THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 264, No. 20, Issue of July 15, pp. 11768-11776,1989

Printed in U.S.A.

The “Regulatory” Sulfhydryl Group of Penicillium chrysogenum ATP Sulfurylase COOPERATIVE LIGAND BINDING AFTER SH MODIFICATION; CHEMICAL AND THERMODYNAMIC PROPERTIES*

(Received for publication, January 9, 1989)

Robert L. Martin$, Lori A. Daleyg, Zdravko Lovricll, Lauren M. WailesII , Franco Renosto, and Irwin H. Segel From the Department of Biochemistry and Biophysics, University of California, Davis, California 95616

ATP sulfurylase from Penicillium chrysogenum is a homohexamer that contains three free sulfhydryl groups/subunit, only one of which (designated SH-1) can be modified by disulfide, maleimide, and halide reagents under nondenaturing conditions. Modifica- tion of SH-1 has only a small effect on kc, but causes the [S]o.6 values for MgATP and SO:- (or MOO:-) to increase by an order of magnitude. Additionally, the velocity curves become sigmoidal with a Hill coeffi- cient ( n ~ ) of about 2 (Renosto, F., Martin, R. L., and Segel, I. H. (1987) J. Biol. Chem. 262,16279-16288). Direct equilibrium binding measurements confirmed that [32P]MgATP binds to the SH-modified enzyme in a positively cooperative fashion (nH = 2.0) if a sulfate subsite ligand (e.g. FSO;) is also present. [“S]Adeno- sine 5’-phosphosulfate (APS) binding to the SH-modi- fied enzyme displayed positive cooperativity (nH = 1.9) in the absence of a PPi subsite ligand. The results indicate that positive cooperativity requires occupancy of the adenylyl and sulfate (but not the pyrophosphate) subsites. [“SIAPS binding to the native enzyme dis- played negative cooperativity (or binding to at least two classes of sites). Isotope trapping profiles for the single turnover of [36S]APS: (a) confirmed the equilib- rium binding curves, (b ) indicated that all six sites1 hexamer are catalytically active, and (c) showed that APS does not dissociate at a significant rate from E*APS-PPi. The MgPPi concentration dependence of [36S]APS trapping was indicative of MgPPi binding to two classes of sites on both the native and SH-modified enzyme. Inactivation of the native or SH-modified en- zyme by phenylglyoxal in the presence of saturating APS was biphasic. The semilog plots suggested that only half of the sites were highly protected. The cu- mulative data suggest a model in which pairs of sites or subunits can exist in three different states desig-

* This work was supported in part by United States Public Health Service Grant GM-26728 and National Science Foundation Grant DMB-88-02731. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Supported by a United States Public Health Service Grant in Molecular and Cellular Biology. Recipient of a Jastro Shields Re- search Scholarship from the College of Agricultural and Environmen- tal Sciences.

$ Supported in part by University of California Biotechnology Research and Education Program 3-440360-19908.

ll Present address: Clinical Hospital Center, Center of Medical Sciences, 4100 Zagreb, Salata 2, Yugoslavia.

11 Recipient of a University of California President’s Undergradu- ate Fellowship.

nated HH (both sites have a high APS affinity, as in the native free enzyme), LL (both sites have a low APS affinity as in the SH-modified enzyme), and LH (as in the APS-occupied native or SH-modified enzyme). Thus, the HH+LH transition displays negative coop- erativity for APS binding while the L h L H transition displays positive cooperativity.

The relative reactivities of like-paired SH-reactive reagents were in the order: N-phenylmaleimide > N- ethylmaleimide; dithionitropyridine > dithionitroben- zoate; thiolyte-MQ > thiolyte-MB. The log k,-, versus pH curve indicates that the pKa of SH-1 is >9. How- ever, the rate constant for SH-1 modification was near constant between pH 6 and 7 suggesting that protona- tion of a second group (pK, about 7) stabilizes ES- or a conformational change in the enzyme blocks access of SH-reactive reagents. Reaction rate versus temper- ature studies yielded the following activation energies for SH-1 modification at pH 8.0: AGS = 14,400 f 300 cal X mol”; = 25,500 f 3,500 cal X mol”; AS* = 37 2 11 cal X mol” X degree”. The results suggest that SH-1 is located in a nonpolar environment and that the transition state for modification is attained only with a substantial decrease in the ordered struc- ture of the enzyme. Overlapping peptides containing the cysteinyl SH- 1 were isolated and sequenced.

No evidence was found for in vivo modification of SH-1 under growth conditions that reduced the re- quirement for sulfate activation.

ATP sulfurylase catalyzes the first step in the biological incorporation of inorganic sulfate.

MgATP + SOj- e MgPPi + APS’

The enzyme from Penicillium chrysogenum is an oligomer ( M I 420,000 f 10%) composed of identical or very similar subunits (Mr 69,000 ? 5%). As isolated (l), the enzyme displays normal hyperbolic kinetics for both forward reaction

The abbreviations used are: APS, adenosine 5’-phosphosulfate (adenyl-5”yl sulfate); SDS, sodium dodecyl sulfate; DTNB, 5,5’- dithiobis-(2-nitrobenzoate); DTNP, 2’,2’-dithiobis-(5-nitropyridine); NEM, N-ethylmaleimide; thiolyte-MQ, monobromotrimethylam- moniobimane; thiolyte-MB, monobromobimane; HEPPS, N-2-hy- droxyethylpiperazine-N’-3-propanesulfonic acid; CAPS, cyclohex- ylaminopropanesulfonic acid; bis-Tris, 2-[bis(2-hydroxyethyl)amino] -2-(hydroxymethyl)propane-1,3-diol; PAPS, adenosine 3’-phos- phate,5‘-phosphosulfate.

11768

Chemical Modification of ATP Sulfurylase SH Groups 11769

substrates and for a variety of oxyanions that serve as alter- native inorganic substrates (e.g. MOO:-, SeOz-) or dead end inhibitors (e.g. FSO;, Cloy). Under nondenaturing condi- tions, one SH group/subunit (designated SH-1) can be mod- ified by DTNB, NEM, and several other sulfhydryl-reactive reagents (2). Modification of SH-1 has little effect on kcat but causes the [SIos for both MgATP and SO:- to increase and the kinetics to become sigmoidal with a Hill coefficient ( n ~ ) of 2. Inactivation protection studies suggested that the sig- moidal velocity curves reflected positive cooperativity in ter- nary E . MgATP. SO:- complex formation, rather than a sub- strate-dependent shift of the reaction flux to an alternative path (2). The kinetic consequence of SH-1 modification is identical to that produced by a "K-type" allosteric inhibitor and consequently, may have physiological significance. Alter- natively, in uitro SH modification may have no in vivo coun- terpart; modification may simply have uncovered a facet of site-site interactions that is "kinetically invisible" with the native enzyme. In either case, it was of interest to ( a ) verify the MgATP binding profiles by more direct methods, ( b ) extend the binding studies to include APS (whose binding could not be analyzed accurately by the inactivation protec- tion method), (c) characterize the chemical properties of SH- 1, ( d ) search for evidence of in uiuo SH modification, and ( e ) determine the amino acid sequence in the region of SH-1.

EXPERIMENTAL PROCEDURES

General Methods and Supplies

The purification of ATP sulfurylase and APS kinase (l), the preparation of "carrier-free'' [35S]APS (l), the enzyme assay methods

and SH modification (2) have been described. [Y-~'P]ATP (4500 Ci (I), and the procedures used to measure enzyme inactivation (3, 4)

X mmol") was obtained from ICN Radiochemicals. The material was mixed with unlabeled ATP to obtain a specific activity of about 150,000 cpm X nmol" for binding studies. Unlabeled APS was obtained from Sigma. The exact concentration of APS in stock solutions was determined with either APS kinase (14) coupled to pyruvate kinase and lactate dehydrogenase, or with ATP sulfurylase coupled to glucose-6-phosphate dehydrogenase. Coupling enzymes were Sigma products. Unlabeled APS was mixed with carrier-free [35S]APS to obtain working solutions of the desired concentrations and specific activity (700-450,000 cpm X pmol"). Unless otherwise indicated, all solutions were prepared in 0.05 M Tris-C1 buffer, pH 8.0, a t 30 "C. Peptide sequencing was performed by Alan J. Smith of the University of California Davis Protein Structure Laboratory.

Equilibrium Binding

Equilibrium binding was measured by centrifugal ultrafiltration through Millipore Ultrafree-MC (30-kDa cutoff) filter cartridges con- taining PTTK polysulfone membrane filters. To avoid formation of a vacuum, the snap-lid of each cartridge was pierced with a needle. Just before use, the filters were pretreated by centrifuging 400 p1 of the standard Tris-C1 buffer (Millipore Personal Centrifuge, XX42- CF060). The walls of the filter cartridge were blotted dry and then 400 pl of sample at 25 "C (enzyme plus radiolabeled ligand at the desired concentration) was added. The free ligand concentration was determined by counting 30 or 40 p1 of filtrate obtained after 6-8 s of centrifugation in an Eppendorf 3200 centrifuge. The short centrifu- gation time allowed only about 15% of the total solution to pass through the filter. The remaining sample was then completely cen- trifuged through the filter (5-8 min in the Eppendorf centrifuge). The cartridge walls were blotted dry and the filters placed in scintil- lation vials where they were washed with 200 p1 of 10% (w/v) SDS plus 1 ml of Tris-C1 buffer. After about 5 min, 9 ml of scintillation fluid (Safety-Solve; RPI Corp.) was added and the bound ligand counted. (The SDS treatment was not necessary for counting [3'P] MgATP.) Nonspecific binding to the filter was determined in the same way using samples from which the enzyme was omitted. Free, bound, and total ligand concentrations were determined from the known specific radioactivities of the substrates under the above counting conditions. The free and total concentrations were calculated

taking into account the small (<5%) decomposition of [35S]APS to 35SO:- or [32P]MgATP to ''Pi. "SOO:- or 32P, were determined by adding 30-pl samples of radioligand solutions to 2 ml of Tris-C1 buffer containing 2% (w/v) charcoal and 0.1 M Na2S04. After centrifugation, an aliquot of the supernatant solution was counted.

For a high affinity ligand (e.g. APS) the above method permitted us to make two separate measurements on the same sample: (a) [s]k,und as the difference between [SI,,, (above the filter) and [S]f,, (in the filtrate) after only 15% of the sample has been filtered, and (b) [S]bound directly (by counting the filter) and [SI, (by counting the filtrate) after complete centrifugation of the sample. Procedure a avoids significant losses of bound ligand from filter-enmeshed (de- natured?) enzyme. The centrifugal ultrafiltration method eliminated all potential artifacts of prolonged equilibrium dialysis at physiolog- ical temperatures, e.g. (i) slow SH-1 oxidation, (ii) slow, nonenzymatic hydrolysis of [35S]APS, (iii) slow enzyme inactivation combined with different degrees of protection at different levels of bound ligand APS (an underappreciated curve-modifying phenomenon), and (iv) possi- ble nonenzymatic modification of the enzyme by highly reactive APS forming E. SSO; or E. NH. AMP.

Single Turnouer Isotope Trapping

Varied r5S]APS Concentration-The "chase" solution (10 pl con- taining 2.2 mM N&PPi, 66 mM MgC12, and at least a 400-fold excess of unlabeled APS in Tris-C1 buffer) was placed in a 12-ml conical centrifuge tube. The solution was vortexed while 100 pl of "equilib- rium" solution was added rapidly. The equilibrium solution contained 60 nM native enzyme sites or 300 nM SH-modified sites, 1 mM EDTA, and the desired concentration of [35S]APS. Within 2 s of adding the chase, 2 ml of Tris-CI buffer containing 2% (w/v) powdered charcoal (Pfansteihl acid-washed Norit) and 10 mM EDTA was added. The samples were vortexed and then centrifuged for 10 min in a table-top clinical centrifuge. "SOO',- produced was determined by counting 0.5 ml of the supernatant solution.

Varied MgPPi Concentration-The equilibrium solution (10 pl) containing either 8.8 or 33 p~ [35S]APS (sufficient to remain satu- rating after dilution with the chase solution), 0.4 p~ enzyme sites, and 1 mM EDTA in Tris-C1 buffer was vortexed while 100 pl of chase solution was added. The chase solution contained the desired concen- tration of PPi (0.2-100 p ~ ) , 5.5 mM and at least a 400-fold excess of unlabeled APS. The charcoal/EDTA/Tris-C1 solution was added within the next 2 s and the mixture processed as described above.

Blanks in which the enzyme was added to the post-chase mixture 2 s before adding the charcoal-EDTA suspension confirmed that the post-chase release of "Soof- was negligible in all cases.

The data were analyzed by Scatchard plots of [s~bstrate]~,.~wd/ [~ubstrate]~, uersus [SUbStrate]t,pped. For varied PPi, the equilibrium concentration of free PPI was assumed to be equal to the concentra- tion of added PPi after 10/11 dilution. [PPiltraPd was assumed to be identical to the measured [35SO:-], i.e. to the trapped [35S]APS. (The reaction stoichiometry is 1:l.) Preliminary experiments a t saturating substrates established that all the bound [35S]APS was trapped as "SOO:-. Thus, when [35S]APS was varied, [APSIre was taken as the difference between total added [35S]APS and the measured [35SO:-].

RESULTS

p2P]MgATP Binding to the Native and SH-modified En- zyme-The marked synergism between MgATP and FSO, was exploited to characterize the binding of [32P]MgATP to ATP sulfurylase. Fig. l a shows the Scatchard plot for the native enzyme at 1 mM FSO,. The plot is linear yielding an apparent Kd of 1.2 pM. If it is assumed that FSO, binds almost, exclusively to E . MgATP. ( 5 ) , the Kd,app is related to the true Kd for MgATP by the relationship shown below where X = FSO; and K, is the FSO: dissociation constant of the ternary E . MgATP . FSO; complex.

Kinetic studies yield a K, of 4 p M ( 5 ) . Thus, Kd is about 0.25 mM which is in reasonable agreement with the analogous

11770 Chemical Modification of ATP Sulfurylase SH Groups

"/.' NEM-MODIFIED

01 05 10 50 10 50 100

[Mg ATPIFBEE PM (LOg Scale)

value of 0.2-0.5 mM determined from inactivation protection ( 2 ) and steady state kinetics (5, 6). The difference between [MgATP]b,,,d at saturation (about 3.5 p ~ ) and total enzyme sites present (4 p M ) is probably a result of experimental error in establishing the exact protein concentration and the pres- ence of small amounts of inactive enzyme in the preparation.

Fig. l b shows the Scatchard plot for [32P]MgATP binding to the SH-modified enzyme. The shape of the plot is charac- teristic of positive cooperativity (7). Fig. IC shows the Hill plots for the native and SH-modified enzyme. The direct equilibrium binding results are in complete agreement with the conclusions drawn from inactivation protection measure- ments: modification of SH-1 increases [S]O.S for MgATP and changes the substrate binding curve in the presence of FSO: from hyperbolic to sigmoidal.

P'SIAPS Binding to the Native and SH-modified En- zyme-Ligand binding analysis based on inactivation protec- tion requires a knowledge of the free ligand concentration. For ligands such as ATP, MgATP, and SO:- whose K d values are >lo0 p ~ , it could safely be assumed that the free ligand concentration was essentially the same as the total added concentration when total enzyme sites were C10 pM. However, APS has an extremely high affinity for ATP sulfurylase ( K , from kinetics studies is CO.l pM). Consequently, inactivation protection could not be used to establish the APS binding profiles,' but direct binding measurements were quite feasible.

Fig. 2a shows the Scatchard plot for [3SS]APS binding to the native enzyme as determined by centrifugal ultrafiltration. The curvature of the Scatchard plot suggests either negative cooperativity or at least two classes of APS binding sites. Fig.

The initial concentration of free APS could be determined inde- pendently by direct binding measurements. However, when [ E ] , = K d = [APS], and [APSIa.. was subsaturating, the concentration of [APS]fm would change as the enzyme was modified, thus complicating the determination o i the binding curve from inactivation protection. (Inactivation would not be a simple first order process.)

FIG. 1. ["PIMgATP binding to ATP sulfurylase in the presence of 1 mM FSOi. u, Scatchard plot for the native enzyme (4 PM sites). The bound ligand was determined directly by count- ing the filter. b, Scatchard plot for the NEM-modified enzyme (4 PM sites). c, Hill plots.

IAPSI FEE nM ILW -1

FIG. 2. ["SIAPS binding to ATP sulfurylase in the absence of a co-ligand. a, Scatchard plot for the native enzyme (75 nM sites). b, Scatchard plot for the NEM-modified enzyme (0.31 PM sites). c, Hill plots (closed symbols: bound = total - free; open symbols: bound determined by counting the filter).

2b shows the plot for [35S]APS binding to the NEM-modified enzyme. In this case, binding displays positive cooperativity with a Hill coefficient (nH) of 2 (Fig. 2c). In our earlier inactivation studies (2) we found that the protection provided by MgATP alone or SO:- alone (or a SO:- analog such as FSO; or SzOg-) was a hyperbolic function of ligand concen- tration for both the native and the SH-modified enzymes.

Chemical Modification of ATP Sulfurylase SH Groups 11771

(b) NEM-MODIFIED NATIVE ENZYME

0 20 40 60 80 IM) 120 0 20 40 60 80 100

INCUBATION TIME (Minutes) INCUBATION TIME (Minutes)

FIG. 3. Inactivation of ATP sulfurylase by phenylyloxal (1 mM) in the 0.05 M Na/HEPPS buffer, pH 8.0. a, native enzyme (0.55 p~ sites) in the absence and presence of APS (10 p ~ ) ; b, NEM- modified enzyme (0.75 p~ sites) in the absence and presence of APS (35 p ~ ) . The broken lines represent the two resolved linear compo- nents of the biphasic plot.

Positive cooperativity for MgATP protection was seen only for the SH-modified enzyme and then, only in the presence of a sulfate subsite ligand. Similarly, positively cooperative protection of the SH-modified enzyme by FSOT or S @ - was seen only in the presence of MgATP. Those results suggested that positive cooperativity required occupancy of two subsites, i.e. the sulfate subsite and either the pyrophosphoryl or ade- nylyl subsite. The present results showing that [35S]APS binding was cooperative in the absence of the reverse reaction co-substrate (MgPPi) or a PPI analog identify the sulfate and adenylyl subsites as the interacting regions.

Enzyme Inactivation by Phenylglyoxal-In the absence of protective ligands, enzyme inactivation by phenylglyoxal was monophasic for both the native and the SH-modified enzyme. However, the first order rate constant for inactivation of the SH-modified enzyme was about twice that of the native enzyme (2). Bound ligands protected the native and SH- modified ATP sulfurylase from inactivation by phenylglyoxal, but protection was not complete. Fig. 3a shows the semilog plot for inactivation of the native enzyme by 1 mM phenyl- glyoxal (50 mM Na/HEPPS buffer, pH 8.0) in the absence of a protective ligand and at saturating (10 p ~ ) APS. In the presence of APS, the plot is clearly biphasic and can be described by ln([E]t/[E]t,o) = 0.5 (e-k1t + where kl is nearly identical to the & for the monophasic inactivation of the unoccupied native enzyme (& kl = (1.71 * 0.03) x loT3 s-’). The rate constant for the slower phase, k2, was 0.37 x

s-’. At saturating MgATP (20 mM), inactivation was monophasic with a k of 0.43 x s-’ (data not shown). In other words, when all enzyme sites were occupied with APS, half of the sites appeared to be essentially unprotected while the other half was protected to about the same degree as provided by MgATP.

The SH-modified enzyme showed essentially the same bi- phasic inactivation curve (Fig. 3b) at near-saturating (35 or 50 PM) APS:3 kl = 1.79 X s-’; kz = 0.48 X s-’. In the absence of APS, & was 3.85 X s-’ (i.e. 2.2 times & for the native enzyme).

Isotope Trapping: Single Turnover of Bound p5S]APS- The curved Scatchard plot for [35S]APS binding (Fig. 2a) and

In the experiment reported in Fig. 3b, the initial [APSIfree was 6-fold greater than the concentration of “low APS affinity” sites (0.5[EIt) and remained about 10-fold greater than the estimated Kd of those sites (Kf. = 3.4 p ~ , see “Appendix”) throughout the inacti- vation period. The same biphasic curve was obtained at 50 p~ total APS. Thus, the nonlinear inactivation plot cannot be attributed to changing [APSIf,.

the biphasic inactivation plot at saturating APS (Fig. 3a) indicated that ATP sulfurylase contains two types of APS binding sites. Isotope trapping (8) was used to determine whether both types of sites are catalytically active. Fig. 4a shows the concentration dependence of [35S]APS trapping at saturating PPi and 5 mM excess M$+. When both substrates were saturating, 60 nM sites yielded 60 nM products indicating that (a) all six sites/hexameric enzyme are catalytically active and (b) APS does not dissociate a t a significant rate from E.APS.MgPPi. The shape of the Scatchard plot is quite similar to that for [35S]APS binding and the [APS]o.5 values obtained by the two methods are almost identical (50 & 4 nM). [35S]APS trapping by the SH-modified enzyme at satu- rating PPi (500 p ~ ) and 5 mM M e was positively cooperative (Fig. 4b) with an [APS]o.5 of 415 nM. Fig. 4c shows the Hill plots. The Scatchard plots of isotope trapping at saturating [35S]APS and different PPi concentrations (5 mM total M e ) displayed negative cooperativity for both forms of the enzyme. The [PPi]o.5 for the native enzyme was 5.4 & 0.4 p ~ , which is nearly identical to the K, for PPi in reverse reaction initial velocity measurements (4.0-6.5 pM; Refs. 6 and 9). [PPi]0,5 increased to 18.5 p~ upon SH modification (Fig. 4d).

“Inactivation” of ATP Sulfuryhe by SH-targeted Re- agents-Covalent modification of an enzyme that results in an increase in substrate K,,, values can be detected by activity measurements only if the assay for residual activity is per- formed at [SI < K,. If the initial concentration of modifying agent, is the integrated rate equation for activity loss is:

In ( A - Am) = -kt ( A , - A m )

(2)

where A = residual activity at any time, t; A. = initial activity; A, = final activity when all the enzyme molecules are modi- fied; k = the pseudo-first order rate constant for the modifi- cation process. If the substrate concentrations in the activity assay mixture are sufficiently low, A, will be indistinguishable from 0, and one can simply plot In A (or A on a log scale) versus time to obtain k ( k = 0.693/tIh where tH is the time required for 50% decrease in activity). Table 1 summarizes the results for six different SH-reactive reagents. In these experiments, residual activity was measured at 1 mM MgATP (about 0.3 [MgATP]o.b) and 1 mM SO:- (about 0.1 Under this assay condition, A , is nearly 0. Comparing anal- ogous reagents, we see that the equally “bulky” but uncharged DTNP was 17 times more effective than negatively-charged DTNB. Of the maleimides, the more bulky (but also more hydrophobic) N-phenylmaleimide was 10 times more effective than NEM. Positively charged thiolyte-MQ was 2.7 times more effective than neutral thiolyte-MB. These results sug- gest that SH-1 may be located in a hydrophobic pocket rather than on the enzyme’s surface. However, the small, but signif- icant advantage of positively-charged thiolyte-MQ over neu- tral thiolyte-MB suggests that other factors (neighboring negatively-charged groups?) may also affect the modification rate. The latter conclusion is supported by the effect of M$+: at pH 8.0, 5 mM M P doubles the apparent rate constant for enzyme modification by DTNB, but has no effect on modifi- cation by DTNP.

Effect of pH on SH-1 Modification-If only the unproton- ated form of the “regulatory” cysteinyl side chain reacts with SH-targeted reagents, the apparent first order rate constant for modification or inactivation will depend on the pH of the incubation medium. The reaction scheme is shown below. In

11 772 Chemical Modification of ATP Sulfurylase SH Groups 3.0 I

0.0 - 0.0 0.1 02 0.3

[AmlTmppcrd pM 0.12 I

0.5

0.1 0.00 0.10 020 0.30 0.40

TABLE I Rate constants for inactivation of ATP sulfuryhe

Inactivation was carried out at 30 "C in 0.05 M K phosphate, pH 7.0, containing 0.2 M KCl. The concentration of enzyme sites was 0.3 uM.

Reagent Concentration k,P k'"

PM S-1 M' x s-'

D T N B ~ 3 4.5 x 10-~ 150 D T N P d 3 8.0 X 2700 NEM 20 3.6 X 18 NPM' 20 3.6 X IOS3 180 Thiolyte-MQ' 300 2.6 X lo-* 8.7 Thiolyte-MB' 300 9.4 X 10" 3.18 Calculated as k.,,/[reagent]. DTNP and thiolyte-MB were dissolved in dimethyl sulfoxide and

the stock solution diluted with reaction buffer immediately before use. The volume of diluted reagent added to the enzyme solution was such that the final dimethyl sulfoxide concentration was 0.6% (v/v), which had no effect on catalytic activity.

The semilog inactivation plots were slightly curved (2). kepp was calculated as 0.693/t,.

dThe N-phenylmaleimide (NPM) was dissolved in ethanol and diluted with reaction buffer before use. The final ethanol concentra- tion was 0.04% (v/v) vhich had no effect on catalytic activity.

e Thiolyte-MQ was dissolved in water and used immediately.

effect, H+ acts as a protective ligand.

ES- %= E' (inactive) + H+

ESH Kall

The rate of inactivation is given by: -d[Elt -=

d t k[ES-]

or

FIG. 4. Isotope trapping of [35S] APS (Scatchard plots). a, native en- zyme (60 nM sites) for varied [35S]APS at saturating PPi (0.5 mM) and 5 mM excess M e . b, NEM-modified enzyme (0.3 pM sites) for varied [36S]APS at saturating PPi (0.5 mM) and 5 mM excess M e . e, Hill plots. d, native enzyme (0.4 p~ sites) for varied PPi at near-saturat- ing [35S]APS (0.8 pM) and 5 mM excess Mg2+ and for the NEM-modified enzyme (0.4 p M sites) for varied PPi at near- saturating [35S]APS (3 pM) and 5 mM excess M P .

'"'I T=30"C DTNp

10-41 6 7 8 9 IO

pH

FIG. 5. pH dependence of SH-1 modification by several dif- ferent SH-targeted reagents. Most of the modification reactions were carried out in 0.05 M KPi buffers containing 0.2 M KCl. Modi- fication with iodoacetamide (IAANH2) was carried out in 0.05 M Na/ CAPS buffer containing 0.05 M NaCl. NBDCI, 7-chloro-4-nitrobenz- 2-oxa-1,3-diazole.

A plot of log kapp uersus pH is horizontal at pH >> pK, and linear with a slope of unity in the range pH << pK,. The intersection of the two limiting lines occurs at pH = pK, (Ref. 7 ) . Fig. 5 shows the log kappuersus pH plots for the inactivation of ATP sulfurylase by a variety of SH-targeted reagents in 0.05 M KPi buffer containing 0.2 M KC1 or in 0.05 M Na/ CAPS containing 0.05 M NaCl. Essentially the same results were obtained in 0.05 M bis-Tris propane-C1 buffers, pH 6.0- 9.5, at constant C1- (data not shown).

Above pH 8, all the plots are linear with a slope of 1.0. There is no indication of an approach to a plateau. (pH values above 10.5 could not be tested because of enzyme instability.) Below pH 8, the plots decrease in slope and become horizontal between pH 7 and 6. The absence of a high plateau indicates that the pKa of SH-1 is >9. This is consistent with a location of SH-1 in a non-polar environment which stabilizes the uncharged ESH. The abnormal plateau between pH 6 and 7 suggests that protonation of SH-1 is affected by an interacting

Chemical Modification of ATP Sulfurylase SH Groups 11773

group whose protonated (ENH+) form stabilizes the ES-. Alternatively, a conformational change in the enzyme below pH 7 may block the access of SH-reactive reagents.

Effect of Temperature on SH-1 Modification-Increasing temperature at constant pH increased the rate constant for SH-1 modification or enzyme inactivation. AGS (free energy of activation) for the process at 30 "C was calculated from the absolute reaction rate theory equation ( 7 ) :

AGt = RT In - k'h ksT

where It' is the second order rate constant for a specific reagent, h is Planck's constant, kB is Boltzmann's constant, and T is the absolute temperature. AHt (enthalpy of activa- tion) was calculated from the slopes of log (It'lT) uersus 1/T plots

where T2 > Tl. ASf (entropy of activation) at 30 "C was calculated from:

For comparative purposes, the activation energies were also obtained for the modification of reduced glutathione (GSH) by DTNB.

Fig. 6 shows the activation energies for SH-1 modification by DTNB and NEM at pH 6, 7 , and 8 in 0.05 M potassium phosphate buffer containing 0.2 M KC1. The calculation of A H t did not take into account the possible effect of varied temperature on SH-1 ionization (i.e. AHi,) for the following reasons: (a) SH-1 does not display normal ionization proper- ties between pH 6 and 8. Thus, we could not assume that the mi,,,, would be that of a typical SH group. ( b ) If the ES- stabilizing group is a histidyl or amino group, increasing temperature would decrease the [ENH+]/[EN"] ratio and thereby suppress the increased ionization of SH-1. ( c ) In a separate series of experiments, the modification reactions were carried out in 0.05 M Tris-C1 buffer adjusted to pH 8.0 at 30 "C and not readjusted as the temperature was changed.

A OTNB MoDlFlCATlON

0 0 A NEM MODIFICATION

AG* 10 3

I 1 - 6 7 8

PH

tion of pH (0.06 M KPi buffers containing 0.2 M KCI). A H * was FIG. 6. Activation energies for SH-1 modification as a func-

determined from modification rate measurements made at 15,20, 25, 30, and 35 "C. The reaction of the enzyme (2 pM sites) with DTNB was monitored spectrophotometrically at 412 nm for full scale chart recorder deflection = 0.05). The DTNB concentration ranged from 40 (at 35 ' C ) to 400 pM (at 15 " C ) . Modification of the enzyme (0.2-0.5 pM sites) by NEM (20-30 pM) was determined from the loss of activity measured at 1 mM MgATP and 1 m M SO:-.

Tris-H+ has a positive A H i o n such that its pKa decreases by about 0.03 units/degree temperature increase. As a result, the pH of Tris buffers decreases as T i s increased, and vice versa. The ratio [RS-]/[RSH] is given by

Thus, if ApH/AT of the buffer is the same as the ApKa/AT of RSH, the fraction of the total RSH in the RS- form would remain constant as T is changed. A normal SH group has a AHi,, of 6500-7000 cal X mol" which corresponds to a ApK, X degree" of about -0.017. Thus Tris overcompensates some- what. Experimentally, the calculated A H S for SH-1 modifi- cation at "pH 8" was the same in the Tris buffer and K P i

buffers (the latter adjusted to pH 8.0 at each temperature). The most striking result of the above experiments is the

positive ASS for enzyme modification. A bimolecular reaction is expected to have a negative ASS because the formation of the transition state is accompanied by a decrease in the degrees of freedom of the two reactants. Thus, the AS* of -8 to -23 cal x mol" X degree" found for the modification of GSH by DTNB is consistent with the nature of the reaction. (The k values for 1 or 2 PM GSH were determined at 15, 20, and 25 "C and pH values of 6, 7 , and 8; the initial DTNB concentration was 20 or 40 PM.) On the other hand, the ASS of +2 to +78 cal x mol" x degree" for enzyme modification suggests that attainment of the transition state involves a substantial conformational change to a less highly ordered state. By way of comparison, ASS for thermal inactivation of ATP sulfurylase at 65 "C is +245 cal X mol" X degree" (9). The ASt for subunit dissociation of APS kinase at 42 "C is +214 cal X mol" X degree-' (4).

Further evidence for a protected location of SH-1 comes from studies on the ATP sulfurylase of the thermophile, Penicillium duponti (9). If SH-1 resided on the outer surface of the enzyme (i.e. solvent accessible), it would be easily modified regardless of the "tightness" of the rest of the en- zyme's structure. Yet, SH-1 of P. duponti ATP sulfurylase is virtually unreactive toward DTNB at 30 "C (a temperature at which the organism does not grow). At 50 "C (ToPC for the organism), SH-1 is just as reactive as the SH-1 of the P. chrysogenum enzyme. However, SH-1 of the enzyme from either source is considerably less reactive than the sulfhydryl group of GSH. For example, at pH 7.0 and 8.0 (0.05 M KPi containing 0.2 M KCl), the k values for GSH modification at 30 "C were 1,875 and 11,500 M-' X s-', respectively. The corresponding values for P. chrysogenum ATP sulfurylase were 135 +- 15 and 360 M-' X s-'.

Is SH-1 Modified in Viuo?-The cells were grown in syn- thetic medium (1) containing either a ''10~'' level (0.1 g X liter-') of L-cysteic acid or L-djenkolic acid or a "high" level (1 g X liter-') of L-methionine, L-cysteine, reduced glutathi- one, or Na2S04. ATP sulfurylase was partially purified from cell-free extracts by blue dextran column chromatography (1) without prior fractionation with ammonium sulfate. In agree- ment with earlier results (ll), the highest enzyme levels were obtained from cysteic acid or djenkolic acid-grown cells.4 Growth on high L-methionine as sole sulfur source repressed the enzyme to about 30% of the maximum level. In all cases, the partially purified enzyme exhibited hyperbolic u uersu [SO:-] and u uersus [MgATP] curves (2). Thus, there is no

Starting with 100 g wet weight of cells, the abbreviated purifica- tion procedure reproducibly yielded 0.30-0.35 molybdolysis units/g of cells after blue dextran chromatography of extracts from low cysteic acid-grown cells. However, yields double the above are obtained by the standard procedure (1) starting with 600-900 g of cells.

11774 Chemical Modification of ATP Sulfurylase SH Groups

evidence for SH-1 modification in vivo under conditions where the need for sulfate activation is diminished. Except for the partial repression, there is no well documented evidence for end product control of the fungal enzyme. The feedback inhibition by sulfide observed earlier (11) is worth reinvesti- gation because it may have resulted from in vitro SH-1 mod- ification by contaminating polysulfides in the Na2S solution.

Amino Acid Sequences in the Region of SH-I-The native enzyme was modified at SH-1 with monobromobimane (thio- lyte-MB). The excess reagent was removed by dialysis, and then the remaining SH groups were modified with NEM in the presence of 0.04% SDS. Overlapping tryptic, chymotryp- tic, and V8 protease peptides were separated by high perform- ance liquid chromatography and, in each case, the single fluorescently labeled (SH-1 containing) peptide was collected and purified on Cla and phenyl reverse phase columns. The peptides were sequenced yielding the following primary S t N C - ture: (Lys or Arg)-Asp-Ala-Val-Ser-Gln-Ala-Gly-Ser-Phe- Phe-Leu-Val-His-Val-Ala-Thr-Pro-Leu-Glu-His-Cys*-Glu- Gln-Ser-Asp-Lys-Arg-Gly-Ile-Tyr. Because (a) we have no results suggesting a regulatory role for SH-1 covalent modi- fication (and hence, for a distinct regulatory domain), and ( b ) MgATP protects against SH-1 modification (2), it is possible that the above peptide sequence is part of, or is in close proximity to, the active site. Precedents include tyrosyl-tRNA synthetase (12) (which, like ATP sulfurylase, is also an ade- nylyl transferase) and phosphoribulokinase (13). In both of these enzymes a “non-essential” Cys residue participates in MgATP binding.

DISCUSSION

Direct equilibrium binding and isotope trapping measure- ments have verified that modification of a single cysteinyl residue (SH-1) per subunit induces positive cooperativity for MgATP binding to P. chrysogenum ATP sulfurylase. Coop- erativity for forward reaction substrates is observed only with respect to the formation of the ternary E . MgATP. SO2- com- plex. On the other hand, APS, a reaction product, binds cooperatively to the SH-modified enzyme in the absence of a co-product or co-product analog. These results indicate that cooperative binding to vacant sites results from occupancy of the adenylyl and sulfate subsites.

Although SH-1 modification has the same effect as a clas- sical K-type allosteric inhibitor, no evidence was found for in uiuo modification even under conditions where inactivation of ATP sulfurylase would have been physiologically prudent. Of course, it could be argued that in vivo SH-1 modification was reversed during the partial purification of the enzyme. But for the present, it seems more likely that in vitro modi- fication either (a) uncovers or exaggerates a facet of the catalytic mechanism that is kinetically invisible in the native enzyme, or ( b ) induces an enzyme conformation that is formed in vivo by the action of a reversibly bound allosteric effe~tor .~

Fig. 7 presents a model that attempts to reconcile the cumulative data, including the ubiquitous factor of 2 and the ability of the enzyme to exhibit both negative and positive cooperativity. The main feature of the model is that a subunit (or site) can exist in either of two conformations which differ with respect to APS affinity (“Appendix”) and accessibility of an essential argininyl group. The following scenario is

In the presence of PAPS (the product of APS kinase, the next enzyme in the sulfate activation sequence), the u uersus [MgATP] and u uersus [SO!-] or [MOO:-] curves are sigmoidal. The site of PAPS binding (active site?, regulatory site?) and the relationship between SH-1 modification and PAPS binding are under investiga- tion.

_____) SH-1 Modification

ARQ-2 ARQ-2 (Conformational Change exposes A~Q-2)

ii + 1st APS (low affinity)

ARQ-2

ARQ-2

(H APS) L H (L APS)

+ 2nd APS (low affinity) (high afflnity)

:. Negative . . Positive Cooperativity Cooperativiry

ARQ-2

ARQ-2

(H .APS) (L-APS)

FIG. 7. Proposed model for the conformational changes in the “minimal dimer” of P. chryeogenum ATP sulfurylase upon APS binding and/or SH-1 modification. HH, native, unoccupied enzyme containing 2 equivalent high APS affinity sites. LL, SH- modified enzyme containing 2 equivalent low APS affinity sites. HL, hybrid containing one high and one low APS affinity sites. HL is formed when APS binds to an H site of the native enzyme or to an L site of the SH-modified enzyme. H. APS . L. APS formed from the SH-modified enzyme is not completely identical to the saturated species formed from the native enzyme as evidenced by (a ) the approximate 15% decrease in kat (2) and ( b ) the 3.5-fold increase in [MgPPi]o.6 of the modified enzyme (Fig. 4c). Native and SH-modified L sites differ in APS affinities. However, native and SH-modified H sites appear to have nearly identical affinities for APS (see “Appen- dix’’). Arg-1 and Arg-2 are functionally essential argininyl groups, one of which (Arg-1) is protected by APS against modification by phenylglyoxal. Arg-2 becomes accessible to phenylglyoxal only after an H+L transition.

proposed (a) the hexameric enzyme contains 2 essential argininyl residues/subunit, one of which (Arg-1) is much more reactive toward phenylglyoxal, but is also highly protected by bound APS. Modification of either Arg group will destroy catalytic activity. (b) Cooperative interactions are restricted to subunit pairs, i.e. the enzyme behaves as a trimer composed of three independent &mew6 ( c ) APS binding to a site on the free native dimer (HH) induces a conformational change in the vacant interacting partner that results in a decreased affinity for APS. Thus, the complete APS binding curve is characteristic of negative cooperativity (Figs. 2a and 4a). ( d ) Arg-2 of an APS-occupied high affinity site (Fig. I, circle) remains in a low phenylglyoxal-reactive state, but Arg-2 of an occupied low affinity site (Fig. 7, square) becomes just as accessible as unprotected Arg-1. Thus, the phenylglyoxal- dependent semilog inactivation plot is biphasic (Fig. 30). The biphasic plot implies that the fully occupied enzyme does not

This is a sufficient, but not absolutely necessary condition. Linear Hill plots with the observed nH values of 0.7 and 1.9 can be obtained with full hexamer models that permit sequential interactions between all sites. The minimal dimer model was chosen because it accommo- dates the inactivation protection data as well as the binding and kinetics results.

Chemical Modification of ATP Sulfurylase SH Groups 11775

rapidly equilibrate between its 2 equivalent conformations (HL and LH). ( e ) Complete SH-1 modification forces the unoccupied enzyme into a new conformation (LL) in which both sites of the dimer have an equally reduced affinity for APS. In this state, Arg-2 of each site is just as accessible to phenylglyoxal as Arg-1. Thus, kaPP for inactivation of the SH- modified enzyme is about double that for the native enzyme (Fig. 3b). ( f ) APS binding to an L state site on the SH- modified enzyme triggers a transition to the hybrid HL state. Thus APS binding to the SH-modified enzyme displays pos- itive cooperativity (Figs. 2b and 4b) and the fully APS- occupied SH-modified enzyme displays the same biphasic phenylglyoxal-dependent inactivation plot as the fully occu- pied native enzyme (Fig. 3, a and b) . The MgPPi-dependent [3SS]APS trapping curves of the native (Fig. 4c) and SH- modified enzyme (Fig. 4d) at saturating APS are also identical in character (indicative of two classes of sites on (H.APS)

The initial velocity kinetics of the native enzyme appears to be normal hyperbolic in both directions (5, 6) so the relevance of the negative cooperativity of APS binding to the catalytic reaction is unclear (especially because MgATP f FSO, displays normal hyperbolic binding to the native en- zyme). One possible explanation is that under steady state conditions, only half of the sites produce APS at any time. APS release from an H site may be promoted by MgATP or SO:- binding to an adjacent vacant L site. In other words, the enzyme may cycle between (HL)3 and (LH)3 and normally never contain more than three APS molecules/hexamer. This alternating site or “flip-flop” mechanism will produce normal hyperbolic kinetics, but the product binding curves will be characteristic of negative cooperativity (lo), as we observe for APS. Saturation of all the sites then is an abnormal situation observable only under equilibrium, nonturnover conditions. Conversion of all six sites of ATP sulfurylase to the L state is also abnormal, occurring only when the unoccupied enzyme is chemically modified at SH-1.

An alternating site mechanism of the type proposed above may be Nature’s solution to the problem of maximizing hff for APS release in the face of a very low Kes of the reaction in the physiological direction (15). In this regard, it is note- worthy that the V,,, for APS formation by the P. chryso- genum enzyme is about 35% of the V,,, for “uncoupled molybdolysis (i.e molybdate-dependent enzymatic hydrolysis of ATP to AMP and PPJ (9). The APS synthesis/molyb- dolysis ratios for the purified spinach leaf7 and rat liver (16) ATP sulfurylases are about 0.06 and 0.1, respectively. Neither of these enzymes possess a cysteinyl residue whose modifica- tion causes cooperative binding.

(L.APS)).

APPENDIX

Ligand binding equations for a “minimal dimer” model of P. chrysogenum ATP sulfurylase are as shown.

Natiue enzyme-The first molecule of APS(S) binds with an intrinsic dissociation constant, KH, to a site on free HH causing the dimer unit to assume an HS .L conformation. The next molecule of S binds to the vacant L site with an intrinsic dissociation constant KL (or aKH where a > 1). The S binding

J. Mazer and I. H. Segel, unpublished results.

equation is given by Equation 10: [s lL[s12

(10)

where: Ys, fractional saturation of sites; [ SIb, total concentra- tion of bound APS; [E], total enzyme (dimer) concentration; [SI, equilibrium concentration of free APS; and u, interaction factor (factor by which the intrinsic dissociation constant of the second site is altered when APS occupies the first site).

Simulations showed that an a of 5 produces a Hill plot that is close to linear between 10 and 90% saturation and has an average slope (nH) of 0.73. [APS]o.5 (50 +. 4 nM; Figs. 2a and 4a) occurs at KH&. Thus, KH = 22 nM; KL = aKH = 112 nM. KH is in the same range as the value of K k (40 nM) determined from initial velocity kinetics (6).

SH-modified Enzyme-The analogous binding equation is:

“I- [Sl [SI2

I + - + - K f . bKL2

where the intrinsic dissociation constants are K t (not neces- sarily equal to KL of the native enzyme) and Kir = bKi (not necessarily equal to KH of the native enzyme). An interaction factor, b, of 0.009 produces a near-linear Hill plot with an nH of 1.9. [APS]o.5 (320 f 90 nM; Figs. 2b and 4b) occurs at K i A . Thus, Ki = 3373 nM, Kir = bKt = 30.4 nM. In other words, SH modification increases the dissociation constant for the first APS molecule by a factor of KL/KH = 153. The relative APS dissociation constants of the SH-modified and native H sites, K ~ / K H is 1.38. Thus, occupancy of the first APS site in the SH-modified dimer returns the second site to near its normal high affinity state.

REFERENCES 1. SegeI, I. H., Renosto, F., and Seubert, P. A. (1987) Methods

2. Renosto, F., Martin, R. L., and Segel, 1. H. (1987) J. Biol. Chem.

3. Renosto, F., Seubert, P. A., Knudson, P., and Segel, I. H. (1985)

4. Renosto, F., Seubert, P. A., Knudson, P., and Segel, I. H. (1985)

5. Seubert, P. A., Hoang, L., Renosto, F., and Segel, I. H. (1983)

6. Seubert, P. A., Renosto, F., Knudson, P., and Segel, I. H. (1985)

7. Segel, I. H. (1975) Enzyme Kinetics, p. 369, Wiley-Interscience,

Enzymol. 143,334-349

262,16279-16288

J. Bioi. Chem. 260 , 11903-11913

J. Biol. Chem. 260 , 1535-1544

Arch. Biochem. Biophys. 226,679-691

Arch. Bwchem. Biophys. 240,509-523

New York

J. Bid. Chem. 249,5163-5168 8. Rose, I. A., O’Connell, E. L., Litwin, S., and Bar-Tana, J. (1974)

9. Renosto, F.. Schultz. T., Re, E.. Mazer. J.. Chandler. C. J.. Barron. A., and Segel, I. H . (1985) J: Bacteriol.’ 164 , 674-683 ’

(1971) Eur. J. Bwchem. 20, 124-139

2446

(1983) Biochemistry 22,3581-3586

10. Lazdunski, M., Petitclerc, C., Chappelet, D., and Lazdunski, C.

11. Tweedie, J. W., and Segel, I. H. (1971) J. Bwl. Chem. 246,2438-

12. Wilkinson, A. J., Fersht, A. R., Blow, D. M., and Winter, G.

13. Porter, M. A,, and Hartman, F. C. (1988) J. Biol. Chem. 2 6 3 ,

14. Renosto, F., Seubert, P. A., and Segel, I. H. (1984) J. Biol. Chem.

15. Schmidt, N. D., Peschon, J. J., Chandler, C. J., Renosto, F., and Segel, I. H. (1982) Sulfate Metabolism and Sulfate Conjugation, pp. 5-12, Taylor & Francis Ltd., London

16. Yu, M., Martin, R. L., Jain, S., Chen, L. J., and Segel, I. H. (1989) Arch. Biochem. Biophys. 269,156-187

14846-14849

259,2113-2123