As should be clear from the previous discussion, enzymes that utilize dinuclear active sites containing two divalent metal ions are both numerous and diverse, and the range of chemical reactions that they catalyze includes a variety of substrates containing both phosphate ester and amide bonds. In contrast, enzymes that utilize a mixed-valence dinuclear center that contains one divalent and one trivalent center exhibit much less diversity in all respects: identity of the metal ions, nature of the reactions catalyzed, and overall protein structure. As we shall see, all such enzymes known to date contain FeIII plus either FeII, ZnII, or MnII, the only physiological
reaction they catalyze is dephosphorylation of proteins containing phosphoserine, -threonine, and possibly -tyrosine residues, and their catalytic domains all consist of afold that contains a phosphoesterase signature motif, Asp–X–{Gly/His}–Xm–Gly–Asp–XX–{Tyr/X}–Xn–Gly–Asn–
His–{Glu/Asp}. Many additional enzymes are known that contain this motif, but which do not utilize a trivalent–divalent dinuclear metal center. Examples of the latter include 50-nucleotidase and the phage proteinphosphatase, which will be discussed at the end of this section.
8.24.3.1 Purple Acid Phosphatases
Purple acid phosphatases (PAPs)123–125 are also called tartrate-resistant acid phosphatases (TRAPs) or type 5 acid phosphatases (Acp 5s) inthe biomedical literature.126 PAPs are typically glycoproteins with a low pH optimum (5–6) for enzymatic activity. They are insensitive to inhibition by tartrate, and exhibit an intense pink or purple color. PAPs have been isolated from mammalian, plant, and microbial sources, and differ in metal content. Essentially nothing is known about the physiological function of PAP in plants, and its function in many mammalian systems remains to be elucidated. Mammals appear to contain only a single copy of the gene encoding PAP, yet its expression levels vary tremendously in different tissues.127As a result, only two physiological roles are well established. High levels of uteroferrin(Uf) are present inthe intrauterine fluids of pregnant sows, whose high pH makes a role as a phosphatase unlikely.
Consequently, Uf has been suggested to play a role in transporting iron to the developing fetus.128,129 In contrast, osteoclasts secrete PAP (along with cathepsin K, which is presumably responsible for its proteolytic activation) into the acidified space between the osteo- clast and the bone surface, where PAP appears to facilitate bone resorption. PAP may help anchor the osteoclast to the bone by dephosphorylation of osteopontin,130,131 and it may also dephosphorylate bone matrix phosphoproteins. A role in bone resorption is supported by the phenotype of double knockout mice that lack both copies of the PAP gene: such mice exhibit osteopetrosis, with short, thick bones whose tissue has not been extensively remodeled.132PAP is also expressed in other cells of the mononuclear phagocyte system, including monocytes and macrophages,133 where its lysosomal locationsuggests a role indegradationof proteins arising from phagocytosis of bacteria and/or senescent erythrocytes. Recently, however, a number of tantalizing observations have been reported that suggest a much wider set of physiological functions for PAP. For example, it has been shown that PAP is expressed at high levels in antigen-presenting dendritic cells in epithelial tissues (skin, gut, lung, thymus, spleen),134,135and that mice lacking PAP exhibit a compromised immune response.133Most intriguing, however, is the observation that high levels of PAP are expressed in the trigeminal ganglion, brain, and spinal cord of rats, suggesting a significant role in the nervous system.136
Plant PAPs are dimers of 55 kDa subunits that contain either 1 Zn and 1 Fe (kidney bean PAP and one isoform of sweet potato PAP) or 1 Fe and 1 Mn per subunit (a second isoform of sweet potato PAP).137–140 The crystal structure of the kidney bean enzyme has been determined,141,142 and shows that the metal ions are located between and at the edge of the sandwiched-sheets, at the bottom of a shallow depressioninthe proteinsurface. As shownschematically inFigure 13A, the metal ions are bridged by a monodentate aspartate carboxylate and a water/hydroxide. In addition, the zinc is coordinated by two histidine imidazoles and an asparagine side-chain amide oxygen, while a histidine imidazole, a monodentate aspartate carboxylate, and a tyrosine pheno- late are coordinated to the iron. In the kidney bean PAP structure, two additional terminal solvent-derived ligands were modeled into the otherwise open coordination sites on the metal ions to give distorted octahedral geometries for both metal ions.
MammalianPAPs are monomers with a molecular mass of 36 kDa; all appear to containa dinuclear oxo- or hydroxo-bridged iron center at the active site. Crystal structures of two different forms of rat bone PAP143,144 and one form of PAP from porcine uterine fluids145 have been reported, and show that the structures are virtually identical to that of the catalytic domain of kidney bean PAP. The structure of the dinuclear metal site in the mammalian enzymes is indistinguishable from that shown in Figure 13A for the kidney bean enzyme, except for replacement of Zn2þ by Fe3þ; the diferric form is stabilized by the presence of a bridging phosphate ion in two of the structures and a bound sulfate in the other. Particularly noteworthy is the presence in all structures of a tyrosinate ligand to the trivalent metal and an O-bonded Asn amide ligand to the divalent metal site. The characteristic purple color of these enzymes is due to a low-energy ligand-to-metal charge transfer band from the tyrosinate to the ferric ion, while the
asparagine ligand plays a crucial role in proteolytic activation of PAP,146,147which is presumed to be a physiologically relevant regulatory mechanism. It seems clear that the distribution of negatively charged residues favors Fe3þ binding to the trivalent site, while the preponderance of neutral ligands favors binding of divalent metal ions at the other site, thus insuring the proper asymmetric distribution of the metal ions. To date, no metals other than Ca have been reported to be present in microbial PAPs, which have received very little attention.
The best characterized PAPs are those from bovine spleen (BSPAP) and porcine uterine fluids (uteroferrinor Uf), which exhibit >90% homology intheir amino acid sequences148,149 and congruent spectroscopic properties.123–125 Studies onthese two systems have beenlargely com- plementary in nature, and have led to the following conclusions. (i) Both enzymes exist in inactive oxidized (FeIIIFeIII) and active reduced (FeIIIFeII) forms; at T<30 K the latter exhibits a g0=1.7 EPR signal diagnostic of the mixed-valence dinuclear center.150–152(ii) The purple or pink color is due to a tyrosinate-to-ferric iron charge transfer transition, whose intensity and energy is modulated by oxidation state, protein conformation, anion binding, etc.151,153,154
(iii) The Fe atoms in the oxidized enzyme are relatively strongly coupled magnetically (2J80 cm1), while those inthe reduced form are much more weakly coupled (2J10–20 cm1), consistent with protonation of a bridging (oxo? hydroxo?) ligand upon reduction.151,155,156
(iv) Isotropically shifted NMR resonances are consistent with coordination of one Tyr phenoxide and one His imidazole to the ferric site and one His imidazole to the ferrous site in the reduced form.156–158(v) Phosphate forms a tight 1:1 complex with the oxidized form of the enzyme, which involves coordination of phosphate to the Fe ions in a bidentate bridging mode.159 (vi) Tetrahedral oxoanions such as phosphate and molybdate bind to the mixed-valence di-iron center in the reduced form of the enzymes.160–162 (vii) The FeII inthe reduced enzyme canbe replaced by ZnIIwith 100% retention of activity,160,163,164
as well as by other divalent metal ions such as CoII. The FeIIIcanbe replaced by Ga3þor Al3þwith retention of activity.165
The use of a mixed-valent, dinuclear iron site, similar to those in hemerythrin and ribonucleo- tide reductase,124,166,167 to catalyze a nonredox reaction such as phosphate ester hydrolysis is novel and unexpected for a variant of the familiar oxo(hydroxo)-bridged diiron center. In contrast to the general agreement that exists regarding the spectroscopic and physical properties of the PAPs, their kinetics properties and especially their mechanism of actionremaincontroversial.
Much of the disagreement stems from the different pH dependences of the catalytic activity of BSPAP and Uf, which is due to the fact that the former is isolated in a proteolytically activated form while the latter is not. Proteolysis results in a substantial increase in optimal pH in addition to anincrease incatalytic activity at the optimal pH.146Current data suggest that many of the spectroscopic studies described inthe literature were performed ona catalytically inactive form of the enzyme.168 As a result, the roles of the trivalent and divalent metal ions in catalysis and in particular the identity of the nucleophilic hydroxide that directly attacks the phosphate ester169 remainunresolved.
The three most likely mechanistic possibilities for catalysis of phosphate ester hydrolysis by PAPs are showninFigure 14. Each mechanism utilizes a different hydroxide as the nucleophile: a
FeIII ZnII O
H(?) O
O His 323 O Asp164
His 286
Asn 201 H2N
O O
Asp 135
His 325 Tyr 167
FeIII ZnII O
H(?) O
O His 323 O Asp 164
His 286
Asn 201
H2N O
O
Asp 135
HO His 92
O O
P
O O
(A) (B)
Figure 13 Schematic drawings of the active sites of (A) kidney bean purple acid phosphatase and (B) the phosphate complex of calcineurin.
terminal OH onthe Fe3þ; an OH inthe second coordinationsphere of the Fe3þ; or the bridging OH, respectively. Mechanism 1 is based on analogy to the rather well-established mechanism of the Zn-containing alkaline phosphatases, and utilizes the greater Lewis acidity of Fe3þ vs. Zn2þto generate a metal-bound OHnucleophile at pH<7, well below the pH range accessible with a divalent transition metal. In this case, the phosphate ester substrate is assumed to bind to the divalent metal site, which acts as a Lewis acid to activate it for nucleophilic attack.141,170 Mechanism 2 invokes the bridging hydroxide as nucleophile, and is based on a plausible mechanism for hydrolysis of urea by urease31 and a variety of spectroscopic studies of Uf.171,172This mechanism assumes that phosphate binds to the mixed-valent dinuclear center in the same way as it does to the diferric center, and it has most recently been invoked to explain the failure to observe a terminal hydroxide or water bound to the Fe3þby ENDOR spectroscopy.173 It is, however, not consistent with results regarding the pH dependence of phosphate or substrate binding in steady-state and pre-steady-state kinetics studies,168and it seems unlikely to be able to account for the high activity of the AlZn form of PAP.174Mechanism 3 utilizes the Fe3þ-bound OHonly as a base to generate a nucleophilic hydroxide in the second coordination sphere of the Fe3þ. It is analogous to a plausible mechanism for the nitrile hydratases, which contain mono- nuclearlow-spinFe3þor Co3þcenters.175,176The rates of ligand exchange of these metal ions are expected to be too slow to account for catalysis, if breaking a bond between the kinetically inert trivalent metal and the product were part of the mechanistic pathway. This mechanism was invoked to explain the fact that the Al3þZn2þ form of BSPAP is as active a catalyst as the Fe3þFe2þ or Fe3þZn2þ forms, despite the 100-fold lower rate of ligand exchange for Al3þ vs.
Fe3þ.174Regardless of which mechanism is correct, the ability to replace Fe2þby Zn2þand vice versa, with full retention of activity, suggests that it is likely to be applicable to both the dinuclear ironsite inthe mammalianPAPs and the FeZnsite inthe plant PAPs. Inall cases, the key feature appears to be the use of a trivalent–divalent dinuclear site for phosphate ester hydrolysis.
8.24.3.2 Protein Phosphatases 1,2A,and 2B
Proteinphosphatases (PPs) are a class of mammalianregulatory enzymes that catalyze the dephosphorylation of phosphoserine and phosphothreonine residues in proteins.177,178 Specific processes known to be controlled by protein phosphatases include glycogen metabolism, muscle contraction, mitosis, and transduction of hormonal signals. In the last 15 years, it has become
Fe3+
HO M2+
CH3 CH3
Fe3+
HO M2+ Fe3+
H O M2+
Fe3+
HO M2+
Fe3+
H O M2+
Fe3+
H O M2+
O
O O P
O O
O O P
O OH O
P O O
O
O P O OH O
O P O O
OH ROH CH3
ROH HOPO32–
HOPO32–
HOPO32–
HOPO32–
ROH
CH3
1 3
2 H
H H
O
Figure 14 Catalytic mechanisms proposed for the purple acid phosphatases: (1) attack of a terminal hydroxide onthe Fe3þon a monodentate phosphate ester substrate coordinated to the divalent metal site;
(2) attack of the bridging hydroxide on a bridging phosphate ester; (3) attack of a hydroxide ion generated in the second coordination sphere of the Fe3þon a monodentate phosphate ester.
clear that protein phosphatases are as important as protein kinases in regulating a wide variety of metabolic processes via dephosphorylation-induced conformational changes with concomitant increases or decreases in the activity of key enzymes.
The known protein phosphatases can be divided into two types, based on their inhibition by two endogenous heat stable proteins (inhibitors 1 and 2) and their preference for dephosphorylating thevs.subunits of phosphorylase kinase.179,180Phosphoproteinphosphatase 1 (PP1), which is inhibited by inhibitors 1 and 2 and preferentially dephosphorylates the -subunit of phosphor- ylase kinase, consists of a 38 kDa catalytic subunit complexed with either a larger glycogen- binding subunit, myosin, or inhibitor 1 or 2, depending on the source. The type 2 enzymes, which are insensitive to inhibitors 1 and 2 and preferentially dephosphorylate the -subunit of phos- phorylase kinase, are further divided into three classes. PP2A is a cytosolic enzyme that possesses broad reactivity, and contains a ca. 36 kDa catalytic subunit. PP2B, also known as calcineurin (CaN), is a Ca- and calmodulin-binding enzyme that dephosphorylates several brain-specific phosphoproteins; it consists of a 60 kDa catalytic subunit and a smaller Ca-binding 19 kDa regulatory subunit.181 In contrast, PP2C consists of a single 46 kDa subunit. The catalytic subunits of PP1, PP2A, and PP2B are apparently closely related; the amino acid sequences deduced from the corresponding DNAs show>50% homology,180,182placing all of these enzymes into the so-called PPP superfamily of phosphoprotein phosphatases; the PPPs are among the most evolutionarily conserved proteins known.177,182The PP2B sequence appears to be the result of fusion of a protein analogous to the PP1 and PP2A catalytic subunits with another domain that is responsible for binding its regulatory subunit, CaNB. In contrast, the amino acid sequence of PP2C is not related to those of the other PPs, and based on its sequence PP2C is a member of the evolutionarily unrelated PPM superfamily of phosphoprotein phosphatases. It contains a diman- ganese center, which was discussed inSection8.24.2.3.2.
The physiological importance of protein phosphatases is indicated by the fact that calcineurin is the primary intracellular target of two different immunosuppressant drugs, cyclosporin and FK506. The complex of cyclosporin with its endogenous binding protein, cyclophilin, and the complex of FK506 with its binding protein, FKBP, form a tight complex with the CaN catalytic/
regulatory subunit complex, which results in complete inhibition of protein phosphatase activity in vitrowith phosphoproteinsubstrates.183,184CaN activity appears to be essential for both T cell receptor and interleukin-2 (IL-2) receptor signaling pathways. Its inhibition by the drug-binding protein complexes interferes with the translocation of a nuclear activating factor, resulting in inhibition of DNA translation in T-lymphocyte nuclei and preventing the immunoglobulin E (IgE)-stimulated release of secretory granules containing tissue inflammation mediators such as histamine.183,184 Most intriguingly, the immunosuppressant-inhibited form of CaN is actually twiceas active as the uninhibited form toward small molecule substrates such aspara-nitrophenyl phosphate.184 This finding suggests that inhibition is due to steric blockage of access of phos- phoprotein substrates to the active site, rather than to binding to the active site.
For many years, no relationship was suspected between the PAPs and PPs, despite the seminal report by King and Huang in 1984 of the presence of equimolar Fe and Zn in bovine brain calcineurin.185 In particular, the amino acid sequences of PPs and PAPs show no evidence whatsoever for an evolutionary relationship between these two classes of enzymes. Based on the existence of regions of the amino acid sequences of PAPs vs. PP1s, PP2As, and PP2Bs that had highly conserved patterns of potential metal ligands, however, in 1990 it was proposed that calcineurin and the homologous PP1s and PP2As would prove to be metalloenzymes with FeZn active site structures similar to those of the plant PAPs.186 This hypothesis was abundantly confirmed by the publication in 1995 of two structures of bovine brain CaN,187,188 whose dinuclear metal site is shown schematically in Figure 13B. Comparisonof this structure to that of the active site of kidney bean PAP inFigure 13A demonstrates that the two structures are highly congruent, with one significant difference: the TyrOligand to the Fe3þhas effectively been replaced by a water-derived species, presumably a terminal hydroxide ion. In particular, the unusual O-bonded Asn ligand to the divalent metal site is present in both structures, as are the two conserved His residues near the dinuclear site that are postulated to be involved in substrate binding. In addition, the CaN structure clearly shows the presence oftwoterminal water-derived species on the trivalent metal: one is that which effectively replaces the TyrOligand, and the other corresponds to the terminal ferric-hydroxide proposed to act as the nucleophile or base in the PAP mechanism. Two crystal structures have also been reported for PP1s,189,190 and show active site structures that are essentially identical to that of CaN. Unfortunately, however, the identity and the oxidation state of the metal ions in the PP1 structures are unclear, with Fe, Mn, and Co reported as possibilities. In both structures, the metal ions were regarded as divalent, but
there is no evidence in the structures to support this assignment. Unexpectedly, given the differences in amino acid sequences between PAPs and PPs, the two sets of enzymes have very similar protein folds and virtually identical locations of the active site within these folds.
Given the similarity in active site structures, one might expect extensive mechanistic similarities between the PAPs and PPPs as well. Unfortunately, however, there is as yet no evidence that this is the case. A major reason is the lack of published mechanistic studies on PPPs. The only such studies have reported that hydrolysis of phosphate esters by CaN occurs via an unprecedented random two-step mechanism, in which either the alcohol product or phosphate is released first.191 In addition, general agreement regarding the oxidation states of the metal ions in the PPP active sites is lacking. An elegant series of studies on bovine brain calcineurin has shown that the Zn2þinCaN canbe replaced by Fe2þto give anactive enzyme with ag01.7 EPR signal that is virtually identical to that of the mixed-valent form of the mammalian PAPs.192,193 Nonetheless, data have been reported showing that superoxide dismutase protects CaN against inactivation by superoxide, which has been interpreted as favoring the presence of an Fe2þZn2þdinuclear center inthe active form of CaN.194 Giventhe tendency of O2 to act as a one-electron reductant in promoting Fenton-type chemistry of H2O2, however, it seems more likely that O2
is actually inactivating CaN byreducingthe ferric ion. This point must be unambiguously established before a detailed chemical mechanism can be considered. Several studies have, however, suggested that calcineurin is subject to redox regulationin vivo, due either to a change in the oxidation state of the ironor redox-active cysteine residues.195,196
8.24.3.3 Related Systems
Two other well-characterized hydrolytic enzymes are known that also possess a very similar protein structure, together with the phosphoesterase signature motif and a dinuclear metal center: the phage proteinphosphatase (PP) and the 50-nucleotidase (50-NT) from E. coli.237Despite the existence of active sites that are either identical (PP) or very similar (50-NT) to those in PP1 and PP2B, these two enzymes apparently utilize a dinuclear center containing two divalent metal ions (Mn2þinPP, Zn2þin50-NT) rather than a trivalent–divalent center.
Of these two enzymes,PP is the better characterized biochemically. Although its physiological function remains obscure, it represents in many ways a minimal model for a PPP, and, more importantly, it lacks the biological complexity of the mammalian enzymes with regard to regu- latory subunits, calcium activation, etc. The enzyme is isolated as an inactive apoenzyme that requires the presence of two Mn2þions for activity. The binding constant for the second Mn2þis about 100 times weaker thanfor the first. The enzyme fully loaded with Mn2þ exhibits anEPR spectra due to anexchange-coupled Mn2þdimer,197and this spectrum is perturbed by phospho- noacetohydroxamic acid and tetrahedral oxoanions that are competitive inhibitors of the enzyme, suggesting that substrates are hydrolyzed at the dinuclear site.198The crystal structure ofPP199 shows that the identity and arrangement of the protein ligands to the dinuclear active site are essentially identical to those in PP2B, which clearly contains Fe3þand Zn2þ. It remains unclear why thePP active site prefers to bind twodivalentmetal ions rather than a trivalent/divalent pair as in the PAPs and the PPs. The crystal structure also contains two types of coordinated sulfate ions in different molecules within the asymmetric unit, which may model the interaction with phosphate and phosphate esters. One sulfate is coordinated to Mn2 (the manganese coordinated to two histidines and an asparagine amide) in a monodentate fashion, but the other sulfate provides a triply bridging ligand to the active site: one O atom is bridging the metals, one O atom is coordinated to Mn1, and a third O atom is coordinated to Mn2. Based on this sulfate binding mode, a mechanism for the action of PP has beenproposed that is very similar to Mechanism 2 discussed above for the PAPs. In this mechanism, the phosphate ester coordinated to Mn2 is attacked by the bridging hydroxide to give a triply bridged phosphate intermediate.
Heavy atom isotope effect studies indicate that the transition state is a dissociative one, involving significant PO bond cleavage.200Finally, site-directed mutagenesis studies have recently identi- fied the Mn2 site as the high affinity binding site.201This led to the suggestionthat the resting enzyme contains a single Mn2þ ion in the Mn2 site, and that the second Mn2þis derived from binding of a Mn2þcomplex of the phosphate ester substrate.
Incontrast, 50-NT is an enzyme that has been known since the 1960s, but its chemistry has not been examined as extensively. 50-NT was initially characterized as a periplasmic UDP-sugar hydrolase. As such, its likely function is to cleave UDP-glucose in the environment into uridine,