Primary Hyperoxaluria Type 1: AGT Mistargeting Highlights the Fundamental Differences Between the Peroxisomal and Mitochondrial Protein Import Pathways

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

Biochimica et Biophysica Acta 1763 (2006) 1776–1784 www.elsevier.com/locate/bbamcr

Review Primary hyperoxaluria type 1: AGT mistargeting highlights the fundamental differences between the peroxisomal and mitochondrial protein import pathways

Christopher J. Danpure

Department of Biology, University College London, Gower Street, London WC1E 6BT, UK Received 28 April 2006; received in revised form 1 August 2006; accepted 18 August 2006 Available online 24 August 2006

Abstract

Primary hyperoxaluria type 1 (PH1) is an atypical peroxisomal disorder, as befits a deficiency of alanine:glyoxylate aminotransferase (AGT), which is itself an atypical peroxisomal enzyme. PH1 is characterized by excessive synthesis and excretion of the metabolic end-product oxalate and the progressive accumulation of insoluble calcium oxalate in the kidney and urinary tract. Disease in many patients is caused by a unique protein trafficking defect in which AGT is mistargeted from peroxisomes to mitochondria, where it is metabolically ineffectual, despite remaining catalytically active. Although the peroxisomal import of human AGT is dependent upon the PTS1 import receptor PEX5p, its PTS1 is exquisitely specific for mammalian AGT, suggesting the presence of additional peroxisomal targeting information elsewhere in the AGT molecule. This and many other functional peculiarities of AGT are probably a consequence of its rather chequered evolutionary history, during which much of its time has been spent being a mitochondrial, rather than a peroxisomal, enzyme. Analysis of the molecular basis of AGT mistargeting in PH1 has thrown into sharp relief some of the fundamental differences between the requirements of the peroxisomal and mitochondrial protein import pathways, particularly the properties of peroxisomal and mitochondrial matrix targeting sequences and the different conformational limitations placed upon importable cargos. © 2006 Elsevier B.V. All rights reserved.

Keywords: Alanine:glyoxylate aminotransferase; Kidney stone; Mitochondrial protein trafficking; Oxalate metabolism; Peroxisomal protein trafficking; Primary hyperoxaluria type 1

1. PH1 and AGT deficiency deposition of insoluble calcium oxalate (CaOx) in the kidney and urinary tract, as various combinations of nephrocalcinosis Primary hyperoxaluria type 1 (PH1, MIM 259900) is a rare (diffuse deposition throughout the renal parenchyma) and/or autosomal recessive disorder of metabolism caused by a urolithiasis (calculi or stones). This eventually leads to renal functional deficiency of the liver-specific, pyridoxal-phos- failure, following which the combined effects of increased phate-dependent enzyme alanine:glyoxylate aminotransferase oxalate synthesis and failure to remove it from the body results in (AGT, EC 2.6.1.44) [1]. In the peroxisomes of normal human the deposition of CaOx almost anywhere. The clinical hepatocytes, AGT catalyses the transamination of the interme- presentation of PH1 is heterogeneous, for example with respect diary metabolite, glyoxylate, to glycine. This can be considered to age of first symptom, levels of urinary oxalate and glycolate to be a detoxification reaction because its dysfunction in PH1 excretion, relative contributions of nephrocalcinosis and uro- allows glyoxylate to escape from the peroxisomes into the lithiasis, and rate of progression of renal pathology. cytosol where it is oxidised to oxalate, catalysed by lactate dehydrogenase, and reduced to glycolate, catalysed by glyox- 2. Relationship between genotype and enzyme phenotype ylate/hydroxypyruvate reductase (Fig. 1) [2,3]. In humans, at least, oxalate cannot be further metabolised and its increased 2.1. The structure of normal AGT synthesis and urinary excretion leads to the progressive Human AGT is a homodimeric protein, each subunit being E-mail address: [email protected]. composed of 392 amino acids with a molecular mass of about

Fig. 1. Metabolic consequences of peroxisomal AGT deficiency in PH1. AGT, alanine:glyoxylate aminotransferase; DAO, D-amino acid oxidase; GO, glycolate oxidase; GRHPR, glyoxylate/hydroxypyruvate reductase; LDH, lactate dehydrogenase. Solid arrows, metabolic conversions; dotted arrows, membrane or other transport.

43 kDa [4]. Its recently-described crystal structure [5] (PDB formed,itisbelievedtoplayasignificantroleinthe 1H0C) (Fig. 2) shows that each subunit can be subdivided into dimerization process per se [5]. Dimerization of AGT is very three recognizable structural/functional domains. The first 20 or important, not only because monomeric AGT has vastly so residues make up an N-terminal extension that wraps over the reduced catalytic activity [8], but also because it is unstable surface of the other subunit. The next 260 or so residues form leading to aggregation and rapid degradation [9]. Knowledge of the ‘large domain’ that contains most of the active site and the the crystal structure of AGT has provided insights into the dimerization interface. The final C-terminal ‘small domain’ processes of AGT folding, dimerization and catalytic activity, as consists of about 110 residues, which contains among other well as allowing rationalisation of the effects mutations and things the principal and ancillary peroxisomal targeting polymorphisms on these processes (see below). information [6,7] (see below). Each subunit binds one pyridoxal phosphate, which forms a Schiff base with Lys209. The large 2.2. Enzyme phenotypes in PH1 surface area of the dimerization interface explains the high stability of the AGT dimer. Although the inter-subunit The clinical heterogeneity of PH1 is matched, or possibly interaction mediated by the N-terminal extension is unlikely even exceeded, by its enzymatic heterogeneity. Three overt to make any major contribution to stability to the dimer once categories can be identified:-deficiency of AGTcatalytic activity

Fig. 2. Crystal structure of human AGT. The two identical subunits are mainly coloured yellow, except for the N-terminal extensions (red) and the PTS1As (green).

NH2, amino termini; KKL, carboxy-terminal PTS1s. The location of the cofactor pyridoxal phosphate is indicated (PLP in orange), as are the locations of the Pro11Leu polymorphism (11 in pink), and the mutations discussed in the text (41, 82, 170, 244 in blue). It is the N-terminal extension (red) that acts as a MTS in the presence of the Pro11Leu polymorphism. 1778 C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784 but not AGT immunoreactivity, deficiency of both catalytic [17]. Some of the most common co-segregate and functionally activity and immunoreactivity, and deficiency of neither interact with the common Pro11Leu polymorphism of the minor catalytic activity nor immunoreactivity. Patients in the latter allele. Many of the better-studied mutations, or mutation– category can have AGT levels up to half the mean normal level, polymorphism combinations, are associated with specific which is well within the normal range and greater than found in enzyme phenotypes, such as loss of AGT catalytic activity, many asymptomatic carriers [2]. Intermediate categories also AGT aggregation, accelerated AGT degradation, and peroxi- exist, in which patients have markedly reduced, but still some-to-mitochondrion AGT mistargeting. The most exten- detectable levels of catalytic activity and/or immunoreactivity. sively studied genotype–enzyme phenotype relationships are More detailed analysis shows the presence of two enzyme described in more detail below. phenotypes unique to PH1. In a small group of patients with almost no AGT catalytic activity and markedly reduced AGT 2.4.1. Gly82Glu immunoreactivity, AGT appears to have aggregated into intra- The Gly82Glu mutation is associated with complete loss of peroxisomal ‘cores’ [9]. In another, much larger, group of AGT catalytic activity, without any effect on enzyme stability, patients who have significant levels of AGTcatalytic activity and dimerization or targeting [18,19]. Structural analysis shows this immunoreactivity, an even more remarkable enzyme phenotype to be due to the mutant glutamate side chain filling the space is found in which AGT is mistargeted from its normal location that should be occupied by the cofactor pyridoxal phosphate [5]. in the peroxisomes to the mitochondria [10]. Although This prevents Shiff base formation with the ε-amino group of mistargeted AGT remains catalytically active, it appears to be Lys209, thereby abolishing catalytic activity [8]. Despite lack of metabolically inefficient. This is because efficient glyoxylate enzyme activity, patients homozygous for Gly82Glu have detoxification must occur at its site of synthesis (i.e. in the completely normal levels of normally compartmentalised peroxisomes), thus preventing its oxidation to oxalate, catalysed immunoreactive AGT protein. Unlike many of the missense by cytosolic LDH. Such organelle-to-organelle mistargeting, in mutations in PH1, the Gly82Glu replacement is always found which both the normal and abnormal compartments require the on the background of the major AGXT allele. protein to contain specific, yet different, targeting information, is without parallel in human genetic disease. 2.4.2. Gly41Arg The Gly41Arg mutation has been found on the background 2.3. The normal AGXT gene of both the major and minor AGXT alleles [9,20]. However, there is some evidence to suggest that when present on the In humans, AGT is encoded by the AGXT gene, which minor allele (i.e. in the presence of the Pro11Leu polymor- consists of 11 exons spanning about 10 kB on chromosome phism) the effects of this mutation are more severe [8]. At least 2q37.3 [11]. There are two main allelic variants, the frequencies on the minor allele, Gly41Arg destabilizes AGT so that it is of which vary between different populations. The less common rapidly degraded. The protein that manages to escape “minor” AGXT allele differs from the more common “major” degradation aggregates into intra-peroxisomal core-like struc- allele by the presence of a 32C→T (Pro11Leu) replacement in tures [9]. Structural analysis shows that Gly41 sits right in the exon 1, a 1020A→G (Ile340Met) replacement in exon 10, and middle of the dimerization interface, making contact with its a 74 bp duplication in intron 1 [12,13]. In uncharacterized opposite number in the other subunit (see Fig. 2) [5]. There is European and North American populations, the minor allele has not enough room at this location to accommodate the large a frequency of about 15–20%. However, it rises to 28% in the arginine side-chain without causing major disruption to the Saami in Northern Sweden and falls to only 2–3% in some far interface, thereby inhibiting dimerization. The monomer has eastern populations in India, China and Japan [14,15]. Although greatly reduced specific catalytic activity [8] and is predicted to the Ile340Met replacement and the intron 1 duplication have no be unstable, one consequence of which would be aggregation obvious effects on the expression or properties of AGT, the due to artefactual polymerization. The presence of aggregates Pro11Leu polymorphism certainly does [8,12]. Firstly, it leads only within the peroxisomal matrix suggests that either to a decrease in AGT stability, which is manifested by slower aggregation is slow and does not occur until after peroxisomal dimerization, especially at elevated temperatures [8,16]. import or, perhaps more probably, aggregates forming in the Secondly, it decreases specific catalytic activity of recombinant cytosol, but not the peroxisomes, are rapidly degraded. purified AGT by a factor of two to three [8]. Thirdly, it redirects a small proportion (∼5%) of AGT away from peroxisomes 2.4.3. Ile244Thr towards the mitochondrial matrix [12]. Fourthly, it sensitises The second most common missense mutation is an Ile244Thr AGT to the untoward effects of many of the mutations found in replacement which segregates with the minor allele (i.e. with the PH1 that are predicted to be innocuous in its absence [8] (see Pro11Leu polymorphism). This mutation has an overall allelic below). frequency of about 6–9% in PH1 patients [21], but in some parts of Spain, such as La Gomera in the Canary Islands, it is 2.4. Mutations in the AGXT gene very much higher [22]. The combined presence of the polymorphism and the mutation appears to destabilize AGT Well over fifty mutations have been identified so far in the leading to its rapid degradation. Homozygous patients have AGXT gene, about half of which are point missense mutations little or no enzyme activity or immunoreactive protein. C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784 1779

Although it has not yet been found in patients liver in situ, AGT Table 1 containing Pro11Leu and Ile244Thr has been found to C-terminal tripeptides of some PTS1 mammalian enzymes aggregate in transfected COS cells [22], as it does when over- Species Enzyme expressed in E. coli [8]. AGT CAT UO ACO GO DAO Human and other primates KKL ANL SRL SKL SKI SHL 2.4.4. Gly170Arg Rat and other murine rodents NKL* ANL SRL SKL SKI SHL By far the most common mutation found so far in PH1 is a Guinea Pig HRL* ANL SKL SHL Gly170Arg replacement [12], with an estimated allelic Rabbit SQL* SHL frequency between 15 and 35%, depending on the particular Cat NKL* Dog HKL* ANL SKL* SKL SKI PHL* population [12,17]. This mutation segregates exclusively with Cow SKL* ANL SRL SKL SKI the minor AGXT allele, and is predicted to have no untoward Pig SRL SHL effects in the absence of the Pro11Leu polymorphism [8]. Koala SKL Gly170Arg and Pro11Leu, together, are responsible for the Number of different PTS1s 6 1 2 1 1 2 mistargeting of 90–95% of the AGT from its normal location in AGT, alanine:glyoxylate aminotransferase; CAT, catalase; UO, urate oxidase; the peroxisomes to the mitochondrial matrix [10].The ACO, acyl CoA oxidase; GO, glycolate oxidase; DAO, D-amino acid oxidase. mechanism by which this remarkable trafficking defect is C-terminal tripeptides different from that in the human sequence are indicated with asterisks. The subcellular distribution of all the peroxisomal enzymes, achieved highlights some of the most fundamental differences except AGT, is presumed to be invariantly peroxisomal (possibly with a between the peroxisomal and mitochondrial protein import proportion in the cytosol). AGT, on the other hand, is peroxisomal in human and pathways (see Section 3 below). some other primates, including the Great Apes, Old World Monkeys, and one New World monkey (saki monkey), guinea pig, rabbit and koala, mainly 3. Molecular and cell biological basis for AGT targeting mitochondrial in cat and dog, and both peroxisomal and mitochondrial in marmoset (a New World monkey), rat and other murine rodents. and mistargeting

3.1. Normal peroxisomal targeting of human AGT influence the efficiency of a PTS1, reference to the crystal structure of human AGT [5] (Fig. 2) shows that the PTS1A is AGT in humans and guinea pigs has been shown to be post- actually in close proximity to the C-terminal KKL. How the translationally imported into the peroxisomal matrix in a Pex5p- PTS1A might work is unclear, but it might involve increasing dependent and Pex7p-independent manner [6,23]. Like many the stability of the KKL-TPR interaction, or possibly directly other peroxisomal enzymes [24–26], AGT folds and dimerizes or indirectly allosterically modifying the TPR domain so that in the cytosol, and is imported into peroxisomes already in a KKL fits. fully active state [27]. Despite these similarities, it is clear that AGT is not a typical PTS1 protein. The C-terminal tripeptide in 3.2. Abnormal mitochondrial targeting of human AGT human AGT is KKL. Although this varies from the prototypical PTS1 SKL only at the − 3 position, all the evidence suggests The peroxisome-to-mitochondrion mistargeting of AGT in that this tripeptide should not be able to act as a PTS1. For PH1 requires the combined presence in cis of the common example, although KKL is necessary for the peroxisomal Pro11Leu polymorphism and the disease-specific Gly170Arg targeting of human AGT [6], it is insufficient to direct the mutation [12]. Surprisingly, however, the polymorphism peroxisomal import of reporter proteins, such as firefly appears to have a much greater effect on the properties of luciferase, jellyfish green fluorescent protein (GFP), and AGT than does the mutation (see above). The Pro11Leu bacterial chloramphenicol acetyltransferase [6,7,28]. KKL is replacement generates a functionally-weak N-terminal mito- even unable to direct the peroxisomal import of non- chondrial matrix targeting sequence (MTS), the effective mammalian AGT, such as that from the amphibian Xenopus strength of which is enhanced by the additional presence of laevis [29]. This difficulty is compounded by the observations the Gly170Arg replacement [16,32]. obtained from the crystal structure of the TPR domain of human MTSs are very different from PTS1s. They are much larger Pex5p that lysine at −3 could not possibly fit [30]. (typically 20–30 residues), N-terminal, and, although in These findings, together with the observations that the different mitochondrial proteins they show little or no sequence peroxisomal import of human AGT can actually be inhibited conservation, they all have a tendency to fold into positively- by over-expression of Pex5p [31] and that the C-terminal charged amphiphilic α-helices [33]. MTSs bind to a hydropho- tripeptides of mammalian AGTs are very poorly conserved bic pocket in the outer-membrane mitochondrial import compared to other PTS1 proteins (see Table 1), suggested that receptor TOM20 [34,35]. Also unlike PTS1s, MTSs are almost human AGT might contain additional peroxisomal targeting invariably cleaved after import [36]. The N-terminal sequence information that allows such an unconventional PTS1 to work. of human AGT (MASHKLLVTPPKALLKPLSI) has some A putative region of human AGT that might contain this similarities to a MTS, for example, absence of acidic residues, ancillary peroxisomal targeting information (termed PTS1A) and a good distribution of basic, neutral or hydrophobic, and has been identified in the small domain, 47–68 residues hydroxyl amino acids. However, the presence of two neigh- upstream of the C-terminus [7] (Fig. 2). Although further bouring helix breakers (i.e. prolines) at residues 10 and 11 is upstream than any other sequence currently known to likely to prevent this sequence from folding into an α-helix. A 1780 C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784 single Pro11Leu replacement is predicted to allow the N- dimerization long enough to allow the import of AGT into the terminus of human AGT to better adopt the conformation mitochondria mediated by the polymorphic MTS. required for a MTS, even though circular dichroism showed that Surprisingly, unlike almost all other proteins targeted to the the tendency of the purified 20-mer polypeptide to form a α- mitochondrial matrix, including AGT in many other species helix was hardly any different with or without the replacement. (see below), the polymorphic MTS of mistargeted human AGT When both prolines were replaced by leucines, the purified 20- is not cleaved after import. The polymorphic MTS is located in mer showed a strong tendency to form an α-helix [16]. the N-terminal segment which wraps over the surface of the The interesting thing about this polymorphic MTS (i.e. with companion subunit and is thought to be important in the process the single Pro11Leu replacement) is that it is only functionally of dimerization (see above and Fig. 2). The fact that it is not weak when attached to AGT, whereas it is functionally strong cleaved means that mistargeted AGT is still able to eventually when attached to GFP. However, when both prolines are fold and dimerize within the mitochondrial matrix, where it is replaced by leucine, it is strong irrespective of what it is able to acquire relatively normal catalytic activity, if not attached to [16]. The reason for this appears to be due to the fact metabolic efficiency. that AGT folds quickly and rapidly forms a stable dimer that This explanation for the peroxisome-to-mitochondrion does not readily exchange subunits [27]. This fully-folded mistargeting of AGT in PH1 highlights two of the fundamental dimeric conformation, although compatible with peroxisomal differences in the peroxisomal and mitochondrial protein import protein import (see above), is incompatible with mitochondrial pathways, namely the differences between the natures and protein import [37,38]. Therefore, despite the presence of an N- locations of the respective targeting sequences, and the different terminal MTS, most of the polymorphic AGT is actually conformational requirements of importable cargos. Interesting- targeted to peroxisomes due to the presence of the C-terminal ly, mistargeted AGT contains both a N-terminal MTS and a C- PTS1 (and PTS1A). In other words, polymorphic AGT still terminal PTS1 (+PTS1A). Yet, 90–95% of the enzyme ends up folds and dimerizes faster than mitochondrial import. This in the mitochondria rather than the peroxisomes. The reason for kinetic partitioning of polymorphic AGT and its perturbation this is unclear, but might be due to there being more due to the additional presence of the Gly170Arg mutation is mitochondria (and mitochondrial import receptors) than crucial to the understanding of the peroxisome-to-mitochondri- peroxisomes (and peroxisomal import receptors), or the on mistargeting of AGT in PH1 (Fig. 3). mitochondrial import process might be intrinsically faster than On its own, the disease-specific Gly170Arg mutation that of the peroxisomes, or the N-terminal location of the MTS provides no additional mitochondrial targeting information, might make it potentially available for operation before the C- does not directly interfere with peroxisomal targeting, and has terminal PTS1 has been synthesized. This apparent hierarchical no major effect on AGT stability or dimerization [8,27]. dominance of the MTS over the PTS1 is also found with more However, the combined effect of the Gly170Arg and Pro11Leu conventional MTSs, such as those in other mammalian AGTs replacements is to severely delay productive folding and (see below).

Fig. 3. Kinetic partitioning of AGT. Most of the newly-synthesized normal AGT encoded by the major and minor AGXT alleles folds, dimerizes and acquires catalytic activity rapidly in the cytosol before import into peroxisomes. About 5% of normal polymorphic (Pro11Leu) AGT and about 90% of mutant (Pro11Leu+Gly170Arg) AGT are imported into mitochondria before having a chance to fold and dimerize. After import into mitochondria, relatively normal folding and dimerization occurs. It is presumed that homologous sets of molecular chaperones mediate folding in the cytosol and mitochondrial matrix, but not the peroxisomes. Retardation of AGT folding and dimerization in the cytosol (*) leads to its redirection along the other pathways, including import into mitochondria (+) if the polymorphic (Pro11Leu) MTS is present. C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784 1781

Luckily, in the absence of the Gly170Arg mutation, the There is a good metabolic reason why the subcellular Pro11Leu replacement does not interfere enough with AGT distribution of AGT might be related to diet. In order to detoxify folding and dimerization to generate a functionally strong MTS. glyoxylate efficiently, AGT must be located at the site of If it did, then PH1 would have a frequency of 1 in 25 (the glyoxylate synthesis. This site is predicted to be related to diet. frequency of Pro11Leu homozygosity) rather than somewhere In herbivores, the main dietary precursor of glyoxylate is between 1 in 105 and 106. The same would apply if it were not thought to be glycolate [44], whereas in carnivores it is thought the case that mitochondria are only able to import unfolded or to be hydroxyproline [45]. Glycolate is converted to glyoxylate loosely-folded monomeric polypeptides [37], unlike peroxi- in the peroxisomes, catalysed by glycolate oxidase. On the other somes that seem to be able to import almost anything as long as hand, the last step in the conversion of hydroxyproline to it contains a PTS [39]. glyoxylate, catalysed by 4-hydroxy-2-ketoglutarate aldolase [46], occurs in the mitochondria. Thus, the best place for AGT 3.3. Evolutionary history of AGT targeting to be is in the peroxisomes in herbivores and mitochondria in carnivores (Fig. 4). The finding that AGT is mistargeted from peroxisomes to In all of the mammals studied so far, AGT is encoded by a mitochondria in a large subset of PH1 patients was surprising single copy gene that can encode a polypeptide varying from 392 not least because there are many mammals which have most of to 414 amino acids [4,11,23,29,47–50]. The archetypal their AGT in mitochondria anyway without apparently mammalian AGT gene, as found in the rat or marmoset, for succumbing to CaOx kidney stone disease [40,41]. example, has the potential to encode an N-terminal mitochon- For the vast majority of non-cytosolic eukaryotic proteins, drial matrix targeting sequence (MTS) of 22 amino acids, and an compartmentalisation is presumed to have occurred early in atypical C-terminal type 1 peroxisomal targeting sequence eukaryotic evolution, the consequence of this being that the (PTS) of three amino acids. There appears to be a hierarchical same protein has the same subcellular distribution in all dominance of the MTS over the PTS, so that if present together organisms. AGT, on the other hand, can be found in the former predominates [51]. The eventual localisation of AGT peroxisomes, mitochondria, peroxisomes and mitochondria, appears to be dependent on whether or not the MTS is contained peroxisomes and cytosol, and mitochondria and cytosol within the open reading frame, and therefore present in the [29,40,42]. During the evolution of mammals, and possibly encoded polypeptide. Expression of the MTS appears to be vertebrates in general, AGT targeting has been under the dependent on the relative use of two transcription start sites and influence of strong, yet episodic, dietary selection pressure. two in-frame translation start sites [52] (Fig. 5). In many species, Thus AGT tends to be mitochondrial in carnivores, peroxisomal including the rat and marmoset, both transcription and in herbivores, and both peroxisomal and mitochondrial in translation start sites have been maintained, so that two omnivores. The peroxisomal location of AGT in humans is transcripts and two polypeptides are produced. The longer, probably a hangover from our mainly herbivorous Great Ape containing the MTS and PTS is targeted mainly to the ancestry [43]. mitochondria, whereas the shorter containing only the PTS is

Fig. 4. Diet and the intracellular compartmentalization of AGT in mammals. The major precursor of glyoxylate in herbivores is glycolate which is converted to glyoxylate in the peroxisomes, catalysed by glycolate oxidase (GO). In carnivores, the major precursor is hydroxyproline, which is converted to glyoxylate by a number of steps, the last of which catalysed by 4-hydroxy-2-ketoglutarate aldolase (HKA) occurs in the mitochondria. Efficient detoxification of glyoxylate must occur at its site of synthesis. Therefore, alanine:glyoxylate aminotransferase (AGT) is best located in the peroxisomes in herbivores and the mitochondria in carnivores. LDH, lactate dehydrogenase. 1782 C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784

Fig. 5. Archetypal mammalian AGT gene. The figure shows a diagrammatic representation of the gene encoding AGT (AGXT) in mammals. The bars indicate the coding regions. The regions encoding the N-terminal mitochondrial matrix targeting sequence (MTS) is shown in blue, the C-terminal PTS1 in red, and the internal PTS1A in red cross-hatched. The numbers 1 and 2 indicate the two in-frame translation start sites 22 codons apart. The letters A and B indicate the locations of the two groups of transcription start sites. In most mammals, the MTS is 19–24 amino acids, the PTS1 3 amino acids, and in humans the PTS1A is ∼22 amino acids long. Transcription start site A is lost in guinea pig and site B is lost in cat. Translation start site 1 is lost in human, rabbit, guinea pig and saki monkey, but site 2 is never lost. A, B, 1, and 2 are all maintained in marmoset and rat. targeted only to the peroxisomes [48,53]. In some species, such above would suggest one possible benefit, but only for those as human, rabbit, guinea pig, and saki monkey, the MTS is individuals who eat a lot of meat, might be the presence of a permanently excluded from the open reading frame by the small amount of AGT in the mitochondria. If true then dietary evolutionary loss of the more 5′ translation start site lifestyle might have contributed to the differential evolution of [4,23,43,48]. In these animals, AGT is exclusively peroxisomal. this polymorphism in different human populations. In this In other species, such as the domestic cat, almost all nascent context, it is interesting to note that the population with the polypeptides contain a MTS due to the loss of the more 3′ highest frequency of Pro11Leu so far studied (i.e. 27% in the transcription start site [47]. AGT in the cat is 90–95% Saami) is also the population with by far the greatest proportion mitochondrial. In certain species, such as the giant panda, of meat in the diet. At the other extreme, populations in India, although AGT contains a MTS, it does not work very efficiently China and Japan have frequencies of only 2–3% [15]. due to the accumulation of interfering mutations that decrease its Monogenic CaOx kidney stone diseases, such as PH1 are net positive charge and weaken its potential to fold into an α- very rare, but multifactorial idiopathic CaOx kidney stone helix [41]. disease is very common [54,55]. We have speculated in the past The unique evolutionary history of AGT targeting might be, that the presence or absence of Pro11Leu polymorphism might at least in part, responsible for the unusual properties of this be a contributory factor in an individual's susceptibility to PTS1 enzyme, especially the poor conservation of its C- idiopathic CaOx kidney stones. The consequences of its terminus and, in some species at least, its lack of sufficiency dysfunction in PH1 clearly demonstrate that AGT is an [6,7]. Although the presence of the PTS1A might remove some important determinant of the rate of oxalate synthesis and of the selection pressure to conserve the PTS1 (see above), it is excretion, and elevated oxalate excretion is a well-known risk equally possible that during the evolution of vertebrates, or factor for CaOx stones [56]. Thus the polymorphism might be mammals in particular, periods of carnivory and strong beneficial to high meat-eaters, but neutral, or possibly mitochondrial targeting have diminished the need for efficient detrimental, to vegetarians. peroxisomal targeting, so that the PTS1 might have had to have evolved on more than one occasion. 5. Postscript

4. AGT and idiopathic calcium oxalate kidney stone disease AGT, or at least human AGT, fits well into a scholarly volume on peroxisomes. However, in a wider evolutionary As described above, the Pro11Leu polymorphism has context it could just as easily fit into a volume on mitochondria. significant effects on the properties of AGT, most of which The variable peroxisomal and mitochondrial targeting of AGT might be considered detrimental. Yet, in apparently normal during mammalian evolution, and the peroxisome-to-mitochon- European and North American populations its allelic frequency drion mistargeting of AGT in human hereditary kidney stone is 20% or more. In such populations, 4% of individuals would disease, place AGT in a unique position in the comparative be expected to be homozygous for Pro11Leu. This high study of peroxisomal and mitochondrial protein import. frequency of what might appear to be a detrimental polymor- Whether AGT is truly unique, or whether there are any other, phism suggests that under certain circumstances it might as yet unidentified, enzymes that behave in a similar fashion actually be beneficial. The evolutionary studies outlined remains to be seen. C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784 1783

Acknowledgements Allsop, Molecular characterization and clinical use of a polymorphic tandem repeat in an intron of the human alanine:glyoxylate aminotransferase gene, Hum. Genet. 94 (1994) 55–64. The work in my laboratory on AGT has been supported over [15] E.F. Caldwell, L.R. Mayor, M.G. Thomas, C.J. Danpure, Diet and the the years by grants from the Medical Research Council, the frequency of the alanine:glyoxylate aminotransferase Pro11Leu polymor- Wellcome Trust, the Leverhulme Trust, and the Oxalosis and phism in different human populations, Hum. Genet. 115 (2004) 504–509. Hyperoxaluria Foundation. I thank Dr. Pia Huber for critically [16] M.J. Lumb, A.F. Drake, C.J. Danpure, Effect of N-terminal alpha helix reading the manuscript and providing helpful suggestions. formation on the dimerization and intracellular targeting of alanine: glyoxylate aminotransferase, J. Biol. Chem. 274 (1999) 20587–20596. [17] M.B. Coulter-Mackie, G. Rumsby, Genetic heterogeneity in primary References hyperoxaluria type 1: impact on diagnosis, Mol. Genet. Metab. 83 (2004) 38–46. [1] C.J. Danpure, P.R. Jennings, Peroxisomal alanine:glyoxylate aminotrans- [18] P.E. Purdue, M.J. Lumb, J. Allsop, Y. Minatogawa, C.J. Danpure, A ferase deficiency in primary hyperoxaluria type I, FEBS Lett. 201 (1986) glycine-to-glutamate substitution abolishes alanine:glyoxylate aminotrans- 20–24. ferase catalytic activity in a subset of patients with primary hyperoxaluria [2] C.J. Danpure, Primary Hyperoxaluria, in: C.R. Scriver, A.L. Beaudet, W.S. type 1, Genomics 13 (1992) 215–218. Sly, D. Valle, B. Childs, K.W. Kinzler, B. Vogelstein (Eds.), The Metabolic [19] C.J. Danpure, P.R. Jennings, Further studies on the activity and subcellular and Molecular Bases of Inherited Disease, 8th edition, vol. II, McGraw- distribution of alanine:glyoxylate aminotransferase in the livers of patients Hill, New York, 2001, pp. 3323–3367. with primary hyperoxaluria type 1, Clin. Sci. 75 (1988) 315–322. [3] C.J. Danpure, G. Rumsby, Molecular aetiology of primary hyperoxaluria [20] D. Pirulli, D. Puzzer, L. Ferri, S. Crovella, A. Amoroso, C. Ferrettini, M. and its implications for clinical management, Experts Rev. Mol. Med. 2004 Marangella, G. Mazzola, F. Florian, Molecular analysis of hyperoxaluria (2004) 1–16. type 1 in Italian patients reveals eight new mutations in the alanine: [4] Y. Takada, N. Kaneko, H. Esumi, P.E. Purdue, C.J. Danpure, Human glyoxylate aminotransferase gene, Hum. Genet. 104 (1999) 523–525. peroxisomal L-alanine: glyoxylate aminotransferase. Evolutionary loss of a [21] C. von Schnakenburg, G. Rumsby, Primary hyperoxaluria type 1: a cluster of mitochondrial targeting signal by point mutation of the initiation codon, new mutations in exon 7 of the AGXT gene, J. Med. Genet. 34 (1997) Biochem. J. 268 (1990) 517–520. 489–492. [5] X. Zhang, S.M. Roe, Y. Hou, M. Bartlam, Z. Rao, L.H. Pearl, C.J. [22] A. Santana, E. Salido, A. Torres, L.J. Shapiro, Primary hyperoxaluria type Danpure, Crystal structure of alanine:glyoxylate aminotransferase and the 1 in the Canary Islands: a conformational disease due to I244T mutation in relationship between genotype and enzymatic phenotype in primary the P11L-containing alanine:glyoxylate aminotransferase, Proc. Natl. hyperoxaluria type 1, J. Mol. Biol. 331 (2003) 643–652. Acad. Sci. U. S. A. 100 (2003) 7277–7282. [6] A. Motley, M.J. Lumb, P.B. Oatey, P.R. Jennings, P.A. De Zoysa, R.J. [23] G.M. Birdsey, C.J. Danpure, Evolution of alanine:glyoxylate aminotrans- Wanders, H.F. Tabak, C.J. Danpure, Mammalian alanine:glyoxylate ferase intracellular targeting. Structural and functional analysis of the aminotransferase 1 is imported into peroxisomes via the PTS1 transloca- guinea pig gene, Biochem. J. 331 (1998) 49–60. tion pathway. Increased degeneracy and context specificity of the [24] J.R. Glover, D.W. Andrews, R.A. Rachubinski, Saccharomyces cerevisiae mammalian PTS1 motif and implications for the peroxisome-to-mito- peroxisomal thiolase is imported as a dimer, Proc. Natl. Acad. Sci. U. S. A. chondrion mistargeting of AGT in primary hyperoxaluria type 1, J. Cell 91 (1994) 10541–10545. Biol. 131 (1995) 95–109. [25] J.A. McNew, J.M. Goodman, An oligomeric protein is imported into [7] P.A. Huber, G.M. Birdsey, M.J. Lumb, D.T. Prowse, T.J. Perkins, D.R. peroxisomes in vivo, J. Cell Biol. 127 (1994) 1245–1257. Knight, C.J. Danpure, Peroxisomal import of human alanine:glyoxylate [26] C.R. Flynn, R.T. Mullen, R.N. Trelease, Mutational analyses of a type 2 aminotransferase requires ancillary targeting information remote from its C peroxisomal targeting signal that is capable of directing oligomeric protein terminus, J. Biol. Chem. 280 (2005) 27111–27120. import into tobacco BY-2 glyoxysomes, Plant J. 16 (1998) 709–720. [8] M.J. Lumb, C.J. Danpure, Functional synergism between the most [27] J.M. Leiper, P.B. Oatey, C.J. Danpure, Inhibition of alanine:glyoxylate common polymorphism in human alanine:glyoxylate aminotransferase aminotransferase dimerization is a pre-requisite for its peroxisome-to- and four of the most common disease-causing mutations, J. Biol. Chem. mitochondrion mistargeting in primary hyperoxaluria type 1, J. Cell Biol. 275 (2000) 36415–36422. 135 (1996) 939–951. [9] C.J. Danpure, P.E. Purdue, P. Fryer, S. Griffiths, J. Allsop, M.J. Lumb, K. [28] P.B. Oatey, M.J. Lumb, P.R. Jennings, C.J. Danpure, Context dependency M. Guttridge, P.R. Jennings, J.I. Scheinman, S.M. Mauer, N.O. Davidson, of the PTS1 motif in human alanine:glyoxylate aminotransferase 1, Ann. Enzymological and mutational analysis of a complex primary hyperox- N. Y. Acad. Sci. 804 (1996) 652–653. aluria type 1 phenotype involving alanine: glyoxylate aminotransferase [29] J.D. Holbrook, C.J. Danpure, Molecular basis for the dual mitochondrial peroxisome-to-mitochondrion mistargeting and intraperoxisomal aggrega- and cytosolic localization of alanine:glyoxylate aminotransferase in tion, Am. J. Hum. Genet. 53 (1993) 417–432. amphibian liver cells, J. Biol. Chem. 277 (2002) 2336–2344. [10] C.J. Danpure, P.J. Cooper, P.J. Wise, P.R. Jennings, An enzyme trafficking [30] G.J. Gatto, B.V. Geisbrecht, S.J. Gould, J.M. Berg, Peroxisomal targeting defect in two patients with primary hyperoxaluria type 1: peroxisomal signal-1 recognition by the TPR domains of human PEX5, Nat. Struct. alanine/glyoxylate aminotransferase rerouted to mitochondria, J. Cell Biol. Biol. 7 (2000) 1091–1095. 108 (1989) 1345–1352. [31] T.G. Knott, G.M. Birdsey, K.E. Sinclair, I.M. Gallagher, P.E. Purdue, C.J. [11] P.E. Purdue, M.J. Lumb, M. Fox, G. Griffo, C. Hamon Benais, S. Povey, C. Danpure, The peroxisomal targeting sequence type 1 receptor, Pex5p, and J. Danpure, Characterization and chromosomal mapping of a genomic the peroxisomal import efficiency of alanine:glyoxylate aminotransferase, clone encoding human alanine:glyoxylate aminotransferase, Genomics 10 Biochem. J. 352 (2000) 409–418. (1991) 34–42. [32] P.E. Purdue, J. Allsop, G. Isaya, L.E. Rosenberg, C.J. Danpure, [12] P.E. Purdue, Y. Takada, C.J. Danpure, Identification of mutations Mistargeting of peroxisomal L-alanine:glyoxylate aminotransferase to associated with peroxisome-to-mitochondrion mistargeting of alanine/ mitochondria in primary hyperoxaluria patients depends upon activation of glyoxylate aminotransferase in primary hyperoxaluria type 1, J. Cell Biol. a cryptic mitochondrial targeting sequence by a point mutation, Proc. Natl. 111 (1990) 2341–2351. Acad. Sci. U. S. A. 88 (1991) 10900–10904. [13] P.E. Purdue, M.J. Lumb, J. Allsop, C.J. Danpure, An intronic duplication [33] G. von Heijne, Mitochondrial targeting sequences may form amphiphilic in the alanine:glyoxylate aminotransferase gene facilitates identification of helices, EMBO J. 5 (1986) 1335–1342. mutations in compound heterozygote patients with primary hyperoxaluria [34] Y. Abe, T. Shodai, T. Muto, K. Mihara, H. Torii, S. Nishikawa, T. Endo, D. type 1, Hum. Genet. 87 (1991) 394–396. Kohda, Structural basis of presequence recognition by the mitochondrial [14] C.J. Danpure, G.M. Birdsey, G. Rumsby, M.J. Lumb, P.E. Purdue, J. protein import receptor Tom20, Cell 100 (2000) 551–560. 1784 C.J. Danpure / Biochimica et Biophysica Acta 1763 (2006) 1776–1784

[35] T. Muto, T. Obita, Y. Abe, T. Shodai, T. Endo, D. Kohda, NMR [46] T. Takayama, K. Fujita, K. Suzuki, M. Sakaguchi, M. Fujie, E. Nagai, S. identification of the Tom20 binding segment in mitochondrial prese- Watanabe, A. Ichiyama, Y. Ogawa, Control of oxalate formation from L- quences, J. Mol. Biol. 306 (2001) 137–143. hydroxyproline in liver mitochondria, J. Am. Soc. Nephrol. 14 (2003) [36] K.N. Truscott, K. Brandner, N. Pfanner, Mechanisms of protein import into 939–946. mitochondria, Curr. Biol. 13 (2003) R326–R337. [47] M.J. Lumb, P.E. Purdue, C.J. Danpure, Molecular evolution of alanine/ [37] M. Eilers, G. Schatz, Protein unfolding and the energetics of protein glyoxylate aminotransferase 1 intracellular targeting: analysis of the feline translocation across biological membranes, Cell 52 (1988) 481–483. gene, Eur. J. Biochem. 221 (1994) 53–62. [38] M. Schleyer, W. Neupert, Transport of proteins into mitochondria: [48] P.E. Purdue, M.J. Lumb, C.J. Danpure, Molecular evolution of translocational intermediates spanning contact sites between outer and alanine:glyoxylate aminotransferase 1 intracellular targeting. Analysis inner membranes, Cell 43 (1985) 339–350. of the marmoset and rabbit genes, Eur. J. Biochem. 207 (1992) [39] P.A. Walton, P.E. Hill, S. Subramani, Import of stably folded proteins into 757–766. peroxisomes, Mol. Biol. Cell 6 (1995) 675–683. [49] M. Mori, T. Oda, K. Nishiyama, T. Serikawa, J. Yamada, A. Ichiyama, A [40] G.M. Birdsey, J. Lewin, J.D. Holbrook, V.R. Simpson, A.A. Cunningham, single serine:pyruvate aminotransferase gene on rat chromosome 9q34– C.J. Danpure, A comparative analysis of the evolutionary relationship q36, Genomics 13 (1992) 686–689. between diet and enzyme targeting in bats, marsupials and other mammals, [50] T. Oda, H. Miyajima, Y. Suzuki, A. Ichiyama, Nucleotide sequence of the Proc. R. Soc. B 272 (2005) 833–840. cDNA encoding the precursor for mitochondrial serine:pyruvate amino- [41] G.M. Birdsey, J. Lewin, A. Cunningham, M.W. Bruford, C.J. Danpure, transferase of rat liver, Eur. J. Biochem. 168 (1987) 537–542. Differential enzyme targeting as an evolutionary adaptation to herbivory in [51] P.B. Oatey, M.J. Lumb, C.J. Danpure, Molecular basis of the variable Carnivora, Mol. Biol. Evol. 21 (2004) 632–646. mitochondrial and peroxisomal localisation of alanine:glyoxylate amino- [42] C.J. Danpure, P. Fryer, P.R. Jennings, J. Allsop, S. Griffiths, A. transferase, Eur. J. Biochem. 241 (1996) 374–385. Cunningham, Evolution of alanine:glyoxylate aminotransferase 1 perox- [52] C.J. Danpure, Variable peroxisomal and mitochondrial targeting of isomal and mitochondrial targeting. A survey of its subcellular distribution alanine: glyoxylate aminotransferase in mammalian evolution and disease, in the livers of various representatives of the classes Mammalia, Aves and Bioessays 19 (1997) 317–326. Amphibia, Eur. J. Cell Biol. 64 (1994) 295–313. [53] T. Oda, T. Funai, A. Ichiyama, Generation from a single gene of [43] J.D. Holbrook, G.M. Birdsey, Z. Yang, M.W. Bruford, C.J. Danpure, two mRNAs that encode the mitochondrial and peroxisomal serine: Molecular adaptation of alanine:glyoxylate aminotransferase targeting in pyruvate aminotransferase of rat liver, J. Biol. Chem. 265 (1990) primates, Mol. Biol. Evol. 17 (2000) 387–400. 7513–7519. [44] T. Noguchi, Amino acid metabolism in animal peroxisomes, in: H.D. [54] J.R. Asplin, Hyperoxaluric calcium nephrolithiasis, Endocrinol. Metab. Fahimi, H. Sies (Eds.), Peroxisomes in Biology and Medicine, Springer- Clin. North Am. 31 (2002) 927–949. Verlag, Berlin, 1987, pp. 234–243. [55] C.J. Danpure, Genetic disorders and urolithiasis, in: M.I. Resnick (Ed.), [45] A. Ichiyama, H.H. Xue, T. Oda, C. Uchida, T. Sugiyama, E. Maeda-Nakai, K. Urolithiasis, The Urologic Clinica of North America, vol. 27, W.B. Sato, E. Nagai, S. Watanabe, T. Takayama, Oxalate synthesis in mammals: Saunders, Philadelphia, 2000, pp. 287–299. properties and subcellular distribution of serine:pyruvate/alanine:glyoxylate [56] W.G. Robertson, M. Peacock, The cause of idiopathic calcium stone aminotransferase in the liver, Mol. Urol. 4 (2000) 333–340. disease: hypercalciuria or hyperoxaluria? Nephron 26 (1980) 105–110.