MINIREVIEW Evolutionary changes to transthyretin: structure and function of a transthyretin-like ancestral protein Sarah C. Hennebry Department of Biochemistry and Molecular Biology, Bio21 Institute, The University of Melbourne, Victoria, Australia Introduction The evolution of the structure and the function of the thyroid hormone (TH) distributor, transthyretin, has been well researched. The primary, secondary, tertiary and quaternary structures of this vertebrate protein are highly conserved. It was therefore hypothesized that the transthyretin gene may have evolved in a nonverte- brate organism. Searches for a transthyretin progenitor led to the identification of a transthyretin homolog, which was found initially in nonvertebrate genomes and subsequently in all major kingdoms. The evolution of the structure and function of the transthyretin homolog [referred to as transthyretin-like protein (TLP)] has been the focus of recent studies by several research groups. TLPs from various organisms have been demonstrated to share remarkable structural similarities to vertebrate transthyretins. Despite this Keywords evolution; purines; structure; transthyretin; transthyretin-like protein Correspondence S. C. Hennebry, Human Neurotransmitters Laboratory, Baker IDI Heart and Diabetes Institute, P.O. Box 6492, St Kilda Road Central Melbourne, Victoria 3008, Australia Fax: +61 3 8532 1100 Tel: +61 3 8532 1734 E-mail: sarah.hennebry@bakeridi.edu.au (Received 2 February 2009, revised 8 June 2009, accepted 8 July 2009) doi:10.1111/j.1742-4658.2009.07246.x The structure of the thyroid hormone distributor protein, transthyretin, has been highly conserved during the evolution of vertebrates. Over the last decade, studies into the evolution of transthyretin have revealed the exis- tence of a transthyretin homolog, transthyretin-like protein, in all king- doms. Phylogenetic studies have suggested that the transthyretin gene in fact arose as a result of a duplication of the transthyretin-like protein gene in early protochordate evolution. Structural studies of transthyretin-like proteins from various organisms have revealed the remarkable conservation of the transthyretin-like protein ⁄ transthyretin fold. The only significant differences between the structures of transthyretin-like protein and transthyretin were localized to the dimer–dimer interface and indicated that thyroid hormones could not be bound by transthyretin-like protein. All transthyretin-like proteins studied to date have been demonstrated to function in purine metabolism by hydrolysing the oxidative product of uric acid, 5-hydroxyisourate. The residues characterizing the catalytic site in transthyretin-like proteins are 100% conserved in all transthyretin-like protein sequences but are absent in transthyretins. Therefore, it was proposed that following duplication of the transthyretin-like protein gene, loss of these catalytic residues resulted in the formation of a deep, negatively charged channel that runs through the centre of the transthy- retin tetramer. The results thus demonstrate the remarkable evolution of the transthyretin-like protein ⁄ transthyretin protein from a hydrolytic enzyme to a thyroid hormone distributor protein. Abbreviations 5-HIU, 5-hydroxyisourate; COG, cluster of orthologous groups; OHCU, hydroxy-4-carboxy-5-ureidoimidazoline; PTS2, type-two peroxisomal sequence; RNAi, RNA interference; TH, thyroid hormone; TLP, transthyretin-like protein. FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS 5367 structural similarity, TLP and transthyretin have dif- ferent functions. TLP is an enzyme functioning in the purine catabolism pathway, where it hydrolyses 5-hy- droxyisourate (5-HIU), the oxidation product of uric acid. Phylogenetic analyses have revealed that it is likely that the transthyretin gene arose as a result of a duplication of the TLP gene in early vertebrate evolu- tion. Thus, the evolution of TLP and transthyretin rep- resents a remarkable case of the divergent evolution from an enzyme to a hormone distributor. This minireview will present and discuss recent find- ings regarding the identification and distribution of TLP genes in nature, the structural and functional characterization of the TLP from various organisms, and the evolution of TLP and transthyretin. The identification of TLPs and transthyretins in nature The evolution of transthyretin and its distribution in nature have been well researched [1,2]. Several studies have demonstrated that all vertebrates synthesize transthyretin at some stage during their development [2–4] and this synthesis is primarily localized to the liver, choroid plexus and retinal pigment epithelium. The expression of the transthyretin gene in vertebrates occurs independently in these tissues [5,6]. There is considerable sequence identity and similarity between the amino acid sequences of transthyretin from various vertebrate organisms. The most divergent transthyretin sequences (for example, human and sea bream trans- thyretin) still retain 67% similarity (48% identity). This consensus of primary structure is also reflected in the highly conserved secondary, tertiary and quater- nary structures of transthyretin from vertebrates. Together, these data suggest that the transthyretin gene may have evolved before the divergence of verte- brates from invertebrates. The genomics era has been characterized by the increasingly rapid sequencing of multiple genomes alongside the development of sophisticated pairwise sequence-alignment search tools. These were the cata- lysts enabling the search for a transthyretin homolog among nonvertebrate organisms. The first evidence for such a homolog was published in 2000 [7], when blast searches [8] revealed the existence of ORFs with the potential to encode a protein of similar length and sequence composition to transthyretin. These ORFs were initially identified in the enteric bacteria Escherichia coli and Salmonella dublin, in the yeast Schizosaccharomyces pombe and in the nematode Caenorhabditis elegans [7]. The predicted protein homo- log of transthyretin was termed TLP because the name transthyretin implied a role in the transport of thyroid hormones and retinol binding protein [9]. Such a func- tion could not be assumed for the nonvertebrate trans- thyretin homolog. Sequence characteristics of TLP and transthyretin With the availability of an increasing number of genomes to mine, Eneqvist et al. [10] used bla st searches to identify a further 49 putative TLP sequences in the genomes of bacteria, plants and invertebrate animals. The TLP genes they identified typically encoded a pro- tein of 114 amino acid residues compared with, on average, 125 residues in transthyretin (the number of residues was species dependent). Furthermore, Eneq- vist et al. [10] observed that all TLP sequences possessed a consensus C-terminal tetrapeptide: Tyr- Arg-Gly-Ser. Alignment of TLP and transthyretin sequences revealed that the regions of greatest similar- ity between the two families of proteins were in the N-terminal and C-terminal regions [11]. In order to distinguish between the two protein families in greater detail, a comparative analysis of TLP and transthy- retin sequences was performed [11]. In this study, a set of bacterial TLP and vertebrate transthyretin sequences was probed for motifs that might be con- served in each group. The study revealed that the transthyretin sequences in this set were so similar that a single motif spanned the entire length of each protein sequence. However, in the set of TLP sequences, five specific motifs were identified, namely motifs A–E (with motif A being the most highly conserved). The motifs in the TLP sequences were found in the follow- ing arrangement (from N-terminal to C-terminal): (E)-B-D-C-A (see Fig. 1A). Motif E was only found in TLPs from plant species and from two alphaproteo- bacteria: Bradyrhizobium japonicum and Magnetospiril- lum magnetotacticum. Motif E is homologous to the proteins of cluster of orthologous groups (COG) 3195, a group of bacterial proteins where the entire protein is made up of this single domain. Motif E has been subsequently identified as a unique protein, 2-oxo-4-hydroxy-4-carboxy-5-ureidoimadolazine (OHCU) decarboxylase, whose function relative to TLP will be discussed later in this review. A combined set of TLP and transthyretin sequences was also probed for motifs to determine whether there were any motifs in common between the two protein families. Three motifs (A’–C’), which highlighted regions of similarity between TLP and transthyretin sequences, were identified and found in the arrange- ment B’-C’-A’ (see Fig. 1B). These motifs were shown The evolution of the transthyretin-like protein S. C. Hennebry 5368 FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS to correspond to regions of structural significance in the transthyretin molecule (see Fig. 1C). Motif A’ cor- responds to residues that line the hydrophobic core of the transthyretin tetramer. Motif B’ corresponds to residues forming the dimer–dimer interface and resi- dues in motif C’ are involved in monomer–monomer interactions (see Fig. 1C). Based on these observations, it was hypothesized that TLP probably has a tertiary structure similar to that of transthyretin [11]. The motifs identified in this study also provided a more accurate means of differentiating between TLP and transthyretin sequences and for the identification of novel TLP ⁄ transthyretin sequences through Hidden Markov searches in protein databases [11]. Interestingly, whilst motifs A’–C’ represent the regions of greatest sequence similarity between TLP and transthyretin, they also contain specific amino acid substitutions that enabled the distinction of one group from the other. For instance, at their C-termini (motif A’ region), nearly all TLP sequences possess a Tyr- Arg-Gly-Ser tetrapeptide. Specifically, the tyrosine and glycine residues were found to be 100% conserved among TLP sequences. Upon sequence alignment with TLP, the residues at the same positions in transthyretin are threonine and valine, respectively. At the N-termini of TLP sequences (the motif B’ region) a conserved his- tidine residue was found. The equivalent residue in transthyretin sequences is lysine (also 100% conserved). Interestingly, the residues involved in TH binding in transthyretin are not conserved in TLP sequences. Rather, it appears that residues involved in the struc- tural integrity of the TLP ⁄ transthyretin molecule have been conserved. The alignment of representative trans- thyretin and TLP sequences in Fig. 2 demonstrates the distribution of residues that are 100% conserved in both TLP and transthyretin sequences as well as those that are 100% conserved solely within the set of TLP sequences. Distribution of TLPs and transthyretins in nature The distribution of TLP in nature and its evolutionary relationship to transthyretin have been studied exten- sively in recent years [10,11]. To date, TLP genes have been identified in over 200 organisms across all king- A B C Motif A′ A′ ACDB C′ ~ 127 amino acids ~ 114 amino acids B′ Motif B′ Motif C′ Fig. 1. Motifs common between TLP and transthyretin indicate conservation of the TLP ⁄ TTR structure through evolution. Motifs identified in (A) TLP sequences and (B) transthyretin+TLP sequences. (A) In the set of TLP sequences, four motifs were identi- fied (A–D). The motifs are found in the order B-D-C-A, with A being the most highly con- served. (B) In the set of transthyretin+TLP sequences, three motifs were identified, A’–C’. Motif A’ is equivalent to motif A from the TLP motif set. Motif B’ is similar but extended in the N-terminal and C-terminal regions to motif B. Motif C’ is shorter than motif C and its location is shifted towards the N-terminus. Motif D is specific to the TLP set of proteins. (C) Motifs A’–C’ were superimposed on the tertiary structure of sea bream transthyretin. Motif A’ lines the hydrophobic core. Motif B’ forms the dimer–dimer interface and the opening of the central channel of the TTR molecule. Residues in motif C’ are involved in mono- mer–monomer interactions. (Modified from [11]). S. C. Hennebry The evolution of the transthyretin-like protein FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS 5369 doms. By contrast, the transthyretin gene is only found in vertebrates. Whilst the TLP gene is widely distributed in nature, there are some notable absences or apparent ‘losses’ of the TLP gene. For instance, no protozoans to date have been found to have a TLP gene, even though related organisms such as the slime mold Dictyosteli- um discoideum and the jakobite Jakoba bahemiensis both express the TLP gene. A TLP gene is absent from the cnidarian and ascidian phyla, despite the fact that organisms before and after these branch points in evo- lution express the TLP gene. This evidence suggests that whilst TLP might have been conserved throughout evo- lution because they have an important functional role, it is by no means essential to all organisms. Subcellular localization of TLP in bacteria In most instances, the TLP gene is present as a single copy in the organisms in which it has been identified. The gene typically encodes a cytoplasmic protein and, in the case of bacteria, is typically located in purine metabolism operons, neighbouring the gene which encodes OHCU decarboxylase [11]. This is consistent with the recently determined role of TLP in this meta- bolic pathway (to be discussed later). A notable excep- tion to this is the case of the enterobacterial TLP genes and a handful of TLP genes from other Gram- negative bacteria. The TLPs from these bacteria have been found to possess an N-terminal extension, namely a periplasmic localization sequence [11]. Interestingly, these TLP genes are not found to be associated with purine metabolism operons [11], and it is therefore tempting to speculate that their primary function is not purine metabolism. Some organisms have multiple copies of a TLP gene (see Table 1) [11]. In these cases, one gene encodes a cytoplasmic TLP and the ‘additional’ TLP gene encodes a periplasmic protein that, similarly to entero- bacterial TLP genes, is not associated on the bacterial chromosome with genes encoding proteins involved in purine metabolism. Indeed, phylogenetic analyses of all periplasmic TLP sequences (S.C. Hennebry, unpub- lished results) suggests that the genes encoding these TLPs were probably obtained through horizontal gene transfer from an enterobacterial ancestor. Subcellular localization of TLPs in eukaryotes In most nonfungal eukaryotic TLP sequences exam- ined to date, an N-terminal extension has also been Fig. 2. Alignment of representative transthyretin and TLP sequences. Mature amino acid sequences for transthyretin and TLP from selected organisms are shown (i.e. with signal peptides removed). The shared secondary structure characteristics of transthyretins and TLPs are indi- cated above the alignment: motifs A’–C’ are indicated with straight lines and are labelled; b-strands are indicated with arrows and are labelled A–H. A single a-helix is indicated with a rectangle. The residues that are strongly conserved between transthyretins and TLPs are indicated with an asterisk (*). Residues 100% conserved among all TLP sequences are indicated with a hash (#). Numbering for human transthyretin is shown directly beneath the alignment. The evolution of the transthyretin-like protein S. C. Hennebry 5370 FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS identified. This N-terminal extension contains a nona- peptide, which is predicted to encode a type-two peroxisomal sequence (PTS2) [11,12]. Recently, a proteomic analysis of leaf peroxisomes confirmed the peroxisomal localization of the Arabidopsis thaliana TLP [13]. Found in all eukaryotic cells, peroxisomes are specialized organelles in which oxidative reactions, such as those associated with purine metabolism, are compartmentalized. The co-localization of purine- metabolism enzymes (e.g. uricase) with TLP in peroxi- somes is therefore in keeping with the function of the A. thaliana TLP hydrolysis of the purine 5-HIU (S. C. Hennebry, unpublished results). These observations contradict those made by Nam and Li [14], where the A. thaliana TLP was reported to be localized only in the cytosol and was unlikely to have a function in pur- ine metabolism. In this study, the authors failed to take into account that the A. thaliana gene At5g58220 encoded two distinct proteins: OHCU decarboxylase and TLP. Therefore, conclusions drawn from yeast two-hybrid studies were based on interactions of the N-terminal region of OHCU decarboxylase with the receptor kinase brassinosteroid-insenitive-1, rather than interactions made by TLP. Furthermore, their conclu- sion that the TLP could not be peroxisomal was largely based on the observation that the TLP did not possess a C-terminal peroxisomal targeting sequence. Splice variants have been detected for most eukary- otic TLP genes and some of these variants result in the truncation of the TLP at the N-terminus. This trunca- tion has no effect on amino acid residues known to be involved in the function of the protein, but result in the deletion of the PTS2 nona-peptide. In the case of Mus musculus, transcript data available at RIKEN Mouse Encyclopedia (genome.gsc.riken.go.jp) suggest that over 90% of TLP gene transcripts possess the region encoding the PTS2 and were isolated from hepatocytes. A small proportion of TLP transcripts (< 10%) do not encode the PTS2 and appear not to be under tissue-specific regulation. Splice variations resulting in deletion of the PTS2 have also been described for plant TLPs [11]. All TLP sequences identified in the Viridiplantae kingdom are encoded by multiple exons [11]. For example, the TLP gene from A. thaliana is encoded by four exons, the last of which encodes the TLP. As pre- viously mentioned, exons 1–3 (motif E) encode a pro- tein from COG 3195, which was recently identified as the enzyme OHCU decarboxylase [12,15]. The functional relationship between TLP and OHCU decarboxylase will be discussed below. Evidence for gene duplication The most primitive organisms found to have a trans- thyretin sequence are the lampreys Petromyzon marinus and Lampetra appendix [16]. By contrast, TLP genes have been identified in all kingdoms. Given their high degree of sequence similarity, it has been hypothesized that the transthyretin gene arose as a result of a dupli- cation of the TLP gene at some stage in early verte- brate evolution [11]. Initial phylogenetic analyses of TLP and transthyretin sequences showed a branching of transthyretin slightly before the separation of the chordates [17]. Subsequent analyses using the recently determined transthyretin sequences from lamprey and recent additions to echinoderm expressed sequence tag (EST) databases, suggest that the TLP gene duplica- tion probably occurred just after the separation of echinoderms (S. C. Hennebry, unpublished results). Table 1. Bacteria with multiple copies of TLP genes. Organism Taxonomy (phylum, class) Genes encoding cytoplasmic TLP Genes encoding periplasmic TLP Rhodococcus Actinobacteria, Actinobacteria 2 0 Bradyrhizobium sp. Proteobacteria, Alphaproteobacteria 2 0 Sinorhizobium meliloti Proteobacteria, Alphaproteobacteria 2 0 Dinoroseobacter shibae DFL 12 Proteobacteria, Alphaproteobacteria 2 0 Loktanella vestfoldensis SKA53 Proteobacteria, Alphaproteobacteria 2 0 Roseovarius sp. HTCC2601 Proteobacteria, Alphaproteobacteria 2 0 Ralstonia eutropha H16 Proteobacteria, Betaproteobacteria 2 1 Comamonas testeroni KF-1 Proteobacteria, Betaproteobacteria 2 1 Klebsiella pneumoniae Kp342 Proteobacteria, Gammaproteobacteria 1 1 Salmonella enterica ssp. I choloraesuis Proteobacteria, Gammaproteobacteria 0 2 Chromohalobacter salexigens DSM3034 Proteobacteria, Gammaproteobacteria 1 1 Acinetobacter sp. (strain ADP1) Proteobacteria, Gammaproteobacteria 1 1 Pseudomonas fluorescens Pf5 ATCC BAA-477 Proteobacteria, Gammaproteobacteria 1 2 S. C. Hennebry The evolution of the transthyretin-like protein FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS 5371 Following the gene-duplication event, profound modifications to the duplicated TLP occurred, leading to the development of a deep channel into which the THs 3¢,3,5-triiodo-L-thyronine (T3) and 3¢,5¢,3,5-tetra- iodo-L-thyronine (thyroxine, T4) could bind. The nature of this structural modification will be discussed below. The function of TLP in purine metabolism To date, three studies have been performed examining the role of TLP in vivo. In a study of the A. thaliana TLP, no phenotype was observed when an insertional mutation was introduced into the TLP gene [14]. How- ever, the lack of phenotype observed may be attributed to the presence of an additional 5-HIU hydrolase in plants (see later discussion regarding TLP functional redundancy). In 2003, Eneqvist et al. [10] performed RNA interference (RNAi) studies in C. elegans to determine a loss-of-function phenotype for R09H10.3 and ZK697.8 TLP genes. RNAi-treated worms were scored for embryonic lethality and for postembryonic phenotypes (sterility, aberrant morphology, uncoordi- nated movements, egg-laying defects or slow growth). No obvious phenotype was detected upon examination of the gross phenotype of the worms using a dissecting microscope [10]. However, more in-depth examination into a possible phenotype was not performed. For example, the worms were not subjected to any type of environmental stress. In addition, RNAi was per- formed using dsRNA for a single TLP gene at a time. As such, the RNAi studies in C. elegans may have been more informative had double-knockdown studies been performed. A role for TLP in purine metabolism was first pro- posed in 2001. In an effort to develop a greater under- standing of purine metabolism in the Gram-positive bacterium, Bacillus subtilis, Schultz et al. [18] generated a series of insertion mutants. One of these mutations was made in the TLP gene (pucM), which is located immediately downstream of the gene encoding uricase. The bacteria harbouring this mutation were character- ized as having a reduced rate of proliferation (com- pared with wild-type bacteria) on media containing uric acid as the principal source of nitrogen [18]. Purines are major components of nucleic acids and nucleotides. Subsequently, de novo and salvage path- ways for purine biosynthesis are important compo- nents in the metabolism of all organisms. The ability to degrade purine compounds, either aerobically or anaerobically, has been identified in all kingdoms [19]. The aerobic degradation of purines is dependent on the oxidation of hypoxanthine and xanthine to uric acid via xanthine dehydrogenase ⁄ oxidase (E.C. 1.1.1.204 ⁄ E.C. 1.1.3.22). In humans, anthropoid apes, birds, uricotelic reptiles and most insects, uric acid is the end product of purine metabolism and is thus excreted [20,21]. Most mammals and gastropods fur- ther degrade uric acid to allantoin [20,22], fish and amphibians completely degrade purines to urea, ammonia and carbon dioxide [20,23,24], whilst most plants degrade purines to carbon dioxide and ammonia [25]. Purine oxidation, in particular that of uric acid, is the major route of ureide biogenesis in nature. Conse- quently, the enzymes involved in the various stages of purine metabolism have been the focus of much inves- tigation. Recently, however, the degradation of uric acid to allantoin has been shown to be more complex than originally thought. Previously, it had been assumed that uricase (EC 1.7.3.3) was the sole enzyme responsible for the oxidation of uric acid to allantoin. However, Tipton’s group [26] showed that the oxida- tion of uric acid by uricase in fact yields the metastable compound, 5-HIU. They observed the spontaneous decomposition of 5-HIU to OHCU within 20 min at neutral pH, followed by the spontaneous decarboxyl- ation of OHCU to racemic allantoin. The spontaneous decomposition of 5-HIU results in the generation of numerous free-radical species, which ultimately con- tribute to lipid oxidation [27]. Given this fact and the observation that only (S)-allantoin is found in nature, Kahn and Tipton [26] proposed the existence of addi- tional enzymes in the uric acid degradation pathway – first to hydrolyse 5-HIU and second to decarboxylate OHCU to (S)-allantoin. As previously discussed, bacterial TLP genes are fre- quently found in close proximity to the uricase gene and to another gene encoding proteins belonging to COG 3195. In 2005, Lee et al. [28] revealed the ability of recombinant TLP from B. subtilis and E. coli to specifi- cally hydrolyse 5-HIU. Importantly, they demonstrated the inability of human transthyretin to hydrolyse the same compound. Ramazzina et al. [12] subsequently showed that mouse TLP hydrolysed 5-HIU and that the COG 3195 proteins were responsible for the decarboxyl- ation of OHCU to (S)-allantoin. Thus, the pathway of the conversion of uric acid to (S)-allantoin via the three enzymes uricase, TLP (5-HIUase) and OHCU decar- boxylase was revealed (see Fig. 3). Whether the three proteins are able to form a multi-enzyme complex remains to be determined. One could speculate that the ability to do so would be favourable given the rapid kinetics of spontaneous decomposition of both 5-HIU and OHCU. The evolution of the transthyretin-like protein S. C. Hennebry 5372 FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS To date, the TLP from three bacteria [28–30], one plant (A. thaliana; S. C. Hennebry, unpublished results) and two vertebrate species [12,17], have been analysed for 5-HIU hydrolytic activity and have all been shown to be 5-HIU hydrolases. Thus, a role for TLP in this purine degradation pathway is evident throughout evolution. In addition, the expression of the TLP gene in some organisms may be uric acid- dependent. For example, in the Gram-positive bacte- rium Deinococcus radiodurans, both the uricase and TLP genes are regulated by a novel uric acid-respon- sive transcriptional regulator of the MarR family [31]. Given the similarities in the structures of purine metabolism operons among Gram-positive bacteria, it is likely that both uricase and TLP genes are similarly regulated in other bacteria. Interestingly, periplasmic TLPs (those from the Enterobacteria) have also been demonstrated to have 5-HIU hydrolase activity [28,30]. Given that in bacteria, purine metabolism is localized in the cytosol, it is pos- sible that the TLP from these organisms acts indepen- dently of the classical purine catabolism pathway. In addition, no enterobacteria have been found to possess homologs of OHCU decarboxylase or uricase genes. Therefore, the question arises as to the in vivo role of periplasmic TLP and whether it is capable of hydroly- sing compounds other than 5-HIU. The fact that TLP has been demonstrated to hydro- lyse 5-HIU results in its inclusion in the superfamily of cyclic amidohydrolases (E.C. 3.5.2). Other cyclic amidohydrolases include hydantoinase, allantoinase and dihydrooratase [32]. Cyclic amidohydrolases share a number of physicochemical characteristics. These characteristics include quaternary, tertiary, secondary and primary structure as well as the reliance on a diva- lent metal cofactor via a conserved metal-binding motif [33]. Studies have also shown the inhibitory action of some divalent cations on cyclic amidohydro- lase activity as well as the ability of many enzymes within this group to bind a variety of cyclic amides with varying affinities [32]. TLP does not appear to share the classic sequence characteristics of cyclic amidohydrolases (S. C. Hennebry, unpublished results). Whilst the E. coli TLP was crystallized in the presence of Zn 2+ , it has been shown that TLP is not a zinc-dependent hydrolase [17]. Structural comparison of Transthyretin and TLP The 3D structures of transthyretin from various organ- isms have been well characterized. The first transthyre- tin crystal structure to be solved (that of human) was published in 1978 [34]. The Protein Database (http:// www.pdb.org) contains multiple crystal structure coor- dinates for human transthyretin (including multiple amyloidogenic forms and with various ligands bound). The crystal structures of transthyretin from rat [35], chicken [36] and sea bream [37,38] have also been solved. All of these structures demonstrate the remark- able conservation of the prealbumin-like fold (as described by SCOP, http://scop.mrc-lmb.cam.ac.uk), which consists of an eight-stranded b-sandwich (strands A-H) with each sheet adopting a greek-key topology. A two-turn a-helix usually (with the exception of chicken transthyretin) exists between strands E and F in trans- thyretin. The two transthyretin dimers associate, via nonpolar interactions, between the loops joining stands G and H with the loops joining strands A and B, mak- ing the transthyretin tetramer a ‘dimer of dimers.’ Recently, the first crystal structures of TLP from various organisms were solved. Within a short period of 3 months, the crystal structures for the TLP from S. dublin (pdb: 2GPZ; [30]), E. coli (pdb: 2G2N; [39]), B. subtilis (pdb: 2H0E; [29]) and Danio rerio (zebrafish; pdb: 2H6U; [17]) were solved. Remarkably, the struc- tures of these proteins, all tetrameric, showed signifi- cant similarity to the published structures of transthyretin. By way of example, a comparison of the structure of S. dublin TLP with the structures of trans- thyretin from various organisms is shown in Figure 4. Generally, the structural deviation between TLPs and transthyretins from various organisms is of the same order of magnitude to that within the set of transthyre- Fig. 3. Schematic of the oxidation of uric acid. Uric acid is oxidized by uricase to 5-HIU, which is subsequently hydrolysed by TLP (5-HIU hydrolase) to OHCU. The enzyme OHCU decarboxylase generates (S)-allantoin. (Adapted from [26]). S. C. Hennebry The evolution of the transthyretin-like protein FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS 5373 tin. For instance, the rmsd between equivalent Ca atoms in the structures of TLPs and human transthy- retin are 1.0 A ˚ and 1.2 A ˚ for the monomer and dimer respectively [17]. The rmsd between equivalent Ca atoms in the structures of transthyretin from various vertebrates is between 0.34 A ˚ and 1.59 A ˚ [30]. The main differences between the structures of TLP and transthyretins are found in the loop connecting b-strands B and C, which is highly exposed to the solvent in TLP [17]. Interruptions in the b-strands A, G and H are also observed in TLP structures as a result of alterations to the formation of hydrogen bonds between strands. The carbonyls of residues V104 and P105 (zebrafish TLP numbering), in the middle of b-strand G, do not form hydrogen bonds with the nitro- gen atoms of H12 and Y116 of b-strand H in TLP. The P105 residue, mainly responsible for the b-strand irregularities, is invariant in TLP sequences, suggesting a crucial role for the particular conformation observed in b-strands A, G and H [17]. Structural nature of the TLP and transthyretin active sites One of the striking features of transthyretin is the cen- tral channel of the protein into which the THs bind. This central channel traverses the entire tetramer. It has previously been postulated [40] and demonstrated [7,41] that the characteristics of the N-termini of trans- thyretin from different organisms account for differ- ences in the affinity of the two main THs (T3 and T4) to the channel by hindering or allowing greater accessi- bility. The central channel is also present in TLP, albeit with quite different structural properties. Previously, it was demonstrated that the regions of greatest similar- ity between TLP and transthyretin were those forming this central channel, namely motifs A’ and B’ (see Fig. 1C). Interestingly, differences between TLP and transthyretin within these motifs also account for sig- nificant physicochemical alterations to the central channel of the protein and provide a structural basis for the differing function compared with transthyretin. The presence of a conserved, bulky tyrosine residue at the C-termini of TLP (part of the Tyr-Arg-Gly-Ser tet- rapeptide) causes the central channel to become blocked (see Fig. 5A). As a result, the dimer–dimer interface of TLP is characterized by two ‘grooves’ on either side of the protein rather than a central channel [30]. Other key residues at the dimer–dimer interface of TLP include H14, R49 and H106 (B. subtilis TLP numbering) [29]. An examination of the active site of B. subtilis TLP with the uric acid analogue 8-azaxan- thine bound, reveals that these residues form impor- tant interactions with the ligand (see Fig. 5B). Indeed, site-directed mutagenesis studies targeting these resi- dues show that substitution at these sites has profound consequences for the 5-HIU hydrolase activity of the TLP [17,29,30] (and see Table 2). Mutagenesis of H14 and R49 showed that these resi- dues are the most sensitive to mutation, with H14A, H14N and R49E substitutions abolishing enzyme activity (B. subtilis numbering) [29,30]. However, the conservative substitution at residue 49 from arginine to lysine had no effect on activity. This suggests the need for a positively charged residue at this site. Sub- stitution of H105 and Y118 also significantly reduced enzyme activity, by approximately 90% [30]. Deletion of the C-terminus tetrapeptide Tyr-Arg-Gly-Ser signifi- cantly affected enzyme activity, but it has been sug- gested that S121 does not influence the reaction [29]. Fig. 4. Comparison of the tertiary structure of TLP with transthyretin. Stereo diagram showing a superimposition of tetramers of Salmonella dublin TLP (magenta) with trans- thyretin from human (1F41, cyan), rat (1KGI, yellow), chicken (1TFP, orange) and sea bream (1SNO, green). Tetramers were superimposed using the A chain only. (Adapted from [30]). The evolution of the transthyretin-like protein S. C. Hennebry 5374 FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS Interestingly, those residues playing a role in enzyme activity are 100% conserved in all TLP sequences and 100% substituted in transthyretin (see Table 2). How- ever, substitutions at position 121 (to threonine or glu- tamate) have been observed. None of the mutations affected the tetrameric assembly of the TLP molecule [30]. Furthermore, the surface charge of the TLP active site is considerably different from the equivalent region in transthyretin [29,30]. An electrostatically positive groove in TLP contrasts the negatively charged TH-binding site in transthyretin (see Fig. 5C). In summary, a comparison of the catalytic cavity of TLP with the equivalent region of transthyretin (the TH-binding channel) revealed that the TLP cavity is significantly shallower and ‘groove-like’ compared with the deep, hollow channel of transthyretin [30]. In par- ticular, the substitution of the C-terminal tyrosine (118) with the much less bulky threonine residue fol- lowing duplication, had a profound effect on the shape of the channel. Loss of the tyrosine residue opened up A B C iii iii iii Fig. 5. The active site of TLP. (A) Compari- son of the ligand-binding cleft at the dimer– dimer interface in (i) human transthyretin with (ii) Salmonella dublin TLP. Residues that contribute to the active site are shown. Hydrogen bonds are shown as broken cyan lines. Thyroxine is shown in stick represen- tation in yellow. For clarity, some elements of secondary structure are not shown. Resi- dues His6, His95 and Y108 (S. dublin TLP numbering) are equivalent, upon structural alignment, to Lys15, Thr106 and Thr119 of human TTR. (Adapted from [30].) (B) The active site of B. subtilis TLP with (i) the uric acid analog, 8-azaxanthine bound and (ii) showing interacting residues (from [29]). Note that the active site of the B. subtilis TLP is depicted at 90° to those depicted for transthyretin and TLP in part A. (C) (i) Elec- trostatic surface potential of human trans- thyretin with thyroxine bound inside the negatively charged and deep channel at the dimer–dimer interface of the protein. (ii) The equivalent region in TLP is shallow and positively charged. (Adapted from [30]). Table 2. Site-directed mutagenesis of conserved residues in TLP. Transthyretin residue Equivalent residue in TLP (S. dublin TLP numbering) Effect of mutation on TLP activity Publication Lys15 His6 Abolishes [30] Ser52 Asp42 Reduces by 50% [17] Glu53 Arg44 Abolishes [29] Thr106 His95 Reduces by 90% [30] Thr119 Tyr108 Reduces by 90% [30] Val122 Ser111 No effect [29] S. C. Hennebry The evolution of the transthyretin-like protein FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS 5375 the central channel of the transthyretin molecule, allowing for the binding of bulkier ligands such as THs. Superimposition of the dimer–dimer interface of TLP with that of transthyretin illustrates the evolu- tionary changes that resulted in the functional transi- tion of the enzyme into a transport protein. Comparison of structures of TLPs from various organisms A comparison of the TLP from three species of bacte- ria with a vertebrate TLP (zebrafish) shows little struc- tural divergence. Major differences between the S. dublin TLP and zebrafish TLP are found in the flex- ible portions of strands B and C that protrude towards the solvent and in the conformation of the long loop connecting strands D and E [17]. Greater differences are observed between the structures of B. subtilis and zebrafish TLPs: loop B-C is significantly shorter in B. subtilis TLP whilst the loop connecting the short a-helix to strand F is extended. The active sites of TLPs from prokaryotes and eukaryotes are nearly identical. The location and ori- entation of the residues present in the catalytic pockets are well maintained, including the putative main cata- lytic residues H12 and R52 (zebrafish TLP numbering). The only significant difference is found in the C-termi- nal serine residue, which assumed different orientations in the three structures. However, the role of this resi- due in catalysis has been shown to be negligible [29]. Evolution of TLP function in the context of urate metabolism Ramazzina et al. [12] eloquently demonstrated the co-evolution of the three proteins [uricase, TLP (5-HIUase) and OHCU decarboxylase] involved in the oxidation of uric acid to allantoin. Certainly, the co-localization of these proteins in the peroxisomes of metazoan and plant species, and the co-regulation of TLP genes in some bacteria, suggests a concerted effort in the rapid generation of allantoin. The co-dis- tribution of uricase, TLP and OHCU decarboxylase genes in nature reveals that whenever an organism is found to have a uricase gene, it always has both TLP and OHCU decarboxylase genes [12]. In vertebrates, the loss of these three genes through evolution is mir- rored. For instance, hominoids lost their ability to degrade uric acid as the result of the inactivation of the uricase gene in a primate ancestor, some 15 Ma [42]. In humans, the TLP gene has several inactivating mutations and the OHCU decarboxylase gene does not appear to be expressed [12]. Uric acid is a potent antioxidant in biological sys- tems. Despite uric acid being the end point of purine metabolism in humans and birds, high levels of allan- toin have been detected in their plasma [43,44]. Uric acid chelates transition metal ions (minimizing metal- catalysed oxidation), scavenges hypochlorous acid, is a potent quencher of peroxynitrite and reduces haemo- globin oxidation by nitrite (for a review, see [45]). It has been suggested that in humans and birds, the allantoin generated in these organisms could be a mea- sure of the levels of oxidative stress [44]. The nonenzymatic oxidation of uric acid generates 5-HIU, just as in the uricase reaction. As previously discussed, 5-HIU is a highly reactive compound, which, if left to spontaneously decompose, is capable of forming numerous free-radical species, which ulti- mately contribute to lipid peroxidation [27]. Therefore, the rapid elimination of 5-HIU would be advantageous to the organism. Whilst birds have lost functional uri- case and OHCU decarboxylase gene products, TLP transcripts have been detected. It is tempting to specu- late that the role of TLP in birds might be to rapidly ‘mop-up’ 5-HIU generated through the nonenzymatic oxidation of uric acid, thereby reducing the potential free-radicals generated when 5-HIU is left to spontane- ously decompose. The role of TLP in scavenging 5-HIU clearly warrants further investigation. Purine metabolism occurs in the cytosol of bacteria (for a review, see [46]). The fact that most bacteria possess a cytosolic TLP is consistent with this. How- ever, it is not clear what the functional role of a TLP localized to the periplasm might be. It is possible that the source of 5-HIU to the periplasm could be from the external environment. Interestingly, all bacteria which possess a periplasmic TLP are found to colonize various animals. Uric acid is secreted on the surface of mucosal epithelial tissues of all animals as part of the innate immune system [47] and is also thought to act as a microbicidal agent in these instances. Because uric acid can easily permeate the outer membrane of these bacteria, it might be that the TLP located in the peri- plasm acts as a primary defence for the bacterium against oxidized uric acid. Alternatively, it could be that 5-HIU is generated in small quantities by the non- specific oxidation of the uric acid by other periplasmic enzymes, such as cytochrome c or peroxidase [48,49]. TLP: an enzyme with functional redundancy? TLP was not the first protein to be identified as having 5-HIU hydrolytic activity. Having hypothesized the need for additional enzymes to contribute to the oxida- The evolution of the transthyretin-like protein S. C. Hennebry 5376 FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS [...]... Brajter-Toth A (1989) On-line mass spectrometric investigation of the peroxidase-catalysed oxidation of uric acid Anal Chem 61, 1709–1717 The evolution of the transthyretin-like protein 50 Raychaudhuri A & Tipton P (2002) Cloning and expression of the gene for soybean hydroxyisourate hydrolase Localisation and implications for function and mechanism Plant Physiol 130, 2061–2068 51 Sarma AD, Serfozo P, Kahn... Munro SLA, Richardson SJ & Schreiber G (1999) Evolution of thyroid hormone binding by transthyretins in birds and mammals Eur J Biochem 259, 534–542 Prapunpoj P, Leelawatwatana L, Schreiber G & Richardson SJ (2006) Change in the structure of the N-terminal region of transthyretin produces change in affinity of transthyretin to T4 and T3 FEBS J 273, 4013–4023 Oda M, Satta Y, Takenaka O & Takahata N (2002)... (prealbumin) J Biol Chem 261, 3475–3478 Prapunpoj P, Yamauchi K, Nishiyama N, Richardson SJ & Schreiber G (2000) Evolution of structure, ontogeny of gene expression, and function of Xenopus laevis transthyretin Am J Physiol Regul Integr Comp Physiol 279, R2026–R2041 Altschul SF, Madden TL, Schaffer AA, Zhang Z, ¨ Miller W & Lipman AJ (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database... functional analysis of PucM, a hydrolase in the ureide pathway and a member of the transthyretin-related protein family Proc Nat Acad Sci USA 103, 9790–9795 30 Hennebry SC, Law RHP, Richardson SJ, Buckle AM & Whisstock JC (2006b) The crystal structure of the transthyretin-like protein from Salmonella dublin, a prokaryote 5-hydroxyisourate hydrolase J Mol Biol 359, 1389–1399 31 Wilkinson SP & Grove A (2004)... analysis of Arabidopsis leaf peroxisomes reveals novel targeting peptides, metabolic pathways and defense mechanisms Plant Cell 19, 3170– 3193 Nam KH & Li J (2004) The Arabidopsis transthyretinlike protein is a potential substrate of BRASSINOSTEROID-INSENSITIVE 1 Plant Cell 16, 2406–2417 Kim K, Park J & Rhee S (2007) Structural and functional basis for (s)-allantoin formation in the ureide pathway J Biol... Experientia 48, 583–593 25 Ashihara H & Crozier A (2000) Biosynthesis and metabolism of caffeine and related purine alkaloids in plants Adv Bot Res 30, 117–205 26 Kahn K, Serfozo P & Tipton P (1997) Identification of the true product of the urate oxidase reaction J Am Chem Soc 119, 5435–5442 27 Santos CXC, Anjos EI & Augusto O (1999) Uric acid oxidation by peroxynitrite: multiple reactions, free radical formation,... of uric acid [26], a 5-HIU hydrolase from soybean (Glycine max) root nodules was purified [50,51] This 5-HIU hydrolase showed greatest homology to b-glucosidases (3.2.1.21) (members of the family of retaining glycosidases) and has quite a different catalytic mechanism to TLP in order to hydrolyse its substrate The fact that two structurally distinct proteins have been identified as sharing the same function. .. (1998) Identification of the structural similarity in the functionally related amidohydrolases acting on the cyclic amide ring Biochem J 330, 295–302 Blake CCF, Geisow MJ & Oately SJ (1978) Structure of prealbumin: secondary, tertiary and quaternary inter˚ actions determined by Fourier refinement at 1.8 A J Mol Biol 121, 339–356 Wojtczak A (1997) Crystal structure of rat transthyretin ˚ at 2.5 A resolution:... substitution of a small number of residues in the active site of TLP appears to have been sufficient for the acquisition of new functional properties of the protein whilst its overall structure was unchanged Furthermore, the distribution of TLPs in all kingdoms, but the representation of transthyretins in vertebrates alone, clearly suggests that the transthyretin-like fold originally functioned in purine metabolism... chicken and turkey plasma Comp Biochem Physiol Pt B 135, 325–335 Ames BN, Cathcart R, Schwiers E & Hochstein P (1981) Uric acid provides an antioxidant defense in humans against oxidant- and radical-causing aging and cancer: a hypothesis Proc Natl Acad Sci USA 78, 6858–6862 FEBS Journal 276 (2009) 5367–5379 ª 2009 The Authors Journal compilation ª 2009 FEBS S C Hennebry 46 Reitzer L (2003) Nitrogen assimilation . MINIREVIEW Evolutionary changes to transthyretin: structure and function of a transthyretin-like ancestral protein Sarah C. Hennebry Department of Biochemistry and. These characteristics include quaternary, tertiary, secondary and primary structure as well as the reliance on a diva- lent metal cofactor via a conserved metal-binding motif