Iron(III)-mediated photolysis of outer arm dynein ATPase from sea urchin sperm flagella.

Irradiation of outer arm dynein ATPase from sea urchin sperm tail flagella at 365-410 nm in the presence of Fe(III)-gluconate complex and ATP produces photolytic cleavage at two distinct sites on the beta heavy chain, located approximately 250 and approximately 230 kDa from its amino terminus. The former cut is close to or identical with the V1 site of the vanadate-mediated photocleavage (Gibbons, I.R., Lee-Eiford, A., Mocz, G., Phillipson, C. A., Tang, W.-J.Y., and Gibbons, B.H. (1987) J. Biol. Chem. 262, 2780-2786. The rate of photolysis shows a hyperbolic dependence on Fe(III)-gluconate concentration with half-maximal rate occurring at 23 microM at pH 6.3. In the presence of 0.1-0.5 mM Fe(III)-gluconate-ATP, approximately 58% of the beta chain becomes cleaved with a half-time of about 34 s; the remainder of the beta chain and almost all of the alpha chain are resistant to cleavage. This photolytic cleavage of the beta chain is accompanied by an approximately parallel loss of the dynein latent ATPase activity, whereas the Triton-activated ATPase is lost to a somewhat greater extent. Mg2+ concentrations above approximately 3 mM inhibit photolysis. Substitution of ADP for ATP changes the pattern of cleavage so that both the alpha and beta heavy chain undergo scission but at the 250-kDa site only. AMP, adenyl-5'-yl imidodiphosphate and Fe(II) do not support cleavage at either site. Trivalent rhodium-ATP complexes, as models of MgATP, can also catalyze photolysis of the beta chain at the 250-kDa site. These results suggest that photolysis results from the activation of an Fe(III)-ATP complex bound to the hydrolytic ATP binding site of the beta chain and that both Fe(III) cleavage sites are located close to the nucleotide binding site in the tertiary folding of the beta heavy chain. The cleavage reaction possibly involves initial photoreduction of Fe(III) bound at the Mg2+ binding site in the dynein.Fe.ATP complex, followed by covalent modification of an amino acid side chain that leads to eventual peptide scission.


Irradiation
of outer arm dynein ATPase from sea urchin sperm tail flagella at 365-410 nm in the presence of Fe(III)-gluconate complex and ATP produces photolytic cleavage at two distinct sites on the 8 heavy chain, located -250 and -230 kDa from its amino terminus.
The rate of photolysis shows a hyperbolic dependence on Fe(III)-gluconate concentration with half-maximal rate occurring at 23 pM at pH 6.3. In the presence of 0.1-0.5 mM Fe(III)-gluconate-ATP, -58% of the j3 chain becomes cleaved with a half-time of about 34 s; the remainder of the j3 chain and almost all of the (Y chain are resistant to cleavage. This photolytic cleavage of the fi chain is accompanied by an approximately parallel loss of the dynein latent ATPase activity, whereas the Triton-activated ATPase is lost to a somewhat greater extent. Mg2+ concentrations above -3 DIM inhibit photolysis. Substitution of ADP for ATP changes the pattern of cleavage so that both the (11 and p heavy chain undergo scission but at the 250-kDa site only. AMP, adenyl-5'-yl imidodiphosphate and Fe(I1) do not support cleavage at either site. Trivalent rhodium-ATP complexes, as models of MgATP, can also catalyze photolysis of the B chain at the 250-kDa site. These results suggest that photolysis results from the activation of an Fe(III)-ATP complex bound to the hydrolytic ATP binding site of the fl chain and that both Fe(II1) cleavage sites are located close to the nucleotide binding site in the tertiary folding of the fl heavy chain. The cleavage reaction possibly involves initial photoreduction of Fe(II1) bound at the Mg2+ binding site in the dynein*Fe*ATP complex, followed by covalent modification of an amino acid side chain that leads to eventual peptide scission.
The vanadate-mediated photocleavage of certain phosphohydrolases has been studied extensively since it was reported that the cx and @ heavy chain subunits of outer arm dynein from sea urchin sperm flagella can be cleaved at a specific site (Vl) by irradiation with near-ultraviolet light in the presence of MgATP and low micromolar concentrations of vanadate, Vi' (Lee-Eiford et al., 1986;Gibbons et al., 1987Gibbons et al., , 1989. The * This work was supported by Grant GM30401 from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. available evidence suggests that monomeric Vi acts as a structural analogue of phosphate that binds to the dynein at the T-Pi locus of the hydrolytic ATP-binding site on each heavy chain, where it acts as a chromophore and catalyzes photolytic scission at a nearby site in the polypeptide backbone. Further exploration of the cleavage reaction has shown that irradiation in the presence of oligomeric Vi species, in the absence of ATP, can induce photolysis of dynein heavy chains at a different (V2) site that is located about 100 kDa toward the amino terminus from the Vl site . Vi-mediated photocleavage occurs in dyneins from numerous other sources including the inner arm dyneins of sea urchin sperm flagella Milgram et al., 1987) and axonemal dyneins from Z'etruhymena cilia and Chlamydomonas flagella as well as cytoplasmic dyneins from brain and nematode (Porter et al., 1987;King and Witman, 1987;Paschal et al., 1987;Lye et al., 1987). Thus, Vi-mediated photocleavage has provided a new experimental approach to gain information on the structure and function of dyneins and has been particularly helpful in mapping the linear structure of their heavy chains King and Witman, 1988).
More recently, it has been reported that oligomeric Vi produces photocleavage at three distinct sites on the heavy chains of myosin from rabbit skeletal muscle and suggested that these cleavage sites may correspond to phosphate-binding sites on the myosin heavy chain (Mocz, 1989). Other similar work has indicated that Vi-mediated photooxidation of myosin subfragment 1 and of ribulose-bisphosphate carboxylase/oxygenase modifies a serine at the active sites to a serine aldehyde and that this can be followed by cleavage of the polypeptide backbone (Cremo et al., 1988;Grammer et al., 1988;Mogel and McFadden, 1989). In addition, unpublished results from this laboratory' indicate that a Vi-mediated photolytic cleavage occurs also in pyruvate kinase from rabbit and chicken muscle.
Since at least four diverse enzymes involving nucleotides are susceptible to Vi-mediated photolysis at specific sites believed to be close to their nucleotide-binding site, photocleavage offers a potentially valuable approach to probing the detailed structure of the catalytic site of these enzymes. The present study was undertaken in an attempt to broaden the scope of photolytic cleavage by exploring the possibility that some transition-metal cations can act as MgZ' analogs capable of functioning as photolytic chromophores at the Mg2+ locus of the catalytic site, just as Vi appears to act as a Pi analog that binds to the y-phosphate locus in the ATP-binding site. In this paper, we report that iron(III)-nucleotide complexes, and to a somewhat lesser extent rhodium(III)-nucleotide complexes, promote photocleavage of the dynein heavy chains. Of particular interest is the fact that the pattern of photolytic Iron(III)-mediated Photolysis of Dynein cleavage catalyzed by iron(III)-nucleotide, unlike the corresponding pattern with Vi-nucleotide, differs depending on whether the nucleotide is ATP or ADP and also distinguishes between the structures of the active sites in the (Y and /3 heavy chains of dynein. Some of these results were presented in a preliminary report . tative work were performed in a darkened room. Other conditions were as described for vanadate cleavage . Gel Electrophoresis-Polyacrylamide gel electrophoresis in the presence of sbdium dodecyl sulfate was carried out on 6.5% gels according to Dreyfuss et al. (1984). Gels were stained with Coomassie Brilliant Blue R-250 (Serva). Quantitation of proteins in gel bands was performed by extracting the protein-associated Coomassie dye with 75% (v/v) dimethvl sulfoxide as described previously (Gibbons et al., 1987j.' -Since irradiation in the presence of Fe(II1) tends to form -S-S-

MATERIALS AND METHODS
Dynein Preparation-Outer arm dynein was isolated from sperm flagella of the sea urchin Tripneustes gratilla as described previously . The extracted dynein was precipitated with 60% saturated (NH&SO4 and then dialyzed for 24 h with three changes against a standard acetate medium containing 0.45 M sodium acetate and 50 mM MES/NaOH buffer, pH 6.3, with or without 7 mM 2mercaptoethanol as required for particular experiments. The (Y and p subunits of dynein were separated by dialysis against a low salt medium, followed by sucrose density gradient centrifugation . Assay Procedures-Standard ATPase assays in the presence of MgATP were performed in 4 mM MgSO+ 1 mM ATP, 0.5 mM EDTA, 30 mM Tris-HCl buffer, pH 8.0. ATPase assays in the presence of Fe(III)-ATP and Rh(III)-ATP were performed in a darkened room in a medium containing 0.25-1.0 mM trivalent metal, 1 mM ATP, 1 mM 2-mercaptoethanol, 30 mM HEPES/NaOH buffer, pH 7.0; the background level of phosphate liberated in the absence of trivalent metal (amounting to 8.1 + 1.8% of the level obtained with MgATP) was subtracted from the values obtained. Activation of the dynein ATPase activity was carried out by incubating the enzyme in assay buffer containing 0.1% Triton X-100 and no ATP for 10 min at room temperature prior to the assay. Inorganic phosphate was determined by the procedure of Fiske and SubbaRow.
Preparation of Fe(IZZ)-Gluconate Complex-The Fe(III)-gluconate complex (FeGH-) was prepared essentially by the method of Pecsok and Sandera (1955), with some modifications that facilitate the preparation of solutions free of hydrated ferric hydroxides. First, ferric perchlorate was prepared by repeated (3 X) precipitation of ferric ammonium sulfate (20 mM) with ammonium hydroxide, followed by washing the precipitate with distilled water and then dissolving in perchloric acid (1 M final concentration of acid). Then, 50 ml of ZO DIM ferric perchlorate were added to 50 ml of 22 mM sodium nluconate. The DH was adiusted to 4.0 with KOH. followed bv filtering at 4 "C to remoie precipitated KClO,. Finally, the pH of the"resultani FeGH-solution was adjusted to 7.0 with NIOH. The Fe(II1) content of the complex was routinely determined spectrophotometrically using molar extinction coefficients of 2800 and 2350 M-' cm-' at 300 and 320 nm, respectively (Pecsok and Sandera, 1955). This procedure was confirmed by volumetric titration of FeGH-with permanganate by the standard Zimmermann-Reinhardt method.

Preparation of Rhodium(M)
Polyphosphates-Rhodium perchlorate was made from rhodium chloride hydrate (Alfa, Morton Thiokol Inc.) according to the method of Shukla (1961). Complexes of rhodium with ATP, ADP, tripolyphosphate, and pyrophosphate were prepared from rhodium perchlorate according to the method of Lin et al. (1984).

Conditions-Irradiation
was usually performed for l-15 min with near-ultraviolet light (365-nm peak) from a model EN-28 lamp (Spectronics Corp., Westbury, NY) as described previously for photolysis of dynein with V, . The standard irradiation medium contained 0.35 mM FeGH-, 0.7 mM ATP, 0.45 M sodium acetate, and 50 mM MES/NaOH buffer, pH 6.3.
During the course of the experiments it was found that, unlike the case with V,, the action spectrum of the cleavage reaction extends well into the visible, with moderately rapid cleavage occurring even in the presence of normal room lighting. A 5-watt visible range "cool white" fluorescent tube (F6/T5/CW, General Electric Co.) promoted the cleavage almost to the same extent as the standard 8-watt near-UV lamp. Although the effective upper wavelength limit has not been determined, preliminary results indicate that cleavage can be obtained at least up to 425 nm. For this reason all manipulations for quanti-bonds that bridge the cleaved peptides, it was found essential to heat the samples to-above 50 "C in the presence of 1% sodium dodecyl sulfate and 1% 2-mercaptoethanol after irradiation. The tendency to form disulfides was confirmed by measuring the cysteine content of dynein samples that had been irradiated with FeGH-in the presence and absence of ATP. In dynein irradiated with FeGH-alone, which does not result in cleavage, approximately 6 cysteine residues/m01 of dynein became oxidized, whereas approximately 11 residues underwent oxidation in the samples containing ATP. Samples irradiated in the presence of 10 mM dithiothreitol do not require subsequent heating to demonstrate cleavage peptides. Zmmunoblot Analysis-Transfer of proteins separated in the gels onto polyvinylidene difluoride membranes and staining with monoclonal antibodies directed toward knownregions of the p heavy chain of dynein were performed as described by .

Photolysk of Dynein by Irradiation of 365 nm in ATP and
FeCH--Preliminary experiments showed that dynein heavy chains can be photocleaved by irradiation in the presence of ATP and any of a variety of Fe(II1) complexes including Fe(III)-gluconate (FeGH-), Fe(III)-citrate, and Fe(III)-nitrilotriacetate or (less reproducibly) with well buffered solutions of inorganic ferric salts. Because of its stability and easy preparation, the gluconate complex FeGH-was selected for most subsequent experiments. It was used routinely in the concentration range 0.1-0.6 mM together with 0.1-1.0 mM ATP in the absence of Me and EDTA. As shown in Fig. 1, irradiation in the presence of FeGH-and ATP yields four new peptides, indicating that photolysis occurs at two distinct sites, which we will term Fe-a and Fe-b. The peptides formed by cleavage at the Fe-a site have molecular masses of 250 and 220 kDa, while those formed by cleavage at the Fe-b site have masses of 240 and 230 kDa. (Evidence for the assignment of these pairs is given in Fig. 6, described below.) The molecular masses of the peptides formed by cleavage at the Fe-a site appear identical with those of the HUVl and LUVl peptides formed by vanadate-mediated cleavage at the Vl site (data not shown). No cleavage was observed in samples irradiated in the standard medium in the absence of ATP (Fig. 1). Samples irradiated in the presence of such structure-changing agents as 8 M urea or 1% sodium dodecyl sulfate showed no detectable cleavage of dynein heavy chains after irradiation for 1 h with 0.5 mM FeGH-and 1 mM ATP indicating that the Fe(III)-mediated photolysis is a selective process and involves a site-directed mechanism.
Presence of a 20-fold molar excess of EDTA over the concentration of FeGH-in the irradiation medium almost completely inhibited cleavage. The cleavage rate with FeGH-and ATP, determined from the rate of loss of intact heavy chains by quantitating the Coomassie Blue-stained polyacrylamide gels, shows an exponential plus constant process, where the constant represents an uncleavable fraction of the heavy chains (Fig. 2, dashed line). From experiments similar to the one shown in Fig. 1, except that more time points were taken in the initial phase of photolysis, a half-time of 34 f 5 s was obtained. The ironmediated photolysis is thus substantially faster than the cleavage by Vi, which has a half-time of about 8 min under the same irradiation conditions. The average conversion efficiency was 58 f 9%, calculated for the p heavy chain alone since it will be shown later that only this chain undergoes cleavage under the conditions used. It is not yet known whether the uncleavable fraction (-40%) of the dynein /3 chain derives from an abortive side reaction that does not result in peptide bond cleavage or whether it corresponds to a fraction of the p chain that is completely insensitive to photoreaction, possibly as a result of being in a different regulation state. ATPase Activity of Photocleaued Dynein-The addition of 0.5 mM FeGH-to standard ATPase assay medium with EDTA 3 The molecular masses of the HUVl and LUVl cleavage peptides were originally reported as 230 and 200 kDa, respectively (Lee-Eiford et al., 1986;Gibbons et al., 1987). However, subsequent revisions to the generally accepted masses of the standard proteins myosin and spectrin have indicated that these values should be revised upward to about 250 and 220 kDa . and thiol-protecting agents omitted has only a small effect on dynein ATPase activity provided the reaction is kept completely dark. In a typical experiment with ATPase assay medium containing 0.35 mM FeGH-, the latent ATPase activity was increased 15% and the Triton-activated ATPase activity decreased by 10%. Irradiation of dynein with FeGHfor 15 min in the absence of ATP produces about a 30% increase in its latent ATPase activity and a 15% increase in its Triton-activated ATPase activity. The same irradiation with FeGH-and ATP leads to 20% loss in latent activity and 40% loss in Triton-activated activity which occurs somewhat more slowly than heavy chain cleavage (data not shown).
Assay of the ATPase activity of dynein subjected to irradiation in the presence of 10 mM 2-mercaptoethanol shows a simpler pattern of changes. Irradiation in the presence of FeGH-with no added ATP produces almost no change in either the latent or the Triton-activated ATPase activity (Fig.  2). The latent ATPase activity of dynein irradiated in FeGHand ATP decreases at approximately the same rate as the cleavage of the heavy chain, resulting in 30% activity loss, whereas the Triton-activated ATPase declines about 70%. These data show that the loss of dynein ATPase activity is largely related to the cleavage process. Part of the ATPase loss is probably due to side reactions, which is indicated by the fact that the decrease of the Triton-activated ATPase activity is larger than the amount of heavy chain cleavage.
The ATPase activity of the pooled p heavy chain fraction from sucrose density gradients that had been irradiated in the presence of 0.35 mM FeGH-with and without ATP before separation was also assayed to clarify what proportion of the total activity is lost by scission of the p chain alone. The latent ATPase activity of the p subunit irradiated for 10 min in the absence of ATP increases 20%, while the Tritonactivated ATPase activity shows a 5% increase. Upon irradiation in the presence of ATP, both the latent and the Tritonactivated ATPase activities decrease to 25 and 15% of that of the unirradiated controls, respectively. At the same time, 62% of the heavy chain is cleaved. Thus is appears that most, if not all, of the changes in the ATPase of intact dynein are due to activity changes in the fl chain. Note that the Triton activation ratio differs in the separated fi chain fraction and in intact dynein, their values being approximately 2 and 10, respectively, and therefore, it is not possible to make a direct comparison between the observed changes in their latent ATPase activity.
Effect of pH-A pH rate profile for the cleavage reaction was determined by quantitation of gel bands. It was found that the rate of cleavage increases appreciably with decreasing pH. This is interesting because nearly the opposite effect has been observed with Vi-mediated photocleavage. The rate of cleavage at pH 7 was about 70% of that at pH 6, and there was no detectable cleavage at pH values above 8.5. Therefore, a buffer of pH 6.3 was adopted for routine use.
Dependence of Initial Rate of Photocleavage on FeGH-and ATP Concentration-At 1 mM ATP the rate of photocleavage follows a hyperbolic dependence on FeGH-concentration between 1 FM and 0.6 mM, with a half-maximal rate occurring at 23 + 10 pM and a plateau leveling off at >200 ELM (data not shown). These values indicate that the iron complex binds to dynein somewhat more weakly than Vi, with which the halfmaximal rate is obtained at about 4 j.tM . The dependence of the rate of cleavage upon ATP concentration over the range ~KM-1.5 mM was determined at a constant concentration of FeGH-(0.35 mM) and showed a similar hyperbolic profile with a half-maximal rate occurring at 46 f 13 ELM ATP. This is consistent with the hypothesis that an equimolar complex of Fe(II1) and ATP is the chromophore responsible for cleavage.

Effects of Diualent Cations-Irradiation
in the presence of ATP and freshly prepared solutions of Fe(I1) sulfate or of FeGH-reduced with 5 mM dithionite prior to addition to dynein gave no cleavage. On the other hand, cleavage was produced upon irradiation with Fenton's reagent (Fe(I1) and hydrogen peroxide).
These data show that iron in the 3+ valence state is required for the photocleavage reaction. The addition of 3 mM Mg*' or 1.0 mM Mn'+, Co'+ Cu'+, or Zr?' to standard irradiation medium containing 0.7 inM ATP and 0.35 mM FeGH-almost completely inhibits photolysis at both cleavage sites (data not shown). This inhibition by divalent cations, together with the fact that no cleavage is observed in the absence of ATP (Fig. l), suggests that the complexation of Fe(II1) with ATP is required for it to promote photocleavage.
The presence of 3 mM Mg*+ together with ATP in excess (5 mM) similarly inhibits photocleavage by 90-100% (Fig. 3). This implies that the presumptive Fe(III)-ATP complex competes with MgATP for binding to the hydrolytic site on the dynein where it acts as the chromophore promoting photolysis. The hypothesis that Fe(III)-ATP binds to the hydrolytic site was examined by testing its ability to function as a substrate. When the ATPase activity of intact dynein was assayed in the absence of Mg*+, the presence of the 0.25-1.0 mM Fe(III)-ATP complex consistently increased the levels of both latent and Triton-activated ATPase activity by 3.4 + 2.1% of the normal MgATPase activity over the background levels observed in the absence of Fe(II1). The fact that this increase in activity was observed with both the latent and Triton-activated ATPase suggests that the Fe(III)-ATP complex is able to function as a substrate.
Effect of Different Nucleotides-Among the nucleotides tested, ATP is unique in supporting photocleavage at both the Fe-a and Fe-b sites to form four cleavage peptides. Substitution of ADP for ATP supports photocleavage at essentially only a single site at or close to the Fe-a site, yielding two principal cleavage peptides of mobilities similar to Ha and La (Fig. 4); a minor amount of cleavage also occurs at the Fe-b site. The rate of cleavage with 0.7 mM ADP is similar to that with ATP, although the extent is usually greater. AMP-PNP or AMP does not support cleavage at either site. Comparison of the ability of other nucleotides to support photocleavage indicated that at a concentration of 1 mM nucleotide and 0.35 mM FeGH-cleavage occurred at one site only in the order CTP > ITP >> GTP, but the extent in all cases was less than with ADP. No cleavage was observed when the Fe(II1) was omitted from the irradiation medium, confirming that the observed cleavage was not due to contamination with Vi. Dynein (1 mg/ml) was subjected to irradiation in standard acetate medium for 3 min with 0.35 mM FeGH-and either 0.7 mM ATP (left) or 0.7 mM ADP (right), respectively. The irradiated samples were dialyzed against a low salt medium and centrifuged on 5-20% sucrose gradients to separate the (Y and p heavy chain polypeptides of dynein. (Fraction 1 is the top of each gradient).
Competing Effect of V, on Photolysis by FeGH---In order to evaluate possible competition between Fe(III)-ATP and V,, dynein was irradiated in the combined presence of both metals. The result obtained depended upon which metal had been added to the dynein first. When Vi was added to preformed dynein. Fe. ATP complex shortly before irradiation, the FeGH-cleavage pattern was obtained, whereas when FeGHwas added to preformed dynein.ATP.V, complex, the Vl cleavage bands were generated (data not shown). This suggests that Fe(III)-ATP and Vi have to compete for binding to dynein.

Effect of Free Radical Scavengers-Among
the free radical scavenging agents tested, the presence in the irradiation medium of lo-100 mM formate, benzoate, ethanol, isopropyl alcohol or t-butyl alcohol, or of l-10 mM mannitol or thiourea had little effect on the rate of photocleavage.
The sulfhydryl protecting agents, 2-mercaptoethanol, 2-mercaptoacetic acid, and dithiothreitol (10 mM), also had no effect on the photocleavage rate. Concentrations of sodium azide greater than 10 mM partially inhibit the cleavage reaction, but this may be due to complex formation with Fe(II1) rather than to radical scavenging. The lack of quenching by free radical scavengers suggests that the cleavage process involves electron transfer without release of kinetically accessible free radicals.

Location of Photocleavage Sites in Dynein Heavy Chains-
In order to identify which of the dynein heavy chains become cleaved, samples of dynein that had been preirradiated in the presence of FeGH-and either ATP or ADP were separated into their a and p heavy chain fractions by sucrose density gradient centrifugation.
Subsequent gel electrophoretic analysis showed that the four cleavage peptides observed after irradiation in the presence of ATP all result from cleavage of the @ heavy chain at the two distinct Fe-a and Fe-b sites, while the a chain remains intact (Fig. 5, left panel). On the other hand, in the samples irradiated with 0.7 mM ADP, both the LY and ~3 chains become cleaved with this cleavage occurring at a single site (Fe-a) on the (Y chain and principally at the Fe-a site on the /? chain (Fig. 5, right panel). Irradiation in the presence of ADP yields nearly complete cleavage of all heavy chain material without the uncleavable fraction of the /3 chain observed with ATP.
In order to locate the two cleavage sites on the p chain with respect to each other and to the polarity of the chain, we used a series of monoclonal antibodies specific for previously mapped epitopes on the /3 heavy chain . The results (Fig. 6) show that the 250-and 230-kDa peptides contain Epitope 1, while the complementary 220-and 240-kDa peptides contain Epitopes 2 and 3b. This localizes the Fe-a and Fe-b cleavage sites on the p chain showing that Fea is located at or close to the Vl site and that Fe-b is located -20 kDa closer to the amino terminus (Fig. 7).
Photolysis with Rhodium Polyphosphates-The photolytic reactions of certain rhodium complexes with ATP and ADP have also been investigated in this study, although in less detail than FeGH-because the tendency of these complexes to undergo epimerization and isomerization limits their useful life to a few days after preparation. Upon irradiation at 365 nm, both the bidentate and tridentate RhATP complexes promote photocleavage of the p heavy chain at a single site close to its Fe-a site (Fig. 8). However, neither of the RhATP complexes nor the bidentate RhADP complex appears to support the photolysis of the cy heavy chain. This difference emphasizes the difference in the structure of the hydrolytic site in the two heavy chains of dynein.
Both Identical Western blots of dynein cleaved by irradiation for 3 min in the presence of 0.35 mM FeGH-and 0.7 mM ATP were reacted with monoclonal antibodies against epitopes El, E2. and E3b (clones 6-31-24. C-241-2. and C-26-2. resnectivelv (Piperno, 1984)): The leftmost iane was'stained with Coomassie" Blue (Co) to show total protein. The diagram shows the relative positions of the Fe-a and Fe-b cleavage sites on the p heavy chain in relation to the sites of Vi-mediated cleavage (Vl and V2) and to the principal sites of tryptic cleavage (Tl and T2). The numbers represent approximate molecular masses (in kilodaltons) of the various cleavage peptides subject to the uncertainties discussed previously by . Experimental conditions are as in Fig. 5 except that FeGH-was replaced by 0.20 mM bidentate RhATP complex and no free ATP was added to the medium. The irradiation time was 20 min. Control experiments showed that the two cleavage peptides coelectrophoresed with those formed by Vi-mediated photocleavage at the Vl site. strongly suggests that the rhodium chromophore in the RhATP complex is located at the Mg binding site in the dynein-RhATP complex.

DISCUSSION
In the Vi-mediated photolysis of dynein heavy chains it is generally thought that Vi replaces the y-phosphate of ATP . The fact that both the normal substrate MgATP and the inhibitor Vi appear to compete with Fe(III)-ATP in the cleavage reaction suggests that the Fe(III)-ATP also binds to the hydrolytic center of the @ chain where it can catalyze photolytic cleavage at either of two distinct positions. However, the competition by MgATP together with the similarity of the photocleavage mediated by bidentate RhATP (in which Mg2f is replaced by Rh(III), Lin et al., 1984) suggests that in Fe(III)-mediated photocleavage the iron is located at the magnesium locus rather than at the y-phosphate locus of the substrate binding site. The fact that a variety of Fe(II1) complexes and even inorganic ferric salts in highly buffered solutions promote photocleavage suggests that a common reactive iron species is involved and that the principal function of the complexing agents is to provide a stable source of inorganic ferric irons for complexing to the nucleotide and the protein. The suitability of the FeGH-complex for this purpose presumably derives from the fact that its formation constant at pH 6.3 is 3 x lOI (calculated from the equilibrium constant given by Pecsok and Sandera, 1955), substantially lower than the value of 2 x 10z7 for EDTA (Sillen and Martell, 1971), for the presence of excess EDTA inhibits cleavage. The reactive Fe(II1) species may well be the monohydroxo complex FeOH*', which possibly replaces Mg2+ to give a P-chain. FeOH. ATP complex. The strongest support for such a model comes from the competing effect of Mg2' on photocleavage. The observed decrease in cleavage rate with increasing pH also favors this model, for an increase in pH enhances formation of the dihydroxo complex Fe(OH)Z that would no longer be an analogue of Mg?+. However, other possible explanations for the effect of pH on photolysis are the involvement of a histidine in Fe(II1) complexation or the quenching of photolysis by OH-at some intermediate step. The similarity of the Rh-mediated photocleavage supports the assignment for the site of iron binding, for bidentate RhATP is a probe for the MgATP binding site with Rh3+ replacing Mg2+ in the exchange-inert Rh-nucleotide complexes (Lin et al., 1984).
The lack of photocleavage upon irradiation with Fe(II), in spite of Fe'+ being able to support dynein ATPase activity (Evans et al., 1986), suggests that the initial step in the cleavage reaction involves photoreduction of the iron in a dynein. Fe(III)OH . ATP complex. The photoactivation of this Fe(II1) might promote the formation of the monohydroxo radical Fe*+. OH via a photoinduced charge transfer from OHto the Fe3+ in the FeOH*+ complex, with subsequent reaction between dynein and Fe'+. OH leading to photooxidation of an amino acid side chain. Further reactions would presumably result in peptide bond cleavage by hydrogen atom abstraction or hydroxylation.
Such a mechanism would be analogous to that suggested for the Vi-mediated photolysis of myosin subfragment 1 in which photooxidation of a serine residue occurs as an intermediate step (Cremo et al., 1988). In the case of dynein, the involvement of amino acids other than serine must be considered possible, for Fe(II1) is frequently complexed to proteins through tyrosine or histidine as in the nonheme iron purple acid phosphatases, catechol deoxygenases, and transferrins (Casella et al., 1987). It is notable that with vanadium, iron, and rhodium the metal appears to act as a photosensitizing agent only when it is in its highest normal valency state.
The major new property of outer arm dynein from sea urchin sperm flagella that emerges from this study is that there is a clear difference in the structures of the hydrolytic sites on the 01 and /3 heavy chains and that this is reflected by their different photocleavage behavior in the presence of Fe(II1) and ATP or ADP. It is notable that the cleavage pattern with Fe(II1) and ADP resembles that obtained in Vimediated photocleavage, in which the o( and p heavy chains are both cleaved at a single site at approximately equal rates , rather than the pattern obtained with Fe(II1) and ATP. One possible explanation is that the difference in cleavage patterns between iron and vanadate is a consequence of the different locations of the chromophores in the hydrolytic sites. However, the similarity of the two patterns obtained with ADP suggests that the major reason for the distinct pattern of cleavage obtained with iron and ATP may be that the structure of the hydrolytic site in the dynein. FeOH . ATP complex is trapped in a state that occurs only transiently during normal ATP hydrolysis, whereas its structure in the dynein. FeOH. ADP complex more closely resembles that of the dynein.MgADP.Vi complex that is formed when dynein and Vi are incubated with either ATP or ADP and that is thought to resemble the natural dynein. MgADP . Pi. product complex.
The results obtained in this study encourage further investigation of the iron-mediated photolysis as an alternative experimental tool to vanadate-mediated photolysis that may bring new data to improve our understanding of the structurefunction relationships in dyneins and perhaps other proteins. As one example, the selective cleavage of the p heavy chain in the presence of FeGH-and ATP described here, used in conjunction with selective cleavage of the /3 heavy chain obtained by irradiation of the dynein with vanadate and ATP at -78 "C , may provide useful information regarding the distinct functions of the (Y and /3 heavy chains in flagellar motility.