

Our work also investigated another molecular difference that has been described in pLL hair cells—expression of the transcription factor Emx2.²⁵ In the pLL, Emx2 is expressed only in A-to-P sensitive cells. In A-to-P sensitive cells, Emx2 is required to establish the orientation of these cells relative to P-to-A sensitive cells, creating two populations of hair cells with opposing orientations. We show that the mechanosensitive responses of hair cells in emx2 lof mutants (all P-to-A sensitive) are larger compared to emx2 gof mutants (all A-to-P sensitive). Based on this result, our work indicates that in A-to-P sensitive cells, Emx2 expression not only acts to establish hair cell orientation but also can influence the mechanosensitive properties of pLL hair cells. As a transcription factor, Emx2 could directly repress tmc2a expression, as Emx2 has been shown to act in this capacity in other biological contexts, such as in development of the central nervous system.³³ Alternatively, it is possible that Emx2 does not directly control the expression of tmc2a but controls Tmc2a localization as Emx2 does for another hair cell-expressed protein, Gpr156.³⁴

What is the physiological need underlying the Tmc- and Emx-dependent reduction in sensitivity to posteriorly directed water flow? It is possible that during normal forward swimming, fish may need lower sensitivity to posterior flow to avoid constant overstimulation. In support of this idea, work has demonstrated that efferent feedback selectively inhibits responses to rearward stimuli³⁵; however, other work indicates that this may not always be the case.³⁶ This inhibition is thought to reduce sensitivity to non-threatening, self-generated stimuli that occurs during forward movement. In previous work, we have found that Tmc-dependent differences also exist in pLL dorsal-ventral sensitive neuromasts.²³ However, it is not clear if a functional asymmetry exists in these neuromasts, or how an asymmetry in ventral versus dorsal directed-fluid detection would impact the natural behavior of larval zebrafish.

Lateral line afferents also carry information to a pair of bilateral Mauthner cells and other neurons of the hindbrain to govern the fast startle reflex and rheotaxis, respectively.^37,38 Asymmetric responses to water flow in pLL hair cells may be incorporated differently in these two behaviors. It has been demonstrated that the information for anterior- and posterior-directed flow detected by the pLL during rheotaxis (forward swimming against a posterior-directed current) is sent to different regions of the hindbrain.³⁹ In the hindbrain, signals of different strengths received from afferents originating from asymmetric mechanotransduction in neuromasts may guide rheotaxis through these anatomical differences in neuronal circuitry. In addition, asymmetric mechanotransduction in pLL neuromasts may also impact the fast startle response. Although different sets of afferent neurons carry anterior and posterior flow information from lateral line hair cells, both sets are presumed to innervate a single Mauthner cell. If these two sets of afferent neurons that carry flow information to the Mauthner cell are equivalent (including factors such as synaptic strength), then it is possible that for a comparable stimulus, anterior-directed flow is more likely to elicit a fast startle response than posterior flow. It has been shown that piscine predators can path-follow fish prey with attacks predominantly from behind.⁴⁰ Adaptations that enhance fast startle in response to anterior flow generated by predators approaching from the rear, such as Tmc2a expression in P-to-A hair cells, could directly assist in predator avoidance and thus survival. Therefore, pLL asymmetry at the level of mechanotransduction may help govern two fundamental fish behaviors: predator avoidance and orienting in a stream of water.

Zebrafish is not the only animal that uses Tmcs to detect flow from different directions. Drosophila larvae rely on a single Tmc gene to detect larval movement from opposing directions. In Drosophila, two different proprioceptive neurons with distinct anatomical positions and morphologies are exposed to opposing forces, allowing differential detection of forward versus backward locomotion.⁴¹ However, neuromasts rely on differential expression of Tmc2 isoforms, along with hair cell orientation, to enhance mechanotransduction in hair cells that sense anterior-directed flow. This strategy of differential expression of Tmc proteins to alter the mechanosensitive properties of sensory cells has implications for mechanotransduction in the ears of fishes and mammals. The ear of zebrafish expresses three Tmc isoforms (Tmc1, Tmc2a and Tmc2b), and the vestibular organs of mice express Tmc1 and Tmc2. Similar to what we have shown in the lateral line, these sensory organs may use differential expression of Tmc proteins to scale the magnitude of mechanotransduction.

Overall, by examining the fundamental unit of the zebrafish pLL, the neuromast, we show that hair cells with opposing orientations express different proteins that yield functional differences in mechanotransduction. Further, these differences are faithfully transmitted to the brain. Our results suggest that aquatic vertebrates integrate this asymmetrical information from neuromasts to reliably establish behaviors necessary for survival.

STAR★Methods

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Brian M. McDermott, Jr. (Bmm30@case.edu)

Materials availability

This study did not generate new unique reagents.

Data and code availability

• Raw data reported in this study have been deposited and are publicly available as of the date of publication. The DOIs are DOI:10.5061/dryad.g79cnp5v2 and DOI:10.5061/dryad.jm63xsjg2 and are listed in the key resources table.

• All data reported in this paper will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-myosin7a	Proteus	Cat#25–6790
Mouse anti-Emx2	Trans Genic	Cat#KO609
Mouse anti-pan-MAGUK (IgG1)	Millipore	Cat#MABN7
Mouse anti-Ribeye b (IgG2a)	Ref. 43
Goat anti-Rabbit IgG Alexa Fluor™ 546	ThermoFisher Scientific	Cat#A-11010
Goat anti-Mouse IgG Alexa Fluor™ 647	ThermoFisher Scientific	Cat#A-21236
Goat anti-Rabbit IgG Alexa Fluor™ 488	ThermoFisher Scientific	Cat#A-11008
Goat anti-Mouse IgG2a Alexa Fluor™ 546	ThermoFisher Scientific	Cat#A-21133
Goat anti-Mouse IgG1 Alexa Fluor™ 647	ThermoFisher Scientific	Cat#A-21240
Chemicals, Peptides, and Recombinant Proteins
Tricaine (0.03 % ethyl 3-aminobenzoate methanesulfonate salt	SigmaAldrich	Cat#886–86-2
alpha-bungarotoxin	Tocris	Cat#11032–79-4
Alexa 488 Phalloidin	ThermoFisher	Cat#A-12379
ProLong™ Gold Antifade Mountant	ThermoFisher	Cat#P36930
Experimental Models: Organisms/Strains
Zebrafish: Tübingen Wildtype	CWRU NIH	ZFIN ID: ZDB-GENO-990623–3
Zebrafish:Tg(myo6b:GCaMP6s-CAAX)idc1Tg	NIH	ZFIN ID: ZDB-ALT-170113–3
Zebrafish: Tg(en.sill,hsp70l:GCaMP6s) idc8Tg	NIH	ZFIN ID: ZDB-ALT-171206–1
Zebrafish: Tg(myo6b:emx2-p2A-nls-mCherry) idc4Tg	NIH	ZFIN ID: ZDB-ALT-170606–4
Zebrafish: emx2idc5	NIH	ZFIN ID: ZDB-ALT-170606–5
Zebrafish: tmc2acwr3	CWRU Chen et al.23	N/A
Software and Algorithms
jClamp	SciSoft	http://www.scisoftco.com/
ImageJ	NIH	RRID:SCR_003070 https://imagej.net/ij/
Clampfit 10.7.0.3	Molecular Devices	RRID:SCR_011323 https://support.moleculardevices.com/s/article/Axon-pCLAMP-10-Electrophysiology-Data-Acquisition-Analysis-Software-Download-Page
GraphPad Prism 8	Graphpad Software	RRID:SCR_002798 https://www.graphpad.com/scientific-software/prism/
MATLAB 9.0 R2016a	MathWorks	RRID:SCR_001622 https://www.mathworks.com/products/compiler/matlab-runtime.html
Deposited Data
Electrophysiology raw data	Dryad	DOI: 10.5061/dryad.g79cnp5v2
Calcium imaging raw data	Dryad	DOI: 10.5061/dryad.jm63xsjg2

Experimental model and subject details

Zebrafish

Zebrafish were bred and cared for in the Case Western Reserve Zebrafish Facility or at the National Institutes of Health (NIH) under animal study protocol #1362–13 following standard protocols.⁴² The previously described mutant and transgenic lines were used in this study: Tg(myo6b:GCaMP6s-CAAX)^idc1Tg, Tg(en.sill,hsp70l:GCaMP6s)^idc8Tg, Tg(myo6b:emx2-p2A-nls-mCherry)^idc4Tg (emx2 gof), emx2^idc5(emx2 lof), and tmc2a^cwr3.^16,24,25,43 The tmc2a^cwr3 mutant has a premature stop codon and results in a functionally null phenotype that manifests in a tmc2b mutant background in lateral line hair cells.²⁴ In tmc2b mutants a subset of lateral line hair cells (primarily P-to-A sensitive) retain mechanosensitive function. This subset relies on Tmc2a, and their mechanosensitive function is lost in tmc2b ^cwr2; tmc2a ^cwr3 double mutants. Sex cannot be determined for larval zebrafish.

Method details

Lateral line microphonic potential recordings

Recordings were made from fish 5–7 dpf, anesthetized at room temperature (~22°C) with tricaine (0.03 % ethyl 3-aminobenzoate methanesulfonate salt, SigmaAldrich) in a bath solution consisting of 120 mM NaCl, 2 mM KCl, 10 mM HEPES, 2 mM CaCl₂, and 0.7 mM NaH₂PO₄, adjusted to pH ~7.3. We examine L1-L4 neuromasts in the pLL. We selected these neuromasts for characterization because their location along the fish makes it possible to deliver a consistent stimulus in both directions along the A-P axis (Figure 1A,B). Larvae were restrained using strands of dental floss in a dish containing the bath solution. Heart rate and blood flow were visually monitored to assess larva viability before, during, and after experiments. The fish were visualized under an upright Olympus BX1WI microscope using a 4✕ 0.1 NA and a 100✕ 1 NA objective. All neuromast images were collected with a Grasshopper3 CMOS camera (Point Grey, Richmond, BC, Canada). jClamp software (SciSoft, Joseph Santos-Sacchi, Yale, New Haven, CT) was used to generate either a sinusoidal stimulus of 50 Hz or sustained deflections for 200 ms, delivered as a fluid jet from a borosilicate glass pipette. The stimulus pipette, with a diameter of approximately 7 μm, was placed about 50 μm away from the neuromast, parallel to the neuromast’s axis of best sensitivity. A HSPC-1 (ALA Scientific instruments, Farmingdale, NY) was used to control the fluid jet. The recording electrode was placed inside a borosilicate glass pipette of approximately 1 μm tip diameter, which was placed near the apical surface of the neuromast. Resistance of the recording pipette was 3–6 MΩ when filled with the bath solution. Due to the layout of the electrophysiology rig, the recording pipette cannot be placed on the same side of the neuromast as the stimulus pipette. Placement of pipettes was controlled by a set of micromanipulators (MPC-325; Sutter Instrument). Microphonic potentials were amplified using a PC-505B amplifier (Warner Instruments, Hamden, CT) in a current clamp mode and SIM983 scaling amplifier (Stanford Research, Sunnyvale, CA) and recorded using a PCI-6221 digitizer (National Instruments, Austin, Texas) and jClamp software. All sinusoidal stimulus-evoked recordings were low-pass filtered at 200 Hz, and each trace shown and used in analysis is an average of at least 500 deflections. Step-evoked recordings were low-pass filtered at 500 Hz, and each step trace is an average of at least 400 deflections. For each sample, stimulation was initiated either from the anterior or posterior, so that half of the initial stimuli were anterior, and the other half were posterior. For sinusoidal stimuli, magnitude was measured from peak to peak; whereas, for step-like deflections, the peak change of the potential was recorded. The average magnitude was calculated for each recording for both P-to-A sensitive and A-to-P sensitive hair cells.

Functional calcium imaging in zebrafish hair bundles

GCaMP6s-based calcium imaging of zebrafish hair bundles and of the afferent processes has been previously described in detail.^16,44 Briefly, individual 5–6 dpf larvae were first anesthetized with tricaine. To restrain larvae, they were pinned onto a Sylgard-filled recording chamber. To suppress the movement, alpha-bungarotoxin (125 μM, Tocris) was injected into the cavity of the heart. Larvae were then immersed in extracellular imaging solution (in mM: 140 NaCl, 2 KCl, 2 CaCl₂, 1 MgCl₂ and 10 HEPES, pH 7.3, OSM 310 +/− 10) without tricaine. A fluid jet similar to the one used for microphonics was used to mechanically stimulate the apical bundles of hair cells of the A-P neuromasts. To stimulate the two orientations of hair cells (A-to-P sensitive and P-to-A sensitive) a 500-ms ‘push’ was delivered. Larvae were rotated 180° to deliver a comparable ‘push’ stimulus to both the A-to-P sensitive and P-to-A sensitive cells. For all calcium imaging experiments, a saturating stimulus was applied (tips of the tallest part of the hair bundle, the kinocilia were deflected at least 4 μm). One exception is Figure S3. Here three stimulus intensities were applied to deflect the tips of the kinocilia 1, 2 and 4 μm. For each sample, stimulation was initiated either from the anterior or posterior, so that half of the initial stimuli were anterior, and the other half were posterior.

To image calcium-dependent mechanosensation in apical hair bundles or calcium-dependent activity in the afferent process, a Bruker swept-field confocal system was used. The system was equipped with a Rolera EM-C2 CCD camera (QImaging) and a Nikon CFI Fluor 60✕ 1.0 NA water immersion objective. Images were acquired in 5 planes along the Z-axis at 0.5 μm intervals (hair bundles) or 1 μm intervals (afferent processes), at a 50 Hz frame rate yielding a 10 Hz volume rate. The 5 plane Z-stacks were projected into one plane for image processing and quantification. The generation of the spatial ∆F heatmaps has been previously described.^29,44 For GCaMP6s measurements, a circular ROI with a ~1.5 μm (hair bundles) or 3 μm (afferent processes) diameter was placed on the center of each individual bundle or afferent hot spot. The mean intensity (∆F/F₀) within each ROI was quantified where F₀ represents the GCaMP6s intensity prior to stimulation. We examined the GCaMP6s signal in each hair bundle to determine its orientation. The GCaMP6s responses for each neuromast were averaged to quantify the magnitude of the A-to-P sensitive and P-to-A sensitive responses.

Immunohistochemistry and confocal imaging

Immunohistochemistry to label myosin7a, Emx2 and actin was performed on whole zebrafish larvae similar to previous work.²⁵ The following primary antibodies were used: rabbit anti-myosin7a (Proteus 25–6790; 1:500); mouse anti-Emx2 (Trans Genic KO609; 1:250); mouse anti-pan-MAGUK (IgG1) (Millipore MABN7; 1:500); and mouse anti-Ribeye b (IgG2a) (Sheets et al 2011; 1:10,000). The following secondary antibodies were used at 1:1000: (#A11010, #A21236, #A-11008, #A-21133, #A-21240, ThermoFisher Scientific) along with Alexa 488 Phalloidin (#A12379, ThermoFisher Scientific). Larvae were fixed with 4 % paraformaldehyde in PBS for 3.5 hr at 4 °C. For pre- and post-synaptic labeling, all wash, block and antibody solutions were prepared with 0.1% Tween in PBS (PBST). For Emx2 labeling, all wash, block and antibody solutions were prepared with PBS + 1% DMSO, 0.5% Triton-X100, 0.1% Tween-20 (PBDTT). After fixation, larvae were washed 5 × 5 min in PBST or PBDTT. For synaptic labeling, prior to block, larvae were permeabilized with acetone. For this permeabilization, larvae were washed for 5 min with H₂O. The H₂O was removed and replaced with ice-cold acetone and placed at −20°C for 5 min, followed by a 5 min H₂O wash. The larvae were then washed for 5 × 5 min in PBST. Larvae were then blocked overnight at 4°C in blocking solution (2% goat serum, 1% bovine serum albumin, 2% fish skin gelatin in PBST or PBDTT). Larvae were then incubated in primary antibodies in antibody solution (1% bovine serum albumin in PBST or PBDTT) overnight, nutating at 4°C. The next day, the larvae were washed for 5 × 5 min in PBST or PBDTT to remove the primary antibodies. Secondary antibodies in antibody solution were added and larvae were incubated for 2 hrs at room temperature. After 5 × 5 min washes min in PBST or PBDTT to remove the secondary antibodies, larvae were rinsed in H₂O and mounted in Prolong Gold (ThermoFisher Scientific). Fixed zebrafish samples were imaged on an inverted Zeiss LSM 780 laser-scanning confocal microscope with Airyscan (Carl Zeiss AG) using an 63✕ 1.4 NA oil objective lens.

Quantification of lateral line synapses

Myosin7a and phalloidin label were used in combination to link hair bundles to their respective cell base and synapses. Images containing synaptic labels were processed in ImageJ as previously described.⁴⁵ To qualify as a ribbon or postsynapse, the following minimum size filters were applied: Ribeye b: 0.025 μm², pan-Maguk: 0.04 μm². A complete synapse was comprised of both Ribeye b and MAGUK puncta. In each neuromast, synapses in four A-to-P and four P-to-A cells were examined for quantification.

Quantification and statistical analysis

Statistics and software

All statistical analyses were performed using GraphPad Prism 8. Data in text and graphs are reported as mean ± SEM. Statistical tests used to compare groups include Mann-Whitney test, Student’s t-test, Kruskal-Wallis test, or ANOVA with Holm-Sidak post hoc testing.

Highlights.

• Lateral line hair cells that function in opposition show asymmetric responses

• Tmc2a expression augments mechanotransduction to create asymmetric responses

• Emx2 is required to establish this functional asymmetry

• Asymmetric responses between hair cells are preserved in afferent processes