As noted above, the precise identity of the metal ions used by an enzyme in vivo is often problematical, especially for those enzymes whose low affinity for their native metal ions results in their being lost during purification. Nonetheless, it is clear that certain hydrolytic metallo- enzymes contain particular metal ions, which raises the interesting question of how the properties of a given metal ion make it peculiarly suited for catalyzing a particular reaction. As we shall see below, for example, urease contains two NiIIions as isolated, and forms containing other metal ions such as MnII are much less active. In contrast, arginase, which catalyzes a reaction that is superficially very similar to that of urease, generally contains MnII as isolated, and achieves maximal activity with two MnIIper subunit. Unlike urease, however, arginase exhibits significant activity when reconstituted with a variety of other metal ions, including NiII! Similarly, amino- peptidases are known that contain either two ZnII or, inone possible case, two CoII ions. In addition, two series of Ser/Thr protein phosphatases are known, one of which utilizes a binuclear MnII center, while the other utilizes a trivalent–divalent dinuclear center containing FeIII and either ZnIIor FeII. Do systems such as these represent independent biochemical solutions to the same chemical problem using randomly selected metals, in which the observed metal ions simply reflect the evolutionary history of the system? Or do the properties of the metal ions somehow make them uniquely suited for these reactions, resulting in evolutionary pressure that selects against other metals? It seems clear that, in many systems at least, it is a gross oversimplification to view the metal ions as Lewis acids whose role in catalysis is dictated simply by their charge- to-radius ratios.
Inthis section, we will beginwith a considerationof urease, for which a great deal of spectro- scopic, kinetics, and structural information is now available, and which in many ways is a paradigm that illustrates the interplay between mechanistic, spectroscopic, and structural studies.
We will then proceed to systems containing dinuclear zinc, manganese, and cobalt centers, which will be discussed in considerably less detail. We will not, however, consider enzymes that employ dimagnesium centers in catalysis.5,6 These constitute a rapidly growing group, which includes ribonucleases,7 endonucleases,8 exonucleases,9 inositol poly- and monophosphate and fructose- 1,6-bisphosphate phosphatases,10 and ribozymes.11 Although these are biologically and in some cases medically important systems, the spectroscopic invisibility of magnesium makes the study of such enzymes a rather separate area, with little overlap of techniques with those typically used to study enzymes containing transition metal ions. The interested reader is referred to the lead references cited above and to Chapter 8.5.
8.24.2.1 Urease,a Dinickel Hydrolase
Urease allows bacteria, fungi, and higher plants to utilize urea as a nitrogen source for growth; it catalyzes the hydrolysis of urea to ammonia and carbamate, which subsequently hydrolyzes spontaneously to ammonia and bicarbonate (Equation(1)):
C + H2O NH4+ + NH2CO2D H2O
2 NH4+ + CO2 O
NH2 H2N
ð1ị
Urease from jack beans was the first enzyme to be crystallized, and based on his chemical analysis of the crystalline enzyme, Sumner in 1926 made the revolutionary proposal that enzymes
were pure proteins that contained only amino acids.12 In1975, however, Zerner et al. demon- strated convincingly that urease was in fact a nickel enzyme that contained two nickel atoms per catalytic site.13,14Subsequent studies have demonstrated that the enzyme consists of an6hexa- mer, with one active site per 90.4 -kDa subunit.15 Initial kinetics and spectroscopic studies were carried out onthe jack beanenzyme, but most recent work has focused onbacterial enzymes, particularly those from Klebsiella aerogenes, Bacillus pasteurii, and, most recently, Helicobacter pylori. Although the latter are trimers of asymmetric trimers, with anoverall ()3architecture and two nickel atoms per subunit, they exhibit >50% sequence homology to the jack bean enzyme.16,17 In addition, all spectroscopic data reported to date indicate that there are no significant differences between the Ni sites of the bacterial and plant enzymes.18
Most surprisingly, the activity of apo-urease cannot be restored by reconstitution with most other divalent metal ions, such as ZnII, CoII, an d CuII, although MnIIgives A˚2% of the activity of the reconstituted NiIIenzyme.19,20The unique role played by nickel in urease is emphasized by the fact that three other proteins are required for incorporation of nickel into urease21. Inaddition, a fourth protein binds nickel and appears to function as a metallochaperone that provides nickel for insertion into urease.22,23 Carbon dioxide is also required for nickel binding to apo-urease,24 which is explained by the incorporation of CO2into a carbamylated Lys side chain that bridges the two metal ions (see below). Not surprisingly, CS2canreplace CO2to give a nickel-containing protein that presumably has a bridging dithiocarbamate ligand, but this form of the enzyme is completely inactive.25 Of the other species examined, only vanadate could replace CO2 inthe activation process, giving an active enzyme with kinetic properties essentially identical to those of the wild-type enzyme; this form of the enzyme presumably contains a vanadylated Lys residue in which the vanadate bridges the two nickels.25
X-ray structure determinations of three bacterial ureases have now been reported, as have those of a number of mutant enzymes and enzyme–inhibitor complexes. All structures of the native enzymes agree on the basic protein architecture: the enzyme is a trimer oftrimers with three- fold symmetry.26–28Thesubunit (60.3 kDa) is the largest and consists of two domains, the larger of which is an(/)8 barrel that contains the dinuclear nickel center. All three structures also contain a mobile flap that covers the active site and contains residues that interact with the active site.
Inthe first structure reported, that ofK. aerogenesurease at 2.2 A˚ resolution,26the dinickel site was well resolved, with the two nickel ions separated by 3.5 A˚ and bridged by the carbamylated lysine residue. Adjacent to the dinickel site was a small, solvent-filled cavity approximately the size of a urea molecule. A more recent structure determination of the B. pasteurii enzyme, with slightly better resolution(2.0 A˚) and a more complete data set, gave a detailed picture of the active site and the solvent structure in the cavity.27As showninFigure 2, the two nickel ions are 3.5 A˚ apart, bridged by the bidentate lysine carbamylate and a water/hydroxide. In addition, Ni1 is coordinated by two histidine imidazoles and a water/hydroxide to give a distorted square pyramidal coordination geometry, while Ni2 is coordinated by two histidine imidazoles, a mono- dentate aspartate carboxylate, and a water/hydroxide, giving a distorted octahedral coordination geometry. These coordination geometries are consistent with a variety of spectroscopic results.18
Asp383 His137 His138
Lys220
His275
His249
Figure 2 The active site of native urease fromBacillus pasteurii(pdb 2UBP). Nickel atoms are indark gray, carbon atoms in light gray, oxygen atoms in white, and nitrogen atoms in black. Hydrogen bonding
interactions are indicated as dashed lines.
In addition, the structure shows the presence of a fourth water molecule hydrogen-bonded to the other three water/hydroxide ligands and to a sulfate ion from the crystallization buffer. Given the importance of water in the reaction that is catalyzed, the presence of an ordered and fully occupied array of water (hydroxide) inthe cavity is likely to be relevant to the catalytic mechanism.
Nickel canbe removed from urease by dialysis at low pH inthe presence of complexing agents to give an apoenzyme; the apoenzyme can also be isolated from bacteria in which one or more of the genes necessary for nickel insertion have been deleted. The structure of apo-urease from K. aerogenes shows several important features.29 As expected, the nickel ions are absent, and a single well-ordered water molecule is hydrogen bonded to the two histidine imidazoles that are ligands to Ni2 in the native enzyme. In addition, the carbamyl group on the active site lysine is also absent if nickel is not present. Most importantly, the basic architecture of the active site is essentially unchanged, indicating that the protein provides a rigid array of amino acid residues that coordinate to the nickel ions with minimal structural rearrangement.
The structures of more than a dozen different mutants of K. aerogenes urease have been determined at atomic resolution, some at multiple pHs and/or in the presence of inhibitors.29–31 Together with the observed effects of the mutations on the kinetics of the enzyme, these structures have yielded important insights into the function of many of the residues at or near the active site.
For example, His320 has variously beenproposed to act as a general acid incatalysis32or as a hydrogen bond acceptor that stabilizes an incipient positive charge on the urea.27Conversion of His320 to Ala, Asn, or Gln gives a 105 decrease in kcat, with little change in Km. For the His320Ala mutant, it is clear that changes in the positions of the waters in the active site result ina major change incoordinationgeometry at Ni1, which becomes pseudotetrahedral.29,31Thus, it remains unclear whether His320 functions as an acid or base in catalysis or whether it simply helps to order the water cluster and ensure the correct geometry at Ni1.
His219 does not interact directly with the active site or its cluster of waters, but it does point directly at the active site cavity with its presumably protonated N" atom, suggesting that it can hydrogen-bond to the urea oxygen and polarize the CẳO bond. Mutation of this residue to Ala, Asn, or Gln results in a 2–20-fold decrease inkcatand an 80–1,000-fold increase inKm, consistent with it playing an important role in both substrate binding and catalysis, butnotvia acid or base chemistry. The structures of these mutants show only modest changes in the conformation of the flap and the water cluster.29,31
Other mutations result in major changes in the structure of the active site and, predictably, completely abolish activity. For example, removal of one of the histidine ligands to Ni2 in the His134Ala mutant produces an inactive enzyme with a mononuclear active site.33 Similarly, replacement of Lys217, which in its carbamylated form provides the bridging ligand between the nickel ions, with other residues such as Cys, Glu, or Ala gave inactive enzymes containing no nickel. These mutants could, however, be ‘‘rescued’’ by prolonged incubation with Ni2þand acids such as formate, which were shown to replace the carbamate in bridging the nickel ions, resulting inenzymes with modest activity.34
Perhaps the most informative structural studies have been those on inhibited forms of the enzyme, which include one containing a potential transition state analogue. A variety of kinetics and spectroscopic data, including EXAFS35 and the observation of strong antiferromagnetic coupling between the nickels,36 had demonstrated that thiolates were competitive inhibitors that bound to and bridged the nickel ions. This was confirmed by the structure of the -mercaptoethanol-inhibited enzyme from B. pasteurii, which showed that the thiolate replaced the bridging water; the -OH of the alcohol forms a chelate with Ni1.37 A similar geometry is observed for the enzyme complexed with acetohydroxamate, in which the hydroxamate oxygen bridges the two nickels, and the carbonyl oxygen coordinates to Ni1. Phosphate, which is a relatively weak, pH-dependent competitive inhibitor of urease (Ki=0.1 mM and 50 mM at pH 5 and 7, respectively), is found to form a complex in which three oxygen atoms are coordinated to the dinickel center: one O atom bridges the two metal ions, while two other O atoms form chelates to Ni1 and Ni2, respectively, with the fourth O atom pointing into the solvent channel.38Analysis of the hydrogen bonding suggests that phosphate binds as H2PO4. Finally, treatment of urease with phenylphosphorodiamidate, (PhO)PO(NH2)2, results inhydrolysis of the phenyl phosphate moi- ety to yield enzyme containing coordinated diamidophosphate, (H2N)2PO2
, which appears to be a transition state analog. The diamidophosphate binds to the dinickel center very similarly to phosphate, with one bridging O atom, the second O atom coordinated to Ni1, and one -NH2
bound to Ni2.
Let us now turn to a consideration of the mechanisms that have been proposed for hydrolysis of urea by urease. Due to the extreme stability afforded by its resonance energy, hydrolysis of urea is a difficult reaction, with an estimated half-life of 3–4 yr at 38C and pH 7. Yet urease accelerates the reactionby a factor of at least 1014, making it one of the most proficient enzymes known. Such a tremendous catalytic enhancement implies a truly extraordinary stabilization of the transition state, and a number of specific proposals have been made as to how this might occur. The first proposed mechanism was due to Zerner and co-workers,39 who suggested that one nickel acted to activate urea and one to activate water (Figure 3(A)), with the protein providing a carboxylate to stabilize a positive charge on one of the urea nitrogen atoms as well as a general acid (a cysteine?) and a general base.
More recently, a modified version of this same mechanism has been suggested based on the structural and kinetics results on theK. aerogenesenzyme.32The key features of this proposal are the interaction of urea with both nickel ions and the presence of a reverse protonation scheme.
In this scheme, a protonated His320 acts as a general acid, while the less acidic water coordinated to Ni2 is deprotonated and acts as the nucleophile that attacks the urea coordinated to Ni1. Thus, it is proposed that the active form of the enzyme consists of the one molecule in approximately 102.5 that contains both a deprotonated water ligand (estimated pKa&9.0) and a protonated His320 (estimated pKa&6.5). Such a reverse protonation scheme results in both a stronger acid and a more potent nucleophile, which in this model accounts for the ability of urease to overcome the lack of basicity of urea (estimatedpKa&2) and its inertness to nucleophilic attack. Thus, if this model is correct, the actual kcat/Km value for urease is approximately 108.5 rather than106, i.e., very near the diffusion-controlled limit.
Most recently, an alternative mechanism has been proposed based on the structural results for the phosphate and diamidophosphate complexes of the B. pasteurii enzyme.18,27 This proposal, which is also supported by computational studies in which inhibitors and potential transition states were docked into the active site,40 is shownschematically inFigure 3B. The key features of this mechanism are binding of urea in a bridging bidentate mode and the identity of the nucleophile that attacks the urea carbonyl group: the bridging hydroxide, which also acts as the required general acid. In addition, His320 is proposed to act as a general base that stabilizes
Ni Ni
O O
C
NH2
NH3
Ni Ni
OH O
C
NH2
NH2 (B)
Ni Ni
O C
O
Ni Ni
O C (A)
H2N
NH2 OH
H3N NH2
Figure 3 The two limiting mechanisms for urease: (A) attack of a monodentate hydroxide coordinated to one Ni on urea coordinated to the other, and (B) concerted attack of a bridging hydroxide on a bidentate
bridging urea.
the developing positive charge on the transition state. Recent results on inhibition of the K. aerogenesenzyme by fluoride, which is a slow-binding inhibitor that binds only to a protonated form of the enzyme, have been interpreted in terms of fluoride replacing the bridging hydroxide.41 These results are consistent with a bridging oxide derived from the bridging hydroxide as the nucleophile. The bridging hydroxide can be replaced by fluoride only when protonated to give a more labile bridging water. This proposal thus results in a picture very similar to that shown in Figure 3B, with the main point of remaining contention being the immediate source of the protonthat is transferred to urea: His320 vs. the bridging hydroxide. It is clear that we now have a rather complete understanding of the mechanism of urease, which we shall utilize frequently as a paradigm inour discussionof other dinuclear hydrolases.
8.24.2.2 Dizinc Hydrolases
Hydrolytic enzymes that contain dinuclear zinc sites are known that hydrolyze both phosphate ester and amide bonds, and the number of such enzymes is growing rapidly.42 Although each of the major types of enzyme to be discussed contains a distinctive protein fold that is unrelated to those of the other dizinc hydrolases, the active sites of these diverse enzymes exhibit a number of structural similarities, which suggests a convergent evolution of similar metal binding sites in otherwise unrelated proteins (analogous to the relationship between the mononuclear zinc hydro- lases, thermolysin and carboxypeptidase A). In addition, as we shall see, some of the dizinc hydrolases exhibit structural similarities to other metalloenzymes (including urease), suggesting an evolutionary process in which they have diverged from a common ancestor.
8.24.2.2.1 Phosphotriesterase
We will begin with phosphotriesterase, which is an unusual (if not unique) enzyme in that it has no naturally occurring substrates. The enzyme was originally isolated from soil bacteria that were able to hydrolyze and detoxify a wide variety of organophosphate insecticides and nerve toxins.
A typical reaction is the hydrolysis of the insecticide paraoxon, shown inEquation(2):
P O EtO
OEt O
NO2 + H2O EtO P OH OEt O
+ HO NO2
ð2ị In addition, phosphotriesterase is a relatively efficient catalyst for the hydrolysis of chemical warfare agents such as Sarin and VX. Because of its obvious relevance in bioremediation and detoxification of potential chemical warfare agents, this enzyme has been the object of intense investigation.43 By far the best-characterized example is the enzyme from Pseudomonas diminuta (henceforth abbreviated as PTE), on which almost all the relevant experiments have been performed. PTE is a dimeric proteinwith a subunit molecular weight of 20 kDa. It is isolated with up to two Zn bound per subunit, suggesting that zinc is the physiologically relevant metal ion. It is, however, possible to remove the bound zinc by treatment witho-phenanthroline to give an apoenzyme, which can be reconstituted with a variety of divalent metal ions, including cobalt, nickel, cadmium, manganese, and copper. All of these derivatives contain 2 mol of divalent metal per subunit, and all except the copper derivative exhibit significant enzymatic activity.44 As for a number of zinc enzymes, the dicobalt derivative is actually a somewhat better catalyst than the presumed native dizinc form. In fact, the order of catalytic efficiency for the substituted species is Co2þ–Co2þ>Ni2þ–Ni2þ>Cd2þ–Cd2þ>Zn2þ–Zn2þ>Mn2þ–Mn2þ,45which shows no obvious correlation with the charge to radius ratio of the metal ions.
High-resolution(1.3 A˚) crystal structures of the dizinc, dicadmium, dimanganese, and zinc–
cadmium forms of PTE are now available,46 which reinforce and expand upon many of the conclusions derived from earlier lower resolution structure determinations47–49 and studies of the EPR spectra of the dimanganese50and dicopper enzymes.44Inparticular, the two metal ions
are located in close proximity to one another at the C-terminal end of an ()8barrel structure.
As showninFigure 4, in the native enzyme the zinc ions are bridged by a water or hydroxide, as well as a carbamylated lysine residue like that found in urease. In addition, both zincs are coordinated by two histidine imidazoles each, with an aspartate carboxylate also coordinated to one zinc and a water to the other, giving distorted trigonal bipyramidal sites.46 Detailed quantum mechanical calculations on the PTE active site have been carried out, and emphasize the importance of additional water molecules in the vicinity of the active site.51,52
Both the overall protein architecture and the structure of the dinuclear center of phosphotries- terase are very similar to those of the nickel-binding domain of urease (cf.Figure 2), suggesting a possible evolutionary relationship between these enzymes. Unlike urease, however, which requires a set of accessory proteins for insertion of nickel, apo-phosphotriesterase readily incorporates zinc inthe presence of CO2to give the fully functional enzyme.53Although the ()8barrel motif is a commonfeature inthe structure of a wide variety of proteins, the elliptical barrel axis of the ()8
barrel of phosphotriesterase and the urease nickel-binding domain are rather rare. Only two other proteins with similar structures known26: adenosine deaminase, which contains a single zinc ion in place of the dinuclear zinc or nickel site,54and a protein of unknown function from Escherichia coli called phosphotriesterase homology protein (PHP), which contains a dinuclear zinc center very similar to that of phosphotriesterase, but inwhich a glutamate carboxylate replaces the carbamylated lysine as a bridging ligand.55Sequence alignments reveal the presence of homologues to PHP in both pathogenic microorganisms and in mammals,55but inno case has a physiological function been established. The fact that all other known members of this class of proteins are hydrolytic metalloenzymes strongly suggests that PHP also catalyzes a hydrolytic reaction, as did the presumed ancestor of this family of enzymes.
A variety of mutagenesis experiments has been performed to characterize the requirements for catalytic activity.56Conversion of Asp253, which is hydrogen-bonded to the His230 ligand to the more exposed zinc, to Asn had essentially no effect on catalytic activity, but replacement of this residue by Ala caused a 500-fold decrease inactivity. These results were interpreted interms of a role for the Asp in orienting the His ligand properly for binding to the dizinc site, rather than in terms of the Asp functioning as a proton donor or acceptor in the catalytic mechanism. The aspartate ligand to the five-coordinate zinc (Asp301) is clearly crucial: replacement by either coordinating residues (His, Asn, Cys) or a noncoordinating residue (Ala) caused a 102104 decrease inactivity. Finally, replacement of the carbamylated Lys residue by Arg, Met, Glu, or Ala caused (103105)-fold decreases inactivity. Insome cases, activity could be partially restored by incubation with bicarbonate or acetate, suggesting that these exogenous ligands can replace the bridging carbamate, as with urease.
Asp 301
His55 His57
Lys105
His201 His230
Figure 4 The active site of phosphotriesterase (pdb 1HZY). The zinc ions are in dark gray, carbon atoms in light gray, oxygen atoms in white, and nitrogen atoms in black. Hydrogen bonding interactions are indicated
as dashed lines.