A major outstanding goal of vestibular neuroscience is to understand the distinctive functional roles of type I and type II hair cells. One important question is whether these two hair cell types differ in bundle structure. To address this, we have developed methods to characterize stereocilia numbers on identified type I and type II hair cells in the utricle of a turtle, Trachemys scripta. Our data indicate that type I hair cells, which occur only in the striola, average 95.9 ±16.73 (SD) stereocilia per bundle. In contrast, striolar type II hair cells have 59.9 ± 8.98 stereocilia, and type II hair cells in the adjacent extrastriola average 44.8 ± 10.82 stereocilia. Thus type I hair cells have the highest stereocilia counts in the utricle. These results provide the first direct evidence that type I hair cells have significantly more stereocilia than type II hair cells, and they suggest that the two hair cell types may differ in bundle mechanics and peak mechanoelectric transduction currents.
Vertebrates use mechanoreceptors called hair cells to detect head motion. Each hair cell bears a hair bundle comprising a single kinocilium and multiple stereocilia. There are two types of vestibular hair cells. Type II hair cells are innervated by bouton terminals of vestibular primary afferents; they occur in all vertebrates. Type I hair cells are innervated by cup-like, calyceal terminals; they occur only in amniotes (reptiles, birds, and mammals). These two hair cell types were recognized almost 50 years ago (Wersäll 1956), but their distinctive functional roles are still poorly understood.
Hair cells are mechanoreceptors; thus one logical suggestion is that type I and type II hair cells differ in bundle mechanics. Because bundle structure is a major determinant of bundle mechanics (Howard et al. 1988; Peterson et al. 1996; Silber et al. 2003), we compared the morphology of type I and type II ciliary bundles. Here we describe a method that allows us to characterize stereocilia numbers on identified hair cell types. We report, for the first time, that type I hair cells have significantly higher stereocilia numbers than type II hair cells. These differences in stereocilia counts suggest that type I and type II hair cells may differ in bundle mechanics and peak mechanoelectric transduction currents.
Our experimental preparation is the utricle of a turtle, Trachemys (Pseudemys) scripta elegans. In this species, type I hair cells are restricted to the striola (Jorgensen 1974, 1988), a crescent-shaped specialization in the neuroepithelium and otoconial membranes of vertebrate otolith organs (Werner 1933). Previously (Peterson and Rowe 2001), we reported that the striola of T. scripta can be distinguished from the extrastriola by quantitative features of its ciliary bundles and that it is subdivided into two parallel bands, which we call zone 2 and zone 3 (Fig. 1A). Zone 3 bundles vary markedly in stereocilia numbers; some of them have the highest stereocilia counts in the utricle (Peterson and Rowe 2001). Type I hair cells are restricted to zone 3, and indirect evidence (based on cuticular plate size) suggested the hypothesis that bundles with the highest stereocilia counts belong to type I hair cells (Moravec et al. 2003). To test this hypothesis, we used confocal microscopy to examine utricular whole mounts that were double-stained to visualize stereocilia and the utricular afferents that define hair cell type. Some of these data have appeared previously in abstract form (Moravec and Peterson 2004).
Four turtles of both sexes (11–13 cm plastron length, 225–329 g; Kons Direct, Germantown, WI; Table 1) provided useful data. Animal care protocols have been published previously (Brichta and Peterson 1994). We followed the Ohio University Animal Care and Use Committee guidelines at all times.
Labeling of primary afferents and hair bundles
We chilled turtles in an ice bath for 15–30 min and killed them via intramuscular injection of 0.7 ml pentobarbitol sodium with phenytoin sodium (Euthasol). We decapitated the turtles, bisected the heads, placed them in oxygenated turtle Ringer solution (Hounsgaard and Nicholson 1990), and removed the brains from the skulls, leaving the labyrinth and a large stump of cranial nerve VIII (CN8) intact.
We labeled afferents by applying gelfoam pellets infiltrated with Micro-Ruby (BDA/Rhodamine 3,000 mW, Molecular Probes; 33 mg/ml in distilled water) to the stump of CN8 and maintained the head in circulating, oxygenated Ringer solution. After 4–6 h, we fixed the heads with 4% paraformaldehyde in phosphate buffer for 24 h, removed the utricles from the bony labyrinths, opened them, and removed the otolithic membranes. We visualized stereocilia with phalloidin, which labels filamentous actin, conjugated to Alexa-Fluor 488 (Molecular Probes; 2U/100 ml 0.1 M phosphate buffer). Finally, we mounted intact utricles on microscope slides that had wells filled with anti-fade reagent (Glycerol/PBS, Molecular Probes) and secured them with coverslips.
To judge whether the band of calyces was adequately labeled, we compared Micro-Ruby–labeled utricles with utricles from a separate study in which we used a different technique to visualize calyces. In that study, we manipulated the bath composition to swell calyces so that they appear as a band of black rings surrounding one to seven hair cells in utricular whole mounts (Moravec et al. 2003). We analyzed utricles only if the band of calyces appeared to be fully labeled.
We used a Zeiss LSM 510 confocal microscope equipped with a Plan-Apochromat 63× oil immersion DIC objective (NA = 1.4) to collect all image stacks, and we controlled on-screen image size via digital magnification. Pinhole size varied from 0.60–0.75 Airy units, which yielded optical sections 0.4–0.6 μm thick. In the first two turtles, we visualized hair bundles and afferents in the same stacks using digitally magnified multi-track scans with 488- and 543-nm laser lines (Fig. 1B) (see Supplemental Material1for confocal stack). In the final two turtles, we also mapped the spatial location of all hair cells in our sample area as follows. First, we made a low-magnification, multi-track image stack of our sample area with 488- and 543-nm laser lines, including all bundles and afferents. Then we used single-track scans with the 488-nm laser to acquire higher magnification images of small groups of bundles within the sample area. Next we combined four to six of these higher magnification confocal stacks to form a montage of the total sample area. Finally, we used MATLAB (ver. 6.5; MathWorks) to convert the coordinates of hair cells in individual high magnification images to a single set of global coordinates that showed the location of all hair cells within our total sample area. To do this, we 1) selected the region to be magnified and recorded the location of its upper left corner on the low-magnification image, 2) specified the location of each cell within the high magnification image as a vector originating in the upper left corner of that image and ending at the kinocilium, 3) rescaled the vector magnitude in proportion to the magnification ratio between the low- and high-magnification images, and 4) translated the origin of the vector to its correct location on the low-magnification image. High-magnification stacks provided the clearest images of hair bundles; the low-magnification stack allowed us to determine the hair cell type for each bundle. An example of one such data set is shown in Fig. 1C.
Except for one control sample (see Control for the effect of sample location), all analyses focused on an ∼200 × 200-μm region of interest (Fig. 1A, solid box) that included the striola (zones 2 and 3) and adjacent lateral (zone 1) and medial (zone 4) extrastriola. We refer to this as our primary sample area. We counted stereocilia in this region because type I hair cells are located in the striola (Jorgensen 1974, 1988; Moravec et al. 2003) and because it is important to contrast stereocilia numbers on type I hair cells with stereocilia numbers on type II hair cells at the same utricular locus (see Known location). Our primary sample area is part of the larger transect on which we focus much of our quantitative analyses of bundle structure (Fontilla and Peterson 2000). Data for our four primary samples are summarized in Table 1.
Measurements and statistical analysis
For hair cells that had bundles clear enough to analyze, we counted stereocilia in optical sections just above the apical surface, directly from the computer monitor. We examined more than one optical section through the base of the bundle if necessary to visualize all stereocilia in the bundle; this is useful if the hair cell surface is curved or not perfectly perpendicular to the scanning axis. We also recorded hair cell type based on the presence or absence of a calyx, the cup-like afferent terminal that surrounds each type I hair cell (Fig. 1B, inset). For turtle 1, we recorded stereocilia numbers and hair cell type only; for turtles 2–4, we also assigned hair cells to one of four zones (Fig. 1, A and C; Table 1). In this paper, zone 1 corresponds to hair cells lateral to the line of polarity reversal (dotted line in Fig. 1C); the kinocilial (tall) ends of utricular bundles face each other across this reversal line. Occasionally, bundles with striolar-like properties, i.e., relatively high stereocilia counts (this report) and low kinocilia-to-stereocilia height ratios (Xue and Peterson 2004) occur immediately lateral to the line of polarity reversal (see Fig. 7 of Fontilla and Peterson 2000). Such bundles properly belong in zone 2 rather than zone 1 (i.e., they should be considered striolar bundles). Because we have not yet developed quantitative criteria for identifying these “errant” striolar bundles, we made no attempt to separate them from the rest of zone 1 in this report. Thus differences we describe between zone 1 (lateral extrastriolar) and zone 2 (striolar) bundles are probably underestimates of the true difference between these two bundle populations. Zone 2 corresponds to hair cells between the reversal line and the band of calyces. Zone 3 corresponds to hair cells innervated by calyces and any hair cells scattered between the calyceal terminals. Collectively, zones 2 and 3 form the striola. Zone 4 corresponds to hair cells medial to the band of calyces.
All four zones contain type II hair cells; type I hair cells are restricted to zone 3 (Moravec et al. 2003). We tested differences between stereocilia numbers for statistical significance using ANOVA and planned or posthoc (Tukey honest significant difference test for unequal sample sizes) comparisons.
In agreement with earlier results (Moravec et al. 2003), labeled calyces formed a 50- to 60-μm-wide band that followed the trajectory of the striola and was displaced 20–30 μm medially from the line of polarity reversal (Fig. 1C). This distribution corresponds to zone 3 as identified from autocorrelation analysis (Rowe and Peterson 2004) of scanning micrographs (Peterson and Rowe 2001). Thus type I hair cells are restricted to striolar zone 3. Stereocilia numbers on hair cells of the striola and adjacent extrastriola were markedly heterogeneous. This is shown for one utricle in Fig. 1, C and D, and summarized for all four cases in Table 1. High magnification confocal micrographs of type I and type II bundles from the striola are shown in Fig. 2. A second type I bundle and its calyx terminal are shown in Fig. 1B and in Supplemental Material.
A two-factor ANOVA revealed significant effects of hair cell type (F = 171.7, df = 1, 315, P < 0.001) and of zone (F = 46.5, df = 3, 315, P < 0.001). Thus there are significant differences between type I and type II hair cells (collapsed across zones) and between zones (collapsed across hair cell types). The R2 value for the overall model (R2 = 0.79; F = 289, df = 4, 315, P < 0.001) indicates that differences between hair cell type and utricular zone account for a significant percent (79%) of the variability in stereocilia number within our primary sample area. We can partition these overall differences in stereocilia number into two sources of variation.
Type I hair cells have significantly more stereocilia than total striolar type II hair cells (95.9 ± 16.7, n = 69 vs. 59.9 ± 9.0, n = 85; F = 597.5, df = 1, 215, P < 0.001) or type II cells within zone 3 alone (55.4 ± 11.2, n = 14; F = 94.1, df = 1, 81, P < 0.001; Figs. 2 and 3 A, filled symbols, and 3B). To control for the possibility that these differences between type I and type II stereocilia numbers are unique to our primary sample location (Fig. 1A, solid box) we compared stereocilia numbers on type I and type II hair cells from a sample at the far posterior striola (Fig. 1A, dotted box; Fig. 3A, open symbols). In this region, stereocilia numbers are nearly identical to those in our primary sample area. Type I hair cells averaged 97 ± 12.6 stereocilia (n = 26; range, 80–119); these stereocilia counts are significantly higher than total striolar type II cells in the posterior striola (57.2 ± 7.9, n = 91, range, 44–74; F = 208.6, df = 1, 61, P < 0.001) or type II hair cells in zone 3 alone (57.3 ± 10.1, n = 9, range, 51–69; F = 25.7, df = 1, 15, P < 0.001). Thus the difference between stereocilia numbers on type I and type II hair cells appears to be a general feature of the utricular striola in T. scripta.
A second source of variation in stereocilia number arises from local differences in type II hair cells (Fig. 3, A, C, and D). A single-factor ANOVA on type II hair cells revealed that striolar type II hair cells have significantly more stereocilia than hair cells in the adjacent lateral and medial extrastriola (Fig. 3C; planned comparison: F = 38.3, df = 1, 247, P < 0.001). In addition, there is a significant effect of zone on stereocilia number (Fig. 3D; F = 63.7, df = 3, 247, P < 0.001). The R2 value for the overall model was 0.44, indicating that zone assignment accounts for 44% of the variability in type II stereocilia numbers. Stereocilia number increased significantly from zone 1 (lateral extrastriola) to striolar zone 2 (posthoc comparison, P < 0.001) then decreased toward the medial extrastriola (zone 4). Comparison of type II bundles in zones 2–4 indicated that there were no significant differences between adjacent zones (zone 2 vs. 3, zone 3 vs. 4), but the difference between zone 2 and zone 4 was significant (posthoc comparison, P < 0.01), as was the regression of stereocilia number against zones 2–4 (F = 17.1, df = 1, 120, P < 0.001; R2 = 0.12). Thus stereocilia numbers on type II hair cells peak in zone 2 and decrease toward both the lateral and medial extrastriola.
Our most important finding is that utricular type I hair cells have more stereocilia than type II hair cells. In addition, type II hair cells in or near the striola differ in stereocilia numbers as a function of their location. To show these differences between bundles, we attempted to satisfy four criteria.
Approximately 50% of bundles in our samples were measurable. We did not count stereocilia on the remaining bundles because they were bent off the optical axis of the microscope, which made it more difficult to resolve their closely spaced stereocilia.
Adequate sample size
Convincing demonstration of differences between bundles requires samples that are large enough for statistical analysis. In our primary sample area (Fig. 1A, solid box), which includes the striola and adjacent lateral and medial extrastriola, we counted stereocilia on 397 bundles from four utricles of four different turtles. For three of these utricles (n = 320), we also assigned type II hair cells to a utricular zone (by definition, type I hair cells are always in zone 3). All the type- and location-specific differences we report here were statistically significant with one exception: medial to the reversal line, type II bundles in adjacent zones did not differ in stereocilia counts. Nevertheless, the gradual decrease in stereocilia numbers from zone 2 to zone 4 was significant. In addition to our four primary samples (Table 1), we also counted stereocilia numbers on 97 hair cells from the far posterior striola (Fig. 1A, dotted box). In this control sample, stereocilia numbers on type I and type II hair cells were also significantly different (Fig. 3A, open circles).
Knowledge of hair cell location is critical for establishing type-specific differences in ciliary bundles. The reason for this is that bundle structure is known to vary with location in amniote vestibular organs (e.g., Jorgensen 1989; Kirkegaard and Jorgensen 2001; Lim 1976, 1979; Lindeman 1969; Peterson et al. 1996), and without holding location constant, it is impossible to know whether observed differences between type I and type II bundles are due to effects of spatial locus or effects of hair cell type. In turtle utricle, type I hair cells have significantly more stereocilia than type II hair cells, even when location is held constant (i.e., when all measured bundles are within the striola or even within zone 3). Thus the differences we observed between hair cell types are not confounded by spatial effects.
Accurate identification of hair cell type
We identified hair cell type by labeling postsynaptic afferents with a dextran amine tracer applied to the utricular nerve. We cannot exclude the possibility that we missed some type I hair cells because their calyces were too weakly labeled, but we believe the number of such misidentified hair cells is small for two reasons. First, we analyzed only those utricles in which the band of calyces was judged to be completely labeled (compared with an alternate method of visualizing calyces; see methods). Second, the pattern of afferent labeling and the differences in stereocilia numbers between hair cell types were remarkably consistent in all four cases (Table 1); such consistency is unlikely if a significant number of type I hair cells were misidentified. Note too that the type I–type II differences in stereocilia counts were so large that a small number of misidentified type I cells, if there were any, would not be likely to change the results of our statistical analysis. One advantage of labeling afferents with dextran amine is that individual calyceal terminals can be isolated (Fig. 1C, red profiles), the number of type I hair cells supplied by each terminal can be counted, and the structural features of their bundles can be analyzed.
A previous study on Daubenton's bat reported statistically significant differences between stereocilia numbers on striolar and extrastriolar hair cells (Kirkegaard and Jorgensen 2001); in contrast to the present results, striolar bundles in the bat had significantly fewer stereocilia than extrastriolar bundles. Two other studies have suggested that type I hair cells may have more stereocilia than type II hair cells based on indirect evidence: comparisons of a mixed population of bundle types with a population of known type II bundles (Peterson et al. 1996) or extrapolations from partial stereocilia counts (Morita et al. 1997). To our knowledge, the present results are the first direct demonstration of significant differences in stereocilia numbers on type I and type II hair cells of any vestibular organ. The picture that emerges is of two superimposed patterns.
Type I hair cells are restricted to one subdivision of the striola, and they form a distinct population because of their high stereocilia counts. We saw no obvious tendency for type I hair cells supplied by the same calyceal terminal to have similar stereocilia counts or for stereocilia numbers to depend on the number of type I receptors supplied by a single calyceal terminal. This result is tentative, however, because often the type I hair cells supplied by a single calyceal terminal were not all measurable with our current techniques (Fig. 1C).
Type II hair cells are present throughout the striola and adjacent extrastriola. Their stereocilia numbers peak just medial to the line of polarity reversal (zone 2) and decrease significantly toward both the lateral and medial extrastriola. The numbers we report for type II hair cells in zones 1 and 4 (Table 1) are typical of type II hair cells in lateral and medial extrastriola. We base this on results of a separate study in which we used scanning micrographs of sonicated utricles and autocorrelation analysis (Rowe and Peterson 2004) to analyze stereocilia arrays (number, spacing, and arrangement of stereocilia on the apical surface of individual hair cells) along a medial-to-lateral strip of macula; this strip includes the primary sample area described in the present report. Data from such sonicated utricles suggest that stereocilia numbers do not vary significantly across either subdivision of the extrastriola except for a slight decrease in the peripheral 25 μm of the lateral extrastriola (Peterson and Rowe 2001; Peterson and M. H. Rowe, unpublished data). We conclude that 1) type I hair cells have higher stereocilia numbers than striolar type II hair cells, 2) striolar type II hair cells have higher stereocilia counts than type II hair cells in the extrastriola, and therefore, 3) type I hair cells have the highest stereocilia numbers in the utricle.
We considered the possibility that the small type II hair cells in zone 3 (Fig. 2B) represent developing bundles, but this seems unlikely for three reasons. First, we and others (Severinsen et al. 2003) have observed apparently immature bundles in turtle utricle, and they are unlike type II bundles in zone 3 because they appear to have fewer stereocilia, smaller apical surfaces, and shorter kinocilia and stereocilia (Xue and Peterson 2004). Second, such apparently immature bundles are more prevalent in the lateral and medial extrastriola than in the striola. This is consistent with observations that bundles are added to the utricular macula much more frequently in the extrastriola than in the striola (cf. Fig 6b of Severinsen et al. 2003). Third, as noted above, type II bundles in zone 3 appear to be part of a gradient in which stereocilia numbers decrease (present results) and the ratio of kinocilia to stereocilia heights increases (Xue and Peterson 2004) from striolar zone 2 into the medial extrastriola.
Stereocilia numbers matter because they affect bundle mechanics and transduction currents. The stiffness of a ciliary bundle, and thus the range of head accelerations it encodes, is approximated by the formula where K is the static linear stiffness of the bundle, N is the number of stereocilia in the bundle, and k is the linear stiffness of a single stereocilium (Howard et al. 1988). Thus stereocilia number is a major (but not the only) determinant of bundle stiffness. Preliminary models of turtle utricular bundles confirm this effect of stereocilia number on steady-state stiffness (Silber et al. 2004).
Stereocilia number is also likely to affect transduction current amplitudes because each stereocilium is associated with one to two mechanotransduction channels (Denk et al. 1995). Interestingly, a preliminary report suggests that peak mechanoelectric transduction currents in type I hair cells of turtle utricle are 492 ± 216 pA compared with 149 ± 47 pA for striolar type II hair cells (Rennie and Ricci 2004). It is unclear why this difference in transduction currents (>3:1) is greater than the difference in stereocilia counts (<2:1); possible reasons include sampling bias and differences in the number of channels per stereocilium or in single transduction channel currents. In turtle auditory hair cells, single channel currents depend on the Ca2+ concentration of the bathing medium, and they decrease with the characteristic frequency of the hair cell (Fig. 5 of Ricci et al. 2003). If this frequency dependence extrapolates to the response range of turtle vestibular hair cells, one would predict a channel current of <3–7 pA (depending on Ca2+ concentration) for these low-frequency receptors and a maximum transduction current of 300–700 pA for type I hair cells (assuming 100 stereocilia, each with 1 channel). This range includes the measured values, but such predictions are uncertain because we do not know single channel currents or the number of channels per stereocilium for turtle utricular hair cells. A related question, as yet unanswered, is whether differences in stereocilia number will affect overall current-displacement relations.
Taken together, these considerations suggest that type I and type II hair cells will differ in bundle mechanics and maximum mechanotransduction currents because they differ markedly in stereocilia numbers. Thus these two hair cell types may play distinctive roles in encoding head movement.
This work was supported by National Institute on Deafness and Other Communication Disorders Grant DC-05063 to E. H. Peterson and a Provost's Undergraduate Research Fellowship and Student Enhancement Award to W. J. Moravec.
We thank G. Monk Adams for invaluable technical assistance, Dr. M. H. Rowe for writing the MATLAB routine for coordinate conversions, and J. Xue for the confocal micrograph in Fig. 1A. Drs. R. A. Eatock, J. W. Grant, and M. H. Rowe reviewed the manuscript and provided helpful comments.
↵1 The Supplemental Material for this article (a video) is available online at http://jn.physiology.org/cgi/content/full/00428.2004/DC1.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 by the American Physiological Society