Shawn D. Newlands, Min Wei, David Morgan, and Hongge Luan Department of Otolaryngology, University of Rochester Medical Center, Rochester, New York

Transcription

1 J Neurophysiol 116: , 216. First published August 3, 216; doi:1.1152/jn Responses of non-eye-movement central vestibular neurons to sinusoidal yaw rotation in compensated macaques after unilateral semicircular canal plugging Shawn D. Newlands, Min Wei, David Morgan, and Hongge Luan Department of Otolaryngology, University of Rochester Medical Center, Rochester, New York Submitted 1 March 216; accepted in final form 27 July 216 Newlands SD, Wei M, Morgan D, Luan H. Responses of noneye-movement central vestibular neurons to sinusoidal yaw rotation in compensated macaques after unilateral semicircular canal plugging. J Neurophysiol 116: , 216. First published August 3, 216; doi:1.1152/jn After vestibular labyrinth injury, behavioral measures of vestibular performance recover to variable degrees (vestibular compensation). Central neuronal responses after unilateral labyrinthectomy (UL), which eliminates both afferent resting activity and sensitivity to movement, have been well-studied. However, unilateral semicircular canal plugging (UCP), which attenuates angular-velocity detection while leaving afferent resting activity intact, has not been extensively studied. The current study reports response properties of yaw-sensitive non-eye-movement rhesus macaque vestibular neurons after compensation from UCP. The responses at a series of frequencies (.1 2 Hz) and peak velocities (15 21 /s) were compared between neurons recorded before and at least 6 wk after UCP. The gain (sp/s/ /s) of central type I neurons (responding to ipsilateral yaw rotation) on the side of UCP was reduced relative to normal controls at.5 Hz, /s [8.3 (SD) normal, ipsilesion; 4.2 contralesion]. Type II neurons (responding to contralateral yaw rotation) after UCP have reduced gain (.27 normal, ipsilesion; contralesion). The difference between responses after UCP and after UL is primarily the distribution of type I and type II neurons in the vestibular nuclei (type I neurons comprise 66% in vestibular nuclei normally; 51% ipsilesion UCP; 59% contralesion UCP; 38% ipsilesion UL; 65% contralesion UL) and the magnitude of the responses of type II neurons ipsilateral to the lesion. These differences suggest that the need to compensate for unilateral loss of resting vestibular nerve activity after UL necessitates a different strategy for recovery of dynamic vestibular responses compared to after UCP. brainstem; threshold; vestibular nucleus NEW & NOTEWORTHY This is the first manuscript to describe the physiology of central vestibular neurons after occlusion of the semicircular canals unilaterally (unilateral canal plugging) and highlights similarities and differences between the physiology underlying compensation after canal plugging and after unilateral vestibular labyrinth ablation. Labyrinth ablation disrupts vestibular processing at rest and with movement, whereas canal plugging primarily impacts function during motion. These findings are important since both canal plugging and labyrinth ablation are techniques used clinically. VESTIBULAR COMPENSATION, the recovery of behavioral deficits over time following vestibular lesion, has been extensively Address for reprint requests and other correspondence: S. D. Newlands, Dept. of Otolaryngology, Univ. of Rochester Medical Center, 1 Elmwood Ave., Box 629, Rochester, NY ( shawn_newlands@urmc.rochester.edu). studied after unilateral destruction of the vestibular labyrinth or nerve. Unilateral labyrinthectomy (UL) involves the removal or destruction of all of the neuroepithelia of the labyrinth on one side. In this model of neural plasticity, deficits seen after lesion are attributed to either the loss of tonic drive from the spontaneously active vestibular nerve (static signs and symptoms such as spontaneous nystagmus and vertigo that are present at rest) or to the loss of modulation of neural drive with head movement (resulting in deficits of vestibular reflexes). Recovery from static deficits is generally more complete and rapid, whereas recovery from dynamic deficits is incomplete and slower (Fetter and Zee 1988; Smith and Curthoys 1989). Alternative models of vestibular loss have clinical importance, as surgical occlusion of the semicircular canals and application of the ototoxic aminoglycoside gentamicin to the middle ear have emerged as treatments for vestibular maladies (Crane et al. 28; Nedzelski et al. 1992; Parnes and McClure 199). Occlusion of the lumen of the semicircular canals (semicircular canal plugging) has been used to attenuate preferentially the modulation of neural drive with movement while retaining spontaneous activity in the vestibular nerve (Goldberg and Fernández 1975; Paige 1983; Rabbitt et al. 1999; Sadeghi et al. 29). In canal plugging, afferent activity from the otolith organs both at rest and with motion are preserved (Goldberg and Fernández 1975), and the endolymphatic and perilymphatic movement driving deflection of the ampullary cupula is reduced (Paige 1983; Rabbitt et al. 1999). Bilateral canal plugging reduces the dominant time constant of the vestibuloocular reflex (VOR), specific for the plane of the canal that is plugged, which results in severe attenuation of the spatially appropriate (canal plane) VOR response for lower frequencies (Böhmer et al. 1985; Yakushin et al. 1998). In rhesus macaques, the attenuation is 9% for frequencies 2 Hz (Yakushin et al. 1998). A subject who has undergone UL must overcome both the loss of vestibular afferent drive to one side and the loss of half of the dynamic information during rotation. In contrast, with unilateral semicircular canal plugging (UCP), the subject only needs to overcome loss of half of the dynamic information during rotation. Thus comparison of neuronal responses during motion after UCP, a unilateral lesion that preserves the resting activity of the vestibular afferents, to responses after UL should reveal the interactions between the static and dynamic components of compensation (Newlands et al. 25). The hypothesis tested in this study is that dynamic responses in central neurons after compensation following UCP are more similar to normal dynamic responses than following compensation from UL. In particular, we expected that ipsilesion type I neuron activity in compensated animals would have the same /16 Copyright 216 the American Physiological Society 1871

2 1872 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING gains as type I neurons before UCP. This prediction was predicated on the premise that recovery of the resting rate of ipsilesion UL neurons, which likely involves the upregulation of excitatory inputs and/or the downregulation of inhibitory inputs, competes with recovery of dynamic vestibular reflexes, which is mediated through the inhibitory vestibular nuclei commissural system (Shimazu and Precht 1966). We also hypothesized that the asymmetry in the number of ipsilesion type I to ipsilesion type II cells found after UL would not be found after UCP since that asymmetry reflects loss of primary afferent drive and the inability of the process of compensation to restore such drive fully (Newlands and Wei 213a). Thus the goals of the current study are to understand better both the degree to which macaques compensate after UCP and the interactions between dynamic and static compensation after UL. Several studies have investigated the impact of UL on central vestibular neurons (Heskin-Sweezie et al. 27; Newlands and Perachio 199a,b; Newlands and Wei 213a; Precht et al. 1966; Ris and Godaux 1998; Sadeghi et al. 21, 211; Smith and Curthoys 1988a,b), however, there is far less information about the responses of central neurons after unilateral semicircular canal plugging (Abend 1977, 1978; Farrow and Broussard 23). In the current study, we investigated the responses of central vestibular neurons to yaw rotation after unilateral occlusion of all three semicircular canals on the left side in rhesus macaques. The study uses the same rotational protocols as a previous study on the effect of UL on central vestibular responses from the same laboratory (Newlands and Wei 213a). Thus we compare responses of central vestibular neurons after unilateral three-canal plugging (UCP) to both prelesion data in the same animals and to the responses after UL collected in other animals (Newlands and Wei 213a). We found that the level of modulation measured 6 wk after lesion was similarly reduced for type I neurons on the lesion side (those responding in phase with ipsilateral rotation) after either UL or UCP. In the earlier study, we found a major difference in the proportion of ipsilaterally vs. contralaterally sensitive neurons between the vestibular nuclei after UL, resulting from a decrease in the proportion of type I (ipsilateral rotation sensitive) on the side of the lesion (Newlands and Wei 213a). We did not find the same relative loss of type I neurons following UCP. The dynamic responses of type II neurons (those responding in phase with contralateral rotation) on the side of the lesion were reduced after UCP, in contrast to our findings after UL. METHODS Animal preparation. Data were collected from one female (Bl) and two male (To and St) rhesus macaque monkeys (Macaca mulatta) with weights ranging from 4. to 6.3 kg. The initial surgical procedures, including placement of recording chambers and eye coils, have been detailed in earlier reports (Newlands et al. 29; Newlands and Wei 213a,b). Subsequent to collection of the control data, the animals underwent unilateral (left, all 3 canals) semicircular canal plugging under isoflurane anesthesia in sterile conditions, similar to the technique previously reported (Newlands et al. 1999). To accomplish this procedure, the left mastoid cavity was exposed via a postauricular incision. A simple mastoidectomy was performed using an electric drill and curette, exposing the otic capsule. All three semicircular canals were exposed. Exposure of the superior (anterior) semicircular canal required gentle retraction of the portion of the paraflocculus that extends through the arc of the superior canal in this species. All three canals are then fenestrated with a 2-mm diamond burr to expose the perilymph. Bone dust and chips are used to pack the canal lumen. The canal plugs were covered with muscle fascia, the wound was closed with subcutaneous interrupted absorbable sutures, and the skin was closed with permanent monofilament sutures that were removed in 1 2 wk. After surgery, the animals were treated with postoperative antibiotics for 7 days and intramuscular buprenorphine (.1 mg/kg) for pain control twice a day for 3 days. All surgical procedures were performed according to institutional and National Institutes of Health guidelines and under a protocol approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch or the University of Rochester Medical Center. Experimental setup and neural recordings. The experimental apparatus and techniques used to perform these experiments were described in detail previously (Newlands et al. 29; Newlands and Wei 213a,b). In short, trained animals were placed into primate chairs on a computer-controlled vestibular stimulation apparatus, which allowed the animal to be rotated in yaw, pitch, or roll or translated in any direction in the horizontal plane. Extracellular recordings were made of rostral vestibular nuclei neurons using stereotaxically placed epoxy-coated tungsten microelectrodes (2- to 1-M impedance; FHC, Bowdoin, ME). In each animal, the abducens nuclei were located bilaterally based on recording their characteristic burst-tonic responses to horizontal eye movements. The region 1 4 mm posterior and 1 4 mm lateral to abducens nuclei yielded all of the recordings in this report. The search stimulus for the current study was yaw rotation at.5 Hz, /s with a head-fixed target. Neurons included in this study were found to be sensitive to rotation in the yaw plane (horizontal head rotation with the head pitched nose down 2 to align the horizontal canals with earth horizontal). Eye movement sensitivity was tested by recording during saccades to 2 targets (spaced every 5 to the left and right of straight ahead) and smooth pursuit of a target moving in the horizontal plane (.25 Hz at 1 ). Translational sensitivity was tested in 4 6 head orientations in the horizontal plane. Based on the sensitivity to eye movements, neurons were classified by standard classifications (Scudder and Fuchs 1992) as vestibular only or non-eye movement (NEM), position-vestibular-pause (PVP; eye movement sensitivity, generally eye position, in the direction opposite to head velocity sensitivity, with or without pauses for eye saccades), or eye-head velocity (EHV; eye movement sensitivity, generally eye velocity, in the same direction as head velocity sensitivity). All neurons were assigned to being either type I, if the timing of the peak of the response was within 9 of ipsilateral rotational peak velocity, or type II, if the neuron responded within 9 of contralateral peak velocity (Duensing and Schaefer 1958). Neurons more responsive to dynamic pitch than yaw were excluded. Yaw-sensitive neurons were recorded during yaw rotation in a frequency series (at frequencies of.1,.2,.3,.5,, 1., 1.5, and 2. Hz, /s peak velocity) and an amplitude series (at.5 Hz, 15, 3,, 9,,,, and 21 /s peak velocity) as previously described (Newlands et al. 29; Newlands and Wei 213a,b). Data were collected before UCP and 6 78 wk after the procedure. Data analysis. The details of the data analysis have been reported previously (Newlands et al. 29; Newlands and Wei 213a,b) except for threshold determination. Neural data were recorded and stored at 4 khz and converted to spike trains with time-amplitude window discrimination and feature analysis. The spike train was converted to instantaneous firing rate, and this rate was compared with the eye and head position and velocity traces. Sensitivity (gain) and phase of the neuronal response were determined by averaging both the instantaneous firing rate and the stimulus (head velocity) over the collected cycles. Both the stimulus and response were fit with a first-order sine wave using a Levenberg-Marquardt least-squares algorithm, and the gain and phase were calculated by comparing the peak-to-peak amplitude of the stimulus to the response and the timing of the respective peaks. Rightward velocity was positive. Neurons recorded before

3 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING 1873 UCP were labeled as prelesion. Neurons recorded after UCP on the left were termed ipsilesion, whereas neurons on the right recorded after the left UCP were labeled contralesion. In the interest of understanding the effect of lesion on the threshold of response at a neuronal level, we calculated a threshold measure using a sign test (Rhyne and Steel 1965). In this methodology, we first used the phase of the response to align the stimulus and response, eliminating the phase difference. We then created bins for the response such that the point of zero velocity was between bins. We then compared the number of spikes in the bin before the zero crossing to the same-sized bin after the zero crossing. A positive sign was assigned if the number of spikes is larger on the excitatory side of the zero crossing compared with the same-sized bin on the inhibitory side. A negative sign was assigned if the number of spikes is the same or fewer on the excitatory side compared with the inhibitory side. Using a sign test, we then tested whether the number of cycles with a positive sign was significantly higher than chance using a one-tailed test, with the level for significance set at P.5. To determine the direction threshold of the neuron, we increased the bin size in increments of 25 ms until the sign test indicated a significant difference (Fig. 1). We assigned the velocity of the head at the end of the smallest bin that satisfied the sign test as a conservative estimate of the threshold of the neuron. For this study, we compared direction thresholds at.5 Hz, /s before and after UCP. Thus, for neurons with a significant sign test for 25-ms bins, the threshold was estimated as 4.7 /s [sin(9 25/5) ]. If the first significant bin size was 5 ms, the threshold was estimated as 9.4 /s, for 75 ms, 14 /s, etc. This method identified a direction threshold measure, rather than a detection threshold, because it compares responses during velocities on either side of zero crossings at the boundaries of inhibitory direction and excitatory direction responses. For comparison, we also calculated the detection threshold of the responses in both directions using the technique of Jamali, Cullen, and colleagues (Jamali et al. 29, 213, 214; Sadeghi et al. 27). For this analysis, the spike train was passed through a Kaiser window filter with a cutoff frequency.1 Hz above the.5-hz stimulus frequency (Cherif et al. 28). Kaiser-window-filtered neural responses were plotted against stimulus velocity after adjusting for the phase relationship between the stimulus and response. We calculated the mean and standard deviation ( ) of the distribution of firing rates within bins of rotational velocity 1 /s in size. For the calculation of detection thresholds, we calculated the degree of overlap between the firing rate distributions at each velocity bin with that at zero velocity as d= from signal detection theory (Green and Swets 1966). The detection threshold is defined for each direction for a given trial as the velocity at which d= 1. All relationships between stimuli and responses were tested for statistical significance using the Rayleigh coefficient calculation, with a criterion for significance set at P.5 (Mardia 1972). Other statistical tests all used a criterion of P.5 for significance (SPSS 14.). Nonparametric ordinal tests were used for firing rate, gain, and phase comparisons between groups. RESULTS All three animals tolerated UCP well. After surgery, they were all able to move around their cages freely and without the typical postural imbalance and unsteadiness seen after UL. All three demonstrated nystagmus in the dark with slow-phase velocity toward the plugged (left) side, as is typical of this procedure (Paige 1983). Data were collected from a total of 297 neurons recorded from the bilateral vestibular nuclei in these 3 animals. Of these, 169 were recorded before UCP and 128 after UCP. The prelesion neurons are the control cell population to which the neurons recorded postlesion are compared. Additionally, previously reported data after UL from 4 different animals under Inhibitory Direction Excitatory Direction [ ] 25 ms [ ] 5 ms [ ] 75 ms [ ] 1 ms Bin Size (ms) Signs Test = + Inhibitory Direction Total Inhibitory Crossings P value Signs Test = + Excitatory Direction Total Pvalue Excitatory Crossings Fig. 1. One example neuron (type I neuron prelesion) analyzed for direction threshold. The top traces are the response of the neuron (after Kaiser window filtering; black) and the head velocity (gray). The timing of the stimulus and response have been aligned by adjusting for the phase difference (in this example, 5 phase lead). deg, Degrees. The raster plot (middle) shows the actual spike times over 23 cycles. The table (bottom) shows the number of cycles in this example where there were more spikes for the bins immediately before or after the transition from excitatory to inhibitory or inhibitory to excitatory, respectively (sign test when the number of spikes is greater in the expected bin when comparing the bins immediately before or after the crossing). P values for 1-tailed sign test.

4 1874 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING the same protocols (Newlands and Wei 213a) are compared to the canal-plugged data in the current report. Neurons characterized. Neurons recorded were categorized based on side of the brain stem, sensitivity to rotation (ipsilaterally, type I and contralaterally, type II), whether collected before or after UCP, whether they were sensitive to translation, and whether their firing is modulated with either eye position or eye velocity. The population of postlesion neurons recorded included 112 NEM cells, 13 PVP cells, and 3 EHV cells. Because the majority of the cells recorded were NEM cells, the analysis will concentrate on this population. Table 1 presents the breakdown of the recorded neurons by side [ipsilesion (left) or contralesion (right)], directional sensitivity (type I or type II), and neuron class (NEM, PVP, or EHV) for the control and UCP conditions. The numbers of cells recorded between sides is not very meaningful because it depends on the number of passes on each side. More telling is the ratio of type I to type II neurons recorded on either side, as this ratio is independent of the number of electrode passes made. In the 3 monkeys before the lesion, 84 out of 127 recorded neurons (66%) were type I. In the UCP animals, on the lesion side 2 out of 39 recorded neurons (51%) of the neurons were type I, and on the intact side 43 out of 73 recorded neurons (59%) were type I. In contrast, in the UL animals, 38% of recorded neurons in the lesion side nucleus were type I, and 65% of recorded neurons in the intact side nucleus were type I (Newlands and Wei 213a). Comparing the prelesion ratio of type I to type II cells to the 2 UCP postlesion ratios, neither the ipsilesion nor the contralesion side has a significantly different ratio of type I to type II neurons ( 2 ; P.9 and P.3, respectively). In contrast, after UL, on the ipsilesion side, the proportion of type I neurons was significantly different from prelesion in those animals (Newlands and Wei 213a). Of the 2 type I neurons on the lesion side, 12 were tested for sensitivity to translation, and all 12 were sensitive to translation. For the type II neurons on the lesion side, 5 out of 9 tested had translational sensitivity. On the intact side, 17 out of 29 type I NEM neurons and 18 out of 22 type II NEM neurons tested were sensitive to translation. In these 3 monkeys, 26 out of 31 recorded type I and 15 out of 15 recorded type II neurons had translational sensitivity, when tested, before UCP. Dynamic responses at.5 Hz and /s peak velocity. We compared the dynamic responses of the NEM neuron with rotation for each of the three unique recording situations (prelesion, ipsilesion, and contralesion). Polar plots for each of the recording situations are presented in Fig. 2. The gain and Table 1. Count of recorded neurons I Prelesion 84 NEM 43 NEM 14 PVP 14 PVP 1 EHV 4 EHV Ipsilesion 2 NEM 19 NEM 3 PVP 5 PVP 1 EHV EHV Contralesion 43 NEM 3 NEM 1 PVP 4 PVP 1 EHV 1 EHV phase at.5 Hz, /s for each cell is represented, with gain being the radial distance from the center (log plot) and angle being the phase of the response peak relative to peak right head velocity. For control animals, all recordings are collapsed into either right (type I) or left (type II) side of the polar plot, to help with the visual comparison. At this frequency and peak velocity, the gain of type I neurons did differ among the three groups (control, ipsilesional, and contralesional: Kruskal-Wallis test; P.3). The gain of the prelesion type I neurons was 8.3 sp/s/ /s (SD); compared with sp/s/ /s (SD) for ipsilesion type I and 4.22 sp/s/ /s (SD) for contralesion type I. The ipsilesion gain was significantly lower than either prelesion or contralesion (Mann-Whitney U test; P.2 for prelesion vs. ipsilesion, P 1. for pre-ucp control vs. contralesion, and P.8 for ipsilesion vs. contralesion). The gain of type II neurons differed among the three groups (prelesion, ipsilesional, and contralesional: Kruskal-Wallis test; P.12). The gain of the prelesion control type II neurons was.27 sp/s/ /s (SD); compared with sp/s/ /s (SD) for ipsilesion type II and sp/s/ /s (SD) for contralesion type II. The gain was significantly lower for comparing contralesion with prelesion (Mann-Whitney U test; P 9 for prelesion vs. ipsilesion, P.13 for prelesion vs. contralesion, and P.12 for ipsilesion vs. contralesion). The phase relationship did not change with UCP for both type I (Kruskal-Wallis test; P.36) and type II neurons (Kruskal-Wallis test; P.85). The mean phases for type I neurons were , , and for prelesion, ipsilesion, and contralesion neurons, respectively. The mean phases for type II neurons were , , and for prelesion, ipsilesion, and contralesion neurons, respectively. In two monkeys, recordings were made over a longer period of time (Monkey Bl, 6 78 wk after UCP and Monkey To, 6 56 wk after UCP). In the third monkey, St, all units were recorded 6 8 wk after UCP. To test whether the compensation we are measuring was stable throughout the period of recording, we compared the gains at.5 Hz, /s of neurons recorded in the first half of the recording period (weeks 7 42 in Bl, weeks 7 31 in To, all data in St) with neurons recorded in the second half of the recording period in Bl and To. Sixty neurons were recorded in the early period and forty-eight in late period. There was no difference in the gain between neurons recorded in the earlier period after UCP compared with later (Mann- Whitney U test for all 4 groups, P.13). Responses with changing frequencies of yaw rotation. Most of the recorded neurons were tested with sinusoidal yaw rotation at frequencies between.1 and 2. Hz while holding peak velocity at /s. Figure 3 plots gain and phase vs. frequency of rotation for NEM neurons. Each of the four classes of post-ucp neurons are shown in separate Bode plots. The gain and phases of prelesion NEM neurons in these same animals is shown for reference. Neuronal typing (type I vs. type II) was based on response phase at.5 Hz. For both normal and post-ucp neurons, a minority of neurons have large phase leads consistent with acceleration sensitivity. Such neurons are seen both with intact labyrinths and with peripheral lesions. Ipsilesion type I neurons had reduced sensitivities compared with prelesion at multiple frequencies of rotation (Mann-Whitney U test, P.5

5 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING 1875 I Pre-Lesion Ipsilesion Fig. 2. Responses of recorded neurons in the prelesion vestibular nuclei (top), ipsilesion nuclei (middle), and contralesion nuclei (bottom) at.5 Hz, /s. Open circles represent individual neurons, phase re: velocity is the measure, and gain is the radial distance from the center. For prelesion, all neurons represented as if on the right side (neurons recorded on the left with phases shifted ). Red squares are the mean of gain and phase for prelesion neurons. Black, filled circles are postlesion mean responses. The contralesion mean type I response is obscured by the red square (lower left plot). Individual Neurons Mean UCP Mean Pre-lesion Contralesion at all frequencies marked with an arrow in Fig. 3). Contralesion type I neurons only had a significant difference from prelesion type I neurons at.3 Hz. I neurons on the contralesion side also had reduced sensitivities compared with prelesion at several frequencies (Mann-Whitney U test, P.5 at the marked frequencies). The sensitivities of ipsilesion type II neurons did not differ from prelesion neurons except at.1 Hz. Responses at.5 Hz with changes in peak velocity. Figure 4 shows a comparison of the gain of responses by peak velocity for type I and type II neurons by side for UCP animals. In these data, frequency was held constant at.5 Hz. For ipsilesion and contralesion type I and type II neurons, the gain of the response to rotation decreases with higher peak velocities of rotation (Kruskal-Wallis test; P.1 for all comparisons). These data demonstrate that the amplitude

6 1876 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING Fig. 3. Bode plots showing gain (upper plots) and phase (lower plots) of type I (left) and type II (right) NEM neurons in the ipsilateral (top) and contralateral (bottom) nuclei after UCP compared with prelesion. Type is determined by response phase at.5 Hz, /s. Peak velocity /s for all data. Individual neuron responses demonstrated by open, red squares for the NEM neurons before lesion and black circles for the NEM neurons after lesion. Mean prelesion values are demonstrated by thick, red line, and the post-ucp values are demonstrated by thick, black line. Prelesion data from both sides of the midline are combined with phases for neurons recorded on the opposite side adjusted by. For prelesion data, there were 33, 41, 36, 79, 37, 46, 37, and 43 type I neurons reported and 18, 26, 23, 42, 23, 29, 23, and 29 type II neurons reported at.1,.2,.3,.5,, 1., 1.5, and 2. Hz, respectively. After UCP, the respective neurons shown at each of these same frequencies are 15, 17, 6, 2, 16, 18, and 8 for ipsilesion type I; 11, 11, 3, 18, 13, 17, 7, and 1 for ipsilesion type II; 29, 32, 15, 42, 32, 34, 19, and 21 for contralesion type I; and 2, 21, 8, 28, 21, 22, 1, and 12 for contralesion type II. Gain plots are log-log, phase plots log on x-axis only. Arrows denote statistically significant differences between prelesion and post-ucp groups. Data points are artificially offset horizontally to improve clarity. compression with yaw rotation, which we previously reported for NEM neurons in normal animals (Newlands et al. 29) and animals after UL (Newlands and Wei 213a) and was demonstrated after UL in cats (Heskin-Sweezie et al. 27), is also seen after UCP. The phase of the response did not change with peak velocity for three of the four groups of neurons after UCP but did change for contralesion type II (type I ipsilesion: Kruskal-Wallis test; P 66; type I contralesion: Kruskal-Wallis test; P.932; type II ipsilesion: Kruskal-Wallis test; P.995; type II contralesion: Kruskal-Wallis test; P.9). Comparisons of gain at each peak velocity were made between pre- and post-ucp responses. Arrows in Fig. 4 note conditions for which the gain after UCP was significantly lower than prelesion gain (Mann-Whitney U test; P.5). Similar to our report on UL results, we also examined whether UCP affected the symmetry of responses by examining the gain of the responses independently in the inhibitory and excitatory directions using the slope of the relationship between the head velocity and neuron firing for phase-adjusted x y plots. As with UL, there is no systematic change in the relationship between the inhibitory and excitatory gains with changes in frequency or amplitude. In the canal-plugged animals, as with normal animals and those following UL, excitatory gain is larger than inhibitory gain. Comparison of gain and phase between UCP and UL. Figures 5 (gain and phase vs. frequency at peak velocity of /s) and 6 (gain and phase vs. peak velocity at frequency of.5 Hz) allow comparison of the mean of responses after UCP to those after UL in our previous report (Newlands and

7 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING 1877 Ipsilesion I Gain (sp/s/ /s) Phase ( ) Gain (sp/s/ /s) Pre-lesion Peak Velocity ( /s) UCP Mean Pre-lesion Mean UCP Contralesion Peak Velocity ( /s) I Fig. 4. Gain and phase vs. peak velocity for the 4 groups of NEM neurons after UCP compared with control data prelesion. Type is determined by response phase at.5 Hz, /s. All data at.5 Hz. Peak velocities at 15, 3,, 9,,,, and 21 /s. In all of the plots, the prelesion control data are shown by open, red boxes (individual responses) and thick, red line (average of responses). For the lesion data, the individual values (open, black circles) and the average values (thick, black lines) are shown. The control values reflect neurons on both sides of the brain stem, with phases for neurons recorded on the opposite side adjusted by. For prelesion data, there were 18, 39, 8, 4, 32, 36, 35, and 3 type I neurons reported and 1, 22, 23, 12, 13, 12, 1, and 7 type II neurons reported at 15, 3,, 9,,,, and 21 /s, respectively. After UCP, the respective neurons shown at each of these same peak velocities are 3, 13, 2, 11, 9, 6, 6, and 6 for ipsilesion type I; 3, 8, 18, 6, 8, 5, 4, and 4 for ipsilesion type II; 1, 24, 42, 26, 17, 13, 13, and 12 for contralesion type I; and 7, 16, 28, 15, 1, 7, 6, and 5 for contralesion type II. Gain plots are log on x-axis only, and phase plots are linear. Arrows denote statistically significant differences between prelesion and post-ucp groups. Data points are artificially offset horizontally to improve clarity. Phase ( ) Peak Velocity ( /s) Peak Velocity ( /s) Wei 213a). After both types of peripheral lesion, the gain of ipsilesion type I neurons are similarly depressed. The increase in type II ipsilesion gain seen after UL, particularly in the plot of gain vs. frequency (Fig. 5), is not seen after UCP. Contralesion type II responses are similar after UL and UCP. Bias firing rates and rectification. One possibility is that the source of long-term deficits in vestibular function after vestibular lesion is increasing directional asymmetry at the single neuron level with increasing peak velocity. However, the proportion of neurons that silenced during rotation was not significantly different after UCP vs. before UCP in the same animals under any combination of velocity or frequency tested ( 2 ; P.5 for all comparisons). Additionally, after UL, there was a change in the baseline firing rate of some neurons, particularly type II neurons (Newlands and Wei 213a). This was not the case after UCP. The baseline (bias) firing rate was unchanged for both type I [ (SD) sp/s ipsilesion, sp/s contralesion] and type II [ sp/s ipsilesion and sp/s contralesion] neurons on both sides compared with prelesion neurons (type I sp/s and type II sp/s; Mann-Whitney U test for all 4 comparisons of postlesion to prelesion; P.259). There was no change in bias firing with frequency or with amplitude of rotation for any of the 6 groups (Kruskal-Wallis test; P.366 for all comparisons).

8 1878 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING Ipsilesion I Gain (sp/s/ /s) Fig. 5. Gain and phase vs. frequency at peak velocity /s comparing prelesion (current study; red), UCP (current study; black), and UL (Newlands and Wei 213a; gray) mean responses. Error bars SE. Data points are artificially offset horizontally to improve clarity. Prelesion and UCP averages as in Fig. 3, but gain scale is linear to improve clarity. Phase ( ) Frequency (Hz) 1 Mean UCP Mean Pre-lesion Mean UL Contralesion Frequency (Hz) 1 I Gain (sp/s/ /s) Phase ( ) Frequency (Hz) 1.1 Frequency (Hz) 1 Threshold determination. We investigated whether, in our data, the threshold of our population of central vestibular NEM neurons was affected by UCP. As described in METHODS, our approach was to use actual spike counts to detect the angular velocity at which there was a statistically significant difference in the number of spikes in the period before zero-velocity crossing compared with after zero-velocity crossing. The methodology is demonstrated in Fig. 1. This measure is a directional threshold measure that compares the period of time with negative velocity with the period with positive velocity. Unlike previous threshold measures, spike counts are compared directly within each cycle rather than being pooled across cycles. We calculated two direction detection velocities, one for rotation in the inhibitory direction and one for rotation in the excitatory direction. This measure was applied to data at.5 Hz with a peak velocity of /s. The data for prelesion and UCP neurons for direction threshold are shown on the upper lefthand side of Table 2. In addition, we also analyzed our data after the technique of Jamali et al. (214) using a Kaiser window filter on the raw spike data before calculating a d= from signal detection theory to determine the detection threshold of each neuron as the rotational velocity at which d= 1. This result is shown in the right-hand side of Table 2 (Detection Threshold). For comparison, we also show the direction (our technique) and detection [Jamali et al. (214) technique] thresholds for the UL data previously reported (Newlands and Wei 213a). We found no statistically significant difference in direction thresholds between prelesion and post-ucp type I or type II neurons (Kruskal-Wallis test; P.37 for all comparisons).

9 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING 1879 Ipsilesion I Gain (sp/s/ /s) Phase ( ) Peak Velocity ( /s) Mean UCP Mean Pre-lesion Mean UL Contralesion Peak Velocity ( /s) I Fig. 6. Gain and phase vs. peak velocity at.5 Hz comparing prelesion (current study; red), UCP (current study; black), and UL (Newlands and Wei 213a; gray) mean responses. Error bars SE. Data points are artificially offset horizontally to improve clarity. Prelesion and UCP averages as in Fig. 3, but gain scale is linear to improve clarity. Gain (sp/s/ /s) Phase ( ) Peak Velocity ( /s) Peak Velocity ( /s) For detection threshold, type II neurons on both sides and ipsilesion type I neurons had higher values after UCP compared with prelesion (Mann-Whitney U test; P.5). DISCUSSION These results present a unique data set exploring the responses of central NEM vestibular neurons after compensation from plugging of the semicircular canals on one side. Because we used the same protocols in a previously published report of the behavior of NEM neurons after compensation from labyrinthectomy, we are also uniquely able to compare the compensated state after these two lesions. The most striking differences between the UCP and UL results are the difference in degree of recovery of type II responses on the ipsilesion side and the relative abundance of type I neurons on the ipsilesion side. Physiology of canal plugging. Much of what is known about cellular activity in the vestibular nuclei after recovery from vestibular lesions has been learned following UL. However, a number of studies have investigated aspects of vestibular physiology after plugging of the semicircular canals. Unlike labyrinthectomy, which results in a complete loss of normal vestibular function in the labyrinth, the deficit seen after canal plugging is dependent on the frequency of stimulation. The frequency dependence of the effect of semicircular canal occlusion has been elaborately modeled by Rabbitt and col-

10 188 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING Table 2. Direction and detection thresholds Direction Threshold, /s Detection Threshold, /s Excitatory Inhibitory Excitatory Inhibitory Variability Mean SD N Mean SD N Mean SD N Mean SD N () SD UCP animals prelesion I prelesion ipsilesion * * I ipsilesion * contralesion I contralesion * * UL animals prelesion I prelesion ipsilesion * I ipsilesion 24.* contralesion I contralesion *Different from prelesion, Mann-Whitney U test; P.5. (), SD of firing activity at velocity [phase-adjusted, Kaiser-window filter spike train after Jamali et al. (214)]. leagues (1999) and confirmed in toadfish (Rabbitt et al. 1999, 21) and primates (Sadeghi et al. 29) at the vestibular afferent level. The VOR after bilateral canal plugging reflects the frequency dynamics of canal plugging, resulting in a much greater attenuation of the VOR at lower frequencies than at higher frequencies (Lasker et al. 1999; Paige 1983; Yakushin et al. 1998). Studies in which not all of the canals are plugged have confirmed that rotational attenuation is specific to the involved canal (Carey et al. 27; Cremer et al. 2; Yakushin et al. 1998), that UCP does not result in the postural symptoms seen after UL (Paige 1983), and that afferent activity is attenuated only for the specific canals that are plugged (Goldberg and Fernández 1975; Paige 1983). Compensation following UCP. Recovery of function after UCP has primarily been studied assessing yaw VOR function. UCP results in an increase in the latency of the VOR response for rotation to the side of the lesion (Lasker et al. 1999). In the same study, they found that VOR asymmetries were not only frequency dependent, but also velocity dependent, as there was no significant asymmetry at any frequency of yaw rotation with a peak velocity of 2 /s, but there was asymmetry if the peak velocity was raised to 1 /s. These authors also examined the VOR after UL under the same conditions in the same species (Lasker et al. 2). The differences noted in the VOR were that there was more nystagmus after UL, which resulted in an increased bias velocity during VOR that altered the dynamics of the VOR at 2 Hz. The time course of recovery for steps of acceleration toward the intact side was longer after UL than UCP. Similar findings were reported in gerbils by Newlands et al. (25). In that study, gerbil VOR gain decreased to a lower level and recovered more slowly following UL than following UCP. Gerbil VOR after UCP dropped acutely to 5% of normal in the 1st day postplugging and was markedly asymmetric at all frequencies tested (.2,.5, and 1. Hz) for 1wk after the surgery. By 3 wk postplugging, the VOR gain recovered to 91% of normal (Newlands et al. 25). Paige (1983) found in squirrel monkeys that VOR gain dropped to 57% of preplugging after unilateral horizontal canal plugging and recovered to a gain of 83% of normal in one month s time at frequencies between.1 and 2. Hz. He also noted that gain of the VOR decreased with peak velocity /s at.2 and.2 Hz, and marked nonlinearity was noted at 24 and 3 /s. In cats, Broussard et al. (1999) noted frequency-dependent asymmetry in the horizontal VOR 2 Hz after UCP independent of velocity and acceleration. The gain of the VOR in cats after UCP dropped to 5% of normal and then recovered to 7% of normal (Broussard et al. 1999). Broussard and Hong (23) also measured VOR recovery after unilateral plugging of the horizontal semicircular canal and found a 3 to 5% decrease in gain at 1. Hz with recovery to within 1% of normal after 3 mo in cats. We did not assess behavioral compensation in our animals. The literature, however, supports a compensation process after UCP similar to that following UL, yet faster and potentially more complete. Comparison to other central neuron studies. Despite the interest in canal plugging as an experimental technique and its use in clinical medicine, there is very little information about the impact of plugging the canals on central vestibular responses. This study represents the most comprehensive assessment to date of vestibular responses in the vestibular nuclei after UCP. Abend (1977, 1978) performed UCP and investigated the responses of neurons in the superior vestibular nucleus (SVN) to angular-velocity trapezoids and constant-velocity rotations. These experiments were performed in barbiturate anesthetized, cerebellectomized squirrel monkeys. These studies established that most SVN neurons receive synergistic bilateral inputs from parallel canals. These acute studies do not shed light on plasticity after canal plugging. Plasticity in the efficacy of vestibular commissural pathways in cats has been studied after recovery from UCP. Farrow and Broussard (23) used responses to electrical stimulation before and after UCP to evaluate changes in the efficacy of commissural pathways in cats. Compared to before UCP, they reported an decrease in excitatory commissure projections from the contralesion to ipsilesion side but no significant difference in inhibitory commissure projections. They found that neurons receiving inhibitory input (nearly all type I) had increased resting firing rates after UCP, and those receiving excitatory commissural input

11 CENTRAL VESTIBULAR NEURONS AFTER UNILATERAL CANAL PLUGGING 1881 (roughly divided evenly between type I and II) had no significant change in firing. Neuronal thresholds. Calculations of neuronal thresholds for individual neurons to sinusoidal vestibular stimuli have been investigated by several authors (Jamali et al. 213, 214; Liu et al. 21; Yu et al. 212, 214). These discussions have centered around two basic types of thresholds, detection thresholds, which ask the question of whether an impartial observer of firing can tell whether the animal is moving, and direction thresholds, which ask whether an impartial observer can discriminate the direction of movement based on firing. Traditional psychometric evaluations of threshold in the vestibular system have concentrated on direction thresholds. The appropriate protocol for evaluating behavioral direction thresholds includes application of subthreshold through suprathreshold stimuli and evaluating the responses of the subject to these stimuli using forced choice (Grabherr et al. 28; Mallery et al. 21). A similar paradigm has been developed by neuronal data in the vestibular system (Liu et al. 21; Yu et al. 212, 214). Jamali, Cullen, and colleagues (Jamali et al. 213, 214) have put forward a methodology for determining neuronal motion-detection thresholds that requires only suprathreshold stimuli. This methodology produces different results, depending on whether the spike train is filtered or interspike intervals are used (Jamali et al. 214). Although that technique is an established one, useful and extensively described, it does suffer two potential shortcomings. The first is that the technique compares all cycles with all other cycles. Thus, if the baseline firing rate drifts (which would increase the variance), this technique will return a higher threshold for the same gain and regularity of firing than if the baseline firing is stable. A second concern is that filtering of the neuron eliminates the information in the timing of spike activity. The technique we describe differs from both of the other techniques in that we compare firing rates on either side of baseline firing for each cycle. Like the technique described by Angelaki and colleagues (Liu et al. 21; Yu et al. 212, 214), we preserve spike-timing information, and the measure is one of direction threshold. Using the two techniques described above, three laboratories investigated thresholds after peripheral lesion in the vestibular system (Jamali et al. 214; Newlands et al. 214; Yu et al. 214). These results from different laboratories have been inconsistent. Jamali et al. (214) presented data demonstrating that UL resulted in an increase in detection threshold in central vestibular NEM neurons contralateral to the side of lesion acutely following UL. With recovery, these detection thresholds remained elevated 3 wk despite a recovery in gain, which was attributed to a persistent elevation in variability. In contrast, Yu et al. (214) found that higher thresholds after recovery from UL were due to decreased gain rather than increased variability. Similarly, we do not see differences between the variability of firing [ (); the standard deviation of the Kaiser-window-filtered, phase-adjusted firing rate at velocity after the technique of Jamali et al. (214) reported in Table 2] of neurons recorded labyrinth intact or after UL or UCP. Thresholds are generally poorer following UCP or UL in our data, consistent with lower gains after peripheral lesion, but these differences are not statistically significant for a majority of the comparisons. Does the compensated state differ between UCP and UL? In our experiments, we only recorded from neurons before lesion and chronically, after the animals compensated from the lesion. Others have recorded acutely after canal plugging (Abend 1977). Acutely after UCP, as acutely after UL (Newlands and Perachio 199a; Sadeghi et al. 21, 211), contralesion type I neurons have reduced gain due to loss of modulation from the contralateral labyrinth (Abend 1977). Our findings support that recovery of contralesion type I gain to head movements occurs with compensation after UCP as it does after UL (Newlands and Wei 213a; Sadeghi et al. 21, 211), likely through increased efficacy of modulation from the contralateral (intact) vestibular afferents. We propose that the recovery from UCP differs from UL in the ipsilesion vestibular nuclei. In UCP, there is no need to balance the resting activity of the vestibular nuclei, which is the earliest stage of recovery after UL (review: Smith and Curthoys 1989). The need for balancing resting activity likely impacts the mechanisms available to the brain stem for balancing dynamic vestibular activity with head movement. In UCP conditions, we found that the bias firing rates were unchanged between groups, whereas the firing rates were increased for type II neurons on both sides in the UL animals. The gain of type II neurons was increased on the ipsilesion side after UL but not after UCP. In both preparations, modulation of ipsilesion type I neurons depends on either direct or indirect commissural pathways; there is no source of this information other than the contralateral labyrinth. I neurons are excited from the contralateral intact labyrinth (excitatory commissural projections; Abend 1977; Shimazu and Precht 1966). neurons can also be modulated from the contralateral labyrinth either through type II interneurons (Shimazu and Precht 1966) or by direct inhibitory commissural projections (Bagnall et al. 27; Goldberg et al. 1987; Kasahara et al. 1968; Kasahara and Uchino 1974; Mano et al. 1968; Wilson et al. 1968). Plasticity that increases the gain of ipsilesion type I neurons to yaw rotation could theoretically be through in- Ipsilesion I Potential sites of plasticity after UCP Potential site of plasticity after UL Contralesion Fig. 7. Schematic of basic circuitry in the vestibular nuclei and the potential sites of plasticity after UCP or UL. Filled triangles, inhibitory connections; open triangles, excitatory connections.