Proc. Natl. Acad. Sci. USA Vol. 88, pp. 6545-6549, August 1991 Biochemistry

Reconstitution of a protein translocation system containing purified SecY, SecE, and SecA from Escherichia coli (membrane proteins/membrane assembly/proteoliposomes/ATP)

JIRO AKIMARU, SHIN-ICHI MATSUYAMA, HAJIME TOKUDA, AND SHOJI MIZUSHIMA* Institute of Applied Microbiology, the University of Tokyo, 1-1-1, Yayoi, Bunkyo-ku, Tokyo 113, Japan

Communicated by Gunter Blobel, April 15, 1991 (received for review January 26, 1991)

Reconstitution of the translocation machinABSTRACT ery for secretory proteins from purified constituents was performed. SecY was solubilized from SecY/SecE-overproducing Escherichia coli cells and purified by chromatography on ion-exchange and size-exclusion columns. Proteoliposomes active in protein translocation were reconstituted from the purified preparations of SecY and SecE. The reconstituted translocation activity was SecA- and ATP-dependent. Although the purified preparations of SecY and SecE were still contaminated with minute amounts of other proteins, the elution profiles of SecY and SecE on column chromatographies coincided with the elution profres of reconstituted translocation activity, indicating that SecY and SecE are the indispensable components in these preparations. We conclude that SecY, SecE, and SecA are essential components of the protein secretion machinery and that translocation activity can be reconstituted from only these three proteins and phospholipids.

Genetic studies have revealed that SecA (1), SecY (2), SecE (3), SecD (4), and SecF (4) play important roles in the translocation of secretory proteins across the cytoplasmic membrane of Escherichia coli. All except SecA are integral membrane proteins (2-5). Biochemical studies, therefore, require purification of the proteins involved and their reconstitution into translocationally active proteoliposomes. Proteoliposomes that exhibit protein translocation activity have been reconstituted from unfractionated components (6-8). Brundage et al. (9) have reported the reconstitution of proteoliposomes active in protein translocation from a fraction containing SecE, SecY, and an uncharacterized protein. It was not clear, however, which components in the fraction were essential for reconstitution. SecA has been overproduced and purified (10, 11) and then characterized in detail (12). SecE has been purified from SecE-overproducing cells and shown to be essential for the reconstitution of active proteoliposomes (13). Furthermore, SecY could be overproduced when SecE was simultaneously overproduced (14). We report here the purification of SecY from the cells overproducing both SecY and SecE and the reconstitution of a protein translocation system from purified preparations of SecY, SecE, and SecA.

MATERIALS AND METHODS Bacterial Strains and Plasmids. E. coli W3110 M25 (ompT-) (15) was transformed with pMAN809 for the overproduction of SecE or with pMAN809 and pMAN510 for the simultaneous overproduction of SecE and SecY (14). The overproduction was induced by isopropyl -D-thiogalactopyranoside as reported (14). pOAD26 carries the ompA-D26 gene encoding proOmpA-D26, which is a derivative of proOmpA The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement"

in accordance with 18 U.S.C. §1734 solely to indicate this fact.

that lacks about 250 amino acid residues at its C terminus (K. Kanamaru, H. Yamada, and S.M., unpublished data). pSI053 (16) carries the ompA gene that codes for intact proOmpA. Preparation of SecE, SecA, and Phospholipids. SecE was purified from the cytoplasmic membrane fraction prepared from SecE-overproducing cells as described (13). SecA was purified from SecA-overproducing cells as described (17). E. coli phospholipids were- prepared as reported (8). Purification of SecY. Cytoplasmic membrane fractions prepared from SecY/SecE-overproducing cells were solubilized at 1 mg of protein per ml on ice for 10 min with 2.5% (wt/vol) n-octyl ,f-D-glucopyranoside (octyl glucoside) (Dojindo Laboratories, Kumamoto, Japan) containing 50 mM potassium phosphate (pH 6.95), 150 mM NaCl, 10% (wt/vol) glycerol, and E. coli phospholipids (2.5 mg/ml). After ultracentrifugation at 140,000 x g for 30 min in a Beckman TLA 100.3 rotor, the supernatant containing 11.5 mg of protein was applied on a Mono S cation-exchanger column (1 cm x 10 cm; Pharmacia), which had- been equilibrated with 2.5% octyl glucoside containing 50 mM potassium phosphate (pH 6.95), 10% glycerol, and 150 mM NaCl. The column was then developed at the flow rate of 4 ml/min with a linear gradient of NaCI (0.15-1 M) in the same buffer. The amount of SecY in each fraction (2 ml) was determined by SDS/PAGE followed by immunoblot analysis with anti-SecY antiserum. The fraction that -contained most of the SecY was concentrated by membrane filtration. A sample (0.5 ml) of the concentrated fraction was further purified by size-exclusion chromatography on a Superose 12 HR column (1 cm x 30 cm; Pharmacia) that had been equilibrated with 2.5% octyl glucoside containing 50 mM potassium phosphate (pH 6.95), 10%O glycerol, and 150 mM NaCI. The column was developed with the same buffer at the flow rate of 0.4 ml/min. The amount of SecY in each fraction was determined by densitometric scanning of the silver-stained gels. Size-exclusion chromatography was performed 10-20 times, and fractions containing SecY with a purity of =70% were combined and concentrated. Reconstitution of Proteoliposomes Exhibiting Protein Translocation Activity. Proteoliposomes were reconstituted by the octyl glucoside dilution method (13). Samples of fractions obtained by cation-exchange chromatography or sizeexclusion chromatography of the solubilized membrane fraction derived from SecE/SecY-overproducing cells were mixed with 1.25 mg of E. coli phospholipids in 2.5% octyl glucoside. Where specified, purified SecE was also added to the mixture. After a 20-min incubation on ice, the mixture was rapidly diluted with 4.6 ml of 50 mM potassium phosphate (pH 7.5) containing 150 mM NaCI and then incubated at room temperature for 5 min with stirring. The proteoliposomes formed were recovered by centrifugation at 160,000 x g for 2 h, suspended in 100 Al of 50 mM potassium phosphate *To whom reprint requests should be addressed.

6545

Biochemistry: Akimaru et al.

6546

Proc. Natl. Acad. Sci. USA 88 (1991)

(pH 7.5) containing 150 mM NaCI, frozen, thawed, and sonicated as described (8). In Vitro Transcription and Translation. In vitro transcription of the ompA-D26 and ompA genes was performed as described (18). The translation reaction was carried out in the presence of Tran35S-Label (0.46 mCi/ml; 1 Ci = 37 GBq; ICN) as described (19). [35S]Methionine-labeled proOmpAD26 and proOmpA were partially purified as reported (20). Protein Translocation by Reconstituted Proteoliposomes. Samples (15 jul) of the reconstituted proteoliposomes were mixed with 1 Al of SecA (1.5 mg/ml) and 5 A.l of 50 mM potassium phosphate (pH 7.5) containing 10 mM MgSO4, 10 mM ATP, 150 mM NaCl, and an ATP generating system composed of 50 mM creatine phosphate and creatine kinase (1.25 mg/ml). After a 3-min preincubation at 370C, the assay was started by the addition of 4 ,.l of [35S]methionine-labeled proOmpA-D26 or proOmpA (2 x 105 cpm). The translocated protein, which was proteinase K-resistant, was detected on an SDS/polyacrylamide gel by means of fluorography, as described (21). Densitometric quantification of band materials was carried out with a Shimadzu (Kyoto) CS-930 chromatoscanner. The amounts of translocated proOmpAs were expressed as percentages of the input precursor protein. Preparation of Anti-SecE and Anti-SecY Antisera. Peptides corresponding to the region from Lys-64 to Lys-81 of SecE (3) and the region from Met-1 to Arg-22 of SecY (22) were synthesized and used to raise antisera as described (8). SDS/PAGE and Immunoblot Analysis of SecE and SecY. A gel containing 13.5% acrylamide and 0.36% N,N'-methylenebisacrylamide was used as described by Laemmli (23). All samples were applied to the gel without boiling. Immunoblot analysis was carried out as described (24). Determinations of Phospholipids and Proteins. Amounts of phospholipids and proteins were determined by the methods of Bartlett (25) and Lowry et al. (26), respectively.

A E 0.8 Go 0

N 0.6 0

0.4-

.0

O 0.2 '

n

Fraction

B SecY

-p

SecE

-

I I

a I-

-

I.

-§

-

.,

0.1

E c

IF

0

co 04

-

co 0

c 0.05 m0 0

.0

15867 43

13.7

30

40

4

440

Vo 0

20

50

Fraction

B . r. C^

^

29

.~~~~~~~~~~~~ *

C

j

_ _ _ ,4.-.34.~o

50

D

0

U)

/

40

1 2 3 4 kDa

0 C C

20~~~~

43

-

-

__r

SecY

29

-

18.4

10 27

29

31

33 35 Fraction

37 39

14.3 -6.2

FIG. 2. Purification of SecY by size-exclusion chromatography. Fractions 11-20 in Fig. 1A were combined and then concentrated by membrane filtration. (A) A sample (500 p1) of the concentrated fraction was purified on a Superose 12 HR column. The elution positions of the following molecular mass markers are also indicated: ferritin (440 kDa), aldolase (158 kDa), bovine serum albumin (67 kDa), ovalbumin (43 kDa), and ribonuclease A (13.7 kDa). VO denotes the void volume. (B) SDS/PAGE was performed with 10 ,ul of each fraction in A followed by silver staining. (C) Proteoliposomes were reconstituted from samples (100 ,ul) of the fractions in A and 1.25 mg of E. coli phospholipids with (e) or without (o) 8 tzg of purified SecE. The translocation of partially purified [35S]methionine-labeled proOmpA-D26 into the proteoliposomes was assayed at 37°C for 20 min. (D) SecY-containing fractions obtained at each step of purification were analyzed by SDS/PAGE, followed by staining with Coomassie brilliant blue. Lanes: 1, membrane fraction (8.0 ,ug); 2, octyl glucoside supernatant (7.6 ,ug); 3, Mono S fraction (1.9 ,ug); 4, Superose 12 HR fraction (0.5 jig).

RESULTS

18.4

|I _-" 10

kDa 43 29

A

-14.3

ln

15

-

20

6.2

25

Fraction Anti-SecY Anti-SecE

FIG. 1. Separation of SecE and SecY by cation-exchange chromatography. (A) The cytoplasmic membrane prepared from SecE/ SecY-overproducing cells was solubilized with 2.5% octyl glucoside. A supernatant was obtained and fractionated on a Mono S column. The column was developed with a linear gradient of NaCI (---). Fractions of 2 ml were collected and Ano was recorded (-). (B) A sample (50 p1) of each fraction in A was analyzed by SDS/PAGE. The gels were either stained with Coomassie brilliant blue or immunoblotted with anti-SecE or anti-SecY antiserum.

The cytoplasmic membrane fraction was prepared from cells overproducing both SecY and SecE and solubilized with octyl glucoside. The soluble fraction, which accounted for 95% of the membrane protein, was recovered by centrifugation and then subjected to cation-exchange chromatography on a Mono S column. The amounts of SecY in the eluates were determined on SDS gels with Coomassie brilliant blue staining and immunoblot analysis with anti-SecY antiserum (Fig. 1). SecY was preferentially eluted immediately after the pass-through fraction in which SecE and most of the other proteins appeared. Proteoliposomes were then reconstituted from these eluates in the presence of purified SecE and analyzed for protein translocation activity in the presence of SecA. The highest activity was reconstituted with fractions

Proc. Natl. Acad. Sci. USA 88 (1991)

Biochemistry: Akimaru et al.

A E 0 CO 00 -W

04

I 0

o0 .0 Fraction

B

kDa -

43

-29

SecE

-

18.4

-.

-14.3 - 6.2 1415 1617

18

19 20 21 22 23

Fraction

Fraction

FIG. 3. Purification of SecE by size-exclusion chromatography. (A) Partially purified SecE was obtained by anion-exchange chromatography as described (13) and then subjected to size-exclusion chromatography under the conditions described in Fig. 2 for SecY. The molecular mass markers used were the same as those in Fig. 2. (B) A sample (25 Al) of each fraction in A was analyzed by SDS/ PAGE, followed by staining with Coomassie brilliant blue, as described in Fig. 1B. (C) Proteoliposomes were reconstituted from 20 41A of each fraction in A and 1.25 mg of E. coli phospholipids with (e) or without (o) 1.9 ,ug of purified SecY and then assayed for translocation of proOmpA-D26.

11-20 (data not shown). These fractions were combined and concentrated. Further purification of SecY was carried out on a Superose 12 HR column (Fig. 2A) and the individual fractions obtained were analyzed by SDS/PAGE and silver staining (Fig. 2B). Reconstitution was then performed with fractions in the presence of purified SecE. The translocation activity peak

A

only coincided with the SecY peak (Fig. 2C). No translocation activity was reconstituted in the absence of SecE. We conclude, therefore, that SecY is the component of this fraction that is essential for the translocation reaction. Samples obtained at each stage of the purification were analyzed by SDS/PAGE followed by Coomassie brilliant blue staining (Fig. 2D). The purity of the final SecY preparation was >70%o. The band near the top of the gel was more prominent in Fig. 2B. This may be due to a lack of uniformity in the silver staining procedure. The SecE preparation used in this and previous (13) studies was still contaminated by minute amounts of other proteins. To exclude the possible involvement of these minor proteins in the translocation, fractions from the Superose 12 HR column (the final step of the SecE purification) were analyzed for SecE and translocation activity (Fig. 3). The reconstitution was carried out with a fixed amount of purified SecY. The elution profile of SecE and the profile of reconstituted activity coincided closely, whereas the elution profile of other proteins was different from the reconstitution profile. We conclude, therefore, that SecE is the essential component in the final preparation. No translocation activity was reconstituted in the absence of SecY. The reconstitution of protein translocation activity from the purified preparations of SecY, SecE, and SecA was then performed, and the translocation kinetics were studied more precisely. The Sec proteins used were of the highest purity obtained (Fig. 4A). Proteoliposomes were reconstituted from 1.8 t.g of SecY, 1.9 ,Ag of SecE, and 1250 Ag of E. coli phospholipids. The reconstituted proteoliposomes recovered by centrifugation contained 1.3, 1.5, and 530 ug of SecY, SecE, and phospholipids, respectively. A time course study with proOmpA-D26 showed that the reconstituted activity was SecA- and ATP-dependent, and the reaction proceeded quite steadily for 15 min (Fig. 4B). The activity was as high as the activity of the proteoliposomes reconstituted from the SecY/SecE-overproducing membrane. The rate of translocation of proOmpA-D26 into everted membrane vesicles is much faster than that of intact proOmpA (M. Kato and S.M., unpublished observation). This was the reason we used proOmpA-D26 in the present study. We found, however, that the reconstituted proteoliposomes were appreciably active in the translocation of intact proOmpA, a natural presecretory protein (Fig. 4C). It should be mentioned that

B SecY

SecE

6547

C

SecA kDa

--

43

@

2

0

a 29

0 C

Ie. 18.4

0'

10

0

5

10 15 Tune (m-)

20

0

5

10 15 rue (m-)

20

FIG. 4. Reconstitution of a protein translocation machinery from purified SecY, SecE, and SecA. (A) SDS/PAGE profiles of the purified components used for the reconstitution. (B) Proteoliposomes were reconstituted from 1.8 Ag (0.037 nmol) of SecY and 1.9 Ag (0.14 nmol) of SecE and then assayed for translocation of proOmpA-D26 for the indicated times in the presence of SecA plus ATP (e), SecA alone (A), or ATP alone (o). Reconstitution was also performed using 20 A.l of the unfractionated octyl glucoside extract, containing 1.2 ,ug (0.024 nmol) of SecY and 1.6 ,g (0.12 nmol) of SecE, prepared from the SecY/SecE-overproducing membrane preparation, and the resultant proteoliposomes were assayed for protein translocation activity (a). (C) Proteoliposomes were reconstituted from 2.0 jig (0.041 nmol) of SecY and 2.8 EAg (0.21 nmol) of SecE and then assayed for translocation of intact proOmpA for the indicated times in the presence of SecA plus ATP (A), SecA alone (a), and ATP alone (A). The translocation profile of proOmpA-D26 in the presence of SecA plus ATP is also presented (-).

6548

Biochemistry: Akimaru et al.

A

Proc. Nad. Acad. Sci. USA 88 (1991)

'B

50e

0 50

'O40 -u4 4

0

0030

0.

0

C

IP 20 -2010

10

0~~~~~~ 0.04 0.08 SecY (nmol)

0.12

0

0.1

0.2 0.3 SecE (nmol)

0.4

FIG. 5. Effects ofthe amounts of SecY and SecE on the translocation activity ofreconstituted proteoliposomes. (A) The translocation activity of proteoliposomes reconstituted from 1.9 ,ug (0.14 nmol) of SecE and the indicated amounts of SecY was assayed in the presence (e) or absence (o) of SecA for 20 min. (B) The translocation activity of proteoliposomes reconstituted from 3.0 ,ug (0.06 nmol) of SecY and the indicated amounts of SecE was assayed for 20 min with (0) or without (o) SecA.

the rate oftranslocation of proOmpA was still 180 times lower than that exhibited by native membrane vesicles on the basis of the amount of SecY. When reconstitution was carried out with various amounts of SecY and a fixed amount of SecE (0.14 nmol), the reconstituted activity exhibited saturation at about 0.03 nmol of SecY (Fig. SA). Reconstitution was also carried out with various amounts of SecE and a fixed amount of SecY (0.06 nmol) (Fig. 5B). The reconstituted activity increased as the amount of SecE increased. The increase was observed even when SecE was >5-fold greater than SecY. These results suggest that SecE is present in excess to SecY in functional stoichiometry. Bieker and Silhavy (27) indicated that SecE precedes SecY in the protein translocation pathway. The importance of SecA in the initial stage of the secretion pathway has been suggested (17, 20, 28). Thus by taking these results (17, 20, 27, 28) and our results (Fig. 5) into consideration, it is likely that SecE functions as a shuttle between SecA and SecY in the secretory pathway. The results shown in Fig. 5 also indicate that SecY, SecE, and SecA are all indispensable components of the protein translocation machinery. This conclusion is inconsistent with a recent report (6) that SecY may be dispensable for the reconstitution of a translocation system. The reasons for this discrepancy are unclear.

DISCUSSION In the present work, a translocation system for secretory proteins was reconstituted from purified protein components (SecY, SecE, and SecA) and phospholipids. The reconstituted activity was ATP-dependent. In the present study, proteinase K resistance was the sole index of protein translocation. Cleavage of the signal peptide, another index of translocation, was not demonstrated, since signal peptidase was not included. The possibility that the observed activity represents only a partial reaction of translocation has not been excluded completely. Although the SecY and SecE preparations used were still contaminated by minute amounts of other proteins, SecY and SecE were the only proteins whose elution profiles coincided with profiles of the reconstituted translocation activity (Figs. 2 and 3). The results strongly suggest that SecY and SecE are the essential components in these preparations. Brundage et al. (9) have demonstrated the reconstitution of translocationally active proteoliposomes from a solubilized membrane fraction containing SecY, SecE, and an uncharacterized protein as major components. We demonstrated that both SecY and SecE are essential but that the third protein is not;

the third protein was not detected in our purified Sec preparations on SDS gels. We also demonstrated that SecY and SecE were not cofractionated by column chromatography. On the basis of translocation activity per unit amount of SecY or SecE, proteoliposomes reconstituted from the purified components were as active as those reconstituted from the crude solubilized membrane from which the purified samples were derived. The activity was, however, far lower than that exhibited by native everted membrane vesicles as mentioned above. Possible reasons for this are as follows. (i) Only small fractions of SecY and SecE might have been functionally reconstituted into proteoliposomes, (it) the reconstituted proteoliposomes were unable to generate the proton motive force that generally enhances the translocation activity (29), and (iii) the reconstitution was carried out in the absence of SecD and SecF, both of which are genetically suggested to be involved in the translocation (4). The absence of SecD and SecF may not be crucial, however, since they might function on the periplasmic side of the cytoplasmic membrane (i.e., inside the everted membrane vesicles) (4) and the translocation was assayed by resistance to proteinase K, which was added externally to the membrane vesicles. Finally, it should be mentioned that the present biochemical study is consistent with a previous genetic study (27) in that both suggest that SecY and SecE interact but can easily be separated and SecE appears to be in functional excess to SecY. Such consistencies support the physiological significance of the translocation activity reconstituted in the present study. Reconstitution techniques have been applied to structurefunction analysis of organelles and complex biological systems. The reconstitution of a secretory machinery from purified components will greatly facilitate elucidation of the molecular mechanisms underlying protein secretion across the cytoplasmic membrane and the mechanisms underlying protein translocation across membranes in general. We thank Y. Kabuyama for technical help and I. Sugihara for secretarial support. This work was supported by grants from the Ministry of Education, Science and Culture of Japan (61060001, 02404013, 02680153, and 02780170).

1. Oliver, D. B. & Beckwith, J. (1981) Cell 25, 765-772. 2. Ito, K., Witterkind, M., Nomura, M., Shiba, K., Yura, T., Miura, A. & Nashimoto, H. (1983) Cell 32, 789-797. 3. Schatz, P. J., Riggs, P. D., Jacq, A., Fath, M. J. & Beckwith, J. (1989) Genes Dev. 3, 1035-1044. 4. Gardel, C., Johnson, K., Jacq, A. & Beckwith, J. (1990) EMBO J. 9, 3209-3216.

Biochemistry: Akimaru et al. 5. Akiyama, Y. & Ito, K. (1985) EMBO J. 4, 3351-3356. 6. Watanabe, M., Nicchitta, C. V. & Blobel, G. (1990) Proc. Natl. Acad. Sci. USA 87, 1960-1964. 7. Driessen, A. J. M. & Wickner, W. (1990) Proc. Natl. Acad. Sci. USA 87, 3107-3111. 8. Tokuda, H., Shiozuka, K. & Mizushima, S. (1990) Eur. J. Biochem. 192, 583-589. 9. Brundage, L., Hendrick, J. P., Shievel, E., Driessen, A. J. M. & Wickner, W. (1990) Cell 62, 649-657. 10. Kawasaki, H., Matsuyama, S., Sasaki, S., Akita, M. & Mizushima, S. (1989) FEBS Lett. 242, 431-434. 11. Cunningham, K., Lill, R., Crooke, E., Rice, M., Moore, K., Wickner, W. & Oliver, D. (1989) EMBO J. 8, 955-959. 12. Oliver, D. B., Cabelli, R. J. & Jarosik, G. P. (1990) J. Bioenerg. Biomembr. 22, 311-336. 13. Tokuda, H., Akimaru, J., Matsuyama, S., Nishiyama, K. & Mizushima, S. (1991) FEBS Lett. 279, 233-236. 14. Matsuyama, S., Akimaru, J. & Mizushima, S. (1990) FEBS Lett. 269, 96-100. 15. Sugimura, K. (1988) Biochem. Biophys. Res. Commun. 153, 753-759. 16. Yamada, H., Tokuda, H. & Mizushima, S. (1989) J. Biol. Chem. 264, 1723-1728.

Proc. Nati. Acad. Sci. USA 88 (1991)

6549

17. Akita, M., Sasaki, S., Matsuyama, S. & Mizushima, S. (1990) J. Biol. Chem. 265, 8164-8169. 18. Melton, D. A., Krieg, P. A., Rebagliati, M. R., Maniatis, T., Zinn, K. & Green, M. R. (1984) Nucleic Acids Res. 12, 70357056. 19. Yamane, K., Matsuyama, S. & Mizushima, S. (1988) J. Biol. Chem. 263, 5368-5372. 20. Tani, K., Shiozuka, K., Tokuda, H. & Mizushima, S. (1989) J. Biol. Chem. 264, 18582-18588. 21. Yamane, K., Ichihara, S. & Mizushima, S. (1987) J. Biol. Chem. 262, 2358-2362. 22. Ceretti, D. P., Deau, D., Davis, G. R., Bedwell, D. M. & Nomura, M. (1983) Nucleic Acids Res. 11, 2599-2616. 23. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 24. Yamada, H., Matsuyama, S., Tokuda, H. & Mizushima, S. (1989) J. Biol. Chem. 264, 18577-18581. 25. Bartlett, G. R. (1959) J. Biol. Chem. 234, 466-468. 26. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. (1951) J. Biol. Chem. 193, 265-275. 27. Bieker, K. L. & Silhavy, T. J. (1990) Cell 61, 833-842. 28. Hartd, F.-U., Lecker, S., Shiebel, E., Hendrick, J. P. & Wickner, W. (1990) Cell 63, 269-279. 29. Mizushima, S. & Tokuda, H. (1990) J. Bioenerg. Biomembr. 22, 389-399.