Reversible, Nonionic, and pH-dependent Association of Cytochrome c

Vol. 261. No. 31, Issue of November 5, PP. 22243-22248, 1992 Printed in U.S A

OF BIOLOGICAL CHEMlSTRY THEJOURNAL

!c)1992 by The Amerlcan Society for Biochemistry and Molecular Biology, Inc

Reversible, Nonionic, and pH-dependent Association of Cytochrome c with Cardiolipin-Phosphatidylcholine Liposomes* (Received for publication, April 7, 1992)

Marjatta Rytomaa, Pekka Mustonen, and PaavoK. J. Kinnunen4 From the Department of Medical Chemistry, University of Helsinki, Siltauuorenpenger I O A, 00170 Helsinki, Finland

Membrane association of cytochrome c (cyt c ) was (e.g. Kimelberg and Lee, 1969; Quinnand Dawson, 1969; monitored bythe efficiency of resonance energy trans- Steinemann and Lauger, 1971; Vanderkooi et at., 1973; Walfer from a pyrene-fatty acid containing phospholipid tham et al., 1986; Mustonen et al., 1987; Kozarac et al., 1988; derivative (l-palmitoyl-2[6-(pyren1-yl)]hexanoyl- Demel et al., 1989). Of the negatively charged lipids cyt c has sn-glycero-3-phosphocholine(PPHPC)) to the heme of highest affinity for phosphatidic acid (Nicholls and Malviya, cyt c. Liposomes consisted of 85 mol% egg phosphati- 1973). It also bindsavidly to cardiolipin which probablyeither dylcholine (egg PC), 10 mol% cardiolipin, and5 mol% as such or complexed with cyt c oxidase provides with the PPHPC. Cardiolipin was necessary for the membrane binding site at the inner mitochondrial membrane (Vik et al., binding of cyt c over the pH range studied, from 4 to 1981; Speck et al., 1983). Association of cyt c with acidic 7.In accordance with the electrostatic nature of the phospholipids is mainly electrostatic in nature andsensitive is membrane associationof cyt e at neutral pH both 2 mM tochangesin ionic strength (Nicholls, 1974; Brown and MgClz and 80 mM NaCl dissociated cyt c from the negative chargedensity due to acidic vesicles completely. At neutral pH also adenine nucle- Wuthrich, 1977). Critical otides in millimolar concentrations were able to dis- phospholipids seems to be required (Mustonen et al., 1987). place cyt c from liposomes, their efficiency decreasing Cyt c appears to partly penetrate intocardiolipin containing in thesequence ATP > ADP > AMP. In addition, both bilayers so as to interactalso hydrophobically with the memCTP and GTP were equally effective as ATP. The brane interior (Brown and Wuthrich, 1977; Szebeni and Tolinduces conformational detachment of cyt c from liposomes by nucleotides is lin, 1988). Bindingtomembranes likely to result from a competition between cardiolipin changes in cytc (Jori et al., 1974; Hildebrandt and Stockburand thenucleotides for a common binding site in cyt c. ger, 1989a, 1989b; De Jough and De Kruijff, 1990; Soussi et When pH was decreased to 4 there was a small yet al., 1990; Heimburg et al., 1991; Muga et al., 1991; Spooner significant increase in the apparent affinity of cyt c to and Watts,1991). It may also leadto the formation of laterally cardiolipin containing liposomes. Notably, at pH 4 the segregated domains enriched in acidic phospholipids (Brown above nucleotides as well as NaCl and MgC12 were no and Wuthrich, 1977; Mustonen et al. 1987; Haverstick and longer able to dissociate cyt c and, on the contrary, Glaser, 1989). These negatively charged membrane domains they slightly enhanced the quenching of pyrene fluo- constitute the binding site interacting with a basic domain of rescence by cyt c. The above results do suggest that the c (Dickerson et al., 1971). Cyt c also inducespH-dependent cyt membrane association of cyt c a t acidic pH was nonionic and presumably due to hydrogen bonding. The fusion of phosphatidylserine/phosphatidylethanoiaminevesfully icles with maximum efficiency at pH 5 and 8 (Lee and Kim, pH-dependent binding of cyt c to membranes was reversible. Accordingly, in the presence of sufficient 1989). Fluorescence resonance energy transfer from pyrene-fatty concentrations of either nucleotides or salts rapiddetachment and membrane association of cyt c could be acid containing phospholipid derivatives to the heme of cyt c induced by varying pH between neutral and acidic has been utilized by us to study cyt c binding to liposomes (Mustonen et al., 1987). Here we have used this technique to values, respectively. investigate the effectsof adenine nucleotides and pH on the binding of cyt c to cardiolipin containing vesicles. Cytochrome c (cyt c)’ is a well characterizedperipheral membraneprotein of theinnermitochondrialmembrane transferringelectrons between themitochondrial enzyme complexes totheterminal enzyme,cytochrome c oxidase. Binding of cyt c to phospholipid monolayers and liposomes containing acidic phospholipids has been extensively studied * This study was supported by the Medical Research Council of the Finnish Academy. 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. $ Senior Investigator of the Finnish Academy. T o whom correspondence should be addressed. Fax: 358-0-1918276. ’ The abbreviationsused are: cyt c, cytochrome c; PA, phosphatidic acidPC,phosphatidylcholine;PG, phosphatidylglycerol; MES, 2(N-morpho1ino)ethanesulfonic acid; PPDPC, l-palmitoyl-2-[10(~yren-l-yl)]decanoyl-sn-glycero-3-phosphocholine; PPHPC,l-palmitoyl-2-[G-(pyren-1-yl)]hexanoyl-sn-glycero-3-phosphocholine.

EXPERIMENTALPROCEDURES

Materials-l-Palmitoyl-2-[G-(pyren-l-yl)]hexanoyl-sn-glycero-3phosphocholine (PPHPC)and l-palmitoyl-2[6-(pyren-l-yl)]decanoyl-sn-glycero-3-phosphocholine (PPDPC) were purchased from KSV Chemical Co (Helsinki, Finland). Nan salts of ATP, ADP, AMP, and GTP were from Boehringer Mannheim. CTP (Na2 salt) type 11, horse heart cyt c (type VI, oxidized form), cardiolipin (from bovine heart), egg PA, egg PC, and egg PG were from Sigma. No impurities were detected in theabove lipids upon thin-layer chromatography on silicic acid using chloroform/methanol/water/ammonia (65:20:2:2,v/ v) as the solvent system and examination of the plates for pyrene fluorescence or after iodine staining. Preparation of Liposomes-Lipids were dissolved either in chloroform or ethanol. After mixing of the desired lipid composition these solvents were removed under a stream of nitrogen. The dry lipid residue was then maintained under reduced pressure for at least 2 h and then hydrated in 20 mM MES-0.1 mM EDTA a t room temperature. The pH of the buffer had been adjusted to 4.0, 5.0, 6.0,or 7.0 with 1 M NaOH. The final concentration of lipid was 25 PM. Unless

22243

22244

Binding of Cytochrome c to Cardiolipin at AcidicpH

otherwise indicated the liposomes consisted of 5 mol% PPHPC, 10 mol% cardiolipin, and 85 mol% egg PC. In some experiments the content of cardiolipin in the vesicles was varied, and the amount of egg PC was adjusted to maintain the total number of acyl chains constant, i.e. one cardiolipin was considered equivalent for two egg P C molecules. Whenindicated, cardiolipin was eitheromittedor substituted for 20 mol% egg PA or egg PG. To obtain unilamellar vesicles, the dispersions were sonicated for 4 min on an ice-water bath with a Branson sonifier equipped with a microtip and operated at a 65-watt output. Minimal exposure of the lipids to lightwas ensured throughout theabove procedure. Fluorescence Spectroscopy-Fluorescence measurements were carriedout with anSLM4800sspectrofluorometerinterfacedto a Hewlett-Packard 85 computer. Excitation was at 344 nm, and the emission intensity was measured at 394 nm. These wavelengths were selected with monochromators. Slits of 1 and 16 nm were used for the excitation and emission beams, respectively. Two ml (50 nmol of lipid or as indicated)of liposome solution was placed into a magnetically stirred four-window cuvette in a holder thermostated with a circulating water bath at 25 "C. Aliquots of 1-8 pl of a 0.02 mM solution of cyt c were added to the liposome solutions. Changes in fluorescencestabilized withinapproximately 30 s, whereafterthe intensity of pyrene monomer fluorescence was recorded. As judged from the absorption spectra the cyt c used was mainly in the oxidized form. We did not carry out an extensive comparison between the reduced and oxidized form of cyt c at thisstage. However, our preliminaryexperiments did notindicateany major redox-statedependent differences. Yet, we wish to emphasize that the present resultsare in strictsense applicable tothe oxidized form only. Subsequently to the addition of cyt c, we included nucleoside phosphates, NaCI, or MgC12 to yield the indicated final concentrations, whereafter changes in pyrene monomer fluorescence intensity were recorded. After completing the fluorescence measurements the pHof the sample was determined. Maximal difference between the initial and final [H+]was 50 p~ at pH4. The pHof the nucleotide solutions in 20 mM MES buffer was adjusted to 4.5 or 6.0 with 1 M NaOH. Some of the advantages aswell as ambiguities of the use of pyrenelabelled lipids in energy transfer measurements have been recently discussed (Kaihovaara et al., 1991). Notably, the efficiency of resonance energy transfer from pyrene-fatty acid containing phospholipid derivatives to the heme of cyt c does not yield direct quantitative data on the binding of cyt c to liposomes. Yet, this technique allows for the rapid semiquantitative assessment of the membrane association of cyt c. In the present study we utilized low concentrations of both lipids and cyt c so as to result in a minimal inner filter effect. Unless otherwise indicated the recorded values for fluorescence were subsequentlycorrected for the intensity decrease due to dilution. Emission spectra were not corrected for instrument response. RESULTS

Effect of pH on the Bindingof cyt c to Cardiolipin-containing Liposomes-Binding of cyt c to liposomes consisting of 10 mol% cardiolipin, 85 mol% egg PC, and 5 mol% PPHPC was monitored by following the quenching of pyrene monomer fluorescence due to resonance energy transfer to cytc (Mustonen et al., 1987). Under these conditions the membrane association of cyt c was efficient and onlyinsignificantly affected by pH varied between 4.0 and 6.0 (Fig. 1).Yet, at p H 4.0 cyt c appearsto have a slightlyincreased affinityto liposomes,whereas at pH 7.0 thebinding was somewhat weaker. Over this pH range very little binding of cyt c to vesicles lacking cardiolipin was observed. Interestingly, the dependency of the degree of vesicle association of cyt c on the content of cardiolipin were strikingly different at pH 7.0 and 4.0 (Fig. 2). In accordancewith our previous studies there appears tobe a requirement for a critical membrane-negative charge densityfor the bindingof cyt c at neutral pH, resulting in a sigmoidal dependency on the content of the acidic lipid in vesicles (Mustonen et al., 1897). At p H 4.0 the affinity of cyt c to the membrane appears to be higher than at neutral pH. The binding of cyt c continues up to an apparent saturation reached at 1/10 molar ratio of cyt c to cardiolipin. Displacement of cyt c from Cardiolipin Containing Vesicles

0

200

0

400

c y t c , nM

FIG. 1. Cyt c-induced quenching of the relative fluorescence intensity of pyrene monomer emission (RFI) from PPHPC. Egg PC vesicles containing 5 mol% PPHPC and 10 mol% cardiolipin were used. Total lipid concentration was 25 pM in 20 mM MES-0.1 mM EDTA, p H 4.0 (O),5.0 (O),6.0 (V), or 7.0 (V).The values for fluorescence emission intensity have been corrected for dilution. 100, "

75

0' 0

5

10

15

20

25

cardiolipin, mol Z

FIG. 2. The binding of cyt c (400 nM) to liposomes at pH 4 (0) and 7 (0)as a function of the content of cardiolipin (mol% of total lipid). Total lipid concentration was between 22.5 and 27.5 in 20 mM MES-0.1 mM EDTA. The total number of acyl chains in liposomes was maintainedconstant (see "ExperimentalProcedures" for details).

PM

by Nucleotides-ATP, ADP, and AMPwere all able todissociate cyt c from liposomes a t neutral pH (Fig. 3a). ATP was most effective and fora half-maximal reversal of pyrene fluorescence quenching 2 mM nucleotide was needed. The corresponding value for ADP was 5 mM, whereas 50% dequenching of pyrene emission could not be achieved in the presence of up to 6.5 mM AMP. The nucleotide-induced detachment of cyt c from vesicles turned out tobe highly dependent on pH. Thus at pH 6.0 the only adenine nucleotide causing 50% reversal of fluorescence quenching was ATP, whereas ADP and AMP had only little effect (Fig. 321).When pH was lowered further to 4.9, neither ATP nor AMPinfluenced the lipid binding of cyt c, and ADP actuallyincreasedthequenching of pyrene fluorescence slightly (Fig. 3c). Thisphenomenon was evenmore pronounced at pH 4.2, where all three adenine nucleotides enhanced pyrene fluorescence quenching by cyt c with weakest effect exerted by ATP (Fig. 3d). The above pH-dependent effects of ATP on the binding of cyt c to vesicles were not specific to cardiolipin and essentially similar data at both pH 7.0 and 4.0were obtained usingvesicles containing insteadof cardiolipin either 20 mol% egg PA or egg PG.' influenced the At pH 6.9 and 4.2 both CTP and GTP membrane association of cyt c in a manner indistinguishable M. Rytomaa and P. K. J. Kinnunen, unpublished results.

Binding of Cytochrome Cardiolipin to Acidic c at 100

pH

22245

100,

a

I

75 50

RFQ 25 0

-25

t

t

RFo

t 6

4

75

2

0

rnM

AT?/ADP/AMP, 100

rk53Ed

-25

I 2

0

I/

25

t

: 0

-25t 0

,

,

I

2

,

,

,

4

ATP/ADP/AMP, -

100

75

RFQ

50 25

"-.-#-"

F ,

I J /

I

1

6 mM

t

.,

-25

1

0

, 40

,

,

,

80

, 120

,

I 160

NaCI, r n M

FIG. 5. Reversal of fluorescence quenching due to NaC1. The pH of 20 mM MES-0.1 mM EDTA was 7.0 (V), 6.0 (v),5.0 (O),or 4.0 (0). Otherwise the conditionswere as described in the legend for Fig. 3.

0

-25 0

2

4

ATP/ADP/AMP, 100

6

rnM

FIG. 4. Reversal of fluorescence quenching induced by CTP (0,V) or GTP (0,V).The pH of 20 mM MES-0.1 mM EDTA was 6.9 (V, V) or 4.2 (0,0 ) .Otherwise the conditions were as described in the legend for Fig. 3.

b

100

RFQ

4

CTP/GTP,

6

rnM

d

RFQ

50% reversal of fluorescence quenching required 10 mM phosphate (data not shown). Dissociation ofcyt c from Membranes by NaCl and MgC& T o exclude the possibility that the counterions (up to12 mM Na+) in the nucleotide solutions were responsible for the displacement of cyt c from liposomes we checked for the effects ofNaC1. In accordance with the electrostatic nature of the membrane association of cyt c salt displacedcyt c effectively from the vesicles at both pH 6.0 and 7.0 (Fig. 5). However, theconcentrations of NaCl needed were much higher than thoseof Na' brought with thenucleotides. Interestingly, the detachmentby NaCl of cyt c from liposomes was highly dependent on pH. At pH 4.0 up to 0.13 M NaCl was unable to dissociate cyt c from membranes, and on the contrary a slightly increased pyrene fluorescence quenching by cyt c was observed in the presence of salt (Fig. 5). Compared with NaCl much lower concentrations of MgC12 were needed to dissociate cyt c from cardiolipin at pH 5.0,6.0, and 7.0 (Fig. 6). Similarlyto NaCl also MgCl, slightlyincreased the quenching of pyrene fluorescence by cyt c at pH 4.0. To conclude at neutral pH bothNaCl and MgC12 dissociate cyt c from liposomes, whereas in an acidic milieu the presence of thesesaltsresultsinanaugmentedpyrene fluorescence quenching by cyt c. To find out if the enhancement of pyrene fluorescence quenching by cyt c at acidic pH in thepresence of nucleotides or NaCl was due to cyt c-lipid interactions changing from electrostatic to hydrophobic we used PPHPC-egg PC liposomes containing no cardiolipin. However, with such vesicles and at both pH 7 and 4, regardless of the presence or absence

ob=;=;

25

-25

0

2

4

6

A T P / A D P / A M Pr, n M

FIG. 3. Reversal of cyt c-induced fluorescence quenching T h e p H of 20 mM MESdue to ATP (O), ADP (O), or AMP (0). 0.1 mM EDTA was 6.9 (a),6.0 ( b ) ,4.9 ( c ) , or 4.2 ( d ) . Egg PC vesicles containing 5 mol% PPHPC and10 mol% cardiolipin were used. Total . initial fluorescence quenching lipid concentration was 25 p ~ The was induced by 470 nM cyt c. The values have been corrected for dilution and are given as a percentage of the fluorescence quenching by cyt c observed prior to the additionof the nucleotides.

from ATP (Fig. 4). Accordingly, the effects by these trinucleotides appear to be related to the phosphate groups of the nucleotides. This is further supported by the dissociation of cyt c due to the additionof phosphate. Thus, at neutral pH a

’ Binding of Cytochrome c to Cardiolipin Acidic at pH

22246 loo

RFO

RFI

0

2

10

6

4

MgCI,, mM

12

min

FIG. 8. Membrane association and detachment of cyt c by FIG. 6. Reversal of fluorescence quenching due to MgClz. subsequent acidification and neutralization, respectively, of The pH of 20 mM MES was 7.0 (’I), 6.0 (V), 5.0 (e),or 4.0 (0).the medium. The first downward-slanted arrow indicates the addiOtherwise the conditions were as described in thelegend for Fig. 3.

RF‘

”

75

-

50

-

25

-

tion of cyt e (final concentration 470nM) and the second upward slanted arrow indicates the addition of ATP (final concentration 4.8 mM). The subsequent upward and downward pointing arrowsindicate the additions of either 90 pl of 1 M NH3 or 30 p1 of 1 M HCI, respectively. The fluorescence intensity levels have not been corrected for dilution.

a

0

o

200 400 c y t c , nM

0

2 4 6 AT?/ADP, m M

560

t 0

wavelength, nrn

0

200

400

cyt c . nM

0

40

80 120 1 6 0

NoCi, m M

FIG. 9. Pyrene fluorescence emission spectra of egg PC liposomes containing 5 mol% PPHPC and 10 mol% cardiolipin in 20 mM MES-0.1 mM EDTA. At pH 4 and 7, prior tothe

FIG. 7. The effect of pH and ATP, ADP, and NaCl on the quenching of pyrene fluorescence by cyt c in egg PCvesicles additions of either cyt c or ATP, the spectra for the liposomes were containing 5 mol% PPHPC and lacking cardiolipin. Total lipid essentially identical ( A ) .No significant membrane association of cyt concentration was 25 pM in 20 mM MES-0.1 rnM EDTA, pH 4.0 (0,c (final concentration470 nM) is evident at pH 7.0 when 6.5 mM ATP is present ( B ) . Yet, subsequent acidification of the medium to pH 4 0 )or pH 7.0 (V,V). a, after the addition of cyt c (to a final concentration of 470 nM)either ATP (0,V) or ADP (0,‘I) were subsequently by 30 p1 1 M HCI leads to the binding of cyt c to the vesicles ( C ) . included. b illustrates a similar experiment but instead of nucleotides aliquots of 1 M NaCl were added to yield the indicated final concen1.ration at pH 4.0 (0) or pH 7.0 (0).

Total lipid concentration was25 pM. The spectra have not been corrected for the 9.5% decrease in fluorescence intensity due to dilution.

of the nucleotides (ATP, ADP, or AMP, up to 6.5 mM) or NaCl (up to 0.17 M) the addition of cyt c did cause only an insignificant quenchingof pyrene fluorescence thus revealing a very weak membrane association of cyt c under these conditions (Fig. 7). Regulation of the Membrane Association of cyt c by p H Taken together the experiments described above suggested that in the presence of sufficient concentrations of either of cyt c to acidic phospholipidnucleotides or salts the binding containing liposomes could be efficiently controlled by pH. This proved to be the case. Accordingly, in the presence of 4.8 mM ATP or 100 mM NaCl rapid association and detachment of cyt c to andfrom cardiolipin-containingvesicles could be triggered by varying the pHbetween 4 and 7, respectively, a s illustrated for the nucleotide in Fig. 8. Essentially identical results were obtained when NaCl was included. At both neutral as well as acidic pH the ratesof detachment and binding of cyt c to membranes were rapid and complete within ap-

proximately 5 s. Accordingly, the recorded fast kinetics are apparent only and are limited by the efficiency of mixing of the cuvette contents. In the absence of cyt c varying pH between acidic and neutral values caused only a slight decrease in fluorescence intensity entirely due to dilution (data notshown). To illustratethemagnitude of fluorescence changesas well asthesignal-to-noiseratio of theactual measured intensities the correspondingemission spectra are shown in Fig. 9. DISCUSSION

The pH dependency of the binding of cyt c to lipid membranes has been investigated previously using monolayers on a n air/water interface (Quinn and Dawson, 1969). The association of cyt c with cardiolipin films was found to remain nearly constant between pH 3 and 8. In the neutral pH range the amount of protein bound tolipid monolayers diminished in the presence of 1 M NaCl. However, at acidic pH the

Binding of Cytochrome c to Cardiolipin at Acidic pH

22247

increment in surface pressureof PC monolayers (initially at to the chargeof ADP at pH6.9. Therefore, thedifferences in the efficiency of the adenine nucleotides to detach cyt c are 10 mN/m)due to the membrane penetration of cyt cwas increased, particularly in the presence of 1 M NaC1. Accord- likely to reflect differences in their net negative charge. This being equally effective ingly, the modes of interaction of cyt c with the lipid mono- is further supported by CTP and GTP layers appear to be different in neutral and acidic milieus. as ATP indissociating cyt c from liposomes. Accordingly, the The present results show that although the tight binding of identity of the base and sugar moieties appears to be of no cyt c to cardiolipin-containing vesicles is insignificantly af- importance. Adenine nucleotides as well as phosphate ions fected by pH varied between4.0 and 6.0 the membrane affinity have been shown todissociate cyt c from cytochromec oxidase of anions to cyt (Ferguson-Milleret al., 1976) and the binding of cyt c seems to be slightly higher at the former pH. The somewhat lower affinity and degree of membrane association c has been experimentally verified (Nicholls, 1974).However, of cyt c at neutral pHcould result from a decrease in the net compared with the adenine nucleotides or phosphate much as positive charge of cyt c due to its deprotonation (Quinn and higher concentrations of the monovalent anions such chloDawson, 1969). These differences in the effects of pH on the ride are needed to dissociate cyt c from cytochrome c oxidase (Ferguson-Miller et al., 1976). Cyt c has also shown to have binding of cyt c to (i) monolayers studied by Quinn and Dawson and our data obtained usingliposomes (ii) couldarise strong affinityfor the negatively charged,heavily phosphorylfrom differences betweenthese two model systems. The initial ated protein, phosvitin (Nicholls, 1974). Thus, the displacesurface pressure in some of the monolayer experiments of ment of cyt c fromcardiolipin-containingliposomes at neutral Quinn and Dawson was rather low, 10 mN/m whereas the pH reported here is likely to result from a competition between nucleotides and cardiolipin for a common binding site equilibrium lateral surface pressure of liposomes has been estimated to be 17-25 mN/m (for a brief review see Konttila on cytc. Somewhat analogously cardiolipin is able todisplace et al., 1988). Accordingly, the extent of penetration of cyt c ADP from theinactive dnaA protein and thus reactivate this into the monolayers at such low pressures is enhanced by protein (Sekimizu and Kornberg, 1988; Sekimizu et al., 1988; Yung and Kornberg, 1988). hydrophobic interactionsandis likely to besignificantly Although cyt c is generally considered as a paradigm forthe higher than that intoliposomal bilayers. Quinn and Dawson (1969) also showed that the binding of cyt c to acidic lipid electrostatically associated peripheral membrane proteins it films a t 40 mN/m was purely electrostatic. In keeping with has also been suggested that nonionic interactions could be involved in the membrane bindingof cyt c (Quinn and Dawprevious studies fluorescence quenching by cyt ccouldbe completely prevented at neutral pH by NaCl as well as by son, 1969). Interestingly, at pH 4.0 salts and nucleotides do MgC12.The latter iseffective already at millimolar concentra- not dissociate cyt c from membranes but instead their prestions. The high affinity and binding of M e to cardiolipin ence leads to a more efficient quenching of fluorescence by cyt c thus indicating a quite drastic change in the nature of could be involved. Interestingly, at neutral pHalso millimolar concentrations interaction of cyt c with the liposome membranes. As sugof nucleotides were found to displace cyt c from cardiolipin- gested by the data inFig. 7 the binding of cyt c to liposomes containing liposomes in the sequence of efficiency of ATP = at pH4.0 in thepresence of either salts ornucleotides should significant C T P = G T P > ADP > AMP. This effect could be shown not not be due to hydrophobic interactionsasno to be due to the counterions present in nucleotides. the At p H association of cyt c with vesicles containing only the zwitter6.0 the only adenine nucleotide capableof causing dissociation ionic PC was observed. The increase in fluorescence quenchof cyt c from cardiolipinis ATP and the degree of fluorescence ing by cyt c in the presence of salts or nucleotides at low pH dequenching is comparable with that caused by ADP at pH is unlikely to involve major changes in the membrane pene6.9. The pKvalues for the secondary phosphate ionizationfor tration of cyt c as only insignificant effects were detected on ATP, ADP, andAMPare 7.6,7.2, and 6.7, respectively the efficiency of pyrene fluorescenceenergy transfer from heme.' Importantly, the depend(Saenger, 1984).Because the deprotonation of the secondary PPDPC and PPHPC to the encies of the membrane association of cyt c on the contentof phosphates of adenine nucleotides commences at p H 6.9 the net negative charge of ATP at pH 6.0 is approximately equal cardiolipin were remarkably different at neutral and acidic values forpH (Fig. 2). We have previously noted thesigmoidal dependency at pH7 (Mustonen et al., 1987). Similar behavior has been recently reported for the membrane association of basic model peptides (Mosior and McLaughlin, 1992). To summarize, the membrane binding of cyt c at pH4.0 requires acidic phospholipids and is not reversed by salts or nucleotides. Accordingly, at pH 4.0 the lipid binding of cyt c is likely to involve hydrogenbonding.At thismoment we cannot distinguish between changes in the protonationof either cyt c or lipid causing the high affinity binding of cyt c to membranes at low pH. This question is currentlybeing addressed in our laboratory. Several reasons could account for the nucleotide and salt induced increase in the quenching of pyrene fluorescence by cyt c observed a t low pH. The absorbanceof cyt c at 395 nm FIG. 10. Model for the binding of cyt c to lipid bilayers at is sensitive to pH and at acidic pH also to the presence of pH 4 and 7.At the latter pH the cyt c cardiolipin interactions are ions (Stellwagen and Babul, 1975). The emission wavelength ionic and canbe reversed by anions. At pH 4 cyt c binds tocardiolipin of pyrene fluorescence monitored in our experiments was at due to nonionic interactions, presumably by hydrogen bonding. At 394 nm. Therefore the increase in fluorescence quenching by acidic pH anions cause a conformational change in cytc thus leading cyt c at pH4.0 caused by the nucleotides and saltscould result t o a more efficient resonance energy transfer from pyrene to the heme of cyt c. See "Discussion" for further details. Thickness of the from conformational changes in cyt c. We are at present in membrane and size of cyt c are not drawn toscale. the favor of a mechanism involving at low pH an anion-

22248

Binding of Cytochrome c to Cardiolipin at Acidic pH

induced alteration in the conformation of cyt c leading to speculate that pumping of protons by the respiratory chain more efficient energytransfer from pyrene to the heme. Such and theredox-linked deprotonation of the phospholipid-cyt c conformational change could result in a change in the mag- complex (Mitchell and Moyle, 1983) generates an effective nitude of the absorption dipole of the heme as well as in its acidic pH in the intracristal spacewhich then allows for the membranebinding of cyt cregardless of the presence of reorientation withrespect to membrane bilayer plane and nucleotides and theprevailing ionic strength. Continuing this thus also with respect to the relaxationdipole of pyrene. of the intracristal In the presence of salts and nucleotides (in sufficient con- line of argument further changes in the pH centrations to cause the membrane dissociation of cyt c a t space could also play a regulatory role in controlling the neutralpH) varying p H between 4 and 7 by subsequent association of cyt c with membrane acidic phospholipids. additions of 1 M HCl and 1 M NH3, respectively, caused rapid Acknowledgments-We thank Dr. Anu Koiv for valuable discusand completely reversible changes in the membraneassociasions and Birgitta Rantala for skillful technical assistance. tion of cyt c. Therefore, pH-induced membrane fusion, forREFERENCES mation of hexagonal phase I1 by cardiolipin, and trapping of cyt c inside the lipid vesicles seem unlikely. Nuclear magnetic Brown, L. R., and Wiithrich, K. (1977) Biochim. Biophys. Acta 468,389-410 Jongh, H. H. J., and de Kruijff, B. (1990) Biochim. Biophys. Acta 1 0 2 9 , resonance studiesalso indicate thateven in thepresence of 1 de 105-112 Demel, R. A,, Jordi, W., Lambrechts, H., van Damme, H., Hovius, R., and de M Ca2+hexagonal phase I1 is not formed by 10 mol% cardiKruijff, B. (1989) J. Biol. Chem. 2 6 4 , 3988-3997 olipin mixed with l-palmitoyl-2-oleoyl-sn-glycero-3-phosphoDickerson, R. E., Takano,T., Eisenberg, D., Kallai, 0. B., Samson, L., Cooper, A,, and Margoliash, E. (1971) J. Bzol. Chem. 2 4 6 , 1511-1535 choline (Macdonald andSeelig, 1987). S.. Brautiaan. - . D. L... and Mareoliash. E. (1976) J. Biol. Chem. To summarize, we have attempted to rationalize the present Fereuson-Miller. 2 3 1 , 1104-1115 Haverstick, D. M., and Glaser, M. (1989) Biophys. J. 5 5 , 677-682 findings in terms of a model illustrated in Fig. 10. Thus, at Heimburg, T., Hildebrandt, P., and Marsh, D. (1991) Biochemistry 3 0 , 9084neutral pH cyt c associates electrostatically with the acidic 9089 phospholipid containing membrane and, accordingly, is dis- Hidebrandt, P., and Stockburger, M. (1989a) B@chem[stry28,6710-6721 Hildebrandt, P., and Stockburger, M.(1989b) Blochemstry 28,6722-6728 placed by increasing the ionic strength. Likewise nucleotides Hochman, J. H., Schindler, M., Lee, J. G., and Ferguson-Miller,S. (1982) Proc. Natl. Acad. Sci. U. S. A. 79,6866-6870 competing with the phosphatemoieties of the acidic lipids for Jori, G., Tamburro, A. M., and Ami, A. (1974) Photochem. Photobiol. 19,337a n electrostatic binding withbasic residues on the surfaceof 345 P., Raulo, E., and Kinnunen, P. K. J. (1991) Biochemistry 30, cyt c can dissociate this proteinfrom the membrane. At acidic Kaibovaara, 8380-8386 Kimelberg, H. K., and Lee, C. P. (1969) Biochem. Biophys. Res. Commun. 34, pH, however, cyt c interacts due to another site with mem784-790 brane acidic phospholipids,presumably by hydrogen bonding. Konttila, R., Salonen, I., Virtanen J. A,, andKinnunen, P. K. J. (1988) Biochemistry 27,7443-7446 The basic residues responsible for attaching cyt c to memKozarac, Z., Dhathathreyan, A., and Mobius, D. (1988) FEBS Lett. 2 2 9 , 372branes at neutral pH should not be involved. Yet, the presence 276 Lee; S . , and Kim, H. (1989) Arch. Biochem. Biophys. 2 7 1 , 188-199 of anions (nucleotides, phosphates, C1-) would cause a con- Macdonald, P. M., and Seelig, J. (1987) Biochemistry 26,6292-6298 formational change in cytc leading to an enhanced quenchingMitchell, P., and Moyle, J. (1983) FEES Lett. 151, 167-178 Mosior, M., and McLaugblin, S. (1992) Biochemistry 31,1767-1773 of the pyrene moieties residing in the hydrocarbon region of Mt~,pg:A.,Mantsch, H. H., and Surewlcz, W. K. (1991) Biochemistry 30,7219, the membrane. Mustonen, P., Virtanen, J. A,, Somerharju, P. J., and Kinnunen,P. K. J. (1987) The possible physiological significance, if any, of the presBiochemistry 26,2991-2997 e n t results remains open and essentiallyspeculative. Con- Nicholls, P. (1974) Biochim. Biophys. Acta 3 4 6 , 261-310 Nicholls, P., and Malviya, A. N. (1973) Trans. Biochem. SOC.1,372-375 cerning the salt-induced release of cyt c from mitochondrial Quinn, P. J., and Dawson! R. M. C. (1969) Biochem. J. 115,,65-75 Saenger, W. (1984) Prlnclples of NucleLc Acld Structure, Springer-Verlag, New membranes, it is noteworthy that the intracristal space beYork comes accessible to salt only after swelling (Hochman et d., Sekimizu, K., and Kornberg, A. (1988) J. Biol. Chem. 2 6 3 , 7131-7135 Sekimizu, K., Yung, B. Y., and Kornherg, A. (1988) J. Biol. Chem. 2 6 3 , 71361982). Likewise, based onelectron microscopy studieson 7140 mitochondria the amountof water in the intracristalspace is Sjostrand, F. S. (1983) in,Membrane Fluidity in Biology (Aloia, R. C., ed) pp. 83-142, Vol. 1,Academlc Press, New York likely to be limited (Sjostrand, 1983).Thus, the conditions in Soussi, B., Bylund-Fellenius, A.-C., Schersthn, T., and Angstrom, J. (1990) Biochem. J . 2 6 5 , 227-232 the mitochondrial intracristal space can be assumed to differ S. H..Neu. C. A,. Swanson. M. S., and Maraoliash, E. (1983) FEES from those prevailing elsewhere in the intermembranespace SDeck. 'Lett 164,'379-382 as well from those outside the mitochondria, in the cytoplasm. Spooner, P. J. R . , and Watts, A. (1991) Biochemistry 30,3871-3879 Steinemann, A,, and Lau er, P (1971) J. Membrane Biol. 4 , 74-86 Nevertheless, if the strong affinity of cyt c to cardiolipin is Stellwagen, E., and BabufJ. (i975) Biochemistry 14,5135-5140 relevant to the mitochondrial electron transport activity the Szebeni, J., and Tollin,G. (1988) Biochim. Biophys. Acta 9 3 2 , 153-159 Yung, B. Y., and Kornberg, A. (1988) Proc. Natl. Acad. Scr. U. S. A. 8 5 , 7202intracristal milieu should, (i) if the effective pH is neutral, 7205 J., Erecinska, M., and Chance, B. (1973) Arch. Biochem. Biophys. have low levels of nucleotides as well as inorganic ions, or (ii) Vanderkooi, 1 5 4 , 219-229 alternatively, intracristal p H would have to be acidic (pH < Vik, S. B., Georgevich, G., and Capaldi, R. A. (1981) Proc.Natl.Acad. Sci. U. S. A. 78,1456-1460 5) so as to ensure the membraneassociation of cyt c even in Waltham, M. C., Cornell, B. A,, and Smith, R. (1986) Biochim. Biophys. Acta 862,451-456 the presence of saltsand nucleotides. Therefore, we may '

LL4

'