ATP enhances repair of hair bundles in sea anemones

Hearing Research 136 (1999) 1^12

ATP enhances repair of hair bundles in sea anemones Glen M. Watson *, Stacy Venable, Renee R. Hudson, Je¡rey J. Repass Department of Biology, P.O. Box 42451, University of Southwestern Louisiana, Lafayette, LA 70504-2451, USA Received 8 October 1998; received in revised form 11 March 1999; accepted 5 May 1999

Abstract Hair bundle mechanoreceptors of sea anemones are similar to those of the acousticolateralis system of vertebrates (Watson, Mire and Hudson, 1997, Hear. Res. 107, 53^63). Anemone hair bundles are repaired by `repair proteins' secreted following a complete loss of structural integrity and loss of function caused by 1 h exposure to calcium free seawater. Exogenously supplied repair proteins (RP) restore structural integrity to hair bundles and restore vibration sensitivity in 7^8 min (Watson, Mire and Hudson, 1998, Hear. Res. 115, 119^128). We here report that exogenously supplied ATP enhances the rate by which RP restore vibration sensitivity. A bimodal dose response to ATP indicates maximal enhancement at picomolar and micromolar concentrations of ATP. At these concentrations of ATP, vibration sensitivity is restored in 2 min. These data suggest that at least two ATPases exhibiting different binding affinities for ATP are involved in the repair process. Whereas the higher affinity site is specific for ATP, the lower affinity site does not discriminate between ATP and ADP. Nucleotidase cytochemistry localizes ATPase activity in isolated repair proteins. In the absence of exogenously added RP, sea anemones secrete and consume ATP during the 4 h recovery period after 1 h exposure to calcium free seawater. In the presence of exogenously added RP, ATP is secreted and then consumed within 10 min. Quinacrine cytochemistry localizes possible stores of ATP in the apical cytoplasm of sensory neurons located at the center of the hair bundle. According to our model, ATP is secreted by the sensory neuron after its hair bundle loses structural integrity. Hydrolysis of ATP by repair proteins is essential to the repair process. ß 1999 Elsevier Science B.V. All rights reserved. Key words: ATP; Ecto-ATPase; Mechanoreceptor; Stereocilia

1. Introduction Hair bundle mechanoreceptors occur in all vertebrate and in some invertebrate phyla (for example, see Budelmann, 1988). Previously, we found that hair bundle mechanoreceptors located on tentacles of sea anemones are surprisingly similar to hair bundles of the acousticolateralis system of vertebrates (Watson et al., 1997 ; Watson and Mire, 1998). Anemone hair bundles are derived from a multicellular complex consisting of a central, sensory neuron surrounded by supporting cells. Small diameter stereocilia from the supporting cells converge onto large diameter stereocilia from the sensory neuron. The large diameter stereocilia surround a single, nonmotile kinocilium (Watson et al., 1997). Anemone hair bundles are

* Correspondingence author. Tel.: +1 (318) 482-5667; Fax: +1 (318) 482-5660; E-mail: [email protected]

composed of actin based stereocilia interconnected by numerous linkages including tip links. Step de£ections of anemone hair bundles induce transients in membrane current that are graded in relation to stimulus strength (Mire and Watson, 1997). The responses saturate at strong de£ections and adapt to prolonged de£ections. Responses are reversibly inhibited by aminoglycoside antibiotics. Anemone hair bundles regulate discharge of nematocysts (stinging capsules) into vibrating targets (Watson and Mire-Thibodeaux, 1994). A biological assay performed on fully intact specimens reveals frequency speci¢c enhancement in discharge of nematocysts into vibrating test probes touched to tentacles (Watson et al., 1998a). Vibration dependent discharge of nematocysts (hereafter referred to as `vibration sensitivity') is reversibly inhibited by aminoglycoside antibiotics. In addition, vibration sensitivity is selectively abolished by treatments thought to attack tip links including calcium depleted bu¡ers or elastase (Assad et al., 1991 ; Preyer et al., 1995; Watson et al., 1997,

0378-5955 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 5 9 5 5 ( 9 9 ) 0 0 0 8 7 - 8

HEARES 3265 24-9-99

2

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

1988b). Discharge of nematocysts into nonvibrating targets is una¡ected by such treatments (Watson et al., 1997). Sea anemones repair hair bundles following a complete loss of structural integrity and loss of vibration sensitivity caused by 1 h exposure to calcium free seawater. Apparently, hair bundles are repaired by large protein complexes named `repair proteins' secreted into the extracellular £uid (Watson et al., 1998b). Whereas recovery of hair bundle structure is ascertained by video microscopy, recovery of function is ascertained by electrophysiology and by the bioassay. Based on the bioassay, the time course of recovery is 7^ 8 min for animals exposed to repair proteins. Using gel ¢ltration chromatography, a fraction named fraction L is identi¢ed having comparable bioactivity to the complete repair protein mixture (Watson et al., 1998b). Negative staining transmission electron microscopy reveals that fraction L includes ¢lamentous structures shaped like the letter Y and decorated by globular subunits. The globular subunits dissociate from the ¢lamentous backbone in the presence of mM ATP or AMP-PNP, an ATP analog that is nonhydrolyzable by most ATPases. The threshold concentration of ATP that induces subunits to dissociate from the ¢lamentous backbone of fraction L is not yet known. The ¢lamentous backbone of fraction L has dimensions comparable to linkages interconnecting stereocilia including tip links. According to our working model, fraction L consists of replacement linkages of all types and accessory proteins (Watson et al., 1998b). In the present paper, we examine the role of ATP in the repair process. 2. Materials and methods 2.1. Reagents and seawater formulae Unless otherwise speci¢ed reagents were obtained from Sigma Chemical Co., St. Louis, MO. Calcium free arti¢cial seawater consisted of: 447 mM NaCl, 26 mM MgSO4 , 24 mM MgCl2 , 9 mM KCl, 2 mM EGTA, and 2 mM NaHCO3 , pH 8.3. High potassium arti¢cial seawater (KSW) consisted of: 382 mM NaCl, 26 mM MgSO4 , 24 mM MgCl2 , 50 mM KCl, 12 mM CaCl2 and 2 mM NaHCO3 . TBS consisted of 10 mM Tris and 150 mM NaCl, pH 7.5. 2.2. Bioassay for vibration sensitivity Vibration sensitivity was assayed as described previously (Watson and Hudson, 1994). Brie£y, 2 cm segments of nylon ¢shing line are coated with 25% gelatin to a thickness of approximately 200 Wm. The uncoated ends of the ¢laments are inserted into a glass capillary

tube attached to a piezo disk induced to vibrate by a function generator equipped with a frequency counter. Each vibrating test probe is inserted into the seawater and allowed to hydrate for several seconds before it is moved into contact with the tentacles of an animal. The probe is then swiftly withdrawn in a direction that will minimize the chance for additional tentacles to touch the probe. Test probes are ¢xed for 30^60 s in 2.5% glutaraldehyde and then stored hydrated in microtiter plates. Test probes are examined using phase contrast microscopy. At 400U total magni¢cation, discharged microbasic p-mastigophore type nematocysts are counted for a representative ¢eld of view. Time courses of recovery were performed as follows. For each concentration of ATP tested ranging from zero to 1034 M ATP, dilutions of ATP in seawater were prepared on the morning of the experiment. A total of four separate bowls of anemones (each containing approximately 15 animals) was exposed to calcium free seawater for 1 h using methods described elsewhere (Watson et al., 1998b). At the completion of the 1 h treatment, calcium free seawater was replaced with seawater containing 10 mM calcium (Watson et al., 1998b). After a 15 min recovery period from the anesthetic e¡ects of calcium free seawater, this seawater was replaced with calcium containing seawater forti¢ed with ATP. At this point, concentrated repair protein (RP; collected and concentrated as described in Watson et al., 1998b) was injected into the dish (20 Wl per 6 ml seawater). Discharge of nematocysts was tested at 1 min intervals after the addition of RP and ATP through 10 min. For each time point, a single animal was touched with a test probe vibrating at 55 Hz, a preferred frequency. Each animal was touched only once. Separate dishes were used in order to keep the time intervals precise. For a given batch of RP, a total of four replicate test probes was obtained at each time point and ATP concentration. A total of four batches of RP was tested to further test reproducibility of experimental results. Final data points plotted in graphs indicate mean responses ( þ S.E.M.) based on two separate batches of RP. Baseline discharge (obtained using nonvibrating test probes and plotted at the t = 0 time point) was subtracted from each data point to yield `enhanced discharge' indicative of vibration sensitivity. Maximal recovery was estimated from the original data. In order to estimate the time for half maximal recovery, response curves were ¢tted using a weighted algorithm (Axum, Trimetrics, Seattle, WA) and half maximal recovery was estimated by eye from the ¢tted curves. This procedure was followed in order to diminish the importance of any single data point in determining the shape of the curve. Data sets for treatments were compared to data for controls receiving RP only using ANOVA and post-hoc comparisons. In addition, the mean half max-

HEARES 3265 24-9-99

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

imal recovery was statistically compared using ANOVA. Post-hoc comparisons among treatments are reported as signi¢cant at a P value 90.05 (CSS Statistica, Statsoft, Tulsa, OK). 2.3. Assay for seawater levels of ATP ATP levels were measured from seawater samples using the luciferase/luciferin assay (LeMasters and Hackenbrock, 1978). A crowded culture of sea anemones was used (3.8^5.5 animals/cm2 ) in which the animals live attached to the bottom of a Corning Ware dish ¢lled with 200 ml seawater, su¤cient to completely cover the animals. The seawater was exchanged with

3

calcium free, arti¢cial seawater (with ¢ve washes to ensure the removal of calcium) for 1 h to disrupt hair bundles as described previously (Watson et al., 1998b). At the completion of the hour, the animals were returned to natural, Millipore ¢ltered (0.45 Wm pore size) seawater containing approximately 10 mM calcium (with ¢ve washes to ensure the removal of EGTA). At speci¢c intervals, small samples of seawater, typically 100 Wl each, were removed from the culture dish. Seawater samples were Millipore ¢ltered and stored on ice. Luminescence was assayed using a luminometer (Model TD 20/20, Turner Designs, Sunnyvale, CA). Luciferase/luciferin was reconstituted in sterile Hepes bu¡er according to the supplier's instructions (Turner Designs) and used the same day. To 25 Wl of the seawater sample, 50 Wl of the luciferase/luciferin stock was added. To this solution, 10 Wl of 1035 M ATP in seawater was added. For each sample reading, luminescence was scored three times as follows : (1) the seawater sample alone ; (2) after the addition of luciferase/luciferin ; and (3) after the addition of the supplement of 1035 M ATP. Each reading consisted of a 15 s integration after a 3 s delay to allow for mixing of reagents. Luciferase dependent luminescence was determined by subtracting the value for (1) from the value for (2). The magnitude of the increase in luminescence following the addition of the ATP supplement was used to normalize data. The ¢nal data points plotted consisted of the mean þ S.D. of three consecutive readings per sample and compared to a standard curve performed on the same day. 2.4. Quinacrine cytochemistry Anemone tentacles were processed for quinacrine cy6 Fig. 1. E¡ects of repair protein on the time course for the recovery of vibration sensitivity. Animals were exposed to calcium free seawater for 1 h, then transferred to calcium containing seawater. After a 15 min recovery, repair protein (RP) was added (20 Wl per 6 ml seawater). At 1 min intervals tentacles were touched with gelatin coated test probes vibrating at 55 Hz, a preferred frequency. To maintain precise time intervals, one animal was touched with a test probe at each minute. Microbasic p-mastigophore nematocysts discharged into the gelatin coating were counted for a representative ¢eld of view using phase contrast optics at a total magni¢cation of 400U (see Section 2 for additional details). Each data point plotted indicates the mean number of mastigophore nematocysts counted based on a sample size of eight replicate test probes ( þ S.E.M.). For presentation, baseline discharge into nonvibrating controls was subtracted from each data point to give `enhanced discharge' indicative of vibration sensitivity. Panels A and B each depict two complete replicate experiments, giving a total of four replicate experiments. In panel C, data are shown for RP ¢rst tested in the combined presence of RP and AMP-PNP (open circles in Fig. 3F). After dialyzing the protein sample to remove the AMP-PNP, bioactivity was tested as described above.

HEARES 3265 24-9-99

4

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

Fig. 2. E¡ects of repair protein with ATP on the time course for the recovery of vibration sensitivity. The experimental design was the same as was described for Fig. 1 except that at time zero, the seawater was exchanged for a solution of ATP in seawater. At this point, RP was added. At 1 min intervals discharge was tested. The ATP concentration was varied as follows: (A) 10314 M ATP; (B) 10312 M ATP; (C) 10310 M ATP; (D) 1038 M ATP; (E) 1036 M ATP; and (F) 1034 M ATP.

tochemistry of adenine nucleotides using modi¢cations of published methods (White et al., 1995). Brie£y, sea anemones were anesthetized in KSW. Excised tentacles were immobilized along a strand of human hair, the ends of which were glued to a 20U40 mm coverslip mounted in a plastic photographic slide holder. Excised tentacles were incubated for 30 min in 50 WM quinacrine dihydrochloride in KSW, rinsed in KSW for 30 min, then imaged using an inverted microscope (Model IMT-2F, Olympus, Tokyo) equipped with UV permissive optics and a band-pass, excitation ¢lter having peak transmittance centered at 340 nm. Images were collected by a regulated, cooled CCD camera (Model ST-6, Santa Barbara Instrument Group, Santa Barbara, CA). To improve the quality of the images, glare was

digitally removed using maximum entropy deconvolution software (Maxim-DL, Di¡raction Limited, Ontario, Canada). 2.5. Nucleotidase cytochemistry ATPase cytochemistry was performed at the transmission electron microscopic level on concentrated repair proteins based on a modi¢cation of published procedures (Watson and Hessinger, 1992). Concentrated repair proteins were carefully injected into the lumen of a small, hollow agar tube prepared using 3% agar and a disposable transfer pipette. The lumen was formed by a segment of ¢shing line present when the agar solution was drawn into the pipette. Upon solid-

HEARES 3265 24-9-99

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

5

Fig. 3. E¡ects of repair protein with ADP, AMP, or AMP-PNP on the time course of recovery of vibration sensitivity. The experimental design was the same as was described for Fig. 2 except that nucleotides other than ATP were used as follows: (A) 10312 M ADP; (B) 1036 M ADP; (C) 10312 M AMP; (D) 1036 M AMP; (E) 10312 M AMP-PNP; and (F) 1036 M AMP-PNP.

ifying, ¢shing line was removed from the mold. After the protein sample was injected into the lumen, the open ends were plugged to retain the protein sample in the agar tube. This preparation was incubated in the appropriate nucleotide solution in 0.1 M TBS and including 20 mM strontium chloride as the trapping agent. After 5 min, the samples were ¢xed in 2.5% glutaraldehyde for 30 min, dehydrated in acetone, and embedded in Spurr resin. Thin sections were counterstained with 1% uranyl acetate for 1 h and then imaged at 75 kV. Because the negatives had low intrinsic contrast, contrast was maximized during the preparation of photographic prints. In addition, photographic prints of electron micrographs were digitally scanned and contrast was further enhanced using image processing software (Adobe Photodeluxe, San Jose, CA).

2.6. Nucleotide dependent elution of repair proteins Repair proteins were obtained and concentrated as described previously (Watson et al., 1998b). Concentrated repair protein comprises a mixture of individual RP complexes and aggregates of RP complexes (unpublished). Concentrated repair protein was injected onto a Sepharose 2B column equilibrated to a mobile phase consisting of 100 mM Tris-HCl and 2 mM MgSO4 , pH 7.8. Individual repair protein complexes are not sieved on a 2B column so that the majority pass through the column in the void volume. These repair proteins were eluted and discarded. In an attempt to elute aggregated repair protein remaining in the column, ATP was added to the mobile phase in a series of increasing ATP gradients as follows : (i) zero (none

HEARES 3265 24-9-99

6

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

added) to 10310 M ATP; (ii) 10310 to 1038 M ATP ; (iii) 1038 to 1036 M ATP and; (iv) 1036 to 1034 M ATP. Protein fractions were collected, dialyzed against the mobile phase lacking ATP, and then tested for bioactivity using the bioassay for vibration sensitivity. Brie£y, after animals were exposed to calcium free seawater for 1 h, they were returned to calcium containing seawater where they were allowed 15 min to recover from the anesthetic e¡ects of calcium free seawater. At this point, the appropriate protein fraction was added to the dish (50 Wl per 6 ml) and discharge was tested 10 min later. Response data were analyzed for statistical signi¢cance as compared to control animals not receiving a protein fraction using methods described above. Experiments on animals were performed in accordance with guidelines appearing in the Declaration of Helsinki. 3. Results 3.1. E¡ects of exogenously added repair proteins on restoring vibration sensitivity Previously, we found vibration dependent discharge of nematocysts to be selectively abolished by exposing animals for 1 h to calcium free seawater (Watson et al., 1997). After such exposure, discharge into vibrating targets is comparable to discharge into nonvibrating targets (i.e., baseline) (Watson et al., 1997). Upon adding RP to the experimental animals, vibration dependent discharge of nematocysts remained at baseline levels during the ¢rst 5 min and then sharply recovered at 7^ 8 min to a maximum of approximately 15^25 nematocysts greater than baseline (Fig. 1A,B). Increasing ambient ATP at the instant that the RP is added to experimental animals signi¢cantly altered the time course of recovery at 10312 M ATP (Fig. 2B), 1036 M ATP (Fig. 2E) and 1034 M ATP (Fig. 2F) as compared to RP only controls (Fig. 1A,B). Maximal enhancement of the recovery rate occurred at 10312 M and 1036 M ATP. At these concentrations of ATP, vibration sensitivity recovered 2 min after the addition of ATP and RP. Other concentrations of ATP tested did not signi¢cantly a¡ect the time course of recovery (Fig. 2A,C,D; ANOVA with LSD post-hoc analysis ; signi¢cance at P90.05). In the combined presence of RP and picomolar ADP (Fig. 3A), the time course of recovery was not signi¢cantly di¡erent from controls receiving RP only (Fig. 1A,B). On the other hand, in the presence of RP and micromolar ADP (Fig. 3B), the data were signi¢cantly di¡erent from RP controls. In micromolar ADP, vibration sensitivity recovered 2 min after the addition of

Fig. 4. Mean half maximal recovery of vibration sensitivity. Half maximal recovery was estimated by eye from computer ¢t lines drawn from the data plotted in Figs. 1^3 (see Section 2 for additional details). Mean half maximal recovery is plotted ( þ S.D.) for n = 2 replicates. Mean half maximal recovery signi¢cantly di¡ering from RP only controls is indicated by asterisks (ANOVA with LSD post hoc analysis, P 6 0.05). A: Half maximal recovery in RP only (C) or in RP with ATP at the concentration indicated on the x-axis. B: Half maximal recovery in RP only or in RP with speci¢c nucleotides as indicated on the x-axis.

ADP and RP. Neither AMP (Fig. 3C,D) nor AMPPNP (Fig. 3E,F) signi¢cantly altered the time course of recovery as compared to RP only controls. Repair protein used to generate the data plotted in Fig. 3F (open circles) was saved after the experiment, dialyzed to remove AMP-PNP, and then tested for repair bioactivity (Fig. 1C). This `used' RP was found to have normal bioactivity (compare Fig. 1A,B with C). To further evaluate nucleotide e¡ects on the rate of

HEARES 3265 24-9-99

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

7

Fig. 5. Mean luminescence of luciferase/luciferin in seawater samples taken from anemone cultures. See Section 2 for additional details. Luminescence values for ATP standards are plotted within each panel (horizontal lines) with the log molar concentration of standards appearing in the right hand margin of each panel. A: Mean luminescence ( þ S.D., n = 3 readings/sample) as a function of time for seawater samples obtained from an experimental anemone culture transferred to calcium free seawater for a 1 h incubation, and then transferred to natural, Millipore ¢ltered, seawater (time zero) for a 5 h recovery. B: Mean luminescence for seawater samples obtained from a control anemone culture treated as described in A, except that natural, Millipore ¢ltered (calcium containing) seawater was used both for the 1 h incubation and for the 5 h recovery. C: Mean luminescence for seawater samples obtained from an experimental culture of anemones treated as in A, except that RP was added at time zero. D: Mean luminescence for a control culture of anemones treated as in B.

recovery, the time required to achieve half maximal recovery (K0:5 ) was estimated from the data shown in Figs. 1^3. Because replicate experiments were performed for each treatment, two di¡erent estimates were obtained per treatment, allowing statistical comparisons. In the event that no discernible recovery occurred, (e.g., Fig. 3F, open circles), the K0:5 was arbitrarily assigned a value of 10 min. Thus, a possible inhibition of recovery was not considered here. In RP alone, the K0:5 ranged from 6.6 to 7.0 min (Fig. 4). In the presence of RP and ATP, the estimated K0:5 was signi¢cantly di¡erent from RP only controls at 10312 M, 1036 M, and 1034 M ATP (Fig. 4A). In 10312 M ATP or in 1036 M ATP, the K0:5 was approximately 1.5 min (Fig. 4A). In addition, the estimated K0:5 for RP with 1036 M ADP (at approximately 1.8 min) was signi¢cantly di¡erent from RP only controls (Fig. 4B).

3.2. Extracellular levels of ATP To determine if ATP is secreted or consumed during the recovery period, experimental animals were exposed to calcium free seawater for 1 h and then returned to calcium containing seawater where they were allowed to recover vibration sensitivity over the next 5 h. One or two (Fig. 5A) transient increases in ATP were detected in seawater from experimental animals within 2^4 h after they were returned to calcium containing seawater. In control experiments, animals were placed in calcium containing seawater for 1 h, then transferred to calcium containing seawater for 5 h. No transient increases in ATP were detected in seawater containing control anemones (Fig. 5B). In a separate experiment, RP was added to seawater containing experimental animals. An ATP transient was detected 7^8 min after

HEARES 3265 24-9-99

8

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

adding RP to seawater from experimental animals (Fig. 5C). No such transient was detected in seawater from controls animals not given RP (Fig. 5D). 3.3. Nucleotidase cytochemistry of repair proteins Concentrated repair protein mixtures or concentrated fraction L processed for nucleotidase cytochemistry showed a honeycombed array of ¢lamentous structures. At 10312 M ATP, the network of ¢laments included electron dense reaction product at numerous, small loci (Fig. 6A). At 1036 M ATP, the electron dense loci were less abundant, but the ¢lamentous network remained obvious (Fig. 6B). Walls on opposite sides of the honeycomb were separated by approximately 134 þ 16 nm (mean þ S.D., n = 10). Occasionally, isolated complexes that appeared to have di¡used away from the primary mass of repair protein were observed shaped like the letter Y (Fig. 7; n = 6 at each ATP concentration). These Y-shaped complexes were morphologically similar to fraction L (Watson et al., 1998b). At 10312 M ATP, the majority of electron dense reaction product indicative of ATPase activity was re-

Fig. 7. Nucleotidase cytochemistry of repair protein. A: An individual Y-shaped complex is shown in 10312 M ATP with reaction product in the head domain (H) and in the proximal region of the tail domain (T). B: A di¡erent example in 10312 M ATP in which the distal distribution of reaction product in the arms of the head domain (H) is more clearly indicated than in the previous example although the tail domain (T) is more di¤cult to see clearly. C: At 1036 M ATP, reaction product occurs distally in the tail domain and in the head domain near the junction (J) interconnecting the head (H) and tail domains (T). D: Overlay indicating complementary distribution of reaction product observed at 10312 M ATP (white) and 1036 M ATP (black). The junction shows enzyme activity in each (dotted). The Y-shaped complexes were rotated around the branch point to align the two complexes. Scale bar = 100 nm.

stricted to the `head' domain where it occurred distally (i.e., away from the junction of the head and tail domains) in both arms of the Y-shaped complex (Fig. 7A,B). At 1036 M ATP, the majority of reaction product occurred distally in the tail domain (Fig. 7C). Mapping reaction product observed at each of these two ATP concentrations (from the best examples) showed that reaction product distributed across the Y-shaped complex in a complementary fashion (Fig. 7D). 3.4. Nucleotide e¡ects on elution of repair protein

Fig. 6. Nucleotidase cytochemistry of aggregated repair protein. A: In 10312 M ATP, electron dense reaction product occurs at numerous punctate loci within the network of aggregated repair protein. B: In 1036 M ATP, reaction product is more evenly distributed along the ¢laments comprising the aggregated repair protein. Scale bar = 200 nm.

Repair protein has a tendency to self interact to form large aggregates. Concentrated repair protein comprises a mixture of individual RP complexes and aggregates of RP complexes (unpublished). To determine whether including ATP in the mobile phase would improve the yield of protein eluted, concentrated repair protein was injected onto a Sepharose 2B column, which does not signi¢cantly sieve individual repair protein complexes (Watson et al., 1998b). We reasoned that the 2B column would likely prevent elution of repair protein aggregates. After the individual RP complexes were

HEARES 3265 24-9-99

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

9

ing a conventional epi£uorescence microscope enhanced the quality of the image by digitally removing glare (Fig. 9B). Deconvolved images revealed labeling in a radial pattern that corresponds in size and shape to sensory cells (compare Fig. 9B with C). The image shown in Fig. 9B showed £uorescence at the base of the hair bundle where the large diameter stereocilia extended from the apical cell surface of the sensory cell.

Fig. 8. E¡ects of ATP on elution of repair protein from Sepharose 2B columns. The mean relative area of peaks eluted is plotted (dark bars þ S.D., n = 2) as a function of estimated ATP concentration in the mobile phase. Bioactivity of the fractions containing eluted protein is indicated by enhanced discharge of nematocysts into vibrating targets (empty bars þ S.E.M.). Asterisks indicate a statistically signi¢cant increase in discharge relative to untreated controls (not receiving RP fractions). ANOVA with LSD post-hoc analysis, P90.05.

eluted using a mobile phase lacking ATP, the column was exposed to a series of ATP gradients ranging from zero to 1034 M ATP. As ATP levels were increased to 1034 M ATP, four peaks of eluted protein were detected. Beginning at approximately 10312 M ATP, a small peak appeared comprising 3.6 þ 0.1% of the total area eluted in the presence of ATP (Fig. 8, n = 2 replicate experiments). A second peak was observed beginning at approximately 10310 M ATP comprising 5.8 þ 1.9% of the total area. A third peak appeared beginning at approximately 1038 M ATP comprising 9.6 þ 0.8% of the total area detected. Finally, a large peak was detected at approximately 3U1035 M ATP comprising 82.5 þ 5.1% of the total area detected in response to ATP (Fig. 8). Using the bioassay for vibration sensitivity, bioactivity was con¢rmed in three of the four peaks indicating the presence of repair protein. Only the peak eluted at 10310 M ATP failed to increase vibration dependent discharge of nematocysts signi¢cantly above control animals. Control animals were anemones which lost vibration sensitivity after 1 h exposure to calcium free seawater but which were not subsequently exposed to any of the RP fractions (Fig. 8). 3.5. Quinacrine labeling of specimens Fluorescence microscopy of excised tentacles labeled with quinacrine showed punctate £uorescence distributed at discrete loci (Fig. 9A). Deconvolution applied to images of quinacrine labeled specimens obtained us-

Fig. 9. Epi£uorescence microscopy of excised anemone tentacles labeled with quinacrine. See Section 2 for a detailed methodology. A: Digital image of a tentacle tip showing punctate £uorescence at discrete loci. B: The same image is shown after maximum entropy deconvolution to remove some of the glare. A sensory neuron (n) is labeled. C: A reference image of sensory neuron (n) shown in cross section after it was prepared for routine transmission electron microscopy. Stereocilia arising from the surrounding supporting cells (s) also are shown. Scale bar = 1 Wm.

HEARES 3265 24-9-99

10

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

Fig. 10. Anemone hair bundle processed for routine transmission electron microscopy shown in an oblique section. Apically, the sensory neuron (n) and supporting cell (s) contain mitochondria (m). The supporting cell contains endosomes (e) and the sensory cell contains electron lucent vesicles (v) of an unknown function. Scale bar = 0.5 Wm.

Transmission electron micrographs of sensory cells revealed an abundance of small vesicles in sensory cells and supporting cells located in the apical cytoplasm adjacent to stereocilia rootlets. Mitochondria also occur in this area of these cell types (Fig. 10). 4. Discussion The structural integrity of anemone hair bundles is completely disrupted and vibration sensitivity is abolished by 1 h exposure to calcium free seawater. However, hair bundles regain normal structural integrity after 4 h in calcium containing seawater. Likewise, vibration sensitivity is restored in 4 h. Such repair of anemone hair bundles is conferred by large protein complexes secreted by the anemone called repair proteins (Watson et al., 1998b). Exogenously supplied RP consistently restore vibration sensitivity in 7^8 min (Fig. 1A,B). 4.1. Evidence that ATP is required for recovery of vibration sensitivity In the present study, we begin to unravel the molec-

ular basis of repair. Several lines of evidence indicate that ATP is required for repair. First, based on the bioassay for vibration sensitivity in the presence of RP, we ¢nd that exogenously supplied ATP signi¢cantly enhances the rate of recovery. The dose responses to ATP suggest the existence at least two di¡erent ATPases, one that is active at 10312 M ATP (named `high a¤nity'), and another that is active at 1036 M ATP (named `low a¤nity'). The ATPases are likely to be distinct from each other based on their dose sensitivity to ATP and preference for ATP versus ADP or AMP. Whereas the high a¤nity ATPase is speci¢c for ATP, showing no enhanced rate of recovery in the combined presence of RP and ADP or AMP, the low a¤nity ATPase does not discriminate between ATP and ADP. Apparently, the low a¤nity ATPase cannot e¤ciently utilize AMP since, in the combined presence of RP and AMP, no enhancement in the recovery rate is observed versus that in RP alone. The poor speci¢city of the lower a¤nity ATPase for ATP versus ADP, but not AMP, is typical of other known ecto-ATPases (Plesner, 1995). Ecto-ATPases comprise a family of large glycoproteins ubiquitous in eukaryotic cells (Plesner, 1995). At least some ectoATPases appear to be cell adhesion molecules (Lin et al., 1991; Plesner, 1995). Our observation of ATPase activity in repair proteins, which apparently function to interconnect stereocilia (Watson et al., 1998b), suggests that repair proteins may be related to certain cell adhesion molecules. Second, nucleotidase cytochemistry con¢rms ATPase activity within isolated repair protein. These results indicate that repair protein directly hydrolyzes ATP. The appearance and distribution of enzymatic reaction product is di¡erent at 10312 M ATP than at 1036 M ATP. Whereas at 10312 M ATP, punctate labeling is observed at numerous loci within aggregated repair protein, at 1036 M ATP, labeling is observed along the length of ¢laments comprising the aggregated repair protein. The possibility that the high a¤nity and low a¤nity ATPases reside together on the same repair protein complexes is supported by limited observations (n = 6) of Y-shaped complexes that di¡used away from the aggregated repair protein during processing of the protein sample for TEM. The Y-shaped complexes show high a¤nity ATPase activity in the head domain where it occurs distally along the two arms and low a¤nity ATPase activity distally in the tail domain. Although these results are tentative, they suggest that high a¤nity and low a¤nity ATPases can coexist on the same molecule. Furthermore, these results suggest that the high a¤nity ATPase may be inactive at relatively high ATP concentrations (i.e., micromolar). Third, based on the luciferin/luciferase biolumines-

HEARES 3265 24-9-99

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

cence assay, anemones with experimentally disrupted hair bundles caused by 1 h exposure to calcium free seawater both secrete and consume ATP during the normal 4 h time course of recovery. Control anemones do not routinely secrete signi¢cant levels of ATP. Fourth, based on quinacrine cytochemistry, it appears that sensory neurons contain stores of ATP. Sensory neurons reside at the center of anemone hair bundles. Anemone hair bundles consist of large diameter stereocilia and a kinocilium from the sensory neuron and numerous small diameter stereocilia arising from the surrounding supporting cells. The apical cytoplasm of sensory cells stains with quinacrine indicating likely stores of adenine nucleotides. Transmission electron microscopy of anemone tentacles indicates that the apical cytoplasm of sensory neurons and supporting cells is ¢lled with numerous vesicles and mitochondria. Although all of these cell types contain mitochondria in their apical cytoplasm, the quinacrine £uorescence arises from sensory cells only. Thus, mitochondria apparently do not contribute signi¢cantly to the £uorescent signal. Assuming that vesicles in sensory cells store ATP, then ATP is released by exocytosis directly into the hair bundle originating from the same cells, namely sensory and supporting cells. 4.2. Possible interactions between extracellular levels of ATP and RP Secreted ATP would di¡use outward from the base of the hair bundles into the seawater. One possible function of the ATP may be to recruit individual RP complexes to the hair bundle from aggregated RP. The rationale behind this idea is based on our experimental observation that ATP enhances elution of individual RP complexes from aggregated RP trapped on Sepharose 2B columns. Perhaps the default state of secreted repair protein is in an aggregated form. The aggregated RP may be more readily retained in the mucous coat covering the tentacle epithelium than smaller, individual RP complexes. In the presence of exogenously supplied, concentrated RP (which consists of a mixture of individual RP complexes and aggregated RP), experimental animals secrete and consume signi¢cant levels of ATP within 10 min after the addition of RP. Thus, RP apparently induces ATP secretion. It appears that ATP may recruit individual RP complexes from aggregated RP and the individual RP complexes may induce further ATP secretion. The ¢nding that RP induces ATP secretion may help to explain relatively high variability observed in data from experiments incorporating nucleotides other than ATP [i.e., compare ATP data (Fig. 2) with non-ATP data (Fig. 3)]. For example, in micromolar AMP one of

11

the replicate experiments shows normal recovery while the other replicate experiment shows essentially no recovery (Fig. 3D). Perhaps RP induced ATP secretion in the replicate for which normal recovery occurred but not in the other replicate in which recovery did not occur (Fig. 3D). 4.3. Extracellular ATP in cochlea ATP is present in cochlear £uids at concentrations of approximately 10^20 nM (Munoz et al., 1995). The ATP may be secreted by cells residing in the stria vascularis (White et al., 1995). In vivo applications of ATP to cochlear £uids results in changes both to resting potentials and sound evoked electrical potentials (Kujawa et al., 1994). ATP exerts it e¡ects on cell behavior via P2 purinoreceptors which can act as ligand gated channels (P2X receptors) or as G-protein coupled receptors (P2Y receptors). Inner hair cells and outer hair cells possess P2X receptors permeable to Na‡ , K‡ and Ca2‡ (Ashmore and Ohmori, 1990; Nakagawa et al., 1990 ; Sugasawa et al., 1996). Recent evidence localizes P2X receptors to stereocilia of inner and outer hair cells (Housley et al., 1998). Presumably, ATP gated channels modulate sensitivity of the hair cells by manipulating the driving force for ions involved in mechanoelectrical transduction (Housley et al., 1998). Whereas a considerable body of information supports a neurohumoral function for extracellular ATP in cochlea, results from the present study suggest that ATP may have an additional role related to the restoration of disrupted linkages between stereocilia. Such a repair mechanism is now known to occur in hair cells of birds (Zhao et al., 1996), but a possible involvement of ATP in this repair mechanism of birds is not yet known. 4.4. Concluding remarks In the present paper evidence is presented that ATP is required for restoring vibration sensitivity to hair bundles on anemone tentacles disrupted by exposure to calcium free seawater. Although ATPase activity localizes to repair protein complexes, the precise function of ATPases in the repair process remains unknown. Nevertheless, the results presented in this paper suggest that repair proteins are active participants in the repair process. Assuming that repair proteins are replacement linkages that interconnect stereocilia to restore structural integrity and function to the hair bundle (as was proposed by Watson et al., 1998b), then the results of this study indicate that linkages interconnecting stereocilia may be enzymatically active biomolecules.

HEARES 3265 24-9-99

12

G.M. Watson et al. / Hearing Research 136 (1999) 1^12

Acknowledgements We thank Dr. Patricia Mire for assistance with quinacrine cytochemistry and for critically reading the manuscript. We appreciate support from NSF (MCB9505844) and NIH (R01GM52334). References Ashmore, J.F., Ohmori, H., 1990. Control of intracellular calcium by ATP in isolated outer hair cells of the guinea pig cochlea. J. Physiol. 428, 109^131. Assad, J.A., Shepherd, G.M.G., Corey, D.P., 1991. Tip link integrity and mechanotransduction in vertebrate hair cells. Neuron 7, 985^ 994. Budelmann, B.U., 1988. Morphological diversity of equilibrium receptor systems in aquatic invertebrates. In: Atema, J., Fay, R.R., Popper, A.N., Travolga, W.N. (Eds.), Sensory Biology of Aquatic Animals, Springer, New York, pp. 757^782. Housley, G.D., Raybould, N.P., Thorne, P.R., 1998. Fluorescence imaging of Na‡ in£ux via P2X receptors in cochlear hair cells. Hear. Res. 119, 1^13. Kujawa, S.G., Erostegui, C., Fallon, M., Crist, J., Bobbin, R.P., 1994. E¡ects of adenosine 5P-triphosphate and related agonists on cochlear function. Hear. Res. 76, 87^100. LeMasters, J.J., Hackenbrock, C.R., 1978. Fire£y luciferase assay for ATP production by mitochondria. Methods Enzymol. 57, 36^ 57. Lin, S.H., Culic, O., Flanagan, D., Hixson, D.C., 1991. Immunochemical characterization of two isoforms of rat liver ecto-ATPase that show an immunological and structural identity with a glycoprotein cell adhesion molecule with Mr 105000. Biochem. J. 278, 155^161. Mire, P., Watson, G.M., 1997. Mechanotransduction of hair bundles arising from multicellular complexes in anemones. Hear. Res. 113, 234^244. Munoz, D.J.B., Thorne, P.R., Housley, G.D., Billett, T.E., 1995. Adenosine 5P-triphosphate (ATP) concentrations in the endolymph

and perilymph of the guinea pig cochlea. Hear. Res. 90, 119^ 125. Nakagawa, T., Akaike, N., Kimisuki, T., Komune, S., Toshio, A., 1990. ATP-induced current in isolated outer hair cells of the guinea pig cochlea. J. Neurophysiol. 63, 1068^1074. Plesner, L., 1995. Ecto-ATPases: Identities and functions. Int. Rev. Cytol. 158, 141^214. Preyer, S., Hemmert, W., Zenner, H.P., Gummer, A.W., 1995. Abolition of the receptor potential response of isolated mammalian outer hair cells by hair bundle treatment with elastase: a test of the tip link hypothesis. Hear. Res. 89, 187^193. Sugasawa, M., Erostegui, C., Blanchet, C., Dulon, D., 1996. ATP activates non-selective cation channels and calcium release in inner hair cells of the guinea pig cochlea. J. Physiol. 491, 707^718. Watson, G.M., Hessinger, D.A., 1992. Receptors for N-acetylated sugars may stimulate adenylate cyclase to sensitize and tune mechanoreceptors involved in triggering nematocyst discharge. Exp. Cell Res. 198, 8^16. Watson, G.M., Hudson, R.R., 1994. Frequency and amplitude tuning of nematocyst discharge by proline. J. Exp. Zool. 268, 177^185. Watson, G.M., Mire, P., 1998. A comparison of hair bundle mechanoreceptors in sea anemones and vertebrate systems. Curr. Topics Dev. Biol. 43, 51^84. Watson, G.M., Mire-Thibodeaux, P., 1994. The cell biology of nematocysts. Int. Rev. Cytol. 156, 275^300. Watson, G.M., Mire, P., Hudson, R.R., 1997. Hair bundles of sea anemones as a model system for vertebrate hair bundles. Hear. Res. 107, 53^66. Watson, G.M., Mire, P., Hudson, R.R., 1998a. Frequency speci¢city of vibration dependent discharge of nematocysts in sea anemones. J. Exp. Zool. 281, 582^593. Watson, G.M., Mire, P., Hudson, R.R., 1998b. Repair of hair bundles in sea anemones by secreted proteins. Hear. Res. 115, 119^128. White, P.N., Thorne, P.R., Housley, G.D., Mockett, B., Billet, T.E., Burnstock, G., 1995. Quinacrine staining of marginal cells in the stria vascularis of the guinea pig cochlea: a possible source of extracellular ATP? Hear. Res. 90, 97^105. Zhao, Y., Yamoah, E.N., Gillespie, P.G., 1996. Regeneration of broken tip links and restoration of mechanical transduction in hair cells. Proc. Natl. Acad. Sci. USA 94, 15469^15474.

HEARES 3265 24-9-99