Kinetic properties of microtubule-activated 13 s and 21 s dynein ATPases: evidence for allosteric behaviour associated with the inner row and outer row dynein arms

The dynein ATPases from cilia or flagella occur in two forms that are distinguishable by their physicochemical properties as well as by their structure and organization. In native Tetrahymena cilia, 21 S dynein or dynein 1 occupies the outer row arm position while 13 S dynein or dynein 2 occupies the inner row arm position (Warner et al. 1985). The 21S particle is structurally complex (Goodenough & Heuser, 1984), comprising three different high molecular weight ATPases (M_r ≅ 400000) and numerous other polypeptides of unknown function (Porter & Johnson, 1983). Each of the three high molecular weight subunits has a binding site for microtubules and an active site recognizing MgATP^2- (Johnson, 1983). The ATPase activity of both 13 S and 21 S dyneins is activated by microtubules (Warner et al. 1985), while substrate binding by 21 S dynein dissociates the resulting tubulin–dynein complex (Mitchell & Warner, 1981; Johnson, 1983).

Rebinding of 21 S dynein to extracted doublet microtubules exhibits sigmoid behaviour (Warner & Mcllvain, 1982), although it is unlikely that this behaviour represents true cooperative interactions between tubulin-binding domains of the dynein particle. Instead, it is probably related to weak interactions between adjacent dynein particles as they reassociate with the microtubule lattice. Goodenough & Heuser (1984) have shown that the structure usually identified as an outer row dynein arm is a composite image constructed from two overlapping 21 S particles. This finding implies the existence of specific interactions between subunits of adjacent dynein arms. The structure and organization of 13 S dynein has not been thoroughly characterized, although recent data indicate that, while the particles may not overlap, they do exhibit considerable structural complexity and probably comprise two high molecular weight ATPases (Piperno & Luck, 1981; Porter & Johnson, 1983) that are related in overall organization to the radial spokes (Goodenough & Heuser, 1985; Warner et al. 1985).

The elaborate structure of the 13 S and 21 S dynein particles implies the existence of interactions between subunit domains that would probably be expressed as positive or negative cooperativity relative to substrate saturation kinetics. However, the soluble or microtubule-free conformations of the dyneins have usually exhibited simple saturation kinetics (no cooperativity). We recently demonstrated the capacity to rebind differentially the isolated 13 S and 21 S dyneins from Tetrahymena cilia to the inner row and outer row arm positions. Rebinding activates the ATPase activity of each dynein. By using differential rebinding of the two ATPases, we are now able to characterize the kinetic properties of each enzyme independently and also describe the mechanism of activation induced by microtubules with respect to cooperativity, enzyme velocity and substrate affinity.

Cilia were isolated from cultures of Tetrahymena thermophila, mucocyst-free strain B-255, by methods described elsewhere (Mitchell & Warner, 1980). All isolation, extraction and resuspension solutions contained 0^*1 mM-EGTA and 0·01 mM-leupeptin.

Cilia were demembranated by one exposure to 0·1 % Nonidet P40, collected by centrifugation, and resuspended in 20mM-HEPES, 0·1 mM-EGTA (pH7·4). In order to prepare twice-extracted doublets, dynein extraction was done in 0·5M-KC1, 10 μM-ATP, 10 μM-vanadate for 25 min, followed by desalting of the subsequent pellet fraction and low ionic strength dialysis against 1 mM-HEPES, 0·1 mM-EGTA, for 24 h. Twice-extracted doublets were collected by centrifugation and resuspended in 6 mM-MgCh, 20mM-HEPES, 0·1 mM-EGTA (pH7·4). The dynein-containing supernatant fraction from the KC1/ATP extraction was dialysed against 10mM-HEPES, 0·1 mM-EGTA and concentrated to 0·5–1 mgml^-1 on an Amicon PM30 membrane. The concentrated extract was fractionated on calibrated linear 0 to 30% sucrose density gradients (Mitchell & Warner, 1981). ATPase-containing fractions from the gradient were redialysed against 10mM-HEPES, 0·1 mM-EGTA and reconcentrated to 1–2 mg ml^-1 on the Amicon membrane.

All dynein ATPase assays were done at 20 °C in a standard assay buffer containing 6 mM-MgClz, 0·l mM-EDTA, 40 mM-HEPES, 1 mM-ATP (pH 7·4). For kinetic experiments, MgCl₂ was added 5 mM in excess of the ATP concentration. The reaction was initiated by addition of ATP and terminated by addition of cold trichloroacetic acid (TCA). The inorganic phosphate released was measured (A₆60_nm) by the method of Taussky & Shorr (1953). Specific activity was calculated from triplicate determinations for the linear region of phosphate release (5 min) at both subsaturating and saturating ATP concentrations. Enzyme concentrations used in the assays were 8 μgml^-1 for both the 13 S and 21 S dyneins.

For kinetic experiments or experiments using subsaturating concentrations of ATP, a stable ATP concentration was maintained by using an ATP-regenerating system consisting of 2mM-phosphoenol pyruvate and pyruvate kinase (23 μgml^-1) added to the standard assay buffer (Mitchell & Warner, 1980). Kinetic data were expressed as μmoles of inorganic phosphate (P _i) released per mg of enzyme or protein per minute. The data were analysed and kinetic constants derived by using the non-linear least squares computer program described previously (Warner & Mcllvain, 1982; Warner et al. 1985).

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 a protein standard. For experiments involving variable concentrations of a given protein, the concentration sequence was derived by serial dilution of a known concentration.

The Hill coefficient is a measure of a curve fit to a rectangular hyperbola. When h= 1·0, the data fit a hyperbola and exhibit simple saturation kinetics (no cooperativity). When h is ⪢1·0, the data exhibit sigmoid kinetics consistent with positive cooperativity between subunits of the enzyme or its oligomer. When h is ⪡1·0, the data exhibit anti-sigmoid kinetics, which, although consistent with negative cooperativity, are more likely to be related to a lack of homogeneity relative to the active sites in the enzyme preparation. The significance of values for h different from 1·0 was determined by an analysis of variance of the residual sum of squares resulting from use of the two non-linear equations (Warner & Mcllvain, 1982).

For graphical illustration, kinetic data are expressed in direct plots (V versus [S]) as well as in Eadie—Scatchard plots (V/ [S] versus V) or Lineweaver—Burke plots (1/V versus 1/[S]), both of which linearize the relationship between the independent [S] and dependent (V) variables when h = 1·0. However, all kinetic constants and estimated dependent variables were determined by equations (1) and (2) and were used to fit the curves in each of the three kinds of plots.

Substrate saturation kinetics (V versus [S]) for the soluble or microtubule-free forms of the 13 S and 21 S dyneins were analysed over an ATP concentration range of 1 (M to 1 mM at an enzyme concentration of 8 μgml^-1. Ten individual preparations of cilia were examined, the resulting velocity data (μmol Pi mg^-1 min^-1) were averaged and then analysed by equations (1) and (2). Both the 13 S and 21 S dyneins exhibit simple saturation kinetics (Fig. 1; h = 1·0). None of the preparations examined showed significant deviation from hyperbolic kinetics and, therefore, the observed dependent variables (V) were best fitted by equation (1), since it has the fewest (two) estimated parameters.

The kinetic data for the soluble 13 S and 21 S dyneins are summarized in Table 1. Enzyme velocity at an infinite substrate concentration (V_max) was ⪡ 1 μmol P _i mg^-1 min^-1 for both enzymes, although 13 S dynein exhibited a considerably greater range of activity between preparations of cilia. The Michaelis constant (K_m) for 21 S dynein (21 μM-ATP) was about twofold lower than the K_m for 13 S dynein (37 μM-ATP) but, again, 13 S dynein exhibited a considerably greater range of values that overlapped with the mean value for 21 S dynein. It should be noted that K_m was determined for total added ATP, not free MgATP^2-, and thus the reported values are slight overestimates (≅17 %) of the true parameter.

We also examined the kinetic behaviour of twice-extracted (2×) doublets because, as will be seen, they represent an apparently activated, albeit residual activity enzyme preparation. Estimated parameters for the 2× doublets are also summarized in Table 1. The doublets exhibit anti-sigmoid kinetics (h = 0·8) and have a considerably greater K_m (> μM-ATP) than the soluble dyneins, consistent with an activated enzyme preparation (see following sections). Although the mean value for h was <1·0, the parameter also had a broad range of frequently significant values both less than and greater than unity.

Before assessing the effects of extracted doublets on dynein ATPase activity, we first had to determine if any of the three enzyme preparations retained an endogenous substrate for the dyneins, or if the two dyneins retained an activator or inhibitor of ATPase activity. An endogenous substrate will result in a non-linear plot of velocity (μmol P_i min^-1) versus protein concentration when [S] is <K_m and linear plots when [S] is >K_m. An endogenous activator that increases V_max or an inhibitor that decreases V_max will also result in non-linear plots of velocity versus [protein] but when [S] is either less than or greater than K_m.

The specific activity of extracted doublet microtubules (μmol P_i mg^-1 min^-1) was determined for serially diluted protein concentrations of 0·02–0·51 mgml^-1 relative to 50 μM and 1 mM-ATP. The results obtained from three preparations of cilia are illustrated in Fig. 2. At protein concentrations <0·3mgml^-1, extracted doublets exhibit a strict linear relationship (constant specific activity) between velocity and concentration at both saturating and subsaturating concentrations of ATP (only data from saturating [ATP] are illustrated). At the outset of these experiments, it became apparent that twice-extracted doublets retained a relatively high orthophosphate concentration (4·8mMmg^-1 protein), which at pH7·4 exists predominantly as free H₂PO₄^-. The concentration of P_i was linear with respect to [2×] and constant with respect to both time and temperature and, thus, was treated as simple background. The source of P_i is unknown, but inasmuch as all processing solutions were free of phosphate, it must be native to the cilium, resulting perhaps from protein phosphatase activity and existing in a form only weakly associated with the axoneme. Although P_i is known to be a weak competitive inhibitor of dynein ATPase that reduces effective [MgATP^2-] by competing with ATP^4- for free Mg²⁺ (Okuno & Brokaw, 1979), its potential effect on our system is negligible because free Mg²⁺ was maintained 5 mM in excess of [ATP^4-), wherein [MgATP^2-] = 83 % and [Mg₂ATP] = 15% of [ATP^4-] (Storer & Cornish-Bowden, 1976), and K_d for MgATP^2- (1× 10⁻⁴) is about two orders of magnitude lower than for MgHPO₄ (1·3 10⁻²)Moreover, it is evident empirically from the linear relationship of activity versus [2×] at saturating [ATP] that P_i is not acting as an inhibitor of residual dynein ATPase. Likewise, and although there is no evidence that microtubule-associated guanine nucleotides are accessible to the dynein ATPases, these data also eliminate the possibility that GTP or GDP may affect enzyme activity,-inasmuch as they are also known to be weak competitive inhibitors of dynein ATPase (Gibbons & Gibbons, 1972).

The specific activity of serially diluted, soluble dynein (μmol P_i mg^-1 min^-1) was also examined relative to enzyme concentration and saturating and subsaturating concentrations of ATP. Like 2× doublets, both 21 S and 13 S dynein retained substantial background concentrations of P_i (T6 and l·3mM mg^-1 protein, respectively). 21 S dynein also exhibited a strict linear relationship (constant specific activity) between V versus [21 S] at both saturating (1mM) and subsaturating (10 μM) concentrations of ATP (Fig. 2). However, in contrast to 21 S dynein, the specific activity of 13S dynein increased as a saturable function of [13 S] at both saturating (ImM) and subsaturating (30μM) concentrations of ATP (Fig. 2), consistent with the presence of an endogenous activator in the enzyme preparation. Using equation (1) but substituting V versus [13 S], maximum specific activity at infinite [13 S] was 0·59 μmol P_i mg^-1 min^-1, which is in the range of values obtained for the activity of soluble 13 S dynein stored at high concentration (Table 1). These values are considerably greater than the activity present at low [13 S] when the concentration was obtained by serial dilution (0·16 μmol P_i mg^-1 min^-1). The concentration-dependent behaviour of 13 S dynein probably accounts for the greater variability encountered in the estimated kinetic parameters for soluble 13 S dynein (Table 1), inasmuch as activity was generally measured relative to small volumes of enzyme removed from concentrated stock solutions.

Previously we reported that when soluble 13 S or 21 S dynein was mixed with twice-extracted doublets, the resulting ATPase activity was stimulated several-fold and was accompanied by restoration of inner row or outer row dynein arms (Warner et al. 1985). To characterize dynein activation by microtubules, 13 S or 21 S dynein (8μgml^-1) was mixed with increasing concentrations of 2× doublets (0·03–0·51 mg ml^-1) and ATPase activity was determined under standard assay conditions (1 mM-ATP). The activity of 2× doublets alone (μmolP_iml^-1 min^-1), within the limits of assay sensitivity, was linear at protein concentrations <0·3mgml^-1 but became decidedly non-linear at concentrations >0_i3mgml^-1 (Fig. 3). As a consequence, ATPase activity of the 13 S + 2× or 21 S + 2× mixtures also exhibited non-linear behaviour at high protein concentrations (Fig. 3), although total ATPase activity was clearly greater than the activity contributed by the 2× doublets or soluble dyneins alone. At present we have no explanation for the non-linear behaviour, although it may relate to substrate depletion owing to ineffectiveness of the ATP-regenerating system when used in combination with extracted doublets. Nonlinearity could limit the accuracy of our kinetic assays done at high protein concentrations; however, most of our data were derived at low protein concentrations (<0·3 mgml^-1) where ATPase activity was linear (see below). After subtracting residual ATPase activity of the 2× doublets from the mixed preparations, the remaining activity associated with both dyneins (μmolPimg^-1 min^-1) appeared to exhibit simple saturation behaviour with respect to [2×], For 21 S dynein, velocity at [2×] = 0·51 mgml^-1 was >60% of V_max estimated by equation (1) and, thus, a doublet concentration ≦0·51 mgml^-1 was sufficient to derive kinetic parameters for 21 S dynein. However, for 13 8 dynein, velocity at [2×] = 0·51 mgml^-1 was only 38% of the estimated V_max. Use of higher concentrations of 2x doublets (≧1mgml^-1) was ineffective because the residual activity of the doublets introduced an unfavourable signal-to-noise ratio into the data, restricting the extent of our kinetic analysis of 13 S dynein (see below). For the data illustrated in Fig. 3, 13 S dynein was activated fivefold and 21 S dynein was activated 2·5-fold at [2×] = 0·51 mgml^-1 and 1 mM-ATP.

In order to determine true kinetic parameters for the microtubule-activated dyneins at infinite [2×] and [ATP], substrate saturation kinetics were determined relative to five concentrations of activating 2x protein obtained by serial dilution (0·032, 0·06, 0·13, 0·26 and 0·51mgml^-1; dynein = 8 μgml^-1). To maintain internal consistency in these experiments, saturation kinetics for 13 S dynein, 21 S dynein, 2× doublets, 13S + 2× doublets and 21S + 2× doublets were done simultaneously from the same preparations of cilia. The dependent variables (V) obtained from each of two preparations were averaged and the mean values analysed by equations (1) or (2). For all experiments using equation (2), curves were fitted using four to eight data points (2–4 points at [ATP] < [S]_0·5).

Because at high protein concentrations residual ATPase activity of the doublets can constitute a significant proportion of total activity (2× + 13 S or 2×+21 S) and thus interfere with analysis of dynein-related activity, we used [2×] ≦ 0·51 mgml^-1 which, as noted above, was sufficient for a complete kinetic analysis of 21 S dynein but insufficient for 13 S dynein. For the following experiments, the relative proportions of ATPase activity contributed to total activity by the added dyneins are summarized in Table 2. At the protein concentrations used, added dynein contributed a maximum of 75 % and a minimum of 25 % to total activity. Although the kinetic analysis that follows depends on subtracting residual ATPase activity of the 2× doublets from total activity to derive kinetic parameters for the microtubule-activated dyneins, we also analysed kinetic parameters relative to total ATPase activity of the combined preparations. For example, as illustrated in Fig. 4, both the 13 S and 21 S dyneins (8μ g ml^-1) when mixed with 2× doublets (0·064 mg ml^-1) exhibit sigmoid saturation kinetics (h ≅ 1·4), in contrast to the simple saturation kinetics (h = 1·0) associated with the 2x doublets or soluble dyneins alone. Moreover, both the Hill coefficient and the Michaelis constant derived for total activity uniformly exceeded the range of values observed for the soluble dyneins (Table 1). This behaviour held for every preparation of axonemes examined and for each concentration of 2× protein used to activate the dyneins.

The results from a typical kinetic experiment done at [2×] = 0·26 mg ml^-1 (after subtracting residual activity) are illustrated in Figs 5 (13 S) and 6 (21 S). Both 13 S and 21 S dynein in the presence of microtubules exhibit a significant shift to sigmoid saturation kinetics (h ≧ 2·0) accompanied by a considerable increase in both V_max and K_m([S]_0·5). For purposes of estimating kinetic behaviour, we must first distinguish between true enzyme velocity (V) and apparent velocity (V_app) because use of equations (1) or (2) relative to [2×] requires that the enzymes have no activity in the absence of extracted doublets. Because the microtubule-free dyneins have significant basal activity, this activity was subtracted from observed velocity at each [2×] to yield V_app and then re-added to yield true V_max (Table 3). In order to estimate the kinetic parameters at infinite [2×] and [ATP], values for V_max at infinite [ATP] and each of four concentrations of 2× doublets were obtained (Table 3) and then analysed by equation (1) (V_maxversus [2×]; Fig. 7). For 21 S dynein, infinite [2×] and [ATP] yield V_max = 2·65 μmol P _img^-1 min^-1, with an apparent K_d = 0·51 mg ml^-1 (see below). Owing to the limitation on 13 S kinetics discussed earlier, we can at present carry the analysis of 13S dynein no further than the data summarized in Table 3. However, by equation (1), V_max = 9·54 μmolP _img^-1 min^-1, although the range of values observed in different experiments on 13 S dynein was very large (4–16μmolP _i mg^-1 min^-1).

However, for 21 S dynein, values for all three kinetic parameters (V_max, [S]_0·5 and h) at infinite [ATP] and [2×] can be derived by replotting velocity (V) from the data used to derive the parameters in Table 3 but at four concentrations of ATP (0·1,0·2, 0·4, 1·0mM) versus three concentrations of doublets (0·06, 0·13, 0·51 mgml^-1) (Fig. 8). Equation (1) then yields V_max at infinite [2×] for each given [ATP], and if V_max obtained for each [ATP] (Fig. 8) is then plotted versus [ATP] (Fig. 9), we obtain by equation (2) estimates of true V_max, [S]_0·5and h. For 21 S dynein, V_max = 2·62 μmol P_i mg^-1 min^-1, [S]]_0·5= 141 μM-ATP and h = 1·9. In principle, V_max obtained by equation (2) (Fig. 9; V_maxversus [ATP]) should agree, as it does, with the value obtained by equation (1) (Fig. 7; V_maxversus [2×]). The average velocity for 21 S dynein (2·64 μmol P_i mg^-1min^-1) converts to a turnover rate of 88 mol ATP per mol E s^-1 (M_r 21 S = 2× 10⁶; Johnson & Wall, 1983). The tentative velocity for 13 S dynein (9·54 μmol P_i mg^-1min^-1) yields a turnover rate of 95 mol ATP per mol Es^-1 (M_T 13S = 6×10⁵; Gibbons & Rowe, 1965). Although the pattern of changes occurring in the kinetic parameters of the activated dyneins is consistent relative to increasing concentrations of activating protein (Table 3), because of the non-linear ATPase activity discussed earlier (Fig. 3) we urge caution when interpreting values for the three parameters since they may be slight underestimates of the true values.

In principle, an approximate dissociation constant (K_d) for the dyneinmicrotubule complex can also be determined from data illustrated in Fig. 8. However, because we do not know the identity of the activating ligand (presumably tubulin), the parameter is meaningful only in view of the changes that it undergoes as a result of dynein activation or ligand binding. Just as activation of 21 S dynein results in a systematic increase in [S]o-s (Table 3), activation also results in a systematic increase in apparent K_d. Using equation (2) but substituting K_dversus [ATP] (from Fig. 8), K_d = 0·42mgml^-1 at infinite [ATP] and [2×],

By permitting the soluble forms of the 13 S and 21 S dynein ATPases to react with extracted doublet microtubules under conditions that restore inner row and outer row dynein arms (Warner et al. 1985), we have shown that rebinding of dynein to the microtubule lattice activates ATPase activity to a turnover rate of about 90 mol ATP per mol Es^-1 and at the same time decreases the affinity of the two enzymes for MgATP^2- and microtubules. These characteristics are accompanied by a shift from hyperbolic saturation kinetics (h = 1·0) to sigmoid kinetics (h= 2—3). It may be inferred that these properties approximate physiological behaviour for the functional or microtubule-activated conformations of the two dynein ATPases.

The soluble dyneins from a variety of organisms have always exhibited considerable activity in their microtubule-free conformations, regardless of the organism or method by which they were isolated, although this basal activity has been termed latent because it can be increased or activated by several different methods (Gibbons & Fronk, 1979). Basal activity typically exhibits simple saturation kinetics (h = TO) and spans a rather broad range of values (0·2–3 μmol P_i mg^-1 min^-1), although antisigmoid kinetics have been reported for soluble Tetrahymena 21 S dynein (Shimizu, 1981). The Michaelis constant (K_m) likewise exhibits a considerable range (10-50 μM-ATP) although higher affinity sites (≅1 μM-ATP)have been reported (Johnson, 1983; Shimizu, 1981). In the case of 13 S dynein, basal activity is related in part to the presence of a co-purifying, activating ligand (Fig. 2). However, the conformation of the soluble dynein particle may also have a considerable effect on its activity inasmuch as it can be quite different from the highly ordered, overlapping array of outer row dynein arms found in native cilia or flagella (Goodenough & Heuser, 1984). The potential effect of conformation on activity is particularly evident when comparing the turnover rates for the soluble forms of the two enzymes. Even though it is partially activated, the basal turnover rate for 13 S dynein (10 mol ATP per mol Es^-1) is considerably lower than the basal rate for 21 S dynein (24 mol ATP per mol Es^-1).

Under our experimental conditions, the kinetic behaviour of soluble 13 S and 21 S dynein from Tetrahymena, with minor exceptions, was not significantly different from behaviour reported previously (Gibbons, 1966). However, saturation kinetics for the activated forms of the dynein ATPases have not been described previously. Because extracted doublet microtubules retain many of their accessory proteins and structures when they are used to activate the soluble dyneins, we cannot be certain that activation is derived solely from the effect of a single ligand or even from the same ligand for both dyneins. However, using microtubules reassembled from highly purified brain tubulin, Omoto & Johnson (1984) have shown that 21 S dynein is activated by polymeric tubulin. In our experiments, the most significant activation-related change in behaviour occurs in the form of sigmoid saturation kinetics. Although sigmoid kinetics can be generated by other properties of an enzyme, in a rapid equilibrium system they are most frequently related to positive cooperativity, particularly if binding of the activating ligand by the enzyme also exhibits sigmoid kinetics. Although saturation of the doublet microtubule lattice with exogenous 21 S dynein exhibits strong sigmoid behaviour (h = 4.5; Db versus Df, derived from Warner & Mcllvain, 1982), as noted, it is unlikely that this behaviour is related to true allosteric interactions between tubulin-binding domains on the dynein particle.

The Hill coefficient for activated 13 S dynein was uniformly determined to be ≅ 2·0 (Table 3), consistent with positive cooperativity between a minimum of two interacting sites and the presence of two ATPase subunits in the inner arm particle (Piperno & Luck, 1981). In contrast, the Hill coefficient for 21 S dynein exhibited a broader range of values and frequently was ⪢ 3·0 (Table 3), consistent with a minimum of three interacting sites and the presence of three ATPase subunits in the outer arm particle (Johnson, 1983). Although for the experiments summarized in Fig. 9, h derived for 21 S dynein at infinite [ATP] and [2×] was only 1·9, this lower value may simply reflect insufficient data at ATP concentrations <[S] _0·5.

The behaviour of the two dynein ATPases may have important implications for understanding how dynein activity is coupled to motility. If this apparent cooperativity is occurring relative to a minimum to two (13 S) and possibly three (21 S) interacting sites, it seems likely to be related to subunit interactions within individual dynein particles. However, in the case of the dynein ATPases, the practical definition of cooperativity may have to be extended to include not only allosteric interactions between subunits of the 13 S or 21 S particles, but also the possibility of specific interactions between subunits of adjacent dynein arms. If we postulate that a row of dynein arms is analogous to an enzyme oligomer, interactions between subunits (arms) of the oligomer may then be true allosteric interactions. The practical consequences of such behaviour could be to provide a (necessary) mechanism for regulating dynein activity at the leading and, or, trailing edges of a propagating bend. The absence of detectable cooperativity in the ATPase activity of the soluble forms of the two dyneins may be related to the absence of conformational stability that otherwise exists with dynein arms in situ, although interaction with an activating ligand may also be required for integrated activity of the different ATPase domains. It is worth noting that soluble 13 S dynein, even though partially activated by a copurifying ligand (Fig. 2), nonetheless exhibits simple saturation kinetics only (Table 1). Sigmoid kinetics occur only after addition of microtubules (Table 3).

The maximum velocity of activated 21 S dynein was stimulated about fourfold at infinite [ATP] and [2×] and was accompanied by a considerable decrease in both substrate and microtubule affinity. Although we were unable to determine kinetic parameters for 13 S dynein at infinite [ATP] and [2×], the kinetic behaviour of activated 13 S dynein appeared to be very similar to that of 21 S dynein (Table 3). Although based on qualitatively similar enzymic mechanisms (Johnson, 1983), the steady-state behaviour of the dynein ATPases is in part very different from that of actin-activated myosin ATPase. Even allowing for variability resulting from different experimental conditions, actin-activated heavy meromyosin exhibits no apparent cooperativity and has a maximum turnover rate of about 30 mol ATP per mol Es^-1 (M_R 4·7×10⁵) and a correspondingly lower K_m (6 μM-ATP) (Eisenberg & Moos, 1970). The rate of product release determines the steady-state turnover rate for both myosin and dynein, and the rate for soluble dynein is about 100-fold faster than the rate for soluble myosin (Johnson, 1983). On the basis of the considerable difference in the Michaelis constant for the soluble and activated forms of dynein, product release by activated dynein is likely to be even faster. [S]_0·5 for activated 21 S dynein (141 μM-ATP)may underestimate the true value owing to limited data at ATP concentrations <[S]_0·5 (Fig. 9); however, it is consistent with the K_m determined for motility-coupled ATPase activity in reactivated sea-urchin sperm flagella (⪢ 100 μM-ATP;Gibbons & Gibbons, 1972). In contrast to microtubule-activated dynein, motility-coupled dynein activity, expressed either as beat frequency or enzyme velocity, has generally exhibited hyperbolic to anti-sigmoid saturation kinetics (Gibbons & Gibbons, 1972; Okuno & Brokaw, 1979). However, apparent saturation kinetics for motility-coupled activity may result from the normal superimposition of several interacting mechanisms or regulatory events rather than resulting solely from the kinetic behaviour of one or both dyneins.

The capability of evaluating independently the kinetic behaviour of the two dynein ATPases when recombined with native microtubules should provide an important means for studying motility-coupled ATPase activity as a function of the two rows of dynein arms.

This study was supported by research grant GM20690 from the National Institutes of Health.