Endoplasmic Reticulum-Associated Secretory Proteins Sec20p, Sec39p, and Dsl1p Are Involved in Peroxisome Biogenesis

Abstract

Two pathways have been identified for peroxisome formation: (i) growth and division and (ii) de novo synthesis. Recent experiments determined that peroxisomes originate at the endoplasmic reticulum (ER). Although many proteins have been implicated in the peroxisome biogenic program, no proteins in the eukaryotic secretory pathway have been identified as having roles in peroxisome formation. Using the yeast Saccharomyces cerevisiae regulatable Tet promoter Hughes clone collection, we found that repression of the ER-associated secretory proteins Sec20p and Sec39p resulted in mislocalization of the peroxisomal matrix protein chimera Pot1p-green fluorescent protein (GFP) to the cytosol. Likewise, the peroxisomal membrane protein chimera Pex3p-GFP localized to tubular-vesicular structures in cells suppressed for Sec20p, Sec39p, and Dsl1p, which form a complex at the ER. Loss of Sec39p attenuated formation of Pex3p-derived peroxisomal structures following galactose induction of Pex3p-GFP expression from the GAL1 promoter. Expression of Sec20p, Sec39p, and Dsl1p was moderately increased in yeast grown under conditions that proliferate peroxisomes, and Sec20p, Sec39p, and Dsl1p were found to cofractionate with peroxisomes and colocalize with Pex3p-monomeric red fluorescent protein under these conditions. Our results show that SEC20, SEC39, and DSL1 are essential secretory genes involved in the early stages of peroxisome assembly, and this work is the first to identify and characterize an ER-associated secretory machinery involved in peroxisome biogenesis.

Peroxisomes are ubiquitous multifunctional organelles limited by a single membrane and devoid of genetic material. The paracrystalline matrix of peroxisomes results from an extremely high concentration of proteins that carry out a number of metabolic processes, the most conserved among species being the β-oxidation of fatty acids and the neutralization of hydrogen peroxide (41). Peroxisomes have a striking capacity to alter their enzyme content according to the metabolic needs of the cell and to increase dramatically in number and size under various environmental conditions. The inability of peroxisomes to assemble correctly or react appropriately to the metabolic demands of the cell is detrimental. This biological requirement for peroxisomal function is evidenced by a group of genetic disorders collectively called the peroxisome biogenesis disorders (PBD). PBD-afflicted individuals fail to assemble functional peroxisomes, and the severity and lethality generally associated with the PBDs demonstrate the necessity for peroxisomes in normal human development and physiology (55).

Historically, peroxisomes have been classified with mitochondria and chloroplasts as semiautonomous organelles that derived from an endosymbiotic event during evolution (9). Characterization of the peroxisomal fission pathway supported an autonomous proliferation model of peroxisomes in which peroxisomes arose by growth and division of preexisting peroxisomes and were partitioned to and inherited by daughter cells (29, 30). Further evidence in support of the autonomous nature of peroxisomes came from the demonstration that peroxisomal proteins are synthesized on free polysomes and in the majority are imported directly to the peroxisome from the cytosol (15-17, 42). Unlike mitochondria and chloroplasts, peroxisomes were observed to form de novo in mutant cells that completely lacked any identifiable peroxisomal structures upon reintroduction of the wild-type version of the mutated gene (33, 46, 51-53). This observation challenged the concept of an endosymbiotic origin for peroxisomes and the view that peroxisomes relied solely on the autonomous partitioning of preexisting peroxisomes for their proliferation and inheritance.

The ability of peroxisomes to form de novo suggests that another organelle must provide peroxisomal membrane components. Previous morphological evidence showing peroxisomes in close association with, and as extensions of, the endoplasmic reticulum (ER) suggested a role for the ER as the donor compartment (18, 36, 57). Biochemical evidence for the ER origin of peroxisomes (60, 61) was controversial and suggested that a functional role for the ER in the de novo synthesis of peroxisomes was dependent on the experimental organism being studied (29). Alteration or loss of protein components of the classical secretory machinery generally had little or no effect on peroxisome biogenesis (37, 47, 53, 73). Recent findings with the yeast Saccharomyces cerevisiae have now provided unambiguous morphological and biochemical evidence that the ER is the site of de novo peroxisome biogenesis (21, 58). These studies took advantage of the requirement for the integral peroxisomal membrane protein Pex3p for the biogenesis of peroxisomes in peroxisome-deficient cells and conclusively showed that Pex3p targets to discrete ER-localized puncta, forming a dynamic ER subcompartment, en route to the peroxisome. Trafficking of Pex3p through the ER is not confined to cells lacking peroxisomes, since the de novo synthesis of peroxisomes from the ER was also found to occur in normal mammalian cells and to require Pex16p, an integral component of the peroxisomal membrane protein import machinery (26).

Given the recent incontrovertible evidence showing de novo peroxisome biogenesis from the ER, we sought to identify proteins involved in the exit of peroxisomes from the ER. Transcriptome profiling, organellar proteomics, comparative gene analysis, and screening of knockout libraries for nonessential genes of S. cerevisiae have so far failed to implicate components of the secretory machinery in the biogenesis of peroxisomes from the ER (44, 49, 50, 59, 69, 70, 74; R. J. Perry and R. A. Rachubinski, unpublished results). Therefore, we screened a library of S. cerevisiae essential genes placed under the control of a regulatable promoter to identify essential secretory (SEC) genes that might be required for the biogenesis of peroxisomes. From this screen we identified SEC20, SEC39, and DSL1 as essential SEC genes encoding ER-associated proteins involved in the early stages of peroxisome biogenesis and assembly.

MATERIALS AND METHODS

Strains and culture conditions.

The S. cerevisiae strains used in this study are listed in Table 1. Strains were cultured at 30°C unless indicated otherwise. The Tet promoter Hughes clone (THC) collection (Open Biosystems) was used to control the expression of essential genes from a regulatable Tet promoter (34). Repression of the TetO₇ promoter was achieved by addition of doxycycline to 10 μg/ml from a 10-mg/ml stock in water to the medium and incubation for 18 h or less. Cells incubated in the presence of doxycycline were seeded at an increased cell number to obtain comparable cell densities at the time of experimentation. Medium components were as follows: YEPD, 1% yeast extract, 2% peptone, 2% glucose; YEP2×D, 1% yeast extract, 2% peptone, 4% glucose; YEPR, 1% yeast extract, 2% peptone, 4% raffinose; YEPG, 1% yeast extract, 2% peptone, 2% galactose; SCIM1, 0.67% yeast nitrogen base without amino acids, 0.5% yeast extract, 0.5% peptone, 0.5% (wt/vol) Tween 40, 1× complete supplement mixture, 0.1% glucose, 0.15% (vol/vol) oleic acid; YPBO, 0.3% yeast extract, 0.5% peptone, 0.5% KH₂PO₄, 0.5% K₂HPO₄, 0.2% (wt/vol) Tween 40, 1% (wt/vol) oleic acid; and SCIM2, SCIM1 containing 0.3% glucose and 0.3% (vol/vol) oleic acid. Solid media were prepared by addition of 2% agar to liquid media.

TABLE 1.

Yeast strains used in this study

Strain	Genotype	Reference
BY4742	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0	19
pex3Δ	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 pex3::KanMX4	19
SEC20-GFP-PEX3-mRFP	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 sec20::SEC20-GFP(HIS5) pex3::PEX3-mRFP(URA3)	This study
SEC39-GFP-PEX3-mRFP	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 sec39::SEC39-GFP(HIS5) pex3::PEX3-mRFP(URA3)	This study
DSL1-GFP-PEX3-mRFP	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 dsl1::DSL1-GFP(HIS5) pex3::PEX3-mRFP(URA3)	This study
PEX3-GFP-RTN1-mRFP	MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 pex3::PEX3-GFP(HIS5) rtn1::RTN1-mRFP(URA3)	This study
R1158-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTa pot1::POT1-GFP(HIS5)	This study
THC-SEC1-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC1::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC7-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC7::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC11-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC11::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC12-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC12::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC13-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC13::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC14-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC14::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC17-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC17::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC18-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC18::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC20-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC20::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC21-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC21::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC26-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC26::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC27-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC27::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC32-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC32::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-COPI-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pCOPI::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC39-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC39::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC59-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC59::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC61-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC61::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC63-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC63::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
THC-SEC65-POT1-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC65::tet07-TATA(KanMX6) pot1::POT1-GFP(HIS5)	This study
R1158-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pex3::PEX3-GFP(HIS5)	This study
THC-SEC14-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC14::tet07-TATA(KanMX6) pex3::PEX3-GFP(HIS5)	This study
THC-SEC61-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC61::tet07-TATA(KanMX6) pex3::PEX3-GFP(HIS5)	This study
THC-SEC20-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC20::tet07-TATA(KanMX6) pex3::PEX3-GFP(HIS5)	This study
THC-SEC39-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC39::tet07-TATA(KanMX6) pex3::PEX3-GFP(HIS5)	This study
THC-DSL1-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pDSL1::tet07-TATA(KanMX6) pex3::PEX3-GFP(HIS5)	This study
THC-SEC39-pGAL1-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC39::tet07-TATA(NAT) pPEX3::pGAL1(KanMX6) pex3::PEX3-GFP(HIS5)	This study
THC-SEC61-pGAL1-PEX3-GFP	MATahis3Δ1 leu2Δ0 met15Δ0 URA3::CMV-tTA pSEC61::tet07-TATA(NAT) pPEX3::pGAL1(KanMX6) pex3::PEX3-GFP(HIS5)	This study

Plasmids and integrative transformation of yeast.

Plasmids and PCR-based integrative transformation used to genomically tag genes and introduce the GAL1 promoter (pGAL1) upstream of the PEX3 gene have been described previously (11, 58). POT1 and PEX3 were genomically tagged with sequences encoding fluorescent proteins to visualize peroxisomes, and genomic introduction of pGAL1 upstream of PEX3 was used to control the biogenesis of peroxisomes.

Induction of Pot1p-GFP expression.

THC strains expressing a fusion of POT1 and the green fluorescent protein (GFP) gene were grown overnight in YEPD and transferred to YEPD with or without 10 μg doxycycline/ml at a cell density to achieve mid-log growth following a further 18 h of incubation. An aliquot was taken at time zero, and the remaining culture was used for oleic acid induction of Pot1p-GFP synthesis by incubation in YPBO with or without 10 μg doxycycline/ml for 90 min. Cells were then washed extensively to remove free oleic acid, and images were acquired by epifluorescence microscopy.

Pulse-chase analysis of Pex3p-GFP transport between the ER and peroxisomes.

THC strains expressing PEX3-GFP under the control of pGAL1 were grown overnight in YEPD, transferred to YEPR at an optical density at 600 nm of 0.005, and incubated for an additional 16 to 18 h in YEPR to allow derepression of pGAL1. The culture was divided and primed in YEPR with or without 10 μg doxycycline/ml for the times indicated. Pulse expression of Pex3p-GFP was achieved by transferring cells to YEPG for 1 h, followed by a chase incubation of cells in YEP2×D for 4 h and 15 h to monitor the transport of Pex3p-GFP from the ER to peroxisomes. An aliquot was taken following each incubation step (prime, pulse, and chase), and cells were fixed in 3.7% formaldehyde and analyzed by epifluorescence microscopy.

Microscopy.

Images of live or fixed cells were captured using one of two methods. In the first, a Plan-apochromat 63×/1.4 NA oil differential interference contrast objective on an Axiovert 200 inverted microscope equipped with a LSM510 META confocal scanner (Carl Zeiss) was used. GFP was excited with a 488-nm laser and its emission collected with a 505-nm long-pass filter or, for colocalization analysis, a 505- to 530-nm band-pass filter, while monomeric red fluorescent protein (mRFP) was excited with a 543-nm laser and its emission collected using a 600-nm long-pass filter. In the second method, a Plan-APO 100×/1.4 NA oil objective on an IX81 inverted epifluorescence microscope (Olympus) equipped with a CoolSNAP HQ digital camera (Roper Scientific) and an ExFo X-Cite 120 PC fluorescent illumination system was used. Final images were prepared using Adobe Photoshop software. To prevent interference of internal structures captured in the transmission images, the internal structures were removed in Adobe Photoshop. For Fig. 2, 6B, 6C, and 6D, images were deconvolved using Huygens Professional software (Scientific Volume Imaging BV, The Netherlands). For this method, three-dimensional (3D) data sets were processed to remove blur through an iterative classic maximum-likelihood estimation algorithm and an experimentally derived point spread function. Imaris software (Bitplane) was subsequently used to prepare a maximum-intensity projection of the deconvolved 3D data set (for Fig. 2 and 6D) or single-plane images (for Fig. 6B and C) before final assembly in Adobe Photoshop. For Fig. 2, further image analysis was performed on deconvolved 3D data sets using the “Blend” viewing mode in Imaris (Bitplane). This generates a black-and-white image in which the fluorescent signal is standardized to minimize the differences between bright and dim fluorescent signals for evaluation of the entire fluorescent signal located within the cell. For electron microscopy, whole cells were processed as described previously (12).

FIG. 2.

The peroxisomal membrane protein Pex3p is mislocalized to the cytosol and undefined tubular-vesicular structures in cells repressed for SEC20, SEC39, or DSL1. THC-SEC strains with genomically integrated PEX3-GFP were incubated for 18 h in YEPD (−doxycycline). Repression of TetO₇ promoter-regulated genes was achieved by addition of doxycycline to a concentration of 10 μg/ml in YEPD (+doxycycline). Cells were fixed with formaldehyde, images were captured by confocal microscopy, and a maximum-intensity projection was created from a deconvolved 3D data set using Huygens software. Black-and-white images were generated using the “Blend” viewing mode in Imaris (Bitplane) to enhance the total fluorescent signal located within the cell compared to the maximum-intensity projection. Repression of SEC20, SEC39, and DSL1 resulted in redistribution of Pex3p-GFP to the cytosol and undefined tubular-vesicular structures (arrows). Bar, 5 μm.

FIG. 6.

Sec20p, Sec39p, and Dsl1p associate with peroxisomes. (A) A 20KgP fraction was prepared from wild-type BY4742 and mutant pex3Δ cells grown in oleic acid for 16 h for the isolation of peroxisomes by isopycnic gradient centrifugation on a discontinuous Nycodenz gradient as described in Materials and Methods. Equal volumes of each fraction were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted for the indicated proteins. P, peroxisome fractions; M, mitochondrial fractions. BY4742 cells coexpressing Pex3p-mRFP and Sec20p-GFP, Sec39p-GFP, or Dsl1p-GFP were grown in SCIM2 (B) for 4 h or in YEPD (C) for 16 h and fixed in 3.7% formaldehyde. Fluorescently tagged proteins were visualized for a single plane by confocal microscopy. Narrow arrowheads highlight areas of colocalization, and wide arrowheads highlight areas where Pex3p-mRFP is closely apposed to Sec20p-GFP, Sec39p-GFP, or Dsl1p-GFP. Bars, 5 μm. (D) BY4742 cells coexpressing Pex3p-GFP and Rtn1p-mRFP were grown in YEPD for 16 h. Cells were fixed with 3.7% formaldehyde, images were captured by confocal microscopy, and a maximum-intensity projection was created from a deconvolved 3D data set using Huygens software. Bar, 5 μm.

Subcellular fractionation and isolation of peroxisomes.

Subcellular fractionation and isolation of peroxisomes were performed essentially as described previously (70). Briefly, strains grown overnight in YEPD were transferred to SCIM2 and incubated for 16 h. Cells were harvested, and spheroplasts were prepared by digestion with Zymolyase 100T (2 mg/g cells). Spheroplasts were homogenized, and the homogenate was subjected to centrifugation to yield a postnuclear supernatant fraction. The postnuclear supernatant was further divided by differential centrifugation at 20,000 × g to yield supernatant and pellet (20KgP) fractions. Organelles in the 20KgP were separated by isopycnic gradient centrifugation using a discontinuous gradient consisting of 17, 25, 35, and 50% (wt/vol) Nycodenz. The gradient was subjected to centrifugation at 100,000 × g for 90 min in a VTi50 rotor (Beckman Coulter). Fractions of 2 ml were collected from the bottom of the gradient, and equal volumes from each gradient were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose, and immunoblotted for the indicated proteins.

Antibodies and immunoblotting.

Polyclonal antibodies to Sec20p and Dsl1p were raised in guinea pig. Antibodies to thiolase and Sdh2p have been previously described (10, 13). Rabbit polyclonal antibodies to Act1p, GFP, Sec39p, Dsl1p, and Sec61p were kind gifts from Gary Eitzen (University of Alberta, Canada), Luc Berthiaume (University of Alberta, Canada), Daniel Unger (University of York, United Kingdom), Hans Dieter Schmitt (Max Planck Institute, Germany), and Randy Schekman (University of California, Berkeley), respectively. Yeast whole-cell lysates were prepared by solubilizing cell pellets in 1.85 M sodium hydroxide-7.4% 2-mercaptoethanol. Proteins were precipitated from lysates by addition of trichloroacetic acid, and the resultant protein pellets were solubilized by successive equal-volume additions of magic A (1 M unbuffered Tris, 13% SDS) and magic B (30% glycerol, 200 mM dithiothreitol, 0.25% bromophenol blue) solutions. Equivalent amounts of protein were resolved by SDS-10% PAGE. Proteins were transferred to nitrocellulose for immunoblotting, and horseradish peroxidase-conjugated secondary antibodies were used to detect the primary antibodies by chemiluminescence.

RESULTS

Criteria for selecting essential secretory genes involved in peroxisome biogenesis.

Analysis of isogenic yeast gene knockout libraries has failed to identify a nonessential SEC gene involved in the biogenesis of peroxisomes from the ER (44, 49, 50, 59, 69, 70, 74; Perry and Rachubinski, unpublished results). To ascertain a role for an essential SEC gene in the ER-to-peroxisome biogenesis pathway, we employed the THC collection library, which allows for the conditional and efficient repression of essential S. cerevisiae genes (2, 3, 34). In this library, the endogenous promoter of an essential gene has been replaced by the regulatable TetO₇ promoter, allowing for repression of the essential gene by addition of doxycycline to the growth medium. Previous reports have shown that the concentrations of doxycycline required for gene repression have little effect on yeast physiology and essentially no effect on global gene expression (23, 39). Using the Saccharomyces Genome Database (http://www.yeastgenome.org), we selected essential SEC genes of unknown function or that had previously been determined to function in ER and/or Golgi apparatus transport. Using these criteria, we identified 19 SEC genes to screen for a role in peroxisome assembly ( Table 2).

TABLE 2.

SEC genes screened for defects in localization of Pot1p-GFP to peroxisomes

Gene name	Systematic name	Alternate name(s)	Description of product or functiona
SEC1	YDR164c		Localized to sites of secretion; dependent on SNARE function
SEC7	YDR170c		GEF involved in intra-Golgi and ER-to-Golgi transport
SEC11	YIR022w		18-kDa subunit of signal peptidase complex
SEC12	YNR026c	SED2	GEF involved in initiation of transport vesicle budding from ER
SEC13	YLR208w	ANU3	COPII complex
SEC14	YMR079w	PIT1	PtdIns/PtdCho transfer protein involved in export of vesicles from the Golgi apparatus; homolog of mammalian PITPs
SEC17	YBL050w		Required for transport of vesicles between ER and Golgi apparatus
SEC18	YBR080c	ANU4	Homolog of mammalian NSF; involved in ER-to-Golgi transport
SEC20	YDR498c		v-SNARE; Golgi-to-ER retrograde transport
SEC21	YNL287w		COPI complex
SEC26	YDR238c		COPI complex
SEC27	YGL137w		COPI complex
SEC32	YLR078c	BOS1	v-SNARE; ER-to-Golgi vesicular transport
SEC33	YDL145c	COPI, RET1, SOO1	COPI complex
SEC39	YLR440c		Unknown function
SEC59	YMR013c		Lipid synthesis; protein N glycosylation
SEC61	YLR378c		Translocon
SEC63	YOR254c	PTL1	Translocon
SEC65	YML105c		Signal recognition particle complex

^aGEF, guanine nucleotide exchange factor; PITPs, phosphatidylinositol transfer proteins; NSF, N-ethylmaleimide-sensitive factor.

Repression of SEC20 and SEC39 results in mislocalization of peroxisomal Pot1p to the cytosol.

Growth of yeast in oleic acid as the sole carbon source induces both the de novo formation and proliferation of peroxisomes (58, 65, 68). In addition, oleic acid activates the expression of a number of genes that code for peroxisomal β-oxidation enzymes (20, 48-50, 68). These enzymes are imported directly from the cytosol into the peroxisomal matrix during the normal assembly of functionally mature peroxisomes. Under conditions where peroxisome assembly is compromised or when peroxisome biogenesis is inhibited, oleic acid-induced β-oxidation enzymes, such as 3-ketoacyl coenzyme A thiolase (Pot1p), are mislocalized to the cytosol instead of the peroxisome matrix. Thus, we integrated the cDNA sequence encoding GFP at the 3′ end of the POT1 locus to serve as an unambiguous binary reporter (punctate versus nonpunctate) to screen for peroxisome assembly defects in each of the selected THC-SEC strains (Table 2). We monitored the intracellular distribution of Pot1p-GFP using epifluorescence microscopy following growth of cells in either glucose medium (YEPD) or oleic acid medium (YPBO) in the presence or absence of doxycycline (Fig. 1). Growth of the THC library control strain, R1158, in YEPD exhibited the expected low expression of Pot1p-GFP. Culturing R1158 cells in YPBO significantly increased Pot1p-GFP expression, where it localized exclusively to discrete punctate peroxisome structures. Addition of doxycycline had a modest effect on the induction of Pot1p-GFP expression in R1158 cells cultured in YEPD but had no effect on Pot1p-GFP expression when these cells were grown in YPBO. This did not affect our screen for peroxisome biogenesis defects in doxycycline-repressed THC-SEC-POT1-GFP strains, since the subsequent intracellular localization of Pot1p-GFP to characteristic punctate peroxisome structures in control R1158 cells was not affected.

FIG. 1.

The peroxisomal matrix protein Pot1p is mislocalized in cells repressed for SEC20 and SEC39 expression. (A) Wild-type strain R1158 and strains THC-SEC7, THC-SEC61, THC-SEC20, and THC-SEC39 harboring genomic POT1-GFP were incubated for 18 h in YEPD, at which time cells were transferred to oleic acid-containing medium (YPBO) for an additional 90 min to induce the expression of Pot1p-GFP. (B) Pot1p localizes to peroxisomes in repressed THC-SEC11, THC-SEC13, THC-SEC17, THC-SEC18, THC-SEC65, and THC-COPI cells. The indicated THC strains expressing genomic POT1-GFP were incubated in YEPD as for panel A and then transferred to YPBO medium for 2 h (SEC13) or 4 h (remaining strains). Repression of essential SEC genes under the control of the TetO₇ promoter in panels A and B was achieved by addition of 10 μg doxycycline/ml to the medium. The localization of Pot1p-GFP to peroxisomes was monitored by epifluorescence microscopy. Bars, 5 μm. (C) Sec20p levels are reduced in doxycycline-treated or untreated THC-SEC20 cells. Equal amounts of protein from whole-cell lysates of doxycycline-treated or untreated THC-SEC20 cells and untreated wild-type R1158 cells were subjected to immunoblot analysis with antibodies to Sec20p. Immunodetection of Act1p was used as a control for protein loading.

Of the THC-SEC-POT1-GFP strains screened, only doxycycline-repressed THC-SEC20-POT1-GFP, and THC-SEC39-POT1-GFP cells exhibited significant mislocalization of Pot1p-GFP to the cytosol (Fig. 1A and B and unpublished results). Interestingly, Pot1p-GFP also mislocalized in THC-SEC20-POT1-GFP cells even in the absence of doxycycline, suggesting that altering the endogenous promoter for SEC20 was sufficient to affect normal peroxisome assembly in these cells. Indeed, expression of Sec20p in THC-SEC20 cells in the absence or presence of doxycycline was significantly reduced compared to that in the parental strain, R1158 (Fig. 1C). Repression of SEC7, SEC13, SEC18, SEC61, or COPI did not affect localization of Pot1p-GFP to peroxisomes; however, a modest effect of gene repression on peroxisome number and size was observed in the THC-SEC61-POT1-GFP and THC-COPI-POT1-GFP strains (Fig. 1A and B). Overall, these results show that the mislocalization of Pot1p-GFP in repressed THC-SEC20-POT1-GFP and THC-SEC39-POT1-GFP cells was specific to these genes and did not result from nonspecific effects caused by the loss of an essential secretory protein associated with the ER and/or Golgi apparatus.

Repression of SEC20, SEC39, and DSL1 mislocalizes Pex3p to tubular-vesicular structures.

Since Pot1p-GFP was mislocalized to the cytosol in cells repressed for SEC20 or SEC39 expression, we wanted to characterize further a role for these SEC genes in peroxisome biogenesis. To do this, a cDNA sequence encoding GFP was integrated at the 3′ end of the PEX3 locus in the THC-SEC20 and THC-SEC39 strains. We choose PEX3 since the biogenesis and maintenance of peroxisomes from ER-derived membrane components require Pex3p and are independent of the oleic acid-induced biogenesis and proliferation of peroxisomes (21, 58). Thus, using this system, we could determine the effects of SEC20 and SEC39 gene repression on ER-dependent peroxisome biogenesis by monitoring the intracellular distribution of Pex3p between the ER and peroxisomes. To control for effects of doxycycline addition and for loss of an essential SEC protein at the Golgi apparatus or the ER, the same integration and experimentation were performed with the control R1158 strain and with the THC-SEC14 and THC-SEC61 strains, respectively.

Although it was not identified initially using our selection criteria (see above), we included the THC-DSL1 strain in this screen, because the gene products of SEC20 and SEC39 are part of an essential secretory protein complex that includes Dsl1p (27). Furthermore, inclusion of the THC-DSL1 strain as part of this screen served to provide a more complete analysis of the potential role of these SEC genes and this complex in the biogenesis of peroxisomes.

As expected, Pex3p-GFP localized to discrete punctate peroxisomal structures in the R1158 strain regardless of the presence or absence of doxycycline (Fig. 2). Similarly, Pex3p-GFP localized normally in the THC-SEC14-PEX3-GFP strain, further supporting that the loss of an essential Golgi apparatus-localized Sec protein does not have a general inhibitory effect on peroxisome biogenesis. Pex3p-GFP in the THC-SEC61-PEX3-GFP, THC-SEC39-PEX3-GFP, and THC-DSL1-PEX3-GFP strains cultured in the absence of doxycycline also localized to discrete peroxisome structures, as was seen in the R1158-PEX3-GFP strain. Pex3p-GFP in the THC-SEC20-PEX3-GFP strain exhibited a heterogeneous distribution between peroxisomes and tubular-vesicular structures under similar conditions. Doxycycline repression of SEC61 resulted in an increased number of these Pex3p-GFP-labeled peroxisome structures, in addition to a weaker localization of Pex3p-GFP to tubular-vesicular structures (Fig. 2). In contrast, repression of SEC20, SEC39, and DSL1 expression resulted in significant localization of Pex3p-GFP to tubular-vesicular structures, in addition to discrete punctate structures of various sizes (Fig. 2). These results corroborate our initial Pot1p-GFP localization screen showing that peroxisome biogenesis is compromised by the loss of SEC20 and SEC39. Moreover, repression of DSL1 similarly affected Pex3p-GFP localization to peroxisomes, supporting a relationship between these genes for peroxisome biogenesis and assembly.

Repression of SEC20, SEC39, and DSL1 alters peroxisome profiles and ultrastructure.

Due to the mislocalization of peroxisomal proteins seen in cells repressed for SEC20, SEC39, and DSL1, we next investigated the peroxisome profiles and ultrastructure in doxycycline-repressed SEC20, SEC39, and DSL1 cells by electron microscopy (Fig. 3). To clearly visualize peroxisomes by electron microscopy, cells were cultured in oleic acid to induce the expression and delivery of matrix proteins to the peroxisome, thereby producing the characteristic peroxisome structure of an electron-dense core surrounded by a single-membrane bilayer. Characteristic peroxisome structures and profiles were seen in each of the strains grown in oleic acid except for THC-SEC20. In this strain, we were unable to detect typical peroxisome structures, but observed expanded vesicular regions of the ER (Fig. 3D). This vesicular expansion of the ER was further enhanced in doxycycline-repressed THC-SEC20 cells and was highly reminiscent of the tubular-vesicular structures seen in Fig. 2 (Fig. 3D′ and D″). Interestingly, repression of SEC20 also resulted in a significant expansion and reticulation of the outer nuclear membrane, which was continuous with the ER. This proliferation of the ER generally resulted in a characteristic honeycomb-like pattern, which has also been observed in yeast cells harboring mutant COPI and COPII coat proteins (1, 45). Repression of SEC39 also led to a honeycomb-like expansion of the outer nuclear membrane and expanded vesicular regions of the ER (Fig. 3E′ and E″ and unpublished results). Unlike for the repression of SEC20, a few peroxisomes were present in doxycycline-treated THC-SEC39 cells; however, these structures were typically smaller than those seen in R1158 control cells. Peroxisomes were also found in cells repressed for DSL1, and these structures were usually associated with ER membranes (Fig. 3F′). In addition, several atypical peroxisome-like structures were found to extend from ER membranes in doxycycline-repressed DSL1 cells (Fig. 3F′ and F″). These structures were smaller and contained a central core of decreased electron density that was similar to that seen for the expanded vesicular structures in cells repressed for SEC20 and SEC39.

FIG. 3.

Altered peroxisome profiles and ultrastructure in cells repressed for SEC20, SEC39, or DSL1. (A to F) Strains R1158 (A), THC-SEC14 (B), THC-SEC61 (C), THC-SEC20 (D), THC-SEC39 (E), and THC-DSL1 (F) were incubated for 16 h in YEPD prior to incubation in YPBO for 7 h. (A′ to F′) Repression of TetO₇ promoter-regulated genes was achieved by addition of doxycycline to a final concentration of 10 μg/ml to the medium. (D″ to F″) Increased magnification of selected areas from panels D′ to F′, respectively, to allow better visualization of the expanded vesicular regions associated with the ER (asterisks) resulting from repression of SEC20, SEC39, or DSL1. M, mitochondrion; N, nucleus; P, peroxisome. Bars, 1 μm in panels A to F′ and 0.5 μm in panels D″ to F″.

The addition of doxycycline had no effect on the peroxisome profiles of R1158 and THC-SEC14 cells (Fig. 3A′ and B′, respectively), as expected based on the light microscopy analysis (Fig. 2). Also, similar to what was observed by epifluorescence (Fig. 1 and 2), repression of SEC61 resulted in an increased number of peroxisomes as seen by the number of peroxisomes captured in a single thin-slice electron microscopy preparation (Fig. 3C′). Ultrastructural analysis further revealed increased variability in the dimensions of this expanded peroxisome profile observed in gene-repressed THC-SEC61 cells.

Loss of Sec39p attenuates the formation of peroxisomes from the ER.

Since the normal distribution of Pex3p-GFP between the ER and peroxisomes was affected in yeast upon loss of the essential secretory genes, SEC20, SEC39, and DSL1 (Fig. 2), we next wanted to determine which point along the ER-to-peroxisome transport pathway was affected. As Pex3p is involved in the early stages of peroxisome biogenesis from the ER (21, 58), we again used Pex3p as the reporter for the following pulse-chase experiments. To do this we integrated the GAL1 promoter (pGAL1) to the 5′ end of the PEX3-GFP locus in THC-SEC39-PEX3-GFP cells, permitting us to regulate the expression of Pex3p-GFP. This system further permitted the study of Pex3p transport out of the ER for de novo peroxisome biogenesis in the absence of preexisting peroxisomes (21).

To ensure that Pex3p-GFP expression driven by pGAL1 was not affected by the transient loss of SEC39 or the presence of doxycycline, we first compared galactose-induced Pex3p-GFP expression in THC-SEC39-pGAL1-PEX3-GFP cells in the presence and absence of doxycycline (Fig. 4A). THC-SEC39-GAL1-PEX3-GFP cells were primed for Pex3p-GFP expression by incubation in raffinose for 18 to 20 h, and SEC39 expression was transiently repressed by inclusion of doxycycline during the last 8 h of incubation. As expected, no Pex3p-GFP was detected in THC-SEC39-pGAL1-PEX3-GFP cells following the priming period, and essentially no Sec39p was detected when these cells were primed in the presence of doxycycline. Importantly, the level of Pex3p-GFP expression during the 1-h pulse in galactose was not altered by the loss of Sec39p, and the repression of Sec39p and the expression of Pex3p-GFP were maintained throughout the chase period.

FIG. 4.

Repression of SEC39 alters Pex3p trafficking. (A) Equal amounts of protein from whole-cell lysates prepared from THC-SEC39 cells containing genomically integrated PEX3-GFP under the control of the GAL1 promoter at the PEX3 locus were resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted for GFP, Sec39p, and Gsp1p as a loading control. (B) THC-SEC39 cells integrated for pGAL1-PEX3-GFP were fixed with formaldehyde prior to image capture by epifluorescence microscopy. Arrows indicate the accumulation of Pex3p-GFP in tubular-vesicular structures with apparent perinuclear localization. Prime, cells were cultured for 8 h in YEPR to derepress the GAL1 promoter for subsequent galactose induction of PEX3-GFP expression. Pulse, cells were then transferred to YEPG and incubated for 1 h to pulse-label the cells for PEX3-GFP expression. Chase, expression of PEX3-GFP was then stopped by changing the growth medium to YEP2×D, and cells were incubated for an additional 4 h and 15 h to monitor the trafficking of Pex3p-GFP between the ER and peroxisomes. Repression of SEC39 was regulated by the addition or omission of 10 μg doxycycline/ml to the culture medium. Bar, 5 μm. (C) THC-SEC39-pGAL1-PEX3-GFP, and THC-SEC61-pGAL1-PEX3-GFP cells were subjected to prime, pulse, and chase incubation steps as described for panel B, except that the prime step was for 18 h and the chase step was for 4 h. Cells were fixed in formaldehyde prior to image capture by epifluorescence microscopy at the indicated steps. Bars, 5 μm. (D) Equal amounts of protein from whole-cell lysates of the THC-SEC39-pGAL1-PEX3-GFP and THC-SEC61-pGAL1-PEX3-GFP cells used for panel C were resolved by SDS-PAGE, transferred to nitrocellulose and immunoblotted for GFP, Sec39p, Sec61p, Gsp1p, and Act1p. Gsp1p and Act1p served as loading controls.

Having determined the validity of our experimental system we next monitored the trafficking of Pex3p-GFP from the ER by epifluorescence microscopy (Fig. 4B). Consistent with our immunoblot analysis, Pex3p-GFP was not detected initially in cells during the priming step. Following the 1-h pulse in galactose to induce pGAL1-PEX3-GFP expression, Pex3p-GFP was detected and localized mainly to the cytosol along with targeting to a few localized foci. This localization was independent of the repression of SEC39 by doxycycline and was consistent with previous reports of an ER localization for Pex3p (21, 58). Upon removal of galactose, the transit of newly synthesized Pex3p-GFP between the ER and peroxisomes was monitored at 4 h and 15 h during the chase period. In the absence of doxycycline, Pex3p-GFP was found to traffic efficiently out of the ER to punctate peroxisome structures at the 4-h chase time point and remained in discrete punctate peroxisome structures for the remaining 15-h chase period. Loss of Sec39p resulted in Pex3p-GFP localizing initially to tubular-vesicular structures, with apparent perinuclear localization at the 4-h chase time point and subsequent localization to discrete punctate peroxisomal structures by 15 h of chase. However, localization of Pex3p-GFP to tubular-vesicular structures was still evident at the 15-h chase time point in the doxycycline-repressed cells. These results show that the loss of Sec39p attenuates the exit of Pex3p-GFP from the ER, thereby delaying the de novo formation of ER-derived peroxisomes.

When the experimental conditions were modified to include doxycycline during the entire priming step, we found that a more prolonged loss of Sec39p led to a more striking localization of Pex3p-GFP to tubular structures following the 1-h pulse incubation in galactose (Fig. 4C). Localization of Pex3p-GFP to tubular structures persisted during the 4-h chase period. Interestingly, the apparent fluorescent signal for Pex3p-GFP was found to be considerably less in THC-SEC39-pGAL1-PEX3-GFP cells when they were treated with doxycycline.

Since the intensity of the fluorescent signal may not be linear with protein abundance, depending on the intracellular localization of the protein, immunoblot analysis was performed on these samples. The decrease in the apparent fluorescence was traced to a significant reduction in the expression of Pex3p-GFP from pGAL1 under these conditions (Fig. 4D). The reduced expression was specific for the prolonged repression of SEC39, since this was not seen with the shorter repression time for SEC39 (Fig. 4A) or in THC-SEC61-pGAL1-PEX3-GFP cells treated under similar conditions (Fig. 4D). Normal Pex3p-GFP trafficking between the ER and peroxisomes was not affected in THC-SEC61-pGAL1-PEX3-GFP cells under these conditions (Fig. 4D). These results demonstrate that gene repression of SEC39 affects the early stages of peroxisome biogenesis by attenuation of Pex3p transit from the ER to peroxisomes and provide additional support for the specificity of SEC39 repression on peroxisome biogenesis, as loss of the essential ER secretory gene SEC61 had no effect on Pex3p trafficking between the ER and peroxisomes.

Oleic acid induces the expression of Sec20p, Sec39p, and Dsl1p.

A number of genes coding for proteins associated with peroxisome biogenesis and proliferation are induced during growth of yeast in oleic acid as the sole carbon source (25, 49, 50, 68, 69). Accordingly, we investigated whether the expression of Sec20p, Sec39p, and Dsl1p was induced by oleic acid (Fig. 5). Indicative of induction of peroxisome biogenesis, Pot1p showed the typical marked increase in expression over time of incubation of cells in oleic acid. The levels of Sec20p, Sec39p, and Dsl1p also increased to various degrees with time of incubation of cells in oleic acid. Sec20p levels increased throughout the 6-h time course, whereas Sec39p and Dsl1p expression peaked at 2 h following oleic acid addition. These results are consistent with a role for these proteins at the earliest stages of peroxisome biogenesis.

FIG. 5.

Oleic acid induces the expression of Sec20p, Sec39p, and Dsl1p. Equal amounts of protein from whole-cell lysates of wild-type BY4742 cells cultured in oleic acid-containing SCIM1 for the times indicated were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted for Sec20p, Sec39p, Dsl1p, the ER-resident protein Kar2p, and Pot1p.

Sec20p, Sec39p, and Dsl1p associate with peroxisomes in vivo.

Having established that Sec20p, Sec39p, and Dsl1p were involved in peroxisome biogenesis and assembly, we next wanted to determine if these proteins were associated with peroxisomes in vivo. To do this, we isolated peroxisomes from cells grown in oleic acid for 16 h by isopycnic gradient ultracentrifugation of a peroxisome-enriched 20,000 × g fraction (20KgP) (Fig. 6A). Similar to results described in previous reports (49, 62, 71), peroxisomes were found in higher-density fractions of the gradient, localizing mainly to fractions 3, 4, and 5 as determined by the presence Pot1p and Pex3p in these fractions. Sec20p, Sec39p, and Dsl1p were also enriched in fraction 4, indicating an association of these proteins with peroxisomes in vivo. The ER-resident proteins Kar2p (luminal) and Sec61p (membrane) were also found in this fraction, raising the possibility that the enrichment of Sec20p, Sec39p, and Dsl1p in the peroxisome fraction was due solely to cofractionation with ER membranes. To determine if the localization of Sec20p, Sec39p, and Dsl1p in the peroxisome fraction was due to a true association with peroxisomes or the indirect result of cofractionation with ER membranes, we performed the same fractional analysis on pex3Δ cells, which lack peroxisomes. Indeed, Sec20p and Sec39p were no longer enriched in fraction 4 and localized to lighter fractions. A similar pattern was also found for Dsl1p but not to the extent of that observed for Sec20p and Sec39p. In contrast, the ER-resident membrane protein Sec61p did not shift to lighter fractions upon loss of peroxisomes, indicating that the enrichment of Sec20p, Sec39p, and Dsl1p in fraction 4 in BY4742 yeast cells was due to the presence of peroxisomes in this fraction. The absence of peroxisomes in pex3Δ cells was verified by the loss of Pot1p in the 20KgP fractions. Interestingly, the ER luminal protein Kar2p also localized to lighter fractions upon loss of peroxisomes. In summary, these data provide the first evidence of an in vivo association of essential secretory proteins with peroxisomes.

To further investigate the association of Sec20p, Sec39p, and Dsl1p with peroxisomes, we coexpressed the peroxisome marker Pex3p-mRFP with Sec20p-GFP, Sec39p-GFP, or Dsl1p-GFP in BY4742 yeast cells, cultured the cells in oleic acid for 4 h, and visualized the intracellular localization of the proteins by confocal microscopy. As expected, the majority of Sec20p-GFP, Sec39p-GFP, and Dsl1p-GFP were found to localize to the ER, as previously described (24, 43). However, a small fraction of each protein colocalized with Pex3p-mRFP as seen by single-plane confocal images (Fig. 6B). Colocalization of Sec20p-GFP, Sec39p-GFP, and Dsl1p-GFP with Pex3p-mRFP was not dependent on growth of cells in oleic acid, as these proteins were also found to colocalize in cells grown in YEPD (Fig. 6C). In addition to the association of Sec20p-GFP, Sec39p-GFP, and Dsl1p-GFP to peroxisomes, Pex3p-mRFP was also found frequently juxtaposed to, interspersed between, or at the terminal ends of Sec20p-GFP-, Sec39p-GFP-, and Dsl1p-GFP-positive ER membranes (Fig. 6B and C). In contrast, Pex3p did not colocalize with the structural ER-localized membrane protein Rtn1p (8, 72), even in a maximum-intensity projection that collapses all confocal planes onto a single image (Fig. 6D), thus showing that the association of Pex3p with Sec20p, Sec39p, and Dsl1p was specific.

DISCUSSION

Recent studies have established that the ER plays an essential role in the biogenesis and maintenance of peroxisomes (21, 26, 58). Disruption of intracellular transport between the ER and peroxisomes leads to the complete loss of identifiable peroxisomes. This absolute requirement for ER-derived membrane components for the biogenesis of peroxisomes predicts the existence of secretory machinery acting in this process, but a role for secretory genes involved in membrane trafficking between the ER and peroxisomes has yet to be described. Previous studies have ruled out a direct role for COPI- and COPII-mediated vesicular transport in peroxisome biogenesis (53, 73). To date, a complete loss of peroxisomes in yeast has been shown to occur only upon deletion of either PEX3 or PEX19. However, the products of these genes exhibit no similarities to proteins that function as secretory machinery. In the current study, we performed a gene repression screen of essential secretory genes implicated in ER-localized trafficking and identified a role for SEC20, SEC39, and DSL1 in the biogenesis of peroxisomes. The results from this study are significant because they are the first to describe secretory machinery involved in the de novo formation and maintenance of peroxisomes. Moreover, these results support the existence of another vesicular ER trafficking pathway in addition to the one dedicated to ER-to-Golgi transport.

Under wild-type conditions, newly derived Pex3p-containing structures from the ER fuse with preexisting peroxisomes and do not serve as templates for the assembly and maturation of new peroxisomes (35). The growth and division of preexisting peroxisomes constitute the primary mechanism for the proliferation of peroxisomes under these conditions, and preexisting peroxisomes serve as the site of import for peroxisomal matrix proteins (35). In the absence of preexisting peroxisomes, ER-derived Pex3p-containing structures serve as a scaffold upon which new peroxisomes assemble (35), and the intracellular pool of peroxisomes cannot be maintained without the continuous production of ER-derived Pex3p-containing structures (21, 22, 58). Based on this current view of peroxisomes, the results of the present study strongly support a role for Sec20p, Sec39p, and Dsl1p in the early stages of peroxisome biogenesis and/or assembly.

Repression of Sec20p and Sec39p in cells resulted in significant mislocalization of the peroxisomal matrix protein Pot1p to the cytosol, indicative of a lack of preexisting peroxisomes or a defect in peroxisomal matrix protein import in these cells (Fig. 1). The peroxisomal membrane protein Pex3p relocalized to tubular-vesicular structures in cells repressed for the expression of SEC20, SEC39, and DSL1 (Fig. 2). This is also indicative of a lack of preexisting peroxisomes in these cells. Analysis of peroxisome ultrastructure by electron microscopy further supported the relocalization of Pex3p to tubular-vesicular structures in cells repressed for SEC20, SEC39, and DSL1 expression and corroborated the lack of mature peroxisome structures in gene-repressed SEC20 and SEC39 cells (Fig. 3). Even in the absence of gene repression, Sec20p levels in THC-SEC20 cells were below the detection limits of our antibody; however, the viability of these cells indicates that a minimum threshold of expression for the essential SEC20 gene was achieved. Of note, under these conditions THC-SEC20 cells also exhibited defects in peroxisome biogenesis, consistent with a role for Sec20p in this process (Fig. 1, 2, and 3).

We also found that the normal trafficking of newly synthesized Pex3p-GFP from the ER to peroxisomes was disrupted in cells depleted of Sec39p (Fig. 4). This disruption was not observed in cells repressed for SEC61 expression and was therefore specific to the loss of Sec39p (Fig. 4). Pulse-chase experimentation provided evidence for a defect in Pex3p trafficking being at a point between the ER and peroxisomes, as delivery of Pex3p to the ER and its insertion into the ER were unaffected in Sec39p-repressed cells. Also consistent with a role for SEC20, SEC39, and DSL1 in peroxisome biogenesis, the expression of these genes was induced in cells grown under conditions that stimulate peroxisome biogenesis and proliferation in wild-type yeast cells (Fig. 5). Finally, a minor fraction of Sec20p, Sec39p, and Dsl1p associated with peroxisomes in vivo (Fig. 6B and C), and subcellular fractionation of wild-type yeast cells cultured in oleic acid confirmed an association of these proteins with peroxisomes (Fig. 6A). Taken together, these results strongly support a role for Sec20p, Sec39p, and Dsl1p in the early stages of the peroxisome biogenic process.

A role for Sec20p, Sec39p, and Dsl1p in the biogenesis of peroxisomes from the ER challenges the current view that these proteins function exclusively in Golgi-to-ER retrograde transport (4, 5, 27, 43, 66). New structural evidence showed that Sec39p and Dsl1p operate together as a vesicle-tethering complex at the ER via interactions with the ER-localized SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) Sec20p (63). A central region in Dsl1p contains binding sites for two different subunits of the COPI vesicle coat complex, which through its interactions with Sec20p serves as a bivalent recognition complex for the recruitment of COPI vesicles to the ER (4, 5, 63). This suggests a mechanism for Sec20p, Sec39p, and Dsl1p in peroxisome biogenesis that involves the recruitment of COPI vesicles to the ER. However, consistent with previous results (53, 73), gene repression of COPI did not significantly affect de novo peroxisome biogenesis as indicated by the localization of Pot1p-GFP to peroxisomes in these cells (Fig. 1), although it did alter the number and size of peroxisomes, suggesting that COPI has an unidentified role in the growth and division of preexisting peroxisomes (37). Overall, the interaction of Sec20p, Sec39p, and Dsl1p with the COPI vesicle coat appears to be dispensable for the de novo biogenesis of peroxisomes and indicates that Sec20p, Sec39p, and Dsl1p function via a different mechanism when acting in peroxisome biogenesis.

The association of Sec20p, Sec39p, and Dsl1p with additional functions is not without precedent. The Sec20p homolog in plants has been implicated in the exit of seed storage proteins from the ER (31), and the mammalian equivalent of Dsl1p, ZW10, was shown to have a role in cell cycle checkpoint control, dynein targeting, and membrane trafficking (64). A possible explanation for these seemingly diverse functions could be the recruitment of dynein by this complex to various intracellular sites. This hypothesis is supported by studies that show that the loss of ZW10 function leads to the release of cytoplasmic dynein from both kinetochores and membranes (54, 67). A role for dynein in peroxisome biogenesis and movement in mammalian cells has also been described, and dynein localizes to peroxisomes in yeast (7, 28, 56). Since the movement of peroxisomes for inheritance in yeast occurs along actin cables and uses a myosin motor (14), the association of microtubular dynein with peroxisomes most certainly functions in another capacity. Our data support a potential role for Sec39p, Dsl1p, and Sec20p in the recruitment of dynein for peroxisome biogenesis, and this requires further investigation. Yet, the recruitment of other factors by Sec20p, Sec39p, and Dsl1p to the ER for the biogenesis of peroxisomes cannot be ruled out.

The site of action for Sec20p, Sec39p, and Dsl1p is most likely at the ER, as these proteins were found localized mainly to this compartment (Fig. 6B and C) (24, 43). Sec20p, Sec39p, and Dsl1p, along with the ER-luminal protein Kar2p, were also enriched in peroxisome-containing fractions from cells grown in oleic acid (Fig. 6A). The presence of Sec20p, Sec39p, Dsl1p, and Kar2p in these fractions was dependent on the presence of peroxisomes, because in pex3Δ cells that lack peroxisomes, these proteins shifted in their localization to fractions of lesser density (Fig. 6A). In contrast, the ER-resident membrane protein Sec61p, which was also found in peroxisome-enriched fractions from wild-type cells, did not exhibit a shift to lighter fractions in pex3Δ cells. Localization of Sec20p, Sec39p, and Dsl1p, together with some amount of soluble Kar2p, to a specific ER subdomain that is involved in the biogenesis of peroxisomes and which excludes most membrane components of the rough ER could explain this observation (58). The shift of Sec20p, Sec39p, Dsl1p, and Kar2p to less dense subcellular fractions of pex3Δ cells further indicates that the integrity of this specialized ER subdomain is dependent on the continuous production of ER-derived Pex3p-containing structures and/or the presence of peroxisomes. Close apposition of peroxisomes with ER membranes containing Sec20p, Sec39p, and Dsl1p supports this conclusion.

Minor amounts of Sec20p, Sec39p, and Dsl1p colocalized with Pex3p in vivo. To determine the validity of this colocalization, we investigated the extent of colocalization between Pex3p and another ER-resident membrane protein, Rtn1p. Rtn1p is localized throughout the cortical ER and functions in providing the characteristic reticular structure of that organelle (8, 72). Rtn1p was observed not to colocalize with Pex3p in vivo, indicating that the pools of Sec20p, Sec39p, and Dsl1p colocalizing with Pex3p, although small, were significant. The function of the fraction of Sec20p, Sec39p, and Dsl1p that associates with peroxisomes remains to be determined. The association of these proteins with peroxisomes could potentially result from their “escape” from the ER subdomain involved in peroxisome biogenesis during the peroxisome biogenic process (38).

The atypical vesicular expansions associated with ER membranes in cells repressed for SEC20, SEC39, or DSL1 are consistent with earlier reports investigating mutant alleles of DSL1 (4, 66). Similar membrane continuities between the ER and peroxisomes have also been described for mammalian intestinal and dendritic cells (18, 36). The electron density of the expanded ER structures in the repressed SEC20, SEC39, and DSL1 cells was considerably less than that of normal peroxisomes, suggesting a lack of matrix protein content and consistent with the mislocalization of the matrix protein Pot1p to the cytosol in these cells. The absence of normal peroxisome profiles in THC-SEC20 cells, even in the absence of doxycycline, suggests that the Pex3p-GFP-labeled structures observed in these cells are not peroxisomes. These results are further support for the presence of a specialized subdomain of the ER involved in the de novo formation of peroxisomes.

The oleic acid-induced expression of SEC20, SEC39, and DSL1 was consistent with a previous report from a global transcriptional analysis to identify genes involved in peroxisome assembly and function (49). Similar to the observation by Smith and colleagues of the robust activation of SEC39 by oleic acid compared to that of SEC20 or DSL1 (50), we found that Sec39p exhibited the greatest increase in protein levels when cells were incubated in oleic acid (Fig. 5). Interestingly, the reported biphasic activation of SEC39 and DSL1 by oleic acid was essentially mirrored by the extent of Sec39p and Dsl1p expression, suggesting that the activation and expression of these genes are tightly regulated. The biphasic nature of the oleic acid-induced gene activation of SEC39 and DSL1 also implies that peroxisome biogenesis in response to a fatty acid carbon source may not be a linear process but occurs in a periodic fashion based on the length of exposure of yeast cells to lipid.

An unexpected finding was that gene repression of SEC61 resulted in an increase in peroxisome number. South and colleagues had reported previously that incubation of yeast cells harboring a temperature-sensitive mutant of SEC61 at the nonpermissive temperature had no effect on peroxisome biogenesis (51). The apparent discrepancy between these results and ours may be explained by the absence of PEX11 in the yeast cells used in their study. Pex11p is a peroxisomal membrane protein involved in peroxisome proliferation (32), and the use of pex11Δ cells most likely masked any effect that the loss of Sec61p activity had on peroxisome division and multiplication. We are currently investigating how Sec61p regulates peroxisome number and size.

The moderate accumulation of Pex3p-GFP in tubular-vesicular structures in cells repressed for SEC61 may have occurred secondarily to the primary loss of Sec61p. Sec20p is predicted to be a type II transmembrane protein, and its insertion into membranes would be dependent on Sec61p (40). Based on this relationship, we expect that decreased expression of Sec61p would affect peroxisome biogenesis by affecting Sec20p's incorporation into membranes, thereby indirectly affecting the newly attributed function of Sec20p in peroxisome biogenesis.

Roles for other essential SEC genes in peroxisome biogenesis cannot be completely excluded by our study. Although our experimental conditions were consistent with those of previous studies involving repression of essential genes using the THC yeast strains (2, 3, 34), subsequent loss of protein was not directly verified for each gene. However, based on experimentally measured protein half-lives in the budding yeast (6), we would expect the majority of the essential Sec proteins screened in our study to have been knocked down sufficiently under our experimental conditions.

In conclusion, we have identified the ER-resident proteins Sec20p, Sec39p, and Dsl1p as being involved in the initial steps of peroxisome biogenesis and assembly. Whether these proteins function in the ER exit of Pex3p-containing structures or in the delivery of peroxisomal membrane components requires further investigation.