2.4. Hydratase/isomerase superfamily of enzymes
The hydratase/isomerase superfamily, also called the crotonase superfamily, was first defined to consist of 2-enoyl-CoA hydratases and Δ3-Δ2-enoyl-CoA isomerases participating in β -oxidation (Müller-Newen & Stoffel 1993). The first members of the superfamily included the rat peroxisomal MFE-1 (Osumi et al. 1985), the rat monofunctional mitochondrial 2-enoyl-CoA hydratase-1 (Minami-Ishii et al. 1989), the α-subunit of the fatty acid degradation multienzyme complex from Escherichia coli (E. coli) (Dirusso 1990) and the rat mitochondrial monofunctional Δ3-Δ2-enoyl-CoA isomerase (Müller-Newen & Stoffel 1991, Palosaari et al. 1991). These enzymes have low but significant similarity in their amino acid sequence, about 25-27 %, and they share at least one common catalytic amino acid (Müller-Newen & Stoffel 1993, Müller-Newen et al. 1995). Since then, the hydratase/isomerase superfamily has expanded to consist of over 30 members acting in a wide range of metabolic pathways but still possessing an amino acid sequence pattern typical of the superfamily (Müller-Newen & Stoffel 1993, Wu et al. 1997, Holden et al. 2001). The reactions catalyzed by the hydratase/isomerase superfamily members nowadays include dehalogenation, hydration/dehydration, isomerization, decarboxylation, formation/cleavage of carbon-carbon bonds, and hydrolysis of thioesters (Table 1). What is common for the enzymes is that, with one exception, they all use coenzyme A esters as substrates and their catalytic mechanisms involve the stabilization of an oxyanion intermediate (Babbitt & Gerlt 1997, Holden et al. 2001). Because of the similarity of their amino acid sequences, the hydratase/isomerase superfamily members are thought to have evolved from a common ancestor and, therefore, to be both mechanistically and structurally related. This assumption has also been confirmed by the five crystal structures solved so far within the hydratase/isomerase superfamily (Benning et al. 1996, Engel et al. 1996, Modis et al. 1998, Benning et al. 2000, Kurimoto et al. 2001). In this section, the members of the hydratase/isomerase superfamily will be described and their structure-function relationships will be discussed, the emphasis being on the enzymes for which the structure is known.
2.4.1. Members of the superfamily
Table 1 summarizes the hydratase/isomerase superfamily members with known sequence and function. Since there is a vast amount of sequence information available in genome databases due to the genome sequencing projects, this is not a complete list. In addition, the list does not contain hydratase/isomerase proteins with unknown function.
Table 1. The hydratase/isomerase superfamily members with known sequence and function.
Enzyme | Reference |
|---|---|
Monofunctional 2-enoyl-CoA hydratases | |
rat mitochondrial | Minami-Ishii et al. 1989 |
human mitochondrial | Kanazawa et al. 1993 |
Clostridium acetobutylicum | Boynton et al. 1996 |
Rhizobium meliloti | Margolin et al. 1995 |
E. coli CaiD, crotonobetainyl-CoA hydratase | Eichler et al. 1994, Elssner et al. 2001 |
2-enoyl-CoA hydratases as a part of a multifunctional proteina | |
rat peroxisomal MFE-1 | Osumi et al. 1985 |
human peroxisomal MFE-1 | Chen et al. 1991, Hoefler et al. 1994 |
guinea pig peroxisomal MFE-1 | Caira et al. 1996 |
α-subunit of rat MTPb | Kamijo et al. 1993 |
α-subunit of human MTP | Kamijo et al. 1994 |
α-subunit of pig MTP | Yang et al. 1994a |
E. coli fadB gene product | Dirusso 1990, Yang et al. 1991 |
Pseudomonas fragi faoA gene product | Sato et al. 1992 |
human AU-specific RNA-binding protein | Nakagawa et al. 1995 |
mouse AU-specific RNA-binding protein | Brennan et al. 1999 |
Δ3-Δ2-Enoyl-CoA isomerases | |
rat mitochondrial | Müller-Newen & Stoffel 1991, Palosaari et al. 1991 |
human mitochondrial | Janssen et al. 1994, Kilponen et al. 1994 |
mouse mitochondrial | Stoffel et al. 1993 |
rat peroxisomal MFE-1c | Osumi et al. 1985 |
human peroxisomal | Geisbrecht et al. 1999b |
mouse peroxisomal | Geisbrecht et al. 1999b |
E. coli fadB gene product | Dirusso 1990, Yang et al. 1991 |
Streptomyces collinus ChcBd | Patton et al. 2000 |
Δ3,5-Δ2,4-Dienoyl-CoA isomerases | |
rat peroxisomal/mitochondrial | FitzPatrick et al. 1995, Filppula et al. 1998 |
S. cerevisiae peroxisomal | Gurvitz et al. 1999 |
4-Chlorobenzoyl-CoA dehalogenases | |
Pseudomonas sp. CBS-3 | Babbitt et al. 1992 |
Arthrobacter sp. SU | Schmitz et al. 1992 |
Enzyme | Reference |
Methylmalonyl-CoA decarboxylase | |
E. coli | Haller et al. 2000 |
3-Hydroxyisobutyryl-CoA hydrolases | |
Human | Hawes et al. 1996 |
Arabidopsis | Zolman et al. 2001 |
Naphthoate synthases | |
E. coli | Sharma et al. 1992 |
Bacillus subtilis | Driscoll & Taber 1992 |
Others | |
Rhodopseudomonas palusris 2-ketocyclohexanecarboxyl-CoA hydrolase | Pelletier & Harwood 1998 |
Rhodococcus sp 6-oxocamphor hydrolasee | Grogan et al. 2001 |
Pseudomonas fluorescens feryloyl-CoA hydratase/HMPHP-CoA lyasef | Gasson et al. 1998 |
a In MFE-1, the α-subunit of MTP and the fadB gene product the 2-enoyl-CoA hydratase-1 activity is located in the N-terminal part of the polypeptide. b MTP, mitochondrial trifunctional protein c The enoyl-CoA isomerase activity is located in the same domain as the 2-enoyl-CoA hydratase-1 activity. d ChcB, 2-cyclohexenylcarbonyl-CoA isomerase e Not CoA-dependent f HMPHP, 4-hydroxy-3-methoxyphenyl-β -hydroxypropionyl-CoA | |
2.4.1.1. Trans-2-enoyl-CoA hydratase-1
Trans-2-enoyl-CoA hydratase-1 (hydratase-1) can occur either as a monofunctional enzyme or as an integral part of a multifunctional enzyme, and these vary in substrate chain length specificity, as described in the section “The β -oxidation cycle”. Hydratase-1 catalyzes the second step of β -oxidation, the addition of water to the trans-2 double bond generating L-3-hydroxyacyl-CoA (Figs 2, 6) (Willadsen & Eggerer 1975). The mitochondrial short-chain-specific, monofunctional 2-enoyl-CoA hydratase-1, also called crotonase, is the most intensively studied hydratase-1, and it has been purified from rat, bovine and pig (Hass & Hill 1969, Fong & Schulz 1977, Furuta et al. 1980). All these hydratases have similar properties. In solution, they are hexamers formed of six identical subunits and they have the highest catalytic activity towards crotonyl-CoA (trans-2-butyryl-CoA) as substrate. The cDNA sequences of rat and human hydratase-1s have been determined (Minami-Ishii et al. 1989, Kanazawa et al. 1993). In fact, the rat hydratase-1 was the first monofunctional member of the hydratase/isomerase superfamily characterized at the molecular level, and it is also the most thoroughly investigated one. The cDNA of rat hydratase-1 encodes a polypeptide of 290 amino acid residues, including a 29-residue N-terminal mitochondrial targeting sequence. The calculated molecular mass of the mature polypeptide is 28.3 kDa, which is in agreement with the size of the polypeptide isolated from rat liver, 26 kDa (Furuta et al. 1980, Minami-Ishii et al. 1989). Rat hydratase-1 is a very fast and efficient enzyme with a reaction rate close to being diffusion-controlled; its kcat towards crotonyl-CoA is 2100-5700 sec-1 and kcat/Km is 2.8 x 108 sec-1M-1 (Furuta et al. 1980, Fersht 1999). The reaction rate, however, decreases almost linearly with increasing substrate chain length (Furuta et al. 1980).
Hydratase-1 activity as part of a multifunctional protein has been characterized in peroxisomal MFE-1 (Osumi et al. 1979), the α-subunit of MTP (Kamijo et al. 1993) and the α-subunit of the E. coli fatty acid degradation complex, encoded by the fadB gene (Dirusso 1990, Yang et al. 1991). In all of these enzymes, the hydratase-1 activity is located in the N-terminal part of the polypeptide, since this region shows significant sequence similarity to rat hydratase-1 (Ishii et al. 1987, Minami-Ishii et al. 1989, Dirusso 1990, Kamijo et al. 1993). The rate of the hydratase-1 reaction catalyzed by multifunctional enzymes is considerably lower than that of rat 2-enoyl-CoA hydratase-1. For example, the kcat of rat MFE-1 towards crotonyl-CoA is 895 sec-1, i.e. up to six-fold lower when compared to hydratase-1 (Furuta et al. 1980). The C-terminal parts of the multifunctional proteins are also similar in sequence with each other, and they contain 3-hydroxyacyl-CoA dehydrogenase activity. Peroxisomal MFE-1 has, in addition, Δ3–Δ2-enoyl-CoA isomerase activity sharing the active site with hydratase-1 (Palosaari & Hiltunen 1990, Palosaari et al. 1991), and the fadB gene product contains the Δ3-Δ2-enoyl-CoA isomerase and 3-hydroxyacyl-CoA epimerase activities (Yang et al. 1988). The multifunctional proteins with hydratase-1 activity characterized from other sources are shown in Table 1. All these proteins share significant sequence similarity.
A 78-kDa gastrin-binding protein (GBP) purified from porcine gastric mucosal membranes (Baldwin et al. 1986) was suggested to be a component of the gastrin receptor controlling gastrin-dependent acid secretion in parietal cells (Mu et al. 1987). Its subsequent sequence determination surprisingly showed it to have significant sequence similarity with MFE-1 in both the N-terminal and the C-terminal parts (Mantamadiotis et al. 1993). The molecular characterization of the α-subunit of the pig MTP, however, showed its sequence to be identical to the sequence reported to belong to GBP (Yang et al. 1994a). This makes the reported GBP unlikely to serve as a gastrin receptor. The ability of the assumed GBP to bind gastrin was explained by the finding that gastrin and the GBP antagonist benzotrip bind to the trifunctional protein and inhibit its enzyme activities (Hashimoto et al. 1996).
The messenger RNA of some lymphokines and proto-oncogenes are rapidly degraded via a signal provided by an AU-rich element within the 3´-untranslated region of the transcripts. Nakagawa and co-workers (1995) purified and cloned a human 32-kDa protein named AUH that binds specifically to AU-rich transcripts. Surprisingly, the protein had sequence homology to 2-enoyl-CoA hydratase-1, but not to the known RNA-binding proteins, and the recombinant protein possessed a low degree of hydratase-1 activity. The physiological substrate of the protein is unlikely to be crotonyl-CoA because the ability of the enzyme to use it as a substrate was about 1000-fold less than that of bovine hydratase-1. The corresponding mouse AUH has also been characterized, and it was shown to be 94 % identical with its human counterpart (Brennan et al. 1999). Interestingly, AUH was found to be located in mitochondria, suggesting that it could provide a link between the mitochondrial metabolic pathways and RNA stability (Brennan et al. 1999).
The CaiD gene of E. coli was previously suggested to encode for a carnitine racemase in the carnitine pathway (Eichler et al. 1994). Recent studies have shown, however, that, in addition to having a 30 % identical amino acid sequence compared to 2-enoyl-CoA hydratase-1, CaiD also catalyzes the hydratase-1 reaction. More specifically, it hydrates crotonobetainyl-CoA to L-carnitinyl-CoA, and it was thus renamed crotonobetainyl-CoA hydratase or carnitinyl-CoA dehydratase (Elssner et al. 2001).
2.4.1.2. Δ3-Δ2-Enoyl-CoA isomerase
The Δ3-Δ2-Enoyl-CoA isomerase is required in the β -oxidation of unsaturated fatty acids, and it catalyzes the reaction converting cis-3-enoyl-CoA or trans-3-enoyl-CoA into trans-2-enoyl-CoA, the substrate of 2-enoyl-CoA hydratase-1 in the β -oxidation cycle (Figs 3, 6, see “β -oxidation of (poly)unsaturated fatty acids”) (Hiltunen & Qin 2000). Similarly to 2–enoyl-CoA hydratase-1, the Δ3-Δ2-enoyl-CoA isomerase can occur either as a monofunctional enzyme or as part of a multifunctional enzyme, as described above. Monofunctional isomerases have been purified from several sources, including the mitochondria of rat (Stoffel & Grol 1978, Palosaari et al. 1990, Müller-Newen & Stoffel 1991), bovine (Euler-Bertram & Stoffel 1990) and human (Kilponen & Hiltunen 1993) as well as the peroxisomes of cucumber seedlings (Engeland & Kindl 1991), human and mouse (Geisbrecht et al. 1999b).
The rat mitochondrial enoyl-CoA isomerase is a basic protein with a subunit size of 29 kDa and pI 9.7 (Palosaari et al. 1990, Müller-Newen & Stoffel 1991, Palosaari et al. 1991). The subunit sizes and the amino acid sequences of the rat and human mitochondrial enoyl-CoA isomerases are highly similar with 74 % sequence identity (Kilponen et al. 1994). Interestingly, however, the pI of the human enzyme is three pH units more acidic compared to its rat counterpart. This difference is explained by the fact that many basic amino acid residues in the rat enoyl-CoA isomerase have been changed to neutral of basic ones in the human enzyme (Kilponen et al. 1994). Another difference is that the human enoyl-CoA isomerase was reported to be a homodimer with a native molecular mass of 70 kDa (Kilponen & Hiltunen 1993), whereas the rat enzyme was found to be a trimer (Palosaari et al. 1990, Zeelen et al. 1992). Recent studies in our laboratory have suggested, however, that the human enoyl-CoA isomerase is also most probably a trimer. The rat and human enoyl-CoA isomerases also differ in their substrate specificity, the rat enzyme being most active towards short-chain 3-enoyl-CoAs (Palosaari et al. 1991) and the human isomerase having, in contrast, no clear-cut substrate chain length specificity (Kilponen & Hiltunen 1993). A finding showing that there is also a long-chain-specific enoyl-CoA isomerase in rat mitochondria (Kilponen et al. 1990) could explain why the rat enoyl-CoA isomerase described here metabolizes only short-chain substrates. In human, no such enzyme has been characterized, suggesting that the mitochondrial enoyl-CoA isomerase should be able to use 3-enoyl-CoAs of all chain lengths as substrates (Kilponen & Hiltunen 1993).
Recently, a monofunctional Δ3-Δ2-enoyl-CoA isomerase has also been characterized from the peroxisomes of human and mouse (Geisbrecht et al. 1999b). This enzyme, called PECI, has a somewhat higher molecular mass, 39.4 kDa, compared to the other monofunctional members of the hydratase/isomerase superfamily. It was found that, in addition to the hydratase/isomerase-like sequence, PECI has an extra domain about 80 amino acids in length at its N-terminus. This domain has significant sequence identity compared to the acyl-CoA binding protein (ACBP). ACBP is believed to be involved in the intracellular acyl-CoA transport and pool formation as well as in the control of fatty acid metabolism. ACBP exists in many isoforms and can also occur as a subdomain in acyl-CoA metabolizing enzymes, as in the case of PECI (Knudsen et al. 2000).
A novel Δ3-Δ2-enoyl-CoA isomerase has been recently characterized from Streptomyces collinus. This enzyme, named 2-cyclohexenylcarbonyl-CoA isomerase (ChcB), is involved in the production of a polyketide antifungal antibiotic called ansatrienin A by catalyzing the isomerization of 2-cyclohexenylcarbonyl-CoA to 1–cyclohexenylcarbonyl-CoA (Patton et al. 2000). ChcB has wide substrate specificity, being able also to metabolize straight-chain 3-enoyl-CoAs and even with higher specific activity than 2-cyclohexenylcarbonyl-CoA. This could imply that 2–cyclohexenyl-carbonyl-CoA is not the physiological substrate for ChcB. However, the disrupted chcB mutant was able to grow on unsaturated fatty acids, whereas ansatrienin biosynthesis was blocked (Patton et al. 2000).
2.4.1.3. Δ3,5-Δ2,4-Dienoyl-CoA isomerase
Δ3,5-Δ2,4-Dienoyl-CoA isomerase takes part in the metabolism of double bonds at odd-numbered positions in unsaturated fatty acids together with 2,4-dienoyl-CoA reductase and Δ3-Δ2-enoyl-CoA isomerase (see “β -oxidation of (poly)unsaturated fatty acids”). It catalyzes the conversion of 3,5-dienoyl-CoA to 2,4-dienoyl-CoA, the substrate of dienoyl-CoA reductase (Figs 3, 8). Only one mammalian Δ3,5-Δ2,4-dienoyl-CoA isomerase, namely the rat one, has been characterized at the molecular level. The cDNA was isolated by FitzPatrick and colleagues (1995) by screening peroxisome proliferator-induced genes in rat liver. The cDNA was proposed to encode for a 2-enoyl-CoA hydratase-1 on the basis of its sequence similarity with rat hydratase-1, and it was subsequently named rECH1 (FitzPatrick et al. 1995). Filppula and co-workers (1998) expressed rECH1 as a recombinant protein, starting from residue 54, and purified and characterized the gene product. It was discovered that rECH1 does not encode for a hydratase-1 but, instead, for a Δ3,5-Δ2,4-dienoyl-CoA isomerase having a specific activity of 0.12 µmol/min/mg towards 3,5,8,11,14-eicosapentaenoyl-CoA as substrate. Low but detectable levels of hydratase-1 activity (5nmol/min/mg) were also recorded, but no Δ3–Δ2-enoyl-CoA isomerase activity was found (Filppula et al. 1998). An antibody raised against the rat Δ3,5-Δ2,4-dienoyl-CoA isomerase recognized two polypeptides with molecular sizes of 32 kDa and 36 kDa, respectively, the smaller one being in the mitochondrial and the larger in the peroxisomal fraction. Dual distribution of the same gene product was further verified by immunoelectron microscopy. The amino acid sequence of Δ3,5-Δ2,4-dienoyl-CoA isomerase contains both the N-terminal mitochondrial targeting sequence and the C-terminal tripeptide targeting it to peroxisomes (Filppula et al. 1998). Thus, it seems likely that, upon mitochondrial import, the targeting sequence is cleaved off, giving rise to the 32-kDa species, and that the full-length 36-kDa polypeptide is taken to the peroxisomes without modifications. A recent study has argued, however, that the peroxisomal form of dienoyl-CoA isomerase is also 32 kDa in size and has a truncated N-terminus (Zhang et al. 2001). The occurrence of the same gene product in two different organelles also explains the finding by He and co-workers (1995) suggesting that both the peroxisomal and the mitochondrial Δ3,5-Δ2,4-dienoyl-CoA isomerase share similar chain length specificities and that the antibody against the mitochondrial isoenzyme also recognizes the peroxisomal one.
The presence of a novel enzyme, Δ3,5,7-Δ2,4,6-trienoyl-CoA isomerase, acting in the β -oxidation of unsaturated fatty acids with conjugated double bonds was detected in rat and pig mitochondria (Liang et al. 1999). Zhang and co-workers (2001) found, however, that the trienoyl-CoA isomerase is actually the same enzyme as the Δ3,5-Δ2,4-dienoyl-CoA isomerase. To study its enzymatic properties in more detail, the mature mitochondrial form of dienoyl-CoA isomerase, starting from residue 35, was expressed as a recombinant protein and its properties were characterized (Zhang et al. 2001). A specific Δ3,5-Δ2,4-dienoyl-CoA isomerase activity of 2450 µmol /min/mg was measured, which was considerably higher than that detected previously (Filppula et al. 1998). In addition, a Δ3-Δ2-enoyl-CoA isomerase activity of 0.034 µmol/min/mg was detected, which had not been recorded earlier (Filppula et al. 1998). Furthermore, the enzyme catalyzed the isomerization of Δ3,5,7- trienoyl-CoA to Δ2,4,6-trienoyl-CoA at a rate of 48 µmol/min/mg (Zhang et al. 2001).
In S. cerevisiae, a gene encoding the Δ3,5-Δ2,4-dienoyl-CoA isomerase has been identified (Gurvitz et al. 1999). The gene product, Yor180cp/Dci1p, catalyzes the dienoyl-CoA isomerase reaction but is dispensable for the β -oxidation of unsaturated fatty acids with odd-numbered double bonds. Apparently, the degradation of odd-numbered double bonds occurs via the isomerase-dependent route in S. cerevisiae (Gurvitz et al. 1999).
2.4.1.4. 4-Chlorobenzoyl-CoA dehalogenase
Halogenated hydrocarbons, such as chloroaromatics, which are used widely as industrial and agricultural agents, constitute a major class of environmental pollutants, which are inefficiently detoxified by conventional methods. A new strategy for the disposal of these pollutants is the use of microbial biodegradation. One of the bacterial strains capable of metabolizing chlorinated hydrocarbons is Pseudomonas sp. strain CBS-3, which can use 4-chlorobenzoate (4-CBA) as its sole carbon source. 4-CBA is converted into 4-hydroxybenzoate (4-HBA) by an enzyme system that catalyzes the aromatic substitution of the chloride with a hydroxyl group from a water molecule (Müller et al. 1984). The enzyme complex, called the 4-CBA dehalogenase complex, consists of three polypeptides, namely the 57-kDa 4-CBA:CoA ligase activating 4-CBA to 4-CBA-CoA, the 30-kDA 4-CBA-CoA dehalogenase catalyzing the aromatic substitution reaction (Fig. 7) and the 16-kDa thioesterase cleaving 4-HBA-CoA into 4-HBA and CoA (Scholten et al. 1991). The cloning and sequencing of the components of the enzyme complex revealed that the sequence of the 30-kDa 4-CBA-CoA dehalogenase is similar to 2-enoyl-CoA hydratase-1 (Minami-Ishii et al. 1989) and the other members of the hydratase/isomerase superfamily characterized so far (Babbitt et al. 1992). The 4-CBA-CoA dehalogenase and 2-enoyl-CoA hydratase-1 were suggested to be both evolutionary and also mechanistically related, since they both require the activation of water for addition across a carbon-carbon bond in catalysis (Babbitt et al. 1992). Similarly to the rat Δ3-Δ2-enoyl-CoA isomerase, the 4-CBA-CoA dehalogenase forms homotrimers (Benning et al. 1996).
2.4.1.5. Methylmalonyl-CoA decarboxylase
The genome-sequencing project of E. coli revealed that its genome contains seven paralogues of the hydratase/isomerase superfamily, four with unknown function. One of the unknown genes was ygfG, a member of a four-gene operon also containing the genes sbm, ygfD and ygfH (Haller et al. 2000). This operon was found likely to encode enzymes forming a metabolic cycle which decarboxylates succinate to propionate. The metabolic context of this pathway was, however, left unresolved (Haller et al. 2000). The reaction catalyzed by YgfG was determined to be the decarboxylation of methylmalonyl-CoA to propionyl-CoA, a new reaction in the hydratase/isomerase superfamily (see also Fig. 9). YgfG was thus renamed methylmalonyl-CoA decarboxylase. Similarly to 2-enoyl-CoA hydratase-1 and dienoyl-CoA isomerase, methylmalonyl-CoA decarboxylase is also a hexamer consisting of six identical 29 kDa subunits (Benning et al. 2000, Haller et al. 2000).
2.4.1.6. Other members
Dihydroxynaphthoate synthase catalyzes a ring closure reaction forming 1,4-dihydroxy-2-naphthoic acid from o-succinylbenzoyl-CoA (Meganathan & Bentley 1979). This is a reaction involved in the biosynthesis of a bacterial electron carrier menaquinone (vitamin K2). The gene encoding dihydroxynaphthoate synthase, menB, has been sequenced from E. coli (Sharma et al. 1992) and Bacillus subtilis (Driscol & Taber 1992).
In contrast to the ring closure reaction of dihydroxynaphthoate synthase, another hydratase/isomerase family member, 2-ketocyclohexanecarboxyl-CoA (2-ketochc-CoA) hydrolase catalyzes a hydrolytic ring cleavage of 2-ketochc-CoA to pimelyl-CoA, which is required in the anaerobic benzoate degradation in Rhodopseudomonas palustris (Egland et al. 1997, Pelletier & Harwood 1998). The 2-ketochc-CoA hydrolase is encoded by the badI gene located in a cluster of anaerobic benzoate degradation genes (Egland et al. 1997) and highly induced when grown on benzoate. Interestingly, 2-ketochc-CoA hydrolase shares the highest amino acid sequence identity, about 45 %, with dihydroxynaphthoate synthase, which reflects the similar nature of the reactions they catalyze, involving either ring closure or cleavage steps (Pelletier & Harwood 1998).
3-Hydroxyisobutyryl-CoA (HIB-CoA) hydrolase is a vital enzyme catalyzing the hydrolytic cleavage of HIB-CoA to free CoA and 3-hydroxyisobutyrate in the valine catabolic pathway. This is an interesting reaction, since the destruction of an activated intermediate is rare in metabolic pathways, especially if the later steps of the pathway also require CoA esters as substrates, as in this case. HIB-CoA hydrolase is, however, essential for the removal of a toxic intermediate of the pathway, methacrylyl-CoA, which is detoxified by two subsequent reactions, hydration to HIB-CoA by hydratase-1 and cleavage by HIB-CoA hydrolase (Shimomura et al. 1994). The toxicity of methacrylyl-CoA is probably due to its ability to react with free thiol groups of proteins, thereby inactivating them (Brown et al. 1982). HIB-CoA hydrolase has been purified from rat liver (Shimomura et al. 1994), and the cDNA sequence of the human enzyme has been determined (Hawes et al. 1996). The enzyme is specifically active with a very high turnover rate towards HIB-CoA and lesser turnover towards 3-hydroxypropionyl-CoA (Shimomura et al. 1994, Hawes et al. 1996). These powerful catalytic properties and the strict substrate specificity are required for efficient removal of methacrylyl-CoA during valine catabolism and for prevention of the interference of HIB-CoA hydrolase with the catabolism of fatty acids, leucine and isoleucine. Surprisingly, the cDNA of HIB-CoA hydrolase had no similarity to other thioesterases but, instead, showed striking homology to the hydratase/isomerase superfamily members (Hawes et al. 1994). Thus, the hydratase/isomerase proteins catalyze two subsequent reactions in the valine catabolic pathway. Recently, a plant HIB-CoA hydrolase encoded by an Arabidopsis gene CHY1 has also been characterized (Zolman et al. 2001). The chy1 mutant is resistant to a plant hormone indole-3-butyric acid and also has impaired β -oxidation (Zolman et al. 2001). CHY1 is located in the peroxisomes, where the β -oxidation of plants also occurs, and shares 43 % sequence identity with the mammalian mitochondrial HIB-CoA hydrolase. The metabolic function of CHY1 was confirmed by expressing it as a recombinant protein and by showing that the expressed protein had specific HIB-CoA hydrolase activity (Zolman et al. 2001). The impaired β -oxidation and hormone sensitivity in the mutant were suggested to be due to the inactivation of the enzymes of the respective pathways by accumulating methacrylyl-CoA (Zolman et al. 2001).
Yet another different reaction catalyzed by an enzyme belonging to the hydratase/isomerase superfamily was characterized from Pseudomonas fluorescens. This novel enzyme was found to be involved in the metabolism of ferulic acid (4-hydroxy-3-methoxy-trans-cinnamic acid) to vanillin by catalyzing both the hydration of feruloyl-CoA to 4-hydroxy-3-methoxyphenyl-β -hydroxypropionyl-CoA (HMPHP-CoA) and its subsequent cleavage to vanillin and acetyl-CoA (Gasson et al. 1998). The enzyme, feruloyl-CoA hydratase/HMPHP-CoA lyase could have potential use in the biotechnological production of vanillin, one of the principal flavouring compounds in the world. Since feruloyl-CoA is a breakdown product of lignin and plant cell wall material, plant waste could be used as starting material for vanillin production (Gasson et al. 1998, Priefert et al. 2001).
The enzyme 6-oxocamphor hydrolase from Rhodococcus makes an interesting exception to the rule that hydratase/isomerase proteins only accept CoA esters as substrates. It catalyzes the asymmetric hydrolysis of 6-oxocamphor to (2R,4S)-α-campholinic acid as well as the hydrolysis of some other bicyclic β -diketones (Grogan et al. 2001). The characterization of 6-oxocamphor hydrolase adds a new enzyme to the group of biocatalysts that could be employed in the synthesis of chemical intermediates (Grogan et al. 2001).
2.4.2. Structure and function of the hydratase/isomerase proteins
2.4.2.1. Structural aspects
As mentioned above, the crystal structures of five members of the hydratase/isomerase superfamily are known, those being the rat 2-enoyl-CoA hydratase-1 (Engel et al. 1996, 1998), the 4-CBA-CoA dehalogenase from Pseudomonas (Bennning et al. 1996), the rat Δ3,5-Δ2,4-dienoyl-CoA isomerase (Modis et al. 1998), the methylmalonyl-CoA decarboxylase from E. coli (Benning et al. 2000) and the very recently published human AUH protein, a RNA-binding homologue of 2-enoyl-CoA hydratase-1 (Kurimoto et al. 2001). All these structures have a similar overall fold consisting of an N-terminal core domain, containing the first 200-250 amino acid residues, and a C-terminal trimerization domain (Fig. 5). The N-terminal domain has a spiral fold formed of four turns, each turn being shaped by two β -strands and an α-helix. The β -strands form two β -sheets, which are almost at right angles with respect to each other and connected by the α-helices. The C-terminal domain contains four α-helices that are mainly used for subunit-subunit contacts and also for shaping the acyl-CoA-binding pocket. In the structures of hydratase-1, dienoyl-CoA isomerase, 4-CBA-CoA dehalogenase and the AUH protein, the C-terminal helices H9 and H10 protrude out of the core domain and cover the active site of the adjacent subunit. In the methylmalonyl-CoA decarboxylase structure, however, the active site is fully contained within one subunit since the C-terminal domain folds over the active site of the same subunit (Benning et al. 2000). All the known hydratase/isomerase structures assemble into disk-like trimers via contacts by the C-terminal domain; in addition, in all except 4-CBA-CoA dehalogenase, two trimers bind together to form a hexamer. Thus, the hexamers are described to be dimers of trimers.
The human AUH protein differs from the other hydratase/isomerase proteins in that its physiological function is not likely to be enzymological but to bind the AU-rich elements of single-stranded RNA (Nakagawa et al. 1995, Kurimoto et al. 2001). The AUH protein is able to catalyze the hydratase-1 reaction, but the reaction rate is so low that the activity is rather residual than of any physiological significance (Nakagawa et al. 1995, Kurimoto et al. 2001). The binding of RNA is accomplished by the clefts between the trimers in the hexamer, the edges of the clefts being formed by the α-helix H1 of each subunit. The 20-residue segment, the R-peptide, which has been found to bind to the AU-rich elements of RNA (Nakagawa & Moroni 1997), is mostly contained within the helix H1. This segment is rich in positively charged lysine residues, which form the “lysine comb” that could bind the negatively charged phosphate groups of RNA. The mutation of the lysines to negatively charged residues abolishes the RNA binding. In addition, the cleft between the trimers is wide enough to accommodate a RNA molecule (Kurimoto et al. 2001). In contrast, in 2-enoyl-CoA hydratase-1, the helix H1 is negatively charged and no RNA-binding activity for it has been detected (Kurimoto et al. 2001).
Of 2-enoyl-CoA hydratase-1, 4-CBA-CoA dehalogenase and methylmalonyl-CoA decarboxylase also structures with an active site ligand exist (Engel et al. 1996, 1998; Benning et al. 1996, 2000). The acyl-CoA molecule is bound in such a way that the 3´-phosphate ADP and pantothenic acid moieties are bound in a curved conformation on the outside of the core domain against a β -sheet, whereas the acyl part slides into the substrate-binding pocket inside the protein (Fig. 5). The shape of the binding pocket is mainly determined by the conformation of the α-helix H2 of the core domain and the helices H9 and H10 of the trimerization domain. The variability in these regions enables the binding of different substrates at the active sites. More detailed discussion on the structures of the hydratase/isomerase superfamily will follow in the ”Discussion” section.
2.4.2.2. Catalytic mechanisms
The first sequenced members of the hydratase/isomerase superfamily, mono- and multifunctional 2-enoyl-CoA hydratases and Δ3-Δ2-enoyl-CoA isomerases, were suggested to be, in addition to their evolutionary relationship, also mechanistically related, since both the hydratase-1 and the isomerase reactions involve protonation or deprotonation at the α-carbon (C2) of the substrate (Fig. 6) (Müller-Newen & Stoffel 1993). The hydratase-1 reaction consists of an addition of a hydroxyl group from an activated water molecule to the β -carbon (C3) and the protonation of the α-carbon (Fig 6). The reaction follows the syn stereochemistry, in which both the proton and the hydroxyl group are added to the same side of the planar α,β -double bond (Willadsen & Eggerer 1975). In the reaction catalyzed by the Δ3-Δ2-enoyl-CoA isomerase, a proton is abstracted from the α-methylene group of the substrate and subsequently donated to the γ -carbon (C4), resulting in a shift of the double bond from the β ,γ -position to the α,β -conformation (Fig. 6) (Müller-Newen & Stoffel 1993). The initial sequence comparisons and mutagenesis studies showed that Glu164 of hydratase-1 was the only protic amino acid residue completely conserved among the characterized hydratase/isomerase superfamily members (Müller-Newen & Stoffel 1993, Müller-Newen et al. 1995). Glu164 (Glu165 in the rat enoyl-CoA isomerase) was confirmed to be essential for the catalytic activity of both the rat 2-enoyl-CoA hydratase-1 and the enoyl-CoA isomerase by site-directed mutagenesis, suggesting that it is probably involved in the protonation/deprotonation step (Müller-Newen & Stoffel 1993, Müller-Newen et al. 1995). The determination of the crystal structure of rat hydratase-1 showed another protic residue, Glu144, to be also involved in catalysis (Engel et al. 1996). This residue is also completely conserved in the proteins catalyzing the hydratase-1 reaction, including the multifunctional proteins, but not in monofunctional Δ3-Δ2-enoyl-CoA isomerases. It was proposed that Glu144 acts as a catalytic base and activates the water molecule for the attack at C3, whereas Glu164 is the catalytic acid protonating C2 (Engel et al. 1996). The water molecule needed for activation is seen to be bound between Glu144 and Glu164 in the crystal structure of 2-enoyl-CoA hydratase-1 (Engel et al. 1996). Further mutational studies (Kiema et al. 1999) confirmed the necessity of Glu144 for the hydratase-1 reaction, the Glu144Ala variant having 2000-fold lower hydratase-1 activity compared to the wild type. Kiema and co-workers (1999) further found that rat hydratase–1 also possesses slight residual Δ3-Δ2-enoyl-CoA isomerase activity. Glu144Ala and Glu164Ala mutants were tested for isomerase activity, and the mutation of Glu164 was found to totally abolish isomerase activity, while the activity in the Glu144Ala variant was only lowered 10-fold. These findings are in line with the suggestion that Glu164 is needed for both the hydratase-1 and the enoyl-CoA isomerase activities, whereas Glu144 is only essential for the water activation reaction of hydratase-1 (Kiema et al. 1999). The amino acid residue responsible for the protonation of the γ -carbon in the isomerization reaction is not known, the proton could be donated by a water molecule or the deprotonation and the protonation could be accomplished by the same residue. A crystal structure is needed to determine the active site architecture of the Δ3-Δ2-enoyl-CoA isomerase. Interestingly, in the mammalian monofunctional peroxisomal Δ3-Δ2-enoyl-CoA isomerase, the glutamate at position 164 (hydratase-1 numbering) is not conserved (Geisbrecht et al. 1999b), nor is it in the novel bacterial isomerase, 2-cyclohexenylcarbonyl-CoA isomerase (Patton et al. 2000).
Spectroscopic studies have shown that, in 2-enoyl-CoA hydratase-1, the substrate is activated for catalysis by forming a strong hydrogen bond from a protein atom to the oxygen of the thioester carbonyl group of the substrate (D´Ordine et al. 1994). In fact, crystallographic and mutation studies have indicated that this hydrogen bond in hydratase-1 is made by the main chain NH group of Gly141, and it is essential for the catalysis to occur (Engel et al. 1996, Bell et al. 2001). Hydrogen bonding from Gly141 together with the NH group of Ala98 form the oxyanion hole, which also stabilizes the negative charge forming on the thioester oxygen during the transition state of the enzymatic reaction (Engel et al. 1996, 1998). The ability to stabilize the intermediate seems to be a common requirement in the hydratase/isomerase superfamily, since the locations of the oxyanion hole residues are conserved in all of the structurally characterized members (Babbitt & Gerlt 1997, Holden et al. 2001).
The dehalogenation of 4-chlorobenzoyl-CoA (4-CBA-CoA) to 4-hydroxybenzoyl-CoA (4-HBA-CoA) by 4-CBA-CoA dehalogenase occurs via three reaction intermediates (Fig. 7) (Yang et al. 1994b). First, an active site carboxylate residue attacks the C4 of the benzoyl ring of the substrate, forming a covalent Meisenheimer complex. Second, the chloride ion leaves, rearomatizing the benzoyl ring and generating the aryl enzyme intermediate. Third, an activated water molecule adds to the acyl carbonyl carbon, leading to a tetrahedral intermediate, from which the product 4-HBA-CoA is formed and the catalytic carboxylate generated (Yang et al. 1994b, Yang et al. 1996). Mutagenesis and structural studies revealed that the active site nucleophilic carboxylate forming the covalent complex is Asp145 (Benning et al. 1996, Yang et al. 1996), a residue not conserved in 2-enoyl-CoA hydratases. The catalytic base activating the water molecule was found to be His90, which is not conserved in the previously characterized hydratase/isomerase family members, either (Benning et al. 1996, Yang et al. 1996). In addition, Trp137 is important in providing a hydrogen bond to Asp145 (Benning et al. 1996, Yang et al. 1996) and thereby correctly positioning the catalytic residue with respect to the substrate (Lau & Bruice 2001). In the Meisenheimer intermediate, a negative charge develops on the thioester carbonyl oxygen. As in 2-enoyl-CoA hydratase-1, this charge is stabilized by the hydrogen bonds in the oxyanion hole, which is formed by the NH groups of Phe64 and Gly114 (Benning et al. 1996). Similarly, these hydrogen bonds are also involved in the activation of the substrate for catalysis by polarizing the 4-CBA-CoA. In addition, the α-helix terminating in Gly114 provides a dipolar electrostatic component, which contributes to the polarization of the thioester carbonyl oxygen and the whole benzoyl moiety, thus facilitating the attack of Asp145 to C4 of the ring (Clarkson et al. 1997, Taylor et al. 1997).
Δ3,5-Δ2,4-Dienoyl-CoA isomerase catalyzes the simultaneous shift of two double bonds from the Δ3,5 position to the Δ2,4 conformation during the β -oxidation of unsaturated fatty acids with odd-numbered double bonds (Luo et al. 1994). According to the crystal structure of the rat Δ3,5-Δ2,4-dienoyl-CoA isomerase, there are three protic amino acid residues at the active site of the enzyme, Asp176, Glu196 and Asp204 (Modis et al. 1998), of which Asp176 and Glu196 are equivalents to the catalytic residues of rat hydratase-1, Glu144 and Glu164 (Engel et al. 1996), and Asp204 corresponds to the active site aspartate-145 of 4-CBA-CoA dehalogenase (Benning et al. 1996, Yang et al. 1996). Glu196 and Asp204 are positioned correctly at the active site of the dienoyl-CoA isomerase for proton exchange from the carbons C2 and C6, whereas Asp176 is probably not directly involved in catalysis but, instead, optimizes the catalytic properties of Glu196 by hydrogen bonding to it. The importance of Asp204 for catalysis was also shown by mutagenesis (Modis et al. 1998). The following reaction mechanism was therefore proposed (Fig. 8): Glu196 would abstract the proton from C2 followed by rearrangements of the double bonds. To prevent the accumulation of charge in the substrate, Asp204 would donate a proton to the C6 of the substrate and the product would be released. As in the other superfamily members, the negative charge on the thioester oxygen forming during the transition state of the reaction is stabilized in the oxyanion hole residues Ile117 and Gly173 (Modis et al. 1998). In a recent mechanistic study on the Δ3,5-Δ2,4-dienoyl-CoA isomerase, Asp176, Glu196 and Asp204 were mutated separately to nonprotic residues, and the characteristics of the mutants were investigated (Zhang et al. 2001). The mutation of either Glu196 or Asp204 lowered the dienoyl-CoA isomerase activity 105-fold compared to the wild type, whereas the Asp176Ala variant had only 10-fold lower activity. Moreover, Glu196 could catalyze the Δ2,Δ3-isomerization of 2,5-dienoyl-CoA to 3,5-dienoyl-CoA when Asp204 was inactivated, and Asp204 was able to catalyze the Δ5,Δ4-isomerization of 2,5-dienoyl-CoA to 2,4-dienoyl-CoA when Glu196 was mutated. However, these monoene isomerizations occurred at a much lower rate than the simultaneous diene isomerization in the wild type (Zhang et al. 2001). The results confirmed the suggestion by Modis and co-workers (1998) that Glu196 and Asp204 are involved in proton exchange from carbons C2 and C6, respectively. The ability of the Δ3,5-Δ2,4-dienoyl-CoA isomerase to catalyze the isomerization of 2,5-dienoyl-CoA to 2,4-dienoyl-CoA without the 3,5-intermediate could imply that a separate monofunctional Δ3–Δ2-enoyl-CoA isomerase is not necessarily needed in the pathway metabolizing odd-numbered double bonds. Zhang and colleagues (2001) also found that the Δ3,5-Δ2,4-dienoyl-CoA isomerase is a Δ3,5,7-Δ2,4,6-trienoyl-CoA isomerase and that Asp204 is the residue also catalyzing the shift of the double bond from Δ7 to Δ6 during the metabolism of fatty acids with conjugated double bonds.
In the amino acid sequence of E. coli, methylmalonyl-CoA decarboxylase Glu144 of rat hydratase-1 is conserved as Glu113 (Haller et al. 2000). This is the first case where only Glu144 but not Glu164 is conserved. The crystal structure of methylmalonyl-CoA decarboxylase showed, surprisingly, that Glu113 is probably not directly involved in catalyzing the decarboxylation of methylmalonyl-CoA to propionyl-CoA, since its location is not appropriate with respect to the substrate and, in addition, it is hydrogen-bonded to Arg86 (Benning et al. 2000). Instead, Tyr140 is positioned in such a way that it could orient the carboxylate group of methylmalonyl-CoA properly for decarboxylation; therefore, Tyr140 is suggested to be the catalytic residue (Fig. 9). The nonpolar environment would destabilize the negatively charged carboxylate group and enhance the decarboxylation (Benning et al. 2000). The reaction also requires activation of the substrate by polarization of the thioester carbonyl oxygen and stabilization of the anionic transition state, which is accomplished by hydrogen bonding by the oxyanion hole residues, His66 and Gly110 (Benning et al. 2000). It is currently not known which residue serves as the general acid delivering a solvent-derived proton to the α-carbon to generate the product, propionyl-CoA.
The Glu144 of rat hydratase-1 is also conserved in the 6-oxocamphor hydrolase (Grogan et al. 2001) and the Glu164 in the HIB-CoA hydrolase (Hawes et al. 1996) and the feruloyl-CoA hydratase/lyase (Gasson et al. 1998). In the absence of structural and mutational data, the possible role of these residues in the reaction mechanisms can only be speculated. Similarly, nothing is known about the catalytic residues of the 2-ketochc-CoA hydrolase (Pelletier & Harwood 1998) and the dihydroxynaphthoate synthase (Sharma et al. 1992), since none of the catalytic residues identified in the other members are conserved in them.
As a conclusion, within the hydratase/isomerase superfamily, polypeptides with considerably low sequence identities (20-30 %) still fold into strikingly similar three-dimensional structures with the same active site design. Three conserved sites for catalytic amino acids have been described so far, the Glu144 and Glu164 of 2-enoyl-CoA hydratase-1 and the Asp145 of 4-CBA-CoA dehalogenase. Some members of the family use none of these residues for catalysis, indicating that more catalytic residues are yet to be identified. The common active site template with multiple catalytic “stations” supports the variety of catalytic reactions to occur within the hydratase/isomerase superfamily, and the evolution of enzymes catalyzing new reactions has probably taken place by mutating residues within the active site template (Xiang et al. 1999). To provide experimental evidence in support of this hypothesis, Xiang and co-workers (1999) engineered the active site residues of 4-CBA-CoA dehalogenase in such a way that the residues corresponding to the Glu144 and Glu164 of hydratase-1 were incorporated. Indeed, the engineered variant of 4-CBA-CoA dehalogenase was able to catalyze the 2-enoyl-CoA hydratase-1 reaction, the syn addition of water to a α,β -double bond. Above all, the evolution of this superfamily has probably been dominated by the need to stabilize the anionic transition state of the enzymatic reaction (Holden et al. 2001). This is highlighted by the fact that the two peptidic NH groups forming the oxyanion hole are conserved in all the structurally characterized members of the superfamily. Furthermore, the latter of the oxyanion hole residues is always at the end of an α-helix. This helix dipole adds to the polarizing effect of the oxyanion hole. Thus, all the proteins in this family have a common structural solution, the oxyanion hole, to the mechanistic problem of how to lower the free energy of the transition state in order for the reaction to proceed (Babbitt & Gerlt 1997, Holden et al. 2001).