Cloning and Expression of the DJP1 Gene
We have identified the DJP1
gene by functional complementation of the peroxisome assembly mutant pas22-1
(see Materials and Methods). DJP1
is identical to YIR004w, a 432–amino acid ORF on chromosome IX, identified during the course of the S
genome sequencing project (Voss et al., 1995
). We deleted the YIR004w gene and allowed the disruption mutant to mate with pas22-1
cells. Both the diploid cells and the haploid gene deletion mutant showed the same growth characteristics as pas22-1
cells. Growth on oleic acid–containing medium was retarded, whereas growth on all other carbon sources (glucose, galactose, glycerol, acetate, and ethanol) was normal (not shown). In pas22-1
cells, we found a single base pair deletion (G1129) in the DJP1
gene. This alteration caused a frame shift at codon 376 resulting in premature termination of the ORF. These results identified YIR004w as the DJP1
gene. A polyclonal antiserum raised against the COOH-terminal 294–amino acid residues of Djp1p recognized Djp1p specifically on a Western blot (Fig. , lane 1
), and this protein band was absent from Δdjp1
cell lysates (Fig. , lane 2
). As expected, pas22-1
cells expressed a low molecular weight version of the DJP1
gene (Fig. , lane 3
Figure 1 Western blot analysis of DJP1 gene products. Lysates from wild-type, Δdjp1, and pas22-1 cells grown overnight on oleate-contaning medium were used for Western blot analysis (left panel) or lysates from wild-type cells grown overnight on medium (more ...)
Since growth of Δdjp1 cells was retarded only on oleate media, we anticipated a role for Djp1p in peroxisome biogenesis. Peroxisome number and peroxisomal protein levels are regulated in response to fatty acids (oleate) in the media. The expression level of Djp1p, however, was constitutive; it was unaffected by culture conditions or heat shock (Fig. , lanes 4–7).
Djp1p contains an NH2-terminal J-domain (Fig. , A and B). This 70–amino acid domain is conserved throughout evolution and is named after E. coli DnaJ. EcDnaJ consists of several additional evolutionarily conserved domains, which Djp1p does not have (Fig. A). Instead, Djp1p contains two predicted domains of unknown function.
Figure 2 Djp1p is a member of a novel subfamily of J-domain– containing proteins. (A) Schematic representation of the modular structure of Djp1p compared with E. coli DnaJ. According to the ProDom database (Gouzy et al., 1996), which makes use of the (more ...)
Localization of Djp1p in the Cytosol
Using subcellular fractionation experiments, we studied the subcellular localization of Djp1p in cells grown overnight on oleic acid medium (Fig. a). A homogenate (H) was prepared from spheroplasts by gentle osmotic lysis and fractionated by successive differential centrifugation into a 2,500 g pellet (P1), a 17,000 g pellet (P2), a 100,000 g pellet (P3) and a supernatant fraction (S) (Fig. a). All four fractions were analyzed for the presence of organellar marker proteins, either by Western blot analyses or by enzymatic activity measurements. F1β-ATPase, Kar2p, and phosphoglucose isomerase (PGI) were used as markers for mitochondria, ER, and cytosol, respectively. We used peroxisomal catalase (catalase A), peroxisomal malate dehydrogenase (Mdh3p) and 3-ketoacyl-CoA thiolase (thiolase) as marker proteins for the peroxisomal matrix and the peroxisomal ABC transporter protein 1 (Pat1p) as marker for the peroxisomal membrane.
Figure 3 Djp1p is located in the cytosol. (a) Subcellular fractionation experiments showing that Djp1p was localized to the cytosolic fraction of oleate-grown wild-type and Δpex6 cells. A cell-free homogenate (H) was fractionated by successive differential (more ...)
The fractionation profile of Djp1p was different from that of mitochondrial, peroxisomal and ER proteins. The bulk of Djp1p cofractionated with the cytosolic marker PGI in the 100,000 g supernatant (S) (Fig. a) or even in a 200,000 g supernatant (not shown), indicating that Djp1p is a cytosolic protein. However, a fraction of Djp1p was pelleted at 17,000 g (P2). P2 contained a very low amount of PGI activity suggesting a marginal cytosolic contamination of P2, but this was not sufficient to explain the presence of Djp1p in P2. Djp1p was not necessarily associated with functional peroxisomes as Djp1p was also found in the 17,000 g pellet from cells that lack morphologically distinguishable peroxisomes, the Δpex6 cells (Fig. a) and Δpex13 cells (not shown). The fractionation behavior of Djp1p was identical in glucose-grown cells.
We next aimed to identify the nature of the pelletable Djp1p fraction. The 17,000 g pellet derived from wild-type cells was therefore subjected to Nycodenz equilibrium density gradient centrifugation followed by Western blot analysis. In this gradient, Djp1p did not colocalize with peroxisomes and mitochondria but remained in the low density fractions (Fig. b). The pelletable fraction of Djp1p apparently associated with a large structure, either as part of a protein complex or associated with membranes. To distinguish between these two possibilities, we prepared a 17,000 g organellar pellet from wild-type cells, resuspended it in 50% Nycodenz (wt/vol) and layered the suspension at the bottom of a Nycodenz floatation step gradient (see Materials and Methods). Membranes were floated to light density by overnight centrifugation at 150,000 g as monitored by the distribution of the organellar marker proteins (Fig. c). The majority of Djp1p molecules migrated up into the gradient (Fig. c, peak in fraction 8) implying its association with membranes. The Djp1p equilibration profile was distinct from that of peroxisomes, which peaked in fraction 5 (Fig. c). Mitochondria and microsomes migrated with the bulk of the membranes to fractions 5–8, with a peak in fractions 7 and 8. We concluded that Djp1p was associated with membranes that overlap in density with, but are distinct from, the bulk of peroxisomes, mitochondria, and microsomes.
Since proteins may dissociate from membranes or from protein assemblies during homogenization of cells and subsequent subcellular fractionation, we wanted to confirm our results on the subcellular localization of Djp1p using a milder method. Spheroplasts prepared from oleate-grown cells were incubated with increasing concentrations of digitonin and the release of marker proteins was measured. At low digitonin concentrations, the plasma membrane was permeabilized selectively as monitored by the release of the cytosolic marker protein PGI (Fig. d) with simultaneous retention of organellar markers. At higher digitonin concentrations intracellular membranes were permeabilized, upon which catalase and NH2-Mdh3p were released from the cells. Djp1p started to be released from cells at low digitonin concentrations, initially coeluting with PGI. However, by increasing the digitonin concentration gradually, more Djp1p was released. This illustrates that most of Djp1p behaved as a soluble cytosolic protein, whereas the rest was retained until digitonin concentrations were reached that solubilized intracellular membranes completely. The use of different carbon sources for growth did not affect the retarded release of Djp1p from digitonin-permeabilized cells (not shown). Furthermore, the more digitonin-resistant pool of Djp1p was not necessarily associated with functional peroxisomes, since the retarded release was also observed in digitonin-permeabilized Δpex6 cells (Fig. d).
To firmly establish the cytosolic localization of Djp1p, and to determine whether the membrane-associated pool of Djp1p was located on the cytosolic side of the membranes, we used a third approach. We tested latency of the protein in a cell homogenate using protease digestion (Fig. e
). We therefore prepared detergent-free homogenates from wild-type cells, and treated them with proteinase K to degrade proteins that are not protected by a membrane. The control peroxisomal matrix protein thiolase was completely protected from digestion by the protease, whereas upon addition of the detergent Triton X-100 to open up the peroxisomal membranes, thiolase was cleaved (Fig. e
). Thiolase contains a single protease-sensitive site that is rapidly cleaved upon exposure to protease, whereas the remaining proteolytic product is protease resistant (Höhfeld et al., 1991
). In contrast to thiolase, Djp1p was degraded rapidly and quantitatively by Proteinase K both in the absence and presence of Triton X-100. These experiments confirm and extend our differential centrifugation experiments and digitonin titration experiments that indicate that Djp1p was not incorporated into organelles. It appears mainly as a cytosolic protein, with a fraction associated with the cytosolic side of membranes.
Mislocalization of PTS-containing GFP in Δdjp1 Cells
The growth deficiency of Δdjp1
cells was restricted to media containing fatty acids as sole carbon source, which is suggestive of a defect in peroxisome functioning. To investigate whether peroxisome biogenesis was indeed affected in Δdjp1
cells, we determined whether proteins with a peroxisomal targeting signal can be imported into peroxisomes in these cells. We used both a PTS1- and a PTS2-containing variant of the GFP from the jellyfish Aequorea victoria
(Kalish et al., 1995
). The cDNAs encoding these proteins were cloned into a yeast expression vector containing the oleate-inducible regulatory sequence of the catalase A (CTA) gene (Elgersma et al., 1993
). Cells transformed with the GFP-PTS1 or PTS2-GFP expression plasmids were grown to late logarithmic phase on selective glucose medium, and then transferred to oleate medium and analyzed 24 h later (Fig. ).
Figure 4 Ddjp1 cells partially mislocalize GFP-PTS1 and PTS2-GFP and show aberrant morphology of peroxisomal structures after prolonged growth on oleate media. Distribution of GFP containing a PTS1 or PTS2 in wild-type cells, Δdjp1 cells, and either (more ...)
In wild-type cells transformed with either construct, a few large clusters of fluorescent spots were present, indicating that both proteins efficiently entered peroxisomes (Fig. ). In Δdjp1 cells, however, a punctate fluorescence pattern reminiscent of peroxisomes was observed against a background of pronounced diffuse labeling for both proteins. Prolonged culturing of wild-type cells on oleate media resulted in the formation of very large fluorescent structures, which were absent from Δdjp1 cells for both GFP-PTS1 (not shown) and PTS2-GFP (Fig. ). Instead, the diffuse labeling of the cells increased and the peroxisomes remained small. We concluded that in Δdjp1 cells, only a fraction of PTS-containing GFP is associated with peroxisomes whereas the rest is mislocalized to the cytosol. After prolonged culture on oleate medium, atypically small peroxisomal structures are present in Δdjp1 cells.
Mislocalization of Peroxisomal Matrix Proteins in Δdjp1 Cells
To extend and confirm the results with GFP for endogenous peroxisomal proteins, and to determine whether the effects of DJP1 deletion are specific for the peroxisome, we examined the subcellular distribution of marker proteins for cytosol (PGI), mitochondria (F1β-ATPase), ER (Kar2p), and peroxisomes (catalase A, NH-Mdh3p as a typical PTS1 marker protein, thiolase as PTS2 marker protein, and Pat1p and Pex13p as markers for the peroxisomal membrane). Cells were grown for 16 h on oleate medium to allow optimal peroxisome induction and then fractionated as shown in Fig. . Whereas marker proteins for ER and mitochondria fractionated in Δdjp1 cells as in wild-type cells, the peroxisomal marker protein catalase A behaved clearly differently (Fig. a). In contrast, NH-Mdh3p, thiolase, and the membrane proteins Pat1p (Fig. a) and Pex13p were located mainly in the 17,000 g pellet (P2) of Δdjp1 cells. Hence, after 16 h growth on oleate, the most pronounced defect in Δdjp1 cells was the mislocalization of catalase A to the cytosolic fraction.
Figure 5 Mislocalization of peroxisomal matrix proteins in Δdjp1 cells depends on growth conditions. Subcellular distribution of marker proteins of wild-type and Δdjp1 cells grown (a) on oleate for 16 h, (b) on oleate for 40 h, and (c) on glycerol (more ...)
Since we had observed that the cytosolic labeling of GFP-PTS1 and PTS2-GFP increased with time after the shift to oleate medium, we wanted to find out whether the mislocalization of endogenous peroxisomal matrix proteins in Δdjp1 cells was dependent on cellular growth conditions as well. We grew Δdjp1 cells on oleate for 40 h and compared the distribution of peroxisomal marker proteins between the 17,000 g organellar pellet and the supernatant fraction (Fig. b). Under these conditions, catalase A was recovered almost completely in a 17,000 g supernatant fraction. Moreover, thiolase was now also present in the supernatant fraction whereas NH-Mdh3p and the peroxisomal membrane proteins remained completely particulate.
The same experiment was performed with cells grown overnight on glycerol-containing medium (Fig. c). These cells showed the most defective phenotype, with extensive mislocalization of NH2-Mdh3p as well as catalase A and thiolase. Our results indicate that the extent to which peroxisomal proteins cofractionated with the cytosolic marker protein varied with growth conditions in Δdjp1 cells. Optimal oleate induction partially restored the Δdjp1 phenotype. The level of Djp1p itself was not regulated in response to oleate in the medium and, inversely, the oleate response was not impaired in Δdjp1 cells as determined by the level of activity of peroxisomal β–oxidation enzymes (not shown). These results suggest that either Djp1p was functionally replaced by an oleate-inducible factor or the function of Djp1p was needed less under conditions of optimal oleate induction. Interestingly, serial analysis of gene expression (SAGE) of wild-type cells grown on oleate for 16 h revealed the induction of several heat-shock proteins that may render Djp1p partially redundant (Kal, A.J., and H.F. Tabak, personal communication).
Under both growth conditions, oleate as well as glycerol, marker proteins for ER, mitochondria and cytosol behaved as in wild-type cells (Fig. a; and results not shown), indicating that the effects of DJP1 gene deletion were limited to peroxisomes (see below).
Low-Density Peroxisomal Structures in Δdjp1 Cells
To further characterize the peroxisomes in Δdjp1 cells, especially considering the punctate pattern shown in Fig. , we determined their density in a Nycodenz gradient. Fractionation of a 17,000 g pellet (P2) of either wild-type or Δdjp1 cells by Nycodenz equilibrium density gradient centrifugation revealed that the peroxisomal matrix protein thiolase comigrated with the peroxisomal membrane proteins Pex13p (Fig. ) and Pat1p (not shown) in Δdjp1 cells. In contrast to peroxisomes in wild-type cells, which equilibrated at high density, the peroxisomes in Δdjp1 cells equilibrated at a broad density range, the bulk of them having a lower density than the peroxisomes in wild-type cells (Fig. ).
Residual Protein Import into Small Spherical Peroxisomes in Δdjp1 Cells
The biochemical experiments revealed a partial association of matrix proteins with peroxisomes in Δdjp1 cells, and direct immunofluorescence suggested a partial peroxisomal localization for the GFP-PTS proteins. To determine whether these matrix proteins were indeed imported into the organelle, or whether they were just associated with the peroxisomal surface, we performed immunogold electron microscopy studies. The localization of CTA and NH2-Mdh3p in Δdjp1 cells was studied (Fig. a). Antisera directed against NH2-Mdh3p (Fig. a, large gold particles) and catalase A (small gold particles) specifically labeled the peroxisomal matrix in Δdjp1 cells, showing that matrix proteins were indeed imported into the peroxisomes of Δdjp1 cells.
Figure 7 Peroxisomes of Δdjp1 cells are small and contain relatively little catalase. (a) Immunogold electron micrograph of wild-type (WT) and Δdjp1 cells expressing NH2-Mdh3p grown on oleate for 16 and 40 h. Small gold particles (5 nm) represent (more ...)
In cross-sections of Δdjp1 cells grown on oleate for 40 h, peroxisomal structures were scarce and relatively small. As our subcellular fractionation studies predicted, the ratio of small gold particles (representing CTA) to large gold particles (NH2-Mdh3p) was much higher in wild-type cells than in Δdjp1 cells (Fig. , a and b). Hence, the amount of CTA relative to the amount of NH2-Mdh3p was low in peroxisomes of Δdjp1 cells when compared with wild-type cells.
Specificity of Djp1p for Peroxisomal Protein Import
Peroxisomes are indispensable for growth on oleate media but are not necessary for growth on other carbon sources such as glucose, glycerol, acetate, and ethanol. Peroxisomal protein import mutants therefore have been identified based on these growth characteristics. Disturbances of protein import into the endoplasmic reticulum (Novick et al., 1980
) or nucleus (Doye and Hurt, 1995
) are either lethal or have been shown to affect the growth rate on glucose media. A complete block in mitochondrial protein import is lethal, whereas mutation of import-stimulating factors has been shown to affect growth on non-fermentable carbon sources (Baker and Schatz, 1991
). Thus, based on the growth characteristics of Δdjp1
cells, it is not likely that Djp1p is required for import of proteins into nuclei, ER, or mitochondria.
Our fractionation data shown in Fig. a for the resident ER protein Kar2p and mitochondrial F1-ATPase confirm the specificity of Djp1p function for peroxisomes. To further exclude a relation between Djp1p and the ER, we expressed GFP fusion proteins containing an ER retention signal at the COOH terminus and either the presequence of invertase (Suc2p) or that of preproα-factor at the NH2 terminus (see Materials and Methods). Expression of the fusion proteins in wild-type cells resulted in a fluorescent pattern similar to that observed for Kar2p by indirect immunofluorescence microscopy (not shown). After growth on oleate for 24 h (Fig. a) or 40 h (not shown), the GFP fusion proteins displayed the same distribution in Δdjp1 cells as in wild-type cells. This experiment indicated that ER morphology was normal and that import into the ER was not disturbed to such an extent that the steady state distribution of the GFP fusion proteins was affected in Δdjp1 cells.
Figure 8 Import of proteins into the ER in Δdjp1 cells. (a) Expression of GFP-HDEL fused to the invertase presequence in wild-type (WT) and Δdjp1 cells grown on oleate medium for 24 h. (b) Pulse-chase experiment studying the processing of pre-Kar2p (more ...)
If proteins are long-lived, a decrease in import rate is not necessarily reflected by a change in their steady state distribution. To address this, we studied the import rate of Kar2p in Δdjp1 cells by monitoring the processing of the precursor to the mature form (Fig. b). Therefore, wild-type and Δdjp1cells were incubated with [35S]methionine and [35S]cysteine for 5 min (pulse), and the label was chased with cold methionine and cysteine during 5 and 30 min. We did not detect any accumulation of preKar2p in wild-type or Δdjp1 cells after 5 min of pulse-labeling, indicating rapid import and processing. As a control we used sec62ts cells, which clearly accumulated preKar2p. We concluded that preKar2p was rapidly imported into Δdjp1 cells. We again showed that Djp1p was not involved in the import of proteins into the ER.
To further exclude a role for Djp1p in protein import into mitochondria we analyzed the import of mitochondrial Hsp60 by following its processing from the precursor form. As a control, we used Δtom20
cells, which accumulate some mitochondrial precursor proteins in the cytosol, including Hsp60 (Ramage et al., 1993
; Moczko et al., 1994
). Western blot analysis of total lysates showed that in contrast to Δtom20
cells (Fig. , lane 3
cells (Fig. , lane 2
) did not accumulate precursor Hsp60, implying that protein import into mitochondria was not affected to such an extent that it led to accumulation of precursor Hsp60 in the cytosol. If import of Hsp60 was only slightly retarded, or if mitochondrial Hsp60 is long-lived, accumulation of the precursor may have escaped detection. Therefore, we performed a pulse-chase experiment as described above for Kar2p. All of the labeled Hsp60 was already processed into the mature form at the end of the pulse (Fig. , lanes 4
); precursor was not detectable at any time. The results were identical for wild-type and Δdjp1
cells, indicating rapid import and processing of mitochondrial proteins in the absence of Djp1p.
Figure 9 Import of Hsp60 into mitochondria in Δdjp1 cells. (a) Western blot analysis of Hsp60 in lysates from wild-type, Δdjp1, and Δtom20 cells grown overnight on glucose-containing medium at 28°C. (b) Pulse-chase experiment (more ...)
Import into neither the ER nor mitochondria was affected in Δdjp1
cells. A common feature of proteins targeted to the ER and mitochondria is the NH2
-terminal location of the organellar targeting sequence (signal sequence). Furthermore, signal-sequence–containing proteins are kept in a (partially) unfolded conformation during the translocation process. Since most peroxisomal matrix proteins contain a COOH-terminal targeting signal sequence, and since for at least two of our reporter proteins (Mdh3p and thiolase) it had been shown that they oligomerize before they are imported into the organelle, we also tested whether import into the nucleus of (folded) GFP fused to a nuclear localization signal was affected in Δdjp1
cells. In wild-type and Δdjp1
cells NLS-GFP accumulated in the nucleus (Fig. ). No morphological abnormalities of the nucleus were observed. To study nuclear protein import kinetics, we made use of a recently published method (Shulga et al., 1996
) which is based on the reversible import of NLS-GFP. Cells were depleted of ATP by incubating them with azide. This resulted in disappearance of the bright nuclear fluorescence and equilibration of NLS-GFP between cytosol and nucleus (Fig. ). The vacuole remained unlabeled and appeared as a dark hole surrounded by a diffuse fluorescence. Upon restoration of import conditions (washing azide-treated cells and addition of glucose-containing medium), NLS-GFP concentrated into a small bright spot in each cell, reminiscent of the nucleus. Relative import rates were quantified by counting the percentage of cells that showed NLS-GFP nuclear accumulation as a function of time. Δdjp1
cells imported NLS-GFP with the same kinetics as wild-type cells. This experiment illustrated that Djp1p was not required for import of proteins into the nucleus.
Figure 10 Nuclear import kinetics of NLS-GFP in wild-type and Δdjp1 cells. Cells in the exponential growth phase expressing NLS-GFP were treated with azide. This resulted in equilibration of NLS-GFP between cytoplasm and nucleus. The vacuole remained (more ...)
Finally, we studied the rate at which Δdjp1 cells can adapt their metabolism to environmental changes, which depends on the synthesis and maturation of a number of (cytosolic) proteins. For this purpose exponentially growing cells were shifted from glucose medium to galactose medium and growth was measured. Δdjp1 cells grew at the same exponential rate as wild-type cells (Fig. ), implying that under these conditions Djp1p was not required. Furthermore, Δdjp1 cells rapidly adapted to growth on galactose medium as the lag phase was similar to that of wild-type cells. To follow maturation of one protein in more detail, the induction of one of the enzymes specifically required for growth on galactose (galactose-1-phosphate uridylyltransferase [Gal-1-PUT]) was measured. In Δdjp1 cells this enzyme was induced with the same kinetics as in wild-type cells (Fig. ). Metabolic adaptation requires the function of many processes, including signal transduction, de novo protein synthesis, folding, and protein sorting. We concluded from our results that Djp1p was specifically required for peroxisomal protein import and was not needed for maturation or import of proteins in any other organelle. Djp1p was therefore dispensible for cell growth under all culture conditions except oleate.
Galactose induction kinetics in wild-type and Δdjp1 cells. (a) Growth curve of cells in galactose-containing medium and (b) induction curve of the galactose-inducible enzyme galactose-1-phosphate uridylyltransferase (Gal-1-PUT).