Hearing Research, 58 (1992) 107-121 © 1992 Elsevier Science Publishers B.V. All rights reserved 0378-5955/92/$05.00

107

HEARES 01694

Neurite regeneration from single primary-auditory neurons in vitro Robin L. Davis 1,2 and William F. Sewell 1,3 ! Department of Otolaryngology, Harrard Medical School, and Eaton-Peabody Laboratory, Massachusetts Eye and Ear Infirmary, Boston, Massachusetts; 2 Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts and ~ Program in Neuroscience, Harrard Medical School, Massachusetts, U.S.A. (Received 3 June 1991; accepted 9 October 1991)

Neurons of the Vlll th cranial nerve in lower vertebrates precisely reconnect with their targets after sustaining injury. It is not known, however, whether the regenerating neurites are guided entirely by external cues or may also be directed by intrinsic mechanisms. To address this issue, single adult primary-auditory neurons were dissected from goldfish and placed in an in vitro environment, devoid of the normal complement of satellite cells, neighboring neurons, and synaptic targets, to observe their patterns of growth. Because acutely isolated neurons showed little neurite outgrowth, neurite regeneration was enhanced by focally crushing the Vlllth cranial nerve 2 to 24 h prior to removal for tissue culture. Neurons that regenerated under identical culture conditions showed growth patterns that could be categorized into three separate groups based on both their morphology and growth patterns. They either 1) remained unbranched (54%), 2)bifurcated or trifurcated into major branches directly from the myelinated stump (V-shaped) (19%), or 3) bifurcated from a regenerated process (Y-shaped), sometimes with a third, smaller branch (27%). Unbranched and V-shaped neurites grew at a constant elongation rate, while Y-shaped neurites grew variably, with alternating retractions and elongations. Neurite elongation was completed in a uniform time period of approximately 15 days despite the differences in elongation rate, maximum length, and latency to growth onset. The neurite branching morphology and manner of growth revealed in this study indicated that adult regenerating neurons can reproduce some elements of the final branching patterns in the absence of extrinsic cues, a capability which may ultimately contribute to the fidelity of reconnection Goldfish; Conditioning lesions; Intrinsic regulation; Endogenous determinants; Regeneration; Peripheral nerve

Introduction

The ability of auditory and vestibular nerve fibers to regenerate their central and peripheral processes with remarkable precision is well documented in non-mammalian species. Restoration of appropriate CNS connections after VIII 'h nerve transection has been demonstrated behaviorally (Sperry, 1945; Sperry, 1963, Newman et al., 1986, 1987, 1989), histologically (Sperry, 1945; Zakon and Capranica, 1981a,b; Zakon, 1983; Marbey and Browner, 1984; Newman et al., 1986, 1987, 1989) and electrophysiologically (Zakon and Capranica, 1981a,b, Zakon, 1983). The evidence to date suggests that the auditory and vestibular components of the V I I I th n e r v e are guided to their appropriate targets primarily by chemotrophic substances rather than by following specific substrate-bound molecules (Sperry,

Correspondence to: Robin L. Davis, (Present address) Department of Biological Sciences, Nelson Biological Laboratories, Rutgers University, Piscataway, NJ 08~55-1059, U.S.A. Fax: 908-932-5870.

1945; Sperry, 1963; Zakon and Capranica, 1981a,b; Newman et al., 1986). While extrinsic cues to axonal guidance are important in establishing the precise innervation pattern of these neurons (Sperry, 1945; Sperry, 1963; Zakon and Capranica, 1981a,b; Newman et al., 1986), many of the fundamental features of the growth patterns and final shapes may be intrinsically determined, as has been shown for a wide variety of other neuronal cell types. Cells in vitro can reproduce certain aspects of their in vivo morphology. For example hippocampal pyramidal neurons placed in vitro maintain their pyramidalshaped cell body and form characteristic tufts of short apical dendrites and thin basal processes (Banker and Waxman, 1988), while retinal ganglion cells in vitro form a complex steilate dendritic arrangement surrounding the entire cell body similar to that observed in the retina (Montague and Friedlander, 1989) It is unknown at present whether primary-auditory neurons have any intrinsically-regulated growth properties that may contribute to the specificity of VIII 'h nerve regeneration. By performing a detailed and quantitative analysis of the neurite branching patterns of individual primary auditory neurons isolated in vitro, our aim was to determine what aspects of the neuronal

108

morphology, if any, were regenerated in vitro. By eliminating or reducing the affects of neural targets, satellite cells, and neighboring neurons, the morphology and growth properties of regenerating neurites could be examined in the absence of extrinsic regulation. The regenerated neurites exhibited simple branching patterns which were similar to those observed in situ and could be further distinguished on the basis of the way in which they grew. Furthermore, the period of active iaeurite outgrowth was completed in approximately 15 days, a time period similar to that observed in vivo. Thus, it appears that some of the key features of neurite regeneration from primary-auditory neurons may be determined by the neuron itself.

Materials and Methods

Animals Goldfish (Carassius auratus) of the comet variety, ranging from 10 to 20 cm in length were obtained from Lillipons Water Gardens, Lilypons, MD., or from local suppliers. Animals were housed in either 30 gallon or 60 gallon filtered aquaria at room temperature. Occasional treatment with Maroxy® (Mardel Laboratories, Inc.) added to the water at the recommended dosage to control fungal and bacterial infections, was often necessary when new fish were first placed into the aquarium. Solutions L-15 Leibovitz medium (Sigma L-4386) was made from powder and supplemented with 10 mM HEPES, 15 mM Glucose, 0.5 mg/ml Claforan, and titrated with 2.5 M NaOH to pH 7.5. Ten percent fetal calf serum (FCS; Sigma F-4010) was added just prior to use. This solution, which will be referred to as supplemented L-15, had a measured vapor pressure osmolarity of 350 mOsm. Supplemented Hank's balanced salt solution (HBSS) was made from powder (Sigma H-2387) and supplemented with 10 mM HEPES, and 0.5 m g / m l Claforan, and titrated with 2.5 mM NaOH to pH 7.5. The dissociation medium was composed of 45% supplemented L-15, 45% supplemented HBSS and 10% FCS. Dissection Fish were anesthetized prior to decapitation by immersion in water containing 200 m g / l of 3-aminobenzoic acid ethyl ester (MS-222). The skull was removed and the brain was retracted rostrally exposing the VIII th nerves. The saccular nerves were crushed with a pair of no. 5 jeweler's forceps and then cut with microdissection scissors at the point where they projected into the brainstem. Once the brain was removed from the cranial cavity, the bony labyrinth and blood vessels

were dissected away to expose the nerve and macula. The nerves were subsequently dissected from each saccular labyrinth along with the attached saccular macula, washed with sterile medium and placed into a sterile Petri dish containing dissociation medium for further dissection. The saccular nerve was separated into the two fiber bundles that innervate either the caudal or the rostral/intermediate portions of the macula. The caudal branch of the nerve innervated the caudal 1/3rd of the saccular macula which appeared pink, delicate and granular. The rostral 2/3rds of the macula had a clear, rigid appearance. The difference in appearance of each portion of the saccular macula was useful as a methodological guideline to separate the saccular nerve. The fiber bundles were peeled away from the connection between these two areas of the macula, and the remainder of the procedure was then performed separately on the caudal and rostral/intermediate fiber bundles. Even though the rostral fibers could be consistently separated from the caudal fibers, some neurons that projected to the intermediate macula were more difficult to separate and occasionally some remained with the caudal branch. Nevertheless, the terms rostral fibers and rostral innervation will refer to both the rostral and intermediate fiber groups.

Dissociation As described by others, the auditory neurons are bipolar (Rosenbluth and Palay, 1961) extending a total length of approximately 1 to 3 mm from the saccular macula to the CNS depending on the size of the goldfish. These neurons either remain unbranched until reaching the periphery where fine ramifications were observed, or the peripheral branch bifurcated upon entering the periphery, sending out ramifications to hair cells that responded to different phases of the sound stimulus (Furukawa, 1978; Sento and Furukawa, 1987). To isolate individual neurons with a minimal number of fibroblasts and non-neuronal tissue, individual neurons were microdissected from the saccular nerve using a pair of no. 5 jeweler's forceps and a right angle microdissection pick. In no case was the cells' bifurcation observed during the dissociation procedure. The cell bodies that were distributed throughout the length of the nerve were removed along with differing lengths of the peripheral and central processes (the total length was usually less than 0.5 mm). Individual neurons were then transferred with a fire-polished micropipette into either Petri (Falcon 1008) or culture dishes (Nunclon & Primaria) coated with poly-L-lysine hydrobromide (Sigma P-7890) and containing approximately 2 ml of supplemented L-15 media (described above). Cultured neurons were kept at 21 to 23°C in a humidified incubator (Lab-Line, Ambi-Hi-Low Chamber).

109

Dye injections The whole-cell configuration of the patch clamp technique (Hamill et. al., 1981) was used for intracellular dye injections. The pipette solution contained 5% Lucifer Yellow in 0.1 M LiCi. The bath solution consisted of 1.7 mM CaCl2, 1.0 mM MgC! 2, 5.4 mM KCI, 137 mM NaCl, 17 mM glucose, 50 mM sucrose, and 10 mM HEPES, pH 7.5. A Nikon epi-fluorescence filter system with an ultraviolet light source was used to visualize the Lucifer Yellow dye. This consisted of a 510 nm dichroic mirror, a 450-490 nm excitation filter, and a 520 nm barrier filter. The fluorescence was observed with a Nikon 40 x objective (0.55 n.a., LWD).

Conditioning lesion We adapted a standard procedure commonly used to evaluate regeneration of the optic nerve, called a 'conditioning lesion' (Landreth and Agranoff, 1976, 1979; McQuarrie and Grafstein, 1981, 1982). Goldfish anesthetized with 200 mg/! MS-222 were placed in a Plexiglas holder which enabled the gills to be peffused with water. The skull was opened and the brain was retracted laterally to expose the Vlll th cranial nerve. The saccular nerve was then crushed or transected with curved no. 5 jeweler's forceps at the point where it entered the brainstem. Once the focal lesion had been made, the skull was replaced over the brain and attached with either polyacrylamide glue or dental cement. The goldfish was then placed in an isolated tank to recover. Saccular nerves were focally lesioned 2 to 24 h prior to removal for tissue culture.

Diameter measurements Neuron fiber diameters were measured from 35 mm negatives projected with a photographic enlarger (Durst Laborator 1200); photographs were taken within the

first 24 h of the cells being placed in culture. Measurements were made on either side of the outer edge of the cell body.

Length measurements Photographs of the cultured neurons were taken with 35 mm film and later imaged and digitized with a Truval scanner connected to an Apple Macintosh II computer. The digital resolution of 300-700 dots per inch (dpi) was more than adequate for the measurements made in this study. Montages, often required to analyze extensive outgrowth, were made digitally with public domain software (Image, version 1.27). This program was also used to measure neurite length. Length measurements were performed by tracing along the middle of a neurite starting from where it emerged out of the myelin sheath and ending at its termination or branch point. For branched neurites, the lengths of individual segments were measured and summed to calculate the neurite length. Therefore, these measurements were representative of the total neurite length, rather than the distance from the end of the cell. Calibrations were obtained with scale bars that were photographed and digitized at the same magnification.

Results

Morphology of primary-auditory neurons in vitro As shown in Fig. 1, long lengths of goldfish primary-auditory neurons could be isolated while maintaining the integrity of the neuron and its surrounding myelin sheath. Nodes of Ranvier were visible along the length of the processes, and the cell body was obvious as an ovoid protrusion within an internodal length of

Fig. 1. Morphological characteristics of goldfish auditory-neurons in vitro. Montage of a single neuron, approximately 2 mm in length, after 1 day in culture. The cell body is centered within the two nodes of Ranvier that are denoted with arrows. The inset shows a detailed view of the rounded, unmyelinated ending. The white calibration bar on the left side of the figure represents 200 ~tm, and 30/zm for main photograph and the inset, respectively.

110

membrane. The most noticeable feature of the cell soma was the asymmetrically positioned nucleus with its prominent nucleolus (Fig. 2), There was a wide range of somata and fiber diameters observed within a single saccular nerve. The two cells shown at the same magnification in Fig. 2 illustrate the extent of this range. The neurons that composed the rostral and caudal branches of the nerve had partia!~/overlapping fiber diameter distributions (Fig. 3). Cell~ 'hat projected to the rostral macula had axon diameters that ranged from 6 to 24/xm, whereas cells that projected to the caudal macula had axon diameters that ranged from 4 to 12 ttm. This distribution was obtained from 163 neurons isolated from both saccular nerves of a single fish (11.5 cm in length), to avoid the necessity of standardizing sizes from different fish. Fiber diameter measurements made from neurons isolated from other fish conformed to the same general size distribution. In the absence of a conditioning lesion (see below), only 4% of the neurons placed in standard tissue culture conditions (without any growth promoting factors or substrates) regenerated neurites, although the neurons remained viable for periods of up to 30 days. This low rate of growth was comparable to the sparse outgrowth of retinal ganglion cells under similar conditions (Landreth and Agranoff, 1976, 1979). A common

50 I~m Fig. 2. Processes and somata of primary-auditory neuron~ shown at the same magnification. The asymmetrically positioned nucleus, prominent nucleolus and granular cytoplasm can be seen in cells that varied greatly in size, yet were removed from fish of similar length (11.5-13 cm). The calibration bar, which represents 50/.tin, applies to both panels.

8° r

!!

caudal rostral

60

,,o

E 40 t--

20

.

4

.

.

.

.

.

.

8 12 16 axon diameter (l~m)

20

24

Fig, 3. A wide range of fiber diameters can be measured from the saccular neurons taken from one fish. This histogram of process diameter evaluates separately the cells that projected to the rostral and intermediate areas of the saccular macula from cells that projected to the caudal saccular macula. Fiber diameter measurements were taken from the area between the tapered lateral perimeter of the cell body and 200 ttm along the length of the ceil. Both central and peripheral processes were analyzed when possible.

feature of the auditory neurons was the appearance of a small rounded protrusion at their cut ends after several days in vitro. In some cases the rounded endings were free of the myelin covering (inset, Fig. 1), but these terminations did not adhere to the substrate or extend processes. Intracellular injections of Lucifer Yellow into the rounded endings showed that they were confluent with the neural cell body since dye injected into these structures spread unimpeded throughout the neuron (N = 5). Individual neurons placed into culture were still ensheathed by the myelinating Schwann cells that surrounded the neurons in vivo. When a single Schwann cell soma located within the internodal segment containing the neuron cell body was filled intracellularly with Lucifer Yellow, dye quickly traveled down the length of the cell, filling the other Schwann cells that myelinate the same internode. This was shown unequivocally in each of the three cases evaluated in this manner. The multiple Schwann cells known to ensheath this internodal segment of the goldfish primary-auditory neuron (Peters et al., 1976) were most likely coupled by gap junctions, which allow Lucifer Yellow to pass from cell to cell (Stewart, 1978). An abrupt cessation of dye spread was observed at the nodes of Ranvier which corresponded to the interruption of the myelin sheath. No dye coupling was noted between the Schwann cells and the neurons, as deter-

111

mined from dye injections into both of these cell types (N = 20). Other than the myelinating Schwann cells, virtually no other satellite cells were observed in vitro.

Conditioning lesions promote neurite outgrowth In an attempt to promote neurite regeneration, experiments were performed in which the saccular nerve was crushed in vivo prior to removal for tissue culture (see Methods), an approach that was successful in promoting regeneration from the goldfish optic nerve (Landreth and Agranoff, 1976). This procedure resulted in increases in the number of auditory neurons showing neurite outgrowth. Processes emanating from the myelinated stump of the primary-auditory neurons extended with a relatively uniform diameter (Fig. 4A) having a ribbed appearance along the length that was in some cases, punctuated with varicosities of similar shape (Fig. 5E). During the early stages of growth, growth cones with actively motile microspikes were observed at the termination of these regenerating neurites. Combined data from three separate experiments (Fig. 4.13) showed that a greater number of neurons regenerated processes after receiving a focal crush 24 h before removal for tissue culture, compared to those neurons dissociated from the contralateral side that had not been crushed. The time course of regeneration under these conditions showed that the maximum outgrowth occurred at approximately 35 days. After this period of time both the number of surviving neurons and the number of regenerated processes detected in culture declined steadily from approximately 3 weeks to 3 months. The percentage of lesioned neurons with regenerating neurites was significantly greater than that of contralateral, unlesioned neurons (chi-square test, P < 0.05; Runyon, 197To) when compared at times when growth was maximal (31 to 36 days in vitro). As expected, the percentage was also greater when lesioned neurons were compared to neurons that had not received a conditioning lesion (22.2%, 28 of 126 neurons; and 4.1%, 3 out of 73 neurons, respectively; P < 0.01). Although the percentage of growing contralateral neurons (8.5%, 5 of 59 neurons), was twice that of neurons that were isolated from an animal that had not received a conditioning lesion, this comparison did not show a significant difference. The analysis, based on an evaluation of 132 neurons, suggested that there was no obvious systemic effect on the neurons taken from an animal that had received a conditioning lesion (on the opposite side), but were not directly crushed. To test for effects on neuronal viability, the percentage of surviving neurons was evaluated for each experimental condition. As shown in Fig. 4C, only slight variations were noted throughout the 95 day time pc-

riod indicating that neuron survival was similar for lesioned and unlesioned conditions (not significantly different at P > 0.05). Furthermore, there were no consistent differences when compared to neurons isolated from animals that had not received a conditioning lesion (data not shown). As with neurons obtained from unoperated animals, myelinating Schwann cells were the predominant nonneuronal cell type. However, in cultures of neurons that had received conditioning lesions, a variety of satellite cells were seen scattered along the substrate, the most prominent of which were cells that demonstrated the flattened, irregular shape of fibroblasts (Fig. 5B)(Bloom and Fawcett, 1968). These cells did not appear to survive in culture for as long a time as the growing neurites, and they were not an element of the substrate on which neudtes grew. Most of the neurites attached directly to the substrate rather than to satellite cells. A small number of cells were also present in vitro that resembled macrophages with their veiled outer membrane, granular inclusions and rounded shape (Fig. 5C) (Bloom and Fawcett, 1968). Bipolar cells, similar in size and shape to non-myelinating Schwann cells (Brockes et al., 1979) (Fig. 5A), were often observed near the myelinated portion of the neurons. These satellite cells also extended processes along the culture dish that terminated with large flattened endings. A unipolar cell type was occasionally observed that was always associated with myelinated neurons, and extended long beaded processes onto the substrate and the myelin of the neuron. Intracellular injections of Lucifer Yellow were necessary to visualize the fine processes of these cells (Fig. 5D). Because these satellite cells elaborated processes which might be confused with neural processes, intracellular injections of Lucifer Yellow were made to establish the characteristics of neural and non-neural growth. The characteristic outgrowth from the myelihated neural stump described earlier was confirmed as neural in origin since dye injected into these processes spread unobstructed back to the neural soma (7 out of 7 cases), an example of which is shown in Fig. 5E. Dye injected into cells that resembled fibroblasts and nonmyelinating Schwann cells ( N = 11)verified that the flattened, irregular, and constantly-changing growth of these satellite cells were features that could be used as reliable criteria to differentiate neuronal from nonneuronal processes.

Growth Characteristics of Regenerating Neurites Branching patterns Neurites elaborated from adult, myelinated, primary-auditory neurons demonstrated a characteristically simple neurite morphology. Rather than showing random neurite outgrowth with varied branching com-

CtJ

(0 tj1

(Jrl

(.11

o

0

0

•1~ 0

03 o

% survival

m. 0 :3 (1) (3. O.

::3

I> •

O0 0 0 0

t,j r l

¢Jrt

tjrll 0"11

Co

0 i

.-s 0

._s ~

I'~ 0

% neurons regenerating PO 011

113

Fig. 5. Morphological features of satellite cells and neurite outgrowth, a. Bipolar cell with a small cell body, splayed endings, and nucleus that was difficult to detect. The cell body size was comparable to the myelinating Schwann cells observed along the length of the neurons, b. Flat cells with constantly changing shapes and nuclei containing two nucleoli were characteristic of fibroblasts, c. Occasionally round cells with a veiled appearance to the outer membrane were observed along the substrate of the tissue culture dish. d. Lucifer Yellow injected into an unidentified cell type revealed fine processes that extended along the length of the myelinated processes and onto the substrate of the culture dish. e. Neurons extended processes with different characteristics from those seen from non-neuronal cells, having a relatively constant diameter, that were sometimes interrupted along the length with beaded varicosities, as shown in the example illustrated. Lucifer Yellow, injected into the process where it bifurcated (arrow), spread to the branched processes and into the myelinated portion of the cell, back to the neuronal cell body. The membrane collapsed in the area where the patch pipette had been removed after the intracellular fill was complete. The calibration bar at the lower right applies to all panels, and represents 50 p.m.

plexity, as observed from the wide variety of cell types grown under similar culture conditions (Banker and Waxman, 1988; Montague and Friedlander, 1989; Ack-

lin and Nicholls, 1990), only three separate branching patterns were observed. Neurites either 1) extended for hundreds of microns without branching (54%), 2) bifur-

Fig. 4. Neurites regenerate more extensively from neurons that had received a conditioning lesion than from the contralateral, unlesioned neurons. A. The neurites that extended from or near the myelinated stump of a neuron (arrow) displayed a characteristic morphology that was later identified in dye studies as neural in origin since it was confluent with the cell soma (arrowhead). This long, constant diameter process, extending from the middle to the right side of the photograph, remained unmyelinated during the course of time that it was maintained in tissue culture. The calibration bar represents 100/zm. B. The number of neurons that regenerated processes was expressed as a percentage of the total number of live cells placed into tissue culture. Since the percentage is based on the initial number of viable cells in culture the decline after day 35 is due to a combination of factors: 1) some cells remained viable yet lost their processes and 2) cells that had regenerated processes were no longer viable. Data were combined from three separate experiments, and binned into 4 day intervals; error bars represent the standard error of the mean, which was calculated from the measurements that fell into each interval. The outgrowth from neurons that had received a conditioning lesion 24 h prior to removal for tissue culture (lesioned: filled diamonds; N = 126) was compared with the outgrowth from neurons taken from the contralateral side of the same animals, (unlesioned: open triangles; N = 58). C. Neuronal survival was evaluated as the number of live neurons evaluated at different times in vitro expressed as a percentage of the total live neurons initially placed into tissue culture. Data were taken from the same three experiments shown above and was also binned into 4 day intervals. Filled diamonds and open triangles represent neurons taken from lesioned and unlesioned conditions, respectively. Error bars represent the standard error of the mean.

Fig. 6. Regenerating neurites dispiay three types of branching patterns. a. Montage of neuron that at 11 days in vitro elaborated an unbranched process from one of its cut ends, and a process that bifurcated (V-shaped) from the other of its cut ends. b. An example of a Y-shaped process emanating from a myelinated stump, 22 days in vitro, the neuronal soma is not shown. The calibration bar applies to both panels, and represents 100 pm.

Al

115

A2

/

/ days in vitro

5

7

l

12

11

18

20

i t/ 25

29

5

B

7

C 13 V-shaped II Y-shaped

9

12

18

25

D

1.5

400

~ A

E A

E

S

1.0

300

.-_= o

A

E.6 E

200 eva

r-

o~ re)

~.

.4

== =,

0

A v

Cm

0

o =_

0.5

10

A v

days in

20

v

A v

~ v

A v

j'

100

0

/ ,/

i

30

vitro

Fig. 7. The two types of branched processes grow differently. (A t) Camera lucida drawings of a Y-shaped process that elongated and changed shape over time in vitro. (A 2) Camera lucida drawings of a V-shaped process that elaborated at a relatively constant rate, retaining its general shape over time in vitro. (B) Length measurements of branches taken at incremental times in culture illustrates the cyclic elongation/retraction of the Y-shaped process shown in A, (filled symbols, diamonds and circles represent the left and right branches, respectively), compared to the steady elongation of the V-shaped process shown in A 2 (open symbols, diamonds and circles represent the left and right branches, respectively). Measurements were made from the edge of the myelinated stump for the V-shaped neurite, and from the end of the central process for the Y-shaped neurite. (C&D) Mean and S.E.M. of the maximum length (defined as the combined length of each segment of a branched neurite, measured at the time in culture when it had reached its longest length) (C) and length to the first branch point (D) from Y-shaped (N = 6) and V-shaped ( N = 8) neurites. While the maximum length of V- and "/-shaped neurites were not significantly different ( P > 0.05; Student's t test, two-tailed), the length to the first branch point was significantly longer for the Y-shaped than the V-shaped measarements (P < 0.001; Student's t test, two-tailed)(Runyon, 1977a).

cated or trifurcated into major branches directly from the myelinated stump (V-shaped) (19%), or 3) bifurcated from a central regenerated process (Y-shaped),

sometimes with a third, smaller branch (27%). Examples of the three different branching patterns are shown in Fig. 6.

116

Of the 127 regenerating neurons that were observed, only 6 of these neurons regenerated processes from both ends. Fig. 6A is an example of one of these cases; this neuron elaborated a single unbranched pro,-:ess from one end, and a V-shaped morphology from the other end. Three neurons were observed with this configuration; two neurons elaborated unbranched and a Y-shaped neurites, whereas one neuron was observed that showed Y-shaped and V-shaped branched neurites from each end. The symmetrical morphology of the goldfish primary auditory neurons precluded a determination as to how these branching patterns corresponded to the cells' original central/peripheral polarity. The three classes of ncurite morphology appeared to represent distinctly different types of neurite outgrowth. Detailed measurements were made from 27 cells which elaborated processes that survived for over 2 weeks in culture and showed stable morphological features. Y-shaped neurites were distinguished from V-shaped and unbranched neurites by different patterns of elongation. The Y-shaped processes alternately extended and retracted, changing shape as they grew, whereas the V-shaped and unbranched processes showed constant elongation, without any noticeable retraction during this phase of growth. This difference is illustrated in Fig. 7 which shows a comparison between a Y-shaped and V-shaped neurite regenerated from neurons obtained from the same animal. Camera lucida drawings were made from single cells over time

A

2.5

"

2 E E v ~" 1.5 t~

E =

1

.E

e

X

E

.5

•

0

0

e

• I

I

.03

.06

,,

I

I

I

.09

.12

.15

rate (mm/day) Fig. 9. The average growth rate was calculated from the maximum length of neurons that remained in vivo for 24 h or less after having received a conditioning lesion. All three types of branching patterns were represented in these measurements. Y-shaped neurons were only included in the graph when the growth curve was constant enough to get an accurate measurement of the average rate (in those cases enhanced central growth compensated for the branch retraction yielding a relatively steady increase in total length).

in culture (Fig. 7, panels A~ and A 2) and length measurements were taken from each of the branches of these fibers over the course of 30 days (Fig. 7B). The

B 2.5

.8

~

"6 I

1.5

J~

c

1

0

t

0

10

,..

20

0

30

days in vitro

40

50

0

5

10

15

20

25

30

35

time (days)

Fig. 8. Growth patterns measured from neurites elaborated by different cells. (A) Measurements of total outgrowth taken from seven neurites over time in vitro. Length measurements were made from the three types of processes: 3 unbranched, 2 V-shaped, 2 Y-shaped. (B) The measurements in panel A were normalized, averaged, and smoothed to construct a mean growth curve. Lengths were normalized to the maximum measurement and overlapped in time at 50% of the maximum length; day 0 is an arbitrary reference point. Averages were taken in 1/2 day intervals and the resulting curve was then smoothed three times with a 3 point running average to remove the minima and maxima. The resulting curve (thick line) represents the average pattern of growth observed from neurites in vitro.

117

Fig. 10. Neurites either remodeled along their length during the elongation, retraction and stabilization growth phases, or remodeled only their endings once the growth had entered into the retraction and stabilization growth phases. The panels on the left show two different neurites that formed rounded endings during the stabilization growth phase. Simple rounded endings, like that shown in (a) were most commonly observed, although endings with more elaborate structures, often accompanied by an angular turn were also observed (b). The panels on the right show remodeling of the same section of neurite at different times in culture: nine (c), fifteen (d), twenty (e), and twenty-nine (f) days in vitro. Both calibration bars at the bottom of the figure represent 50/Lm; the calibration bar in b applies to panels a and b; the calibration bar in f applies to panels c-f.

cyclic pattern of branch elaboration and retraction was evident for the Y-shaped branches, whereas the Vshaped process elongated constantly until reaching a maximum length. This same pattern of elongation was observed for all 27 cells; neurites that grew at a constant elongation rate all corresponded to the V-shaped (N = 8) or unbranched ( N = 13) morphology, whereas neurites that grew by alternatingly lengthening and

retracting all corresponded to the Y-shaped morphology ( N = 6). The Y- and V-shaped processes could also be distinguished from one another by measurements from the end of the myelin to the bifurcation. Branching either occurred close to the myelin (mean distance = 3.8 ttm) or after a considerable distance (mean distance = 304.6 /~m). A statistical analysis of this branching distance

118

from the myelinated neural stump revealed that the Yand V-shaped neurites were significantly different (Student's t test, two-tailed; P < 0.001) (Fig. 7D), although the total neurite length was not (P > 0.05; Fig. 7C). It therefore seems likely that the V-shaped neurites were not just an abbreviated version of the Yshaped processes (without the newly formed central growth), but a different type of process. It also appeared that unbranched neurites were not simply Y-shaped neurites with incomplete or aborted outgrowth. Unbranched processes attained lengths comparable to the Y-shaped processes (average length: 613.7 /~m and 788.0 /~m, respectively), showing no significant difference (P > 0.05) and survived for similar times in culture. Furthermore, even at the maximal length, there was no evidence of even small side branches on any unbranched process (N = 13).

Growth phases Neurite regeneration was categorized into five different phases: latency period, growth initiation, neurite elongation, retraction and stabilization. This was evaluated by measuring the total length of neurite outgrowth from individual neurons over time (Fig. 8A). Neurons included in this analysis had received a conditioning lesion 2 to 24 h prior to removal for tissue culture and were selected because they survived long enough and were measured at frequent enough intervals to show all the phases of growth. From this type of analysis the latency, elongation rate, maximum outgrowth and pattern of growth were evaluated. The first stage of growth, or latency period was quite variable and no strict relationship between branching pattern and latency was observed. Neurite regeneration was initiated from as early as 5 days in vitro to as late as 27 days in vitro (Fig. 8A). The second stage, growth initiation, was marked by an actively remodeling growth cone that emerged from the cut end of the neuron; this also occurred for a variable amount of time, of up to one week, before the third stage, elongation, was observed. In contrast to the previous stages, neurite elongation occurred within a restricted amount of time (approximately 15 days), independent of the maximum length achieved. The Y-shaped processes retracted and lengthened during this phase; the two other types of outgrowth (unbranched and Vshaped) maintained a steady elaboration of newly synthesized membrane. The maximum growth achieved was a transition point to the fourth phase, or retraction, where the neurite decreased in length. Stabilization characterized the final, or fifth phase, which was observed in most of the cases that were followed for a sufficient period of time. These characteristics of the stages of growth can be most clearly seei: in a normalized and averaged curve

(Fig. 8B). The measurements shown in Fig. 8A were scaled (thin lines) according to maximum outgrowth and aligned in the time domain at 50% of the maximum length. The abscissa of the graph in panel B represents relative time, indicating that despite the differences in neurite length and the latency preceding outgrowth, neurites elongated for approximately 15 days in vitro. The difference in maximum length was accounted for by variable growth rates; neurites that achieved longer lengths grew at faster rates (Fig. 9). During the later growth stages, remodeling of the neurites was evident. This process was observed either at the ends of neurites where simple or more complex terminal structures were elaborated (Fig. 10A and B), or along the length of a neurite where varicosities took on an elongated and rigid appearance (Fig. 10C-F). The terminal structures were observed during the retraction and stabilization growth phases, whereas remodeling along the length of neurites could be observed even during the elongation phase. Perhaps these morphological changes correspond to processes such as the establishment of synaptic specializations or the assemblage of cytoskeletal components, yet without more detailed analysis, such as electron microscopy, we can only conjecture about the underlying structures.

Discussion

Neuron morphology in vitro The gross morphological appearance of the primary-auditory neurons was maintained in vitro. The cells' myelin, and the position and relative size of the nucleus corresponded to the characteristics of these same cells observed in fixed specimens from in vivo preparations (Rosenbluth and Palay, 1961). The range of fiber diameters were also comparable with that reported by Rosenbluth and Palay (1961) and by Furukawa and Ishii (1967), with larger fiber diameters innervating the rostral macula and neurons with smaller diameters innervated the more caudal macula. This evaluation should not be thought of as a comprehensive representation of the cell sizes in situ, however, since no attempt was made to isolate all of the cells that compose the saccular nerve. One might expect, for example, that the dissociation procedure would most likely under-represent the smaller-diameter neurons since these neurons are more difficult to visualize and manipulate mechanically. On the other hand, few of the neurons that were maintained in culture over long periods of time were likely to be those Furukawa and his colleagues described as S1 fibers (Furukawa and Ishii, 1967; Sento and Furukawa, 1987), since the largest neurons (diameters > 16 /~m) generally did not survive.

119

Conditioning lesions enhanced neurite regeneration in vitro Similar to findings from optic nerve (Landreth and Agranoff, 1976, 1979), conditioninglesions of the auditory nerve increased the amount of neurite outgrowth and the number of satellite cells present in vitro, without affecting neuronal survival. Observations made from 24-hour conditioning lesions showed that the increase in outgrowth was limited to the lesioned neurons; contralateral cells were not affected. The mechanisms underlying these effects have not been investigated for the auditory nerve; however, from extensive work done on the goldfish optic nerve a neurite-promoting protein has been isolated and is thought to mediate regeneration (Caday et al., 1989). Whether the auditory nerve releases or responds to such a factor has not been tested; however the similarities observed from conditioning lesions for both the optic nerve and the primary-auditory nerve suggest that both may be regulated by similar mechanisms. Since non-neuronal cells have been shown to release both trophic and tropic substances (for review see Schwartz, 1987; Gage and Varon, 1988), it may be the case that satellite cells might have contributed to enhanced neurite regeneration.

Characteristics of neurite outgrowth Because central and peripheral processes were not distinguished in tissue culture, only general comparisons of the morphology observed in vitro can be made with what is known about the branching patterns of the primary-auditory neurons in vivo. Yet even without making this distinction, it is clear that the simple branching patterns elaborated in tissue culture look very much like the gross morphology of primary-auditory neurons in vivo. These bipolar neurons extend peripherally into the saccular macula to either remain unbranched or divide into two equal branches before reaching the sensory epithelium (Furukawa, 1978; Sento and Furukawa, 1987). Furthermore, the central extent of these same types of neurons in vertebrates as diverse as fish and man bifurcate once they enter the CNS, sometimes with a third branch (Lorente de N6, 1933; Ramon y Cajal, 1956). Therefore both sides of these bipolar neurons show relatively simple branching patterns, much like what was observed in the tissue culture dish. If the different branching classes represent central or peripheral processes, additional observations may lead to some insights into whether the polarity of these bipolar neurons is reestablished or re-formed in vitro. The observations that regenerated neurites in vitro bore a fundamental resemblance to their in vivo morphology, and neurite morphology fell into three distinct categories in a homogeneous tissue culture environment suggested that the morphology of regenerating

neurons may be endogenously regulated. Thus, much like developing mammalian retinal ganglion cells (Montague and Friedlander, 1989), hippocampal neurons (Banker and Waxman, 1988), and cranial sensory ganglion cells (Davies, 1989) in tissue culture, primaryauditory neurons also showed their own unique growth and branching patterns even in the absence of their normal environment. Although we have observed intrinsic regulation of certain morphological and growth features in regenerating neurons it is unclear whether these same types of intrinsic mechanisms are also a feature of developing, primary-auditory neurons. In contrast to the gross morphology, the fine structural features, such as terminal endings, showed limited correspondence to those observed from primaryauditory neurons in situ. Regenerating neurites formed either a single rounded ending or more complex rounded clusters at the termination of the process. In some cases these neurites formed angular turns on the uniform substrate of the culture dish, close to the end of the process, comparable to what has been observed from peripheral branches in vivo that" ave reached the macula (Sento and Furukawa, 1987). An obvious difference between the appeara~ce of the neurons in vivo and those in vitro was that the terminal morphology could be much more elaborate in the animal, although sim~le endings were also observed (Sento and Furukawa, i987; Nakajima, 1974). This suggested that target cells or some other extrinsic feature must be present in order to induce the complex fine structure observed in vivo. Apart from the morphological classes that were elaborated in the culture dish, growth features also displayed a stereotyped quality. Neurites displayed active outgrowth within a restricted time period of approximately 15 days, despite the differences in lengths and outgrowth latencies. This time period corresponds roughly with what others had observed from regeneration in vivo. Gleisner and Wers~ill (1975) showed with ultrastructural studies that the horizontal and anterior vertical ampullary branches of the Vlll th cranial nerve in frog regenerated back to their peripheral targets in 13 to 15 days and formed synapses 20 days after being transected distal to their branching point. Furthermore, Sperry (1945) noted that the earliest functional recovery occurred 21 days after V l l l th nerve transection. The growth rates of the primary-auditory neurons in culture varied from approximately 0.01 to 0.02 mm/day. This was about a factor of 10 slower than regenerating goldfish optic nerves growing on a similar substrate at identical temperatures (Hopkins et al., 1985). Since the nerve crush enhanced growth rate and protein synthesis in regenerating neurons (McQuarrie and Grafstein, 1981, 1982), and since the time interval that the nerve remained in vivo after the crush enhanced outgrowth

120

(Landreth and Agranoff, 1979), it is possible that the protocol for the conditioning lesion could account for these differences. For example, the crushed nerve remained in vivo for up to 24 h in the present studies, whereas the crushed optic nerve remained in vivo for 7 to 14 days before explantation. Another feature consistently observed in regenerating neurons in vitro was that modes of growth correlated with branching patterns. Unbranched and Vshaped neurites grew to their maximum length in a manner different from that displayed by Y-shaped neurites. These growth patterns not only served to differentiate the two types of branched neurites from one another, but also provided additional evidence for the intrinsic determination of the growth patterns from these regenerating cells. The stereotyped growth characteristics observed in the homogeneous environment of tissue culture conditions illustrated features of neuronal growth in the absence of normal exogenous cues. Under these conditions neurons grew in a predictable manner both in their final morphology, as well as in their timing and mode of elongation. The similarities between the morphological features in vitro to those of in vivo preparations lead one to speculate that the endogenous properties of the cells may contribute to the precision observed from regenerating VIII th nerve fibers in vivo (Zakon and Capranica, 1981a,b; Zakon, 1983; Newman et al., 1989). For example, during regeneration the gross morphology of the neuron would be reiterated, thus the maintenance of stereotyped branching patterns may be of some advantage. Features of regeneration such as the precise direction of growth and the formation of synaptic specializations,on the other hand, would then fall within the domain of exogenously regulated interactions (Letourneau, 1975; Bentley and Caudy, 1983; Raper et al., 1983; Eisen et al., 1986; Harris, 1986; Kuwada, 1986; Dodd and Jessel, 1988).

Acknowledgements We appreciate the contribution of E.A. Mroz during the course of this work. We thank J.C. Adams, M.C. Brown, and M.C. Liberman for their comments on earlier versions of the manuscript. This work has been supported by NIH grants DC00119, DC00767 and DC00006.

References Acklin, S.E. and Nicholls, J.G. (1990) Intrinsic and extrinsic factors influencing properties and growth patterns of identified leech neurons in culture. J. Neurosci. 10, 1082-1090. Banker, G.A. and Waxman, A.B. (1988) Hippocampal neurons generate natural shapes in cell culture. In: R.J. Lasek and M.M.

Black (Eds.), Intrinsic determinants of neuronal form and function, Alan R. Liss, Inc., New York, NY, pp. 61-82. Bentley, D. and Caudy, M. (1983) Pioneer axons lose directed growth after selective killing of guidepost cells. Nature 304, 62-65. Bloom, W. and Fawcett, D.W. (1968) A Textbook of histology. Saunders, PA. Broches, J.P., Fields, K.L. and Raft, M.C (1979) Studies on cultured rat Schwann cells. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res. 165, 105-118. Caday, C.G., Apostolides, P.J., Benowitz, L.I., Perrone-Bizzozero, N.I. and Finklestein, S.P. (1989) Partial purification and characterization of a neurite-promoting factor from the injured goldfish optic nerve. Mol. Brain Res. 5, 45-50. Davies, A.M. (1989) Intrinsic differences in the growth rate of early nerve fibers related to target distance. Nature 337, 553-555. Dodd, J. and Jessell, T.M. (1988)Axon guidance and the patterning of neuronal projections in vertebrates. Science 242, 692-699. Eisen, J.S., Myers, P.Z. and Westerfield, M. (1986) Pathway selection by growth cones of identified motoneurones in live zebra fish embryos. Nature 320, 269-271. Furukawa, T. (1978) Sites of termination on the saccular macula oi? auditory nerve fibers in the goldfish as determined by intracellular injection of procion yellow. J. Comp. Neurol. 180, 807-814. Furukawa, T. and Ishii, Y. (1967) Neurophysiological studies on hea.-ing in goldfish. J. Neurophysiol. 30, 1337-1403. Gage, F.H. and Varon, S. (1988) Trophic hypothesis of neuronal cell death and survival. In: S. Finger, T.E. Levere, C.R. Alm|i and D.G. Stein (Eds.), Brain Injury and Recovery, Plenum Press, New York, NY, pp. 201-214. Gleisner, L. and Wers[ili, J. (1975) Experimental studies on the nerve-sensory cell relationship during degeneration and regeneration in apullar nerves of the frog labyrinth. Acta Oto-laryngoIogica (Suppl.) 333, 1-28. Grumbacher-Reinert, S. (1989) Local influence of substrate molecules in determining distinctive growth patterns of identified neurons in culture. Proc. Natl. Acad. Sci. USA 86, 7270-7274. Hamill, O.P., Marty, A., Neher, E., Sakmann, B. and Sigworth, F.J. (1981) Improved patch-clamp techniques for high resolution current recording from cells and cell-free membrane patches. Pfliigers Arch. 391, 85-100. Harris, W.A. (1986) Homing behavior of axons in the embryonic vertebrate brain. Nature 320, 266-269. Hopkins, J.M., Ford-Holevinski, T.S., McCoy, J.P., Agranoff, B.W. (1985) Laminin and optic nerve regeneration in the goldfish. J. Neurosci. 5, 3030-3038. Kuwada, J.Y. (1986) Cell recognition by neuronal growth cones in a simple vertebrate embryo. Science 233, 740-746. Landreth, G.E. and Agranoff, B.W. (1976) Explant culture of adult goldfish retina: effect of prior optic nerve crush. Brain Res. 118, 299-3O3. Landreth, G.E. and Agranoff, B.W. (1979) Explant culture of adult goldfish retina: A model for the study of CNS regeneration. Brain Res. 161, 39-53. Letourneau, P.C. (1975) Cell-to-substratum adhesion and guidance of axonal elongation. Dee. Biol. 44, 92-101. Lorente de N6, (1933) Anatomy of the eighth nerve. The central projection of the nerve endings of the internal ear. Laryngoscope 43, 1-38. Marbey, D. and Browner, R.H. (1984) The reconnection of auditory posterior root fibers in the red-eared turtle, Chrysemys scripta elegans. Hear. Res. 15, 89-94. McQuarrie, I.G. and Grafstein, B. (198D Effect of a conditioning lesion on optic nerve regeneration in goldfish. Brain Res. 216, 253-264. McQuarrie, I.G. and Grafstein, B. (1982) Protein synthesis and axonal transport in goldfish retinal ganglion cells during regeneration accelerated by a conditioning lesion. Brain Res. 251, 25-37.

121 Montague, P.R and Friedlander, M.J. (1989) Expression of an intrinsic growth strategy by mammalian retinal neurons. Proc. Natl. Acad. Sci. 86~ 7223-7227. Nakajima, Y. (1974) Fine structure of the synaptic endings on the mauthner cell of the goldfish. J. Comp. Neurol. 156, 375-402. Newman, A., Honrubia, V. and BelJ, T. (1987) Regeneration of the eigh~,h cranial nerve I1. Physiologic verification in the bullfrog. Laryngoscope 97, 1219-1232. Newman, A., Kuruvilla, A., Pereda, A. and Honrubia, V. (1986) Regeneration of the eighth cranial nerve. I. Anatomic verification in the bullfrog. Laryngoscope 96, 484-493. Newman, A., Suarez, C., Kuruvilla, A. and Honrubia, V. (1989) Regeneration of the eighth cranial nerve. !II. Central projections of the primary afferent fibers from individual vestibular receptors in the bullfrog. Laryngoscope 99, 162-173. Peters, A., Palay, S.L. and Webster H. de F. (1976) The Fine Structure of the Nervous System. Saunders, PA. Ramon y Cajal (1956) Degeneration and Regeneration of Nervous System. Vol 1, R.M. May (transl.) Hafner Publ. Co., New York, NY. Raper, J.A., Bastiani, M. and Goodman, C.S. (1983) Pathfinding by neuronal growth cones in grasshopper embryos: II. Selective fasciculation onto specific axonal pathways. J. ".."/eurosci.3, 31-41. Rosenbluth, J. and Palay, S.L. (1961) The fine structure of nerve cell bodies and their myelin sheaths in the eighth nerve ganglion of the goldfish. J. Biophys. Biochem. Cytol. 9, 853-877. Rousselet, A., AutUlo-Touati, A., Araud, D. and Prochiantz A (1990) In vitro regulation of neuronal morphogenesis and polarity by astrocyte-derived factors. Dev. Biol. 137, 33-45. Runyon, R.P. (1977a) Descriptive and Inferential Statistics, A Contemporary Approach. Addison-Wesley Publishing Company, Reading, MA, pp. 179-184.

Runyon, R.P. (197To) Nonparametric Statistics, A Contemporary Approach. Addison-Wesley Publishing Company, Reading, MA, pp. 45-53. Sento, S. and Furukawa, T. (1987) intra-axonal labeling of saccular afferents in the Goldfish, Carassius auratus: Correlations between morphological and physiological characteristics. J. Comp. Neurol. 258, 352-367. Schwartz, M. (1987) Molecular and cellular aspects of nerve regeneration. CRC Crit. Rev. Biochem. 22, 89-110. Sperry, R.W. (1945)Centripetal regeneration of the 8th cranial nerve root with systematic restoration of vestibular reflexes. Am. J. Physiol. 144, 735-741. Sperry, R.W. (1963) Chemoaffinity in the orderly growth of nerve fiber patterns and connections. Proc. Natl. Acad. Sci. 50, 703-710. Stewart, W.W. (1978) Functional connections between cells revealed by dye coupling with a highly fluorescent naphthalimide tracer. Cell 14, 741-759. Szpir, M.R., Sento, S. and Ryugo, D.K. (1990)Central projections of cochlear nerve fibers in the alligator lizard. J. Comp. Neurol. 295, 530-547. Yasuda. T., Sobue, G., Ito, T.. Mitsuma, T. and Takahashi, A. (1990) Nerve growth factor enhances neurite arborization of adult sensory neurons: a study in single-cell culture. Brain Res. 52, 54-63. Zakon, H.H. and Capranica, R.P. (1981a) Reformation of organized connections in the auditory system after regeneration of the eighth nerve. Science 213, 242-244. Zakon, H.H. and Capranica, R.P. (1981b) An anatomical and physiological study of regeneration of the eighth nerve in the leopard frog. Brain Res. 209, 325-338. Zakon, H.H. (1983) Reorganization of connectivity in amphibian central auditory system following Vii! th nerve regeneration: time course. J. Neurophysiol. 49, 1410-1427.