`
`The Key-Lock Theory and the Induced Fit Theory
`
`Daniel E. Koshland, Jr.
`
`It is a great pleasure for me to contribute to this symposium
`honoring the great scientist Emil Fischer. My graduate thesis
`required me to synthesize [1-14C]glucose, which introduced me
`to the famous Fischer-Kiliani synthesis of glucose and man-
`nose from arabinose and HCN. [1] I was also particularly in-
`trigued with his classic key-lock (or template) theory of enzyme
`specificity,[2, 3] which like all great theories seemed so obvious
`once one understood it.
`This symposium in his honor allows me to pay tribute to
`Fischer’s great contributions to biochemistry varying from nat-
`ural products chemistry to the key-lock theory, to review some
`of the history and significance of our induced fit theory, to
`illustrate the ramifications of those theories in our present era of
`protein-ligand interactions, and to discuss recent work in our
`laboratory which is helping to clarify conformational changes
`and their function. These theories have assumed again a central
`role in modern health research where the need for drug design
`requires taking into account the complementarity of fit of
`Fischer’s principle and the flexibility and regulatory implica-
`tions of the induced fit theory.
`The induced fit theory is no more a refutation of Fischer’s
`key-lock principle than the Heisenberg atom was of the Bohr
`atom or the modern DNA sequences are of the one gene-one
`enzyme hypothesis. A new theory must explain all the existing
`facts that pertain to it at the time of its enunciation. Gradually
`the new theory becomes accepted and then acquires anomalies
`due to the new facts uncovered after its enunciation. That in
`turn generates a newer theory which elicits new techniques to
`test it and its predictions. These new techniques then uncover
`facts which eventually require further new theories and so on.
`The new theories are built on components of the old principles.
`It is said that each scientist stands on the shoulders of the giants
`who have gone before him. There can be no more honored place
`than to stand on the shoulders of Emil Fischer.
`
`Limits of the Key-Lock Theory
`
`My first inkling that the Fischer key-lock model needed
`modification really arose from my consideration of the role of
`
`[*] Prof. Dr. D. E. Koshland, Jr.
`Department of Molecular and Cell Biology
`University of California
`229 Stanley Hall 3206, Berkeley, CA 94720 (USA)
`Telefax: Int. code + (510) 643.6386
`[**I Based on a commemorative lecture at the European Reseah Conference
`“Supramolecular Chemistry: 100 Years “Lock-and-Key” Principle in Mainz
`on August 12, 1994.
`
`water in biological reactions. I was preparing a lecture for
`a scientific meeting and decided to consider why some
`proteins were kinases and others ATPases. The more I
`thought about the protein, the more astonishing it seemed that
`water could be prevented from reacting at the active site
`of a kinase.
`In hexokinase, which I took as a typical kinase, the OH group
`of water was known to be as good a nucleophile as the OH
`group of a sugar. If glucose is bound very tightly then it could
`exclude water, and a basic group on the protein would generate
`a glucosyl oxyanion nucleophile which could attack the ATP.
`But glucose would not normally saturate the site and could in
`many physiological circumstances fall to very low levels. Water,
`at 5 5 M , would fill up an empty site, and therefore water would
`be constantly competing with glucose in the nucleophilic attack
`on ATP. The existence of kinases in the absence of substrate
`or with only partially filled template type active sites would
`result in great ATPase activity and an enormous waste of
`energy.
`Once I started thinking along these lines other anomalies
`came to mind. One example was “noncompetitive inhibition”,
`which was explained by saying that the inhibitor blocked
`enzyme action but did not affect the binding of the
`substrate. No key-lock concept was available to explain such a
`result.
`The key-lock (or template principle) could explain why
`smaller sugars might not react: they would not be attracted to
`the active site strongly enough to form significant amounts of
`the ES complex. However, we found that cc-methylglucoside was
`not a substrate but was a tightly bound competitive inhibitor of
`the enzyme. Thus it was tightly bound, could fit into the site, had
`the right chemical stereochemistry, but did not react. (“Sub-
`strate analog” was used for those chemicals whose chemistry is
`similar to a substrate but fail to react on the enzyme’s surface as
`in a-methylglucoside, a substrate analog of the enzyme amylo-
`maltase.)
`Other reactions raised the same question of smaller chemical-
`ly logical molecules that nevertheless did not react. And
`there were also cases in which a bigger substrate analog
`did not react. As another example, we found that cyclo-
`hexaamylose was an inhibitor of
` (an enzyme that
`cleaved glucosyl bonds in long amylose chains). One could
`try to explain this on the basis of the key-lock principle
`by saying that the cyclic amylose was too big and couldn’t bind,
`but we showed it did, in fact, bind (and tightly) but failed to
`react.[4]
`
`Angew. Chem. Inl. Ed. Engl. 1994, 33, 2375-2378
`
`(C) VCH Verlagsgesellschaft mbH, D-69451 Weinheim, 1994
`
`0570-0833/94/2323-2375 $ 10.00 + .25/0
`
`2315
`
`TEVA - EXHIBIT 1022
`
`
`
`REVIEWS
`
`Introduction of the Induced Fit Theory
`
`So the induced fit theory [5] was proposed in the following
`terms “a) the precise orientation of catalytic groups is required
`for enzyme action, b) the substrate causes an appreciable change
`in the three-dimensional relationship of the amino acids at the
`active site, and c) the changes in the protein structure caused by
`the substrate will bring the catalytic groups into the proper
`alignment, whereas a nonsubstrate will not.”
`
`Key-Lock Theory and the Induced Fit Theory
`
`Pictures to illustrate this concept and how it could explain the
`previous anomalies are shown in Figure 1 taken from papers
`published at the time. [6] The theory of Emil Fischer was
`deep in the hearts of scientists and journal editors, so I had great
`difficulty getting the original ideas published or convincing
`skeptics, but we did obtain more evidence from my own labora-
`tory, and soon others joined in. One of the predictions that
`results from the assumption of a flexible enzyme, namely that a
`small nonreactive molecule could make up for a structural defi-
`ciency in a nonsubstrate (Fig. 2), was established for us by two
`
`a
`
`Fig. 2. Activator molecules can, according to the flexible model of enzyme action,
`help to make a deficient molecule act as a substrate by altering the shape of the
`enzyme. For example, in the case of a molecule (unshaded) that by itself is too small
`to induce the proper alignment of catalytic groups, A and B [shown in a)]. a second
`molecule (shaded) can bind immediately adjacent to the deficient molecule [shown
`in b)] or (not shown here) to a more distant site. thereby inducing a stable shape with
`the proper ahgnment of catalytic groups.
`
`laboratories. Sols et al. showed that xylose, a pentose (similar to
`glucose but lacking the 6-CH2OH group), made hexokinase a
`better ATPase,[7a] and Murachi et a1.[7b] showed that the non-
`substrate for trypsin, glycine ethyl ester, could react appreciably
`if ethylamine was added to the incubation mixture. These “reg-
`ulatory” molecules which did not themselves undergo chemical
`changes could induce the further conformational changes in an
`enzyme needed for reaction (as illustrated in Fig. 2).
`These indirect chemical assays added to the credibility of the
`hypothesis, but we needed direct evidence for the predicted in-
`duced conformational change in the protein (a proof which was
`easy later when protein crystallography became available). So
`Yankeelov and I said we must get a result with protein reactivity
`
`Fig. 1. Schematic model of the induced fit mechanism. Black lines indicate protein
`chains containing catalytic groups A and B and binding group C. Upper left:
`substrate and enzyme dissociated. Upper right: substrate with induced change of
`protein chains to bring A and B into proper alignment for reaction. Lower left:
`bulky group added to substrate prevents proper alignment of A and B. Lower right:
`deletion o f a group eliminates buttressing action on the chain containing A, so the
`thermodynamically stable complex has incorrect alignment of A and B.
`
`Daniel E. Koshland was born in New York City in 1920. He earned his B. S. degree from the
`University of California, Berkeley, in 1941, and Ph. D from the University of Chicago in 1949.
`After two postdoctoral years at Harvard he joined the staff of Brookhaven National Laborato-
`ry, and later also of Rockefeller University. In 1965 he joined the faculty of the University of
`California, Berkeley, where he is currently Professor of Biochemistry and Molecular Biology.
`He became Editor of Science in 1985. Among his honors are the National Medal of Science, the
`Edgar Fahs Smith and Pauling Awards of the American Chemical Society, the Rosenstiel
`Award of Brandeis University, the Waterford Prize, and the Merck Award of the American
`Society of Biochemistry and Molecular Biology. Included in his fields of interest are the role of
`conformational changes in enzyme regulation and the elucidation of the cutulytic power of
`enzymes. He demonstrated that bacteria have short-term memory and that purified mammalian
`cell lines show rudimentary memory. His recent w o r k has emphasized the chemical mechanism
`of short-term and long-term memory, and the structure-function relationship of receptors.
`
`2376
`
`Angew Chem. Int. Ed. Engl. 1994, 33. 2375-2378
`
`
`
`D. E. Koshland, Jr.
`
`that tested the key-lock template hypothesis and induced fit.
`We argued that adding a ligand to a template type enzyme can
`bury groups but it cannot expose them, whereas an induced fit
`conformational change could bury some groups and expose
`others. We picked the enzyme phosphoglucomutase (whose re-
`action had similarities to hexokinase, and thus we expected it to
`be an induced fit enzyme)[8] and used the reactivity of its SH
`group as a test. The experiment illustrated in Figure 3 gave the
`result we wanted. [8] Ligand binding induced the exposure of an
`SH group, a result incompatible with the key-lock theory. We
`likened it to the flexibility of a “hand in glove”, which included
`Fisher’s idea of a fit but added the flexibility concept.
`
`Fig. 3. Schematic illustration of flexibility in the action of phosphoglucomutase.
`The upper part of the figure represents the enzyme molecule in the absence of
`substrate. The lower part of the figure represents the change in conformation lead-
`ing to exposure of -SH and burying of X, Y, Z, and W.
`
`Further support came when the structures of lysozyme[9] and
`ribonuclease[10] were published because there were definite con-
`formational changes; however, these were small and did not
`impress many. (Many biologists forgot that C-C and C-O
`bonds are only 1.5
` long, so small changes can easily disrupt a
`catalytic alignment needed to catalyze changes in the bonds.)
`Then Steitz et al. with carboxypeptidase[11] and Steitz et al. with
`hexokinase[12] showed conformational changes that were
`breathtakingly large and highly convincing. Steitz showed that
`the engulfing of the substrate glucose by hexokinase occurred
`precisely as the induced fit predicted, thus giving visual proof
`that ligand-induced conformational changes were real and sig-
`nificant.[12]
`
`REVIEWS
`
`Today almost every enzyme has been shown to undergo sig-
`nificant ligand-induced changes. A recent review by Gerstein,
`Lesk, and Chothia [13] divides these changes into “hinge do-
`main” and “shear” motions and lists 42 enzymes that illustrate
`major conformational changes. The enzymes that show the least
`conformational changes are the hydrolases such as the proteases
`and nucleases--and they are precisely those one might expect to
`fall in this category, since they do not need to exclude water. The
`finding of extensive conformational changes in many enzymes is
`logical, since most enzymes exist in a cytoplasm with many
`pathways that contain many smaller substrate analogs, for ex-
`ample trioses, which must be prevented from reacting at sites of
`larger analogous substrates, for example hexoses. If the
`specificity failed to exclude smaller analogs, poor yields and bad
`side reactions would occur.
`The question then arises as to how big the conformational
`changes have to be in order to be considered “significant”. Some
`recent evidence indicating some answers to this problem is dis-
`cussed below.
`
`Isocitrate Dehydrogenase and Small Conformational
`Changes
`
`We have recently been studying isocitrate dehydrogenase to
`obtain some clues to the size and significance of conformational
`changes. We found, for example, that the enzyme was inactivat-
`ed by phosphorylation, [14] but this phosphorylation, unlike the
`case of glycogen phosphorylase, [15] involved phosphorylation
`right at the active site with little resulting change in conforma-
`tion of the protein. [I6 19] We also found very little change in
`conformation induced by the substrate isocitrate on binding to
`the protein and were about to conclude that the enzyme was one
`of those that approximated the Fischer key-lock model. How-
`ever, we did one more experiment and tested the protein in the
`presence of the product, a-keto glutarate.[20] In that case the
`ligand-induced conformational changes were wide spread.
`Many atoms moved though each movement was rather small.
`The protein did not tit into the “hinge domain” category of
`Gerstein, Lesk, and Chothia nor even into the “shear” category,
`but rather into what might be called a “spider web” category,
`that is, small interconnected changes occurring over an exten-
`sive surface. The changes in each atom were less than an
`angstrom but many atoms moved, which suggested that sub-
`tleties in alignment were capable of turning an enzyme off or on.
`We have also measured the changes in the aspartate receptor
`of chemotaxis in collaboration with Sung Hou Kim et a1.[21]
`This case fits the shear model more closely, as we postulate that
`the small changes at the binding site for aspartate can cause a
`sliding of one helix past another.[22-24] The changes that are
`generated in the cytoplasmic domain are relatively small may-
`be an average change of 0.5
` the conformational change
`is transmitted from one side of the dimer to the other. In
`addition we have shown that the receptor shows negative coop-
`erativity [25, 26] in which binding of the first aspartate to a dimer
`completely blocks aspartate binding to a second aspartate site.
`The two sites are initially identical. but the ligand-induced
`changes in the second site reduce the size of the second site so it
`becomes too small to bind aspartate. The changes are quite
`
`Angew. Chem. Int. Ed. Engl. 1994, 33, 2375-2378
`
`2377
`
`
`
`REVIEWS
`
`Table 1. Distances between side chains in binding sites in the Salmonella aspartate
`receptor-ligand binding domain.
`
`Amino acids
`
`Separation [A]
`in unbound
`receptor [a]
`
`Separation [A]
`in empty site of
`Asp-bound receptor[a]
`
`Reduction
`in distance
`
`Ser-68, Thr-154
`Tyr-149, Arg-73
`Tyr-149, Arg-64
`Phe-150. Arg-73
`Ser-68. Arg-69
`
`8.9
`6.9
`4.1
`4.8
`7.4
`
`8.1
`6.0
`3.2
`3.5
`6.6
`
`[a] Distance between closest non-hydrogen atoms.
`
`0.x
`0.9
`0.9
`1.3
`0.8
`
` (Table 1) but are enough to prevent binding of the
`small in
`aspartate molecule.[26]
`Our conclusion is that big movements are important but so
`are small ones. The important feature from the induced tit the-
`ory is that the alignment of catalytic groups and binding groups
`must be optimized for the transition state, and the attainment of
`the state is unfavorable energetically unless it is supplied with
`the energy of the substrate binding. If the protein movements
`were easy to attain, they would occur spontaneously often
`enough to have little effect on catalysis. However, a small move-
`ment can also be energetically unfavorable, as in the shift of a
`ferrous atom 0.7
`into and out of the plane of the heme in
`hemoglobin. [27] When the small movement needed for catalysis,
`in the case of an enzyme, is generated by the binding of the
`substrate, enzyme action occurs.
`A second conclusion is that the conformation of the protein
`is undoubtedly selected during evolution to optimize both the
`unliganded state and the liganded state. Allosteric sites are often
`distant from the active site by 20
` or more, and the conforma-
`tional changes are far larger than can be explained by a distor-
`tion that propagates from one site to another by pure chemical
`torsions. Those nonbonded forces dampen out too rapidly.
`Therefore, the conformational change induced by the substrate
`binding has a long-range effect because it generates and cata-
`lyzes the transition from one evolutionarily selected conforma-
`tion to another.
`
`Summary and Outlook
`
`The basic concept of Emil Fischer’s key-lock theory, which
`explained enzymatic properties of specificity and action for 60
`years, required modification to explain discrepancies such as the
`lack of hydrolytic activity of kinases, noncompetitive inhibition,
`and other apparent inconsistencies. The new theory, the induced
`fit theory, incorporated Fisher’s concepts of the complementar-
`ity of enzyme and substrate but introduced the concept of a
`flexible enzyme, likened to the tit of a hand in a glove. The
`flexible enzyme concept not only explained the discrepancies but
`set the stage for further understanding of regulation,
`cooperativ-
`a n d specificity as described in papers by Pardee,[28]
`ity,
`Monod,[29] and our own laboratory, as well as many others.
`Thus the great work of Emil Fischer lives on in an extension of
`
`Key-Lock Theory and the Induced Fit Theory
`
`the theory and application to new problems of chemistry and
`biochemistry that were impossible to visualize in the 1900s. The
`new studies focus on the importance of conformational changes,
`both large and small, and the manner in which they control
`enzymatic reactions. The findings from modern X-ray crystal-
`lography that essentially all enzymes undergo conformational
`changes induced by substrate binding has made the induced fit
`theory universally accepted in textbooks and by scientists.
`These theories are of increasing importance because of the
`rise in drug-resistant strains of organisms. Computer-assisted
`drug design is what we and many others are now developing to
`prevent the ravages of the new virulent organisms. For that
`purpose the key-lock theory with a relatively rigid enzyme
`would be an easier basis for computer designs, but unfortunate-
`ly the evidence that induced fit theory is closer to reality means
`that computer programs will have to be a little more sophisticat-
`ed. However, the modern computer seems clearly up to the
`challenge, and a rigid enzyme is a good starting point for initial
`assumption. The flexibility can then be built into subsequent
`calculations. Moreover, the flexible enzyme allows binding to a
`“regulatory” or “allosteric” site, which may be a better target
`for drug therapy in many cases. The finding that very small
`changes can “turn on” or “turn off’ an enzyme is very encour-
`aging in this regard.
`
`received financial support from the National Insti-
`This work
`tutes of Health and the National Science Foundation.
`
`[1]
`
`[2]
`[3]
`[4]
`[5]
`[6]
`
`[7]
`
`[8]
`[9]
`
`[10]
`
`[1 1]
`
`[12]
`[13]
`[14]
`[15]
`[16]
`
`[17]
`[18]
`
`[19]
`[20]
`[21]
`
`[22]
`[23]
`[24]
`[25]
`[26]
`[27]
`[28]
`[29]
`
`F. W. Lichtenthaler, Angew. Chem. 1992. 104, 1577; Angew. Chem. Int. Ed. Engl.
`1992, 31, 1541.
`E. Fischer, Ber. Dtsch. Chem. Ges. 1890, 23, 2611.
`E. Fischer, Ber. Dtsch. Chem. Ges. 1894. 27. 2985.
`J. A. Thoma, D. E. Koshland. Jr.. J. Am. Chem. Soc. 1960, 82, 3329.
`D. E. Koshland. Jr., Proc. Natl. Acad. Sci. USA 1958, 44, 98.
`a) D. E. Koshland. Jr., Science 1963. 142. 1533: b) Cold Spring Harbor Symp.
`Quant. Bid. 1963, 28, 473.
`a) G. Dela Fuente, R. Lagunas, A. Sols, Eur. J. Biochem. 1970. 16, 226: b) T
`Inagami, T. Murachi. J Biol. Chem. 1964, 229, 1395.
`J. A. Yankeelov, Jr., D. E. Koshland, Jr.. J. Bid. Chem. 1965. 240, 1593.
`C. C. F. Blake, D. F. Koenig, G. A. Mair. A. C. J. North. D. C. Phillips, V. R.
`Sarma, Nature 1965. 206, 757.
`H. W. Wyckoff, K. D. Hardmann, N. M. Allewell. T. Inagami. L. N. Johnson.
`F. M. Richards, J. Bid. Chem. 1967. 242, 3984.
`T. A. Steitz, M. I. Lu dwig, F. A Guiocho, W. N. Lipscomb. J. Biol. Chem. 1978.
`242, 462.
`C. M. Anderson, F. H. Zucker, T. A. Steitz, Science 1979. 204, 375.
`M. Gerstein, A. M. Lesk, C. Chothia, Biochemistry 1994, 33. 6739.
`D. C. LaPorte, D. E. Koshland, Jr., Nature 1982, 300, 458.
`L. N. Johnson, D. Barford. Annnual Reviews 1993, 22, 199.
`J. H. Hurley. A. M. Dean, J. L. Sohl. D. E. Koshland, Jr, R. M. Stroud. Science
`1990,249, 1012.
`A. M. Dean, D. E. Koshland. Jr., Science 1990. 249, 1044.
`J. H. Hurley, A. M. Dean, D. E. Koshland. Jr., R. M. Stroud. Biochemistry
`1991, 30, 8671.
`B. L. Stoddard, A. Dean, D. E. Koshland, Jr., Biochemistry 1993. 32. 9310.
`B. L. Stoddard, D. E. Koshland, Jr., Biochemistry 1993. 32. 9317.
` D. L. Milligan. W. G. Scott. J. Yeh. J. Jancarik. D.
`M. V. Milburn, G. G.
`E. Koshland, Jr., S.-H. Kim. Science 1991. 254. 1342.
`D. L Milligan. D. E. Koshland. Jr.. Science 1991, 254. 1651.
`B. A Lynch, D. E. Koshland, Jr., FEBS Letters 1992. 307, 3.
`S. L. Mowbray, D. E. Koshland. Jr., Cell 1987, 50, 171.
`D. L. Milligan, D. E. Koshland, Jr., J. Bid. Chem. 1993. 268, 19991.
`H.-P. Biemann, D. E. Koshland, Jr., Biochemistry 1993, 33, 629.
`M. Perutz, Proc. Roy. Soc. Lond. Ser. B. 1990, 208, 135.
`J. Gerhart, A. Pardee, J. Bid. Chem. 1962, 237, 891.
`J. Monod, J. Wyman, J. P. Changeux. J. Mol. Biol. 1965. 12. XX.
`
`2378
`
`Angew Chem. Int. Ed. Engl. 1994, 33, 2375-2378