Carboxyl-terminal Consensus Ser-Lys-Leu-related Tripeptide of Peroxisomal Proteins Functions in Vitro as a Minimal Peroxisome-targeting Signal*

The minimal sequence requirement for a peroxi- some-targeting signal was investigated using an in vitro import system. Carboxyl-terminal sequences Ser-Lys-Leu (SKL) and Leu-Gln-Ser-Lys-Leu (LQSKL) of acyl-CoA oxidase (AOX) directed to per- oxisomes the fused proteins with import-incompetent forms of AOX and catalase that had been truncated, implying that the SKL tripeptide functions as a targeting signal. Elimination of the entire SKL sequence or deletion of any 1 or 2 amino acids in the sequence abolished the import activity of AOX. Substitution of alanine for serine did not affect the import activity. Topogenic activity was retained when lysine was mutated to either arginine or histidine, whereas mutation to glutamic acid completely abolished the activity. A synthetic peptide comprising the carboxyl-terminal 10 amino acid residues of AOX inhibited the import of the authentic AOX polypeptide, whereas other peptides in which SKL was mutated, deleted, or internally located were not effective. The uptake of AOX was little af- fected by the peptide with an amidated a-carboxyl group. These results strongly suggest that the car- boxyl-terminal SKL motif sequence

tides into peroxisomes resides in the internal sequence of the protein. Recently, a peroxisome-targeting signal was noted in vivo and in vitro for several peroxisomal enzymes, including luciferase and acyl-CoA oxidase (AOX)' (Gould et al., 1988;Miyazawa et al., 1989). The topogenic signal is suggested to comprise a Ser-Lys-Leu (SKL)-containing sequence located at the extreme carboxyl terminus (for a review, see Osumi and F'ujiki (1990)). In a yeast system such as Candida tropicalk, however, the signal for AOX (PXP4) appears to be different (Small et al., 1988).
To study further the minimal sequence and positional requirement of the SKL translocation signal and to examine if the analogous carboxyl-terminal sequences such as SHL of porcine kidney D-amino-acid oxidase (Ronchi et al., 1982) and SRL of rat liver urate oxidase (Reddy et al., 1988) function as targeting signals, we have carried out in vitro import experiments on AOX variants in which the SKL sequence was modified by site-directed mutagenesis. Each residue in SKL was found to be essential for import activity, but the lysine residue was replaceable by arginine or histidine. Studies on the effect on AOX import of synthetic peptides containing SKL suggested the presence of a putative component(s) interacting with an SKL peroxisome-targeting signal.

EXPERIMENTAL PROCEDURES
Synthesis of Oligonucleotides and Peptides-Oligonucleotides were synthesized by the phosphoramidite method (Sinha et al., 1984) on an Applied Biosystems Model 380A DNA synthesizer. The oligonucleotides were purified by reverse-phase high-performance liquid chromatography or column chromatography on an oligonucleotide purification cartridge (Applied Biosystems, Inc.).
Peptides were synthesized by stepwise solid-phase peptide synthesis with tert-butoxycarbonyl chemistry (Merrifield, 1963) on an Applied Biosystems Model 430A peptide synthesizer using phenylacetamidomethyl resin that had been attached by a carboxyl-terminal amino acid. Peptides were cleaved from the resin by treatment with trifluoromethanesulfonic acid (Tam et al., 1986) and purified by reverse-phase high-performance liquid chromatography using a CIS column. The purified peptides were hydrolyzed in uacuo with 5.7 N HCI at 110 "C for 22-24 h and then analyzed for purity and quantity by amino acid analysis on a Beckman Model 6300E amino acid analyzer. Amino acid sequences of some peptides were verified by an Applied Biosystems Model 470A Protein/Peptide Sequencer equipped with a Model 120A phenylthiohydantoin analyzer. For synthesis of a peptide in which the a-COOH group of the carboxyl-terminal amino acid was amidated, p-methylbenzhydrylamine resin (Mitchell et al., 1978) was used.
Site-directed Mutagenesis-The full-length AOX cDNA had previously been cloned into the SacI site of the pTZ18R vector (Miyazawa et al., 1989). After digestion of the AOX plasmid with SacI The abbreviations used are: AOX, acyl-CoA oxidase; Hepes, 4-(2-hydroxyethy1)-1-piperazineethanesulfonic acid. endonuclease, the fragment containing full-length AOX cDNA was separated by agarose gel electrophoresis and isolated using a DEAEcellulose membrane filter (NA45,Schleicher & Schuell). This fragment was subcloned into the SacI site of the M13mp19 vector. For deletions and mutations in the carboxyl-terminal SKL sequence of AOX, mutant AOX cDNAs were prepared by means of site-directed mutagenesis using antisense synthetic oligonucleotides of 20-24 bases according to the procedure of Kunkel (1985). Substitutions or deletions of 1-9 bases were incorporated into the central region of each oligomer and were flanked by 9-12 bases of exact complementarity, as shown in Table I. Transformants were screened for mutations, and the cDNAs obtained were verified by restriction enzyme mapping and nucleotide sequencing. Mutant AOX cDNAs were cloned into the SacI site of the pTZl8R vector downstream of the T7 promoter and then used for in uitro transcription using T7 RNA polymerase to produce mutant AOX transcripts as described previously (Miyazawa et al., 1989).
Isolation of Peroxisomes from Rat Liuer-A male Sprague-Dawley rat weighing 200-250 g that had been injected intraperitoneally with Triton WR-1339 (Leighton et al., 1968) was fasted overnight and killed under anesthesia with ethyl ether. The liver was excised and homogenized in 2 volumes of 0.25 M sucrose, 10 mM Hepes/KOH, pH 7.5, 1 mM EDTA, 0.1% ethanol, 25 p M leupeptin (buffer A) by two strokes of an Elvehjem-Potter homogenizer. All of the following steps were done at 4 "C. A postnuclear supernatant fraction, prepared by centrifugation of the homogenates at 750 X g for 10 min, was centrifuged at 2500 X g for 10 min. The supernatant (post-heavy mitochondrial fraction) was further centrifuged at 25,000 X g for 10 min; the resulting pellet (light mitochondrial fraction) was washed twice with buffer A and suspended in the same buffer but without leupeptin.
The light mitochondrial fraction was centrifuged at 50,000 rpm (228,000 X g) for 1.5 h in a Beckman Model L8-70M ultracentrifuge using a VTi-65.2 rotor in a sucrose density gradient containing 30-60% sucrose, 50 mM glycylglycine/KOH, pH 7.6, 1 mM EDTA, 0.1% ethanol. The gradient was collected from the bottom of the centrifuge tube into 20 fractions, which were then assayed for marker enzymes (Leighton et al., 1968). Highly pure peroxisomes almost free from mitochondria, which were obtained usually in fraction 4 or 5, were used for in uitro import assay.
In Vitro Import-Wild-type and mutant AOX polypeptides were synthesized by translation of cDNA transcripts in a nuclease-treated rabbit reticulocyte lysate cell-free protein-synthesizing system with ["S]methionine as label (Miyazawa et al., 1989). In uitro import of AOX was carried out with freshly isolated peroxisomes essentially as described Miyazawa et al., 1989). Briefly, the translation mixture was adjusted to 0.25 M sucrose and centrifuged for 10 min at 18,000 X g; the resulting supernatant (75 pl) was incubated for 60 min at 26 "C with peroxisomes (18.75 pg, calculated on the basis of catalase activity) . The import reaction mixture was then divided into three aliquots; two were treated at 0 "C for 30 min with proteinase K (100 pg/ml) in the presence or absence of 1% Triton X-100. Digestion was terminated by addition of 20 pg of phenylmethylsulfonyl fluoride, 250 p~ antipain, 250 p M chymostatin, 250 p M leupeptin, 250 p M pepstatin. The remaining aliquot was placed at 0 "C for 30 min in the absence of proteinase K and detergent. The supernatants and peroxisome pellets were separated by centrifugation at 18,000 X g for 10 min and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis using 10% gel (Laemmli, 1970), followed by fluorography (Bonner and Laskey, 1974). In the experiments examining the effect on AOX import of various synthetic SKL-related peptides, the import reaction was started by adding peroxisomes to the import assay mixture to which the peptides had been added at the concentrations indicated. The quantity of peptides in the import assay mixtures was determined as follows. Following incubation for 0, 10, 30, and 60 min at 26 "C of the import assay mixture containing all components except for 35Slabeled AOX, 1% octyl glucoside, 10 mM N-ethylmaleimide, 100 pg of phenylmethylsulfonyl fluoride was added. The solubilized mixture was filtered by centrifugation through a membrane (Millipore Ultrafree C3LCC, M, cutoff = 5000); the filtrate was applied to an Applied Biosystems Model 130A reverse-phase high-performance liquid chromatography CIS column, with scanning at 230 nm.
Other Methods-DNA sequencing was carried out with [cY-~'P] dCTP by the dideoxy sequencing method of Sanger et al. (1977). Radioactivity of 35S-labeled polypeptides was quantitated by counting protein bands on polyacrylamide gels (Mori et al., 1981).

RESULTS
Minimal Peroxisome-targeting Signal of AOX-Our previous studies on the in vitro import of AOX mutants into peroxisomes suggested that the peroxisome-targeting signal of AOX resides at the extreme carboxyl terminus, comprising <5 amino acid residues (Miyazawa et al., 1989). To investigate the minimal targeting information, various AOX mutants as well as fusion proteins were constructed and assayed for in vitro import.
Wild-type full-length AOX was synthesized in a cell-free system, which was then post-translationally incubated for 60 min at 26 "C with freshly and highly purified rat liver peroxisomes. AOX was recovered, to a significant extent (46%), in the peroxisome fractions (Fig. 1, lane 3 ) , nearly 20% of which were resistant to externally added proteinase K ( l a n e 5 ) , thereby implying that import activity of AOX was represented as -10% incorporation. Treatment with 1% Triton X-100 prior to protease digestion abolished protease resistance (lanes TABLE I Carboxyl-terminn1 nucleotide sequences of wild-type and SKL AOX mutants Antisense mutagenic oligonucleotide primers were synthesized with exact complementarity to the desired mutant sequences. A dash in the sequence indicates a deletion; triple asterisks above the sequence indicate a desired amino acid substitution. An asterisk under the sequence shows base substitution to create (+) or eliminate (-)  6 and 7). No AOX polypeptide was found in the peroxisome fractions immediately after mixing the translocation product and peroxisomes followed by treatment with proteinase K (data not shown). These results suggest that full-length AOX is imported into peroxisomes, consistent with previous data Imanaka et al., 1987;Miyazawa et al., 1989). An AOX variant (AOGss) in which the tripeptide SKL was removed from wild-type AOX was found with peroxisomes, nearly to the same extent as noted in the wild type, when assayed for import activity (Fig. 1, lane 10). No AO&s polypeptides remained in the peroxisome fraction after proteinase K treatment ( l a n e 12), indicating that the polypeptide is not taken up by peroxisomes.
A 35S-labeled polypeptide of -50 kDa was detected in wildtype AOX without treatment with proteinase K (Fig. 1, lane  3, downward-pointing arrowhead). This protein was resistant to proteinase K (lane 5, upward-pointing arrowhead), suggesting that it is most likely to be a 52-kDa B component of AOX derived from imported AOX by endogenous proteolytic cleavage. The same conversion was noted both i n vivo and in vitro (Furuta et al., 1982;Miura et al., 1984;Miyazawa et al., 1989). In the case of AOXm, however, the -50-kDa 35Slabeled protein, immunocross-reactive with the anti-AOX antibody (data not shown), was likewise noted in lanes 11-13, but not in lane 10, implying that this polypeptide is a cleavage product by proteinase K of AO&s and is partly recovered in the peroxisome fraction, presumably due to nonspecific binding (lanes 10-14). A similar cleavage fragment was seen in wild-type AOX (lanes 4 and 6).
AOXssl, in which 70 amino acid residues were deleted from the carboxyl terminus, was not detected in peroxisomes after digestion with proteinase K ( Fig. 2A, P2 lane). When the tripeptide SKL was linked to the carboxyl terminus of AOX591, the resulting polypeptide (AOX691-C3) was recovered in peroxisomes and was resistant to proteinase K, indicating that AOX591-C3 is translocated into peroxisomes, consistent with our previous observation for a deletion mutant of AOX in which residues 592-658 were eliminated (Fig. 2B) (Miyazawa et al., 1989). Furthermore, AOX was truncated -200-230 amino acid residues from the carboxyl termininus; neither of the resulting polypeptides (AOX462 and AOX4P8) was proteaseresistant in the peroxisome fraction after the import assay ( Fig. 2, C and E ) . However, via addition of SKL to the carboxyl termini of AOX462 and AOX428, both truncated proteins became import-competent as verified by resistance to protease digestion (Fig. 2, D and F). Linkage of pentapeptide LQSKL to the carboxyl terminus of AOX28 likewise restored the import of AOX428 (Fig. 2G), consistent with the deletion analysis of AOX described previously (Miyazawa et al., 1989).
Next, a truncated rat liver catalase was used as a heterologous protein for fusion with AOX carboxyl-terminal sequences of 3 and 5 amino acid residues. Catalase from rat liver is a peroxisomal matrix protein of 60 kDa, consisting of FIG. 2. In vitro import of fusion proteins. Fusion polypeptides containing SKL and LQSKL at their carboxyl termini were synthesized as described under "Experimental Procedures." I n uitro import was carried out and analyzed as described for Fig. 1. AOX,, carboxylterminal deletion constructs of AOX in which n specifies the residue number at the newly formed carboxyl-terminal position; AOX,-C3 and -C5, constructs in which three or five carboxyl-terminal amino acid residues (SKL or LQSKL) of wild-type AOX are linked to the carboxyl terminus of AOX,. Rat liver catalase consists of 527 amino acid residues (Furuta et al., 1986); the carboxyl-terminal part of catalase was deleted at position 370 (Cat). In Cat-C3 and -CS, the carboxyl-terminal SKL and LQSKL sequences of AOX are fused to the carboxyl terminus of Cat, respectively. Import activity is represented as a plus or minus sign (see P2 lanes). The arrowhead in the P2 lane indicates a fusion protein (AOX162-C3); the bands seen in the S2 and S g lanes of C-G are due to an undigested endogenous polypeptide in this region. 527 amino acids with no extreme carboxyl-terminal SKL or its related sequence (Furuta et al., 1986); catalase truncated at position 370 (Cat) was assayed for in vitro import into peroxisomes. Cat seen in peroxisome fractions after the reaction was digested with proteinase K, suggesting that Cat is import-incompetent (Fig. 2H). Accordingly, Cat was used as a heterologous passenger protein. When the carboxyl terminus of Cat was fused to SKL and LQSKL, the resulting polypeptides (Cat-C3 and Cat-C5) were both recovered, nearly to the same extent as Cat, with peroxisomes and were found to be partially resistant (lesser in the case of Cat-C3) to externally added protease (Fig. 2, I and J, PI and Pz lunes).
They were, however, alp digested with the protease after treatment with detergent prior to digestion (P3 and S3 lunes).
These results indicate that SKL and LQSKL possess the topogenic activity to translocate the import-negative truncated catalase into peroxisomes, suggesting that the minimal sequence of a peroxisome-targeting signal is within an SKL tripeptide.
Mutation and Deletion of SKL Motif-In an attempt to identify the minimal sequence of the peroxisome-targeting signal, various AOX mutants in which the SKL sequence was either mutated or deleted by site-directed mutagenesis were examined for import activity; only the peroxisome fractions (see Figs. 1 and 2, P2 lunes) after digestion of the import reaction mixtures with proteinase K are shown (Fig. 3). None of the AOX mutants with deletion of any 1 or 2 residues in SKL were found in peroxisomes (lunes 2-7) into which wildtype AOX was efficiently imported (lune 1 ), as seen in Fig. 1. The AOX variants with lysine mutated to arginine or histidine in SKL were recovered in peroxisomes and found to be protease-resistant, with lesser import activity (34%) of AOX with SHL than with SRL (86%) and the wild type (100%) (Fig. 3, compare lunes 1,8, and 9). By substitution of glutamic acid for lysine in SKL, however, the AOX mutant became import-negative (lune 10).
A carboxyl-terminal sequence (-Ala-Lys-Leu) similar to SKL has been noted for nonspecific lipid transfer protein (see Table 11) (Morris et al., 1988), a protein identical to sterol carrier protein 2, which is mostly, if not completely, localized in peroxisomes from rat liver (Tsuneoka et al., 1988. We tested if the AKL sequence functions as a  FIG. 3. Effects of deletions and mutations in SKL sequence on AOX import. AOX mutants in which the carboxyl-terminal SKL sequence was either deleted or mutated were constructed by sitedirected mutagenesis as described under "Experimental Procedures." Import assay of AOX mutants was done as described for Fig. 1. Only the peroxisome fraction after proteinase K treatment of each import assay mixture is shown (see Figs. 1 and 2, P2 lanes). Lune 1, wildtype (WT) AOX containing the SKL sequence; lanes 2-7, AOX mutants with deletions of any 1 (lanes 2-4) or 2 (lanes 5-7) amino acids; lanes 8-10, mutants in which the lysine residue (K) in SKL was changed to arginine (R), histidine (H), or glutamic acid (E), respectively; lane 11, serine in SKL was replaced by alanine (A). The radioactivity of the AOX polypeptide in each lane was quantitated by counting as described under "Experimental Procedures"; the relative import activities of AOX mutants are represented as percentages of that (as 100%) of the wild type, where 11% input radioactivity was imported (the dash indicates 4 % activity of the wild type). peroxisome-targeting signal by mutating serine to alanine in the SKL sequence of AOX. The 35S-labeled AOX with AKL sequence was recovered in the peroxisome fraction to the same extent as the wild type (78%) (Fig. 3, lune 11 ).
These results strongly suggest that the topogenic signal requires all residues in SKL and that lysine is replaceable by other basic amino acids such as arginine and histidine, but cannot be substituted with an acidic residue such as glutamic acid. The sequence AKL also appears to be functional as a targeting signal.
Inhibition of Import by Synthetic Peptides-To confirm the finding that SKL is a minimal peroxisome-targeting sequence, we studied the effect on AOX import of various synthetic peptides in which SKL was modified. In vitro import of authentic 35S-labeled AOX was performed in the presence of synthetic peptides at concentrations of 10, 25, 50, 100, and 200 p~ (Fig. 4). The wild-type peptide comprising the carboxyl-terminal 10 amino acid residues of AOX (KHLKPLQ-SKL) apparently reduced (at 25 p~) 40% of the import of full-length AOX, and a further 70% was reduced as the concentration of peptide was increased to 200 p~ (Fig. 4 FIG. 4. Inhibition of AOX import by synthetic peptides. Import of full-length AOX was carried out as described for Fig. 1, except that synthetic peptides were present at the concentrations indicated. Protease-resistant AOX polypeptides recovered in the peroxisome fractions are shown as described for Fig. 3. A, a 10-amino acid peptide comprising the authentic carboxyl-terminal sequence of wild-type AOX ( K H L K P L Q E ) ; B, a peptide in which SKL was deleted but 3 residues were extended to the amino terminus in the authentic sequence of AOX (SYHKHLKPLQ); C, a peptide in which SKL was inserted in the middle of the sequence, yielding K H L K E P L Q ; 0, an authentic peptide, but the a-COOH group was amidated; E, a peptide in which lysine in SKL was changed to glutamic acid ( K H L K P L Q a ) ; F, a peptide of 5 amino acids with the carboxylterminal AOX sequence L Q e , G, a tripeptide of =. AOX polypeptide bands were quantitated as described in Fig. 3. Upper right, AOX imported in the presence of peptides is plotted relative to that in the absence of peptides. 0, wild-type peptide AOX-CdSKL); 0, peptide AOX-Clo(SEL). Lower right, stability of the SKL-related peptides in the import assay mixture. The peptides used in the inhibition assay (A-E) (200 p~) were studied as described under "Experimental Procedures"; -70% were recovered immediately after adding the peptides to the import assay mixture and were expressed as 100%. The representatives AOX-Clo(SKL) and AOX-CdSEL) are shown. and upper right, AOX-Clo (SKL)). No apparent difference in the inhibition of AOX import was noted between pretreatment of the peroxisomes at 0 "C for 30 min with the peptide (25 and 100 p~) and simultaneous incubation in the import reaction mixture (data not shown). In the import assay in the presence of the AOX-Clo(SKL) peptide, AOX was found in a lesser amount (nearly one-third) in peroxisome fractions prior to protease digestion (corresponding to P1 lanes in Figs. 1 and   2) compared with in the absence of the peptide (data not shown). This suggests that the inhibition by the peptide of AOX import may occur at the step of binding of AOX to the peroxisomes.
Other peptides were likewise tested to determine if they had any effect on AOX import. The peptide of AOX sequence 649-658 (SYHKHLKPLQ), lacking carboxyl-terminal SKL but containing 10 amino acid residues, did not show such an apparent reducing effect even at 200 p~ (Fig. 4B). To find clues as to whether SKL needs to be located at the carboxyl terminus, SKL was placed in the middle of 10 residues. The resulting peptide apparently did not inhibit AOX import (Fig.   4C). The authentic peptide, in which the a-carboxyl group of the carboxyl-terminal leucine was amidated, did not decrease the peroxisomal uptake of AOX, an -10% decrease in uptake was apparent only at 200 p~ (Fig. 40). The peptide with a substitution of glutamic acid for lysine in the SKL sequence did not affect the import of wild-type AOX (Fig. 4, E and  upper right). Pretreatment of peroxisomes with the AOX-Clo(SEL) peptide, performed as with AOX-C,o(SKL), showed no apparent difference in the effect on AOX import (data not shown). To exclude the possibility that the synthetic peptides showing no apparent inhibitory effect are degraded in the in vitro import assay mixture, we studied the stability of the peptides (Fig. 4, A-E) in the reaction mixture. All the peptides (200 PM) remained at 3040% levels after the import reaction for 30 min when the inhibitory or noninhibitory effect of the peptides on AOX import was apparent. A nearly identical effect on AOX import was noted among the peptides at this concentration, i.e. -100 p~. At 60 min of incubation, only (10% of the peptides were detected, although the effect on AOX import was evident (Fig. 4, lower right). This may be explained by the finding that import of AOX reaches maximal levels in vitro at 15-30 min of incubation Miyazawa et al., 1989; this study (data not shown)). In comparison with AOX-Clo(SKL), the peptide with the sequence LQSKL showed much less influence on the import of wild-type AOX, i.e. -10% reduction at 100 p~; no apparent effect of the tripeptide SKL was noted (Fig. 4, F and G).
Taken together, the SKL-containing peptides appear to requires some structural context to exert their effects, indicating that the SKL motif functions as the peroxisomal topogenic signal when (and more likely only when) located at the extreme carboxyl terminus.

DISCUSSION
In this study using an in vitro import system, we identified and characterized the minimal sequence of a peroxisometargeting signal. The AOX carboxyl-terminal sequence of 5 (LQSKL) or 3 (SKL) amino acids functions as a peroxisomal topogenic signal when linked to homologous and heterologous proteins. The SKL tripeptide signal is functional only when it is in full sequence. In SKL, serine can be changed to alanine; lysine is replaceable by other basic amino acids such as arginine and histidine, but not by the acidic residue glutamic acid, implying the importance of a positive charge. All the in vitro findings reported here are essentially in good agreement with those obtained in in vivo experiments, where the mutated and fused firefly luciferase genes were transfected to mammalian cells (CV-1) (Gould et al., 1989). It is noteworthy that any mutation of leucine in SKL abolishes the targeting activity in vivo; serine can be replaced by cysteine and alanine.
Studies either in an in vivo or in vitro system have some advantages and disadvantages per se and are mutually complementary. The immunofluorescence assay, employed in the in vivo experiments such as those of Gould et al. (1989), is qualitative. It is therefore rather difficult to evaluate the import efficiency among the various proteins. Failure in vivo to detect low efficiency import may result in accumulation in the cytosol of the polypeptides examined. On the other hand, the in vitro import assay that we carried out here allows us to determine the activities of the polypeptides tested, although the fundamental issue may remain obscure as to what extent the findings in in vitro studies reflect physiological occurrence. Nevertheless, investigations both in vivo and in vitro have reached the same conclusion: a peroxisome-targeting signal is located at the extreme carboxyl termini of several peroxisomal proteins and comprises a tripeptide with the consensus sequence of serine or alanine at the first position in the sequence; lysine, arginine, or histidine at the second; and leucine at the third.
Over one dozen proteins among -40 peroxisomal polypeptides thus far sequenced contain a carboxyl-terminal SKL motif (Table 11). The SKL signal has been proven functional for the import of luciferase and AOX by independent in vivo (Gould et al., 1989) and in vitro (Miyazawa et al., 1989; this report) studies, respectively. The notion of the SKL motif sequence present in mammals, plants, insects, and yeast implies that the mechanism of protein translocation into peroxisomes using the SKL signal has been conserved throughout eukaryotic evolution (Osumi and Fujiki, 1990), as suggested by the observation that insect luciferase is translocated into peroxisomes in mammalian cells (Keller et al., 1987) and yeast (Gould et al., 1990). It is also noteworthy that glyceraldehydephosphate dehydrogenase from Trypanosoma brucei glycosome, a peroxisome-like organelle, contains a carboxyl-terminal AKL sequence (Michels et al., 1986). Taken together, as noted by Gould et al. (1990), we propose here that the SKL motif functions as a peroxisome-targeting signal in eukaryotic organisms. The targeting signal with a short and well-conserved motif is rather unique in contrast to other topogenic signals responsible for the transport of newly synthesized polypeptides to organelles such as mitochondria and the endoplasmic reticulum that is mediated by amino-terminal cleavable extra peptides with longer sequences, but no consensus sequences (Verner and Schatz, 1988). A short carboxylterminal signal is reminiscent of a retention signal in the endoplasmic reticulum (KDEL) (Munro and Pelham, 1987). However, the processes of protein retention in the endoplasmic reticulum and polypeptide import into peroxisomes appear to be fundamentally different.
The authentic 10-amino acid synthetic peptide of AOX inhibits the import of AOX in a concentration-dependent fashion. However, all the other 10-amino acid peptides examined such as those with the sequence SEL instead of SKL or with SKL but located in the middle of the sequence, do not reduce the uptake of AOX, which is similar to the finding of Gould et al. (1989) that mutation of SKL to SKLS or SKLIK abolishes the translocation of luciferase into peroxisomes in CV-1 cells. The importance of the free carboxyl group of the carboxyl-terminal leucine can be inferred from the loss by amidation of the inhibitory effect of the peptide. Taken together, we conclude that the SKL signal functions Acyl-CoA oxidase, rat -Leu-Gln-Ser-Lys-Leu Miyazawa et al. (1987) Bifunctional protein, rat -His-Gly-Ser-Lys-Leu Osumi et al. (1985) Luciferase, firefly -Gly-Lys-Ser-Lys-Leu de Wet et al. (1987)
Citrate synthase 2 encoded by the CZT2 gene have been shown to be peroxisomal (Lewin et al., 1990). A Deroxisomal membrane-associated Drotein with a molecular mass of -20 kDa.
only when located at the extreme carboxyl terminus with a free a-COOH group. Not all peroxisomal proteins contain the SKL motif sequence at their carboxyl termini. Some possess it internally instead, although it is not known if the internal SKL motif is functional as topogenic information. It is of interest to note that the carboxyl-terminal 27-amino acid segment of human catalase directs the fusion protein to peroxisomes in CV-1 cells (Gould et al., 1988). This sequence contains the SKL motif Ser-His-Leu at an internal location. This tripeptide, however, is changed to Ser-His-Ile in rat liver catalase (Furuta e t al., 1986), which is less likely to function as a targeting signal as deduced from mutation studies on luciferase, i.e. leucine cannot be changed to any other amino acid (Gould e t al., 1989). We should await further investigations of whether it acts as a signal in certain specific contexts. The glycosomal phosphoglycerate kinase from a Kinetoplastida, Crithidia fasciculata, contains an extra carboxyl-terminal sequence compared with its cytosolic counterpart (Swinkels e t al., 1988). This region is proposed to possess a glycosome-targeting signal, but an SKL motif is not found in this sequence. Another type of signal may be functional.
Small e t al. (1988) suggested by in vitro import assay that AOX (PXP4) from the yeast C. tropicalis contains topogenic signal(s) in two regions. A similar conclusion was obtained in in vivo experiments (Kamiryo e t al., 1989). An SKL sequence has been noted internally in this protein, but not in the segments that appear to contain the targeting information. Other type(s) of peroxisome-targeting signals may exist in C. tropicalis; further studies would identify them. Aitchison e t al. (1991) recently suggested that a carboxyl-terminal tripeptide (AKI) of the C. tropicalis trifunctional enzyme, similar to but distinct from the SKL motif, functions as a targeting signal.
The concentration of peptide used in this assay appears to be rather high as compared to that of radiolabeled AOX synthesized in a cell-free system and even to the amount of peroxisomes. It is noteworthy that similar concentrations of synthetic presequence peptides have been used (up to 100 PM) to determine if they block the translocation of precursors into mitochondria (Gillespie e t al., 1985; Furuya e t al., 1987; Chu e t al., 1989; Glaser and Cumsky, 1990) and chloroplasts (Perry e t al., 1991). It is possible that only a fraction of the peptide with a given amino acid sequence has a conformation effective for the inhibition of import. To our surprise, the synthetic peptides studied here were much less stable in the import assay mixture than we anticipated, probably because of endogenous proteolytic activity such as that in the reticulocyte lysate (Rechsteiner, 1987). This kind of experiment, regarding inhibition by synthetic peptides of protein translocation in an in vitro system, should be done with caution. Nevertheless, a clear difference was noted in the inhibition of AOX import between the peptide with carboxyl-terminal SKL and the other peptides with modified SKL, suggesting that the effect of synthetic peptides is not likely to be an artifact caused by high concentrations of peptides in the assay. Competition of AOX import with a synthetic peptide (AOX-Clo(SKL)) is concentration-dependent, implying that some component(s) may be present, presumably such as those interacting with the SKL signal. Several approaches such as those using protein cross-linking, affinity chromatography, and anti-idiotype antibody, in combination with sensitive in vitro import assay systems, would be needed to detect the signal receptor(s1, as successfully used for the identification of import sites of mitochondria and chloroplasts (Vestweber e t al., 1989;Sollner e t al., 1989Sollner e t al., , 1990Pain e t al., 1988Pain e t al., , 1990). An approach of genetic complementation of cell mutants that are detective in assembly of peroxisomes has also been successful in identifying genes essential for biogenesis of peroxisomes, such as mammalian peroxisome assembly factor-1 (Tsukamoto e t al., 1991; Shimozawa e t al., 1992) and pas-I and pas-3 from