Effects of photo-oxidizing analogs of fluorescein on the sarcoplasmic reticulum Ca2+-ATPase. Functional consequences for substrate hydrolysis and effects on the partial reactions of the hydrolytic cycle.

Erythrosin B was used to photo-oxidize the sarcoplasmic reticulum Ca2+-ATPase. The ATPase activity is rapidly and irreversibly inhibited by photo-oxidation with erythrosin. This inhibition is protected by the presence of ATP during the photo-oxidation period. After photo-oxidation, the steady-state phosphorylation by ATP remains almost unchanged, whereas phosphorylation by inorganic phosphate is impaired. The pseudo-first order rate constants for phosphorylation by 15 μM ATP at 25°C are strongly inhibited when starting from either a Ca2+-bound or a Ca2+-free enzyme form, decreasing from 145 to 23 s−1 for the Ca2+-bound form and from 50 to 18 s−1 for the Ca2+-free form. Concurrently, the rate constants for dephosphorylation are also severely inhibited, changing from a fast double exponential to a very slow single exponential decay in the reverse direction and from a moderately slow single to a very slow single exponential decay in the forward direction. Ca2+ binding data show that the phosphorylated intermediate formed by the photo-oxidized enzyme contains two occluded Ca2+, and TNP-ATP fluorescence measurements indicate that it accumulates in a E1-P·;Ca2-like conformation. Protection by ADP against glutaraldehyde-induced cross-linking indicates that ADP binding to Ca2+-ATPase is not impaired by photo-oxidation nor by free erythrosin. These data support the view that an ADP-insensitive, Ca2+-bound, slowly interconverting phosphoenzyme is formed. Thus, photo-oxidation with erythrosin B leads to impairment of phosphoryl transfer reactions and related conformational changes.

The sarcoplasmic reticulum (SR) 1 Ca 2ϩ -ATPase is a 115-kDa transmembrane protein that couples ATP hydrolysis to removal of Ca 2ϩ from the cytosol in skeletal muscle fibers, triggering the end of muscular contraction (Hasselbach and Makinose, 1961;Ebashi and Lipmann, 1962). The primary structure of this enzyme has been determined and has led to predictions of secondary structure and enzyme topology (Brandl et al., 1986;Clarke et al., 1989). Molecular imaging of ATPase crystals seems to agree with general structural predictions (Stokes et al., 1994), but questions concerning the role of specific amino acids in binding and transporting Ca 2ϩ as well as the structure of the nucleotide binding site(s) remain unanswered. Efforts to map the enzyme functionally have used chemical modification by specific reagents and site-specific mutagenesis. This latter technique has been used recently to identify amino acids that affect kinetic parameters related to the catalytic site and to the Ca 2ϩ binding sites (Maruyama and MacLennan, 1988;Maruyama et al., 1989;Clarke et al., 1990;Vilsen et al., 1991). In addition, many amino acids have been labeled, identified, and located within the primary structure using a variety of probes for the nucleotide binding site. These experiments have almost always involved probes that bind covalently to the enzyme, such as FITC (Mitchinson et al., 1982), 8-azido-ATP (Lacapère et al., 1993), pyridoxal 5Ј-phosphate (Yamagata et al., 1993), oxidized ATP (Mignaco et al., 1990), or chemical reagents that cause radical structural changes, such as glutaraldehyde (McIntosh, 1992) or dithiothreitol (Daiho and Kanazawa, 1994). Thus, kinetic parameters such as those involved in enzyme phosphorylation and dephosphorylation could not always be assessed with the labeled enzyme, and the precise functional role of the labeled amino acids remains uncertain.
It has been reported that Ca 2ϩ -ATPase modified with FITC loses all the activities related to ATP binding (Pick, 1981;Pick and Bassilian, 1981). Because Lys-515, which is the point of attachment of FITC to the enzyme, is neither unusually reactive toward FITC (Murphy, 1988) nor essential for ATP hydrolysis (Maruyama et al., 1989), it has been suggested that it is the fluorescein moiety of FITC that directs the reaction toward the ATP site. Thus, affinity binding of fluorescein would bring the isothiocyanate group close to Lys-515, promoting a specific attachment, and fluorescein would then remain anchored to the putative ATP binding site, thus leading to inhibition of the Ca 2ϩ -ATPase activity.
In this study we use erythrosin B, a halogenated derivative of fluorescein that is considered to bind to enzyme nucleotide binding sites with high affinity and specificity (Lundblad and Noyes, 1984;Neslund et al., 1984), to photo-oxidize residues on SR Ca 2ϩ -ATPase, and we examine its influence on the partial reactions of the hydrolytic cycle. Erythrosin is capable of promoting selective photo-oxidation of amino acids, which can be useful in the identification of groups involved in catalysis (Lundblad and Noyes, 1984;Neslund et al., 1984;Halliwell and Gutteridge, 1989), and has two advantages: even after photooxidation the sites would still be able to bind ATP, and no bulky group would remain covalently bound to the enzyme.

MATERIALS AND METHODS
Enzyme Preparations-Sarcoplasmic reticulum vesicles were obtained from rabbit hind leg skeletal muscle as described by Eletr and Inesi (1972). Ca 2ϩ -ATPase was purified by method number 2 of Meissner et al. (1973) and stored in liquid nitrogen. Protein determinations were performed according to Lowry et al. (1951), using bovine serum albumin as standard.
Photo-oxidation of Ca 2ϩ -ATPase-Unless otherwise stated, Ca 2ϩ -ATPase was incubated with the dyes for 10 min at 25°C under room light in medium containing 20 mM Tris-HCl (pH 7.4), 0.05 mM CaCl 2 , 80 mM KCl, 5 mM MgCl 2 , and the dye concentrations specified in the legends before starting the reactions. The protein concentration on the assays could vary from 10 to 50 g/ml. In this range there was no significant variation in the sensitivity of the ATPase toward photoinhibition by erythrosin. However, when the protein concentration was raised to 0.2 mg/ml for the experiments of phosphorylation by P i , approximately three times as much erythrosin was needed to reach the same extent of inhibition.
Enzyme Hydrolytic Activities-ATPase activities were assayed at 37°C in reaction medium containing 20 mM Tris-HCl (pH 7.4), 0.05 mM CaCl 2 , 80 mM KCl, 5 mM MgCl 2 , 10 -20 g/ml purified Ca 2ϩ -ATPase, and 2 mM ATP. In most cases, activity was determined by counting the amount of radioactive P i released from [␥-32 P]ATP. The reaction was quenched and free nucleotide was trapped by the addition of an equal volume of activated charcoal suspended in 0.1 N HCl (Grubmeyer and Penefsky, 1981). After centrifugation the P i remaining in the supernatant was counted in a liquid scintillation counter. In some experiments, release of nonradioactive P i was measured using molybdovanadate reagent (Lin and Morales, 1977).
Differential Absorbance Spectrophotometry-The differential absorbance spectrum for control versus erythrosin-treated samples was recorded using a Milton-Roy Spectronic 3000 Array Spectrophotometer. The samples contained 0.2 mg/ml Ca 2ϩ -ATPase, 1 M erythrosin, 20 mM Tris-HCl (pH 7.4), 80 mM KCl, 5 mM MgCl 2 , and 50 M CaCl 2 . The absorbance spectra were recorded before photo-oxidation and then after a 60-min incubation period, either in the presence or the absence of light, and the differential absorbance spectrum was calculated.
Steady-state Phosphorylation of Ca 2ϩ -ATPase with [␥-32 P]ATP-Ca 2ϩ -ATPase (20 -50 g/ml) was incubated at 25°C in medium with the same composition used to measure ATPase activity but using 15 M [␥-32 P]ATP. After photo-oxidation, phosphorylation of the enzyme was started by the addition of [␥-32 P]ATP, and the reaction was quenched after 3 s by adding 10 volumes of an ice-cold solution of 250 mM perchloric acid and 5 mM P i . Samples were then filtered through Millipore filters (0.45-m pore size) and washed five times with the same volume of quenching solution. The amount of radioactivity on the filters was counted in a liquid scintillation counter, and phosphoenzyme was determined according to Knowles and Racker (1975).
Phosphorylation of Ca 2ϩ -ATPase with [ 32 P]P i -Ca 2ϩ -ATPase (0.1 mg) was phosphorylated in 0.5 ml of a medium containing 20 mM MES-NaOH (pH 6.0), 1 mM EGTA, 10 mM MgCl 2 , and 4 mM [ 32 P]P i . After 5 min, the reaction was quenched by adding 10 volumes of ice-cold 250 mM perchloric acid and 5 mM P i . The resulting samples were treated as described above to measure incorporated [ 32 P]P i .
Rapid Kinetics Experiments-The formation or decay of phosphoenzyme was followed using a rapid mixing device coupled to either two or three syringes. For phosphorylation experiments, one syringe contained Ca 2ϩ -ATPase (40 g/ml), 20 mM Tris-HCl (pH 7.4), 80 mM KCl, 5 mM MgCl 2 , 0.2 mM EGTA, and varying erythrosin, whereas a second syringe contained 20 mM Tris-HCl (pH 7.4), 80 mM KCl, 30 M [␥-32 P]ATP, 5 mM MgCl 2 , 0.3 mM CaCl 2 . In some experiments, both syringes contained 0.05 mM CaCl 2 , and the same concentration of erythrosin. The protein solution of syringe 1 was preincubated for 10 min at 25°C under room light before initiating the phosphorylation reactions. The solutions were forced through a mixing chamber connected to a capillary tube, whose length was varied to obtain different reaction times. This capillary emptied into a quenching tube containing 5 volumes of a mixture of 250 mM perchloric acid and 5 mM P i . For dephosphorylation experiments, the capillary was connected to a second mixing chamber, to which a third syringe delivered the dephosphorylation solution (a volume equivalent to 88% of the sum of volumes of syringes 1 and 2). Syringe 3 contained 22.8 mM Tris-HCl (pH 7.4), 91.2 mM KCl, 5.7 mM MgCl 2 , 6.84 mM EGTA, 0 or 2.28 mM ADP, and 1.14 times the dye concentration used in syringes 1 and 2. Reactions were quenched in acid, and phosphorylated protein samples were treated as above.
Ca 2ϩ Dependence of Enzyme Activity-Free Ca 2ϩ concentrations were calculated with the Ca 2ϩ -EGTA dissociation constants of Schwarzenbach et al. (1957), using an algorithm developed by Fabiato and Fabiato (1979). Reaction media were the same as that used for ATPase activity measurements, except that 2 mM CaCl 2 and various EGTA concentrations were added.
Steady-state Ca 2ϩ Binding-Ca 2ϩ binding to control and erythrosinphoto-oxidized Ca 2ϩ -ATPase was measured in medium containing 10 M [ 45 Ca]CaCl 2 , 50 g/ml Ca 2ϩ -ATPase, 2 mM ATP, 20 mM Tris-HCl (pH 7.4), 80 mM KCl, and 5 mM MgCl 2 . Reactions were stopped by Millipore filtration, and the retained radioactivity was counted. Blanks obtained either by processing the reaction at pH 5.0 or by omission of the enzyme were discounted to correct for nonspecific binding of [ 45 Ca]Ca 2ϩ to the enzyme and for the dead volume of the filters, respectively. Both blanks were identical, and the radioactivity corresponding to specific binding was always equivalent to approximately twice the radioactivity of the blanks.
Cross-linking with Glutaraldehyde-After photo-oxidation with 1 M erythrosin for 60 min, the Ca 2ϩ -ATPase was further incubated with glutaraldehyde for 90 min in the presence of 0 -100 M ADP as described by McIntosh (1992). The medium contained 0.2 mg/ml Ca 2ϩ -ATPase, 20 mM MOPS-Tris (pH 8.1), 80 mM KCl, 5 mM MgCl 2 , 0.05 mM CaCl 2 , and 0.142 mM glutaraldehyde. The cross-linking reaction was quenched with one volume of a denaturing buffer containing 2% SDS, 8 M urea, 200 mM Tris-HCl (pH 6.8), and 500 mM 2-mercaptoethanol, and submitted to SDS-polyacrylamide gel electrophoresis (Laemmli, 1970). In another set of experiments, photo-oxidation was omitted, and ADP was replaced by equivalent concentrations of erythrosin.
Fluorescence of TNP-ATP-Fluorescence was measured at room temperature in a continuously stirred 1-cm light path quartz cuvette using a Hitachi F-3010 spectrofluorimeter. Excitation was set at 408 nm, and emission was set at 540 nm. The emission values were corrected for erythrosin fluorescence and filter effects.
Computer Analysis-Kinetic reactions were simulated in an IBM PC computer, using a program described by Hecht et al. (1990), modified with the introduction of input-output subroutines. The program, based in an optimization of the fourth order Runge-Kutta-Gill algorithm, was run using an integration time interval of 1 ϫ 10 Ϫ5 s.
Reagents-ATP and ADP (sodium salts), 2-mercaptoethanol, glutaraldehyde, Tris, MOPS, and EGTA were from Sigma; fluorescein (free acid or sodium salt), erythrosin B, and eosin Y were from Riedel; [ 45 Ca]Ca 2ϩ was from DuPont NEN; [ 32 P]P i was purchased from IPEN (Sã o Paulo-Brasil) and purified on a Dowex 50-W column according to Kessler et al. (1986). [␥-32 P]ATP was prepared according to Walseth and Johnson (1979). All other reagents were of analytical grade.

RESULTS
Photo-inhibition of Ca 2ϩ -ATPase with Erythrosin-After a 10-min incubation with the enzyme, erythrosin B proved to be the best photo-inhibitor when compared with fluorescein and eosin Y (Fig. 1A). Erythrosin presented the lowest K 0.5 (0.5-1.0 M), followed by eosin (K 0.5 of 4 -5 M) and fluorescein (K 0.5 Ͼ 100 M). The extent of inhibition by erythrosin increased with the time of exposure to light and also increased with dye concentration. The addition of ATP ( Fig. 1B) or ADP (not shown) during exposure to light partially protected the enzyme against photo-oxidation with erythrosin. The photo-inhibition elicited by erythrosin was irreversible, because several centrifugation-washing cycles or dilution of the phospholipids with large excess of detergents such as polydocanol or C 12 E 8 did not restore the ATPase activity (not shown). Erythrosin (and eosin, not shown) also induced the inhibition of ATP hydrolysis in the absence of light. The K 0.5 for this effect was 5-10-fold higher than in the light (Fig. 1C). Unlike the photo-inhibition, the inhibition in the dark was not time-dependent and was almost fully reversible after the washing or detergent treatments indicated above (not shown).To determine whether the protein had indeed been covalently modified by photo-oxidation with erythrosin, we measured the differential absorbance of the protein samples prior to and after reaction in the light. The differential absorption spectrum resulting from the photo-oxi-dation of Ca 2ϩ -ATPase with erythrosin is shown in Fig. 2. This spectrum agrees with the previous observations of Stuart et al. (1992) for the photo-oxidation of heavy sarcoplasmic reticulum vesicles with rose bengal and strongly suggests that the inhibition in the light is the result of a modification of some amino acid residue of the enzyme, because in the absence of light no variation in absorbance of the samples was detected. However, the photo-oxidation did not significantly modify the isoelectric point of the protein, as revealed by isoelectric focusing of the normal and oxidized protein and by nonionic slab-gel polyacrylamide gel electrophoresis (data not shown).
The ATP concentration dependence of Ca 2ϩ -ATPase activity was studied using low erythrosin concentrations (0.2 and 0.5 M) for photo-inhibition in order to achieve measurable ATP hydrolysis together with observable kinetic effects. The photoinhibition of the ATPase activity induced by erythrosin was not reversed by increasing the ATP concentration. Both V max1 and V max2 were significantly and proportionally reduced, with no change in the affinity for ATP at the catalytic site. However, a significant decrease was observed in the apparent affinity for the secondary activation induced by ATP. These data are summarized in Table I. A similar inhibitory behavior with eryth-rosin was observed for the erythrocyte Ca 2ϩ -ATPase by Mugica et al. (1984). Those authors concluded that inhibition with erythrosin was noncompetitive with ATP for the catalytic site but that the dye modified the affinity and competed with ATP for the regulatory site of the enzyme.
Ca 2ϩ Binding to Ca 2ϩ -ATPase-The affinity of the enzyme for Ca 2ϩ was not modified by photo-inhibition, considering both the high and low affinity Ca 2ϩ binding sites, although the apparent cooperativity may have been slightly decreased (Fig.  3). Equilibrium binding of Ca 2ϩ to the photo-oxidized enzyme reached the same values as in the control, attaining approximately 8 nmol of [ 45 Ca]Ca 2ϩ bound per milligram of Ca 2ϩ -ATPase with 10 M Ca 2ϩ in the medium. This result rules out the possibility that the observed decrease in ATPase activity could be due to destruction of the Ca 2ϩ binding sites transforming the inhibited enzyme into an incompetent silent form. It also establishes that inhibition is not a result of nonsaturation of these sites due to a decrease in affinity.
Effects of Photo-inhibition on the Steady-state Levels of Phosphoenzyme-The same steady-state values for phosphorylation with ATP were obtained with control enzyme and photo-inhibited Ca 2ϩ -ATPase after illumination with 5 M erythrosin during 10 min. Because ATPase activity was inhibited up to 80% in these conditions, this observation indicates that photo-oxidation led to a slower turnover rate. A slow and probably nonspecific reaction finally impaired phosphorylation by ATP (not shown).  2. Differential absorbance spectrum of Ca 2؉ -ATPase after photo-oxidation. The absorbance spectrum of a sample containing a protein concentration of 0.2 mg/ml, 1 M erythrosin, 20 mM Tris-HCl (pH 7.4), 80 mM KCl, 5 mM MgCl 2 , and 50 M CaCl 2 was recorded. After irradiation for 60 min, the spectrum of the sample was re-run, and the differential spectrum was calculated.

TABLE I
Kinetic parameters of the Ca 2ϩ -ATPase photo-inhibited with erythrosin The values are calculated from the best fit to the experimental points, considering a biphasic response to the ATP concentration using the nonlinear regression program Enzfitter. The curve was generated assuming a model of two interconverting ATP binding sites, one with high affinity that behaves as the catalytic site and one with low affinity that acts as the regulatory site. The rates and affinities for ATP hydrolysis are calculated according to the equation where V m1 is the maximal velocity at the catalytic site and V m2 is the velocity calculated for the sum of the effects of both sites. A, Ca 2ϩ -ATPase was photo-oxidized for 10 min under fluorescent room light in presence of either fluorescein (ç), eosin (É), or erythrosin (q). In this and the following experiments, activity was measured in medium containing 20 mM Tris-HCl (pH 7.4), 80 mM KCl, 5 mM MgCl 2 , 0.05 mM total or free CaCl 2 , 10 g/ml Ca 2ϩ -ATPase, 2 mM ATP, and the indicated concentration of dye at 37°C. B, erythrosin was fixed at 1 M, and oxidation was done for the designated times in presence of 0.5 mM EGTA (q) or 0.5 mM EGTA ϩ 2 mM ATP (E). Hydrolysis was started by the addition of either 0.55 mM Ca 2ϩ ϩ 2 mM ATP (q) or 0.55 mM Ca 2ϩ (E). C, inhibition of ATPase activity induced by erythrosin with (F) or without (É) photo-oxidation. E 1 ⅐Ca 2 form remained fully phosphorylatable by ATP, phosphorylation of Ca 2ϩ -free enzyme by orthophosphate was gradually impaired, showing that reactions attributed to the E 2 conformation of the enzyme were affected dramatically. The same results were obtained regardless of whether the pre-incubation with erythrosin was done in the presence or the absence of Ca 2ϩ . Therefore, the site where photo-inhibition takes place seems to be accessible to erythrosin in both the E 1 and E 2 conformations. As also shown in Table II, the effects observed with erythrosin were similarly obtained with fluorescein, in spite of fluorescein being a less effective photo-inhibitor and consequently requiring much higher dye concentrations. It is also worth noting that these results are totally different from those obtained with Ca 2ϩ -ATPase after labeling with FITC, where phosphorylation by ATP is drastically impaired (because the putative ATP binding site is occupied by the covalently bound label) and phosphorylation by P i is not affected at all (Pick and Bassilian, 1981).
Partial Reactions of the Ca 2ϩ -ATPase Photo-inhibited with Erythrosin-As described above, the steady-state levels of ATP-derived phosphoenzyme, as well as the apparent affinity of the catalytic site for ATP, were not modified by photo-inhibition with erythrosin, although overall enzyme activity was markedly decreased. The time course for phosphorylation of Ca 2ϩ -ATPase by ATP is depicted in Fig. 4. In Fig. 4, A and B, control and photo-oxidized enzyme were preincubated in the presence of Ca 2ϩ and Mg 2ϩ , and the phosphorylation reaction was started by the addition of ATP, whereas in Fig. 4, C and D, enzyme phosphorylation was initiated by the addition of ATP and Ca 2ϩ to the enzyme previously exposed to light in the presence of EGTA and Mg 2ϩ with or without the dye. In both cases, a marked reduction of the phosphorylation rate was observed in the photo-inhibited enzyme. However, inhibition with the dye induced a proportionally greater decrease (k Х 145 s Ϫ1 to kЈ Х 23 s Ϫ1 ) in the initial apparent phosphorylation rate when the phosphorylation reactions were started with the Ca 2ϩ -bound enzyme (Fig. 4, A and B) than when enzyme was preincubated in EGTA (k Х 50 s Ϫ1 , kЈ Х 18 s Ϫ1 ) beforehand (Fig. 4, C and D). Conversely, maximal phosphorylation levels similar to control tended to be reached both by the Ca 2ϩincubated and EGTA-incubated enzyme. It must be stressed that the reaction rates observed for phosphorylation of the photo-oxidized enzyme are very similar regardless of the initial (i.e. whether Ca 2ϩ was bound or not) enzyme conformation. At the concentrations used in this work, Ca 2ϩ and ATP binding to the E 1 form of the enzyme are expected to be very fast when compared with the E 2 -E 1 conformational transition rate (Scofano et al., 1979). Thus it is assumed that isomerization from E 2 to E 1 would be the rate-limiting step for phosphorylation of the native Ca 2ϩ -ATPase when beginning with the E 2 form of the enzyme. However, after photo-oxidation this rate-limiting step seems to be superseded due to slowing of a subsequent step, which makes the overall phosphorylation reaction rates very slow, and similar irrespective of the initial conformer. Changing the rate of ATP binding would be expected to change the K d for ATP and shift the K m to higher values, a change that was not observed for the catalytic site of the photo-oxidized enzyme (see Table I). Thus, it is likely that the phosphoryl transfer reaction was affected by reaction with erythrosin.
Photo-oxidation had a pronounced effect on the dephosphorylation rates of ATP-formed phosphoenzyme. The dephosphorylation reaction was much slower than control both for the forward (EGTA-induced) and backward (ADP-sensitive) directions. The EGTA-induced phosphoenzyme decay showed a pronounced decrease in the observed rate constants for the photooxidized enzyme, decreasing from k Х 3.36 s Ϫ1 to kЈ Х 1.1 s Ϫ1 (Fig. 5, A and B). The extent of inhibition was higher the higher the dye concentrations used (not shown). Although photo-inhibition severely impaired phosphoenzyme decay in both the forward and reverse directions, the inhibition was much more accentuated on the ADP-sensitive pathway, and the typical two-exponential, fast ADP-sensitive decay (k 1 Х 183 s Ϫ1 ; k 2 Х 20 s Ϫ1 ) became so inhibited that it could be simply fitted with a slow (kЈ Х 0.75 s Ϫ1 ), single exponential decay (Fig. 5, C and  D). Binding of ADP to the phosphoenzyme was probably not impaired by photo-oxidation, because in these experiments the enzyme was already phosphorylated by ATP (evidencing nucleotide binding) and also because ADP was still able to protect against the well described glutaraldehyde-induced intramolecular cross-linking (McIntosh, 1992) in the same conditions and to the same extent for both the control and the photo-oxidized enzyme (Fig. 6A). Interestingly, erythrosin, in the absence of illumination, efficiently protected against this glutaraldehydeinduced cross-linking (Fig. 6B), suggesting that both ADP and erythrosin (and probably ATP also), may bind to a common site or induce similar enzyme conformations. Therefore, our results point toward the accumulation of a trapped, ADP-insensitive phosphoenzyme. This could be the E 2 -P⅐Ca 2 or the E 2 -P form or a modified form of E 1 -P⅐Ca 2 , which binds ADP but can not efficiently resynthesize ATP due to the lack of one or more amino acids essential to the synthesis reaction.  Determination of the Conformational Phosphoenzyme Intermediate Accumulated-As cited earlier in this text, [ 45 Ca]Ca 2ϩ binding at equilibrium to a static, nonhydrolyzing enzyme showed the same levels for control and erythrosin photo-inhibited enzyme. If the accumulated phosphoenzyme generated upon ATP phosphorylation were a Ca 2ϩ -free E 2 -P form, one would expect a decrease in the total bound [ 45 Ca]Ca 2ϩ and also in the ratio between bound Ca 2ϩ and phosphoenzyme formed. This indeed was not the case, because in all the conditions tested, with varying degrees of inhibition, the ratio of Ca 2ϩ bound to the phosphoenzyme formed was always around 2. This ratio was maintained even when ATPase activity was reduced to less than 10% of the control and even when the levels of phosphorylation with ATP were already partially impaired by photo-oxidation (Table III). Therefore, a possible accumulation of a Ca 2ϩ -free phosphoenzyme can be discarded, and the observed phosphoenzyme must thus be a Ca 2ϩ -bound intermediate.
TNP-ATP, a fluorescent ATP derivative that binds strongly but noncovalently to Ca 2ϩ -ATPase, was used to distinguish among the possible Ca 2ϩ bound, phosphorylated enzyme forms. TNP-ATP fluorescence is enhanced upon binding to the enzyme, and fluorescence is further enhanced when phosphorylation by ATP or P i occurs. This phenomenon, classically known FIG. 4. Effect of photo-oxidation on the rate of phosphoprotein formation. Ca 2ϩ -ATPase (40 g/ml) was exposed to light for 10 min at 25°C in medium containing 20 mM Tris-HCl (pH 7.4), 80 mM KCl, 5 mM MgCl 2 , and 0.05 mM CaCl 2 (A and B) or 0.3 mM EGTA (C and D), either with no added dye (open symbols) or with 5 M erythrosin (filled symbols). Reactions were started by direct mixing of a syringe containing the above solutions with a syringe containing Ca 2ϩ and ATP or ATP only (see "Materials and Methods"). The curves were fitted with a computer simulation program (as described under "Material and Methods") in which the set of constants was adjusted to fit all the partial reactions and steadystate kinetic results. The simulations for the most important intermediates formed in each reaction are also shown (see Table  IV for the rate constants used). The open symbols correspond to control, and the filled symbols correspond to photo-oxidized Ca 2ϩ -ATPase. as superfluorescence, is attributed to the accumulation of E 2 -P⅐Ca 2 and E 2 -P forms (Berman, 1986;Bishop et al., 1987). Binding of TNP-ATP to the Ca 2ϩ -ATPase was not disabled by the photo-inhibition nor by the remaining free erythrosin, because the addition of 2 M TNP-ATP to both control and modified enzyme caused a fluorescence enhancement much greater than that observed following addition of the probe to enzymefree solutions and because the fluorescence of the bound TNP-ATP was reduced to the same extent by the addition of ATP in the absence of Ca 2ϩ with both control and photo-inhibited enzyme (not shown). The affinity for TNP-ATP (or possibly the conformation of its binding site) appeared to be modified, because the enhancement of TNP-ATP fluorescence upon binding to the ATPase was consistently slightly higher for the control than for the photo-oxidized enzyme (not shown). The superfluorescence of the enzyme-bound TNP-ATP was entirely suppressed when an 80% phosphorylatable but less than 10% catalytically competent photo-oxidized Ca 2ϩ -ATPase was phosphorylated with ATP (Fig. 7), showing that perhaps a hydrophilic/hydrophobic transition in the TNP-ATP binding site, linked to the transition from E 1 -P⅐Ca 2 to E 2 -P ⅐ Ca 2 and responsible for the superfluorescence, was impaired. Indeed, the addition of ATP to the photo-oxidized enzyme (Fig. 7, right panel) induced a small decrease in fluorescence that could be due to binding of the nucleotide and accumulation of E 1 -P⅐Ca 2 , because (as cited above) the superfluorescence is observed when the accumulation of E 2 -P⅐Ca 2 or E 2 -P occurs.

DISCUSSION
The purpose of this study was to correlate the modification of amino acids of the SR Ca 2ϩ -ATPase by a dye that behaves as a nonhydrolyzable substrate analog with the effects observed on the different steps of substrate hydrolysis. This was achieved using the halogenated fluorescein derivative erythrosin B, which can induce photo-inhibition of the Ca 2ϩ -ATPase in a reaction that has been described as oxidizing residues located in particular environments, possibly within putative nucleotide binding site(s) (Lundblad and Noyes, 1984;Ray and Koshland, 1962).
Photo-oxidation with methylene blue (Yu et al., 1967) or rose bengal (Yu et al., 1974;Coffey et al., 1975) has been used to investigate ATP hydrolysis and Ca 2ϩ transport by the Ca 2ϩ -ATPase, and early findings led to the conclusion that histidyl residues, which in these cases were shown to be the main target, are critical for Ca 2ϩ binding and participate in ATP hydrolysis and phosphoenzyme formation and turnover. Some xanthene dyes are structurally similar to adenine nucleotides and have been shown to bind to several nonmembranous enzymes that have nucleotide binding sites, such as lactate dehydrogenase (Wassarman and Lentz, 1971), aspartate transcarbamylase (Jacobsberg et al., 1975), hexokinase (Yip and Rudolph, 1976), and creatine kinase (Sommerville and Quiocho, 1977). They also bind to membraneous ion-transporting enzymes, including shark rectal gland Na ϩ ,K ϩ -ATPase (Skou and Esmann, 1981;Skou and Esmann, 1983), pig stomach H ϩ ,K ϩ -ATPase (Helmich-de Jong et al., 1986), mitochondrial F 1 ATPase (Neslund et al., 1984), erythrocyte Ca 2ϩ -ATPase (Mugica et al., 1984;Gatto and Milanick, 1993), plant plasma membrane Ca 2ϩ -ATPase (De Michelis et al., 1993), yeast plasma membrane H ϩ -ATPase (Wach and Grä ber, 1991), and SR Ca 2ϩ -ATPase (Coffey et al., 1975;Morris et al., 1982;Murphy, 1988). For most of these ATPases, xanthene dyes were shown to reversibly inhibit ATP hydrolysis in the micromolar range, provided that photo-oxidation was avoided. The structural similarity of erythrosin with adenine nucleotides is well discussed in Neslund et al. (1984). Therefore, it is quite possible that fluorescein and its derivatives are binding to nucleotide sites of these enzymes. The nucleotide site or sites in different P-type ion-transporting ATPases are likely to be very similar in structure, as evidenced by amino acid homologies within P-ATPases, which reveal a highly conserved glycine-rich fragment in the segments that are predicted to form the ATP catalytic site (Taylor and Green, 1989).  2-7). In lanes 9 -14, conditions were similar but the enzyme was previously photo-oxidized for 60 min in the presence of 1 M erythrosin. Lanes 1 and 8 correspond to native Ca 2ϩ -ATPase and photooxidized control, respectively. ADP concentrations were 0, 5, 10, 20, 50, and 100 M (increasing from lanes 2 to 7 and from lanes 9 to 14). In B, the protocol was the same as for A, but 0, 1, 2, 5, 10, 20, 50, and 100 M erythrosin was added instead of ADP (lanes 2-9). Lane 1 corresponds to native Ca 2ϩ -ATPase. In this experiment the enzyme was not exposed to light, in order to avoid photo-oxidation.  7. Effect of photo-oxidation on the fluorescence of TNP-ATP bound to Ca 2؉ -ATPase. Enzyme (0.2 mg/ml) was exposed to light for 60 min either without (control, left) or with (photo-oxidized, right) 1 M erythrosin in medium containing 0.05 mM CaCl 2 , 5 mM MgCl 2 , 80 mM KCl, and 20 mM Tris-HCl (pH 7.4), and then 2 M TNP-ATP was added. Recordings show fluorescence after TNP-ATP addition, and the changes that occur when 100 M ATP is added.
A reagent frequently used for the study of ion-transporting pumps is the fluorescein derivative FITC, which in most cases binds covalently and stoichiometrically to a lysine residue in the putative nucleotide binding site, inducing a concomitant loss of the ability to bind ATP (Pick, 1981). Murphy (1988) demonstrated that Lys-515, which is specifically labeled in SR Ca 2ϩ -ATPase (Maruyama et al., 1989), is not unusually reactive toward FITC, indicating that attachment of the affinity label must be directed by the fluorescein moiety of the compound. However, Murphy's data and our own results show that the Ca 2ϩ -ATPase affinity for free fluorescein is not particularly high, being strongly increased (at least two orders of magnitude) by the addition of halogenated substitutes to the molecule. Many authors have used eosin isothiocyanate (Papp et al., 1987;Munkonge et al., 1988) and erythrosin isothiocyanate (Papp et al., 1987;Birmachu and Thomas, 1990;Voss et al., 1991) rather than FITC, due to their higher affinity and particular fluorescence characteristics, to study rotational movements of the SR Ca 2ϩ -ATPase. These authors concluded that both compounds bind to the same site and react with the same lysyl residue as FITC (Papp et al., 1987;Birmachu and Thomas, 1990). Consequently, it is expected that erythrosin itself must bind noncovalently and with high affinity to a nucleotide binding cleft. Furthermore, as the oxygen species generated upon excitation of the dye with light is rapidly deactivated in aqueous solutions (overall rate constant 5 ϫ 10 5 M

Ϫ1
. s Ϫ1 at pH 7.0, Halliwell and Gutteridge, 1989), the free radical must be produced very near the target modified amino acid. Thus, our reasoning is that erythrosin exerts photo-inhibition by modification of some residue(s) within the nucleotide binding site(s) of the SR Ca 2ϩ -ATPase, through a reaction that depends on the generation of singlet oxygen reactive species inside the site(s) by the light-excited, noncovalently bound dye (Halliwell and Gutteridge, 1989). Yu et al. (1974) demonstrated that photo-inhibition of SR Ca 2ϩ -ATPase by rose bengal was due to the modification of histidyl residues, which were preferentially lost upon photooxidation, and not to peroxidation of phospholipids, consistent with Martonosi et al. (1972). However, Yu et al. (1974) reported that Ca 2ϩ transport was more sensitive than ATPase activity to photo-inhibition with rose bengal. Morris et al. (1982) found a similar behavior using erythrosin but assumed that no photooxidation occurred in their experiments. We find that net Ca 2ϩ transport is indeed inhibited more effectively than ATPase activity, but this is due to a Ca 2ϩ leakage induced after some minutes of incubation with erythrosin, even in the absence of light. 2 Therefore, it is not clear whether the apparently higher sensitivity of Ca 2ϩ transport to halogenated dyes is due to uncoupling of the enzyme by photo-oxidation of amino acids directly related to Ca 2ϩ transport or to membrane permeabilization.
Our results show that the events related to phosphoryl transfer to and from the Ca 2ϩ -ATPase are drastically affected by photo-oxidation. Coffey et al. (1974) had already observed that both phosphorylation and dephosphorylation rates were impaired by photo-oxidation of SR vesicles. It does not seem that the residue(s) modified in our work, are directly involved in either Ca 2ϩ binding (Coan and DiCarlo, 1990) or substrate binding (Lacapère et al., 1990), because neither the Ca 2ϩ dependence of ATP hydrolysis nor the K m1 for nucleotide hydrolysis is impaired by photo-oxidation. This is further supported by the fact that photo-oxidation also impairs phosphorylation by P i , which does not depend on Ca 2ϩ binding and does not need a tight fit to a nucleotide cleft.
The time course for phosphorylation of the photo-oxidized Ca 2ϩ -ATPase by ATP is adequately simulated by a model in which only the phosphoryl transfer and the rate constants related to the E 2 -E 1 transition are changed. These simulations were based on the hydrolytic cycle of the Ca 2ϩ -ATPase presented in Scheme I, as originally proposed by Carvalho et al. (1976) and do not distinguish transitional subconformations, such as the sequential binding of Ca 2ϩ ions (Inesi, 1987), and The constants presented here are apparent values that apply to the experimental conditions used in this work (pH 7.4, 25°C, 80 mM KCl, 5 mM MgCl 2 , 0.05 mM CaCl 2 , and 15 M ATP). k ϩn and k Ϫn are the rate constants calculated for the forward and reverse reactions in the absence of erythrosin. kЈ ϩn and kЈ Ϫn are the rate constants after photo-oxidation with 5 M erythrosin. The rate constants omitted are assumed not to be modified by photo-oxidation. Step Reaction k ϩn k Ϫn kЈ ϩn kЈ Ϫn  Pickart and Jencks, 1984. b Hecht, 1990 (with modifications from Teruel et al., 1987). c Petithory andJencks, 1988. d Fujimori andJencks, 1988. e Pickart and Jencks, 1982. f Nakamura et al., 1986. g Inesi et al., 1982 h This work. i Stahl and Jencks, 1987. j Inesi and de Meis, 1989. intermediate steps after nucleotide binding and before the phosphoryl transfer reactions (Stahl and Jencks, 1987;Petithory and Jencks, 1988). Most of the rate constants used in Table IV for the unmodified Ca 2ϩ -ATPase are based on other publications (see Table IV), and after small adjustments are adequate for the simulation of the curves shown in this work (see Figs. 4 and 5). All of the results obtained with the photo-oxidized Ca 2ϩ -ATPase presented in Figs. 4 and 5 can be reasonably simulated by assuming that the rate constants related to the phosphoryl transfer reactions (steps 3 and 7 in Scheme I) and to the interconversions between the E 1 -P and E 2 -P and E 2 and E 1 (steps 5 and 9 on Scheme I) are lower than in the control (Table  IV). These values also predict the steady-state ATPase activity for the photo-oxidized enzyme at 15% of the control values, which is in good agreement with the observed experimental values (15-20%, see Fig. 1). The values for the constants after photo-oxidation were obtained by considering: (a) the low phosphorylation rate of E 1 ⅐Ca 2 by ATP (k 3 ); (b) the very slow dephosphorylation induced by ADP and EGTA (k Ϫ3 ); (c) the accumulation of a Ca 2ϩ -bound phosphoenzyme together with the observed loss of the superfluorescence of TNP-ATP (k 5 and k Ϫ5 ); and (d) the inhibition of phosphorylation by P i (k 7 and k Ϫ7 ). The values of k 9 and k Ϫ9 were adjusted so as to account for the differences in phosphorylation rates between the Ca 2ϩbound enzyme and the Ca 2ϩ -depleted enzyme (estimated as 25% of the enzyme initially in the E 2 conformation).
There are several ways to explain the results obtained with the photo-oxidized Ca 2ϩ -ATPase. One or more of the oxidized residues could be involved in: (a) the coordination of the phosphate group to be transferred, (b) the binding and transfer of water molecules within the catalytic site, and (c) the nucleophilic attack of the phosphoryl and/or aspartyl group(s), which is the basis for phosphoryl transfer or hydrolysis. Taken together with results obtained elsewhere using halogenated photo-oxidants, our data are consistent with the notion that one or more amino acids susceptible to photo-oxidation by xanthene dyes is involved in substrate hydrolysis in the catalytic site of some, if not all, P-type ATPases.