OUP user menu

The biochemistry of peroxisomal β-oxidation in the yeast Saccharomyces cerevisiae

J. Kalervo Hiltunen, Anu M. Mursula, Hanspeter Rottensteiner, Rik K. Wierenga, Alexander J. Kastaniotis, Aner Gurvitz
DOI: http://dx.doi.org/10.1016/S0168-6445(03)00017-2 35-64 First published online: 1 April 2003


Peroxisomal fatty acid degradation in the yeast Saccharomyces cerevisiae requires an array of β-oxidation enzyme activities as well as a set of auxiliary activities to provide the β-oxidation machinery with the proper substrates. The corresponding classical and auxiliary enzymes of β-oxidation have been completely characterized, many at the structural level with the identification of catalytic residues. Import of fatty acids from the growth medium involves passive diffusion in combination with an active, protein-mediated component that includes acyl-CoA ligases, illustrating the intimate linkage between fatty acid import and activation. The main factors involved in protein import into peroxisomes are also known, but only one peroxisomal metabolite transporter has been characterized in detail, Ant1p, which exchanges intraperoxisomal AMP with cytosolic ATP. The other known transporter is Pxa1p–Pxa2p, which bears similarity to the human adrenoleukodystrophy protein ALDP. The major players in the regulation of fatty acid-induced gene expression are Pip2p and Oaf1p, which unite to form a transcription factor that binds to oleate response elements in the promoter regions of genes encoding peroxisomal proteins. Adr1p, a transcription factor, binding upstream activating sequence 1, also regulates key genes involved in β-oxidation. The development of new, postgenomic-era tools allows for the characterization of the entire transcriptome involved in β-oxidation and will facilitate the identification of novel proteins as well as the characterization of protein families involved in this process.

  • Metabolic compartmentalization
  • Unsaturated fatty acid
  • Oleic acid induction
  • Eci1p
  • Dci1p
  • Ehd3p/Ydr036cp
  • Cta1p

1 Introduction to fungal β-oxidation

Saccharomyces cerevisiae cells are able to grow on a variety of carbon sources, including fatty acids. Supplementation of yeast growth medium with fatty acids as a sole carbon and energy source elicits the transcriptional up-regulation of genes encoding enzymes of the β-oxidation process [1], which represents the main pathway for degrading fatty acids. This response is additionally accompanied by a remarkable proliferation of the peroxisomal compartment in which β-oxidation is exclusively housed [2]. By now, most of the genes involved in fatty acid degradation have been identified (Table 1), and the major breakdown routes characterized (Fig. 1). The genetic mechanism associated with the metabolic switch permitting growth on fatty acids has also been elucidated to a large extent. Although several important questions relating to fatty acid degradation remain open, most of these actually refer to ancillary processes, including transport of fatty acids and their metabolites across cell membranes, cross-talk between peroxisomes and other subcellular compartments, and regulation of cellular functions by acyl-CoA esters. In contrast to the situation in yeast, peroxisomal β-oxidation in mammalian cells is augmented by an additional mitochondrial pathway. It is precisely the relative simplicity of a β-oxidation process confined to a single organelle that has helped turn S. cerevisiae into an attractive model organism for studying the degradation of fatty acids.

View this table:
Table 1

Genes and gene products involved in the breakdown of fatty acids

Gene (standard name/alias)ORFFunctionRegulationLocalization
I. Transport factors
ANT1YPR128cATP carrierPip2p–Oaf1pperoxisomal membrane
PXA1 (SSH2, PAL1, PAT2, LPI1)YPL147wABC transporterPip2p–Oaf1pperoxisomal membrane
PXA2 (PAT1)YKL188cABC transporterPip2p–Oaf1pperoxisomal membrane
CRC1YOR100ccarnitine/acylcarnitine carrierPip2p–Oaf1pmitochondrial inner membrane
SFC1 (ACR1)YJR095wsuccinate/fumarate antiporterPip2p–Oaf1pmitochondrial inner membrane
AGP2YBR132ccarnitine carrierPip2p–Oaf1pplasma membrane
II. Peroxisomal protein import factors
PEX3 (PAS3)YDR329cperoxisomal protein targetingN.D.peroxisomal membrane
PEX5 (PAS10)YDR244wPTS1 peroxisomal targeting signal receptorN.D.cytosol, peroxisomal membrane/matrix
PEX7 (PAS7, PEB1)YDR142cPTS2 peroxisomal targeting signal receptorN.D.cytosol, peroxisomal membrane/matrix
PEX11 (PMP24, PMP27)YOL147cperoxisome proliferation and free fatty acid transport pathwayPip2p–Oaf1pperoxisomal membrane
PEX14YGL153wperoxisomal targetingN.D.peroxisomal membrane
PEX17 (PAS9)YNL214wperoxisomal targetingN.D.peroxisomal membrane
PEX19 (PAS12)YDL065cperoxisomal targetingN.D.cytosol, peroxisomal membrane
III. Enzymes involved in fatty acid activation and import
ACB1YGR037cacyl-CoA binding proteinN.D.cytosol? ER?
FAA1YOR317wlong-chain fatty acyl-CoA ligaseN.D.internal cellular locations
FAA2 (FAM1)YER015wmedium-chain fatty acyl-CoA ligasePip2p–Oaf1pperoxisomal matrix
FAA3YIL009wlong-chain fatty acyl-CoA ligaseN.D.?
FAA4YMR246wlong-chain fatty acyl-CoA ligaseN.D.cytosol
FAT1YBR041wvery long-chain fatty acyl-CoA ligaseN.D.plasma/peroxisomal/microsomal membrane
IV. Enzymes involved in β-oxidation
DCI1 (EHD2, ECI2)YOR180cΔ3,52,4-dienoyl CoA isomerasePip2p–Oaf1pperoxisomal matrix
ECI1 (EHD1)YLR284cΔ32-enoyl-CoA isomerasePip2p–Oaf1pperoxisomal matrix
FOX2 (POX2), MFE2YKR009c(3R)-hydroxyacyl-CoA dehydrogenase, 2-enoyl-CoA hydratase 2Pip2p–Oaf1pperoxisomal matrix
POT1 (FOX3, POX3)TIL160cβ-ketoacyl-CoA thiolasePip2p–Oaf1pperoxisomal matrix
POX1 (FOX1)YGL205wacyl-CoA oxidasePip2p–Oaf1pperoxisomal matrix
SPS19 (SPX19)YNL202w2,4-dienoyl-CoA reductasePip2p–Oaf1pperoxisomal matrix
V. Glyoxylate cycle and related enzymes
ACO1 (GLU1)YLR304caconitate hydrataseRtg2p, Hap2/3/4p?cytosol, mitochondrial matrix
CIT1 (CS1, LYS6)YNR001Cmitochondrial citrate synthasePip2p–Oaf1pmitochondrial matrix
CIT2YCR005cperoxisomal citrate synthaseRtg1p, Rtg3pperoxisomal matrix
CIT3YPR001wmitochondrial citrate synthaseRtg1p, Rtg2p, Rtg3pmitochondrial matrix
DAL7 (MLS2, MSL2)YIR031cmalate synthase involved in allantoin degradationGln3p, Gat1p, Dal82pperoxisomal matrix
ICL1YER065cisocitrate lyaseCat8p, Ino80pcytosol
MDH1YKL085wmitochondrial malate dehydrogenaseHap2/3/4pmitochondrial matrix
MDH2YOL126cNAD+ regenerationCat8p, Sir4pcytosol
MDH3YDL078cNAD+ regenerationPip2p–Oaf1pperoxisomal matrix
MLS1YNL117wcarbon-catabolite sensitive malate synthase involved in glyoxylate cycleCat8pcytosol and peroxisomal matrix
VI. Transcription factors
ADR1YDR216wpositive transcriptional regulator of ADH2 and peroxisomal protein genesReg1pnucleus
OAF1 (YAF1)YAL051wtranscription factor/activatorN.D.nucleus
PIP2 (OAF2)YOR363ctranscription factor/activatorPip2p–Oaf1pnucleus
VII. Other enzymes involved in fatty acid breakdown
AAT2 (ASP5)YLR027cNAD+ regeneration? (aspartate aminotransferase)N.D.cytosol, peroxisome
CAT2 (YCAT)YML042wcarnitine O-acetyltransferasePip2p–Oaf1pperoxisomal matrix
CTA1YDR256ccatalase APip2p–Oaflpperoxisomal matrix
IDP2YLR174wNADPH regeneration (cytosolic isocitrate dehydrogenase)Cat8pcytosol
IDP3YNL009cNADPH regeneration (peroxisomal isocitrate dehydrogenase)Pip2p–Oaf1pperoxisomal matrix
TES1 (PTE1)YJR019Cacyl-CoA thioesterasePip2p–Oaf1pperoxisomal matrix
  • Not all of the listed yeast peroxins are mentioned in the text. N.D.: not determined.

Figure 1

Schematic representation of β-oxidation in yeast peroxisomes. The metabolic pathways occurring in yeast peroxisomes during fatty acid β-oxidation are indicated by solid arrows. The broken arrows show the (3S)-hydroxyacyl-CoA-dependent route, which occurs in many other β-oxidation systems but not in yeast (see text).

The methodological progress currently made in investigating biological molecules as well as the advent of genome-wide analyses of cell function have allowed dissection and scrutiny of metabolic processes in ever finer detail, right down to the molecular and atomic levels. These new developments further contribute to the use of yeast cells not only as a favorite model organism in biological research, but also as a target for metabolic engineering in preparation for novel biotechnological applications. We will review here the progress that has been made in understanding fatty acid metabolism in the yeast S. cerevisiae, with emphasis on oxidative degradation pathways and the role of peroxisomes in accommodating them. Additional emphasis will be placed on lessons learnt from the crystal structure of several enzymes involved in β-oxidation, especially those belonging to the hydratase/isomerase family of proteins. Finally, short experimental sections aimed at addressing some open questions have also been added, including a brief characterization of the remaining S. cerevisiae enoyl hydratase/isomerase protein Ehd3p/Ydr036cp, peroxisomal targeting of Eci1p and Dci1p, and the role of catalases in fatty acid degradation.

2 Uptake and activation of fatty acid

Although S. cerevisiae can synthesize de novo of all the fatty acids that it requires [3], the ability to take up fatty acids or derivatives thereof from the environment is vital when alternative nutrients are not available. Even in the presence of more favorable carbon sources than fatty acids, the ability of cells to import fatty acids may be advantageous, as they can use existing molecules instead of needlessly expending the energy required for biosynthesis. The mechanisms involved in the uptake of fatty acids in eukaryotes are still not well understood. One model has been put forth proposing that fatty acid import is solely due to diffusion across the cell membrane [47]. However, there is physiological evidence that fatty acid transport is a saturable, protein-dependent process in higher eukaryotic cells [810], at least for long-chain fatty acids (LCFAs) [11]. This has evoked a second model based on the situation in prokaryotes, which proposes that a membrane-bound transporter protein is responsible for the uptake of exogenous fatty acids. For example, in Escherichia coli, fatty acid import depends on two proteins, the outer membrane-associated transporter FadL, and the cytosolic fatty acyl-CoA synthetase FadD. In this second model, import and activation are thought to be intimately linked to each other, with the process being described as ‘vectorial acylation’[12, 13] whereby activation of the imported fatty acid ensures its retention. An additional model for fatty acid uptake has been put forth that combines the passive diffusion model with protein-dependent import [13, 14]. This third model states that while passive diffusion via flip-flop of the uncharged fatty acid from the outer to the inner leaflet of the cell membrane is likely, protein-mediated import plays an equally important role. In S. cerevisiae, Fat1p as well as the Faa1p and Faa4p fatty acyl-CoA synthetases appear to be the principal players in protein-mediated fatty acid import and activation [13].

Fat1p was identified as a homologue (54% similarity) of the mammalian adipocyte fatty acid transport protein FATP [15, 16]. The FATP family of proteins is highly conserved in eukaryotes and appears to have two functions. The first is associated with fatty acid import, while the second is associated with very long-chain acyl-CoA synthetase activity. Murine FATP can complement the fatty acid uptake deficiency phenotype of a yeast strain carrying a disruption in FAT1. When yeast cells are grown on medium containing the fatty acid synthetase inhibitor cerulenin, growth can be restored to wild-type levels by adding exogenous oleic acid. This form of phenotype rescue is dependent on the presence of Fat1p [17], presumably due to its requirement for fatty acid import.

There is an ongoing debate regarding the primary function of Fat1p in fatty acid uptake, as it appears to be localized to both the endoplasmic reticulum (ER) and the peroxisomal membrane rather than the plasma membrane [18]. Moreover, Fat1p is mainly involved in very long-chain fatty acid (VLCFA) homeostasis, and the effect of deleting FAT1 on fatty acid import is attributed only to indirect consequences of the gene's absence [19, 20]. A subsequent study based on site-directed mutagenesis demonstrated that the VLCFA synthetase activity of Fat1p could be separated from its function in fatty acid import [14]. While this result appears to confirm the conclusion that Fat1p plays a direct role in fatty acid import, the discrepancy between the observed localization (ER and peroxisomes) and the localization required for fatty acid import (the cell membrane) needs to be resolved. A yeast protein with high similarity to Fat1p, Fat2p/Pcs60p, is exclusively localized to the luminal side of the peroxisomal membrane [21], and might also be involved in the uptake of fatty acids into the peroxisomes rather than into the cytosol. The function of this protein remains elusive, since a strain with a disruption of FAT2 displayed no growth phenotype on oleic acid, and introduction of fat2 deletion into a fat1Δ background does not cause additional loss of acyl-CoA synthetase activity if measured with oleic acid [19, 21].

In addition to VLCFA synthetase activity, fatty acid import in yeast is also dependent on two LCFA synthetases encoded by FAA1 and FAA4 [13]. The corresponding protein products of these latter two genes together account for 99% of the total cellular myristoyl-CoA and palmitoyl-CoA synthetase activities [22, 23]. A double deletion of these genes results in impaired fatty acid uptake [13]. Ultimately, it is the conversion of the fatty acids into acyl-CoA esters by Faa1p and Faa4p that establishes a concentration gradient of fatty acids across the membrane. This process of activation helps to extract fatty acids from the membrane and creates the thermodynamic sink that governs the process of their transport [13, 19]. Contradictory to this model, there has been a report claiming that no differences could be determined between wild-type and faa1Δfaa4Δ strains with respect to uptake of myristic and palmitic acids, arguing against coupling of activation to import [23]. However, the validity of this result has since been disputed on grounds of faulty methods used in the experiments [13]. Therefore, more work is needed to clarify the issue of uptake of exogenous fatty acids.

In mammalian cells, a family of soluble, low molecular mass (14–15 kDa) fatty acid-binding proteins (FABPs) has been implicated in the metabolism of free fatty acids (reviewed in [24]). Although the exact role of FABPs remains to be established, they are thought to divert free fatty acids into a readily available pool, thereby preventing the deleterious effects of their binding non-specifically to proteins and/or improperly inserting into membranes [25]. Despite some efforts to identify FABP homologues in yeast, these have not been found [26], and no such activity has been detected in S. cerevisiae [27].

Cytosolic transport of activated fatty acids and regulation of the intracellular acyl-CoA pool is thought to be mediated, at least in part, by acyl-CoA binding proteins (ACBPs). ACBPs typically have a mass of about 10 kDa, and comprise a highly conserved protein family in eukaryotes that bind activated fatty acids of a length of 14–22 carbons and play an important role in the mediation of intermembrane transport of fatty acyl-CoAs [28]. In yeast, ACBP is encoded by the ACB1 gene [29, 30]. Although the consequences of Acb1p depletion for yeast β-oxidation have not been studied directly, it is assumed that such depletion has only an indirect effect due to the ability of Acb1p to regulate intracellular fatty acyl-CoA concentrations [28, 31, 32]. Other data, however, suggest a more specific role for Acb1p in the transport of acyl-CoA to the fatty acid elongation system. Acb1p depletion experiments in yeast show severe negative effects on cell growth and membrane organization, probably caused by impaired fatty acid elongation and sphingolipid synthesis [33]. The import of fatty acids into peroxisomes and activation of medium-chain fatty acids (MCFAs) by Faa2p in combination with Ant1p in the peroxisomal compartment is discussed below.

3 Peroxisomes and their impermeable boundary

3.1 Translocation of peroxisomal matrix proteins

As mentioned in Section 1, fatty acid β-oxidation in fungi is confined to the peroxisomes [25], which are organelles bounded by a single membrane [2] that is impermeable to soluble factors, including proteins. Matrix proteins are synthesized on free ribosomes in the cytosol and are then folded and usually even oligomerized prior to their import into peroxisomes [34]. Peroxisomal import is mediated by a specific protein translocation machinery that is composed of peroxins. This class of proteins is involved in the biogenesis of peroxisomes, acting not only in the import of matrix proteins or insertion of membrane proteins at the organellar membrane, but also in the maintenance and expansion of the peroxisomal compartment.

Almost all of the enzymes involved in peroxisomal β-oxidation require Pex5p to reach their correct destination. Pex5p acts as receptor for a specific type of peroxisomal targeting signal (PTS) appearing at the extreme C-terminal end of matrix proteins, usually SKL or a variant thereof, termed PTS1 [3538]. Once a receptor–cargo complex is formed, it is delivered to the cytosolic face of the peroxisomal membrane, where it attaches to a docking site represented by Pex14p [39, 40], a peroxisomal membrane protein (PMP). Additional PMPs then act to translocate the cargo protein into the matrix and eject Pex5p back into the cytosolic pool of cycling receptors [41].

However, there are several exceptions to this form of protein import. Targeting of the S. cerevisiae acyl-CoA oxidase Pox1p/Fox1p [42] is also Pex5p-dependent, but in this case the receptor–cargo interaction is mediated by an internal region of the PTS1-less enzyme and the N-terminal half of Pex5p [43], which lacks the tetratricopeptide repeat domain involved in PTS1 binding [37]. Another exception is Pot1p/Fox3p [44, 45], which is preceded by an N-terminal signal PTS2 that is recognized by Pex7p. The resulting Pot1p–Pex7p complex is similarly docked at the peroxisomal boundary with the help of Pex14p to translocate its cargo [4648]. It should be stressed that although most of the principal players in protein import have now been identified, the exact mechanism of how matrix enzymes are translocated across the peroxisomal membrane is completely unknown. A novel perspective on what the events leading to import of peroxisomal proteins might look like is provided by Gould and Collins [49].

3.2 Insertion of peroxisomal membrane proteins

Targeting and insertion of PMPs into the peroxisomal membrane relies on a pathway that is independent of that of matrix proteins. PMPs are hypothesized to contain a unique type of targeting signal (mPTS) directing them to the peroxisomal membrane. Unlike the situation with matrix enzymes, PMPs are still targeted to peroxisomal remnants or ghosts in most of the pex mutants [50]. Only in cells lacking Pex3p or Pex19p (and in mammals also Pex16p, [51]) are the import routes of both membrane and matrix proteins impaired, indicating that these two peroxins are specifically involved in the targeting of membrane proteins in S. cerevisiae. Pex3p is an integral membrane protein [52], whereas Pex19p is localized to both the cytosol and the peroxisomal membrane [53]. Pex19p interacts with many PMPs, thereby making it a good candidate for an mPTS receptor [54]. However, since in a number of PMPs the mPTS was apparently dissectable from the Pex19p binding site [55, 56], further work will be required to elucidate the exact role of Pex19p in PMP insertion.

3.3 Transport of metabolites across the peroxisomal membrane

The peroxisomal membrane is also impermeable to small metabolites [57]. However, if the carbon flux through β-oxidation is to be maintained, metabolites must continually cross the peroxisomal membrane. This property of the peroxisomal membrane has triggered extensive studies into cross-membrane traffic of fatty acids, ATP, acetyl-CoA, NADH, NADPH, and other metabolites. To date, only a single peroxisomal solute carrier, Ant1p [58, 59], has been functionally reconstituted, whereas a transporter function has been revealed for just one additional protein complex, Pxa1p–Pxa2p (Pat2p–Pat1p) [6063]. Notably, both transporters are involved in providing activated fatty acids for peroxisomal β-oxidation.

LCFAs are predominantly transported in their activated form into the peroxisomal lumen by the Pxa1p–Pxa2p heterodimer (Fig. 1). Its constituents Pxa1p and Pxa2p belong to the family of ATP binding cassette (ABC) transporters that translocate a variety of molecules unidirectionally across biological membranes in an ATP-driven manner. ABC transporters can be formed by a single polypeptide chain (full-size ABC transporter), or can be composed of a dimeric or multimeric protein complex. Disruption of either of the genes encoding Pxa1p or Pxa2p affects growth on medium containing LCFAs as the sole carbon source, and mutant levels of β-oxidation are only 50% of those of wild-type cells, although peroxisomal biogenesis and function are not affected [6163]. Experimental evidence has led to the suggestion that LCFAs cannot enter peroxisomes in pxa1Δ or pxa2Δ mutants as β-oxidation is normal in mutant cell lysates. Since Pxa1p–Pxa2p is only required for the β-oxidation of LCFAs, which are activated in the cytosol, the candidate substrate for import is apparently the activated LCFA [61]. This conclusion is confirmed by the observation that metabolism of LCFAs is completely dependent on Pxa1p–Pxa2p in faa2Δ mutant cells (see below) lacking peroxisomal acyl-CoA synthetase activity [61]. The mechanism by which Pxa1p–Pxa2p actually transfers its substrate across the membrane remains unclear, but it has been hypothesized that the polar CoA group of the activated fatty acid is moved across the bilayer by Pxa1p–Pxa2p in an ATP-dependent manner [64].

Pxa1p–Pxa2p is homologous to four structurally related human ABC half-transporters: the adrenoleukodystrophy protein ALDP, the 70-kDa peroxisomal membrane protein PMP70, the ALDP-related protein ALDRP, and a PMP70-related protein PMP70R. Concerning the topology of peroxisomal ABC transporters, it has been demonstrated that at least ALDP has its nucleotide binding fold on the cytoplasmic surface of peroxisomal membrane [65]. Mutations in the ALD gene are responsible for the X-linked form of adrenoleukodystrophy (X-ALD), a neurodegenerative disorder that also affects the adrenal gland and is characterized by the accumulation of VLCFAs in the serum due to a block in peroxisomal VLCFA degradation [66, 67]. Mutations have also been identified in the the gene for PMP70 in two patients with Zellweger syndrome [68], an inborn error of peroxisome biogenesis. These transporters might homo- or heterodimerize to allow transport of various substrates which, in analogy to yeast, could also turn out to be activated VLCFAs. However, the intraperoxisomal localization of mammalian VLCFA synthetase [69, 70] points to other substrate specificities for the ALDP transporter family.

Unlike the situation with LCFAs, which in yeast are activated in the cytosol, MCFAs are activated to their CoA derivative inside peroxisomes. This is performed by Faa2p, a peroxisomal acyl-CoA synthetase [61]. Mislocalization of Faa2p to the cytosol causes MCFA β-oxidation to become completely dependent on Pxa1p–Pxa2p [61]. The ATP-consuming activation reaction catalyzed by Faa2p relies on a peroxisomal pool of ATP, which is maintained by the adenine nucleotide transporter Ant1p [58, 59].

Ant1p belongs to the mitochondrial carrier family, whose members are usually localized to the inner mitochondrial membrane [71], albeit peroxisomal representatives have also been identified among this family of proteins in most other species, including PMP34 in humans [72], and Pmp47p in Candida boidinii [73, 74]. The specific transport of adenine nucleotides by Ant1p could be unequivocally demonstrated by functionally reconstituting the purified carrier into liposomes [59]. Ant1p exchanges ATP for ADP, but also for AMP, which is a clear difference from the substrate specificity of the mitochondrial ADP/ATP exchanger [75]. This property is significant since AMP is generated intraperoxisomally upon activation of MCFAs, and therefore it has been suggested that the major function of Ant1p is probably to exchange AMP for cytosolic ATP across the peroxisomal membrane. The growth phenotype of a strain lacking Ant1p concurs with such a function, as it is able to grow on LCFAs but not on MCFAs [58, 59]. A similar phenotype is also observed for the corresponding C. boidinii mutant strain devoid of Pmp47p [76], a protein that can functionally replace Ant1p [59].

An additional protein, Pex11p, has also been suggested to play a role in the transport of MCFAs across the peroxisomal membrane. Pex11p is a PMP involved in the partitioning of expanded peroxisomes into several smaller ones [77, 78]. However, it has also been implicated in β-oxidation [79] by binding fatty acids via a domain that is highly similar to the ligand binding domain of a nuclear hormone receptor [80]. Subsequent investigations reiterated that S. cerevisiae Pex11p is directly involved in peroxisomal proliferation irrespective of the carbon flux through β-oxidation, since over-expression of Pex11p induces peroxisomal proliferation in the absence of fatty acids from the growth medium or an intact β-oxidation process [81]. It has therefore been suggested that the observed effects of deleting the corresponding PEX11 gene on fatty acid metabolism may be indirect. There remains the formal possibility that Pex11p is actually a multifunctional protein involved in both processes, i.e. peroxisomal proliferation and β-oxidation.

4 Classical and auxiliary enzymes of peroxisomal β-oxidation

4.1 The enzymes of classical β-oxidation: Pox1p/Fox1p, Mfe2p/Fox2p, and Pot1p/Fox3p

The core reactions of peroxisomal and mitochondrial β-oxidation can be considered as a variation of the tricarboxylic acid (TCA) cycle steps involved in converting succinate to oxaloacetate via a sequence of dehydrogenase, hydratase, and dehydrogenase. In analogy to the process metabolizing succinate, in which intermediate a double bond is introduced to generate fumarate using an FAD-containing succinate dehydrogenase, the β-oxidation process also begins with oxidation of the acyl-CoA substrate to trans-2-enoyl-CoA by FAD enzymes representing acyl-CoA oxidase in peroxisomes and acyl-CoA dehydrogenase in mitochondria, respectively. Although acyl-CoA dehydrogenases and acyl-CoA oxidases are related with respect to their amino acid sequences [82], mitochondrial FAD enzymes pass on electrons to oxygen via the electron transport chain, whereas peroxisomal oxidases such as Pox1p/Fox1p in S. cerevisiae pass electrons directly to oxygen to generate H2O2. Hydrogen peroxide decomposition to water and oxygen is facilitated by peroxisomal catalase A Cta1p (see below) with concomitant release of energy as heat.

Unlike the situation in the yeast Yarrowia lipolytica, which has several acyl-CoA oxidase isoenzymes with differing chain length preferences [83], in S. cerevisiae Pox1p/Fox1p, an 84-kDa protein, is the only such oxidase. Furthermore, it has been demonstrated that pox1Δ cells are unable to grow on fatty acids as the sole carbon source [42]. Investigations into the substrate specificities of Pox1p/Fox1p and mammalian acyl-CoA oxidases have also helped answer an important question regarding why β-oxidation in the peroxisomes of yeast proceeds to completion, whereas the analogous process in mammalian peroxisomes does not (it does so only in the mitochondria). This is due to the fact that while the mammalian peroxisomal acyl-CoA oxidase metabolizing straight-chain acyl-CoA esters shows preference towards medium-chain acyl-CoAs and is also active with long-chain ones, it has only a very low catalytic rate with short-chain substrates. On the other hand, acyl-CoA oxidase from S. cerevisiae also accepts short-chain substrates, thereby allowing β-oxidation to be completed. Future investigations into the substrate binding preferences of acyl-CoA oxidases will no doubt receive a high priority once the structures of these proteins become available. The medium-chain acyl groups produced in peroxisomal β-oxidation can be transported out of mammalian peroxisomes as acyl carnitines generated in the reaction catalyzed by carnitine octanoyl-transferase [84, 85]. This activity is lacking in fungal peroxisomes.

In contrast to the peroxisomes of S. cerevisiae, those in mammalian cells are more versatile in that they are capable of metabolizing not only straight-chain acyl-CoAs but also 2-methyl branched acyl-CoAs, by having multiple acyl-CoA oxidases with specificities commensurate with these types of substrates: two acyl-CoA oxidases have been characterized in human peroxisomes [86, 87] and three in rat peroxisomes [8892].

The mitochondrial and peroxisomal β-oxidation processes are also not identical with respect to their stereochemistry: mitochondrial β-oxidation of mammals proceeds from trans-2-enoyl-CoA to 3-ketoacyl-CoA via (3S)-hydroxyacyl-CoA esters; fungal peroxisomes utilize (3R)-hydroxy intermediates; and mammalian peroxisomes metabolize both enantiomers [9395]. The two types of enoyl-CoA hydratases (1 and 2) and NAD+-dependent dehydrogenases (specific for (3S)- or (3R)-hydroxyl groups) catalyzing the conversion of trans-2 to 3-keto intermediates (Figs. 1 and 2) are not related at the level of their amino acid sequences [96]. The lack of homology has led to the suggestion that this step of the β-oxidation pathway has been invented twice during the course of evolution. In the fungal system, the hydratase 2 and (3R)-hydroxy-specific dehydrogenase reactions are catalyzed by Mfe2p/Fox2p, which is a homodimeric multifunctional enzyme (160 kDa). In addition to possessing an MFE2-type protein, mammals also have a second peroxisomal multifunctional enzyme, MFE1, which has both 2-enoyl-CoA hydratase 1 and (3S)-hydroxyacyl-CoA dehydrogenase activities [97] as well as a Δ32-enoyl-CoA isomerase activity [98]. The issue of maintaining the peroxisomal pool of NAD+ required for the dehydrogenase step is addressed in Section 5.

Figure 2

Organization of the enzymes catalyzing the second and third reaction of β-oxidation in various subcellular compartments and species. If 3-ketoacyl-CoA thiolase is present, it is contained in the β-subunits of MFE1s. Abbreviations: (R)DH, (3R)-hydroxyacyl-CoA dehydrogenase; (S)DH, (3S)-hydroxyacyl-CoA dehydrogenase; H2, 2-enoyl-CoA hydratase 2; H1, 2-enoyl-CoA hydratase 1; SCP, sterol carrier protein type 2-like; I, Δ32-enoyl-CoA isomerase; E, 3-hydroxyacyl-CoA epimerase; X, C-terminal domain of unknown function; THIOL, 3-ketoacyl-CoA thiolase. References: [9698, 104, 106, 108, 109, 141, 251260].

Mfe2p/Fox2p, first isolated from Candida tropicalis [99], has a duplicated domain organization in its N-terminal half, thereby revealing a partial gene duplication during evolution [96, 100]. Subsequent studies on S. cerevisiae Mfe2p/Fox2p demonstrated that the duplicated regions contain two dehydrogenase domains, termed A and B [101]. Inactivation of the domains by introducing point mutations into the DNA encoding the corresponding dinucleotide binding sites showed that both domains are enzymatically active. Furthermore, domain A has the highest activity with long- and medium-chain substrates whereas domain B is most active with short-chain substrates [101]. The evolution of the two domains with different chain-length preferences is an intriguing method to solve the problem of having to metabolize a large variety of substrates. Interestingly the dehydrogenase domain of the mammalian MFE2 has a broad substrate specificity accepting both long- and short-chain substrates [102].

The fragment in C. tropicalis Mfe2p/Fox2p containing the two dehydrogenase domains is linked at its carboxy-terminus to the 2-enoyl-CoA hydratase 2 domain. Although the yeast and mammalian hydratase 2 enzymes are related with respect to their amino acid sequences, they have different kinetic properties: the turn-over rate of the mammalian enzyme decreases rapidly with substrates shorter than C8 [103, 104], whereas the yeast enzyme additionally hydrates short-chain substrates. The mammalian peroxisomal MFE2 was first identified as 17β-hydroxysteroid dehydrogenase type IV that is proposed to be involved in steroid hormone metabolism [105]. However, subsequent studies including both in vitro assays using the isolated enzyme [106110] and in vivo analysis of MFE2−/− knockout mice [111] have challenged this assertion. Mammalian MFE2 actually contributes to the metabolism of isoprenoid-derived acyl-CoAs such as those released following cleavage of the cholesterol tail during bile acid synthesis as well as to shortening of acyl-CoAs with very long hydrocarbon chains. It is possible that the different physiological functions of yeast and mammalian MFE2s are also due to the existence of a sterol carrier protein 2-like domain in the mammalian enzyme, which is lacking in the yeast protein.

At the next reaction of the β-oxidation cycle the ketoacyl-CoA intermediate undergoes thiolytic cleavage by Pot1p/Fox3p, which represents 3-ketoacyl-CoA thiolase [44, 45]. The products of this last step are acetyl-CoA and a C2-shortened acyl-CoA, the latter acting as substrate for Pox1p/Fox1p. The possible fates of acetyl-CoA generated in the peroxisomes are discussed below. In contrast to yeast, mammals have three different peroxisomal thiolases. 3-Ketoacyl-CoA thiolases A and B [112114] have similar broad substrate specificities and thiolase B presents the enzyme which is highly inducible by peroxisome proliferators in rodents. The third thiolase, sterol carrier protein 2/3-ketoacyl-CoA thiolase (sterol carrier protein X) also accepts 2-methyl-branched substrate in addition to the conventional straight-chain substrates and it carries an N-terminal sterol carrier protein 2 domain [115, 116].

Determination of the the crystal structure of Pot1p/Fox3p [117, 118] has unveiled the structural basis for the different properties of various thiolases. Pot1p/Fox3p is a dimeric protein with a subunit size of 45 kDa. A single Pot1p/Fox3p subunit is composed of three domains: two core domains, which have the same fold, and a loop domain of 120 residues [117]. This loop is an insertion in the N-terminal domain (Fig. 3). The core domains fold into a mixed five-stranded β-sheet, which is covered on each side by α-helices. Pot1p/Fox3p is targeted to peroxisomes using the N-terminal PTS2 [46] but this region is disordered in the thiolase structure [117, 118]. The active site of yeast thiolase is shaped by residues from the two core domains and surrounded by the loop domain [117, 118]. The catalytic triad, formed by Cys125, His375 and Cys403 situated at the bottom of the cylindrical pocket comprising the active site, is responsible for catalysis [117, 118].

Figure 3

The structure of the yeast Pot1p/Fox3p homodimer. The view is along the dimer twofold axis towards the active sites of both subunits. The active site of each subunit is marked by the three catalytic side chains (in white), Cys125, His375 and Cys403. The right subunit is in blue, whereas for the left subunit the N-terminal domain is green, the C-terminal domain is red and the loop domain is yellow. The white asterisk marks the position of the predicted binding pocket of the fatty acid tail.

The active site pocket is extended beyond the three catalytic residues so as to be able to bind the fatty acid tail, which can be of variable length [117, 118]. The crystal structure of a bacterial thiolase that functions in the synthesis of acetoacetyl-CoA from two molecules of acetyl-CoA is devoid of such a pocket. This synthetic thiolase can also degrade acetoacetyl-CoA but not 3-ketoacyl-CoA with longer fatty acid tails. It has not been possible to obtain crystal structures of yeast thiolase complexed with active site ligands; however, structures are available of several liganded bacterial synthetic thiolase complexes, and these structures have improved our understanding of the thiolase catalytic mechanism to a large extent. It turns out that His375 is important for activating Cys125. Cys125 becomes acetylated in the reaction cycle and the role of Cys125 is to shuttle the acetyl group from the initial substrate, 3-ketoacyl-CoA, to the CoA molecule. Cys403 is shuttling protons during the reaction cycle (Kursula and Wierenga, unpublished observation).

4.2 Peroxisomal catalase A Cta1p

As mentioned previously, the FAD enzyme Pox1p/Fox1p transfers electrons derived from the conversion of acyl-CoA to trans-2-enoyl-CoA directly to oxygen to yield hydrogen peroxide. Catalase then breaks down H2O2 to water and oxygen, and represents the hallmark enzyme for peroxisomes as well as other analogous structures. For example, catalase is found in the peroxisomes of mammals and fungi [2], in the glyoxysomes of plants [119], and in the glycosomes of Leishmania donovani [120]. In the yeast S. cerevisiae, peroxisomal catalase A Cta1p in combination with cytosolic catalase Ctt1p is thought to play an important role in the survival of mother cells under conditions of increased oxygen [121].

The issue of longevity notwithstanding, the main physiological role of Cta1p appears to be to remove the H2O2 formed by acyl-CoA oxidase during fatty acid breakdown [122]. Expression of both Cta1p and Pox1p/Fox1p is up-regulated in S. cerevisiae cells grown under fatty acid medium conditions [42, 123]. The presumed close metabolic association between these two peroxisomal enzymes has been used previously as the rationale for applying H2O2 to yeast cells in positive selection procedures for identifying mutants with defective peroxisomes [124, 125]. Moreover, it has also been chronicled in the literature that a strain lacking Cta1p grows poorly on solid medium containing oleic acid plus glycerol compared to a related strain in which this activity was intact [125]. However, the oleic acid growth phenotype associated with catalase mutants appears not to be reproducible using other laboratory strains.

To examine further the requirement for catalase during β-oxidation, a set of isogenic yeast strains was used that lack catalase activity [121]. The strains were applied to solid medium containing oleic acid (cis-C18:1(9)) as a sole carbon source (Fig. 4A). The results demonstrated that W303-derived strains lacking Cta1p, Ctt1p, or both, formed clear zones in the opaque medium that were similar to those formed by the wild-type strain. Formation of such clear zones surrounding regions of cell growth indicates utilization of the fatty acid substrate, since mutant cells with dysfunctional peroxisomes that are additionally defective in β-oxidation, such as the BY4741-derived pex5Δ cells shown here, fail to take up oleic acid from the medium to generate clear zones (Fig. 4A, lower panel). Application of strains to solid medium supplemented with petroselinic acid (cis-C18:1(6)) or with palmitic acid (C16), a saturated fatty acid that is not able to mop up reactive oxygen species, also failed to reveal a significant impairment in fatty acid uptake (Fig. 4B,C). Hence, this indicated that cta1Δ mutants utilized the fatty acid substrate as efficiently as the wild-type and, therefore, catalase activity was dispensable for this process.

Figure 4

Growth of catalase mutants on fatty acid media is unaffected. The indicated strains were spotted or streaked on solid medium supplemented with (A) oleic acid, (B) petroselinic acid, or (C) palmitic acid, prepared as described [131]. Strains derived from the W303 genetic background used in A, B, and C have been published [121] and were kindly supplied by Dr. Christoph Schüller, whereas those from the background of BY4741, used in A, were obtained from EUROSCARF.

The lack of an essential metabolic link between catalase and β-oxidation in peroxisomes is supported by additional evidence. An investigation of catalase-less peroxisomes obtained from a mutant Chinese hamster ovary cell line showed that the capacity for β-oxidizing VLCFAs was comparable to that of wild-type cells [126]. Also, Trypanosoma brucei, the etiologic agent of African sleeping sickness, contains enzymes for β-oxidation in its glycosomes [127], but peroxisomal catalase is absent from these structures [120]. We reason that β-oxidation per se does not depend on a functional peroxisomal catalase.

4.3 The auxiliary enzymes of β-oxidation: Eci1p, Sps19p, and Dci1p

Double bonds in naturally occurring (poly)unsaturated fatty acids are mostly in the cis configuration, but certain fatty acids have their double bonds in the trans configuration due to their enzymatic synthesis by microorganisms or due to partial hydrogenation of fats and vegetable oils during industrial food processing [128]. S. cerevisiae cells can utilize both cis- and trans-unsaturated fatty acids as a sole carbon and energy source [129].

Classical β-oxidation is sufficient for the complete breakdown of saturated fatty acids or those containing trans double bonds at even-numbered positions. However, for degrading cis or trans double bonds at odd-numbered positions, or cis double bonds at even-numbered positions, auxiliary enzymes are required (Fig. 5). By combining the action of auxiliary enzymes with those of the main β-oxidation pathway, metabolism of (poly)unsaturated enoyl-CoA esters can proceed, at least in principle, via alternative routes [130].

Figure 5

Auxiliary enzymes participating in the β-oxidation of (poly)unsaturated enoyl-CoAs in S. cerevisiae.

Following two rounds of β-oxidation, the original double bond of oleic acid at position Δ9 is moved to position Δ5, with the concomitant introduction of a trans-2 double bond by Pox1p/Fox1p. This cis-5,trans-2 intermediate has several possible fates, one of which includes a third round of β-oxidation that moves the original double bond to position Δ3, resulting in cis-3-enoyl-CoA. This intermediate acts as substrate for Δ32-enoyl-CoA isomerase Eci1p [131, 132], catalyzing the transfer of the Δ3 double bond (cis and trans) to a trans double bond at position Δ2. This new arrangement is now suitable for further degradation by Mfe2p/Fox2p. An alternative metabolic route is explained later.

All of the characterized Δ32-enoyl-CoA isomerases belong to the low-homology hydratase/isomerase protein family. Three mammalian isoforms of Δ32-enoyl-CoA isomerase have been identified [133135]: two isomerases that are apparently monofunctional enzymes, and one that is an integral part of the peroxisomal multifunctional enzyme type 1, perMFE1 [135138]. One of the two monofunctional isomerases (MECI in rat) is a predominantly mitochondrial enzyme thought to additionally occur in peroxisomes (see Section 7) and is most active in catalyzing cis-3 to trans-2 isomerization, whereas the other dually localized monofunctional isomerase (ECI in rat) is optimal for trans-3 to trans-2 isomerization [139]. Interestingly, glyoxysomes of cucumber seedlings representing a modified peroxisomal compartment have also both monofunctional Δ32-enoyl-CoA isomerase and isomerase activity as part of a multifunctional enzyme [140, 141].

Analysis of yeast eci1Δ mutant cells devoid of Δ32-enoyl-CoA isomerase activity demonstrated that they were defective for growth on media containing unsaturated fatty acids such as oleic acid but were otherwise unaffected when a saturated fatty acid such as palmitic acid was supplied as the sole carbon source. When combined as a fusion with green fluorescent protein (GFP), Eci1p could be shown to be located in peroxisomes. The kinetic properties of Eci1p have been well characterized [131, 132] and it was found to be a monofunctional Δ32-enoyl-CoA isomerase.

Degradation of cis double bonds at even-numbered positions requires a further auxiliary enzyme. Following two rounds of β-oxidation, the breakdown of petroselinic acid containing a double bond at position Δ6 yields a cis-4,trans-2 dienoyl-CoA intermediate, which is metabolized slowly in β-oxidation. This is facilitated by an NADPH-dependent reduction at carbons 2 and 4 that is catalyzed by 2,4-dienoyl-CoA reductase Sps19p [142, 143]. The resulting trans-3-enoyl-CoA intermediate is further metabolized to trans-2-enoyl-CoA by the above-mentioned Eci1p prior to re-entry into β-oxidation at the Mfe2p/Fox2p-catalyzed step. Reduction of NADP+ back to NADPH is explained in a separate section below.

The characterized eukaryotic 2,4-dienoyl-CoA reductases are members of a large short-chain alcohol dehydrogenase/reductase superfamily. In contrast to the eukaryotic counterparts, bacterial 2,4-dienoyl-CoA reductase [144] is a flavin-containing enzyme that produces trans-2-enoyl-CoA as the end-product, and not trans-3-enoyl-CoA. At least three isoforms of 2,4-dienoyl-CoA reductase exist in the rat, two mitochondrial and one peroxisomal [145, 146]. As indicated above, Sps19p [147], an oleic acid-inducible peroxisomal protein, represents the sole 2,4-dienoyl-CoA reductase in S. cerevisiae [142, 143]. Sps19p has 28% identity to the human mitochondrial reductase, and an sps19Δ mutant strain is unable to utilize petroselinic acid as the sole carbon source, but its growth remains unaffected on oleic acid. Purified Sps19p exists as a homodimeric enzyme with a native molecular mass of 69 kDa.

Studies comparing the breakdown of linoleic acid (cis,cis-C18:2(9,12)) with that of linolelaidic acid (trans,trans-C18:2(9,12)) revealed that trans double bonds at even-numbered positions can be metabolized by β-oxidation without the involvement of Sps19p [129]. However, like the situation with cis double bonds at odd-numbered positions, odd-numbered trans double bonds also require Eci1p to be repositioned to Δ2 so as to be accepted by Mfe2p/Fox2p [129, 148].

An alternative way for degrading oleic acid is hypothesized to proceed on the line that following two rounds of β-oxidation and the participation of Pox1p/Fox1p, a trans-2,cis-5 intermediate arises which, with the help of Eci1p, is converted to cis/trans-3,cis-5-dienoyl-CoA. Since this postulated 3,5 intermediate of oleic acid degradation will be recalcitrant to further degradation, cells employ a third auxiliary enzyme, Δ3,52,4-dienoyl-CoA isomerase Dci1p [149, 150], which is a member of the hydratase/isomerase protein family [149151]. This enzyme converts 3,5-dienoyl-CoAs to the corresponding 2,4-dienoyl-CoA. If the latter intermediates are in the trans,trans configuration, they will be accepted as substrate directly by Mfe2p/Fox2p [129, 148]. However, if trans-2,cis-4-dienoyl-CoAs are generated, a combination of Sps19p and Eci1p would metabolize them to the permissible trans configuration at position Δ2.

The enzymatic reactions catalyzed by Dci1p and Eci1p are very similar in that they both involve an isomerization of a double bond between carbon atoms 2 and 3 of the substrate molecule. In addition, Dci1p can simultaneously isomerize the double bond from carbon atom 5 to carbon atom 4. The similarity of Eci1p and Dci1p is highlighted also by their amino acid sequences; they are 50% identical to each other [149151]. Although the enzyme activity of Dci1p can be demonstrated in soluble extracts derived from yeast cells overproducing it, thus far not one of the groups working on Dci1p has been able to purify this yeast protein to homogeneity while still retaining enzyme activity.

Dci1p additionally contains an intrinsic Δ32-enoyl-CoA isomerase activity; mutant eci1Δ cells overproducing Dci1p could generate 3,5,8,11,14-eicosapentenoyl-CoA from 2,5,8,11,14-eicosapentenoyl-CoA [150]. Amplification of this intrinsic activity restored the growth of an eci1Δ strain on oleic acid for which di-isomerase is not required whereas Eci1p is. The possibility that this effect was due to di-isomerase activity was excluded by the observation that heterologous expression of rat Δ3,52,4-dienoyl-CoA isomerase, which lacks Δ32-enoyl-CoA isomerase activity, was not able to re-establish growth of the eci1Δ mutant on oleic acid. It is unlikely that the intrinsic Δ32-enoyl-CoA isomerase activity in Dci1p actually participates in vivo in degrading fatty acids, since the dci1Δ strain grows on oleic acid to wild-type levels, whereas cells lacking both Eci1p and Dci1p grow as poorly as the eci1Δ mutant [150].

Evidence for the function of a Dci1p-dependent pathway in yeast is still fragmentary. Depending on the strain background used, oleic acid-grown yeast cells devoid of Dci1p are essentially symptomless [149, 150], as are those without a functional Sps19p [142, 143]. On the other hand, investigations of yeast grown on conjugated linoleic acid (CLA) isomers (cis,trans-C18:2(9,11) and trans,cis-C18:2(10,12)) revealed that cells do not grow efficiently when a 3,5-dienoyl-CoA intermediate arises during the breakdown of cis-9,trans-11 CLA, whereas cells grow abundantly when supplied with (cis-9) oleic acid [148]. The ambiguity of the data relating to the Dci1p pathway could possibly be resolved in the future by considering creative ways for quantifying the intermediates thought to be generated by the carbon flux through it. Heterologous generation of polyhydroxyalkanoates in S. cerevisiae using a bacterial synthase showed that these are produced from the 3-hydroxyacyl-CoA intermediates formed by Mfe2p/Fox2p of peroxisomal β-oxidation [152]. This diverts β-oxidation intermediates from the spiral so as to be captured in the form of polyhydroxyalkanoates, which are resistant to further fungal degradation. Extending these studies to the production of polyhydroxyalkanoates in yeast cells devoid of the auxiliary enzymes Eci1p, Dci1p, and Sps19p is expected to supply the sorely needed data to complete the picture of the carbon flux through the alternative pathways in peroxisomal β-oxidation.

4.4 Putative cryptic PTSs in Eci1p and Dci1p

Contradictory data have appeared in the literature as to how Eci1p and Dci1p reach the peroxisomes. Both proteins contain sequences resembling PTS1 at their C-termini; Eci1p is extended with HRL whereas Dci1p ends with HKL [131, 132, 149, 150]. Karpichev and Small [153] suggested that the two proteins additionally contain potential PTS2s at their N-termini, but that neither PTS1 nor PTS2, or their cognate receptors, is essential for the import of Eci1p and Dci1p. It was thus claimed that the two proteins can be targeted to peroxisomes in an unidentified fashion. They also reported that Eci1p and Dci1p are transported independently of each other. However, other groups have found a two-hybrid interaction between Eci1p and Dci1p [149, 154] and, furthermore, Eci1p appears to require the presence of Dci1p for efficient peroxisomal import. It was suggested that these two proteins might be imported as a complex [149, 155].

In agreement with the findings by Karpichev and Small [153], Yang and co-workers showed that the PTS2 of Eci1p is not used for peroxisomal targeting [155]. Since targeting was dependent on Pex5p, this implied that PTS1 is the one actually needed for the import. Interestingly, the PTS1 of Eci1p was not required for peroxisomal import when Dci1p was present. Only when Dci1p was not available could small amounts of Eci1p be imported into the peroxisome by relying on its own PTS1. This indicated that Eci1p preferably uses the PTS1 of Dci1p to enter peroxisomes and reiterated that these two proteins enter as a complex [155]. It should be noted that, at least in yeast, PTS1 is sometimes associated with proteins additionally containing redundant, cryptic PTSs, e.g. Cta1p [156] and Cat2p [157].

Using N-terminal GFP fusions of Eci1p and Dci1p [131, 150], we examined here (i) whether PTS1-terminating Eci1p and Dci1p depend solely on the cognate receptor Pex5p for peroxisomal import and, (ii) in light of their mutual interaction in the two-hybrid system [149, 154], whether Eci1p and Dci1p depend on each other for correct subcellular localization. Yeast eci1Δ cells over-expressing GFP-Eci1p or GFP-Dci1p were able to form clear zones on oleic acid plates that were indistinguishable from that formed by the wild-type strain (Fig. 6A). Hence, these GFP fusions were functional, indicating that they were able to oligomerize to a hexameric structure and possibly also to interact with native Eci1p or Dci1p.

Figure 6

Peroxisomal import of Eci1p and Dci1p primarily depends on their PTS1s. A: The ability of the Eci1p- or Dci1p-extended GFP fusions to rescue the phenotype of an eci1Δ mutant was examined on oleic acid medium. Yeast eci1Δ cells expressing GFP-Eci1p or GFP-Dci1p were compared to those expressing GFP-SKL and an isogenic wild-type strain for clear zone formation on oleic acid medium. The eci1Δ mutant expressing GFP-SKL did not utilize the fatty acid, whereas those mutants expressing GFP-Eci1p or GFP-Dci1p formed clear zones that were indistinguishable from that formed by the wild-type strain. B: The requirement of Eci1p and Dci1p for Pex5p during import into peroxisomes was determined using the corresponding GFP fusions that were expressed in a pex5Δ mutant strain. The two GFP fusions gave rise to punctate fluorescence in wild-type cells but not in the pex6Δ mutants. Arrows point to infrequent GFP-Dci1p punctation in pex5Δ cells. C: The two GFP fusions were expressed in eci1Δ and dci1Δ cells. Punctate green fluorescence could be demonstrated in all of the strains tested.

To determine whether Eci1p and Dci1p require Pex5p for import into peroxisomes, the corresponding GFP fusions were expressed in a pex5Δ mutant strain. As a comparison, cells lacking Pex6p were also used [158]. In the situation with cells devoid of Pex5p, expression of GFP-Dci1p resulted in punctation in less than 10% of the pex5Δ mutants (arrows, Fig. 6B), whereas GFP-Eci1p remained diffuse (corresponding left panel). If these rare structures indeed represented peroxisomes loaded with GFP-Dci1p, then this might point to an inefficient Pex7p-mediated import of the fusion protein. The more important conclusion drawn from this experiment, however, was that import of Eci1p and to a large extent also that of Dci1p depended on Pex5p. This implies that like in the case of Cat2p, which contains an additional Pex5p-dependent internal PTS [157], the proposed additional internal PTS in Dci1p and possibly in Eci1p also relies on Pex5p for efficient function. An internal Pex5p-dependent PTS could resemble that of Pox1p, which interacts with Pex5p independently of the latter's TPR domain [43].

To address the second issue of whether Eci1p and Dci1p interact with one another to facilitate peroxisomal import, the two GFP fusions were expressed in eci1Δ and dci1Δ cells. Examination of living cells for green fluorescence demonstrated that all of the strains tested contained ample punctation (Fig. 6C), indicating that neither Dci1p nor Eci1p had an important role in the peroxisomal import of itself or its homologue. This result was in agreement with observations made following cell fractionation experiments conducted on eci1Δdci1Δ double mutants expressing tagged truncations of Eci1p and Dci1p [153], but contradicted previous suggestions that Dci1p acts as a chaperone during peroxisomal import of Eci1p [149], or that Eci1p preferably uses the PTS1 of Dci1p to enter peroxisomes [155]. Hence, using functional GFP reporters, peroxisomal import of Eci1p and Dci1p could be shown to rely primarily on Pex5p, presumably by interacting with their resident PTS1s represented by HRL and HKL [159]. Only when these signals are artificially clipped, redundant topogenic internal signals or a detour via piggybacking might come into action to translocate these proteins.

4.5 Comments on hydratase/isomerase proteins and structure–function relationships

More than 30 members currently comprise the hydratase/isomerase superfamily, including Eci1p and Dci1p. The enzymes of this family possess a typical amino acid sequence pattern, although their overall sequence identity is very low [160162]. The superfamily members are thought to have evolved from a common ancestor and, therefore, to be both mechanistically and structurally related. The subunit molecular mass of all family members is approximately 30 kDa. Hydratase/isomerase proteins catalyze a wide range of metabolic reactions, including dehalogenation, hydration/dehydration, isomerization, decarboxylation, formation or cleavage of carbon–carbon bonds, and hydrolysis of thioesters. With one exception, all family members use coenzyme A esters as substrates, and their catalytic mechanisms involve the stabilization of an oxyanion intermediate [162, 163].

The active sites of hydratase/isomerase proteins commonly display residual catalytic activities that are related to the actual enzyme activity. For example, rat Δ3,52,4-dienoyl-CoA isomerase contains traces of enoyl-CoA hydratase activity [164], and yeast Dci1p has some of Δ32-enoyl-CoA isomerase [150] as does rat enoyl-CoA hydratase [165]. Unlike the situation with Dci1p, Eci1p only has Δ32-enoyl-CoA isomerase activity but no detectable hydratase or dienoyl-CoA isomerase activities [131, 132].

The crystal structures of six hydratase/isomerase proteins are currently known. Of these, five are enzymes, including 4-chlorobenzoyl-CoA dehalogenase from Pseudomonas [166], rat enoyl-CoA hydratase [167, 168], rat Δ3,52,4-dienoyl-CoA isomerase [169], methylmalonyl-CoA decarboxylase from E. coli [170], and yeast Δ32-enoyl-CoA isomerase, Eci1p [171] (Fig. 6). The structure of the sixth protein, a human RNA binding protein that is also referred to as the AUH protein, has also been determined [172]. The AUH protein was shown to have low enoyl-CoA hydratase activity [172].

When compared to the other known structures within the hydratase/isomerase superfamily, the overall topology of the Eci1p subunit is highly similar. Nevertheless, a superposition of the structures of the five enzymes highlights considerable structural differences, in particular the positioning of α-helices with respect to the β-sheets (Fig. 7). Such structural diversity agrees well with the low sequence identity in the hydratase/isomerase superfamily. The structure consists of an N-terminal core domain, containing the first 200 amino acid residues, and a C-terminal trimerization domain (approximately 100 residues). The N-terminal domain has a spiral fold topology formed of four turns, each turn being shaped by two β-strands and an α-helix. The β-strands form two β-sheets, which are almost at right angles with respect to each other and connected by the α-helices. The C-terminal domain contains four α-helices, H7, H8, H9 and H10. In particular H8 and H9 are used for subunit–subunit contacts when three Eci1p subunits assemble into a trimer. All the known hydratase/isomerase structures assemble into disk-like trimers via contacts made by the C-terminal domain. In addition, the two trimers assemble together to form a hexamer. These hexamers can thus be described as dimers of trimers.

Figure 7

The comparison of the Cα traces of the Δ32-enoyl-CoA isomerase Eci1p (yellow) with crotonase (red), dienoyl-CoA isomerase (cyan), dehalogenase (green) and methylmalonyl-CoA decarboxylase (gray). The view is from the bulk solvent into the active site in which the ligand for the dehalogenase is shown as a ball-and-stick model. In all the figures helix H8 of the neighboring isomerase subunit is shown and labelled as H8′. Helices H1, H2, H9 and H8′ form a continuous surface facing the intertrimer space. The extreme 10 C-terminal amino acid residues including PTS type 1 are disordered and not visible in the structure. The N- and C-termini of the visible Eci1p structure are identified. For the superposition the Cα atoms of 25 residues were used: 19–23, 56–60, 115–119, 122–124, 136–139 and 145–147, which are residues of β-strands A1, A2, A3, B3, A4 and B4, respectively, of Eci1p. The largest structural variability is seen for H2 and H9. These two helices are near the active site and are important determinants for defining the binding pocket of the fatty acid tail.

The most notable difference in the yeast Eci1p structure when compared to the structures of 2-enoyl-CoA hydratase [167], dienoyl-CoA isomerase [169] and 4-chlorobenzoyl-CoA dehalogenase [166] is the structural switch of the helices H9 and H10 of the C-terminal domain. This is an example of domain swapping that was first described for the seminal ribonuclease dimer [173] and the diphtheria toxin dimer [174], and is reviewed by Newcomer [175]. In Eci1p, H9 and H10 are positioned so that they fold over the core domain and cover the active site of the same subunit [171]. In the structures of enoyl-CoA hydratase, dienoyl-CoA isomerase and dehalogenase, however, H9 and H10 protrude away from the core domain and cover the active site of the neighboring subunit. The structural switch of H9 and H10 in Eci1p is also seen in methylmalonyl-CoA decarboxylase [170]. Despite the domain swapping, in all trimers of the hydratase/isomerase enzymes, the positions of H9 and H10 are equivalent, and the active sites are always covered by the C-terminal domain of either the same or the adjacent subunit.

The catalytic amino acid residue of Eci1p is Glu158 [171], which is conserved also as Asp204 in dienoyl-CoA isomerase [169] and as Asp145 in 4-chlorobenzoyl-CoA dehalogenase [166]. Glu158 is probably the only catalytic residue involved in proton shuttling in Eci1p since the Glu158Ala mutant protein is totally inactive and in the crystal structure no other protic residues can be seen in the vicinity of Glu158 [171]. Other important residues for the function of Eci1p are Ala70 and Leu126. The main-chain NH groups of these residues are seen to form the so-called oxyanion hole by making hydrogen bonds to the thioester oxygen of the enoyl-CoA substrate molecule during catalysis. These hydrogen bonds stabilize the negative charge developing on the thioester oxygen during the transition state of the enzymatic reaction [171].

4.6 Preliminary characterization of Ehd3p/Ydr036cp

In addition to Eci1p and Dci1p (formerly Ehd1p and Ehd2p), S. cerevisiae contains a third, novel, hydratase/isomerase protein, Ydr036cp [150, 176], which we call Ehd3p. This raises the issue of whether Ehd3p might also function in peroxisomal β-oxidation of fatty acids. Ehd3p contains neither an obvious PTS nor an oleate response element (ORE) in the promoter of its corresponding gene, which in yeast are hallmarks of genes encoding enzymes involved in β-oxidation [123, 177]. To investigate whether transcription of EHD3/YDR036c is affected in cells grown on fatty acid medium, Northern analysis was performed as described [178] using RNA obtained from wild-type cells as well as pip2Δ mutants in which oleic acid induction is abolished [178]. Cells were grown on glucose, ethanol, or oleic acid medium. As a control, the Northern filter was also probed with the oleic acid-inducible genes POT1/FOX3 [44, 45] and SPS19 [142]. The results showed that the POT1/FOX3 and SPS19 transcripts were detectable predominantly in the oleic acid-grown wild-type; however, the transcripts of EHD3/YDR036c were mostly detected under ethanol medium conditions (Fig. 8A). Since the oleic acid-dependent signal for this gene was very much pronounced in the pip2Δ lane (Fig. 8A), which was not the case with POT1/FOX3 and SPS19, it was reasoned that the transcriptional pattern of EHD3/YDR036c is consonant with the lack of an ORE in the gene's promoter. Hence, EHD3/YDR036c is probably not regulated like other oleic acid-inducible genes.

Figure 8

Preliminary characterization of Ehd3p indicates that it is not involved in fatty acid degradation. A: Northern analysis was performed as described [178]. ACT1 encoding actin served as a loading control. B: Mutant ehd3/ydr036cΔ cells were compared to otherwise isogenic pex6Δ cells or to the wild-type strain for clear zone formation on solid oleic acid medium. YDR036c is reserved in SGD as MRP5.

To exclude a direct role for Ydr036cp in the structural integrity of peroxisomes, a yeast strain was generated with a deletion in the gene encoding it [179]. Analysis of ehd3/ydr036cΔ cells was performed for presence of peroxisomes as well as for integrity of nuclei, mitochondria, and general morphology; however, the strain was found to be wild-type for all the features examined. Finally, to determine whether β-oxidation was affected in ehd3/ydr036cΔ mutants, these were compared to the wild-type strain for clear zone formation on solid oleic acid medium. Mutant cells devoid of Pex6p served as a control for impaired β-oxidation [158]. The results showed that ehd3/ydr036cΔ mutant cells gave rise to clear zones that were indistinguishable from those generated by the wild-type strain (Fig. 8B).

To study further the role of Ehd3p, it was over-expressed in E. coli, and the resulting recombinant protein was purified to apparent homogeneity. The purified Ehd3p was found not to catalyze any of the known β-oxidation reactions. Based on sequence alignments, the amino acid sequence of Ehd3p was shown to have the highest identity with the human 3-hydroxyisobutyryl-CoA hydrolase [180], with 35% identity in an overlap of 240 residues. 3-Hydroxyisobutyryl-CoA hydrolase is involved in the valine catabolic pathway where it has an important role in destroying a toxic intermediate, methacrylyl-CoA, together with crotonase [181]. Ehd3p was actually able to catalyze the specific hydrolysis of 3-hydroxyisobutyryl-CoA. However, because the turnover rate of Ehd3p (7.1 μmol min−1 mg protein−1 [182]) was only a fraction of that of the mammalian enzyme and because no clear phenotype for the ehd3/ydr036cΔ could be found, the physiological function of Ehd3p remains unclear.

5 Ancillary factors of peroxisomal β-oxidation

In addition to those enzymes directly involved in β-oxidation, a number of other peroxisomal matrix proteins as well as some cytosolic factors are needed for trafficking metabolites between the peroxisomal compartment and other subcellular sites in S. cerevisiae grown on fatty acids. These proteins include carnitine acetyltransferase Cat2p (found in both peroxisomes and mitochondria), peroxisomal citrate synthase Cit2p [57, 157, 183, 184], peroxisomal malate dehydrogenase Mdh3p [185, 186] as well as cytosolic and peroxisomal isocitrate dehydrogenases Idp2p and Idp3p, respectively, and glucose-6-phosphate dehydrogenase Zwf1p [187]. Another yeast protein postulated to be important for optimal β-oxidation is the peroxisomal acyl-CoA thioesterase Tes1p/Pte1p [188, 189].

5.1 Reduction equivalents and dinucleotide cofactors linked to peroxisomal fatty acid degradation

Two oxidation–reduction equivalents, FAD and NAD+, are necessary for β-oxidation of fatty acids and an additional third one, NADPH, for the reductase-linked metabolism of unsaturated fatty acids. The issue of FAD has been dealt with above in the context of Pox1p/Fox1p. The Mfe2p/Fox2p-dependent enzyme activity of (3R)-hydroxyacyl-CoA dehydrogenase reduces NAD+. However, in light of the fact that the peroxisomal membrane is impermeable to (di)nucleotides, this raises the question of how the intraperoxisomal pool of NAD+ is maintained for repeated dehydrogenation by Mfe2p/Fox2p. One plausible theory is that peroxisomal Mdh3p restores NAD+ levels by converting oxaloacetate to malate in an NADH-dependent manner, with a cytosolic malate dehydrogenase Mdh2p converting malate back to oxaloacetate [57, 186]. Although the issue of whether an aspartate/malate cycle contributes to the shuttling of NADH across the peroxisomal membrane is not yet settled, nevertheless the presence of a dually localized cytosolic and peroxisomal aspartate aminotransferase Aat2p that could facilitate malate translocation adds credence to this suggestion. However, despite the headway made in identifying PMPs [190], to date no positive proof has been presented for the existence of the metabolite transporter proteins that would be needed for this putative NADH shuttle to operate. It is worth noting that when generated in the cytosol, NADH can be reoxidized in the mitochondria of S. cerevisiae via several mechanisms: by extramitochondrial NADH dehydrogenases, the glycerol-3-phosphate shuttle, or oxidation by intramitochondrial NADH dehydrogenases. Because the inner mitochondrial membrane is impermeable to NADH, the latter mechanism requires shuttles [191].

Metabolism of dienoyl-CoA esters with double bonds at even-numbered positions depends on NADPH acting as cofactor for peroxisomal 2,4-dienoyl-CoA reductase, Sps19p [25, 142, 192, 193]. A further shuttle mechanism has been postulated for this reduction equivalent. According to the proposed model, 2-ketoglutarate is reduced in the cytosol to isocitrate by Idp2p using NADPH. Isocitrate is then imported into the peroxisomes, where it is oxidized back to 2-oxoglutarate by peroxisomal Idp3p [194, 195]. In this way NADP+ is reduced to NADPH. The resulting 2-oxoglutarate is transported out of the peroxisome to complete the NADPH regeneration cycle in the cytosol. As in the case of reoxidation of NADH, none of the shuttle transporters postulated to be involved in the regeneration of NADPH has been identified to date.

5.2 Export of peroxisomally generated acetyl-CoA

In yeast, straight-chain fatty acids can be completely degraded by peroxisomal β-oxidation, which results in the production of acetyl-CoA units. In the case of odd-numbered fatty acids, the final thiolytic step generates propionyl-CoA. Export of these compounds is thought to be accomplished through their conjugation to carnitine by the carnitine acetyl transferase Cat2p, which is localized to both peroxisomes and mitochondria [157]. The mitochondrial form catalyzes the opposite reaction, thereby supplying acetyl-CoA units to the TCA cycle.

The carrier protein facilitating transport of acetyl-carnitine across the inner mitochondrial membrane has now been identified as Crc1p, and its substrate specificity was determined by functional reconstitution in proteoliposomes [196, 197]. Crc1p transports carnitine, acetyl-carnitine, and propionyl-carnitine, but not medium- or long-chain derivatives into mitochondria, in exchange for free carnitine [197]. The peroxisomal counterpart so far remains unidentified. The possibility of a dually localized Crc1p has been addressed with a myc-tagged version of the protein [196]; however, no evidence was found for a separate peroxisomal pool of Crc1p. It is worth noting that the human mitochondrial carnitine carrier had been additionally localized to peroxisomes [198].

Interestingly, disruption of the gene for either Cat2p, which generates acetyl-carnitine, or Crc1p, which transfers acetyl-carnitine across the inner mitochondrial membrane, does not have any obvious growth phenotype on fatty acid media [157, 196]. This is due to an alternative pathway for exporting acetyl-CoA out of the peroxisomes relying on the peroxisomal isoform of citrate synthase Cit2p [157]. Cit2p catalyzes the conversion of acetyl-CoA and oxaloacetate into citrate, which is presumed to be able to leave the peroxisomal compartment. This is thought to be facilitated by a carrier that exchanges citrate for either isocitrate or 2-ketoglutarate. Subsequent conversion of citrate into succinate and glyoxylate by a combination of aconitase followed by cytosolic Icl1p allowing a carnitine-independent export of acetyl units out of peroxisomes. The generated C4 metabolites can be used for glucosegenesis (see below) or oxidized via pathway(s) having pyruvate as an intermediate to CO2 in the TCA cycle.

Genetic evidence argues in favor of this route, since a cit2Δ mutant strain is not impaired for growth on ethanol or oleic acid whereas cit2Δcat2Δ and cit2Δcrc1Δ double mutants are [157, 196]. Furthermore, the overall β-oxidation rates in intact mutant cells are drastically reduced. These data were interpreted to mean that in the absence of Cit2p, the mitochondrial isoform of the citrate synthase, Cit1p, is able to substitute for Cit2p while still performing its function in the TCA cycle. However, this only works as long as both cycles are continuously fed with acetyl-CoA, which would become scarce in the mitochondria in the absence of the acetyl-carnitine transporter or the mitochondrial carnitine acetyl-transferase.

As mentioned previously, fatty acids are degraded to completion in yeast peroxisomes, and so there is no formal requirement for transporting shortened acyl-CoAs to the mitochondria, which in yeast are anyway not competent for β-oxidation. The situation in mammals is different, and it has been reported that a carnitine-independent pathway [199] for transporting acyl-CoAs from peroxisomes to mitochondria might also exist in addition to the carnitine octanoyl-transferase-dependent route. In the former route, CoA esters are first hydrolyzed within the peroxisomes, and the free fatty acids are transported to the mitochondria where they are reactivated back to their CoA esters. Hydrolysis of acyl-CoAs within the peroxisomes is catalyzed either by acyl-CoA thioesterases or by hydrolases [200]. Additional roles played by thioesterases potentially also include the modulation of cellular levels of acyl-CoAs versus free CoA [201]. Thioesterases might also be required to prevent CoA sequestration by intermediates that can be further β-oxidized only very slowly or not at all.

The role of S. cerevisiae Pte1p/Tes1p in β-oxidation is not well understood, particularly since a strain devoid of thioesterase is essentially symptomless for growth on oleic acid medium [188, 202]. In mammalian cells, over-expression of thioesterase activity leads to a block in the carbon flux through this process [189]. Since the preference of Pte1p/Tes1p was highest towards medium-chain fatty acids (acetyl-CoA was not a substrate for Pte1p/Tes1p [202]), this fact largely excludes a role for Pte1p/Tes1p in the export of acetyl units. It is not entirely impossible that a mammalian-like route for transporting acetyl-CoAs that are first hydrolyzed intraperoxisomally to acetate might be accomplished in S. cerevisiae by engaging a hitherto unidentified peroxisomal hydrolase with an appropriate substrate specificity. Once in the cytosol, acetate would be activated by acetyl-CoA synthetase ACS1 prior to further processing.

5.3 Diversion of acetyl-CoA for biosynthesis: the glyoxylate cycle and its compartmentalization

Although a separate subcellular compartment is devoted to β-oxidation, fatty acid degradation should not be seen to exist in isolation from the rest of cellular needs for energy and building blocks. Indeed, yeast can grow on fatty acids as a sole carbon and energy source specifically because of their ability not only to direct acetyl-CoA to the mitochondria for further oxidation and energy release, but also towards biosynthesis via the glyoxylate cycle.

The glyoxylate cycle represents an abridged version of the TCA cycle in which the two decarboxylation steps are bypassed. This enables yeast cells to form C4 carbon skeletons from acetyl-CoA that can then be diverted to the production of more complex carbohydrates via gluconeogenesis. Due in part to the multiplicity of isoenzymes potentially representing those involved in the glyoxylate cycle (see below), the subcellular site of this anabolic process has long remained an open issue.

The glyoxylate cycle in S. cerevisiae is composed of five enzyme activities, some of which, as mentioned above, are represented by more than one enzyme (Table 1). These activities include isocitrate lyase Icl1p; malate synthase Mls1p and Dal7p; malate dehydrogenase Mdh1p, Mdh2p, and Mdh3p; citrate synthase Cit1p, Cit2p, and Cit3p; and aconitase Aco1p [57, 183185, 203208]. Isocitrate lyase and malate synthase are unique to the glyoxylate cycle, whereas some of the remaining listed enzymes are shared with the TCA cycle, e.g. Mdh1p, Cit1p, and Aco1p, all three of which are mitochondrial proteins.

The data on the location of the glyoxylate cycle are not clear cut. Icl1p is not a peroxisomal protein, unlike Mdh3p and Cit2p, which terminate with the tripeptide SKL representing a PTS1 [35, 209, 210]. Moreover, both malate synthases Mls1p and Dal7p also end with SKL, albeit only Mls1p is thought to be involved in the glyoxylate cycle, since the DAL7 gene is transcriptionally quiescent in cells grown on non-fermentable carbon sources and is dispensable under these medium conditions [204].

Adding to this confusion is the observation that it is not peroxisomal Mdh3p that represents the glyoxylate cycle malate dehydrogenase [157] but rather cytosolic Mdh2p [206], and that peroxisomal Cit2p is not required for yeast growth on acetate [57]. Moreover, pex mutants devoid of functional peroxisomes grow efficiently on ethanol and acetate [211], and pex diploid cells undergo abundant meiosis and sporulation on acetate medium [212], for which a functional glyoxylate cycle is imperative [213].

To refine the location of the glyoxylate cycle in wild-type cells, studies were conducted on Mls1p [214], which is essential for cell growth on non-fermentable carbon sources [204]. It transpired from these studies that although Mls1p is extended by SKL, and enters peroxisomes in cells supplied with oleic acid, the protein remains cytosolic in cells grown on ethanol. It should be noted that peroxisomes in ethanol-grown cells are both prevalent enough and import-competent to accommodate most of the peroxisomal Cta1p that is produced fairly extensively on non-fermentable carbon sources [214]. When the C-terminal SKL tripeptide was removed from Mls1p, rendering the truncated protein unable to enter peroxisomes, cells could continue to grow on oleic acid [214]. Only very careful analysis revealed that such cells are less efficient than those expressing wild-type Mls1p at utilizing oleic acid.

Although some uncertainties still exist, the overriding conviction is that despite appearing to meander into the peroxisomal matrix, strictly speaking the glyoxylate cycle is a cytosolic process in wild-type cells with normal peroxisome biogenesis [214]. The reason for the advantage of a peroxisomal Mls1p to cells grown on oleic acid is not known. If indeed acetyl-CoA generated by β-oxidation reacts peroxisomally in an Mls1p-dependent manner with glyoxylate produced cytosolically by Icl1p to yield malate, then there must be a way for glyoxylate to be able to reach the peroxisomes.

6 Regulation of gene expression

6.1 Pip2p–Oaf1p and ORE

As mentioned previously, when S. cerevisiae cells are supplied with fatty acids as the sole carbon source, this causes them to increase dramatically the number and size of their peroxisomes as well as to induce the transcription of genes encoding proteins involved in fatty acid β-oxidation by about 10-fold [1]. This form of response, dubbed oleic acid induction, is mediated to a very large extent, albeit not exclusively, by the Pip2p–Oaf1p transcription factor [178, 215217], which binds to OREs in the promoter of target genes [123, 177]. Cells devoid of either component of this transcription factor fail to degrade oleic acid or expand their peroxisomal compartment. The ORE consensus is currently viewed as two inverted CGG triplets spaced by 14 (formerly 15) to 18 intervening nucleotides (N), i.e. CGGN3TNAN8–12CCG [218]. In almost all of the known OREs at least one halfsite contains conserved A and T residues (CGGN3TNA) that are important for the function of this element [123, 177]. Such halfsites can confer limited transcriptional activation on a basal reporter gene, and weakly bind Pip2p–Oaf1p in vitro.

6.2 Adr1p and UAS1

It has been known for some time that adr1Δ mutant cells are unable to break down oleic acid or expand their peroxisomal compartment [219]. Adr1p was originally identified as a regulator of the alcohol dehydrogenase gene ADH2 [220]. It binds to the consensus sequence CYCCRDN4–36HYGGRG, termed upstream activating sequence 1 (UAS1) [221, 222]. Adr1p regulates the transcription of the CTA1 gene by binding to UAS1CTA1 [219]; however, the influence of Adr1p on the transcription of genes encoding peroxisomal β-oxidation enzymes, such as Mfe2p/Fox2p and Pot1p/Fox3p, was reported to be less pronounced and possibly indirect [219, 223]. Renewed interest in the role of Adr1p in peroxisome function arose by the finding that Adr1p regulates the peroxisomal 2,4-dienoyl-CoA reductase gene SPS19 by binding to UAS1SPS19 [224]. However, this still did not explain the precise role of Adr1p in regulating either the carbon flux through the β-oxidation spiral or the process of peroxisome proliferation, because neither CTA1 nor SPS19 is essential for growth on oleic acid.

Clues to the causes of the dual β-oxidation and peroxisome proliferation phenotype of adr1Δ mutant cells came with the discovery of a canonical UAS1 in the promoter of POX1/FOX1 and from the fact that also the PEX11 promoter contains UAS1-like sequences [225]. The roles of these two genes in β-oxidation and peroxisome proliferation have been discussed in detail previously. It turns out that UAS1POX1 is able to bind Adr1p in vitro. On the other hand, demonstration of a direct role for Adr1p in regulating PEX11 proved elusive. The issue of the PEX11 promoter sequence binding Adr1p notwithstanding, at the transcriptional level, both POX1/FOX1 and PEX11 are as tightly regulated by Adr1p as they are by Pip2p–Oaf1p [225]. This strict dependence of the β-oxidation spiral and peroxisome proliferation on Adr1p explains why adr1Δ mutants fail to degrade oleic acid or to expand their peroxisomal compartment.

Another novel aspect relating to gene regulation under oleic acid medium conditions was the discovery of synergy between the Pip2p–Oaf1p and Adr1p transcription factors. The question of the significance of overlapping UAS1/ORE elements prevalent among oleic acid-inducible genes was addressed using a basal CYC1-lacZ reporter gene driven by promoter elements from the CTA1 gene [225]. The combined potential for transcriptional activation of the CTA1 ORE or UAS1CTA1 in mutual isolation was about 2.5-fold less than that conferred on the basal promoter by the overlapping arrangement found in the native promoter [225]. ORE-dependent transcriptional activation is also influenced by other factors [226, 227], but less is known about them. The issue of how ORE-regulated genes are repressed in cells grown on glucose medium and derepressed in the presence of non-fermentable carbon sources is beyond the scope of this review [178, 215217].

6.3 Characterization of the fatty acid signal

Not much is known about the identity of the signal eliciting the response in yeast to the presence of fatty acids in the medium. One approach, which has been used rather recurringly to try and gain insight into the source of this signal, is to compare transcriptional activation in cells grown on different fatty acid substrates. This type of investigation disclosed that induction of peroxisomal functions is not restricted to just a single or only a few selected types of fatty acids. In fact, peroxisomal enzymes are likely to be induced en bloc in the presence of a variety of fatty acids, irrespective of the role single enzymes play in the metabolism of the fatty acid supplied in the medium.

For example, lauric acid does not preferentially activate genes such as ANT1 that are required for utilizing MCFAs [218]. Similar observations were made with Sps19p. Despite being solely required for breaking down unsaturated fatty acids with cis double bonds at even-numbered positions, such as petroselinic acid, the corresponding SPS19 gene is transcriptionally activated by oleic acid (among other fatty acids), which actually contains a cis double bond at an odd-numbered position [142, 228]. The widely accepted model is that Pip2p–Oaf1p is the exclusive sensor for a variety of fatty acids. However, this raises the issue of whether fatty acids from different classes are first converted to a common compound prior to activating Pip2p–Oaf1p. A reasonably straightforward concept is that were this compound to represent a fatty acid breakdown intermediate, then impairment of β-oxidation or of peroxisomal structures would probably prevent this putative metabolite from being formed, thereby obstructing subsequent induction of OREs. However, expression of an oleic acid-inducible lacZ reporter gene was not impaired in mutant cells lacking either Pox1p/Fox1p, which are blocked in the first and rate-limiting step of β-oxidation [42], or various peroxins, which do not assemble functional peroxisomes. It is therefore unlikely that the signalling molecule originates from the β-oxidation of fatty acids [218]. Since de novo protein synthesis is not required for triggering oleic acid-dependent transcriptional up-regulation [218], it is an attractive idea that activation of Pip2p–Oaf1p is accomplished by the direct binding of a variety of fatty acids. This scenario is reminiscent of the situation in higher eukaryotes, where peroxisome proliferator-activated receptors are capable of binding a number of lipids, which in turn lead to transcriptional responses serving to achieve lipid homeostasis [229234].

6.4 Genome analysis, SAGE, transcriptome profiling

The publication of the complete genomic sequence of S. cerevisiae in 1996 made this organism amenable for genome-wide analyses. The sequences of complete genomes are routinely analyzed in silico for homologues of proteins from other species [176]. This allows researchers to formulate rapidly an opinion regarding the possibility for the existence of certain biochemical processes in the organism under investigation. However, for such in silico searches to be valid, they must be substantiated experimentally. An example of the shortcomings of such searches with respect to fatty acid oxidation is the fruitless pursuit of the key enzyme involved in 3-methyl-branched-chain fatty acid degradation, phytanoyl-CoA hydroxylase. Despite containing an open reading frame with similarity to the enzyme catalyzing the subsequent step in this pathway, 2-hydroxyphytanoyl-CoA lyase [235], none of the laboratories engaged in studying this form of degradation could demonstrate that wild-type cells were able to grow on phytanic acid. It now appears as if the hydroxylase is absent from S. cerevisiae, and that the pathway for degrading 3-methyl-branched-chain fatty acids probably does not exist in yeast. Another attractive facet offered by analyzing whole genome sequences is the possibility to establish the actual members of a protein family. For instance, the S. cerevisiae genome was found to contain three members of the hydratase/isomerase family which could then be specifically analyzed for function in β-oxidation (see Section 4).

In addition to analyzing open reading frames, also promoter regions can be scrutinized for the occurrence of transcription factor binding sites on a genome-wide basis. This strategy was employed to identify novel ORE-regulated genes [131, 143, 151, 202]. The PatMatch program has since been made available at SGD for this purpose (http://genome-www.stanford.edu/Sacch3D/patmatch.html).

However, as in the previous case, it is necessary to confirm the data obtained by experimental means, since these types of searches are usually conducted on the basis of rather loose definitions for element consensus. In reference to the ORE consensus, it could turn out to be premature to claim that expression of 14 members of the mitochondrial carrier family is under ORE control solely because of the presence of ORE-like sequences in their promoters [58]. The accuracy of predictions regarding ORE-regulated genes would increase by combining promoter analyses with searches for potential PTSs in the corresponding gene products [143, 202]. Having said that, there are of course cases of peroxisomal proteins lacking obvious PTSs, e.g. Pox1p/Fox1p, and investigations based on Northern blot analysis reveal that not only genes encoding peroxisomal matrix proteins are induced in cells grown on oleic acid medium but also PMPs with uncharacterized mPTSs and even a few mitochondrial proteins such as Cit1p and Crc1p [151].

Novel experimental approaches aimed at identifying oleic acid-inducible genes on a genome-wide scale include serial analysis of gene expression (SAGE) and microarray-based transcriptome profiling. Both methods rely on the proportional amplification of the mRNA pool present in the cell. Changes in the overall composition of this pool under varying conditions can then be analyzed and interpreted. SAGE involves a very large number of tags that have to be sequenced before a representative subset is obtained to give a reliable overview [236]. The principal feasibility of this approach in reference to oleic acid induction was demonstrated by Kal et al. [237], however, the overall picture remained fragmentary since several known oleic acid-inducible genes such as ECI1 [131, 132, 151] and DCI1 [149, 150, 151] failed to come up in the search. An interesting feature to emerge from SAGE of oleic acid-treated pip2Δoaf1Δ mutant cells was the up-regulation of a number of known stress-induced genes. This implies that failure to utilize oleic acid is a stressful condition for which the cell tries to compensate.

This activation of a transient oxidative stress response is also observed in wild-type cells shifted to oleic acid medium from a glucose-limited chemostat [238]. Only after the stress response had terminated were cells in a position to initiate oleic acid induction. This latter study monitored the alterations in mRNA abundance using DNA microarrays, thereby enabling the identification of most of the previously reported oleic acid-inducible genes. In total, 269 out of 6013 genes were significantly regulated, but only 40 could be grouped in the cluster of oleic acid-inducible genes. When the raw data were analyzed with the REDUCE algorithm, which searches for common regulatory elements in the promoters of co-expressed genes, a consensus sequence (CGGN17CCG) that abides to that of the ORE was predicted for the cluster of oleic acid-induced genes [238].

Using a similar approach, the kinetics of oleic acid induction was studied using derepressed cultures of a wild-type diploid strain [239]. This led to a final list of 217 candidate oleic acid-inducible genes, including many PEX genes but also a number of genes whose products are not involved in carbon source metabolism [238]. It is also worth noting that the use of diploid strains probably introduced drawbacks associated with starving cells being synchronized for meiosis and sporulation. These concerns notwithstanding, the data collected to date allow us to look at a relatively complete set of genes that are highly induced in cells grown on oleic acid medium. On the other hand, genes that are only moderately induced by oleic acid, say by a factor of 2–3, such as ANT1 (YPR128c), appear to evade detection in such genome-wide studies. It seems as if regulation of moderately inducible genes is best investigated by gene expression studies dedicated to individual open reading frames, meaning that there is still some way to go before a truly global picture of oleic acid-induced genes emerges.

7 Yeast as a test organism

The strength of S. cerevisiae as a test organism for examining the biology of peroxisomes as well as the metabolic pathways contained therein has been demonstrated in numerous investigations. One of the milestones of research on peroxisomes was the isolation of a collection of S. cerevisiae mutants that were unable to grow on oleic acid as the sole carbon source [211]. This paved the way for the identification of a number of novel peroxins and other gene products involved in peroxisomal lipid metabolism. Thereafter, mammalian counterparts could be identified for most of these proteins by homology probing. Since many of the human peroxisomal proteins can be linked to inborn errors of metabolism, such mutant yeast cells have also served as unique models for cellular pathophysiology studies, thereby providing an opportunity for testing the functional significance of protein variants as causative agents of metabolic diseases.

In addition, heterologous expression of mammalian genes encoding peroxisomal proteins in S. cerevisiae can be used to replace the function of endogenous yeast genes, allowing engineered proteins to be tested in vivo [101]. Characterization of Mfe2p/Fox2p as an enzyme catalyzing the (3R)-specific route of hydration and dehydrogenation of β-oxidation ruled out the previous assumption that Mfe2p/Fox2p is a trifunctional enzyme having the activities of 2-enoyl-CoA hydratase 1, (3S)-specific 3-hydroxyacyl-CoA dehydrogenase, and 3-hydroxyacyl-CoA epimerase, whose latter activity was postulated to convert (3R)-hydroxyacyl-CoA to (3S)-hydroxyacyl-CoA and vice versa [96]. However, it was possible to complement yeast mutants devoid of Mfe2p/Fox2p by expressing the rat peroxisomal MFE1 that catalyzes the same reactions via (3S)-hydroxy intermediates [94]. This implies that both stereochemical alternatives allow yeast cells to grow on oleic acid in vivo. The biological perspective of this experiment is intriguing because once metabolism is established, the stereochemical course of the reactions involved is generally highly conserved. It is noteworthy that the mammalian MFE1 also catalyzes hydratation of short-chain 2-enoyl-CoA substrates [240], whereas the activity of the mammalian MFE2 drops rapidly with substrates shorter than C8.

As mentioned previously, to reposition cis double bonds in unsaturated fatty acids prior to β-oxidation, both the mitochondrial and peroxisomal compartments engage a comparable set of reaction steps that are executed by β-oxidation auxiliary enzymes [241, 242]. Herein lies the dispute over the functionality of auxiliary enzymes in β-oxidation, for the settlement of which yeast was mobilized. Human 2,4-dienoyl-CoA reductase [243] is a mitochondrial protein thought to be an auxiliary enzyme involved in the β-oxidation of unsaturated fatty acids; however, its function during this process was first demonstrated in vivo using a yeast mutant devoid of the peroxisomal 2,4-dienoyl-CoA reductase Sps19p, as explained below.

Investigations showed that mutant sps19Δ cells expressing human 2,4-dienoyl-CoA reductase ending with the native C-terminus did not grow on petroselinic acid medium (for which Sps19p is essential), but when the protein was altered to terminate with a PTS tripeptide SKL, wild-type levels of growth were restored [244]. We are not aware of other exclusively mitochondrial β-oxidation enzymes that have been made to function in yeast peroxisomes. Up to this point it was at least conceivable that, were the accumulated C16 intermediate of petroselinic acid to leak out of the peroxisomes, the SKL-less human 2,4-dienoyl-CoA reductase, in cooperation with cytosolic NADPH generation, could then restore growth of the sps19Δ strain from outside the compartment. Following the metabolism of the Δ2,4-conjugated double bonds in the cytosol, the corresponding Δ3-enoyl-CoA intermediate could have re-entered the peroxisomes to re-establish the carbon flux through β-oxidation, since such activated LCFAs have previously been shown to enter yeast peroxisomes from the cytosol [245]. Based on the cited work [244] performed in yeast, it could be shown that C16 fatty dienoyl-CoAs do not permeate out and subsequently back into this compartment in sufficient quantities to support growth of a mutant expressing the missing peroxisomal activity in a mislocalized form. Hence, beyond simply tricking yeast cells into processing human 2,4-dienoyl-CoA reductase as a peroxisomal protein, this work helped to underscore a fairly fundamental issue regarding peroxisome membrane permeability.

Use was also made of an eci1Δ strain devoid of peroxisomal Δ32-enoyl-CoA isomerase to examine in vivo the corresponding isomerase activity of the above-mentioned rat peroxisomal MFE1 [246]. Mutant eci1Δ cells expressing this protein from a plasmid could grow on oleic acid medium, thereby demonstrating that the in vitro Δ32-enoyl-CoA isomerase activity previously measured in this rat protein was functional in vivo [131]. A further peroxisomal Δ32-enoyl-CoA isomerase has since been identified in rat liver cells [139], and there is evidence for a third rat peroxisomal Δ32-enoyl-CoA isomerase, monofunctional ECI.

Rat monofunctional ECI contains a mitochondrial leader sequence, and has been shown to be predominantly mitochondrial [135]. It lacks an obvious PTS. When monofunctional ECI was expressed in a yeast eci1Δ strain, this re-established growth on oleic acid medium irrespective of whether an SKL extension was appended to its C-terminus. Since it was shown that Δ2,4 double bonds could not be metabolized extra-peroxisomally (see above) to restore growth of the sps19Δ strain, monofunctional ECI very likely acted on the Δ3 unsaturated metabolite of oleic acid by replacing the missing activity of the mutant from within the peroxisomes. Immunoblotting of fractionated yeast cells expressing monofunctional ECI in combination with electron microscopy supported the proposal that the protein additionally functions in peroxisomes [244]. In this respect the situation resembles a number of other mammalian fatty acid-metabolizing enzymes, which show both mitochondrial and peroxisomal localization: Δ3,52,4-dienoyl-CoA isomerase [164], α-methylacyl-CoA racemase [247], and the aforementioned mammalian Δ32-enoyl-CoA isomerase [139].

8 Future aspects

S. cerevisiae is one of the most widely used model organisms for developing new tools and carrying out genome-wide analysis of gene functions by high-throughput techniques. The data obtained by these approaches are indispensable for the understanding of protein–protein interactions, the establishment of cellular regulatory networks, and the identification of functions of unknown open reading frames in the genome. Pioneering work has often been completed via international cooperation, but further development of these robust techniques so as to enable their use by separate research laboratories remains a challenge for the future.

One of the goals of future research will be to understand how proteins and other cofactors associated with fatty acid degradation are transported into the peroxisomes. A key issue will be the identification of additional genes and their products, other than ANT1, PAT1/PXA2, and PAT2/PXA1, that act at the level of peroxisomal membranes, as well as the potential reconstruction of transport processes under in vitro conditions. In fact, many peroxisomal transporters remain to be identified, including those postulated to be involved in the shuttling of metabolites associated with maintaining the intraperoxisomal pools of NADPH and NADH. Analysis of the protein content of the S. cerevisiae peroxisomal membrane by mass spectrometry [190] could help reveal such novel proteins. In addition, screening available collections of S. cerevisiae knockout strains for mutants defective in growth on fatty acids might also lead to the identification of novel peroxisomal carriers. However, it is likely that dually localized carriers exist, which would exhibit a phenotype that is less specific than that anticipated for a β-oxidation mutant, since in this latter case, cell growth would probably also be affected on non-fermentable carbon sources other than fatty acids. It is also possible that cycling metabolites do not require transporters altogether.

The fact that S. cerevisiae cells contain an expandable peroxisomal compartment under fatty acid medium conditions has been well demonstrated by now but the observation that peroxisomes are also packaged into yeast spores formed under starvation conditions [212, 248] has drawn much less attention. The importance of the demonstration of meiotic inheritance of peroxisomes is that at all times in their life cycle, wild-type yeast cells harbor peroxisomes. Future investigations into the mechanism ensuring permanent peroxisome presence in cells could shed important new light on the biogenesis of this organelle, and might result in the identification of novel peroxins.

Humans have been using S. cerevisiae cells as biological agents in food processing for thousands of years; the making of wine and (un)leavened bread is described in the Bible. However, the more modern techniques of molecular biology and genetics have made yeast cells amenable to metabolic engineering aiming to exploit them for the generation of novel compounds. Concerning lipid-based metabolites, this is exemplified by the production of dicarboxylic acid [249, 250] or by manipulating lipid-metabolizing enzymes and changing β-oxidation intermediates into polyhydroxyalkanoates in yeast [152]. A crucial factor for structure–function studies is easy access to proteins of interest. When studying peroxisomal proteins from higher eukaryotes, yeast cells provide a convenient host because heterologously expressed proteins are often targeted to the right subcellular compartment and a high expression level can be achieved at a low cost.


We dedicate this review to the memory of our colleague, Professor Helmut Ruis (who died on September 1, 2001). We were not able to cite all of the contributions made to the field of peroxisomal fatty acid degradation in yeast, and apologize to those authors whose publications might have been omitted. Our original research work was supported in part by grants from the Academy of Finland, Sigrid Juselius Foundation, and the Fonds zur Förderung der wissenschaftlichen Forschung (FWF), Vienna, Austria. We thank Dr. Vasily Antonenkov for critically reading the manuscript, Dr. Christoph Schüller for helpful discussions on yeast catalases, and MA Virpi Hannus for her skilful help in the preparation of the manuscript.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
  34. [34].
  35. [35].
  36. [36].
  37. [37].
  38. [38].
  39. [39].
  40. [40].
  41. [41].
  42. [42].
  43. [43].
  44. [44].
  45. [45].
  46. [46].
  47. [47].
  48. [48].
  49. [49].
  50. [50].
  51. [51].
  52. [52].
  53. [53].
  54. [54].
  55. [55].
  56. [56].
  57. [57].
  58. [58].
  59. [59].
  60. [60].
  61. [61].
  62. [62].
  63. [63].
  64. [64].
  65. [65].
  66. [66].
  67. [67].
  68. [68].
  69. [69].
  70. [70].
  71. [71].
  72. [72].
  73. [73].
  74. [74].
  75. [75].
  76. [76].
  77. [77].
  78. [78].
  79. [79].
  80. [80].
  81. [81].
  82. [82].
  83. [83].
  84. [84].
  85. [85].
  86. [86].
  87. [87].
  88. [88].
  89. [89].
  90. [90].
  91. [91].
  92. [92].
  93. [93].
  94. [94].
  95. [95].
  96. [96].
  97. [97].
  98. [98].
  99. [99].
  100. [100].
  101. [101].
  102. [102].
  103. [103].
  104. [104].
  105. [105].
  106. [106].
  107. [107].
  108. [108].
  109. [109].
  110. [110].
  111. [111].
  112. [112].
  113. [113].
  114. [114].
  115. [115].
  116. [116].
  117. [117].
  118. [118].
  119. [119].
  120. [120].
  121. [121].
  122. [122].
  123. [123].
  124. [124].
  125. [125].
  126. [126].
  127. [127].
  128. [128].
  129. [129].
  130. [130].
  131. [131].
  132. [132].
  133. [133].
  134. [134].
  135. [135].
  136. [136].
  137. [137].
  138. [138].
  139. [139].
  140. [140].
  141. [141].
  142. [142].
  143. [143].
  144. [144].
  145. [145].
  146. [146].
  147. [147].
  148. [148].
  149. [149].
  150. [150].
  151. [151].
  152. [152].
  153. [153].
  154. [154].
  155. [155].
  156. [156].
  157. [157].
  158. [158].
  159. [159].
  160. [160].
  161. [161].
  162. [162].
  163. [163].
  164. [164].
  165. [165].
  166. [166].
  167. [167].
  168. [168].
  169. [169].
  170. [170].
  171. [171].
  172. [172].
  173. [173].
  174. [174].
  175. [175].
  176. [176].
  177. [177].
  178. [178].
  179. [179].
  180. [180].
  181. [181].
  182. [182].
  183. [183].
  184. [184].
  185. [185].
  186. [186].
  187. [187].
  188. [188].
  189. [189].
  190. [190].
  191. [191].
  192. [192].
  193. [193].
  194. [194].
  195. [195].
  196. [196].
  197. [197].
  198. [198].
  199. [199].
  200. [200].
  201. [201].
  202. [202].
  203. [203].
  204. [204].
  205. [205].
  206. [206].
  207. [207].
  208. [208].
  209. [209].
  210. [210].
  211. [211].
  212. [212].
  213. [213].
  214. [214].
  215. [215].
  216. [216].
  217. [217].
  218. [218].
  219. [219].
  220. [220].
  221. [221].
  222. [222].
  223. [223].
  224. [224].
  225. [225].
  226. [226].
  227. [227].
  228. [228].
  229. [229].
  230. [230].
  231. [231].
  232. [232].
  233. [233].
  234. [234].
  235. [235].
  236. [236].
  237. [237].
  238. [238].
  239. [239].
  240. [240].
  241. [241].
  242. [242].
  243. [243].
  244. [244].
  245. [245].
  246. [246].
  247. [247].
  248. [248].
  249. [249].
  250. [250].
  251. [251].
  252. [252].
  253. [253].
  254. [254].
  255. [255].
  256. [256].
  257. [257].
  258. [258].
  259. [259].
  260. [260].
View Abstract