REBINDING OF TETRAHYMENA 13 S AND 21 S DYNEIN ATPases TO EXTRACTED DOUBLET MICROTUBULES

Transcription

1 J. Cell Sd. 77, (1985) 263 Printed in Great Britain The Company of Biologists Limited 1985 REBINDING OF TETRAHYMENA 13 S AND 21 S DYNEIN ATPases TO EXTRACTED DOUBLET MICROTUBULES THE INNER ROW AND OUTER ROW DYNEIN ARMS FRED D. WARNER, JOHN G. PERREAULT AND JANE H. McILVAIN Department of Biology, Biological Research Laboratories, Syracuse University, Syracuse, New York 13210, U.SA. SUMMARY Ciliary axonemes from Tetrahymena contain a second salt-extractable ATPase distinguishable from outer arm 21 S dynein by sedimentation velocity (= 13 S), electrophoretic mobility and substrate specificity. As characterized by turbidimetric assay, gel electrophoresis in the presence of sodium dodecyl sulphate, ATPase activity and electron microscopy, the 13 S dynein ATPase rebinds to extracted doublet microtubules. Compared to structural-side (ATP-insensitive) 21 S dynein binding, which is moderately specific for the 24 nm outer row arm position, rebinding of 13 S dynein is highly specific but for the inner row arm position. However, 13 S dynein rebinds to the A subfibre with a spacing that coincides with the triplet spacing of the radial spokes ( nm periods; 96 nm repeat). All of the major protein components present in the 13 S or 21 S fractions rebind to extracted doublets under conditions that both restore and activate dynein ATPase activity. Unlike active-side (ATP-sensitive) rebound 21 S dynein, rebound 13 S dynein is completely insensitive to dissociation by ATP-vanadate and does not independently decorate the B subfibre. The saturation profile for rebinding of 13 S dynein exhibits a lack of cooperativity between binding events (h = 1-0) similar to structural-side rebinding of 21 S dynein. At low 21 S/doublet stoichiometry there is no measureable competition between the 13 S and 21 S dyneins for binding sites on the A subfibre lattice, although at saturating concentrations of 21 S dynein, rebinding of 13 S dynein is blocked completely. INTRODUCTION Axonemal doublet microtubules slide along one another in an active, tipward direction (Sale & Satir, 1977) as a result of dynein ATPase crossbridge cycling. Sliding is regulated by one or more unknown mechanisms to produce an effective-recovery stroke cycle in cilia or to propagate principal and reverse bends in flagella or sperm tails. All of the information required to express this behaviour is contained within doublet-doublet interactions (Nakamura & Kamiya, 1970), interactions mediated solely by dynein arms and interdoublet links (Warner, 1983a). 21 S dynein ATPase has been thought to be the principal force-generating enzyme in both Tetrahymena cilia and sea-urchin sperm, although structurally distinct outer row and inner row dynein arms have always been recognized. The 21 S particle repeats at 24 nm along the A subfibre and appears to be restricted to the outer row Key words: dynein ATPases, inner and outer arms, cilia.

2 264 F. D. Warner, J. G. Perreault andj. H. Mcllvain arm position (Ogawa, Negishi & Obida, 1982). The isolated 21 S particle is a large, multipeptide aggregate of uncertain, but probably species-specific, composition containing three high molecular weight ATPases in Tetrahymena (Porter & Johnson, 1983) and two in sea-urchin (Bell & Gibbons, 1982). Both sea-urchin and Tetrahymena 21 S dynein readily recombine with extracted doublet microtubles (Gibbons & Gibbons, 1979; Mitchell & Warner, 1980). In contrast to Tetrahymena 21 S dynein, outer arm dynein from Chlamydomonas dissociates during isolation into 18 S and 12 S subunits containing two and one high molecular weight ATPases, respectively (Pfister, Fay & Witman, 1982; Piperno & Luck, 1979). Rebinding of outer arms to extracted doublets requires both subunits (Fay & Witman, 1977). Analysis of Chlamydomonasflagellar mutants lacking either outer or inner row arms shows that the outer and" inner arm dyneins are different proteins, although inner arm dynein also may comprise two different subunits sedimenting at S (Piperno & Luck, 1981). Unlike outer arm dynein, which is easily obtained at high ionic strength extraction, inner arm dynein is somewhat resistant to standard extraction procedures in both Chlamydomonas and sea-urchin sperm. In Tetrahymena, however, inner arms and a S dynein ATPase are readily extracted along with 21 S outer arm dynein (Gibbons, 1966; Gibbons & Rowe, 1965; Mitchell & Warner, 1981). Each ATPase has a different electrophoretic profile (Mabuchi & Shimizu, 1974; Porter & Johnson, 1983; Zanetti & Warner, 1982), although the function of inner arm dynein, if different from outer arm dynein, is unknown. Extraction of seaurchin 21 S dynein and outer row arms reduces the beat frequency of reactivated axonemes but only to about half its normal value (Gibbons & Gibbons, 1973), suggesting that both inner and outer row ATPases have comparable capabilities in the generation of sliding. Rebinding of 21 S dynein restores outer arms and returns beat frequency (Gibbons & Gibbons, 1979) or sliding velocity (Yano & Miki-Noumura, 1981) to near normal levels. Rebinding of Tetrahymena 21 S dynein to extracted doublets also returns arms to the outer row position (Mitchell & Warner, 1980). Heretofore, specific rebinding of either inner arms or Tetrahymena 13 S dynein has not been successful (Gibbons, 1965; Porter & Johnson, 1983; Shimizu, 1975). Consequently, we do not know the location or action of the enzyme or its potential for interaction with 21S dynein or other components to produce regulated motion. This limitation has been exacerbated by the lack of a functional assay to distinguish between the contributions of each ATPase. We now report specific rebinding of 13 S dynein ATPase to the inner row arm site of extracted doublet microtubules. MATERIALS AND METHODS Dynein and microtubule preparations Cilia were isolated from litre cultures of Tetrahymena thermopfnla, mucocyst-free strain B-255, by methods described elsewhere (Mitchell & Warner, 1980). All isolation, extraction and resuspension solutions contained 0-1 mm-egta and 0-1 mm-phenylmethylsulphonyl fluoride (PMSF) or 0-01 mm-leupeptin. Cilia were demembranated by one exposure to O'l % Nonidet P-40, collected by centrifugation,

3 Inner arm and outer arm dynein ATPases 265 and resuspended in 20mM-HEPES, O-lmM-EGTA (ph7-4). Sequential dynein extraction was done in 0-5 M-KC1 for 25 min, followed by desalting and low ionic strength dialysis against 1 mm- HEPES, O-lmM-EGTA, for 24 h. Some preparations were extracted for a third time in 0-5 M-KC1 for 18 h. Twice-extracted doublets were collected by centrifugation and resuspended in 6 mm-mgck, 20mM-HEPES, 0-1 mm-egta (ph7-4). Dynein containing supernatant fractions from all three extractions were dialysed to lomm-hepes, 0-1 mm-egta and concentrated to mgml" 1 on an Amicon PM-30 membrane. The concentrated extracts were fractionated on calibrated linear 0 to 30% sucrose density gradients (Mitchell & Warner, 1981). ATPase-containing fractions were redialysed to lomm-hepes, 0-1 mm-egta and reconcentrated to l-2mgml~' on the Amicon membrane. Dynein fractions from the first KC1 extraction only in combination with twice-extracted (2 X) axonemes were used in subsequent experiments. Dynein binding assays Rebinding of 13 S and 21 S dynein fractions was monitored by the quantitative turbidimetric assay (350nm) characterized earlier (Mitchell & Warner, 1980, 1981; Warner & Mcllvain, 1982). All turbidity measurements were made in a total reactant volume of 0-5 ml maintained at 16 C in the spectrophotometer. Readings were standardized at 10 min post-dynein addition and 1 min post- ATP addition. All dynein binding assays were done in 6mM-MgCl2, 0-lmM-EDTA, 20 mm- HEPES (ph7-4). The reactions were initiated by adding samples of 13 S or 21 S dynein to lofig of twice-extracted doublets. Potential dissociation of the resulting tubulin-dynein complex was tested by addition of 20/JM-ATP, 10/iM-sodium metavanadate. For 21 S dynein, turbidity changes at 350 nm are directly proportional to changes in bound versus soluble protein and, hence, are a quantitative measure of dynein binding to the microtubule lattice after correcting for volume changes and the inherent turbidity of the isolated dynein fractions (Mitchell & Warner, 1981; Porter & Johnson, 1983). Standard curves for 13 S binding were generated by cosedimentation of sucrose-gradient purified 13 S dynein with twice-extracted doublets (Mitchell & Warner, 1981). The range of the turbidity increase that results from dynein binding to the microtubules, after subtracting the turbidity contributed by protein alone, iso unit for 13 S dynein and unit for 21 S dynein relative to 20 n% ml" 1 of twice-extracted (2 X) doublets. Protein concentrations for all procedures were determined by the Coomassie Blue dye binding technique (Bradford, 1976) as modified by Read & Northcote (1981), using bovine serum albumin (BSA) as protein standard. Dynein ATPase assays were done at 20 C in a standard assay buffer containing 6mM-MgCl2, 0-lmM-EDTA, 40mM-HEPES, 1 mm-atp (ph7-4). The reaction was initiated by addition of ATP and terminated by addition of cold trichloroacetic acid. The inorganic phosphate released was measured (Awonm) by the method of Taussky & Shorr (1953). Specific activity was calculated from the linear region of phosphate release after correcting for phosphate concentration at to. Gel electrophoresis Discontinuous polyacrylamide gel electrophoresis was done in the presence of sodium dodecyl sulphate (SDS) using a slab gel apparatus. The stacking gel contained 3 % acrylamide, 0-1 % SDS, 125 mm-tris-hcl buffer (ph6-8) and was polymerized with 0-11% ammonium persulphate, 0-5 % TEMED. The separating gel consisted of a 5 % to 15 % acrylamide gradient in 0-1 % SDS, 375 mm-tris-hcl (ph8-8) and polymerized with 0-02% ammonium persulphate, 0-048%TEMED. Protein samples were denatured in a solution of 1%SDS, 1%/J-mercaptoethanol, 62-5 mm-tris-hcl (ph6-8), and heated in a boiling water bath for 5 min. Electrophoresis was carried out in a reservoir buffer containing 1%SDS, l-92m-glycine, 250mM-Tris-HCl (ph8-3). Gels were stained in 0-25% Coomassie Blue, 50% methanol, 10% acetic acid and destained in methanol/acetic acid (v/v, 5:1). Molecular weight standards used were ovalbumin (Af r 43OO0) and glutaraldehyde-crosslinked oligomers of bovine serum albumin (Af r 68000; n = 2-8) (Warner et al. 1977). Electron microscopy For thin-section electron microscopy, 13 S and 21 S dynein-decorated doublet preparations were pelleted at gi, f r 30 min and fixed for 1-5 h in 2% ultrapure glutaraldehyde in 20 ITIM-

4 266 F. D. Warner, J. G. Perreault andj. H. Mcllvain HEPES (ph7-4), and containing 6mM-MgCfe and 1 % tannic acid. Fixation solutions for the 21 S preparations also contained 20 ^M-ATP, 10 /au-vanadate. The pellet was rinsed in buffer and postfixed for 1 h in 1 % OsC^ in the same buffers. Silver-gold interference colour sections were stained in 4% aqueous uranyl acetate and Reynolds' lead citrate. In addition to thin-section microscopy, dynein-decorated doublet preparations and twice-extracted axonemes suspended in 10 mm-hepes, 4mM-MgCl 2, O-lnuuEGTA, 10/M-ATP, 10/an-vanadate (ph7-4) were kindly prepared by Dr John E. Heuser for rapid-freeze, deep-etch electron microscopy (Goodenough & Heuser, 1982; Heuser, 1980). RESULTS Characterization of dynein and tubulin reactants Nonidet-demembranated axonemes were sequentially extracted by 0-5M-KC1 (KC1-1) and low ionic strength dialysis (LIS) to remove soluble ATPases from the axoneme. This was sometimes followed by a third extraction in 0-5 M-KC1 (KC1-2) for 18 h. On the basis of the concentration of 21 S dynein calculated to be present in Tetrahymena axonemes (Table 1), double (2x) extraction removes 83% of total ATPase and most components recognizable as dynein arms from the axonemal profile, resulting in groups of 9 or fewer doublets held into tight arrays by the interdoublet links (Warner, 1983a). Typical sucrose-gradient profiles from a triple extraction procedure are illustrated in Fig. 1. KC1-1 extraction removed about 68 % of total 13 S and 21 S ATPase from the axoneme. Subsequent LIS extraction removed an additional 15 % of the ATPase, and an additional 8 % was removed by the KC1-2 extraction; 9 % of total ATPase protein is unextractable by these methods. Extraction stoichiometry is summarized in Table 1. The mass ratio of extractable 21 S/13 S dynein is l-6± 0-01 S.E.M. (n= 15 preparations). The polypeptide composition of Tetrahymena dyneins has not been thoroughly characterized, although several studies have clearly demonstrated the absence of homogeneity between proteins found in the 21 S and 13 S fractions (Mabuchi & Shimizu, 1974; Porter & Johnson, 1983; Zanetti& Warner, 1982). Both 21 Sand 13 S dyneins isolated by a single high-salt extraction and purified on a linear sucrose-density gradient, exhibit considerable apparent purity and generally are free from copurifying Table 1. Structural-side stoichiometry for the 13 S and 21S dynein ATPases Predicted Extracted Rebound concentration* concentration! concentrationj Enzyme (mg D/mg T) (mg D/mg T) (mg D/mg T) 13S ± ± S ± ±0-042 Concentrations in vivo, predicted on the basis of single row saturation and M r 13S = 6xlO s, 32 nm periodicity; M T 21 S = 2x 10*, 24nm periodicity. f Concentrations derived from sucrose-gradient fractionated triple-extraction procedure. All values are the mean ± S.E.M. for 15 preparations. Saturation values (6mu) estimated by eqn (1), h = 1-0. Concentrations expressed as mg dynein (D)/mg extracted doublets (T).

5 Inner arm and outer arm dynein ATPases S o 1 1 i KCI / O o o o o / / 1 \ J/ r 13 S o o 1I \ / Q _ A A " A \ 1 I\ A V A A A L1S / / / / / »-- a N> a - D D D O a o y _ KCI-2 V. - 1 (bottom) 1 I i i Fraction number I Fig. 1. Linear 0 to 30 % sucrose density gradient profiles of the three sequential dynein extracts from Tetrahymena axonemes. Both protein (A280nm, (^)) and ATPase activity (A660nm, (O)) are illustrated. ATPase containing fractions separate into peaks at 21 S and 13 S for each of the three successive extractions. KCI-1, 0-5M-KC1, 25 min extraction; LIS, low ionic strength, 24h extraction; KC1-2, 0-SM-KC1, 18h extraction. tubulin when examined by SDS/polyacrylamide gel electrophoresis. As described below, nearly all of the polypeptides contained within the 21 Sand 13 S fractions rebind to extracted doublets under conditions that restore both dynein arms and ATPase activity.

6 268 F. D. Warner, J. G. Perreault andj. H. Mclhain Recombination of 13 S and 21S dynein with extracted axonemes Both 13 S and 21 S dynein were allowed to react with twice-extracted axonemes after suspending each reactant in 6mM-MgCl2, 20mM-HEPES, (MmM-EGTA (ph7-4). The reaction was monitored at 350 run and terminated by addition of 20/iM-ATP, 10/iM-vanadate. For initial comparisons of rebinding of the 13 S and 21 S dynein fractions, proportions of dynein and doublets were adjusted to yield similar turbidimetric responses. Typical spectrophotometric tracings of the responses at low stoichiometry are illustrated in Fig. 2. Addition of 21 S dynein to the axoneme sample (2/ig21 S/10/ig doublets) results in a turbidity increase of 25-30% over baseline (Ai, Fig. 2A). When ATP and vanadate were added, a considerable portion of the turbidity increase was eliminated (A2) as a result of dissociation of ATPsensitive bound dynein from the doublet lattice (Mitchell & Warner, 1980; Warner, 19836). The remainder of the turbidity increase represents ATP-insensitive 21 S dynein, rebound principally to structural or outer arm sites on the A subfibre (Warner & Mcllvain, 1982). In comparison, addition of 13 S dynein (10jug 13 S/10/ig doublets) yields a comparable increase in turbidity (Fig. 2B). However, 13 S dynein rebound to the extracted doublets was completely insensitive to dissociation by ATP and vanadate, regardless of dynein-doublet stoichiometry and regardless of whether the substrate was added before or after dynein addition. As an initial test for possible competition between 13 S and 21 S binding, the two fractions were added sequentially and in either order to a single axoneme sample, using the same stoichiometry used in the above experiments. The resulting individual turbidity increases (Ai, AT) were the same as the increases observed when the fractions were added independently (Fig. 2c). If the 13 S and 21 S fractions were mixed before addition to the axoneme sample, the resulting turbidity increase was the same as the sum of the sequential or individual additions (Fig. 2D). The proportion of ATP-sensitive binding in the mixed preparations remained the same as for 21 S dynein alone. Because 21 S dynein has the capability nearly to saturate the doublet lattice (Warner, 19836; Warner & Mcllvain, 1982), a simple experiment was done to measure potential competition as a function of the concentration of 21 S dynein. 13 S dynein (15/ig dynein/10/ig doublets) was added after addition of 21 S dynein (2-30/ig dynein/10/ig doublets) (as in Fig. 2c). As the concentration of 21 S dynein was increased, the amount of 13 S dynein that would rebind to the doublets progressively declined, being inversely proportional to the amount of 21 S dynein bound (Db) (Fig. 3). The decline in rebinding of 13 S dynein was detected only at a 21 S concentration (D6>2/ig) that is equivalent to in vivo stoichiometry for 21 S dynein (Fig. 3, arrow on thej-axis). Because the lower limit on the sensitivity of the turbidity assay is about 0-2/ig protein bound per 10/ig doublets, we would expect to have detected competition at considerably lower concentrations of bound 21 S dynein if it had occurred. To ensure that the major polypeptides within the 13 S and 21 S dynein fractions were rebinding to extracted doublets, KC1-1 dyneins were mixed with twiceextracted doublets in the presence of 10/iM-ATP and 10/iM-vanadate, following

7 Inner arm and outer arm dynein ATPases 269 A. T A, Av ATP.v 21 S B A, ATP.v T S y C ATP.v 13 S 21 S T A,. A 2 - A," D ' \ A, T -H ATP.v I i, Time (min) i 13/21 S Fig. 2. Spectrophotometric tracings (350 nm) that record (reading right to left) additions of 2/ig21 S dynein or 10/^g 13 S dynein to 10/ig of twice-extracted doublets (T). Binding equilibrium is established in min (A i). Binding reversibility was tested by addition of 20 ^M-ATP, 10/iM-vanadate (v). The spikes in the tracings result from opening the optical chamber and mark addition of the reactants. A. Extracted doublets plus 21 S dynein plus ATP, V; B, extracted doublets plus 13 S dynein plus ATP, v; c, extracted doublets plus 21 S dynein plus 13 S dynein plus ATP, v; D, extracted doublets plus 21 S/13 S dynein (mixed) plus ATP, v.

8 270 F. D. Warner, J. G. Perreault andj. H. Mcllvain Fig. 3. Competition experiment in which 13 S dynein (1S jug dynein/10 fig doublets) was added to twice-extracted doublets that had first been mixed with increasing concentrations of 21 S dynein (as for Fig. 2c). 13 S binding (O) declines, in inverse proportion to the amount of 21 S dynein that has bound to the doublets ( ). The arrow on they-axis marks the stoichiometric saturation point for single row 21 S rebinding to the A subfibre. Scale values on the right ordinate indicate the absolute turbidity change resulting from dynein binding to the microtubules. Each data point is the mean ± S.E.M. for four experiments (n = 6 8). The curve for 21 S binding was fitted by a non-linear least-squares computer program, h = 1-0. which free protein was separated from the doublets by centrifugation. Reactant stoichiometry was adjusted so that rebound dynein was approximately equal to in vivo stoichiometry (Tablel; 13 S = 0-8/igD6/10^g doublets, 21 S = 1-7/ig Db/10/xg doublets). The doublet-dynein preparations were then analysed by SDS/polyacrylamide gel electrophoresis. The principal high, middle and low molecular weight polypeptides of each dynein fraction were restored to the extracted doublets upon incubation with the fractions (Fig. 4). Although not visible because of the reduced

9 Inner arm and outer arm dynein ATPases CO CO 271 ^3 x CM X <N CO CO X CM x CM I I I I I I b ^ 6 -^- -^ b Pi/mg 'min~' Fig S and 13 S dynein fractions from a sucrose gradient recombined with twice (2 X) extracted doublets in the presence of 10/fltf-ATP, 10/M-vanadate and centrifuged to separate bound and free dynein. Pelleted protein was then electrophoresed on an SDS/polyacrylamide gel. Reactant stiochiometry was adjusted so that the concentrations of bound dynein were approximately equal to the concentrations occurring in native axonemes (Table 1). In this preparation, three high molecular weight bands that comigrate with dynein polypeptides (lines in A) are incompletely extracted from the 2 X doublets. All of the principal bands in the high molecular weight region of the gel rebind to extracted doublets. Each dynein lane was loaded at a protein concentration of 5 fig. Each axoneme lane or axoneme plus dynein lane was loaded at 15 /Jg. The ATPase activity for each preparation (/anol P,mg protein" 1 min" 1 ) is indicated at the bottom of each lane in B.

10 272 F. D. Warner, J. G. Perreault andj. H. Mcllvain loading concentrations of protein, only two minor components in the 13 S fraction and three minor components in the 21 S fraction did not rebind to the extracted doublets. The specific ATPase activity of each preparation (/imolpimg protein" 1 min" 1 ) is indicated in Fig. 4B. Not only is ATPase activity restored to the doublets, the activity of the recombined preparations is considerably greater than the activity of the individual fractions (see below). In order to be certain that both the 13 S and 21 S ATPases were interacting specifically with the extracted doublets, we examined ATPase activity as a function of the concentration of extracted doublets added to a fixed concentration of dynein but without separating bound from free dynein. 13 S or 21 S dynein (8 fig ml" 1 ) was mixed with twice-extracted doublets ( mgrnp 1 ) in the absence of ATP and vanadate, following which ATPase activity was determined by standard assay conditions but using an ATP-regenerating system to maintain the ATP concentration at lmm (Mitchell & Warner, 1981). The data were corrected for residual ATPase activity of the extracted doublets. The specific activity of each dynein-doublet suspension (/anolpimg enzyme" 1 min" 1 ) increased in proportion to the concentration of microtubule protein added (Table 2). At the highest concentration of microtubules tested (64 :1), 13 S dynein was activated fourfold while 21 S dynein was activated threefold above the activity of microtubule-free dynein. 13 S dynein binding stoichiometry The stoichiometry of 13 S dynein binding to twice-extracted doublets was measured over most of its saturable range (1-60 fig dynein/10 fig doublets). Binding properties were evaluated by sigmoid or hyperbolic curve-fitting programs derived from the Hill equation for non-linear ligand-binding analysis. The equation for nonlinear analysis was misprinted in our original publication (Warner & Mcllvain, 1982). The correct equation (eqn (1)) is printed below. Db is the amount of dynein bound and Table 2. Effects of extracted doublet microtubules on dynein ATPase activity 2 X doublets (mgmt 1 ) S dynein (funol Pi mg E" 1 min" 1 ) (a = 4-0) 21 S dynein (/imol Pi mg E~' min~') l-84(a = 3-5) 13Sor21S dynein (8 fjg ml"') was titrated with increasing concentrations of twice-extracted (2 X) doublets. Specific activity (/imol Pi mg E~' min~'; E, enzyme) was determined by standard assay conditions after correcting for residual ATPase activity of the 2 X doublets. The activation ratio (a) is the specific activity at 0-51 mg/0 mg doublets. All data are the mean values obtained from seven preparations of cilia.

11 Inner arm and outer arm dynein ATPases Df is the amount of free dynein. The estimated parameters are b (related to affinity constants), and h (the Hill coefficient). 273 (saturation), KH Db = 13 S dynein rebinds to extracted doublets with a simple hyperbolic profile, saturating at 0-096mg dynein/mg doublets (Fig. 5). For both dyneins, use of equation (1) gives values for h that do not differ significantly from 1(13 S, h = 0-81 ±0-47; 21 S, h = 0-79 ± 0-27) and, therefore, the data are best fitted by a hyperbola (Warner & Mcllvain, 1982). However, interpretation of the 13 S curve is difficult because there (1) E co Dynein free (/ig) Fig. 5. Saturation curve for rebinding of 13 S dynein ( ) to 10 fig of twice-extracted doublets. For comparison, data for ATP-insensitive 21 S binding (unbroken line) also are included (Warner & Mcllvain, 1982). Each curve has a hyperbolic profile (h = 1*0), characteristic of a lack of cooperativity between binding events. Scale values on the right ordinate indicate the absolute turbidity change resulting from dynein binding to the microtubules. Each data point is the mean ± S.E.M. for seven preparations (n = 5-11). The curves were fitted by a non-linear least-squares computer program.

12 274 F. D. Warner, J. G. Perreault andjf. H. Mcllvain are only two data points below 6max/2 and potential sigmoid properties may have gone undetected. The amount of 13 S protein that will rebind to the doublets at saturation is 2-3-fold over the stoichiometry predicted for a single row of inner arms (Table 1; mg per mg doublets), based on an average periodicity of 32 nm (see below) and am- of 6 X 10 5 (Gibbons & Rowe, 1965). ATP-insensitive 21 S dynein saturates at a 2-6-fold excess over predicted single row saturation (Table 1; 0-179mg per mg doublets) based on an average periodicity of 24 nm and a M t of 2 X 10 6 (Johnson & Wall, 1983). Structural specificity of 13 S dynein binding The structural specificity of 13 S dynein rebound to extracted doublet microtubules was examined initially by thin-section electron microscopy. Preparations of 13 S and 21 S dyneins were mixed separately with twice-extracted axonemes in dynein/doublet proportions of 2 :1 (13 S = 0-7/igDfe/lO/ig doublets; 21 S = 1-3/igDfe/lOjUg doublets). 21 S dynein was processed in the added presence of 20/iM-ATP, 10yaAvanadate, which supports only structural-side binding to the subfibres. Outer and inner row dynein arm positions on extracted doublets are easily distinguished by the curvature of the doublet profiles and radial spoke positions. Extracted doublets remain connected by prominent interdoublet links near the inner row arm position (Figs 6A, 7A), although regions of incompletely extracted outer arms can occasionally be seen. The links are easily distinguished from dynein arms or radial spokes on the basis of their structure and organization (Warner, 1983a). However, some of the material originally described as a subunit of the interdoublet links (e.g. Fig. 3D of Warner, 1983a) may be, in retrospect, incompletely extracted 13 S dynein. As described earlier in this paper, twice-extracted axonemes retain by mass ^ 17 % of total axonemal ATPase, which, if divided equally between the inner row and outer row arms, represents an extraction efficiency of 92% for each arm position. For the recombined preparations wheredb in vivo stoichiometry, this means that < 1 out of 10 arms visible^in cross-section at each position (as in Fig. 6) results from incompletely extracted dynein. In order to confirm this value, we have counted the number of residual arms visible in cross-sectional images of twice-extracted doublets. From a sample size of 48 groups of doublets representing three preparations and comprising 270 individual doublets, < 7 % of the doublets retained arms at the outer row position and < 22 % retained arms at the inner row position. Because good structural criteria exist for outer row arms, the value for this position is in excellent agreement with extraction stoichiometry. However, because no comparable structural criteria exist Fig. 6. Cross-sections of twice-extracted axonemes (A) and extracted axonemes mixed with either 21 S dynein (B) or 13 S dynein (c) in proportions of 20/ig dynein to 10 \xg doublets. Before fixation, 20/iM-ATP, 10/iM-vanadate was added to the 21 S-doublet suspension. 21 S dynein rebinds predominantly to outer row arm positions on the doublet microtubules (B, arrowheads). In contrast, 13 S dynein rebinds predominantly to inner row arm positions (c, arrowheads), except at high stoichiometry (> 3 : 1) where it rebinds to several positions on the doublet lattice (d, arrowheads). Radial spokes are indicated by rr. x

13 Inner arm and outer arm dynein ATPases 275 Fig. 6

14 F. D. Warner, J. G. Perreault andjf. H. Mcllvain Fig. 7. Longitudinal sections of twice-extracted axonemes (A) and extracted axonemes mixed with either 21S dynein (B) or 13 S dynein (c) in proportions of 20 ^ig dynein to 10/ig doublets. Although often missing from thin sections, radial spoke heads are frequently superimposed over the doublets as single rows of triplet periodicities (96 nm repeat). Interdoublet links also repeat at 96nm but connect adjacent A and B subfibres. 13 S dynein apparently rebinds coincident with the radial spokes, often resulting in clusters (brackets) or rows of particles superimposed over the doublets (c). X for inner row arms, the value obtained for this position may be unreliable, even though it also is in reasonable agreement with extraction stoichiometry. 21 S dynein mixed with the doublet arrays in the presence of ATP rebinds principally to the outer row arm position on the A subfibre (Fig. 6B), although at higher dynein concentrations, supernumerary arms will bind to a variety of sites on the A or B subfibre lattice, including the region of the inner row arms (Warner & Mcllvain, 1982). 21 S binding near the outer row site probably represents binding specific for the positions in situ from which the dynein was extracted. In transverse axoneme sections, 13 S dynein mixed with the doublet arrays rebinds primarily near the centripetal side of the interdoublet link on the A subfibre (Fig. 6c). At higher dynein concentrations (> 3 :1), supernumerary material also rebinds to a variety of sites on the doublet lattice (Fig. 6D) but not nearly to the extent that occurs with 21 S dynein. In contrast to extracted doublets, > 75 % of the outer row positions and > 70% of

15 Inner arm and outer arm dynein ATPases 111 the inner row positions are occupied after recombination (2:1) with the 21 S and 13 S dynein fractions, respectively. Inner row dynein does not appear to form stable crossbridges between adjacent A and B subfibres, although interpretation is difficult because of the presence of the interdoublet links connecting the subfibres. Consistent Fig. 8. Rapid-freeze, deep-etch replica of two groups of twice-extracted doublets. When viewed from outside the axoneme profile (left three doublets), the surface of the microtubules is clean except for a number of randomly distributed particles. The interdoublet links (96 nm) are only faintly visible at the lower limit of platinum deposition. When viewed from inside the axoneme profile (remaining doublets), the triplet radial spoke heads generally are superimposed over the doublets as bright bands or particles repeating at 96 nm. X

16 278 F. D. Warner, J. G. Perreault andj. H. Mcllvain with the common appearance of arms in cross-sections of native cilia or flagella, the structure of inner row 13 S dynein is poorly defined compared with the prominent hook-shape of outer row 21 S dynein. In longitudinal sections, 21 S dynein repeats in the characteristic 24 nm array associated with outer arms (Fig. 7B). However, 13 S dynein rebound to the extracted doublets does not form discrete arm-like projections along the microtubules (Fig. 7c). Instead, it appears to have reassociated near the triplet grouping of the radial spokes, although again the images are confusing because of the proximity of both the spokes and interdoublet links. In extracted axonemes, spoke heads have the appearance of Fig. 9 Interdoublet links of twice-extracted doublets viewed in the longitudinal plane and from outside the axoneme profile by rapid-freeze, deep-etch electron microscopy. The links repeat at 96 nm, connecting adjacent A and B subfibres, and exhibit a recognizable substructure in the deep-etch replica. X

17 Inner arm and outer arm dynein ATPases 279 Fig. 10. Twice-extracted doublets recombined with 13 S dynein at low stoichiometry (13 SDb = 0-2/*g/10/*g doublets). When the doublets are viewed from outside the axoneme profile (A), their surfaces are clean except for randomly distributed particles visible also in Fig. 8. The interdoublet links (96 nm) are faintly visible at the lower limit of platinum deposition. When viewed from inside the axoneme profile (B), the bright spoke heads and 13 S particles are frequently superimposed over the doublets or sometimes over the interdoublet space. The 13 S particles (arrowheads) can sometimes be distinguished from the spoke heads (brackets; 96nm) by their thin connections with the adjacent B subfibre. (cf. Fig. 11B). x

18 180 F. D. Warner, J. G. Perreault andj. H. Mcllvain opaque knobs superimposed over an A or B subfibre, or sometimes over the interdoublet space (Fig. 7A). In comparison, 13 S dynein has the appearance of multiple rows of clustered particles that appear to coincide with the 96 nm repeat of the radial spokes (Fig. 7c). Consequently, numerous rows of particles can be superimposed over the doublet arrays; the number of rows sometimes exceeded the number of doublets by twofold. Because this characterization is imprecise with respect to the organization and absolute periodicity of rebound 13 S dynein, in an effort to resolve the structure of 13 S dynein we have examined extracted doublets and doublets recombined with each of the dyneins by the rapid-freeze, deep-etch technique of Goodenough & Heuser (1982, 1985). Using this technique, platinum replicas of extracted doublets exhibit the same general organization visible in thin sections (Fig. 8). However, the radial spoke heads have a unique bright appearance, owing to excessive accumulation of platinum that easily distinguishes them from subsequently rebound dynein (Figs 8, Fig. 11. High-magnification view of twice-extracted doublets decorated with 21 S dynein (Db = 1-3 /ig/10 fig doublets) and viewed from outside the axoneme profile (A) and twiceextracted doublets decorated with 13 S dynein (Db = 0-2/*g/10 \>% doublets) and viewed from inside the axoneme profile (B). 13 S particles have rebound to the A subfibre with a nm triplet spacing (brackets). The triplet groups of particles repeat at 96nm. Both outer and inner row dyneins are connected by thin strands to the adjacent B subfibre (lower microtubule in both figures) even though each preparation was processed in ATP and vanadate. A, X164000; B, X

19 Inner arm and outer arm dynein ATPases B). Interdoublet links also have a recognizable substructure and repeat at 96 nm, connecting adjacent A and B subfibres (Fig. 9). Extracted doublets were decorated with 13 S dynein at two different stoichiometries (Z)6 = 0-2/ig and 0-8/ig/lO/ig doublets), the latter value again being equivalent to the concentration of 13 S dynein extracted from the doublets (Table 1). After reaching binding equilibrium, each dynein preparation was maintained in 10/iM-ATP, 10/iM-vanadate to stabilize structural-side interactions, and then processed for the deep-etch procedure. After rebinding 13 S dynein, the doublets also have an appearance similar to their image in thin sections, particularly when the 13 S particles and radial spoke heads are viewed collectively above the doublets (Fig. 10B; compare with Fig. 7c). However, numerous regions in these images also occur where the 13 S particles are seen in the interdoublet space (Figs 10B, 11B). At low stoichiometry (0-ZfJgDb), the spacing of the rebound dynein appears to coincide with the spacing of the radial spokes (Fig. 11B; nm periods, 96nm repeat; 32nm average period). Each 'arm' comprises at least two subunits positioned near the base of each spoke and is connected by one or two thin strands to the adjacent B subnbre. This connection also distinguishes the 13 S particles from the radial spoke heads when the images of both structures are seen in the interdoublet space. When groups of decorated doublets are viewed from the outside the axoneme profile, there is no evidence of significant association of the 13 S particles with the outer surfaces of the A or B subfibres (Fig. 10A; compare with Fig. 8). Although it is impractical to measure bound dynein in longitudinal profiles Fig. 12. Twice-extracted doublets decorated with 13 S dynein at high stoichiometry (Db = 0-8^g/10/ig doublets) and viewed from inside the axoneme profile. 13 S particles are clustered in groups that repeat at 96 nm. The B subfibre is the lower microtubule in both figures. X

20 282 F. D. Warner, J. G. Perreault andjf. H. Mcllvain of decorated doublets, in contrast to low-binding stoichiometry; at high stoichiometry (0-8/igZ)6) the 13 S particles are more densely packed along the A subfibre, often filling in the space between successive spoke groups (Fig. 12). Although regions of particle grouping and 96 nm repeat are still apparent, the resulting periodicity is difficult to assess, owing to the number of visible subunits that cannot be specifically assigned to a given arm. If we were to describe periodicity on the basis of subunits visible in the interdoublet space rather than their association as a group of two to three subunits with each radial spoke, the resulting periodicity in this region could sometimes appear to be as low as 16 nm. Apart from spacing, their is no evidence for any order or structural polarity of inner row dynein that would be comparable to the tilted configurations associated with the outer row arms. DISCUSSION Isolated cilia or flagella from each of the three organisms Tetrahymena, Chlamydomonas and sea urchin contain a second salt-extractable ATPase that can be distinguished from outer arm 21 S dynein on the basis of sedimentation coefficient (== 13 S), electrophoretic profile, and substrate specificity (Ca/Mg ratio) (Gibbons, 1966; Ogawa& Gibbons, 1976; Piperno& Luck, 1979). The 13 S ATPase (dynein 2) has been called an isoenzyme of 21 S dynein (dynein 1) (Ogawa & Gibbons, 1976), although there is no genetic evidence or compelling biochemical evidence in support of this interpretation. Unlike 21 S dynein, the location and functional properties of 13 S dynein are unknown, owing to previous inability to rebind the enzyme to extracted microtubules. We have now shown that sucrose gradient-fractionated 13 S dynein ATPase isolated from Tetrahymena cilia will rebind to extracted doublet microtubules as A subfibre inner row dynein arms. The specificity of rebinding was confirmed by SDS/polyacrylamide gel electrophoresis, ATPase activity and electron microscopy. Ionic conditions supporting rebinding of 13 S dynein are similar to those that support rebinding of 21 S dynein as outer row arms (Mitchell & Warner, 1981). However, major differences were observed between the recombination behaviour of the two ATPases relative to ATP sensitivity, binding site specificity and structural organization. Unlike 21 S outer arm dynein, rebinding of 13 S dynein is saturated at a relatively low concentration and is largely specific for only a single row of binding sites that appear to be coincident with the periodicity of the radial spokes. Furthermore, rebound 13 S dynein is not sensitive to dissociation by ATP and vanadate under conditions that readily dissociate 21 S dynein from the B subfibre. Each of these differences may help to explain previous inability to detect rebinding of the isolated 13 S ATPase to microtubules or inability to characterize the structural organization of inner row dynein arms. The effects of extracted doublet microtubules on ATPase activity of the 13 S and 21 S dyneins indicate that rebinding of each enzyme to the microtubules is a highly specific reaction. By using preparations in which we controlled the concentration of bound dynein (Fig. 4) as well as preparations in which soluble dynein was mixed

21 Inner arm and outer arm dynein ATPases 283 with extracted doublets (Table 2), we have shown that ATPase activity of both the 13 S and 21S dyneins is enhanced by relatively low concentrations of microtubules. Although we have not identified the activating ligand (tubulin, an accessory protein or an ion) or its specific effect on the ATPases, activation of the dyneins has presumably occurred, because we have restored their potential for active-side interaction with adjacent B subfibres in the doublet arrays and thereby elevated the local concentration of tubulin seen by the active side of the dynein particle. We cannot eliminate the possibility that structural-side interaction with the A subfibre may also have an effect on enzyme activity, as is suggested by our finding that active-side rebinding of 13 S dynein cannot be detected independent of structural-side rebinding. Stoichiometry of rebound 13 S and 21S dynein ATPases The stoichiometry of extracted 21 S dynein is in excellent agreement with the stoichiometry predicted for the native cilium (Table 1) and taken together with its immunological specificity (Ogawa et al. 1982), clearly establishes that in vivo 21S dynein occupies only a single row of 24 run binding sites per doublet. In contrast, 2-5- fold more 13 S protein is extracted from the doublets than is predicted for saturation of a single row of 32 nm binding sites and zm r of 6 X 10 s. However, the 13 S dynein fraction from Tetrahymena cilia has recently been shown to comprise two (1:1) morphologically distinct particles having a combined M r of about 1 X 10 6 (Marchese- Ragona et al. 1984). Neither particle contains polypeptides that comigrate on SDS/polyacrylamide gels with 21 S dynein polypeptides (Porter & Johnson, 1983). Similar properties have been found for Chlamydomonas inner arm dynein (Piperno & Luck, 1981). These data suggest that the inner arm dynein fraction contains nonidentical but co-sedimenting subunits that reassociate to form the inner arms. Consistent with this view, both rebound and native inner arms (see below) appear to comprise multiple (2-3) subunits (Fig. 12) and binding stoichiometry {bma) for 13 S dynein is in near unity with the stoichiometry of 13 S dynein extracted from the microtubules (Table 1). The observation that Tetrahymena 21 S dynein rebinds at a stoichiometry that far exceeds its concentration in vivo (Table 1) has led to questions regarding the specificity of the reaction (Gibbons & Fronk, 1970). However, the resulting saturation curve for 21 S structural-side binding (Fig. 5) indicates that binding is highly specific for a single class of sites. If 21 S dynein were rebinding to high-affinity structural sites at low stoichiometry and to lower-affinity, non-specific sites at high stoichiometry, then the binding curve should not have fitted a hyperbola. Therefore, it appears that expression of potential binding sites on the microtubules may be controlled by simple accessibility of the site. Every tubulin dimer may conserve only a single class of structural binding sites for 21 S dynein, but expression of the sites may be regulated either directly or indirectly by accessory proteins on the microtubule lattice. This seems to contradict our interpretation that low stoichiometric 21 S rebinding is specific for in situ structural sites (Fig. 6B). However, this interpretation may have resulted as much from expectation as from actuality, particularly when outer row structural sites represent nearly 40 % of the total available ATP-insensitive binding sites (i *)

22 284 F. D. Warner, J. G. Perreault andjf. H, Mcllvain and the remaining non-structural sites appear to be distributed randomly with respect to all classes of microtubules (Mitchell & Warner, 1980). Although our analysis is preliminary, as a probable consequence of these properties there is little or no measureable competition between the 21 S and 13 S dyneins when both are added simultaneously, but at low 21 S concentrations, to extracted doublets. At high concentrations of 21 S dynein, apparent competition does occur, but this probably results from simple steric intrusion into the limited space on the microtubule lattice. The unique periodicity of 13 S binding sites suggests that these sites are chemically different from 21 S binding sites, even though 21 S dynein will rebind near the inner arm position. It is not surprising that 21 S binding appears to be largely unregulated. First, it is unnecessary: structural-side specificity is established developmentally and thereby limits subsequent active-side interactions to the B subfibre. Second, motility-coupled activity is probably more efficient if active-side interactions can occur with respect to any tubulin dimer in the B subfibre. In comparison to 21 S dynein, the low tmax for 13 S rebinding suggests that all positions other than the inner row position are considerably more restricted with respect to potential 13 S interactions. The apparent inability of 13 S dynein to undergo active-side binding independent of structural-side binding also suggests difference in the enzymological behaviour of the two ATPases. Structural specificity of rebound US and 21S dynein ATPases Although inner row dynein arms have been widely represented as having a 24 nm repeat and structure comparable to outer row arms, there is little morphological evidence to support this interpretation. Rebound 13 S dynein bears only a slight resemblance to projections or arms, particularly when compared with outer row 21 S dynein. 13 S dynein is positioned between the radial spokes and interdoublet links and the major part of its mass does not lie near the B subfibre. Goodenough & Heuser (1985) have recently described native inner arm structure from several unrelated species, including flagella from a spoke-free mutant of Chlamydomonas as well as Tetrahymena cilia, and found the same subunit structure and organizational relationship of inner row arms to the radial spokes as we describe for our recombined preparations. The apparent coupling of inner arm periodicity to radial spoke periodicity, while surprising, is consistent with long-standing inability to describe the structure and organization of inner row dynein arms by any means other than axonemal crosssections. Moreover, it is uncertain how much the structure traditionally identified in cross-sections as an inner row dynein arm may have been contributed to by the interdoublet links (compare Fig. 6A and c). We might also consider the extent to which the interdoublet links and inner row dynein arms may form part of a single functional complex, from which only the 13 S subunits are solubilized by high salt solutions. The thin connections observed between the 13 S particles and adjacent B subfibre are similar to those observed between outer row (21 S) arms and the B subfibre (Goodenough & Heuser, 1982). For 21 S dynein, the connections are apparently related to the stalk or projection that emanates from each of the three major subunits of the dynein particle (Goodenough & Heuser, 1984). However, the fact that

23 Inner arm and outer arm dynein ATPases 285 the connections occur under substrate conditions (ATP, vanadate) that dissociate 21 S dynein from the B subfibre and under conditions where 13 S dynein does not associate by itself with the B subfibre, suggests that their apparent attachment to the B subfibre may be related to the rapid-freeze technique. The relative contributions of the inner row and outer row dynein ATPases to the sliding bending mechanism are unknown, although the coincident distribution of inner row arms, radial spokes and interdoublet links suggests that these three components may interact at some fundamental level. Our present view of the function of inner row dynein comes from experiments showing that sea-urchin axonemes from which 21 S outer arms had been extracted, beat normally but with about half their usual frequency (Gibbons & Gibbons, 1973,1979). This result has been widely interpreted to mean that all dynein arms contribute equal amounts of force to the slidingfilament mechanism and, consequently, inner and outer row dynein arms must contribute equally to the production of that force. Having potentially greater significance, however, is the conclusion that normalflagellar beating can be produced by inner row dynein arms alone (Gibbons & Gibbons, 1979; Mitchell & Rosenbaum, 1985). The ability to rebind separately the 13 S and 21 S dynein ATPases to the inner and outer row arm positions of extracted axonemes should provide an important methodological approach for investigating the relative contributions of each ATPase to the activation and control of ciliary motion (Warner & Zanetti, 1980; Zanetti & Warner, 1982) as well as provide a means to study independently the enzymological properties of the activated ATPases. This study was supported by research grant GM20690 from the National Institutes of Health. We thank Dr John E. Heuser for his preparation of the deep-etch replicas of our dy nein-decorated doublet preparations and Dr Ursula G. Goodenough for her help with interpretation of the resulting images. REFERENCES BELL, C. W. & GIBBONS, I. R. (1982). Structure of the dynein-1 outer arm in sea urchin sperm flagella. II. Analysis by proteolytic cleavage. J. biol. Chem. 257, BRADFORD, M. M. (1976). A rapid and sensitive method for the quantitation of microgram quantities of protein utili2ing the principle of protein dye-binding. Analyt. Biochem. 72, FAY, R. B. & WITMAN, G. B. (1977). The localization of flagellar ATPases in Chlamydomonas reinhardii.j. Cell Biol. 75, 286 (abstr.). GIBBONS, B. H. & GIBBONS, I. R. (1973). The effect of partial extraction of dynein arms on the movement of reactivated sea urchin sperm. J. Cell Set. 13, GIBBONS, B. H. & GIBBONS, I. R. (1979). Relationship between the latent adenosine triphosphatase state of dynein 1 and its ability to recombine functionally with KCl-extracted sea urchin sperm flagella. J. biol. Chem. 254, GIBBONS, I. R. (1965). Chemical dissection of cilia. Arch Biol. (Liege) 76, GIBBONS, I. R. (1966). Studies on the adenosine triphosphatase activity of 14 S and 30 S dynein from cilia of Tetrahymena. J. biol. Chem. 241, GIBBONS, I. R., COSON, M. P., EVANS, J. A., GIBBONS, B. H., HOUCK, B., MARTINSON, K. H., SALE, W. S. & TANG, W.-J. Y. (1978). Potent inhibition of dynein adenosine triphosphatase and of the motility of cilia and sperm flagella by vanadate. Proc. natn. Acad. Sci. U.SA. 75, GIBBONS, I. R. & FRONK, E. (1979). A latent adenosine triphosphatase form of dynein 1 from sea urchin. J. biol. Chem. 254, 187-1%.

24 286 F. D. Warner, J. G. Perreault andj. H. Mcllvain GIBBONS, I. R. & ROWE, A. J. (1965). Dynein: a protein with adenosine triphosphatase activity from cilia. Science 149, GOODENOUGH, U. G. & HEUSER, J. E. (1982). Substructure of the outer dynein arm. J. CellBiol. 95, GOODENOUGH, U. G. & HEUSER, J. E. (1984). Structural comparison of purified dynein proteins with in situ dynein arms. J. molec. Biol. 180, GOODENOUGH, U. W. & HEUSER, J. E. (1985). Structure of inner dynein arms, radial spokes, and the central pair/projection complex of cilia and flagella. J. CellBiol. 100, HEUSER, J. (1980). Three-dimentional visualization of coated vesicle formation on fibroblasts. J. CellBiol. 84, JOHNSON, K. A. & WALL, J. S. (1983). Structure and molecular weight of the dynein ATPase. J. Cell Biol. 96, MABUCHI, I. & SHIMIZU, T. (1974). Electrophoretic studies on dynein from Tetrahymena cilia. J. Biochem. 76, MARCHESE-RAGONA, S. P., BELLES ISLES, M., GAGNON, C, WALL, J. S. & JOHNSON, K. A. (1984). Structures and masses of 14 S dynein from Tetrahymena and 19 S dynein from bull sperm. J. CellBiol. 99, 45a (abstr.). MITCHELL, D. R. & ROSENBAUM, J. L. (1985). A motile Chlamydomonas mutant lacking outer dynein arms.j. CellBiol. 100, MITCHELL, D. R. & WARNER, F. D. (1980). Interactions of dynein arms with B subfibers of Tetrahymena cilia: quantitation of the effects of magnesium and adenosine triphosphate. J. Cell Biol. 87, MITCHELL, D. R. & WARNER, F. D. (1981). Binding of dynein 21 S ATPase to microtubules. Effects of ionic conditions and substrate analogs. J. biol. Chem. 2S6, NAKAMURA, S. & KAMIYA, R. (1978). Bending motion in split flagella of Chlamydomonas. Cell Struct. Fund. 3, OGAWA, K. & GIBBONS, I. R. (1976). Dynein 2. A new adenosine triphosphatase from sea urchin sperm flagella. J. biol. Chem. 251, OGAWA, K., NEGISHI, S. & OBIDA, M. (1982). Immunological dissimilarity in protein component (Dynein 1) between outer and inner arms with sea urchin sperm axonemes. J. Cell Biol. 92, PFISTER, K. K., FAY, R. B. & WITMAN, G. B. (1982). Purification and polypeptide composition of dynein ATPases from Chlamydomonas flagella. CellMotil. 2, PIPERNO, G. & LUCK, D. J. L. (1979). Axonemal adenosine triphosphatases from flagella of Chlamydomonas reinhardtii.j. biol. Chem. 254, PIPERNO, G. & LUCK, D. J. L. (1981). Inner arm dyneins from flagella of Chlamydomonas reinhardtii.cell 27, PORTER, M. E. & JOHNSON, K. A. (1983). Characterization of the ATP-sensitive binding of Tetrahymena 30 S dynein to bovine brain microtubules. J'. biol. Chem. 258, READ, S. M. & NORTHCOTE, D. H. (1981). Minimization of variation in the response to different proteins of the Coomassie Blue G dye-binding assay for protein. Analyt. Biochem. 116, SALE, W. S. & SATIR, P. (1977). Direction of active sliding of microtubules in Tetrahymena cilia. Proc. natn. Acad. Sci. U.SA. 74, SHIMIZU, T. (1975). Recombination of ciliary dynein of Tetrahymena with the outer fibers. J. Biochem. 78, TAUSSKY, H. H. & SHORR, E. (1953). A microcolorimetric method for the determination of inorganic phosphate. J. biol. Chem. 202, WARNER, F. D. (1983a). Organization of interdoublet links in Tetrahymena cilia. CellMotil. 3, WARNER, F. D. (19836). Differences between ATP-sensitive and ATP-insensitive 21 S dynein binding to ciliary microtubules. J. submicrosc. Cytol. 15, WARNER, F. D. & MCILVAIN, J. H. (1982). Binding stoichiometry of 21 S dynein to A and B subfiber microtubules. CellMotil. 2, WARNER, F. D., MITCHELL, D. R. & PERKINS, C. R. (1977). Structural conformation of the ciliary ATPase dynein. J. molec. Biol. 114, WARNER, F. D. & ZANETTI, N. C. (1980). Properties of microtubule sliding disintegration in isolated Tetrahymena cilia. J. CellBiol. 86,

25 Inner arm and outer arm dynein ATPases 287 YANO, Y. & MIKI-NOUMURA, T. (1981). Recovery of sliding ability in arm-depleted flagellar axonemes after recombination with extracted dynein Cell Set. 48, ZANETTI, N. C. & WARNER, F. D. (1982). Evidence for a role of 13 S axonemal ATPase in modulation of ciliary microtubule sliding. Cell Motil. 2, {Received 12 February Accepted 11 April 1985)