Micelle-assisted Protein Folding

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 267, No. 9, Issue of March 25, pp. 5811-5816,1992 Printed in U.S A .

Micelle-assisted Protein Folding DENATURED RHODANESE BINDING TO CARDIOLIPIN-CONTAINING LAURYL MALTOSIDE MICELLES RESULTSIN SLOWER REFOLDING KINETICSBUT GREATER ENZYME REACTIVATION* (Received for publication, November 12, 1991)

Gustavo Zardeneta and Paul M. Horowitz$ From the Department of Biochemistry, - . Uniuersitv - of, Texas Health Science Center, San Antonio; Teras 78284-7760

Unfolded (inactive) rhodanese (thiosu1fate:cyanide sulfurtransferase, EC 2.8.1.1) can be reactivated in the presence of detergents, e.g. lauryl maltoside (LM). Here, we report the reactivationof urea-unfolded rhodanese in the presence of mixed micelles containing LM and the anionic mitochondrial phospholipid, cardiolipin (CL).Reactivation times increased as the number of CL molecules/micellewas increased. A maximum of 94%of the activity wasrecovered at 2.2 CL/micelle. Only 71%of the activitywas recovered in the absence of CL. The major zwitterionic mitochondrial phospholipid, phosphatidylcholine (PC), had no effect on the LM-assisted reactivation of rhodanese. Size exclusion chromatography showed that denatured, but not native, rhodanese apparently binds to micellar amounts of LM andCL/LM,but not to PC/LM micelles. The lifetime of the enzyme-micelle complex increased with the number of CL molecules/micelle. Furthermore, chromatographic fractions containing micelle-bound enzyme had no activity, while renatured rhodanesecontaining fractions were active.These results suggest that transient complexes form between enzyme and both LM and CL/LM micelles, and that this complex formation may be necessary for reactivation. For CL/ LM micelles, interactions may occur between the positively charged amino-terminal sequence of rhodanese and the negatively charged CL phosphate. Finally, this work shows that there are similarities between “micelle-assisted” and chaperonin-assisted rhodanese refolding.

The in vitrodenaturation andreactivation of the mitochondrial enzyme rhodanese (thiosu1fate:cyanide sulfurtransferase, EC 2.8.1.1) has been extensively studied. It has been shown that detergents, present at concentrations above their respective critical micelle concentration (CMC)’ (I), can assist in the proper refolding of the enzyme, yielding high (about 90%) activity (2). Recently, conditions have been described

* This research was supported by Research Grants GM25177 and ES05729 from the National Institutes of Health, and Welch Grant AQ723. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed Dept. of Biochemistry, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, TX 78284-7706. Tel.: 512-567-3737; Fax: 512567-2490. The abbreviations used are: CMC, critical micelle concentration; LM, lauryl maltoside (dodecyl-P-D-maltoside); CL, cardiolipin (diphosphatidylglycerol); PC, phosphatidylcholine; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

whichmaximize the spontaneous (unassisted) refolding of rhodanese to approximately 80% (3). A possible way in which rhodanese may be foldedcorrectly in vivo is by the assistance of molecular chaperonins. We have recently reported the in vitro reactivation of denatured rhodanese by the chaperonins cpn60 and cpnlO (4), the bacterial analogs of the eukaryotic heat shock proteins hsp60 (5) and hsplO (6), respectively, which are themselves molecular chaperones. Theseapproaches to rhodanese reactivation offer insight intothe mechanisms and kinetics of protein folding, yet the entire process is still poorly understood. For instance, an outstanding question, we address in this report, is the following: why does the minimum concentration of LM required for detergent-assisted refolding of rhodanese coincide with its CMC (0.08 mg/ml)? Rhodanese is coded by the nuclear genome, synthesized in vivo in thecytoplasm, and then transportedto themitochondrial matrix throughthe inner and outer mitochondrial membranes. An interesting feature of rhodanese is that the only processing of the translation product is removal of the initiating methionine (7).Thus,the NHZ-terminal leader sequence, a 23-amino acid stretch which contains several positively charged residues (8), is left intact in the final enzyme. A current hypothesis is that thechaperonin hsp70 transports the newly synthesized rhodanese from the ribosome to the mitochondria (9). Once inside the mitochondria, it is believed that the chaperonins hsp6O and hsplO aid in the folding of the enzyme into an active form (6). Currently,it is not known whether the phospholipids in the mitochondrial membrane have an active role in the transport andfolding of rhodanese. Since it is believed that proteins must have an unfolded or nonglobular conformation to traverse biological membranes (lo), we decided to investigate the possible interactions of unfolded and native rhodanese with the eukaryotic mitochondrial phospholipid, cardiolipin (CL). There areseveral reasons why cardiolbin, themajor anionic and the third most prominent phospholipid component of the mitochondrial membrane ( l l ) , may interact with rhodanese. The net negative charge of the diphosphatidylglycerol head of CLmay interact electrostatically with the positively charged NH2terminus of rhodanese. CLhas been found primarily in mitochondrial and bacterial membranes (11). Previous studies have shown that CL interacts with, and modulates, mitochondrial proteins (e.g. CL is required for optimal activity of cytochrome c oxidase (12). Inthis report, we show that unfolded rhodanese binds transiently to CL-containing LM micellesand that this interaction modulates the refolding/reactivation of the enzyme. Additionally, we find evidence for the formation of a transient complex between unfolded rhodanese and LMmicelles during reactivation. We explore the possibility that the former may

581 1

Micelle-assisted Protein Folding

5812

be due to electrostatic interactions which may be unique to Dicyanohemin-stained micelles were prepared by mixing LM with either CL or negatively chargedphospholipids, while the latter dicyanohemin at concentrations of 1 mg/ml and 50 pM, respectively, in a final volume of 400 pl. may be due to hydrophobic binding. PolyacrylamideGel Electrophoresis-Discontinuous minigels (HoeEXPERIMENTAL PROCEDURES

Materials-Cardiolipin (from beef heart) andphosphatidylcholine purchased from Avanti Polar Lipids, urea (electrophoresis grade) was purchased from Bio-Rad, and lauryl maltoside was obtained from Behring Diagnostics. All other reagents were of analytical grade. Rhodanese PreparationlAssay-Rhodanese was prepared from bovine liver as previously described (13), stored at -70 "C as a crystalline suspension in 1.8 M ammonium sulfate and 1 mM thiosulfate. Purified rhodanese were kept a t 4 "C in the presence of 0.2 M Na phosphate (pH 7.5). Rhodanese concentration was determined using a molecular mass of 33 kDa and a value of A (0.1%, 280 nm) = 1.75. Enzymatic activity was assayed using the substrate thiosulfate and quantified by spectrophotometric measurement (at 460 nm) of the colored complex formed by the product, thiocyanate, with ferric ions

fer Scientific Instruments) were prepared as described (16), using a 3.5 and 12% acrylamide in the stacking and resolving gels, respectively. Protein bands were visualized by silver stain.

(L-CX lecithin from egg) were

RESULTS

Effect of Cardiolipin-containing Lauryl Maltoside Micelles on Rhodanese Reactivation-Denatured rhodanese was found to be reactivated to approximately 71% (relative to a native enzyme under similar conditions) or 69%(relative to its initial activity) in the presence of 1 mg/ml LM, phosphate buffer, (3-mercaptoethanol, and thiosulfate as described under "EXperimental Procedures.'' Maximum activity was maintained between t = 5 and 24 h, while 50% reactivation (tLh)occurred a t approximately 40 min (Fig. 1). When rhodanese was reac(14). tivated under similar conditions, but in the presence of CLDenaturation and Reactiuation-In all experiments, except those containing LM micelles,a CL-dependent increase in the t2,+ of involving the gel filtration of CL/LM micelle-reactivated rhodanese, activation was seen, i.e. higher amounts of CL/micelle rethe enzyme was denatured as follows. Rhodanese was incubated at a sulted in longer t%. When the amount of CL/micelle was concentration of 0.6 mg/ml in 100 pl of a solution containing 9 M urea and 14 mM 8-mercaptoethanol at 24 "C for 30 min. Under these between 0.4 and 4.4, maximal activity was obtained at t = 24 conditions, residual enzymatic activity was found to be less than 1% h. This time increased to 48 and 66 h when the CL/micelle of original activity. Native enzyme, which served as a control, was were 7.3 and 10.2, respectively. Fig. 1 shows the activity of diluted to 0.6 mg/ml in 0.1 M Na phosphate (pH 7.5) prior to use in rhodanese versus time upon reactivation in the presence of the reactivation reactions. Reactivation of enzyme, for use in non-gel various CL/micelle solutions. It shows that a solution confiltration experiments, was initiated by dilution of 0.6 mg/ml of denatured rhodanese to a concentrationof 0.03 mg/ml into a solution taining micelles with an average number of 2.2 CL/micelle containing 0.1 M Na phosphate (pH 7.5), 50 mM Na thiosulfate, 0.2 gave the best enzyme activity recovery (94%relative to native enzyme incubated under similar conditions and 93.3%relative M 0-mercaptoethanol, 1 mg/ml LM, and the indicated amounts of phospholipids, in a final volume of 100 pl, followed by incubation a t to its original activity). This activity was 34% higher than 24 'C as designated in the text. In all gel filtration experiments, that seen when reactivation was done in the absence of CL. except when CL was involved, the reactivation mixture composition Fig. 2 summarizes the optimum activity recovered in the was as described above, except that the totalvolume was 900 pl. It is presence of various amounts ofCL. It clearly shows that important to note that the LM micelles, which contained a variable rhodanese reactivation was CL-dependent up to anoptimum number of phospholipids per micelle, were formed by dilution of CL/ LM or PC/LM stock solutions into thereaction mixture, followed by CL concentration. At higher concentrations of CL, less reacbrief vortexing, prior to theaddition of denatured enzyme. tivation was seen; probably due to thedecay of enzyme activity When denatured rhodanese was reactivated with CL/LM micelles over time as it is held longer by the micelle in an unfolded and then chromatographed by gel filtration, denaturation of enzyme conformation, and perhaps due to irreversible oxidation (17). was as described above, except that denaturation was carried out a t Under the above reactivation conditions and in the absence a protein concentration of 1 mg/ml, and thereactivation mixture had of CL and LM, there was a maximum of 4% regain of activity, an enzyme concentration of 0.2 mg/ml in a total volume of 900 p1. Mixed Micelle Preparations-Two stock solutions of LM micelles, apparently due to spontaneous reactivation. The substrate, each of which contained either 15 CL molecules/micelle or 15 PC thiosulfate, was found to be necessary for this activation, molecules/micelle were prepared. The former was prepared by mixing since negligible amounts of activity were detected in samples 2 mgof CL in 4 ml of a 1 mg/ml LM solution, and the latter by containing both CL and LM but not thiosulfate. Cardiolipindissolving 1 mgof PC in 5.05 ml of a 1 mg/ml LM solution. Both containing LM micelles did not have a significant effect on solutions were sonicated for 2 min and then stored a t -70 "C until needed. It was assumed that a mixedmicelle retaineda similar geometry as apure LM micelle and, therefore, the value of 150 monomers/micelle of LM (15) was used in the calculations for the preparation of these micelles. However, for purposes of calculation, we assumed that the prepared LM/lipid mixed micelles had 150 "hydrophobic tails"/micelle, rather than 150 monomers/micelle. This rationale was employedto accommodate the fact that a CL monomer has 4 tails, while each LM has only one tail. Thus, in the case of CL/ LM micelles, the mixed micelles wereconsidered to be constructed of 150 tails; 60 from CL and 90 from LM. Hence, there were 15 CL and 90 LM molecules/micelle. Similar geometrical assumptions were made in considerations of the PC/LM micelles, except that in these calculations we took into account that each PC monomer has 2 hydrophobic tails. Gel Filtration Chromatography-Size exclusion chromatography was carried out at room temperature in a column (0.7 X 47 cm) containing Sephadex G-75 (Pharmacia LKB Biotechnology Inc.) equilibrated with a solution containing 0.1 M Na phosphate (pH 7.5) and 1 mg/ml LM. In all experiments, the sample loaded was 400 pl, the flow rate was set at3 ml/h with the aid of a peristaltic pump,and 425-p1 fractions were collected. The absorbance (at 275 nm) of the eluate was monitored with an LKB (2138 Uvicord S) flow-through detector. The column's void volume ( Vo)was determined to he 6.3 ml by using a Blue Dextran 2000 (Pharmacia) solution. Enzyme fractions were assayed for rhodanese activity and analyzed by SDS-PAGE.

o.*t

Time (h)

FIG. 1. Reactivation kinetics of rhodanese in the presence of cardiolipin. Rhodanese was denatured in urea as outlined under "Experimental Procedures," and then3 pg ofthe enzyme were diluted into 100 pl of a solution containing the reactivation buffer (see "Results") in the presence of the following amounts of CL/micelle: 0 (a), 2.2 (A),4.4 (A), 7.3 (O),and 10.2 (0).The activity of the enzyme was quantified by assaying a 5-pl aliquot from the reactivation mixture at theindicated times (abscissa). Time length refers to the time between dilution of denatured rhodanese into reactivation buffer and start of activity assay. The ordinate represents the absorbance at 460 nm, which measures the presence of ferric thiocyanate; the reaction product in the KCN/thiosulfate rhodanese activity assay.


5813

Binding of Rhodanese to Micelles-Previous work in our laboratory showed that various detergents, including zwitterionic or nonionic detergents (e.g. LM), could aid in the reactivation of denatured rhodanese (1).Interestingly, the minimum concentration of detergent needed for this effect coincided with the detergent‘s CMC. These previous observations suggested that denatured or partially folded enzyme might interact with an amount of detergent equivalent to whole micelles, rather than with individual detergent molecules. To Cordiolipin/rnicelle test this hypothesis, we performed a series of size-exclusion FIG. 2. Maximum rhodanese reactivation in the presence of chromatography experiments using a Sephadex G-75 matrix various amounts of CL/micelle. The reactivation experiments and (see “Experimental Procedures”) in order to detect micelle activity monitoring were performed as outlined under “Results” and binding by either native or denatured rhodanese at different in thelegend to Fig. 1. Percent reactivation was calculated by dividing times during LM, CL/LM, and PC/LM reactivation. In all the experimentally obtained A,, value by 0.90; the theoretically gel filtration experiments the columns were equilibrated with expected value for a similar amount of 100%active rhodanese assayed a similar time. The theoretically expected amount was within 5% of 1 mg/ml LM; a concentration above the detergent’s CMC. Chromatography of Micelles and Native Rhodanese-The the experimentally determined amount of native rhodanese incubated in the presence of CL and the other reactivation components. The elution volume (V,) of LM micelles determined by chromaabscissa shows the number of CL monomers/micelle in the reactiva- tography of dicyanohemin-stained LM micelles was found to tion mixture. be6.94 ml,while its K,, was = 0.057. The V, of native rhodanese, when chromatographed in a solution containing 0.61 I 0.2 mg/ml enzyme, 15 CL/LM micelles, and the usual reactivation mixture components was found to be 9.06 ml, and its K., was 0.244. Native rhodanese which was loaded onto the same column, in a solution containing only column equilibration buffer, eluted at the same position as that loaded in the presence of CL/LM micelles and reactivation components. This suggests that native rhodanese does not interact with LM or CL/LM micelles. Elution profiles of LM micelles and native rhodanese are shown in Fig. 4. The small peak at Vo Time ( h ) in the rhodanese sample is probably due to light scattering by FIG. 3. Reactivation kinetics of rhodanese in the presence the CL/LM micelles and/or entrapmentof P-mercaptoethanol of phosphatidylcholine. The experiments were done as indicated molecules in these micelles. under “Results” and aliquots assayed as explained in the legend to Chromatography of Denatured Rhodanese during CL/LM Fig. 1. Shown are plots where the number of PC/micelle are as Reactivation-Rhodanese was denatured at 1mg/ml and reacfollows: 0 (O), 2 (A),4 (A), 6 (O), and 10 (0). tivated in a total volume of 900 pl at a concentration of 0.2 native rhodanese. Native rhodanese in the reactivation reac- mg/ml in the presence of CL/LM micelles (4.4 CL/micelle) tion mixture, in the absence of CL or in the presence of 10.2 and reactivation components (see “Experimental ProceCL/micelle, showed a maximum increase in activity of 14% dures’’). Recovery of enzymatic activity in the reactivation mixture was monitored for 36 h. It was found that after 30 after 2 h followedby a steady but slow decrease in basal activity over time. At 24 h, the activities of native rhodanese min, only 6% of the total enzyme was active, whereas a in the reactivation solutions, in the absence or presence of maximal reactivation of 31% was detected at 29 h. It should be noted that the maximal reactivation obtained was not as CL, were within 5% of each other. Effect of Phosphutidykholine on the Reactivation of Dena- high (94%) as in the above experiments; this was due to the tured Rhodanese-We hypothesized that perhaps the negative higher concentration of rhodanese in the reactivation mixture charges in the CL head groups were interacting electrostati- which has previously been shown to favor aggregation and cally with the positively charged amino-terminal sequence of reduce the yields of reactivation. After 30 min in the reactirhodanese, thus, trapping the molecule and thereby prevent- vation solution, 400 pl were chromatographed on Sephadex ing aggregation with other unfolded enzymes. As generally believed, prevention of aggregates will lead to higher yields of proper refolding, and thus, better recovery of enzymatic activity (18,19).We decidedto see whether a neutralphospholipid could give a similar reactivation pattern. We chose the zwitterionic phospholipid phosphatidylcholine (PC),the most abundant mitochondrial membrane phospholipid, for this study. We tested the effect in a similar reactivation mixture, using PC-containing LM micelles in a range from 0.25 to 10 PC monomers/micelle. The resulting reactivation curves were Froction number nearly identical for all samples except the one with 10 PC/ micelle. We found no significant difference in thet.,? of any of FIG. 4. Chromatographic profile of LM micelles and native the samples assayed (Fig. 3). Additionally, we found no effect rhodanese. Shown is the Sephadex G-75 chromatographic profile of inthe maximum percent reactivation reached in samples dicyanohemin-stained LM micelles (0)and native rhodanese (0).In containing between 0 and 6 PC/micelle. The sample contain- each case, a 400-pl solution was loaded onto the column and 425-pl fractions collected. The abscissa represents the fraction number and ing 10 PC/micelle attained only 77% of the activity reached the ordinate is the absorbance at 275 nm. Details of the chromatogby the sample without any PC.Possible reasons for this result raphy and the preparation of the stained micelles are given under are discussed below. “Experimental Procedures.”

5814

Folding

ProteinMicelle-awisted

G-75, asdetailed above, andasecond 400-p1 aliquot was tal Procedures”), and then chromatographed aliquots of this mixture on the same size exclusion column as above, at 30 chromatographed 29 h after dilution of enzyme into reactivation solution. The elution profile obtained after 30 min of min and 22 h after the startof reactivation. The reactivation of the enzyme mixture was monitored, andit was found that at30 min there reactivation (Fig. 5 A ) showed that the majority eluted at the void volume (6.3 ml), indicating that this enzyme was 37% reactivation, whereas maximal (68%) reactivation was presumably boundto themicelles. SDS-PAGE of column was achieved a t 22 h. In these experiments, the concentration fractions verified that >90% of rhodanese was present in the of rhodanese in the reactivation mixture was only 0.03 mg/ Vo fractions (Fig. 6 A ) . After 29 h of reactivation, the elution ml, compared to0.2 mg/ml in the CL experiments, due to the profile (shown in Fig. 5 R ) showed two similar peaks; the first reduced solubility of the unfolded enzyme in the absence of at VI)and the second at the V,. of native rhodanese. However, CL (see below). At this rhodanese concentration, it was not SDS-PAGE analysis of column fractions (Fig. 6 R ) showed possible togeta reliable elution profile. However, colthat virtually all of the rhodanese was present in the peak lected fractions were analyzed by SDS-PAGE, followed hy corresponding to nativeenzyme. Only traces of protein were silver staining. Fig. 7 A shows the protein analysis by SDSdetected at thevoid volume fractions. Thus, theobserved V, PAGE of fractions from the chromatograms after30 min and peak was mainly due to light scattering from CL/LM micelles 22 h. At 30 min, approximately 25% of rhodanese eluted with or trace amountsof other chemicalswhich absorb lighta t 275 the micelles a t VI), while the rest eluted at the V,. for native nm, such as P-mercaptoethanol, trapped in these micelles. rhodanese. After 22 h, all of the rhodmese eluted as micelleThe fractionsfrom the above experiments were assayed for free rhodanese (Fig. 7 R ) . These results indicate that denaresidual enzymatic activity. Fig. 5 shows that, in both cases, tured rhodanese binds transiently to LM micelles, although all of the activity was found a t V, = 9.06. This indicates that for a shorter time than to CL/LM micelles. Rased on the micelle-hound enzyme was completely inactive and that en- monitoring of activity of the reactivation mixture at 30 min, zymatic activitywas recovered only afterrelease of rhodanese one might expect a greater amount of rhodanese at V,. However, the observed results are expected if the half-time for from micelles (i.e. in micelle-free fractions). Chromatography of Denatured Rhodanese during Reactiua- reactivation in thepresence of LM (about 2.5 h ) andthe tion with LM-To determine whether rhodanese might hind amount of time it takes for the micelle-bound rhodanese to t o LM micellesin theabsence of CL, we denaturedand elute from the column (about 3.5 h) are taken into considerreactivated the enzyme in a 900-pl solution (see “Experimen- ation. In agreement with the findings involving the CL/gel filtration experiments above, residual enzymatic activity was found only in the fractions eluting at the V,. for native rhodanese and no detectable activity was found in V,, fractions (data not shown). Chromatography of Denatured Rhodancw during PC/LM Reactivation-We investigated the possibility that PC/LM micelles might also hind rhodanese. We prepared denatured rhodanese and reactivated theenzyme in the same manner as detailed above for reactivation using only LM. After 30 min 1 13 16 19 22 25 2.9 of reactivation, a 400-pl aliquot was applied to the Sephadex Fraction number G-75 column. At this point,assaying of the reactivation mixture showed that 38% of rhodaneseactivityhad been In I recovered. A maximal recovery of 72% was obtained after 4.5 x ._ h. Analysis of column fractions by SDS-PAGE, followed by .-> E? silver staining (Fig. 7 C ) revealed that all of the rhodanese W (i.e. notraces of protein were elutedasnativerhodanese >. ._ “ detected in V, fractions). Analysis of rhodanese enzymatic -0W n activity in the column fractionsshowed that all of the activity was present in the same protein-containing fractions.Again. no activity was detected in V,, fractions. c a

Fraction number

FIG.5. Chromatographic profiles of rhodanese at different times duringCL/LM reactivation. Urea-denatured rhodanese was reactivat,ed in the presence of CI,/LM micelles as descrihed under “Results.” A 400-pl aliquot was chromat.ographed on SephadexG-75, as outlined under “Experimental Procedures,” 30 min ( A ) and 29 h ( R )after dilution into reactivation mixture. Shownis the absorbance a t 275 nm (0)and the rhodanese activity ( 0 )of collected fractions.

- vo

1

A A

2

3

.x

2

“

5 - 6

* I ,cl

(?k)

7

P

L“

‘1

1‘

-

C

FIG.7. Protein analysis of gel filtration fractions from L M and PC/LM reactivation. Urea-unfoltled rhodanese was react ivated in the presence o f I,M only micelles as indicated in the text. Then, 400 pl were chromatographed on Sephadex ( ; - 7 5 at 30 min and 22 h 1 2 3 4 5 6 7 8 9 1 0 (see “I.:xperimentnl I’roceafter dilution into reactivation mixture A .dures”). Forty-pl aliquots from collected fractions were analyzed hy A 0 “ SDS-PAGE, followed hy silver staining. Shown in lnnm I - 1 0 are the FIG.6. Protein analysis of gel f i l t r n t i o n fractions from CIA/ stained rhodanese hands pertaining tn fractions 17-26. respectively, LM-rhodanese reactivation. J;orty-pl aliqunts from fractions colfrom the chromatography performed: 30 min ( A ) and 2’2 h ( I { ) after 5 were analyzed by SDS- reactivation. The protein hands in ( ’ represent analysis of rhndnnew lected from the chromatographies in Fig. PAGE in 12% polyacrylamide gels, followed hy silver staining. Shown in fractions from a similar experiment. except that the reartivat inn are the stained rhodanese hands pertaining to Fig. 5A ( A ) and Fig. was done in the presence I’C/IAl of micelles and the chromatography 5 H (N).1,onP.q 1-10 represent column fractions 17-26, respect.ively. was performed 30 min after the onsetof reactivation (see “ l f w r l l t s “ t . Indicated are the experiment,ally determined void volume and elutionIndicated are fractions representing the vnidvnltrme and thr. pllltinn volume of native rhodanese ( V , ( R h ) ) . volume of native rhodanese ( V,. ( R h ) .

vo

I

. V,

(Rh)


5815

between enzymatic reactivation and LM’s CMC is not coincidental. Gel filtration studies of rhodanese, in theprocess of being refolded by LM micelles, show an apparent binding of inactive protein species and a subsequent release of enzyme, which is active when assayed. The protein may not necessarily be active immediately after release. The binding of enzyme to micelles may be required to keep the denatured protein from aggregating with other denaturedproteins. The nature of this interaction is most likely hydrophobic since LM is a nonionic detergent. There probably are hydrophobic interactions between denatured rhodanese and single LM molecules at LM concentrations under the CMC, however, these interactions must not suffice to prevent aggregation since no reactivation is seen under these conditions. These observations suggest that there are probably two or more hydrophobic sites on unfolded rhodanese which need to be anchored to a relatively large structure (in this case, a micelle) which holds certain sections of the denatured molecule apart until some degree of folding has occurred. After this hypothetical partial folding, we presume that a stable kinetic intermediate has been formed. At this point, the number of hydrophobic areas in rhodanese have probably been reduced, thus weakening interactions with LM and allowing for the release of rhodanese from the micelles. It is possible that a final refolding step(s) occurs following release of this “stable” intermediate. The interaction of unfolded rhodanese with micelles was seen more clearly when the LM micelles contained several molecules of CL. Cardiolipin-containing LM micellesresulted in greater reactivation yields of unfolded rhodanese over the basal level seen in the presence of LM only micelles. This increase in yield and the correlated increase of the tlh of reactivation are presumably due to the tighter binding of unfolded protein to CL/LM micelles. The stronger interaction may be due to the electrostatic interaction of the positive NHQ-terminal sequence of rhodanese to the negatively charged CL head groups, in addition to hydrophobic interactions with LM and/or CL tails. Gel filtration experiments showed that virtually all of the unfolded rhodanese was bound to the CL/LM mixedmicelles 30 min after thestart of reactivation. Furthermore, after refolding had progressed for 29 h, rhodanese was found to elute from Sephadex G-75 as a micelle-free enzyme. In addition, we have found that micellebound rhodanese is completely inactive, whereas free enzyme retains activity after size exclusion chromatography. We specDISCUSSION ulate that as mixed micelle-bound rhodanese folds, the elecPrevious studies have shown that enzymatic activity can be trostaticinteraction is the last one to be broken asthe recovered after dilution of denatured rhodanese into a deter- partially folded enzyme is released because, apparently, this gent-containing reactivation solution. The basic elements in interaction is stronger than thehydrophobic interactions seen the reactivation mixture include 8-mercaptoethanol and thi- with LM only micelles. Furthermore, the presence of hydroosulfate, which prevent oxidation of sensitive sulfhydryl phobic binding sites further intensify the electrostatic intergroups; phosphate buffer which provides the appropriate pH action. Since crystallographic analysis shows that thecharged for optimal activity; and the detergent LM which prevents NH2-terminal of rhodanese binds to the surface of the native aggregation and thereby promotes correct folding/reactiva- rhodanese molecule (21), it maybe postulated that while tion of enzyme. Rhodanese refolding can be easily monitored micelle-bound, a kinetic intermediate form of rhodanese may because its measured activity is directly related to its degree need to have a near “native” conformation in order to pull of proper folding (2). In detergent-assisted refolding studies, the amino terminus away from CL’s phosphate groups. it had been observed that reactivation was accomplished only To support the theory that theobserved reactivation kinetwhen the detergent was present at concentrations above its ics, enzymatic activity yields, and strong mixed micelle bindCMC (1).Furthermore, the degree of reactivation increased ing was due to theelectrostatic charge interactions, we tested as the concentration of detergent increased. However, this the effect of the zwitterionic phospholipid, phosphatidylchoeffect reached an optimum, after which increasing amounts line, on denatured rhodanese. PC-containing LMmicelles had of detergent resulted in lower reactivation yields. In the case the same t3,*of reactivation, and the same level of enzyme of lauryl maltoside, the reactivation effect begins at precisely reactivation as that seen with LM only micelles, except at its CMC (0.08 mg/ml), peaks at 5 mg/ml, and declines as this high concentrations of PC (e.g. 10 PC/micelle). In light of the concentration is increased (1). gel filtration experiments in the text, which suggest the posIn the present study, we have shown that the correlation sible transient complex formed between enzyme and micelle,

Solubility of Denatured Rhodanese in CL/LM SolutionsThe solubility of urea-unfolded rhodanese when diluted into various reactivation solutions was subjectively studied. Previous reports have shown that dilution of unfolded rhodanese, at a final concentration of 0.03 mg/ml, into a solution containing buffer, thiosulfate, and reducing agents, but no detergents, results in the immediate precipitation/aggregation of protein, while addition of 1 mg/ml LM to thediluting solution allowed for the solubilization of unfolded rhodanese at concentrations between 0.03 and 0.05 mg/ml (20). This solubilization effect is essentially the prevention of denatured enzyme aggregates by detergent molecules or detergent micelles. However, we have found that denatured enzyme at 0.2 mg/ml precipitates in this detergent solution immediately upon dilution. Interestingly, the solubility of unfolded enzyme was increased to atleast 0.2 mg/ml when CL was present (in the form of CL/LM micelles) in the above reactivation mixture. This solubilization suggests that there is anincreased affinity of the CL/LM micelles for the unfolded enzyme, thus, there is more efficient sequestering of denatured species and better net prevention of aggregates. These observations may help explain why greater enzymatic reactivation was obtained in the presence of CL. Furthermore, these results may help clarify data from the gel filtration experiments. An obvious concern was that the elution of denatured rhodanese at the void volume may not be due to its binding to micelles but rather to the fact that unfolded proteins have a greater Stokes radius and, thus, migrate as higher molecular weight species. Attempts to chromatograph a solution containing unfolded rhodanese at 0.2 mg/ml in an LM only reactivation mixture resulted in the immediate precipitation of the protein at the top of the column, and no protein eluted from the column even after exhaustive washing. This observation shows that unfolded rhodanese which was not bound to CL/LM micelles would precipitate and would not be eluted from the column. Thus, in the gel filtration experiments, rhodanese which was detected at thevoid volume is likely bound to CL/LM micelles. Furthermore, it may be safe to assume that in the LM only micelle experiments, elution of rhodanese at thevoid volume reflects an actual protein-micelle complex, since unbound unfolded protein would aggregate and precipitate in the column.

5816

Micelle-assisted Folding Protein

the lower reactivation yield obtained at 10 PC/micelle may be explained by a decreasing affinity of unfolded enzyme as the amount of PC in the micelles increases. This would result in a shorter binding time of rhodanese to these micelles and subsequently result in the premature release of a significant amount of incorrectly folded intermediates which would not regain activity. Protein analysis of fractions from the chromatography of PC/LM micelle-assisted reactivation of rhodanese support the assumption that unfolded rhodanese has less affinity for PC/LM micelles than it does for LM only micelles. In gel filtration experiments, about 25% of total rhodanese eluted as LMmicelle-bound after 30 min in an LM reactivation solution, whereas, in a similar experiment no rhodanese was seen eluting as PC/LM micelle-bound when reactivated in a PC/LM solution. This implies that unfolded rhodanese binds to PC/LM micelles for a shorter duration than to LM only micelles. These findings are consistentwith the proposal that the NH2-terminal sequence of rhodanese plays a role in interacting with phospholipids in LM micelles,because it suggests that theearlier release of enzyme from the mixed micelle may be due to a repulsion between the positively charged rhodanese leader sequence andthe positive charged tertiary amine in the head group of PC. Results which suggest an earlier release of unfolded rhodanese from PC/LM micelles, relative to LM micelles, while no significant difference is seen in the tlhof reactivation and maximum activity values suggests that rhodanese is not completely folded (active) upon release by the micelle. Since, if rhodanese was released from the PC/LM micelles in an active state, thenits tIhwould be shorter than with LM only micelles due to the difference in binding duration to the micelles. An alternate explanation, which is not necessarily mutually exclusive, would be that the folding pathway of enzyme, while micelle bound, is different in the three studied cases (ie.LM only, CL/LM, and PC/LMmicelles) and, therefore, rhodanese kinetic intermediates at the time of release are not identical. Thus, after release, the step(s) and time necessary for full reactivation would vary for each case. Experiments presented in this work show that the mitochondrial phospholipid, cardiolipin, may have an active role in the transport and perhaps refolding of rhodanese. A current view is that chaperonins within the mitochondria may influence the refolding/reactivation of rhodanese by binding to partially folded inactive interactive intermediates. Here we present evidence that CLmay influence refolding of rhodanese without the aid of chaperonins, although it is possible

that, in uiuo, these two systems may complement each other. Nevertheless, we have shown that the CL-assisted protein folding has several similarities to the chaperonin-assisted rhodanese folding model. These include 1) binding of rhodanese to a high molecular weight structure (ie. micelle or 14-mer macromolecules); 2) both structures bind only unfolded enzyme and not native enzyme; 3) presumably they both increase enzyme reactivation yields by preventing aggregation; and 4) rhodanese is completely inactive when bound to either micelles or to the 60-kDa chaperonin (cpn60) 14mer (4). REFERENCES 1. Tandon, S.,and Horowitz, P. M. (1987) J. Biol. Chem. 262, 4486-4491 2. Tandon, S.,and Horowitz, P. M. (1989) J. B i d . Chem. 2 6 4 , 9859-9866 3. Mendoza, J. A., Rogers, E., Lorimer, G . H., and Horowitz, P. M. (1991) J. Biol. Chem. 2 6 6 , 13587-13591 4. Mendoza, J. A., Rogers, E., Lorimer, G. H., and Horowitz, P. M. (1991) J , Bid. Chem. 266,13044-13049 5. McMullin, T. W., and Hallberg, R. L. (1988) Mol. Cell. Biol. 8, 371-380 6. Lubben, T. H., Gatenby, A. A., Donaldson, G. K., Lorimer, G. H., and Viitanen, P. V. (1990) Proc. Natl. Acad. Sci. U.S. A . 87,7683-7687 7. Miller, D. M., Delgado, R., Chirgwin, J. M., Hardies, S. C., and Horowitz, P. M. (1991) J. Biol. Chem. 266,4686-4691 8. Boggaram, V., Horowitz, P., and Waterman, M. R. (1985) Biochem. Biophys. Res. Commun. 130,407-411 9. Pain, D., Schnell, D. J., Murakami, H., and Blobel, G . (1991) Genet. Eng. 13,153-166 10. Schleyer, M., and Neupert, W. (1985) Cell 43,339-350 11. Rouser, G., Nelson, G . J., Fleischer, S., and Simon, G . (1968) in Biological Membranes (Chapman, D., ed) Vol. I, pp, 37-42,

Academic Press, New York 12. Robinson, N.C., Strey, F., and Talbert, L. (1980) Biochemistry 19,3656-3661 13. Horowitz, P. M. (1978) Anal. Biochem. 86,751-753 14. Sorbo, B. H. (1953) Acta Chem. S c a d . 7, 1123-1127 S. (1983) Biochemistry 15. Thomuson. D. A.. and Fermson-Miller. 22,3178-3187 16. Laemmli. U. K. (1970) Nature 227.680-685 17. Horowitz, P. M.; and Bowman, S.-(1989) J. Biol. Chem. 2 6 4 , 3311-3316 18. Tandon, S.,and Horowitz, P. (1986) J. Biol. Chem. 2 6 1 , 1561515618 19. Fischer, G., and Schmid, F. X. (1990) Biochemistry 29, 22052212 20. Horowitz. P. M.. and Simon. D. (1986) J. Biol. Chem. 261, 13tm-i3891 21. Ploeeman. J. H.. Drent. G.. Kalk. K. H., Hol, W. G . J., Heinrikso;, R. L., Keim, P.,'Weng, L.; and Russell, J. (1978) Nature 273,124-129 '