key: cord-0287370-orb4ay6n
authors: Liu, Zhikai; Kimura, Yukiko; Higashijima, Shin-ichi; Hildebrand, David G.; Morgan, Joshua L.; Holy, Timothy E.; Bagnall, Martha W.
title: Central vestibular tuning arises from patterned convergence of otolith afferents
date: 2020-02-15
journal: bioRxiv
DOI: 10.1101/2020.02.14.948356
sha: fa9702ab7eb787147ca9cb44545821255a99f830
doc_id: 287370
cord_uid: orb4ay6n

As sensory information moves through the brain, higher-order areas exhibit more complex tuning than lower areas. Though models predict this complexity is due to convergent inputs from neurons with diverse response properties, in most vertebrate systems convergence has only been inferred rather than tested directly. Here we measure sensory computations in zebrafish vestibular neurons across multiple axes in vivo. We establish that whole-cell physiological recordings reveal tuning of individual vestibular afferent inputs and their postsynaptic targets. An independent approach, serial section electron microscopy, supports the inferred connectivity. We find that afferents with similar or differing preferred directions converge on central vestibular neurons, conferring more simple or complex tuning, respectively. Our data also resolve a long-standing contradiction between anatomical and physiological analyses by revealing that sensory responses are produced by sparse but powerful inputs from vestibular afferents. Together these results provide a direct, quantifiable demonstration of feedforward input convergence in vivo.

Neurons compute information from many different synaptic inputs. A central 31 challenge in understanding neuronal circuits is determining how the tuning and 32

connectivity of these inputs affect the resulting computations. For example, neurons in 33 visual cortex exhibit simple or complex orientation tuning, which is thought to derive 34 from the convergence of presynaptic inputs with distinct tuning properties (Hubel and 35 Wiesel, 1962, Alonso and Martinez, 1998) . Computational models of such input-output 36 relationships have fundamentally shaped the way we think of information processing in 37 the brain (Felleman and Van Essen, 1991, LeCun et al., 2015) . However, these models 38

generally require assumptions about many parameters that can only be measured with 39

incompatible approaches: the tuning of the presynaptic population, input connectivity, 40

and synaptic strengths, as well as the activity of the postsynaptic neuron itself. Direct 41 measurements of these parameters simultaneously are prohibitively difficult in most 42 systems, making it hard to define neuronal computations in vivo. 43

from peripheral vestibular afferents (Boyle et al., 1992) and project to the spinal cord 45 (Boyle and Johanson, 2003) . Understanding the neuronal computations of VS neurons 46

would not only inform how vestibular sensory signals are processed in the brain, but 47 also provide a mechanistic view of sensorimotor transformation. VS neurons, like other 48 central vestibular neurons, produce diverse responses to head movement. During head 49 tilt or acceleration, some central vestibular neurons exhibit simple cosine-tuned 50

responses, similar to those of the afferents: the strongest activity is evoked by 51 movements in a preferred direction, with little or no response in the orthogonal direction. 52

In contrast, other central vestibular neurons exhibit more complex responses, including 53 bidirectional responses (Peterson, 1970) and spatiotemporally complex tuning (Angelaki 54 et al., 1993) . A vectorial model predicts that convergence of several simple cosine-55 tuned afferents can fully account for the response of either a simple or a complex 56 central vestibular neuron, depending on whether those afferents are similarly tuned or 57 differently tuned (Angelaki, 1992) . However, as in other systems, this model has been 58 technically challenging to test experimentally. 59

We chose to address this question in the small brain of the larval zebrafish. The 60

VS circuit was previously identified in the larval zebrafish as the homolog of mammals 61 (Kimmel et al., 1982) , which becomes functional as early as 3 days post fertilization 62

(dpf) (Mo et al., 2010) . The accessibility of the larval zebrafish brain for intracellular 63 recording from identified VS neurons allows us to investigate how central vestibular 64

neurons compute sensory signals in vertebrates. 65

Here we establish a novel approach to record sensory evoked responses in vivo 66 from VS neurons in the larval zebrafish. We find that individual afferents evoke large 67 amplitude-invariant excitatory postsynaptic currents (EPSCs), allowing us to separate 68 distinct afferent inputs that converge onto a given VS neuron. This provides a 69 mechanism to simultaneously measure the sensory tuning and synaptic strength of 70 each converging afferent, as well as the response of the postsynaptic neuron. We show Fig. 2A) . Stimulation evoked a synaptic current with two components. The first 117 component had fast kinetics with short latency (0.56 ± 0.28 ms, n=8), low jitter (0.05 118 ± 0.04 ms, n=8), and invariant EPSC amplitude (SD: 6.7±3.9%, normalized to peak) 119 across trials. In contrast, the second component had slower kinetics and variable 120 amplitudes (Fig. 2B) . We dissected the two components of evoked EPSCs 121

pharmacologically. Bath application of the gap junction blocker carbenoxolone (CBX, 122 500 µM) during afferent stimulation substantially reduced the first component of the 123 EPSC (Figs. 2C, E). In contrast, bath application of the AMPA receptor antagonist 124 NBQX (10 µM) abolished the second component of synaptic current (Fig. 2D ). 125

Furthermore, the fast EPSCs were not reversed by changing the holding potential ( Fig.  126 S2), a signature behavior of electrical synaptic transmission (Akrouh and 127

Kerschensteiner, 2013). Thus, the early and late components of afferent-evoked 128 synaptic currents are mediated by gap junctions and AMPA receptors, respectively. 129

Across VS neurons, the NBQX-sensitive currents accounted for 27.1±20.2% of total 130 charge transfer (n=7, Fig. 2F ), demonstrating that gap junctional current is the major 131 component mediating synaptic transmission. 132

To evaluate ultrastructural evidence for mixed synaptic transmission, we re-133 imaged existing serial ultrathin sections of a 5.5 dpf larval zebrafish (Hildebrand et al., 134 2017) at sufficiently high resolution (1-4 nm/px) to identify synaptic contacts between 135 myelinated utricular afferents and VS neurons, identified anatomically. We found both 136 tight junction structures (Fig. 2G ), and vesicles apposed to a postsynaptic density ( Fig.  137 2H) at appositions between utricular afferent and VS neurons, consistent with 138 anatomical evidence for mixed electrical / chemical transmission at this synapse in adult 139

fish (Korn et al., 1977) and rat (Nagy et al., 2013 correlogram of all EPSC event times in this example neuron did not display a refractory 157 period (Fig. 3C, top) . In contrast, an auto-correlogram within each EPSC cluster 158 exhibited a clear refractory period around 0 ms (Fig. 3C , bottom). Furthermore, cross-159 correlograms between EPSC clusters did not show this structure, consistent with the 160 notion that they arise from independent inputs (Fig. S3) . Accordingly, we can interpret 161 these three EPSC clusters as deriving from the activity of three distinct presynaptic 162 afferents (Fig. 3D, left) . Because of the high fidelity of electrical transmission, each 163

EPSC cluster effectively reads out the spiking of an individual afferent, allowing us to 164 measure presynaptic activity via postsynaptic recording (Fig. 3D, right) . 165

To test this interpretation of electrophysiological data with a completely 166 independent approach, we reconstructed the whole volume of myelinated utricular 167

inputs onto 11 VS neurons from a high resolution re-imaged serial section EM dataset 168 acquired from the right side of one 5.5 dpf larval zebrafish (Fig. 4A , B). We found that 169 the connection between myelinated utricular afferents and VS neurons was relatively 170

sparse. All VS neurons were contacted by at least two utricular afferents, but some 171 afferents did not innervate any VS neurons (Fig. 4C ). These reconstructions showed 172 that a range of 2-6 afferents (mean±std: 3.4±1.4) converged onto each VS neuron ( Fig.  173 4D). We compared these numbers to those derived from whole-cell physiology, where 174

we inferred the number of convergent afferents from the number of EPSC clusters. 175

Across all VS neuron recordings, we found a range of 0-5 afferents (1.7±1.3) converged 176 onto each VS neuron (Fig. 4E ). The result from anatomical reconstruction is largely 177 consistent with the overall distribution of afferent contacts as measured by whole-cell 178 physiology, presumably with some small-amplitude EPSCs elicited by the afferents not 179 successfully clustered. Therefore, these results demonstrate that synaptic inputs from 180

individual vestibular afferents can be separated by their stereotypic EPSC waveforms, 181

yielding inferred afferent convergence consistent with high-resolution anatomical 182

connectivity. 183

By recording from one VS neuron, we can infer the activity of its presynaptic 185

afferents. This approach thus offers a unique opportunity to measure the sensory tuning 186

of several convergent afferents simultaneously. To determine the spatial tuning of 187 convergent afferent inputs, we delivered 2 Hz, ±0.02 g sinusoidal translational stimuli 188 on four axes in the horizontal plane and recorded the sensory-evoked EPSCs, as 189

shown for an example VS neuron (Fig. 5A ). In this example neuron, the inferred 190 utricular afferent (EPSC cluster) with the largest synaptic amplitude responded best to 191 caudally-directed acceleration, while two others responded with varying sensitivities to 192

rostrally-directed acceleration, in all cases with phase leads relative to peak 193 acceleration (Fig. 5B ). With these measurements, we can derive the preferred tuning 194 direction, gain and phase of each afferent, as represented by the direction and length of 195 a vector (Fig. 5B , right). To validate the consistency of the vectorial representation, we 196 used a previously established approach (Schor et al., 1984) to quantify the tuning 197 vectors with separately measured responses to two circular stimuli ( Fig. S4 B) , which 198 afferent tuning vectors (Fig. 5C ). When fish were oriented dorsal-up, the axes tested 203

were rostral-caudal and ipsilateral-contralateral (motion along an axis from one ear to 204 the other). In this position, most afferents were strongly tuned to acceleration towards 205 the contralateral direction (31/60), some exhibited preferential tuning to the acceleration 206

to the rostral (4/60) and caudal (20/60) directions, and only 5/60 afferents were tuned to 207 the ipsilateral direction (Fig. 5D) . These results showed that each afferent in the larval 208 zebrafish exhibits selective responses to different translational stimuli. Afferents overall 209 responded best to acceleration towards the contralateral, rostral and caudal directions, 210

which correspond to ipsilateral, nose-up and nose-down tilts in postural change 211 (Angelaki and Cullen, 2008) , consistent with the distribution of hair cell polarity in the 212 utricular macula (Haddon et al., 1999) . 213

The sensitivity and phase of vestibular afferents varies for motion at different 215

frequencies (Fernandez and Goldberg, 1976b) . The tuning of otolith afferents ranges 216

from typically more jerk-encoding (derivative of acceleration) at low frequencies to more 217

acceleration-encoding at high frequencies. What temporal tuning profile do afferents in 218 larval zebrafish exhibit? We applied translational stimuli with different frequencies (0.5-8 219

Hz, ±0.02 g) on the rostral-caudal axis. In the example neuron, all three inferred otolith 220

afferents showed similar tuning, with progressively stronger responses with increasing 221 frequencies of stimulation (Fig. 6A ). Across group data acquired at both ±0.02 g and 222

±0.06 g, the average tuning gain increased 3.3-fold (0.02 g) and 2.3-fold (0.06 g) from 223 0.5 Hz to 8 Hz (Fig. 6B ). Most afferents (39/48) showed at least 2-fold increase from 0.5 224

Hz to 4 Hz in tuning gain at either 0.02 g or 0.06 g. Only one afferent had relatively flat 225 gain (< 50% increase) at both 0.02 g and 0.06 g, and its tuning was overall weak (mean 226 gain: 1.88 and 2.24 EPSC/s respectively), suggesting it was less sensitive or not tuned 227 on the rostral-caudal axis. Regardless of tuning direction (rostral: 44%, 21/48; caudal: 228 56%, 27/48), afferents exhibited a phase lead relative to peak acceleration at various 229 tested stimulus magnitudes and frequencies ( (Fig. 6C, S5 ). On average, the phase lead 230 at low frequency (0.5 Hz) was 84.0° for 0.02 g and 78.6° for 0.06 g. At high frequency (8 231

Hz), the phase lead was reduced to 33.6° for 0.02 g and 39.3° for 0.06 g. The temporal 232 dynamics of the afferents resembled those of previously reported irregular units 233 (Goldberg et al., 1990) , with low spontaneous firing rates (10.28±9.1 EPSC/s) and 234

larger coefficients of variation (CV). The average CV across inferred afferents was 235 0.97±0.24, and the smallest CV was 0.5 (Fig. S6) , indicating that no regular-firing otolith 236 afferents were detected synapsing onto VS neurons. We conclude that the otolith 237 afferents act as a high-pass filter, encoding a mixture of acceleration and jerk, similar to 238 otolith afferents in primates (Laurens et al., 2017) . 239 Preferential convergence 240 Figure 6 : Temporal tuning of inferred otolith afferents A. Sensory tuning of afferent inputs to one VS neuron during translational movement at 5 different frequencies in the rostral(+)-caudal(-) axis. Left, EPSC waveforms of three different clusters recorded from one VS neuron. Right, temporal tuning profile of each EPSC cluster on the rostral-caudal axis. B. Gains of inferred afferents across different frequencies of translational acceleration. Gray, individual afferents; colored, afferents from A; black, mean and standard deviation of gains from all afferents (0.02 g, 48 afferents; 0.06 g: 46 afferents; 25 neurons, 20 fish) C. Phases of inferred afferents across frequencies, relative to sinusoidal stimulus. 180°( 0.5 cycle in A) represents the peak of acceleration towards rostral direction; 360°r epresents the peak of acceleration towards caudal direction (0 or 1 cycle in A). Data were thresholded to only include afferents whose gain was > 5 EPSC/s (0.02 g, 36 afferents; 0.06 g, 38 afferents; 25 neurons, 20 fish) Individual VS neurons can receive inputs from afferents with similar (Fig. 6A) or 241 different tuning (Fig. 5B) . Is afferent tuning convergence random or structured? The 242

responses of inferred afferents that converge onto the same VS neuron were 243

represented by their tuning vectors (Fig. 7A) . The angle between the vectors indicates 244 the similarity of convergent inputs, with a small angle for a VS neuron with similarly 245 tuned inputs and a large angle for a VS neuron with differently tuned inputs. From 43 246

VS neurons recorded in the side-up orientation, 60% (38/63) of converging afferent 247 pairs had small angles (<45°) and 27% (17/63) had large angles (>135°). Compared to 248 a random pairing angle distribution generated by bootstrapping, the percentage of 249 similarly tuned convergent afferent pairs was significantly higher than chance (Fig. 7B,  250 left). From 36 VS neurons recorded in the dorsal-up orientation, there were 71% (37/52) 251

of inferred afferent pairs with a converging angle smaller than 45°, and only 2% (1/52) 252 with a converging angle larger than 135° due to the small number of ipsilaterally tuned 253 afferents (Fig. 5D) . Nonetheless, the probability of similarly tuned afferent convergence 254 (<45°) was significantly higher than that chance (Fig. 7B, right) . For afferent pairs with 255 converging angle larger than 45° (45°-90°, 90°-135°, 135°-180°), their probabilities was 256 slightly lower than their respective estimated distribution by bootstrapping. Accordingly, 257 on a given body axis (R-C or I-C), convergent afferents are also more likely to encode 258 similar tuning directions (Fig. 7C) . These results suggest that afferents with similar 259 tuning direction preferentially converge at rates exceeding what would be expected by 260 random connectivity. , for all non-spiking VS neurons with multiple convergent afferents. Sensory tuning of afferent inputs and EPSPs was measured on the R-C axis (black, n=27) and I-C axis (grey, n=5). Dashed, unity line. Pearson's R: 0.67 (Peterson, 1970) and broadly tuned sensory responses (Angelaki, 1992) of central 284 vestibular neurons can be computationally reconstructed from multiple modelled cosine-285 tuned inputs. However, directly measuring these inputs has been technically difficult, 286

and it is unclear whether such models can sufficiently explain the activity of central 287

neurons. Therefore, we took advantage of the inferred afferent spiking to examine 288

whether the tuning of VS neurons can be constructed from the convergence of otolith 289

afferents. 290

We observed that different VS neurons showed simple or complex membrane 291 potential responses to translational stimuli on the rostral-caudal axis. An example 292

simple cell was only depolarized during a specific phase of acceleration (Fig. 8A ), 293

whereas a complex cell exhibited multiple depolarized periods during the stimulus (Fig.  294 8C). Next, we measured the EPSC tuning in the same VS neurons. In the example 295 simple neuron, sensory evoked EPSCs exhibit three distinct amplitudes (Fig. 8B) , 296

indicating three afferents converge onto the cell. These three afferents showed similar 297 tuning to each other, with strongest responses for rostrally-directed acceleration. In 298 contrast, the four inferred afferents that converge onto the example complex cell 299 exhibited a different tuning pattern. Two afferents were tuned to rostrally-directed 300 acceleration and the other two to caudally-directed acceleration (Fig. 8D) . 301

To examine this relationship across the population, we defined an afferent inputs 302 similarity index for multiply innervated VS neurons, based on the phase of afferent 303

inputs and their EPSC amplitudes. The index ranges from 0-1, with smaller index 304

representing more divergent ESPC input tuning and larger index representing more 305 similar tuning (see Methods). A classifier originally developed for visual cortical neurons 306 was used to quantify the tuning complexity of the postsynaptic neuron's membrane 307 potential responses to sensory stimuli (Skottun et al., 1991) . In this metric, neurons with 308 simple tuning show large AC and small DC responses, whereas complex cells exhibit 309

small AC and large DC responses (Fig. 8E) . We found that the AC/DC ratio of the 310 membrane potential was strongly correlated with the similarity index of afferent inputs 311 (Fig. 8F) . In other words, convergence of more similarly tuned afferents yields a more 312 simple VS neuron response, and the convergence of more differently tuned afferents 313

generates a more complex postsynaptic response. 314

We next extended the comparison of presynaptic to postsynaptic tuning by 316 measuring the spiking responses of VS neurons during sensory stimulation. In a subset 317 of VS neurons, the largest translational stimuli that we could deliver while holding the 318 cell was sufficient to evoke postsynaptic firing; in other neurons, a small bias current 319 was injected to evoke spiking during sensory stimulation (see Methods). Most VS 320

neurons exhibited simple spike tuning, and received convergent inputs from similarly 321 tuned afferents (Fig. 9A) . Some VS neurons with simple spike responses received 322 convergent inputs from differently tuned afferents (Fig. 9B) . Finally, complex spike 323 tuning in VS neurons was always generated by inputs from differently tuned afferents 324 (Fig. 9C) . These three categories of input-output transformation (similar to simple, 325 different to simple, different to complex) were identified across all recordings from VS 326 neurons (Fig. 9D ). In total, most recordings (R-C, 6/13; I-C, 14/19) exhibited simple 327 spike tuning, of which 69% (20/29) received inputs from similarly tuned afferents, and 328 31% (9/29) received inputs from differently tuned afferents (Fig. 9E) . Thus, convergence 329 of similarly tuned afferents yields simple spike tuning, but convergence of differently 330 tuned afferents can yield either simple or complex spike tuning. Interestingly, complex 331

spike tuning was only observed on the R-C axis (3/13 recordings), which subserves 332 pitch movements, not the I-C axis (0/19), which subserves roll (Fig. 9F) Similarly tuned otolith afferents preferentially converge onto VS neurons (Fig. 7) , 357

demonstrating that feedforward excitation can generate central neurons with simple 358 response properties. In a similar vein, thalamocortical inputs with similar angular tuning 359 also preferentially project onto the same site in somatosensory cortex, and the preferred 360 tuning direction of the cortical neuron can be predicted by that of the presynaptic 361 thalamic neuron (Bruno et al., 2003) . Furthermore, we found that convergence of 362 differently tuned afferents can yield a more complex postsynaptic response in central 363 vestibular neurons, similar to bidirectional or complex tuning observed previously in cats 364 (Peterson, 1970) and primates (Angelaki and Dickman, 2000) . This result generally 365 supports the hypothesized model (Angelaki, 1992 ) that the tuning of central vestibular 366

neurons can be constructed from cosine tuned inputs with varying tuning properties. 367

However, we find that convergence of differently tuned afferents can also yield simple 368 tuning in VS neurons (Fig. 8B) , suggesting other factors such as inhibition (Straka and 369 Dieringer, 1996) and thresholding (Priebe et al., 2004 ) might be involved. We found no 370 evidence for polysynaptic excitatory circuits during afferent stimulation (Fig. S9 A and  371 B), and modelling indicates that excitatory synaptic input is sufficient to predict 372 subthreshold membrane potential and tuning (Fig. S9 C-F) . However, stronger stimuli 373 might elicit inhibition and other nonlinearities. Across brain regions, sensory tuning of 374 central neurons is constructed by a variety of mechanisms. These include afferent 375 convergence pattern (Alonso and Martinez, 1998, Priebe and Ferster, 2012), local 376 excitatory or inhibitory modulation (Wilent and Contreras, 2005) , and nonlinear dendritic 377

computation (Lavzin et al., 2012) . Our results demonstrate that sensory response of a 378 central neuron can be constructed from the afferent inputs in a direct feedforward 379 manner. 380

The derived spatial tuning profile of afferents in the larval zebrafish is similar to 382 the polarity of the hair cells in otolith macula, consistent with results in fish (Fay, 1984 , 383 Platt, 1977 and primates (Fernandez et al., 1972) . Notably, tuning to dorsal or ventral 384 acceleration was relatively weak for most afferents, presumably due to the horizontal 385 orientation of the utricular membrane in larval zebrafish inner ear. Afferents were 386

preferentially tuned to contralateral acceleration (ipsilateral tilt) in the roll axis, consistent 387

with the dearth of ipsilaterally tuned hair cells in larval zebrafish (Haddon et al., 1999) . 388

In species with more centrally located line of polarity reversal (Fernandez and Goldberg, 389 1976a, Tomko et al., 1981) , we would predict more convergence of oppositely tuned 390 afferents, and correspondingly more complex response of VS neurons in the roll axis, 391

as seen in cats (Peterson, 1970) . 392 A significant question in vestibular systems is whether central vestibular neurons 393 receive selective projections from afferents with regular as opposed to irregular firing. 394

Both regular and irregular afferents are thought to converge on VS and vestibulo-ocular 395 reflex neurons in mammals, based on studies comparing recruitment thresholds of 396 afferent inputs (Boyle et al., 1992) . Our data provide direct evidence that vestibular 397 inputs to VS neurons exhibit classic characteristics of irregular afferents (Eatock and 398

Songer, 2011): high-pass tuning, low spontaneous firing rate, and high CV of firing (Fig.  399 S6). It is unknown whether regular utricular afferents exist in the larval zebrafish. 400

Although regular utricular afferents were observed in guitarfish (Budelli and Macadar, 401 1979), they appear absent in toadfish (Maruska and Mensinger, 2015) and sleeper goby 402 (Lu et al., 2004) . Based on serial section EM, many afferents make no contacts with VS 403 neurons (Fig. 4C) , leading us to conclude that either regular afferents have not yet 404 developed or that they do not contact VS neurons in the larval zebrafish. synapse may be a conserved mechanism across species to implement fast, frequency-420 independent transmission in the lateral vestibular nucleus. The amplitude invariance of 421 this connection allowed us to examine whether there was any relationship between an 422 afferent's sensory gain or firing rate and the synaptic amplitude it evokes in a VS 423

neuron. No correlation appeared in either of these measures (Fig. S8) , indicating that at 424

least within this population, synapse size is not "normalized" by firing rate. 425

The swim bouts to regain balance (Ehrlich and Schoppik, 2017) . 440

The high-pass tuning and phase lead of otolith afferents innervating VS neurons 441 will make larvae most sensitive to ongoing changes in tilt or acceleration, especially at 442 high frequency. These data are consistent with behavioral observations that larvae 443 become more likely to swim to correct their position in the pitch axis when angular 444 velocity (i.e., changing tilt) reaches a critical threshold (Ehrlich and Schoppik, 2017 Animals were raised and maintained in the Washington University Zebrafish Facility at 475

28.5ºC with a 14:10 light:dark cycle. Larval zebrafish (4-7 dpf) were housed either in 476 petri dishes or shallow tank with system water. Adult animals were maintained up to 1 477

year old with standard procedure. larvae (4-7 dpf) were paralyzed by 0.1% -bungarotoxin and embedded in a 10 mm 482

FluoroDish ( axes) and a series of frequency-varying sinusoidal translational stimulus was applied. 517

The stimulus amplitude was set at 0.02 g or 0.06 g (min to max: 0.04 g or 0.12 g 518 respectively), and stimulus frequency range was 0. WaferMapper (Hayworth et al., 2014) . The resulting images were aligned onto the 18.8 541 nm/px dataset using linear affine transformations in FIJI with the TrakEM2 plug-in 542 (Cardona et al., 2012) and will be freely available after publication. In a small subset of 543 identified synapses, we carried out further re-imaging at 1nm/px to visualize the 544 hallmarks of gap junctions. 545

The existing tracings of VS neurons and utricular afferents were extended to cover 546 branches that had been missed or untraced in the original dataset. Afferent/VS neuron 547

appositions were considered to be synaptic contacts if the presynapse contained 548

vesicles, the membranes were tightly apposed and straight, and there were signs of a 549 postsynaptic density. In cases where appositions were more difficult to determine, such 550

as those parallel to the plane of section, vesicle clustering at a tight apposition was used 551

as the criterion for a synapse. 552

All analysis are implemented in Matlab (Mathworks). 554

Event detection: 555 EPSC events were detected by a derivative method (Bagnall and McLean, 2014) . 556

Tuning index of all EPSCs was calculated as the vectoral sum of all events' phase, 557

weighted by the EPSC amplitude. 558 = | ∑ 1 * 3 * 4 5 | ∑ 1 , = √−1 559

( 1 is the amplitude of each EPSC event , and 1 is the phase of that event relative to 560 the sinusoidal stimulus on each axis.) 561

We assumed that the signals we observed on voltage clamp were majorly composed of 563 electrical EPSCs and chemical EPSCs from afferents, based on our observation from 564 the pharmacology data. Clusters with refractory period (threshold: probability < 0.003 within 1 ms) in auto-585 correlograms (100 ms) were considered from an individual afferent. 586

For each cluster, the tuning vector of inferred afferent on each axis was quantified as: respectively.) 592

Tuning in four axes was fitted into a 2-dimensional spatiotemporal model (Angelaki, 593 1992) to obtain the maximum tuning direction, the tuning gain and phase in that 594 direction. 595

Afferent inputs similarity index for a VS neuron was determined as: 596

are the average EPSC amplitude, tuning vector and number of events for 598

cluster .) 599 AC/DC response quantification 600 AC of membrane potential and spiking response were defined as the amplitude of 601 sinusoidal fit (2 Hz) of the membrane potential, and the spike vectorial sum during 602 sensory stimulation, respectively. DC of membrane potential and spiking response were 603 defined as the average membrane potential during sensory stimulation above baseline 604 (no stimulation), and the total spike number during sensory stimulation above baseline. 605

For spiking responses, VS neurons with firing rate > 4 spike/cycle and spike AC or 606 DC >1 spike/cycle were included in the analysis. 607

Bootstrapping: e S f V g afferent pairs were counted for VS neuron with 3 distinct afferent 608

inputs ( 3 ≥ 2). The same number of total afferent pairs ∑ e S f V g i 3jk from all VS neurons 609

was randomly selected among all ∑ 3 i 3jk inferred afferents to determine the 610 convergence angle or phase difference distribution by chance, and such selection was 611

performed 5000 times to calculate mean and standard deviation. 612

Richard Roberts for helping set up the electrophysiology recording rig

Bello Rojas for thoughtful 615 critiques of the paper. We are grateful to Drs. Daniel Kerschensteiner and David 616 Schoppik for insightful comments on the manuscript

Center 618 for Cellular Imaging (WUCCI) for supporting the confocal imaging experiments. This 619 work is supported by funding through the National Institute of Health

621 and the National BioResource Project in Japan (S.H.). M.W.B. is a Pew Biomedical 622 Scholar and a McKnight Foundation Scholar

Author contributions

UAS:GFP) fish line. Z.L and M.B conceived 626 the project. Z.L performed the electrophysiology, confocal imaging experiments and 627 analyzed the data. T.H helped develop the deconvolution algorithm for sorting EPSC 628 events

B wrote the manuscript with input from all other authors

Intersecting Circuits Generate Precisely Patterned 636 Retinal Waves

Functional connectivity between simple cells and 638 complex cells in cat striate cortex

SPATIOTEMPORAL CONVERGENCE (STC) IN OTOLITH NEURONS. 640 Biological Cybernetics

2-DIMENSIONAL SPATIOTEMPORAL 642 CODING OF LINEAR ACCELERATION IN VESTIBULAR NUCLEI NEURONS

Vestibular system: The many facets of a multimodal 645 sense

Spatiotemporal processing of linear acceleration: 647 Primary afferent and central vestibular neuron responses

The contribution of single 650 synapses to sensory representation in vivo

Frequency-independent 652 synaptic transmission supports a linear vestibular behavior

Modular organization of axial microcircuits in zebrafish

A Fast Iterative Shrinkage-Thresholding Algorithm for Linear 656 Inverse Problems

Inputs from regularly and irregularly 658 discharging vestibular nerve afferents to secondary neurons in squirrel monkey 659 vestibular nuclei. III. Correlation with vestibulospinal and vestibuloocular output 660 pathways

Morphological properties of vestibulospinal neurons in 662 primates

Thalamocortical angular tuning 664 domains within individual barrels of rat somatosensory cortex

STATO-ACOUSTIC PROPERTIES OF UTRICULAR AFFERENTS

TrakEM2 Software for 670 Neural Circuit Reconstruction. Plos One

Synaptic diversity enables temporal coding of coincident multisensory inputs in single 673 neurons

A Fully Automated Approach 676 to Spike Sorting

Vestibular processing during natural self-motion: implications for 678 perception and action

Vestibular hair cells and afferents: two channels for head 680 motion signals

Control of Movement Initiation Underlies the 682 Development of Balance

A primal role for the vestibular sense in the development 684 of coordinated locomotion

Cellular-Resolution Imaging of Vestibular Processing across the Larval Zebrafish 687 Brain

THE GOLDFISH EAR CODES THE AXIS OF ACOUSTIC PARTICLE MOTION IN 3 689 DIMENSIONS

Distributed Hierarchical Processing in the Primate 691

PHYSIOLOGY OF PERIPHERAL NEURONS 693 INNERVATING OTOLITH ORGANS OF SQUIRREL-MONKEY .1. RESPONSE TO STATIC TILTS 694 AND TO LONG-DURATION CENTRIFUGAL FORCE

PHYSIOLOGY OF PERIPHERAL NEURONS 697 INNERVATING OTOLITH ORGANS OF SQUIRREL-MONKEY .3. RESPONSE DYNAMICS

RESPONSE TO STATIC TILTS OF 700 PERIPHERAL NEURONS INNERVATING OTOLITH ORGANS OF SQUIRREL-MONKEY

THE VESTIBULAR NERVE 703 OF THE CHINCHILLA .4. DISCHARGE PROPERTIES OF UTRICULAR AFFERENTS

Hair cells without supporting cells: further studies in the ear of the zebrafish mind bomb 707 mutant

Imaging ATUM ultrathin section libraries with WaferMapper: a 710 multi-scale approach to EM reconstruction of neural circuits

Whole-brain serial-section electron microscopy in larval zebrafish

RECEPTIVE FIELDS, BINOCULAR INTERACTION AND 719 FUNCTIONAL ARCHITECTURE IN CATS VISUAL CORTEX

Dendritic organization of sensory 722 input to cortical neurons in vivo

Brain neurons which project to the 724 spinal cord in young larvae of the zebrafish

Efficient generation of knock-727 in transgenic zebrafish carrying reporter/driver genes by CRISPR/Cas9-mediated genome 728 engineering

Graded co-expression of ion channel, neurofilament, and synaptic genes in 731 fast-spiking vestibular nucleus neurons

The lateral vestibular nucleus of the toadfish 733 Opsanus tau: Ultrastructural and electrophysiological observations with special 734 reference to electrotonic transmission

736 Transformation of spatiotemporal dynamics in the macaque vestibular system from 737

Nonlinear dendritic 739 processing determines angular tuning of barrel cortex neurons in vivo

Deep learning

Coding of acoustic particle motion by utricular fibers in 743 the sleeper goby, Dormitator latifrons

Directional sound sensitivity in utricular afferents in 746 the toadfish Opsanus tau

748 Implementation of linear sensory signaling via multiple coordinated mechanisms at 749 central vestibular nerve synapses

Spinal Interneurons Differentiate Sequentially from Those 751 Driving the Fastest Swimming Movements in Larval Zebrafish to Those Driving the 752 Slowest Ones

Whole-Brain Calcium Imaging during 755 Physiological Vestibular Stimulation in Larval Zebrafish

Quantification of vestibular-757 induced eye movements in zebrafish larvae

Morphologically mixed chemical-759 electrical synapses formed by primary afferents in rodent vestibular nuclei as revealed 760 by immunofluorescence detection of connexin36 and vesicular glutamate transporter-1

CENTRAL DISTRIBUTION OF CERVICAL PRIMARY 763 AFFERENTS IN THE RAT, WITH EMPHASIS ON PROPRIOCEPTIVE PROJECTIONS TO 764 VESTIBULAR, PERIHYPOGLOSSAL, AND UPPER THORACIC SPINAL NUCLEI

Central projections of the vestibular nerve: a review 767 and single fiber study in the Mongolian gerbil

ACTIVITY OF VESTIBULOSPINAL NEURONS DURING LOCOMOTION

The functional organization of the barrel cortex

Distribution of neural responses to tilting within vestibular nuclei of the 772 cat

HAIR CELL DISTRIBUTION AND ORIENTATION IN GOLDFISH OTOLITH ORGANS

Mechanisms of Neuronal Computation in Mammalian Visual 776 Cortex

The contribution of spike 778 threshold to the dichotomy of cortical simple and complex cells

Development of utricular otoliths, but not saccular 781 otoliths, is necessary for vestibular function and survival in zebrafish

Impair Vestibular Circuit Formation in Zebrafish

Comprehensive mapping of whisker-evoked 786 responses reveals broad, sharply tuned thalamocortical input to layer 4 of barrel cortex

Input-output relations of Deiters' lateral vestibulospinal neurons with 789 different structures of the brain

Central Vestibular Neurons Project Asymmetrically to Extraocular Motoneuron Pools

Responses to head tilt in cat central 795 vestibular neurons. I. Direction of maximum sensitivity

Thalamic relays and cortical functioning

Classifying simple and complex cells on the basis of response modulation

Vesicular glutamate transport at a central synapse limits 803 the acuity of visual perception in zebrafish

Uncrossed disynaptic inhibition of second-order vestibular 805 neurons and its interaction with monosynaptic excitation from vestibular nerve afferent 806 fibers in the frog

RESPONSES TO HEAD TILT IN CAT 8TH NERVE 808 AFFERENTS

Synaptotagmin 7 confers frequency 810 invariance onto specialized depressing synapses

Specificity and strength of retinogeniculate 812 connections

Automated Reconstruction of 814 Three-Dimensional Fish Motion, Forces, and Torques

Dynamics of excitation and inhibition underlying 816 stimulus selectivity in rat somatosensory cortex

Vestibulospinal contributions to mammalian locomotion. 818 Current Opinion in Physiology