We made flattened neuroepithelial preparations of horizontal and vertical (anterior and posterior) cristae from mouse, rat, gerbil, guinea pig, chinchilla, and tree squirrel. Calretinin immunohistochemistry was used to label the calyx class of afferents. Because these afferents are restricted to the central zone of the crista, their distribution allowed us to delineate this zone. In addition to calyx afferents, calretinin also labels ∼5% of type I hair cells and 20% of type II hair cells throughout the mouse and rat crista epithelium. Measurements of the dimensions of the cristae and counts of hair cells and calyx afferents were determined on all species. Numbers of calyx afferents, hair cells, area, length, and width of the sensory epithelium increase from mouse to tree squirrel. As in the companion paper, we obtained additional data on vestibular end organ dimensions from the literature to construct a power law function describing the relationship between crista surface area and body weight. The vertical cristae of the mouse, rat, and gerbil have an eminentia cruciatum, a region located transversely along the midpoint of the sensory organ and consisting of nonsensory cells. Apart from this eminentia cruciatum, there are no statistical differences between horizontal and vertical cristae with regard to area, width, length, the number and type of hair cells, and number of calretinin-labeled calyx afferents.
The crista ampullaris was first described in mammals as a saddle-shaped organ by Scarpa (1789). Retzius (1881, 1884) was the first to describe the pattern of hair cells and supporting cells in cristae of the cat and rabbit. An anatomical feature in vertical cristae in reptiles, birds, and some mammals, called the septum cruciatum, was originally described in reptiles and birds by Steifensand (1835), the person who also first named the planum semilunatum. The septum cruciatum in birds (Igarashi and Yoshinobu 1966; Landolt et al. 1972) and bats (Ramprashad et al. 1980) is an expansive plate-like projection emanating from the midpoint of the crista. It is referred to as an eminentia cruciatum in the vertical cristae of species where it is less prominent, including the mouse, rat, gerbil (present study), cat, otter (Retzius 1884), and dog, and it is not found in primates (Igarashi and Yoshinobu 1966; cf. Lewis et al. 1985). In reptiles and birds, a distinct swelling on top of the septum or eminentium cruciatum is termed the torus (Brichta and Peterson 1994; de Burlet 1935; Landolt et al. 1972; Scarpa 1789). Both the septum cruciatum and eminentia cruciatum contain neither hair cells nor nerve fibers, only supporting cells. Lindeman (1969b) studied the structure of the crista using flattened preparations prepared by microdissection of the sensory epithelium from the underlying stroma. Based on his survey of hair-cell density, Lindeman (1969b) was able to divide the crista into three regions of approximately equal area. The innermost zone is called the central zone, the outermost zone is called the peripheral zone. The two zones are separated by a transitional area called the intermediate zone.
In mammals, type I and type II hair cells are located throughout the sensory epithelium. There are three morphologically distinct types of afferents that innervate these hair cells and carry information about head rotation to the brain (Fernández et al. 1988). There are calyx afferents innervating type I hair cells, bouton afferents innervating type II hair cells, and dimorphic afferents innervating both types of hair cells. The three groups of afferents differ in their regional distribution (Fernández et al. 1988), in their physiology (Baird et al. 1988), and in their immunocytochemical profiles (Desmadryl and Deschesne 1992; Kevetter and Leonard 2002; Lysakowski et al. 1999). In this, as in the companion paper, we focus on the calyx class of vestibular afferents.
Although earlier papers (Dechesne et al. 1991; Rogers 1989) had shown that calretinin labels some subset of vestibular afferents, Desmadryl and Dechesne (1992) were the first to make the correlation that calretinin antibody selectively labels the calyx class of afferents in the guinea pig and chinchilla cristae. Lysakowski and Goldberg (1997), in examining the ultrastructure of the crista, found that the zonal boundaries delineated previously by Fernández et al. (1988) were inadequate to describe the regional variations in ultrastructure found in their study, that in fact the central apical zone was narrower than in the Fernández study. A similar suggestion was made by Purcell and Perachio (1997) and by Leonard and Kevetter (2002). We wanted to know if our original suggestion of a redefined central zone in the crista was robust enough to extend to other species, so in this paper, we adopted a comparative approach, examining several rodent species and obtaining detailed quantitative data in several specimens from each species. We found that the calyx-containing region corresponds to a central zone with low hair-cell density and a high proportion of type I hair cells.
Our cristae were enzymatically flattened to minimize their significant three-dimensional component. This allowed us to make a type of surface preparation and to perform the analysis directly on the sensory epithelium instead of having to reconstruct it from sectioned material. The combination of this and other techniques allowed us to quantitatively describe the morphology of horizontal and vertical cristae in several rodent species, including those commonly used for vestibular research. The material in this publication is based in part on a thesis presented by S. S. Desai in partial fulfillment of the requirements for the doctoral degree at the Graduate College of the University of Illinois at Chicago. We have previously presented this research in abstract form (Desai et al. 2003).
At least eight horizontal and eight vertical cristae were obtained from male and female wild-type adults from six species representing the three suborders of Rodentia: Myomorpha: mouse, rat, gerbil; Caviomorpha: guinea pig, chinchilla; and Scuriomorpha: tree squirrel. Tissue was obtained as previously described (Desai et al. 2005). All animal tissue was harvested in accordance with the University of Illinois at Chicago Institutional Animal Care and Use Committee (IACUC) approved protocols.
Details on the immunohistochemical staining, postfixation, embedding, and quantitative methods are given in the companion paper (Desai et al. 2005).
After immunochemistry, three horizontal and three vertical cristae from each species were flattened using protease X (Sigma, St. Louis, MO). Protease X, also known as thermolysin, was first characterized by Matsubara et al. (1965) and subsequently used in unfixed otolith organ tissue to cleave specific amino acid sequences found at the level of the basement membrane of the sensory epithelium just above the underlying stroma (Corwin et al. 1995; Suzuki et al. 1992). Some of the cristae used in this study had been stored in fixative for at least two years before they were incubated in protease X. We have successfully used this technique in recently fixed tissue as well.
Whole organs were incubated for 12 h at 55°C in a 1% aqueous solution of protease X (Sigma) containing 0.15% CaCl2. If the sensory epithelium was not already floating at the top of the protease solution, it was easily dissociated from the stroma using a pair of fine forceps and flattened by coverslipping in glycerol on a clean glass slide. Sensory organs were then photographed and digitized for quantitative morphometry. These morphometric measurements were not corrected for shrinkage artifact, which we have previously determined to be ∼10%. Camera lucida drawings were made of the labeled calyx afferents in each sensory organ.
Western blots were done as previously described (Desai et al. 2005). For this study, cristae, vestibular ganglion, cerebellum (positive control), and kidney (negative control) were harvested from adult mice, rats, and chinchillas within 10 min of the animal's death.
All measurements were done using Sigma Scan Pro (Jandel Scientific, San Rafael, CA) as previously described (Desai et al. 2005). Measurements of area, length, and width of the crista are illustrated in Fig. 2 for a standard reconstruction taken from the chinchilla. Quantitative comparisons and regressions were done with Igor Pro (Wavemetrics, Lake Oswego, OR). The crista was divided into two regions, the central zone (CZ) which contained >99% of the calretinin-stained calyx afferents (shaded region), and the peripheral zone (PZ), which surrounds the central zone; these borders were verified by examining specimens in the microscope.
The location and degree of complexity of each calyx afferent, and the location of every calretinin-labeled type I and II hair cell in the mouse and rat were drawn with camera lucida, photographed, prepared for computer-assisted morphometric analysis, and coded for complexity as previously described (Desai et al. 2005).
The species in this study can, with the exception of the rat, be ordered by their body weights (BW; Table 1). Based on the latter, the gerbil and rat have been interchanged.
Comparative crista morphology
Previously, the crista was divided into three zones: central, intermediate, and peripheral. In this study, the central zone was defined as the region containing the vast majority of the calretinin-labeled calyx afferents. Based on calretinin staining patterns and hair-cell density, we could not distinguish separate intermediate and peripheral zones in this study, hence the remaining sensory epithelium was considered the peripheral zone. The area of each region was measured. ACZ and ASE increase from mouse, rat, gerbil, guinea pig, chinchilla, to tree squirrel (Table 1 and Fig. 3) and the area of the central zone (ACZ) is approximately one-third (range, 31–34%) of the entire sensory epithelium area (ASE) (Table 1). Note particularly in Fig. 3 that although the rat weighs 3.5 times more than the gerbil, their cristae are similar in size.
Every measure of the central zone increases in these six rodent species as their body weight increases. The only exceptions to this trend are, for the rat, the length of the central zone, and for the guinea pig, the width of the central zone and the width of the crista at the midpoint.
Other general features to note in Table 1 are that the central zone is about half of the width of the crista at the midpoint, and about one-third of the width at the planum, the widest point of the crista. The central zone ends near the widest point of the planum region (Fig. 2).
Similar to our analysis in macular organs, the area of the crista sensory epithelia in mammals can be predicted by body weight. This can be seen in Fig. 3, which plots ASE versus body weight (BW) in log-log coordinates for the six species of the present study, to which were added values from the literature for the lemur, hare, cat, dog, sheep, pig, dugong (a manatee-like animal), horse (Gray 1907), squirrel monkey (Fernández et al. 1995), rhesus monkey (Igarashi et al. 1975), and humans (Merchant et al. 2000; Rasmussen 1940; Rosenhall 1972; Watanuki and Schuknecht 1976).
The data fell into two groups based on body weight. Points for the smaller animals (BW ≤ 4 kg) can be fit with a power law, ASE = aBWb where a = 0.025 ± 0.013 and b = 0.44 ± 0.07 (means ± SD). The only small animal to deviate from the power law is the rat, but the deviation failed to reach statistical significance. For larger animals, ASE increases only 1.4 times even though body weight increases 300-fold.
The trends for the cristae are remarkably similar to those found in the preceding paper for the otolith organs. The slope for the power-law relationship between BW and ASE in the crista is 0.44, very similar to that of the macular organs (saccular macula = 0.45, utricular macula = 0.44, see companion paper). Large animals (>4 kg) are found on the flat portion of the graph in Fig. 3. As illustrated in Fig. 3B, the areas of the crista, the saccular macula, and the utricular macula parallel one another. Taking areal ratios, the area of the saccular macula is 2.2 ± 0.13 times that of the crista. Ratios are similar for small and large animals. However, the ratio of areas of the utricular macula is disproportionately larger than the saccular macula for larger animals (3.1 ± 0.17 vs. 2.1 ± 0.08). As implied by these values, the utricular macula is larger than the saccular macula for species >5 kg (paired t-test, P < 0.001).
In the mouse, rat, and gerbil, there is an eminentia cruciatum located near the midpoint of each vertical crista. Based on our semi-thin sections, the eminentia cruciatum contains one or two layers of nonsensory cells and is devoid of hair cells and afferent innervation. This gives the region a more translucent appearance in flattened preparations (Fig. 4). There is also evidence that the eminentia cruciatum is devoid of efferent innervation (Purcell and Perachio 1997). This nonsensory region has a longitudinal dimension of 50 ± 2 μm in the mouse, 20 ± 1 μm in the rat, and 30 ± 1 μm in the gerbil. Transversely, the eminentia cruciatum extends slightly into the peripheral zone in the mouse, having a width of 103 ± 18 μm. It is restricted to the central zone in the rat (75 ± 5 μm) and extends to the edges of the sensory epithelium in the gerbil (225 ± 5 μm).
The eminentia cruciatum was not considered part of the central zone because it does not contain any calretinin-labeled calyx afferents; this is reflected in the relevant central zone measurements (Fig. 5). However, it was included in the measurements of the sensory epithelium.
The eminentia occupies ∼3% of the area of the sensory epithelium in the mouse, 2% in the rat, and 7% in the gerbil. The mouse has 186 ± 12 nonsensory cells in this region, the rat has 59 ± 4, and the gerbil has 953 ± 38. Regardless of these differences in morphology among the horizontal and vertical cristae of the mouse, rat, and gerbil, there are no statistically significant differences between the two types of crista (P < 0.01) with regard to any of their morphometric attributes. For this reason, values have been pooled for horizontal and vertical cristae (Table 1).
Hair cell and supporting cell counts
The numbers of type I, type II, and supporting cells increase from mouse to squirrel (Table 2). The number of type I hair cells ranged from 761 ± 57 in the mouse cristae to 3310 ± 94 in tree squirrel; the number of type II hair cells ranged from 653 ± 61 in the mouse cristae to 3,025 ± 16 in tree squirrel; and the number of supporting cells ranged from 1,858 ± 220 in mouse cristae to 7,587 ± 165 in tree squirrel (Table 2). Values are given by region in Table 2.
While the area of the central zone was approximately one-third that of the entire sensory epithelium, the total number of hair cells in the central zone varied between 20% (guinea pig) and 29% (gerbil) of the total. Supporting cell numbers varied between 18% (guinea pig) and 26% (gerbil) of the total (Table 2). The density of hair cells and supporting cells is ∼30% less in the central zone compared with the peripheral zone (P < 0.01). The density of hair cells per 100 μm2 in the central zone ranged from 0.82 ± 0.03 in the tree squirrel to 2.33 ± 0.04 in the gerbil. The density of hair cells per 100 μm2 in the peripheral zone ranged from 1.14 ± 0.04 in the tree squirrel to 2.80 ± 0.05 in the rat (Table 3). Supporting cells outnumbered hair cells 1.2:1 (Table 3). The density of supporting cells per 100 μm2 is also less in the central zone versus the peripheral zone (P < 0.01).
The ratio of type I to type II hair cells varies by region. Type I hair cells outnumber type II hair cells in the central zone, ranging from 1.2:1 in chinchilla and tree squirrel cristae to 1.8:1 in guinea pig cristae (Table 3). The difference is less apparent in the peripheral zone, where the ratio ranges from 0.76:1 in the guinea pig to ∼1:1 in the mouse and tree squirrel (Table 3). The ratio of type I to type II hair cells in the central zone is significantly greater than the ratio in the peripheral zone (P < 0.01).
Number of calyx afferents
The number of calyx afferents and the type I hair cells they contact are similar between the horizontal and vertical cristae (Table 4, Fig. 5). The number of calyx afferents increases from the mouse (58 ± 3) to the tree squirrel (326 ± 14); this increase is proportional to the increase in area of the central zone.
Calyx afferents were broken down by the number of hair cells each terminal contacts (Table 4). Calyx afferents in the mouse contact up to three type I hair cells, up to four type I hair cells in the rat, gerbil and chinchilla, and up to five type I hair cells in the guinea pig and tree squirrel. In all rodents studied, with the exception of the tree squirrel, approximately half of the calyx afferents contact only one hair cell (Table 4).
The number of type I hair cells contacted by calyx afferents increased from mouse (95 ± 4) to tree squirrel (601 ± 23). The density of type I hair cells contacted by calyx afferents per 100 μm2 ranged from 0.27 ± 0.01 in the guinea pig to 0.48 ± 0.02 in the rat (Table 4). The number of type I hair cells contacted by calyx afferents can be compared with the total number of type I hair cells in the central zone: between 32% (gerbil) and 66% (tree squirrel) of type I hair cells in the central zone are contacted by calyx afferents. The remaining type I hair cells belong to dimorphic afferents (Fernández et al. 1988). On average, ∼40% of all type I hair cells in the central zone are contacted by calyx afferents.
Hair cell staining in the mouse and rat
In the rat macular organs, calretinin-labeling of type II hair cells in the extrastriola region has been noted (Dechesne et al. 1991; Sans et al. 2001). We found that in the mouse and rat crista, ∼5% of type I hair cells and 15–20% of type II hair cells are calretinin-labeled (Table 2 and Fig. 3). No other species had any calretinin-labeled type I or type II hair cells, and no species had calretinin-labeled dimorphic or bouton afferents (Fig. 3). Results for either hair cell type are similar for the central zone and peripheral zone (Table 2 and Fig. 3). When the corresponding zones for the crista and the maculae are compared (see Desai et al. 2005), however, percentages of labeled type II hair cells are higher in the macular central region (striola vs. crista central: 35 vs. 17%) and, especially, in the macular periphery (extrastriola vs. crista peripheral: 70–80 vs. 15–20%). As described in the companion paper, Western blots for calretinin confirmed the specificity of our calretinin antibody (Fig. 6).
Based on hair-cell density, Lindeman (1969b) divided the crista into three zones of almost equal areas. Fernández et al. (1988) used simple arithmetic proportions, (±1/√3, ±1/√3), (±√2/√3, ±√2/√3), and (±1, ±1), to set the corners of the central, intermediate, and peripheral zones of a “standard reconstruction.” While this approach has been used in subsequent studies, it is important to note that the reconstruction was based on formal, rather than anatomical, criteria. Because of its formal nature, it would be surprising if the standard reconstruction precisely matched boundaries based on structural considerations. One indication that there was a need for revision came by considering the distribution of calretinin-positive calyx fibers. These fibers fell into a zone that was longer and narrower than the original central zone. A second indication arose when we considered hair cell densities for the original three zones. Values for the intermediate and peripheral zones were more similar to each other and different from those for the central zone. Thus there was little reason in this study to consider the intermediate zone separately from the peripheral zone (see, however, discussion below on Intermediate and juxtastriolar zones). These considerations led us to define the central zone based on the staining pattern of calretinin-labeled calyx afferents and to combine the remaining parts of the neuroepithelium into a single peripheral zone. It is interesting to note that although Lindeman presented data from the crista in terms of three regions, he shows a sketch (Fig. 19A in Lindeman 1969b) and mentions in his discussion that it is more appropriate to divide cristae into only two regions, central and peripheral, due to the reaction of the sensory epithelium to aminoglycoside toxicity such as streptomycin and kanamycin (Lindeman 1966, 1969a, b). Recent work using gentamicin-treated animals has confirmed Lindeman's findings; it appears that type I hair cells in the central zone of cristae and the analogous striolar region of macular organs show the greatest damage after treatment with ototoxic antibiotics (Lopez et al. 1997; Twine 1985).
Boundaries of the central zone
In the guinea pig crista, Lindeman's sensory epithelium zones (Lindeman 1969b) were three regions of approximately equal area (central zone 29.5%, intermediate zone 38%, peripheral zone 32.5%). Our delineation of the central zone in guinea pig cristae, as defined by calretinin immunostaining, is similar to that of Lindeman (1969b). Our central zone in the guinea pig cristae is just 2% longer and 1% wider than that found by Lindeman (Tables 2 and 5).
For the chinchilla crista, our results are also consistent with, but not identical to, earlier reports. Fernández et al. (1988) plotted 41 calyx fibers they had labeled via extracellular HRP injections in the vestibular nerve on their standard reconstruction. Thirty six of those calyx afferents were located in their central zone, 4 were located in their intermediate zone (3 of these would fall within our longer, narrower central zone), and 1 was located in their peripheral zone (outside our central zone). In their study, Fernández and colleagues extrapolated their calyx data to estimate that there were 128 total calyx afferents in chinchilla crista. In a later study (Fernández et al. 1995), chinchilla hair cell counts were used and the estimate was revised to 250 calyx afferents. If the central zone in their study is defined as the region containing the majority of their labeled calyces, the dimensions of such an area are comparable to those found in the present study (Table 5).
Desmadryl and Dechesne (1992) were the first to use calretinin immunochemistry as a neurochemical method to identify this anatomically and physiologically distinct class of afferents. They were able to estimate that there were a total of 122 calyx afferents (Table 5). Our data, using a flattened surface preparation that allowed us to pinpoint the location of every calyx afferent, indicate that the actual number of calretinin-labeled calyx afferents in crista is closer to 207. Our method allowed us to easily separate adjacent calyx terminals and precisely count the number of type I hair cells contacted by each afferent.
In our ultrastructural study (Lysakowski and Goldberg 1997), we examined the cytoarchitectonic and synaptic features of type I and II hair cells by region, using the standard reconstruction of Fernández et al. (1988) to delineate zonal boundaries in the crista. There were indications that a division based on this reconstruction did not fit the ultrastructural data, thus we further divided the central zone along the transverse axis of the crista into an inner part called the “central apical” zone (CA) and a surrounding region called the “central slope” (CS).
Within the CA, hair-cell density was lower and type I hair cells were wider, shorter and had thicker calyces, more synaptic ribbons and calyceal invaginations (Lysakowski and Goldberg 1997). CA also differed from CS because type I hair cells in complex calyces, which were much more common in CA and had many more ribbon synapses and calyceal invaginations, while those in simple calyces (more common in CS) had more outer face ribbon synapses. We extended the analysis into the intermediate zone near the planum, IP, which corresponds to the longitudinal ends of the central zone in the present study. IP could be distinguished in the ultrastructural study because, similar to CS, simple calyces in IP also had more outer face ribbon synapses. We found that the intermediate and peripheral zones, in general, were more alike than they were different when considering these traits. Type II hair cells follow the trends found in type I hair cells except that type II hair cells were longer and synaptic ribbons were larger in the central apical zone (Lysakowski and Goldberg 1997). Because the central apical region defined by Lysakowski and Goldberg (1997) is fully contained within the region of the calretinin-labeled calyx afferents, one would expect to find the ultrastructural features described in the previous study within the central zone defined in the present study.
Further supporting evidence for divisions based on calretinin-labeling of calyx afferents comes from a study of efferent innervation in the gerbil. Purcell and Perachio (1997) determined the innervation patterns of ipsilaterally and contralaterally projecting vestibular efferent neurons in gerbil cristae. They reconstructed efferent terminal fields in 28 gerbil cristae. Although the dimensions of their central zone were initially determined using equal areas similar to the method used by Fernández et al. (1988), they redefined their central zone as the region that did not contain the terminal fields of contralaterally projecting efferent neurons. As a result, their central zone is very similar to that obtained in the gerbil crista with calretinin staining in the present study. They stated that the ipsilaterally projecting efferent neurons preferentially innervate their redefined central zone (Purcell and Perachio 1997).
Finally, recent preliminary reports (Lysakowski and Price 2002; Syeda et al. 2002) indicate that the distribution of two different potassium channels, KCNQ4 and Herg1, respect the boundaries of the central zone as delineated by calretinin-labeling of calyx afferents. KCNQ4 is heaviest within, and Herg1 is heaviest outside, this central zone.
Intermediate and juxtastriolar zones
Although we have just suggested a reorganization of the rodent vestibular sensory epithelium into two (central/peripheral and striolar/extrastriolar) rather than three (central/intermediate/ peripheral and striolar/juxtastriolar/extrastriolar) zones, we also propose that this division, although clear and well-delineated, may not be complete. There is compelling evidence from recent studies for an intermediate or transitional zone just outside the central zone as defined in this study.
First, Fernández and colleagues (1988, 1990) provided evidence for a transitional zone within which only dimorphic afferents terminate. Second, Fernández et al. (1990) provided evidence for a juxtastriolar zone, a zone transitional between the striola and extrastriola. Dimorphic afferents in the juxtastriolar zone had thicker axons, smaller terminal-field diameters, fewer complex calyces, and fewer boutons per afferent than those in the extrastriola. On the other hand, juxtastriolar dimorphic afferents also had more bifurcations than those in the striola comparable in number to those in the extrastriola. The latter study also described a sharp morphological boundary between the juxtastriola and the striola, which we have found to be co-extensive with the area of calyx innervation in the present study (companion paper). The crista has similar hair-cell densities in the intermediate and peripheral zone in both the monkey and chinchilla, although the ratio of type I to II hair cells is more similar between the central and intermediate zone in the monkey (Fernández et al. 1995).
Finally, recent studies in the gerbil by Leonard and Kevetter (2002) provide evidence for distinct boundaries between the juxtastriolar and extrastriolar zones using calbindin antibody staining. This last finding is particularly interesting because the studies of Fernández and colleagues (1990) were unable to distinguish this boundary, not having such chemical markers. Peripherin may also be useful to distinguish this boundary and the intermediate/peripheral boundary (Leonard and Kevetter 2002; Lysakowski et al. 1999). If the presence of a distinct boundary can be verified in a number of species, it could be hypothesized that the intermediate/ juxtastriolar zone serves a distinct purpose, as yet unknown. Thus we feel that there may be good reason to preserve the intermediate zone as a distinct region, although we do not yet have a unique marker for it.
We compared the number of calyx afferents stained in our study with the total number of ganglion cells in the mouse, gerbil, and chinchilla to determine the contribution of calyx afferents to the overall number of afferents innervating the crista. Given a population of ganglion cells, approximately equal numbers of afferents emanating from these cell bodies innervate each of the five vestibular organs in the guinea pig (Gacek and Rasmussen 1961) and the chinchilla (Boord and Rasmussen 1958); in the quantitative reconstructions in the following text, we assume this to be the case for the mouse and gerbil.
In B6CBA and C57BL/6J mice, there are ∼3,400 total ganglion cells with an estimated 680 vestibular afferents contacting each of the five sensory organs (Bäurle and Guldin 1998). Of 680 afferents innervating each crista ampullaris, we found that 58 ± 3 (8.5%, mean ± SE) are calyx afferents. In the gerbil, there are 3,962 ± 358 (mean ± SD) ganglion cells with an estimated 792 afferents innervating each vestibular organ (Leonard and Kevetter 2002). We found 100 ± 4 calyx afferents in the gerbil, ∼13% of the estimated total number of afferents innervating the crista ampullaris. In the chinchilla, there are 7,772 total afferents (Boord and Rasmussen 1958) with ∼2,100 innervating each crista (Fernández et al. 1995; Hoffman and Honrubia 2002). We found 207 ± 4 (mean ± SE) calyx afferents, again ∼10% of the estimated total number of afferents innervating the crista. Thus in each species for which we have afferent counts, ∼10% of the vestibular afferents to the crista are calyx afferents.
Comparison to macular organs
Even though the area of the sensory epithelium in cristae is smaller by one- to two-thirds compared with the macular organs (see companion paper), the area of the central zone in crista is similar to the area of the striola in macular organs. The extrastriola, however, is four times larger in area than the peripheral zone of the crista, both in terms of percentage and absolute area. Thus one might suspect that the terminal fields and the number of calyx and bouton endings are larger in the extrastriola than in the peripheral zone. Terminal field diameters are larger (62 ± 29 in the extrastriola vs. 48 ± 24 in peripheral zone) and the numbers of type I cells innervated are larger (2.8 ± 1.6 in the extrastriola vs. 1.7 ± 1 in the periphery) but neither is proportionately larger (Fernández et al. 1990). Moreover, the numbers of boutons per peripheral dimorphic afferent and per bouton afferent are equal in both organs (23 and 34, respectively), so it has been suggested (Fernández et al. 1990) that there are at least two arrangements responsible for the higher innervation ratio in the macula: individual type II hair cells could be provided with fewer bouton endings or each afferent could make more bouton endings. The conclusion of the 1990 paper was that it was apparently the second arrangement that occurred and that the average number of bouton endings per hair cell was the same in the crista and macula. Ultrastructural data from the chinchilla utricular macula suggest that the first arrangement is also tenable, as there are ∼12 boutons per type II hair cell (Lysakowski, unpublished observations), rather than the 23 suggested by the quantitative reconstruction of Fernandez et al. (1990).
Although cristae have significantly fewer hair cells than macular organs (companion paper), the number of afferent fibers innervating cristae is approximately the same as that innervating macular organs (Boord and Rasmussen 1958; Gacek and Rasmussen 1961). This also suggests that afferent sensory information coming from the cristae has more resolution than it does in macular organs.
We did cell counts and measured surface areas of the otolithic maculae and the cristae ampullares to obtain morphometric data and to determine the proportion of hair cells that are calretinin-positive. When each of the morphometric measures in these closely related (rodent) species appeared to increase with body weight, we were curious as to whether they would continue to increase in very large animals, e.g., would a horse or a cow have a much larger sensory epithelium? So, we went to the literature to obtain more data from other species (Gray 1907; Igarashi et al. 1975; Kirkegaard and Jorgensen 2001; Lindenlaub et al. 1995; Retzius 1884; Watanuki and Schuknecht 1976; see also companion paper). We found that in small animals (<4 kg), such as the rodents used in our study, the surface area of the three sensory organs increased with body weight but that the surface area of all three organs reached a plateau in larger animals.
It is well-known that a power law relationship of brain to body weight increases continuously in mammals with an exponent of 0.75 (Jerison 1986; Martin 1983). We knew of this relationship and wanted to see if it also held for the animals used in our study. It did (see following text). Thus our finding of an exponent of ≈0.5 for crista surface area versus body weight was unexpected. Similarly unexpected was that the slope would asymptote >4 kg of body weight. A further correlation between the three types of organs showed that the saccular macula increases proportionately compared with the crista, as does the utricular macula, but that the utricular macula becomes disproportionately larger >4 kg. A similar suggestion had been made by (Watanuki and Schuknecht 1976), based on comparing human to guinea pig and squirrel monkey material.
We considered two explanations for the increase in the surface areas of the vestibular endorgans with body mass. The first was architectural. Simply put, small animals have small skulls, and it makes sense that a small skull could not accommodate the labyrinth of, e.g., an elephant. If the overall dimensions of the labyrinth are constrained in such a manner, it is likely that the sensory epithelia would be similarly constrained. To explore this idea, we needed to know how the skull and labyrinth grow with body weight. Labyrinthine dimensions for a large number of diverse species have been studied in two major works (Jones and Spells 1963; Spoor and Zonneveld 1998). Jones and Spells (1963) measured the radius (R) of curvature of the membranous semicircular canals and found that it varied with body mass (BM) according to a power law, R = aBM0.1. Spoor and Zonneveld (1998) obtained a similar coefficient, 0.14. From the geometry of the labyrinth, the canal ducts are the most prominent elements and it is likely that the overall dimensions obey a similar power law. At the same time, the linear dimensions of the skull should grow in proportion to those of the brain. One notable exception was the cetaceans, whose semicircular canals and overall labyrinth dimensions are small for their body mass (Spoor et al. 2002). Brain volume grows as BM0.75 (Jerison 1986; Martin 1983), so the growth in linear dimensions should be proportional to BM0.25. If there was an architectural constraint, one might expect that the power laws relating labyrinth and skull sizes should have similar exponents, which they do not. Another difficulty is that, while the skull continues to grow with body mass, neuroepithelial surface areas reach an almost constant value for animals >4 kg (present studies).
The second explanation supposes that the neuroepithelial cross-sectional area is matched to the volume of the vestibular nuclei. From available fiber counts (Hoffman and Honrubia 2002; Velazquez-Villasenor et al. 2000), the number of afferent fibers should grow with animal size. It is known that the growth of the brain stem (specifically the vestibular nuclear complex) with body mass parallels that of the brain (Matano 1986; Stephan et al. 1981, 1991). One might expect that the vestibular nuclei would share in this growth and that, as a result, more afferent fibers would be needed to innervate the increased volume of their central targets. To explore this idea, we compared the surface area of the crista neuroepithelium (ASEcrista) and the volumes of the brain (BR) and vestibular nuclei (VVN) with body mass (Blinkov and Ponomarev 1965; Crile and Quiring 1940; Matano 1986; Ramprashad et al. 1984; Stephan et al. 1981, 1991; Suarez et al. 1997; Yoshikawa 1968). Because of the relatively proportional growth of all five organs, the growth of the crista can be taken as representative of that of the entire labyrinth. For the animals in our series, the brain volume grows as a power law, VBR = BM0.78±0.09 (Fig. 7A). The exponent is similar to that found in two larger series of mammals, i.e., 0.75 (Jerison 1986; Martin 1983). Brain volume continues to grow with body mass throughout the series. In contrast, both the surface area of the crista and the vestibular nuclear volume grow with body mass for small animals, but reach an asymptote for large animals. A power-law regression with BM for the small animals in this analysis (BM <4 kg) gave an exponent for ASEcrista (0.49 ± 0.086) that was significantly smaller than the exponent for VVN (0.75 ± 0.13; Fig. 7A). To determine the relation between ASEcrista and VVN, we plotted the two variables against one another (Fig. 7B) and obtained VVN = (80.9 ± 1.16)ASEcrista1.53±0.13 (r2 = 0.95). The relation holds for both large and small animals. An exponent greater than unity implies that VVN grows faster than ASEcrista. We take this to mean that, as the vestibular nuclei grow with body weight, each vestibular afferent has a more extensive set of central terminations and covers a larger volume of the vestibular nuclei. That the central terminations expand makes sense since many vestibular afferents extend to the borders of the vestibular nuclei (Sato et al. 1989). The vestibular nuclei grow in volume with body weight among small animals. This requires that the innervation provided by the vestibular nerve also increase. Taking the reciprocal of the power-law exponent of 1.5, i.e., 0.67, we can suggest that two-thirds of the increase in innervation may be due to an increase in the number of vestibular afferents and that 1/3 may be attributed to the more extensive innervation provided by individual afferents. Further studies may test this hypothesis.
The crista central zone in rodents can be morphologically delineated as a region having more type I hair cells (Spoendlin 1965; Wersäll 1956), lower hair cell and supporting cell density (Lindeman 1969a; Lysakowski and Goldberg 1997), shorter and wider type I and type II hair cells (Lysakowski and Goldberg 1997), more complex calyx terminals (Engström and Wersäll 1958; Lysakowski and Goldberg 1997), thicker fibers (Cajal 1908; Lorente de Nó 1926; Wersäll 1956), and no contralaterally projecting efferent fibers (at least in the gerbil, see Purcell and Perachio 1997). The results of this paper and the companion paper show that the central zone of cristae and the striolar zones of the macular organs can also be neurochemically identified using calretinin immunochemistry; the method is consistent with prior definitions of the central zone and striolar region.
As body weight increases, the area of the sensory epithelium increases. At a species level, this involves parallel increases in the area of the central zone, the length and width of the central zone and sensory epithelium, the number of type I and II hair cells and supporting cells and the number of calyx afferents. Horizontal and vertical cristae are more alike than they are different with regard to their morphological characteristics, number of type I and type II hair cells, supporting cells, and calretinin-stained calyx afferents, even though the vertical cristae of the mouse, rat, and gerbil possess an eminentia cruciatum. We confirm that there are differences in hair-cell density between the central zone and peripheral zone, and our study provides further evidence that the cristae are most ideally divided into two morphologically distinct regions, a central region demarcated by the presence of calretinin-labeled calyx afferents, lower cell density, and more type I hair cells, and a peripheral region lacking calretinin-labeled calyx afferents. A caveat to this suggestion, however, is that future studies may find evidence for additional subdivisions. Finally, using an enzyme-based flattening technique makes it easier to work with a sensory organ that otherwise has a very significant three-dimensional component.
This work was supported by National Institute of Deafness and Communication Disorders Grant R01 DC-02521 and National Aeronautics and Space Administration Grant NAG 2–1358, both to A. Lysakowski, and by a National Institute of Deafness and Communication Disorders Predoctoral NRSA Fellowship F30 DC-05451 to S. S. Desai.
We are very grateful for the excellent photographic and technical assistance of S. D. Price and the assistance on Western blots by D. Nahey. Dr. James Artwohl assisted us with the tree squirrels. We thank Drs. Ruth Anne Eatock and J. Christopher Holt for insightful comments on an earlier version of this manuscript. We also thank Dr. Jay M. Goldberg for helpful discussions on the allometry portion of the paper. Dr. Catherine Zeh participated in early experiments.
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 © 2005 by the American Physiological Society