Differential Affinity Cross-linking of Phosphorylase Kinase Conformers by the Geometric Isomers of Phenylenedimaleimide*

Phosphorylase b kinase (PbK) from skeletal muscle is a highly regulated oligomer consisting of four copies of four distinct subunits (αβγ)δ4. The γ subunit is catalytic, and the remaining subunits are regulatory. To characterize effector-induced changes in the quaternary structure of the enzyme, we utilized the ortho-, meta, andpara-isomers of phenylenedimaleimide (PDM), which in addition to having different geometries, also vary 2.5-fold in their cross-linking spans. Even at concentrations equivalent to the αβγδ protomers of PbK, all three isomers caused specific, rapid, and extensive cross-linking of the holoenzyme to form primarily αβ dimers, plus smaller amounts of βγγ and αγγ trimers. The formation of these three conjugates was nearly totally inhibited by a 10-fold molar excess over PDM of N-(o- andp-tolyl)succinimide, which are chemically inert structural analogs of PDM. This inhibition suggests that PbK has binding sites for PDM and that PDM acts as an affinity cross-linker in binding to these sites prior to forming cross-linked conjugates. The largest effect on cross-linking in progressing from o- to p-PDM was on the αγγ trimer, which is preferentially formed by thep-isomer. Activation of the enzyme by either phosphorylation or the allosteric activators ADP and GDP resulted in large increases in the amount of αγγ formed, small increases in βγγ, and little change in αβ. When cross-linked in the presence of the reversibly activating nucleoside diphosphates, PbK remained activated after their removal, indicating that cross-linking had locked it in the active conformation. Our results provide direct evidence for perturbations in the interactions of the catalytic γ subunit with the regulatory α and β subunits upon activation of PbK.

units clearly control the activity of the catalytic ␥ subunit, little is known concerning the mechanisms through which they exert this control, including the extent to which their regulatory influence on ␥ is direct versus indirect; this is especially the case for the larger regulatory subunits, ␣ and ␤. Phosphorylation (3) or proteolysis (4) of ␣ causes increased activity of ␥ within the (␣␤␥␦) 4 holoenzyme, but evidence for a direct ␣-␥ interaction that is altered by this activation has not been observed. Likewise, multiple means of activating the holoenzyme cause common conformational changes in the ␤ subunit (5-7); but again, no evidence for alteration of a direct ␤-␥ interaction has been observed. In two previous studies, cross-linking was used successfully to detect changes in the cross-linking patterns of both the ␣ (8) and ␤ (5) subunits upon activation of the holoenzyme; but with the cross-linkers used, the observed changes involved only a second ␣ or ␤ subunit to form homodimers and did not involve the catalytic ␥ subunit. Even though cross-linking has been shown to be a potentially promising approach to probing structural changes associated with activation of the hexadecameric PbK holoenzyme, it has not been exploited widely because of the intrinsic difficulty in identifying unambiguously the subunit composition of cross-linked complexes (techniques used previously with PbK to analyze such complexes have been limited to approximate mass, relative susceptibility to proteolysis, and specific radioactivity of individual subunits (5,9,10)). The recent availability of a panel of subunit-specific monoclonal antibodies against PbK's four different subunits (11)(12)(13) has greatly aided analysis of the subunit composition of cross-linked complexes (8), which in turn, has encouraged us to evaluate additional cross-linkers, seeking to discover one capable of detecting changes in the interaction of the catalytic and regulatory subunits of PbK upon its activation.
The geometric isomers (ortho, meta, and para) of the homobifunctional cross-linker phenylenedimaleimide (PDM) were evaluated with the rationale that any effector-induced perturbations in their cross-linking of PbK could be interpreted solely on the basis of geometry and span (4.8 Å for o-PDM to 12 Å for p-PDM (14)). In addition to forming an ␣␤ dimer, each isomer cross-linked two ␥ subunits with an ␣ and with a ␤ subunit; moreover, this cross-linking was differential, in two senses. First, ␣␥␥ and ␤␥␥ were preferentially formed with activated conformers of the kinase; and second, in progressing from o-, to m-, to p-PDM, there was a reversal in the relative amounts of ␣␥␥ and ␤␥␥ conjugates formed with the activated conformers (from predominantly ␤␥␥ to predominantly ␣␥␥). The three isomers also behaved as affinity cross-linkers, i.e. binding to PbK prior to cross-linking. Based on relative efficiencies of cross-linking, two major sets of binding sites for PDM were observed, with cross-linking at the lower efficacy sites reflecting the extent of the enzyme activation.

EXPERIMENTAL PROCEDURES
Enzymes and Proteins-PbK was isolated from fast-twitch skeletal muscle of New Zealand White rabbits (15), dialyzed against a solution of Hepes buffer (50 mM, pH 6.8), sucrose (10%), EDTA (0.2 mM), and either used immediately or stored frozen at Ϫ60°C. All experiments described in this study were repeated a minimum of three times using three different PbK preparations. When autophosphorylated PbK was required for cross-linking studies, the phosphorylation was carried out at pH 8.2 in Hepes buffer for 1 min using the methodology of King et al. (16). The extent of phosphate incorporation with different kinase preparations ranged from 1.2 to 2.1 mol/␣ subunit and 0.8 -1.0/␤ subunit. The enzyme was also phosphorylated by the catalytic subunit of protein kinase A, as described previously (12), to the extent of 1 mol of phosphate each per ␣ and ␤ subunit. Prior to cross-linking, PbK phosphorylated through either mechanism was purified by gel filtration over a Sepharose 6B column (1.5 ϫ 112 cm) developed with Hepes buffer (50 mM, pH 6.8), 0.2 mM EDTA, and 10% sucrose. Fractions that eluted at the position of native holoenzyme were collected, buffer was exchanged, and the enzyme was concentrated to 4.5 mg/ml by ultrafiltration using a Centricon-30 concentrator (Amicon). Nonphosphorylated enzyme used as the control to determine the effects of phosphorylation was subjected to the same incubation, gel filtration, and concentration protocols. Phosphorylase b was isolated from rabbit skeletal muscle (17), and residual AMP was adsorbed with activated charcoal (Sigma, C-4386). The concentrations of PbK and phosphorylase b were determined spectrophotometrically using their respective absorbance indices (18,19).
Bovine serum albumin (A-9647) was from Sigma, and the catalytic subunit of cAMP-dependent protein kinase was from Promega. Melittin was from Sigma, with its concentration determined as described previously (20). Monoclonal antibodies (mAbs) against the ␣, ␤, and ␥ subunits of phosphorylase kinase were generated in mice against the holoenzyme as antigen (12,13). The anti-calmodulin mAb was generated in mice against a peptide antigen corresponding to residues 107-148 in the COOH-terminal region of calmodulin (11). Detection conjugates for immunoblots were from Southern Biotechnology.
Quantification of Cross-linkers-The concentrations of the PDM cross-linkers (Aldrich) were determined from their absorbance in dry acetone at 332 nm, the absorption maximum characteristic of the maleimido functionality. The extinction coefficient at this wavelength was determined for each isomer through its alkylation of a known excess of the free thiolate (TNB Ϫ ) of DTNB, with this and all related reactions carried out in capped quartz cuvettes. The TNB Ϫ was generated by thiol exchange of DTNB (441 M) and dithiothreitol (59 M) in acetone (21). The release of TNB Ϫ was monitored at 412 nm and reached a plateau after 12 min. Solutions of known TNB Ϫ concentrations (⑀ TNB ϭ 14,150 M Ϫ1 cm Ϫ1 (22)) were then alkylated at room temperature with limiting amounts of each PDM isomer. The absorbance of TNB Ϫ stopped decreasing after 1 h, and final measurements were made after 3 h of alkylation. The concentration of PDM in the reaction was taken to be equal to half the final decrease in the concentration of TNB Ϫ . The ⑀ 332 values for o-, m-, and p-PDM determined in this manner were 0.42, 0.59, and 0.66 M Ϫ1 cm Ϫ1 , respectively. Concentrations of PDM solutions determined spectrophotometrically using these extinction coefficients agreed well with concentrations based on triplicate dry weight measurements and the manufacturer's stated purity for the cross-linkers.
Cross-linking-Cross-linking was carried out at 30°C at the indicated time intervals with o-, m-, or p-PDM. Final concentrations of protein and reagents in the standard reaction were: PbK, 1.73 M ␣␤␥␦ protomer, which is the concentration used in generating Figs. 1-5; Hepes, 50 mM, pH 8.2; EDTA, 1.0 mM; and cross-linker, 17.3 M, unless otherwise indicated. Prior to the cross-linking of PbK in the presence or absence of effectors, the enzyme was incubated for 2 min at 0°C with the indicated concentrations of GDP, ADP, heparin, and melittin. Cross-linking was initiated by the addition of PDM dissolved in dry acetone, with the final amount of acetone in the cross-linking reactions never exceeding 1% (v/v); in control experiments, this concentration of acetone was found to have no influence on enzymatic activity. Crosslinking was quenched by dilution of an aliquot of the reaction mixture into an equal volume of SDS buffer (0.125 M Tris, pH 6.8, 20% glycerol, 5% ␤-mercaptoethanol, 4% SDS) followed by brief mixing. After heating at 80°C for 10 min, the samples (14 g/lane) were run on SDS-polyacrylamide gradient gels (4 -15%) (23) and stained with Coomassie Blue. All gels were destained in 40% methanol, 10% acetic acid (2 h) and 7% acetic acid, 4% methanol (15 h). All conclusions regarding the relative amounts of cross-linked species formed were based on the integrated optical density of the protein bands determined using a BioImage whole band analyzer.
The N-(tolyl)succinimides used as inhibitors of cross-linking were obtained from Aldrich and were used without further purification.
Activity Assays-To measure the effects of cross-linking on the phosphorylase conversion activity of PbK, cross-linking (at pH 8.2) was first arrested by an 80-fold dilution in cold Hepes buffer (50 mM, pH 6.8) containing 3.0 mM dithiothreitol. Control experiments showed that inclusion of this concentration of dithiothreitol in the dilution buffer did not influence the activity of PbK in subsequent activity assays. An aliquot of the diluent was then measured for activity at pH 6.8 by following the incorporation for 5 min at 30°C of 32 P from [␥-32 P]ATP into phosphorylase b using phosphocellulose strips (24). Final concentrations in the assay mixture were: PbK, 0.7 g/ml; buffer (50 mM Tris, 50 mM ␤-glycerophosphate, pH 6.8); phosphorylase b, 6.0 mg/ml; EGTA, 0.1 mM; CaCl 2 , 0.2 mM; ␤-mercaptoethanol, 13 mM; [␥-32 P]ATP (NEN Life Science Products), 1.5 mM, 0.17 Ci/mol; Mg(CH 3 CO 2 ) 2 , 10 mM; and sucrose, 2-3%. linking was intramolecular, i.e. within hexadecamers, as opposed to between. Based on their mobilities in SDS-polyacrylamide gels, three general sets of conjugates (a high, intermediate, and low mass set) were formed by cross-linking the enzyme with each PDM isomer. The formation of these sets was dependent upon the time of cross-linking and concentration of cross-linker used, as is discussed in the following sections.

Time-and Concentration-dependent Cross-linking of Nonac
Formation of High Mass PDM Conjugates-When nonactivated enzyme was incubated either with very small amounts of any of the three PDM isomers (1 PDM:2 ␣␤␥␦) or with an excess of the cross-linkers (10 PDM:1 ␣␤␥␦) for very short times (0.5 min) (Fig. 1), the predominant cross-linked conjugate formed was a doublet with an average mass of 258 kDa, which corresponds to an ␣␤ dimer by mass (mass theor ϭ 264 kDa) and by cross-reactivity with subunit-specific mAbs (Fig. 2). Two additional high mass conjugates with apparent molecular masses corresponding to a ␤␤ dimer (mass theor ϭ 250 kDa; Ϫ3.0% error) and an ␣␤␥ trimer (mass theor ϭ 308 kDa; Ϫ1.7% error) were also observed in trace and minor amounts, respectively, in Western blots of the cross-linked enzyme (Fig. 2). Because the mass of the latter fell outside of the range of our mass standards (29 -264 kDa), it was not considered further in this study.
When cross-linking was carried out at pH 6.8, but under otherwise standard conditions (2 min at 30°C with a 10-fold molar excesss of cross-linker over ␣␤␥␦ protomer), the only significant product was the pair of ␣␤ dimers (data not shown). Because the reaction of sulfhydryls with maleimides is approximately 1,000-fold faster than with amines at pH values below 7.0 (25), cysteine residues on ␣ and ␤ may be the primary nucleophiles for PDM that result in ␣␤ formation. The presence of two ␣␤ dimers with slightly different mobilities on SDSpolyacrylamide gels presumably arises either from the crosslinking of different regions on ␣ and ␤ or from intrasubunit cross-linking of one or both of the individual subunits (26).
Either possibility indicates more than one region of cross-linking by PDM on the ␣ and/or ␤ subunits. Similarly, the crosslinking of ␤ subunits to form ␤␤ dimers with different migrations has been observed previously when activated forms of the kinase are treated with 1,5-difluoro-2,4-dinitrobenzene (5), indicating that more than one region of the ␤ subunit is also subject to chemical modification by this bifunctional aryl reagent.
Based on the densities of the bands in Fig. 1, greater than 90% of the ␣ and ␤ subunits lost formed ␣␤ dimers when the nonactivated holoenzyme was treated with the three PDM isomers under those conditions (low concentrations of crosslinkers or short incubation times). Not only was ␣␤ formation relatively specific, it was also an efficient process, given that 15% of the ␣ and ␤ subunits were cross-linked during a short period of cross-linking (2 min) at the low ratio of 1 PDM:2␣␤␥␦ (Fig. 1A). This high yield of cross-linking under these conditions indicates minimal competition from monosubstitution reactions or hydrolysis of the active maleimide, a significant reaction at pH 8.2 (27), our standard pH of cross-linking. The specificity, rapidity, and extent of cross-linking are consistent with the presence of binding sites for PDM on the ␣ and/or ␤ subunits, which have significant sequence similarity (28).
Formation of PDM Conjugates of Intermediate Mass-Increasing the concentration of PDM or the time of cross-linking resulted in increased formation of two additional conjugates with apparent masses of 225 and 205 kDa, corresponding to the masses of ␣␥␥ (mass theor ϭ 228 kDa; Ϫ1.3% error) and ␤␥␥ trimers (mass theor ϭ 215 kDa; Ϫ4.7% error), respectively (Fig.  1). The subunit composition of these new heteromers was verified further by cross-reactivity with subunit-specific mAbs (Fig. 2). As was the case with the ␣␤ dimer, the ␤␥␥ trimer was also present as a doublet. In contrast, however, to the ␣␤ dimer, the rate of formation of the trimers, especially of ␣␥␥, varied with the cross-linker used: the p-isomer caused the greatest formation of ␣␥␥, although the amount of it formed using this isomer and nonactivated enzyme was still less than that of ␤␥␥ (Fig. 1). Also, the extent of ␣␥␥ and ␤␥␥ formation increased over a 10-min period at excess PDM concentrations (10 PDM:1 ␣␤␥␦) (Fig. 1B), whereas the amount of discrete ␣␤ dimer reached a maximum within 0.5 min and then slowly decreased upon further cross-linking. Because only ␣␤ dimers are observed when the enzyme is cross-linked below pH 7.5 (data not shown), the formation at higher pH values of ␣␥␥ and ␤␥␥ trimers presumably results either from the cross-linking of different types of side chains or from a conformational change induced by alkaline pH, which is known to stimulate the activity of phosphorylase kinase greatly and to induce a conformational change (13).
Formation of a Low Mass PDM Conjugate-A minor crosslinked species corresponding to the known mass of an ␣␦ dimer (mass theor ϭ 155 kDa, 0.6% error) was observed under conditions identical to those under which ␣␥␥ and ␤␥␥ trimers formed (Fig. 1). As was observed for ␣␥␥, the rate of formation of this putative ␣␦ dimer increased in order of cross-linking by o-, m-, and p-PDM. In blots of enzyme cross-linked by either oor p-PDM, the conjugate interacted with the anti-␣ mAb but not with the anti-calmodulin mAb (the ␦ subunit is endogenous calmodulin) (Fig. 2, ࡗ). Although the anti-calmodulin mAb, which targets the COOH-terminal region of calmodulin (11), has been used successfully with other cross-linkers of PbK to detect trace amounts of calmodulin-containing conjugates (29), it is possible that the calmodulin epitope in this particular conjugate is masked as a result of the cross-linking by PDM. Analogs N-(o-and p-Tolyl)succinimide-The rapid and rela-  (16,29)), which is a commonly observed degradation product resulting from storage of the enzyme (18), was observed in the anti-␣ blot. The position of the putative ␣␦ dimer is indicated in that blot by ࡗ. Lane BMW contains the following broad range biotinylated molecular weight markers (Bio-Rad) in the indicated order of descending mass: myosin, ␤galactosidase, phosphorylase b, bovine serum albumin, ovalbumin, carbonic anhydrase, soybean trypsin inhibitor, lysozyme, and aprotinin. tively specific cross-linking with concentrations of PDM below PbK suggested that these reagents might be functioning as affinity cross-linkers, i.e. that they bind to the enzyme prior to covalently cross-linking its subunits. To test this possibility, we asked whether nonfunctional structural analogs of PDM could competitively inhibit cross-linking. The first PDM analogs tested were their corresponding disuccinimides, produced by catalytic hydrogenation of the three PDM isomers; however, none of these disuccinimides was sufficiently soluble under our standard cross-linking conditions to evaluate as an inhibitor. We turned, therefore, to the more soluble monosuccinimide structural analogs o-NTS and p-NTS, both of which were found to be potent inhibitors of cross-linking by o-and p-PDM (Fig. 3). Based on the residual amounts of the ␣, ␤, ␥, and ␦ subunits that were protected from cross-linking, a 10-fold excess of o-NTS (the more potent inhibitor) over PDM inhibited crosslinking by o-and p-PDM (10 PDM:1 ␣␤␥␦) by 95 and 91%, respectively; under identical conditions p-NTS inhibited crosslinking by the respective PDM isomers by 86 and 87%. The potent inhibition of o-and p-PDM cross-linking by their nonreactive succinimide structural analogs indicates the presence of binding sites on PbK for the PDM isomers. Furthermore, the nearly identical inhibition of both PDM cross-linkers by either monosuccinimide isomer suggests that these four compounds bind to a common site or sites.

Activators of PbK Perturb Its Cross-linking Pattern-Be-
cause activation of PbK is associated with conformational changes, some of which can be detected by cross-linkers (5,7,8,29), we wished to determine whether PDM could also distinguish the activated conformation(s) of the holoenzyme. The allosteric activators ADP and GDP caused a large increase over controls in the formation of ␣␥␥, especially by m-and p-PDM, and a small increase in ␤␥␥ (Fig. 4A). Likewise, activation by autophosphorylation also caused a large increase in the formation of ␣␥␥ and a small increase in ␤␥␥ (Fig. 4B); activation via phosphorylation by the catalytic subunit of protein kinase A caused similar effects (data not shown). The isomer selectivity for trimer formation with these activated conformers of PbK induced by either nucleoside diphosphates or phosphorylation changed from predominantly ␤␥␥ with o-PDM, to approximately equivalent amounts of ␣␥␥ and ␤␥␥ with m-PDM, to predominantly ␣␥␥ with p-PDM. Thus, the relative selectivity for increased ␣␥␥ formation in progressing from o-to p-PDM is the same as was described previously for the nonactivated enzyme (Fig. 1); of course, it is possible that the basal formation of these trimers with the nonactivated enzyme may simply reflect the fact that cross-linking is carried out at the stimulatory pH of 8.2. Unlike the very large increase in ␣␥␥ and the small increase in ␤␥␥, there were only modest increases in the ␣␤ and putative ␣␦ dimers upon activation.
In contrast to the significant amounts of ␣␥␥ and ␤␥␥ conjugates induced by nucleoside diphosphates and by phosphorylation, only trace amounts of these trimers were formed when the enzyme was cross-linked in the presence of the activator heparin. Instead, heparin protected the ␥ subunit while promoting extensive cross-linking of the ␤ subunit (Fig. 4A). These results suggest that heparin, which also promotes dissociation of the ␦ subunit (30), activates the enzyme through a different mechanism than ADP, GDP, or phosphorylation. Melittin, a model calmodulin-binding peptide that is an inhibitor of PbK (31), did not enhance, with respect to control, the formation of either ␣␥␥ or ␤␥␥ (Fig. 4A), which is consistent with the notion that formation of these trimers is character- istic of activated conformations or at least those activated conformations induced by nucleoside diphosphates or phosphorylation.
Trapping Activated Conformation of PbK by Cross-linking-Because activation of PbK influenced its cross-linking pattern, resulting in increased formation of ␣␥␥ and ␤␥␥ trimers, we were curious if the cross-linked activated enzyme would remain activated when subsequently assayed under nonactivating conditions, i.e. whether cross-linking could lock it in the activated conformation. Because m-PDM formed approximately equal amounts of ␣␥␥ and ␤␥␥, it was the isomer chosen for these experiments. Cross-linking with m-PDM at the stimulatory pH of 8.2, but in the absence of any specific allosteric activators, caused a 2.5-fold activation in subsequent assays at the control pH of 6.8 (Fig. 5); this activation was not observed when the cross-linking was performed at pH 6.8, which resulted in ␣␤ dimers being the only significant conjugate formed (data not shown). When the allosteric activator ADP was included during cross-linking at pH 8.2, the activation in subsequent assays was increased an additional 60% to 4-fold over control. The additional activation was not the result of carryover ADP in the activity assay because the dilution step between cross-linking and assay was sufficient to dilute the activator to well below its K a value, as shown by the lack of activation by ADP in the non-cross-linked control (Fig. 5). Cross-linking in the presence of GDP, the other known purine nucleoside diphosphate activator, caused an activation of 3.25-fold. When both the crosslinking and activity assays were performed at pH 8.2, this irreversible activation was not detected, even when effectors were included in the cross-linking reaction; in fact, cross-linking resulted in a slight inhibition of the activity measured at this pH value (data not shown).
As was observed previously with perturbation of the crosslinking pattern (Fig. 4A), heparin and melittin also acted differently than the other effectors in their influence on the crosslinking-dependent activation. In the case of heparin, activation was not completely reversed by dilution, even though its assay concentration was well below its K a (32), and cross-linking in its presence did not cause a significant increase over the noncross-linked control activity (Fig. 5). These results further suggest that heparin brings about activation through a mechanism that is distinct from that of the other activators. Rather than stimulate the activity of enzyme cross-linked in its presence, the inhibitor melittin, if anything, caused a slight inhibition (Fig. 5), further indicating that PDM is a reporter of active conformations of the enzyme. DISCUSSION PbK underwent intramolecular differential affinity crosslinking by the three geometric isomers of PDM to form three major conjugates (␣␤, ␣␥␥, and ␤␥␥), plus smaller amounts of a probable ␣␦ complex. With all forms of the enzyme tested, there was an increased formation of ␣␥␥ in going from o-, to m-, to p-PDM, which not only changes the geometry of cross-linking, but also increases the cross-linking span from 4.8 Å to 12 Å (14). The rate, efficiency, and concentration dependence of cross-linking suggested the possible presence of binding sites for PDM on PbK, which was confirmed through the inhibition of cross-linking by the chemically inactive structural analogs oand p-NTS. These results identify PbK as one of numerous proteins that bind and subsequently undergo modification by reactive phenyl derivatives that would not generally be considered as affinity labels (for discussion, see Ref. 33). Given that the cross-linking of PbK by both o-and p-PDM was inhibited to the same extent by either o-or p-NTS, it is probable that the isomers of PDM bind to the same site or sites. In a recent study that compared the binding of benzene, o-xylene, and p-xylene within a buried nonpolar cavity of T4 lysozyme, it was shown that the same site accommodated each xylene isomer through different ligand binding interactions: the o-and p-substituted isomers were bound in orientations resulting from rotation and translation of the ring with respect to its orientation in bound benzene (34). Similarly, the differential formation of ␣␥␥ and ␤␥␥ could possibly be caused by different orientations of each PDM isomer within a given site as opposed to simply the different cross-linking spans of the isomers.
Although the number of binding sites for PDM/NTS, their relative affinities for ligand, and their subunit location are certainly issues of interest, consideration of these issues is complicated by the fact that affinity cross-linking has two underlying components, binding and reactivity, with the latter comprising two separate reactions by the bifunctional reagent. Thus, even though ␣␤ forms much more readily than ␣␥␥ and even though higher concentrations of NTS are required to inhibit the formation of ␣␤ than of ␣␥␥, one cannot say with certainty that PDM binds with higher affinity to the site(s) involved in formation of ␣␤ than to those involved in ␣␥␥ trimer formation because at any given site low affinity could be offset by high reactivity. In analyzing the number and location of PDM/NTS binding sites, one must also take into account the high sequence homology of the ␣ and ␤ subunits (28). Based on the formation of ␣␤, which is inhibited by NTS, there is clearly at least one binding site for PDM/NTS on the ␣ or ␤ subunit. In addition, there must also be one other class of binding site on the holoenzyme, because neither ␣␥ nor ␥␥ accumulates as an intermediate of the ␣␥␥ trimer when cross-linking is carried out in the presence of NTS, i.e. both cross-linking events necessary for trimer formation are inhibited by NTS, demonstrating the presence of at least two distinct binding sites in the trimer. The second site could be on either the ␣ or ␥ subunit, and the trimer could theoretically be formed either as ␥-␣-␥ or as ␣-␥-␥. The latter configuration seems unlikely because neither ␥ 2 nor higher order oligomers of ␥ have been observed after cross-linking of the PbK holoenzyme with a variety of crosslinkers spanning from 0 to 17 Å (5, 8 -10); nevertheless, it cannot be completely ruled out, especially since there is indirect evidence suggesting that ␥ subunit from rat soleus exists as an oligomer when expressed by itself (35). Although for the sake of simplicity only the ␣␥␥ trimer was considered in the above discussion, each consideration also applies equally to the ␤␥␥ trimer. Moreover, the ␣ and ␤ subunits, being homologs, could contain similar binding sites for PDM.
The cross-linking of PbK by PDM was conformation-depend- The enzyme was then diluted 80-fold in Hepes buffer, pH 6.8, containing 3.0 mM dithiothreitol as cross-linking quencher, followed by an additional 10-fold dilution into the activity assay at pH 6.8. The error bars represent standard deviations of assays performed in triplicate. This protocol is described in more detail under "Experimental Procedures." ent in that stimulators of activity (ADP, GDP, and phosphorylation) caused a large increase in the formation of ␣␥␥ (especially with p-PDM) and a small increase in ␤␥␥. Thus, activation of PbK is associated with changes in quaternary structure involving the ␣, ␤, and ␥ subunits and/or with the unmasking of nucleophilic residue(s) on at least one of these subunits. Regardless, however, of the exact mechanism through which activation by nucleoside diphosphates and phosphorylation occurs, cross-linking with PDM provides direct structural evidence for alterations in the interactions of the catalytic ␥ subunit with the regulatory ␣ and ␤ subunits upon activation. The structural changes resulting in increased formation of ␣␥␥ and ␤␥␥ trimers caused by the above well characterized activators of the kinase were not observed with the polyanionic activator heparin, which has previously been shown to have different structural effects on PbK than other activators (32), including causing dissociation of the ␦ subunit (30). Melittin, an inhibitor of PbK (31), likewise did not cause increased formation of ␣␥␥ and ␤␥␥ trimers.
The ability of PDM to lock the enzyme in the active form(s) induced by ADP and GDP further indicates that these crosslinkers are effective reporters of active conformations of the holoenzyme. Intramolecular cross-linking has been shown to stabilize given forms of proteins (36), effectively locking them in specific conformations. For example, cross-linking by bis (3,5dibromosalicyl)fumarate has been reported to lock hemoglobin in its T-state (37). In an oligomer the size and complexity of PbK, there is undoubtedly a large number of intersubunit contacts that define its active conformation(s) (38), and only a small fraction of these are likely to be targeted by PDM. Nevertheless, our results indicate that in the case of activation by ADP and GDP, PDM cross-links at least several sites in the activated enzyme that allow it to remain activated after removal of the activators. Of the PDM conjugates formed, ␣␥␥, and to a lesser extent ␤␥␥, are apparent indicators of the activated conformer(s), in that both trimers are preferentially formed with activated kinase. The extent of formation of the ␣␤ dimer was essentially the same with both activated and nonactivated forms of the kinase; moreover, when cross-linking was performed at a low pH so that the only conjugate formed to a significant extent was the ␣␤ dimer, no activation was observed in subsequent assays. Similarly, zero length cross-linking with transglutaminase, which resulted in ␣␤ dimers as the predominant conjugate, did not give rise to activation (8). Based on the time-and concentration-dependent cross-linking of PbK by PDM under different conditions, there are two major sets of PDM binding sites on the PbK holoenzyme: high efficacy site(s) involving the formation of ␣␤, which is generated in greater amounts than the remaining conjugates with all forms of the enzyme tested, and low efficacy site(s) involving ␣␥␥ and ␤␥␥ formation. It is the cross-linking between the catalytic ␥ subunits and the inhibitory ␣ and ␤ subunits (39 -41) at the low efficacy site(s) that reflects the activation state of the kinase.