Advances in Physical Organic Chemistry Volume 44 Editor JOHNP.RICHARD DepartmentofChemistry UniversityatBuffalo,SUNY Buffalo,NY,USA Amsterdam–Boston–Heidelberg–London–NewYork–Oxford Paris–SanDiego–SanFrancisco–Singapore–Sydney–Tokyo AcademicPressisanimprintofElsevier Contributors to Volume 44 Claude F. Bernasconi Department of Chemistry and Biochemistry, University ofCalifornia, SantaCruz,CA 95064, USA W.W. Cleland Department of Biochemistry and Institute for Enzyme Research, University ofWisconsin-Madison, Madison WI 53726, USA Ronald Kluger Davenport Chemistry Laboratories, Department of Chemistry, University of Toronto, Toronto, Ontario M5S 3H6, Canada Scott O.C. Mundle Davenport Chemistry Laboratories, Department of Chemistry, University of Toronto, Toronto, Ontario M5S 3H6, Canada Rory More O’Ferrall School of Chemistry and Chemical Biology, University College Dublin,Belfield, Dublin 4, Ireland Charles L. Perrin Department of Chemistry & Biochemistry, University of California—San Diego,La Jolla,CA 92093-0358, USA Jakob Wirz Department of Chemistry, University of Basel, Klingelberg- strasse 80,CH-4056 Basel, Switzerland Hiroshi Yamataka Department of Chemistry, College of Science and ResearchInstitute for Future Molecules, Rikkyo University, Tokyo, Japan xi The low-barrier hydrogen bond in enzymic catalysis W.W. CLELAND Department ofBiochemistry and Institute forEnzyme Research, University ofWisconsin-Madison,Madison,WI 53726, USA 1 Introduction 1 2 Propertiesofhydrogenbonds 1 3 Roleoflow-barrierhydrogenbondsinenzymaticreactions 3 Enolizationreactions 3 Facilitatedtetrahedralintermediateformation 6 Facilitatedprotonionization 10 Asparticproteases 12 Miscellaneousenzymes 13 Acid–Basecatalysis 14 4 Conclusion 15 References 15 1 Introduction The term ‘‘low-barrier hydrogen bond’’ was introduced by me in 1992 to describe hydrogen bonds between groups of equal pK that showed low deuteriumfractionationfactors(aslowas0.3).1ItwasnotuntilanEnzyme MechanismsconferenceinKeyLargo,however,thatanumberofusfinally realized how such bonds can help catalyze enzymic reactions and papers describing this appeared in 1993 and 1994.2–5 Since then such bonds have been shown to play a role in many enzymic reactions and a Google search under‘‘low-barrierhydrogenbond’’turnsupover5000hits.InthisreviewI shall describe the properties of low-barrier hydrogen bonds and then give a number of examples. I have not tried to cover the entire literature and apologize to those whose works are not mentioned. 2 Properties of hydrogen bonds Hydrogen bonds come in a continuum of bond lengths and strengths. Those in water which hold it together as a liquid are (cid:1)2.8A˚ between oxygens and are weak (only a few kcalmol(cid:3)1). Since the pK of water as an acid is above 15 and its pK as a base is less than –1, the pK’s of the two 1 ADVANCESINPHYSICALORGANICCHEMISTRY (cid:1)2010ElsevierLtd. VOLUME44ISSN:0065-3160DOI:10.1016/S0065-3160(08)44001-7 Allrightsreserved 2 W.W.CLELAND oxygens in the hydrogen bond are drastically different and the hydrogen is covalently bound to one oxygen with a bond distance of 1A˚ and weakly bonded electrostatically to the other oxygen. When the pK’s of the two groups are the same, as in a hydrogen bond between formic acid and formate ion, the bond is shorter (2.5–2.6A˚ ) and the zero point energy level of the hydrogen is at or above the barrier (thus ‘‘low-barrier hydro- gen bond,’’ Fig. 1).6–8 Neutron diffraction of crystals containing such bonds show a diffuse distribution centered between the two heavy atoms.9 In certain cases where the bond is especially short, there is no barrier as in the F–H–F(cid:3) or HO–H–OH(cid:3) ions which are only 2.3A˚ long.10,11 Low-barrier hydrogen bonds are quite strong (as much as 27kcal mol(cid:3)1 in the gas phase and perhaps 12 in aqueous solution7), but in a medium with a dielectric constant of (cid:1)7 (similar to what occurs in an enzyme active site) the strength decreases by (cid:1)1kcalmol(cid:3)1 per pH unit mismatch in the pK’s of the groups involved.12 Thus there is a continuum between the very strong ones with matched pK’s and the weak ones with very different pK’s and the distances similarly differ as well. Low-barrier hydrogen bonds have considerable covalent character,6,13 which decreases as the bonds weaken and lengthen, so that the weak ones are only electrostatic in nature. As noted in 1992, low-barrier hydrogen bonds show low fractionation factors, with up to threefold discrimination against deuterium. They show downfield chemical shifts in proton nuclear magnetic resonance (NMR) of 18–20ppm. At first it was thought that they only occur in the gas phase or organicsolvents,butitisnowclearthattheycanoccurinsolutionscontaining ahighmolefractionofwater,evenatroomtemperature.14,15Whatlimitstheir determination in aqueous solution is rapid exchange with solvent protons. Hydrogen bonds can occur between two oxygens, two nitrogens, or one of each. We will show examples of O–O and O–N bonds in the discussion below. (a) (b) (c) O H O O H O O H O O H O Fig. 1 Energy diagrams for hydrogen bonds between groups of equal pK. (a) Weak hydrogen bond; O–O distance 2.8A˚ . (b) Low-barrier hydrogen bond (2.55A˚ ); the hydrogen diffusely distributed. (c) Single-well hydrogen bond (2.29A˚ ). Horizontal linesarezeropointenergylevelsforhydrogen(upper)anddeuterium(lower). THELOW-BARRIERHYDROGENBONDINENZYMICCATALYSIS 3 3 Role of low-barrier hydrogen bonds in enzymatic reactions ENOLIZATIONREACTIONS The first examples of enzymatic reactions where low-barrier hydrogen bonds played a role involved enolization of the substrate to change the pK of a key group in the reaction. Mandelate racemase enolizes R or S mandelate to convert the carboxyl group into an aci-carboxylate which can be protonated onoppositesidestogivetheRorSforms.Inthegroundstate,oneoxygenof thecarboxylgroupofmandelateiscoordinatedtoMg2þandtheotheroxygen ishydrogenbondedtoGlu317whichisprotonated.16ThepKofaCTOgroup is low, so this is a weak hydrogen bond. In the aci-carboxylate intermediate, however, the pK of its oxygen will be similar to that of Glu317 and the hydrogen bond becomes a low-barrier one (Fig. 2). The energy liberated by formationofthestronghydrogenbondlowerstheactivationforformationof the intermediate. The 105 reduction in k for the E317Q mutant supports cat this model.17 A similar situation occurs with triose-P isomerase, where Glu165 abstracts a proton from either glyceraldehyde-3-P or dihydroxyacetone-P to give an enediolate intermediate. The carbonyl group of the substrate is hydrogen bonded to a neutral imidazole in the active site; this will be a weak hydrogen bond because of the huge mismatch in pK’s.18 The pK of both the imidazole and the enediol intermediate, however will be (cid:1)11, and thus this hydrogen bond becomes a low-barrier one in the intermediate Fig. 3). An isoenergetic shift of the imidazole from one OH to the other shifts the strong hydrogen bond to the oxygen destined to become a carbonyl group when the intermediate isprotonated by Glu165 tocomplete thereaction. Ketosteroid isomerase is another enzyme in which enolization of the substrate changes the pK of a key atom so that a low-barrier hydrogen bond forms and helps stabilize the intermediate. Asp38 is the general base that removes a proton from the substrate, and Tyr14 is hydrogen bonded to the carbonyl oxygen of the substrate. The pK’s of a ketone and of tyrosine are Mg Mg HO HO O O C C C C O H Glu O H Glu H Lys166 H-base (bases) His297 Fig.2 Mechanismofmandelateracemase.16,17Lys166andHis297arethetwogeneral basesandareonoppositesidesofmandelate. 4 W.W.CLELAND H H Glu– HC OH C OH GluH C O HN N C O H N N CH OPO 2– CH OPO 2– 2 3 2 3 H H C O HN N C O H N N GluH Glu– HC OH C OH CH OPO 2– CH OPO 2– 2 3 2 3 Fig. 3 Mechanism of triose-P isomerase.4 Notethe isoenergetic shift of the histidine between the two OH groups of the enediolate intermediate; a low-barrier hydrogen bondispresentinbothstructures. drastically different, but in the dienolate intermediate, the pK’s become more similar. An analog aromatic in the A ring and containing a phenolic hydroxyl in place of the ketone bound at least 1000-fold tighter to the D38N mutant than to wild-type isomerase.19 The neutral Asn38 mimics the proto- nated state of Asp38 after the formation of the intermediate dienolate. In the inhibitorcomplexprotonNMRpeakswereat18.15and11.6,withtheproton at 18.15 having a deuterium fractionation factor of 0.34 and the hydrogen bondhavingastrengthof7.1kcalmol(cid:3)1morethanonebetweeninhibitorand water. This increase in hydrogen bond strength corresponds to over 5 orders ofmagnituderateaccelerationandmatchesthedecreaseinrateof4.7ordersof magnitude inthe Y14F mutant. Subsequent work has shown that Asp99 is involved in the hydrogen bond network in this enzyme and the 18.15ppm NMR peak is from a hydrogen bond between it and Tyr14.20 The 11.6ppm peak comes from the hydrogen bond between the intermediate and Tyr14. Despite this complexity, it is still true that formation of a strong hydrogen bond in the presence of the intermediatedecreasestheactivationenergyofthereactionandthusprovides catalysis. Aconitase contains a 4Fe–4S center with citrate or isocitrate binding with one of their carboxyl groups and the OH group coordinated to the Fe at one corner of the Fe–S cluster.21,22 A water molecule is also coordinated to this THELOW-BARRIERHYDROGENBONDINENZYMICCATALYSIS 5 Feandishydrogenbondedtoafreecarboxylgroup.Thegeneralbaseforthe elimination reaction is Ser642, which donated its proton to the Fe-bound hydroxide when the substrate bound. Proton removal by Ser642 produces an aci-carboxylate from the carboxyl next to the carbon from which the proton was removed, and the pK of the aci-carboxylate now is a close match to the pK of the Fe-bound water to which it is hydrogen bonded. This hydrogen bond thus becomes a low-barrier one, its formation providing part of the energy needed to form the aci-carboxylate (Fig. 4). His101 then protonates the Fe-coordinated OH of the substrate to allow it to be eliminated to give cis-aconitate. IntheE-isocitrateX-raystructurethehydrogenbondbetweentheFe-bound water and the carboxyl of isocitrate is 2.7A˚ long, while in a similar structure with the nitro analog of isocitrate bound as an aci-nitronate the distance is 2.5A˚ .21 Citrate synthase catalyzes the enolization of acetyl-CoA and attack of the enolate on oxaloacetate to form citryl-CoA, which is then hydrolyzed. Asp375 takes the proton from the methyl group of acetyl-CoA and neutral His274 hydrogen bonds to the carbonyl oxygen to stabilize the enolate.23 X-ray structures of carboxyl or amide analogs of acetyl-CoA showed 2.4–2.5A˚ hydrogen bonds between the carboxyl or amide group of the inhi- bitor(replacingthemethylofacetyl-CoA)andAsp375.24TheK oftheamide i inhibitor was pH independent, while that of the carboxylate decreased as the pH decreased, showing that the protonated form was the inhibitor. The carboxyl inhibitor binds 4 orders of magnitude tighter than acetyl-CoA and thus the low-barrier hydrogen bond (chemical shift 20ppm25) contributes at least this much to binding. During the catalytic reaction, the low-barrier hydrogen bondshouldbe between His274 and the enolate oxygen, since their pK’s will be similar, and the energy from formation of the stronger hydrogen bond will help catalyze the enolization (Fig. 5). Vitamin K-dependent carboxylase uses vitamin K epoxidation to drive the carboxylationofglutamategroupsinGladomains.Itisthoughtthatreaction of oxygen with reduced vitamin K produces a strongly basic form of an H H H OH O H OH Fe O Fe O Fe O O HO HHis O HO HHis O HOH His C O C O– C O C C C C C C C C C CH2 CH2 CH2 O H H COO– O H COO– O H COO– –O–Ser HO–Ser HO–Ser Fig. 4 Mechanism of aconitase.4 The aci-carboxylate intermediate shares a low-barrierhydrogenbondwiththeFe–OHgroup. 6 W.W.CLELAND COO– O HN N COO– O H N N Arg Arg O C CH CSCoA O C CH CSCoA 3 2 His His CH CH 2 Asp– 2 AspH COO– COO– COO– O HN N COO– O HN N Arg Arg HO C CH CSCoA HO C CH CSCoA 2 2 His His CH CH O H 2 Asp– 2 COO– COO– H Asp COO– O H N N COO– O HN N Arg Arg HO C CH CSCoA HO C CH C 2 2 His His CH O H CH O H + HSCoA 2 2 COO– H Asp COO– Asp– Fig.5 Putativemechanismofcitratesynthase.4Alow-barrierhydrogenbondhelpsto stabilizetheenolandtetrahedralintermediates. epoxidethatremovesaprotonfromaglutamateresiduetogiveacarbanion intermediate that reacts with CO . It was recently found that a H160A 2 mutant carried out epoxidation readily, but carboxylation very poorly.26 It was postulated that His160 forms a hydrogen bond to one oxygen of the carboxyl group of glutamate. This will be a weak hydrogen bond before enolization, but proton removal will give an aci-carboxylate whose pK is a close match to that of neutral histidine. Thus the authors proposed that a low-barrier hydrogen bond between aci-carboxylate and His160 helped to stabilizetheintermediate.Asyetthereisnostructuralevidenceinsupportof this attractive hypothesis. FACILITATEDTETRAHEDRALINTERMEDIATEFORMATION A low-barrier hydrogen bond forms between Asp102 and His57 in the tetrahedralintermediateofthereactioncatalyzedbychymotrypsinandsimilar serineproteases.InthefreeenzymethepKofAsp102andtheneutralformof His57 are quite different, but when the Ser195 proton is transferred to His57 during formation of the tetrahedral intermediate, the pK’s of Asp102 and protonated His57 now become matched and the hydrogen bond between THELOW-BARRIERHYDROGENBONDINENZYMICCATALYSIS 7 them becomes a low-barrier one, thus providing the energy for formation of the unstable intermediate (Fig. 6).5 Transfer of the proton from His57 to the leaving amino group gives an acyl enzyme and dissipates the strength of the His57 O C Asp102 Ser195 O H :N N H O O C Peptidyl NHR O C His57 O Peptidyl NHR C Asp102 Ser195 O H :N N H O – O Peptidyl His57 O C NHR y– C Asp102 Ser195 O HN y+ N H O H NR 2 O Peptidyl His57 O C C Asp102 Ser195 O :N N H O Fig. 6 Mechanism of chymotrypsin.5 A low-barrier hydrogen bond between Asp102 andHis57helpsstabilizethetetrahedralintermediate. 8 W.W.CLELAND low-barrier hydrogen bond. Clear evidence for this mechanism is provided by observation of tetrahedral adducts of trifluoromethyl ketone inhibitors with the enzyme. In these complexes the proton chemical shift of the proton in the Asp102–His57 hydrogen bond is 18–19ppm and the fractionation factor is 0.3–0.4. The exchange rate of the proton with the solvent ranges from282s(cid:3)1forN–AcF–CF withaK of26mMto12.4s(cid:3)1forN–AcLF–CF 3 i 3 with a K of 1.8mM. The pK of His57 in these complexes is 10.7 or 12.1. i ThepKof12.1is5pHunitshigherthanthatinfreeenzyme,correspondingto 5 orders of magnitude rate acceleration.27,28 This situation was mimicked by observing complexes of N-alkylimidazoles with carboxylic acids in chloroform.29 As the pK of the acid increased, the chemicalshiftoftheprotoninthehydrogenbondmoveddownfieldto18ppm and then moved back upfield. With 2,2-dichloropropionate the chemical shift of 18ppm did not change with dilution, suggesting a strong hydrogen bond between the two molecules. The chemical shifts of complexes with more upfield protons moved further upfield on dilution, showing that they were weaker. Calorimetric measurements of complexes between 2,2-dichloropropionate and N-methyl or N-t-butylimidazole gave values of 12 or 15kcalmol(cid:3)1 for the enthalpy of formation.30 The IR spectrum of a complex with 2,2-dichloropropionate showed two peaks for the CTO stretch at 1700cm(cid:3)1 for the low-barrier hydrogen bond ((cid:1)2/3 of the complex) and 1647cm(cid:3)1fortheedge-onionpairwherethecarboxylgroupisperpendicular totheringoftheimidazoleandbothoxygensareincontactwiththepositively charged nitrogens ((cid:1)1/3 of the complex). The NMR shift of 18ppm is an average for the two species, which are in rapid equilibrium on the NMR timescale. An 0.78A˚ structure of subtilisin resolved the proton between His64 and Asp32 of the catalytic triad.31 The distance of the hydrogen bond was 2.62A˚ with the proton 1.2A˚ from His64 and 1.5A˚ from Asp32. The authors felt that this was not a low-barrier hydrogen bond because His64 was not protonated, but the short distance suggests that when His64 does become protonated during formation of the tetrahedral intermediate, it will become a low-barrier one. For the reaction catalyzed by cytidine deaminase, an analog of cytidine where the 3–4 bond is a single one and there is a hydroxy group at C4 (zebularine 3,4-hydrate) is a competitive inhibitor with a K of 10(cid:3)12M.32 An i X-raystructureofthisinhibitorboundtotheenzymeshowsa2.45A˚ hydrogen bondbetween theOHgroup atC4and thecarboxylateofGlu104.33TheOH group is also coordinated to a Zn2þ ion and the other oxygen of Glu104 is hydrogen bonded to N3 (2.74A˚ ). This structure corresponds to the putative tetrahedralintermediateformedbyattackoftheZn-boundhydroxylgroupon C4 of the pyrimidine ring, but with the amino group at C4 replaced with hydrogen (Fig. 7). It appears that the formation of a low-barrier hydrogen bond between the OH group and Glu104 may provide some of the energy