Proc. Nati. Acad. Sci. USAVol. 91, pp. 10541-10545, October 1994Biochemistry

Saccharomyces cerevisiae peroxisomal thiolase is imported asa dimer

(protein targeting/lntraorganeile assembly/epitope tagging/protein conformation)

JOHN R. GLOVER*, DAVID W. ANDREWS*, AND RICHARD A. RACHUBINSKIt*Department of Biochemistry, McMaster University, Hamilton, Ontario L8N 3Z5, Canada; and tDepartment of Anatomy and Cell Biology, University ofAlberta, Edmonton, Alberta T6G 2H7, Canada

Communicated by David D. Sabatini, July 7, 1994

ABSTRACT The active conformation of native peroxiso-mal 3-ketoacyl-CoA thiolases (EC 2.3.1.16) is homodimeric.We have previously shown that a truncated Saccharomycescerevsiae thiolase lacking its first 16 N-terminal amino acidsfails to be translocated into peroxisomes but assembles into anenzymatically active form in the cytoplasm of a strain with adisrupted nuclear thiolase gene. We now report that whentruncated thiolase is cosynthesized with full-length thiolase,-50% of truncated thiolase cofractionates with the full-lengththiolase to fractions enriched for peroxisomes and is translo-cated into peroxisomes as shown by its protection from theaction of external proteases. We constructed an immunologi-cally distinct cytosolic variant of thiolase by adding an influ-enza hemagglutin epitope tag to the N terminus of thetruncated thiolase. In a strain simultaneously expressing thefull-length, truncated, and epitope-tagged truncated thiolases,we demonstrated that normally untargeted thiolase subunitsare efficiently translocated into peroxisomes by dierizationwith full-length thila subunits. Even though truncated andepitope-tagged truncated thiolase subunits are translocatedinto peroxisomes in this strain, only the full-length thiolasesubunit can be coimmunoprecipitated with the epitope-taggedtruncated thiolase subunit from the peroxisomal matrix. Thisobservation suggests that interactions between thiolase sub-units are not disrupted during translocation.

Proteins destined for secretion, membrane integration, orassembly into organelles are sorted with high fidelity to theirrespective intracellular sites by virtue of targeting signalsencoded within the primary structures of the nascent poly-peptides themselves. The principal role of targeting signals isto mediate the engagement of the targeted protein withcomponents of a specific translocation machinery. There isabundant evidence that the translocation-competent confor-mations ofposttranslationally translocated proteins contain aminimum of tertiary structure (1-4). For peroxisomal matrixproteins, two types of peroxisomal targeting signal (PTS)have been identified. PTS-1 variants are conserved tripeptidemotifs found at the C termini of many peroxisomal matrixproteins. They are identical to or conserved variants of theprototypical Ser-Lys-Leu PTS-1 of firefly luciferase (5, 6).The N-terminal 11 amino acids of rat peroxisomal thiolaseconstitute a second type of PFS, designated PTS-2, whichresides on a cleavable presequence (7). We have recentlyidentified a similar uncleaved form ofPTS-2 at theN terminusof Saccharomyces cerevisiae peroxisomal thiolase (EC2.3.1.16) (8). Here we report that cytosolic forms of S.cerevisiae peroxisomal thiolase made by deletion of its PTS-2are translocated with high efficiency into peroxisomes bydimerization with full-length thiolase. Our results suggest anunusual translocation-competent protein conformation.

MATERIALS AND METHODSCulture Media. Yeast strains carrying plasmids were cul-

tured and maintained in YNBD [0.67% yeast nitrogen basewithout amino acids (YNB)/2% (wt/vol) glucose]. To inducethe expression ofgenes encoding peroxisomal proteins and toproliferate peroxisomes, cells grown in YNBD were pelleted,washed in sterile water, and transferred to either inductionmedium (0.67% YNB/0.5% yeast extract/0.5% Peptone/0.1% Tween 40/0.1% oleic acid/0.1% glucose; ref. 9) orYNOmedium (0.67% YNB/0.1% oleic acid/0.05% Tween 40/0.1%glucose). All YNB media were appropriately supplementedto satisfy auxotrophic requirements. For transformation,yeast strains were cultured in YEPD [1% yeast extract/2%(wt/vol) Peptone/2% glucose].

Transformation of Yeast. Plasmid DNA was introducedinto 20-pl aliquots of cells by electroporation with a BRLCell-Porator equipped with a voltage booster set to a resis-tance of 4 kfl and delivering a pulse with a field strength of=7.5 kV-cm-' (10). Cells were diluted into 100 p4 of 1 Msorbitol and spread onto selective YNBD agar plates.

Subcellular Fractionation. Cells grown in induction me-dium were harvested and converted to spheroplasts by di-gestion with Zymolyase 100T. Spheroplasts were harvested,resuspended in disruption buffer (5 mM Mes-KOH, pH6.0/0.6 M sorbitol/0.5 mM EDTA/0.1% ethanol/i mM phe-nylmethylsulfonyl fluoride; ref. 11), and homogenized with10 strokes of a motor-driven Potter-Elvehjem homogenizer.A postnuclear supernatant (PNS) fraction was prepared andrecentrifuged at 20,000 x g for 20 min to obtain a pelletenriched for peroxisomes and mitochondria and a superna-tant enriched for cytosol (12). The pellet was sometimesfurther fractionated on a Nycodenz gradient to yield apurified peroxisomal fraction (8, 11).

Protease Protection. In protease-protection experiments,500 ,ug of protein from the pellet was treated with 1% TritonX-100/1% deoxycholate for 15 min on ice, followed byprotease digestion with various amounts of trypsin or ther-molysin for 20 min on ice. Duplicate samples were treated asabove but in the absence of detergents. Digestions wereterminated by addition of hot SDS/polyacrylamide gel elec-trophoresis sample buffer and immediate boiling.

SDS/Polyacrylamide Gel Electrophoresis and ImmunoblotAnalyses. SDS/polyacrylamide gel electrophoresis was per-formed on 10%o or 7-15% gradient polyacrylamide gels (13).Immunoblot analysis was performed essentially as described(14). Nitrocellulose was blocked with 1% skim milk in 20mMTris-HCl, pH 7.5/150 mM NaCl/0.05% Tween 20. Totalthiolase was detected with a rabbit anti-thiolase polyclonalserum, and the antigen-antibody complexes were detectedwith either 175I-labeled protein A or alkaline phosphatase-

Abbreviations: fl-thiolase, full-length thiolase; HA, influenza hemag-glutinin; PNS, postnuclear supernatant; PFS, peroxisomal targetingsignal; t-thiolase, A1-16 thiolase lacking its N-terminal PIS-2.

10541

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

Proc. Natl. Acad. Sci. USA 91 (1994)

conjugated antibodies specific for rabbit IgG. Influenza he-magglutinin (HA) epitope-tagged truncated thiolase was de-tected with a mouse monoclonal antibody to the HA epitope(12CA5mAb), and the antigen-antibody complexes weredetected with alkaline phosphatase-conjugated antibodiesspecific for mouse IgG. Anti-SKL-reactive peroxisomal pro-teins were detected with a rabbit polyclonal serum (anti-SKL) directed against a synthetic peptide corresponding tothe C terminus of firefly luciferase (15).In Vivo Labeling of Cells and Immunoprecipitation. Strain

DL1[AN/HA] (Table 1, for a description of strains) wasgrown overnight in YNBD. Cells were harvested by centrif-ugation, washed with sterile water, transferred to YNOmedium, and grown for an additional 6 hr. Approximately 5x 108 cells were harvested, washed with water, and trans-ferred to 1 ml of YNO medium containing 500 uCi ofL-[35S]methionine (1 Ci = 37 GBq). Cells were labeled for 15min at 300C, washed in YNO/20 mM unlabeled methionine,and incubated for a further 20 min in this medium. The cellswere processed for subcellular fractionation as describedexcept that buffers contained a mixture ofprotease inhibitorsand 2mM methionine. The 20,000 x g supernatant and pelletwere brought to 10mM Tris HCl, pH 8.5/2.5 mM EDTA/0.5M NaCl and incubated for 30 min on ice. The solubilizedfractions were centrifuged at 100,000 x g for 30 min at 40C.The resultant supernatants were divided equally, mixed witheither 2.5 j. of rabbit anti-thiolase serum or 10 Aul of12CA5mAb ascites fluid, and incubated with gentle rockingfor 1 hr at room temperature. The immune complexes wereadsorbed to fixed Staphylococcus aureus cells (Pansorbin,Calbiochem) by gentle agitation at room temperature for 1 hr.The S. aureus cells were pelleted by centrifugation andwashed four times in the same buffer. The final pellet wasresuspended in 100 A1 of SDS/polyacrylamide gel electro-phoresis sample buffer and boiled for 5 min. The S. aureuscells were removed by centrifugation, and the immunopre-cipitated radiolabeled proteins were subjected to electropho-resis on 7-15% gradient polyacrylamide/SDS gels. The driedgels were exposed to a phosphor image storage screen.

RESULTSA Truncated Form of Thioase L Ig Its P7S-2 Is Trans-

loated into Peroxisomes When Costhesized with the Full-Length Thiolase (fl-thlole). Our initial efforts at localizingthe targeting signal of S. cerevisiae peroxisomal thiolaseinvolved the construction of a strain, DL1[AN], expressing aplasmid-borne mutated thiolase gene (POT); ref. 20) thatencodes a truncated form of thiolase lacking its N-terminal

16-amino acid PIS-2 (t-thiolase; ref. 8). Strain DL1[AN]carried a nuclear copy of the POT) gene encoding thefl-thiolase. The subcellular locations of the two forms ofthiolase were determined by fractionation of homogenizedspheroplasts into a 20,000 X g supernatant consisting essen-tially of cytosol and a pellet enriched for peroxisomes (12),followed by SDS/polyacrylamide gel electrophoresis andimmunoblot analysis with anti-thiolase serum (Fig. 1 A andB). As expected, fl-thiolase (Fig. 1B, open arrow) was foundprimarily in the organellar fraction. Surprisingly, about halfof t-thiolase (Fig. 1B, solid arrow) was also found in thisfraction. Thiolase, one of the leakiest components of theperoxisomal matrix (21-23), was broadly distributed whenthe organellar pellet was subfractionated on a discontinuousNycodenz gradient (Fig. 1 C and D). Nevertheless, thespecific occurrence of thiolase was 4-fold higher in theperoxisome-enriched fractions relative to the mitochondria-enriched fractions. Significantly, the distribution oft-thiolasematched that of fl-thiolase in all fractions.To determine whether t-thiolase and fl-thiolase were inside

the peroxisome or associated with the cytoplasmic face oftheperoxisomal membrane, samples ofthe organellar pelletweresubjected to digestion with proteases in the absence orpresence ofdetergents (Fig. 1E). Both forms ofthiolase wereprotease-sensitive only in the presence of detergents andexogenous protease, showing that both proteins were insideperoxisomes and protected by the peroxisomal membrane.

In contrast to the results seen with DL1[ANJ, t-thiolase wasfound in the cytosolic compartment in a strain (STUD[AN];ref. 8) with the POT) gene disrupted (Fig. 2A Upper, com-pare lanes 3 and 4 to lanes 7 and 8). The native conformationof a number of peroxisomal thiolases has been shown to behomodimeric (24-26). We hypothesized that t-thiolase mightbe recruited for import into peroxisomes by the formation ofheterodimers between newly synthesized monomers of t-thi-olase and fl-thiolase. Ideally, a random dimerization of trun-cated and full-length monomers expressed approximatelyequally would result in translocation of 50% of t-thiolase,with the remaining 50%o of t-thiolase forming homodimerstrapped in the cytosol.Truncated Thiolase Monomers Are Transocated into Per-

oxsomes Through the Formation of Heterodimers with thefi-Thiolase Monomers. To test our hypothesis, an immuno-logically distinct version of thiolase was constructed byadding two copies of the influenza HA epitope tag (18) ontothe N terminus of t-thiolase. Sorting of epitope-tagged t-thi-olase to the pellet fraction was dependent on the presence(strain DL1[HA]) of fl-thiolase (Fig. 2A Lower, compare

Table 1. Yeast strains used in this studyStrain Chromosomal genotype Plasmid genotype

DL1* ura3, his2, leu2 NoneDL1[AN] LEU2, CEN4, ARS6,t SCTA)-16*DL1[HA] LEU2, CEN4, ARS6, SCTA1-16(HA)2§DL1[AN+HA] LEU2, CEN4, ARS6, SCTAI-16/HIS3, CEN4, ARS6,

SCTAJ-16(HA)2DL1[AN+HA]HC LEU2, 2 MmO SCTAJ-16/HIS3, 2 mn, SCTAJ-16(HA)2STUD* ura3, his2, leu2, potl::URA3 NoneSTUD[AN] LEU2, CEN4, ARS6, SCTAJ-16STUD[HA] LEU2, CEN4, ARS6, SCTA)-16(HA)2*Strain previously described (16).tFor low-copy-number plasmid expression, thiolase gene constructs were made in pRS315 and pRS313 (17).tConstruction of the gene encoding thiolase with a 16-amino acid N-terminal deletion (t-thiolase) and the thiolase genedisruption strain (STUD) has been described (8).§Complementary synthetic oligonucleotides encoding the peptideMXEYDVPDYAA were annealed and ligated into the NcoI site ofpSCTAN (8). The recombinant used in these experiments contains two tandem repeats ofthe influenza HA epitopetag (underlined residues; ref. 18).IFor high-copy number plasmid expression, constructs were transferred to plasmids pRS423 and pRS425 (19).

10542 Biochemistry: Glover et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

Proc. Natl. Acad. Sci. USA 91 (1994) 10543

A MWm 1 2 3 CMwM 1 2 3 5 6 8 9 '10 I

106 1 06ao:.9. X

80 _ae..8C...s0xvx4@

49.5~ ~ 95

495. -~~~~~~2

322.5o-

2 / 5_ -0,:.. .......2.18 1855o

PXM 7MAIT

E

e-. ,-;: .-

N N,

K- K-A ;$1.$. -.

SF3;: S -s.

B0-

< I-..\

:.-,--.

.; Z,- -

F-'C. -.,- s - -,- cj 'C', s

TRYPSIN THERMOLYSIN

DE - - - - + + + - - - ++ +PRT 0 5 20 50 5 20 50 5 20 50 5 20 50

............. A _'-1UWS .......4I, 4-0

FiG. 1. Cytosolic form of thiolase is targeted to and translocatedinto peroxisomes in the yeast strain DL1[AN] expressing fl-thiolase.(A and C) SDS/polyacrylamide gels. (B and D) Correspondingimmunoblots with anti-thiolase serum. (A) Lanes: MWM, molecularmass standards (in kDa); 1, PNS; 2, 20,000 x g supernatant; 3, 20,000x g pellet. (C) Lanes: MWM, molecular mass standards (in kDa);1-11, fractions 1 (most dense) to 11 (least dense) ofa fractionation toenrich for peroxisomes. Open arrows, fl-thiolase; solid arrows,t-thiolase. PXM, fractions enriched for peroxisomes; MIT, fractionsenriched for mitochondria. (E) Protease protection of fl-thiolase andt-thiolase in the 20,000 x g pellet of DL1[AN]. DL1[AN] was grownin induction medium and fractionated. Equivalent portions of PNS,supernatant, and pellet (A) and supernatant and pellet (B) and equalvolumes (25 A1) of each gradient fraction (C and D) were subjectedto electrophoresis on 10% SDS/polyacrylamide gels. Immunoblotswere probed with rabbit anti-thiolase polyclonal serum (1:5000dilution). Antigen-antibody complexes were detected with m"I-labeled protein A followed by exposure to x-ray film at -700C. Inprotease-protection experiments (E), 500 pg of protein from pelletwas treated with 1% Triton X-100/1% deoxycholate (DET) for 15 minon ice (lanes +), followed by protease digestion (PRT) with 5, 20, or

50 pg of trypsin or thermolysin as indicated for 20 min on ice.Digestions were terminated by addition of hot SDS/polyacrylamidegel electrophoresis sample buffer and immediate boiling. Duplicatesamples were treated as above but without detergent (lanes -). Tenpercent of the reaction products from each reaction was analyzed onan immunoblot as described above.

lanes 1 and 2 to lanes 5 and 6), in a manner comparable to thatof t-thiolase.When both t-thiolase and epitope-tagged t-thiolase were

expressed together in the same cell with fl-thiolase (strainDL1 [AN + HA]), the amounts of translocated epitope-tagged t-thiolase and t-thiolase were quantitatively in agree-ment with the random formation of dimers among all avail-able subunits (Fig. 2B Upper, lanes 1-3). Expression ofepitope-tagged t-thiolase and t-thiolase from high-copy num-ber plasmids (DL1[AN + HA]HC) resulted in levels of thesemonomers 5-7 times that ofthe fi-thiolase monomer (Fig. 2B,lane 9). Under these conditions, virtually all fl-thiolase mono-mers should be dimerized with one or the other of theuntargeted monomers. As expected, the combined quantityof the two cytosolic forms of thiolase fractionating to thepellet was roughly equal to that of fl-thiolase (Fig. 2B, lane11). This result also suggested that individual subunits asso-ciated in translocation-competent dimers were translocatedinto peroxisomes with equal efficiency, since preference for

49.5_/ 4 5. 910 I1i32 1

A\N + H f.!C\N A`

FIG. 2. Cytosolic forms of thiolase are targeted to peroxisomesand assembled normally in the peroxisomal matrix only in strainsexpressing fl-thiolase. (A) Immunoblot analysis of 20,000 x g super-natant (anes S) and pellet (lanes P) fractions of strains STUD[HA](lanes 1 and 2), STUD[AN] (lanes 3 and 4), DL1[HAJ (lanes 5 and 6),and DL1[AN] (lanes 7 and 8). (Upper) Anti-thiolase serum (anti-SCT). (Lower) Anti-HA epitope antibodies (12CA5mAb). (B) Im-munoblot analysis of thiolase (Upper) and anti-SKL peroxisomalproteins (Lower) in DL1[AN/HA] (lanes 1-7) and DL1[AN/HA]HC(lanes 9-15). Open arrow, fl-thiolase; solid arrow, t-thiolase; shadedarrow, epitope-tagged t-thiolase. MWM, molecular mass markers (inkDa). Cells of various yeast strains were cultured, converted tospheroplasts, and fractionated. (A) Duplicate immunoblots wereprobed with either rabbit anti-thiolase serum (Upper) (anti-SCT,1:5000 dilution) or with a mouse monoclonal antibody to the HAepitope (Lower) (12CA5mAb, 1:15,000 dilution). Antigen-antibodycomplexes were detected with alkaline phosphatase-conjugated an-tibodies specific for either rabbit (1:75,000 dilution) or mouse(1:10,000 dilution) IgG. Color development was performed accordingto the manufacturer's instructions (Promega). (B) Cells of strainsDL1[AN + HA] Oanes 1-7) and DL1[AN + HA]HC (lanes 9-15) werefractionated into PNS, supernatant, and pellet. Each pellet wasresuspended and divided equally into three aliquots. One aliquot ofeach pellet was analyzed directly (lanes 3 and 11). A second aliquotwas treated with 1% Triton X-100 for 30 min on ice and centrifugedat 20,000 x g for 20 min to yield a supernatant (S, lanes 4 and 12) anda pellet (P, lanes 5 and 13). The third aliquot was treated with 10 mMTris HCl, pH 8.5/2.5 mM EDTA/0.5 M NaCl for 30 min on ice andcentrifuged at 100,000 x g for 30 min to yield a supernatant (S, lanes6 and 14) and a pellet (P, lanes 7 and 15). An equivalent portion ofeach fraction was loaded onto the gel to facilitate visual comparisonof band intensity. Duplicate blots were probed with anti-thiolaseserum (Upper) (anti-SCT; 1:5000 dilution) and an antiserum directedagainst a synthetic peptide corresponding to the C terminus of fireflyluciferase (Lower) (anti-SKL; 1:200 dilution). Antigen-antibodycomplexes were detected with 125I-labeled protein A and exposure ofthe blots to a Phosphorlmager storage screen (Molecular Dynamics).

the targeted subunit or selective exclusion of an untargetedsubunit would have been detected in the profiles of translo-cated thiolases.Treatment of mammalian peroxisomes with agents that

disrupt the peroxisomal membrane solubilizes components ofthe peroxisomal matrix to varying extents (21, 22). The lowsolubility of some components of the peroxisomal matrix

Biochemistry: Glover et al.

80 --,P-- .' -* -It - -NNW MW - -

or.

1111.11W.-

.42-. -i.,

49.5 --4- 4 -1,11,.-,, Z1.4-

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

Proc. Natl. Acad. Sci. USA 91 (1994)

indicates that they may be organized at the suborganellarlevel into a larger complex of proteins. When the organellarpellet from strain DL1[AN + HA] was treated with TritonX-100, -50%o of all the forms of thiolase was solubilized (Fig.2B Upper, lanes 4 and 5), typical of the thiolase in wild-typecells (data not shown). Treatment of the pellet with 10 mMTris HCl, pH 8.5/2.5 mM EDTA/0.5 M NaCl resulted in thecomplete solubilization and liberation to the supernatant ofall forms of thiolase (Fig. 2B Upper, lanes 6 and 7). Twoperoxisomal matrix proteins corresponding in molecularmass to a multifunctional enzyme (96 kDa; ref. 27) and theperoxisomal isozyme of citrate synthase (52 kDa; ref. 28)were detected by an antiserum generated against the Ser-Lys-Leu PTS-1 of firefly luciferase (15). Neither protein wassolubilized by Triton X-100/deoxycholate (Fig. 2B Lower,lanes 4 and 5 and lanes 12 and 13), indicating that theseproteins associate more strongly with the peroxisomal matrixthan does thiolase; however, like thiolase, these proteinswere completely solubilized by 10 mM Tris HCI, pH 8.5/2.5mM EDTA/0.5 M NaCI (Fig. 2B Lower, lanes 6 and 7 andlanes 14 and 15). Thus these results suggest that in the strainsDL1[AN + HA] and DL1[AN + HA]HC, the translocation ofat least two other peroxisomal proteins is not impaired bytranslocation of untargeted thiolase subunits and that theuntargeted thiolase subunits are incorporated like fl-thiolaseinto the normal suborganellar structure of the peroxisome.The Various Monomeric Forms of Thiolase Physically As-

sociate to Form Homo- and Heterodimers. To demonstratedirectly the physical association of the various forms ofthiolase, cells of strain DL1[AN + HA] expressing all threeforms of thiolase were labeled with L-[35S]methionine for 15min, chased for 20 min with unlabeled methionine, andsubjected to subcellular fractionation. The supernatant andpellet fractions were brought to 10 mM Tris HCl, pH 8.5/2.5mM EDTA/0.5 M NaCl to solubilize peroxisomal matrixproteins, and the thiolases were immunoprecipitated witheither anti-thiolase serum or a monoclonal antibody thatrecognizes the HA epitope tag (12CA5mAb). When theimmunoprecipitated complexes were analyzed by SDS/polyacrylamide gel electrophoresis, the pattern of distribu-tion for the three forms of thiolase (Fig. 3, lanes 1 and 2) wassimilar to that seen in immunoblots with anti-thiolase serum(cf. Fig. 2, lanes 2 and 3). 12CASmAb immunoprecipitatedthe epitope-tagged t-thiolase, fl-thiolase, and t-thiolase fromthe supernatant and the epitope-tagged t-thiolase and fl-thiolase, but not the t-thiolase, in a nearly 1:1 ratio from thepellet (Fig. 3, lanes 3 and 4, respectively), consistent with the

anti-SCT 12CA5mAb

SUP PEL SUP PFL66 --*

45 --*

36 --o

29 --.b

1 2 3 4

FIG. 3. Cytosolic forms of t-thiolase and fl-thiolase are physicallyassociated in the cytoplasm and in peroxisomes. Immunoprecipita-tion ofthiolase complexes from the cytosolic (SUP, lanes 1 and 3) andorganellar (PEL, lanes 2 and 4) fractions of DL1[I N/HA] with eitheranti-thiolase serum (anti-SCT) or antibodies specific for the HAepitope (12CA~mAb). Cells were labeled in vivo with L-[35S]methio-nine and immunoprecipitations were performed. Arrows indicatefl-thiolase (open), t-thiolase (solid), and epitope-tagged t-thiolase(shaded).

-e

idea that only dimers containing at least one targeting signalare imported into peroxisomes and that interactions betweenthiolase subunits are not disrupted during translocation.

DISCUSSIONt-thiolase and epitope-tagged t-thiolase associate with and aretranslocated into peroxisomes only when coexpressed withfl-thiolase. This result can be best explained by a model inwhich a randomly mixed population of dimers of all availablethiolase subunits is formed in the cytoplasm, and only dimerscontaining at least one targeting signal are then translocatedinto peroxisomes. When two distinct thiolase subunit typesare expressed at approximately equal levels, 50%16 of eachsubunit type will form heterodimers and 50% will formhomodimers. Fig. 2A shows that t50% ofthe t-thiolase or ofthe epitope-tagged t-thiolase, presumably as heterodimerswith fl-thiolase, was associated with the organellar pelletfraction, whereas 50% of each t-thiolase was cytosolic,presumably as homodimers with itself.When t-thiolase and epitope-tagged t-thiolase are ex-

pressed at high levels relative to the endogenous fl-thiolase,as in the strain DL1[AN + HA]Hc, essentially all of thefl-thiolase subunits will be heterodimerized with one or theother of the normally untargeted truncated thiolase subunits.Dissociation of subunits at some early step of translocationmight allow monomers lacking a PTS to be released back intothe cytoplasm. Such a scenario would be reflected as apronounced difference between the amount of fl-thiolase andthe total amount of truncated thiolases detected in thetranslocated fraction. The observation that the total amountof t-thiolase and epitope-tagged t-thiolase is roughly equal tothe amount of fl-thiolase in the translocated fraction (Fig. 2B,lane 11) suggests that subunits lacking their own targetingsignal are nevertheless efficiently translocated. Furthermore,the relative amounts of epitope-tagged t-thiolase and t-thio-lase in the imported fraction indicate that there is no prefer-ential interaction of fl-thiolase with either t-thiolase or epi-tope-tagged t-thiolase in regard to translocation.

Immunoprecipitation with the HA epitope-specific mono-clonal antibody was used to determine the composition ofdimers containing at least one epitope tag. Determination ofthe composition of epitope-tagged dimers in the translocatedfraction of DL1[AN + HA] can be used to discriminatebetween two possible modes of translocation. In one sce-nario, both subunits of a heterodimer are engaged simulta-neously by the translocation apparatus on the cytoplasmicside of the peroxisomal membrane but are subsequentlydissociated and translocated across the membrane indepen-dently. Monomers released into the peroxisomal lumenwould then be free to dimerize with any available subunits.In this case, epitope-tagged dimers should be composed of arandom mixture of epitope-tagged t-, t-, and fl-thiolase sub-units. However, epitope-tagged dimers in the translocatedfraction are composed only of epitope-tagged t-thiolase andfl-thiolase in an approximately 1:1 ratio (Fig. 3). This obser-vation suggests a model in which thiolase subunits aretranslocated as dimers, thereby preventing equilibration withother translocating thiolases in the peroxisomal matrix. How-ever, we cannot exclude the possibility that epitope-taggedt-thiolase-t-thiolase dimers are underrepresented in the im-munoprecipitations, either because of the inherent instabilityof such dimers or because epitope-tagged t-thiolase subunitsredimerize in a nonrandom fashion with fl-thiolase.

Oligomerization of glyoxysomal malate synthase (29) andCandida boidinii peroxisomal alcohol oxidase (30) into oc-tamers occurs only after the monomeric subunits have beensynthesized in the cytoplasm and translocated into theirrespective organelles. Cytosolic catalase is redistributed toperoxisomes when complementary Zellweger fibroblasts,

10544 Biochemistry: Glover et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

Proc. Natl. Acad. Sci. USA 91 (1994) 10545

which are deficient in peroxisome assembly, are fused.However, this redistribution is inhibited when the cells aretreated with 3-aminotriazole, which covalently modifies cat-alase and stabilizes its folded structure (31). These observa-tions suggest that peroxisomes cannot always import foldedoligomerized proteins. Moreover, oligomerization of pro-teins appears to inhibit import into mitochondria. Disruptionofa domain required for tetramerization ofa chloramphenicolacetyltransferase passenger domain was required for the invivo import into mitochondria of an apoiso-l-cytochromec-chloramphenicol acetyltransferase fusion (32). Similarly,deletion of an internal segment of the (3 subunit of mitochon-drial F1-ATPase needed for tetramerization eliminates therequirement for ATP in in vitro import (33).

In contrast, purified octameric alcohol oxidase from per-oxisomes of the yeast Pichia pastoris was rapidly incorpo-rated into punctate structures after microinjection into mam-malian cells (34). Some, but not all, of these punctatestructures costained with anti-catalase antibodies. Similarly,fusion of some complementary Zellweger fibroblast linesresulted in the very rapid assembly of preexisting activecatalase (i.e., heme-containing tetramers) into peroxisomes(35). Although there is no direct evidence that the rapidimport of these proteins does not involve at some point theunfolding and disassembly of subunits that are then translo-cated independently, the possibility that preassembled oligo-mers are translocated into peroxisomes cannot be ruled out.

Interestingly, the translocation into peroxisomes ofhumanserum albumin decorated with a number of covalently linked12-amino acid peptides containing the PTS-1 tripeptide Ser-Lys-Leu has been observed in both microinjected (36) andstreptolysin 0-permeabilized (37) mammalian cells. Thetranslocon of the mitochondrion apparently can also accom-modate such "branched" polypeptides, as shown by theimport of cytochrome c oxidase chemically cross-linked to amitochondrially targeted dihydrofolate reductase-fusion pro-tein construct (38).The aspect of translocation of such complexes that is

pertinent to the modeling of thiolase translocation is that atleast two polypeptide segments are obliged to occupy thetranslocation machinery simultaneously. Since, to ourknowledge, nothing is currently known about the structuralrequirements for dimerization ofthiolase monomers, one canspeculate that the bulk of the thiolase dimer could be un-folded without disrupting the interface between monomers.Nevertheless, for monomers to remain associated throughoutthe translocation event, at least the segments of both mono-mers containing the dimer interface must cooccupy thechannel of the translocon.

This work was supported by a grant from the Medical ResearchCouncil ofCanada (MRC) to R.A.R. J.R.G. is the recipient ofaMRCStudentship. D.W.A. is the recipient of a MRC Scholarship. R.A.R.is a MRC Scientist.

1. Verner, K. & Schatz, G. (1988) Science 241, 1307-1313.2. Pfanner, N. & Neupert, W. (1990) Annu. Rev. Biochem. 59,

331-353.3. Craig, E. A., Kang, P. J. & Boorstein, W. (1990) Antonie van

Leeuwenhoek 58, 137-146.4. Hard, F.-U., Martin, J. & Neupert, W. (1992) Annu. Rev.

Biophys. Biophys. Chem. 21, 293-322.

5. Subramani, S. (1992) J. Membr. Biol. 125, 99-106.6. Aitchison, J. D., Nuttley, W. M., Szilard, R. K., Brade,

A. M., Glover, J. R. & Rachubinski, R. A. (1992) Mol. Micro-biol. 6, 3455-3460.

7. Swinkels, B. W., Gould, S. J., Bodnar, A. G., Rachubinski,R. A. & Subramani, S. (1991) EMBO J. 10, 3255-3262.

8. Glover, J. R., Andrews, D. W., Subramani, S. & Rachubinski,R. A. (1994) J. Biol. Chem. 269, 7558-7563.

9. Erdmann, R., Veenhuis, M., Mertens, D. & Kunau, W.-H.(1989) Proc. Natl. Acad. Sci. USA 86, 2432-2436.

10. Nuttley, W. M., Brade, A. M., Gaillardin, C., Eitzen, G. A.,Glover, J. R., Aitchison, J. D. & Rachubinski, R. A. (1993)Yeast 9, 507-517.

11. Lewin, A. S., Hines, V. & Small, G. M. (1990) Mol. Cell. Biol.10, 1399-1405.

12. Aitchison, J. D., Murray, W. W. & Rachubinski, R. A. (1991)J. Biol. Chem. 266, 23197-23203.

13. Laemmli, U. K. (1970) Nature (London) 227, 680-685.14. Burnette, W. M. (1981) Anal. Biochem. 112, 195-203.15. Gould, S. J., Krisans, S., Keller, G.-A. & Subramani, S. (1990)

J. Cell Biol. 110, 27-34.16. Van Loon, A. P. G. M., van Eijk, E. & Grivell, L. A. (1983)

EMBO J. 2, 1765-1770.17. Sikorski, R. S. & Hieter, P. (1989) Genetics 122, 19-27.18. Kolodziej, P. A. &Young, R. A. (1991) Methods Enzymol. 194,

508-519.19. Christianson, T. W., Sikorski, R. S., Dante, M., Shero, J. H.

& Hieter, P. (1992) Gene 110, 119-122.20. Igual, J. C., Matallana, E., Gonzalez-Bosch, C., Franco, L. &

Perez-Ortin, J. E. (1991) Yeast 7, 379-389.21. Hayashi, H., Hino, S. & Yamasaki, F. (1981) Eur. J. Biochem.

120, 47-51.22. Alexson, S. E. H., Fujiki, Y., Shio, H. & Lazarow, P. B.

(1985) J. Cell Biol. 101, 294-304.23. Thompson, S. L. & Krisans, S. K. (1990) J. Biol. Chem. 265,

5731-5735.24. Zeelen, J. P., Wierenga, R. K., Ermann, R. & Kunau, W.-H.

(1990) J. Mol. Biol. 215, 211-213.25. Kurihara, T., Ueda, M. & Tanaka, A. (1989) J. Biochem.

(Tokyo) 106, 474-478.26. Miyazawa, S., Furuta, S., Osumi, T., Hashimoto, T. & Ui, N.

(1981) J. Biochem. (Tokyo) 90, 511-519.27. Hiltunen, J. K., Wenzel, B., Beyer, A., Erdmann, R., Fossa,

A. & Kunau, W.-H. (1992) J. Biol. Chem. 267, 6646-6653.28. Rosenkrantz, M., Alan, T., Kim, K. S., Clark, B. J., Srere,

P. A. & Guarente, L. P. (1986) Mol. Cell. Biol. 6 4509-4515.29. Kruse, C. & Kindl, H. (1983) Arch. Biochem. Biophys. 223,

629-638.30. Goodman, J. M., Scott, C. W., Donahue, P. N. & Atherton,

J. P. (1984) J. Biol. Chem. 259, 8485-8493.31. Middelkoop, E., Strijland, A. & Tager, J. M. (1991) FEBS Lett.

279, 79-82.32. Nye, S. H. & Scarpulla, R. C. (1990) Mol. Cell. Biol. 10,

5753-5762.33. Chen, W.-J. & Douglas, M. G. (1988) J. Biol. Chem. 263,

4997-5000.34. Walton, P. A., Gould, S. J., Rachubinski, R. A., Subramani,

S. & Feramisco, J. R. (1992) J. Cell Biol. 118, 499-508.35. Brul, S., Wiemer, E. A. C., Westerveld, A., Strijland, A.,

Wanders, R. J. A., Schram, A. W., Heymans, H. S. A.,Schutgens, R. B. H., van den Bosch, H. & Tager, J. M. (1988)Biochem. Biophys. Res. Commun. 152, 1083-1089.

36. Walton, P. A., Gould, S. J., Feramisco, J. R. & Subramani, S.(1992) Mol. Cell. Biol. 12, 531-541.

37. Wendland, M. & Subramani, S. (1993) J. Cell Biol. 120,675-685.

38. Vestweber, D. & Schatz, G. (1988) J. Cell Biol. 107, 2045-2049.

Biochemistry: Glover et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0