Glutathione transferases in the genomics era: New insights and perspectives

Biomolecular Engineering 23 (2006) 149–169 www.elsevier.com/locate/geneanabioeng

Review

Glutathione transferases in the genomics era: New insights and perspectives Carla Frova * Department of Biomolecular Sciences and Biotechnology, University of Milano, Via Celoria 26, 20133 Milano, Italy Received 9 February 2006; received in revised form 12 May 2006; accepted 12 May 2006

Abstract In the last decade the tumultuous development of ‘‘omics’’ greatly improved our ability to understand protein structure, function and evolution, and to define their roles and networks in complex biological processes. This fast accumulating knowledge holds great potential for biotechnological applications, from the development of biomolecules with novel properties of industrial and medical importance, to the creation of transgenic organisms with new, favorable characteristics. This review focuses on glutathione transferases (GSTs), an ancient protein superfamily with multiple roles in all eukaryotic organisms, and attempts to give an overview of the new insights and perspectives provided by omics into the biology of these proteins. Among the aspects considered are the redefinition of GST subfamilies, their evolution in connection with structurally related families, present and future biotechnological outcomes. # 2006 Elsevier B.V. All rights reserved. Keywords: Omics; Protein families; Thioredoxin fold; Evolution; Forced evolution; Transgenics

Contents 1. 2.

3. 4.

5.

6.

7.

8.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cytosolic GSTs (cGSTs). . . . . . . . . . . . . . . . . . . . 2.1. Nomenclature. . . . . . . . . . . . . . . . . . . . . . . 2.2. Genome organization . . . . . . . . . . . . . . . . . 2.3. General structure and functional implications. Kappa GSTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microsomal GSTs . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Structure and membrane topology . . . . . . . . 4.2. Extravagant microsomal GSTs . . . . . . . . . . . Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Proteomics studies . . . . . . . . . . . . . . . . . . . 5.2. In vivo functions by knock out studies . . . . . Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Soluble GSTs. . . . . . . . . . . . . . . . . . . . . . . 6.1.1. Phase 1 . . . . . . . . . . . . . . . . . . . . . 6.1.2. Phase 2 . . . . . . . . . . . . . . . . . . . . . 6.2. Microsomal GSTs. . . . . . . . . . . . . . . . . . . . Biotechnological applications . . . . . . . . . . . . . . . . 7.1. Forced evolution. . . . . . . . . . . . . . . . . . . . . 7.2. Transgenics . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

* Tel.: +39 02 50315012; fax: +39 02 50315044. E-mail address: [email protected]. 1389-0344/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.bioeng.2006.05.020

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

150 150 151 151 152 154 155 156 157 157 159 159 160 160 160 161 162 163 163 164 165 166 166

150

C. Frova / Biomolecular Engineering 23 (2006) 149–169

1. Introduction Genome sequencing projects have started a new era in genetics and molecular biology, leading to the understanding of the sequence and organization of genes in many organisms, from the simplest bacteria to complex eukaryotes including man. With the development of high throughput technologies and sophisticated computational tools a change in perspective, from mere sequence description to the understanding of the structure and function of multiple genes and proteins in simple and complex organisms, took place. The ultimate goal is to go beyond the gene/protein level to elucidate the pathways leading to the organization and function of macromolecular complexes, organelles, cells, organs and whole organisms. All this is defined as systems biology and the new biotechnological tools that make it possible, genomics, transcriptomics, proteomics, functional/structural genomics, metabolomics, etc., go collectively under the name of ‘‘omics’’. Besides being invaluable in understanding structures and functions, omics are extremely powerful for detecting evolutionary relationships between sets of genes and, at a higher level, organisms. One field that can particularly benefit from omics is the study of gene families, often conserved across organisms, and of their evolution in terms of structure and functions. In this review the glutathione transferase gene/protein family is considered. Glutathione transferases (GSTs) (EC 2.5.1.18) are a superfamily of multifunctional proteins with fundamental roles in the cellular detoxification of a wide range of exogenous and endogenous compounds. Although several of them are not transferases at all (see below), these proteins are still collectively called GSTs, probably because the first discovered ones were indeed transferases, and as such they will be referred to throughout this review. In plants and animals, GSTs are the principal phase II enzymes in metabolic detoxification processes. Their main chemistry is to catalyze the conjugation of the tripeptide glutathione (GSH: g-Glu-Cys-Gly) with compounds containing an electrophilic centre, to form more soluble, nontoxic peptide derivatives, ready to be excreted or compartimentalized by phase III enzymes (Coleman et al., 1997). In addition, GSTs can serve as peroxidases, isomerases and thiol transferases (Jensson et al., 1986; Bartling et al., 1993; Fernandez-Can˜on and Pen˜alva, 1998; Board et al., 2000), or have non-catalytic functions among which binding of non-substrate ligands and modulation of signaling processes (Listowsky, 1993; Marrs, 1996; Mueller et al., 2000; Smith et al., 2003; Axarli et al., 2004; Adler et al., 1999; Loyall et al., 2000; Cho et al., 2001). The original view of GSTs as solely detoxication enzymes has thus gradually changed, and their roles extended to non-stress metabolism, as leukotriene and prostaglandin biosynthesis (Tsuchida et al., 1987; Kanaoka et al., 1997; Jakobsson et al., 1999) and the catabolism of aromatic aminoacids (FernandezCan˜on and Pen˜alva, 1998; Thom et al., 2001). Three main subfamilies of GSTs are generally recognized, each encoded by distinct multigene families: (1) the soluble or cytosolic GSTs (also termed canonical by some authors);

(2) the microsomal GSTs, now termed MAPEG (membraneassociated proteins involved in eicosanoid and glutathione metabolism); (3) the plasmid-encoded bacterial fosfomycin-resistance GSTs. Recent evidence suggests that Kappa GSTs, previously considered as just one class of the soluble GSTs, constitute a distinct subfamily instead (Jowsey et al., 2003; Ladner et al., 2004; Robinson et al., 2004). In addition, microsomal GSTs with primary and tertiary structure more similar to the Alpha class soluble GSTs than to MAPEGs have been identified (Prabhu et al., 2001, 2004; K.S. Prabhu and C.C. Reddy, personal communication). It is thus possible that the number of GST subfamilies is actually larger than so far thought. Furthermore, genomics and postgenomics studies are continuously highlighting the links with ‘‘related families’’ which borders with the GSTs can be quite thin and disputable. Such related families include glutaredoxins (GRX) (Xia et al., 2001; Collison and Grant, 2003), chloride intracellular channels (CLIC) (Harrop et al., 2001; Dulhunty et al., 2001; Cromer et al., 2002), dehydroascorbatereductases (DHAR) (Dixon et al., 2002a), selenocysteine glutathione peroxidases (SecGPX) (Epp et al., 1983), bacterial DsbA (Martin et al., 1993), eukaryotic protein elongation factors (eEF1Bg) (Jeppesen et al., 2003), all of which share with GSTs a basic structural motif, the thioredoxin fold. In this review I will summarize current knowledge on the main eukaryotic GST subfamilies (the soluble, the Kappa, the microsomal GSTs, and some related families), focusing on how structural and functional new acquisitions contributed to trace possible evolutionary scenarios of this protein superfamily. Bacterial GSTs are in some way a world apart, so far less characterized especially in terms of functions and diversification. For this reason here I will touch upon them only when appropriate to complete and explain general aspects of GSTs. For a more detailed description and discussion of bacterial GSTs readers are referred to specific reviews (Vuilleumier, 1997; Vuilleumier and Pagni, 2002). Biotechnological perspectives, in terms of both transgenics and directed evolution, will also be discussed. 2. Cytosolic GSTs (cGSTs) This subfamily, ubiquitously found in all aerobic organisms, is by far the most abundant, often counting tens of members in each species. For instance, in man and other mammalian species 15–20 different cGST genes have been identified (Hayes et al., 2005), 40–60 in plants (McGonigle et al., 2000; Wagner et al., 2002; Frova, 2003; Soranzo et al., 2004), 10–15 in bacteria (Vuilleumier and Pagni, 2002), over 10 in insects (Ranson et al., 1998 and refs. herein). Based on several criteria, including aminoacid/nucleotide sequence identity, physical structure of the genes (i.e. intron number and position) and immunoreactivity properties, cGSTs have been grouped into numerous classes, some of which are ubiquitous throughout taxa and even kingdoms, while other are organism-specific. Presently, seven classes of cytosolic GSTs are recognized in mammals, namely the specific Alpha, Mu, Pi

C. Frova / Biomolecular Engineering 23 (2006) 149–169 Table 1 Common and specific cytosolic GST classes in major taxa Taxa

Common

Specific

Mammals

Zeta (Z) Theta (T) Omega (O) Sigma (S)

2 2 1 1

Alpha (A) Mu (M) Pi (P)

4–6 5–6 1–2

Insects

Zeta (Z) Theta (T) Omega (O) Sigma (S)

1 1 1 1

Delta (D)

2–4

Plants

Zeta (Z) Theta (T)

2 1–3

Phi (F) Tau (U) Lambda (L) DHAR

?–16 ?–39 1–2 1–3

Bacteria

Theta (T) ?

?

Beta (B)

6–15

Number of genes/species are given in separate columns.

and the common Sigma, Theta, Zeta and Omega. Plants have six classes, Lambda, Phi, Tau, DHAR (dehydroascorbate reductases) (specific), Theta and Zeta, while five classes have been recognized in insects, the Delta (specific), Sigma, Theta, Zeta and Omega. In bacteria the picture is less clear: they certainly possess a specific class named Beta, in addition to other enzymes more related to the common Theta and possibly other classes (Table 1). In most organisms the specific cGST genes outnumber the common ones. This, together with the tendency of the specific GST genes to cluster within limited genomic regions (see ‘‘genome organization’’ below), provides interesting clues into the evolution of this gene family. Almost all soluble GSTs are active as dimers, of either identical (homodimers) or different (heterodimers) subunits, each encoded by independent genes. Heterodimerization is restricted to subunits of the same class, as monomers of different classes are unable to dimerize because of incompatibility of the interfacial residues. Rare monomeric active forms, as the plant Lambda and DHAR, have been reported (Dixon et al., 2002a), but these isolated cases do not question the general ‘‘dimer’’ rule. The monomeric enzymes lack GSHdependent conjugating and peroxidase activities with standard substrates (Dixon et al., 2002a). Nonetheless, they are considered GSTs on the basis of sequence and structure similarity and because they share with some sensu strictu GSTs other catalytic properties such as thiol transferase activity. 2.1. Nomenclature cGSTs were initially discovered and characterized in mammals, followed later by a few enzymes from plants, insects and other organisms. This originated a confusing situation, as for each organism or even species a different nomenclature was adopted. Furthermore, the first nonmammalian enzymes had very limited identities with the well-defined human Alpha, Mu and Pi classes, and thus were all allocated to the Theta class, at the time the most heterogeneous and comprehensive class. With the explosion

151

of sequencing studies a wealth of GST genes were recognized in many different organisms, which prompted the need of a more homogeneous and unequivocal classification and nomenclature system. This was developed essentially by adopting the existing mammalian system also for nonmammalian GSTs. According to the recommendations of the Committee for Human Gene Nomenclature, for each GST gene or subunit, the species of origin should be indicated by a three-letter prefix (Hsa for Homo sapiens, Mmu for Mus musculus, etc.), followed by GST and a letter indicating the class (see Table 1), and a progressive number indicating the order of gene discovery in the species. For enzymes, subunit composition is indicated. For instance, HsaGSTM1 and HsaGSTM2 define the first and second gene, or encoded subunit, of Mu class reported in man, while HsaGSTM1-2 indicates the heterodimer of human M1 and M2 subunits. Some variations of this scheme are still in use. So, for mammalian GSTs, the species is frequently indicated by a single letter (h for human, r for rat, m for mouse, etc.), while for other organisms the species is most often indicated by a two-letter prefix (Zm for Zea mays, Lc for Lucilia cuprina, etc.) (Edwards et al., 2000; Chelvanayagam et al., 2001). 2.2. Genome organization The completion of several genome sequencing projects, together with the development of powerful bioinformatics tools, has led to extensive gene annotation and construction of physical maps, that revealed the genome distribution of GST genes in an increasing number of organisms. Most cGST genes appear to form tight clusters, preferentially of the same class: the human Alpha, Mu and Theta (Morel et al., 2002 and refs. herein) and the mouse Pi (Henderson et al., 1998) genes are all grouped in specific and limited chromosomal regions. In insects too GST genes are grouped in clusters (Ranson et al., 1998; Toung et al., 1993). Bacterial GST genes often occupy nearly adjacent positions in operons or gene clusters involved in specific degradative pathways (Vuilleumier and Pagni, 2002). But the most striking examples of gene clustering are found among plant GSTs. The fully sequenced Arabidopsis thaliana and rice genomes provide an accurate picture of a dicot and a monocot species. The 52 Arabidopsis GST genes belong to six classes: 28 Tau (U), 13 Phi (F), 3 Theta (T), 2 Lambda (L) and 4 DHAR (Dixon et al., 2002a; Wagner et al., 2002). Excluding the DHAR genes, not precisely located yet, 34 of the remaining genes are present in tight class-specific clusters: for instance, 7 Tau genes lay in tandem in a 14 kb segment on chromosome 2 (Lin et al., 1999). Isolated genes account for less than 30% of all GST sequences (Dixon et al., 2002b). In rice, a bioinformatic approach has led to the identification of 61 cGST sequences: 39 Tau, 16 Phi, 3 Zeta, 2 Theta, 1 Lambda (Soranzo et al., 2004). A gene coding for a dehydroascorbate reductase, not represented in the abovementioned study, has also been cloned (Urano et al., 2000). All the 61 genes identified in silico were mapped and only 15 (<25%) occupy isolated positions in the genome. The remaining 46 are grouped in clusters of various dimensions, the most prominent of which are a large cluster of 23 Tau genes in a

152

C. Frova / Biomolecular Engineering 23 (2006) 149–169

239 kb contig on chromosome 10 and one of 8 Phi genes in a 318 kb segment on chromosome 1 (Soranzo et al., 2004). The systematic presence of clusters of GST genes in genomes of both plant and animal species is indicative of a common organizational theme within this gene family and reflects their evolutionary history. 2.3. General structure and functional implications As mentioned above, soluble GSTs are, as a general rule, biologically active as dimers of subunits of 23–30 kDa and an average length of 200–250 aminoacids. Sequence mean identity within class is typically >40%. However, pairwise comparisons often indicate a much broader range. For instance, in rice Tau and Phi GSTs, protein identities range between 17 and 98% and 23 and 82%, respectively (Soranzo et al., 2004). Interclass identities are significantly lower, usually <25% in mammals and <20% in plants. To date, the crystal structures of over a dozen soluble GSTs, belonging to the main plant, animal and bacterial classes, have been resolved. Their analysis clearly demonstrates that, despite pronounced overall sequence divergence, all GST proteins show striking levels of structural conservation, displaying a common 3D-fold and a dimeric organization. As an example, the 3D structure of a Tau class GST from wheat is shown in Fig. 2. Each subunit is composed of two spatially distinct domains: an N-terminal domain (domain I), consisting of b strands and a helices as secondary elements, and an all helical C-terminal domain (domain II).

Domain I adopts a thioredoxin-like fold (bababba) (Fig. 1a), which consists of two typical structural motifs, the N-ter b1a1b2 and the C-ter b3b4a3, linked together by a long loop containing an a-helix (a2). Together the two regions form a b-sheet of three parallel (b1b2b4) and one antiparallel (b3) b strands, sandwiched between the a2 helix on one side and helices a1 and a3 on the other side. Helix a2 and strand b3 are connected by a loop containing a cis-Proline which is highly conserved in all GSTs. The cis-Pro loop, though not directly involved in catalysis, is important in maintaining the protein in a catalytically competent structure (Allocati et al., 1999). Domain II consists of a variable number (4–7) of a-helices positioned downstream the thioredoxin structure and is connected to domain I by a short linker sequence (around 10 aa) (Figs. 1b and 2a). There are two ligand-binding sites per subunit. A very specific glutathione-binding site (G-site) constructed mainly from residues of domain I, and the hydrophobic substrate binding site (H-site), which is formed primarily by residues with non-polar side chains lying in domain II. The two sites together constitute the catalytically active site. The N-terminal domain is quite conserved, and contains specific residues critical for GSH binding and catalytic activity. In particular, the highly conserved Tyr7 of the mammalian Alpha/Mu/Pi classes and Ser17 of the ubiquitous Theta and Zeta, of the plant specific Phi and Tau and of insect Delta classes, have a crucial role in the catalytic activation of GSH (Fig. 2a). The Tyr/Ser hydroxyl group acts as hydrogen bond donor to the thiol group of GSH, promoting the formation and stabilization of the highly reactive thiolate anion which is the target for nucleophilic attack of an

Fig. 1. Diagram of TRX proteins secondary structure. Arrows indicate b strands, rectangles indicate a helices. Dotted squares in (a) indicate the N-terminal and the C-terminal motifs of the thioredoxin fold, connected by the a2 helix. In (b and c), ovals mark the position of the second domain, while dashed lines indicate extra domains in some of the proteins listed below the diagrams. The grey thick line in (b) indicates the short linker between domains I and II of GSTs. The nature of the second domain is indicated in parenthesis near the proteins.

C. Frova / Biomolecular Engineering 23 (2006) 149–169

153

Fig. 2. A ribbon diagram of the wheat Tau class GST TaGSTU4 monomer (a) and TaGSTU4-4 dimer (b). a-Helices are drawn as ribbons, b-strands as arrows. The inhibitor S-hexylglutathione is represented in ball-and-stick colored according to atom type. (a) a-Helices and b-strands are numbered. Thick light blue and red arrows indicate the glutathione binding (G-site) and substrate binding (H-site) sites. Thin black arrows indicate the position of the catalytic residue Ser (S), cis-Pro in the omonimous loop and the short sequence linking domains I and II (see text for details). (b) The dimeric structure of the enzyme, the two-fold axis relating to the dimer subunits normal to the plane (left) and in the plane of the page (right).The two subunits are colored lilac and light blue, respectively (adapted from Thom et al., 2002, with permission from American Chemical Society).

electrophilic substrate (Dirr et al., 1994a; Armstrong, 1997, 1998). The essential role of these residues in GST catalysis has been confirmed by site-directed mutagenesis of the tyrosine of porcine class Pi (Dirr et al., 1994b) and the serine of human Theta (Tan et al., 1996), insect Delta (Caccuri et al., 1997) and plant Zeta (Thom et al., 2001) enzymes: in all cases the catalytic activity of the enzyme was lost. Some GST classes have, instead of a Ser or Tyr, a cysteine at the usual active site position, a residue that promotes the formation of mixed disulphides with glutathione rather than the formation of the thiolate anion. This feature is shared by Omega, Beta, Lambda, DHARs, in addition to glutaredoxins and CLICs. All these enzymes have poor, when not null, conjugation activity to GSH, and are implicated in redox reactions instead. Whereas Lambda, DHARs, glutaredoxins and CLICs are biologically active as monomers, Omega and Beta class GSTs are dimers. Omega GSTs exist in several mammalian species, in insects and in nematodes. As glutaredoxins, they have glutathione-dependent thiol transferase activity and can also catalyze dehydroascorbate reduction. By contrast, GSH conjugating activity is negligible (Board et al., 2000). As for the Beta

GSTs, studies on the Proteus enzyme showed it exhibits both a conjugating activity and a redox activity towards disulphides. The Cys10 active residue is essential for the redox activity, but not for the conjugation activity. For these characteristics, Beta GSTs have been proposed as intermediate enzymes bridging the evolutionary pathway from thiol-disulphide oxidoreductase to glutathione transferases (Caccuri et al., 2002). Unlike domain I, domain II is quite variable both in sequence and topology, and this diversity determines the ample and distinct hydrophobic substrate specificities observed for the different enzymes. Each subunit of dimeric GSTs is catalytically independent, in that it possess a G-site and an H-site. Yet, with the abovementioned exceptions, all catalytically active GSTs found in nature are dimers. The reasons for this apparent contradiction have long remained obscure and only recently a number of studies have started elucidating the molecular basis for dimerization and the reasons why the dimeric organization has been so highly conserved by evolution. In all dimeric GSTs the two subunits are related by a two-fold axis (Fig. 2b). The main intersubunit interactions occur between domain I of one

154

C. Frova / Biomolecular Engineering 23 (2006) 149–169

subunit and domain II of its partner. Different types of interactions are involved in assembling and maintaining the quaternary structure. In Theta/Sigma/Beta/Tau classes the interactions are basically hydrophilic, whereas in the Alpha/ Mu/Pi/Omega and Phi classes the protein surfaces and the interactions engaged in dimerization are hydrophobic. One important hydrophobic interaction in Alpha/Mu/Pi and Phi GSTs is constituted by a lock and key motif, in which the side chain of an aromatic aminoacid, termed the key residue (usually a tyrosine or a phenylalanine), protruding from the loop preceeding the b3 strand from domain I in one subunit is fit into a hydrophobic pocket formed by domain II a4 and a5 helices of the other subunit (Armstrong, 1997; Hegazy et al., 2004). Mutation of the key residue in HsGSTP1-1, GSTA1-1 and GSTM1-1 leads to decreased dimerization and/or dimer stability and reduced catalytic activity (Stenberg et al., 2000; Sayed et al., 2000; Hornby et al., 2002). In the case of GSTP1-1, dimerization has been suggested to stabilize the tertiary structure of each subunit (Erhardt and Dirr, 1995) and to constitute the structural basis for positive or negative intersubunit cooperativity for GSH binding (Caccuri et al., 1999). More recently Hegazy et al. (2004) elegantly showed that the two active sites of GSTP1-1 work synergistically. They generated a homodimeric mutant where the key residue Tyr50 was substituted by Ala in both subunits (Y50A/ Y50A) and a heterodimer composed by one wild-type and one mutated subunit (GSTP1/Y50A). By testing the specific activity of the purified native enzyme (GSTP1-1), the homozygous mutant and the heterodimer using the model substrate CDNB (1-chloro-2,4-dinitrobenzene), they found that the mutant Y50A/Y50A was nearly inactive, 25,000-fold activity reduction with respect to the native enzyme. Interestingly, the heterodimer also showed a very marked activity reduction of 27-fold, instead of the two-fold which would have been expected if the two subunits were working independently. Thus, at least for this enzyme, a dimeric organization greatly increases the enzyme efficiency. Although it is not clear whether this behavior is shared by other cytosolic GSTs, it is reasonable to expect that it may apply to other classes displaying the lock and key motif, such as the Mu, Alpha and Phi class enzymes. In most GST dimers a solvent-accessible deep V-shaped crevice is observed at the intersubunit interface, with a buried area of 2700–3400 Angstrom. However, in Theta, Omega and Kappa GSTs, the crevice is more open and the buried area reduced. Differences in the openness of the subunit interface and in the polarity of the H-site are elements contributing to tailor the functional specificities of the enzymes. For instance, it has been proposed that Omega GSTs might be able to bind relatively large and partially hydrophilic molecules such as proteins (Board et al., 2000). Given the presence of a cysteine as active site residue at the G-site, it is conceivable that Omega GSTs have a role in reducing the S-thiol adducts with GSH or cysteine formed by a number of cellular proteins as a consequence of oxidative stress (Board et al., 2000). Also for Kappa GSTs, binding to another protein as part of its function has been hypothesized (Ladner et al., 2004).

3. Kappa GSTs The first traces of what now appears to be a distinct GST subfamily date back to 1991, when a GST enzyme was purified from the matrix of rat liver mitochondria (Harris et al., 1991). This enzyme was initially allocated to the Theta-class GSTs, on the basis of limited N-terminal sequence analysis. Upon full determination of the cDNA and protein sequence, major differences with the other known mammalian GSTs emerged and a new class, termed Kappa, was introduced to allocate this enzyme (Pemble et al., 1996). Porcine and human homologs were also reported, but these enzymes were not characterized at that time. Molecular cloning and biochemical characterization of a mouse (Jowsey et al., 2003) and human (Robinson et al., 2004; Morel et al., 2004) Kappa GST and the 3D structure determination of the rat (Ladner et al., 2004) and human (Li et al., 2005) enzymes, added new elements for the definition of this new GST subfamily. All mammalian Kappa enzymes are encoded by single copy genes, highly similar in the three species both in sequence (approximately 70% aminoacid identity) and in organization: they all consist of eight exons and seven introns at conserved positions (Jowsey et al., 2003; Morel et al., 2004). Blast searches have revealed similar sequences in ESTs from B. taurus, S. scrofa, G. gallus, X. laevis and C. elegans (Morel et al., 2004), and GSH transferases related to the Kappa enzymes are also found in bacteria (Vuilleumier and Pagni, 2002). Their presence in evolutionary distant species is suggestive of a fundamental and conserved biological function. Kappa enzymes exhibit GSH-dependent conjugating and peroxidase activity with model substrates, thus sharing common catalytic features with the canonical GSTs. Kappa GST genes have a wide tissue expression profile (Jowsey et al., 2003; Morel et al., 2004), and the enzymes are typically found in mitochondria and not in the cytosol. In addition, the human enzyme was detected also in peroxisomes (nothing is known so far about the other enzymes) (Morel et al., 2004). Since mitochondria and peroxisomes are involved in lipid metabolism and are cellular sites where high amounts of reactive oxygen species are produced, the localization of the human Kappa GST in both organelles is suggestive of a role in b-oxidation of fatty acids and in detoxication of lipid peroxides. Aminoacid sequence comparisons indicate that Kappa GST enzymes share no overall significant homology with any of the other GST classes, but are closer to E. coli DsbA, a protein disulphide bond isomerase, and to bacterial 2-hydroxychromene-2-carboxylate (HCCA) isomerase, a GSH-dependent oxidoreductase involved in naphthalene degradation (Ladner et al., 2004; Robinson et al., 2004). Also the secondary structure topology of Kappa GSTs is more similar to the bacterial HCCA and DsbA than to all other cytosolic GSTs. While in the latter group the helical domain II is added downstream the thioredoxin fold, in Kappa GSTs, HCCA and DsbA it is inserted within the thioredoxin fold, between the N-ter (bab) and the C-ter (bba) motifs (Fig. 1c). These findings brought two independent research groups to hypothesize an

C. Frova / Biomolecular Engineering 23 (2006) 149–169

evolutionary pathway of Kappa GSTs, distinct from that of canonical GSTs (Ladner et al., 2004; Robinson et al., 2004). However, despite the marked sequence divergence with canonical GSTs, an alignment of secondary structural elements of the thioredoxin fold is possible. When comparisons are limited to the bab and bba motifs, the human Kappa enzyme (hGSTK) is more similar to the Theta-class GSTs than to any other GST class, or DsbA (Li et al., 2005). In addition, as Thetaclass GSTs, the dimeric human and rat Kappa enzymes have an open-wings butterfly shape where the deep V-shaped intersubunit crevice is absent (Ladner et al., 2004; Li et al., 2005). 4. Microsomal GSTs Microsomal GSTs, now designated MAPEG, are also ubiquitous, with a well documented presence in a wide spectrum of organisms occupying all positions in the evolutionary scale (Pflugmacher et al., 2000; Bresell et al., 2005). However, they are less numerous than the cytosolic GSTs and, with the exception of the mammalian ones, far less characterized. Most MAPEG proteins are involved in the synthesis of eicosanoids, leukotrienes and prostglandins, catalyzing GSHdependent transferase or isomerase reactions. Others have different functions. For instance, the first microsomal GST characterized, the human MGST1 (Morgenstern et al., 1982) has no part in either leukotriene or prostglandin biosynthesis, but exhibits catalytic activities more typical of cytosolic GSTs. It catalyzes GSH conjugation to several halogenated arenes, including CDNB, as well as various polyhalogenated unsaturated hydrocarbons (Andersson et al., 1994). In addition to the transferase activity, MGST1 also catalyzes the GSH-dependent reduction of lipid hydroperoxides such as organic hydroperoxides, fatty acid hydroperoxides and phospholipids hydroperoxides (Morgenstern and DePierre, 1983; Mosialou et al., 1995). For these characteristics MGST1 is thought to be essentially a detoxication enzyme involved in the cellular defense against toxic xenobiotics as well as metabolites produced as a consequence of oxidative stress.

155

For a long time MGST1 remained the only characterized microsomal GST. Additional MAPEG proteins were then identified, namely leukotriene C4 synthase (LTC4S), 5lipoxigenase activating protein (FLAP) and prostaglandin E synthase 1 (PGES1). FLAP is a non-catalytic protein that binds non-enzymatically to arachidonic acid and is required to activate 5-lipoxigenase (5-LO), the enzyme that catalyzes the conversion of arachidonic acid to leukotriene A4 (LTA4). FLAP is likely involved also in the synthesis of leukotriene B4 (Mandal et al., 2004). LTC4S has glutathione transferase activity and catalyzes leukotriene C4 (LTC4) synthesis from LTA4 and reduced glutathione. PGES1, formerly named MGST1-L1 due to its homology to MGST1, has no or very limited glutathione transferase activity with LTA4 or CDNB, respectively, but, in contrast to LTC4S and FLAP which have no peroxidase activity, PGES1 is able to reduce cumene hydroperoxide and other substrates (Thore´n et al., 2003). However, the main chemistry of PGES1 is to catalyze the GSHdependent isomerization of PGD2 to PGE2 (Fig. 3). Additional roles for PGES1 in cancer development, apoptosis and in Alzheimer’s disease are suggested by its strong upregulation by p53 in a colorectal cancer cell line and in b-amyloid treated rat astrocytes, respectively (Polyak et al., 1997; Satoh et al., 2000). Bioinformatic searches of human EST database led to the identification of two additional enzymes, MGST2 and MGST3 (Jakobsson et al., 1996, 1997). Upon cloning, purification and characterization, both enzymes were found to catalyze the conjugation of GSH to LTA4 and thus the formation of LTC4 (Fig. 3). When CDNB was used as substrate, however, transferase activity was detected only for MGST2. In addition, MGST2 and MGST3 possess peroxidase activity in that they catalyze the GSH-dependent reduction of 5-hydroxyperoxyeicosatetraenoic acid (5-HPETE) to 5-hydroxyeicosatetraenoic acid (5-HETE). Due to their catalytic activities MGST2 and MGST3 may play roles in the detoxification of xenobiotics and in cell protection against oxidative damage. A first analysis of the MAPEG family included 13 proteins, 6 mammalian and 7 from plants, fungi and bacteria. According to

Fig. 3. Involvement of MAPEG in leukotrienes and prostglandins biosynthetic pathways. MAPEGs are enclosed in grey dotted rectangles.

156

C. Frova / Biomolecular Engineering 23 (2006) 149–169

multiple sequence alignments, four groups were recognized, termed I, II, III and IV (Jakobsson et al., 2000). Group I includes MGST2, LTC4S and FLAP, while MGST1 and PGES1 form group IV. Group III is exclusively bacterial (one E. coli and one V. cholerae sequence). A third bacterial protein, the Synecocistis MAPEG, has no clear collocation, although it appears more related to group I. Finally, group II is the most heterogeneous as to the origin of the sequences, including the human MGST3 as well as members from plants and fungi. The discrepancy in numbering between the human microsomal GSTs and the groups (MGST1 is in the fourth group, MGST2 in the first, MGST3 in the second) may be confusing. For the sake of clarity I propose, and will adopt in the present paper, a different nomenclature, where classes (instead of groups) are indicated, and numbered according to the human MGST they include. As shown in Table 2, group IV becomes class 1 (from MGST1), group I class 2, group II class 3. The third group is renamed as class B1 (from Bacteria) and the Synecocistis sequence as class B2. Subclasses are also indicated. These 13 proteins are distantly homologous, with only 2 residues strictly conserved in all, Asn78 and Arg114, plus, with the exception of FLAP, Glu81 (MGST1 numbering). Other almost conserved aminoacids are Pro47 (no FLAP and B1), Arg74 (no FLAP and B1), Pro85 (no B2) and Tyr122 (no B1) (Jakobsson et al., 1999). For some of these aminoacids, a specific role in catalysis has been hypothesized. Arg51 of LTC4S (corresponding to Arg74 in MGST1) is thought to be the proton donor for the opening of the LTA4 epoxide ring. In the same enzyme Tyr93 was identified as the residue responsible for the formation of the thiolate anion of glutathione (Lam et al., 1997). In other MAPEGs the conserved Tyr122 or Tyr118 could serve the same role. Homology is a lot higher within class, with the highest aminoacid identity (44%) between LTC4S and MGST2. The original set of 13 MAPEG members has meanwhile substantially expanded thanks to bioinformatic screenings of EST and genome database. A total of 131 distinct MAPEG proteins were found, 52 from prokaryotes and 79 from a wide spectrum of eukaryotes (Bresell et al., 2005), and an additional 24 members were detected in plants (C. Frova, unpublished results). Also in this extended set, a limited number of residues, among which those mentioned above, is conserved in nearly all members. A first phylogenetic analysis reveals clustering of the members in six eukaryotic and at least two, possibly more, Table 2 Proposed MAPEG new classification Old group (Jakobsson et al., 2000)

New class

Subclass

IV I II III Synecocistis

1 2 3 B1 B2

MGST1; PGES1 MGST2; LTC4S; FLAP MGST3

prokaryotic branches, suggesting that the original four classes are inadequate to explain MAPEG diversity. The eukaryotic branches (subclasses in Table 2) correspond each to the six human proteins, i.e. MGST1, 2, 3, LTC4S, FLAP, PGES1. The bacterial proteins reflect the original scheme, with one (or more) specific clusters entirely distinct from the eukaryotic MAPEGs, and a Synecocistis branch more related to the MGST2/FLAP/LTC4S branches. All eukaryotic clusters are supported by sequence specific patterns which are 100% unambiguous and thus constitute clusters’ fingerprints (Bresell et al., 2005). Bacterial and animal species often possess more than one MAPEG member, either of the same or different classes (Bresell et al., 2005). On the contrary, plant and fungus species express only one MAPEG protein each, always of class 3 (C. Frova, unpublished results). Functional characterization of the newly identified proteins is so far limited to very few members, namely the pike and Drosophila (MGST1-like), Arabidopsis (MGST3-like), E. coli (B1) and Synecocistis (B2) proteins (Bresell et al., 2005). All display GSH conjugating activity with CDNB, whereas only the MGST1-like enzymes are able to reduce cumene hydroperoxide. Thus the Arabidopsis MGST3-like enzyme, in contrast with its human counterpart, does not seem to have peroxidase activity. However, the discrepancy could be due to the different substrate tested, cumene hydroperoxide and 5HPETE for the plant and the human enzymes, respectively. 4.1. Structure and membrane topology Microsomal GSTs are strongly divergent from the cytosolic enzymes: sequence identity with cGSTs is less than 10% and subunits are shorter, with an average length of 150 aminoacids. More important, the structure is completely different. MAPEG proteins characterized so far have transmembrane domains, the amino and carboxyl termini of the protein protruding into the luminal side of the membrane, while putative sites for GSH and substrate binding are located in loops facing the cytosol (Lam et al., 1997; Busenlehner et al., 2004). The number of transmembrane domains were initially predicted to be three (Jakobsson et al., 1996, 1997; Lam et al., 1997). A new interpretation of hydropathy plots, corroborated by 3D structural maps, now suggests that four membrane spanning domains are more likely (Bresell et al., 2005; Schmidt-Krey et al., 2000, 2004). The quaternary structure of MAPEGs appears to be nonunivocal. The 3D map of MGST1 (Schmidt-Krey et al., 2000) indicates the enzyme is a homotrimer. Contrary to cGSTs, both the human and rat MGST1 multimers bind only one substrate glutathione molecule, indicating that the monomers are not kinetically independent, but rather interact to constitute the active site of the enzyme (Sun and Morgenstern, 1997; Lengqvist et al., 2004). Projection structures of LTC4S and PGES1 support a homotrimeric quaternary structure for these enzymes too (Thore´n et al., 2003; Schmidt-Krey et al., 2004), and LTC4S homo-oligomers were detected also in vivo (Svartz et al., 2003).

C. Frova / Biomolecular Engineering 23 (2006) 149–169

Analyses of deletion/substitution mutants identified a Cterminal domain (residues 114–150) necessary/sufficient for LTC4S oligomerization and activity (Lam et al., 1997; Svartz et al., 2003), suggesting that oligomerization is important for the catalytic and structural stability of the active enzyme. By contrast, FLAP can exist as monomer, dimer or trimer (Mandal et al., 2004). Furthermore, there is evidence that FLAP and LTC4S subunits can interact, forming heterodimers or heterotrimers, although the subunit composition of heterotrimers is not clear yet (Mandal et al., 2004). The quaternary structures of MGST2 and MGST3 are presently unknown. In summary, a number of uncertainties remain as to the spatial organization, subunit composition and membrane topology of MAPEGs, and further studies are needed to clarify these points. 4.2. Extravagant microsomal GSTs In addition to the classical MAPEGs described above, two microsome-associated GSTs with quite different characteristics, a sheep liver and a human liver microsomal GSTs, were recently cloned and characterized (Prabhu et al., 2001, 2004). The sheep liver enzyme is associated with the inner microsomal membrane. Its molecular mass (25.6 kDa), subunit length (222 aa) and primary structure are similar to Alpha class cytosolic GSTs (Prabhu et al., 2001). Conserved with cGSTs is also the N-terminal domain, with a tyrosine residue at the active site. The enzyme, designated sheep liver microsomal S-transferase A1-1 (SLM GSTA1-1), exists as a homodimer and is highly active as GSH transferase and selenium-independent peroxidase (Prabhu et al., 2001). Interestingly, it also exhibits LTC4 synthase activity. Despite some differences in the C-terminal domain, the overall 3D structure and the active site topology of SLM GSTA1-1 are identical to the cytosolic GSTs of class Alpha (K.S. Prabhu and C.C. Reddy, personal communication). The human enzyme, designated M-GSTA (Prabhu et al., 2004), shares many characteristics with SLM GSTA1-1, including the molecular mass and sequence (99% aminoacid identity) (K.S. Prabhu and C.C. Reddy, personal communication). Conversely, it has no sequence similarity or immunological cross-reactivity with the human liver microsomal MGST1. As the sheep liver enzyme, M-GSTA is intrinsic to the microsomes and at the same time is a class Alpha enzyme (Prabhu et al., 2004). Contrary to MAPEGs, neither SLM GSTA1-1 nor M-GSTA have transmembrane domains. Therefore their membrane topology must be different from that of MAPEG proteins. The exact mechanism of insertion into the membrane is not precisely defined. However, based on the 3D structure of the sheep liver monomer and in analogy with cyclooxigenases (Smith et al., 2000), the a-helix close to the C-terminus could be responsible for anchoring the protein in the microsomal membrane (K.S. Prabhu and C.C. Reddy, personal communication). 5. Functions GST functions are quite diversified not only between but also within classes. During the course of evolution substantial or more subtle changes in sequence and structure have resulted

157

in functional diversification in terms of chemistry, activity and/ or substrate specificity. Thus the panel of GSTs now present in nature encompasses enzymes that catalyze conjugation, reduction and isomerase reactions, as well as proteins that act non-enzymatically as ligandins or signal transduction modulators (Table 3). A variety of approaches, including the creation and analysis of mutants, biochemical/kinetic assays with various substrate classes, expression/activity profiling in different tissues/developmental stages and under a range of conditions, have contributed to elucidate broad GST functions. Such studies have brought, for instance, to the definition of major roles in herbicides detoxification for the plant Phi GSTs, or in chemotherapeutic drugs detoxification for mammalian GSTs. With the aid of structural biology approaches it has been possible to understand, sometime to a great detail, the links between protein structure and functions. By defining the role of specific aminoacid residues in tailoring protein domains, these studies unraveled the principles governing the chemistry and substrate selectivity of single enzymes, and thus have been fundamental for the elucidation of the molecular bases of functional diversification of closely related GSTs, especially those belonging to the same class. This knowledge has profound applicative consequences, in that it provides guidelines for manipulating the original function through forced evolution. A paradigmatic example are the mammalian Alpha class GSTs. Of the five human enzymes, four (GSTA1-1, GSTA2-2, GSTA3-3 and GSTA4-4) have been thoroughly characterized. Despite sharing pronounced sequence homology and the presence of a C-terminal helix (a9) which folds back onto the N-terminal domain, thereby contributing to the binding of both glutathione and the electrophilic substrate, these four enzymes display rather distinct chemistries and substrate preferences. In particular, GSTA1-1 shows high catalytic activity with CDNB but much lower efficiency with alkenal substrates. By contrast GSTA4-4 has evolved a specific catalytic efficiency towards alkenals several orders of magnitude higher than that of other GSTs, approximately 200-fold higher than that of GSTA1-1. A comparison between the two enzymes (Burns et al., 1999) indicates that the different substrate specificities depend essentially from the shape and characteristics of the substratebinding pocket, which are determined by the positioning and packing of amino acid residues in the surrounding structural elements, namely the b1–a1 loop, the C-terminal part of the a4 helix and the C-terminus. Mutagenesis and structural studies (Bijo¨rnestedt et al., 1995; Burns et al., 1999) identified a limited number of residues lying within these elements, that are crucial for the high activity of GSTA4-4 with alkenals, that is Gly12 (Ala in GSTA1-1) in the b1–a1 loop, Ile107 (Leu), Met108 (Leu) and Phe111 (Val) at the end of the a4 helix, Pro208 (Met), Tyr212 (Ser), Val213 (Leu), Val 216 (Ala) and Pro222 (Phe) at the C terminus. Most of these residues are specifically conserved in all GSTs with high activity towards alkenals. In particular, Tyr212 is the key catalytic residue and is positioned to interact with the aldehyde group of the alkenal substrate and polarize it to facilitate the Michael addition to GSH. The correct placement of Tyr212 would be prevented by

158

C. Frova / Biomolecular Engineering 23 (2006) 149–169

Table 3 Main structural and functional characteristics of selected TRX proteins Family/class

Extra domain

Quaternary structure

Active site residue

Catalytic activity a

b

c

C-ter C-ter C-ter C-ter C-ter

Dimer Dimer Dimer Dimer Dimer

Tyr Tyr Tyr Ser Ser

+ + + () ()

+ +  + +

+ + + +

Phe/Tyr catabolism; DCA dechlorination

Delta Phi Tau

C-ter C-ter C-ter

Dimer Dimer Dimer

Ser Ser Ser

+ + +

+ +

+

Ligandin Ligandin; signaling modulation

Beta Lambda DHAR Omega Kappa

C-ter C-ter C-ter C-ter Inserted

Dimer Monomer Monomer Dimer Dimer

Cys(Ser) Cys Cys Cys Ser

+

Inserted C-ter

Monomer Monomer

Cys Cys

C-ter None None C-ter Inserted

Monomer Monomer Monomer Dimer Tetramer

Cys Cys Cys Ser/Tyr SeCys

GST Alpha/Mu/Pi Microsomal a Sigma Theta Zeta

Other TRX DsbA CLIC GRX2 (E. coli) GRX1 (E. coli) GRX1/2 (yeast) eEFB1g SecGPX

() +

e

Signaling modulation – Prostaglandin synthesis

+ + + +

+ +

Ion channels modulation

+ +

+

d

Non-catalytic and/or specific functions

+

+ ?

Ion channels formation/ modulation

+ + +

/+ +

Protein elongation H2O2 reduction

Catalytic activity is as follows: a, GSH transferase; b, peroxidase; c, isomerase; d, thiol transferase; e, dehydroascorbate reductase.

any side chain at residue 12, thus explaining the essential requirement of Gly at that position, indicated also by mutation studies (Bijo¨rnestedt et al., 1995). Proline at position 208 seems also important for the positioning of Tyr212. The other two enzymes, GSTA2-2 and GSTA3-3, are 89% identical in amino acid sequence and have comparable activities towards CDNB. However, while the former is highly efficient in reducing cumene hydroperoxide (CuOOH) and has negligible isomerase activity with ketosteroids like D5androstene-3,17-dione (AD), the latter has opposite characteristics (Pettersson et al., 2002). Both enzymes feature the active site Tyr9 at the G-site. However, in the AD isomerization reaction typical of GSTA3-3, Tyr9 promotes acid–base catalysis with the thiolate of glutathione serving as catalytic cofactor, while in the peroxidase reaction catalyzed by GSTA22 the same thiolate makes a nucleophilic attack on the hydroperoxide substrate. Thus the two enzymes make a different utilization of the reactivity of the glutathione sulphur atom. Of the 25 amino acids that are variant in the two enzymes, 5 are sufficient to determine the catalytic specificities of GSTA2-2 and GSTA3-3. They are all located in the H-site, and lie in the b1–a1 loop (positions 10 and 12), in the C-terminal portion of helix a4 (position 111) and in the C-terminal region (positions 208 and 216). These examples demonstrate how in GSTs a small number of selected mutations in the right position (the H-site) have played a crucial role in the evolution of enzymes with new activities, and constitute a valuable background for attempting the engineering of new useful ones.

Most of the abovementioned approaches have dealt with one or very few genes/enzymes at a time. Thus progress, although substantial, has been relatively slow, since a large amount of data had to be pooled and/or confronted in order to draw some solid conclusions. With omics and bioinformatics irrupting as major players on the scene, an upshift into a higher level of analytical procedures occurred, impressing an accelerated pace to functional studies. Screening of increasingly large EST and genome database, besides bringing to the identification of an unexpectedly high number of new GSTs, revealed unsuspected links (at the sequence, structure and functional levels) with other protein families previously thought to be unrelated. A good example is the CLIC family. Sequence alignment revealed high similarity to plant DHAR and, to a less extent, to Omega class GSTs, and the conservation of several key residues that are conserved through almost all GSTs. Also the 3D structure of CLIC is very similar to that of Omega GSTs (Harrop et al., 2001; Dulhunty et al., 2001). CLIC function is to form ion channels in intracellular membranes in addition of being putative regulators of chloride channels. While CLIC do not display GST or PDX activity, Omega GSTs are able to modulate calcium release channels, the ryanodine receptors, thus highlighting a functional link between the CLIC and the GST families (Dulhunty et al., 2001). In other cases, as for bacterial GSTs, the definition of the genes’ genomic context has been instrumental in the attribution of functions to otherwise uncharacterized GSTs (Vuilleumier and Pagni, 2002).

C. Frova / Biomolecular Engineering 23 (2006) 149–169

It is beyond the scope of this review to analyse in detail GST functions, briefly described above and in previous sections. This topic has been dealt with by several reviews covering human (Hayes and Pulford, 1995; Hayes et al., 2005), plant (Marrs, 1996; Frova, 2003; Edwards et al., 2000; Dixon et al., 2002b) and bacterial (Vuilleumier and Pagni, 2002) GSTs. Here I will simply highlight novel insights on GST functions that have come from omics approaches, in particular proteomics and reverse genetics. 5.1. Proteomics studies In plants, the two prevalent classes, the Phi and the Tau have long been considered to function primarily as herbicide detoxicant and in the defense against oxidative stress, respectively. Proteomic analysis of A. thaliana GSTs now question this strict role division. Smith et al. (2004) analysed GSTs expressed by plants exposed to copper, a promotor of oxidative stress, and to the safener benoxacor. Safeners are compounds that enhance crop tolerance to herbicides, often by selective induction of GSTs that detoxify specific herbicides. Eight Arabidopsis GST enzymes, six Phi and two Tau, were found to be expressed in seedlings grown in control conditions. Of these, four, three Phi and one Tau were significantly induced by copper, only one Tau by benoxacor. These results indicate that Phi GSTs have diffused antioxidant functions, in addition to their classical herbicide detoxification roles. In another study (Sappl et al., 2004), a class-specific dose– response to salicylic acid (SA) was observed. SA has several physiological roles in plants, including defense responses to pathogens. Although the mechanism of action of SA is not clearly understood, it is possible that at low doses it may act as a signaling molecule, inducing pathogenesis-related gene expression during systemic acquired resistance (SAR) or, at high levels, inducing programmed cell death such as in the hypersensitive response (HR). High doses of SA also inhibit the ROS (reactive oxygen species) scavenging system (catalases and ascorbate peroxidases), thus causing an accumulation of toxic ROS. In their study, Sappl et al. (2004) found that Tau GSTs were maximally induced by low levels of SA (100 mM for 6 h), while Phi enzymes were induced mainly by a stronger treatment (1 mM for 24 h). This differential response indicates a different regulation of Phi and Tau GSTs by SA, and suggests a role as plant defense proteins against oxi-stress for Phi enzymes, which is consistent with the results from copper treatment reported above. 5.2. In vivo functions by knock out studies Due to the high number of members of this family, partial functional redundancy is likely, especially for the most abundant classes. Accordingly, disruption of GST genes has been most informative for GST classes represented by few members, such as the common Sigma and Zeta classes, the mammalian Alpha and Pi classes or MAPEGs. Mammalian Sigma GST catalyzes the synthesis of prostaglandin D2, a key eicosanoid involved in inflammatory response.

159

Knock out mice for this gene show weaker allergic reaction with respect to wild-type animals (Urade et al., 2004). The murine Pi class GSTs includes two genes with putative major roles in the detoxification of a number of carcinogenic compounds, especially polycyclic aromatic hydrocarbons (PAH) metabolites. In vitro studies with several mouse cell lines showed that Pi enzymes are by far the most active GSTs of all in the detoxification of (+)-anti-7,8-dihydroxy-9,10-oxy7,8,9,10-tetrahydrobenzo[a]pyrene, a highly carcinogenic metabolite of benzo[a]pyrene (Hu et al., 1997). This role has been confirmed in vivo by analysis of a mouse Pi double knockout line, Pi1 / Pi2 / . Whereas in control growth conditions the null animals showed no obvious phenotype, upon treatment with a PAH compound associated with a TPA (tumor promoting activity) substance, they developed approximately three-fold more skin papillomas with respect to the wild-type mice. The time of tumor onset was also anticipated (Henderson et al., 1998). These results indicate that Pi GSTs have no critical physiological functions, but rather a role in xenobiotic defence, in particular as protectants towards cancer. Murine GSTA4-4 is particularly active in the conjugation of 4-hydroxynonenal (4-HNE), a strong electrophile, Michael acceptor lipid peroxidation product, that can form covalent adducts with proteins, nucleic acids and phospholipids. GSTA4 homozygous null mice are viable in normal growth condition, but are more susceptible to bacterial infection and have a shorter survival time when treated with paraquat, a well known oxidative stress inducer. Knock out of GSTA4 results in decreased ability to conjugate 4-HNE with a consequent increase of the steady-state level of this aldehyde in several tissues. This decrease is compensated by an mRNA increase of other GSTs, in particular Mu class enzymes, as well as of other antioxidant defense enzymes such as SOD 1 and 2, CAT and GPX1 (Engle et al., 2004). Taken together these data indicate major functions of GSTA4 in the defense against oxidative stress. The subcellular localization of GSTA4 (approximately 25% of the cellular mGSTA4-4 is localized in the mitochondrial matrix) (Raza et al., 2002) and the presence of ARE (antioxidant response elements) in the gene promoter (Hayes et al., 2005) are consistent with this role. Zeta GSTs are involved in the phenylalanine/tyrosine pathway, catalyzing the isomerization of maleylacetoacetate to fumarylacetoacetate, the penultime step of tyrosine catabolism. The MAAI activity of mGSTZ1 is confirmed in vivo: null GSTZ1 animals lack activity toward maleylacetone and chlorofluoroacetic acid. These mice do not suffer under normal dietary conditions, but rapidly loose weight and ultimately die when fed with supplemental 2% phenylalanine (Fernandez-Can˜on et al., 2002). Knock out of mGSTZ1 stimulates upregulation of GSTs of other classes (Alpha, Mu and Pi) (Lim et al., 2004), as well as of NAD(P)H:quinone oxidoreductase (NQO1), probably through accumulated tyrosine degradation products (Hayes et al., 2005). Notably all these genes have an ARE or EpRE (electrophilic responsive element) in their promoters (Hayes et al., 2005). Thus, also Zeta GSTs seem to functionally contribute to an antioxidant and electrophile defense network.

160

C. Frova / Biomolecular Engineering 23 (2006) 149–169

Disruption of non-mammalian cGST genes is limited to a bacterial gene, the cytosolic GSTB gene of Proteus mirabilis. The null strain is more sensitive to hydrogen peroxide, CDNB, fosfomycin and minocycline, indicating that the enzyme is involved in protective mechanisms against oxidative as well as chemical stress (Allocati et al., 2003). As for MAPEGs, mice with disrupted FLAP, LTC4S and PGES1 genes have been created. FLAP knockout animals do not synthesize leukotrienes. Neither LTC4 nor the preceding metabolites of the 5-LO pathway, such as 5-HETE and LTA4 are apparently formed, indicating that FLAP is essential for the synthesis of all leukotrienes (Byrum et al., 1997). Disruption of the LTC4S gene results in a marked reduction of LTA4 conjugation to GSH and in the level of LTC4 metabolites, such as LTD4 and LTE4, while the precursor of LTC4, 5-HETE, and other leukotrienes (LTB4) and PGD2, which are formed through different pathways, are apparently not affected (Kanaoka et al., 2001). These effects are consistent with the very restricted substrate specificity of the enzyme and with its position in the leukotriene synthesis pathway (see Fig. 3). Disruption of the PGES1 gene precludes the synthesis of PGE2 and of related responses normally triggered by this prostglandin during inflammatory processes and infection (Trebino et al., 2003). On the whole, the specific functions of these three MAPEG proteins are confirmed by the analysis of null animals. Unfortunately, no such studies have yet been performed with other MAPEGs, in particular MGST1, which functions as antioxidants are so far only generically inferred. To be true, Drosophila flies bearing a disrupted MGST1-like gene have been generated, but the only phenotype observed is a reduction in life span. No further characterization has been carried out (Toba and Aigaki, 2000). As for MGST3 and the corresponding enzymes in plants and fungi, functions are so far uncharacterized. We are currently growing Arabidopsis plants carrying a TDNA insertion into the MGST3-like gene promoter, and hopefully future analysis of these mutants will give some clues on the function of this microsomal GST class. 6. Evolution 6.1. Soluble GSTs The first accounts of the evolution of this GST subfamily were based mainly on sequence comparisons. Following this criterion, and considering their ubiquitous presence from aerobic bacteria to higher eukaryotes, Theta GSTs were proposed as the ancient progenitors of the family, their evolution predating the prokaryote–eukaryote split, as a response to oxygen toxicity. Plant and mammalian specific GST classes would then have evolved from a Theta-class gene duplication that occurred before the divergence of fungi and animals (Pemble and Taylor, 1992). Subsequent description of the overall structure of GSTs highlighted a striking folding similarity not only among all soluble GST classes, but also with a number of other protein families, and prompted to consider GST evolution in a larger context, as part of the evolution of the thioredoxin fold (TRX) superfamily. An increasing number of

proteins have now been detected that share this common fold. Some of these protein families share also functional similarities. For instance thioredoxins, glutaredoxins and DsbA are all redox proteins catalyzing thiol-disulphide exchange reactions, while glutathione peroxidases and peroxiredoxins are more specialized in the reduction of peroxides (Martin, 1995; Copley et al., 2004). Other seems to have acquired very specialized functions, such as calsequestrins (binding of calcium ions), CLICs (formation of ion channels), DHARs (reduction of dehydroascorbate), eEF1Bg (elongating factors in protein synthesis). GSTs are unique, in that their functions are multiple and quite diversified, but share a number of functional links with other members of the TRX family. It is now clear that tracing the evolution of the TRX superfamily by looking at sequence, function or structure alone is limitative and may lead to ambiguous results, and that it is necessary to integrate sequence, structure, functional and active site residues information. A summary of the main characteristics of TRX proteins that may help in the reconstruction of their evolutionary history is reported in Table 3. Soluble GST evolution can be broadly divided into two phases: phase 1 includes the more ancient events and is now relatively well understood. Phase 2 encompasses the appearance and diversification of the multiplicity of GST classes and is less clear. Herebelow they are briefly described. 6.1.1. Phase 1 There is a wide consensus that thioredoxins/glutaredoxins are the ancestors from which all soluble GSTs, i.e. the cytosolic and mitochondrial (Kappa) subfamilies, as well as other TRX proteins originated. The basic event of phase 1 is the recruitment of a second, all helical domain (domain II) either as C-terminal addition to (E. coli GRX2, CLICs, eEF1Bg, cytosolic GSTs) or as insertion into the thioredoxin fold (DsbA, HCCA, Kappa GSTs). Glutathione peroxidases and peroxiredoxins (PRX) also have an insertion, although not entirely helical, at the same position as DsbA, HCCA and Kappa GSTs (Fig. 1c). On the basis of their structural features, it has been proposed (Ladner et al., 2004; Robinson et al., 2004) that cytosolic and Kappa GSTs evolved independently from thioredoxin/glutaredoxin proteins via two parallel pathways, A and B (Fig. 4a). This model substitutes two previous ones (Armstrong, 1997; Sheehan et al., 2001), in which Kappa GSTs were placed on the same evolutionary pathway as Theta enzymes, likely preceding them. In Fig. 4a, phases 1 and 2 are shown at the left and the right side of a hypothetical border line. All left side enzymes have thiol transferase activity and feature a cysteine as essential active site residue. By contrast, cytosolic and Kappa GSTs (on the right side) have a serine as active site residue, do not form mixed disulphides with glutathione and act as conjugating rather than redox enzymes. E. coli GRX2 and bacterial DsbA are the most likely intermediates towards GSTs in the two pathways. A variant scheme, indicated by a dotted arrow, proposes that DsbA derives from E. coli GRX2 through a repositioning of the helical domain from the C-terminus to the internal position of the TRX fold (Ladner et al., 2004).

C. Frova / Biomolecular Engineering 23 (2006) 149–169

161

Fig. 4. Model of TRX proteins evolution based on structural and functional data. The most ancient steps are illustrated in (a). (b) Phase 2 differentiation of cytosolic GSTs. Thick arrows in (b) indicate the likely sequence of critical evolutionary steps, dimerization and changes in active site residues (see text for details), these last represented by geometric symbols. Circle = Cys, diamond = Ser and triangle = Tyr.

To complicate this scenario, two yeast glutaredoxins with no extra a-helical domain, ScGRX1 and ScGRX2, have been reported as having a remarkable glutathione transferase activity with CDNB, and peroxidase activity (Collison and Grant, 2003). These two enzymes present significant sequence homology with Omega GSTs, with a relevant number of conserved residues in the N-terminal part of the proteins, including the Cys-Pro active site residues. Thus a third pathway resulting in the evolution of GST functions cannot be excluded. 6.1.2. Phase 2 In pathway A, CLIC, plant DHAR and Lambda GSTs conserve the monomeric structure of E. coli GRX2. Then a dimerization step occurred, as Omega and Beta GSTs are found as dimers. Furthermore, all these families/classes maintain the GRX2 cysteine as active site residue and show none or negligible conjugating activity. With the exception of CLIC, that have specialized in the formation of ion channels, they share thiol transferase and/or dehydroascorbate reductase activity. Thus they form a functionally and structurally distinct group from the canonical conjugating GSTs and can be reasonably predicted to have evolved earlier. The exact order of appearance is not clear yet. However, considering that Beta class enzymes act equally well as thiol-disulphide oxidoreductase and as glutathione transferase (Caccuri et al., 2002), their evolution should be posterior to Omega GSTs, but precede

the second crucial event in GST evolution, namely the shift from cysteine to serine chemistry. What happened next? Among the ‘‘serine’’ GSTs, Theta, Zeta GSTs and eEF1Bg are present in all eukaryotes, including the lower ones such as fungi and nematodes. Thus this group seems to have evolved anciently and the Theta class, represented also in bacteria, is the most likely progenitor. Despite their functional diversification Zeta GSTs and eEF1Bg are similar to Theta GSTs in sequence and biochemical properties, as both have conjugating activity with CDNB. Alltogether these data suggest that Zeta GSTs and eEF1Bg diverged early in the eukaryotic lineage although their relative position in the pathway is difficult to establish. Several lines of evidence indicate that the plant Phi and Tau and the insect Delta classes have evolved later. First they are specific of these phylogenetic groups. Secondly, the large number of the members of these classes, the frequent clustering of the genes in short genomic regions and the often high sequence similarity, are all indications of recent duplication events. A change in the critical G-site residue, from serine to tyrosine, in the mammalian Alpha, Mu, Pi classes and in Sigma GSTs, marks another evolutionary separation of the cytosolic GSTs. This separation is supported by phylogenetic trees based on sequence data (Board et al., 1997). The presence of Sigma GSTs in both vertebrates and invertebrates, suggests they

162

C. Frova / Biomolecular Engineering 23 (2006) 149–169

diverged before the mammalian Alpha/Mu/Pi group, evolving distinct functional properties in the diverse lineages. In cephalopods two functionally distinct Sigma proteins exist: a classical GST with high GSH binding activity, and S-crystallin, the major eye lens protein, which has no enzymatic activity (Tomarev et al., 1993). S-crystallin is though to have evolved from a duplicated Sigma GST through the insertion of 11 residues between the a4 and a5 helices that caused the loss of enzymatic activity (Sheehan et al., 2001). By contrast, in vertebrates and in multicellular parasites, Sigma GSTs have acquired specialized functions in prostaglandin synthesis (Sheehan et al., 2001). In pathway B, there is a gap between DsbA and Kappa GSTs. DsbA are bacterial monomeric enzymes with a cysteine chemistry, while Kappa GSTs are essentially eukaryotic dimeric enzymes that have adopted a serine chemistry, and their functions are quite different. Sequence comparisons suggest bacterial HCCA isomerase as a possible intermediate, but resolution of the 3D structure of this enzyme is required to confirm the hypothesis. Similarities between Kappa GSTs and the human Theta GSTs in the dimeric structure, in the TRX-like domain sequence (19% identity) (Li et al., 2005) and in the active site residue (a serine in both) could suggest a common evolutionary descent. However, the absence of any significant overall similarity in primary structure and the position of the all helical domain insertion are in contrast with this hypothesis. Thus it appears that dimerization and the shift from cysteine to serine chemistry occurred independently in the two pathways leading to the soluble GSTs. On the other hand, reutilization of the same schemes is not uncommon in nature’s strategy for enzyme evolution. A possible phase 2 evolutionary model of cytosolic GSTs is described in Fig. 4b. The picture as a whole suggests some general considerations. The ubiquitously distributed Theta, Zeta, Omega and Sigma classes are represented by a maximum of two to three members in each species, indicating that these genes have undergone few duplications or that the duplicated copies were subsequently lost. These genes/proteins often share a high degree of sequence similarity even over long evolutionary periods. For instance, Zeta GSTs display 49 and 38% aminoacid identity between human–C. elegans and human– carnation, respectively (Board et al., 1997). A monophiletic origin of Zeta and Theta GSTs, preceding the plant–animal split, has been suggested, and is consistent with the conservation of intron number and position, active site residue and the function of these enzymes in all eukaryotes (Frova, 2003; Dixon et al., 1998). The taxon specific GSTs are much more numerous. This indicates that whereas the family on the whole is ancient, the great expansion occurred more recently and independently in the diverse lineages. A phylogenetic analysis of plant GSTs showed that the two prevalent classes, the Phi and the Tau, have undergone extensive duplication and divergence after the mono–dicot split, approximately 200 My ago (Soranzo et al., 2004). The higher evolutionary rate of the specific GST classes likely reflects an adaptive response to environmental insults.

In fact the specific GST classes are those typically involved in detoxication of xenobiotics, being the principal phase II enzymes in animals and plants. The adaptive advantages of these systems depend on their versatility, i.e. the ability to detoxify a wide range of compounds, and on the rapidity by which new environmental challenges can be counteracted. During the course of evolution these needs have been met by two means: (1) by increasing the number of GSTs through multiple cycles of gene duplication, and (2) by favoring rapid diversification of the newly originated genes/enzymes with regard to their second-substrate specificity, while conserving the main chemistry allowing GSH conjugation with toxic electrophiles. In other words by combining a strictly conserved G-site with a highly variable H-site. Functional diversification via H-site variation has been observed for instance in Alpha GSTs and even between enzymes derived by recent duplication events and sharing extremely high aminoacid identity (Frova et al., 2004). That significant variations in substrate specificity and/or catalysis can be brought about by substitution of a very limited number of residues at the H-site, is further demonstrated by numerous examples of forced evolution (see below). In conclusion, the overall protein scaffold in GSTs and related families is characterized by a modular architecture in which variations in specificity determinants can arise without the disruption of GSH binding and catalytic activity. This combinatorial strategy, that allows a high degree of functional plasticity to be obtained in a fast and ‘‘economical’’ way, is recurrent in nature: in fact, structural conservation of shared functional domains combined with a multiplicity of diverse chemistries is a very common theme within protein superfamilies in which the different members have evolved to perform a wide array of functions (Gerlt and Babbitt, 1998; Nagano et al., 2002; Gerlt and Raushel, 2003). 6.2. Microsomal GSTs Very little is known about the evolution of microsomal GSTs. As described in a previous section, several classes and subclasses of MAPEG are now recognized (see Table 2), and their distribution in different phylogenetic groups has been determined (Bresell et al., 2005). Bacteria contain at least two potential ancestral classes, one of which is unique and unrelated to the eukaryotic enzymes. Class 3 is found in all eukaryotes, and in plants and fungi is the only one represented (C. Frova, unpublished results). Insects too possess just one class, similar but distinct from class 1 MAPEG. Tunicates and echinoderms have class 1 enzymes in addition to class 3. From fishes on, all classes and subclasses are represented (Table 4). Thus class 3 MAPEG could be the progenitors of this GST subfamily in eukaryotes. A first duplication possibly occurred in the animal lineage after the plant–animal split. While in insects a specific class evolved with the concomitant loss of the original enzyme, in tunicates and echinoderms both enzymes have been maintained. Further duplications and diversification along the vertebrate lineage likely brought to the present family complexity in fishes, amphibia, birds and mammals. A

C. Frova / Biomolecular Engineering 23 (2006) 149–169

163

Table 4 MAPEG classes distribution in major taxa Classes and subclasses

Fungi/algae

Plants

Insects

Tunicates/echinoderms

Fishes

Amphibia/birds

Mammals

+?

+ +

+ +

+ +

+ +

+ + +

+ + +

+ + +

+

+

+

1 MGST1 PGES 2 MGST2 LTC4S FLAP 3 MGST3

+

+

more detailed reconstruction of MAPEG evolution will be possible as structural and functional information on the nonmammalian enzymes, to date extremely limited, will become available. The only microsomal GSTs which appear distinct from MAPEG are two highly homologous enzymes from human and sheep liver. These two enzymes are very similar in sequence and structure to the Alpha class cytosolic GSTs (Prabhu et al., 2001, 2004; K.S. Prabhu and C.C. Reddy, personal communication), and their belonging to this class is fully confirmed by phylogenetic analyses conducted on a wide set of cytosolic and microsomal GSTs (C. Frova, unpublished results). However, the sheep liver enzyme, as the mammalian LTC4S, MGST2 and MGST3 MAPEG, has leukotriene C4 synthase activity. As the totally different tertiary and quaternary structure of mammalian cytosolic GSTs and MAPEG de facto excludes a common origin for these subfamilies, this appears to be a case of convergent functional evolution. 7. Biotechnological applications Due to their modular structure, GSTs represent a very adaptable platform for engineering enzymes with novel or enhanced catalytic activities through forced evolution. This is a first area of biotechnological research in GSTs, that is attracting wide interest, especially for the search of new biocatalysts and drugs. A second area is transgenesis. GST transgenic organisms may serve several purposes, among which (a) understanding gene functions in vivo (a few examples have been described in ‘‘functions’’), (b) setting up ‘‘cell factories’’ for the production of compounds for medical or technological application and (c) engineering organisms with new desirable characteristics. With regard to the last point plants and bacteria are so far the main targets. For plants, transformation with novel GSTs is a very promising and fast way to engineer tolerance to herbicides and pollutants in important crops or to generate varieties suitable for phytoremediation. Current research in bacteria is aimed at creating strains capable to degrade toxic/carcinogenic compounds. Although the two abovementioned areas are often not independent, since transgenics can be created utilizing genes previously manipulated by forced evolution, here they are described separately.

+

7.1. Forced evolution Two main different strategies can be adopted, commonly defined as rational and stochastic design. Rational design consists in introducing specific residue changes on the basis of detailed information on protein structure, functions and catalytic mechanisms. Knowledge of the spatial organization of the active site and of the roles of the different residues involved in catalysis is required for the choice of the target residues. The first striking success of GST engineering through this strategy has been the redesign of the human a-class enzyme, GSTA1-1. As already mentioned, GSTA1-1 is catalytically very active with CDNB, but shows low efficiency with alkenal substrates. By contrast, another Alpha class enzyme, GST A4-4, has natural high activity towards alkenals. The two enzymes share an overall 53% aminoacid identity. Careful comparison of their active sites indicated that four residues (Gly12, Ile107, Met108, Phe111) and the C-terminal helix of GSTA4-4 (residues 208–222) are critical for the high specificity of this enzyme for alkenals. The corresponding regions of GSTA1-1 were changed accordingly, and a new enzyme was created with high catalytic efficiency with alkenals and a shift in the catalyzed reaction from nucleophilic substitution (as in the original GSTA1-1) to Michael addition (as in natural GSTA4-4). The new enzyme alkenal activity is very specific, as indicated by a sharp drop (20-fold) in catalytic efficiency with CDNB (Nilsson et al., 2000). Many successful studies followed soon, some of which are mentioned here. An entirely new, non-natural protein catalyst for thiol-ester hydrolysis was obtained by incorporation of a single His residue in a ‘‘rationally’’ chosen position in human GSTA1-1 (Hederos et al., 2004). In another study the redesign of the Hsite, with the substitution of five residues, led to a dramatic functional conversion of the human GSTA2-2, naturally characterized by high peroxidase activity towards cumene hydroperoxide, into an efficient steroid isomerase (Pettersson et al., 2002). In plants, engineering of the maize GSTF1-1 by mutating selected G-site residues resulted in substantial changes in the pH-dependence of kinetic parameters of the enzyme (Labrou et al., 2004). Mutation of a key residue in the H-site of the same enzyme (Ile118Phe) led to a four-fold improved specificity of the enzyme towards the herbicide alachlor (Labrou et al., 2005). Finally, a very interesting output

164

C. Frova / Biomolecular Engineering 23 (2006) 149–169

of rational design protein engineering has been the creation of a GST capable to reduce hydrogen peroxide, a peroxidase activity typical of GPXs. This ability is dependent on the presence of a selenocysteine (Sec) residue at the active site of GPXs. Since Sec is encoded by a stop codon, attempts to introduce such residue in GSTs by traditional recombinant DNA techniques had met with little success. Yu et al. (2005) adopted a winning strategy to engineer the Delta GST from L. cuprina into a selenium containing enzyme. They first replaced the active site Ser9 with a cysteine, and then biosynthetically substituted it with a selenocysteine in a bacterial cysteine auxotrophic system. The new seleno-LcGST1-1 enzyme displayed a significantly high efficiency in the reduction of hydrogen peroxide by glutathione, comparable with that of natural GPXs. The requirement of structural knowledge in the rational design approach can be limitative. An alternative way for the choice of the target residues is an ‘‘evolutionary’’ approach, based on the concept that positive selection is an important force driving diversification of proteins during natural evolution. In practice, it consists in the screening of sequences of closely related paralogs, in search of positively selected residues, as estimated by the nonsynonymous/synonymous mutation rate. Such an analysis was conducted on 19 Mu class GSTs, some of which, as the human GSTM1-1 and GSTM2-2, share a high aminoacid identity but display major differences in their substrate selectivity. Among the 19 Mu GSTs considered, 6 polymorphic residues were identified to be under positive selection. Three, at positions 210, 104 and 130 (M1 numbering), distinguish GSTM1-1 and GSTM2-2 and were thus targeted for mutation to verify if and which could be responsible for the different substrate selectivities of the two enzymes. Interestingly the only changes markedly influencing catalytic activity were Ser210Thr and vice versa, an interconversion normally considered functionally conservative (Ivarsson et al., 2003). Such a result indicates that the ‘‘evolutionary’’ approach can identify residues for rational design targeting that would not otherwise be obvious. Although this approach does not need any a priori structural information about the active site or other functional regions, a minimum number of sequences of closely related proteins is required for its application. In that sense it is very promising for engineering the specific plant and insect GSTs that are numerous and derived from recent duplication events. Stochastic approaches basically reproduce in vitro natural evolution mechanisms, i.e. random mutagenesis and selection. The starting point is usually uniparental or multiparental DNA shuffling. By this way large recombinant libraries are created, which are then expressed and screened for enhanced or novel enzymatic activity. The system is very powerful because the DNA diversity that can be obtained is almost unlimited, but the screening process is laborious and time consuming, thus constituting a bottleneck. Despite this limitation several examples from the literature demonstrate the success of a stochastic approach in the development of enzymes with altered functions (Tobin et al., 2000 and refs. herein). For mammalian GSTs, enzymes with novel catalytic properties were obtained

for the Mu (Hansson et al., 1999) and the Theta classes (Broo et al., 2002). This last study, in particular, highlights the power of the method. A recombinant DNA library was created by shuffling sequences of the human GSTT1 and the rat GSTT2 genes. Functional characterization of less then 100 randomly picked recombinant clones led to the identification of over 40 enzyme variants with activity profiles towards 6 different electrophilic substrates that differ substantially from those of the parental human GSTT1-1 and rat GSTT2-2 enzymes. These first results were then further improved. As in nature, forced evolution optimally proceeds through recurrent cycles of mutagenesis and isolation of improved mutants. However, the choice of the best mutants from one cycle to initiate the following one is not always obvious. Larsson et al. (2004) showed how the use of statistical tools can aid in the rational sampling of the mutants to be used to parent the next generation. By applying multivariate cluster analysis to the variants identified in the hT1/rT2 library mentioned above they identified the best five clones to generate a second generation library, from which a new human Theta enzyme with 65-fold enhanced alkyltransferase activity was eventually isolated. This is just an example of a mixed forced evolution strategy, where a basically stochastic design is combined with a rational step. A number of such combinations have been used. In plants, forced evolution aimed at obtaining enzymes with improved herbicide detoxification activity involved two maize Tau GSTs, ZmGSTU1-1 and ZmGSTU2-2. A first step of random fragmentation and recombination of ZmGSTU1 and ZmGSTU2 cDNAs was used to create a mutant library. Five thousand clones of the library were screened for activity with the herbicide fluorodifen and seven mutants were identified that had over five-fold higher activity than the parental enzymes. Upon sequencing, all seven clones resulted to be chimeras of GSTU1 and GSTU2 fragments. The best performing one presented also a point mutation in the C-ter domain, with leucine substituting glutamine at position 115 (Q115L), and this mutation resulted responsible of the increased activity. Further ‘‘rational’’ mutation of this residue to alanine led to additional activity improvement. Overall the combination of DNA shuffling and directed mutagenesis of the key residue 115 yielded an enzyme with 29-fold increase in detoxifying activity toward the herbicide fluorodifen relative to the most active parent enzyme ZmGSTU2-2 (Dixon et al., 2003). 7.2. Transgenics The first plant GST transgenics were tobacco plants overexpressing an endogenous Tau class GST (Nt107) with peroxidase activity. Overexpression of Nt107 resulted in higher GST and GPX activity, and the transgenic plants were more tolerant to high and low temperature stress, salt stress and exposure to herbicides (Roxas et al., 1997, 2000). However, transgenic cotton plants expressing Nt107, while exhibiting increased GST/GPX activity, did not show increased tolerance to oxi-stress factors such as low temperature, salinity or the herbicides atrazine and imazethapyr (Light et al., 2005). The authors suggest that the insertion of the foreign GST gene might

C. Frova / Biomolecular Engineering 23 (2006) 149–169

disrupt the endogenous cotton stress adaptation system, and thus fail to provide protection towards stress-induced oxidative damage (Light et al., 2005). Whatever the causes, these results indicate that caution must be exerted when designing transgenic organisms, and that endogenous factors must be taken into account. One of the most promising fields in plant GST transgenesis is engineering the ability to detoxify or sequester xenobiotics such as herbicides or pollutants. In some crop species, i.e. maize and soybean, GSTs have key roles in determining the metabolism and selectivity, i.e. the ability to control weeds without harming the crop, of various classes of herbicides. However for many other crops the use of herbicides is severely restricted, because the crops lack the full GST complement necessary for detoxification. For instance, wheat and tobacco are relatively susceptible to the widely used chloroacetanilide and thiocarbamate herbicides. Maize GSTIV, a Phi class GST very active against the chloroacetanilide alachlor, was expressed in tobacco. The transgenic plants showed a significant increase in tolerance to chloroacetanilide as well as thiocarbamate herbicides (Jepson et al., 1997). ZmGSTIV was also used to transform wheat. Again transgenic plants exhibited a marked tolerance increase against two chloroacetanilides and one thiocarbamate, the level of tolerance in different T2 homozygotes being correlated with the level of expression of the transgene (Milligan et al., 2001). While these studies exploited the natural detoxifying capability of a maize Phi GST, even more promising results were obtained by transforming A. thaliana with the engineered form of the maize Tau GST reported above (Dixon et al., 2003). Tau GSTs protect plants from photobleaching caused by diphenylether herbicides such as fluorodifen. Arabidopsis is quite susceptible to fluorodifen, probably because the expression level of Tau GSTs in this species is modest. Transgenic A. thaliana plants expressing the engineered maize enzyme were markedly more resistant to fluorodifen than untransformed controls or plants transformed with the original maize enzymes (Dixon et al., 2003). Bioremediation (phytoremediation if plants are involved rather than bacteria) is a fast growing new technology that uses bacteria or plants to remove or degrade pollutants from soil and water. Contaminants that can be eliminated by exploiting the GSH–GST detoxification system include toxic metals, radioactive elements, herbicidal residues and a variety of organic toxic compounds. For some of these compounds, suitable bacterial or plant species either do not exist or display low accumulation/degradation efficiency. Their remediation capacity, however, can be significantly increased by genetic manipulation. Chlorinated ethenes (CE) are highly toxic and potentially carcinogenic compounds and are among the most prevalent groundwater contaminants. Remediation of CE by bacteria may be difficult because the first step of their detoxication, oxidation by monooxigenases, generates epoxide intermediates that are toxic to the host bacteria. Since epoxiethanes are electrophilic, their conjugation to GSH through GSTs could be a valid mechanism for their biological detoxification. Based on this

165

reasoning Rui et al. (2004) constructed an E. coli strain in which TOM-green, a DNA shuffling variant of toluene orthomonooxigenase from Burkholderia cepacia with enhanced degradation rates for cis-dichloroacetylene (cis-DCE), a novel GST from Rhodococcus and a mutant g-glutamylcysteine synthetase gene from E. coli to increase GSH levels, were coexpressed. By this way a recombinant bacterial strain was obtained that expressed eight genes simultaneously creating an engineered pathway that efficiently degrades cis-DCE and other chlorinated ethenes without harm to the host microorganism. This rational combination of catabolic segments from different organisms is a perfect example of the power of metabolic engineering in bioremediation of environmental pollutants. When soils rather than water need to be decontaminated, the use of bacteria might be inadequate because the detoxification process is too slow or too superficial. In such cases high biomass plants with an extended root system have been suggested as a suitable tool for a fast and efficient removal of heavy metals or toxic herbicidal residues from contaminated soils. For instance, transgenic poplar plants were created to overexpress a mercuric reductase gene originally isolated from mercury-resistant bacteria (Rugh et al., 1998). Mercuric reductase converts HgII, the highly toxic ionic mercury, to much less toxic elemental mercury, Hg(O). However, bioremediation of large mercury contaminated sites through bacteria is inefficient because bacterial Hg(O) release is far too slow (Summers and Lewis, 1973). By contrast, transformed poplar plants grown on media containing high levels of ionic mercury were able to convert it into elemental mercury, and to release appreciable amounts of this non-toxic form into the air (Rugh et al., 1998). Tobacco transformed with the maize GSTF1 gene showed substantially higher tolerance to alachlor, compared to untransformed plants. These transgenic plants are potentially useful for the phytoremediation of agricultural fields contaminated with herbicidal residues (Karavangeli et al., 2005). 8. Concluding remarks Our understanding of GSTs has greatly improved since the appearance of omics. Unexpected links with other protein families were highlighted by large-scale sequence and structural data, and have shed new light into the evolution of GSTs. The availability of whole genome sequences is beginning to reveal the consistency and distribution of microsomal GSTs, so far characterized only in mammals. The power of forced evolution for engineering enzymes with novel properties and/or specificities has been demonstrated by numerous recent studies. Transgenics, in particular plants and bacteria with great ecological potential in bioremediation and sustainable agriculture, are beginning to be successfully created. Structural genomics is maintaining the promise to reveal the molecular function of protein domains. Novel exciting progresses are around the corner. In the field of functional genomics, for instance, the latest developments in protein bioinformatics are expected to open significant

166

C. Frova / Biomolecular Engineering 23 (2006) 149–169

breakthroughs in the definition of GST expression maps in metabolic networks. Programs are being developed that exploit comprehensive protein expression databases (such as for instance those from humans and model plants) that will likely identify metabolic pathways in which the different GSTs are involved. Acknowledgements I wish to thank K.S. Prabhu and C.C. Reddy for providing the 3D structure of sheep liver microsomal GST (SLM GSTA11) and the sequence of the human liver microsomal GST (MGSTA) prior to public release, and for useful discussion about microsomal GSTs evolution. Research in my laboratory was supported by grants of the Italian Ministry of University and Research (COFIN programs). References Adler, V., Yin, Z., Fuchs, S.Y., Benezra, M., Rosario, L., Tew, K.D., Pinkus, M.R., Sardana, M., Henderson, C.J., Wolf, C.R., Ronai, Z., 1999. Regulation of JNK signaling by GSTp. EMBO J. 18, 1321–1334. Allocati, N., Casalone, E., Masulli, M., Ceccarelli, I., Carletti, E., Parker, M.W., Di Ilio, C., 1999. Functional analysis of the evolutionarily conserved proline 53 residue in Proteus mirabilis glutathione S-transferase B1-1. FEBS Lett. 445, 347–350. Allocati, N., Favaloro, B., Masulli, M., Alexeyev, M.F., Di Ilio, C., 2003. Proteus mirabilis glutathione S-transferase B1-1 is involved in protective mechanisms against oxidative and chemical stress. Biochem. J. 373, 305– 311. Andersson, C.E., Mosialou, E., Weinander, R., Morgenstern, R., 1994. Enzymology of microsomal glutathione S-transferase. Adv. Pharmacol. 27, 19– 35. Armstrong, R.N., 1997. Structure, catalytic mechanism, and evolution of the glutathione transferases. Chem. Res. Toxicol. 10, 2–18. Armstrong, R.N., 1998. Mechanistic imperatives for the evolution of glutathione transferases. Curr. Opin. Chem. Biol. 2, 618–623. Axarli, I.A., Ridgen, D.J., Labrou, N.E., 2004. Characterization of the ligandin site of maize glutathione S-transferase I. Biochem. J. 382, 885–893. Bartling, D., Radzio, R., Steiner, U., Weiler, E.W., 1993. A glutathione Stransferase with glutathione-peroxidase activity from Arabidopsis thaliana. Eur. J. Biochem. 216, 579–586. Bijo¨rnestedt, R., Tardioli, S., Mannervik, B., 1995. The high activity of rat glutathione transferase 8-8 with alkene substrates is dependent on a glycine residue in the active site. J. Biol. Chem. 270, 29705–29709. Board, P.G., Baker, R.T., Chelvanayagam, G., Jermiin, L.S., 1997. Zeta, a novel class of glutathione transferases in a large range of species from plants to humans. Biochem. J. 328, 929–935. Board, P.G., Coggan, M., Chelvanayagam, G., Easteal, S., Jermiin, L.S., Schulte, G.K., Danley, D.E., Hoth, L.R., Griffor, M.C., Kamath, A.V., Rosner, M.H., Chrunyk, B.A., Perregaux, D.E., Gabel, C.A., Geoghegan, K.F., Pandit, J., 2000. Identification, characterization, and crystal structure of the Omega class glutathione transferases. J. Biol. Chem. 275, 24798– 24806. Bresell, A., Weinander, R., Lundqvist, G., Raza, H., Shimoji, M., Sun, T.-H., Balk, L., Wiklund, R., Eriksson, J., Jansson, C., Persson, B., Jakobsson, P.J., Morgenstern, R., 2005. Bioinformatic and enzymatic characterization of the MAPEG superfamily. FEBS J. 272, 1688–1703. Broo, K., Larsson, A.-K., Jemth, P., Mannervik, P., 2002. An ensemble of Theta class glutathione transferases with novel catalytic properties generated by stochastic recombination of fragments of two mammalian enzymes. J. Mol. Biol. 318, 59–70. Burns, C.M., Hubatsch, I., Ridderstro¨m, M., Mannervik, B., Tainer, J.A., 1999. Human glutathione transferase A4-4 crystal structures and mutagenesis

reveal the basis of high catalytic efficiency with toxic lipid peroxidation products. J. Mol. Biol. 288, 427–439. Busenlehner, L.S., Codreanu, S.G., Holm, P.J., Bhakat, P., Hebert, H., Morgenstern, R., Armstrong, R.N., 2004. Stress sensor triggers conformational response of the integral membrane protein microsomal glutathione transferase 1. Biochemistry 43, 11145–11152. Byrum, R.S., Goulet, J.L., Griffith, R.J., Koller, B.H., 1997. Role of the 5lipoxygenase-activating protein (FLAP) in murine acute inflammatory responses. J. Exp. Med. 185, 1065–1075. Caccuri, A.M., Antonini, G., Allocati, N., Di Ilio, C., De Maria, F., Innocenti, F., Parker, M.W., Fasulli, M., Lo Bello, M., Tirella, P., Federici, G., Ricci, G., 2002. GSTB1-1 from Proteus mirabilis. A snapshot of an enzyme in the evolutionary pathway from a redox enzyme to a conjugating enzyme. J. Biol. Chem. 277, 16777–16784. Caccuri, A.M., Antonini, G., Ascenzi, P., Nicotra, M., Nuccetelli, M., Mazzetti, A.P., Federici, G., Lo Bello, M., Ricci, G., 1999. Temperature adaptation of glutathione S-transferase P1-1. A case of homotropic regulation of substrate binding. J. Biol. Chem. 274, 19276–19280. Caccuri, A.M., Antonini, G., Nicotra, M., Battistoni, A., Lo Bello, M., Board, P.G., Parker, M.W., Ricci, G., 1997. Catalytic mechanism and role of hydroxyl residues in the active site of theta class glutathione S-transferases. Investigation of Ser-9 and Tyr-113 in a glutathione S-transferase from the Australian sheep bowfly, Lucilia cuprina. J. Biol. Chem. 272, 29681–29686. Chelvanayagam, G., Parker, M.W., Board, P.G., 2001. Fly fishing for GSTs: a unified nomenclature for mammalian and insect glutathione transferases. Chem. Biol. Interact. 133, 256–260. Cho, S.G., Lee, Y.H., Park, H.S., Ryoo, K., Kank, K.W., Park, J., Eom, S.J., Kim, M.J., Chang, T.S., Choi, S.Y., Shim, J., Kim, Y., Dong, M.J., Kim, S.G., Ichijo, H., Choi, F.J., 2001. Glutathione S-transferase mu modulates the stress-activated signals by suppressing apoptosis signal-regulating kinase 1. J. Biol. Chem. 276, 12749–12755. Coleman, J.O.D., Blake-Kalff, M.M.A., Davies, T.G.E., 1997. Detoxification of xenobiotics by plants: chemical modification and vacuolar compartmentation. Trends Plant Sci. 2, 144–151. Collison, E.J., Grant, C.M., 2003. Role of yeast glutaredoxins as glutathione Stransferases. J. Biol. Chem. 278, 22492–22497. Copley, S.D., Novak, W.R.P., Babbitt, P.C., 2004. Divergence of function in the thioredoxin fold superfamily: evidence for evolution of peroxiredoxins from a thioredoxin-like ancestor. Biochemistry 43, 13981–13995. Cromer, B.A., Morton, C.J., Board, P.G., Parker, M.W., 2002. From glutathione transferase to pore in a CLIC. Eur. Biophys. J. 31, 356–364. Dirr, H., Reinemer, P., Huber, R., 1994a. X-ray crystal structures of cytosolic glutathione S-transferases. Implications for protein architecture, substrate recognition and catalytic function. Eur. J. Biochem. 220, 645–661. Dirr, H., Reinemer, P., Huber, R., 1994b. Refined crystal structure of porcine ˚ resolution. J. Mol. class Pi glutathione S-transferase (pGST P1-1) at 2.1 A Biol. 243, 72–92. Dixon, D.P., Cummins, I., Cole, D.J., Edwards, R., 1998. Glutathione mediated detoxification systems in plants. Curr. Opin. Plant Biol. 1, 258–266. Dixon, D.P., Davies, B.G., Edwards, E., 2002a. Functional divergence in the glutathione transferase superfamily in plants. J. Biol. Chem. 277, 30859– 30869. Dixon, D.P., Lapthorn, A., Edwards, R., 2002b. Plant glutathione transferases. Genome Biol. 3, 3004.1–3004.10. Dixon, D.P., McEwen, A.G., Lapthorn, A.J., Edwards, R., 2003. Forced evolution of a herbicide detoxifying glutathione trasnsferase. J. Biol. Chem. 278, 23930–23935. Dulhunty, A., Gage, P., Curtis, S., Chelvanayagam, G., Board, P., 2001. The glutathione transferase structural family includes a nuclear chloride channel and a ryanodine receptor calcium release channel modulator. J. Biol. Chem. 276, 3319–3323. Edwards, R., Dixon, D.P., Walbot, V., 2000. Plant glutathione S-transferases: enzymes with multiple functions in sickness and in health. Trends Plant Sci. 5, 193–198. Engle, M.R., Singh, S.P., Czernik, P.J., Gaddy, D., Montagne, D.C., Ceci, J.D., Yang, Y., Awasthi, Y.C., Zimniak, P., 2004. Physiological role of mGSTA4-4, a glutahione S-transferase metabolizing 4-hydroxynonenal: generation and analysid of mGSTA4 null mouse. Toxicol. Appl. Pharmacol. 194, 296–308.

C. Frova / Biomolecular Engineering 23 (2006) 149–169 Epp, O., Ladenstein, R., Wendel, A., 1983. The refined structure of the selenoenzyme glutathione peroxidase at 0.2 nm resolution. Eur. J. Biochem. 133, 51–69. Erhardt, J., Dirr, H.W., 1995. Native dimer stabilizes the subunit tertiary structure of porcine class Pi glutathione S-transferase. Eur. J. Biochem. 230, 614–620. Fernandez-Can˜on, J.M., Pen˜alva, M.A., 1998. Characterisation of a fungal maleylacetoacetate isomerase genes and identification of its human homologue. J. Biol. Chem. 273, 328–337. Fernandez-Can˜on, J.M., Baetscher, M.W., Finegold, M., Burlingame, T., Gibson, K.M., Grompe, M., 2002. Maleylacetoacetate isomerase (MAAI/ GSTZ)-deficient mice reveal a glutathione-dependent nonenzymatic bypass in tyrosine catabolism. Mol. Cell. Biol. 22, 4943–4951. Frova, C., 2003. The plant glutathione transferase gene family: genomic structure, functions, expression and evolution. Physiol. Plant. 119, 469– 479. Frova, C., De Toma, G., Mizzi, L., Sari Gorla, M., 2004. Plant GSTs: diversification and evolution. In: Schnarrenberger, C., Wittmann-Liebhold, B. (Eds.), Genes, Gene Families and Isozymes: Medimond International Proceedings, Italy, pp. 199–205. Gerlt, J.A., Babbitt, P.C., 1998. Mechanistically diverse enzyme superfamilies: the importance of chemistry in the evolution of catalysis. Curr. Opin. Chem. Biol. 2, 607–612. Gerlt, J.A., Raushel, F.M., 2003. Evolution of function in (b/a)8-barrel enzymes. Curr. Opin. Chem. Biol. 7, 252–264. Hansson, L.O., Bolton-Grob, R., Massoud, T., Mannervik, B., 1999. Evolution of different substrate specificities in Mu class glutathione transferase probed by DNA shuffling. J. Mol. Biol. 287, 265–276. Harris, J.M., Meyer, D.J., Coles, B., Ketterer, B., 1991. A novel glutathione transferase (13-13) isolated from the matrix of rat liver mitochondria having structural similarity to class theta enzymes. Biochem. J. 278, 137–141. Harrop, S.J., DeMaere, M.Z., Fairlie, W.D., Reztova, T., Valenzuela, S.M., Mazzanti, M., Tonini, R., Qiu, M.R., Jankova, L., Warton, K., Bauskin, A.R., Wu, W.M., Pankhurst, S., Campbell, T.J., Breit, S.N., Curmi, P.M.G., 2001. Crystal sructure of a soluble form of the intracellular chloride ion ˚ resolution. J. Biol. Chem. 276, 44993– channel CLIC1 (NCC27) at 1.4-A 45000. Hayes, J.D., Flanagan, J.U., Jowsey, I.R., 2005. Glutathione transferases. Annu. Rev. Pharmacol. Toxicol. 45, 51–88. Hayes, J.D., Pulford, D.J., 1995. The glutathione S-transferase supergene family. Regulation of GST and the contribution of the isoenzymes to cancer chemoprotection and drug resistance. CRC Crit. Rev. Biochem. Mol. Biol. 30, 445–600. Hederos, S., Broo, K.S., Jakobsson, E., Kleywegt, G., Mannervik, B., Baltzer, L., 2004. Incorporation of a single His residue by rational design enables thiol-ester hydrolysis by human glutathione transferase A1-1. Proc. Natl. Acad. Sci. U.S.A. 101, 13163–13167. Hegazy, U.M., Mannervik, B., Stenberg, G., 2004. Functional role of the lock and key motif at the subunit interface of glutathione transferase P1-1. J. Biol. Chem. 279, 9586–9596. Henderson, C.J., Smith, A.G., Ure, J., Brown, K., Bacon, E.J., Wolf, C.R., 1998. Increased skin tumorigenesis in mice lacking pi class glutathione S-transferases. Proc. Natl. Acad. Sci. U.S.A. 95, 5275–5280. Hornby, J.A., Codreanu, S.G., Armstrong, R.N., Dirr, H.W., 2002. Molecular recognition at the dimer interface of a class Mu glutathione transferase: role of a hydrophobic interaction motif in dimer stability and protein function. Biochemistry 41, 14238–14247. Hu, X., Benson, P.J., Srivastava, S.K., Xia, H., Bleicher, R.J., Zaren, H.A., Awasthi, S., Awasthi, Y.C., Singh, S.V., 1997. Induction of glutathione Stransferase p as a bioassay for the evaluation of potency inhibitors of benzo(a)pyrene-induced cancer in a murine model. Int. J. Cancer 73, 897– 902. Ivarsson, Y., Mackey, A.J., Edalat, M., Pearson, W.R., Mannervik, B., 2003. Identification of residues in glutathione transferase capable of driving functional diversification in evolution. J. Biol. Chem. 278, 8733–8738. Jakobsson, P.-J., Mancini, J.A., Ford-Hutchinson, A.W., 1996. Identification and characterization of a novel human microsomal glutathione S-transferase with leukotriene C4 synthase activity and significant sequence identity to 5-

167

lipoxygenase-activating protein and leukotriene C4 synthase. J. Biol. Chem. 271, 22203–22210. Jakobsson, P.-J., Mancini, J.A., Riendeau, D., Ford-Hutchinson, A.W., 1997. Identification and characterization of a novel microsomal enzyme with glutathione-dependent transferase and peroxidase activity. J. Biol. Chem. 272, 22934–22939. Jakobsson, P.-J., Morgenstern, R., Mancini, J., Ford-Hutchinson, A., Persson, B., 1999. Common structural features of MAPEG—a widespread superfamily of membrane associated proteins with highly divergent functions in eicosanoid and glutathione metabolism. Protein Sci. 8, 689–692. Jakobsson, P.-J., Morgenstern, R., Mancini, J., Ford-Hutchinson, A., Persson, B., 2000. Membrane-associated proteins in eicosanoid and glutathione metabolism (MAPEG). A widespread protein superfamily. Am. J. Respir. Crit. Care Med. 161, S20–S24. Jensson, H., Guthenberg, C., Alin, P., Mannervik, B., 1986. Rat glutathione transferase 8-8, an enzyme efficiently detoxifying 4-hydroxyalk-2-enals. FEBS Lett. 203, 207–209. Jeppesen, M.G., Ortiz, P., Shepard, W., Kinzy, T.G., Nyborg, J., Andersen, G.R., 2003. The crystal structure of the glutathione S-transferase-like domain of elongation factor 1Bg from Saccharomyces cerevisiae. J. Biol. Chem. 278, 47190–47198. Jepson, I., Holt, D.C., Roussel, V., Wright, S.Y., Greenland, A.J., 1997. Transgenic plant analysis as a tool for the study of maize glutathione Stransferases. In: Hatzios, K. (Ed.), Regulation of Enzymatic Systems Detoxifying Xenobiotics in Plants. Kluwer Academic Publishers, The Netherlands, pp. 313–323. Jowsey, R.I., Thomson, R.E., Orton, T.C., Elcombe, C.R., Hayes, J.D., 2003. Biochemical and genetic characterization of a murine class Kappa glutathione S-transferase. Biochem. J. 373, 559–569. Kanaoka, Y., Ago, H., Inagaki, E., Nanayama, T., Miyano, M., Kikuno, R., Fujii, Y., Eguchi, N., Toh, H., Urade, Y., Hayaishi, O., 1997. Cloning and crystal structure of hematopoietic prostglandin D synthase. Cell 90, 1085– 1095. Kanaoka, Y., Maekawa, A., Penrose, J.F., Austen, K.F., Lam, B.K., 2001. Attenuated zymosan-induced peritoneal vascular permeability and IgEdependent passive cutaneous anaphylaxis in mice lacking leukotriene C4 synthase. J. Biol. Chem. 276, 22608–22613. Karavangeli, M., Labrou, N.E., Clonis, Y.D., Tsaftaris, A., 2005. Development of transgenic tobacco plants overexpressing maize glutathione S-transferase I for chloroacetanilide herbicides phytoremediation. Biomol. Eng. 22, 121– 128. Labrou, N.E., Ridgen, D.J., Clonis, Y.D., 2004. Engineering the pH-dependence of kinetic parameters of maize glutathione S-transferase I by site-directed mutagenesis. Biomol. Eng. 21, 61–66. Labrou, N.E., Karavangeli, M., Tsaftaris, A., Clonis, Y.D., 2005. Kinetic analysis of maize glutathione S-transferase I catalyzing the detoxification from chloroacetanilide herbicides. Planta 222, 91–97. Ladner, J.E., Parson, J.F., Rife, C.L., Gilliland, G.L., Armstrong, R.N., 2004. Parallel evolutionary pathways for glutathione transferases: structure and mechanisms of the mitochondrial class Kappa enzyme rGSTK1-1. Biochemistry 43, 352–361. Lam, B.K., Penrose, J.F., Xu, K., Baldasaro, M.H., Austen, K.F., 1997. Sitedirected mutagenesis of human leukotriene C4 synthase. J. Biol. Chem. 272, 13923–13928. Larsson, A.-K., Emere´n, L.O., Bardsley, W.G., Mannervik, B., 2004. Directed enzyme evolution guided by multidimensional analysis of substrate-activity space. Protein Eng. Des. Select. 17, 49–55. Lengqvist, J., Svensson, R., Evergren, E., Morgenstern, R., Griffith, W., 2004. Observation of an intact noncovalent homotrimer of detergent-solubilized rat microsomal glutathione transferase-1 by electrospray mass spectrometry. J. Biol. Chem. 279, 13311–13316. Li, J., Xia, Z., Ding, J., 2005. Thioredoxin-like domain of human k class glutathione transferase reveals sequence homology and structure similarity to the u class enzyme. Protein Sci. 14, 2361–2369. Light, G.G., Mahan, J.R., Roxas, V.P., Allen, R.D., 2005. Transgenic cotton (Gossypium hirsutum L.) seedlings expressing a tobacco glutathione Stransferase fail to provide improved stress tolerance. Planta 222, 346– 354.

168

C. Frova / Biomolecular Engineering 23 (2006) 149–169

Lim, C.E.L., Matthaei, K.I., Blackburn, A.C., Davis, R.P., Dahlstrom, J.E., Koina, M.E., Anders, M.W., Board, P.G., 2004. Mice deficient in glutathione transferase Zeta/Maleylacetoacetate isomerase exhibit a range of pathological changes and elevated expression of Alpha, Mu, and Pi class glutathione transferases. Am. J. Pathol. 165, 679–693. Lin, X., Kaul, S., Rounsley, S., Shea, T.P., Benito, M.I., Town, C.D., Fujii, C.Y., Mason, T., Bowman, C.L., Barnstead, M., Feldblyum, T.V., Buell, C.R., Ketchum, K.A., Lee, J., Ronning, C.M., Koo, H.L., Moffat, K.S., Cronin, L.A., Shen, M., Pai, G., Van Aken, S., Umayam, L., Tallon, L.J., Gill, J.E., Adams, M.D., Carrera, A.J., Creasy, T.H., Goodman, H.M., Somerville, C.R., Copenhaver, G.P., Preuss, D., Nierman, W.C., White, O., Eisen, J.A., Salzberg, S.L., Fraser, C.M., Venter, J.C., 1999. Sequence and analysis of chromosome 2 of the plant Arabidopsis thaliana. Nature 402, 761–768. Listowsky, I., 1993. High capacity binding by glutathione S-transferases and glucocorticoid resistance. In: Tew, K.D., Pickett, C.B., Mantle, T.J., Mannervik, B., Hayes, J.D. (Eds.), Structure and Function of Glutathione Transferases. CRC Press, Boca Raton, FL, pp. 199–209. Loyall, L., Uchida, K., Brown, S., Furuya, M., Frohnmeyer, H., 2000. Glutathione and a UV-light induced glutathione S-transferase are involved in signalling to chalcone synthase in cell cultures. Plant Cell 12, 1939–1950. Mandal, A.K., Skoch, J., Bacskai, B.J., Hyman, B.T., Christmas, P., Miller, D., Yamin, T.-D., Xu, S., Wisniewski, D., Evans, J.F., Soberman, R.J., 2004. The membrane organization of leukotriene synthesis. Proc. Natl. Acad. Sci. U.S.A. 101, 6587–6592. Marrs, K.A., 1996. The functions and regulation of plant glutathione Stransferases. Annu. Rev. Plant Physiol. Plant Mol. Biol. 47, 127–158. Martin, J.L., 1995. Thioredoxin—a fold for all reasons. Structure 3, 245–250. Martin, J.L., Bardwell, J.C.A., Kuriyan, J., 1993. Crystal structure of the DsbA protein required for disulfide bond formation in vivo. Nature 365, 464–468. McGonigle, B., Keeler, S.J., Lau, S.-M.C., Koeppe, M.K., O’Keefe, D.P., 2000. A genomics approach to the comprehensive analysis of the glutathione Stransferase gene family in soybean and maize. Plant Physiol. 124, 1105– 1120. Milligan, A.S., Daly, A., Parry, M.A.J., Lazzeri, P.A., Jepson, I., 2001. The expression of a maize glutathione S-transferase gene in transgenic wheat confers tolerance, both in planta and in vitro. Mol. Breed. 7, 301–315. Morel, F., Rauch, C., Coles, B., Le Ferrac, E., Guillouzo, A., 2002. The human glutathione transferase alpha locus: genomic organisation of gene cluster and functional characterisation of the genetic polymorphysm in the hGSTA1 promoter. Pharmacogenetics 12, 277–286. Morel, F., Rauch, C., Petit, E., Piton, A., Theret, N., Coles, B., Guillouzo, A., 2004. Gene and protein characterization of the human glutathione Stransferase Kappa and evidence for peroxisomal localization. J. Biol. Chem. 279, 16246–16253. Morgenstern, R., DePierre, J., 1983. Microsomal glutathione transferase: purification in unactivated form and further characterization of the activation process, substrate specificity and aminoacid composition. Eur. J. Biochem. 134, 591–597. Morgenstern, R., Guthenberg, C., DePierre, J.W., 1982. Microsomal S-transferase purification, initial characterization and demonstration that it is not identical to the cytosolic glutathione S-transferases A, B and C. Eur. J Biochem. 128, 243–248. Mosialou, E., Piemonte, F., Andersson, C., Vos, R.I., van Bladeren, P.J., Morgenstern, R., 1995. Microsomal glutathione transferase: lipid-derived substrates and lipid dependence. Arch. Biochem. Biophys. 320, 210–216. Mueller, L.A., Godman, C.D., Silady, R.A., Walbot, V., 2000. AN9, a petunia glutathione S-transferase required for anthocyanin sequestration, is a flavonoid binding protein. Plant Physiol. 123, 1561–1570. Nagano, N., Orengo, C.A., Thornton, J.M., 2002. One fold with many functions: the evolutionary relationships between TIM barrel families based on their sequences, structures and functions. J. Mol. Biol. 321, 741–765. Nilsson, L.O., Gustafsson, A., Mannervik, B., 2000. Redesign of substrateselectivity determining modules of glutathione transferase A1-1 installs high catalytic efficiency with toxic alkenal products of lipid peroxidation. Proc. Natl. Acad. Sci. U.S.A. 97, 9408–9412. Pemble, S.E., Taylor, J.B., 1992. An evolutionary perspective on glutathione transferases inferred from class-Theta glutathione transferase cDNA sequences. Biochem. J. 287, 957–963.

Pemble, S.E., Wardle, A.F., Taylor, J.B., 1996. Glutathione S-transferase class Kappa characterization by the cloning of rat mitochondrial GST and identification of a human homologue. Biochem. J. 319, 749–754. Pettersson, P.L., Johansson, A.-S., Mannervik, B., 2002. Transmutation of human glutathione transferase A2-2 with peroxidase activity into an efficient steroid isomerase. J. Biol. Chem. 277, 30019–30022. Pflugmacher, S., Sandermann Jr., H., Schroeder, P., 2000. Taxonomic distribution of plant glutathione S-transferases acting on xenobiotics. Phytochemistry 54, 267–273. Polyak, K., Xia, Y., Zweier, J.L., Kinzler, K.W., Vogelstein, B., 1997. A model for p53-induced apoptosis. Nature 389, 300–305. Prabhu, K.S., Reddy, P.V., Gumpricht, E., Hildenbrandt, G.R., Scholz, R.W., Sordillo, L.M., Reddy, C.C., 2001. Microsomal glutathione S-transferase A1-1 with glutathione peroxidase activity from sheep liver: molecular cloning, expression and characterization. Biochem. J. 360, 345–354. Prabhu, K.S., Reddy, P.V., Jones, E.C., Liken, A.D., Reddy, C.C., 2004. Characterization of a class alpha glutathione-S-transferase with glutathione peroxidase activity in human liver microsomes. Arch. Biochem. Biophys. 424, 72–80. Ranson, H., Collin, F., Hemingway, J., 1998. The role of alternative mRNA splicing in generating heterogeneity within the Anopheles gambiae class I glutathione S-transferase family. Proc. Natl. Acad. Sci. U.S.A. 95, 14284– 14289. Raza, H., Robin, M.A., Fang, J.K., Avadhani, N.G., 2002. Multiple isoforms of mitochondrial glutathione S-transferases and their differential induction under oxidative stress. Biochem. J. 366, 45–55. Robinson, A., Huttley, G.A., Booth, H.S., Board, P.G., 2004. Modelling and bioinformatics studies of the human Kappa-class glutathione transferase predict a novel third glutathione transferase family with similarita` to prokaryotic 2-hydroxychromene-2-carboxylate isomerases. Biochem. J. 379, 541–552. Roxas, V.P., Smith Jr., R.K., Allen, E.R., Allen, R.D., 1997. Overexpression of glutathione S-transferase/glutathione peroxidase enhances the growth of transgenic tobacco seedlings during stress. Nat. Biotechnol. 15, 988–991. Roxas, V.P., Lodhi, S.A., Garrett, D.K., Mahan, J.R., Allen, R.D., 2000. Stress tolerance in transgenic tobacco seedlings that overexpress glutathione Stransferase/glutathione peroxidase. Plant Cell Physiol. 41, 1229–1234. Rugh, C.L., Senecoff, J.F., Meagher, R.B., Merkle, S.A., 1998. Development of transgenic yellow poplar for mercury phytoremediation. Nat. Biotechnol. 16, 925–928. Rui, L., Kwon, Y.M., Reardon, K.F., Wood, T.K., 2004. Metabolic pathway engineering to enhance aerobic degradation of chlorinated ethenes and to reduce their toxicity by cloning a novel glutathione S-transferase, an evolved toluene o-monooxigenase, and g-glutamylcysteine synthetase. Environ. Microbiol. 6, 491–500. Sappl, P.G., Onate-Sanchez, L., Singh, K.B., Millar, A.H., 2004. Proteomic analysis of glutathione S-transferases of Arabidopsis thaliana reveals differential salicilic acid-induced expression of the plant-specific phi and tau classes. Plant Mol. Biol. 54, 205–219. Satoh, K., Nagano, Y., Shimomura, C., Suzuki, N., Saeki, Y., Yokota, H., 2000. Expression of prostglandin E synthase mRNA is induced in b-amyloid treated rat astrocytes. Neurosci. Lett. 283, 221–223. Sayed, Y., Wallace, L.A., Dirr, H.W., 2000. The hydrophobic lock-and-key intersubunit motif of glutathione transferase A1-1: implications for catalysis, ligandin function and stability. FEBS Lett. 465, 169–172. Schmidt-Krey, I., Mitsuoka, K., Hirai, T., Murata, K., Cheng, Y., Fujiyoshi, Y., Morgenstern, R., Hebert, H., 2000. The three-dimensional map of micro˚ resolution. EMBO J. 19, 6311–6316. somal glutathione transferase 1 at 6 A Schmidt-Krey, I., Kanaoka, Y., Mills, D.J., Irikura, D., Haase, W., Lam, B.K., Austen, K.F., Kuhlbrandt, W., 2004. Human leukotriene C4 synthase at ˚ resolution in projection. Structure 12, 2009–2014. 4.5 A Sheehan, D., Meade, G., Foley, V.M., Dowd, C.A., 2001. Structure, function and evolution of glutathione transferases: implications for classification of nonmammalian members of an ancient enzyme superfamily. Biochem. J. 360, 1–16. Smith, A.P., DeRidder, B.P., Guo, W.-J., Seeley, E.H., Regnier, F.E., Goldsbrough, P.B., 2004. Proteomic analysis of Arabidopsis glutathione S-transferases from benoxacor- and copper-treated seedlings. J. Biol. Chem. 279, 26098–26104.

C. Frova / Biomolecular Engineering 23 (2006) 149–169 Smith, A.P., Nourizadeh, S.D., Peer, W.A., Xu, J., Bandyopadhyay, A., Murphy, A.S., Goldsbrough, P.B., 2003. Arabidopsis AtGSTF2 is regulated by ethylene and auxin, and encodes a glutathione S-transferase that interacts with flavonoids. Plant J. 36, 433–442. Smith, W.L., DeWitt, D.L., Garavito, R.M., 2000. Cyclooxygenases: structural, cellular and molecular biology. Annu. Rev. Biochem. 69, 145–182. Soranzo, N., Sari Gorla, M., Mizzi, L., De Toma, G., Frova, C., 2004. Organisation and structural evolution of the rice glutathione S-transferase gene family. Mol. Gen. Genomics 271, 511–521. Stenberg, G., Abdalla, A.-M., Mannervik, B., 2000. Tyrosine 50 at the subunit interface of dimer human glutathione transferase P1-1 is a structural key residue for modulating protein stability and catalytic function. BBRC 271, 59–63. Summers, A.O., Lewis, E., 1973. Volatilization ofmercuric chloride by mercury-resistant plasmid-bearing strains of Escherichia coli, Staphylococcus aureus, and Pseudomonas aeroginosa. J. Bacteriol. 113, 1070–1072. Sun, T.H., Morgenstern, R., 1997. Binding of glutathione and an inhibitor to microsomal glutathione transferase. Biochem. J. 326, 193–196. Svartz, J., Blomgran, R., Hammarstrom, S., Soderstrom, M., 2003. Leukotriene C4 synthase homo-oligomers detected in living cells by bioluminescence resonance energy transfer. BBA 1633, 90–95. Tan, K.L., Chelvanayagam, G., Parker, M.W., Board, P.G., 1996. Mutagenesis of the active site of the human Theta-class glutathione transferase GSTT2-2: catalysis with different substrates involves different residues. Biochem. J. 319, 315–321. Thom, R., Dixon, D.P., Edwards, R., Cole, D.J., Lapthorn, A.J., 2001. The structure of a Zeta class glutathione S-transferase from Arabidopsis thaliana: characterisation of a GST with novel active-site architecture and a putative role in tyrosine catabolism. J. Mol. Biol. 308, 949–962. Thom, R., Cummins, I., Dixon, D.P., Edwards, R., Cole, D.J., Lapthorn, A.J., 2002. Structure of a Tau class glutathione S-transferase from wheat active in herbicide detoxification. Biochemistry 41, 7008–7020. Thore´n, S., Weinander, R., Saha, S., Jegerschold, C., Pettersson, P.L., Samuelsson, B., Hebert, H., Hamberg, M., Morgenstern, R., Jakobsson, P.-J., 2003. Human microsomal prostglandin E synthase-1. J. Biol. Chem. 278, 22199–22209. Toba, G., Aigaki, T., 2000. Disruption of the microsomal glutathione Stransferase-like gene reduces life span of Drosophila melanogaster. Gene 253, 179–187.

169

Tobin, M.B., Gustafsson, C., Huisman, G.W., 2000. Directed evolution: the ‘‘rational’’ basis for ‘‘irrational’’ design. Curr. Opin. Struct. Biol. 10, 421– 427. Tomarev, S.I., Zinovieva, R.D., Guo, K., Piatigorsky, J., 1993. Squid glutathione S-transferase. J. Biol. Chem. 268, 4534–4542. Toung, Y.P., Hsieh, T.S., Tu, C.P., 1993. The glutathione S-transferase D genes. A divergently organized, intronless gene family in Drosophila melanogaster. J. Biol. Chem. 268, 9737–9746. Trebino, C.E., Stock, J.L., Gibbons, C.P., Naiman, B.M., Wachtmann, T.S., Umland, J.P., Pandher, K., Lapointe, J.-M., Saha, S., Roach, M.L., Carter, D., Thomas, N.A., Durtschi, B.A., McNeish, J.D., Hambor, J.E., Jakobsson, P.-J., Carty, T.J., Perez, J.R., Audoly, L.P., 2003. Impaired inflammatory and pain responses in mice lacking an inducible prostglandin E synthase. Proc. Natl. Acad. Sci. U.S.A. 100, 9044–9049. Tsuchida, S., Izumi, T., Shimizu, T., Ishikawa, T., Hatayama, I., Satoh, K., Sato, K., 1987. Purification of a new acidic glutathione S-transferase, GSTYn1Yn1, with a high leukotriene-C4 synthase activity from the rat brain. Eur. J. Biochem. 170, 159–164. Urade, Y., Eguchi, N., Aritake, K., Hayaishi, O., 2004. Functional analyses of lipocalin-type and hematopoietic prostaglandin synthases. Folia Pharmacol. Jpn. 123, 5–13. Urano, J., Nakagawa, T., Maki, Y., Masumara, T., Tanaka, K., Murata, N., Ushimaru, T., 2000. Molecular cloning and characterization of a rice dehydroascorbate reductase. FEBS Lett. 466, 107–111. Vuilleumier, S., 1997. Bacterial glutathione S-transferases: what are they good for? J. Bacteriol. 179, 1431–1441. Vuilleumier, S., Pagni, M., 2002. The elusive roles of bacterial glutathione Stransferases: new lessons from genomes. Appl. Microbiol. Biotechnol. 58, 138–146. Wagner, U., Edwards, R., Dixon, D.P., Mauch, F., 2002. Probing the diversity of the Arabidopsis glutathione S-transferase gene family. Plant Mol. Biol. 49, 515–532. Xia, B., Vlamis-Gardikas, A., Holmgren, A., Wright, P.E., Dyson, H.J., 2001. Solution structure of Escherichia coli glutaredoxin-2 shows similarity to mammalian glutathione S-transferases. J. Mol. Biol. 310, 907–918. Yu, H.-J., Liu, J.-Q., Bock, A., Li, J., Luo, G.-M., Shen, J.-C., 2005. Engineering glutathione transferase to a novel glutathione peroxidase mimic with high catalytic efficiency. J. Biol. Chem. 280, 11930–11935.