LL_HO_JMB_2003.PDF

Crystal Structures of the Ferric, Ferrous, andFerrous–NO Forms of the Asp140Ala Mutant ofHuman Heme Oxygenase-1: Catalytic Implications

Latesh Lad1, Jinling Wang2, Huiying Li1, Jonathan Friedman1

B. Bhaskar1, Paul R. Ortiz de Montellano2 and Thomas L. Poulos1*

1Department of MolecularBiology and BiochemistryDepartment of Physiology andBiophysics, and Program inMacromolecular StructureUniversity of California atIrvine, Irvine, CA 92697-3900USA

2Department of PharmaceuticalChemistry, University ofCalifornia, San Francisco, CA94143-2280, USA

Site-directed mutagenesis studies have shown that Asp140 in both humanand rat heme oxygenase-1 is critical for enzyme activity. Here, we reportthe D140A mutant crystal structure in the Fe(III) and Fe(II) redox statesas well as the Fe(II)–NO complex as a model for the Fe(II)–oxy complex.These structures are compared to the corresponding wild-type structures.The mutant and wild-type structures are very similar, except for the distalheme pocket solvent structure. In the Fe(III) D140A mutant one watermolecule takes the place of the missing Asp140 carboxylate side-chainand a second water molecule, novel to the mutant, binds in the distalpocket. Upon reduction to the Fe(II) state, the distal helix running alongone face of the heme moves closer to the heme in both the wild-type andmutant structures thus tightening the active site. NO binds to both thewild-type and mutant in a bent conformation that orients the NO O atomtoward the a-meso heme carbon atom. A network of water molecules pro-vides a H-bonded network to the NO ligand, suggesting a possible protonshuttle pathway required to activate dioxygen for catalysis. In the wild-type structure, Asp140 exhibits two conformations, suggesting a dynamicrole for Asp140 in shuttling protons from bulk solvent via the water net-work to the iron-linked oxy complex. On the basis of these structures, weconsider why the D140A mutant is inactive as a heme oxygenase butactive as a peroxidase.

q 2003 Elsevier Science Ltd. All rights reserved

Keywords: heme oxygenase-1; peroxidase; heme; CO; P450*Corresponding author

Introduction

In humans, approximately 75–80% of the totalbody iron is in the form of heme bound to proteins,with the remaining complexed with the ironstorage protein, ferritin.1 Heme is the prostheticgroup of numerous proteins such as hemoglobin,peroxidases, cytochromes, and various oxidasesand hence is essential for diverse biologicalprocesses such as oxygen transport, electrontransfer, energy production and biotransformationof xenobiotics. Free heme, however, is highlytoxic. Humans and other organisms have evolveda unique enzyme for the removal of excess heme,heme oxygenase (HO), which utilizes heme as

both a prosthetic group and substrate.2 – 6 Using atotal of three molecules of O2 and seven electrons,7

HO is able to degrade heme to biliverdin, throughthe intermediates a-meso-hydroxyheme and verdo-heme, expelling iron from the heme ring andreleasing a one carbon atom fragment as carbonmonoxide (Figure 1).3,8 – 10 Finally, biliverdin is con-verted to bilirubin in the presence of biliverdinreductase. In mammals the transfer of electronsfrom NADPH to HO is mediated by cytochromeP450 reductase, the redox partner that is respon-sible for electron transfer to the cytochrome P450enzymes.11,12 Mammalian HO exists as two iso-forms, HO-1 and HO-2, with essentially identicalenzyme activities. Inducible HO-1 is involvedprimarily in heme catabolism in the liver andspleen, while the constitutive HO-2 may serve as asource of CO, a putative messenger molecule invarious physiological functions.10,13,14

Several aspects of the HO mechanism remain

0022-2836/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved

E-mail address of the corresponding author:[email protected]

Abbreviations used: HO, heme oxygenase; HO-1,mammalian HO isozyme 1; heme, Fe-protoporphyrin IX.

doi:10.1016/S0022-2836(03)00578-3 J. Mol. Biol. (2003) 330, 527–538

unclear, although it is known that the HO reactionproceeds through a mechanism different from thatcurrently accepted for other hemoproteins such ascytochrome P450s and peroxidases.4,6,15 – 17 In thecatalytic cycle of HO, the enzyme first binds oneequivalent of heme, resulting in the formation ofthe heme–enzyme complex, which exhibits opticalabsorption spectral properties similar to those ofthe globins.8,9 The first electron provided by theP450 reductase reduces the ferric heme iron to theferrous state, and a molecule of oxygen binds toform a metastable, O2-bound complex that isreduced by a second electron to generate ferric-hydroperoxy HO (Fe(III)–OOH).12 The next step isvery likely the electrophilic addition of the reactiveFe(III)–OOH species to the a-meso carbon atom ofthe heme group. This assumption is made on thebasis of the following results: (a) ability of H2O2 tosubstitute for NADPH and P450 reductase in theproduction of verdoheme;18 (b) formation of ana-ethoxyheme when human HO-1 is exposed toethylhydroperoxide;19 (c) electronic effects of meso-methyl and meso-formyl heme substituents on thereaction;20,21 and (d) electron paramagnetic reson-ance spectroscopy (EPR)/electron nuclear doubleresonance (ENDOR) studies that have shown con-clusively that the hydroperoxy-ferric-HO self-hydroxylates to form the a-meso hydroxyheme.22

This is in contrast to peroxidases and cytochromeP450, which proceed through a high-valent ferryl(Fe(IV)–O) active intermediate4,19 as the reactivehydroxylating species rather than the peroxoligand capable of attacking the a-meso bridge ofthe heme (Figure 1). The exact stoichiometricrequirements involved in the formation of ferricverdoheme from a-meso-hydroxyheme have beencontroversial.23 – 25 Although there is general agree-ment that this process is oxygen-dependent, thesuggestion that additional reducing equivalentsare required23,26 has been questioned.24,25 The con-version of verdoheme to biliverdin is the least

well-characterized step of the overall reaction,although a mechanism has been proposed.4

In mammals, HO is membrane-bound, a compli-cation that initially hindered detailed mechanisticstudies. However, the expression of soluble, trun-cated forms of HO that lack the membrane-bindingdomain while retaining full activity18,27,28 greatlyfacilitated detailed mechanistic studies of theenzyme. Most notable were crystallization andstructure determination of the human and ratHO-1 enzymes complexed with heme,29,30 whichprovided significant insights into both the mechan-ism and regiospecificity of the HO reaction.Crystallographic studies revealed a novel proteinfold that consists primarily of a-helices with theheme sandwiched between the proximal and distalhelices. In the human HO-1 structure, conservedglycine residues in the distal pocket provideflexibility that allows the two molecules in thecrystallographic asymmetric unit cell to differ. Inone molecule, the active-site pocket is more openwith a relatively loose distal helix–heme contact,while in the other molecule the contact betweenthe heme and distal helix is tighter. This, togetherwith the high crystallographic thermal factors andthe publication of the apo structure,31 suggest thatflexibility of the distal helix enables the hemepocket to be opened and closed to bind the hemesubstrate and to permit dissociation of thebiliverdin product. Early suggestions that theregioselectivity in the formation of a-meso-hydroxyheme might be controlled by steric con-straints imposed upon the heme–O2 complex aresupported by the crystal structures.29 The distalhelix passes across the entire face of the heme andphysically restricts access to all the meso positionsexcept for a-meso carbon. Indeed, spectroscopicmeasurements have shown that the Fe–O–O bondis bent,32 which, combined with the results fromthe recently reported crystal structure of rat HO-1in a complex with heme bound to azide,33 suggests

Figure 1. The overall reaction catalyzed by heme oxygenase that results in oxidation of the heme and release of thea-meso carbon atom as CO.

528 Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1

that the Fe–O–O bond is in the direction of thea-meso carbon atom.

Even though the present structural informationprovides a fairly accurate interpretation of theregioselectivity in terms of steric blocking of allbut the a-meso position by the distal helix, theenvironmental influences that stabilize the Fe(III)–OOH rather than Fe(IV)vO species remainunclear. Mutagenic studies with human34 and ratHO-135 have shown that elimination of thecarboxylate group of Asp140 on the distal helixdecreases or suppresses HO activity drasticallyand leads to the formation of a ferryl species,which is able to turn over a traditional peroxidasesubstrate, guaiacol, in the presence of H2O2.

34 Acloser analysis of the human HO-1 structureidentifies a water molecule that H-bonds withAsp140 and another water molecule coordinatedto the heme and this network of interactions hasbeen proposed to play a crucial role in stabilizingthe Fe(III)–OOH species.34,35 In addition, NMRstudies on human HO-1 identified an extensiveH-bond network, with very strong H-bonds,which involves the catalytically critical Asp140and extends from the distal pocket through theopposite end of the enzyme, with the peptideprotons of Arg85, Lys86, Ala 165 and Phe66 andthe side-chain labile protons of Tyr58, Trp96,His 132 and Arg136, serving as donors to fourcarboxylate groups (Glu62, Asp92, Asp140 andGlu202).36,37 It has been suggested that the likelyrole of the strong H-bonding network is to providea scaffold to support the ordered water molecules,which, in turn, stabilize the hydroperoxy inter-mediate in the reaction pathway and serve as a“water channel” to funnel protons to the catalyticsite in a controlled manner.38 To obtain a betterunderstanding of the oxygen activation mechanismin HO enzymes, we have determined the crystalstructures of the human HO-1 D140A mutant and,as mimic of the oxy complex, the NO complexesof wild-type and the D140A mutant. These struc-tures indicate that the precise H-bonding andordered solvent structure within the distal pocketare essential for HO function.

Results

Asp140Ala structure

The mutant crystallized in two forms, mono-clinic space group P21 and orthorhombic spacegroup P212121. Since the monoclinic form diffractsto 1.59 A and the orthorhombic to 2.59 A, thefollowing discussion is confined to the monoclinicform. We refer to the two molecules in the asym-metric unit as open and closed. In the closed con-formation, the distal helix is closer to the heme,thus forming a slightly tighter active-site pocket.

Figure 2 shows the 2Fo 2 Fc electron densityaround the site of the mutation and the position ofkey water molecules in the closed molecule. Acomparison of the wild-type and D140A closedmolecules is presented in Figure 3. In the closedmolecule of the wild-type structure, the carboxyl-ate side-chain of Asp140 forms a set of H-bonds/ionic interactions with Tyr58, Arg136 and a nearbywater molecule, Wat1 (Figure 3). Wat1 is thoughtto be important in HO catalysis, since in the wild-type structure Wat1 is only 3.3 A from the distalwater molecule (Wat0) ligated to the heme iron(Figure 3),29,31 suggesting that Wat1 may be part ofa proton shuttle system that participates in proto-nation of the Fe–oxy complex. In the closedmolecule of the D140A mutant, these interactionsare disrupted (Figure 3). The loss of H-bondsbetween Asp140-Tyr58 and Asp140-Arg136 in theD140A mutant results in a 0.8–1.3 A movement ofTyr58 and Arg136 away from the site of mutation.Moreover, the additional space in the active-sitepocket of the mutant is taken up by two newordered solvent molecules, Wat3 and Wat4. Wat3takes the place of the missing Asp140 carboxylategroup. The second new water molecule, Wat4, ispositioned 3.28 A from Wat0, and contacts theside-chain of Leu147 (Figure 3). Wat1, Wat2, andthe distal ligand, Wat0, are essentially the same inthe wild-type and D140A mutant, and form similarH-bonding interactions, although Wat1 is furtherfrom Wat0, 4.35 A in the mutant compared to3.34 A in wild-type. As with wild-type HO-1,31 the

Figure 2. Stereo diagram of the 2Fo 2 Fc electron density contoured at 1.5s around part of the distal helix of D140Ain the Fe(III) oxidation state. All the Figures were prepared with MOLSCRIPT51 and Raster3D.52

Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1 529

open conformation in the mutant shows a similaractive-site structure although the solvent structureis not as well defined.

In addition to changes in solvent structure, thereare small but important differences in the protein,even though the overall root-mean-squaredifference (rmsd) in backbone atoms between themutant and wild-type structures is only 0.3 A.Replacing Asp140 with Ala changes the positionof nearby residues Leu141, Ser142 and Gly143 rela-tive to the heme. Together with Gly139, Gly143 isone of two conserved glycine residues in HOthought to be responsible for providing the flexi-bility in the distal helix that enables the openingand closing of the active site required to bindheme and release biliverdin.29 On the other hand,Ser142 stabilizes the kink in the distal helixthrough an array of H-bonds with Gly143, Leu141and Lys179. In the closed molecule of D140A,Gly143 moves away from the heme plane, whileLeu141 moves closer to the heme. This movementcauses the loss of H-bonding interaction betweenSer142 and Lys179. In the open molecule, Gly143and Ser142 have moved closer to the heme, somuch so, that the carbonyl oxygen atom of Ser142interacts directly with the distal water moleculecoordinated to the heme iron (Figure not shown).

To obtain an objective estimate of the variousheme and ligand parameters in the mutant struc-ture, none of the Fe to ligand distances was

restrained during refinement. Table 1 gives variousparameters relevant to heme and ligand geometry,and bond distances. The most noticeable differenceobserved between the two structures is the geome-try of the heme. The heme in the mutant structureis flatter than that observed in the wild-type struc-ture (Figure 3). Furthermore, the His–Fe bond islonger in the D140A structure, 2.10 A in the closedand 2.11 A in the open molecule compared to thecorresponding distances in the wild-type structure,1.98 A and 1.90 A, respectively.

Another notable difference is the Fe–H2O dis-tances. In the wild-type structure the Fe–H2Obond distance is about 2.05 A in both the openand closed molecules. In the closed molecule ofD140A, the Fe–H2O bond distance is shorter,1.93 A, while in the open molecule of D140A theFe–H2O distance is 2.2 A. These differences maybe accounted for by the presence of an extraH-bonding interaction between the carbonyloxygen atom of Ser142 and the distal watermolecule that is present in the open molecule ofD140A but not present in either closed moleculesof D140A or in the wild-type structure.

Structures of D140A and wild-type HO-1 in theferrous Fe(II) state

Reduction of the iron from Fe(III) to Fe(II) causessmall changes in both the wild-type and D140A

Table 1. Heme parameters

Fe–His (A) Fe–pyrrole N (average) (A) Fe–pyrrole N plane (A) Fe–H2O (A) Fe–NO (A)

A. ClosedWT 1.98 1.97 0.01 2.05 –D140A 2.10 1.98 0.01 1.93 –WT-Fe2þ 2.12 1.99 0.16 – –D140A–Fe2þ 2.12 1.99 0.14 – –WT-NO 2.12 1.98 0.05 – 1.65D140A–NO 2.12 1.98 0.01 – 1.50

B. OpenWT 1.90 1.96 0.08 2.06 –D140A 2.11 1.99 0.06 2.19 –WT-Fe2þ 2.15 1.99 0.23 – –D140A–Fe2þ 2.13 1.98 0.20 – –WT-NO 2.10 1.99 0.06 – –D140A–NO 2.12 1.99 0.05 – 1.76

Figure 3. A comparison of the Fe(III) wild-type and D140A human HO-1 structures. Key hydrogen bonding inter-actions are shown as broken lines.

530 Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1

structures that result in the movement of the distalhelix toward the heme (Figure 4). In the closedmolecule of the reduced wild-type HO-1, the back-bone atoms of Gly143 have shifted approximately1 A closer to the heme (Figure 4). The His–Febond increases from 1.98 A to 2.12 A in goingfrom Fe(III) to Fe(II) in the closed molecule. In thisinstance, the change in the His–Fe bond distanceresults from the movement of the Fe atom out ofthe porphyrin core, which lengthens the His–Febond and the pyrrole N–Fe bonds (Table 1). Themovement of Gly143 results in other residues inthis region changing as well (i.e. Gly144, Glu145and Leu147); in particular Gly144, which movesabout 0.9 A closer to the heme (Figure 4). However,in the open molecule the structures of reduced andoxidized forms are essentially the same eventhough the His–Fe bond increases from 1.90 Ato 2.15 A. In D140A we observe conformationalchanges similar to that seen in the wild-typestructure, although not as large (Figure 4 right).Interestingly, the carbonyl oxygen atom of Ser142in the open molecule of ferrous D140A has flippedback by approximately 458 and now adopts a con-formation similar to that in the closed molecule.Also, in both D140A molecules reduction of theheme iron results in longer His–Fe bonds (Table 1).No ordered water molecule is found near the

vacant sixth heme coordination position in eitherset of reduced structures.

Structure of the ferrous wild-type-NO complex

While NO binds to both molecules in the asym-metric unit of the mutant, NO binds in only thewild-type closed molecule and, hence, only theclosed molecule of wild-type and D140A can becompared directly. Figure 5 provides a comparisonbetween the wild-type and mutant NO complexes.In the wild-type structure, NO binds end-on to theheme iron with a bend angle of <1388 (/Fe2þ–N–O) such that NO leans toward the a-meso-carbonatom of the heme (Figure 5), which is due, in largepart, to an especially close 2.7 A contact betweenthe NO oxygen atom and the carbonyl O atom ofGly139. (Figure 5). The bend angle of NO is similarto the bend angle reported for the oxy complex ofP450-cam (1328)39 and for the oxy complex inhorseradish peroxidase (HRP) (1268).40 The dis-tance between the terminal NO oxygen atom andthe a-meso-carbon atom of the heme is about4.0 A. The bond distance between the heme ironatom and the nitrogen atom of NO is 1.65 A,which is much shorter than the bond distance(2.2 A) between azide and iron in the rat HO–heme–N3

2 structure.33 Upon binding of NO, theHis–Fe bond increases from 1.98 A to 2.12 A. Resi-dues and water molecules in the heme pocket ofthe NO complex are positioned very similarly tothose in the wild-type structure. The NO oxygenH-bonds with Wat1, which, in turn forms a 2.6 AH-bond with the catalytically critical Asp140(Figure 5). Through Wat1 and Wat2, NO is able tointeract with the carbonyl group of Thr135(Figure 5).

The most important change observed in theactive site upon NO binding is the alternate con-formations for the phenyl ring of Tyr58 and theside-chain of the highly conserved but catalyticallycritical Asp140 residue (Figure 6). One con-formation is the same as that observed in allprevious human HO-1 structures published todate, with the carboxylate side-chain pointing “in”

Figure 4. Comparison of the distal and proximal hemeregions between reduced Fe(II) (grey) and Fe(III) (black)structures in both the wild-type and D140A structures(closed molecules only).

Figure 5. The 2Fo 2 Fc electron density maps contoured at 1.5s of the wild-type and D140A Fe(II)–NO complexes.Key hydrogen bonding interactions are shown as broken lines.

Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1 531

towards the heme where it forms an H-bond withArg136, the hydroxyl group of Tyr58, Wat1 andWat2 (Figure 6). The alternate “out” conformationresults in the movement of Asp140 about 3.5 Afrom the heme, while motion of Tyr58 positionsthe phenol ring about 2 A closer to the heme. Thismotion results in a weakening of the electrostaticinteractions between Asp140 and its surroundingneighbors, Arg136 and Tyr58. Most importantly,Asp140 loses the H-bonding interaction with Wat1and the close contact distance with Wat2 (Figure 5).

Structure of the ferrous D140A–NO complex

In the D140A–NO complex structure, weobserve the binding of NO in both open and closedmolecules compared to the closed molecule in thewild-type NO complex, although it should benoted that NO was modeled with a partial occu-pancy of 0.65 in the open molecule. Since NObound in only the closed molecule in the wild-type-NO complex, we limit the following discus-sion on the mutant–NO complex to the closedmolecule. As in the wild-type NO complex, theNO ligand is bent and leans towards the a-meso-carbon atom of the heme. The distance betweenthe oxygen atom of NO and the a-meso-carbonatom of the heme is the same as in the wild-typestructure, about 4.0 A. The Fe–NO bond is 1.50 Ain the mutant compared to 1.65 A in the wild-type. Neither of these distances was restrainedduring refinement. As in the wild-type structure,residues and solvent structure in the active site ofD140A–NO are located in positions almost identi-cal with those in D140A (Figure 5).

In the wild-type NO complex, NO is stabilizedby a network of H-bonds involving Tyr58, Asp140,Wat1, Wat2 and Thr135 (Figure 5). The main differ-ence in this network in D140A–NO is the missingH-bond between Wat1 and Ala140. In D140A–NO, the extra water molecule, Wat3, replacesAsp140 (Figure 5) and is essentially in the same

position as Wat3 observed in D140A. Wat3 main-tains the same H-bonding interactions as Asp140in wild-type-NO, except Wat3 does not H-bondwith the heme water ligand but instead interactswith Wat2.

The main difference in the open molecule ofD140A–NO is that the NO is bent more towardpyrrole ring B than in wild-type-NO or the closedD140A–NO molecule. The lack of water moleculesnear the NO ligand in the open form may explainwhy we observe leaning of the NO ligand towardspyrrole B of the heme rather than the a-meso car-bon atom. The H-bond that is observed betweenthe carbonyl oxygen atom of Ser142 and the distalwater ligated to the heme in the open molecule ofD140A is lost in D140A–NO. In the open moleculeof D140A–NO, the carbonyl oxygen atom of Ser142has flipped away by 908 and now occupies a pos-ition similar to that in the wild-type structure andclosed molecule of D140A. The lack of solventstructure near NO in the open molecule ofD140A–NO suggests that the opening and closingof the distal helix traps water molecules in theactive site that participate in catalysis and help tosteer the ligand toward the a-meso carbon atom.

Discussion

The key to understanding the mechanism of HOis first, identification of the chemical functionalitythat delivers a proton to the iron-linked dioxygenand second, the structural requirements thatfavor/stabilize the formation of an FeIII –OOHintermediate capable of electrophilic attack on thea-meso heme carbon atom rather than heterolyticcleavage of the O–O bond to give a traditional per-oxidase Fe(IV)–O oxyferryl center. In peroxidases,a distal histidine residue acts as a general acid–base catalyst that promotes heterolytic cleavage ofthe peroxide O–O bond to give the Fe(IV)–Ointermediate.15,16 In HO, there is no strategicallypositioned acid–base catalyst such as a histidineresidue that can serve a similar acid–base catalyticfunction. However, the absence of an acid/basecatalytic side-chain cannot be the basis for sup-pressing O–O bond cleavage over electrophilicattack, since P450s also lack a catalytic residueand yet P450s cleave the peroxide O–O bond.

Mutagenesis work has shed considerable lighton this problem. Mutation of conserved Gly143 inthe distal helix rendered human HO-1 inactive,41

whereas mutation of the other conserved glycineresidue, Gly139, gives an HO exhibiting a newperoxidase activity.41 Most recently, mutagenicstudies,34,35 which were later investigated by EPRand ENDOR studies,42 have shown that Asp140 inboth rat and human HO-1 enzymes is crucial inthe oxygen activation mechanism in HO-1.Replacement of Asp140 with Ala in both humanand rat HO-1 enzymes converts HO-1 into a per-oxidase with formation of Fe(IV)–O. Furthermore,these mutants were unable to degrade heme and,

Figure 6. The 2Fo 2 Fc electron density map contouredat 1.0s of the wild-type Fe(II)–NO structure showing thealternate conformations of Asp140 and Tyr136.

532 Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1

most importantly, unable to activate the iron-bound dioxygen and hydroperoxide moietiesproperly. On the basis of these findings, an oxygenactivation mechanism has been postulatedinvolving a hydrogen-bonding network through abridging water molecule and Asp140 side-chain inHO-1.35 Our current work was designed to furtherexplore the structural consequences of the D140Amutant as well as determine the structure of theNO complex. While the recent structural work onthe azide rat HO-1 complex has proveninsightful,33 NO is a much better mimic of thedioxygen complex and should provide a moreaccurate picture on possible interactions in theoxy-complex.

NO complexes

To separate the effects of NO binding from ironreduction, we first solved the structures of theFe(II) structures for both wild-type and D140A. Inboth wild-type and D140A the iron moves slightlyabove the heme plane resulting in an increase inthe His–Fe bond. In the closed wild-type structure,the distal helix moves closer to the heme, whichhelps to sterically shield the b-, g-, and d-mesocarbon atoms of the heme by forcing any incomingligand (i.e. oxygen) to direct its terminal atomtowards the a-meso carbon (Figure 5). Thus, whenNO binds the ligand must bend toward the a-mesocarbon atom. In addition to steric control, theD140–Wat1–NO H-bonds further stabilizethe bending NO toward the a-meso carbon. TheD140A–NO complex is similar, except Wat3 takesthe place of the Asp140. That we did not observeNO binding in the open wild-type structure isprobably related to the absence of ordered solventrequired to stabilize NO. It thus appears that theactive site must close down around the boundligand and trap internal ordered solvent moleculesnecessary for both stabilizing the iron-linkedligand and in forming the proper H-bondednetwork required for catalysis.

Implications for catalysis

Proton transfer is required to initiate regio-specific hydroxylation of the heme. A protontransfer step in HO has been proposed recentlyusing the crystal structure of rat HO-1 with hemebound to azide.33 In rat HO-1, Sugishima andco-workers predict from the rat azide structurethat the amide group of Gly143 stabilizes thenegatively charged terminal oxygen atom of theoxy complex and through a conserved H-bondingnetwork, involving two water molecules, side-chain of Asp140, Arg136, Tyr58 and Thr135, makea nearby water molecule (similar to Wat1 inhuman HO-1) highly anionic, i.e. capable ofdonating a proton to the iron-bound dioxygenmoiety.33

Our current work provides a similar picture,although the Gly143 peptide NH is too far from

the NO oxygen, 3.9 A, and is tied up in a helicalH-bond. Therefore, the Gly143 peptide NH isunlikely to provide much stabilization of anegative charge on the oxygen ligand as proposedby Sugishima et al.33 However, the conformationsof the distal helix in the rat and human HO-1structures are different, thus freeing the Gly143peptide NH to interact with ligands in the ratenzyme. This could reflect the differences inligands, azide versus NO, and redox state of theiron. Alternatively, these differences may simplyreflect the established flexibility of the distal helixalthough the Fe(II)–NO complex more closelymimics the Fe(II)–O–O complex than the Fe(III)–azide complex. Despite these subtle differences,the H-bonded network in the distal pocketinvolving ordered solvent is very similar in the ratHO-1-azide and the human HO-1 and D140A-NOcomplexes. These structures provide a consistentpicture of a water molecule H-bonded directly tothe distal oxygen atom serving as the protonsource required to give Fe(III)–O–OH which thenattacks the a-meso heme carbon atom.

Such similarity in solvent structure leaves openthe question of why the D140A mutant loses HOactivity but gains peroxidase activity by hetero-lytically cleaving the O–O bond thus giving theperoxidase-like Fe(IV)–O intermediate. The cryoreduction work of Davydov and co-workers22 hasshed considerable light on this question. Oneelectron radiolytic cryoreduction of the oxy-ferrousform of rat heme–HO-1 Fe(II)–oxy complex at 77K yields a ferric-hydroperoxy species, indicatingfacile proton transfer via the solvent H-bondednetwork. However, in D140A, proton transfer todioxygen occurs only at temperatures greater than180 K, clearly showing that proton transfer ismore difficult in D140A.42 These results implicateAsp140 as a key component in rapid protonationof the oxy complex and stabilization of the hydro-peroxy intermediate. The conformational hetero-geneity of Asp140 we observe in the NO complexmay be relevant. The slight motion of Asp140between the in and out positions (Figure 6) mayenhance the ability of Asp140 to shuttle a protonfrom bulk solvent to dioxygen via Wat1 (Figure 7).A critical aspect of this mechanism is that Asp140can serve as only an H-bond acceptor. Thus, in thehydroperoxy complex, Wat1 is an H-bond acceptorfrom the hydroperoxy intermediate, whichprovides sufficient lifetime to the hydroperoxycomplex to enable reaction with the a-meso carbonatom. Such stabilization of the hydroperoxy inter-mediate is not possible in D140A. Wat3, whichtakes the place of Asp140, can operate as either anH-bond donor or acceptor so the rigid H-bondedwater network is free to reorient in D140A. Thus,the stabilizing H-bond by Wat1 to the hydroperoxyintermediate is lost in D140A. Such instability ofthe hydroperoxy intermediate in D140A results inthe “default” reaction of heterolytic fission of theO–O bond giving Fe(IV)–O. In addition, the distalpocket may have greater access to bulk solvent

Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1 533

Figure 7. A representation for the proposed mechanism of oxygen activation in human heme oxygenase-1. Note, in the absence of a stabilizing hydrogen bond from theAsp140 side-chain to Wat1, a default peroxidase (Fe(IV)–O) intermediate is formed. Black arrows represent the interactions that enhance the hydrogen-donating ability ofWat1.

and, hence, a non-selective source of protons thatwould further promote heterolysis.

One final difference observed in the D140A isheme ruffling. Rivera et al.43 recently pointed outthat differences in heme ruffling may be part ofwhat controls heme oxygenase chemistry. Thus, itis interesting to note that in the ferric structures,the D140A heme is substantially flatter than in thewild-type structure. Heme ruffling is expected toincrease spin density on the meso carbon atoms,thus promoting electrophilic attack by the ferric-hydroperoxy intermediate.43 Therefore, in additionto altering the proton shuttle machinery in the dis-tal pocket, the D140A mutant may have alteredthe ability of the heme to adopt the propergeometry and electronic state for rapid electro-philic addition of the peroxide OH group, thus tip-ping the balance in favor of heterolytic cleavage ofthe peroxide O–O bond.

Materials and Methods

Bacterial expression and protein purification

The D140A mutant and wild-type enzyme of the trun-cated water-soluble form of recombinant human HO-1were expressed, purified, and reconstituted with heminaccording to published procedures.28,34,44

Crystallization

Human HO-1 D140A and wild-type crystals weregrown under conditions similar to those described.29,44

Using the sitting-drop, vapor-diffusion method with awell solution of 2.08 M ammonium sulphate, 100 mMHepes (pH 7.5), and 0.9% (v/v) 1,6-hexanediol, dropsconsisted of protein stock (5 ml) at 45 mg ml21 in 20 mMpotassium phosphate (pH 7.4), mixed with well solution(5 ml) on siliconized cover-slips. The only differencehere is that all crystals were grown at room temperatureinstead of at 28 8C, and one round of touch seedingusing crushed wild-type crystals was necessary toinitiate D140A crystal growth. D-(þ)-Trehalose was usedas the cryoprotectant for cryogenic data collectioncrystals. Cryogenic data collection involved a seven steptransfer to artificial precipitant solution with increasedD-(þ)-trehalose concentration up to 35% (w/v). Wild-type human HO-1 crystals belong to the monoclinicspace group P21, whereas D140A crystals grew in twodifferent forms: one form belonging to the orthorhombicspace group P212121, the other being a monoclinic spacegroup P21. Cell dimensions of all crystal forms are listedin Table 2.

Generation of NO complexes

Crystals of wild-type and D140A human HO-1 werepassed stepwise through the normal (as describedabove) seven step transfer to artificial precipitantsolution with increased concentration of D-(þ )-trehaloseup to 35% (w/v). The cryoprotectant solution and thecryo-soaked crystals were subjected to anaerobic treat-ment by cycling between extensive degassing andpurging with pure N2 gas for 45–60 minutes. Freshsolutions of dithionite and sodium nitrate were made

with the degassed cryoprotectant solution in sealedserum vials. NO-complexed crystals were prepared bysoaking with 20 mM dithionite and 10–20 mM sodiumnitrite for at least 40 minutes inside a glovebox beforebeing flash-frozen for data collections.

Data collection

In-house data were collected using an R-AXIS IVimaging plate detector equipped with a rotating copperanode X-ray generator with Osmic optics (Rigaku).Crystals were maintained at 2160 8C in a steam of nitro-gen (Crystal Logic, Los Angeles). High-resolution datacollection was performed at ALS beamline 5.0.2 with aan ADSC CCD detector. Optimization of data collectionwas guided by the STRATEGY function of MOSFLM.45

All data were reduced using HKL 2000,46 and rejectionswere performed with ENDHKL (Louis Sanchez,California Institute of Technology) in conjunction withSCALEPACK. For all data collection, a 1808 scan using18 frames was collected.

Model building and refinement

The D140A human HO-1 structure in space groupP212121 was determined by the method of molecularreplacement using AmoRe.47 A monomer of the humanheme–HO-1 crystal structure (PDB accession number1qq8),29 with the heme and water molecules removed,was used as the probe with searches being carried outat 5 A in the P212121 space group. The best cross-rotationand translation function solutions were rigid bodyrefined and fixed in place followed by a search for theremaining molecule in the asymmetric unit. Two sol-utions were found, corresponding to the expected twomonomers per asymmetric unit. The final R factor was44% with a correlation coefficient of 62%. This structureand the other structures presented here were refinedfurther in CNS.48 Protein atoms were initially refined bysimulated annealing, followed by a few cycles of conju-gate gradient minimization and water picking. Finally,temperature factors were refined. No restraint for non-crystallographic symmetry was applied. The program Owas used for further adjustment and modeling of proteinatoms, ligands and water molecules.49 The backbonegeometry was checked in PROCHECK,50 and none ofthe residues was in the disallowed region.

Diffraction and refinement statistics for all structuresare summarized in Table 2.

Protein Data Bank accession codes

PDB coordinates have been deposited in the RCSBProtein Data Bank with accession codes 1OYK, 1OYL,1OZR, 1OZE, 1OZW and 1OZL.

Acknowledgements

This work was supported by NIH grantsGM33688 (to T. L. P.) and DK30297 (to P. R. O. M.)We thank Jayshali Lad and George Meigs duringprotein purification and for technical assistanceduring data collection at ALS.

Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1 535

References

1. Joshi, J. G. & Clauberg, M. (1988). Ferritin: an ironstorage protein with diverse functions. Biofactors, 3,207–212.

2. Tenhunen, R., Marver, H. S. & Schmid, R. (1968). Theenzymatic conversion of heme to bilirubin by micro-somal heme oxygenase. Proc. Natl Acad. Sci. USA,61, 748–755.

3. Tenhunen, R., Marver, H. S. & Schmid, R. (1969).Microsomal heme oxygenase. Characterization ofthe enzyme. J. Biol. Chem. 244, 6388–6394.

4. Ortiz de Montellano, P. R. (1998). Heme oxygenasemechanism: evidence for an electrophilic ferric per-oxide species. Acc. Chem. Res. 31, 543–549.

5. Ortiz de Montellano, P. R. (2000). The mechanism ofheme oxygenase. Curr. Opin. Chem. Biol. 4, 221–227.

6. Ortiz de Montellano, P. R. & Wilks, A. (2001). Hemeoxygenase structure and mechanism. Advan. Inorg.Chem. 51, 359–407.

7. Liu, Y. & Ortiz de Montellano, P. R. (2000). Reactionintermediates and single turnover rate constants forthe oxidation of heme by human heme oxygenase-1.J. Biol. Chem. 275, 5297–5307.

8. Yoshida, T. & Kikuchi, G. (1978). Features of thereaction of heme degradation catalyzed by thereconstituted microsomal heme oxygenase system.J. Biol. Chem. 253, 4230–4236.

9. Yoshida, T. & Kikuchi, G. (1978). Purification andproperties of heme oxygenase from pig spleen micro-somes. J. Biol. Chem. 253, 4224–4229.

10. Maines, M. D. (1988). Heme oxygenase: function,multiplicity, regulatory mechanisms, and clinicalapplications. FASEB J. 2, 2557–2568.

11. Schacter, B., Nelson, E., Marver, H. & Masters, B. S.(1972). Immunochemical evidence for an associationof heme oxygenase with the microsomal electrontransport system. J. Biol. Chem. 247, 3601–3607.

12. Yoshida, T., Noguchi, M. & Kikuchi, G. (1980).Oxygenated form of heme-heme oxygenase complexand requirement for second electron to initiate hemedegradation from the oxygenated complex. J. Biol.Chem. 255, 4418–4420.

13. Maines, M. D. (1992). Heme Oxygenase: ClinicalApplications and Functions, CRC Press, Boca Raton,FL pp. 203–266.

14. Suematsu, M. & Ishimura, Y. (2000). The hemeoxygenase-carbon monoxide system: a regulator ofhepatobilary function. Hepatology, 31, 3–6.

15. Dunford, H. B. (1999). Heme Peroxidases (Dunford,H. B., ed.), Wiley/VCH, New York.

16. Poulos, T. L. & Kraut, J. (1980). The stereochemistryof peroxidase catalysis. J. Biol. Chem. 255,8199–8205.

17. Poulos, T. L., Cupp-Vickery, J. & Li, H. (1995). Struc-tural studies on prokaryotic cytochromes P450. InCytochrome P450: Structure, Mechanism andBiochemistry (Ortiz de Montellano, P. R., ed.), 2ndedit., pp. 125–150, Plenum Press, New York.

18. Wilks, A. & Ortiz de Montellano, P. R. (1993). Ratliver heme oxygenase. High level expression of atruncated soluble form and nature of the

Table 2. Data collection and refinement statistics

D140A-L D140A-H WT WT-FeII D140A–FeII WT-NO D140A–NO

PDB code 1OYK 1OYL 1QQ8 1OZR 1OZE 1OZW 1OZLA. Crystal dataRadiation source In-house Synchrotron Synchrotron Synchrotron In-house Synchrotron SynchrotronSpace group P212121 P21 P21 P21 P21 P21 P21

Cell parametersa (A) 55.67 61.59 61.44 61.85 61.66 61.68 61.44b (A) 77.66 54.70 54.53 54.65 54.37 54.50 54.42c (A) 100.32 71.14 70.97 72.14 71.75 70.92 71.00b (deg.) 99.57 99.05 99.28 99.82 98.81 99.23

B. Data collectionDetector distance (nm) 130 140 130 150 150 130 140Molecules per asymmetric unit 2 2 2 2 2 2 2Resolution (A) 2.59 1.59 1.50 1.74 2.19 1.55 1.58Mosaicity (deg.) 1.43 0.57 0.46 0.78 0.61 0.81 0.74Total observations 135,214 297,413 273,199 233,738 157,747 355,458 208,425Unique reflections 33,028 63,070 74,186 96,379 30,463 1,45,019 63,001Completeness (%) 98.7 98.4 99.7 99.7 95.1 99.5 99.1Mean I/s 7.5 (2.01) 11.1 (2.51) 11.5 (2.2) 11.4 (2.12) 14.9 (2.29) 14.0 (2.14) 13.0 (2.14)Rsym (%) 6.6 (53.2) 3.8 (42.4) 5.1 (48.4) 4.6 (50.1) 10.5 (60.3) 4.7 (51.6) 3.8 (40.1)

C. Refinement statisticsRcryst

a 0.235 0.224 0.159 0.209 0.209 0.212 0.205Rfree 0.291 0.243 0.193 0.232 0.260 0.237 0.222rmsdb bond lengths (A) 0.008 0.007 0.006 0.006 0.007 0.005 0.007rmsdb angles (deg.) 1.3 1.1 1.1 1.2 1.2 1.1 1.2Water molecules 99 426 494 366 404 438 459

Ramachandran anglesMost favored (%) 89.6 93.0 93.8 93.7 94.3 93.5 92.5Additionally allowed (%) 10.4 7.0 6.2 6.3 5.7 6.5 7.5Generously allowed (%) 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Values in parentheses are for the outermost shell.a Rcryst ¼

PðlFobsl2 lFcalclÞð

PlFobsl: The Rfree is the Rcryst calculated on the 5% reflections excluded from refinement.

b rms bond and rms angle represent the root-mean-squared deviation between the observed and ideal values.

536 Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1

meso-hydroxylating species. J. Biol. Chem. 268,22357–22362.

19. Wilks, A., Torpey, J. & Ortiz de Montellano, P. R.(1994). Heme oxygenase HO-1. Evidence for electro-philic oxygen addition to the porphyrin ring in theformation of alpha-meso-hydroxyheme. J. Biol. Chem.269, 29553–29556.

20. Torpey, J. & Ortiz de Montellano, P. R. (1996). Oxi-dation of the meso-methylmesoheme regioisomersby heme oxygenase. J. Biol. Chem. 271, 26067–26073.

21. Torpey, J. & Ortiz de Montellano, P. R. (1997).Oxidation of a-meso-formylmesoheme by hemeoxygenase. J. Biol. Chem. 272, 22008–22014.

22. Davydov, R. M., Yoshida, T., Ikeda-Saito, M. &Hoffman, B. M. (1999). Hydroperoxy-hemeoxygenase generated by cryoreduction catalyzes theformation of a-meso-hydroxyheme as detected byEPR and ENDOR. J. Am. Chem. Soc. 121,10656–10657.

23. Matera, K. M., Takahashi, S., Fujii, H., Zhou, H.,Ishikawa, K., Yoshimura, T. et al. (1996). Oxygen andone reducing equivalent are both required for theconversion of alpha-hydroxyhemin to verdoheme inheme oxygenase. J. Biol. Chem. 271, 6618–6624.

24. Liu, Y., Moenne-Loccoz, P., Loehr, T. M. & Ortiz deMontellano, P. R. (1997). Heme oxygenase-1, inter-mediates in verdoheme formation and the require-ment for reduction equivalents. J. Biol. Chem. 272,6909–6917.

25. Sakamoto, H., Omata, Y., Palmer, G. & Noguchi, M.(1999). Ferric a-hydroxyheme bound to hemeoxygenase can be converted to verdoheme bydioxygen in the absence of added reducing equiva-lents. J. Biol. Chem. 274, 18196–18200.

26. Migita, C. T., Fujii, H., Matera, K. M., Takahashi, S.,Zhou, H. & Yoshida, T. (1999). Molecular oxygen oxi-dizes the porphyrin ring of the ferric a-hydroxyhemein heme oxygenase in the absence of reducingequivalents. Biochim. Biophys. Acta, 1432, 203–213.

27. Ishikawa, K., Takeuchi, N., Takahashi, S., Matera,K. M., Sato, M., Shibahara, S. et al. (1995). Hemeoxygenase-2. Properties of the heme complex of thepurified tryptic fragment of recombinant humanheme oxygenase-2. J. Biol. Chem. 270, 6345–6350.

28. Wilks, A., Black, S. M., Miller, W. L. & Ortiz deMontellano, P. R. (1995). Expression and characteriz-ation of truncated human heme oxygenase hHO-1and a fusion protein of hHO-1 with human cyto-chrome P450 reductase. Biochemistry, 34, 4421–4427.

29. Schuller, D. J., Wilks, A., Ortiz de Montellano, P. R. &Poulos, T. L. (1999). Crystal structure of human hemeoxygenase-1. Nature Struct. Biol. 6, 860–867.

30. Sugishima, M., Omato, Y., Kakuta, Y., Sakamoto, H.,Noguchi, M. & Fukuyama, K. (2000). Crystal struc-ture of rat heme oxygenase-1 in complex with heme.FEBS Letters, 471, 61–66.

31. Lad, L., Schuller, D. J., Shimizu, H., Friedman, J., Li,H., Ortiz de Montellano, P. R. & Poulos, T. L. (2003).Comparison of the heme-free and -bound crystalstructures of human heme oxygenase-1. J. Biol.Chem.278, 7834–7843.

32. Takahashi, S., Ishikawa, K., Takeuchi, N., Ikeda-Saito,M., Yoshida, T. & Rousseau, D. L. (1995). Oxygen-bound heme–heme oxygenase complex: evidencefor a highly bent structure of the coordinatedoxygen. J. Am. Chem. Soc. 117, 6002–6006.

33. Sugishima, M., Sakamoto, H., Higashimoto, Y.,Omato, Y., Hayashi, S., Noguchi, M. & Fukuyama,K. (2002). Crystal structure of rat heme oxygenase-1

in complex with heme bound to azide: implicationfor regiospecific hydroxylation of heme at the a-mesocarbon. J. Biol. Chem. 277, 45086–45090.

34. Lightning, L. K., Huang, H., Moenne-Loccoz, P.,Loehr, T. M., Schuller, D. J., Poulos, T. L. & Ortiz deMontellano, P. R. (2001). Disruption of an active sitehydrogen bond converts human heme oxygenase-1into a peroxidase. J. Biol. Chem. 276, 10612–10619.

35. Fujii, H., Zhang, H., Tomita, T., Ikeda-Saito, M. &Yoshida, T. (2001). A role for highly conserved car-boxylate, aspartate-140, in oxygen activation andheme degradation by heme oxygenase-1. J. Am.Chem. Soc. 123, 6475–6484.

36. La Mar, G. N., Asokan, A., Espiritu, B., Yeh, D. C.,Auclair, K. & Ortiz de Montellano, P. R. (2001). Sol-ution 1H NMR of the active site of substrate-bound,cyanide-inhibited, human heme oxygenase: compari-son to the crystal structure of the water-ligated form.J. Biol. Chem. 276, 15676–15687.

37. Li, Y., Syvitiski, R., Auclair, K., Wilks, A., Ortiz deMontellano, P. R. & La Mar, G. N. (2002). SolutionNMR characterization of an unusual distal H-bondnetwork in the active site of the cyanide-inhibitedhuman heme oxygenase complex of the symmetricsubstrate 2,4-dimethyldeuterohemin. J. Biol. Chem.277, 33018–33031.

38. Syvitski, R. T., Li, Y., Auclair, K. & L Mar, G. N.(2002). 1H NMR detection of immobilized watermolecules within a strong hydrogen-bonding distalnetwork in substrate-bound human heme oxygenase.J. Am. Chem. Soc. 124, 14296–14297.

39. Schlichting, L., Berendzen, J., Chu, K., Stock, A. M.,Maves, S. A., Benson, D. E. et al. (2000). The catalyticpathway of cytochrome P450 at atomic resolution.Science, 287, 1615–1622.

40. Berglund, G. I., Carlsson, G. H., smith, A. T., Szoke,H., Henriksen, A. & Hajdu, J. (2002). The catalyticpathway of horseradish peroxidase at highresolution. Nature, 417, 463–468.

41. Liu, Y., Lightening, L. K., Huang, H., Moenne-Loccoz, P., Schuller, D. J., Poulos, T. L. et al. (2000).Replacement of the distal glycine 139 transformshuman heme oxygenase-1 into a peroxidase. J. Biol.Chem. 275, 34501–34507.

42. Davydov, R., Kofman, V., Fujii, H., Yoshida, T., Ikeda-Saito, M. & Hoffman, B. M. (2002). Catalytic mechan-ism of heme oxygenase through EPR and ENDOR ofcryoreduced oxy-heme oxygenase and its Asp 140mutants. J. Am. Chem. Soc. 124, 1798–1808.

43. Rivera, M., Caignan, G. A., Astashkin, A. V.,Raitsimring, A. M., Shokhireva, T. K. & Ann Walker,F. (2002). Models of the low-spin iron (III) hyder-peroxide intermediate of heme oxygenase: magneticresonance evidence for thermodynamic stabilizationof the dxy electronic state at ambient temperatures.J. Am. Chem. Soc. 124, 6077–6089.

44. Schuller, D. J., Wilks, A., Ortiz de Montellano, P. R. &Poulos, T. L. (1998). Crystallization of recombinanthuman heme oxygenase-1. Protein Sci. 7, 1836–1838.

45. Leslie, A. G. W. (1992). Joint CCP þ ESF-EAMCB.Newsletter Protein Crystallog, No. 26.

46. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.Methods Enzymol. 276, 307–326.

47. Navanza, J. (1994). AMoRe: an automated packagefor molecular replacement. Acta Crystallog. sect. A,54, 157–163.

48. Brunger, A. T., Adams, P. D., Clore, G. M., Delano,W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).

Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1 537

Crystallography and NMR system: a new softwaresuite for macromolecular structure determination.Acta Crystallog. sect. D, 54, 905–921.

49. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kieldgaard,M. (1991). Improved methods for binding proteinmodels in electron density maps and the location oferrors in these models. Acta Crystallog. sect. D, 54,1017–1019.

50. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &

Thorton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein struc-tures. J. Appl. Crystallog. 26, 283–291.

51. Kraulis, P. J. (1991). MOLSCRIPT: a program to pro-duce both detailed and schematic plots of proteinstructure. J. Appl. Crystallog. 24, 946–950.

52. Merritt, E. A. & Bacon, D. J. (1997). Raster3D—photo-realistic molecular graphics. Methods Enzymol. 277,505–524.

Edited by D. Rees

(Received 28 February 2003; received in revised form 15 April 2003; accepted 24 April 2003)

538 Structure of the Asp140Ala Mutant of Human Heme Oxygenase-1