Enzyme Flexibility and Enzyme Action’
`
`D. E. KOSHLAND, JR.
`Biology Department, Brookhaven National Laboratory,
`Upton, New York
`
`It has long been clear that enzyme ac-
`tion is intimately involved with the three-
`dimensional arrangement of amino acids.
`The specificity of the enzymes led Fischer
`(1894) to propose a “key-lock model for
`the steric relations at the active site, and
`others showed that denaturation was cor-
`related with changes in shape of the en-
`zyme. The tools for measuring protein
`shapes are still primitive, and the physical
`description of that ill-defined area called
`“the active site” is even more complicated.
`Nevertheless, unsuspected tools are fre-
`quently uncovered under the impetus of
`pertinent questions and we should like,
`therefore, to attempt an answer to the
`question “Is the active site rigid or flexible
`during enzyme action?”.
`One source of information comes from
`the general studies on protein shape, which
`have followed almost from the discovery
`of proteins and protein denaturation. The
`first evidence that proteins were flexible
`as well as fragile probably comes from
`the studies of Anson and Mirsky (’34) on
`the reversible denaturation of
`trypsin.
`Since then the able and original work of
`a number of workers (e.g., Karush, ’50;
`Kauzmann, ’54; Lumry and Eyring, ’54;
`Doty and Yang, ’56; Linderstrom-Lang and
`Schellman, ’59) has led to a far greater
`understanding of this vital area. In addi-
`tion to temperature, other reagents such
`as urea, pH, salt concentration, and or-
`ganic solvents can be used to induce
`changes in the three-dimensional geometry
`of a protein. The changes in shape caused
`by these stresses can be measured by
`a variety of tools of which optical rotation,
`viscosity, sedimentation constant, deuter-
`ium exchange, and solubility are exam-
`ples. A brief and over-simplified summary
`of all of this work is that certainly large
`portions of many proteins are flexible in
`the sense that they can be reversibly de-
`
`formed. Urea, for example, will produce
`reversible changes in viscosity and enzyme
`activity in both trypsin and chymotrypsin
`(Harris, ’56). With almost all these re-
`agents, there appears to be a point of no
`return, after which irreversible changes
`are induced. If this limit is not exceeded,
`however, the evidence supports the con-
`clusion that removal of
`the stress re-
`turns the flexible portions of the protein
`to their natural conformations.
`The fact that large portions of the pro-
`tein are flexible is by no means evidence
`that all portions of the protein are. Actu-
`ally, fragmentary evidence exists that a
`change in certain portions of the protein
`always results in irreversible denaturation.
`We are left, therefore, with the conclusion
`that either a flexible or a rigid active site
`would be compatible with the general stud-
`ies of protein properties.
`Let us, therefore, examine the evidence
`for the template model of enzyme spe-
`cificity that argues for a relatively hard
`and inflexible active site. To a chemist,
`one of the most astounding properties of
`enzymes is their specificity. The fact that
`a small group in the substrate, far from
`the bond to be cleaved, can decide whether
`an enzyme acts or does not act on a par-
`ticular compound is almost incredible.
`Since studies of physical organic chem-
`istry have clearly indicated the magnitude
`of inductive effects, we can conclude that
`such effects cannot explain the observed
`changes in velocity from substrate to non-
`substrate in most cases. Fischer therefore
`proposed the template model with which
`he postulated that the substrate must be
`able to fit on the surface of the enzyme
`in order to get in close proximity with
`the catalytic groups there. If the groups
`
`1 Research carried out at Brookhaven National
`Laboratory under the auspices of the U.S. Atomic
`Energy Commission.
`
`245
`
`Koshland JR., D. E. (1959). "Enzyme flexibility and enzyme action."
`Journal of Cellular and Comparative Physiology 54(S1): 245-258.
`
`Columbia Ex. 2081
`Illumina, Inc. v. The Trustees
`of Columbia University in the
`City of New York
`IPR2020-00988, -01065,
`-01177, -01125, -01323
`
`

`

`Enzyme Flexibility and Enzyme Action’
`
`D. E. KOSHLAND, JR.
`Biology Department, Brookhaven National Laboratory,
`Upton, New York
`
`It has long been clear that enzyme ac-
`tion is intimately involved with the three-
`dimensional arrangement of amino acids.
`The specificity of the enzymes led Fischer
`(1894) to propose a “key-lock model for
`the steric relations at the active site, and
`others showed that denaturation was cor-
`related with changes in shape of the en-
`zyme. The tools for measuring protein
`shapes are still primitive, and the physical
`description of that ill-defined area called
`“the active site” is even more complicated.
`Nevertheless, unsuspected tools are fre-
`quently uncovered under the impetus of
`pertinent questions and we should like,
`therefore, to attempt an answer to the
`question “Is the active site rigid or flexible
`during enzyme action?”.
`One source of information comes from
`the general studies on protein shape, which
`have followed almost from the discovery
`of proteins and protein denaturation. The
`first evidence that proteins were flexible
`as well as fragile probably comes from
`the studies of Anson and Mirsky (’34) on
`the reversible denaturation of
`trypsin.
`Since then the able and original work of
`a number of workers (e.g., Karush, ’50;
`Kauzmann, ’54; Lumry and Eyring, ’54;
`Doty and Yang, ’56; Linderstrom-Lang and
`Schellman, ’59) has led to a far greater
`understanding of this vital area. In addi-
`tion to temperature, other reagents such
`as urea, pH, salt concentration, and or-
`ganic solvents can be used to induce
`changes in the three-dimensional geometry
`of a protein. The changes in shape caused
`by these stresses can be measured by
`a variety of tools of which optical rotation,
`viscosity, sedimentation constant, deuter-
`ium exchange, and solubility are exam-
`ples. A brief and over-simplified summary
`of all of this work is that certainly large
`portions of many proteins are flexible in
`the sense that they can be reversibly de-
`
`formed. Urea, for example, will produce
`reversible changes in viscosity and enzyme
`activity in both trypsin and chymotrypsin
`(Harris, ’56). With almost all these re-
`agents, there appears to be a point of no
`return, after which irreversible changes
`are induced. If this limit is not exceeded,
`however, the evidence supports the con-
`clusion that removal of
`the stress re-
`turns the flexible portions of the protein
`to their natural conformations.
`The fact that large portions of the pro-
`tein are flexible is by no means evidence
`that all portions of the protein are. Actu-
`ally, fragmentary evidence exists that a
`change in certain portions of the protein
`always results in irreversible denaturation.
`We are left, therefore, with the conclusion
`that either a flexible or a rigid active site
`would be compatible with the general stud-
`ies of protein properties.
`Let us, therefore, examine the evidence
`for the template model of enzyme spe-
`cificity that argues for a relatively hard
`and inflexible active site. To a chemist,
`one of the most astounding properties of
`enzymes is their specificity. The fact that
`a small group in the substrate, far from
`the bond to be cleaved, can decide whether
`an enzyme acts or does not act on a par-
`ticular compound is almost incredible.
`Since studies of physical organic chem-
`istry have clearly indicated the magnitude
`of inductive effects, we can conclude that
`such effects cannot explain the observed
`changes in velocity from substrate to non-
`substrate in most cases. Fischer therefore
`proposed the template model with which
`he postulated that the substrate must be
`able to fit on the surface of the enzyme
`in order to get in close proximity with
`the catalytic groups there. If the groups
`
`1 Research carried out at Brookhaven National
`Laboratory under the auspices of the U.S. Atomic
`Energy Commission.
`
`245
`
`Koshland JR., D. E. (1959). "Enzyme flexibility and enzyme action."
`Journal of Cellular and Comparative Physiology 54(S1): 245-258.
`
`

`

`246
`
`D. E. KOSHLAND, JR.
`
`were too bulky to allow this association,
`no catalysis resulted. If the groups neces-
`sary for binding the substrate to the en-
`zyme were absent, the substrate was not
`held to the enzyme and again no catalysis
`would result. Since the two postulated
`phenomena for the template hypothesis,
`i.e., steric hindrance and affinity by the
`formation of noncovalent complexes, were
`well substantiated in organic chemistry
`and since the development of enzyme ki-
`netics supported the presence of an en-
`zyme-substrate
`intermediate, this theory
`became widely accepted. In fact, it does
`suffice to explain the vast majority of the
`observed specificity patterns of enzymes.
`Our own feelings that all was not quite
`so well with the template theory as might
`appear on the surface came when we were
`trying to explain the failure of muscle
`phosphorylase to catalyze an exchange be-
`tween P320a and glucose 1-phosphate
`(Koshland, '54). In muscle phosphorylase,
`it had been shown that an acceptor was
`needed to observe exchange whereas with
`sucrose phosphorylase no acceptor was
`required (Doudoroff et al., '47; Cohn and
`Con, '48). Let us assume that the same
`mechanism is operating for muscle phos-
`phorylase as was indicated for the sucrose
`synthesizing enzyme; i.e., that a group on
`the enzyme attacks from the back of the
`carbon atom to form a glucosyl-enzyme
`intermediate. The existence of this mech-
`anism does not necessarily mean that ex-
`change must occur. It could be said that
`the glucosyl-enzyme
`intermediate exists
`for so short a time that the inorganic
`phosphate is unable to leave and be re-
`placed by a radioactive phosphate before
`the new covalent glucose-phosphate bond
`is formed. There is good analogy for this
`kind of kinetic variation in the neighbor-
`ing group effect, in which the gamut from
`the formation of a completely stable bond,
`as in epoxides, to the transient interaction
`of a neighboring methoxyl group is ob-
`served. However, if such a process were
`going on and there were repeated forma-
`tions of a glucosyl-enzyme
`intermediate,
`we might expect that periodically the
`water in the adjacent site would be able
`to react. Water should certainly be about
`as nucleophilic as the 4-hydroxyl group
`of the glycogen polymer, and it seemed
`
`inter-
`unlikely that the glucosyl-enzyme
`mediate being formed so rapidly and re-
`versibly would not occasionally react with
`the adjacent water molecule. An alterna-
`tive mechanism based on the SNi reaction
`(Koshland, '54) leads to almost precisely
`the same difficulty.
`It would seem that either the displace-
`ment mechanism or the template theory
`was inadequate. This would hardly be
`sufficient basis for questioning the tem-
`plate hypothesis, but on reflection we
`thought the failure of water to react in a
`number of other instances (e.g., the hexo-
`kinase reaction) was equally puzzling.
`Moreover, evidence in support of the dis-
`placement mechanism increased, and an
`intensive search of
`the literature was
`therefore made for examples that could
`not be reconciled with the template hypo-
`thesis. An amazingly large number of
`instances were found (Koshland, '55, '58,
`'59), and since this material has already
`been published, only one example will be
`used to illustrate the type of reasoning
`involved.
`Amylomaltase is a purified enzyme that
`catalyzes the hydrolysis of maltose but
`does not act on a-methylglucoside (Wies-
`meyer and Cohn, '57). a-Methylglucoside
`has the same stereochemistry at the C-1
`as maltose and the same type of bond
`to be broken; it differs only in that the
`methyl group has two hydrogen atoms
`where the remaining part of the second
`glucose ring would be placed. Since it
`could hardly be argued that these two hy-
`drogen atoms would be unable to fit into
`the area on the template reserved for the
`full glucose ring, the failure of a-methyl-
`glucoside to react would, on the template
`hypothesis, have to be explained by a fail-
`ure to be attracted to the enzyme surface.
`However, a-methylglucoside has been
`shown
`to be a competitive inhibitor.
`Hence it is known to be present at the
`enzyme surface and in the appropriate
`position, and yet no reaction occurs.
`From examination of these and other
`examples, it was clear that the template
`theory would have to be modified. The
`reasoning that led Fischer to conclude that
`a steric interaction was required seemed
`unassailable. The theory was modified,
`therefore, to give the substrate a more-
`
`

`

`ENZYME FLEXIBILITY AND ENZYME ACTION
`
`247
`
`1 0 - ~
`
`
`
`0
`
`\
`
`I
`
`positive role. It was assumed that the
`active site was not initially a negative of
`the substrate but became so only after
`interaction with substrate. This change
`in conformation of the protein occurred
`with the result that the final enzyme-
`substrate complex had the catalytic groups
`on the enzyme in the proper alignment
`with each other and with the bonds to be
`broken in the substrate molecules. This
`retained the idea of a steric fit proposed
`by Fischer but modified it in such a way
`that the failure of either too large or too
`small a compound
`to react could be
`readily explained. For example, the fail-
`ure of water to react in the phosphorylase
`reaction would be explained by the fact
`that the small size of the water molecule
`did not provide sufficient buttressing action
`to lead to the proper alignment of catalyt-
`ic groups. This, moreover, is in line with
`the observation of a minimum size for
`the primer in the phosphorylase reaction.
`This mechanism requires a flexible action
`at the active site, i.e., the protein changes
`shape under the influence of the substrate
`and returns to its original shape after the
`products have been released from the en-
`zyme surface. This "induced fit" hypothe-
`sis also explained a number of other ob-
`servations such as the synthetase-type en-
`zymes that require the simultaneous pres-
`ence of a number of substrates on the en-
`zyme surface before any partial reaction
`occurs (Koshland, '55, '58, '59).
`Some evidence of a different nature
`obtained by Dr. Harvey Levy and Dr.
`Nathan Sharon (Levy et aE., '59a) supports
`the postulated flexibility of the active site
`and the specific modification of protein
`conformation by the substrate itself. This
`evidence grew out of temperature studies
`on the enzyme myosin, the data for which
`are shown in figure 1. Recording the rate
`data on an Arrhenius plot gives a straight
`line if the activation energy and the PZ
`factor are constant. Such is observed to
`be the case for the myosin-catalyzed hy-
`drolysis of ATP, which is linear over the
`experimental range of 30" to 0°C. It is
`to be noted, however, that both ATP in the
`presence of dinitrophenol (DNP) and ITP
`show a pronounced curvature. This is
`not a function of the experimental error or
`of the method of graphing. Careful repeti-
`
`5
`30 25 20 16
`10
`TEMPERATURE("C)
`Fig. 1 Arrhenius plot for myosin with ATP,
`ITP, and ATP in the presence of DNP. Solution
`contained 0.01 M MgClz, 0.05 M Tris, 0.1 M
`KCI pH 7.3, and 0.005 M nucleotide triphosphate.
`tion of the experiment and replotting of
`the data with various abscissa and ordi-
`nant ratios leads to the same conclusion.
`Actually, the two curves can be very well
`approximated by straight lines at each of
`the extremes of temperature, and these
`straight lines intersect in each case very
`near 16°C. This extrapolation should not
`be taken to imply that an abrupt discon-
`tinuity exists. However, the rather good
`agreement of the straight lines over a
`considerable range of temperature does
`tend to indicate that a shift occurs from a
`process at the higher temperature having
`an activation energy of -12 kcal to a
`process at lower temperatures having an
`activation energy of -25 kcal.
`Curves that show such a change in ac-
`tivation energy have been observed before,
`and a number of different explanations
`have been proposed. Dixon and Webb
`('58) have summarized these as follows.
`(a) There is a phase change in the sol-
`vent. This idea is supported by the ob-
`servation that the point of inflection ap-
`parently occurs in the same place for a
`number of different enzyme systems and
`by the existence of a transition point near
`0°C. where a phase change in the solvent
`water is known to occur.
`(b) There are two parallel reactions
`with different active centers. Dixon and
`Webb ('58) pointed out that such a mech-
`anism could explain Arrhenius plots that
`
`

`

`248
`
`D. E. KOSHLAND, JR.
`an enzyme having activation energies simi-
`lar to those of myosin with ITP are shown
`in figure 3. Since the observed rates for
`such a consecutive-step mechanism are
`seen to be a very good rough approximation
`for the ITP-myosin curve, it is clear that a
`consecutive-step mechanism cannot be ex-
`cluded simply by the argument that the
`transition observed experimentally is too
`abrupt.
`(d) The enzyme exists in two forms
`having differing activities. This sugges-
`tion was originally advanced by Sizer ('43)
`and can be formalized as shown in equa-
`tions ( 3 ) , (4), and (5). In this mechan-
`ism the protein changes from a high-tem-
`perature form, El, to a low-temperature
`form, E,, in a reversible manner over a
`fairly narrow temperature range. The
`velocity of the over-all reaction is given by
`equation (6). This mechanism can give
`Arrhenius plots that are either concave
`
`are concave upward but could not explain
`those that are concave downward.
`(c) The enzymic process involves two
`successive reactions having different tem-
`perature coefficients. The idea that such
`a shift from one rate-mastering step to
`another was responsible for the transition
`in the Arrhenius plot of physiological proc-
`esses was originally suggested by Croz-
`ier ('24) but was later seriously attacked
`by Burton ('39) and others. Burton showed
`that, in some systems having activation
`energies of the order of those found by
`Crozier, a consecutive-step mechanism
`would not lead to so abrupt a transition
`as was observed in the experimental cases.
`Burton was dealing with a system involv-
`ing two different enzymes, however, and
`the mathematics accordingly is not pre-
`cisely the same as that for two consecu-
`tive steps on a single enzyme surface.
`The sequence of events in the latter case
`are illustrated by equation ( 1 ) (fig. 2).
`Since the experiments reported here were
`done at enzyme saturation, we need only
`consider the steps in equation ( 1 ) having
`the constants h, and k,. When the kinetics
`for this case are derived by using only the
`steady-state assumption and the condition
`that
`shown in equation (2) is obtained. This
`turns out to be different from the kinetics
`treated by Burton; the observed values for
`
`(ES)l + (ES)B = ET, the relation
`
`-
`
`k 2
`
`E + P
`
`E + 5
`
`k l
`(E$J1 ---+ (ES)p
`
`d(P) - kikZET
`- -
`k , + k
`df
`2
`
`
`
`001 I-
`
`
`33
`
`3.4
`
`3 5
`
`(A) x t ~ - 3
`
`3 6
`
`I
`30
`
`1
`43
`5
`21
`TEMPERATURE(T)
`Fig. 3 Calculated velocities for a consecutive-
`step mechanism of the type shown in equation
`( 1 ) (fig. 2). The true activation energies of kl
`and kz are assumed to be 33 and 7 kcal, re-
`spectively. The absolute magnitudes of the rate
`constants for these two steps are taken from the
`solid lines drawn with these activation energies.
`The rate of appearance of product at any temper-
`ature is obtained from these values by equation
`(2) and is shown by the points on the graph.
`These velocities, which would be the observed
`velocities in an experimental case, give apparent
`activation energies of 12.6 and 29 kcal in two
`rather linear portions of the curve.
`
`kg
`d(P) - k l K l +
`1 + K1
`dt
`Figure 2
`
`

`

`ENZYME FLEXIBILITY AND ENZYME ACTION
`
`249
`
`downward (see fig. 4) or concave upward
`(see fig. 5).
`(e) There is a reversible inactivation of
`the enzyme. Kistiakowsky and Lumry
`('49) explained a transition in the urease
`curve by a reversible inactivation involv-
`ing sulfite. In a sense this is different
`
`3 3
`
`3.4
`(&)x!~-3
`
`3 5
`
`36
`
`Fig. 4 Calculated velocities for the protein
`change mechanism. Theoretical rates of product
`appearance are calculated for the mechanism
`shown in equations (3-5) by using an activation
`energy for kl of 7 kcal and for kz of 33 kcal.
`The AH for the protein transition was taken to be
`49 kcal with the assumption that Ei=Ea at
`16°C. Values for the observed rate at any
`temperature are obtained by solving equation
`(6) (fig. 2 ) when values for kl and kz taken
`from the solid lines in the figure are used.
`
`!I: *-
`
`- +
`u- 10-
`t-
`a
`K
`_I i
`4
`LT
`W >
`
`I G -
`
`0 5
`
`from the previous case because an added
`reagent is involved. In the absence of
`added inhibitor, inactivated enzyme can
`be viewed as a special case of a changed
`enzyme in which h, happens to be zero.
`( f ) There is a discontinuity affecting
`the forward reaction only. This alterna-
`tive is advanced to explain the case of
`fumarase (Massey, '53) that shows a
`transition for the forward reaction but
`none for the reverse.
`By combining the data of figure 1 with
`the literature, it can be shown that all
`these alternatives are unlikely if a tem-
`plate-type specificity is required. Detailed
`arguments will be presented elsewhere
`(Levy, Sharon, and Koshland, unpub-
`lished) but a typical line of reasoning can
`be presented here. It involves the assump-
`tion that a single explanation will suffice
`for all cases in which a transition is ob-
`served in the Arrhenius plot.
`Thus the myosin behavior clearly ex-
`cludes alternatives ( a ) and (e). In each
`of these cases, it is postulated that an
`external change occurs (i.e., a solvent
`change or protein denaturation) that is
`independent of the substrate and hence
`should affect the two substrates in a quali-
`tatively similar manner. It is, of course,
`not necessary that the solvent change af-
`fect the two rates equally, but it could not
`dramatically change the velocity of the
`ITP hydrolysis without making at least
`some change in the ATP rate. A protein
`inactivation would, of course, result in a
`similar effect in the two cases. Alterna-
`tive (f) is excluded in the present case
`since both substrates are operating in the
`forward direction and alternative (b) is
`excluded because the ITP is concave down-
`ward. Alternative (c) is compatible with
`the ITP data but is inconsistent with the
`fumarase data (Massey, ' 5 3 ) .
`This leaves alternative (d), the protein
`change mechanism. If this enzyme obeys
`template-type specificity, this alternative
`can be excluded by the same arguments
`used to exclude alternatives ( a ) and (e).
`Whether the new form of the protein is
`active or inactive, a transition for one
`substrate should be accompanied by a
`transition for the other. This conclusion
`is predicated on two experimental condi-
`
`loo\
`50
`
`x;
`Et-E2
`
`AH.49 kcal
`
`EZ+ P EA = 7 kwl
`( E S ) ~ ~
`
`*\*
`
`f 4 *x:*
`
`

`

`250
`
`D. E. KOSHLAND, JFi.
`
`tions: (i) that we are dealing with satur-
`ated enzyme and (ii) that the two sub-
`strates are competing for the same site.
`Both these conditions are true in the pres-
`ent case. The enzyme saturation was
`shown by the usual type of kinetics, and
`the competition for the same active site
`was shown by the hydrolysis of ATP'z in
`the presence of nonlabeled ITP. It is of
`interest that in the latter experiment the
`ITP, which by itself is hydrolyzed ten
`times as fast as ATP, is not hydrolyzed at
`all in the presence of ATP because of the
`far greater binding affinity of the latter
`substrate.
`For those who distrust complicated ar-
`guments, it may be worth emphasizing
`that the qualitative difference in Arrhen-
`ius plots of ITP and ATP is by itself dif-
`ficult to explain by a template-type spe-
`cificity. The 6 position in the purine ring
`is many bonds removed from the bond
`being split in the enzymic reaction. The
`-OH
`and -NH2 groups both have an un-
`shared pair of electrons, both have a hy-
`drogen available for hydrogen bond for-
`mation, and both have very similar vol-
`umes. A change in the template that
`dramatically affects the decomposition of
`the enzyme-ITP complex without visible
`effect on the enzyme-ATP complex is dif-
`ficult to imagine.
`The protein change mechanism, how-
`ever, can explain the existing facts if a
`flexible active site having different con-
`formations in the presence of ITP and
`ATP is assumed. A schematic illustration
`of such an interaction is shown in figure
`6. At the higher temperature the enzyme
`exists in a rather loose structure, with a
`
`Fig. 6 Schematic illustration of induced-fit
`behavior to explain the temperature dependence
`of ITP and ATP in myosin hydrolysis.
`
`chain of considerable length leading from
`the active site. It is assumed that there
`is a group X in this chain that is strongly
`attracted to the 6-amino group of ATP
`but has no affinity for the 6-hydroxy
`group of ITP. This attraction leads to a
`constriction of
`the amino acids at the
`active site when ATP is adsorbed and this
`constriction is presumed to make the active
`site less favorable for enzyme action. Ac-
`cordingly, the ITP is hydrolyzed far more
`rapidly than ATP. When the enzyme is
`cooled, a change occurs leading to a coiling
`in an area adjacent to, but not immediately
`at, the active site. This coiling, however,
`shortens the chain leading from the active
`site so that it no longer has the flexibility
`at the lower temperature that it had near
`25°C. This coiling, which proceeds rather
`abruptly in the region of 16"C., therefore
`causes a marked decrease in the freedom
`of the ITP active site with concomitant
`change in enzyme activity. However, the
`same coiling has essentially no effect on
`the activity of ATP since the 6-amino
`group attraction already has caused a con-
`striction of the active site. Thus the fold-
`ing of the protein (perhaps into an a-helix)
`has different effects on ITP and ATP hy-
`drolyses because the interaction of the
`two substrates with the protein does not
`result in identical protein conformations.
`This model is illustrative and is de-
`lineated here mainly to show that the in-
`duced-fit type of behavior can explain why
`two substrates can have qualitatively dif-
`ferent temperature dependences. Never-
`theless, there is considerable added infor-
`mation that it is a good model for the
`action of myosin. First, the difference be-
`tween ITP and ATP does not establish
`whether the -OH or the -NH,
`group is
`playing the positive role. In further stud-
`ies, tripolyphosphate gave a curve similar
`to that of ITP, showing that it is the ab-
`sence of the -NH2 group rather than the
`presence of the -OH
`group that is re-
`sponsible for the difference. That the tri-
`polyphosphate curve shows a transition at
`16°C. similar to ITP further establishes
`that it is not a difference in size between
`-NH,
`and -OH
`that is responsible for
`the transition. Second, the role of DNP
`in this system can be readily explained
`with this model. DNP makes the hydroly-
`
`

`

`ENZYME FLEXIBILITY AND ENZYME ACTION
`
`25 1
`
`extrapolation, therefore, would be that it
`is the enzyme at 0" that is the denatured
`form! This apparent dilemma is only
`semantic but it emphasizes an important
`conclusion; i.e., that the folding of the
`protein is not optimum at any tempera-
`ture. A reagent that could improve this
`folding would therefore catalyze the en-
`zyme activity even though it had no func-
`tion in bond breaking or electron polariza-
`tion. Such a role was proposed for DNP
`in the preceding model, and such a role
`may well be played by other activators in
`this and other systems.
`A particularly intriguing activator of
`the myosin system is actin. This protein
`is not only essential for contraction but
`also greatly accelerates the rate of myosin-
`ATPase activity in the presence of Mg++.
`Chappell and Perry ('55) originally ob-
`served a DNP-actin competition, and this
`was confirmed in our
`laboratory.
`It
`seemed logical to expect that part of the
`actin function, therefore, would be to
`loosen the 6-NHz protein bond in the same
`way as postulated for DNP. To support
`this possibility, Levy and Sharon meas-
`ured the actin-ATP-myosin temperature
`curve, and the results are shown in figure
`7. The similarity to the ITP and DNP-ATP
`
`sis of ATP have the same qualitative ap-
`pearance as that of ITP, which suggests
`that it is competing with the 6-NH2 group
`of ATP for the group X in the protein
`chain. The presence of DNP as the phe-
`noxide ion at pH 7 supports this idea. The
`DNP and 6-NHz will both, therefore, be
`in their basic forms in this solution, where-
`as the 6-OH of ITP will be present as the
`uncharged acid. The role of DNP is to
`release the constriction caused by the at-
`traction between the 6-NHa and group X.
`Moreover, the model readily explains why
`DNP does not activate the myosin-cata-
`lyzed hydrolysis of ITP (Greville and Need-
`ham, '55). Since there is no attraction of
`the -OH
`group for the side chain, the
`DNP cannot release this inhibition, and
`hence there is no effect on the rate. Third,
`the model allows an explanation of a wide
`variety of apparently unrelated and con-
`fusing phenomena on a simple basis. For
`example, the observed activations of myo-
`sin-ATPase activity by low concentrations
`of p-chloromercuribenzoate and ethylene-
`diaminetetraacetic acid (EDTA) also can
`be rationalized on the basis of a competi-
`tion with the 6-NHz for the group X on the
`protein chain. Moreover, EDTA activates
`ATP but has a negligible effect on ITP
`(Bowen and Kenvin, '54) as would be pre-
`dicted by this mechanism.
`Although theories that explain existing
`anomalies are pleasant, those that suggest
`new experiments are pleasanter. I should
`like, therefore, to conclude with a descrip-
`tion of experiments that give us some clue
`to the role of enzyme activators and per-
`haps even of hormones. Let us look for a
`moment at figure 4 in the light of the de-
`naturation theory of protein '%breaks" and
`ask which form of the protein is denatured.
`If we start with the enzyme at 0°C. and
`extrapolate along the straight line to the
`expected rate at 30", we will see that this
`is much higher than the rate actually ob-
`served for this mechanism. Our conclu-
`sion from this comparison would be that
`the high-temperature form of the enzyme
`is clearly the dentured form. If we start
`our studies at 30"C., however, and ex-
`trapolate linearly from the initial rates ob-
`served there to O " , we will find that the
`observed rate at zero is far less than ex-
`pected. Our conclusion from this second
`
`3 3 0
`
`(+) i o - ~
`
`3 5 0
`
`3 6 0
`
`340
`T K
`Fig. 7 Arrhenius plot for the hydrolysis of
`ATP by myosin ( 0 ) and actomyosin ( 0 ) . Myo-
`sin conditions given in figure 1. Actomyosin
`conditions were: 0.025 M KC1, 0.005 M MgC12,
`0.025 M Tris pH 7.3, 0.002 M ATP.
`
`

`

`252
`
`D. E. KOSHLAND, JR.
`
`curves is striking. Again, the rate is ac-
`celerated when actin is added and a tran-
`sition from one protein form to another is
`indicated, the transition seeming to occur
`at about 16°C. The concordance is at
`least a strong indication that one of the
`activating properties of actin is its effect
`on the conformation of the myosin active
`site (Levy et al., '59b).
`The previous discussion has involved
`two activators of the myosin system. One
`of these, actin, is a protein just as are a
`number of the already-known hormones
`like ACTH, and the other is a nonprotein
`chemical, DNP, which in many ways is
`analogous to thyroxine. It is tempting to
`speculate, therefore, that the hormones
`may act not as coenzymes or intermedi-
`ates but as materials that favorably alter
`the conformation of the active sites of the
`appropriate enzymes.
`The accumulated evidence indicates
`that some enzymes must have a flexible
`active site that plays a key role in enzyme
`action. Whether all enzymes must show
`this flexibility at the active site is a logical
`question. It is conceivable that they do
`since, as mentioned, an induced-fit type of
`behavior can explain the failure of both
`larger and smaller molecules to react.
`However, general studies on proteins in-
`dicate a rather wide range in flexibility.
`The most logical prediction at this time
`would seem to be that many enzymes will
`have a flexible active site and hence will