Tryptophan hydroxylase (TrpH) and tyrosine hydroxylase (TyrH) catalyze the hydroxylation of the side chains of their aromatic amino acid substrates in the first step in the biosyntheses of serotonin and the catecholamine neurotransmitters, forming 5-hydroxytryptophan and 3,4-dihydroxyphenylalanine, respectively (Fig. 1). Together with phenylalanine hydroxylase (PheH), the two enzymes comprise the aromatic amino acid hydroxylase family of enzymes (1, 2). All three enzymes catalyze the iron-mediated incorporation of one atom of molecular oxygen into both the amino acid substrate and the reducing substrate tetrahydrobiopterin (BH4) to give the hydroxylated products. This review will focus on the present understanding of the catalytic mechanisms of TrpH and TyrH; the mechanism of PheH is the subject of a separate contribution to this issue.
TrpH and TyrH are both tetramers of identical subunits. Each monomer consists of an N-terminal regulatory domain, a highly conserved catalytic domain, and a C-terminal tetramerization domain (3–5). The catalytic domains are approximately 300 residues in length with greater than 60% sequence identity between the enzymes and essentially identical structures. All of the residues required for activity and substrate binding lie within the catalytic domain, so that the isolated catalytic domains are fully active (4, 6). Structures are available of the catalytic domains of rat (7) and human (pdb file 2XSN) TyrH and human (8) and chicken (9) TrpH, establishing their common structure (Fig. 2). The small C-terminal domains contain an α-helix of 25–27 residues responsible for tetramerization. The N-terminal regulatory domains are more variable; no high-resolution structures are presently available for the regulatory domains of either TrpH or TyrH. There are four splice variants of human TyrH, differing in the length of an insertion after Met30 (10, 11). The shortest form, with no insertion, has a regulatory domain of approximately 155 residues and is identical in length to the more-studied rat enzyme. The other forms contain inserts of 4, 27, or 31 residues. There does not appear to be any significant difference in the catalytic properties of the four forms (12); it is likely that differences have to do with regulation. Ser31 of isoform one of human TyrH is phosphorylated in vivo, so that insertions before this residue would alter the identity of the kinase responsible for phosphorylating it. There are two isoforms of human TrpH encoded by two different genes (13). TrpH1, the first to be discovered and the most studied, is found primarily in the gastric system, while TrpH2 is found mainly in the brain. The two human enzymes are 70% identical, with most of the differences in the regulatory domain, which contains approximately 88 residues in TrpH1 and is 46 residues longer in TrpH2. Mechanistic studies of TrpH have generally been restricted to the isolated catalytic domain of TrpH1 because of difficulties in expressing the intact protein in E. coli (14).
The active sites of TrpH and TyrH are highly conserved (Fig. 3). Both are nonheme iron enzymes, with two histidines (272 and 277 in TrpH, 361 and 366 in TyrH) and a glutamate (317/376) as the ligands to the iron (7, 8), which must be in the FeII form for activity (15, 6). While there is no structure of either TrpH or TyrH with both BH4 and an amino acid substrate bound, there are separate structures of chicken TrpH with tryptophan bound and human TrpH with dihydrobiopterin bound (8, 9). These structures and the structure of PheH with both BH4 and an amino acid bound (16) make it possible to describe the interactions of both TrpH and TyrH with substrates. BH4 interacts with the side chains of a glutamate (273/362) and a phenylalanine (241/330); the remaining interactions are with backbone atoms. The carboxylate of the amino acid substrate interacts with an arginine (257/346) and an aspartate (269/358). The side chain of the amino acid substrate is held in a hydrophobic pocket made up of a proline (268/357), a histidine (272/361) that is also a metal ligand, a phenylalanine (318/407), and either a phenylalanine (313) in TrpH or a tryptophan (402) in TyrH.
Given the similarities of the active sites, it is not surprising that the substrate specificities of the two enzymes are not absolute. TrpH readily hydroxylates both tryptophan and phenylalanine with comparable kinetics (14, 17, 18). In contrast, hydroxylation of tyrosine is approximately 5,000-fold slower (14). TyrH is able to hydroxylate all three aromatic amino acids (19, 20). Based on Vmax/KM values, TyrH prefers tyrosine as a substrate over phenylalanine by an order of magnitude and over tryptophan by 25-fold (21, 20). Both tyrosine and 3-hydroxyphenylalanine are formed when phenylalanine is the substrate for TyrH (22). A number of efforts have been made to change the substrate specificities of these enzymes. An obvious choice is the tryptophan in TyrH, Trp402, that is replaced by Phe313 in TrpH. The F313W mutation in TrpH results in an enzyme with a preference of about twofold for tryptophan over phenylalanine as substrate and no increase in activity with tyrosine as substrate (17, 20), while the reverse mutation in TyrH has no effect for the preference of that enzyme for tyrosine over tryptophan (20), so that this residue does not play a dominant role in determining the specificities of the two enzymes. As shown in Fig. 3, another difference between the active sites of the two enzymes is that Asp455 of human TyrH is replaced by Ile366 in human TrpH; this difference is found in all sequences of the two enzymes that are available. In PheH, this residue is replaced by a valine. The residue at this position does not directly contact the amino acid substrate but instead appears to play a role in controlling the shape of the hydrophobic pocket for the side chain of the substrate. Mutation of this aspartate in rat TyrH to valine converts the enzyme to a highly active PheH and greatly decreases its activity with tyrosine as a substrate (21). Mutagenesis of the corresponding isoleucine residue in TrpH does not appear to have been reported, but changing the valine in PheH to aspartate results in only a slight decrease in the preference of that enzyme for phenylalanine (21).
Mechanism of Oxygen Activation
As shown in Fig. 1, the pterin product of the reactions of TrpH and TyrH is a 4a-hydroxypterin. 4a-Hydroxypterins dehydrate relatively rapidly to quinonoid dihydropterins (Fig. 1) (23); this initially hindered the direct identification of the product of the enzyme-catalyzed reaction. The similarity of the UV absorbance spectra of the initial pterin products of the TrpH and TyrH reactions to that of synthetic and stable 4a-hydroxy-5-deazapterin, the NMR spectrum of the product of the PheH reaction, and the evidence that the product contains an atom of oxygen from O2 are fully consistent with the product of the enzyme-catalyzed reaction being a 4a-hydroxypterin (6, 24, 25).
All studies to date are consistent with a common catalytic mechanism for the three aromatic amino acid hydroxylases (26). The proposed mechanism is shown in Fig. 4 for tryptophan hydroxylation. The reaction can be described in two parts: 1) reaction of the tetrahydropterin, oxygen, and the active site iron to form the reactive hydroxylating intermediate and 2) insertion of oxygen into the amino acid substrate. These two partial reactions are tightly coupled in the wild-type enzymes, so that one molecule of the amino acid is hydroxylated for each molecule of BH4 oxidized. However, under certain conditions these enzymes can become uncoupled, catalyzing tetrahydropterin oxidation without catalyzing hydroxylation of the amino acid substrate. The pterin product of these uncoupled reactions varies. In some cases, the 4a-hydroxypterin is formed (6, 27), while in others the only detectable product is the quinonoid dihydropterin (28, 29). Oxygen kinetic isotope effects of the TyrH reaction have established that the initial step in the reaction of oxygen with that enzyme is the same whether or not amino acid hydroxylation occurs (30)
Neither enzyme will react productively with oxygen unless both an amino acid substrate and a tetrahydropterin are bound; instead the iron is simply oxidized to the unreactive FeIII form (29, 31, 32). Binding of the substrates results in several structural changes that generate the form of the enzyme that can react with oxygen to form the hydroxylating intermediate. In the TyrH reaction, the tetrahydropterin must bind before the amino acid substrate (33). Tetrahydropterin binding results in the closing of an active site loop over the active site (34); this likely results in conversion of the amino acid binding site to the productive form. Structures of TrpH also show movement of active site loops when both the amino acid and pterin sites are occupied (9). In addition, the ligands of the active site iron rearrange when both the amino acid substrate and tetrahydropterin are bound. In the absence of substrates, the ligands to the metal are two histidines, a glutamate, and three water molecules. This changes when the substrates are bound, with two of three water molecules dissociating and the glutamate changing from a monodentate ligand to a bidentate interaction (29). This decreases the number of metal ligands from six to five, opening up a site for oxygen. At present there is no information regarding the structures of subsequent intermediates due to their high reactivity.
The mechanism of the reaction of molecular oxygen with the enzyme–substrate complex to form the hydroxylating intermediate is not fully understood. The formation of a 4a-hydroxypterin as a product of the reaction establishes that there must be a reaction between the pterin and oxygen and that the tetrahydropterin does not simply serve as a passive source of two electrons. The nonenzymatic reaction of tetrahydropterins with oxygen (35, 36) serves as a model for the enzymatic reaction. The autoxidation reaction (Fig. 5, upper path) is initiated by the slow transfer of a single electron from the pterin to O2 to produce superoxide and a pterin radical. The pterin radical either reacts with another pterin radical to form a dihydropterin plus a tetrahydropterin or recombines with the superoxide to produce a 4a-peroxypterin. The latter species can eliminate H2O2 to form the quinonoid dihydropterin. In the enzyme active site, a 4a-peroxypterin could instead react with the active site FeII, either through a stepwise mechanism in which an FeII-peroxypterin intermediate is formed or in a single step via the direct transfer of an oxygen atom. A peroxypterin has not been detected in the reaction of any of the aromatic amino acid hydroxylases. However, when tyrosine is used as a substrate for PheH, very little tyrosine is formed and the tetrahydropterin is oxidized to yield quinonoid dihydropterin and hydrogen peroxide (37); this is consistent with the unproductive decay of a peroxypterin intermediate. The similar formation of quinonoid dihydropterin in uncoupled reactions of TyrH can also be attributed to breakdown of a peroxypterin intermediate. Flavoprotein phenol hydroxylases catalyze a similar reaction to that of TyrH and use a hydroperoxyflavin as the hydroxylating intermediate (38, 29). This raises the possibility of a peroxypterin as the hydroxylating intermediate in the TrpH and TyrH reactions. However, this can be ruled out by the formation of a 4a-hydroxypterin in some uncoupled reactions of TyrH (27) and by the ability of hydrogen peroxide to support hydroxylation of phenylalanine by TyrH and TrpH (39).
The reaction of TyrH and TrpH with oxygen is several orders of magnitude faster than the nonenzymatic autoxidation reaction (29, 40, 32). This suggests that oxygen may initially react with the active site iron to form an FeIII-superoxide intermediate that would then react with the tetrahydropterin to form an FeII-peroxypterin species (Fig. 5, lower path). An intermediate in which oxygen is bound to both the iron and the pterin is attractive, in that heterolytic cleavage of the oxygen–oxygen bond in such a species would generate directly the 4a-hydroxypterin product and an FeIVO intermediate. Mössbauer spectroscopy of the TyrH reaction has shown directly that an FeIV species forms within the first 100 msec of the reaction (41). This species decays as tyrosine is hydroxylated, consistent with it being the hydroxylating intermediate in TyrH. A similar intermediate has been detected in the reaction of a bacterial PheH (42). The spectral properties of these intermediates resemble those of the FeIVO intermediate formed during the reactions of the α-ketoglutarate-dependent family of nonheme iron hydroxylases (43). At present there is no direct information regarding the identity of any intermediates preceding this FeIVO intermediate in the reaction of an aromatic amino acid hydroxylase with oxygen, although a candidate has been detected in stopped-flow studies of the TrpH reaction (32).
Mechanism of Amino Acid Hydroxylation
The mechanism of hydroxylation of the amino acid is better understood. Early studies of TrpH contributed to the discovery of the “NIH shift” (44), a 1,2 migration of the functional group at the site of substitution in aromatic hydroxylation reactions. The hydroxylation of 5-3H-tryptophan by TrpH results in >85% retention of the tritium label in the product 5-HO-tryptophan (45). The migration of the label to the 4-position of the indole ring was initially established from the lability of the tritium in 10% trichloroacetic acid but not at neutral pH. More recently, migration of deuterium in 5-2H-tryptophan to the adjacent carbon was demonstrated directly by NMR (6). Similarly, when 4-3H-phenylalanine is used as the substrate for TyrH, the resulting tyrosine contains tritium, and the subsequent hydroxylation of the labeled tyrosine to 3,4-dihydroxyphenylalanine by TyrH results in the loss of half of the tritium (46). These results can be explained by a 1,2 shift upon hydroxylation of 4-3H-phenylalanine to form 3-3H-tyrosine followed by loss of the label at the site of hydroxylation upon formation of 3,4-dihydroxyphenylalanine. Phenylalanine containing Cl, Br, or CH3 at the 4-position also shows a 1,2 shift of the substituent when used as a substrate for TyrH (19). TrpH will catalyze a similar shift of a CH3 group when 4-CH3-phenylalanine or 5-CH3-tryptophan is used as substrate for that enzyme (47, 48).
Formation of an arene oxide intermediate was originally proposed as one explanation for the NIH shift (44). However, isotope effects have provided evidence against such an intermediate in either the TrpH or the TyrH reaction. Attack of the hydroxylating intermediate on the aromatic ring of the substrate to form an arene oxide would be expected to show an isotope effect with either 4-2H- and 5-2H-tryptophan as substrate for TrpH, as there is a reaction at both carbons of the aromatic ring. In contrast, there is a significant isotope effect only with 5-2H-tryptophan as the substrate, establishing that the reaction with oxygen occurs only at the site of hydroxylation (6). In a similar analysis using 4-2H- and 3,5-2H2-phenylalanine as substrates for TyrH, an isotope effect was again only seen when the carbon being hydroxylated was deuterated (22).
The reaction of an electrophilic FeIVO species with a single carbon on the aromatic ring of the amino acid substrate would be expected to form a cationic intermediate. Several lines of evidence are consistent with the formation of such an intermediate in the TyrH and TrpH reactions. When para-substituted phenylalanines are used as substrates for TyrH, tetrahydropterin oxidation and amino acid hydroxylation become uncoupled. This is consistent with the FeIVO intermediate partitioning between hydroxylation of the amino acid and unproductive breakdown and allowed the extent of uncoupling to be used to measure directly the rate of hydroxylation (19). Plots of the rate of hydroxylation as a function of the electron-donating ability of the substituent showed that the para-substituted phenylalanines with more electron-withdrawing substituents were hydroxylated more slowly, yielding a ρ value of approximately −5. Such a large negative ρ value is most consistent with formation of a cationic intermediate in the hydroxylation reaction and inconsistent with mechanisms involving an amino acid radical.
A cationic intermediate and a 1,2-shift implicate electrophilic aromatic substitution as the mechanism of hydroxylation of the amino acid. In such a mechanism (Fig. 4), the FeIVO intermediate reacts directly with the aromatic ring to form the new CO bond. The formation of the CO bond results in the loss of aromaticity of the ring and a change in the hybridization of the carbon with which the new bond is formed. This change in hybridization has been confirmed using deuterium isotope effects for both enzymes. Substitution of deuterium at the site of hydroxylation would be expected to show an approximately 10% increase in the rate of CO bond formation. For the TrpH reaction, an isotope effect of 0.93 was found with 5-2H-tryptophan as substrate (6). A similar effect was found using 3,5-2H2-tyrosine as a substrate for uncoupled mutants of TyrH (49), supporting electrophilic aromatic hydroxylation as a mechanism for this family.
The reaction of the aromatic ring with the FeIVO intermediate to form the new CO bond followed by a 1,2 shift of the hydrogen results in an intermediate with two hydrogens attached to the carbon adjacent to the site of oxygen addition (Fig. 4). Subsequent loss of either hydrogen results in rearomatization of the ring. The possibility that this step is catalyzed by the enzyme has been investigated with TrpH (6). With either 4-2H- or 5-2H-tryptophan as substrate, a 1,2 shift will result in carbon-4 of the indole ring being bonded to both a hydrogen and a deuterium. The 5-HO-tryptophan product showed little difference in the retention of deuterium regardless of its position of origin, consistent with little preference for the loss of one hydrogen over the other. This lack of stereochemistry suggests the absence of an active site base to catalyze the rearomatization, although an active site base accessible to both positions could also explain these results.
The rate constants for individual steps in the reactions of both TrpH and TyrH have been measured using rapid-reaction methods. The results establish that the actual hydroxylation is not rate-limiting for either enzyme. With TyrH there is a burst of 3,4-dihydroxyphenylalanine formation prior to the steady state (40). This establishes that there is a slow step after the hydroxylated amino acid is formed. The turnover of the enzyme slows as the solvent viscosity increases, confirming product release as the slow step in the reaction. A similar burst of formation of the hydroxylated amino acid is also seen with TrpH with either tryptophan or phenylalanine as substrates (32), indicating that rate-determining product release is a common feature of these enzymes.
Both TyrH and TrpH will catalyze the hydroxylation of nonaromatic carbons. When 4-methylphenylalanine is used as a substrate for TyrH, as much as 30% of the product is 4-(hydroxymethyl)phenylalanine in addition to the expected products of hydroxylation of the aromatic ring, while TrpH produces approximately 40% of this product (19, 48, 50). Similarly, with 5-methyltryptophan as a substrate for TrpH, the major product (>99%) is 5-hydroxymethyltryptophan (47). Thus, both enzymes are able to catalyze benzylic hydroxylation. The mechanism of benzylic hydroxylation has been investigated by measuring deuterium isotope effects on the mixture of products with 4-methylphenylalanine as substrate for both enzymes. Use of 4-C2H3-phenylalanine as substrate for either enzyme results in a large decrease in the product from hydroxylation of the methyl group and an increase in the amount of aromatic hydroxylation, such that the total product formation remains the same (48). This is consistent with a single intermediate partitioning between hydroxylation of the methyl group and hydroxylation of the aromatic ring, so that the same FeIVO intermediate is the reactive species in both cases. The intrinsic primary and secondary isotope effects for benzylic hydroxylation by both enzymes are essentially identical; the primary deuterium isotope effect is 10 and the secondary deuterium isotope effect is 1.1 (50, 48). The normal secondary isotope effect indicates a change in the hybridization state of the benzylic carbon from sp3 to sp2, as would occur upon removal of a hydrogen atom to form a carbon radical. The large primary isotope effect suggests the involvement of quantum mechanical tunneling in the hydrogen transfer; this was verified by measurement of the temperature dependence of the intrinsic isotope effect (48). The values of the intrinsic primary and secondary isotope effects are most consistent with benzylic hydroxylation involving hydrogen atom transfer from the methyl group to the FeIVO, followed by rebound of a hydroxyl radical to the benzylic carbon. This is similar to the reactions of cytochrome P450 (51, 52), a family of enzymes also proposed to catalyze hydrogen atom abstraction. The ability of TrpH and TyrH to catalyze both aromatic and benzylic hydroxylation suggests that the FeIVO intermediate in these enzymes is as reactive as the intermediate in the more common cytochrome P450 family.
Significant progress has been made in recent years toward an understanding of the complex catalytic mechanisms of TrpH and TyrH. This progress has required a range of experimental approaches, including protein crystallography, kinetic isotope effects, single-turnover techniques, and a variety of spectroscopies. The present evidence supports a mechanism involving the formation of an iron-peroxypterin, its decay to a reactive FeIVO intermediate, and the subsequent hydroxylation of the amino acid via electrophilic aromatic substitution. There is still much to be done. There is no direct evidence for the proposed iron-peroxypterin intermediate. The protein structural changes that occur during the reaction and their roles in directing the reaction along a productive pathway are still unclear. The structural basis for the substrate specificities of these enzymes is also far from understood.
The experiments from the authors' laboratory described here have been supported by grants from the NIH and The Welch Foundation.