Structure of the Dynein-1 Outer Arm in Sea Urchin Sperm Flagella

The 21 S latent activity dynein-1 (LAD-1) extracted from flagellar axonemes by a 0.6 M NaCl/Mg’ solution at pH 7.0 is dissociated into smaller particles upon dialysis against 5 m~ imidazole/HCl, pH 7.0, 0.5 mM EDTA, 7 m~ 2-mercaptoethanol. Zonal centrifugation separates this dissociated dynein-1 into two major fractions: one, containing the AD heavy chain and the inter- mediate chains 1,2,3, sedimenting at 9 to 10 s, and the other, containing predominantly aggregates of the A, heavy chain sedimenting over the range 12 to 30 S. Zonal centrifugation of the Ap/intermediate chain (Afl/ IC) fraction after dialysis back into 0.6 M NaCl/Mg+ solution, shows that the AD heavy chain and interme- diate chain 1 now co-sediment as a distinct complex, forming a peak at about 14 S, whereas intermediate chains 2 and 3 constitute a separate complex sedimenting in two peaks of 10 S and 15 to 17 s. The A, chain fraction, after dialysis back into 0.6 M NaCl/Mg’ solution, sediments at about 12 S. Although neither the AD/ IC fraction nor the A, chain fraction appears able to reform a 21 S particle alone, dialysis of the pooled A, and Ao/IC fractions back into 0.6 M NaCl/M&’ solution causes a partial re-formation of a 21 S particle which contains the A, and Ap chains and intermediate chains 1, 2, and 3 in approximately the same proportions as found in the original LAD-1

The 2 1 S latent activity dynein-1 (LAD-1) extracted from flagellar axonemes by a 0.6 M NaCl/Mg' solution at pH 7.0 is dissociated into smaller particles upon dialysis against 5 m~ imidazole/HCl, pH 7.0, 0.5 m M EDTA, 7 m~ 2-mercaptoethanol. Zonal centrifugation separates this dissociated dynein-1 into two major fractions: one, containing the AD heavy chain and the intermediate chains 1 , 2 , 3 , sedimenting at 9 to 10 s, and the other, containing predominantly aggregates of the A, heavy chain sedimenting over the range 12 to 30 S. Zonal centrifugation of the Ap/intermediate chain (Afl/ IC) fraction after dialysis back into 0.6 M NaCl/Mg+ solution, shows that the AD heavy chain and intermediate chain 1 now co-sediment as a distinct complex, forming a peak at about 14 S, whereas intermediate chains 2 and 3 constitute a separate complex sedimenting in two peaks of 10 S and 15 to 17 s. The A, chain fraction, after dialysis back into 0.6 M NaCl/Mg' solution, sediments at about 12 S. Although neither the AD/ IC fraction nor the A, chain fraction appears able to reform a 2 1 S particle alone, dialysis of the pooled A, and Ao/IC fractions back into 0.6 M NaCl/M&' solution causes a partial re-formation of a 21 S particle which contains the A, and Ap chains and intermediate chains 1, 2, and 3 in approximately the same proportions as found in the original LAD-1 particle. Neither the A, fraction, the AD/IC fraction, nor a mixture of the two appears able to restore the beat frequency of outer arm-depleted sperm flagella. However, a 1:l mixture of the A, and AD/IC fractions, and (less effectively) the AD/IC fraction alone, appear to block the manifestation of frequency restoration of subsequently added LAD-1 in a way that is reversed gradually over 2 to 8 min in the presence of ATP. The different enzymatic properties of the ATPases in the A, and AD/IC fractions suggest that the outer arms of sea urchin sperm flagella each contain two distinct dynein ATPases.
The normal oscillatory beating of cilia and flagella is widely thought to be the result of coordinated sliding movements between the doublet microtubules of the axoneme that lead to the formation of bends at the proximal end of the flagellum and their propagation to the tip (1-3). A variety of evidence indicates that this relative sliding of the doublet tubules is largely the result of a shear stress produced by the dynein arms through a mechanochemical cycle involving changes in angular orientation of the arms and their cyclic detachment * This work was supported in part by grant HD loo02 from the National Institute of Child Health and Human Development and by a Helen Hay Whitney Fellowship to W.S.S. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and reattachment to the B-tubule concomitant with the binding and hydrolysis of molecules of ATP (4)(5)(6)(7).
Although the overall action of the dynein arms in the generation of shear stress between doublet tubules seems clear, the details of the mechanochemical cycle are largely speculative, partly because so little is known about the functional substructure within the dynein arms. For some time, it has been evident from electron microscopy that the outer and inner arms have distinct structures, with the former being longer and more complex (8,9). More recent studies have attributed to the outer arm a variety of structures, including one resembling a bent hammer (9,10) and one in the shape of a Y (11,12), but all seem to agree that both the inner and outer arms are to some extent hooked in appearance when viewed in axonemal cross-sections. Warner et al. (13) have shown by the negative staining technique that the dynein arms of Tetrahymena cilia appear, in side view, to be composed of three or four uniform, globular subunits, although the relationship of these subunits to the hooked profiie in cross-sections is at present uncertain.
Recent evidence from high resolution Na dodecyl SO4-polyacrylamide gel electrophoresis has shown that the complex electron microscopic structure of the dynein arms is paralleled by a correspondingly complex polypeptide composition (14).
For instance, an 18 S form of dynein which is solubilized from Chlamydomonas flagellar axonemes by high salt concentrations has been found by Piperno and Luck (15) to contain up to 13 different polypeptides with molecular weights between 330,000 and less than 15,000. Treatment of sea urchin sperm flagellar axonemes with an 0.6 M NaCl extraction medium solubilizes a 21 S form of dynein 1, LAD-1, that is thought to represent most or all of the outer arm on the doublet tubules of the axoneme (16). Purification of LAD-1' by sucrose density gradient sedimentation and subsequent electrophoresis in the presence of Na dodecyl so4 has revealed that the 21 S particle contains a t least nine different polypeptides: two heavy chains, designated A, and AD, with apparent molecular weights of 330,000 and 320,000, three intermediate chains with apparent molecular weights of 122,000, 90,000, and 76,000, and at least four light chains with apparent molecular weights between 24,000 and 14,000 (17). So far, however, there is relatively little data on the stoichiometry of these subunits (16,17) and virtually none on their substructural arrangement within the outer arm.
A more complete understanding of the role of the dynein arms in flagellar movement will require a detailed analysis of their structural and functional organization derived from a comprehensive characterization of their component parts. In the present study, we report the results of a partial separation and characterization of the polypeptide subunits constituting the 21 S LAD-1 particles that represent the solubilized outer arms of sea urchin sperm flagella.

MATERIALS AND METHODS
Preparation of LAD-I and its Subunits-Sperm were obtained from the sea urchin Tripneustes gratilla by the injection of 0.5 M KC1 into the body cavity. Flagellar axonemes were isolated from the sperm as previously reported (16) or by a modified sucrose procedure that avoids the use of Triton X-100 (18). Crude LAD-1 was extracted by suspending the axonemes, usually at 3 mg of axonemal protein/ ml, in 0.6 M NaC1/Mg2' solution (0.6 M NaCl, 4 mM MgSO4, 0.1 m~ EDTA, 1 m~ dithiothreitol, 7 mM 2-mercaptoethanol, 5 mM imidazole/HCl, pH 7.0) at 4 "C for 10 min as previously reported (16). Whenever necessary, the enzyme was concentrated by ultrafiltration under NZ with an Amicon UM20 membrane.
The 21 S LAD-1 was dissociated into subunits by dialysis against a low ionic strength solution containing 5 m~ imidazole/HCl, pH 7.0, 0.5 m~ EDTA, and 14 mM 2-mercaptoethanol (IEM buffer) for 24 h at 4 "C with at least two changes of 100 volumes each of the IEM buffer. Table I shows typical recoveries of protein and ATPase activity at various stages.
Zonal centrifugation was carried out on 11-m15 to 20% w/v sucrose density gradients prepared in an appropriate buffer. After sedimentation in an SW 41 rotor (Beckman Instruments, Spinco Division) at 35,000 rpm for 15 h at 4 "C, each gradient was separated into about 20 fractions of equal volume by carefully lowering a glass capillary tube nearly to the bottom of the tube, and removing the contents, densest fraction first, with a peristaltic pump. The fractions are numbered 1 to 20 according to their position from top to bottom of the gradient. With this method of fractionation, any material that has been pelleted to the bottom of the tube is caught up in the last fraction as the tube is sucked dry, causing this pelleted material to appear mostly in fraction 1. Approximate sedimentation rates on sucrose density gradients were determined by reference to marker proteins consisting of catalase (11.3 S ) and LAD-1 (21.4 S ) (16).
Determination of ATPase Activity and Protein Concentration-Two procedures were used to determine ATPase activity. Routine assays of dynein extracts and of gradient fractions were performed by adding the enzyme to 2 ml of an assay buffer containing 0.1 M NaCl, 2 mM MgS04, 0.1 mM EDTA, 30 mM Tris-HC1 buffer, pH 8.1, and 1 mM ATP, and incubating for 10 to 20 min at room temperature (23-24 "C). The reaction was terminated by the addition of 20 pl of 10% w/v Na dodecyl SO,, and inorganic phosphate was determined by the method of Fiske and SubbaRow (19). For steady state kinetic studies, a more sensitive coupled assay containing an ATP-regenerating system was used (20). In this case, 4 to 10 pg of enzyme were added to 2.0 ml of assay solution containing 0.1 M NaC1,30 mM KC1, 4 mM MgSO,, 0.5 m~ EDTA, 10 mM Tris-HC1 buffer, pH 8.1, 0.26 mM NADH, 1.5 mM phosphoenolpyruvate, 80 pg (42 units) of pyruvate kinase, 40 pg (36 units) of lactate dehydrogenase, and the desired concentration of ATP, and the rate of the reaction, at 24.0 -+ 0.1 "C, was determined from the decrease in absorbance at 340 nm with time. Control experiments indicated that at both high (1 m~) , and low (1 p~) , levels of ATP, the reaction rate was proportional to the concentration of dynein, and that 2-fold increases in the concentrations of phosphoenolpyruvate, pyruvate kinase, and lactate dehydrogenase had no TABLE I Recovery ofprotein and ATPase activity a t various stages during a typicalpreparation of LAD-I subunits Stages are: 1) after extraction of LAD-1 from flagellar axonemes by 0.6 M NaC1/Mg2', 2) after concentration of LAD-1 in ultrafiltrat,ion cell; 3) after dialysis for 24 h against IEM buffer. Latent (i.e. non-Triton-treated) ATPase activity was measured. Recovery of ATPase activity from sucrose gradients is described in the appropriate figure legends. effect on the measured reaction rate. The amounts of ATPase activity present in coupled assays were calculated using an absorbance for NADH at 340 nm of 6.2 X lo3 M" cm" (21).
Triton treatment of LAD-1 or its subunits was carried out by incubating the enzyme with 0.1% w/v Triton X-100 at room temperature for 10 min prior to determination of the ATPase activity. This treatment activates the latent ATPase of LAD-1, and with fresh preparations the specific ATPase activity increases about 10-fold. However, the activation ratio of LAD-1 decreases upon storage or upon the handling involved in further purification (16,18).
Protein concentration was determined by the method of Lowry et al. (22) with bovine serum albumin used as a standard.
Gel Electrophoresis-Polyacrylamide gel electrophoresis in the presence of Na dodecyl SO, was performed essentially according to the method of Laemmli (23). Polyacrylamide gradients from 3 to 6% w/v and 5 to 15% w/v were employed in the separating gels, which were cast 3 mm thick for optimal resolution of the high molecular weight polypeptides (17). Gels were stained overnight with 0.05% w/ v Coomassie Brilliant Blue R-250 in water/methanol/acetic acid (5:5: 1 by volume), and destained in 10% v/v methanol and 7.5% v/v acetic acid, using a rocking platform for agitation and a piece of polyurethane foam to adsorb the dye. Densitometry of protein bands was done by cutting strips from stained slab gels and scanning them at 560 nm essentially as reported previously (17). Integrated absorbances were determined by weighing peak areas traced onto paper. Care was taken to use only scans with peak heights below about 0.8 A, since staining with Coomassie blue is not linear above this level (24). The varying lane widths of different bands (heavily loaded bands tend to have greater lane widths than faint bands) were taken into account when calculating relative quantities of polypeptides in different bands by multiplying the peak weight of a band by its lane width. Bands with significantly distorted shapes were not used for densitometry. Baseline absorbances were determined at the top of the gel lanes being measured, where polypeptide bands were absent.
The determination of apparent molecular weights of polypeptide chains by their relative electrophoretic mobilities is described in the following paper (25).
Assay for Functional Activity-The capability of the separated A heavy chain fractions to recombine functionally with outer armdepleted, demembranated sperm flagella of Colobocentrotus atratus was measured as described earlier (7) with only slight modifications. In general, incubations were carried out at room temperature (23 "C) for 7 min with -10 pg of the protein fraction in question/ml of reactivating solution (0.15 M KCl, 2 m~ MgS04,O.l mM EGTA, 1 mM dithiothreitol, 2% (w/v) polyethylene glycol (M, = 20,000), and 10 mM Tris-HC1 buffer, pH 8.1). In some cases the sperm were incubated fist for 7 min with one or more of the separated A heavy chain fractions followed by a second 7-min incubation with an aliquot of intact LAD-1 before reactivation was initiated by addition of 1 mM ATP. Because of the possibility that some of the recombined proteins might be released rapidly upon the addition of ATP, measurements of beat frequency were begun as quickly as possible after mixing.
Chemicals and Reagents-Imidazole was obtained from Sigma and recrystallized twice from 80% v/v ethanol containing 1 m~ EDTA before use. Tris(hydroxymethy1)aminomethane was obtained from Sigma and, for the purpose of ATPase assays and sperm reactivation, was recrystallized first from 1 mM EDTA and then from 80% v/v methanol before use. For electrophoresis, Tris was used without recrystallization. Pyruvate kinase, lactate dehydrogenase, bovine serum albumin, Coomassie Brilliant Blue R-250, and "Na lauryl SO," (approximately 75% dodecyl , 9 0 4 ) were all purchased from Sigma Chemical Co. (The use of this relatively impure grade of Na dodecyl SO4 was found to be important for obtaining good resolution of the high molecular weight electrophoretic bands.) Highly purified Na dodecyl SO, from BDH Biochemicals was used to terminate ATPase reactions. ATP, phosphoenolpyruvate, and NADH were products of Boehringer Mannheim Biochemicals. Acrylamide, N,N'-methylenebisacrylamide and ammonium peroxydisulfate were from Eastman Kodak Co. All other chemicals and reagents were of analytical grade. Distilled, deionized water was used throughout. neme. In addition to the previously identified C, A,, and Ai< heavy chains (17), there are three polypeptides in whole axonemes that migrate more slowly than the C chain and these will be named sky chains 1,2, and 3. Upon extraction of axonemes with 0.6 M NaCl/Mg" solution, sky chains 2 and 3 are solubilized, along with the A, and the Alr chains and some of the C chain material, while sky chain 1 remains bound to the axoneme. When this 0.6 M NaCl/Mg" extract (crude LAD-1) is subsequently sedimented through a 5 to 20% sucrose density gradient made up in the same buffer, the major protein peak sediments at approximately 21 S (16). Electrophoretic analysis of the fractions from such a sucrose density gradient ( Fig. 1) shows that the 21 S particle contains two species of heavy chain, A, and A($; three species of intermediate chain, numbered 1, 2, and 3, and at least four distinct light chains (17). Both the latent ATPase activity and that activated by treatment with Triton X-100 (16) parallel the distribution of the 21 S LAD-1 polypeptides (Fig. 1).

Improvements in Na dodecyl
Behavior of LAD-1 a t Low Salt Concentration-Preliminary experiments showed that the sedimentation coefficient of LAD-1 increased when the salt concentration was lowered, to about 25 S in 0.1 M NaCl and to 30 to 40 S in 10 mM NaCl. Partial dissociation of the LAD-1 particle occurred a t this lower salt concentration, with peak of the AP chain sedimenting more slowly than that of the A, chain.
Dialysis of crude LAD-1 against the low salt IEM buffer causes a complete dissociation of the 21 S particle, and fractions containing the separated A, and AD heavy chains can then be obtained by sedimentation through a sucrose density gradient made up in the same solution. Under these conditions, the AP chain and the three intermediate chains appear to sediment approximately together as a fairly compact peak at a position corresponding to a sedimentation coefficient of 8 to 10 S (Fig. 2). However, examination of many preparations indicates that the distribution of intermediate chains 2 and 3 is reproducibly broader than that of the intermediate chain 1 which parallels closely that of the As chain. The A, chain is largely aggregated and sediments over a broad region of the gradient, corresponding to about 12 to 30 S. The peak of ATPase activity usually coincides with the peak of the A,< and the three intermediate chains, and there appears to be little ATPase activity associated with the fractions containing the A,, chains under these conditions. Combining samples from appropriate regions of such gradients yields two fractions: one (comprising fractions 12 to 19), designated the A, chain fraction contains the A, chain, as well as some sky chain and C chain; while the second (comprising fractions 7 and 8), designated the Air/IC fraction, contains the A/< chain and intermediate chains 1, 2, and 3, along with small amounts of polypeptides not derived from the 21 S LAD-1 partic!e.
Re-formation of the 21 S Particle-Experiments have been performed to determine whether the 21 S LAD-1 particles that have been dissociated by dialysis a t low ionic strength are able to re-form if the salt concentration is increased. When either the A, or the Air/IC fraction is dialyzed separately back into 0.6 M NaCl/Mg" solution, essentially no re-formation of the 21 S particle occurs. Density gradient centrifugation of the redialyzed A, chain fraction shows a more compact peak (Fig. 3a) than was seen in low salt solution (Fig. Z), with an approximate sedimentation coefficient of 10 to 12 S. ATPase activity is now present and parallels the distribution of the A, chain. The corresponding redialyzed Air/IC fraction sediments a t 10 to 14 S, with the distribution of intermediate chain 1 closely paralleling that of the A/# chain in fractions 8 to 11 (Fig. 36), while intermediate chains 2 and 3 are distributed more broadly across the gradient from fractions 5 to 13, forming two peaks on either side of the AB peak, the slower a t approximately 9 to 10 S and the faster at about 15 to 17 S.
The distribution of ATPase activity parallels approximately that of the AP chain and intermediate chain 1. When pooled fractions containing the separated A, and Ai{ chains are combined and dialyzed against 0.6 M NaCl/Mg" solution for 24 h, their distribution after density gradient sedimentation (Fig. 3c) shows partial re-formation of a 21 S  Fig. 1, for a sample of LAD-1 (1.2 mg; 0.75 pmol of Pi min") that had been dialyzed against IEM solution for 24 h and subsequently sedimented through a 5 to 20% sucrose density gradient prepared in IEM solution. Fifty pl of each fraction were loaded on this slab gel. The latent ATPase activities of the individual fractions are shown superimposed on gel. Recovery of latent ATPase was 0.73 pmol of P, min" (81%). Fractions 12 to 19 from this gradient were combined as the A,. fraction and fractions 7 and 8 were combined as the A,,/ IC fraction. and some A,, chain sediment in the 10 to 14 S region (fractions 8 to 10). The proportion of re-formed 21 S particles, determined by the relative electrophoretic band intensities of the A,, chain in the 10 to 14 S and the 21 S peaks, has varied between preparations, with more than 50% re-formation occurring in the best cases. Assays of ATPase activity reveal two peaks, one in the region of the 10 to 14 S peak and a second at about 21 S. The material in the re-formed 21 S peak usually shows a higher degree of Triton-activation (2-to 3-fold) than that in the 10 to 14 S peak (less than 2-fold). The distribution of the intermediate chains in such recombined preparations is shown more clearly in the more heavily loaded gel shown in Fig. 3d. Intermediate chain 1 shows two peaks, one of which co-sediments with the All chain in the 10 to 14 S peak, while the second co-sediments with the two A chains in the 21 S peak. The distribution of intermediate chains 2 and 3 also shows two peaks, one co-sedimenting with the re-formed 21 S peak, the other sedimenting at about 7 to 10 S. Sedimentation Properties of AI1/ZC Fraction-Analytical ultracentrifugation of the AI,/IC fraction in IEM buffer after dialysis to remove sucrose shows a symmetrical peak with a sedimentation coefficient(s;,),,J of 9.3 f 0.5 S (Table 11). Small amounts of contaminant material form a "skirt" at the leading and the trailing edges of the main peak. The A,/IC fraction that has been dialyzed back into the 0.6 M NaCl/Mg" solution prior to sedimentation usually shows a less symmetrical distribution, with a main peak with as;,),,,. of 14.2 f 0.3 S and a fairly substantial trailing shoulder.
Analytical ultracentrifugation of the pooled A,, fraction has so far been unsatisfactory because of persistent aggregation of the A,. chain.
Enzymatic Properties-In order to minimize the aggregation of the A', chain, the separated A,, or AIJIC fractions were usually dialyzed back into 0.6 M NaCl/Mg" solution prior to study of their enzymatic properties. Double reciprocal plots of ATPase activity against ATP concentration for intact LAD-1 have given a K,, of 1.3 f 0.6 p~ (Table 11). In some preparations, a second kinetic component with a higher apparent K , of 12 to 24 p~ was observed; this is probably attributable to contamination with a small percentage of activated dynein-1, for treatment of LAD-1 with Triton X-100 causes an increase in the K,, to 45 f 9 p~ along with an approximately 10-fold increase in V,,,.,.
The specific MgATPase activity of the A,, chain fraction depended on the presence or absence of MgSO, during the dialysis back into 0.6 M NaCl buffer. After dialysis into 0.6 M NaCl/Mg2+ solution, an average specific MgATPase activity of 0.24 f 0.12 pmol of Pi mg-l min" was obtained, whereas after dialysis in the absence of MgS04, the activity averaged only 0.06 * 0.01 pmol Pi mg" min-l (Table I). These data suggest that the ATPase activity of the A, chain fraction is labile in the absence of Mg'+. The specific MgATPase activity of the AdIC fraction, on the other hand, was independent of the presence or absence of MgS04 during dialysis, and an average value of 0.7 f 0.3 pmol Pi mg-I min" was found in both cases. The ATPase activities of both the A, fraction (dialyzed into 0.6 M NaCl/Mg" solution), and the AdIC fraction increased 1.5-to 2-fold upon treatment with Triton X-100 prior to the ATPase assay (Fig. 3, a and b ) .
The divalent cation dependence and the effect of NaCl concentration on the ATPase activities of the separated A., and A,/IC fractions are qualitatively similar to those on the ATPase activity of LAD-1 (Fig. 4). In most cases, the ATPase activity in the presence of either Mg2+ or Ca2+ is approximately the same, and the activity increases with increasing concentrations of NaCl up to about 1 M. This increase in the Mg"+-activated activity is less pronounced when either the A,/IC fraction or LAD-1 have been treated with 0.1% Triton X-100 prior to the assay. The activity of the A,, fraction is affected less by increasing concentrations of NaCl, and treatment of the A, fractions with Triton X-100 showed little effect under these conditions. The dependence of the Ca"-activated ATPase activity of LAD-1 and the A,, and AI1/IC fractions upon NaCl concentration is generally similar to that of the MgATPase, except that treatment of LAD-1 with Triton X-100 tends to decrease the Ca'+-activated ATPase activity observed a t high concentrations of NaCl.
Effect of Triton X-100 on the Sedimentation of LAD-1-Exposure of LAD-1 in 0.6 M NaCl/Mg'+ solution to 0.1% w/v Triton X-100 for 10 min a t room temperature (23 "C), with subsequent sedimentation a t 4 "C in sucrose gradients made up in the same solvent causes substantial changes in the sedimentation pattern of ATPase activity, as described earlier by Gibbons and Fronk (16). Electrophoretic analysis of such gradients (Fig. 5a) Fig. 2, and were dialyzed back into 0.6 M NaCI/Mg" solution prior to loading on the second density gradient. The Ap/IC fraction ( b ) was contaminated with a small amount of A, chain which sediments as an (AJA,,) complex peaking in fractions 12 and 13. d is the same type of sample as c, but shows a different preparation of pooled A,, and A/,/IC fractions. In this experiment, a smaller proportion of the A heavy chains reassembled into NUMBER a 21 S particle (as compared with c), but the higher protein load enables the distribution of the intermediate chains to be seen more clearly. Latent ATPase activities of 0.04 pmol of Pi min" and 0.14 pmol of P, min" were recovered from gradients c and d, respectively. These activities correspond respectively to 100% and 135% recovery relative to an equal weight of fresh LAD-I (0.15 mg for gradient c, 0.40 mg for gradient d ) . High ATPase recoveries are caused by the incomplete reconstitution of low latent activity 21 S particles. Only the gel lanes corresponding to fractions 2 to 16 are shown in these and 120 p1 were loaded in b and d . Other details of electrophoresis figures. One hundred fifty p1 of each sample were loaded in a and c, and ATPase activity are as in Fig. 1.

Properties of the A., fraction, the A/,/IC fraction and LAD-1
The A. and Ap/IC fractions, obtained by dialysis of LAD-1 against For determination of K,, the amounts of protein per 2.0-ml assay IEM solution and density gradient separation in the same medium, solution were 17 pg and 3.4 pg for latent and Triton-activated dyneinwere dialyzed back into 0.6 M NaCI/Mg" solution for 24 h before 1, respectively, and 12 pg and 6 pg for the A. and Ap fractions, ATPase activities were measured. The numbers in parentheses rep-respectively. resent the number of experiments performed for that type of sample.

FRACTION NUMBER
FIG. 5. Same as Fig. 1, but showing electrophoretic patterns of fractions from a sucrose density gradient of LAD-1 treated with 0.1% Triton X-100. Gradient was prepared in 0.6 M NaCI/Mg'' solution with either ( a ) 0.1% Triton X-100, or ( b ) no Triton. One and one-half mg of protein were loaded on each gradient and 50 pl of each fraction were electrophoresed. fractions 8 and 9 but their distribution is somewhat broader than that of the AB chain and intermediate chain 1. The peak of the ATPase activity is in fraction 7 where both the A, and the AB chain are present, and the activity falls off at approximately the same rate on both the A, and AB sides of the peak.
Visual comparison of band intensities suggests that the specific activity of fractions containing predominantly the A, chain is comparable to those containing predominantly the AB and intermediate chains under these conditions, which stands in contrast to the much lower ATPase activity associated with the A, chain after low salt dialysis (Fig. 2).
If Triton-treated LAD-1 is sedimented into a sucrose density gradient prepared in 0.6 M NaC1/Mg2+ solution containing no Triton X-100, a partial re-formation of a 21 S peak containing the A, and AB heavy chains and intermediate chains 1,2, and 3 occurs (Fig. 5b). The efficiency of the reformation of the 21 S particle, judged in terms of the amount of A, and AB chains sedimenting at a velocity of 21 S, appears to be greater than 50%.
Test of Functional Activity of A Chain Fractions using Outer Arm-depleted Axonemes-Assessment of the functional capability of LAD-1 obtained from axonemes isolated from the new sucrose procedure to restore the beat frequency of outer arm-depleted sperm (le), showed that most preparations raised the average beat frequency from about 15 Hz to 24 Hz, in agreement with our earlier report using LAD-1 from axonemes prepared by the Triton procedure (16). However, some preparations of LAD-1 made by the new procedure raised the beat frequency to 28 to 29 Hz, which almost equals that of standard reactivated sperm.
The pooled A. fractions and the pooled A@/IC fractions from sucrose density gradients, either still in the IEM sucrose solution or dialyzed back into 0.6 M NaCl/Mg" solution, were tested for their effect on the beat frequency of outer armdepleted sperm. Incubation of sperm samples with the A, fraction, the AB/IC fraction, or with a 1:l mixture of these fractions, had no effect on the beat frequency when the sperm were subsequently reactivated with 1 mM ATP.
Preincubation of outer arm-depleted sperm with a 1:l mixture of the separated A, and AB/IC fractions that had been dialyzed back into 0.6 M NaCl/Mg'+ solution (designated A,-AB/IC mixture) resulted in a substantial inhibition of the frequency increase obtained upon incubation with intact LAD-1. The beat frequency of such sperm observed immediately after addition of ATP (-15 Hz) was nearly the same as that of the original outer arm-depleted sperm to which no LAD-1 has been added. However, this inhibition was partially reversed in the presence of ATP; their beat frequency rose quite rapidly with time after addition of the ATP, until by 6 to 8 min, it had increased to about 22 Hz. A similar result was obtained when outer arm-depleted sperm were preincubated with the AB/IC fraction before incubation with intact LAD-1, but in this case the increase in beat frequency following the addition of ATP occurred more rapidly so that the beat frequencies were as high as 17 to 18 Hz when initially measured and rose to 22 Hz within 2 min. Preincubation with the A, fraction alone, on the other hand, appeared to have no effect upon the restoration of beat frequency by subsequently added LAD-1. Samples of A,-Ap/IC mixture that had been heated at 50 "C for 5 min before assay no longer blocked the restoration of beat frequency by LAD-1. For reasons that are as yet unexplained, the restoration of beat frequency by LAD-1 added after preincubation with the A,-AB/IC mixture or the AB/IC fraction proceeds only to a value of -22 Hz, significantly lower than the value of -25 Hz obtained with LAD-1 alone under otherwise identical conditions.

DISCUSSION
The nature of the interactions between the two species of heavy chain, A, and AB, in the 21 S LAD-1 particle has been partially elucidated from their changes in sedimentation be-

Polypeptide Subfractions
of Dynein-1 havior as the composition of the medium is changed. In the 0.6 M NaC1/Mg2+ extraction solution which maintains the functional capability of LAD-1 and constitutes "standard conditions'' for its physicochemical investigation, the two heavy chains co-sediment in apparently equimolar quantities as a compact peak a t 21 S (16,17). In the low ionic strength IEM solution, however, the A, and A, chains are dissociated, suggesting that the interactions between them are predominantly hydrophobic (26). This is supported by the similar dissociation resulting from addition of Triton X-100 to the 21 S LAD-1 particle in 0.6 M NaCl/Mg+ solution, for such nonionic detergents mediate the exposure of hydrophobic surfaces to the external medium (27). Conversely, the aggregation of the A, chain a t low ionic strength, and the dispersion of these aggregates upon restoring to 0.6 M NaC1/Mg2+ solution (Figs. 2 and 3a), suggest that the A, polypeptide chain contains one or more regions of ionic character that tend to associate in the absence of solution ions, but that are dissociated by ionic competition in sufficiently high concentrations of salt. This hypothesis is supported by the observation that at pH 8, even a t low ionic strength, the A, chains are much less aggregated than at pH 7, presumably because they have a greater net negative charge at the higher pH (data not shown). The location of this ionic region on the A, chain is unknown, but since the outer arms are solubilized by 0.6 M NaC1, it may be the same region as that involved in the salt-sensitive binding of the 21 S LAD-1 particle to the A tubule of doublet to form the outer dynein arm (28). If this is the case, then in the absence of A-tubule binding sites, the A, chains might well interact with each other nonspecifically, which would explain the heterogeneous aggregates of the solubilized A, chain at low ionic strength (Fig. 2).
The isolated A,/IC fraction in IEM buffer sediments in the analytical ultracentrifuge with a sedimentation coefficient of 9.3 S, while its sedimentation coefficient increases to about 14.3 S after it is dialyzed back into 0.6 M NaC1/Mg2+ solution. While it cannot be completely excluded that this change in sedimentation coefficient is caused by a secondary charge effect at the lower salt concentration (29), it seems more likely to be a result of either a major conformational change in the A, chain, or a monomer-dimer transition. Sucrose density gradient centrifugation of the Ap/IC fraction indicates that the sedimentation of intermediate chain 1 closely parallels that of the AB chain under all conditions examined, suggesting that the A, chain and intermediate chain 1 occur as a complex (AB/IC1) that is maintained by relatively strong interactions over a wide range of salt concentration. Intermediate chains 2 and 3 have likewise been found to co-sediment under all conditions examined, suggesting that they also occur as a relatively stable complex (IC2/IC3).
The A, chain fraction displays no definite peak in either sucrose density gradient or analytical centrifugation at low ionic strength at pH 7.0. In 0.6 M NaC1/Mg2+ solution, it still displays no definite peak in the analytical ultracentrifuge, although when sedimented in sucrose density gradients, most of the protein in the A, chain fraction sediments as a peak with a sedimentation coefficient of about 10 to 12 S. The only other condition in which the separated A, chains sedimented as an apparently compact peak without signs of aggregation was when LAD-1 in 0.6 M NaC1/MgZ' solution was treated with 0.1% Triton X-100 and sedimented in a sucrose density gradient prepared in the same solvent (Fig. 5), and under these conditions the A, chains also sedimented at about 11 S.
The reconstituted 21 S particles obtained by mixing the separated A, and A,/IC fractions and dialyzing against 0.6 M NaCl/Mg2+ solution contain the A chains and intermediate chains in approximately the same proportions as occur in the original 21 S particle. The fact that no 21 S particle is formed by either the A, or the A,/IC fraction alone provides the fist direct evidence that LAD-1 is a single species of particle containing both the A, and A, chains in equal quantity, rather than a mixture of fortuitously co-sedimenting particles containing the A, chain and Ap chain separately. The apparent lack of interaction between the IC2/IC3 complex and either of the A, chain or the AB/IC1 complex separately suggests that the IC2-IC3 complex becomes incorporated into the 21 S particle within the interface region between the A, and Alj chains, and that its affinity for the separated A, chain and AB/ICl complex is relatively low. It is notable that when the A,/IC fraction is dialyzed alone back into 0.6 M NaCl/Mg2+ solution, the IC2/IC3 complex appears to aggregate with itself to give a peak of 15 to 17 S rather than associating with the AO/ICl complex (Fig. 3b).
The distribution of ATPase activity upon Centrifugation of preparations of LAD-1 dissociated with Triton X-100 in 0.6 M NaCl/Mg"+ solution suggests that there are approximately equal amounts of ATPase activity associated with the A,, and the A, chain fractions (Fig. 5 ) , and appears in conflict with the distribution of ATPase activity in centrifuged preparations of LAD-1 dissociated by dialysis against IEM solution, which show ATPase activity associated with the Al</IC fraction but little or no ATPase activity in the A, chain fraction. A possible explanation for this discrepancy is provided by the results showing that the dialysis-separated A, chain fraction regains a substantial amount of ATPase activity when it is dialyzed against 0.6 M NaCl/Mg" solution, whereas its activity after dialysis against an 0.6 M NaCl solution lacking Mg2+ remains minimal (Table I). The simplest interpretation of the data is that the fractions containing the A, and the A, chains initially have comparable amounts of ATPase activity, but that the activity of the A, chain fraction is labile in the absence of Mg2+ while that of the A, chain fraction is not. A subsequent dialysis of the separated A, chain fraction against 0.6 M NaC1/ Mg2+ then presumably enables a partial recovery of its ATPase activity. This interpretation is consistent with the results showing that the 21 S particles formed by recombination of the A, and AB/IC fractions in 0.6 M NaC1/Mg2' solution regain some but not all of the original properties of LAD-1, including a somewhat greater degree of ATPase latency than is present in the A, and A,/IC fractions dialyzed separately against 0.6 M NaC1/Mg2' solution. However, it has not yet been possible to obtain reconstituted 21 S particles with a full ATPase latency of approximately 10-fold, and possibly for this reason they have lacked the functional capability to restore the beat frequency of outer arm-depleted sperm flagella.
In intact axonemes in the absence of ATP, the outer arms are thought to form rigor cross-bridges between the doublet tubules, with one end of each arm being attached via an 0.6 M NaC1-sensitive binding to the A-tubule of a doublet and the other end of the arm attached via an ATP-sensitive binding to the B-tubule of the adjacent doublet (5,30). The fact that preincubation of outer arm-depleted flagella with the A,-A,d IC mixture is capable of blocking the restoration of beat frequency by subsequently added intact LAD-1 suggests that the reconstituted 21 S particles, although nonfunctional themselves, are nevertheless capable of binding to the A-tubule and/or €3-tubule sites with an affinity comparable to that of the intact functional LAD-1 particles. Since this inhibition of the beat frequency increase is largely relieved within 6 to 8 min after addition of ATP, it seems likely that the strongest interaction of the nonfunctional reconstituted 21

Subfractions of Dynein-1 515
occurs with ATP-sensitive sites on the B-tubule and that their affinity for the salt-sensitive site on the A-tubule is weak compared to that of native outer arms.
The similar inhibition of the beat frequency increase by preincubation of outer arm-depleted flagella with the separated AB/IC fraction, together with rapid relief of this inhibition by ATP, suggests that the 14 S particles in this fraction also became attached to the doublet tubules by ATP-sensitive bonds, and block either the binding or the function of subsequently added intact LAD-1 particles. The failure of the A, fraction to show a similar blocking effect may be associated with its reduced ATPase activity after dialysis and sedimentation in the absence of Mg2'.
More detailed information regarding the localization of binding of the various components will hopefully come from electron microscopy experiments now in progress.
Of the three distinct components of the LAD-1 particle identified here, two, the AB/ICl complex and the A, chain, possess ATPase activity. The substantial differences in their basic enzymatic properties (e.g. specific activity, K,,, for ATP, divalent cation dependence, etc.) argue strongly against the possibility that the two ATPases arise from cross-contamination of the fractions by a single ATPase. Thus it appears that, as in the case of Chlamydomonas flagella, where genetic and biochemical evidence indicate the presence in the outer arms of two distinct ATPases consisting of nonoverlapping sets of polypeptides (15), the outer arms of the sea urchin sperm flagella likewise contain two distinct ATPases. The different enzymatic properties of the A,-and AB-associated ATPases suggest that, rather than being equivalent to the two apparently identical ATPase-containing heads in myosin cross-bridges of muscle, the two ATPases in the flagellar outer arm serve functionally distinct purposes in motility. It is possible, for instance, that the two ATPases may control, at least partially, the direction and waveform of flagellar beating by causing tubule sliding in opposite directions, although in trypsin-treated axonemes disintegration by sliding occurs in one direction only (31). Alternatively, one of the ATPases may be the main power producer in a single direction of tubule sliding interaction while the second ATPase adds power to this interaction under specific conditions where it is required. In either case, it might be expected that the regulatory systems controlling the activity of the two ATPases would be distinct in their actions.