Varicosities of Intraretinal Ganglion Cell Axons in Human and Nonhuman Primates

Abstract

purpose. To describe varicosities of intraretinal ganglion cell axons in the nerve fiber layer of human and nonhuman primate retinas.

methods. Intraretinal ganglion cell axons of seven human donors (1–85 years old) and two nonhuman primates (Macaca mulatta, 15 and 17 years old) were immunohistochemically stained with an antibody of neurofilament on flatmounted retinas and examined with light microscopy. In addition, the axons within the retinal nerve fiber layer were examined with transmission electron microscopy in one human and one nonhuman retina. The variations of diameters of single axons were measured on transverse- and parallel-cut sections, and the frequency distributions of the diameters were statistically evaluated.

results. Varicosities of the intraretinal ganglion cell axons were found throughout the retinas in both nonhuman primate and human eyes of all ages examined. The varicosities were rich in mitochondria and had desmosome- and hemidesmosome-like junctions with other axons and retinal glial cells. Measured on parallel-cut axons, the mean diameter (±SD) of varicosities was 2.7 ± 0.9 μm, whereas the mean diameter of intervaricosity regions was 0.7 ± 0.3 μm. The diameter distribution for transverse-cut axons was also bimodal, but the two peaks were much closer because the peak of the larger-diameter group decreased.

conclusions. The results demonstrated that intraretinal ganglion cell axons are predominantly varicose fibers in both human and nonhuman primates. Size variations exist within a single axon’s diameter and thereby affect the patterns of diameter distribution seen in transverse-cut preparations. The mitochondria-rich varicosities and the presence of intercellular junctions suggest that the varicosities may be functional sites that serve local high-energy demands of unmyelinated fibers and signal transmission.

The diameters of intraretinal ganglion cell axons vary between different retinal regions and between individual axons. ^{1 2 3 4 5} Data from previous reports, describing the frequency distribution of axonal diameter, are often skewed, with a tail of larger axons and a bimodal or trimodal distribution. ^{5 6} Variations in the intraretinal axonal size may represent purely anatomic differences, but may also underlie functional differences between retinal ganglion cells. ⁷ Indeed, axons are sometimes divided into different functional categories according to their size. ⁴  

A recent observation in normal, aged human retinas in our laboratory demonstrated wide variations in the diameters of single intraretinal ganglion cell axons along their course within the retinal nerve fiber layer (NFL). When examined by immunohistochemical techniques and light microscopy, these size variations along the course of an axon appear as a chain of bulb-shaped varicosities. The varicosities seen in our original tissue samples were frequent in number and present throughout the NFL. Under light microscopy, these varicosities have an appearance similar to bouton de passage, the chemical synapses for neuronal communications between axons and their target cells. Similar bulb-shaped varicosities of ganglion cell nerve fibers have been reported in rat, ^{8 9} rabbit, ¹⁰ and feline ^{11 12} retinas and may be sites of communication between glial cells and ganglion cell axons. ^{10 11} Investigators in these studies, all using electron microscopy, reported on only a small number of these structures in nonprimate eyes. Whether these structures are also present in the normal primate retina and how their density varies with eccentricity and age remains unknown. 

Our original observation of these varicosities has lead to further systematic investigations, which are presented in this report. The present study characterizes the morphologic features of the intraretinal ganglion cell axon varicosities, using both light and transmission electron microscopy in human donor retinas of various ages and perfusion-fixed nonhuman primate retinas. The influence of the varicosities, and the plane of retinal sectioning, on the frequency distributions of intraretinal nerve fiber diameters are also described. 

Materials and Methods

Seven human donor eyes without any recorded history of eye disease were obtained from the local branch of the Lions Eye Bank (Oregon Lions Sight & Hearing Foundation, Portland, OR). Human tissue was obtained and used in accordance with the tenets of the Declaration of Helsinki. All the eyes were fixed in 4% paraformaldehyde within approximately 3 hours after enucleation. In addition, the normal eyes from two rhesus monkeys (Macaca mulatta) were obtained after perfusion fixation. Perfusion fixation was used to help rule out postmortem artifact as a source of the axonal findings described later. These animals were involved in a separate experiment and were scheduled to be killed as part of a separate protocol. The procedures of the experiment were in accordance with to the Institutional Animal Care and Use Committee Guidelines and to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The animals were killed by intravenous injection (Euthasol; Diamond Animal Health, Inc., Des Moines, IA) and perfused immediately with 4% buffered paraformaldehyde through precannulated carotid arteries. The perfusion lasted approximately 30 to 45 minutes, and the eyes were enucleated. After the eyes were hemisected, the retinas (both human and monkey) were dissected from the eyecups by trephining around the optic nerve and cutting along ora serrata. The retinas were transferred into 0.1 M phosphate-buffered saline (PBS, pH = 7.4) after manual removal of the vitreous body. 

Neurofilament Immunohistochemical Stain on Flatmounted Retina

Either the whole retina or a wedge of the retina (extending from the peripapillary region to the periphery) was used. The tissue was washed in 0.1 M PBS containing 0.2% Triton X-100 for 1 hour and incubated in blocking serum (1% horse serum and 1% BSA) for 12 to 24 hours at 4°C. The tissue was transferred into a solution of 1:200 monoclonal mouse primary antibody to neurofilament (200 kDa; Novacastra Laboratories Ltd., Newcastle-upon-Tyne, UK) diluted with 1% bovine serum albumin for 48 hours at 4°C. The tissue was then washed in 0.01 M PBS for three changes (1 hour each). Fluorescein isothiocyanate conjugated secondary horse anti-mouse immunoglobulin (Vector Laboratories, Inc., Burlingame, CA) was incubated for another 12 to 24 hours at 4°C. After a thorough wash with 0.01 M PBS, the tissue was mounted and viewed by fluorescence microscope (LB 100T; Leica Microsystems Wetzlar GmbH, Wetzlar, Germany) or a confocal microscope (TCS SP II; Leica Microsystems Heidelberg GmbH, Heidelberg, Germany). Negative controls were performed by omitting the antiserum from the primary antibody solution. 

Transmission Electron Microscopy

Tissue blocks containing approximately 1 to 2 mm³ of retina were postfixed in 5% glutaraldehyde in PBS (pH 7.4) for 3.5 hours and rinsed in PBS before being postfixed in 2% osmium tetroxide for 3 hours. The tissue was rinsed again, dehydrated in an ethanol-acetone series and embedded in Epon 812 (Polyscience, Inc., Warrington, PA). Semithin sections (1 μm) were cut and stained with toluidine blue to assure that further thin sections (∼50 nm) could be cut either transverse or parallel in relation to the ganglion cell axons in the NFL after adjusting the angle of the tissue block on the microtome. Uranium acetate (2%) and 0.3% lead citrate were used as positive stains on the thin sections mounted on polyvinyl butyral-coated grids (Butvar; Structure Probe Inc., West Chester, PA). The sections were observed by electron microscope (EM 10; Carl Zeiss, Oberkochen, Germany). 

Quantification of the Axon Diameters and Statistics

The diameter of the intraretinal ganglion cell axons was measured in both parallel-cut and transverse-cut axons from one of the human retinas (85 years old) in a region approximately 2 to 3 mm inferior to the macular region. For parallel-cut axons, 15 semithin sections photomicrographs were taken at 100× with a digital camera attached to an image-analysis system (Bioquant; R&M Biometrics, Inc., Nashville, TN). In an attempt to measure the largest diameter of each of the bulbs, only the axons that contained both the bulbs (n = 102) and the immediate interbulb regions (n = 99), as illustrated in Figure 8A (inset), were included for the measurement. For transverse-cut axons as illustrated in Figure 8B inset, four digital electron microscopy photomicrographs were used. The measurement was made on all the axons (n = 318) in the photomicrographs, which were viewed with graphic software (Photoshop, ver. 4.0; Adobe Systems, Inc., Mountain View, CA). Using the Line Tool feature in the program, a scaled line can be drawn to measure a distance between two points—in this case, the distance between the membranes on two sides of an axon in the photomicrographs. The actual length of the axonal diameter was then calculated according to the ratio of the line scale to the magnification with which the photomicrographs were taken. 

To determine whether the mitochondria were more prevalent within the bulbs, diameters of all the transverse-cut axons containing mitochondria and those without mitochondria were measured and compared in seven electron microscopy photographs of one nonhuman primate retina. Normal distribution fitting was evaluated with the Kolmogorov-Smirnov test, and the differences between mean axon diameters were compared with unpaired Student’s t-test. 

Results

The human specimens were obtained from seven donors, who were ages 1, 2, 13, 22, 41, 75, and 85 years at the time of death. The nonhuman primates were ages 15 and 17 years. The presence of numerous varicosities along the intraretinal course of ganglion cell axons within the NFL was striking in all human and nonhuman primate specimens. The morphologic features of the intraretinal ganglion cell axon varicosities are characterized below for both light and electron microscopic studies. 

Light Microscopy

Figures 1 2 and 3 demonstrate the pattern of results observed in all specimens stained with an antibody to neurofilament and viewed from the vitreous side of the flatmounted retina under a fluorescence light microscope. There was a high density of bulb-shaped varicosities along the axons in the NFL in both human and nonhuman primate retinas. The long axis of these varicosities, or bulbs, was oriented along the axis of the axons. In more peripheral retinal regions the bulbs are seen more distinctly, probably because the NFL is thinner (e.g., compare Fig. 2D , periphery with Fig. 2A , peripapillary regions). Conversely in the peripapillary region, the bulbs were slightly harder to distinguish under the light microscope. Nevertheless, the bulbs could be seen in this region too (Fig. 2A) . 

The maximal diameter of the bulbs varied widely, even for bulbs along a single axon’s course. Some bulbs were as large as 10 μm in diameter, which may be 5 to 10 times larger than the diameters of the axons in the interbulb regions (Fig. 1) . The bulbs were divided into two types based on their appearance, by using a microfilament immunohistochemical stain and light microscopy. The first type (Fig. 1D) , was much more frequent and had the appearance of a loosely twisted rope. The space within these bulbs appeared to be filled with a fine meshwork of fibers. The second type of bulb was rarely observed, was typically larger in size, and had a more homogenous staining appearance, with fine particles or granules within the bulb (Figs. 1A 1B) . Both types of bulbs could be observed within a single fiber (Fig. 1A) . 

Varicosities in the NFL were seen in all human retinas, from specimens of all ages (1–85 years) with no obvious morphologic differences under the light microscope for different ages (Fig. 3A 3B 3C) . Although there are insufficient data in this study to formally evaluate the effect of age, there appeared to be a tendency of increased density of varicosities in the older specimens. In the nonhuman retina (Fig. 3D) , the density of the bulbs appeared similar to that of the older human retinas (Fig. 3C) . Both nonhuman primate retinas were from older animals (ages 15 and 17 years). 

In addition to the primary observation of numerous bulbs located along axons with the regularly radiating pattern toward the optic nerve, bulbs were also observed on sparse, irregularly distributed nerve fibers (Figs. 4 5) . Adjustment of the microscope’s focal plane revealed that these irregularly oriented fibers lie both superficially and deep in relation to the axons within the NFL. The superficial fibers were very rare and quite long (Fig. 4) , whereas the deeper irregularly distributed fibers were more common and short. It was impossible to determine precisely which retinal layer the latter group belonged to, only that they were located distally in relation to the bulk of the axons in the NFL, perhaps within the inner plexiform layer. 

Transmission Electron Microscopy

The varicosities of the intraretinal ganglion cell axons, in both the human and nonhuman retinas were examined with transmission electron microscopy, along the length of the parallel-cut axons (Fig. 6) . Chains of multiple bulbs along the course of a single axon were observed with electron microscopy when the section included both interbulb fibers and the bulbs themselves. This is similar to the chains visualized by light microscopy. Transverse-cut axons demonstrated a large variance of fiber diameter which is representative of the combined sampling of both the bulb regions and the interbulb regions along the population of axons (Fig. 7D) . We will elaborate on this later. 

As previously reported, astrocytes and Müller cells were present in the extra-axon space of the NFL. Astrocytes were differentiated from Müller cells by the appearance of the cytoplasm, which is less electron dense in astrocytes, as observed by Hollander et al. ¹¹ In transverse-cut axons in one of the nonhuman primate retinas, desmosome-like, or hemidesmosome-like junctions were often observed along the axolemma, adjacent to neighboring glial cells or the other axons (Fig. 7) . Similar junctions were also found between pairs of neighboring glial cells. However, no chemical synaptic vesicles were observed in the axons. 

In both the human and nonhuman retinas, mitochondria were frequently present within the bulbs. Occasionally, mitochondria located within a bulb were so long that they extended into the interbulb region. The mean diameter of the axonal regions that contained mitochondria was 1.40 ± 0.55 μm (n = 240), whereas the mean diameter of the axonal regions without mitochondria was 0.87 ± 0.33 μm (n = 132) measured on all the transverse-cut axons in seven photomicrographs taken from one nonhuman primate retina. The difference between the two groups was statistically significant (P < 0.001, t-test) 

On the 15 photomicrographs taken from the semithin sections in the human retina, the diameter of the parallel-cut axons was measured for both the bulb region and the interbulb region for axons containing both regions. The mean diameter measurement for the bulbs was 2.7 ± 0.9 μm (n = 102; range, 0.6–5.9 μm), whereas the mean diameter measurement for interbulb regions was 0.7 ± 0.3 μm (n = 99; range, 0.2–1.9 μm). The frequency distribution of the bulb diameters was not statistically different from a normal distribution (Kolmogorov-Smirnov test, P > 0.20). The distribution of interbulb diameters was skewed with a short tail on the right and thus was not a normal distribution. Figure 8A shows both frequency distributions. 

In the four photomicrographs for diameter measurement of transverse-cut axons, it was impossible to distinguish whether the section was through a bulb- or interbulb region along any given axon, although the appearance of numerous mitochondria was suggestive of a bulb location. The diameters of all the transverse-cut axons in the photomicrographs were measured, plotted, and analyzed. The frequency distribution of the axon diameters (n = 318) exhibited a bimodal pattern with two peaks at approximately 0.7 μm (65% of the total was <1.2 μm) and 1.6 μm, with a total range of 0.3 to 3.6 μm (Fig. 8B) . 

Discussion

This study demonstrates that unmyelinated intraretinal ganglion cell axons in the primate eye (human and nonhuman) contain numerous bulb-shaped varicosities at all retinal eccentricities. These varicosities were observed in all human specimens studied, ranging in age between 1 and 85 years. To limit the possibility of postmortem and fixation artifact, this finding was confirmed in two adult nonhuman primate retinas that were fixed by perfusion at the time of death. The diameter of the bulbs, or varicosities, was approximately four times larger, on average, than the interbulb regions of the axons. These bulb regions had no vesicle-containing chemical synaptic structures, but were rich in mitochondria, suggesting a possible zone of specialization and/or metabolic demand. The bulb regions also contained electron-dense membrane regions that appeared to be junctions with other axons and with glial cells. Such junctions were observed in the nonhuman retinas; however, the human tissue was of insufficient quality to evaluate such fine detail under the electron microscope. 

Previous studies have shown between- and within-region differences for diameters of intraretinal ganglion cell axons. ^{1 2 3 4 5 7} Accordingly, axons are often divided into different categories based on their size and speed of conductivity. ^{4 7} The present results demonstrated that in the primate retina the measured diameter of a single axon could vary approximately four times on average (2.7–0.7 μm). This ratio is close to the mean maximum/minimum (max/min) diameter ratio seen in the feline retina. ¹² Using consecutive electron microscopic photomicrographs, Greenberg et al. ¹² reconstructed portions of 19 intraretinal axons in feline retina (up to 0.8 mm maximum length) and found the most common max/min diameter ratio was approximately 2, with a mean ratio of approximately 4. Of course, there may be other species and/or retinal-location-dependent differences, but the general similarity is interesting. There are other previously published photomicrographs that illustrate similar variations. However, in many of these reports there are little or no specific comments about these varicosities. ^{13 14}  

The presence of these varicosities complicates the evaluation of intraretinal axonal diameters and simple classifications of large versus small axons. As demonstrated in Figure 8A , when the diameter measurement was performed on parallel-cut axons, the two peaks for the bimodal distributions of interbulb and bulb regions were 0.7 and 2.7 μm, respectively. However, if the measurement was performed on transverse-cut axons (i.e., across the axons, Fig. 8B ), the peak on the right representing the diameter of the varicosities shifted from 2.7 to 1.6 μm, whereas the peak representing the interbulb regions remained the same. The resultant distribution tended to be unimodel. The most likely explanation is that the diameter sampling from transverse-cut axons are random in regard to the distance between the location of maximal bulb diameter and the interbulb fiber region. Consequently, it is less likely that the diameter was sampled at its maximal point along a bulb, resulting in a shift toward smaller diameters, on average. In the case in which the interbulb regions are long, the distribution of the bulb diameters was undersampled, thus resulting in a more unimodal distribution of diameters with a long tail toward larger axons. These results also indicate that classification of intraretinal axons, at least in human and nonhuman primates, should not be based on diameter measurements obtained from transverse-cut axons due to the presence of these varicosities. 

The structural and functional significance of these axonal varicosities is not yet clear. Given that the bulbs were observed in all the normal human retinas from donors of widely varying ages at death (1–85 years), it is not likely that they represent pathologic and/or age-related changes. It is possible, however, that a larger study sample, with donor eyes from each of several different age groups would reveal a potential age effect. 

In the central nervous system, bulb-shaped varicosities (often found in chains along axons) are structures of chemical synapses, known as bouton de passage. ¹⁵ However, the chemical synapses of ganglion cells in the NFL were reported to be limited to the region close to the retinal ganglion cell layer. ¹⁶ The findings of the present study are in agreement, in that no chemical synaptic structures (with vesicles) were observed in the NFL axons under the electron microscope. It should be noted, however, that one of the morphologic types of bulbs defined by light microscopy (Fig. 2B) was only scarcely distributed. It cannot be ruled out that this bulb type may contain vesicles and hence be similar to bouton de passage, but this is also much less likely to be observed with electron microscopy. Nevertheless, there is little doubt that most of the varicosities in the NFL were not sites of conventional chemical synapses. 

Previous studies have also described bulb-shaped structures on intraretinal ganglion cell axons in other mammalian species ^{8 9 10 11 12} and goldfish. ¹⁷ The size of the varicosities from these studies was similar to that observed in this study. These prior studies in rats, ^{8 9} rabbits, ¹⁰ and cats ¹¹ also found foci of electron-dense undercoatings on the axolemmal membrane in the bulb regions, which were surrounded by fine glial processes. Using freeze-fracture techniques in the rat retina, Black et al. ^{9 18} showed that the sites corresponding to the bulbs had more particles and represented hot spots of structural elements presumed to mediate ionic currents, similar to the nodes of Ranvier. It was suggested that the bulbs were probably the structural specialization of communication between the axons and glia. ^{8 9 10 11} One recent study demonstrated that desmosome-like densities in the outer plexiform layer of the nonhuman primate retina were actually unconventional chemical synapses. ¹⁹ In the present study, similar desmosome-like and hemidesmosome-like membrane densities were found in bulb regions, forming junctions with neighboring axons and glial cells. These findings suggest an intrinsic relationship between the bulbs and the junctions. 

In this context, the observation of numerous mitochondria housed within the bulbs in this, and one other study by Heppelmann et al., ²⁰ is interesting. In that study, an accumulation of mitochondria was found in the periodical varicosities in the afferent nerve fibers of knee joint in the cat. It was proposed that these varicosities were the sites of the membrane receptors where more energy was required. Moreover, Greenberg et al. ¹² showed that mitochondria often made contact with groups or “baskets” of microtubules at varicosity sites, which has been suggested as substrate for the movement of mitochondria within the axons. ^{21 22} Taken together, the results showing the presence of numerous mitochondria within the bulbs, as well as the membrane densities, suggests that the bulbs are a specialized functional site, perhaps for intercellular communication between adjacent intraretinal axons and glial cells. 

The varicosities observed in this and other studies in which human donor tissue was used may reflect some aspect of bulk transport frozen in time. ¹⁷ The accumulation of mitochondria in the bulbs on the unmyelinated fibers may also be explained by locally increased energy demands. In myelinated nerve fibers, depolarization occurs only at the nodes of Ranvier, with the impulse jumping from node to node, known as saltatory conduction. This type of conduction conserves metabolic energy, because only the region of the nodes have to be repolarized. In unmyelinated axons, action potentials propagate by depolarization along the membrane, which consumes more energy than does saltatory conduction. Quantitative comparisons have shown that the number of mitochondria in normal unmyelinated retinal fibers is 2.5 times higher than in myelinated fibers in feline optic nerves, and the concentration of mitochondria in experimentally demyelinated fibers may increase and become close to that of unmyelinated fibers. ²³ Higher concentration of mitochondria in unmyelinated fibers in the anterior optic nerve was also found in human and other species. ²⁴ Taken together, these results suggest that the bulbs found on intraretinal (unmyelinated) nerve fibers may be functional sites where the high-energy demands of signal transmission and/or intercellular communication are powered. 

In summary, in the present study the intraretinal ganglion cell fibers in human and nonhuman primates were demonstrated to be varicose. The diameter of the varicosities was approximately four times larger than the diameter of the intervaricosity regions. The frequency distribution of both constituted a clear bimodal pattern, but tended to be unimodel in diameters sampled in transverse-cut axons. The rich collections of mitochondria and the electron-dense membrane regions in the varicosities suggest that these varicosities are likely functional sites, possibly power stations, serving the high-energy demands of unmyelinated fibers and acting as intercellular junctions for signal transmission. 

 Supported by National Eye Institute Grant R01-EY05231 (GAC).

 Submitted for publication April 4, 2002; revised July 15, 2002; accepted July 24, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Images of flatmounted human retinas with neurofilaments (200 kDa) stained immunohistochemically were obtained from the vitreous side with a fluorescence light microscope. (A) Two axons with varicosities of different sizes (confocal fluorescence microscope). (B, arrowhead) A typical varicosity of the rare type, which has a homogeneous appearance in contrast to the common type (D). (C) The frequency of the varicosities was high, and the varicosities appeared on almost all the intraretinal axons. (D, arrow) One of the most common types of varicosity, with an appearance of a loosely twisted rope. Magnification, ×40.

Figure 2.

 

Different eccentricities of a neurofilament-stained, wholemounted human retina from an 85-year-old donor. (A) Peripapillary region, 0 to 2 mm; (B) between the peripapillary and midperipheral regions, 3 to 4 mm; (C) between the midperipheral and peripheral regions, 5 to 6 mm; (D) peripheral retina, >6 mm. Eccentrically, the bulbs were generally larger. In the peripapillary area, the bulbs were still recognizable on the much thinner nerve fibers compared with the peripheral area under the light microscope. In this photograph, there are also a few nerve fibers with unusually large bulbs at the peripapillary area (A). Original magnification, ×20.

Figure 3.

 

Three normal human retinas for donors of (A) 1 year, (B) 22 years, and (C) 85 years and one nonhuman primate retina (D) of 15 years taken from the superior area at a similar eccentricity showed varicosity fibers. The varicosities in the retina of the 85-year-old human retina appeared to be similar to that of the retina of the 15-year-old monkey (C, D). Magnification, ×40.

Figure 4.

 

In this montage of micrographs of a human retina, a few ramified varicose fibers with larger bulbs extended in different directions. Inset, bottom left: the point of joining shown in boxed area. Magnification: montage, ×5; bottom left: ×40.

Figure 5.

 

By focusing down from the ganglion cell axons in the NFL (A) deep into approximately the inner plexiform layer (B), varicose fibers were distributed irregularly (arrowheads). Micrographs were taken in the peripheral region of a 75-year-old human retina. Magnification, ×40.

Figure 6.

 

In a nonhuman primate (A) and a human (B) retina, the sections for transmission electron microscopy were cut parallel to the axons. Some bulbs and interbulb regions were cut simultaneously through the axis and appeared to be a chain of bulbs (⋆), with an appearance similar to that observed by light microscope in Figure 1 . Magnification: (A) ×20,000; (B) ×8,000.

Figure 7.

 

Electron micrographs were taken from parallel-cut axons (A, B) and transverse-cut axons (C, D) of monkey retina. The photographs show desmosome- or hemidesmosome-like junctions between the axons (C, D, open arrows), between the glial cells (B, D, open arrowheads) and between axons and glial cells (A, D, black arrows). (B, circled area) An axon surrounded by a glial cell with junction between the two cells. Ax, axons; G, glial cells.

Figure 8.

 

Frequency distribution of the axon diameters measured on the parallel-cut axons (A) and transverse-cut axons (B) in one human retina. (A) The diameter frequency distribution of interbulb region and bulbs. The diameters of the two were plotted individually. The peaks of the two regions were 0.7 and 2.7 μm, respectively. The corresponding two peaks in (B) are marked with double-headed arrows at 0.7 and 1.6 μm, respectively. (A) Measurements of the bulb diameter and the interbulb region. (B, inset) Measurements of axon diameters from photomicrographs of all transverse-cut axons.

The authors thank Robert J. Kanyon, PhD, and Ken Tiekotter for technical assistance, and the Lions Sight and Hearing Foundation, Portland, Oregon, for providing donor tissues for this study.