Solute Traffic Across the Mammalian Peroxisomal Membrane—The Role of Pxmp2

A 517 OULU 2008 A 517

UNIVERSITY OF OULU P.O.B. 7500 FI-90014 UNIVERSITY OF OULU FINLAND ACTA UNIVERSITATISUNIVERSITATIS OULUENSISOULUENSIS ACTA UNIVERSITATIS OULUENSIS ACTAACTA

SERIES EDITORS SCIENTIAESCIENTIAEA A RERUMRERUM Aare Rokka NATURALIUMNATURALIUM

ASCIENTIAE RERUM NATURALIUM Aare Rokka Professor Mikko Siponen SOLUTE TRAFFIC ACROSS BHUMANIORA THE MAMMALIAN University Lecturer Elise Kärkkäinen CTECHNICA PEROXISOMAL MEMBRANE Professor Hannu Heusala —THE ROLE OF PXMP2 DMEDICA Professor Olli Vuolteenaho ESCIENTIAE RERUM SOCIALIUM Senior Researcher Eila Estola FSCRIPTA ACADEMICA Information officer Tiina Pistokoski GOECONOMICA University Lecturer Seppo Eriksson

EDITOR IN CHIEF Professor Olli Vuolteenaho PUBLICATIONS EDITOR

Publications Editor Kirsti Nurkkala FACULTY OF SCIENCE, DEPARTMENT OF BIOCHEMISTRY, UNIVERSITY OF OULU; ISBN 978-951-42-8971-2 (Paperback) BIOCENTER OULU, ISBN 978-951-42-8972-9 (PDF) UNIVERSITY OF OULU ISSN 0355-3191 (Print) ISSN 1796-220X (Online)

ACTA UNIVERSITATIS OULUENSIS A Scientiae Rerum Naturalium 517

AARE ROKKA

SOLUTE TRAFFIC ACROSS THE MAMMALIAN PEROXISOMAL MEMBRANE—THE ROLE OF PXMP2

Academic dissertation to be presented, with the assent of the Faculty of Science of the University of Oulu, for public defence in Raahensali (Auditorium L10), Linnanmaa, on December 12th, 2008, at 12 noon

Supervised by Professor Kalervo Hiltunen Professor Vasily Antonenkov

Reviewed by Doctor Cécile Brocard Doctor Tiina Kotti

ISBN 978-951-42-8971-2 (Paperback) ISBN 978-951-42-8972-9 (PDF) http://herkules.oulu.fi/isbn9789514289729/ ISSN 0355-3191 (Printed) ISSN 1796-220X (Online) http://herkules.oulu.fi/issn03553191/

Cover design Raimo Ahonen

OULU UNIVERSITY PRESS OULU 2008 Rokka, Aare, Solute traffic across the mammalian peroxisomal membrane—the role of Pxmp2 Faculty of Science, Department of Biochemistry, University of Oulu, P.O.Box 3000, FI-90014 University of Oulu, Finland; Biocenter Oulu, University of Oulu, P.O. Box 5000, FI-90014 University of Oulu, Finland Acta Univ. Oul. A 517, 2008 Oulu, Finland

Abstract Peroxisomes are small oxidative organelles found in all eukaryotes. They contain a matrix which is surrounded by a single membrane and consists mainly of soluble proteins. Peroxisomal enzymes are involved in a broad spectrum of metabolic pathways including conversion of lipids, amino- and hydroxyacids, purines and reactive oxygen species. The carbon fluxes through peroxisomal pathways require a continuous metabolite crossing of the peroxisomal membrane. A long-standing and still unresolved problem of the physiology of mammalian peroxisomes is the role of the membrane of these organelles as a permeability barrier to solute molecules. In this study, we have shown that the peroxisomal membrane represents a type of biomembrane where channel-forming proteins coexist with solute transporters. Disruption of the mouse Pxmp2 gene, encoding the peroxisomal integral membrane protein Pxmp2 also known as PMP22, leads to partial restriction of peroxisomal membrane permeability to solutes in vitro and in vivo. Multiple-channel recording of liver peroxisomal preparations revealed that the channel-forming components with a conductance of 1.3 nS in 1.0 M KCl were lost in Pxmp2-/- mice. The channel- forming properties of Pxmp2 were confirmed with recombinant protein expressed in insect cells and with native Pxmp2 purified from mouse liver. The Pxmp2 channel, with an estimated diameter of 1.4 nm, shows weak cation selectivity and no voltage dependence. The long-lasting open states of the channel indicate its functional role as a protein forming a general diffusion pore in the membrane. Hence, Pxmp2 is the first peroxisomal pore-forming protein identified, and its existence suggests that the mammalian peroxisomal membrane is permeable to small solutes, while transfer of bulky metabolites, e.g., cofactors (NAD/H, NADP/H, and CoA) and ATP, requires specific transporters. In addition, the phenotypic characterisation of Pxmp2-/- mice has revealed a role for Pxmp2 during the development of the epithelia in the mammary glands of female mice. The disruption of Pxmp2 leads to the impairment of ductal outgrowth of mammary glands at puberty, which is followed by the inability of Pxmp2-/- mice to nurse their offspring.

Keywords: channel-forming proteins, mammary gland, peroxisomal membrane, Pxmp2

Acknowledgements

This work was carried out at the Department of Biochemistry and Biocenter Oulu at the University of Oulu, Finland. I am deeply grateful to my supervisor Professor Kalervo Hiltunen for guiding me through the years to finally reach this goal. I would also like to thank my supervisor Professor Vasily Antonenkov for everything. This work would have never been accomplished without your effort and expertise. I owe my sincere thanks to Dr. Raija Soininen for her contribution to this work in generating and analysing the knock-out mouse model. I would like to thank the group leaders and staff at the department for creating a well functioning environment for research. I thank Dr. Tiina Kotti and Dr. Cécile Brocard for valuable comments on the thesis and Dr. Deborah Kaska for careful revision of the language. My deepest gratitude goes to all the people who have contributed to this work over the years. Especially, Dr. Raija Sormunen, Dr. Helmut Pospiech, Dr. Werner Schmitz, Dr. Päivi Pirilä, Dr. Ulrich Bergmann, Professor Roland Benz and Professor Matti Weckström are each acknowledged for providing their special expertise to this project. I am especially grateful to Miia Vapola whose contribution has been crucial for the finishing of this thesis work. I wish to thank Kalle, Jukka, Ilkka, Seppo, Dmitry, Tomi, Anu, Mari, Juha, Tiina, Tiila, Antti H., Hanna, Eija, Kristian, Alex, Miia, Samuli, François, Zhi-Jun, Kaija, Tatu, Maija, Antti I., Fumi, Tonja and all the other present and former members of the KH-group for companionship and support. Dr. Ari-Pekka Kvist deserves my thanks for helping me with computers, statistical analyses and other things. My special thanks go to Maija Risteli for guidance with insect cells. I thank Terttu Keskitalo, Hilkka Lehto and Katri Näppä for keeping my mice happy, Leena Ollitervo, Piia Mäkelä, Marika Kamps, Ville Ratas and Johanna Moilanen for excellent technical help, Anneli Kaattari, Pia Askonen and Tuula Koret for taking care of the official matters and Maire Jarva, Petra Heikkinen, Kyösti Keränen, Jaakko Keskitalo, Jari Karjalainen, Maila Konttila and Pirjo Mustaniemi for making everything work. The members of TOKU and Shellfish Graveyard are acknowledged for providing interesting experiences outside work. Finally, I want to thank my family members and parent in-laws. My warmest thanks belong to my wife Anu, who has managed to keep me on track to reach this destination. Without you it would have been impossible.

5 This work was supported by the Academy of Finland, Sigrid Juselius Foundation, Jenny and Antti Wihuri Foundation, Scientific Foundation of Instrumentarium and the FPG European Union Project “Peroxisome” (LSHG-CT- 2004-512018).

Oulu, November 2008 Aare Rokka

6 Abbreviations

ABC ATP-binding cassette ACOX acyl-CoA oxidase ALDP adrenoleukodystrophy protein ALDR adrenoleukodystrophy related protein AMP adenosine monophosphate ATP adenosine triphosphate cDNA complementary deoxyribonucleic acid CoA coenzyme A DHAP dihydroxyacetone phosphate DHCA 3α, 7α-dihydroxycholestanoic acid DNA deoxyribonucleic acid ER endoplasmic reticulum ES embryonic stem FAD flavine adenine dinucleotide kb kilobase kDa kilodalton LACS long-chain fatty acyl-CoA synthetase lacZ gene encoding β-galactosidase protein LCFA long-chain fatty acids LDH lactate dehydrogenase MCFA medium-chain fatty acids MCT monocarboxylate transporter MFE multifunctional enzyme NAD+ nicotinamide adenine dinucleotide NADP+ nicotinamide adenine dinucleotide phosphate PCR polymerase chain reaction PEX peroxin PGK phosphoglycerine kinase PMP peroxisomal membrane protein Pol DNA polymerase PTS peroxisomal targeting signal Pxmp2 22 kDa peroxisomal membrane protein RNA ribonucleic acid ROS reactive oxygen species SDS-PAGE sodium dodecyl sulphate polyacrylamide gel electrophoresis

7 THCA 3α, 7α, 12α-trihydroxycholestanoic acid UOX uric acid oxidase VDAC voltage-dependent anion channel VLACS very long-chain fatty acyl-CoA synthetase VLCFA very long-chain fatty acids X-ALD X-linked adrenoleukodystrophy

8 Contents

Abstract Acknowledgements 5 Abbreviations 7 Contents 9 1 Introduction 11 2 Review of the literature 13 2.1 Peroxisomes ...... 13 2.2 Peroxisomal biogenesis...... 13 2.3 Protein import into peroxisomes ...... 14 2.3.1 Matrix protein import ...... 14 2.3.2 Membrane protein import...... 15 2.4 Peroxisomal metabolism...... 16 2.4.1 β-Oxidation...... 16 2.4.2 α-Oxidation...... 21 2.4.3 Polyamine oxidation...... 23 2.4.4 Ether phospholipid biosynthesis...... 24 2.4.5 Metabolism of amino acids and glyoxylate...... 25 2.4.6 Metabolism of reactive oxygen species...... 26 2.4.7 Purine catabolism ...... 26 2.5 Transport of metabolites across the peroxisomal membrane ...... 27 2.5.1 Permeability properties of the peroxisomal membrane ...... 27 2.5.2 Peroxisomal pH, current view ...... 28 2.5.3 Two channel concept ...... 29 2.5.4 Channels in the peroxisomal membrane...... 29 2.5.5 Peroxisomal membrane transporters ...... 30 2.5.6 Transport of reducing equivalents by shuttle mechanisms...... 34 2.6 Peroxisomal membrane proteins...... 35 2.6.1 PMP22 protein family ...... 35 3 Aims of the study 39 4 Materials and methods 41 4.1 Generation of Pxmp2−/− mice...... 41 4.2 Southern and Northern analyses...... 42 4.3 Quantitative Real-Time PCR ...... 42 4.4 Analyses of body fluids of Pxmp2−/− mice...... 43 4.5 Blood pressure telemetry ...... 43

9 4.6 Histological analyses...... 44 4.7 Whole-mount analysis...... 44 4.8 Immunohistochemistry...... 44 4.9 Subcellular fractionation and isolation of mouse liver peroxisomes...... 45 4.10 Protein analysis and immunochemistry...... 46 4.11 Enzyme assays ...... 47 4.12 Dietary studies...... 48 4.13 Production of recombinant Pxmp2 in insect cells...... 48 4.14 Purification of native Pxmp2 protein from mouse liver peroxisomes ...... 49 4.15 Reconstitution assay into planar lipid bilayer ...... 50 4.16 Statistical analyses...... 51 5 Results 53 5.1 Channel-forming activity in the mouse peroxisomal membrane...... 53 5.1.1 Characterisation of the purified peroxisomal fraction ...... 53 5.1.2 Detection of channel-forming activity...... 56 5.1.3 Properties of the peroxisomal channels ...... 58 5.2 Pxmp2 functions as a channel in the mammalian peroxisomal membrane...... 59 5.2.1 Studies with Pxmp2-deficient mice ...... 59 5.2.2 Channel-forming activity of recombinant Pxmp2 ...... 67 5.2.3 Purification and characterisation of native Pxmp2...... 70 5.2.4 Channel-forming activity of purified native Pxmp2...... 70 5.2.5 The diameter of the pore of the Pxmp2 channel...... 74 5.2.6 Multiple-channel analysis of purified Pxmp2 using different organic anions as electrolytes ...... 77 5.3 Phenotypic analysis of Pxmp2−/− mice ...... 80 6 Discussion 97 6.1 Pore forming properties of the peroxisomal membrane proteins ...... 97 6.2 The function of Pxmp2 as a peroxisomal membrane channel...... 99 6.3 The phenotype of Pxmp2-deficient mice ...... 102 6.4 Future prospects ...... 104 7 Conclusions 107 References 109

10 1 Introduction

Peroxisomes are organelles that are found in almost all eukaryotic cells and harbour a wide range of metabolic functions. They have a single membrane surrounding an interior containing over 60 specialized enzymes catalysing, for example, the pathways of α- and β-oxidation of fatty acids, biosynthesis of ether phospholipids and bile acids and degradation of purines and polyamines. The peroxisomal membrane possesses various mechanisms for metabolite transport to facilitate these reactions. A wide range of proteins are known to reside in this membrane and these proteins are involved in peroxisomal biogenesis, i.e. they participate in the import of peroxisomal matrix- and membrane proteins. Peroxisomal membrane proteins are also involved in the transport of ATP and fatty acids across the peroxisomal membrane. However, there are still many open questions concerning the permeability properties of the peroxisomal membrane. In the present study a 22 kDa peroxisomal membrane protein, Pxmp2, has been characterised in mice as a non-specific channel for small metabolites in the peroxisomal membrane. The function of Pxmp2 was identified by using Pxmp2- deficient mice, recombinant Pxmp2 and purified native Pxmp2 protein. The results indicate that the peroxisomal membrane represents a new type of biomembranes, where non-selective channels co-exist with solute transporters.

12 2 Review of the literature

2.1 Peroxisomes

Peroxisomes, which are present in nearly all eukaryotic cells, are ubiquitous organelles characterised by a single limiting membrane and a finely granular matrix. These organelles range in size from 0.1 to 1µm in diameter (De Duve & Baudhuin 1966). The term peroxisome was introduced after the discovery that they contain H2O2-producing oxidases as well as catalase, an H2O2-degrading enzyme (De Duve & Baudhuin 1966). Since then mammalian peroxisomes have been shown to participate in various metabolic pathways including the oxidation of fatty acids, biosynthesis of ether phospholipids and bile acids, catabolism of D-amino acids, polyamines and purines, and metabolism of glyoxylate and reactive oxygen species (ROS)(Wanders & Waterham 2006a). The importance of peroxisomes for humans has been exemplified by the existence of several disorders caused by deficiency in peroxisomal functions, such as Zellweger syndrome, neonatal adrenoleukodystrophy, rhizomelic chondrodysplasia punctata, X-linked adrenoleukodystrophy and primary hyperoxaluria type 1 (Steinberg et al. 2006, Wanders & Waterham 2006b).

2.2 Peroxisomal biogenesis

The origin of peroxisomes has been matter of debate for several decades. The involvement of endoplasmic reticulum (ER) was proposed by suggesting that all the matrix and membrane proteins are synthesised on the ER and are then sequestered into expanding vesicles budding from the reticulum to produce functional peroxisomes (Beevers 1979). This view has been questioned later and a new model has been based on the observations that peroxisomal proteins are not synthesised on ribosomes bound to the ER but instead on free polyribosomes in the cytosol (Lazarow & Fujiki 1985). In this model peroxisomes are believed to increase in size by the import of proteins from the cytosol and to proliferate by division while the ER serves as a source for membrane lipids (Lazarow 2003). According to this concept peroxisomes are considered to represent structurally similar, homogenous and singular cellular compartments. However, it has been shown that there is also a heterogeneous population of peroxisomes differing in size, protein content and import capabilities within a

13 single cell, and that these diverse peroxisomal subcompartments are related to each other and participate in a multistep process leading to assembly of mature peroxisomes (Titorenko & Rachubinski 2001a). In the yeast Yarrowia lipolytica early peroxisomal precursors fuse with each other and, by means of a stepwise import of matrix proteins and the uptake of phospholipids, these intermediate organelles are converted into mature peroxisomes (Titorenko et al. 2000). In mammalian cells the earliest form of peroxisomal precursor, the so-called preperoxisomal vesicle, receives two peroxisomal membrane proteins (PMPs) Pex3 and Pex16, in a Pex19 dependent manner (South & Gould 1999, Gould & Valle 2000). Peroxins (Pex-proteins) are defined as proteins required for various aspects of peroxisome biogenesis: peroxisome growth, including import of peroxisomal matrix proteins and assembly of the peroxisomal membrane, as well as proliferation and inheritance of the organelles (Titorenko & Rachubinski 2001b). After obtaining an additional subset of membrane proteins the nascent peroxisome is converted to an early peroxisomal precursor, which grows owing to the import of matrix proteins from the cytosol (South & Gould 1999, Gould & Valle 2000). It has been suggested that the peroxisomal endomembrane system might originate from the ER and that the first subset of peroxisomal membrane proteins are targeted to the ER from which the preperoxisomal template emerges by budding (Titorenko & Mullen 2006).

2.3 Protein import into peroxisomes

The import of proteins into the ER or mitochondria is known to proceed before protein folding and oligomerisation (for review, see Wickner & Schekman 2005, MacKenzie & Payne 2007). In contrast, in the case of peroxisomes, fully folded and oligomeric proteins are imported into the matrix of the organelle (Walton et al. 1995).

2.3.1 Matrix protein import

The peroxisomal matrix proteins are synthesised on free polyribosomes in the cytoplasm and directed to the particles by two distinct peroxisomal targeting signals, PTS1 and PTS2 (Holroyd & Erdmann 2001). PTS1 was identified as the tripeptide Serine-Lysine-Leucine (-SKL) at the extreme C-terminal end of the protein (Gould et al. 1989). The definition of the PTS1-consensus sequence was later extended to (S/A/C)-(K/R/H)-(L/M) (Lametschwandtner et al. 1998). PTS2

14 is located near the N-terminus and has the general consensus sequence

(R/K)(L/V/I)X5(Q/H)(L/A) where X is any amino acid. This sequence is present only in a limited number of matrix proteins (Legakis & Terlecky 2001). Most matrix proteins contain the PTS1, which is recognized in the cytosol by the Pex5 protein, serving as a PTS1 receptor (Hettema et al. 1999). PTS2 containing matrix proteins are recognized by cytosolic Pex7 protein (Marzioch et al. 1994, Hettema et al. 1999). The PTS2 import pathway is believed to require additional PTS2 co-receptors comprising a group of species- specific proteins that includes Pex5, Pex18, Pex20 and Pex21 (Schliebs & Kunau 2006). Recently a transient pore model was proposed (Erdmann & Schliebs 2005) to describe the mechanism of matrix protein import into peroxisomes. In this model the complex consisting of a Pex5 protein and a PTS1-containing protein is directed from the cytosol to the peroxisomal membrane where it interacts with a docking complex constructed from Pex17, Pex13 and Pex14 proteins. Pex5 is believed to oligomerise forming a pore in the peroxisomal membrane (Erdmann & Schliebs 2005). The PTS1-cargo is released through this pore into the peroxisomal matrix with the help of the intraperoxisomal Pex8 protein. After that the pore closes and the Pex5 proteins dissociate from the membrane for another round of transport. The disassembly and recycling of Pex5 is believed to be facilitated by a complex of RING-finger motif-containing proteins Pex2, Pex10 and Pex12. The disassembly of Pex5 is also facilitated by an ubiquitin conjugating enzyme Pex4 that is anchored to the peroxisomal membrane by Pex22, and ATPases Pex1 and Pex6 that are anchored to the peroxisomal membrane by Pex26 (Erdmann & Schliebs 2005).

2.3.2 Membrane protein import

As in the case of matrix proteins, most if not all peroxisomal membrane proteins (PMPs) are translated on free polyribosomes in the cytoplasm and imported by as yet poorly understood mechanisms into the peroxisomal membrane (Fujiki et al. 2006, Van Ael & Fransen 2006). Several studies have been conducted to define the targeting information in the PMPs that leads to peroxisomal membrane localisation. Such peroxisomal membrane targeting signals (mPTS) have been described for Pex3p from mammals and yeast (Baerends et al. 2000, Purdue & Lazarow 2001), Candida boidinii Pmp47 (Dyer et al. 1996, Wang et al. 2001), human Pmp34 (Honsho & Fujiki 2001, Jones et al. 2001), cottonseed ascorbate

15 peroxidase (Mullen & Trelease 2000) and Pxmp2 (previously known as Pmp22) from mammals and plants (Pause et al. 2000, Murphy et al. 2003). These sequences do not exhibit a clear consensus, but in all cases they contain basic amino acids in conjunction with at least one transmembrane domain (Heiland & Erdmann 2005). In some cases the PMPs contain several targeting signals (Brosius et al. 2002, Wang et al. 2004). It has been proposed that Pex19p functions as a chaperone and an import receptor for the newly synthesised PMPs and after recognising and binding the targets delivers them to the peroxisomal membrane where Pex3p acts as a docking factor (Hettema et al. 2000, Snyder et al. 2000, Pinto et al. 2006). Pex19p is known to bind multiple PMPs (Sacksteder et al. 2000, Fransen et al. 2001, Fransen et al. 2004) with the exception of Pex3p, which is targeted to the peroxisomal membrane in a non Pex19p-dependent fashion (Jones et al. 2004). The existence of two classes of peroxisomal membrane proteins have been proposed. The class I PMPs, containing the mPTS1 signal, require Pex19p for their import to the peroxisomal membrane, and the class II PMPs, containing the mPTS2, are targeted to peroxisomes independently of Pex19p (Jones et al. 2004). Several studies have shown that the class II PMPs may be delivered to peroxisomes via the ER (Heiland & Erdmann 2005, Titorenko & Mullen 2006). Such protein is Pex16p, which has been shown to be present in the ER before sorting to peroxisomes (Kim et al. 2006). Also, cottonseed ascorbate peroxidase in plant cells (Mullen et al. 1999) and Pex13p in mouse dendritic cells (Geuze et al. 2003) may be targeted from the cytosol to a pre-existing subdomain of the ER. The structural features that target these proteins to the ER and subsequently to the peroxisomal membrane are currently unknown.

2.4 Peroxisomal metabolism

2.4.1 β-Oxidation

In higher eukaryotes fatty acid β-oxidation takes place in both mitochondria and peroxisomes, while in lower eukaryotes such as the yeast Saccharomyces cerevisiae β-oxidation occurs only in peroxisomes. During the β-oxidation cycle the fatty acid carbon chain is cleaved yielding chain-shortened acyl-CoA and acetyl-CoA (Fig. 1) (Wanders et al. 2001, Wanders & Waterham 2006a). It is well established that mammalian peroxisomes can only accomplish chain shortening of

16 fatty acids and do not fully degrade them (Wanders et al. 2001). The substrates for the mammalian peroxisomal β-oxidation include long-chain fatty acids (LCFA), very-long-chain fatty acids (VLCFA), branched-chain fatty acids, dicarboxylic acids, precursors of bile acids, certain leukotriens and prostaglandins, carboxylic derivatives of some xenobiotics, vitamins E and K (Wanders & Waterham 2006a). These substances with a long hydrophobic carbon chain, which are often poorly soluble in water, are converted to more polar metabolites in the peroxisomal β-oxidation (Poirier et al. 2006). In order to enter in the β-oxidation cycle, the fatty acids are first activated by coupling with CoA. The formation of CoA thioesters is catalysed by acyl-CoA synthetases (Watkins 1997). The peroxisomal localisation has been described for long-chain fatty acyl-CoA synthetase (LACS) and for very- long-chain fatty acyl- CoA synthetase (VLACS) (Steinberg et al. 1999). The isoforms of LACS have also been found in the ER and in mitochondria (Coleman et al. 2002) and the isoform of VLACS has been detected in the ER (Steinberg et al. 1999). The sequence of reactions in peroxisomal β-oxidation is shown in figure 1. The first step is the oxidation of acyl-CoA to 2-enoyl-CoA by acyl-CoA oxidase in a FAD-dependent reaction in which hydrogen peroxide is generated as a side product. Three different acyl-CoA oxidases (ACOX 1-3) have been identified in mammals: ACOX1 acts only on straight chain acyl-CoAs while ACOX2 and ACOX3 can desaturate straight and 2-methyl branched-chain acyl-CoAs (Van Veldhoven et al. 2001, Wanders et al. 2001). The second step in β-oxidation is the hydration of 2-enoyl-CoAs to either 3S- or 3R-hydroxyacyl-CoAs. The subsequent third step is the dehydrogenation of the corresponding 3-hydroxyacyl-CoAs with the production of 3-oxoacyl-CoAs. The second and the third steps are both catalysed by multifunctional enzyme 1 (MFE-1) or multifunctional enzyme 2 (MFE-2). MFE-1 forms and dehydrogenates 3S-hydroxyacyl-CoAs and MFE-2 converts the 2-enoyl-CoAs into the corresponding 3-oxo-compounds via a 3R-hydroxyacyl-CoA intermediate (Huyghe et al. 2006). It was shown that MFE-2 is the main participant in the β-oxidation of long- and very-long-chain fatty acids, branched-chain fatty acids and bile acid intermediates (Wanders & Waterham 2006a). The physiological role of MFE-1 is still unclear, although it may be a primary enzyme involved in the oxidation of dicarboxylic acids (Ferdinandusse et al. 2004). The last step in β-oxidation is the thiolytic cleavage of 3-oxoacyl-CoAs with the production of chain-shortened acyl-CoAs and acetyl-CoA catalysed by 3-oxoacyl-CoA thiolases (Huyghe et al. 2006). Two thiolase isoforms have been

17 described in mammalian peroxisomes: 3-oxoacyl-CoA thiolase that acts on straight chain oxoacyl-CoAs, and sterol carrier protein 2/3-oxoacyl-CoA thiolase catalysing the cleavage of both straight chain and 2-methyl-branched chain oxoacyl-CoAs (Antonenkov et al. 1997). In the last case, the final product of the reaction is propionyl-CoA instead of acetyl-CoA. The core enzymes of the β-oxidation pathway are capable of the complete breakdown of saturated fatty acids or those acids containing trans double bonds at even-numbered positions. However, for the degradation of fatty acids with cis or trans double bonds at odd-numbered positions, or cis double bonds at even- numbered positions, several additional or auxiliary enzymes are present in peroxisomes (Hiltunen et al. 2003). The β-oxidation of unsaturated fatty acids with double bonds extending from even-numbered carbons requires the subsequent action of two auxiliary enzymes: NADP+-dependent 2,4-dienoyl-CoA reductase and Δ3, Δ2-enoyl-CoA isomerase (Geisbrecht et al. 1999). Unsaturated fatty acids with double bonds at odd-numbered carbons can enter the β-oxidation cycle through two pathways: (i) the first one involves Δ3,Δ2-enoyl-CoA isomerase (Geisbrecht et al. 1999), and (ii) the second one requires the participation of three auxiliary enzymes, Δ3,5,Δ2,4-dienoyl-CoA isomerase, 2,4-dienoyl-CoA reductase, and Δ3,Δ2-enoyl-CoA isomerase, respectively (He et al. 1995).

18 Fig. 1. Peroxisomal β-oxidation pathway.

β-Oxidation of straight chain fatty acids and biosynthesis of docosahexaenoic acid

Saturated long- and very-long-chain fatty acids enter the peroxisomal β-oxidation cycle after their activation by LACS and VLACS respectively. The number of rounds of β-oxidation of these substrates in peroxisomes is undetermined, but it is believed that the oxidation proceeds down to medium- and/or short chain fatty

19 acyl-CoAs, which are then delivered to mitochondria for further degradation (Wanders et al. 2001, Wanders 2004). The peroxisomal fatty acid β-oxidation system is involved in the biosynthesis of docosahexaenoic acid, an essential metabolite for normal growth and functional development of the human brain. It is proposed that the synthesis of docosahexaenoic acid proceeds via a process called retroconversion involving enzymes in both the smooth ER and peroxisomes and requires the transfer of this fatty acid between these two cellular compartments (Reddy & Hashimoto 2001).

β-Oxidation of pristanic acid

Pristanic acid is the end product of α-oxidation as described below. It can be activated to pristanoyl-CoA and enter the peroxisomal β-oxidation cycle. It has been shown that the pristanoyl-CoA is subjected to 3 cycles of β-oxidation (Verhoeven et al. 1998) yielding two molecules of propionyl-CoA, one molecule of acetyl-CoA, and one molecule of 4,8-dimethylnonanoyl-CoA as end products.

Metabolism of bile acids

The C27-intermediates in bile acid biosynthesis, 3α,7α-dihydroxycholestanoic acid (DHCA) and 3α,7α,12α-trihydroxycholestanoic acid (THCA), are produced in the liver from cholesterol through a set of reactions catalysed mainly by cytochromes localised in the ER (Russell 2003). These bile acid precursors are then apparently activated to CoA esters by two enzymes: VLACS, localised in the ER and in the peroxisome, or bile acid-CoA synthetase (BACS), localised exclusively in the ER (Mihalik et al. 2002). Whether the activation of bile acid precursors takes place in the ER or after transfer into peroxisomes is currently unclear (Ferdinandusse & Houten 2006). The first step in peroxisomal metabolism of the bile acid precursors is the conversion of 25R-isomers of CoA esters to 25S-isomers that can enter the peroxisomal β-oxidation cycle. This reaction is catalysed by the peroxisomal

α-methylacyl-CoA racemase (Schmitz et al. 1995). The CoA-esters of the C27-bile acid intermediates are then subjected to 1 cycle of β-oxidation catalysed subsequently by acyl-CoA oxidase, MFE-2 and sterol carrier protein 2/3-oxoacyl- CoA thiolase. In these reactions the side chains of DHCA and THCA are shortened by 3 carbon atoms producing propionyl-CoA, chenodeoxycholoyl-CoA

20 derived from DHCA and choloyl-CoA derived from THCA, respectively (Ferdinandusse & Houten 2006). The next step in the biosynthesis of bile acids is the conjugation of chenodeoxycholoyl-CoA and choloyl-CoA with the amino acids taurine and glycine by the enzyme bile acyl-CoA: amino acid N-acyltransferase (BAAT), described in mice (Falany et al. 1997), human (Falany et al. 1994) and rat

(O'Byrne et al. 2003). Conjugated C24-bile acids are secreted from hepatocytes after transfer out of the peroxisomes (Ferdinandusse & Houten 2006).

Dicarboxylic acids

Dicarboxylic acids are produced during the ω-oxidation of monocarboxylic fatty acids. The reactions of ω-oxidation take place in the ER and cytosol (Sanders et al. 2005). Further metabolism of dicarboxylic acids is believed to proceed in peroxisomes and mitochondria with peroxisomes being the preferred site for the β-oxidation of medium and long-chain dicarboxylic acids. The end products of peroxisomal dicarboxylic acid oxidation are acetyl-CoA, medium-chain dicarboxylic acids (adipic acid, C6; suberic acid, C8 and sebacic acid, C10) and succinate. The medium-chain dicarboxylic acids are secreted in the urine and succinate may be further metabolised in mitochondria (Westin et al. 2005).

Other substrates for β-oxidation

Peroxisomes are involved in the β-oxidation of prostaglandins and related compounds such as leukotrienes, thromboxanes and prostacyclins (Masters & Crane 1995). Also vitamins E, which are described as a family of lipophilic micronutrients, consisting of four tocopherols and four tocotrienols and different variants of vitamin K such as phylloquinone (vitamin K1) and a group of menaquinones (vitamin K2), are metabolised by side chain degradation via an initial ω-hydroxylation and subsequent peroxisomal β-oxidation (Landes et al. 2003).

2.4.2 α-Oxidation

Fatty acids such as phytanic acid (3,7,11,15-tetramethylhexadecanoic acid) containing a methyl group at the carbon 3 position are unable to undergo β-oxidation because of the inability of 3-hydroxyacyl-CoA dehydrogenase to

21 convert the corresponding hydroxyacyl-CoA to 3-oxoacyl-CoA (Van Veldhoven et al. 1991). To overcome this problem a catabolic pathway named α-oxidation exists in mammalian peroxisomes (Fig. 2). The main substrate for this pathway is phytanic acid, a derivative of phytol which is consumed with food (Jansen & Wanders 2006). Phytanic acid enters the α-oxidation pathway as phytanoyl-CoA. The activation has been shown to take place outside peroxisomes by LACS (Watkins et al. 1996) or by VLACS located inside peroxisomes (Steinberg et al. 1999). The first step in the α-oxidation pathway is catalysed by phytanoyl-CoA hydroxylase. This reaction involves the hydroxylation of phytanoyl-CoA to 2-hydroxyphytanoyl-CoA. Phytanoyl-CoA hydroxylase belongs to the Fe2+ -dependent oxygenase enzyme family and the conversion of phytanoyl-CoA to 2-hydroxyphytanoyl-CoA is coupled with the decarboxylation of 2-oxoglutarate. (Mihalik et al. 1995). In the second step, 2-hydroxyphytanoyl-CoA is cleaved to the fatty aldehyde pristanal and formyl-CoA by the action of 2-hydroxyphytanoyl-CoA lyase, an enzyme which is dependent on thiamine pyrophosphate (TPP) and Mg2+ (Foulon et al. 1999). Formyl-CoA, which is a very unstable compound, is believed to split non-enzymatically into formic acid and CoA (Croes et al. 1997). It has been suggested that the formic acid is subsequently oxidised by catalase (see below) (Jansen & Wanders 2006). The last step in the phytanic acid α-oxidation is the conversion of pristanal to pristanic acid by NAD(P)+-dependent enzyme aldehyde dehydrogenase (Jansen et al. 2001). The pristanic acid is activated to pristanoyl-CoA, probably by VLACS (Steinberg et al. 1999) and enters the β-oxidation cycle. As the α-oxidation pathway is not stereoselective, both naturally occurring stereoisomers 3R- and 3S-phytanic acid are converted to 2R- and 2S-pristanic acid respectively. In contrast, only 2S-acyl-CoA esters are accepted in the following β-oxidation steps. The conversion of 2R-pristanoyl-CoA to 2S-pristanoyl-CoA is catalysed by the peroxisomal enzyme α-methylacyl-CoA racemase (Ferdinandusse et al. 2000).

22 Fig. 2. Phytanic acid α-oxidation pathway in mammalian peroxisomes (modified from Jansen & Wanders 2006).

2.4.3 Polyamine oxidation

The polyamines butane-1,4-diamine (putrescine), N,N'- bis(3-aminopropyl)butane-1,4-diamine (spermine) and N-(3-aminopropyl)butane- 1,4-diamine (spermidine) are required for numerous fundamentally important cellular processes including wound healing and tissue growth and differentiation (Heby & Persson 1990, Nishioka 1993, Morgan 1998). One of the degradation pathways for spermine and spermidine involves their transformation by the cytosolic enzyme acetyl-CoA:spermine/spermidine N1-acetyltransferase into

23 N1-acetyl-spermine and N1-acetyl-spermidine respectively, followed by oxidation of these products by the peroxisomal enzyme polyamine oxidase (Wanders & Waterham 2006a). Mammalian peroxisomal polyamine oxidase is a FAD 1 dependent, H2O2-producing oxidase that converts N -acetyl-spermine to spermidine and 3-acetamidopropanal, and N1-acetyl-spermidine to putrescine and

3-acetamidopropanal, respectively (Seiler 1995). This enzyme has been proposed to play a role in triggering and/or progression of apoptosis (Hu & Pegg 1997, Kramer et al. 1999). The fate of 3-acetamidopropanal in peroxisomes is unclear. However, it has been suggested that this compound is further metabolised by peroxisomal aldehyde dehydrogenase (Antonenkov et al. 1985) and/or alcohol dehydrogenase (Sakuraba & Noguchi 1995).

2.4.4 Ether phospholipid biosynthesis

Peroxisomes play an important role in the biosynthesis of ether phospholipids such as plasmalogens and platelet activating factor (Hajra 1995). Ether-linked glycerolipids are widely presented as important structural and functional components of biological membranes (Gorgas et al. 2006). The biosynthesis of plasmalogens (Fig. 3) starts with the acylation of dihydroxyacetone phosphate in a reaction catalysed by dihydroxyacetone phosphate acyltransferase. The ether- bond is then introduced by replacement of the fatty acid with a long-chain fatty alcohol by alkyl-dihydroxyacetone phosphate synthase, yielding alkyl- dihydroxyacetone phosphate as a product. (Wanders & Waterham 2006b). The two enzymes: dihydroxyacetone phosphate acyltransferase and alkyl- dihydroxyacetone phosphate synthase are associated with the peroxisomal membrane at the inner matrix side and are suggested to form a metabolon-like complex (Thai et al. 1997, Biermann et al. 1999). The bulk of long chain fatty alcohols, required for the alkyl-dihydroxyacetone phosphate synthase reaction, is believed to be synthesised from acyl-CoAs by the action of long chain acyl-CoA reductases that are localised on the peroxisomal membrane facing the cytosol (Burdett et al. 1991). The final peroxisomal step of the whole pathway is the conversion of alkyl-dihydroxyacetone phosphate into 1-alkyl-sn-glycero-3- phosphate by the NADH-dependent enzyme acyl/alkyl-dihydroxyacetone phosphate reductase which is located on the surface of the peroxisomal and ER membranes facing the cytoplasm (Wanders & Waterham 2006a). The following steps of the ether phospholipid synthesis are carried out in the ER (Brites et al. 2004).

24 Fig. 3. Biosynthesis of plasmalogens. DHAP = dihydroxyacetone phosphate, see text for details. (Modified from Wanders & Waterham 2006a)

2.4.5 Metabolism of amino acids and glyoxylate

Mammalian peroxisomes also contribute to amino acid metabolism, and house some amino acid oxidases and aminotransferases (Masters & Crane 1995). The oxidation of D-amino acids in peroxisomes results in the production of the corresponding 3-oxo-acids. Some L-amino acids, e.g. L-lysine, may be degraded via the L-pipecolic acid pathway by L-pipecolic acid oxidase, a peroxisomal enzyme identified in human liver (Wanders et al. 1989) The importance of glyoxylate metabolism in peroxisomes is illustrated by the development of primary hyperoxaluria I, a human disease, characterised by the accumulation of Ca2+-oxalate in the kidneys (Wanders & Waterham 2006b). Glyoxylate is produced from glycolic acid by peroxisomal glycolate oxidase.

25 Glyoxylate is then converted to glycine by alanine:glyoxylate amino transferase, an enzyme also located in the peroxisomal matrix (Holmes & Assimos 1998). Primary hyperoxaluria I is caused by the mislocalisation of alanine:glyoxylate amino transferase to mitochondria or by the inactivity of the enzyme. In this scenario, glyoxylate leaks out of peroxisomes into the cytoplasm where it is converted into oxalate by cytosolic lactate dehydrogenase (Wanders & Waterham 2006b).

2.4.6 Metabolism of reactive oxygen species

Peroxisomes, together with mitochondria and ER, are the major sites of oxygen consumption in cells. In peroxisomes, oxygen is used in the reactions catalysed by oxidases such as acyl-CoA oxidases, D-amino acid oxidase and urate oxidase.

The by-product of these reactions is H2O2, which is degraded by catalase.

However, some H2O2 molecules may escape decomposition by catalase and enter the set of non-enzymatic reactions leading to the production of reactive oxygen − species including the superoxide anion radical (O2· ), hydroxyl radical (·OH) and nitric oxide (·NO). Mammalian peroxisomes have been shown to harbour a set of enzymes for scavenging ROS (Schrader & Fahimi 2006). These include in addition to catalase (Oshino et al. 1973), glutathione peroxidase (Singh et al. − 1994), Zn- and Mn-dependent superoxide dismutases converting O2· to H2O2 (Singh et al. 1999), and glutathione S-transferase (Morel et al. 2004).

2.4.7 Purine catabolism

The purine degradation pathway can be separated in well-defined steps (Berry & Hare 2004). The last steps of this pathway in humans are the conversion of hypoxanthine to xanthine and xanthine to uric acid. Both of these reactions are catalysed by xanthine oxidoreductase (XOR), a molybdenum-containing flavoprotein that may have two interconvertible forms, xanthine dehydrogenase (XDH) and xanthine oxidase (XO) (Hille & Nishino 1995). The subcellular localisation of XOR has been studied extensively but it still remains controversial. XOR has been localised to the cytoplasm of liver endothelial cells and the peroxisomal core of liver parenchymal cells in rat (Angermuller et al. 1987). However, other authors found the enzyme in the cytoplasm of rat hepatocytes (Ichikawa et al. 1992). Recently, XOR has been identified by electron microscopy in the peroxisomes and cytoplasm of hepatocytes and in various organelles of

26 Kupffer and sinusoidal cells, including the rough ER, lysosomes, and endocytic vesicles (Frederiks & Vreeling-Sindelarova 2002). The catabolism of purines to uric acid is a common pathway for all vertebrates, but further degradation of uric acid varies between species. Most mammals oxidise uric acid to allantoin by urate oxidase located in the crystalloid core of hepatic and renal peroxisomes (Angermuller & Fahimi 1986, Volkl et al. 1988). The exceptions are humans and hominoid primates, which are unable to oxidise uric acid and excrete it with urine as the end product of purine degradation (Usuda et al. 1994). According to recent studies the conversion of uric acid to allantoin follows several previously unknown steps: the first step is the oxidation of uric acid by urate oxidase to 5-hydroxyisourate that is then converted to 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline by 5- hydroxyisourate hydrolase and finally to S-(+)-allantoin by 2-oxo-4-hydroxy-4- carboxy-5-ureidoimidazoline decarboxylase (Ramazzina et al. 2006). In the case of primates urate oxidase has lost its activity owing to mutations in the gene coding for this enzyme (Yeldandi et al. 1991).

2.5 Transport of metabolites across the peroxisomal membrane

The diversity of peroxisomal metabolic pathways implies that a relatively large number of different metabolites must be transported across the peroxisomal membrane. Mechanisms by which these processes are carried out may depend on specific transporters that facilitate the transfer of different compounds or membrane channels providing free movement of metabolites between the peroxisomal lumen and surrounding cytoplasm. Currently these mechanisms are poorly understood.

2.5.1 Permeability properties of the peroxisomal membrane

The biological membrane is a lipid barrier that regulates the transfer of solutes and macromolecules by means of specific membrane proteins. Some hydrophobic and amphipathic molecules are able to cross the lipid bilayer by diffusion or by a flip-flop mechanism, while polar, hydrophilic solutes and large molecules are dependent on specific protein transport systems (Theodoulou et al. 2006). The problem of the permeability of the peroxisomal membrane has long been unresolved. Based on early studies (De Duve & Baudhuin 1966) peroxisomes were considered to be freely permeable to low molecular mass compounds. This

27 conclusion has been drawn from the observations that certain peroxisomal oxidases are freely accessible to substrates meaning they did not exhibit any structure-linked latency (Baudhuin 1966). Later studies with isolated rat liver peroxisomes and liposomes reconstituted with peroxisomal membrane proteins have led to the proposal that the peroxisomal membrane contains porin-like proteins (Van Veldhoven et al. 1987) and is open to solutes (Van Veldhoven et al. 1983, Van Veldhoven & Mannaerts 1986, Verleur & Wanders 1993). Porins are β-barrel proteins forming general diffusion channels in the membrane. They were found in the outer membrane of Gram-negative bacteria and in the outer mitochondrial membrane (for reviews see Nikaido 2003 and Young et al. 2007, respectively). Porin-like proteins are presumed to have a role in the transfer of metabolites across peroxisomal membranes in plants (Reumann et al. 1995, Reumann et al. 1998), yeast (Sulter et al. 1993) and mammals (Van Veldhoven et al. 1987). However, several lines of evidence contradict the concept that peroxisomal membranes are open to solutes. Structure-linked latency was observed for peroxisomal enzymes glucose-6-phosphate dehydrogenase (Antonenkov 1989) and acyl CoA:dihydroxyacetonephosphate acyltransferase (Wolvetang et al. 1990). Besides, the presence of metabolic shuttles described in more detail below has been considered as a proof that the peroxisomal membrane in vivo forms compartment closed to solutes. This suggestion has been strengthened by the discovery of ATP transporters in the peroxisomal membranes of yeast (Palmieri et al. 2001, van Roermund et al. 2001) and mammals (Wylin et al. 1998) and also by some observations indicating the presence of a pH-gradient across the peroxisomal membrane (Dansen et al. 2000).

2.5.2 Peroxisomal pH, current view

Several experiments to measure intra-peroxisomal pH in situ have provided contradicting results indicating that the lumen of yeast peroxisomes is acidic (Nicolay et al. 1987, Waterham et al. 1990, Lasorsa et al. 2004) or alkaline (van Roermund et al. 2004). Studies with human fibroblasts using a fluorescence reporter peptide have detected alkaline intra-peroxisomal pH (Dansen et al. 2000) while experiments with Chinese hamster ovary cells demonstrated a neutral pH inside the organelles (Jankowski et al. 2001). The later observation led to the conclusion that the peroxisomal membrane is highly permeable to protons (Jankowski et al. 2001). This inconsistency in the results may indicate that the

28 intraperoxisomal pH depends on the physiological status of the cell. The pH- gradient between the peroxisomal lumen and the cytosol could be created by the mechanism called Donnan equilibrium which can be described as a specific behaviour of ions in the presence of charged macromolecules surrounded by a membrane impermeable to macromolecules but permeable to solutes (Antonenkov & Hiltunen 2006).

2.5.3 Two channel concept

Recent studies by Antonenkov et al. (2004b) have led to a new assessment of the peroxisomal membrane permeability to solutes. In experiments on highly purified rat liver peroxisomes, it was found that the membrane of these organelles is permeable to small solutes such as urate, glycolate and others, but at the same time strongly restricts permeation of bulky solutes such as NAD/H, NADP/H and CoA (Antonenkov et al. 2004b). These observations have led to the conclusion that in vivo peroxisomes contain a separate pool of cofactors, but share a common pool of small metabolites with the cytoplasm. To explain the experimental results, a two channel hypothesis has been proposed according to which the peroxisomal membrane contains at least two types of channels. One type has a diameter that allows passage of small solutes only. The other type is also accessible to bulky solutes, but passage of these compounds is highly limited (Antonenkov et al. 2004b).

2.5.4 Channels in the peroxisomal membrane

In the late 1980s, pore-forming channels with an apparent diameter of 1.5–3.0 nm were demonstrated in the rat liver peroxisomal membrane by electrophysiological analyses (Labarca et al. 1986, Lemmens et al. 1989). The physiological role for these cation-selective and voltage-gated channels was suggested to provide substrates and cofactors for peroxisomal enzymes. The existence of active peroxisomal channels in vivo was later challenged mainly by observations of the specific transporter proteins in the peroxisomal membrane (Hettema & Tabak 2000, van Roermund et al. 2003). Isolated peroxisomal preparations have also shown channel-forming activities in yeasts and plants. A major 31 kDa peroxisomal integral membrane protein from the yeast Hansenula polymorpha was observed to form pores after insertion into liposomes preloaded with sucrose. This protein showed cross-reactivity with an antibody against a mitochondrial

29 porin (voltage-dependent anion channel, VDAC) of S.cerevisiae (Sulter et al. 1993). In addition, 34 and 36 kDa integral membrane proteins in membranes of oilseed glyoxysomes have been shown to react with an antibody against potato tuber VDAC (Corpas et al. 2000). In membranes of spinach leaf peroxisomes (Reumann et al. 1995) and glyoxysomes of castor bean endosperm (Reumann et al. 1997) small anion-selective channels with single-channel conductance from 0.33 to 0.35 nS in 1 M KCl have been observed. These plant peroxisomal channel proteins are believed to have a role in the passage of photorespiratory metabolites and intermediates of the glyoxylate cycle (Reumann et al. 1998).

2.5.5 Peroxisomal membrane transporters

ABC transporter family

Fatty acids can enter peroxisomes as either non-esterified fatty acids or as CoA esters. The coupling of a free fatty acid to CoA generates a large charged molecule that is unable to cross biomembranes in the absence of a transport protein. It is currently believed that the members of the ATP-binding cassette (ABC) protein superfamily function as transporters in the peroxisomal import of β-oxidation substrates (Theodoulou et al. 2006, Kemp & Wanders 2007, Wanders et al. 2007). A functional ABC protein consists of the transmembrane domain (TMD) and the nucleotide binding fold (NBF) (Higgins 1992). These units in eukaryotes are frequently represented within a single polypeptide with the topology TMD-NBF- TMD-NBF described as a “full size” transporter. Alternatively, functional ABC protein may be formed as a dimer of two “half size” transporters (TMD-NBF). Peroxisomal ABC transporters are almost all half size transporters (Dean et al. 2001). Human and mouse peroxisomes contain four members of the ABCD superfamily: adrenoleukodystrophy protein (ALDP) (Mosser et al. 1993), adrenoleukodystrophy related protein (ALDR) (Holzinger et al. 1997a), 70 kDa peroxisomal membrane protein (PMP70) (Kamijo et al. 1990) and 69 kDa peroxisomal membrane protein (PMP69) (Holzinger et al. 1997b), which are all integral membrane proteins and encoded by ABCD1, ABCD2, ABCD3 and ABCD4 genes, respectively.

30 A defect in ALDP causes X-linked adrenoleukodystrophy (X-ALD), a neurodegenerative disease characterised by the accumulation of VLCFAs (Wanders & Waterham 2006b). To date, 866 mutations in the ABCD1 gene have been described (Wanders & Waterham 2006b), but the precise relationship between the clinical symptoms, biochemical phenotype, and these mutations have not been established. It has been suggested that ALDP participates in β-oxidation of VLCFAs by catalysing transport of VLCFAs or VLCFA-CoAs (Wanders 2004). PMP70 is the first peroxisomal ABC transporter identified (Kamijo et al. 1990, Lombard-Platet et al. 1996). Overexpression studies in Chinese hamster ovary cells indicate that PMP70 mediates the transport of long chain fatty acyl- CoA esters across the peroxisomal membrane (Imanaka et al. 1999). Mutations in the ABCD3 gene have not been shown to associate with any abnormal phenotype in humans leaving the physiological significance of PMP70 to be determined (Wanders et al. 2001, Wanders 2004). PMP69 is ubiquitously expressed in human tissues (Holzinger et al. 1997b, Shani et al. 1997) and based on sequence homology to PMP70 it is believed to have a similar role as a transporter, although it has not yet been characterised functionally. ALDR has high sequence homology to ALDP and these proteins show distinct but overlapping expression patterns in mouse tissues (Lombard-Platet et al. 1996). Strong evidence supports the view that ALDR and ALDP are functionally redundant since both of these proteins were able to restore VLCFA β-oxidation in X-ALD fibroblasts (Braiterman et al. 1998, Kemp et al. 1998). Also, the neurological and biochemical phenotypes of X-ALD mice (ALDP−/−) can be ameliorated by overexpression of ALDR in these mice and ALDP−/− and ALDR−/− double mutants show a more severe phenotype than the ALDP−/− single mutant, indicating that besides sharing functionality with ALDP, the ALDR protein has additional roles (Pujol et al. 2004). In the yeast S. cerevisiae two peroxisomal ABC transporters, Pxa1p and Pxa2p form functional heterodimers in the peroxisomal membrane (Shani et al. 1995, Hettema et al. 1996, Shani & Valle 1996, Swartzman et al. 1996). Studies with deletion mutants have shown that these transporters are vital for the β-oxidation of LCFA. It is thought that MCFAs are delivered into the yeast peroxisome as free fatty acids and activated by the acyl-CoA synthetase Faa2p inside the organelle (Hettema et al. 1996) while LCFA are first activated in the cytosol and transported across the peroxisomal membrane as LCFA-CoA esters by the Pxa1p/Pxa2p acyl-CoA transporter (Hettema & Tabak 2000).

31 ATP/AMP transporter

To date the best characterised peroxisomal transporter is the adenine nucleotide transporter, Ant1p of S. cerevisiae, which belongs to the family of mitochondrial carrier proteins (Palmieri et al. 2001, van Roermund et al. 2001, Palmieri 2004). This transporter may be required for the activation of MCFAs with formation of acyl-CoAs inside peroxisomes in an ATP consuming reaction (Hettema et al. 1996). The reaction is catalysed by peroxisomal acyl-CoA synthetase Faa2p which is localised inside the organelles. Upon the formation of medium chain acyl-CoAs, AMP is formed, which, due to the lack of an ATP regeneration system inside peroxisomes, has to be delivered back to the cytoplasm. This transport of ATP in and AMP out of the peroxisomal lumen is catalysed by Ant1p. (Palmieri et al. 2001, van Roermund et al. 2001). It has been shown also that Ant1p exhibits two transport modes, the ATP-AMP exchange and the uniport of adenine nucleotides (Lasorsa et al. 2004). A functional homolog for Ant1p, Pmp47p has been identified in the yeast C.boidinii (Sakai et al. 1996) and evidence was provided that this protein functions as an ATP carrier in the peroxisomal membrane (Nakagawa et al. 2000, van Roermund et al. 2001). A mammalian ortholog for the yeast Ant1p is PMP34, described in human and mice (Wylin et al. 1998). This protein has been shown to partially complement the MCFA β-oxidation defect of the ant1Δ mutant. The purified recombinant PMP34 showed ATP exchange activity in reconstituted liposomes (Visser et al. 2002).

Other transporters

The activation of fatty acids inside peroxisomes by acyl-CoA synthetases generates pyrophosphate that probably decomposes into two molecules of orthophosphate that need to be delivered outside peroxisomes to regenerate ATP. Studies with proteoliposomes reconstituted from peroxisomal membranes have provided evidence for the existence of a phosphate transporter in mammalian peroxisomes (Visser et al. 2005). This transporter could be distinguished from mitochondrial phosphate transporters, but the molecular identity of this transporter remains to be determined (Visser et al. 2005). A similar experimental approach has provided evidence for the existence of a transporter for 2-oxoglutarate, a metabolite known to be involved in the shuttle mechanism for the reduction of NADP+ to NADPH in peroxisomes (Visser et al. 2006). Recently it has also been shown by proteoliposome assays that the

32 peroxisomal membrane contains a yet unidentified protein or protein complex allowing transport of bile acid conjugates glyco- and taurocholate across the membrane. This protein is suggested to function in the export of glycine- and taurine-conjugated bile acids after biosynthesis inside peroxisomes (Visser et al. 2007a). Monocarboxylate transporters (MCT) previously identified in the plasma membrane and mitochondrial membrane may also function as transporters in the peroxisomal membrane to transfer pyruvate and lactate in the shuttle conversion of NADH to NAD+ (McClelland et al. 2003). An export from peroxisomes of acetyl- and acyl units generated by β-oxidation of fatty acids is believed to take place by utilising a carnitine shuttle (Ramsay 1999). Studies on cultured skin fibroblasts led to the suggestion that acetyl-CoA and medium-chain acyl-CoAs can be converted to the corresponding carnitine esters by peroxisomal carnitine acetyltransferase (Crat) and carnitine octanoyltransferase (Crot), respectively (Jakobs & Wanders 1995, Verhoeven et al. 1998). These carnitine esters are then exported from peroxisomes to mitochondria for further metabolism. In mammals the carnitine conjugates are believed to be transported through the peroxisomal carnitine transporter Octn3 which belongs to the organic cation transporter family (Lamhonwah et al. 2003, Lamhonwah et al. 2005). Acyl-CoA thioesterases (Acots) are enzymes capable of hydrolysing acyl- CoA esters to free fatty acid and CoA, and several Acot-enzyme activities have been localised to peroxisomes (Wilcke & Alexson 1994). In mice, peroxisomes harbour acyl-CoA esterases such as Acot5 and Acot3 that hydrolyse medium- chain and long-chain acyl-CoAs, respectively (Westin et al. 2004), Acot4 for succinyl- and glutaryl-CoAs (Westin et al. 2005), Acot6 for pristanoyl- and phytanoyl-CoAs (Westin et al. 2007) and Acot8, which has been shown to function as a general acyl-CoA esterase (Hunt et al. 2002). Short chain acyl-CoAs, including acetyl-CoA, are believed to be substrates for Acot12, originally identified as an extra-mitochondrial short chain thioesterase (Suematsu et al. 2002) and recently confirmed as a peroxisomal protein (Wiese et al. 2007). The presence of short- and medium-chain carnitine acetyltransferases Crat and Crot and the short- and medium-chain acyl-CoA esterases Acot12 and Acot5 has raised the question of whether these enzymes compete for the same substrates in mammalian peroxisomes. However, a recent study has shown that they have distinct tissue expression patterns suggesting that carnitine acetyltransferases and

33 acyl-CoA esterases provide complementary systems for the export of β-oxidation products from peroxisomes (Westin et al. 2008).

2.5.6 Transport of reducing equivalents by shuttle mechanisms

Each cycle of peroxisomal β-oxidation is accompanied by reduction of one molecule of NAD+ in the dehydrogenase reaction catalysed by the multifunctional enzymes MFE-1 or MFE-2 (Hiltunen et al. 2003). NAD(H) is a bulky solute that is apparently unable to penetrate the peroxisomal membrane (see above). Therefore, it has to be reoxidised inside the particles to continue the β-oxidation process. In mammals the reoxidation of NADH is proposed to take place via lactate-pyruvate and glycerophosphate shuttles by the action of peroxisomal/cytosolic lactate dehydrogenase (LDH) (Baumgart et al. 1996, McClelland et al. 2003) and peroxisomal/mitochondrial glycerol-3-phosphate dehydrogenase (G-3-PDH) (Antonenkov & Hiltunen 2006) respectively. The presence of the shuttles presupposes transfer of the shuttle molecules (lactate/pyruvate and glycerol-3-phosphate/dihydroxyacetone phosphate) across the peroxisomal membrane. The mechanism of transfer has not yet been resolved, although some authors suggested participation of the peroxisomal membrane channels in this process (Antonenkov & Hiltunen 2006). NADPH is a cofactor for 2,4-dienoyl-CoA reductase. This auxiliary enzyme of peroxisomal β-oxidation is required for the degradation of unsaturated fatty acids with double bonds at even numbered positions (Hiltunen et al. 2003). Reduction of NADP+ is believed to proceed via glucose-6-phosphate dehydrogenase (G-6-PDH) (Antonenkov 1989) and isocitrate dehydrogenase (IDH) shunts (Geisbrecht & Gould 1999). Both of these enzymes are located in the peroxisomal matrix and in the cytoplasm and they are able to convert isocitrate to α-ketoglutarate and glucose-6-phosphate to 6-phosphogluconate, respectively. The products formed are metabolised further in the reactions catalysed by the tricarboxylic acid cycle and pentose phosphate pathway, respectively. The presence of the shunt and shuttle mechanisms in peroxisomes is recognised as a solid indication in favour of the impermeability of the peroxisomal membrane to bulky solutes (Antonenkov & Hiltunen 2006).

34 2.6 Peroxisomal membrane proteins

Several studies have described the isolation of the total fraction of membrane proteins from mammalian liver or kidney peroxisomes (Hartl & Just 1987, Gouveia et al. 1999, Wiese et al. 2007). About half of the total protein of the peroxisomal membrane from rat liver is concentrated in two bands after SDS- PAGE. These bands contain mainly PMP70 and Pxmp2 (Hartl & Just 1987). In the recent proteomics characterisation of mouse kidney peroxisomes, 22 proteins were identified from the peroxisomal membrane. They are classified into peroxins, ABC-transporters and other membrane proteins (Wiese et al. 2007). The function of most peroxisomal membrane proteins remains elusive.

2.6.1 PMP22 protein family

Pxmp2-protein

A 22 kDa peroxisomal membrane protein, Pxmp2 (formerly named Pmp22), has been shown to represent the most abundant protein at the peroxisomal membrane of rat liver (Hartl et al. 1985, Hartl & Just 1987). The rat Pxmp2 open reading frame contains 584 base pairs encoding 194 amino acid residues and the hydropathy analysis predicts 4 potential transmembrane segments spanning the peroxisomal membrane with both the amino- and the carboxy-termini facing the cytosol. The expression of Pxmp2 is not stimulated by peroxisomal proliferators (Kaldi et al. 1993). A protein similar to Pxmp2 has also been identified in the plant Arabidopsis thaliana. This protein is also localised to peroxisomes and protease treatment experiments show that it is buried in the peroxisomal membrane. Apparently, the protein may form homo- or hetero-complexes in the membrane (Tugal et al. 1999). The translation of mammalian Pxmp2 proceeds on free polysomes and the protein is inserted into the peroxisomal membrane via a non ATP-dependent mechanism (Diestelkotter & Just 1993). The insertion occurs directly from the cytoplasm and is facilitated by the formation of complexes with some cytoplasmic protein factors (Just & Diestelkotter 1996, Pause et al. 1997). The targeting signals for membrane insertion of the mammalian Pxmp2 have been studied by two groups (Pause et al. 2000, Brosius et al. 2002). In the more recent study (Brosius et al. 2002) it has been shown that both, human and rat

35 Pxmp2 contain two distinct targeting regions at the amino- and carboxy terminus consisting of two transmembrane domains and an adjacent basic cluster of five amino acids. The amino acid sequence of these clusters are RRALA (amino acids 19–23) or KRALA (aa 19–23), for the amino-terminal signal and RRLLL (aa 108–112) or KRLLL (aa 107–111) for the carboxy-terminal signal in human and rat, respectively. These targeting regions interact with human PEX19, a protein that apparently acts as a chaperone for the insertion of peroxisomal membrane proteins (Brosius et al. 2002). Mouse Pxmp2 cDNA has been cloned and shows high similarity to rat Pxmp2 (Bryant & Wilson 1995). The murine Pxmp2 is encoded by the Pxmp2 gene, which is localised on chromosome 5. The expression of Pxmp2 has been detected mainly in liver, kidney and heart (Luers et al. 2001). The first exon of Pxmp2 is adjacent to the first exon of the gene PoleI, encoding the catalytic subunit of the DNA polymerase ε. However, the genes are in opposite orientation. The 393 base- pair sequence that separates these genes has been shown to act as a bidirectional promoter for Pxmp2 and PoleI with independent regulation mechanisms for each (Otte et al. 2003).

Mpv17-protein

Mouse Mpv17 is a 176 amino acid hydrophobic protein showing 26.5% identity to the rat Pxmp2 (Kaldi et al. 1993). A mouse line was developed that was unable to express Mpv17 owing to a retroviral insertion in the corresponding gene. At the age of 2–3 months, the Mpv17-deficient mice developed kidney disease, characterised as glomerulonephrosis with formation of glomerular lesions (Weiher et al. 1990). Electron microscopy examination of the glomeruli in kidneys of Mpv17-deficient mice showed podocyte foot process fusions and mesangiolysis (O'Bryan et al. 2000). Kidney failure was accompanied by proteinuria and hypertension (Clozel et al. 1999). In addition, defects in the inner ear of the Mpv17-deficient mice were detected (Meyer zum Gottesberge et al. 1996, Meyer zum Gottesberge et al. 2001). Degeneration of the stria vascularis and spiral ligament, loss of cochlear neurons and degeneration of the organ of Corti led to severe hearing loss at the age of 2 months (Muller et al. 1997). Electron microscopic examination of tissues from Mpv17−/− mice (Meyer zum Gottesberge & Felix 2005) showed that the basement membranes of the kidney glomeruli and the inner ear stria vascularis were affected. The changes include irregular distribution of the collagen IV subunits and accumulation of laminin γ1-

36 (inner ear) and β1chain (kidney) proteins. The morphology of the basement membrane in Mpv17−/− mice resembles that found in patients suffering from an inherited disease called Alport syndrome (Arnold 1984), a disorder characterised by progressive nephritis and neurosensory deafness. The function of Mpv17 is still unknown. It has been suggested that this protein participates in ROS metabolism (Zwacka et al. 1994) and somehow regulates the expression of matrix metalloproteinase 2 (MMP-2) (Reuter et al. 1998). MMP-2 is a member of the subfamily of proteinases that control the content and structure of the extracellular matrix (Meyer zum Gottesberge et al. 2001). Previous studies on human Mpv17 suggested peroxisomal localisation for this protein (Zwacka et al. 1994). Recently, Mpv17 was identified as a candidate gene in which mutations cause hepatocerebral mitochondrial DNA depletion syndrome, a genetic disorder characterised by a severe, tissue specific decrease in mitochondrial DNA copy number, leading to organ failure (Karadimas et al. 2006, Spinazzola et al. 2006). A recent report indicates that mouse Mpv17 localises nearly exclusively to the inner mitochondrial membrane with no sign of peroxisomal localisation (Spinazzola et al. 2006). In addition, mitochondrial DNA in the livers of Mpv17−/− mice was markedly decreased and mitochondrial DNA- dependent respiratory chain activities (complex I and complex II) were significantly lower, indicating a mitochondrial function for Mpv17 (Spinazzola et al. 2006).

Mpv17-like protein

The third member of the Pxmp2 family is known as the Mpv17-like protein (M-LP). It shows high sequence homology to Pxmp2 and Mpv17 proteins and is also similar to them in hydrophilicity/hydrophobicity profiles. Mouse M-LP contains 194 amino acids and is expressed in kidney and spleen, having lower expression levels in heart, lung, brain and liver. (Iida et al. 2001). M-LP has been localised in peroxisomes and has been suggested to be involved in ROS metabolism based on the finding that the gene SOD2, coding superoxide dismutase (Li et al. 1995, Lebovitz et al. 1996), is upregulated in cell culture by transfection with mouse M-LP (Iida et al. 2003). The two alternatively spliced transcripts generated from the M-LP gene, designated M-LPL and M-LPS, produce proteins of 22 kDa and 10 kDa, respectively. M-LPS is expressed in kidney and is localised in the cytoplasm of renal cells (Iida et al. 2005).

37 A human homologue of the M-LP protein (M-LPH) has been characterised recently (Iida et al. 2006) The gene for that protein contains two open reading frames that may code for two separate isoforms M-LPH1 and M-LPH2, composed of 196 and 147 amino acids, respectively. The mRNAs for these variants have been detected ubiquitously in human tissues, but only M-LPH1 was found at the protein level. M-LPH1 was localised in peroxisomes (Iida et al. 2006).

Yeast proteins homologous to mammalian Pxmp2

Sym1p, the yeast orthologue to human Mpv17, has been functionally characterised to some extent (Trott & Morano 2004, Spinazzola et al. 2006). Sym1Δ cells exhibit a growth defect when grown on ethanol at an elevated temperature and when the cells were cultured at 37 °C, several ethanol-repressed genes were dysregulated compared with wild-type (Trott & Morano 2004). Sym1 was localised in the inner membrane of mitochondria and expression of the murine Mpv17 in Sym1Δ cells could complement the mutant phenotype (Trott & Morano 2004, Spinazzola et al. 2006). Another Pxmp2-related yeast protein is the putative S.cerevisiae open reading frame YOR292c which bears significant sequence homology with Pxmp2 and related genes (Visser et al. 2007b). A yor292cΔ knock-out strain exhibited no growth defects on nonfermentable carbon sources at 37 °C. Furthermore, a sym1Δ yor292cΔ double-mutant strain showed no additional ethanol sensitivity relative to sym1Δ cells indicating a different function for YOR292C (Trott & Morano 2004). Based on the presence of a putative oleate-responsive element in the promoter region of YOR292c, a role in fatty acid metabolism has been suggested for this protein (Visser et al. 2007b).

38 3 Aims of the study

The purpose of this work was to:

1. Reveal whether the mammalian peroxisomal membrane contains channel- forming proteins; 2. Analyse the function of the peroxisomal membrane protein Pxmp2 as a potential channel-forming protein. For this purpose: (i) recombinant Pxmp2 protein was expressed in insect cells; (ii) native Pxmp2 protein was isolated from mouse liver peroxisomes and (iii) a Pxmp2-deficient mouse model was generated; 3. Characterise the phenotype of Pxmp2−/− mice.

40 4 Materials and methods

4.1 Generation of Pxmp2−/− mice

To disrupt the mouse Pxmp2, a lacZneo cassette was inserted into exon 2. This strategy was chosen to leave the regulatory elements unaltered as much as possible and to avoid changes in the expression of PoleI, which is located in close proximity to Pxmp2, but transcribed in the opposite direction (see chapter 2.6.1 in literature part). The Pxmp2 gene was partially sequenced and analysed for exon/intron junctions (data not shown). The restriction map of the Pxmp2/PoleI structure was a gift from Prof. J. Syväoja (University of Oulu, Finland). The genomic BAC ES-clone (from the 129/SvJ mouse strain) containing Pxmp2 was obtained from Genome Systems Inc. (St Louis, MO, USA). A 1.2 kb fragment spanning the region from the EcoRI site downstream of exon 1 to the XbaI site in exon 2 was generated by PCR from a Pxmp2 genomic BAC ES-clone as the 5`homology arm of the targeting construct and subcloned in the pBluescript II SK vector (Stratagene, La Jolla, CA, USA). A lacZneo cassette containing the lacZ reporter gene without an ATG start codon and the neomycin resistance gene for positive selection (PGKneo) in a head to head transcriptional orientation was ligated downstream from the 5`homology arm. As the 3`homology arm a 4.7 kb XbaI-BamHI fragment containing the second half of the exon 2, exon 3 and intron 3 was inserted downstream the lacZneo cassette. The targeting vector was linearised with NotI and electroporated into 129/SvJ RW4 embryonic stem (ES) cells. After 24h cell growth, 200 µg/ml G418 (Invitrogen, Carlsbad, CA, USA) was added to the medium and the neomycin selection was carried out 4–5 days. The surviving ES cell colonies were screened for the correct insertion of the lacZneo cassette by PCR using 3 primers: the forward primer for normal and mutated alleles (GGTCAGAAGCACAGAGAAGAGAAGC) corresponding to the sequence from intron 1, upstream of the 5`flanking region; the reverse primer for the normal allele (CGCCCAGCTTCTCTGATGCTTCTTA) from intron 2, and the reverse primer for the mutated allele (GCGGGCCTCTTCGCTATTACG) from the lacZ-reporter gene. PCR generated products of 1.7 kb and 1.5 kb for wild-type and targeted allele, respectively. Positive ES cell clones were verified by Southern blotting (see below). Recombinant (Pxmp2+/−) ES cell clones were used for aggregation with C57BL6/J morulas. The resulting chimeras were mated with

41 C57BL/6J mice. The Pxmp2+/− germline offspring, identified by PCR analysis of tail-tip genomic DNA, were backcrossed with C57 BL6/J mice for 6 generations after which Pxmp2−/− progeny were generated. Experiments with animals were approved by the University of Oulu committee of animal experimentation.

4.2 Southern and Northern analyses

Genomic DNA was isolated from mouse liver or ES cells by using a Blood and Cell Culture Midi Kit (Qiagen, Hilden, Germany). DNA was digested with SacI restriction endonuclease and hybridisation of the Hybond-N+ nylon membrane (Amersham Pharmacia Biotech AB, Uppsala, Sweden) was carried out following established procedures using a 32P-labelled (Amersham Pharmacia Biotech AB) 670 bp Pxmp2 DNA fragment from intron 1 upstream of the 5`flanking regions as a probe. The resulting DNA fragments were 5.1 kb and 4.2 kb from the wild-type and recombinant allele, respectively (See legend to Fig. 24 for details). Total RNA was isolated from mouse tissues by a Quick Prep Total RNA Extraction kit (Amersham Pharmacia Biotech AB). For Northern analysis 45–50 µg of RNA was separated by using an agarose gel and blotted for hybridisation with mouse Pxmp2 full-length cDNA as a probe.

4.3 Quantitative Real-Time PCR

Total RNA from liver, kidney and spleen tissue was isolated with an RNeasy Mini Kit (Qiagen). cDNA was synthesised from total RNA with a Revert Aid First Strand cDNA Synthesis Kit (MBI Fermentas, Burlington, Canada). Expression levels of selected genes in mouse liver and kidney and DNA polymerase epsilon (PoleI) in spleen were detected with the ABI PRISM 7700 Sequence Detection System (Perkin Elmer, Boston, MA, USA) and Taq Man chemistry by the relative standard curve method as described (Majalahti-Palviainen et al. 2000). The results were normalised to 18S-rRNA quantified from the same samples. Real- time quantitative PCR analysis of Mpv17, M-LP and Pex11 family members in liver and kidney of mouse was performed with a 7500 Real Time PCR System using the fluorogenic probe-based Taq Man chemistry according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA, USA). Glyceraldehyde-3-phosphate dehydrogenase was used as an endogenous control for normalisation and the results were analysed with 7500 System Software (Applied Biosystems).

42 4.4 Analyses of body fluids of Pxmp2−/− mice

Blood was harvested by orbital puncture of anaesthetised mice and urine was collected in metabolic cages (model 3700M021, Tecniplast, Buguggiate, Italy) over a time period of 24 h. The levels of serum total cholesterol, total protein, uric acid, sodium, potassium, chloride, total bilirubin, triglycerides and urinary uric acid, sodium, potassium and chloride were measured using a Cobas Integra 700 automatic analyser (Hoffman-LaRoche, Basel, Switzerland) by the Oulu University Hospital. Urine osmolality was determined by an osmometer based on freezing point depression. Serum progesterone was measured by an automated chemiluminescence system (Advia Centaur, Bayer Corporation, NY, USA) and estradiol by radioimmunoassay (Orion Diagnostica, Espoo, Finland). The levels of bile acid derivatives in mouse bile fluid and serum fatty acids were measured as described elsewhere (Savolainen et al. 2004). Allantoin was determined from mouse serum and urine using a method as described (Berthemy et al. 1999) with modifications. Samples were supplemented with the stable isotope of allantoin ([1-15N, 5-13C] DL-allantoin, Sigma-Aldrich, Helsinki, Finland) as an internal standard and proteins were precipitated by acetone treatment. Supelco Discovery DSC-18 column (Supelco, Bellefonte, PA, USA) was used for sample preparation and the extracted samples were analysed on a PolarityTMdC18 HPLC column interfaced with a Micromass Quattro II mass spectrometer (Waters, Milford, MA, USA). For the determination of oxalic acid, urine was collected for 24 h and mice were then administered i.p. 400 mg/kg body mass glycolic acid (Sigma-Aldrich). After injection, separate urine samples were collected for two consecutive 24 h time intervals and oxalic acid was determined with an Oxalate kit (Trinity Biotech, Co Wicklow, Ireland).

4.5 Blood pressure telemetry

The left carotid artery of an anesthetisised mouse was cannulated and a TA11PA- C20 telemetry device (Data Sciences International, St. Paul, MN, USA) was implanted under the skin. Buprenorphine (Temgesic, Schering-Plough, Kenilworth, NY, USA), 0.05 mg/kg s.c., was used for postoperative analgesia. The mean arterial pressure and heart rate were measured continuously through the experiment. All hemodynamic data were recorded after a one week recovery period from the surgical operation. The recording of hemodynamics was carried

43 out every 10 minutes and averaged for every hour throughout the experiment which lasted for 4 days.

4.6 Histological analyses

Tissue samples for light microscopy were fixed in 4% (w/v) paraformaldehyde, embedded in paraffin and sections were stained with hematoxylin and eosin. For transmission electron microscopy, samples were fixed in 2.5% (w/v) glutaraldehyde, postfixed in 1% (w/v) osmium-tetroxide followed by dehydration in acetone and embedding in Epon Embed 812 (Electron Microscopy Science, Hatfield, PA, USA). Isolated peroxisomes were fixed in 1% (w/v) glutaraldehyde as described by Antonenkov et al. (2004b). After fixation the organelles were sedimented by centrifugation, the pellets were postfixed with 1% (w/v) osmium- tetroxide, stained with 1% (w/v) uranyl acetate and then dehydrated and embedded in Epoxy-embedding medium (Sigma-Aldrich). Thin sections were contrasted with uranyl acetate and lead citrate before examination in a Philips EM410 transmission electron microscope (FEI company, Eindhoven, The Netherlands).

4.7 Whole-mount analysis

For whole-mount analysis inguinal mammary glands were removed from mice and fixed in Carnoy`s fixative (absolute ethanol, chloroform, acetic acid – 6:3:1) for 2–4 hours. Tissues were rehydrated through an ethanol series and stained overnight in Carmine alum stain (0.2% (w/v) carmine and 0.5% (w/v) AlK(SO4)2, both reagents are from Sigma-Aldrich). Samples were then dehydrated in an ethanol series and cleared in xylene after which the slides were mounted with Permount (Electronic Microscopy Sciences) for microscopic evaluation.

4.8 Immunohistochemistry

For immunohistochemical staining, inguinal mammary glands were fixed overnight in 4% paraformaldehyde at 4 °C, followed by dehydration and paraffin embedding. Paraffin sections (5 µm) were deparaffinized in xylene, rehydrated and incubated in Antigen Unmasking Solution (Vector Labs, Burlingame, CA, USA). Sections were treated in 50% (v/v) methanol containing 1% (v/v) hydrogen peroxide to eliminate the endogenous peroxidase. For detection of Whey Acidic

44 Protein (WAP), sections were incubated with 1.5% (v/v) rabbit serum for 30 min to block unspecific binding and then with goat-WAP antibody (SC-14832, Santa Cruz Biotech, Santa Cruz, CA, USA) at +4 °C over night. A Vectastain ABC Elite Kit (Vector Labs) was used for immunohistochemical detection of antibodies and, after hematoxylin counterstaining, the antibody staining was developed with 3`3` diaminobenzidine tetrahydrochloride (Sigma-Aldrich).

4.9 Subcellular fractionation and isolation of mouse liver peroxisomes

For measurements of enzyme activity and Western blot analysis, liver tissue collected from male 6-month-old mice after an overnight fast was homogenised in 0.25 M sucrose, 20 mM MOPS, pH 7.4, 1 mM EDTA. For a comparative study of peroxisomes from Pxmp2−/− and wild-type mice, purification of the organelles was performed at the same time for both groups of animals (8–10 mice in each group per isolation experiment). For morphological examination of peroxisomes and for detection of the latency of peroxisomal enzymes, the organelles were isolated using Nycodenz gradient centrifugation according to the following procedure. Perfused livers were homogenised in isolation medium 1 containing (1:4, w/v): 0.25 M sucrose, 10 mM MOPS, pH 7.4, 1 mM EDTA, 1 mM EGTA, 2 mM dithiothreitol (DDT), 0.1% (v/v) ethanol and 0.1 mM phenylmethylsulphonyl fluoride (PMSF). The light mitochondrial fraction, most enriched in lysosomes and peroxisomes, was isolated from the homogenates by differential centrifugation as described previously (Verheyden et al. 1992), and this fraction was then loaded on a multistep Nycodenz gradient [(two tubes, each containing 32 ml of 16–50% (w/v) Nycodenz (Sigma-Aldrich) prepared on isolation medium 2 (isolation medium 1 without sucrose)]. The gradients were stored overnight at +4 °C before use. The samples were centrifuged in a vertical rotor VTi50

(Beckman, Palo Alto, CA) at 100,000 gmax for 90 min with slow acceleration and deceleration. Purified peroxisomes (total volumes from two tubes, 12 ml) were then slowly diluted to 50 ml by adding isolation medium 2 dropwise. The peroxisomes were then resedimented at 40,000 gmax for 30 min in order to remove soluble proteins that had escaped from broken organelles. To obtain highly purified peroxisomes for channel-forming activity analyses the Optiprep gradient method was used as follows. Mouse livers were homogenised in an isolation medium 3 (1:4, w/v): 0.16 M sucrose, 12% (w/v) PEG1500, 10 mM MOPS, pH 7.4, 1 mM EDTA, 1 mM EGTA, 2 mM DTT, 0.1%

45 (v/v) ethanol, 0.1 mM PMSF and, in some experiments, a cocktail of protease inhibitors (Merck, Darmstadt, Germany). To prevent the admixture of purified peroxisomes with erythrocytes, the homogenate was centrifuged at 500 gmax and then at 1200 gmax. Mitochondria and peroxisomes were sedimented from the resultant supernatant by centrifugation at 23000 gmax. The pellet was resuspended in the isolation medium 3 (total volume 80 ml) and the suspension was loaded on a layer of 50% (w/v) Percoll solution (100 ml, Amersham Pharmacia Biotech AB).

The Percoll gradient was generated by centrifugation at 23,000 gmax for 60 min with slow acceleration and deceleration. Fractions enriched in intact peroxisomes were collected from the bottom of the gradient (total volume 12 ml) and diluted with 6 ml isolation medium 3. An aliquot (8–9 ml) of this suspension was layered on the top of an Optiprep gradient and centrifuged for 1 h at 65,000 gmax (slow acceleration and deceleration) using 60 ml centrifugation tubes and a fixed-angle rotor. The gradient consisted of 6 ml 20%, 6 ml 25%, 6 ml 30%, 6 ml 35%, 4 ml 40% and 4 ml 50% (w/v) Optiprep (60% (w/v) solution of iodixanol in water, Sigma). All solutions used to prepare the gradient also contained, 10 mM MOPS, pH 7.4, 1 mM EDTA, 2 mM DTT, 0.1% (v/v) ethanol and 0.1 mM PMSF. The Optiprep gradient was fractionated from the bottom of the tube into 14 fractions. Fractions enriched in peroxisomes (detected by measuring activity of the marker enzyme L-α-hydroxyacid oxidase), were combined, and diluted four-fold with the isolation medium-3. The organelles were sedimented by centrifugation at 100,000 gmax for 45 min, resuspended in 10 mM MOPS, pH 7.2, and stored at −80 °C. The peroxisomal ghosts (peroxisomes devoid of matrix proteins) were prepared in some experiments by sonication of purified peroxisomes in 50 mM Tris-Cl, pH 8.2 containing 0.15 M KCl. The ghosts were separated from matrix proteins by centrifugation at 100,000 gmax or by multistep sucrose gradient centrifugation as described (Antonenkov 1989). For the isolation of purified mitochondria from mouse liver the procedure described by Antonenkov et al. (2005) was used.

4.10 Protein analysis and immunochemistry

Protein content was determined according to Bradford (Stoscheck 1990). The SDS/PAGE method (Laemmli 1970) was used for separation of proteins by using 15% (w/v) Criterion Precast Gels (Bio-Rad, Hercules, CA, USA) or 15% (w/v) home made polyacrylamide gels. Visualisation of protein bands were done by silver or Coomassie blue staining.

46 SDS/PAGE gels were electroblotted onto nitrocellulose membranes according to standard procedures (Towbin et al. 1979). The blots were incubated with the appropriate primary antibodies, and detected with alkaline phosphatase-labelled anti-rabbit or anti-goat IgG as described (Novikov et al. 1994) or horse radish peroxidase–conjugated anti-mouse secondary antibody (Chemicon, Temecula, CA, USA). In the latter case, the signal was developed with SuperSignal West Pico Chemiluminescent substrate solutions (Pierce, Rockford, IL, USA) and detected with Hyperfilm ECL (Amersham Pharmacia Biotech AB). Polyclonal antibodies were generated in rabbits against: catalase and glutamate dehydrogenase from bovine liver (Chemicon), the recombinant N-terminal segment of human NADPH:cytochrome P450 reductase (Santa Cruz Biotech), a synthetic peptide corresponding to amino acid residues 185–197 of the human voltage-dependent anion channel, isoform-1 (VDAC1, Oncogene, Cambridge, MA, USA), rat peroxisomal 3-oxoacyl-CoA thiolase (Antonenkov et al. 1997), rat sterol carrier protein 2 (SCP-2, a gift of Prof. K. Wirtz, University of Utrecht, The Netherlands), a synthetic peptide corresponding to a predicted cytosolic domain (amino acids 403–417) of the rat 70 kDa peroxisomal membrane protein (PMP70) sequence (a gift of Prof. S. Alexson, Karolinska Institutet, Stockholm, Sweden). Antibodies against a synthetic peptide corresponding to the N-terminus of mouse Pxmp2 (NH2-APAASRLRVESELG) and the C-terminus of rat liver inner mitochondrial membrane marker D-3- hydroxybutyrate dehydrogenase (THFPGAISDKIYIH-COOH) were prepared in our laboratory by standard procedures (Sambrook 1989) .

4.11 Enzyme assays

Activities of marker enzymes for subcellular organelles: mitochondria – glutamate dehydrogenase (Leighton et al. 1968), lysosomes – acid phosphatase (Van Veldhoven et al. 1996) and β-galactosidase (Graham 1993), ER – esterase (Beaufay et al. 1974)), and peroxisomes – L-α-hydroxyacid oxidase (Baker & Tolbert 1966) and catalase (Baudhuin et al. 1964), were measured spectrophotometrically according to standard procedures. The activity of urate oxidase was measured by direct registration at 292 nm (pH 8.2) of the disappearance of urate (Leighton et al. 1968) in latency determination or by a coupled assay described elsewhere (Van Veldhoven et al. 1996), for detection of the activity in gradient fractions. Activities of fructose phosphate isomerase (Noltmann 1964) and glycerol-3-phosphate dehydrogenase (Gee et al. 1974) were

47 measured with fructose-6-phosphate and dihydroxyacetone phosphate as substrates, respectively. Alanine-glyoxylate aminotransferase activity was measured by detection of pyruvate formation after trapping glyoxylate by Tris-Cl buffer (Richardson & Thompson 1970). Units of enzyme activity are given as μmol of substrate consumed or product formed per minute. The unit of activity for the catalase, which shows first-order reaction kinetics, is expressed as the amount of enzyme which causes the decadic logarithm of the H2O2 concentration to decrease by one unit per min in a volume of 50 ml units defined by (Leighton et al. 1968). The latency of the enzymes was detected in freshly prepared peroxisomal fractions after resedimentation of the organelles as described (Antonenkov et al. 2004b). Free enzyme activity was measured before disruption of the peroxisomal membrane by Triton X-100 (0.05%, w/v, final concentration) which was used to reveal total enzyme activity (Antonenkov et al. 2004b).

4.12 Dietary studies

To investigate the effects of possible accumulation of phytol derivatives, 0.5% (w/w) phytol (Sigma Aldrich) was added to mouse chow for 12 weeks. 0.5% (w/w) Clofibrate (ICN Biomedicals, Aurora, OH, USA) was added to mouse chow for 14 days to investigate the effect of the peroxisome proliferator. To study water balance and renal function of Pxmp2-deficient mice, animals were deprived of water in the cycle of 17h without and 7h with free access to drinking water for 5 weeks after which urine was collected in metabolic cages with free access to water. In order to test the effect of excessive uric acid on the kidneys of Pxmp2-deficient mice, animals were provided with drinking water containing 1g/l uric acid (Sigma-Aldrich). After 4 weeks, a 3 week water deprivation cycle of 17h without water and 7 h with water was initiated after which urine was collected in urine cages with constant access to drinking water also containing 1g/l uric acid.

4.13 Production of recombinant Pxmp2 in insect cells

Mouse kidney total cDNA was used to amplify Pxmp2 cDNA by PCR with the forward (CCGGAATTCACCATGGCAACCTGCGGG) and reverse (CCGGAATTCTCACTTCCCCAGAGACC) primers containing the EcoRI restriction sites (shown in bold). The PCR product was cloned into a pFastBac1

48 donor vector (Invitrogen) and a BAC-TO-BACTM Baculovirus Expression System (Invitrogen) was used for generation of recombinant baculovirus and transfection of Spodoptera frugiperda (Sf9) insect cells. Mock transfection was performed with recombinant baculovirus containing the gene coding for human lysyl hydroxylase (gift of Prof. R. Myllylä, University of Oulu, Finland). After 72 h of infection the infected Sf9 cells were homogenised in 20 mM MOPS, pH 7.2 containing 0.25 M sucrose and 1 mM EDTA. The homogenate was centrifuged at

800 gmax for 10 min to remove nuclei and cell debris. The resulting postnuclear supernatant was centrifuged at 100,000 gmax for 45 min to obtain the total membrane fraction and the cytosol. The postmitochondrial particles fraction (PPF) was isolated by centrifugation of the post-nuclear homogenate at 6000 gmax for 20 min and the resulting supernatant was centrifuged at 100,000 gmax for 45 min. The compositions of the isolated fractions were determined using marker enzymes for different subcellular organelles.

4.14 Purification of native Pxmp2 protein from mouse liver peroxisomes

For purification of native Pxmp2, mice were kept on a standard diet containing 0.3% (v/w) clofibrate (Sigma-Aldrich) for 2 weeks after which peroxisomes were isolated from the liver tissue by the Optiprep gradient method as described in section 4.9 with the exception that the Percoll gradient centrifugation step was omitted. The purified organelles were diluted with 20 mM MOPS, pH 7.2 to 2.0–

2.5 mg/ml, sonicated and sedimented at 120,000 gmax. The pellet containing peroxisomes poor in matrix proteins (peroxisomal ghosts) was resuspended in 1.0 M KCl, 20 mM Tris-Cl, pH 9.8, sonicated and sedimented as before. The resulting peroxisomal preparation was suspended in 0.1M Na2CO3, pH 11.3 and rotated for 2 h at 4 °C to remove peripheral membrane proteins. The peroxisomal membranes were collected by centrifugation at 200 000 gmax and the integral membrane proteins were solubilised by incubating the membrane preparation in 20 mM MOPS, pH 7.2 containing 5% (v/v, final concentration) Genapol X-080 (Sigma-Aldrich) overnight at 4 °C. Peroxisomal membrane proteins solubilised by Genapol were separated from insoluble material by centrifugation and subjected to ion-exchange chromatography by using cartridges consisting of 1 Mono-Q Econo-Pac cartridge connected in head to tail configuration to 2 Mono-S cartridges (Bio-Rad). The cassette was equilibrated with 20 mM MOPS, pH 7.2, 0.2% (v/v) Genapol. After

49 a washing step with the equilibration buffer, the Mono-Q cartridge was removed and the Mono-S cartridges were washed once more with the same buffer containing 0.4 M NaCl. The bound proteins were eluted with a 0.4–2.0 M linear NaCl gradient at a flow rate of 1.4 ml/min. Immunoreactive fractions were initially analysed by dot blot by using an antibody against Pxmp2 and the fractions of interest were further investigated by Western blot and by SDS-PAGE followed by silver-staining. 8–10 fractions enriched with Pxmp2 were pooled, concentrated using Amicon membranes with a cut off 10 kDa (Millipore, Billerica, MA, USA) and applied to a SuperdexTM 200 column (Amersham Pharmacia Biotech AB) equilibrated with 20 mM MOPS, pH 7.2, 0.2% (v/v) Genapol, 0.15 M NaCl. Protein samples were chromatographed at a flow rate 0.8 ml/min and fractions of 2.0 ml were collected. 2–3 Pxmp2-immunoreactive fractions were pooled, concentrated and treated with 20 mM MOPS, pH 7.2 containing 4% (w/v) DDM (Sigma-Aldrich) overnight at 4 °C. The sample was then re- chromatographed on the SuperdexTM200 column equilibrated with 20 mM MOPS, pH 7.2, 0.1% (w/v) DDM, 0.15 M NaCl. The Pxmp2-immunoreactive fractions were collected, concentrated and stored on ice for detection of the channel-forming activity. The purity of Pxmp2 was determined by SDS-PAGE followed by silver staining or immunoblotting. To confirm the authenticity of Pxmp2, the protein band was cut from a gel stained with Coomassie Blue dye, followed by in-gel digestion with trypsin and identified from the peptide mass fingerprint by matrix- assisted laser desorption-ionisation mass spectrometry (MALDI-MS)(ABI Voyager-DETM STR Biospectrometry Workstation, Applied Biosystems). A SuperdexTM200 column connected to an AKTA-prime (Amersham Pharmacia Biotech AB) system and calibrated with protein standards (Sigma- Aldrich) was used to determine the molecular mass of native Pxmp2. Chemical cross-linking experiments using purified Pxmp2 were conducted at room temperature for 20 min with 0.2 mM ethylene glycolbis(sulphosuccinimidylsuccinate) (Sulpho-EGS, Pierce).

4.15 Reconstitution assay into planar lipid bilayer

Lipid bilayer experiments were conducted as described elsewhere (Benz et al. 1978) using 1% (w/v) diphytanoyl phosphatidylcholine (Avanti Polar Lipids, Alabaster, AL, USA) dissolved in n-decane as a membrane-forming lipid. Membrane across a circular hole (0.2 mm2) in the thin wall separating two

50 compartments in a Teflon chamber was formed by the painting technique. The aqueous salt solutions (analytical grade) were unbuffered (unless otherwise stated) with a pH around 6. Peroxisomal channels were inserted into the lipid bilayer at a high frequency in a 1.0 M KCl bathing solution and this salt concentration was used in all experiments with some exceptions. Solubilised membrane proteins or the purified Pxmp2 protein were introduced in both compartments of the chamber for incorporation into the bilayer, which took place spontaneously within 5–10 min. The temperature was kept at 20 °C and the applied voltage was +20 mV unless otherwise stated. A Planar Lipid Bilayer Workstation with a BC-535 amplifier and 4 pole low-pass Bessel filter (Warner Instruments, Hamden, CT, USA) was used for detection of reconstituted channels and pCLAMPS 9.2 software (Axon Instruments, Union City, CA, USA) was used for acquisition and analysis. Reversal potential measurements were performed by establishing a two- fold salt gradient across a bilayer containing a single channel and the current was measured at different membrane potentials. Current amplitudes were detected by cursor measurements at current increments that indicated insertion of a new channel in the artificial membrane. Current amplitudes were divided by the applied transmembrane voltage in order to calculate single-channel conductance. Data from measurements of 180–220 current increments were used to construct histograms of probability of the insertion events relative to their conductivity. For each histogram, the number of insertion events with a certain conductivity (bin size 0.1 or 0.25 nS) is presented as a percentage of the total amount of all insertion events registered.

4.16 Statistical analyses

Significance was determined using a two-tailed Student’s t test. When data from the measurements of blood or urine components with frequent deviations from the normal distribution were analysed, a non-parametric U-test was used.

52 5 Results

5.1 Channel-forming activity in the mouse peroxisomal membrane

This section is based on the published work (Antonenkov et al. 2005).

5.1.1 Characterisation of the purified peroxisomal fraction

Peroxisomes were isolated from the livers of C57BL/6J mice by using a medium containing PEG 1500 and Optiprep gradient centrifugation as described in the section 4.9. The purity of the isolated peroxisomes was confirmed by determination of marker enzyme activity: L-α-hydroxyacid oxidase (peroxisomes), glutamate dehydrogenase (mitochondria), esterase (microsomes) and acid phosphatase (lysosomes) (Fig. 4A). In addition, the distribution of organelles in the final Optiprep gradients were monitored by SDS-PAGE analysis (Fig. 4B) and by immunoblotting with antibodies raised against Pxmp2 and catalase (peroxisomal membrane and matrix marker proteins, respectively), VDAC1 (mitochondrial outer membrane protein), D-3-hydroxybutyrate dehydrogenase (D-3-HD, mitochondrial inner membrane protein) and NADPH cytochrome P450 reductase (P450- red, ER) (Fig. 4C). Based on the enzymatic analysis, contamination of gradient fractions enriched with peroxisomes (fractions 3–6) by mitochondria and microsomes was very limited. Furthermore, the protein pattern of these fractions was similar to that described previously for highly purified peroxisomal preparations from rodent liver (Alexson et al. 1985, Bodnar & Rachubinski 1991, Antonenkov et al. 2004a) and immunoblot signals of mitochondrial and microsomal marker proteins in peroxisomal gradient fractions were not detected by Western blotting under standard conditions (fractions 4–7, Fig. 4C). However, overexposed immunoblots of D-3-HD and P450-red indicated some traces of these marker proteins in the peroxisomal fractions (Fig. 4D).

53 A a b 40 30 1.2 30 3 20 20 1.1 10 Relative activity (% total) of 10 Density (g/cmDensity )

0 0 1.0 1 4 7 10 13 1 4 7 10 13 Top Fraction number Fraction number B C Pxmp2 97 66 Catalase 45

31 D-3-HD

21 VDAC1 14

1 4710 13 Top P 450-red Fraction number

1471013 D D-3-HD Fraction number

P 450-red

2 4 6 8 10 12 Fraction number

Fig. 4. Separation of peroxisomes from other cellular organelles by Percoll and Optiprep gradient centrifugation. A, Fractions from the Optiprep gradient were analysed for the following marker enzyme activities: glutamate dehydrogenase (a, filled bars), acid phosphatase (a, grey bars), esterase (a, open bars), L-α-hydroxyacid oxidase (b, filled bars), or protein content (b, open bars). Results are expressed as relative activity (relative protein content) in each fraction (% of the total activity loaded on the gradient). Enzyme (protein) recoveries varied between 84–106%. B and C, Proteins from equal volumes of each fraction were separated by SDS-PAGE and silver- stained (B) or immunoblotted (C). Molecular mass markers are shown on the left in B. D, Immunoblots of D-3-HD and P 450-red were overexposed to detect traces of mitochondria and microsomes in the gradient fractions containing mainly peroxisomes (fractions 3–7). (Antonenkov et al. 2005). Reprinted by permission from Birkhäuser Verlag AG.

54 Electron microscopic examination of the gradient fractions confirmed the results obtained using marker enzyme activities and immunoblotting (Fig. 5). The peroxisomal fraction consisted almost exclusively of peroxisomes, some of which were intact (structures with high electron density, Fig. 5A, B) while some had low levels of matrix proteins (peroxisomal ghosts) because of partial damage to the organelles during isolation. Only a low level of mitochondrial contamination was detected (indicated on Fig. 5B). The top of the Optiprep gradient (fractions 9–11, Fig. 4) contained a mixture of different organelles including mitochondria, ER, lysosomes and structures that resembled peroxisomal ghosts (Fig. 5C). A purified mitochondrial fraction with a low admixture of peroxisomes was isolated for measurements of channel-forming activities (described below). As can be seen in Fig. 5D, this fraction consisted mainly of mitochondria with some admixture of lysosomes and barely detectable peroxisomes.

Fig. 5. Electron micrographs of subcellular fractions obtained after Optiprep gradient centrifugation. A and B, The bottom gradient fractions (see Fig. 4A) contain highly purified peroxisomes, the few mitochondria are marked as mt. C, The fractions from the top of a gradient are enriched in mitochondria and fragments of ER, the fractions contain also structures resembling peroxisomal ghosts. D, Purified mitochondrial fraction. Original magnification was: 3800-fold, bar 1000 nm. (Antonenkov et al. 2005). Reprinted by permission from Birkhäuser Verlag AG.

55 5.1.2 Detection of channel-forming activity

Multiple-channel analyses were conducted with mouse liver peroxisomal preparations to investigate whether purified peroxisomes contain any channel- forming components. The results indicated two major types of channel-forming activity in the peroxisomal fraction with average conductances of 1.3 nS and 2.5 nS, respectively, in 1M KCl (Fig. 6A, B, panel 1). To further characterise the peroxisomal nature of these channels, the testing was performed also on the fraction enriched with mitochondria (pooled Optiprep gradient fractions 9–11, see Fig. 4). The most frequent channel-forming events detected in this fraction were with a conductance of 4.0 nS in 1M KCl (Fig. 6B, panel 2) which is known to represent the mitochondrial porin at its open state (Benz 1994). It was also noticed that the frequency of the channel-forming events with conductance of 1.3 nS and 2.5 nS was significantly lower in the mitochondrial fraction than in the peroxisomal fraction. Another channel-forming event observed in the mitochondrial fraction was with a conductance of 2.0 nS that is presumed to be the mitochondrial porin in its closed state (Benz 1994). However the 2.0 nS event was also detected in the peroxisomal preparations (Fig. 6B, panels 1 and 3) suggesting the possible existence of a third peroxisomal channel-forming activity similar to mitochondrial porin in the closed state. A fifth, unknown channel– forming activity with the conductance of 0.5 nS was detected in the mitochondrial but not in the peroxisomal fraction (Fig. 6B, panel 2). In order to confirm the peroxisomal origin of the channel-forming activities with conductances of 1.3 nS and 2.5 nS in 1 M KCl, a purified mitochondrial fraction was tested (Fig. 6B, panel 5). Channel-forming events in this fraction represented the activity of VDAC in its open (4.0 nS in 1M KCl) and closed (2.0 nS in 1M KCl) state, as was expected. Channel-forming events with a conductance of 1.3 nS and 2.5 nS in 1M KCl were also observed but with much lower frequency relative to the activity of mitochondrial porin. Additional experiments were carried out to localise the channel-forming activities in peroxisomes. Peroxisomal ghosts were generated from the isolated particles and membrane proteins were solubilised and analysed for channel- forming activity. Activities with a conductance of 1.3 nS and 2.5 nS in 1 M KCl were detected, similar to the findings with the peroxisomal preparations (Fig. 6B, panel 3). Separate experiments with peroxisomal matrix proteins failed to register any channel-forming activities indicating that these activities are localised in the peroxisomal membrane.

56 Fig. 6. Detection of the channel-forming activity. A, Multiple-channel recording of an artificial membrane in the presence of detergent-solubilised peroxisomes purified from mouse liver. Insertion events with an average conductance of 1.3 nS (*), 2.5 nS (**) and 4.0 nS (***) in 1 M KCl are indicated. B, Histograms of the multiple-channel conductance events observed in the presence of freshly prepared detergent- solubilised peroxisomes (panel 1; combined gradient fractions 3–6, see Fig. 4); fraction enriched in mitochondria and microsomes (panel 2; combined gradient fractions 9–11, see Fig. 4); peroxisomal ghosts (panel 3); purified peroxisomes which were solubilised by 0.5% (w/v) Genapol X-080 and then kept overnight at +4oC (panel 4); purified mitochondrial fraction (panel 5). C, Detection of the channel-forming activity of the sample consisting of aliquots of the gradient fractions 3–8 (see Fig. 4) at 1 M KCl (panel 1) and 2 M KCl (panel 2) in an aqueous phase, respectively. (Antonenkov et al. 2005). Reprinted by permission from Birkhäuser Verlag AG.

57 5.1.3 Properties of the peroxisomal channels

As shown in Fig. 6A, the measurements on freshly prepared peroxisomal preparations solubilised with the detergent Genapol X-080 indicated step changes in the membrane conductance mainly with average levels of 1.3 nS and 2.5 nS in 1 M KCl. The majority of steps in membrane conductance were directed toward an increase in conductance indicating that the peroxisomal channels were stable in the lipid bilayer. Overnight storage at +4 °C of the peroxisomal preparations solubilised by detergents (Genapol X-080 or Triton X-100) was followed by a large decline in the frequency of the insertion events with conductances of 1.3 nS and 2.5 nS in 1M KCl. The pattern of peroxisomal proteins detected by silver staining was not visibly affected in these preparations indicating that factors other than proteolysis may affect the activity of the solubilised channels. The decrease in peroxisomal channel-forming activities led to the detection of more frequent channel-forming events belonging to channels from contaminating organelles (Fig. 6B, panel 4 compare to panel 1). Channel-forming activities exhibited no voltage dependence at voltages as high as ±100 V, indicating that the 1.3 nS and 2.5 nS activities are provided by separate pore forming proteins instead of different states of the same channel. The overall channel-forming activity in the peroxisomal fraction was shown to be cation selective with a PK/PCl ratio in a KCl gradient close to 4. Experiments on the channel-forming activities of the two peroxisomal channels at different KCl concentrations revealed a linear dependence of a single-channel conductance on the KCl concentration (Fig. 6C). To study the effect of the KCl concentration on the relative frequency of peroxisomal channel-forming events, fractions 3–8 of the Optiprep gradient (see Fig. 4) were combined and the resulting preparation that contained, in addition to peroxisomes, some admixture of mitochondria and other organelles was used for the multiple-channel recording measurements in the presence of 1 M KCl (Fig. 6C, panel 1) or 2 M KCl (Fig. 6C, panel 2). The pattern of insertion events at 1 M KCl shows the presence of both peroxisomal and mitochondrial channel-forming activities in the preparation at approximately similar proportions. However, the pattern of insertion events at 2 M KCl indicated clear predominance of peroxisomal channel-forming activities with an average conductance of 2.5 nS and 5.0 nS, respectively over the activities with conductance 4.0 nS and 7.5–8.0 nS which can be caused by mitochondrial porin (VDAC from mammalian mitochondria shows linear dependence of conductivity on KCl concentration (De

58 Pinto et al. 1987)). The results suggest that the peroxisomal channel-forming activity may be latent at low KCl concentrations but it is easily detectable at higher salt concentrations.

5.2 Pxmp2 functions as a channel in the mammalian peroxisomal membrane

5.2.1 Studies with Pxmp2-deficient mice

This chapter describes the results with isolated peroxisomes from Pxmp2- deficient mice. The generation and phenotypic characterisation of Pxmp2−/− mice is presented in chapter 5.3.

Pxmp2-deficient peroxisomes are fragile

Nycodenz gradients were used to purify Pxmp2-deficient liver peroxisomes and enzyme activities were measured for peroxisomal matrix proteins L-α- hydroxyacid oxidase and catalase. In addition enzyme activities were determined for alkaline phosphatase (AP, lysosomal marker) and β-galactosidase (marker protein for the lysosomal matrix) and glutamate dehydrogenase (marker protein for the mitochondrial matrix). The results indicate increased leakage of matrix proteins from the peroxisomes of Pxmp2-deficient mice during the isolation compared to the wild-type controls (Fig. 7 and Fig. 8). The leakage is confined only to peroxisomes since no differences were observed in the activities of lysosomal and mitochondrial enzymes in the cytosolic fraction of Pxmp2−/− and wild-type peroxisomes (Fig. 8).

1.3

)

g a

t 1.2

(

i t

1.1 e

D (

1.0 1 4 7 1013 16 Fraction number Fig. 7. Effect of Pxmp2-deficiency on the integrity of peroxisomes in vitro. The postnuclear homogenates of liver samples were centrifuged using Nycodenz gradients. The activity of L-α-hydroxyacid oxidase was determined as a marker for soluble peroxisomal matrix proteins in both wild-type (dark grey bars) and Pxmp2- deficient (light grey bars) preparations. The ordinate axis (left) represents relative enzyme activities in each fraction (percentage of the total activity loaded on the gradient). The line connecting open circles indicates the density of the gradient (shown on the right ordinate axis). 60 *

40 **

20 Relative activity Relative of total) (% 0 se se la ta ta a Protein a h GDH HAOXC sp o actosidase h al p d G ci ß- A Fig. 8. Peroxisomes from Pxmp2-deficient mouse liver are fragile. Contents of protein and activities of the soluble matrix enzymes of different organelles: peroxisomes (catalase, L-α-hydroxyacid oxidase /HAOX/), lysosomes (Acid phosphatase, β-galactosidase), and mitochondria (glutamate dehydrogenase /GDH/) were measured in the cytosolic fraction and are presented as a percentage of the total amount in the postnuclear homogenate from wild-type (dark grey bars) and Pxmp2−/− mice (light grey bars). *P < 0.001, **P < 0.05 compared with control group, n = 3.

60 Similar results were obtained when several peroxisomal matrix proteins in the cytosolic fraction isolated from liver homogenates of Pxmp2−/− and wild-type mice were studied by Western blot (Fig. 9). Samples from cytosol and homogenate (marked as c and h, respectively in Fig. 9) were used for immunodetection of peroxisomal proteins: 3-oxoacyl-CoA thiolase (thiolase), sterol carrier protein 2 [Scp2, this protein shows dual, peroxisomal/cytoplasmic, localisation in the liver of rodents (Gallegos et al. 2001)], sterol carrier protein 2/3-oxoacyl-CoA thiolase (Scp2/thiolase), peroxisomal membrane protein 70 (Pmp70), and Pxmp2. The peroxisomal membrane proteins Pmp70 and Pxmp2 were not detected in the cytosolic fraction indicating that only soluble matrix proteins leaked out of the particles during homogenisation. As can be seen in Fig. 9, an increased leakage of matrix proteins from Pxmp2-deficient peroxisomes is observed.

Fig. 9. Effect of Pxmp2 deficiency on the integrity of peroxisomes in vitro. Immunodetection of peroxisomal proteins 3-oxoacyl-CoA thiolase (thiolase), sterol carrier protein 2 (Scp2), sterol carrier protein 2/3-oxoacyl-CoA thiolase (Scp2/thiolase), peroxisomal membrane protein 70 (Pmp70), and Pxmp2 from the cytosolic fraction (c) or homogenate (h) of Pxmp2-deficient (Pxmp2−/−) and wild-type (control) livers.

61 Furthermore, in electron microscopic studies, while peroxisomes in liver cross- sections of Pxmp2-deficient mice were normal (Fig. 10, panels 1 and 2), the isolated Pxmp2-deficient peroxisomes contained a matrix with a lower electron density when compared to control preparations (Fig. 10, panels 3 and 4).

Fig. 10. Morphological examination of Pxmp2-deficient peroxisomes. (1) Peroxisome visible in liver cross section of wild-type mouse, p = peroxisome; (2) cluster of Pxmp2- deficient peroxisomes; (3,4) peroxisomes from livers of Pxmp2−/− (3) and wild-type (4) mice isolated using a Nycodenz gradient (see Fig. 4; fractions 3–4 were collected for analysis). Bar: 500 nm (1,2); 1000 nm (3,4).

These results indicate an increased fragility of Pxmp2-deficient peroxisomes, which may be caused by abnormal osmotic behaviour by the organelles due to apparent limitations in the membrane permeability to solutes.

Latency of peroxisomal enzymes in Pxmp2-deficient peroxisomes

Latency is a phenomenon described for enzymes which are located inside cellular organelles and the organellar membrane forms a permeability barrier for the

62 substrate(s) of this enzyme. In this case, the activity of the enzyme is low when measured in the reaction mixture containing isolated intact organelles in the absence of membrane disrupting agents. This activity is defined as the free activity. When the membranes are disrupted by a detergent or by sonication, the enzyme activity, defined as total activity, shows increase and thus indicates latency. Cofactor-independent peroxisomal oxidases, such as urate oxidase, show no latency in peroxisomes from wild-type rodents, unlike catalase and cofactor- dependent enzymes (De Duve & Baudhuin 1966, Antonenkov et al. 2004b). However, activities of urate oxidase (Fig. 11) and L-α-hydroxyacid oxidase (Fig. 12) in the liver peroxisomes isolated from Pxmp2−/− mice showed latencies. Only a part of the total activities of the oxidases was latent, whereas activities of catalase and cofactor-dependent enzymes determined from the same peroxisomal preparations showed high latency. These observations demonstrate that the Pxmp2 deficiency leads to a partial restriction in the peroxisomal membrane permeability to metabolites in vitro.

1 1 1 min 120 *

v )

2 2 i l

c 80

3 3 t

a 40

(

R OD 0.02

- -/ 0 l 2 o y tr e it mp n re v x o ti F c P C a

Fig. 11. Effect of Pxmp2 deficiency on the latency of peroxisomal urate oxidase. Left panel: absorbance traces at 292 nm indicating oxidation of uric acid. Incubation medium contained: purified peroxisomes (40 μg) only (1); + uric acid (2); + 0.05% (w/v) Triton X-100 (3). Right panel: free activities of urate oxidase in peroxisomes from livers of control (dark grey column) and Pxmp2−/− (light grey column) mice, *P < 0.001, n = 12.

63 100 * 80 ty i v i 60 40 (% of total) of (% Free act 20 0 X H DH L PD AO G H Catalase Fig. 12. Permeability of Pxmp2-deficient peroxisomal membrane to substrates of peroxisomal enzymes in vitro. The latency of: HAOX, catalase, lactate dehydrogenase (LDH), and glycerol-3-phosphate dehydrogenase (GPDH) was detected. The last two enzymes are NADH-dependent. Free activity of the enzymes in peroxisomes isolated from wild-type (dark grey columns) and Pxmp2−/− (light grey columns) mouse livers is presented as a percentage of the total activity, *P < 0.001, n = 7–8.

Determination of uric acid, allantoin and oxalic acid in body fluids of Pxmp2-deficient mice

The diminished access of substrates into peroxisomes might affect their rate of metabolism by the corresponding peroxisomal enzymes such as, oxidation of uric acid to allantoin by urate oxidase. Concentrations of uric acid and allantoin in body fluids of Pxmp2−/− mice were measured to test this possibility. The level of uric acid in the serum of Pxmp2−/− mice was 103 ± 30 µmol/l compared to 68 ± 33 µmol/l in the wild-type control. In urine the ratio of uric acid concentration to creatinine was 0.85 ± 0.4 in Pxmp2-deficient mice compared to 0.18 ± 0.04 in the wild-type mice (Fig. 13A and Table 1). Concomitantly, the levels of allantoin were lower in body fluids of Pxmp2−/− mice than in wild-type controls (Fig. 13B), which indicates that the elevated levels of uric acid are due to a decrease in the degradation of this compound rather than to enhancement of purine catabolism in Pxmp2−/− mice. Since the total activity of uric acid oxidase was shown to be the same in liver homogenates of Pxmp2−/− and wild-type animals, it can be assumed that the decreased uric acid catabolism is caused by restriction in peroxisomal membrane permeability.

64 A B

)

e * i

** l

t ) n

a o

l 160 i 1.5 200 12 /

n l

r m

i o t

( m

e 120 n 10

r µ

c (

160 o

/ 1 t

o e

n t

n a

a 8 a

80 l

a r

l u

a 120

y 0.5 m

r 40 r 6

a u

a r

i e

e S

U 0 U 0 80 4

Fig. 13. Effect of Pxmp2 deficiency on the content of uric acid and allantoin. A, Determination of uric acid in blood (left panel) and urine (right panel) of wild-type (●) and Pxmp2−/− (○) mice. The excretion of uric acid and allantoin into urine is presented as molar ratios to creatinine. B, Allantoin content in blood (left panel) and urine (right panel) of wild-type (●) and Pxmp2−/− (○) mice. *P < 0.05; **P < 0.01.

To further test the permeability properties of the Pxmp2-deficient peroxisomal membrane we measured the content of oxalic acid, which is a peroxisomal metabolite and the final product of glycolic acid oxidation. The metabolism of glycolic acid in mammals takes place in the cytoplasm resulting in the formation of oxalic acid, and in peroxisomes with the production of glycine (Baker et al. 2004). If movement of glycolic acid into peroxisomes were restricted, an elevation in the oxidation of this compound in the cytoplasm would occur, leading to an increase in the production of oxalic acid. Pxmp2−/− and wild-type mice were injected with glycolic acid and the level of oxalic acid in urine was determined (Fig. 14). Both groups of mice displayed similar levels of urinary oxalic acid before the injection of glycolic acid. However, after the injection of glycolic acid, a higher rate of oxalic acid excretion in Pxmp2-deficient mice was detected compared to wild-type control. Measurements of the total activity of enzymes participating in the oxidation of glycolic acid in peroxisomes (HAOX, alanine- glyoxylate aminotransferase) or in the cytoplasm (LDH) in liver homogenates did not reveal any difference between Pxmp2−/− and wild-type mice, respectively.

65 ) )

g g

0 0

0 0 6 1 12 1 / / * h h

4 4

2 2

/ /

l l

o 10 o 4

m m

µ µ

( (

e e

t t

a a 2

l 8 l

a a

x x

o o

y y r r 0 a 6 a

n n

i i

r r

U U 123 123

Fig. 14. Detection of oxalic acid in urine. Wild-type (●) and Pxmp2−/− (○) mice were injected with glycolic acid and the excretion of oxalic acid in urine was measured. The absolute quantity of excreted oxalic acid was calculated as μmoles in the total urine collected within 24 h per 100 g body mass (left panel). The urine was collected for 24 h before glycolate injection (1), for 24h after injection (2) and from 24 h to 48 h after injection (3). The difference between the content of oxalic acid excreted in urine collected 24 h before and 24 h after the injection of glycolic acid is shown in the right panel. The difference was calculated for each mouse. *P < 0.05.

Detection of channel-forming activities in peroxisomes of Pxmp2−/− and wild-type mice

The restriction in peroxisomal membrane permeability in Pxmp2-deficient mice can readily be explained by the presence in the peroxisomal membrane of general diffusion pores that are formed by the Pxmp2 protein. Therefore, channel-forming activities in peroxisomes isolated from Pxmp2−/− mice by the Optiprep gradient method were analysed by an electrophysiological approach using a reconstitution assay in lipid bilayers. As was expected the preparations from wild type mice showed the most frequent insertion events with conductances of 1.3 nS and 2.5 nS in 1M KCl (Fig. 15A, B). When peroxisomal preparations from Pxmp2-deficient mice were subjected to multiple-channel recording, the fluctuations with a conductance of 1.3 nS in 1M KCl were lost. In order to confirm the link between Pxmp2 deficiency and channel-forming activity in peroxisomes a polyclonal antibody against the N-terminal part of Pxmp2 was utilised. As a result the pore- forming activity of a wild-type peroxisomal preparation treated with immune antiserum showed a clear decrease in insertion events with a conductance of 1.3 nS in 1M KCl compared to the control sample treated with pre-immune serum (Fig.15C).

66 A B C Wild-type Pre-immune 30 30 IgG 2.0 nS * 20 20

2 min 10 10 **** 0 0 Pxmp2-/- 30 30 anti-Pxmp2 ** 20 20 Insertion events (%) Insertion events 10 *** 10 0 0 0.25 1.25 2.5 4.0 0.25 1.25 2.5 4.0 G (nS) G (nS)

Fig. 15. Detection of channel-forming activity in membrane preparations. A, Multiple- channel recording of an artificial membrane in the presence of detergent-solubilised liver peroxisomes. Insertion events with a conductance of 1.3 nS (*), 2.0 nS (**), 2.5 nS (***) and 4.0 nS (****) are marked. B, Histogram of multiple-channel recordings registered in a peroxisomal fraction isolated from livers of wild-type (upper panel) or Pxmp2−/− (lower panel) mice. The 1.25 nS conductance level is marked by an arrow on each panel. C, Effect of anti-Pxmp2 antibodies on the channel-forming activity in peroxisomal preparations. The activity was registered after addition of the IgG fraction (5 μg) isolated from pre-immune (upper panel) or immune (lower panel) serum.

5.2.2 Channel-forming activity of recombinant Pxmp2

Mouse Pxmp2 cDNA was expressed in the insect cell line Sf9 using baculovirus expression system. The expressed Pxmp2 was located in postmitochondrial particles fraction (PPF, see chapter 4.13) of infected insect cells (Fig. 16A and Fig. 17A). Multiple-channel recordings of the PPF from mock-transfected cells indicated several types of channel-forming activities (Fig. 16B, C). Cells expressing Pxmp2 showed an additional type of channel-forming activity with a conductance of 1.3 nS in 1M KCl (Fig. 16C). When antibodies against Pxmp2 were applied this activity was suppressed (Fig. 17B). In addition, partially purified recombinant Pxmp2 showed channel-forming activity with a conductivity pattern similar to that found in preparations of native Pxmp2 isolated from mouse liver peroxisomes (Fig. 17C). These results indicate that recombinant Pxmp2 shows channel-forming activity similar to native Pxmp2 (see below).

67 Fig. 16. Detection of channel-forming activity in membrane preparations from insect cells producing recombinant Pxmp2. (A) Isolation of a Pxmp2-enriched membrane fraction from insect cells (see Methods, chapter 4.13). Based on marker enzyme activity detection (Fig. 17A), the postmitochondrial particle fraction (PPF) was enriched with fragments of ER (microsomes) and (micro)peroxisomes. Protein (40 μg) from each fraction were subjected to SDS/PAGE and immunoblotted with antibodies against mouse Pxmp2. Control – mock transfected cells. (B) Multiple-channel recordings of the PPF-preparations isolated from mock-transfected (left trace) insect cells or those expressing recombinant Pxmp2 (right trace). Insertion events with a conductance of 1.3–1.5 nS in 1M KCl are shown by asterisks. (C) Histogram of insertion events observed in the presence of PPF preparations isolated from mock- transfected (upper panel) and Pxmp2-expressing (lower panel) insect cells.

68 A Catalase GDH Esterase

Relative activity (%) activity Relative PPF osol AP LDH embrane Cyt Homogena Total m fraction

B Insertion events events (%) Insertion

0.25 1.0 2.0 3.0 4.5 0.25 1.0 2.0 3.0 4.5 C G (nS) G (nS) Insertion events (%) Insertion

0.1 1.0 2.0 3.0 4.0 G (nS)

Fig. 17. (A) Characterisation of subcellular fractions isolated from insect cells as described in section 4.13. Catalase, GDH, esterase, AP, and LDH were used as markers for (micro)peroxisomes, mitochondria, ER, lysosomes, and cytosol, respectively. Specific activities of the enzymes were determined in the corresponding fractions isolated from mock-transfected (dark grey columns) and Pxmp2-transfected (light grey columns) insect cells and are presented as a percentage of the activity in postnuclear homogenates (100%). Note that in addition to the marker for endoplasmic reticulum (esterase), the postmitochondrial particle fraction (PPF) also shows a high catalase activity, implying that this fraction contains a significant number of (micro)peroxisomes. The PPF also contains the highest concentration of recombinant Pxmp2 (see Fig. 16) and was used for detection of channel forming activity. (B) Effect of anti-Pxmp2 antibody on the channel-forming activity. The PPF was isolated from insect cells expressing Pxmp2, and an aliquot of solubilised protein (400 μg) was

69 incubated (30 min at +4oC) with IgG (5 μg) purified from pre-immune (left panel) or immune (right panel) serum. The preparations were then used for multiple-channel recording. Note the decrease in the relative amount of insertion events with a conductance of 1.3–1.6 nS in 1.0 M KCl after incubation of the samples with anti- Pxmp2 antibodies. (C) Multiple-channel recording of fractions containing recombinant Pxmp2 partially purified from insect cell PPF-preparations. The purification includes ion-exchange chromatography using Mono-S and Mono-Q2 columns. Note the abundance of insertion events with a conductance of 0.4 nS, 0.9 nS and 1.3–1.5 nS in 1.0 M KCl, respectively (see section 5.2.3 for details).

5.2.3 Purification and characterisation of native Pxmp2

In order to obtain further confirmation that Pxmp2 is active as a channel, the native protein was isolated from mouse liver (Fig. 18). Based on size-exclusion chromatography the predicted size of Genapol solubilised native Pxmp2 is 200 kDa (Fig. 18B, upper panel and Fig. 18C). This 200 kDa form is dissociated by the detergent n-dodecyl β-D-maltoside to the form with an estimated molecular mass of 66 kDa (Fig. 18B, lower panel and Fig. 18C). Two forms of the protein were detected after thermal treatment and size-exclusion chromatography (66 and 22 kDa, respectively) suggesting that purified Pxmp2 is a homotrimer (Fig. 18E). This was confirmed by crosslinking experiments (Fig. 18F). The identity of the isolated Pxmp2 protein was confirmed by western blot analysis (Fig. 18D) and by mass spectrometry.

5.2.4 Channel-forming activity of purified native Pxmp2

When purified Pxmp2 was analysed in reconstitution experiments a channel- forming activity with characteristic conductance fluctuations was registered (Fig. 19A) which showed three main conductance levels of 0.45 nS, 0.9 nS and 1.3–1.4 nS in 1M KCl (Fig. 19B). Different conductance sub-states of the purified Pxmp2 channel are not caused by voltage-gating since no voltage-induced closure at positive and negative potentials was observed (Fig. 19C). The 1.3 nS channel demonstrated a linear current-voltage relationship (Fig. 19D) and is moderately cation selective (PK+/PCl− = 2.3). When measurements were conducted with the low-conductance channel (0.45 nS in 1.0 M KCl) a similar ion selectivity was observed (PK+/PCl− = 2.4). Both, high (1.3 nS) and low-conductance (0.45 nS) channels were shown to depend nearly linearly on the KCl concentration (Fig. 19E).

70 A B

1 2 3 4 Genapol 97 66 45 DDM 31 10 14 18 22 26 21 * Fraction number

14 C D I 5.5 97 66 1 II 5.0 2 45

3 III 31 4.5 4

Log mol. mass Log 21 * 4.0 5 0 10 20 30 40 14

Fraction number er nk li E F _ Cross- + _ 97 66 * + 45

Heat treatment Heat 22 24 26 28 30 32 34 36 31

Fraction number 21

Fig. 18. Purification and characterisation of the quaternary structure of Pxmp2. (A) Partial purification of Pxmp2 (see section 4.14). Proteins were separated by SDS-PAGE and stained with silver nitrate. Lane 1, isolated peroxisomes (10 μg); lane 2, peroxisomal membrane preparation after treatment at pH 11.3 (8 μg); lane 3, peroxisomal membranes solubilised by Genapol (5 μg); lane 4, fraction enriched with Pxmp2 obtained after ion-exchange chromatography and first size-exclusion chromatography steps (0.6 μg). The protein identified by MALDI-TOF mass spectrometry as Pxmp2 is marked with an asterisk. The migration of molecular mass

71 (in kDa) markers (LMW Calibration kit, Bio-Rad) is indicated on the left. (B) Size- exclusion chromatography of partially purified Pxmp2 before (upper panel) and after (lower panel) treatment of the protein with n-dodecyl β-D-maltoside (DDM, see section 4.14). Proteins in 200 μl column fractions were separated by SDS-PAGE, transferred to nitrocellulose, and incubated with anti-Pxmp2 antibodies. (C) Mobility of Pxmp2 on a SuperdexTM200 column. The calibration curve was generated by plotting the log of the molecular mass of the standards versus mobility on the column. The molecular mass markers are: (1) catalase, 240 kDa; (2) lactate dehydrogenase, 140 kDa; (3) bovine serum albumin, 66 kDa; (4) carbonic anhydrase, 29 kDa; and (5) cytochrome c, 12.4 kDa. The void volume was determined using Blue Dextran (2000 kDa), which was detected in fractions 1–3. The mobilities of the different forms of Pxmp2 are marked by arrows: (I) Pxmp2 partially purified in the presence of Genapol (see Fig. 18B, upper panel); (II) partially purified Pxmp2 treated with DDM (see Fig. 18B, lower panel and Fig. 18E, upper panel); (III) Pxmp2 after thermal treatment (see Fig. 18E, lower panel) or after solubilisation from the membrane using 2.0% (w/v) SDS (blot not shown). (D) SDS-PAGE and immunoblot analysis of the purified Pxmp2. Left panel: fractions obtained after the final purification step (second size-exclusion chromatography step, see section 4.14) and which were enriched with Pxmp2, were concentrated and proteins (~20 ng) were subjected to SDS-PAGE followed by silver staining. Right panel: immunoblot analysis of the purified Pxmp2. The Pxmp2 band is marked by an asterisk. (E) Mobility of purified Pxmp2 on a SuperdexTM200 column before (upper panel) and after (lower panel) treatment of the protein at 75oC for 6 min. Traces of Pxmp2 visible in fraction 36 are probably caused by retention of some molecules of denatured protein on the column. (F) Cross-linking and immunoblot analysis of purified Pxmp2. The protein eluted from a SuperdexTM200 column was treated with ethylene glycolbis(sulphosuccinimidylsuccinate) (Sulpho-EGS, right lane). The control sample was not treated with the cross-linker (left lane). The immunoreactive product of the cross-linking is marked by an asterisk.

Spontaneous, voltage–independent transition of the 1.3 nS channel to subconductance states close to 0.45 and 0.9 nS were observed (Fig. 19F) indicating that the structure of Pxmp2 might be a trimer of three small channels, each of them with a conductance around 0.45 nS in 1M KCl. To further confirm this suggestion, experiments were conducted to stabilise the structure of the purified Pxmp2. Pxmp2 (~50 μg protein/ml) was incubated at 4oC for 10 days in 10 mM MOPS buffer, pH 7.2 or in 1.0 M Tris-Cl buffer, pH 8.0. Additionally, some samples that had been incubated in Tris-Cl buffer were then diluted with 1.0 M Tris base, final pH 11.2. After incubation of these samples for 30 min at 4oC, the channel-forming activity was measured. A significant increase in the relative number of insertion events at an average conductance of 1.4 nS in 1.0 M KCl was observed after stabilisation of Pxmp2 by Tris-Cl buffer (Fig. 20). In contrast,

72 incubation at basic pH led to the appearance of channels with a conductance of 0.3–0.4 nS in 1.0 M KCl. These observations further strengthened the suggestion that the Pxmp2 channel has a trimeric structure.

A B

12 10 8 6 4 2 Insertion events (%) 0 0.1 1.0 2.0 3.0 4.0 C D G (nS)

150 100 50 current (pA) 0 -120 -80 -40 0 40 80 120 -50 voltage (mV) -100 -150

EF3

G (nS) 1

0 0.25 0.5 1.0 2.0 KCl (M)

Fig. 19. Channel-forming activity of purified Pxmp2. (A) Multiple-channel recording of an artificial membrane after addition of purified Pxmp2 (~10 ng/ml, final concentration). (B) Histogram of conductance events of purified Pxmp2. (C) Dependence of the current produced by a single Pxmp2 channel (1.3–1.4 nS in 1.0 M KCl) on the applied voltage. The insertion of the Pxmp2 channel is indicated by an arrow. The channel with a conductance of 0.45 nS in 1.0 M KCl shows a similar dependence, indicating that both high- and low-conductance Pxmp2 channels are not voltage-gated. (D) Current-voltage relationship of the single channel; the slope of the linear regression is 1.38 nS. (E) Channel conductance as a function of KCl concentration. The figure presents measurements of the increments in current, which shows peaks in channel activity at an average conductance of 1.3 nS (upper line) or 0.45 nS (lower line) in 1.0

73 M KCl (see Fig. 19B). The mean conductance ± SD of 30–40 single events is shown. (F) Spontaneous closure of the high conductance (1.3 nS in 1.0 M KCl) channel. The applied voltage was +20 mV. Insertion events (%) events Insertion

0.1 1.0 2.0 3.0 4.0 5.0 0.1 1.0 2.0 3.0 4.0 5.0 0.1 1.0 2.0 3.0 4.0 5.0 G (nS)

Fig. 20. Stabilisation of purified Pxmp2 protein. Pxmp2 was incubated at 4oC for 10 days in 10 mM MOPS buffer, pH 7.2 (left panel) or in 1.0 M Tris-Cl buffer, pH 8.0 (middle panel). Some samples that had been incubated in Tris-Cl buffer were then diluted with 1.0 M Tris base, final pH 11.2. After incubation of these samples 30 min at 4oC, the channel-forming activity was measured (right panel). In these series of experiments, single-channel recording was performed using 10 mM MOPS, pH 7.2 in the bath solutions to prevent a shift in pH after addition of the protein samples.

5.2.5 The diameter of the pore of the Pxmp2 channel

Additional reconstitution experiments with salts other than KCl were conducted in order to obtain information on the diameter of the channels formed by Pxmp2 (Fig. 21A, B). The sequence of conductance for the salt solutions for high and low conductance channels (1.3 nS and 0.45 nS in 1.0 M KCl, respectively) was RbCl > KCl > NaCl > LiCl > TrisCl > Tetraethylammonium chloride (TEACl) indicating that the single-channel conductances followed the mobility order of the cations in aqueous solution. The low conductance observed with the large cations (Tris+ and TEA+) indicated that the rate of their penetration is limited by the size of the pore. This was confirmed when the logs of the relative conductance rates (G/σ) for different cations were plotted against the hydrated radii of these cations (Fig. 21B). This curve indicates linear dependence between the plotted parameters and predicts that the rates of penetration for cations with hydrated radii between 0.7–0.8 nm are only few percentage points of that for Rb+. This observation indicates that these dimensions are appropriate values for the Pxmp2 pore size. The conductivity rates of the cations for the low conductance channel (0.45 nS in 1M KCl) were shown to be comparable with the levels detected for the high

74 conductance channel (1.3 nS in 1M KCl) suggesting similar size for both types of channels (Fig. 21B).

A B

120 2 + Rb

100

G 50 80 )/( / ) x 100] 1 25 cation

σσ (mS/cm, ) (mS/cm, 40

G σ

G (nS, , ) 10 [( 5 0 0 l l Log / -C 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Cl s-C bCl a iCl R L KCl N Tri TEA r (nm)

Fig. 21. (A) Average conductivity of high (1.3 nS in 1.0 M KCl, light grey bars) and low (0.45 nS in 1.0 M KCl, open bars) conductance Pxmp2 channels in different salt solutions (1.0 M, final concentration). The average single-channel conductance ± SD (G) is shown on the right ordinate axis. Bulk electrolyte conductivity (σ, left ordinate axis) of the corresponding salt solutions is shown as dark grey bars. (B) Estimation of the size of the Pxmp2 channel using cations of different hydrated radii. The relative conductivity rates (G/σ) for high (●) and low (○) conductance channels (see Fig. 21A) were normalized to those of Rb+ (100 arbitrary units) and logarithms of the resulting values were plotted against the hydrated radii of the corresponding cations: Rb, 0.105 nm; KCl, 0.111 nm; NaCl, 0.163 nm; LiCl, 0.216 nm; Tris-Cl, 0.321 nm; TEA-Cl, 0.426 nm. The hydrated radii of cations were taken from the corresponding publication (Trias & Benz 1994).

In order to obtain more information on the channel size, the channel conductivity was measured in the presence of highly concentrated solutions of non-electrolytes with different hydrated radii (Fig. 22). The experimental setup is based on methodology described elsewhere (Krasilnikov et al. 1998). Shortly, high concentrations of non-electrolytes increase the viscosity of aqueous solutions and thus the mobility of ions decreases. The conductance of a channel will decrease when the size of non-electrolytes are small enough to enter the pore and increase the viscosity there. If the non-electrolytes are too large to enter the pore, the conductance of a channel is close to its value in the absence of the non- electrolytes. The lowest conductance rates were observed for small non- electrolyte molecules: ethylene glycol, glycerol, and arabinose, which indicates that these compounds can easily enter the Pxmp2 channel (Fig. 22). An increase

75 in conductance was observed in the presence of polyethylene glycols (PEGs) 200–400 indicating a growing restriction in the penetration of these molecules into the channel. A further increase in the size of PEGs, however, was not followed by a smooth growth in the conductivity, but instead, a plateau was observed for PEGs 600–3400, with a conductance that was significantly lower than the control (sample measured without non-electrolytes). Measurement with PEG6000 yielded a conductance comparable to the control (Fig. 22). These results with two levels of conductivity changes in the presence of non-electrolytes suggest that the structure of the Pxmp2-channel is funnel-like, resembling an hourglass and the dimensions of the narrowest space of the channel (channel friction) corresponds to the size of the smallest impermeable non-electrolyte. The channel conductance reached a plateau with PEG400 and PEG600 suggesting that these compounds are prevented from entering the narrowest pore of the channel. The hydrated radii of PEG400 and PEG600 are 0.7 nm and 0.78 nm, respectively, which are in agreement with a channel radius of 0.7–0.8 nm, predicted from the cation permeability experiments (Fig. 21). The conductivity measurements in the presence of PEG6000 are close to that of control preparations (without polyethyleneglycols) indicating that PEG6000 hardly penetrates the Pxmp2 channel mouth. This suggests that the dimensions of the channel mouth are comparable to or slightly larger than PEG6000 (radius = 2.5 nm).

1.4

1.2 ) S n ( 1.0 G

0.8

0.6 0.2 0.6 1.0 1.4 1.8 2.4 2.6 r (nm) Fig. 22. Estimation of the size of the Pxmp2 channel using non-electrolytes. The average conductivity ± SD (high conductance channel) was plotted against the hydrated radii of non-electrolytes: ethylene glycol, 0.26 nm; glycerol, 0.31 nm; arabinose, 0.34 nm; PEG200, 0.43 nm; PEG300, 0.60 nm; PEG400, 0.70 nm; PEG600, 0.78 nm; PEG1000, 0.94 nm; PEG2000, 1.22 nm; PEG3400, 1.63 nm; PEG6000, 2.50 nm. 20% (w/v) solutions of non-electrolytes containing 1.0 M KCl (final concentration) were added to both halves of the chamber. The solid line indicates the conductivity in control samples without non-electrolytes (dotted line shows the standard deviation).

76 5.2.6 Multiple-channel analysis of purified Pxmp2 using different organic anions as electrolytes

The channel-forming activity of the purified Pxmp2 protein was measured using various organic anions as electrolytes in order to further analyse the function of the Pxmp2 channel (Fig. 23). The results showed that small mono- and divalent anions, such as glycolate, pyruvate, 2-oxoglutarate, and other known peroxisomal metabolites can be transferred through the Pxmp2 channel (Fig. 23). It was found that if the size of the anion is over 200 Da, e.g. glucose-6-phosphate (260 Da), lactobionic acid (358 Da) or AMP (347 Da), the single-channel conductance of Pxmp2 is significantly decreased, indicating a restricted movement of these compounds inside the channel. In the presence of NAD (663 Da) only traces of channel-forming activity were detected with a conductance well below 20 pS. No channel-forming activities were observed in experiments conducted with ATP (507 Da), probably caused by the high net negative charge, that prevents the entry of ATP through the cation-selective Pxmp2 channel. We did not analyse in depth the dependence between the size of the anions and their conductance level since the hydrated radii of most of these anions are not known. However, as can be seen from the data presented in Fig. 23, the conductivity of the Pxmp2 channel is clearly dependent on the size of the anions if their molecular mass exceeds 200 Da (compare, e.g., panels a-c with panels d, g, and h), indicating partial restriction in the diffusion of these anions through the channel.

78 Fig. 23. Multiple-channel recording of purified Pxmp2 using different organic anions as electrolytes. The recordings were made using 1.0 M solutions of potassium or sodium salts of the anions as a bath. The solutions were buffered with 10 mM MOPS, pH 7.2. Molecular masses of the corresponding anions are indicated in brackets.

79 5.3 Phenotypic analysis of Pxmp2−/− mice

Generation and characterisation of Pxmp2-deficient mice

Pxmp2-deficient mice were generated by introducing a lacZneo targeting vector into exon 2 of the Pxmp2 gene (Fig. 24A). The targeting strategy was designed to leave as much as possible of the regulatory elements unaltered and not interfere with the expression of the catalytic subunit of DNA polymerase ε (PoleI), a DNA replication enzyme transcribed in the opposite direction with only 392 bp separating the translation initiation codons of these two genes. The inactivation of Pxmp2 was confirmed by Southern, Northern and Western analyses (Fig. 24B, C and D, respectively). Quantitative-real-time-PCR (see below) also demonstrated the absence of Pxmp2 transcripts in liver and kidney, which are tissues known to express Pxmp2 (Otte et al. 2003) (data not shown). The expression of PoleI in spleen tissues of Pxmp2−/− and wild type mice showed no difference indicating that the disruption of Pxmp2 does not affect the expression of PoleI (Table 5). The spleen was chosen for PoleI detection due to its substantial expression level of PoleI compared to nearly no detectable signal in liver and kidney (Otte et al. 2003). No external phenotype was observed in Pxmp2+/− mice and their litters were of normal size. Breeding of Pxmp2+/− mice provided offspring with a near normal Mendelian ratio with 22% Pxmp2−/−; 52.6% Pxmp2+/− and 25.3% Pxmp2+/+ out of 150 pups analysed. After 6 generations of back-crossing with C57BL/6 the Pxmp2−/− offspring were normal in appearance and generally healthy throughout the follow-up period of 2 years. Age and sex-matched Pxmp2−/− and C57BL/6 mice had similar body weights at 4 months of age as well as similar motor coordination and muscular performance, estimated by hanging and grid tests described elsewhere (Huyghe 2003). Pxmp2−/− mice are fertile, but female Pxmp2−/− mice were unable to nurse their pups due to their inability to produce milk (see below). Major tissues and organs of the Pxmp2−/− mice were examined by light microscopy at the age of 4 months and they appeared morphologically normal. Liver, heart and kidney were also examined by electron microscopy with no abnormalities found except flattening and fusion of podocyte foot processes that were occasionally observed in the kidney glomerulae in Pxmp2−/− mice (Fig. 25). In some cases liver peroxisomes of Pxmp2-deficient mice formed clusters containing organelles of different sizes although their shape was normal (Fig. 10,

80 panel 2). Tissues of Pxmp2-deficient mice were stained for β-galactosidase, an enzyme for the lacZ reporter gene which was inserted into the reading frame of exon 2 of Pxmp2. Specific staining with X-gal was detected in heart, kidney (Fig. 25) and liver (data not shown) tissues of Pxmp2−/− mice confirming that the reporter gene in Pxmp2-deficient mice is functional.

81 A 1 kb PoleI Pxmp2

S E X S B 1 2 3 4 Wild-type allele

2 2 3 lacz neo Targeting vector PoleI Pxmp2

S E X S X S B

I 1 2 2 3 lacz neo

Probe B C +/+ –/– +/+ +/– –/– Pxmp2 Pxmp2 Pxmp2 Pxmp2 Pxmp2

5.1 kb 4.2 kb

D +/+ –/– +/– Pxmp2 Pxmp2 Pxmp2 31 kDa

21.5 kDa 14.5 kDa

Fig. 24. Pxmp2 disruption strategy and verification of gene inactivation. (A) Schematic representation of the mouse Pxmp2 targeting vector and structure of the locus following gene targeting. The direction of transcription of Pxmp2 and PoleI are shown by angled arrows. The first exon of PoleI is denoted as a black box and the exons of Pxmp2 are presented as numbered boxes. Relevant restriction sites are shown: S = SacI, E = EcoRI, X = XbaI and B = BamHI. The arrows for lacZ (β-galactosidase

82 gene) and neo (neomycin phosphotransferase gene) indicate the direction of transcription of the corresponding genes. The locations of a 5´ external probe (Probe) for Southern analysis and primers used for PCR genotyping (small arrows) are shown. (B) Southern blot analysis of genomic DNA isolated from livers of wild-type (Pxmp2+/+), heterozygous (Pxmp2+/−) and homozygous (Pxmp2−/−) mice. DNA was digested with SacI and hybridised with the 5’ probe (shown in A). The 5.1 kb and 4.2 kb fragments represent wild-type and targeted alleles, respectively. (C) Northern blot analysis using total RNA isolated from liver and probed with Pxmp2 cDNA. (D) Western blot analysis of postnuclear liver homogenates using antibodies against Pxmp2. The molecular mass markers are indicated on the left.

Fig. 25. X-gal staining of heart (A,B) and kidney (C,D) tissues for β-galactosidase activity in wild-type (A,C) and Pxmp2 −/− (B,D) mice. E–G, Electron microscopy images from kidneys from wild-type (E) and Pxmp2−/− (F,G) mice. Scale bars: 20 µm in A–D, 2 µm in E and 1 µm in F–G.

Clinical parameters of Pxmp2-deficient mice

Laboratory values for several clinical parameters were measured from the serum and urine of Pxmp2−/− and wild-type mice and the results are presented in Table 1. In addition to elevated levels of uric acid, discussed above, no differences in the clinical values were observed between Pxmp2−/− and wild-type mice.

83 Fatty acid and bile acid profiles and dietary studies.

The fatty acid profile and levels of bile acid intermediates of Pxmp2-deficient mice were analysed in the blood serum and bile fluid, respectively, under normal dietary conditions (Tables 2 and 3). The results indicate that the deficiency in Pxmp2 in mice does not affect the levels of fatty acids and bile acid intermediates. To further investigate the phenotype of Pxmp2-deficient mice under abnormal dietary conditions, animals were fed diets containing phytol, which is a natural methyl-branched acyl alcohol metabolised through α-oxidation only in peroxisomes (Jansen & Wanders 2006) and clofibrate, a known peroxisomal proliferator (Wanders & Waterham 2006a). No weight gain or loss was observed in the Pxmp2−/− animals after maintaining the mice 7 weeks on a 0.5% phytol- containing diet and no visible effect on liver tissue was observed using histological (data not shown) or electron microscopic (Fig. 26) examination. Serum samples of phytol fed Pxmp2-deficient and control mice were analysed for phytanic acid, pristanic acid and other fatty acids, as mentioned in Table 4. The level of phytanic acid in the serum of phytol fed wild-type mice was 93.0 μM compared to 8.2 μM for animals fed a normal diet. The concentration of phytanic acid showed a similar increase in the Pxmp2−/− mice being 100 μM and 6.7 μM in the experimental and control group, respectively. In addition, no significant differences were observed between Pxmp2-deficient and wild-type mice in the levels of other fatty acids measured after administration of the phytol-containing diet (Table 4). No effect on the phenotype of Pxmp2-deficient mice was observed after 14 days of a 0.5% clofibrate diet. Animal and liver weights of clofibrate fed Pxmp2−/− mice did not differ from those of the wild-type control (data not shown) and electron microscopic examination revealed no differences in the appearance and number of liver peroxisomes in Pxmp2-deficient and wild-type mice (Fig. 26).

84 Table 1. Values of clinical tests in Pxmp2-deficient and wild-type mice.

Unit Wild-type Pxmp2−/− Na+ (serum) mmol/l 149 ± 5 (7) 149 ± 2 (6) K+ (serum) mmol/l 4.2 ± 1 (7) 3.9 ± 0.7 (6) Cl− (serum) mmol/l 132 ± 6 (7) 130 ± 5 (6) Bilirubin (serum) µmol/l 3.4 ± 1.4 (7) 2.9 ± 1 (6) Creatinine (serum) µmol/l 10.4 ± 3.2 (11) 10.2 ± 2.2 (11) Triglycerides (serum) mmol/l 0.74 ± 0.24 (7) 0.77 ± 0.23 (6) Cholesterol (serum) mmol/l 2.86 ± 0.41 (7) 2.47 ± 0.48 (6) Total protein (serum) g/l 52.2 ± 3.9 (7) 52.6 ± 1.6 (6) Uric acid (serum) µmol/l 68 ± 33 (12) 103 ± 30 (11)* Uric acid (urine) molar ratio to creatinine 0.18 ± 0.04 (5) 0.85 ± 0.4 (5)** Allantoin (serum) µmol/l 135 ± 28 124 ± 19 Na+ (urine) mmol/24h 0.23 ± 0.08 (5) 0.16 ± 0.07 (5) K+ (urine) mmol/24h 0.47 ± 0.10 (4) 0.29 ± 0.11 (4) Cl− (urine) mmol/24h 0.39 ± 0.14 (5) 0.29 ± 0.12 (5) Creatinine (urine) µmol/24h 6.21 ± 1.81 (5) 5.28 ± 1.39 (5) Protein (urine) mg/24h 1.73 ± 0.57 (4) 1.73 ± 1.24 (4) Osmolality (urine) mosm/kg 3290 ± 630 (5) 3560 ± 1460 (5) Estradiol (serum) nmol/l 19 days 4.0 ± 0.4 (8) 3.7 ± 0.5 (8) 5 weeks 4.1 ± 0.8 (8) 3.3 ± 0.3 (7)* 8 weeks 3.5 ± 0.9 (16) 3.6 ± 0.5 (15) E 15.5 2.5 ± 0.6 (7) 3.1 ± 0.5 (7) Progesterone (serum) nmol/l 19 days 5.3 ± 1.7 (5) 5.4 ± 0.4 (2) 5 weeks 1.6 ± 0.6 (7) 1.6 ± 0.8 (6) E 15.5 208 ± 95 (6) 133 ± 27 (7) Values represent means ± SD. Values in the parentheses indicate the number of independent measurements. * P < 0.05, ** P < 0.01.

85 Table 2. Concentration of bile acids in the bile fluid of Pxmp2-deficient and wild-type mice.

Bile acid (BA) Concentration in bile fluid (mM) wild-type (n = 5) Pxmp2−/− (n = 6) Total free BA 1.28 ± 0.25 1.75 ± 0.53 Total G-BA1 0.49 ± 0.23 0.43± 0.21 Total T-BA2 169 ± 84 156± 77

Total C24-BA 170 ± 84 157 ± 77

Total C27-BA 0.59 ± 0.31 0.50± 0.18 Total BA 170 ± 85 158 ± 78 C24-BA/C27-BA 297 321 Values represent means ± SD. 1 glycine conjugated bile acids, 2 taurine conjugated bile acids.

Table 3. Serum free fatty acids in Pxmp2-deficient and wild-type mice.

Fatty acid Concentration (µM) wild-type (n = 4) Pxmp2−/− (n = 4) C20:0/ph 8.2 ± 2.2 6.7± 1.0 C16:0 1870 ± 160 1310± 280 C21:0 2.7 ± 0.3 1.9± 0.6 C22:0 30.3 ± 4.3 22.1± 5.4 C23:0 1.5 ± 0.2 1.4± 0.3 C24:0 18.3 ± 2.9 13.6± 3.1 C25:0 0.7 ± 0.2 0.4± 0.1 C26:0 2.7 ± 0.2 1.9± 0.6 C20:0/ph /C16:0 0.004 0.005 C24:0/C22:0 0.60 0.62 C26:0/C22:0 0.09 0.09 Values represent means ± SD. C20:0/ph = 3,7,11,15-tetramethyl-hexadecanoic acid (phytanic acid) C16:0 = hexadecanoic acid; C21:0 = heneicosanoic acid; C22:0 = docosanoic acid; C23:0 = tricosanoic acid; C24:0 = tetracosanoic acid; C25:0 = pentacosanoic acid; C26:0 = hexacosanoic acid.

86 Table 4. Serum free fatty acids in Pxmp2-deficient and wild-type mice after consumption of a phytol diet.

Fatty acid Concentration (µM) wild-type (n = 7) Pxmp2−/− (n = 7) C20:0/ph 93.3 ± 12.6 100 ± 69 C16:0 1100 ± 140 770 ± 160 C18:x 2120 ± 350 1520 ± 360 C19:0/pr 12.5 ± 4.1 15.2 ± 13.3 C20:x 823 ± 109 629 ± 151 C22:n 207 ± 23 177 ± 59 C22:0 8.9 ± 1.3 7.9 ± 2.6 C23:0 6.4 ± 1.5 4.6 ± 1.4 C24:n 15.6 ± 2.6 12.1 ± 5.4 C24:0 8.0 ± 1.2 6.7 ± 2.1 C25:0 1.0 ± 0.2 0.80 ± 0.24 C26:0 0.38 ± 0.08 0.31 ± 0.07 Values represent means ± SD. C20:0/ph = 3,7,11,15-tetramethyl-hexadecanoic acid (phytanic acid); C16:0 = hexadecanoic acid; C18:x = all saturated and unsaturated octadecanoic/octadecenoic acids; C19:0/pr = 2,6,10,14-tetramethyl-pentadecanoic acid (pristanic acid); C20:x = all saturated and unsaturated eicosanoic/eicosenoic acids; C22:n = all unsaturated docosenoic acids; C22:0 = docosanoic acid; C23:0 = tricosanoic acid; C24:n = all unsaturated tetracosenoic acids; C24:0 = tetracosanoic acid; C25:0 = pentacosanoic acid; C26:0 = hexacosanoic acid.

87 Fig. 26. Analysis of liver peroxisomes from wild-type (A, C and E) and Pxmp2−/− (B, D and F) mice after dietary stress. Electron microscopy images of liver samples from mice fed with normal diet (A, B), clofibrate (C, D) and phytol (E, F). Scale bars: A,B and F – 2000 nm; C, D and E – 1000 nm.

Kidney function of Pxmp2-deficient mice

In order to test whether the elevated uric acid levels might have negative effects on the kidney, Pxmp2-deficient mice were tested for water balance and several parameters indicative of renal function. The ability to concentrate urine was not impaired in Pxmp2−/− mice, since urine osmolality levels showed a similar increase in Pxmp2-deficient and wild-type mice in response to water deprivation and no differences were observed in water consumption, urine output and Na+, K+, Cl− levels between experimental and control groups (data not shown). In addition, a water deprivation cycle did not cause any visible abnormalities in the kidneys as observed by light or electron microscopy (data not shown). Studies with urate oxidase (Uox) knock-out mice have shown that the levels of uric acid in Uox−/− mice are 10 and 9 fold compared to wild-type in the serum and urine, respectively, leading to hyperuricemia and urate nephropathy (Wu et al. 1994). In the case of Pxmp2-deficient mice the uric acid levels are 1.5 and 4.5

88 fold higher in the serum and urine, respectively, compared to the wild-type control (Table 1, Fig. 13), but these levels cause no visible effect on the kidneys of Pxmp2−/− mice. Our attempts to increase the serum uric acid levels in Pxmp2−/− mice providing drinking water with 1g/l urate failed. No effect of the water containing uric acid was observed on the life time of the animals, their weight, kidney histology and kidney-related parameters in the urine of Pxmp2−/− mice (data not shown). Since Mpv17−/− mice have been shown to develop systemic hypertension (Clozel et al. 1999), the hemodynamic parameters of Pxmp2-deficient mice were measured as described in section 4.5. The heart rate and mean arterial pressure of Pxmp2−/− and wild-type controls were followed, but no differences were observed between the groups (data not shown).

Quantitative real-time PCR analysis of selected genes in Pxmp2-deficient mice

Quantitative real-time PCR was used to study the effect of the disruption of Pxmp2 on the expression levels of selected peroxisomal genes such as urate oxidase (Uox), xanthine oxidoreductase (Xor), 70 kDa peroxisomal membrane protein (Pmp70), hydrogen peroxide degrading enzyme catalase (Cat), carnitine acetyltransferase (Crat), straight- and branched-chain fatty acid oxidases (Acox1 and Acox2, respectively) and multifunctional enzymes type 1 and 2 (Mfe1 and Mfe2, respectively) in liver and kidney tissues of the Pxmp2-deficient mice. In addition, mRNA levels of DNA polymerase epsilon (PoleI) was measured in spleen. No significant changes in the mRNA levels using samples from liver or kidney were observed. The exception was a slight elevation in the expression level of Xor (p < 0.05). The mRNA levels for PoleI in the spleen from Pxmp2−/− mice showed no difference when compared to the wild-type control. We also investigated whether the absence of the Pxmp2 transcript in Pxmp2−/− mice increased the mRNA expression of the two other members of the Pxmp2 family: Mpv17 and M-LP, and also the Pex11 family of proteins that have recently been suggested to have a role in the transfer of metabolites across the peroxisomal membrane (Thoms & Erdmann 2005). Quantitative Real-Time PCR experiments did not, however, reveal any significant differences in the mRNA levels of Mpv17, M-LP, Pex11α, Pex11β and Pex11γ in the liver and kidney of Pxmp2−/− and wild-type mice (Table 5).

89 Table 5. Expression levels of selected genes in liver, kidney and spleen from Pxmp2- deficient mice relative to wild-type controls.

Gene Accession no. Relative expression 1 liver kidney spleen PoleI XM_194234 1.1 Mpv17 NM_008622 1.1 1.0 Mlp AF305634 0.7 1.0 Pex11α NM_011068 1.1 0.9 Pex11β NM_011069 1.1 1.0 Pex11γ NM_026951 0.9 1.0 Uox NM_009474 0.6 Xor XM_192827 0.8 1.5* Cat NM_009804 0.7 1.3 Pmp70 NM_008991 0.6 1.5 Crat NM_007760 1.0 1.2 Acox1 NM_015729 0.7 1.2 Acox2 NM_053115 0.8 1.4 Mfe1 AK004867 0.8 1.6 Mfe2 NM_008292 0.6 1.6 1 For the relative quantification of gene expression the results were normalised using 18S or Gapdh as an endogenous control for each sample and the data from wild-type mice was set to 1. Values represent means of 3–6 independent measurements. * P < 0.05.

Mammary gland phenotype of the Pxmp2−/− mice

Pxmp2−/− male mice are fertile. When Pxmp2−/− female mice were mated with wild-type, Pxmp2+/− or Pxmp2−/− male mice, normal pregnancy was observed in each combination. The number and weight of the fetuses in the pregnant Pxmp2−/− dams at embryonic day 17.5 was comparable to wild-type controls (data not shown). Litter size and the weight of the newborn pups produced from Pxmp2−/− mothers also showed no differences from the wild-type mice [(litter size (an average number of pups): wt 5.8 ± 1.2, n = 6 litters; Pxmp2−/− 6.1 ± 1.5, n = 31 litters; weights of the newborn pups: wt ♂1.32 ± 0.1g, n = 26 pups (12 litters), ♀ 1.25 ± 0.1g, n = 30 (12 litters), Pxmp2−/− ♂1.32 ± 0.1g, n = 36 (12 litters), ♀1.28 ± 0.1g, n = 31 (12litters)]. However, after breeding, Pxmp2−/− dams were unable to maintain their newborn pups alive since they were frequently found dead or eaten after 24 h from birth. The genotype of the pups themselves did not influence this lethality since the outcome from matings of wild type ♂ and Pxmp2−/− ♀ mice was not different from that of matings of Pxmp2−/− ♂ and Pxmp2−/− ♀

90 animals (from 14 wt ♂ Pxmp2−/− ♀ matings in 12 cases all pups were lost and from 9 Pxmp2−/− ♂ x Pxmp2−/− ♀ matings in 7 cases all pups were lost, first pregnancies). In a few cases when the litter or part of the litter from the Pxmp2−/− mother survived the pups were clearly smaller and thinner during weaning, but seemed to gain size after switching to solid food. Newborn pups from Pxmp2−/− mothers were transferred immediately after birth to wild-type foster mothers, in order to test if the lethality of the pups is caused by the inability of Pxmp2−/− mothers to nourish their offspring. Newborn litters from 10 Pxmp2−/− dams were transferred to wild-type mice and over 60% of the pups survived and reached adulthood. In cases were wild-type newborn pups were transferred to Pxmp2−/− mothers, none of the pups survived after 24–48 h, indicating a dysfunction in the nursing of the Pxmp2−/− dams. Mammary gland morphology of female Pxmp2−/− mice was analysed by whole-mount (Fig.27) and histological staining (Fig. 28). We found that virgin Pxmp2−/− mice showed a complete absence of ductal morphogenesis in the mammary gland while the mammary glands of age matched virgin wild-type mice displayed extensive ductal elongation and branching. The dysfunction appears to occur at puberty since no differences in the mammary gland morphology between Pxmp2−/− and wild-type can be observed at 2 weeks of age (Fig. 27A, B), suggesting that the embryonic development of Pxmp2−/− mammary gland is normal. Puberty in female mice is accompanied by the appearance of terminal end buds followed by ductal outgrowth and branching of mammary epithelia into the underlying fat pad, eventually filling the entire fat pad area in the adult mouse (Hinck & Silberstein 2005). Terminal end buds can be observed in the wild-type epithelium (Fig. 27C, E) and the wild-type mammary ducts elongate until the epithelia fills the entire stroma by the age of 12 weeks (Fig. 27G). In Pxmp2−/− epithelium, however, no terminal end buds can be observed and the mammary tree remains as a ductal rudiment (Fig. 27D, F, H).

91 Fig. 27. Morphological analyses of mammary glands of Pxmp2−/− mice during puberty and pregnancy. (A–L) Whole-mount preparations of inguinal glands from wild-type (A, C, E, G, I, K) and Pxmp2−/− (B, D, F, H, J, L) females were prepared from the following developmental stages: day 14 (A and B), 5 weeks (C and D), 8 weeks (E and F), 12 weeks (G and H), 12.5 days of pregnancy (I and J) and 15.5 days of pregnancy (K and L). Scale bar: A–L 1 mm.

92 Another type of mammary gland branching arises through recurrent estrous cycles and during pregnancy by the formation of alveolar buds (Brisken 2002). During lactogenesis, towards the end of pregnancy, the lumen of the alveoli increase dramatically leading to the initiation of full lactation (Neville et al. 2002). The ductal rudiments of mammary glands in pregnant Pxmp2−/− mice seem to follow the lobuloalveolar development (Fig. 27J, L), but the density of alveoli is smaller than in the wild-type as can be seen from the whole-mount (Fig. 28A, B) and hematoxylin-eosin staining (Fig. 28C, D) of mammary gland sections from lactation day 1. However, hematoxylin-eosin staining and immunohistochemical staining for whey acidic protein (protein component of milk) show the presence of milk in the lumen of lactating Pxmp2−/− mammary glands (Fig. 28E–H). A delay in the movement of lipid droplets into the alveolar lumina was observed in some parts of the lactating mammary glands of Pxmp2-deficient mice (Fig. 28F). Taken together, these observations suggest that the mammary phenotype is mainly caused by impaired pubertal growth of mammary epithelia, even though secretion of lipids into the milk might also be affected in Pxmp2−/− mice. The female reproductive hormones estrogen and progesterone that are known to be required for the pubertal development of the mammary gland (Sternlicht et al. 2006) were measured from the serum of Pxmp2−/− and wild-type mice (Table 1). A slight decrease in the level of estradiol was observed in Pxmp2-deficient mice (3.3 ± 0.3 nmol/l) compared to the wild-type (4.1± 0.8 nmol/l) at the age of 5 weeks. No differences were observed at 19 days, 8 weeks and at pregnancy day 15. The levels of progesterone in Pxmp2−/− mice were comparable to the wild- type animals at 19 days, 5 weeks and at pregnancy day 15. The presence of Pxmp2 in the mouse mammary gland was investigated by western blot. The signal for Pxmp2 in the homogenates of intact mammary glands was detected at several developmental time points (Fig. 29A) which indicates that Pxmp2 is expressed in the mammary glands of both virgin and pregnant mice. Furthermore, Pxmp2 was present in the homogenate of the cleared fat pad in 3 week old mice (Fig. 29A, right panel), suggesting that Pxmp2 is expressed in the adipocytes of the mammary fat pad. PoleI protein was determined by western blot analysis from mammary gland homogenates of Pxmp2−/− and wild-type mice to confirm that the synthesis of DNA-polymerase ε has not been affected by the disruption of Pxmp2 (Fig. 29B). Levels of PoleI showed no difference when the wild-type and Pxmp2-deficient mice in both virgin and pregnant animals were compared (Fig. 29B left and right

93 panel, respectively), indicating that the disruption of Pxmp2 has no influence on the protein level of PoleI in the mammary glands of Pxmp2-deficient mice.

Fig. 28. Morphological and histological analyses of mammary glands of Pxmp2−/− mice on lactation day 1. (A–B) Whole-mount preparations of inguinal glands from wild-type (A) and Pxmp2−/− (B) females, scale bar: 1 mm. (C–H) Histological sections from wild- type (C, E, G) and Pxmp2−/− (D, F, H) mammary glands stained with hematoxylin and eosin (C–F) or immunostained with whey acidic protein (G, H) to highlight milk protein production. Scale bars: C, D: 500 µm, E, F: 50 µm, G, H: 100 µm.

94 A -/- -/- Wild-type Pxmp2-/- Control Pxmp2 E17.5 Pxmp2 12 wk Wild-type 8 wk Wild-type E12.5 Wild-type E17.5 Control

21.5 kDa

B 5 wk 8 wk E15.5 E18.5 Wt -/- Wt -/- Wt -/- Wt -/- PoleI

β-tubulin

Fig. 29. Western blot analysis of Pxmp2 and PoleI proteins in the mammary glands of wild-type and Pxmp2-deficient mice. A, Detection of Pxmp2 in intact mammary gland at several developmental ages (left panel) and in cleared fat pads from 3 weeks old Pxmp2−/− and wild-type mice (right panel). Control – sample from liver homogenate from wild-type mouse. B, detection of protein level of PoleI in the mammary gland of Pxmp2−/− (−/−) and wild-type (wt) mice from 5 and 8 weeks old virgin (left panel) or from 15.5 and 18.5 days pregnant animals (right panel). β-Tubulin was used as a loading marker in PoleI detection.

96 6 Discussion

The data presented here lead to the conclusion that the mammalian peroxisomal membrane contains channel (pore)-forming proteins that may be responsible for the traffic of solutes across the membrane. One of these proteins, Pxmp2, was isolated and its channel-forming properties were analysed. Moreover, a Pxmp2- deficient mouse model was generated and the phenotype of Pxmp2−/− animals was characterised. The results show that deficiency of the peroxisomal membrane channel Pxmp2 leads to abnormal development of epithelia in mammary glands of female mice.

6.1 Pore forming properties of the peroxisomal membrane proteins

Using an electrophysiological approach (reconstitution of peroxisomal membrane proteins into an artificial lipid membrane followed by current measurements) we show, that the mouse peroxisomal membrane contains channel-forming activities with conductances of 1.3 nS and 2.5 nS in 1M KCl, respectively. The properties of these peroxisomal channels are different from those of other cellular channels such as, for example VDAC of the outer mitochondrial membrane. Multiple- channel recordings with isolated mitochondrial fractions revealed channel- forming activities with conductances of 4.0 nS and 2.0 nS which are characteristic for VDAC in an open and closed configuration, respectively (Benz 1994). It has been shown that mitochondrial VDAC forms a large diffusion pore with an estimated diameter of 2.5–3.0 nm that allows the passage of not only small metabolites but also ATP and cofactors such as NAD/H, NADP/H and CoA through the mitochondrial outer membrane (Benz 1994). Many indirect observations have suggested the presence of channels in mammalian peroxisomes. Several peroxisomal oxidases such as urate oxidase, D-amino acid oxidase, and L-α-hydroxyacid oxidase, show no structure linked latency (Leighton et al. 1968, Baudhuin 1969, Antonenkov et al. 2004b). Rapid, temperature-independent incorporation of radioactive compounds such as carnitine, sucrose, NAD+, ATP, and others into isolated peroxisomes and leakage of radioactive sucrose from liposomes reconstituted with peroxisomal membrane proteins have been observed (Van Veldhoven et al. 1987). However, some experimental results, such as an apparent existence of a pH gradient across the peroxisomal membrane (see section 2.5.2), can be interpreted as inconsistent with the presence of porin-like channels in peroxisomes from mammals and yeasts.

97 A two channel hypothesis has been suggested recently to explain the unusual permeability properties of the peroxisomal membrane (Antonenkov et al. 2004b). According to this concept the mammalian peroxisomal membrane contains two types of channels in addition to membrane transporters which are specific for bulky solutes (NADH, NADPH, CoA, ATP and others) or lipid-soluble metabolites (fatty acids, bile acids and others). The transfer of bulky solutes by means of channels is restricted while small metabolites with a size up to 300 Da can diffuse freely through peroxisomal channels. In this way peroxisomes may be able to maintain in vivo their own, functionally independent pool of cofactors and, at the same time, share a common pool of small metabolites with the surrounding cytoplasm. Currently, two peroxisomal membrane transporters have been identified. One of these transporters was described as an ATP-dependent ATP/AMP antiporter (Palmieri et al. 2001, Visser et al. 2002). Other transporters are believed to be represented by the proteins related to the superfamily of ABC transporters (Theodoulou et al. 2006). The substrate specificity of the peroxisomal ABC transporters is not clear, but it is believed to be the translocation across the peroxisomal membrane of lipid compounds poorly soluble in water, e.g., very long chain fatty acids (see chapter 2.6.5). Therefore, the ATP/AMP antiporter and ABC transporters may execute functions that are not carried out by the peroxisomal membrane channels. Some reports suggest the presence of transporters specific for small solutes in the peroxisomal membrane. A carrier for acylcarnitines (Fraser et al. 1999) and a monocarboxylate transporter (McClelland et al. 2003) were detected in rat liver peroxisomal fractions by using antibodies generated against the corresponding mitochondrial proteins. However, the purity of the peroxisomal fraction was not characterised and no additional analysis of the transporters was reported. In addition, phosphate transport activity has been detected in peroxisomes isolated from bovine kidney (Visser et al. 2005). However, the molecular nature of the protein responsible for this activity has not been established yet. Recent reports have described that peroxisomes harbour Ca2+ which may regulate peroxisomal metabolism by currently unknown mechanisms (Drago et al. 2008, Lasorsa et al. 2008). The results describing the intraperoxisomal Ca2+ concentration compared to the cytoplasm and the possible mechanisms of influx and efflux of Ca2+ are, however, contradicting. According to Lasorsa et al. (2008) mammalian peroxisomes sustain a Ca2+ gradient based on the observations of 20- fold higher Ca2+ concentration in the peroxisomal lumen compared to the

98 cytoplasm in the resting state of the cell. Also, a large transient Ca2+ uptake is observed in peroxisomes after stimulation of cells with agonists that induce Ca2+ release from intracellular stores. In addition, experimental data in this study suggests the existence of H+ and Na+ gradients for the peroxisomal Ca2+ transport. On the other hand, Drago and co-workers (2008) have shown that the Ca2+ concentration in living cells at resting state is similar to that of the cytoplasm, increase in the cytosolic Ca2+ concentration is followed by a slow increase in the intraperoxisomal Ca2+ level, and that the Ca2+ influx is driven neither by an ATP- dependent pump nor by membrane potential nor by a H+(Na+) gradient. Conclusions from this study were drawn that the peroxisomal membrane forms a permeability barrier for Ca2+ diffusion as a low pass filter preventing the organelle from taking up short lasting cytosolic Ca2+ transients but allowing equilibration of peroxisomal Ca2+ levels with that of the cytoplasm during prolonged Ca2+ increases (Drago et al. 2008). These conflicting results may be explained by the presence of channels for small charged solutes in the peroxisomal membrane which are required for generating a Donnan equilibrium. As discussed in chapter 2.5.2, the explanation for the spectacular differences in the results from measurements of peroxisomal pH may be the existence of channels through which protons and hydroxyl ions may pass more or less freely following the overall charge difference of molecules (such as proteins) between peroxisomal lumen and the cytoplasm. The same mechanism of Donnan equilibrium may also explain the contradicting results about apparent Ca2+ gradient across peroxisomal membrane.

6.2 The function of Pxmp2 as a peroxisomal membrane channel

Our results demonstrate that the peroxisomal membrane protein Pxmp2 is responsible for channel-forming activity with a conductance of 1.3 nS in 1M KCl that was registered in the total peroxisomal fraction (see section 5.2). The evidence for the channel function of Pxmp2 are the following: 1) The inactivation of Pxmp2 in mice leads to partial restriction in the traffic of small metabolites through the peroxisomal membrane in vitro and in vivo; 2) Peroxisomal membrane preparations isolated from Pxmp2-deficient mice do not display the characteristic channel-forming activity with a conductance of 1.3 nS in 1M KCl, described for wild-type mice; 3) The activity with a conductance of 1.3 nS in 1M KCl is inhibited after treatment of solubilised peroxisomal membrane proteins from wild-type mice with antibodies against Pxmp2; 4) The activity with a

99 conductance of 1.3 nS in 1M KCl can be detected in insect cells expressing recombinant Pxmp2 and this activity is inhibited with an antibody against Pxmp2; 5) Native Pxmp2 isolated from mouse liver shows channel-forming activities with three conductance levels, the highest one being 1.3–1.4 nS in 1M KCl. There are two possibilities to explain the nature of the sub-conductance levels of the isolated Pxmp2 channel. The sub-conductance states may be a result of channel-gating, described, for instance, for the mitochondrial VDAC (Benz 1994). Another possibility is that the large channel represents a cluster of small channels. Examples of such clusters are: some antibiotics forming membrane pores (Kaulin et al. 1998) or bacterial porins (Trias & Benz 1994). We observed spontaneous transition of the 1.3 nS conductance level (in 1.0 M KCl) to lower conductance levels (0.9 nS and 0.45 nS in 1 M KCl, respectively) in lipid bilayer experiments with isolated Pxmp2 (Fig. 19F). If the 0.45 nS conductance reflected the closure of the 1.3 nS channel, the pore radii of the large and small channels should be different and the closure event would affect the cation-anion selectivity of the channel, as has been shown for VDAC (Benz 1994). On the other hand, if the large conductance channel (1.3 nS in 1.0 M KCl) represents a cluster of smaller ones, then the 1.3 nS and 0.45 nS conductance channels should demonstrate very similar pore diameter and ion selectivity. Our data favour the cluster organisation model for the Pxmp2 channel since a high similarity in ion selectivity and predicted pore diameter between the 1.3 nS and 0.45 nS channels have been observed. One can assume that the discrete conductivity levels of 0.45 nS, 0.9 nS and 1.3–1.4 nS in 1.0 M KCl observed for the Pxmp2 channel indicates the presence of three pores per one cluster. In this model the channel-forming activities with conductances of 0.45 nS and 0.9 nS in 1.0 M KCl would indicate the insertion of Pxmp2 channel clusters in which only one or two pores, respectively, are in an open configuration. A dependence of the single-channel conductance of 1.3 nS and 0.45 nS channels on the KCl concentration was observed (Fig. 19E) and at all KCl concentrations tested, the conductance of the 0.45 nS channel was always about one third that of the 1.3 nS channel, which further confirms the `cluster` model. Our results indicate that Pxmp2 has an oligomeric composition forming a trimer. Such trimeric structures have been described for bacterial porins (Benz & Bauer 1988) and the preprotein translocation channel of the outer membrane of mitochondria (TOM-complex) (Kunkele et al. 1998). Physical treatments in the isolation steps have been shown to inactivate one of the pores in the TOM complex and similar findings have been made with the isolated Pxmp2 channel.

100 We can assume that in the purification process, the Pxmp2 channel is subjected to conditions, such as high salt and detergent concentrations, change in pH, and others, which may lead to partial deformation of the structure of the protein and to closure of one or two pores in the cluster. Our experiments with protein stabilisers (Fig. 20) confirm this hypothesis since the treatment of Pxmp2 protein with 1.0 M Tris-Cl buffer, pH 8.0 led to a significant increase in the number of insertion events with a conductance of 1.3–1.5 nS in 1.0 M KCl at the expense of ‘small’ channels with a conductance of 0.3–0.4 nS in 1.0 M KCl. Based on the amino acid sequence, the secondary structure of Pxmp2 consists of α-helices, but not β-sheets which are apparent structural elements for porins of the outer mitochondrial membrane and the plastid envelope (Saier 2000). Also, porins from the outer membrane of Gram-negative bacteria are β-barrel proteins. Instead, Pxmp2 might be related to α-type channels, which are described as channel-forming proteins in which the pore is surrounded by α-helices (Saier 2000). However, it seems that Pxmp2 forms a general diffusion pore, which makes it clearly different from members of the α-type channel superfamily. The α-type channels form a narrow pore for inorganic ions (H+, K+, Ca2+ or Cl−) although some channels such as aquaporins allow transmembrane diffusion of water and small neutral molecules like urea or glycerol (King et al. 2004). The lipid reconstitution experiments with purified Pxmp2 using cations and non-electrolytes of different size (see section 5.2.5) suggested that the diameter of the channel pore is 1.4 nm. These dimensions of the pore may allow free passage of metabolites with sizes up to 200 Da across the membrane. Diffusion of solutes with sizes of 200–400 Da would be restricted and the access of metabolites over 400 Da such as cofactors and ATP would be prevented. Our experiments with organic anions confirm this supposition since several small peroxisomal metabolites such as glycolate, pyruvate and 2-oxoglutarate were shown to pass through the Pxmp2 channel (see section 5.2.6). These results are in accordance with the suggestion that the peroxisomal membrane contains channels accessible for small metabolites side by side with transporters for bulky solutes. Given that the Pxmp2 channel does not restrict permeation of small solutes, it is possible that peroxisomes share a common pool of small solutes with the surrounding cytoplasm. In contrast, the bulky solutes are unable penetrate the membrane through channels and require specific transporters, which may generate gradients of these solutes between the peroxisomal lumen and the cytoplasm.

101 How the peroxisomal metabolic machinery may exploit the Pxmp2 channel to carry out some specific functions is easy to predict. It is generally accepted that the metabolic conversion of peroxisomal cofactors proceeds via shuttle mechanisms resembling such systems in the inner mitochondrial membrane (van Roermund et al. 1995). However, in contrast to those in mitochondria, the shuttle molecules in peroxisomes do not need specific transmembrane transporters owing to the presence of pore-forming proteins in the membrane. Several members of the family of Nudix hydrolases are located in peroxisomes and active towards cofactors (CoA, NAD/P), cleaving them into two parts of near equal size (Gasmi & McLennan 2001, Abdelraheim et al. 2003). The reaction leads to the formation of molecules that are able to cross the membrane using peroxisomal channels and provides a route for the removal of cofactors from peroxisomes. Thus, cleavage of NAD+ (663 Da) by the corresponding peroxisomal Nudix hydrolase NUDT12 produces NMNH (334 Da) and AMP (347 Da)(Abdelraheim et al. 2003). Permeation of NAD+ through the Pxmp2 channel is negligible (see Fig. 23). However, the channel is permeable to AMP and apparently also to NMNH based on the similar size of these molecules (NMNH has no net charge, preventing use of this compound for multiple-channel recording).

6.3 The phenotype of Pxmp2-deficient mice

Pxmp2-deficient mice have been generated in this study and the disruption of Pxmp2 has been shown at the DNA, mRNA and protein level. Due to the close proximity of PoleI, the catalytic subunit of DNA polymerase ε, to the translation initiation codon of Pxmp2 (Otte et al. 2003), the expression level for PoleI was investigated and confirmed to be unaltered in Pxmp2−/− mice. In addition, the protein level of PoleI was identical to the wild-type in the mammary gland tissue of Pxmp2-deficient mice. Based on these results we can assume that the disruption of Pxmp2 by a lacZneo insertion has not interfered with the expression of PoleI. The absence of the Pxmp2-channel from the peroxisomal membrane leads to near symptomless mice since Pxmp2−/− animals are healthy and live a normal life span. The histological examination revealed no abnormalities in the tissues in which Pxmp2 is mainly expressed such as heart, liver and kidney, although some differences were observed in the appearance of kidney podocytes in electron microscopic analyses (flattening of podocyte foot processes). Pxmp2-deficient

102 mice did not show any phenotypic differences from the wild-type animals under altered dietary conditions, such as the addition of phytol and clofibrate in the diet (see section 5.3). The expression levels of selected peroxisomal genes were unaffected in the liver and kidney of Pxmp2−/− mice. Because of the flattening of podocyte foot processes, kidney functions were analysed in more details (see section 5.3), but no major differences were found between the wild-type and mutant mouse strains. Among the biochemical parameters analysed, a clear consequence of the absence of the Pxmp2-channel may be the increased levels of uric acid in serum and urine of Pxmp2−/− mice relative to wild-type animals. Although the increase is significant, (in the serum: 103 ± 30 µmol/l in Pxmp2-deficient mice compared to 68 ± 33 µmol/l in the wild-type animals and in the urine the ratio of uric acid concentration to creatinine was 0.85 ± 0.4 in Pxmp2-deficient mice compared to 0.18 ± 0.04 in the wild-type controls) the content of uric acid in Pxmp2−/− mice did not seem to reach a level that would precipitate and cause damage to tissues such as described for urate oxidase-deficient mice (Wu et al. 1994) or patients with an acute gout attack. The observed differences in uric acid serum concentrations in urate oxidase and Pxmp2-deficient mouse strains indicate that inspite of the absence of Pxmp2, a significant amount of uric acid is still degraded in peroxisomes (as also indicated by excretion of allantoin into urine) and suggest the existence of a second peroxisomal transmembrane route for this metabolite. This suggestion is also in line with the notion that the mammalian peroxisomal membrane contains at least two channels showing different properties (Antonenkov et al. 2004b). The mild phenotype of Pxmp2−/− mice could be explained by the overlapping functions of peroxisomal channels. A candidate for another channel could be M-LP, which is also a member of the same protein family. Although Pxmp2-deficient mice are fertile, Pxmp2−/− females are not able to nurse their pups; the pups die during the first day after birth. By studying the mammary glands of Pxmp2−/− mice, the surprising discovery was made, that the deficiency of Pxmp2 leads to the impairment of ductal outgrowth of mammary glands at puberty. Similar pubertal mammary phenotypes that display an absence or impairment of ductal outgrowth have been described for several knock-out mouse model (Howlin et al. 2006). The targeted genes include the estrogen receptor α (Couse & Korach 1999), epidermal growth factor receptor (EGFR) (Wiesen et al. 1999), amphiregulin – a ligand for EGFR (Ciarloni et al. 2007), actin cytoskeleton regulating protein gelsolin (Crowley et al. 2000) and the

103 transmembrane metalloproteinase ADAM17 (Sternlicht et al. 2005). Most of the deleted genes mentioned above are involved in the endocrine and paracrine pathways of steroid hormone action and epithelial-stromal crosstalk regulating mammary branching morphogenesis (Sternlicht et al. 2006). Our preliminary data on tissue transplantation studies showed that the primary defect in the mammary dysfunction of the Pxmp2-deficent mice might be located in the stromal compartment since the Pxmp2−/− epithelia grows in the wild-type fat pad in both a wild-type and also in a Pxmp2−/− host (Vapola, Rokka, Antonenkov, Soininen and Hiltunen unpublished results). This suggests that the systemic environment in the Pxmp2-deficient mice is normal. In addition, the concentrations of estradiol and progesterone in the blood of Pxmp2-deficient and wild-type mice were the same except for a slight decrease in the level of blood estradiol (Table 1), suggesting that the observed phenotype of Pxmp2-deficient mice is not a consequence of an inadequate amount of reproductive hormones. The mammary gland epithelium has an inherent requirement for adipose tissue in order to form ductal branches. In addition, besides providing a matrix for the growth and development of epithelia the mammary fat pad functions as an active store for lipids in the development of the mammary gland (Hovey et al. 1999). Several studies have indicated that the mammary fat pad is a significant site for systemic hormone action and growth factor synthesis and thus provides signals for ductal morphogenesis and alveolar differentiation (Neville et al. 1998, Hovey et al. 1999, Sternlicht et al. 2006). We have shown by western blot that Pxmp2 is present in both the intact mammary gland and the cleared fat pad devoid of epithelial cells. These results suggest that Pxmp2 is expressed in the mammary adipocytes. Further studies are required to provide more insight into the possible role of Pxmp2 in the mammary fat pad and to reveal the link between the peroxisomal channel protein and the observed mammary gland phenotype in Pxmp2−/− mice.

6.4 Future prospects

The work in this thesis describes the characterisation of channel-forming properties of mammalian peroxisomal membrane proteins. In addition, this work describes the function for Pxmp2, a 22 kDa abundant membrane protein in mammalian peroxisomes. Our data suggests that Pxmp2 is a channel-forming protein that is partially responsible for the peroxisomal membrane permeability to solutes. The concept of peroxisomal membrane as a permeability barrier has been

104 a matter of debate for several decades and our results provide new insight to this topic. Our data led to suggestion that peroxisomal membrane in mammals contains other channels besides Pxmp2 which would explain the poor phenotype of Pxmp2-deficient mice in this study. The aim in the future research is to identify what is the molecular nature of these peroxisomal membrane channels which are responsible for the rest of the membrane permeability to small solutes. Also the possible role of the quaternary structure of Pxmp2 in maintaining and regulation of its activity remains to be established by crystallisation and structural three- dimensional analysis at atomic resolution. The ultimate goal is to uncover mechanisms responsible for the cytosolic- peroxisomal communication and to reveal the role of these mechanisms in the pathogenesis of disorders associated with the mild malfunctioning of peroxisomal metabolic machinery.

105

106 7 Conclusions

These data show that the mammalian peroxisomal membrane contains large non- selective channels in addition to specific transporters. By using the Pxmp2- deficient mouse model along with recombinantly produced Pxmp2 and native Pxmp2, several lines of evidence indicate that this protein can act as a membrane channel. Pxmp2 was shown to function as a slightly cation selective channel that permits the free passage of solutes with a size up to 200 Da. The transfer of metabolites with the size of 200–400 Da is partially restricted and solutes with a molecular mass over 400 Da (such as NAD+ and ATP) are prevented from passing through Pxmp2. Channel proteins for small metabolites in the mammalian peroxisomal membrane have not been identified before. Identification of channel proteins in the peroxisomal membrane will open new insights into the peroxisome-cytosol communication and into the metabolism related to this organelle. In addition, the shuttling of reducing equivalents across the peroxisomal membrane can be easily explained in terms of the channel-forming activity of Pxmp2. The mouse model for Pxmp2 revealed that the channel function of Pxmp2 can be partially compensated by other channels that remain to be identified. Possible candidates for these channels include other members of the Pxmp2 protein family. An unexpected linkage to mammalian reproductive biology was revealed from the phenotypic characterisation of Pxmp2-deficient mice. Namely, disruption of Pxmp2 gene leads to impairment in the ductal outgrowth of mammary epithelia. The mammary gland phenotype will lead to new areas in peroxisomal research and follow-up research is being currently undertaken in our laboratory.

107 108 References

Abdelraheim SR, Spiller DG & McLennan AG (2003) Mammalian NADH diphosphatases of the Nudix family: cloning and characterization of the human peroxisomal NUDT12 protein. Biochem J 374: 329–335. Alexson SE, Fujiki Y, Shio H & Lazarow PB (1985) Partial disassembly of peroxisomes. J Cell Biol 101: 294–304. Angermuller S, Bruder G, Volkl A, Wesch H & Fahimi HD (1987) Localization of xanthine oxidase in crystalline cores of peroxisomes. A cytochemical and biochemical study. Eur J Cell Biol 45: 137–144. Angermuller S & Fahimi HD (1986) Ultrastructural cytochemical localization of uricase in peroxisomes of rat liver. J Histochem Cytochem 34: 159–165. Antonenkov VD (1989) Dehydrogenases of the pentose phosphate pathway in rat liver peroxisomes. Eur J Biochem 183: 75–82. Antonenkov VD & Hiltunen JK (2006) Peroxisomal membrane permeability and solute transfer. Biochim Biophys Acta 1763: 1697–1706. Antonenkov VD, Pirozhkov SV & Panchenko LF (1985) Intraparticulate localization and some properties of a clofibrate-induced peroxisomal aldehyde dehydrogenase from rat liver. Eur J Biochem 149: 159–167. Antonenkov VD, Rokka A, Sormunen RT, Benz R & Hiltunen JK (2005) Solute traffic across mammalian peroxisomal membrane–single channel conductance monitoring reveals pore-forming activities in peroxisomes. Cell Mol Life Sci 62: 2886–2895. Antonenkov VD, Sormunen RT & Hiltunen JK (2004a) The behavior of peroxisomes in vitro: mammalian peroxisomes are osmotically sensitive particles. Am J Physiol Cell Physiol 287: C1623–1635. Antonenkov VD, Sormunen RT & Hiltunen JK (2004b) The rat liver peroxisomal membrane forms a permeability barrier for cofactors but not for small metabolites in vitro. J Cell Sci 117: 5633–5642. Antonenkov VD, Van Veldhoven PP, Waelkens E & Mannaerts GP (1997) Substrate specificities of 3-oxoacyl-CoA thiolase A and sterol carrier protein 2/3-oxoacyl-CoA thiolase purified from normal rat liver peroxisomes. Sterol carrier protein 2/3-oxoacyl- CoA thiolase is involved in the metabolism of 2-methyl-branched fatty acids and bile acid intermediates. J Biol Chem 272: 26023–26031. Arnold W (1984) Inner ear and renal diseases. Ann Otol Rhinol Laryngol Suppl 112: 119– 124. Baerends RJ, Faber KN, Kram AM, Kiel JA, van der Klei IJ & Veenhuis M (2000) A stretch of positively charged amino acids at the N terminus of Hansenula polymorpha Pex3p is involved in incorporation of the protein into the peroxisomal membrane. J Biol Chem 275: 9986–9995. Baker AL & Tolbert NE (1966) Glycolate oxidase (ferredoxin-containing form). Methods Enzymol 9: 338–342. Baker PR, Cramer SD, Kennedy M, Assimos DG & Holmes RP (2004) Glycolate and glyoxylate metabolism in HepG2 cells. Am J Physiol Cell Physiol 287: C1359–1365.

109 Baudhuin P (1966) [Lysosomes and cellular autophagy]. Brux Med 46: 1059–1070. Baudhuin P (1969) Liver peroxisomes, cytology and function. Ann N Y Acad Sci 168: 214–228. Baudhuin P, Beaufay H, Rahman-Li Y, Sellinger OZ, Wattiaux R, Jacques P & De Duve C (1964) Tissue fractionation studies. 17. Intracellular distribution of monoamine oxidase, aspartate aminotransferase, alanine aminotransferase, D-amino acid oxidase and catalase in rat-liver tissue. Biochem J 92: 179–184. Baumgart E, Fahimi HD, Stich A & Volkl A (1996) L-lactate dehydrogenase A4- and A3B isoforms are bona fide peroxisomal enzymes in rat liver. Evidence for involvement in intraperoxisomal NADH reoxidation. J Biol Chem 271: 3846–3855. Beaufay H, Amar-Costesec A, Feytmans E, Thines-Sempoux D, Wibo M, Robbi M & Berthet J (1974) Analytical study of microsomes and isolated subcellular membranes from rat liver. I. Biochemical methods. J Cell Biol 61: 188–200. Beevers H (1979) Microbodies in higher plants. Annu Rev Plant Physiol 30: 159–193. Benz R (1994) Permeation of hydrophilic solutes through mitochondrial outer membranes: review on mitochondrial porins. Biochim Biophys Acta 1197: 167–196. Benz R & Bauer K (1988) Permeation of hydrophilic molecules through the outer membrane of gram-negative bacteria. Review on bacterial porins. Eur J Biochem 176: 1–19. Benz R, Janko K, Boos W & Lauger P (1978) Formation of large, ion-permeable membrane channels by the matrix protein (porin) of Escherichia coli. Biochim Biophys Acta 511: 305–319. Berry CE & Hare JM (2004) Xanthine oxidoreductase and cardiovascular disease: molecular mechanisms and pathophysiological implications. J Physiol 555: 589–606. Berthemy A, Newton J, Wu D & Buhrman D (1999) Quantitative determination of an extremely polar compound allantoin in human urine by LC-MS/MS based on the separation on a polymeric amino column. J Pharm Biomed Anal 19: 429–434. Biermann J, Just WW, Wanders RJ & Van Den Bosch H (1999) Alkyl-dihydroxyacetone phosphate synthase and dihydroxyacetone phosphate acyltransferase form a protein complex in peroxisomes. Eur J Biochem 261: 492–499. Bodnar AG & Rachubinski RA (1991) Characterization of the integral membrane polypeptides of rat liver peroxisomes isolated from untreated and clofibrate-treated rats. Biochem Cell Biol 69: 499–508. Braiterman LT, Zheng S, Watkins PA, Geraghty MT, Johnson G, McGuinness MC, Moser AB & Smith KD (1998) Suppression of peroxisomal membrane protein defects by peroxisomal ATP binding cassette (ABC) proteins. Hum Mol Genet 7: 239–247. Brisken C (2002) Hormonal control of alveolar development and its implications for breast carcinogenesis. J Mammary Gland Biol Neoplasia 7: 39–48. Brites P, Waterham HR & Wanders RJ (2004) Functions and biosynthesis of plasmalogens in health and disease. Biochim Biophys Acta 1636: 219–231. Brosius U, Dehmel T & Gartner J (2002) Two different targeting signals direct human peroxisomal membrane protein 22 to peroxisomes. J Biol Chem 277: 774–784.

110 Bryant DD & Wilson GN (1995) Differential evolution and expression of murine peroxisomal membrane protein genes. Biochem Mol Med 55: 22–30. Burdett K, Larkins LK, Das AK & Hajra AK (1991) Peroxisomal localization of acyl- coenzyme A reductase (long chain alcohol forming) in guinea pig intestine mucosal cells. J Biol Chem 266: 12201–12206. Ciarloni L, Mallepell S & Brisken C (2007) Amphiregulin is an essential mediator of estrogen receptor alpha function in mammary gland development. Proc Natl Acad Sci U S A 104: 5455–5460. Clozel M, Hess P, Fischli W, Loffler BM, Zwacka RM, Reuter A & Weiher H (1999) Age- dependent hypertension in Mpv17-deficient mice, a transgenic model of glomerulosclerosis and inner ear disease. Exp Gerontol 34: 1007–1015. Coleman RA, Lewin TM, Van Horn CG & Gonzalez-Baro MR (2002) Do long-chain acyl- CoA synthetases regulate fatty acid entry into synthetic versus degradative pathways? J Nutr 132: 2123–2126. Corpas FJ, Sandalio LM, Brown MJ, del Rio LA & Trelease RN (2000) Identification of porin-like polypeptide(s) in the boundary membrane of oilseed glyoxysomes. Plant Cell Physiol 41: 1218–1228. Couse JF & Korach KS (1999) Estrogen receptor null mice: what have we learned and where will they lead us? Endocr Rev 20: 358–417. Croes K, Van Veldhoven PP, Mannaerts GP & Casteels M (1997) Production of formyl- CoA during peroxisomal alpha-oxidation of 3-methyl-branched fatty acids. FEBS Lett 407: 197–200. Crowley MR, Head KL, Kwiatkowski DJ, Asch HL & Asch BB (2000) The mouse mammary gland requires the actin-binding protein gelsolin for proper ductal morphogenesis. Dev Biol 225: 407–423. Dansen TB, Wirtz KW, Wanders RJ & Pap EH (2000) Peroxisomes in human fibroblasts have a basic pH. Nat Cell Biol 2: 51–53. De Duve C & Baudhuin P (1966) Peroxisomes (microbodies and related particles). Physiol Rev 46: 323–357. De Pinto V, Ludwig O, Krause J, Benz R & Palmieri F (1987) Porin pores of mitochondrial outer membranes from high and low eukaryotic cells: biochemical and biophysical characterization. Biochim Biophys Acta 894: 109–119. Dean M, Rzhetsky A & Allikmets R (2001) The human ATP-binding cassette (ABC) transporter superfamily. Genome Res 11: 1156–1166. Diestelkotter P & Just WW (1993) In vitro insertion of the 22-kD peroxisomal membrane protein into isolated rat liver peroxisomes. J Cell Biol 123: 1717–1725. Drago I, Giacomello M, Pizzo P & Pozzan T (2008) Calcium dynamics in the peroxisomal lumen of living cells. J Biol Chem 283: 14384–14390. Dyer JM, McNew JA & Goodman JM (1996) The sorting sequence of the peroxisomal integral membrane protein PMP47 is contained within a short hydrophilic loop. J Cell Biol 133: 269–280. Erdmann R & Schliebs W (2005) Peroxisomal matrix protein import: the transient pore model. Nat Rev Mol Cell Biol 6: 738–742.

111 Falany CN, Fortinberry H, Leiter EH & Barnes S (1997) Cloning, expression, and chromosomal localization of mouse liver bile acid CoA:amino acid N-acyltransferase. J Lipid Res 38: 1139–1148. Falany CN, Johnson MR, Barnes S & Diasio RB (1994) Glycine and taurine conjugation of bile acids by a single enzyme. Molecular cloning and expression of human liver bile acid CoA:amino acid N-acyltransferase. J Biol Chem 269: 19375–19379. Ferdinandusse S, Denis S, L IJ, Dacremont G, Waterham HR & Wanders RJ (2000) Subcellular localization and physiological role of alpha-methylacyl-CoA racemase. J Lipid Res 41: 1890–1896. Ferdinandusse S, Denis S, Van Roermund CW, Wanders RJ & Dacremont G (2004) Identification of the peroxisomal beta-oxidation enzymes involved in the degradation of long-chain dicarboxylic acids. J Lipid Res 45: 1104–1111. Ferdinandusse S & Houten SM (2006) Peroxisomes and bile acid biosynthesis. Biochim Biophys Acta 1763: 1427–1440. Foulon V, Antonenkov VD, Croes K, Waelkens E, Mannaerts GP, Van Veldhoven PP & Casteels M (1999) Purification, molecular cloning, and expression of 2-hydroxyphytanoyl-CoA lyase, a peroxisomal thiamine pyrophosphate-dependent enzyme that catalyzes the carbon-carbon bond cleavage during alpha-oxidation of 3-methyl-branched fatty acids. Proc Natl Acad Sci U S A 96: 10039–10044. Fransen M, Vastiau I, Brees C, Brys V, Mannaerts GP & Van Veldhoven PP (2004) Potential role for Pex19p in assembly of PTS-receptor docking complexes. J Biol Chem 279: 12615–12624. Fransen M, Wylin T, Brees C, Mannaerts GP & Van Veldhoven PP (2001) Human pex19p binds peroxisomal integral membrane proteins at regions distinct from their sorting sequences. Mol Cell Biol 21: 4413–4424. Fraser F, Corstorphine CG, Price NT & Zammit VA (1999) Evidence that carnitine palmitoyltransferase I (CPT I) is expressed in microsomes and peroxisomes of rat liver. Distinct immunoreactivity of the N-terminal domain of the microsomal protein. FEBS Lett 446: 69–74. Frederiks WM & Vreeling-Sindelarova H (2002) Ultrastructural localization of xanthine oxidoreductase activity in isolated rat liver cells. Acta Histochem 104: 29–37. Fujiki Y, Matsuzono Y, Matsuzaki T & Fransen M (2006) Import of peroxisomal membrane proteins: the interplay of Pex3p- and Pex19p-mediated interactions. Biochim Biophys Acta 1763: 1639–1646. Gallegos AM, Atshaves BP, Storey SM, Starodub O, Petrescu AD, Huang H, McIntosh AL, Martin GG, Chao H, Kier AB & Schroeder F (2001) Gene structure, intracellular localization, and functional roles of sterol carrier protein-2. Prog Lipid Res 40: 498– 563. Gasmi L & McLennan AG (2001) The mouse Nudt7 gene encodes a peroxisomal nudix hydrolase specific for coenzyme A and its derivatives. Biochem J 357: 33–38. Gee R, McGroarty E, Hsieh B, Wied DM & Tolbert NE (1974) Glycerol phosphate dehydrogenase in mammalian peroxisomes. Arch Biochem Biophys 161: 187–193.

112 Geisbrecht BV & Gould SJ (1999) The human PICD gene encodes a cytoplasmic and peroxisomal NADP(+)-dependent isocitrate dehydrogenase. J Biol Chem 274: 30527– 30533. Geisbrecht BV, Zhang D, Schulz H & Gould SJ (1999) Characterization of PECI, a novel monofunctional Delta(3), Delta(2)-enoyl-CoA isomerase of mammalian peroxisomes. J Biol Chem 274: 21797–21803. Geuze HJ, Murk JL, Stroobants AK, Griffith JM, Kleijmeer MJ, Koster AJ, Verkleij AJ, Distel B & Tabak HF (2003) Involvement of the endoplasmic reticulum in peroxisome formation. Mol Biol Cell 14: 2900–2907. Gorgas K, Teigler A, Komljenovic D & Just WW (2006) The ether lipid-deficient mouse: Tracking down plasmalogen functions. Biochim Biophys Acta 1763: 1511–1526. Gould SJ, Keller GA, Hosken N, Wilkinson J & Subramani S (1989) A conserved tripeptide sorts proteins to peroxisomes. J Cell Biol 108: 1657–1664. Gould SJ & Valle D (2000) Peroxisome biogenesis disorders: genetics and cell biology. Trends Genet 16: 340–345. Gouveia AM, Reguenga C, Oliveira ME, Eckerskorn C, Sa-Miranda C & Azevedo JE (1999) Alkaline density gradient floatation of membranes: polypeptide composition of the mammalian peroxisomal membrane. Anal Biochem 274: 270–277. Graham JM (1993) The identification of subcellular fractions from mammalian cells. Methods Mol Biol 19: 1–18. Hajra AK (1995) Glycerolipid biosynthesis in peroxisomes (microbodies). Prog Lipid Res 34: 343–364. Hartl FU & Just WW (1987) Integral membrane polypeptides of rat liver peroxisomes: topology and response to different metabolic states. Arch Biochem Biophys 255: 109– 119. Hartl FU, Just WW, Koster A & Schimassek H (1985) Improved isolation and purification of rat liver peroxisomes by combined rate zonal and equilibrium density centrifugation. Arch Biochem Biophys 237: 124–134. He XY, Shoukry K, Chu C, Yang J, Sprecher H & Schulz H (1995) Peroxisomes contain delta 3,5,delta 2,4-dienoyl-CoA isomerase and thus possess all enzymes required for the beta-oxidation of unsaturated fatty acids by a novel reductase-dependent pathway. Biochem Biophys Res Commun 215: 15–22. Heby O & Persson L (1990) Molecular genetics of polyamine synthesis in eukaryotic cells. Trends Biochem Sci 15: 153–158. Heiland I & Erdmann R (2005) Biogenesis of peroxisomes. Topogenesis of the peroxisomal membrane and matrix proteins. Febs J 272: 2362–2372. Hettema EH, Distel B & Tabak HF (1999) Import of proteins into peroxisomes. Biochim Biophys Acta 1451: 17–34. Hettema EH, Girzalsky W, van Den Berg M, Erdmann R & Distel B (2000) Saccharomyces cerevisiae pex3p and pex19p are required for proper localization and stability of peroxisomal membrane proteins. Embo J 19: 223–233. Hettema EH & Tabak HF (2000) Transport of fatty acids and metabolites across the peroxisomal membrane. Biochim Biophys Acta 1486: 18–27.

113 Hettema EH, van Roermund CW, Distel B, van den Berg M, Vilela C, Rodrigues-Pousada C, Wanders RJ & Tabak HF (1996) The ABC transporter proteins Pat1 and Pat2 are required for import of long-chain fatty acids into peroxisomes of Saccharomyces cerevisiae. Embo J 15: 3813–3822. Higgins CF (1992) ABC transporters: from microorganisms to man. Annu Rev Cell Biol 8: 67–113. Hille R & Nishino T (1995) Flavoprotein structure and mechanism. 4. Xanthine oxidase and xanthine dehydrogenase. Faseb J 9: 995–1003. Hiltunen JK, Mursula AM, Rottensteiner H, Wierenga RK, Kastaniotis AJ & Gurvitz A (2003) The biochemistry of peroxisomal beta-oxidation in the yeast Saccharomyces cerevisiae. FEMS Microbiol Rev 27: 35–64. Hinck L & Silberstein GB (2005) Key stages in mammary gland development: the mammary end bud as a motile organ. Breast Cancer Res 7: 245–251. Holmes RP & Assimos DG (1998) Glyoxylate synthesis, and its modulation and influence on oxalate synthesis. J Urol 160: 1617–1624. Holroyd C & Erdmann R (2001) Protein translocation machineries of peroxisomes. FEBS Lett 501: 6–10. Holzinger A, Kammerer S, Berger J & Roscher AA (1997a) cDNA cloning and mRNA expression of the human adrenoleukodystrophy related protein (ALDRP), a peroxisomal ABC transporter. Biochem Biophys Res Commun 239: 261–264. Holzinger A, Kammerer S & Roscher AA (1997b) Primary structure of human PMP69, a putative peroxisomal ABC-transporter. Biochem Biophys Res Commun 237: 152–157. Honsho M & Fujiki Y (2001) Topogenesis of peroxisomal membrane protein requires a short, positively charged intervening-loop sequence and flanking hydrophobic segments. study using human membrane protein PMP34. J Biol Chem 276: 9375– 9382. Hovey RC, McFadden TB & Akers RM (1999) Regulation of mammary gland growth and morphogenesis by the mammary fat pad: a species comparison. J Mammary Gland Biol Neoplasia 4: 53–68. Howlin J, McBryan J & Martin F (2006) Pubertal mammary gland development: insights from mouse models. J Mammary Gland Biol Neoplasia 11: 283–297. Hu RH & Pegg AE (1997) Rapid induction of apoptosis by deregulated uptake of polyamine analogues. Biochem J 328 ( Pt 1): 307–316. Hunt MC, Solaas K, Kase BF & Alexson SE (2002) Characterization of an acyl-coA thioesterase that functions as a major regulator of peroxisomal lipid metabolism. J Biol Chem 277: 1128–1138. Huyghe S, Mannaerts GP, Baes M & Van Veldhoven PP (2006) Peroxisomal multifunctional protein-2: the enzyme, the patients and the knockout mouse model. Biochim Biophys Acta 1761: 973–994. Ichikawa M, Nishino T & Ichikawa A (1992) Subcellular localization of xanthine oxidase in rat hepatocytes: high-resolution immunoelectron microscopic study combined with biochemical analysis. J Histochem Cytochem 40: 1097–1103.

114 Iida R, Yasuda T, Tsubota E, Matsuki T & Kishi K (2001) Cloning, mapping, genomic organization, and expression of mouse M-LP, a new member of the peroxisomal membrane protein Mpv17 domain family. Biochem Biophys Res Commun 283: 292– 296. Iida R, Yasuda T, Tsubota E, Takatsuka H, Masuyama M, Matsuki T & Kishi K (2003) M-LP, Mpv17-like protein, has a peroxisomal membrane targeting signal comprising a transmembrane domain and a positively charged loop and up-regulates expression of the manganese superoxide dismutase gene. J Biol Chem 278: 6301–6306. Iida R, Yasuda T, Tsubota E, Takatsuka H, Masuyama M, Matsuki T & Kishi K (2005) A novel alternative spliced Mpv17-like protein isoform localizes in cytosol and is expressed in a kidney- and adult-specific manner. Exp Cell Res 302: 22–30. Iida R, Yasuda T, Tsubota E, Takatsuka H, Matsuki T & Kishi K (2006) Human Mpv17- like protein is localized in peroxisomes and regulates expression of antioxidant enzymes. Biochem Biophys Res Commun 344: 948–954. Imanaka T, Aihara K, Takano T, Yamashita A, Sato R, Suzuki Y, Yokota S & Osumi T (1999) Characterization of the 70-kDa peroxisomal membrane protein, an ATP binding cassette transporter. J Biol Chem 274: 11968–11976. Jakobs BS & Wanders RJ (1995) Fatty acid beta-oxidation in peroxisomes and mitochondria: the first, unequivocal evidence for the involvement of carnitine in shuttling propionyl-CoA from peroxisomes to mitochondria. Biochem Biophys Res Commun 213: 1035–1041. Jankowski A, Kim JH, Collins RF, Daneman R, Walton P & Grinstein S (2001) In situ measurements of the pH of mammalian peroxisomes using the fluorescent protein pHluorin. J Biol Chem 276: 48748–48753. Jansen GA, van den Brink DM, Ofman R, Draghici O, Dacremont G & Wanders RJ (2001) Identification of pristanal dehydrogenase activity in peroxisomes: conclusive evidence that the complete phytanic acid alpha-oxidation pathway is localized in peroxisomes. Biochem Biophys Res Commun 283: 674–679. Jansen GA & Wanders RJ (2006) Alpha-oxidation. Biochim Biophys Acta 1763: 1403– 1412. Jones JM, Morrell JC & Gould SJ (2001) Multiple distinct targeting signals in integral peroxisomal membrane proteins. J Cell Biol 153: 1141–1150. Jones JM, Morrell JC & Gould SJ (2004) PEX19 is a predominantly cytosolic chaperone and import receptor for class 1 peroxisomal membrane proteins. J Cell Biol 164: 57– 67. Just WW & Diestelkotter P (1996) Protein insertion into the peroxisomal membrane. Ann N Y Acad Sci 804: 60–75. Kaldi K, Diestelkotter P, Stenbeck G, Auerbach S, Jakle U, Magert HJ, Wieland FT & Just WW (1993) Membrane topology of the 22 kDa integral peroxisomal membrane protein. FEBS Lett 315: 217–222. Kamijo K, Taketani S, Yokota S, Osumi T & Hashimoto T (1990) The 70-kDa peroxisomal membrane protein is a member of the Mdr (P-glycoprotein)-related ATP- binding protein superfamily. J Biol Chem 265: 4534–4540.

115 Karadimas CL, Vu TH, Holve SA, Chronopoulou P, Quinzii C, Johnsen SD, Kurth J, Eggers E, Palenzuela L, Tanji K, Bonilla E, De Vivo DC, DiMauro S & Hirano M (2006) Navajo neurohepatopathy is caused by a mutation in the MPV17 gene. Am J Hum Genet 79: 544–548. Kaulin YA, Schagina LV, Bezrukov SM, Malev VV, Feigin AM, Takemoto JY, Teeter JH & Brand JG (1998) Cluster organization of ion channels formed by the antibiotic syringomycin E in bilayer lipid membranes. Biophys J 74: 2918–2925. Kemp S & Wanders RJ (2007) X-linked adrenoleukodystrophy: Very long-chain fatty acid metabolism, ABC half-transporters and the complicated route to treatment. Mol Genet Metab 90: 268–276. Kemp S, Wei HM, Lu JF, Braiterman LT, McGuinness MC, Moser AB, Watkins PA & Smith KD (1998) Gene redundancy and pharmacological gene therapy: implications for X-linked adrenoleukodystrophy. Nat Med 4: 1261–1268. Kim PK, Mullen RT, Schumann U & Lippincott-Schwartz J (2006) The origin and maintenance of mammalian peroxisomes involves a de novo PEX16-dependent pathway from the ER. J Cell Biol 173: 521–532. King LS, Kozono D & Agre P (2004) From structure to disease: the evolving tale of aquaporin biology. Nat Rev Mol Cell Biol 5: 687–698. Kramer DL, Vujcic S, Diegelman P, Alderfer J, Miller JT, Black JD, Bergeron RJ & Porter CW (1999) Polyamine analogue induction of the p53-p21WAF1/CIP1-Rb pathway and G1 arrest in human melanoma cells. Cancer Res 59: 1278–1286. Krasilnikov OV, Da Cruz JB, Yuldasheva LN, Varanda WA & Nogueira RA (1998) A novel approach to study the geometry of the water lumen of ion channels: colicin Ia channels in planar lipid bilayers. J Membr Biol 161: 83–92. Kunkele KP, Heins S, Dembowski M, Nargang FE, Benz R, Thieffry M, Walz J, Lill R, Nussberger S & Neupert W (1998) The preprotein translocation channel of the outer membrane of mitochondria. Cell 93: 1009–1019. Labarca P, Wolff D, Soto U, Necochea C & Leighton F (1986) Large cation-selective pores from rat liver peroxisomal membranes incorporated to planar lipid bilayers. J Membr Biol 94: 285–291. Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227: 680–685. Lametschwandtner G, Brocard C, Fransen M, Van Veldhoven P, Berger J & Hartig A (1998) The difference in recognition of terminal tripeptides as peroxisomal targeting signal 1 between yeast and human is due to different affinities of their receptor Pex5p to the cognate signal and to residues adjacent to it. J Biol Chem 273: 33635–33643. Lamhonwah AM, Ackerley CA, Tilups A, Edwards VD, Wanders RJ & Tein I (2005) OCTN3 is a mammalian peroxisomal membrane carnitine transporter. Biochem Biophys Res Commun 338: 1966–1972. Lamhonwah AM, Skaug J, Scherer SW & Tein I (2003) A third human carnitine/organic cation transporter (OCTN3) as a candidate for the 5q31 Crohn's disease locus (IBD5). Biochem Biophys Res Commun 301: 98–101.

116 Landes N, Birringer M & Brigelius-Flohe R (2003) Homologous metabolic and gene activating routes for vitamins E and K. Mol Aspects Med 24: 337–344. Lasorsa FM, Pinton P, Palmieri L, Scarcia P, Rottensteiner H, Rizzuto R & Palmieri F (2008) Peroxisomes as novel players in cell calcium homeostasis. J Biol Chem 283: 15300–15308. Lasorsa FM, Scarcia P, Erdmann R, Palmieri F, Rottensteiner H & Palmieri L (2004) The yeast peroxisomal adenine nucleotide transporter: characterization of two transport modes and involvement in DeltapH formation across peroxisomal membranes. Biochem J 381: 581–585. Lazarow PB (2003) Peroxisome biogenesis: advances and conundrums. Curr Opin Cell Biol 15: 489–497. Lazarow PB & Fujiki Y (1985) Biogenesis of peroxisomes. Annu Rev Cell Biol 1: 489– 530. Lebovitz RM, Zhang H, Vogel H, Cartwright J, Jr., Dionne L, Lu N, Huang S & Matzuk MM (1996) Neurodegeneration, myocardial injury, and perinatal death in mitochondrial superoxide dismutase-deficient mice. PNAS 93: 9782–9787. Legakis JE & Terlecky SR (2001) PTS2 protein import into mammalian peroxisomes. Traffic 2: 252–260. Leighton F, Poole B, Beaufay H, Baudhuin P, Coffey JW, Fowler S & De Duve C (1968) The large-scale separation of peroxisomes, mitochondria, and lysosomes from the livers of rats injected with triton WR-1339. Improved isolation procedures, automated analysis, biochemical and morphological properties of fractions. J Cell Biol 37: 482– 513. Lemmens M, Verheyden K, Van Veldhoven P, Vereecke J, Mannaerts GP & Carmeliet E (1989) Single-channel analysis of a large conductance channel in peroxisomes from rat liver. Biochim Biophys Acta 984: 351–359. Li Y, Huang TT, Carlson EJ, Melov S, Ursell PC, Olson JL, Noble LJ, Yoshimura MP, Berger C, Chan PH, Wallace DC & Epstein CJ (1995) Dilated cardiomyopathy and neonatal lethality in mutant mice lacking manganese superoxide dismutase. Nat Genet 11: 376–381. Lombard-Platet G, Savary S, Sarde CO, Mandel JL & Chimini G (1996) A close relative of the adrenoleukodystrophy (ALD) gene codes for a peroxisomal protein with a specific expression pattern. PNAS 93: 1265–1269. Luers GH, Otte DM, Subramani S & Franz T (2001) Genomic organization, chromosomal localization and tissue specific expression of the murine Pxmp2 gene encoding the 22 kDa peroxisomal membrane protein (Pmp22). Gene 272: 45–50. MacKenzie JA & Payne RM (2007) Mitochondrial protein import and human health and disease. Biochim Biophys Acta 1772: 509–523. Majalahti-Palviainen T, Hirvinen M, Tervonen V, Ilves M, Ruskoaho H & Vuolteenaho O (2000) Gene structure of a new cardiac peptide hormone: a model for heart-specific gene expression. Endocrinology 141: 731–740.

117 Marzioch M, Erdmann R, Veenhuis M & Kunau WH (1994) PAS7 encodes a novel yeast member of the WD-40 protein family essential for import of 3-oxoacyl-CoA thiolase, a PTS2-containing protein, into peroxisomes. Embo J 13: 4908–4918. Masters C & Crane DI (1995) The peroxisome: a vital organelle. Cambridge University Press, Cambridge. McClelland GB, Khanna S, Gonzalez GF, Butz CE & Brooks GA (2003) Peroxisomal membrane monocarboxylate transporters: evidence for a redox shuttle system? Biochem Biophys Res Commun 304: 130–135. Meyer zum Gottesberge AM & Felix H (2005) Abnormal basement membrane in the inner ear and the kidney of the Mpv17−/− mouse strain: ultrastructural and immunohistochemical investigations. Histochem Cell Biol 124: 507–516. Meyer zum Gottesberge AM, Felix H, Reuter A & Weiher H (2001) Ultrastructural and physiological defects in the cochlea of the Mpv17 mouse strain. A comparison between young and old adult animals. Hear Res 156: 69–80. Meyer zum Gottesberge AM, Reuter A & Weiher H (1996) Inner ear defect similar to Alport's syndrome in the glomerulosclerosis mouse model Mpv17. Eur Arch Otorhinolaryngol 253: 470–474. Mihalik SJ, Rainville AM & Watkins PA (1995) Phytanic acid alpha-oxidation in rat liver peroxisomes. Production of alpha-hydroxyphytanoyl-CoA and formate is enhanced by dioxygenase cofactors. Eur J Biochem 232: 545–551. Mihalik SJ, Steinberg SJ, Pei Z, Park J, Kim DG, Heinzer AK, Dacremont G, Wanders RJ, Cuebas DA, Smith KD & Watkins PA (2002) Participation of two members of the very long-chain acyl-CoA synthetase family in bile acid synthesis and recycling. J Biol Chem 277: 24771–24779. Morel F, Rauch C, Petit E, Piton A, Theret N, Coles B & Guillouzo A (2004) Gene and protein characterization of the human glutathione S-transferase kappa and evidence for a peroxisomal localization. J Biol Chem 279: 16246–16253. Morgan DM (1998) Polyamine oxidases–enzymes of unknown function? Biochem Soc Trans 26: 586–591. Mosser J, Douar AM, Sarde CO, Kioschis P, Feil R, Moser H, Poustka AM, Mandel JL & Aubourg P (1993) Putative X-linked adrenoleukodystrophy gene shares unexpected homology with ABC transporters. Nature 361: 726–730. Mullen RT, Lisenbee CS, Miernyk JA & Trelease RN (1999) Peroxisomal membrane ascorbate peroxidase is sorted to a membranous network that resembles a subdomain of the endoplasmic reticulum. Plant Cell 11: 2167–2185. Mullen RT & Trelease RN (2000) The sorting signals for peroxisomal membrane-bound ascorbate peroxidase are within its C-terminal tail. J Biol Chem 275: 16337–16344. Muller M, Smolders JW, Meyer zum Gottesberge AM, Reuter A, Zwacka RM, Weiher H & Klinke R (1997) Loss of auditory function in transgenic Mpv17-deficient mice. Hear Res 114: 259–263. Murphy MA, Phillipson BA, Baker A & Mullen RT (2003) Characterization of the targeting signal of the Arabidopsis 22-kD integral peroxisomal membrane protein. Plant Physiol 133: 813–828.

118 Nakagawa T, Imanaka T, Morita M, Ishiguro K, Yurimoto H, Yamashita A, Kato N & Sakai Y (2000) Peroxisomal membrane protein Pmp47 is essential in the metabolism of middle-chain fatty acid in yeast peroxisomes and Is associated with peroxisome proliferation. J Biol Chem 275: 3455–3461. Neville MC, McFadden TB & Forsyth I (2002) Hormonal regulation of mammary differentiation and milk secretion. J Mammary Gland Biol Neoplasia 7: 49–66. Neville MC, Medina D, Monks J & Hovey RC (1998) The mammary fat pad. J Mammary Gland Biol Neoplasia 3: 109–116. Nicolay K, Veenhuis M, Douma AC & Harder W (1987) A 31P NMR study of the internal pH of yeast peroxisomes. Arch Microbiol 147: 37–41. Nikaido H (2003) Molecular basis of bacterial outer membrane permeability revisited. Microbiol Mol Biol Rev 67: 593–656. Nishioka K (1993) International Symposium on Polyamines in Cancer. Critical role of polyamines in cancer: basic mechanisms and clinical approaches. Cancer Res 53: 2689–2692. Noltmann EA (1964) Isolation of Crystalline Phosphoglucose Isomerase from Rabbit Muscle. J Biol Chem 239: 1545–1550. Novikov DK, Vanhove GF, Carchon H, Asselberghs S, Eyssen HJ, Van Veldhoven PP & Mannaerts GP (1994) Peroxisomal beta-oxidation. Purification of four novel 3-hydroxyacyl-CoA dehydrogenases from rat liver peroxisomes. J Biol Chem 269: 27125–27135. O'Bryan T, Weiher H, Rennke HG, Kren S & Hostetter TH (2000) Course of renal injury in the Mpv17-deficient transgenic mouse. J Am Soc Nephrol 11: 1067–1074. O'Byrne J, Hunt MC, Rai DK, Saeki M & Alexson SE (2003) The human bile acid- CoA:amino acid N-acyltransferase functions in the conjugation of fatty acids to glycine. J Biol Chem 278: 34237–34244. Oshino N, Chance B, Sies H & Bucher T (1973) The role of H 2 O 2 generation in perfused rat liver and the reaction of catalase compound I and hydrogen donors. Arch Biochem Biophys 154: 117–131. Otte DM, Schwaab U & Luers GH (2003) The Pxmp2 and PoleI genes are linked by a bidirectional promoter in an evolutionary conserved fashion. Gene 313: 119–126. Palmieri F (2004) The mitochondrial transporter family (SLC25): physiological and pathological implications. Pflugers Arch 447: 689–709. Palmieri L, Rottensteiner H, Girzalsky W, Scarcia P, Palmieri F & Erdmann R (2001) Identification and functional reconstitution of the yeast peroxisomal adenine nucleotide transporter. Embo J 20: 5049–5059. Pause B, Diestelkotter P, Heid H & Just WW (1997) Cytosolic factors mediate protein insertion into the peroxisomal membrane. FEBS Lett 414: 95–98. Pause B, Saffrich R, Hunziker A, Ansorge W & Just WW (2000) Targeting of the 22 kDa integral peroxisomal membrane protein. FEBS Lett 471: 23–28. Pinto MP, Grou CP, Alencastre IS, Oliveira ME, Sa-Miranda C, Fransen M & Azevedo JE (2006) The import competence of a peroxisomal membrane protein is determined by Pex19p before the docking step. J Biol Chem 281: 34492–34502.

119 Poirier Y, Antonenkov VD, Glumoff T & Hiltunen JK (2006) Peroxisomal beta-oxidation- A metabolic pathway with multiple functions. Biochim Biophys Acta 1763: 1413– 1426. Pujol A, Ferrer I, Camps C, Metzger E, Hindelang C, Callizot N, Ruiz M, Pampols T, Giros M & Mandel JL (2004) Functional overlap between ABCD1 (ALD) and ABCD2 (ALDR) transporters: a therapeutic target for X-adrenoleukodystrophy. Hum Mol Genet 13: 2997–3006. Purdue PE & Lazarow PB (2001) Peroxisome biogenesis. Annu Rev Cell Dev Biol 17: 701–752. Ramazzina I, Folli C, Secchi A, Berni R & Percudani R (2006) Completing the uric acid degradation pathway through phylogenetic comparison of whole genomes. Nat Chem Biol 2: 144–148. Ramsay RR (1999) The role of the carnitine system in peroxisomal fatty acid oxidation. Am J Med Sci 318: 28–35. Reddy JK & Hashimoto T (2001) Peroxisomal beta-oxidation and peroxisome proliferator- activated receptor alpha: an adaptive metabolic system. Annu Rev Nutr 21: 193–230. Reumann S, Bettermann M, Benz R & Heldt HW (1997) Evidence for the Presence of a Porin in the Membrane of Glyoxysomes of Castor Bean. Plant Physiol 115: 891–899. Reumann S, Maier E, Benz R & Heldt HW (1995) The membrane of leaf peroxisomes contains a porin-like channel. J Biol Chem 270: 17559–17565. Reumann S, Maier E, Heldt HW & Benz R (1998) Permeability properties of the porin of spinach leaf peroxisomes. Eur J Biochem 251: 359–366. Reuter A, Nestl A, Zwacka RM, Tuckermann J, Waldherr R, Wagner EM, Hoyhtya M, Meyer zum Gottesberge AM, Angel P & Weiher H (1998) Expression of the recessive glomerulosclerosis gene Mpv17 regulates MMP-2 expression in fibroblasts, the kidney, and the inner ear of mice. Mol Biol Cell 9: 1675–1682. Richardson KE & Thompson JS (1970) L-Alanine: Glyoxylate aminotransferase (human liver). Methods Enzymol 17: 163–166. Russell DW (2003) The enzymes, regulation, and genetics of bile acid synthesis. Annu Rev Biochem 72: 137–174. Sacksteder KA, Jones JM, South ST, Li X, Liu Y & Gould SJ (2000) PEX19 binds multiple peroxisomal membrane proteins, is predominantly cytoplasmic, and is required for peroxisome membrane synthesis. J Cell Biol 148: 931–944. Saier MH, Jr. (2000) Families of proteins forming transmembrane channels. J Membr Biol 175: 165–180. Sakai Y, Saiganji A, Yurimoto H, Takabe K, Saiki H & Kato N (1996) The absence of Pmp47, a putative yeast peroxisomal transporter, causes a defect in transport and folding of a specific matrix enzyme. J Cell Biol 134: 37–51. Sakuraba H & Noguchi T (1995) Alcohol:NAD+ oxidoreductase is present in rat liver peroxisomes. J Biol Chem 270: 37–40. Sambrook J, Fritsch EF and Maniatis T (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press. Cold Spring Harbor, NY.

120 Sanders RJ, Ofman R, Valianpour F, Kemp S & Wanders RJ (2005) Evidence for two enzymatic pathways for omega-oxidation of docosanoic acid in rat liver microsomes. J Lipid Res 46: 1001–1008. Savolainen K, Kotti TJ, Schmitz W, Savolainen TI, Sormunen RT, Ilves M, Vainio SJ, Conzelmann E & Hiltunen JK (2004) A mouse model for alpha-methylacyl-CoA racemase deficiency: adjustment of bile acid synthesis and intolerance to dietary methyl-branched lipids. Hum Mol Genet 13: 955–965. Schliebs W & Kunau WH (2006) PTS2 Co-receptors: Diverse proteins with common features. Biochim Biophys Acta 1763: 1605–1612. Schmitz W, Albers C, Fingerhut R & Conzelmann E (1995) Purification and characterization of an alpha-methylacyl-CoA racemase from human liver. Eur J Biochem 231: 815–822. Schrader M & Fahimi HD (2006) Peroxisomes and oxidative stress. Biochim Biophys Acta 1763: 1755–1766. Seiler N (1995) Polyamine oxidase, properties and functions. Prog Brain Res 106: 333– 344. Shani N, Jimenez-Sanchez G, Steel G, Dean M & Valle D (1997) Identification of a fourth half ABC transporter in the human peroxisomal membrane. Hum Mol Genet 6: 1925– 1931. Shani N & Valle D (1996) A Saccharomyces cerevisiae homolog of the human adrenoleukodystrophy transporter is a heterodimer of two half ATP-binding cassette transporters. PNAS 93: 11901–11906. Shani N, Watkins PA & Valle D (1995) PXA1, a possible Saccharomyces cerevisiae ortholog of the human adrenoleukodystrophy gene. Proc Natl Acad Sci U S A 92: 6012–6016. Singh AK, Dhaunsi GS, Gupta MP, Orak JK, Asayama K & Singh I (1994) Demonstration of glutathione peroxidase in rat liver peroxisomes and its intraorganellar distribution. Arch Biochem Biophys 315: 331–338. Singh AK, Dobashi K, Gupta MP, Asayama K, Singh I & Orak JK (1999) Manganese superoxide dismutase in rat liver peroxisomes: biochemical and immunochemical evidence. Mol Cell Biochem 197: 7–12. Snyder WB, Koller A, Choy AJ & Subramani S (2000) The peroxin Pex19p interacts with multiple, integral membrane proteins at the peroxisomal membrane. J Cell Biol 149: 1171–1178. South ST & Gould SJ (1999) Peroxisome synthesis in the absence of preexisting peroxisomes. J Cell Biol 144: 255–266. Spinazzola A, Viscomi C, Fernandez-Vizarra E, Carrara F, D'Adamo P, Calvo S, Marsano RM, Donnini C, Weiher H, Strisciuglio P, Parini R, Sarzi E, Chan A, DiMauro S, Rotig A, Gasparini P, Ferrero I, Mootha VK, Tiranti V & Zeviani M (2006) MPV17 encodes an inner mitochondrial membrane protein and is mutated in infantile hepatic mitochondrial DNA depletion. Nat Genet 38: 570–575. Steinberg SJ, Dodt G, Raymond GV, Braverman NE, Moser AB & Moser HW (2006) Peroxisome biogenesis disorders. Biochim Biophys Acta 1763: 1733–1748.

121 Steinberg SJ, Wang SJ, Kim DG, Mihalik SJ & Watkins PA (1999) Human very-long- chain acyl-CoA synthetase: cloning, topography, and relevance to branched-chain fatty acid metabolism. Biochem Biophys Res Commun 257: 615–621. Sternlicht MD, Kouros-Mehr H, Lu P & Werb Z (2006) Hormonal and local control of mammary branching morphogenesis. Differentiation 74: 365–381. Sternlicht MD, Sunnarborg SW, Kouros-Mehr H, Yu Y, Lee DC & Werb Z (2005) Mammary ductal morphogenesis requires paracrine activation of stromal EGFR via ADAM17-dependent shedding of epithelial amphiregulin. Development 132: 3923– 3933. Stoscheck CM (1990) Quantitation of protein. Methods Enzymol 182: 50–68. Suematsu N, Okamoto K & Isohashi F (2002) Mouse cytosolic acetyl-CoA hydrolase, a novel candidate for a key enzyme involved in fat metabolism: cDNA cloning, sequencing and functional expression. Acta Biochim Pol 49: 937–945. Sulter GJ, Verheyden K, Mannaerts G, Harder W & Veenhuis M (1993) The in vitro permeability of yeast peroxisomal membranes is caused by a 31 kDa integral membrane protein. Yeast 9: 733–742. Swartzman EE, Viswanathan MN & Thorner J (1996) The PAL1 gene product is a peroxisomal ATP-binding cassette transporter in the yeast Saccharomyces cerevisiae. J Cell Biol 132: 549–563. Thai TP, Heid H, Rackwitz HR, Hunziker A, Gorgas K & Just WW (1997) Ether lipid biosynthesis: isolation and molecular characterization of human dihydroxyacetonephosphate acyltransferase. FEBS Lett 420: 205–211. Theodoulou FL, Holdsworth M & Baker A (2006) Peroxisomal ABC transporters. FEBS Lett 580: 1139–1155. Thoms S & Erdmann R (2005) Dynamin-related proteins and Pex11 proteins in peroxisome division and proliferation. Febs J 272: 5169–5181. Titorenko VI, Chan H & Rachubinski RA (2000) Fusion of small peroxisomal vesicles in vitro reconstructs an early step in the in vivo multistep peroxisome assembly pathway of Yarrowia lipolytica. J Cell Biol 148: 29–44. Titorenko VI & Mullen RT (2006) Peroxisome biogenesis: the peroxisomal endomembrane system and the role of the ER. J Cell Biol 174: 11–17. Titorenko VI & Rachubinski RA (2001a) Dynamics of peroxisome assembly and function. Trends Cell Biol 11: 22–29. Titorenko VI & Rachubinski RA (2001b) The life cycle of the peroxisome. Nat Rev Mol Cell Biol 2: 357–368. Towbin H, Staehelin T & Gordon J (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Natl Acad Sci U S A 76: 4350–4354. Trias J & Benz R (1994) Permeability of the cell wall of Mycobacterium smegmatis. Mol Microbiol 14: 283–290. Trott A & Morano KA (2004) SYM1 is the stress-induced Saccharomyces cerevisiae ortholog of the mammalian kidney disease gene Mpv17 and is required for ethanol metabolism and tolerance during heat shock. Eukaryot Cell 3: 620–631.

122 Tugal HB, Pool M & Baker A (1999) Arabidopsis 22-kilodalton peroxisomal membrane protein. Nucleotide sequence analysis and biochemical characterization. Plant Physiol 120: 309–320. Usuda N, Hayashi S, Fujiwara S, Noguchi T, Nagata T, Rao MS, Alvares K, Reddy JK & Yeldandi AV (1994) Uric acid degrading enzymes, urate oxidase and allantoinase, are associated with different subcellular organelles in frog liver and kidney. J Cell Sci 107 ( Pt 4): 1073–1081. Walton PA, Hill PE & Subramani S (1995) Import of stably folded proteins into peroxisomes. Mol Biol Cell 6: 675–683. Van Ael E & Fransen M (2006) Targeting signals in peroxisomal membrane proteins. Biochim Biophys Acta 1763: 1629–1638. van Roermund CW, de Jong M, L IJ, van Marle J, Dansen TB, Wanders RJ & Waterham HR (2004) The peroxisomal lumen in Saccharomyces cerevisiae is alkaline. J Cell Sci 117: 4231–4237. van Roermund CW, Drissen R, van Den Berg M, Ijlst L, Hettema EH, Tabak HF, Waterham HR & Wanders RJ (2001) Identification of a peroxisomal ATP carrier required for medium-chain fatty acid beta-oxidation and normal peroxisome proliferation in Saccharomyces cerevisiae. Mol Cell Biol 21: 4321–4329. van Roermund CW, Elgersma Y, Singh N, Wanders RJ & Tabak HF (1995) The membrane of peroxisomes in Saccharomyces cerevisiae is impermeable to NAD(H) and acetyl-CoA under in vivo conditions. Embo J 14: 3480–3486. van Roermund CW, Waterham HR, Ijlst L & Wanders RJ (2003) Fatty acid metabolism in Saccharomyces cerevisiae. Cell Mol Life Sci 60: 1838–1851. Van Veldhoven P, Debeer LJ & Mannaerts GP (1983) Water- and solute-accessible spaces of purified peroxisomes. Evidence that peroxisomes are permeable to NAD+. Biochem J 210: 685–693. Van Veldhoven PP, Baumgart E & Mannaerts GP (1996) Iodixanol (Optiprep), an improved density gradient medium for the iso-osmotic isolation of rat liver peroxisomes. Anal Biochem 237: 17–23. Van Veldhoven PP, Casteels M, Mannaerts GP & Baes M (2001) Further insights into peroxisomal lipid breakdown via alpha- and beta-oxidation. Biochem Soc Trans 29: 292–298. Van Veldhoven PP, Just WW & Mannaerts GP (1987) Permeability of the peroxisomal membrane to cofactors of beta-oxidation. Evidence for the presence of a pore-forming protein. J Biol Chem 262: 4310–4318. Van Veldhoven PP & Mannaerts GP (1986) Coenzyme A in purified peroxisomes is not freely soluble in the matrix but firmly bound to a matrix protein. Biochem Biophys Res Commun 139: 1195–1201. Van Veldhoven PP, Vanhove G, Vanhoutte F, Dacremont G, Parmentier G, Eyssen HJ & Mannaerts GP (1991) Identification and purification of a peroxisomal branched chain fatty acyl-CoA oxidase. J Biol Chem 266: 24676–24683. Wanders RJ (2004) Peroxisomes, lipid metabolism, and peroxisomal disorders. Mol Genet Metab 83: 16–27.

123 Wanders RJ, Romeyn GJ, Schutgens RB & Tager JM (1989) L-pipecolate oxidase: a distinct peroxisomal enzyme in man. Biochem Biophys Res Commun 164: 550–555. Wanders RJ & Waterham HR (2006a) Biochemistry of mammalian peroxisomes revisited. Annu Rev Biochem 75: 295–332. Wanders RJ & Waterham HR (2006b) Peroxisomal disorders: The single peroxisomal enzyme deficiencies. Biochim Biophys Acta 1763: 1707–1720. Wanders RJ, Visser WF, van Roermund CW, Kemp S & Waterham HR (2007) The peroxisomal ABC transporter family. Pflugers Arch 453: 719–734. Wanders RJ, Vreken P, Ferdinandusse S, Jansen GA, Waterham HR, van Roermund CW & Van Grunsven EG (2001) Peroxisomal fatty acid alpha- and beta-oxidation in humans: enzymology, peroxisomal metabolite transporters and peroxisomal diseases. Biochem Soc Trans 29: 250–267. Wang X, McMahon MA, Shelton SN, Nampaisansuk M, Ballard JL & Goodman JM (2004) Multiple targeting modules on peroxisomal proteins are not redundant: discrete functions of targeting signals within Pmp47 and Pex8p. Mol Biol Cell 15: 1702–1710. Wang X, Unruh MJ & Goodman JM (2001) Discrete targeting signals direct Pmp47 to oleate-induced peroxisomes in Saccharomyces cerevisiae. J Biol Chem 276: 10897– 10905. Waterham HR, Keizer-Gunnink I, Goodman JM, Harder W & Veenhuis M (1990) Immunocytochemical evidence for the acidic nature of peroxisomes in methylotrophic yeasts. FEBS Lett 262: 17–19. Watkins PA (1997) Fatty acid activation. Prog Lipid Res 36: 55–83. Watkins PA, Howard AE, Gould SJ, Avigan J & Mihalik SJ (1996) Phytanic acid activation in rat liver peroxisomes is catalyzed by long-chain acyl-CoA synthetase. J Lipid Res 37: 2288–2295. Weiher H, Noda T, Gray DA, Sharpe AH & Jaenisch R (1990) Transgenic mouse model of kidney disease: insertional inactivation of ubiquitously expressed gene leads to nephrotic syndrome. Cell 62: 425–434. Verheyden K, Fransen M, Van Veldhoven PP & Mannaerts GP (1992) Presence of small GTP-binding proteins in the peroxisomal membrane. Biochim Biophys Acta 1109: 48–54. Verhoeven NM, Roe DS, Kok RM, Wanders RJ, Jakobs C & Roe CR (1998) Phytanic acid and pristanic acid are oxidized by sequential peroxisomal and mitochondrial reactions in cultured fibroblasts. J Lipid Res 39: 66–74. Verleur N & Wanders RJ (1993) Permeability properties of peroxisomes in digitonin- permeabilized rat hepatocytes. Evidence for free permeability towards a variety of substrates. Eur J Biochem 218: 75–82. Westin MA, Alexson SE & Hunt MC (2004) Molecular cloning and characterization of two mouse peroxisome proliferator-activated receptor alpha (PPARalpha)-regulated peroxisomal acyl-CoA thioesterases. J Biol Chem 279: 21841–21848. Westin MA, Hunt MC & Alexson SE (2005) The identification of a succinyl-CoA thioesterase suggests a novel pathway for succinate production in peroxisomes. J Biol Chem 280: 38125–38132.

124 Westin MA, Hunt MC & Alexson SE (2007) Peroxisomes contain a specific phytanoyl- CoA/pristanoyl-CoA thioesterase acting as a novel auxiliary enzyme in alpha- and beta-oxidation of methyl-branched fatty acids in mouse. J Biol Chem 282: 26707– 26716. Westin MA, Hunt MC & Alexson SE (2008) Short- and medium-chain carnitine acyltransferases and acyl-CoA thioesterases in mouse provide complementary systems for transport of beta-oxidation products out of peroxisomes. Cell Mol Life Sci 65: 982–990. Wickner W & Schekman R (2005) Protein translocation across biological membranes. Science 310: 1452–1456. Wiese S, Gronemeyer T, Ofman R, Kunze M, Grou CP, Almeida JA, Eisenacher M, Stephan C, Hayen H, Schollenberger L, Korosec T, Waterham HR, Schliebs W, Erdmann R, Berger J, Meyer HE, Just W, Azevedo JE, Wanders RJ & Warscheid B (2007) Proteomics characterization of mouse kidney peroxisomes by tandem mass spectrometry and protein correlation profiling. Mol Cell Proteomics 6: 2045–2057. Wiesen JF, Young P, Werb Z & Cunha GR (1999) Signaling through the stromal epidermal growth factor receptor is necessary for mammary ductal development. Development 126: 335–344. Wilcke M & Alexson SE (1994) Characterization of acyl-CoA thioesterase activity in isolated rat liver peroxisomes. Partial purification and characterization of a long-chain acyl-CoA thioesterase. Eur J Biochem 222: 803–811. Visser WF, Van Roermund CW, Ijlst L, Hellingwerf KJ, Wanders RJ & Waterham HR (2005) Demonstration and characterization of phosphate transport in mammalian peroxisomes. Biochem J 389: 717–722. Visser WF, van Roermund CW, Ijlst L, Hellingwerf KJ, Waterham HR & Wanders RJ (2006) First identification of a 2-ketoglutarate/isocitrate transport system in mammalian peroxisomes and its characterization. Biochem Biophys Res Commun 348: 1224–1231. Visser WF, van Roermund CW, Ijlst L, Waterham HR & Wanders RJ (2007a) Demonstration of bile acid transport across the mammalian peroxisomal membrane. Biochem Biophys Res Commun 357: 335-340. Visser WF, van Roermund CW, Ijlst L, Waterham HR & Wanders RJ (2007b) Metabolite transport across the peroxisomal membrane. Biochem J 401: 365–375. Visser WF, van Roermund CW, Waterham HR & Wanders RJ (2002) Identification of human PMP34 as a peroxisomal ATP transporter. Biochem Biophys Res Commun 299: 494–497. Volkl A, Baumgart E & Fahimi HD (1988) Localization of urate oxidase in the crystalline cores of rat liver peroxisomes by immunocytochemistry and immunoblotting. J Histochem Cytochem 36: 329–336. Wolvetang EJ, Tager JM & Wanders RJ (1990) Latency of the peroxisomal enzyme acyl- CoA:dihydroxyacetonephosphate acyltransferase in digitonin-permeabilized fibroblasts: the effect of ATP and ATPase inhibitors. Biochem Biophys Res Commun 170: 1135–1143.

125 Wu X, Wakamiya M, Vaishnav S, Geske R, Montgomery C, Jr., Jones P, Bradley A & Caskey CT (1994) Hyperuricemia and urate nephropathy in urate oxidase-deficient mice. PNAS 91: 742–746. Wylin T, Baes M, Brees C, Mannaerts GP, Fransen M & Van Veldhoven PP (1998) Identification and characterization of human PMP34, a protein closely related to the peroxisomal integral membrane protein PMP47 of Candida boidinii. Eur J Biochem 258: 332–338. Yeldandi AV, Yeldandi V, Kumar S, Murthy CV, Wang XD, Alvares K, Rao MS & Reddy JK (1991) Molecular evolution of the urate oxidase-encoding gene in hominoid primates: nonsense mutations. Gene 109: 281–284. Young MJ, Bay DC, Hausner G & Court DA (2007) The evolutionary history of mitochondrial porins. BMC Evol Biol 7: 31. Zwacka RM, Reuter A, Pfaff E, Moll J, Gorgas K, Karasawa M & Weiher H (1994) The glomerulosclerosis gene Mpv17 encodes a peroxisomal protein producing reactive oxygen species. Embo J 13: 5129–5134.

126 ACTA UNIVERSITATIS OULUENSIS SERIES A SCIENTIAE RERUM NATURALIUM

500. Raunio, Janne (2008) The use of Chironomid Pupal Exuvial Technique (CPET) in freshwater biomonitoring: applications for boreal rivers and lakes 501. Paasivaara, Antti (2008) Space use, habitat selection and reproductive output of breeding common goldeneye (Bucephala clangula) 502. Asikkala, Janne (2008) Application of ionic liquids and microwave activation in selected organic reactions 503. Rinta-aho, Marko (2008) On monomial exponential sums in certain index 2 cases and their connections to coding theory 504. Sallinen, Pirkko (2008) Myocardial infarction. Aspects relating to endogenous and exogenous melatonin and cardiac contractility 505. Xin, Weidong (2008) Continuum electrostatics of biomolecular systems 506. Pyhäjärvi, Tanja (2008) Roles of demography and natural selection in molecular evolution of trees, focus on Pinus sylvestris 507. Kinnunen, Aulis (2008) A Palaeoproterozoic high-sulphidation epithermal gold deposit at Orivesi, southern Finland 508. Alahuhta, Markus (2008) Protein crystallography of triosephosphate isomerases: functional and protein engineering studies 509. Reif, Vitali (2008) Birds of prey and grouse in Finland. Do avian predators limit or regulate their prey numbers? 510. Saravesi, Karita (2008) Mycorrhizal responses to defoliation of woody hosts 511. Risteli, Maija (2008) Substrate specificity of lysyl hydroxylase isoforms and multifunctionality of lysyl hydroxylase 3 512. Korsu, Kai (2008) Ecology and impacts of nonnative salmonids with special reference to brook trout (Salvelinus fontinalis Mitchill) in North Europe 513. Laitinen, Jarmo (2008) Vegetational and landscape level responses to water level fluctuations in Finnish, mid-boreal aapa mire – aro wetland environments 514. Viljakainen, Lumi (2008) Evolutionary genetics of immunity and infection in social insects 515. Hurme, Eija (2008) Ecological knowledge towards sustainable forest management. Habitat requirements of the Siberian flying squirrel in Finland 516. Kaukonen, Risto (2008) Sulfide-poor platinum-group element deposits. A mineralogical approach with case studies and examples from the literature

Book orders: Distributed by OULU UNIVERSITY PRESS OULU UNIVERSITY LIBRARY P.O. Box 8200, FI-90014 P.O. Box 7500, FI-90014 University of Oulu, Finland University of Oulu, Finland A 517 OULU 2008 A 517

UNIVERSITY OF OULU P.O.B. 7500 FI-90014 UNIVERSITY OF OULU FINLAND ACTA UNIVERSITATISUNIVERSITATIS OULUENSISOULUENSIS ACTA UNIVERSITATIS OULUENSIS ACTAACTA

SERIES EDITORS SCIENTIAESCIENTIAEA A RERUMRERUM Aare Rokka NATURALIUMNATURALIUM

EDITOR IN CHIEF Professor Olli Vuolteenaho PUBLICATIONS EDITOR