Identification of peroxisomal targeting signals in cholesterol biosynthetic enzymes. AA-CoA thiolase, hmg-coa synthase, MPPD, and FPP synthase

At least three different subcellular compartments, including peroxisomes, are involved in cholesterol synthesis. The peroxisomal targeting signals for phosphomevalonate kinase and isopentenyl diphosphate isomerase have been identified. In the current study we identify the peroxisomal targeting signals required for four other enzymes of the cholesterol biosynthetic pathway: acetoacetyl-CoA (AA-CoA) thiolase, 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) synthase, mevalonate diphosphate decarboxylase (MPPD), and farnesyl diphosphate (FPP) synthase. Data are presented that demonstrate that mitochondrial AA-CoA thiolase contains both a mitochondrial targeting signal at the amino terminus and a peroxisomal targeting signal (PTS-1) at the carboxy terminus. We also analyze a new variation of PTS-2 sequences required to target HMG-CoA synthase and MPPD to peroxisomes. In addition, we show that FPP synthase import into peroxisomes is dependent on the PTS-2 receptor and identify at the amino terminus of the protein a 20-amino acid region that is required for the peroxisomal localization of the enzyme.These data provide further support for the conclusion that peroxisomes play a critical role in cholesterol biosynthesis.     

Role of peroxisomes in isoprenoid biosynthesis.

Our group and others have recently demonstrated that peroxisomes contain a number of enzymes involved in cholesterol biosynthesis that previously were considered to be cytosolic or located in the endoplasmic reticulum (ER). Peroxisomes have been shown to contain HMG-CoA reductase, mevalonate kinase, phosphomevalonate kinase, phosphomevalonate decarboxylase, isopentenyl diphosphate isomerase, and FPP synthase. Four of the five enzymes required for the conversion of mevalonate to FPP contain a conserved putative PTS1 or PTS2, supporting the concept of targeted transport into peroxisomes. To date, no information is available regarding the function of the peroxisomal HMG-CoA reductase in cholesterol/isoprenoid metabolism, and the structure of the peroxisomal HMG-CoA reductase has yet to be determined. We have identified a mammalian cell line that expresses only one HMG-CoA reductase protein, and which is localized exclusively to peroxisomes, to facilitate our studies on the function, regulation, and structure of the peroxisomal HMG-CoA reductase. This cell line was obtained by growing UT2 cells (which lack the ER HMG-CoA reductase) in the absence of mevalonate. The surviving cells exhibited a marked increase in a 90-kD HMG-CoA reductase that was localized exclusively to peroxisomes. The wild-type CHO cells contain two HMG-CoA reductase proteins, the well-characterized 97-kD protein localized in the ER, and a 90-kD protein localized in peroxisomes. We have also identified the mutations in the UT2 cells responsible for the lack of the 97-kD protein. In addition, peroxisomal-deficient Pex2 CHO cell mutants display reduced HMG-CoA reductase levels and have reduced rates of sterol and nonsterol biosynthesis. These data further support the proposal that peroxisomes play an essential role in isoprenoid biosynthesis.     

Localization of the pre-squalene segment of the isoprenoid biosynthetic pathway in mammalian peroxisomes

Previous studies have indicated that the early steps in the isoprenoid/cholesterol biosynthetic pathway occur in peroxisomes. However, the role of peroxisomes in cholesterol biosynthesis has recently been questioned in several reports concluding that three of the peroxisomal cholesterol biosynthetic enzymes, namely mevalonate kinase, phosphomevalonate kinase, and mevalonate diphosphate decarboxylase, do not localize to peroxisomes in human cells even though they contain consensus peroxisomal targeting signals. We re-investigated the subcellular localization of the cholesterol biosynthetic enzymes of the pre-squalene segment in human cells by using new stable isotopic techniques and data computations with isotopomer spectral analysis, in combination with immunofluorescence and cell permeabilization techniques. Our present findings clearly show and confirm previous studies that the pre-squalene segment of the cholesterol biosynthetic pathway is localized to peroxisomes. In addition, our data are consistent with the hypothesis that acetyl-CoA derived from peroxisomal β-oxidation of very long-chain fatty acids and medium-chain dicarboxylic acids is preferentially channeled to cholesterol synthesis inside the peroxisomes without mixing with the cytosolic acetyl-CoA pool.     

Phosphomevalonate kinase is a cytosolic protein in humans

In the past decade, a predominant peroxisomal localization has been reported for several enzymes functioning in the presqualene segment of the cholesterol/isoprenoid biosynthesis pathway. More recently, however, conflicting results have been reported raising doubts about the postulated role of peroxisomes in isoprenoid biosynthesis, at least in humans. In this study, we have determined the subcellular localization of human phosphomevalonate kinase using a variety of biochemical and microscopic techniques, including conventional subcellular fractionation studies, digitonin permeabilization studies, immunofluorescence, and immunoelectron microscopy. We found an exclusive cytosolic localization of both endogenously expressed human phosphomevalonate kinase (in human fibroblasts, human liver, and HEK293 cells) and overexpressed human phosphomevalonate kinase (in human fibroblasts, HEK293 cells, and CV1 cells). No indication of a peroxisomal localization was obtained. Our results do not support a central role of peroxisomes in isoprenoid biosynthesis.     

Central role of peroxisomes in isoprenoid biosynthesis

Peroxisomes contain enzymes catalyzing a number of indispensable metabolic functions mainly related to lipid metabolism. The importance of peroxisomes in man is stressed by the existence of genetic disorders in which the biogenesis of the organelle is defective, leading to complex developmental and metabolic phenotypes. The purpose of this review is to emphasize some of the recent findings related to the localization of cholesterol biosynthetic enzymes in peroxisomes and to discuss the impairment of cholesterol biosynthesis in peroxisomal deficiency diseases.     

Absence of functional peroxisomes does not lead to deficiency of enzymes involved in cholesterol biosynthesis.

To unravel the conflicting data concerning the dependence of human cholesterol biosynthesis on functional peroxisomes, we determined activities and levels of selected enzymes involved in cholesterol biosynthesis in livers of PEX5 knockout mice, a well-characterized model for human Zellweger syndrome. We found that all enzymes measured, including putative peroxisomal enzymes, are at least as active in the peroxisome-deficient Zellweger mice as in control mice, indicating that mislocalization of enzymes to the cytosol does not lead to decreased activity or degradation. Prompted by these results, we re-examined this aspect in human subjects by specific enzyme activity measurements and immunoblotting with highly specific antisera. Our results show that the previously reported deficiencies of mevalonate kinase and phosphomevalonate kinase activity in livers from human Zellweger patients reflect the bad condition of the livers, rather than mislocalization to the cytosol.Our data provide an explanation for the conflicting findings in the literature and show that great care should be taken in the interpretation of data obtained in postmortem material.     

Characterization of phosphomevalonate kinase: chromosomal localization, regulation, and subcellular targeting

Phosphomevalonate kinase catalyzes the conversion of mevalonate-5-phosphate to mevalonate-5-diphosphate and was originally believed to be a cytosolic enzyme. In this study we have localized the phosphomevalonate kinase gene to chromosome 1p13-1q22-23 and present a genomic map indicating that the gene spans more than 8.4 kb in the human genome. Furthermore, we show that message levels and enzyme activity of rat liver phosphomevalonate kinase are regulated in response to dietary sterol levels and that this regulation is coordinate with 3-hydroxy-3-methylglutaryl coenzyme A reductase, the rate-limiting enzyme of cholesterol biosynthesis. In addition, we demonstrate that phosphomevalonate kinase is a peroxisomal protein which requires the C-terminal peroxisomal targeting signal, Ser-Arg-Leu, for localization to the organelle.     

Differential binding of proteins to peroxisomes in rat hepatoma cells: unique association of enzymes involved in isoprenoid metabolism.

Farnesyl diphosphate synthase (FPPS: EC2.5.1.10), a key enzyme in isoprenoid metabolic pathways, catalyzes the synthesis of farnesyl diphosphate (FPP) an intermediate in the biosynthesis of both sterol and non-sterol isoprenoid end products. The localization of FPPS to peroxisomes has been reported (Krisans, S. K., J. Ericsson, P. A. Edwards, and G. A. Keller. 1994. J. Biol. Chem. 269: 14165;-14169). Using indirect immunofluorescence and immunoelectron microscopic techniques we show here that FPPS is localized predominantly in the peroxisomes of rat hepatoma H35 cells. However, the partial release of 60;-70% of cellular FPPS activity is observed by selective permeabilization of these cells with digitonin. Under these conditions, lactate dehydrogenase, a cytosolic enzyme, is completely released whereas catalase, a known peroxisomal enzyme, is fully retained. Digitonin treatment of H35 cells differentially affects the release of other peroxisomal enzymes involved in isoprenoid metabolism. For instance, mevalonate kinase and phosphomevalonate kinase are almost totally released (95% and 91%, respectively), whereas 3-hydroxy-3-methylglutaryl-CoA reductase is fully retained. Indirect immunoflourescence studies indicate that FPPS is localized in peroxisomes of Chinese hamster ovary (CHO)-K1 cells but is dispersed in the cytosol of ZR-82 cells, a mutant that lacks peroxisomes. Unlike in H35 cells, FPPS is completely released upon digitonin permeabilization of CHO-K1 and ZR-82 cells. In contrast, under the same permeabilization conditions, catalase is fully retained in CHO-K1 cells but completely released from ZR-82 cells. These studies indicate that FPPS and other enzymes in the isoprenoid biosynthetic pathways, involved in the formation of FPP, are differentially associated with peroxisomes and may easily diffuse to the cytosol. Based on these observations, the significance and a possible regulatory model in the formation of isoprenoid end-products are discussed.     

Mevalonate kinase is localized in rat liver peroxisomes.

Recent data suggest that rat liver peroxisomes play a critical role in cholesterol synthesis. Specifically, peroxisomes contain a number of enzymes required for cholesterol synthesis as well as sterol carrier protein-2. Furthermore, peroxisomes are involved in the in vitro synthesis of cholesterol from mevalonate and contain significant levels of apolipoprotein E, a major constituent of several classes of plasma lipoproteins. In this study we have investigated the subcellular localization of mevalonate kinase (EC 2.7.1.36; ATP:mevalonate-5-phosphotransferase). Mevalonate kinase is believed to be a cytosolic enzyme and catalyzes the phosphorylation of mevalonate to form mevalonate 5-phosphate. Mevalonate kinase has been purified from rat liver cytosol and a cDNA clone coding for rat mevalonate kinase has also been isolated and characterized. In this study, utilizing monoclonal antibodies made against the purified rat mevalonate kinase, we demonstrate the presence of mevalonate kinase in rat liver peroxisomes and in the cytosol. Each of these compartments contained a different form of the protein. The pI and the Mr of the peroxisomal protein is 6.2 and 42,000, respectively. The pI and Mr of the cytosolic protein is 6.9 and 40,000, respectively. The peroxisomal protein was also significantly induced by a number of different hypolipidemic drugs. In addition, we present evidence for the unexpected finding that the purified mevalonate kinase (isolated from the cytosol and assumed to be a cytosolic protein) is actually a peroxisomal protein.     

Peroxisomal cholesterol biosynthesis and Smith-Lemli-Opitz syndrome

Smith-Lemli-Opitz syndrome (SLOS), caused by 7-dehydrocholesterol-reductase (DHCR7) deficiency, shows variable severity independent of DHCR7 genotype. To test whether peroxisomes are involved in alternative cholesterol synthesis, we used [1-(14)C]C24:0 for peroxisomal beta-oxidation to generate [1-(14)C]acetyl-CoA as cholesterol precursor inside peroxisomes. The HMG-CoA reductase inhibitor lovastatin suppressed cholesterol synthesis from [2-(14)C]acetate and [1-(14)C]C8:0 but not from [1-(14)C]C24:0, implicating a peroxisomal, lovastatin-resistant HMG-CoA reductase. In SLOS fibroblasts lacking DHCR7 activity, no cholesterol was formed from [1-(14)C]C24:0-derived [1-(14)C]acetyl-CoA, indicating that the alternative peroxisomal pathway also requires this enzyme. Our results implicate peroxisomes in cholesterol biosynthesis but provide no link to phenotypic variation in SLOS.     

Enzymes of the mevalonate pathway of isoprenoid biosynthesis

The mevalonate pathway accounts for conversion of acetyl-CoA to isopentenyl 5-diphosphate, the versatile precursor of polyisoprenoid metabolites and natural products. The pathway functions in most eukaryotes, archaea, and some eubacteria. Only recently has much of the functional and structural basis for this metabolism been reported. The biosynthetic acetoacetyl-CoA thiolase and HMG-CoA synthase reactions rely on key amino acids that are different but are situated in active sites that are similar throughout the family of initial condensation enzymes. Both bacterial and animal HMG-CoA reductases have been extensively studied and the contrasts between these proteins and their interactions with statin inhibitors defined. The conversion of mevalonic acid to isopentenyl 5-diphosphate involves three ATP-dependent phosphorylation reactions. While bacterial enzymes responsible for these three reactions share a common protein fold, animal enzymes differ in this respect as the recently reported structure of human phosphomevalonate kinase demonstrates. There are significant contrasts between observations on metabolite inhibition of mevalonate phosphorylation in bacteria and animals. The structural basis for these contrasts has also recently been reported. Alternatives to the phosphomevalonate kinase and mevalonate diphosphate decarboxylase reactions may exist in archaea. Thus, new details regarding isopentenyl diphosphate synthesis from acetyl-CoA continue to emerge.     

Peroxisomal targeting signals in green algae

Peroxisomal enzymatic proteins contain targeting signals (PTS) to enable their import into peroxisomes. These targeting signals have been identified as PTS1 and PTS2 in mammalian, yeast, and higher plant cells; however, no PTS2-like amino acid sequences have been observed in enzymes from the genome database of Cyanidiochyzon merolae (Bangiophyceae), a primitive red algae. In studies on the evolution of PTS, it is important to know when their sequences came to be the peroxisomal targeting signals for all living organisms. To this end, we identified a number of genes in the genome database of the green algae Chlamydomonas reinhardtii, which contains amino acid sequences similar to those found in plant PTS. In order to determine whether these sequences function as PTS in green algae, we expressed modified green fluorescent proteins (GFP) fused to these putative PTS peptides under the cauliflower mosaic virus 35S promoter. To confirm whether granular structures containing GFP–PTS fusion proteins accumulated in the peroxisomes of Closterium ehrenbergii, we observed these cells after the peroxisomes were stained with 3, 3′-diaminobenzidine. Our results confirm that the GFP–PTS fusion proteins indeed accumulated in the peroxisomes of these green algae. These findings suggest that the peroxisomal transport system for PTS1 and PTS2 is conserved in green algal cells and that our fusion proteins can be used to visualize peroxisomes in live cells.     

Ether lipid biosynthesis: alkyl-dihydroxyacetonephosphate synthase protein deficiency leads to reduced dihydroxyacetonephosphate acyltransferase activities.

Recent studies have indicated that two peroxisomal enzymes involved in ether lipid synthesis, i.e., dihydroxyacetonephosphate acyltransferase and alkyl-dihydroxyacetonephosphate synthase, are directed to peroxisomes by different targeting signals, i.e., peroxisomal targeting signal type 1 and type 2, respectively. In this study, we describe a new human fibroblast cell line in which alkyl-dihydroxyacetonephosphate synthase was found to be deficient both at the level of enzyme activity and enzyme protein. At the cDNA level, a 128 base pair deletion was found leading to a premature stop. Remarkably, dihydroxyacetonephosphate acyltransferase activity was strongly reduced to a level comparable to the activities measured in fibroblasts from patients affected by the classical form of rhizomelic chondrodysplasia punctata (caused by a defect in peroxisomal targeting signal type 2 import). Dihydroxyacetonephosphate acyltransferase activity was completely normal in another alkyl-dihydroxyacetonephosphate synthase activity-deficient patient. Fibroblasts from this patient showed normal levels of the synthase protein and inactivity results from a point mutation leading to an amino acid substitution.These results strongly suggest that the activity of dihydroxyacetonephosphate acyltransferase is dependent on the presence of alkyl-dihydroxyacetonephosphate synthase protein. This interpretation implies that the deficiency of dihydroxyacetonephosphate acyltransferase (targeted by a peroxisomal targeting signal type 1) in the classic form of rhizomelic chondrodysplasia punctata is a consequence of the absence of the alkyl-dihydroxyacetonephosphate synthase protein (targeted by a peroxisomal targeting signal type 2).     

Investigation of cellular targeting of carotenoid pathway enzymes in Pichia pastoris

Cellular targeting of lycopene biosynthetic enzymes was investigated in Pichia pastoris X-33. Three lycopene pathway enzymes, CrtE, CrtB, and CrtI, were fused to fluorescent EGFPs with or without a peroxisomal targeting sequence (PTS1) and then expressed in P. pastoris. When P. pastoris was grown in YPD, the PTS1 fusion enzymes were found to be localized in peroxisomes, whereas the enzymes not fused with PTS1 were equally distributed throughout the entire cell. A similar targeting pattern was also observed in P. pastoris strains that were grown in peroxisome-proliferating medium, YPOT. Analysis of the fluorescent images of isolated peroxisomes showed that the PTS1 fused enzymes were dominantly present in peroxisomes whereas small amount of the enzymes not fused with PTS1 were non-specifically sent to peroxisomes. These results indicate that PTS1 specifically target lycopene pathway enzymes into peroxisomes and this targeting pathway was strong enough to overcome their inherent targeting program. In conclusion, we first showed that carotenogenic enzymes can be targeted into the specific cellular location of recombinant hosts and this targeting strategy can serve as the basis for the subsequent development of sophisticated pathway engineering in microorganisms.     

Purification of brain peroxisomes and localization of 3-hydroxy-3-methylglutaryl coenzyme A reductase.

At least three different subcellular compartments, including peroxisomes, are involved in cholesterol biosynthesis. Because proper CNS development depends on de novo cholesterol biosynthesis, peroxisomes must play a critical functional role in this process. Surprisingly, no information is available on the peroxisomal isoprenoid/cholesterol biosynthesis pathway in normal brain tissue or on the compartmentalization of isoprene metabolism in the CNS. This has been due mainly to the lack of a well-defined isolation procedure for brain tissue, and also to the presence of myelin in brain tissue, which results in significant contamination of subcellular fractions. As a first step in characterizing the peroxisomal isoprenoid pathway in the CNS, we have established a purification procedure to isolate peroxisomes and other cellular organelles from the brain stem, cerebellum and spinal cord of the mouse brain. We demonstrate by use of marker enzymes and immunoblotting with antibodies against organelle specific proteins that the isolated peroxisomes are highly purified and well separated from the ER and mitochondria, and are free of myelin contamination. The isolated peroxisomal fraction was purified at least 40-fold over the original homogenate. In addition, we show by analytical subcellular fractionation and immunoelectron microscopy that HMG-CoA reductase protein and activity are localized both in the ER and peroxisomes in the CNS.     

AtPex14p maintains peroxisomal functions by determining protein targeting to three kinds of plant peroxisomes

We previously isolated an Arabidopsis: peroxisome-deficient ped2 mutant by its resistance to 2,4-dichlorophenoxybutyric acid. Here, we describe the isolation of a gene responsible for this deficiency, called the PED2 gene, by positional cloning and confirmed its identity by complementation analysis. The amino acid sequence of the predicted protein product is similar to that of human Pex14p, which is a key component of the peroxisomal protein import machinery. Therefore, we decided to call it AT:Pex14p. Analyses of the ped2 mutant revealed that AT:Pex14p controls intracellular transport of both peroxisome targeting signal (PTS)1- and PTS2-containing proteins into three different types of peroxisomes, namely glyoxysomes, leaf peroxisomes and unspecialized peroxisomes. Mutation in the PED2 gene results in reduction of enzymes in all of these functionally differentiated peroxisomes. The reduction in these enzymes induces pleiotropic defects, such as fatty acid degradation, photorespiration and the morphology of peroxisomes. These data suggest that the AT:Pex14p has a common role in maintaining physiological functions of each of these three kinds of plant peroxisomes by determining peroxisomal protein targeting.     

Cycloheximide as a tool to investigate protein import in peroxisomes: a case study of the subcellular localization of isoprenoid biosynthetic enzymes

Cytosolic background fluorescence is often observed when native low-abundance peroxisomal proteins carrying a weak peroxisomal targeting sequence are expressed as fluorescent fusion protein using a strong constitutive promoter in transiently transformed plant cells. This cytosolic fluorescence usually comes from the strong expression of the low-abundance proteins exceeding the peroxisome import efficiency. This often results in a misinterpretation of the protein subcellular localization, as there is doubt as to whether proteins are dually targeted to the cytosol and peroxisome or are exclusively localized to peroxisomes. To circumvent this experimental difficulty, the protein peroxisome import study can be optimized by de novo protein synthesis inhibition in transiently transformed cells using the translation inhibitor cycloheximide. This approach was used here successfully for the study of the subcellular localization of distinct plant isoprenoid biosynthetic enzymes, allowing us to clearly demonstrate that 5-phosphomevalonate kinase, mevalonate 5-diphosphate decarboxylase and a short isoform of farnesyl diphosphate synthase from Catharanthus roseus are exclusively localized to peroxisomes.     

Novel targeting assay uncovers targeting information within peroxisomal ABC transporter Pxa1

The mechanism behind peroxisomal membrane protein targeting is still poorly understood, with only two yeast proteins believed to be involved and no consensus targeting sequence. Pex19 is thought to bind peroxisomal membrane proteins in the cytosol, and is subsequently recruited by Pex3 at the peroxisomal surface, followed by protein insertion via a mechanism that is as-yet-unknown. However, some peroxisomal membrane proteins still correctly sort in the absence of Pex3 or Pex19, suggesting that multiple sorting pathways exist. Here, we studied sorting of yeast peroxisomal ABC transporter Pxa1. Co-localisation analysis of Pxa1-GFP in a collection of 86 peroxisome-related deletion strains revealed that Pxa1 sorting requires Pex3 and Pex19, while none of the other 84 proteins tested were essential. To identify regions with peroxisomal targeting information in Pxa1, we developed a novel in vivo re-targeting assay, using a reporter consisting of the mitochondrial ABC transporter Mdl1 lacking its N-terminal mitochondrial targeting signal. Using this assay, we showed that the N-terminal 95 residues of Pxa1 are sufficient for retargeting this reporter to peroxisomes. Interestingly, truncated Pxa1 lacking residues 1–95 still localised to peroxisomes. This was confirmed via localisation of various Pxa1 truncation and deletion constructs. However, localisation of Pxa1 lacking residues 1–95 depended on the presence of its interaction partner Pxa2, indicating that this truncated protein does not contain a true targeting signal.     

Alkyl-dihydroxyacetonephosphate synthase

The initial steps of ether phospholipid biosynthesis take place in peroxisomes. Alkyl-dihydroxyacetonephosphate synthase, the peroxisomal enzyme that actually introduces the ether linkage, has been purified from guinea pig liver in this laboratory. With the amino acid sequences obtained from this protein, the authors were able to clone the cDNAs encoding this enzyme from both guinea pig and human liver. In both cases, the enzyme appears to be synthesized as a precursor protein with a N-terminal cleavable presequence containing a peroxisomal targeting signal (PTS) type 2. Levels of the enzyme protein were found to be strongly reduced in human fibroblasts derived from Zellweger syndrome and rhizomelic chondrodysplasia punctata patients. The molecular basis of an isolated alkyl-dihydroxyacetonephosphate synthase deficiency was resolved. A clone encoding a Caenorhabditis elegans homolog of the mammalian enzymes was characterized. In contrast to the mammalian enzymes, this C. elegans enzyme lacks a N-terminal PTS type 2 motif, but carries a C-terminal PTS type 1.     

Improvement of Aspergillus nidulans penicillin production by targeting AcvA to peroxisomes

Aspergillus nidulans is able to synthesize penicillin and serves as a model to study the regulation of its biosynthesis. Only three enzymes are required to form the beta lactam ring tripeptide, which is comprised of l-cysteine, l-valine and l-aminoadipic acid. Whereas two enzymes, AcvA and IpnA localize to the cytoplasm, AatA resides in peroxisomes. Here, we tested a novel strategy to improve penicillin production, namely the change of the residence of the enzymes involved in the biosynthesis. We tested if targeting of AcvA or IpnA (or both) to peroxisomes would increase the penicillin yield. Indeed, AcvA peroxisomal targeting led to a 3.2-fold increase. In contrast, targeting IpnA to peroxisomes caused a complete loss of penicillin production. Overexpression of acvA, ipnA or aatA resulted in 1.4, 2.8 and 3.1-fold more penicillin, respectively in comparison to wildtype. Simultaneous overexpression of all three enzymes resulted even in 6-fold more penicillin. Combination of acvA peroxisomal targeting and overexpression of the gene led to 5-fold increase of the penicillin titer. At last, the number of peroxisomes was increased through overexpression of pexK. A strain with the double number of peroxisomes produced 2.3 times more penicillin. These results show that penicillin production can be triggered at several levels of regulation, one of which is the subcellular localization of the enzymes.