Acc. CTwm. Res. 1996, £9, 331-339
`
`331
`
`Carbonic Anhydrase: Evolution of the Zinc Binding Site by
`Nature and by Design
`
`Department o£ Chemistzy, University of Pennsylvania, Philadelphia, Pennsylvania 19104 6323
`
`DAVID W. CHRISTIANSON*
`
`Department of Biochemistry, Box 3711, Duke University Medical Centez; Durham, North Carolina 27710
`
`Received December 18, 1995
`
`CAROL A. FIERKE*
`
`Ever since its discovery at the University of Penn-
`sylvaniaJ and the University of Cambridge2 more than
`60 years ago, the zinc enzyme carbonic anhydrase has
`occupied a prominent position at the frontiers of
`biochemistry, medicinal chemistry, and protein engi-
`neering. Seven genetically-distinct forms of this en-
`zyme (known as isozymes I-VII) have evolved in
`numerous tissues and cellular locations, and each
`contains a catalytically-obligatory zinc ion.3,4 This
`diversity reflects the ubiquitous biological requirement
`for rapid hydration of carbon dioxide to yield bicarbon-
`ate ion plus a proton.5-9 Although a deceptively
`simple reaction, the chemical and structural aspects
`of this mechanism have only recently been delineated
`for an isozyme found in the red blood cell, carbonic
`anhydrase II (CAII) (Figure 1). The three-dimensional
`structure of CAIIm has stimulated research probing
`the determinants of the substrate association site and
`the pathway the product proton traverses through the
`enzyme active site. On the basis of phylogenetic
`comparisons,3,4 the substrate and zinc binding sites
`are mainly conserved among all catalytic carbonic
`anhydrase isozymes found in n~ammals. However, the
`trajectory of catalytic proton transfer to bulk solvent
`has diverged during the evolution of the seven
`isozymes.
`We begin this Account with a brief review of the
`CAII mechanism, emphasizing recent developments
`(see previous Accounts for additional information6,r).
`Notably, more X-ray crystallographic and enzymologi-
`cal studies have been performed on CAII and its site-
`specific variants than any other metalloenzyme, and
`these studies uniquely illuminate the molecular de-
`tails of catalysis. Importantly, this work provides an
`elegant example of how the complementary methods
`
`After receiving A.B., A.M., and Ph.D. degrees in Chemistry from Harvard University,
`David W. Christianson moved to the University of Pennsylvania in 1988 where he is
`currently Associate Professor of Chemistry. While at Penn he received a Searle
`Scholar Award, the Young Investigator Award from the Office of Naval Research, an
`Alfred P. Sloan Research Fellowship, and a Camille and Henry Dreyfus Teacher-
`Scholar Award. As a protein crystallographer, Christianson has spent more than a
`decade studying structure-function relationships in metalloenzymes such as carbonic
`anhydrase, and a 1989 Account outlines his work on the related zinc enzyme
`carboxypeptidase A.
`Carol A. Fierke received her B.A. and Ph.D. degrees from Carleton College and
`Brandeis University, respectively. After a postdoctoral fellowship in chemistry at
`Pennsylvania State University she joined the faculty of the Duke University Medical
`School where she is currently Associate Professor of Biochemistry and Chemistry
`as well as a member of the Duke University Cancer Center. While at Duke she
`received an American Heart Association Established Investigator Award, a David
`and Lucile Packard Foundation Fellowship, and an American Cancer Society Junior
`Faculty Research Award. During her career she has investigated molecular
`determinants of biological catalysis in metalloenzymes such as carbonic anhydrase
`and ribonuclease P.
`
`of molecular biology and structural biology can be used
`to probe the structure, function, and stability of a zinc
`binding site in a metalloenzyme. Additionally, these
`studies set a useful foundation for understanding the
`evolution of carbonic anhydrase into noncatalytic
`biological roles, such as signal transduction. For
`example, in the nervous system an extracellular, CAII-
`like domain of receptor protein tyrosine phosphatase
`f! binds an axonal cell recognition molecule (contactin)
`important for neuronal development and differ-
`entiation. J J- J 5 Intriguingly, the putative zinc binding
`site of this domain has partially evolved away from
`that found in CAII.
`We then review recent progress in the "directed
`evolution"--i.e., the structure-based redesign--of the
`CAII zinc binding site, following Nature’s example. We
`describe novel structural determinants of protein-
`metal affinity and the chemical reactivity of zinc-
`bound solvent. Not only does this work represent the
`first molecular dissection of structure-function rela-
`tionships in a protein-zinc binding site (and therefore
`serves as a paradigm for the design of de novo metal
`sites in other proteins), it also sets the foundation for
`the development and optimization of CAII as a metal
`ion biosensor. Remarkably, this ubiquitous metallo-
`protein can be engineered and exploited as a sensitive
`tool for analytical chemistry and biotechnology.
`
`(1) Stadie, W. C.; O’Brien, H. J. Bio]. Chem. 1933, 103, 521-529.
`(2) Meldrum, N. U.; Roughton, F. J. W. ~. Physiol. 1933, 80, 113-
`142.
`(3) Tashian, R. E. Adv. Genet. 1982, 30, 321-356.
`(4) Hewett Emmett, D.; Tashian, R. E. Moi. Phyiogenet. ~voi. 1996,
`5, 50-77.
`(5) Coleman, J. E. J. Biol. Chem. 1967, 242, 5212-5219.
`(6) Bertini, I.; Luchinat, C. Ace Chem. Res. 1983, 16, 272-279.
`(7) Silverman, D. N.; Lindskog, S. Acc. Chem. Res. 1988, 21, 30-36.
`(8) Lindskog, S.; Liljas, A. Cur±: Op#~. Ntruct. Biol. 1993, 3, 915-
`920.
`(9) Silverman, D. N. Methods ~nzymol. 1995, 249, 479-503.
`(10) Liljas, A.; Kannan, K. K.; Bergsten, P. C.; Waara, I.; Fridborg,
`K.; Strandberg, B.; Carlbom, U.; Jarup, L.; Lovgren, S.; Peter, M. Nature
`NewBiol. 197Z, 235, 131-137.
`(11) Krueger, N. X.; Saito, H. Proc. Natl. Acad NcL {ZN.A. 199Z, 89,
`7417-7421.
`(12) Barnea, G.; Silvennoinen, O.; Shaanan, B.; Honegger, A. M.;
`Canoll, P. D.; D’Eustachio, P.; Morse, B.; Levy, J. B.; LaForgia, S.;
`Huebner, K.; Musacchio, J. M.; Sap, J.; Schlessinger, J. Mol. Cell. Biol.
`1993, 13, 1497-1506.
`(13) Levy, J. B.; Cannol, P. D.; Silvennoine, O.; Barnea, G.; Morse,
`B.; Honegger, A. M.; Haung, J. T.; Cannizzaro, L. A.; Park, S. H.; Druck,
`T.; Huebner, K.; Sap, J.; Ehrlich, M.; Musacchio, J. M.; Schlessinger, J.
`~L Biol. Chem. 1993, 26& 10573-10581.
`
`(14) Barford, D.; Jia, Z.; Tonks, N. K. Nature St±~ct. Biol. 1995, 2,
`1043-1053.
`(15) Peles, E.; Nativ, M.; Campbell, P. L.; Sakurai, T.; Martinez, R.;
`Lev, S.; ClaW, D. O.; Schilling, J.; Barnea, G.; Plowman, G. D.; Grumet,
`M.; Schlessinger, J. Cell 1995, 82, 251-260.
`
`S0001-4842(95)00123-3 CCC: $12.00
`
`© 1996 American Chemical Society
`
`Akermin, Inc.
`Exhibit 1016
`Page 1
`
`

`

`332 Acco CT~em. Res., Vol. 29, No. 7, 1996
`
`C7~ristianson and FierCe
`
`o
`
`Zn2÷
`
`H~s 94
`
`His 119
`H is-96
`
`~Z /
`
`H2°
`
`HC03"
`
`Zrl2+
`
`. ./.\.I
`
`Hm 119
`His-16
`
`H~s 94
`
`.. .
`
`His I~9
`His-96
`
`His 94
`
`(a)
`
`(b)
`
`H
`
`HIS 94
`
`HIS 119
`His-96
`
`H* to buffer via His-64 shuttle
`
`Figure 1. Summary of the CAII mechanism. Zinc bound hydroxide attacks the carbonyl carbon of CO2 to form zinc bound bicarbonate.
`The initial mode of bicarbonate binding (a) may reflect the structure of either a discrete intermediate or the transition state.
`Bicarbonate may then isomerize (b), representing either a productive or a nonproductive complex. Following the exchange of a water
`molecule for zinc bound bicarbonate, a proton is transferred flom zinc bound water to solvent via His 64 to regenerate the zinc
`hydroxide species.
`
`~:.~,....~.." ~_,, ~.~(cid:128) ~.~
`
`,.
`
`"i; li’t’4i " " ’.’" xi~" ~li~°:1;/
`
`,..,,.; .....
`! .... ....
`
`Figure 2. CAII active site, complexed with the competitive inhibitor phenol bound in the hydrophobic locket.31 Important active
`site residues are labeled. Note that inhibitor (and therefore substrate) binding does not require the displacement of zinc bound
`solvent (unlabeled gray sphere). Figure generated with MOLSCRIPT.89
`
`Mechanism of Catalysis: Substrate Binding
`and Proton Transfer
`
`The substrate association site of CAII is a hydro-
`phobic pocket adjacent to zinc-bound hydroxide,1°,16
`formed in large part by residues Val-143 at its base
`and Val- 121, Trp-209, and Leu- 198 at its neck (Figure
`2). This pocket is highly conserved among all active
`isozymes on the basis of phylogenetic comparisons,3,4
`although in isozyme III the pocket is somewhat
`constricted by the bulky side chain of Phe-198.lz The
`structure-based dissection of this pocket in CAII
`delineates the minimum size and shape of the pocket
`required for enzymatic activity; for instance, the
`occlusion of the pocket by the radical Val-143 ~ Phe
`substitution obliterates catalytic activity due to the
`
`(16) Lindskog, S. In Z#2c ~nzymes; Bertini, I., Luchinat, C., Maret,
`W., Zeppezauer, M., Eds.; Birkhauser: Boston, 1986; pp 307-316.
`(17) Eriksson, A. E.; Liljas, A. Proteins: St±~ct., Funct. Genet. 199a,
`16, 29-42.
`
`loss of the substrate association site.18,19 More moder-
`ate amino acid substitutions in the pocket (e.g., Val-
`143 ~ Gly or Val-121 ~ Leu) modestly affect the
`stability of the transition state for CO2 hydration
`(rather than substrate affinity) and alter the structure
`of active site solvent, leading to slight changes in the
`rate constant for proton transfer from zinc-bound
`water to His-64 (see Figure 1). The hydrophobic
`pocket therefore has a minimum width2°-23 and
`depth18,19 for efficient catalysis, and linear free energy
`relationships indicate that the volume of the amino
`
`(18) Alexander, R. S.; Nair, S. K.; Christianson, D. W. BiochemistW
`1991, 30 11064-11072.
`(19) Fierke, C. A.; Calderone, T. L.; Krebs, J. F. ~iochemist±’y 1991,
`30, 11054-11063.
`(20) Nair, S. K.; Calderone, T. L.; Christianson, D. W.; Fierke, C. A.
`~L £~1ol. Chem. 1991, 266, 17320-17325.
`
`(21) Nair, S. K.; Christianson, D. W. £~iochemist±’y 1993, 32, 4506-
`4514.
`(22) Krebs, J. F.; Rana, F.; Dluhy, R. A.; Fierke, C. A. £~iocheraist±y
`1993, 32, 4496-4505.
`(23) Nair, S. K.; Krebs, J. F.; Christianson, D. W.; Fierke, C. A.
`£~iochemist±y 1995, 34, 3981-3989.
`
`Akermin, Inc.
`Exhibit 1016
`Page 2
`
`

`

`Carbonlc Anhydrase
`
`Acc. CTwm. Res., Vol. 29, No. Z 1996 333
`
`acid residue at the base of the pocket is critical for
`activity,~9 whereas the hydrophobicity of residues at
`the neck of the pocket is likewise important.2°,22,23
`These results are consistent with molecular dynam-
`ics simulations that point to the hydrophobic pocket
`as a substrate association site.24-26 Experimental
`evidence supporting the catalytic role of the hydro-
`phobic pocket comes from a variety of spectroscopic
`studies,at,aS including Fourier transform infrared spec-
`troscopy.32,39 The shift of the asymmetric stretching
`vibration of COa m a lower wavenumber upon associa-
`tion with CAII is consistent with the transfer of COa
`from aqueous solution to a hydrophobic environment.
`These experiments also indicate that the affinity of
`CAII for COa is low (~0.1 M), as required by the high
`turnover number of the enzyme which necessitates a
`rapid product dissociation rate constant. Finally,
`although the precatalytic enzyme-substrate complex
`is too short-lived to be observed by traditional X-ray
`crystallographic methods, the structure of the only
`known competitive inhibitor of CAII-catalyzed CO2
`hydration, phenol,3° has been solved in complex with
`the enzyme:a~ phenol binds in the hydrophobic pocket
`and makes van der Waals contacts with Val- 121, Val-
`143, Leu-198, and Trp-209 while its hydroxyl group
`hydrogen bonds with zinc-bound hydroxide (Figure 2).
`Since a competitive inhibitor must bind in the same
`location as the substrate, COa must therefore bind in
`the hydrophobic pocket prior to catalysis. Impor-
`tantly, COa binding does not displace zinc-bound
`hydroxide, although long-range (i.e., >3 A) weakly-
`polar interactions with zinc may contribute to sub-
`strate orientation.
`Product bicarbonate ion is formed by the nucleo-
`philic attack of zinc-bound hydroxide at COa im-
`mobilized in the hydrophobic pocket, and the binding
`mode of bicarbonate ion to zinc has been the subject
`of some controversy. Central to this controversy is the
`possible role of Thr-199 as a "doorkeeper".aa-a4 That
`is to say, the side chain hydroxyl group of this residue
`is proposed to allow only anions capable of donating
`a hydrogen bond to Thr-199 access to zinc binding.
`X-ray crystallographic structures of Co2+-substituted
`and Thr-199 ~ Ala CAIIs in complex with bicarbonate
`are consistent with this interpretation.35,36 However,
`the binding of azide anion, a competitive inhibitor of
`bicarbonate dehydration, demonstrates that Thr- 199
`is not a doorkeeper all the time:37,38 azide anion binds
`to zinc but does not hydrogen bond with Thr-199.
`
`(24) Merz, K. M. ~L Mol. ;3iol. 1990, 214, 799-802.
`(25) Liang, J. Y.; Lipscomb, W. N. Proc Natl. Acad. ScL [ZS.A. 1990,
`87, 3675-3679.
`(26) Merz, K. M. ~L Am. Chem. Soc. 1991, 113, 406-411.
`(27) Williams, T. J.; Henkens, R. W. ;3iochemist±y 1985, 24, 2459-
`2462.
`(28) Bertini, T.; Luchinat, C.; Monnanni, R.; Roelens, S.; Moratal, J.
`M. J. Am. Chem. Soc. 1987, 109, 7855-7856.
`(29) Riepe, M. E.; Wang, J. H. J. ;3~ol. Chem. 1968, 243, 2779-2787.
`(30) Simonsson, I.; Jonsson B. H.; Lindskog, S. ;3ioc!~em. ;3iop!~ys. Res.
`Commun. 1982, 108, 1406-1412.
`(31) Nair, S. K.; Ludwig, P. A.; Christianson, D. W. ft. Am. Chem.
`1994, 116, 3659 3660.
`(32) HAkansson, K.; Carlsson, M.; Svensson, L. A.; Liljas, A. ft.
`;3ioi. 1992, 227, 1192 1204.
`(33) Liljas, A.; HAkansson, K.; Jonsson, B. H.; Xue, Y. Eu±: J. ;3iochem.
`1994, 219, 1-10.
`(34) Lindahl, M.; Svensson, L. A.; Liljas, A. P[’oteJ±~s: Struct.,
`Genet. 1993, 15, 177 182.
`(35) HAkansson, K.; Wehnert, A. ft. Moi. ;3~oi. 1992, 228, 1212-1218.
`(36) Xue, Y.; Liljas, A.; Jonsson, B. H. P±’ote#~s: StFuct., Funct. Genet.
`1993, 17, 93-106.
`(37) Nair, S. K.; Christianson, D. W. ~u±’. J. ;3iochem. 1993, 213, 507-
`515.
`
`Instead, the zinc-bound azide nitrogen makes a non-
`hydrogen-bonded, van der Waals interaction with the
`Thr-199 hydroxyl group. Such a binding mode is
`likewise possible for product bicarbonate, as illus-
`trated in Figure 1. Although such an interaction does
`not contribute to enzyme-product affinity, it is never-
`theless likely m contribute to efficient catalysis by
`facilitating rapid product dissociation. Consistent
`with this interpretation are studies of Thr-199 vari-
`ants showing that deletion of the Thr-199 hydroxyl
`group stabilizes bicarbonate binding39,4° and alters the
`structure of the bound bicarbonate ion.36 Therefore,
`the Thr-199 side chain promotes maximum catalytic
`efficiency by destabilizing the product complex to
`provide for rapid dissociation and by selecting for a
`catalytically competent bicarbonate-zinc complex (Fig-
`ure 1). Thus, it is clear that the zinc binding site and
`its environment have evolved for optimal catalytic
`activity.
`Following bicarbonate dissociation is the rate-
`determining step of proton transfer to regenerate the
`active catalyst, zinc-bound hydroxide (Figure 1).5-9
`The product proton is not transferred from zinc-bound
`water directly to bulk solvent; instead, it is first
`transferred to an intermediate "shuttle" residue, and
`then transferred to bulk solvent. Because of its
`greater exposure to solvent, the shuttle residue can
`more efficiently transfer a proton to a variety of
`acceptors with pK~ values higher than that of water]
`Interestingly, the proton transfer pathway has diver-
`gently evolved among the carbonic anhydrase isozymes.
`In isozyme II, His-64 is the catalytic proton shuttle,41,42
`and it exhibits significant conformational mobility
`which accompanies its function.43-45 This residue is
`too far from zinc-bound solvent m allow for direct
`proton transfer; instead, proton transfer is achieved
`across two intervening, hydrogen-bonded solvent mol-
`ecules in the native enzyme (Figure 3). Isozymes IV,
`VI, and VII also contain His-64, which may similarly
`function as a proton shuttle (isozyme I contains His-
`64, but His-200 is the major proton shuttle group in
`this isozyme46 ). However, isozyme III contains Lys-
`64, and the main proton transfer pathway is direct
`transfer to bulk solvent.4r Intriguingly, isozyme V
`contains Tyr-64, which plays no major role in proton
`transfer48 due in part to steric effects arising from the
`bulky adjacent side chain of Phe-65.49 The three-
`
`(38) Jonsson, B. M.; Hfikansson, K.; Liljas, A. FEBSLelr. 1993, 322,
`186-190.
`(39) Krebs, J. F.; Ippolito, J. A.; Christianson, D. W.; Fierke, C. A. ~L
`;3~oi. Chem. 1993, 268, 27458-27466.
`
`(40) Liang, Z.; Xue, Y.; Behravan, G.; Jonnson, B. H.; Lindskog, S.
`~u±’. d. Biochem. 1993, 211, 821-827.
`(41) Steiner, H.; Jonsson, B. H.; Lindskog, S. Eu±: J. t3iochem. 1975,
`59, 253-259.
`(42) Tu, C. K.; Silvennan, D. N.; Forsman, C.; Jonsson, B.H.;
`Lindskog, S. Biochemist±y 1989, 28, 7913-7918.
`(43) Krebs, J. F.; Fierke, C. A.; Alexander, R. S.; Christianson, D. W.
`Biochemist±y 1991, 30, 9153-9160.
`(44) Nair, S. K.; Christianson, D. W. ~ Am. Chem. $oc. 1991, 113,
`9455-9458.
`(45) Taoka, S.; Tu, C.; Kistler, K. A.; Silverman; D. N. J. Biol. Chem.
`1994, 269, 17988 17992.
`(46) Engstrand, C.; Johnson, B. H.; Lindskog, S. Eur. J. t3iochem.
`1995, 229, 696-702.
`(47) Jewell, D. A.; Tu, C.; Paranawithana, S. R.; Tanhauser, S. M.;
`LoGrasso, P. V.; Laipis, P. J.; Silvennan, D. N. Biochemist±y 1991, 30,
`1484-1490.
`(48) Heck, R. W.; Tanhauser, S. M.; Manda, R.; Tu, C. K.; Laipis, P.
`J.; Silverman, D. N. J. Biol. Chem. 1994, 269, 24742-24746.
`(49) Boriack Sjodin, P. A.; Heck, R. W.; Laipis, P. J.; Silvennan, D.
`N; Christianson, D. W. Proc. N~tl. Acid. Sci. ~S.A. 1995, 92, 10955-
`10959.
`
`Akermin, Inc.
`Exhibit 1016
`Page 3
`
`

`

`334 Acc. CT~em. t~es., Vol. 29, No. 7, 1996
`
`CT~ristianson and Fierke
`
`Lessons learned in the structure-based redesign of
`the CAII zinc binding site are also relevant to the
`design and construction of de novo transition metal
`binding sites in other proteins,5J-57 since only a few
`examples are structurally characterized by atomic
`resolution X-ray crystallographic methods to provide
`ultimate proof-of-design.58,59 Indeed, examples char-
`acterized by other biophysical techniques sometimes
`illustrate the limitations of current structure-based
`design approaches despite the best attempts of com-
`putational structure analysis--for example, where
`"designed" metal ligands do not coordinate to the
`target metal ion,6°’61 or where zinc affinity is at least
`4 orders of magnitude weaker5~ than that measured
`for CAII.62,63 Furthermore, none of the designed sites
`function as efficient catalysts. Hence, there is much
`yet to be learned regarding structure-activity and
`structure-stability relationships in protein-metal
`binding sites, and the zinc binding site of CAII is a
`universally-recognized and easily-characterized para-
`digm. We note that there is a novel application of
`redesigned CAII zinc binding site variants in analyti-
`cal chemistry and biotechnology: a CAII-based metal
`ion biosensor.64 Following Nature’s example in the
`evolution of carbonic anhydrase for different biological
`functions, the "directed evolution", or structure-based
`redesign, of the CAII zinc binding site allows for the
`optimization of its properties65 in the development of
`a zinc biosensor.
`Altering the Direct Zinc Ligands. A thorough
`understanding of three factors must precede the
`molecular dissection and redesign of direct metal
`ligands in any protein-metal binding site. First, the
`chemical nature of the target metal must be consid-
`ered. It could be "hard", like the small, highly-charged
`magnesium ion, it could be "soft", like the large,
`highly-polarizable mercury ion, or it could be inter-
`mediate between these two extremes (i.e., "border-
`line").66 An optimized metal binding site in a protein
`molecule will exhibit hardness complementary to that
`of its target metal ion. Since zinc is a metal of
`borderline hardness, it is satisfactorily coordinated by
`the harder ligands aspartate and glutamate, the
`borderline ligand histidine, and the softer ligand
`cysteine. Second, the conformations of the amino acid
`side chains that comprise the metal coordination site
`
`(51) Tainer, J. A.; Roberts, V. A.; Getzoff, E. D. Cu±’~: Opin. Biotechnol.
`199Z, 3, 378-387.
`(52) Regan, L. Trends Biochem. ScL 1995, 20, 280-285.
`(53) Christianson, D. W. Adv. ])rotein Chem. 1991, 42, 281-355.
`(54) Higaki, J. N.; Fletterick, R. J.; Craik, C. S. TrendsBiochem. ScL
`199Z, 17, 100-104.
`(55) Klemba, M.; Gardner, K. H.; Marino, S.; Clarke, N. D.; Regan,
`L. Nature Stz~ct. Biol. 1995, 2, 368-373.
`(56) Elling, C. E.; Nielsen, S. M.; Schwartz, T. W. Nature 1995, 374,
`74-77.
`(57) Mfiller, H. N.; Skerra, A. Biochemistry 1994, 33, 14126-14135.
`(58) McGrath, M. E.; Haymore, B. L.; Summers, N. L.; Craik, C. S.;
`Fletterick, R. J. BJochemJst±y 1993, 32, 1914-1919.
`(59) Browner, M. F.; Hackos, D.; Fletterick, R. J. Nature Struct. Biol.
`1994, 1, 327-333.
`(60) Hellinga, H. W.; Richards, F. M. J. Mol. Biol. 1991, 222, 763-
`785.
`(61) Hellinga, H. W.; Caradonna, J. P.; Richards, F. M. ~/. Mol. Biol.
`1991, 222, 787-803.
`(62) Lindskog, S.; Nyman, P. O. Biochim. Biophys. Acta 1964, 85, 462-
`474.
`(63) Kiefer, L. L.; Krebs, J. F.; Paterno, S. A.; Fierke, C. A. Biochem
`istzy 1993, 32, 9896 9900.
`(64) Thompson, R. B.; Jones, E. R. Anal. Chem. 1993, 65, 730-734.
`(65) Kiefer, L. L.; Paterno, S. A.; Fierke, C. A. ~ Am. Chem. Soc. 1995,
`117, 6831-6837.
`(66) Pearson, R. G. J. Am. Chem. Soc. 1963 85, 3533-3539.
`
`Figure 3. Proton shuttle His 64 adopts the "in" conforrnation
`in native CAII and engages in a hydrogen bonded solvent
`network (red) with zinc bound solvent at pH 8.5.32 Proton
`transfer across this network regenerates the reactive zinc bound
`hydroxide species in catalysis (Figure 1). Upon protonation at
`lower pH values, His 64 swings away from the active site to
`the "out" conformation.44 Thus, His 64 is conformationally
`mobile, as indicated. Figure generated with MOLSCRIPT.89
`
`dimensional structure of isozyme V suggests that a
`residue in the vicinity of Tyr-131 (e.g., the phenolic
`side chain of this residue or one of three flanking
`lysines) may serve as a proton shuttle group,49 but
`further experiments are necessary to confirm this
`speculation. Rate constants for proton transfer in the
`carbonic anhydrases are dependent on the difference
`in pKa between zinc-bound water and the proton
`shuttle group, and are limited by organization of the
`intervening hydrogen-bonded solvent molecules.9,45,5°
`Individual components of the zinc binding site may
`therefore affect catalytic proton transfer by modulat-
`ing the pKa of zinc-bound water.
`
`Structure-Based Protein Engineering:
`Redesigning the Zinc Binding Site
`
`As summarized in the preceding section and in
`Figure 1, the zinc ion in the CAII active site plays a
`central role in the mechanism of CO2 hydration. The
`principal role of zinc is that of an electrostatic catalyst,
`since it stabilizes the negatively-charged transition
`state leading to bicarbonate formation. It also de-
`presses the pK~ of its bound water molecule to provide
`an ample supply of nucleophilic hydroxide ion for
`catalysis at neutral pH values. Logically, the protein
`environment of zinc is critical for optimizing the
`electrostatic properties of the metal ion for catalysis,
`and this environment includes not only the direct
`metal ligands but also the residues hydrogen bonding
`with these ligands (i.e., "indirect" or "second shell"
`ligands) (Figure 4). By combining the techniques of
`molecular and structural biology, we have dissected
`the determinants of affinity and catalysis in the zinc
`binding site of CAII.
`
`(50) Silverman, D. N.; Tu, C.; Chen, X.; Tanhauser, S. M.; Kresge, A.
`J.; Laipis, P. J. BJochemJstzy 1993, 32, 10757-10762.
`
`Akermin, Inc.
`Exhibit 1016
`Page 4
`
`

`

`CaFbonlc Anhydz~se
`
`Acc. ~7;em. Res., Vol. 29, No. 7, 1996 335
`
`Direct Metal Ligands
`Indirect Metal Ligands
`
`Figure 4. Scheme of the CAII zinc binding site. Variants with characterized properties and three dimensional structures are indicated
`by italics at the site of substitution.
`
`Table 1. Properties of Selected CA Variants with
`Altered Zinc Binding Sitesa
`
`variant
`
`kcatlKM/uM 1 s 1)
`
`wild type
`H94D
`H94C
`H96C
`Hll9C
`H119D
`
`110
`0.11
`0.11
`0.073
`0.11
`3.8
`
`pKa
`
`6.8
`_>9.6
`_> 9.5
`8.5
`nd
`8.6
`
`zinc Kd (nM)
`
`0.004
`15
`33
`60
`50
`25
`
`Data from ref 75. nd = no data.
`
`must be favorable.54,67 If an engineered ligand must
`incur too high an energetic cost to coordinate to a
`metal ion, it will not do so. Even so, certain regions
`of a protein structure--e.g., loops, ~-helices, or f!-
`strands--can be sufficiently pliable to allow for modest
`conformational changes or segmental shifts which
`optimize metal coordination by an engineered ligand.
`Finally, and perhaps most importantly, the separation
`and stereochemistry of ligand-metal coordination
`must be reasonable. The geometric preferences of
`carboxylate,68,69 imidazole,7° and thiolate71 ligands for
`protein-metal ion coordination have been outlined,
`and the results of these studies provide structural
`criteria by which metal binding site designs can be
`evaluated and optimized.
`The alteration of direct zinc ligands in CAII yields
`important structural and functional insight as the
`properties of selected variants (Table 1) are inter-
`preted in light of their three-dimensional struc-
`
`(67) Ponder, J. W.; Richards, F. M. ~. Mol. J3~o1. 1987, 193, 775-791.
`(68) Christianson, D. W.; Alexander, R. S. ~. Am. Chem. Soc. 1989,
`111, 6412-6419.
`(69) Chakrabarti, P. P±’ote#~ Eng. 1990, 4, 49-56.
`(70) Chakrabarti, P. P±’ote#~ ~ng. 1990, 4, 57-63.
`(71) Chakrabarti, P. J3iochemist±y 1989, 28, 6081-6085.
`
`Lures.72-76 As a point of reference, recall that the
`protein ligands to zinc in wild-type CAII are His-94,
`His-96, and His-ll9 (Figures 2 and 4). A notable
`feature in the CAII zinc binding site is that metal
`binding to the histidine ligands is very cooperative:
`the deletion of any one protein ligand by a histidine
`~ alanine substitution decreases the zinc affinity75 by
`a factor of ~105. This is much larger than the 10-
`100-fold decrease in metal affinity observed for remov-
`ing one ligand from Cys2His2 zinc binding sites,
`including a zinc finger peptide77 and a de novo
`designed zinc coordination polyhedron in the 4~-
`helical bundle protein designated Z~4.7~
`Similarly, substitution of a neutral histidine ligand
`by the negatively-charged side chains of aspartate,
`glutamate, or cysteine decreases the zinc affinity ~ 104-
`fold.75 The structure of His-94 ~ Asp CAII reveals
`that this loss of protein-metal affinity arises prima-
`rily from the 0.9 ik movement of zinc that accom-
`modates the shorter side chain of the ligand. Intrigu-
`ingly, the structure of His-94 ~ Cys CAII demon-
`strates that the zinc ion occupies the same site in the
`His-94 ~ Asp and His-94 ~ Cys variants;75,76 ad-
`ditionally, a 1 ik deformation of the f!-strand contain-
`ing Cys-94 ensures that this shortest ligand side chain
`coordinates to zinc (Figure 5). Notably, this f!-strand
`
`(72) Alexander, R. S.; Kiefer, L. L.; Fierke, C. A.; Christianson, D. W.
`;3iochemist±’y 1993, 32, 1510-1518.
`(73) Ippolito, J. A.; Nair, S. K.; Alexander, R. S.; Kiefer, L. L.; Fierke,
`C. A.; Christianson, D. W. ]3±’ote#~ ~ng. 1995, 8, 975-980.
`(74) Kiefer, L. L.; Ippolito, J. A.; Fierke, C. A.; Christianson, D. W. ~
`Am. Chem. $oc. 1993, 115, 12581-12582.
`(75) Kiefer, L. L.; Fierke, C. A. BJochemJsr±’y1994, 33, 15233-15240.
`(76) Ippolito, J. A.; Christianson, D. W. BJochemJsr±y1994, 33, 15241-
`15249.
`(77) Merkle, D. L., Schmidt, M. H.; Berg, J. M. ~ Am. Chem. Soc.
`1991, 113, 5450-5451.
`(78) Klemba, M.; Regan, L. ;3iochemisr±y 1995, 34, 10094-10100.
`
`Akermin, Inc.
`Exhibit 1016
`Page 5
`
`

`

`336 Acc. CTwm. Reso, Vol. £9, No. 7, 1996
`
`C7~rlstlanson and Fler~e
`
`Figure 6. log of the pH independent second order rate constant
`for CO2 hydration versus the zinc-water pKa for wild type and
`variants of CAII,43,65,r~ including indirect ligand substitutions
`at position 92 (Gln ~ Ala, Leu, AsH, Glu) and position 117 (Glu
`~ Asp, Ala) (@), direct ligand substitutions (His 94 ~ Cys, His
`94 ~ Asp, His 119 ~ Asp) (~), and position 199 variants (Thr
`~ Set, Ala, Val, Pro) (~). The data are fit to a line yielding R =
`0.92 and slope -0.65.
`
`small molecule nucleophile attacking an amide, ester,
`or carbonyl carbon where nucleophilicity increases
`with basicity.8° The negative slope observed for CAII
`variants indicates that electrostatic stabilization of the
`transition state by the positively-charged zinc ion,
`rather than the nucleophilicity of the zinc hydroxide,
`is a dominant factor in catalysis. One possible inter-
`pretation of these data is that the protein ligand-zinc
`separation changes to facilitate electrostatic stabiliza-
`tion of the transition state by maintaining the bond
`valence sum of zinc.8~,82 Therefore, any diminution of
`the positive charge in the zinc binding site has dire
`consequences for catalysis; however, variants with a
`substituted neutral ligand, such as His-119 ~ Gin,
`retain reasonable catalytic activity (C.-c. Huang and
`C. A. Fierke, unpublished results) and maintain active
`site structure (C. A. Lesburg and D. W. Christianson,
`unpublished results). The neutral histidine zh~c ligands
`in CAII are essential for attaining high catalytic
`efficiency by enhancing the net positive charge of the
`metal ion site.75,79 Intriguingly, the CAII-like domains
`of receptor protein tyrosine phosphatases contain
`histidine ~ glutamine substitutions, but neither zinc
`binding properties nor catalytic activity has been
`documented for these systems.~-~5
`Amino acid substitutions elsewhere in the direct
`metal coordination polyhedron of CAII reveal that not
`all metal ligands reside on pliable segments of active
`site/9-strands: the His-119 ~ Cys substitution yields
`an anomalously long zinc-sulfur separation of 2.8 ~,
`and the His-96 ~ Cys substitution leads to an unex-
`pected conformational change for the Cys-96 side
`chain that places it 6.1 ~k away from zinc (a water
`molecule moves into the intended coordination position
`of Cys-96, thereby maintaining the tetrahedral geom-
`etry).76 Each of these amino acid substitutions also
`yields a 104-fold loss of protein-metal affinity.75
`Despite the unexpected consequences of the His-96 ~
`Cys and His-119 ~ Cys substitutions, it is clear that
`(1) the less buried an engineered metal ligand is in
`the protein scaffolding, the more capable it will be of
`achieving metal coordination through a plastic struc-
`tural change and (2) no matter what amino acid
`substitution is made in the protein-zinc coordination
`polyhedron, the coordination geometry of zinc remains
`
`(80) Fersht, A. R. E±2zyme Structure and Mechanism, 2rid ed.; W. H.
`Freeman & Co.: New York, 1985; pp 83 85.
`(81) Thorp, H. H. I±*org. Chem. 1992, 31, 1585-1588.
`(82) Xiang, S. B.; Short, S. A.; Wolfenden, R.; Carter, C. W. BJochem
`Jst±y 1996, 35, 1335-1341.
`
`Figure 5. Superposition of wild type (yellow), His 94 ~ Asp
`(cyan), and His 94 ~ Cys (red) CAIIs. Note the plasticity of the
`f! strand containing residue 94, as well as the movement ofZn2+
`which accommodates amino acid substitutions. Zn2+ moves to
`an identical location in the His 94 ~ Asp and His 94 ~ Cys
`variants. Reprinted with permission from ref 76. Copyright 1994
`American Chemical Society.
`
`deformation results in only an additional 2-fold loss
`of protein-metal affinity relative to that of His-94 ~
`Asp CAII. Thus, the f!-strand containing metal ligands
`His-94 and His-96 is sufficiently pliable to accom-
`modate various metal ligand residues at position 94.
`However, it is the movement of zinc from its naturally-
`evolved wild-type position--and not the compensatory
`plasticity of the f!-strand containing the metal
`ligands--that most seriously compromises protein-
`metal affinity. It is intriguing that f!-sheet structure
`and stability can be regulated by the construction of
`transition metal binding sites, and this feature can
`be exploited in protein engineering experiments with
`many other systems.
`The introduction of a negative charge into the zinc
`coordination polyhedron yields a significant increase
`in the pKa of zinc-bound water as assayed by the pH
`dependence of esterase activity (in addition to cata