structure that have the best potential of domain movements, the crystallographic structure of cat muscle pyruvate kinase was examined. Each enzyme is ...
THEJOURNALOF BIOLOGICAL CHEMISTRY
Vol. 263. No. 6, Issue of February 25, pp. 2794-2601,1988 Printed in U.S. A.
Domain Interactionin Rabbit Muscle Pyruvate Kinase 11. SMALL ANGLE NEUTRONSCATTERING
AND COMPUTER SIMULATION* (Received for publication, July 21, 1987)
Thomas G. ConslerS, Edward C. Uberbacherj, Gerard J. BunickjT, MichaelN. LiebmanlJ,and James C. Lee$** From the #E. A . Doisy Department of Biochemistry, St. Louis University School of Medicine, St. Louis, Missouri 63104, ?lThe National Center for Small Angle ScatteringResearch, Solid State Division, Oak Ridge National Laboratory and the §University of Tennessee, Oak Ridge Graduate School of Biomedical Sciences and Biology Division, Oak Ridge, Tenessee 37830, and the !!Departmentsof Physiology and Biophysics and of Pharmacology, Mt. Sinai School of Medicine, New York, New York10029
The effects of ligands on the structureof rabbit musIn earlier studies (Oberfelder et al., 1984a, 1984b), it was cle pyruvate kinase were studiedby small angle neu- shown that the hydrodynamic properties of rabbit muscle tron scatteripg. The radius of gyration, Rc,decreases pyruvate kinase are altered by metabolites. Binding of subby about 1 A in the presence of the substrate phos- strate and metal ions required for activity causes the enzyme phoenolpyruvate, but increases by about the same mag- to assume a more symmetric structure, whereas the allosteric nitude in thepresence of the allosteric inhibitorphen- inhibitor would induce pyruvate kinase to become moreasymylalanine. With increasing pH or in the absence of Mg2+ metric. The detailed structural features associated with these and K+, theeRGof pyruvate kinase increases. Hence, global conformational changes are not defined. It is conceivthere is a 2-A difference in Rcbetween two alternative able that these global structural changes are associated with conformations. Length distribution analysis indicates the protein domains detected by low resolution x-ray crystalthat, under all experimentalconditions which increase the radius of gyration, there isa pronounced increase lography of cat muscle pyruvate kinase (Stuart et al., 1979 observed in the probability for interatomic distance Stammers and Muirhead, 1975). In the preceding paper (Consler and Lee, 1988), evidence between 80 and 110 A. These small angle neutron a “contraction” and“expan- was presented to show the independent solution behavior of scattering results indicate sion’’ of the enzyme when it transforms between its these domains. The effect of Phe,’ the allosteric inhibitor, on active and inactiveforms. Using the a-carbon coordi- the interaction between structural elements is manifested as toward denaturation by guanidine nates of crystalline cat muscle pyruvatekinase, a changes in protein stability hydrochloride (Consler and Lee, 1988). In addition, it was lengthdistribution profile was calculated, andit shown that a site susceptible to tryptic digestion lies in an matches the scattering profile of the inactive form. interdomain region and becomes preferentially exposed under These observations are expected since thecrystals were grown in the absence of divalent cations(Stuart, solution conditions that favor the inactive state of pyruvate D. I., Levine, M., Muirhead, H., and Stammers, D. K. kinase. Incontrast,thissite becomes protected when the (1979) J. Mol. Biol. 134, 109-142). Hence, results enzyme assumes its active conformation. Thus, itwas inferred from neutron scattering, x-ray crystallographic, and that this domain-domain interaction is involved in the transedimentation studies(Oberfelder, R. W., Lee, L. sition between active and inactive enzymatic forms. L.-Y., and Lee, J. C. (1984) Biochemistry 23, 3813In order to identify the locations in the pyruvate kinase 3821) are totally consistent with each other. structure that have the best potential of domain movements, With the aidof computer modeling, the crystal struc- the crystallographic structure of cat muscle pyruvate kinase ture has been manipulated in order to effect changes was examined. Each enzyme is composed of three major that are consistent with the conformational change domains, one of which protrudes out into the solvent. This described by the solution scattering data. The struc- exposed domain, identified as domain B, forms a cleft with tural manipulation involves the rotation of the B domain relative to the A domain, leading to the closure domain A adjacent to the active site. Domain B is attached of the cleft between these domains. These manipula- to domain A by what appears to be a flexible hinge region. It tions resulted in the generation of new sets of atomic is this particularregion that is identified as a trypticdigestion (C-CY)coordinates, which were utilized in calculations, site whose accessibility correlates with the enzymatic activity the result of which compared favorably with the solu- and conformation of pyruvate kinase. The fact that changes in conformation are correlated to changes in enzymatic activtion data. ity allows one to envision an alteration in structure at the active site as being responsible for the functional change. Thus,it is conceivable that the movement of domain B * This work wassupported by National Institutes of Health Grants relative to domain A may open or close the active site cleft in NS-14269 and DK-21489 (to J. C. L.), by ImClone Systems, Inc. response to substrates or effectors. (supporting M. N. L.), and by National Science Foundation Grant Since no crystallographic data areavailable to characterize DMR-7724459 through Interagency Agreement 40-636-77 with the the active and inactive forms of pyruvate kinase, an effort to United States Department of Energy under Contract DE-ACO5probe the molecular dimensions and overall shape of the 840R21400 with the MartinMarietta Energy Systems, Inc. The costs enzyme as a function of solution conditions was initiated. of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adver- Small angle neutron scattering (SANS) was employedto map tisement” in accordance with 18U.S.C. Section 1734 solelyto indicate this fact. ** To whom correspendence should be addressed.
The abbreviations used are: Phe, L-phenylalanine; SANS, small angle neutron scattering.
2794
in Kinase Pyruvate
Movements Domain structural changes that occur upon binding of ligands to pyruvate kinase. This experimental approach was selected because of the potential for comparison of the observed scattering profile with length distribution profiles calculable from the crystallographic structure. In addition, the most probable structural change that modulates the activity of pyruvate kinase will be computationally modeled and compared with the experimental results obtained by SANS and displayed using computer graphics.
2795
of Moore (1980) and yields information which describes the size, shape, and frequency distribution of all the point-to-point pair distances between scattering centersof the particle. It also yields values for RG,evaluated using the relationship:
and for Z(0)using Equation 4.
MATERIALS ANDMETHODS
Pyruvate kinase, Tris-HC1, Tris base, ADP, P-enolpyruvate, and NADH were obtained from Boehringer Mannheim. Lactate dehydrogenase, MgSO, and KC1 were purchased from Sigma, Phe was from Schwarz/Mann. D,O was supplied by Oak Ridge National Laboratory. Buffers were made up using Tris-HC1 and Tris base, each a t 50 mM, titrated together to achieve the desired pH. If cations were to be present, they were added before the titration to both Tris-HC1 and Tris base to yield MgSO, and KC1 concentrations of 7.2 and 72 mM, respectively. The same procedure was used for both H,O and D,O buffers. For D,O buffers, the value ofpD,, refers to the reading obtained with the glass electrode. Pyruvate kinase was obtained as anammonium sulfate suspension. It was concentrated and desalted before use by sequential centrifugation, resuspension, and gel filtration through a Sephadex G-25 column (fine) equilibrated with the buffer in which the experiment would be carried out. Samples were prepared using the desalted pyruvate kinase stock solution described above. Each pyruvate kinase sample was dialyzed for 18 h at 5 "C against three changes of the buffer in which the experiment would beconducted. A finalbuffer change was made, and the samples were further dialyzed for 4-6 h at 23 "C. Pyruvate kinase is stable for the entire length of the sample preparation and experimental period, as monitored by steady-state enzyme kinetic activity. Protein concentrations were determined by UV absorbance a t 280 nm using an absorptivity of 0.54 ml/(mg.cm) (Boyer, 1962). Samples, typically 2 ml of protein ranging from 7 to 15mg/ml, were transferred to 5- or I-mm path length round quartz cells for the samples in D20 and water, respectively. The corresponding dialysate was placed in a matched cell to be used as correction for background scattering. Transmissions for each sample and buffer were measured in the neutron beam. Exchange of solvent was determined to be complete when each sample and its dialysate had nearly identical transmissions. Experiments were carried out with the 30-m SANS instrument a t the High Flux Isotope Reactor of Oak Ridge National Laboratory (The National Center for Small Angle Neutron- Scattering). The wavelength of the incident neutron beam was 4.75 A, and thesampleto-detector distance was set a t 3.0m. The detector is a 64 X 64 position-sensitive detector. Samples were placed into the automatic liquid sample changer and were counted typically for 4 h (usually cycled twice for 2 h each). Multiple data sets for each sample were added together, and thiscollective run was radially averaged, yielding scattering curves of intensity uersus k, where k = 4(n/X)sinB and X is the wavelength, which assumes a value of 4.75 A, whereas 28 is the scattering angle. All SANS experimentswere conducted a t 23 "C. Data Analysis-SANS data were subjected to two separate analyses. The first method relies upon the data obtained at the lower scattering angles. Data in this region conform to the Guinier relationship: In Z(k) = -k2R~'/3 + In Z(0)
(1)
where Z(k) and Z(0) are the scattering intensities at angles 26 and 0, respectively, and RGis the radius of gyration. From a Guinier plot of In Z(k) uersw k2, one is able to deterTine Rc and Z(0). Data points from low angle scattering up to k = 0.07 A" were subjected to weighted linear least-squares fittingto yield values for these two parameters. The second method of analysis takes into account all of the SANS data up to the point where the intensity approaches 0, e.g. up to k = 0.13 A". The analysis yields the length distribution function, which is defined by Equation 2. P(r) = 2r 77
l-
kZ(k)sin(kr)dk
(2)
This function can be approximated by the indirect transform method
Z(0) = l P ( r ) d r Since this analysis utilizes more of the scattering data than the Guinier analysis, it contains more information. Scattering of Pyruuate Kinase Predicted from X-ray Crystallographic Coordinates-Using the a-carbon coordinates determined by x-ray crystallography,2 the solution scattering behavior of the observed conformation of pyruvate kinase was predicted. The scattering curve and P ( r ) distribution were calculated from the known a-carbon coordinates as follows. The coordinates for the crystallographically symmetric tetramer were constructed by applying the appropriate symmetry operations to the monomer coordinates. For an object assumed to consist of discrete density points, the Debye(1915) relation can be used to calculate directly the scattering curve:
where Z(k) is the scattering intensity at k = 4nsin6/X, 6 is the halfscattering angle, pi is the scattering length density at position i, and r,, is the distance between points i and j . All coordinate pairs were used to directly calculate P ( r ) for pyruvate kinase. Also, in the calculation, the scattering length density was assumed uniformly constant throughout the molecule. Comparison of Solution and Crystal Structures-The fact that the observed scattering data contain information pertaining to interatomic distances enables one to generate length distributions from both SANS and x-ray crystallographic data. The changes in length distributions, reflecting conformational changes induced by ligands, can then be compared between the two sets of experimental observations. It must be stressed that separate comparisons were conducted for the structural parameters derived from solution (using the SANS data)andthe crystal structure (using the x-ray coordinates and molecular modeling as described below) since the solution data (SANS) contain contributions from all scattering centers, including that of the side chains, whereas the available crystallographic data are only that of the a - c a r b o n ~ .Our ~ approach therefore involves generating the difference distribution, A P ( r ) , which is ideal for the comparison of data sets obtained under different conditions. This function ( A P ( r ) )enables one to determine the (length distribution) changes that occur in solution. Having determined those changes, they are used as a guideline for the modelingwhichinvolves the ~
* H. Muirhead, personal communication. A more qualitative, but equally significant observation is that the shape of the distribution of interatomic distances, which approximates the shape of the probability of length distribution computed from thescattering observations, differs only in a scaling factor. Otherwise, the shape of length distribution is independent of choice of a-carbons, &carbons, main-chain atoms, orfull atoms for the structural representation. The details of trial computations performed on hemoglobin are not presented here, but are available from the authors. In addition, previous studies on phosphoglycerate kinase (Pickover et al., 1979)have suggested the suitability of approximation of the full atom set with radius of gyration computations performed using only the @-carbon coordinates (e.g. differences between Rc values computed using &carbons, main-chain atoms, and full atoms differ within less than a total of 1%).In this study, computation was extended to using a-carbon atomsonly and found the approximation to be valid to within 1%.Hence, it was concluded that the approximation of the full protein structure by a-carbon atoms only should not significantly affect the computations related to radius of gyration or length distribution functions for qualitative analysis to aid in identifying the major conformational change that can account for the experimental observations.
2796
Domain Movements in Pyruvate Kinase
manipulation of the a-carbon coordinates. The computation is deemed satisfactory when it yields A P ( r ) function that compares favorably with that determined from solution data. A series of calculations and simulations using Equations 1-5 were carried out as outlined in Fig. 1. It is evident that the common element in analysis is the relation of I versus k . This is obtained directly from SANS and indirectly from x-ray crystallographic coordinates through the Debye (1915) relationship using Equation 5. Once the three-dimensional coordinate data were converted to theform of an I versus k distribution, the path of data analysis converges since both solution and crystal data aretreated identically. Simulation of conformational changes was conducted by manipulation of the x-ray crystal coordinates as indicated by the step of computer modeling, as shown in Fig. 1. This simulation step employed computer graphics, which enabled one to conduct interactive rotation and translation, symmetry operations, and inter-a-carbon distance calculations. Having decided on a simulated conformational change, the information was then reintroduced into the pathof data analysis in the form of altered a-carbon coordinates. All manipulations onthe a-carbon coordinates are performed on isolated monomers. Subsequently, these newly modeled structures are used to reconstruct the tetramer by the same symmetry operations that yielded the original tetramer. The point at which the comparison between experimental and simulated data takes place is at the level of P ( r ) distribution. The difference between solution conformations, hP(r)solution, and crystal conformations, AP(r)crystal, was compared; when AP(r)solution = hP(r)crystal, the conformational change was considered adequately modeled. It is especially useful for the comparison of data setsobtained under different solution conditions. Hence, P ( r ) distributions were employedto illustrate changes in molecular dimensions that are the result of these solution variations.
r
I
10-
0
0
0
04
08 12 PEP coneenfration (mM)
I6
-
20
FIG. 2. Steady-state kinetics of rabbit muscle pyruvate kinase, with P-enolpyruvate (PEP)as variable substrate. Initial velocities are normalized to a value of 1.0. Experiments were performed in TKM buffer (50mM Tris, 72 mM KCI, and 7.2 mM MgSO, (pH 7.5)) (filled symbols) or pD,, 8.5 (open symbols) a t 23 "C. Assay buffer contained 2 mM ADP, 0.3 mM NADH, and 10 gg/ml lactate dehydrogenase, and the assay was carried out as described (Bucher and Pfleiderer, 1955). The assays were performed in the absence (circks) or presence (squares)of 12 mM Phe.
tering experiments in deuterium oxide, conditions were determined so that resultsin D 2 0 could be compared with those in HzO. Hence, steady-state kinetic studies were performed to determine conditions in which the kinetic properties of pyruvate kinase are identical in DzO and HzO. It was empirically determined that pyruvate kinase hadsimilar kinetic properties at pH7.5 and pD.,, = 8.5, as shown in Fig. 2. It is evident that the values of are identical in H,O and D,O. Furthermore, 12 mM Phe exerts the same amount of inhibiRESULTS tory effect on pyruvate kinase as indicated by an identical In order to achieve maximal contrast, the SANS experishift of the activityversus P-enolpyruvate concentration plot, ments were initially performed in buffers containing DzO. although V,,, is dramatically decreased as D,O content is The rationale for this approach is that by examining pyruvate increased, a most likely consequence of deuterium solvent kinase at a greater contrast with respect to the solvent, the isotopeeffect. For the purpose of the SANS experiments, ligand-induced structural changes, if any, would be detected three pD,,, values were chosen, 7.5,8.5,and 9.5. Based on the with a greater sensitivity andprecision since replacement of results shown in Fig. 2, it may be assumed that theseexperiH,0 with D,O would lead to an increase in signa1:noise ratio mental conditions should mimic those of pH values 6.5, 7.5, as a consequence of the lower incoherent scatteringof DzO. and 8.5, respectively. SANS Carried out in D,O-Prior to carrying out the scat- SANS experimentswere conducted in D20 asa function of pD,,, and solution variables. Under all experimental conditions, no upward curvatureswere observed at small angles in the In I ( k ) uersus k plots; thus, they indicated that there is no concentration-dependent aggregations of pyruvate kinase. EXPERIHENTAL This conclusion is substantiated by the resultsfrom sedimentation velocity experiments which showed asimple linear decrease in sedimentation coefficient with increasing pyruvate kinase concentration up to 20 mg/ml in the presence or absence of ligands (Oberfelder, 1982).4Values of Rc, corrected for protein concentrationdependence, were calculated by both SIINWTION Guinier plots andprobability of length distribution a!alysis. At pD,,, 8.5, Rc; for pyruvate kinase is 39.6 f 0.1 A. The presence of P-enolpyruvate resulted !i a decrease in the Rc 4. of about 0.5 A to a value of39.0 f 0.1 A, whereas Phe caused a n increase to 40.1 f 0.2 A. Hence, P-enolpyruvate and Phe exert opposite effects on the structure of pyruvate kinase, a conclusion in total agreement with the sedimentation data published earlier (Oberfelder et al., 1984a, 1984b; Oberfelder, 1982). The effect of cations was also tested. If the cations were omitted from the buffer, the Rc value also increased to 40.3 k 0.2 A, a value identical to that induced by Phe. In TRANSFOWTION addition, 12 mM Phe does not seem t o induce any structural changes in pyruvate kinase in the absence of cations, thus implying that the presence of 12 mM Phe or the absence of cations may induce the enzyme t o assume the same gross conformation. Similar Phe effects are reported at pD.,, 7.5 and 9.5, namely, addition of 1 2 mM Phe induced a significant
L
I I
,&,
FIG.1. Algorithm for comparison of solution and crystalline structure. PK, pyruvate kinase.
T. G . Consler, unpublished results.
Kinase Pyruvate Movements in Domain
2797
increase in Re, as shown in Table I. The effect of pD on the structure of pyruvate kinase was also tested, With increasing pD.,, from 7.5 to .9.5, the value of Rc increases from 39.4 f 0.1 to 40.1 f 0.1 A. Hence, increasing pD seems to favor an expansion of the pyruvate kinase molecule in a manner potentially analogous to that of introducing Phe. In order to gain more insight into thecoupling between the effects of Phe and pD, the changes in Re are shown in Fig. 3. The reference states of this comparison are the values for pyruvate kinase at each pD.,,. As pD.,, is increased from 7.5 to 9.5, the change in RG induced by Phe at a particular pD.,, decreases. Hence, it maybe concluded that the pD.,, and the Phe-induced change are most likely linked. Values of Re in D20 are summarized in Table I. In addition to Re values, the length distribution functions were also calculated from the scattering curves. Comparison TABLE I Summary of RG values in D,O The buffer was 50 mM Tris, 7.2 mM MgSO,, and 72 mM KC], exceut where noted.
of these curves allows the detection of changes in the shape of the scattering particle. Fig. 4 shows the length distribution functions of pyruvate kinase at pD,,, 8.5. The presence of Phe induced a small increase of P ( r ) at longer lengths, and there is a shift in the peak maximum to longer lengths. On the other hand, P-enolpyruvate caused only a slight shift of the peak maximum to shorter lengths. These results are in agreement with the changes in Rc values. These P ( r ) distributions also serve to illustrate that the shape of pyruvate kinase is sensitive to pD,,,. In order to characterize more fully the changes, results are presented as AP(r) versus r. The effect of increasing pD is shown in Fig. 5. Using the results at pD,,, 7.5 as the reference, it is evident that, with increasing pD,,,, the peak maximum is shifted toward lvnger lengths with a pronounced change between 90 and 110 A at pD.,, 9.5. This P ( r ) distribution resembles that seen for the pyruvate
-
oo;i#t.E5i" 0 , 0 ,
0 : i '
O x .
06. D
8
0 "
i
o i
-8
6~ P(r)
A 7.5 12 mM Phe 8.5 2 mM PEP' 1 mM Phe 6 mM Phe 12 mM Phe No Mg2+, K+ No Mg2+, K', 12 mM Phe 9.5 12 mM Phe
39.4 f 0.1 40.2 & 0.1 39.6 &39.5 0.1 0.1 39.0 f39.0 40.1 & 0.1 40.1 & 0.1 40.2 & 0.1 0.1 40.1 f40.5 40.1 f40.4 0.1
39.2 40.7
4
4-
0 .
:.a .
1.
I
5.
I..
I
40.5 40.2 40.4
6 . 8
I
"I;.
Ex..
L
. , ,
I
I
I
1
respectively. PEP, P-enolpyruvate.
c
I
72
24 Length
40.5 41.3
.
EE'.
I
0 48
40.1 & 0.1 40.3 -+ 0.2
n
1
!'
~
2 -
bRC values were calculated according to Equations 1 and 3,
l~
1
1
96
(A)
l a b " l
120
FIG.4. Probability of length distribution for pyruvate kinase in TKM buffer: effects of ligands at pD,,
8.5. The probability of length is in arbitrary units, and the length is in angstroms. The buffer was TKM. X, enzyme alone; U, enzyme with 12 mM Phe; 0,enzyme with 2 mM P-enolpyruvate.
-8
-1 60
-1
L
24
48 Length
24
48
2 96 (i)
120
8
I
L
-2.
FIG.3. Changes in radius of gyration of rabbit muscle pyruvate kinase,measured in 100% DzO. Filled bars refer to values obtained using the Guinier analysis, and open bars are for the values obtained by the length distribution function analysis. Differences are relative to theRc of pyruvate kinase in TKM buffer at each respective pD.,,, except for E. A-C are for pyruvate in the presence of 12 mM Phe atpD.,, 7.5,8.5, and 9.5, respectively. D is for the enzyme in the presence of 2 mM P-enolpyruvate a t pD.,, 8.5, and E is the effect of MgZ+and K+ relative to pyruvate kinase in Tris atpD.,, 8.5.
-I 6
0
Length
72
(A)
96
120
FIG.5. Difference length distribution function: effect of pD.,,. A , pyruvate kinase in TKM buffer as a function ofpD.,,: "- , pD., 8.5; X, pD.,, 9.5. B , pyruvate kinase in TKM buffer and 12 mM Phe asa function of pD.,,: . . . ., pD.,, 7.5; - - -, pD,,, 8.5. All difference plots are relative to pyruvate kinase in TKM buffer at PD.,, 7.5.
PyruvateMovements in Domain
2798
Kinase
kinase-Phe complex at either pD,,, 7.5 or 8.5 (Fig. 5B). theGuinieranalysis of samplesinHzOarein complete When the cations are omittedfrom the buffer at pD,,, 8.5, agreement with those in D,O. the P ( r ) distribution for pyruvate kinase is nearly identical Transformation of the scattering curves yielded P ( r ) proto thatof the enzyme-Phecomplex at pD,,, 8.5. These results files for the samples measured in water. Unexpectedly, the are presented asA P ( r ) versus r, as shown in Fig. 6. Further- changes induced by varying the solution conditions in HzO more, in the absence of cations a t pD,,, 8.5, Phe does not were more pronounced than those seen in D,O. The P ( r ) induce any further structural changes in pyruvate kinase (Fig. distributions clearly demonstratedthatpH,Phe,and P6). enolpyruvate each affectthe overall conformation of pyruvate SANS in H20-In order to correlate structural data with kinase. Increasing p H resulted ina constant shiftof the length functional information, the scattering experimentswere also distribution profile toward larger length values, e.g. at pH8.5, carriedout inbuffersmade up in HzO. Inherent in these a pronounced increase was observed in theoprobability for experiments is a signa1:noise ratio which is lower for samples interatomicdistance between 75 and 110 A in a manner in HzO than in D20. Results of these experiments are sumsimilar to that in DzO (Fig. 5A). Increase in probability for marized in Table 11. It was found that the absolutevalues for this interatomic distanceis also observed at both pH7.5 and Rc varied from experiment to experiment, but that differthe 8.5 in thepresence of Phe, as shown in Fig. 7. Phe also induces ences induced by ligands remained consistent. Results based a significant change in theprobability forinteratomic distance on the Guinier analysis showed that, at each pH measured, between 20 and 40 A; in this instance, theprobability in this the value of Rc decreased in the presence of P-enolpyruvate length range is reduced. A similar change also occurs for all and increased in the presence of Phe, relative to that for samples in the presence of P-enolpyruvate, although there is pyruvate kinase aloneat each pH. The pHdependence of the no significapt perturbation for interatomic distance between Phe-induced Rc change is clearly obvious. It is similar to the 75 and 110 A in thiscase. These changes in P( r ) distributions relationship found inD20,namely,aspHincreases,the substantiate those results for pyruvatekinasein D20, almagnitude of the Phe-induced changedecreases. In addition, though thechanges are amplified over those observed in D,O. it was found that a similar pH dependence exists for the PHaving made observations indicative of a change in the enolpyruvate-induced Rc;change. AS pH increases, the change pyruvatekinasestructure as induced by interaction with in Rc caused by P-enolpyruvate decreases. Hence, results of ligands, it is of interest to identify the specific nature of the conformational change. One way of achieving such a goal involves manipulating theobserved three-dimensional crystal structure of pyruvate kinase, as determined by Stuart et al. 16 (1979) and then comparing the observed and calculated scattering properties of the computationally altered pyruvateki8 nase structure. Toward the goal of defining the conformational changes withina conservative,well-defined framework, the following guidelines are utilized ( a ) The conformational change thatis selected has to accountfor the observed differences in physicochemical properties and should involve the simplest manipulation of the observed crystalline structure. ( b ) Analysis of potential steric hindrance based on a-carbon -I 61 I I I 0 24 48 7,2 96 120 Length ( A )
1
1
i
FIG. 6. Difference length distribution function: effect of cations at pD.,, 8.5. Units are asdescribed for Fig. 3. All difference plots are relative to pyruvate kinase in TKM buffer a t pD.,, 8.5: . . . .,pyruvate kinase in Tris;(- - -), pyruvate kinase in TKM buffer and 12 mM Phe; X , pyruvate kinase in Tris and 12 mM Phe.
-1 6
2j?
0
TABLE I1 Summary of Rc values in H20 The buffer was 50 mM Tris, 7.2 mM MgS04, and72 mM KCI.
48
72
120
96
Length ( A )
RG
pH
Ligands
I"
IIb
A 6.5 15 mM Phe 2 mM PEP' 7.5 15 mM Phe 2 mM P E P 39.8
8.5 15 mM Phe 2 mM P E P
40.8 ? 0.6 41.9 f40.1 0.7 39.8 &39.2 0.6 39.7 f 0.6 40.6 & 0.4 38.8 & 0.6 ? 0.8 40.0 f42.3 0.4 39.3 f 0.4
39.6 40.2 41.4 38.8 41.2
40.3 RGvalues was calculated according to Equations 1and 3, respectively. PEP, P-enolpyruvate. E
-";
-16
1
0
1
,
1
1
1
48 Length
'
'
72
'
1
1
/
120
(%I FIG. 7. Difference length distribution function. A , pH 7.5; B, p H 8.5. X , enzyme in 15 mM Phe minus enzyme in 2 mM P-enolpyruvate; . . . ., enzyme in 15 mM Phe minus enzyme in TKM buffer; (- - -), enzyme in2 mM P-enolpyruvate minus enzyme in TKM buffer.
Kinase PyruvateMovements in Domain
2799
chosen as the siteof movepositions only will be made toward the accommodation of the pyruvate kinase structural domain ment is the B domain of each subunit. conformational change. The first stepin performing the simulation is the choice of In an initial attempt to identify the nature of structural change in pyruvate kinase, a comparison was made between a hinge or rotation axis about which the B domain is to be the crystalline structureof the enzyme and hemoglobin since rotated. Since the B domain is formed by looping out of the the tetramerof pyruvate kinase exhibitsa similarity with the A domain, it is attached by twopolypeptide strands, each organization of the hemoglobin tetramer. A simple rotational consisting of 3 or 4 amino acid residues. These strands form perturbation solely atthedimerdimerinterface does not a stemlike structure which supports the B domain. The end of the stem is the most likely source of flexibility as this induce any change in the Rc value of the pyruvate kinase tetramer. The only motion that would lead to a change of RO would most easily accommodate a cleft closure. The midpoint from 41 to 39 A (Fig. 3) is the introductio? of a translational of the residue pair (residues 115 and 223) at the end of the movement between the dimers by 6-7 A. Even when this stem ( i e . closest to the A domain) was therefore chosen as a hinge point. Bothresidues lie in regions of unordered structure rotation and translation are coupled, the structural change (dimer compaction) is incompatible with the experimentally between segments of more regular secondary structure. Other hinge points were also examined, but were found to be less observed changes becaust the altered structures require interatomic distances of
