Proc. Nati. Acad. Sci. USAVol. 77, No. 8, pp. 4439-4443, August 1980Biochemistry

Mapping of anion binding sites on cytochrome c by differentialchemical modification of lysine residues

(carboxydinitrophenyl-cytochrome c/trinitrophenyl-cytochromecarbonate binding site)

!c/phosphate binding sites/chloride binding sites/

NEIL OSHEROFF*, DAVID L. BRAUTIGANt, AND E. MARGOLIASHtDepartment of Biochemistry and Molecular Biology, Northwestern University, Evanston, Illinois 60201

Contributed by Emanuel Margoliash, April 11, 1980

ABSTRACT The carbonate binding site on horse cyto-chrome c was mapped by comparing the yields of carboxydi-nitrophenyl-cytochromes c, each with a single carboxydini-trophenyl-substituted lysine residue per molecule, when themodification reaction was carried out in the presence and ab-sence of carbonate. The site is located on the "left surface" ofthe protein and consists of lysine residues 72 and/or 73 as wellas 86 and/or 87 (Carbonate Site). Although one of the bindingsites for phosphate on cytochrome c (Phosphate Site 1) is locatednear the carbonate site, the sites are distinctly different sincecarbonate does not displace bound phosphate, as monitored by31P NMR. Furthermore, citrate interacts with Phosphate SiteI with high affinity, whereas chloride, acetate, borate, and ca-codylate have a much lower affinity for this site, if they bindto it at all. The affinity of phosphate for Phosphate Site I (KD= 2 X 10-4 M) is at least 1 order of magnitude higher than it isfor other sites of interaction.

Cytochrome c is a highly basic protein and it is not surprisingthat it interacts with anions (1). The earliest direct evidence foranion binding came from the observation that certain anionsdecreased the electrophoretic mobility of the protein (2, 3).Binding stoichiometries, affinities, and locations have since beenexamined by a variety of techniques. Among these are directmeasurements by gel filtration (4) as well as studies of the effectsof anions on (i) the reduction potential of cytochrome c (5, 6),(ii) theJH NMR spectrum of the protein (7), (iii) the chroma-tographic mobilities of carboxydinitrophenyl cytochromes c,each with a single carboxydinitrophenyl-substituted lysineresidue per molecule [monoCDNP-cytochrome(s) c] (8,9), (iv)the interaction of various native and chemically modified cy-tochromes c with cytochrome c oxidase (10-13, §), cytochromec peroxidase (14, 15), or nonphysiological oxidants and reduc-tants (16-22), (v) the interaction of phosphate (monitored by31P NMR) (12, §) or of chloride (monitored by 31C1 NMR) (23)with the protein, and (vi) the oxidation of threonyl residues afterthe reduction of ferricytochrome c by ferrous ion (22).

Notwithstanding this work, there is no consensus as to thelocations at which various anions bind to cytochrome c or theirbinding affinities. Since lysines constitute most of the protein'scationic side chains, our studies of lysine modification by re-action with 4-chloro-3,5-dinitrobenzoate (8, 9, 24) and 2,4,6-trinitrobenzene sulfonate (25) have afforded an opportunityto modify the e-amino groups directly involved in anionbinding. The relative proportions of the monoCDNP-cyto-chromes c, modified at lysine residue 7, 8, 13, 22, 25, 27, 39, 60,72, 73, 86, 87, or 99 in the presence and in the absence of car-bonate indicated that the side chains of lysine residue 72 and/or73 as well as 86 and/or 87 were involved in the binding of thisanion. Although lysine 87 appears to be within one of the

The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be hereby marked "ad-vertisement" in accordance with 18 U. S. C. §1734 solely to indicatethis fact.

4439

binding sites for phosphate (Phosphate Site I) (8, 9, 12, 22),which is also a binding site for citrate (12), the Carbonate Siteis distinctly different. Moreover, chloride, acetate, sulfate,borate, and cacodylate did not displace phosphate from Phos-phate Site I.

MATERIALS AND METHODSHorse cytochrome c was prepared by the method of Margoliashand Walasek (26) as modified by Brautigan et al. (27). Trini-trophenyl (TNP)-cytochrome c was prepared by the procedureof Wada and Okunuki (28) as modified by Osheroff et al. (25).Cytochrome c. (1 mM) was treated with a 5-fold molar excessof 2,4,6-trinitrobenzene sulfonate (Pierce) in either 200 mMsodium phosphate or 200 mM sodium cacodylate (pH 7.0) at230C. The reaction was stopped after approximately 1.5 TNPgroups had been incorporated per molecule of cytochrome c,as determined from the absorbance of TNP-lysine at 345 nm(28). MonoTNP-cytochrome c substituted at lysine residue 13was purified by cation-exchange chromatography on columnsof CM-cellulose (microgranular, Whatman) as described(25).MonoCDNP-cytochrome c was prepared by the procedure

of Brautigan et al. (8, 9). Cytochrome c (1 mM) was treatedwith a 5-fold molar excess of 4-chloro-3,5-dinitrobenzoate in200mM sodium borate or 200 mM sodium carbonate (pH 9.0)at 230C. The reaction was stopped after approximately 1.5CDNP groups had beenm incorporated per molecule of cyto-chrome c, as determined by the absorbance of CDNP-lysineat 450 nm (8). MonoCDNP-cytochromes c substituted at lysineresidue 7, 8, 13, 22, 25, 27, 39, 60, 72, 73, 86, 87, or 99 werepurified on columns of CM-cellulose (9) in sodium boratebuffers (pH 8.6) that contained 2.5 M ethanol (24).The substitued lysine residues were identified by peptide

mapping of chymotryptic and tryptic digests and by deter-mining the amino acid compositions of the modified peptides(9, 24). The distributions of reaction products were determinedas described (8).

Pulsed Fourier transform 31P NMR spectra were the averagesof 100 to 500 transients and were recorded by using 10 mm

Abbreviations: TNP-, 2,4,6-trinitrophenyl; CDNP-, 4-carboxy-2,6-dinitrophenyl; monoCDNP-cytochrome(s) c, cytochrome(s) c with asingle CDNP-substituted lysine residue per molecule; monoTNP-cytochrome(s)c, cytochrome(s)c with a singleTNP-substituted lysineresidue per molecule.* Present address: Department of Biochemistry, Stanford UniversitySchool of Medicine, Stanford, CA 94305.

t Present address: Department of Biochemistry, University of Wash-ington, Seattle, WA 98195.

* To whom reprint requests should be addressed.§ Osheroff, N. & Margoliash, E. (1977) 174th American ChemicalSociety Meeting, Division of Biological Chemistry, Chicago, IL,Abstr. 83; unpublished data.

4440 Biochemistry: Osheroff et al.

tubes with a JEOL FX-90Q 90 MHz NMR spectrometer. Thebinding of phosphate to cytochrome c was examined in 8 mMphosphate, over a range of 0.5 to 20 mM cytochrome c. Com-petition between other anions and phosphate was examined in8 mM phosphate/10 mM ferricytochrome c with 2-32 mMcarbonate, citrate, chloride, acetate, sulfate, borate, or caco-dylate. All samples were adjusted to pH 7.8 with Tris; all con-tained 0.2 mM EDTA. 2H20 (final concentration, 10%, vol/vol)was added to secure a proper lock. The signal-to-noise ratio wasapproximately 40:1. The downfield shift of the phosphate peakin the presence of cytochrome c from the position of freephosphate at pH 7.8 was used as a measure of binding. Theresults were analyzed according to the method of Hughes andKlotz (29).

RESULTSAnions have a profound influence on the reactivity of lysineresidues in cytochrome c (Fig. 1). Indeed, the rate of trinitro-phenylation of the protein at pH 7.0 was 2.6 times faster incacodylate than in phosphate (Fig. 1A), and the rate of car-boxydinitrophenylation of horse cytochrome c at pH 9.0 was4.6 times faster in borate than in carbonate (Fig. 1B).

Although the proportion of monoTNP-cytochrome c formedwas the same when the reaction was carried out in phosphate(40.6% of total cytochrome c) or in cacodylate (40.3%), thedistribution of products was markedly different. In the elutionprofiles of the trinitrophenylation reaction mixtures, one seesthat the yield of fraction I in the cacodylate buffer (Fig. 2A)was 21% greater than its yield in the phosphate buffer (Fig. 2B).This occurred at the expense of fraction II, whose yield de-creased by 34%. Unfortunately, no additional information couldbe obtained from the trinitrophenylation. Despite multiplechromatographic separations, the only reaction product thatcould be isolated in pure form was monoTNP-cytochrome csubstituted at lysine residue13, which made up 82% of fractionI in the cacodylate- and 77% in the phosphate-buffered reac-tion.To determine the influence of anions on the reactivity of

many of the protein's 19 lysine residues, horse cytochrome cwas reacted with 4-chloro-3,5-dinitrobenzoate in carbonate andin borate buffers at pH 9.0. Thirteen different monoCDNP-cytochromes c were separated in pure form and their relativeyields determined (8, 24). The proportion of monoCDNP-cytochromes c in the reaction mixture obtained in the carbonatebuffer (38.0% of total cytochrome c) was essentially the sameas that obtained in the borate buffer (40.3%). However, the

1.4 1.4

1.2 1.2o 0

1.0 1.0 .

>' 0.8 0.8

r- 0.6 0.6a

0.4 B

Z 0.2 0.2 Z

0 2 4 6 8 10 0 0.2 0.4 0.6 0.8 1.0Reaction time, hr

FIG. 1. The effect of anions on the reactivity of lysine residuesin cytochrome c. (A) Trinitrophenylation of horse cytochrome c inphosphate (o) and in cacodylate (-) at pH 7.0. (B) Carboxydinitro-phenylation of horse cytochrome c in carbonate (0) and in borate (-)at pH 9.0.

10toC6uNko

+

IZ

Elution volume, litersFIG. 2. Elution profiles of trinitrophenylation reaction mixtures

containing 0.5 g of total cytochrome c. Chromatography was on col-umns of CM-cellulose (2.5 X 115 cm) in linear gradients of sodiumphosphate buffer, pH 7.5. Trinitrophenylation reaction was in caco-dylate buffer (A) and in phosphate buffer (B).

distribution of products differed greatly (Fig. 3). The moststriking difference was that the monoCDNP-cytochromes csubstituted at lysine residues 73 and 86 were obtained in sig-nificant quantities in the presence of borate but were com-pletely absent when carbonate was used. Additionally, thepercent yield of the monoCDNP-cytochromes c substituted atlysine residues 72, 87, and 13 increased 52%, 46%, and 12%,respectively, in borate buffer, compared with carbonate buffer.These increases took place at the expense of modification atevery other lysine residue in the protein.The simplest explanation of the ability of carbonate to shield

specific lysine residues from chemical modification is that theanion interacts with their e-amino groups. Thus, the CarbonateSite must be located on the left side of cytochrome c (Fig. 4),with the center of the binding domain being equidistant fromlysine residues 73 and 86 (6.8 A from the /3-carbons of either

30

20 H

lo1

BC BC BC BC BC BC BC BC BC BC BC BC13 72 86G 87 8 27 73 7I25 39 60 99 22

C'DNP- cvtochronme c

FIG. 3. Distribution of monoCDNP-cytochromes c. Reactionswere carried out in borate (B, open bars) and carbonate (C, stippledbars) buffers. The percent yields were calculated by integratingchromatographic elution profiles as described (8). The yields in thepresence of carbonate were taken from Brautigan et al. (8,9). BecausemonoCDNP-cytochromes c substituted at lysine residues 7 and 25were not resolved in the original study, they are listed together.

2D

Q,F. ;.0 -, 7[x .-r.: F- I-fl -fl

Proc. Natl. Acad. Sci. USA 77 (1980)

Proc. Natl. Acad. Sci. USA 77 (1980) 4441

FIG. 4. Location of the Carbonate Siteon cytochrome c. Diagram of ferricytochromec from an electron-density map of the tunaprotein at a resolution of 2.0 A, according toSwanson et al. (30). The larger circles rep-resent the a-carbons and the smaller circlesrepresent the side chain atoms. The proteinis viewed from the left side. The front of theprotein, the surface which contains the ex-posed heme edge, is toward the right. Theheavier circles and filled circles mark lysineresidues 72, 73, 86, and 87, which make up theCarbonate Site. The approximate size andlocation ofbound carbonate are representedby the large stippled circle.

lysine) and 8.0 and 9.4 A from the /3-carbons of lysine residues72 and 87, respectively. Since the van der Waals radius of car-bonate is about 2.5 A (31) and the fl-carbon-to-E-amino distancein lysine is 5 A, these four lysines are all within bonding distanceof the anion. The fact that this site has the highest net positivecharge on the surface of the protein (32, 33) enhances theprobability of its serving as an anion binding site. The only otherlysine residue that showed some shielding by both phosphate(Fig. 2) and carbonate (Fig. 3) was lysine 13. This effect wassmall and may have resulted from either a weak interaction ofthe anions with this residue or an indirect influence of thepolyvalent anion binding.The anion binding site described above is in the same general

region of the protein as one of the two sites ascribed to phos-phate (Phosphate Site I) (8, 9), to citrate (12), and to a phos-phate-ferrous ion complex (22). To determine whether theCarbonate Site was indeed the same as this phosphate/citratesite, 31P NMR competition studies were carried out. In thepresence of the protein, it was found that the phosphate peakshifted downfield by a maximum of about 0.4 ppm. Upon ad-dition of citrate, the phosphate peak shifted back about a thirdof the way toward the position of unbound phosphate. In-creasing the competing anion concentration beyond 12 mM(compared to 8 mM phosphate) caused no further shift (Fig.5A). Carbonate, chloride, borate, acetate, and cacodylate didnot shift the phosphate peak, even when concentrations 4 timesthat of the phosphate were used. This indicates that citrate cancompete with phosphate at only one of the phosphate bindingsites, while the other anions must interact with the phosphate

binding sites with much lower affinities, if they interact at all.Clearly, even though the Carbonate Site and Phosphate Site Iare both located in the area of lysine-87, they are different,because carbonate does not compete with bound phosphate.The binding of phosphate was titrated by varying the con-

centration of ferricytochrome c at a constant concentration ofthe anion at pH 7.8 (Fig. SB). This revealed the presence of onehigh-affinity phosphate binding site with an apparent KD ofabout 2 X 10-4 M and one or more sites of lower affinity withapparent KD values in the millimolar range. In the presenceof citrate, the high-affinity binding of phosphate was greatlydecreased, whereas the low-affinity binding was largely un-affected (Fig. 5B). This demonstrates that the one commonbinding domain for citrate and phosphate (Phosphate Site I)is in fact the high-affinity phosphate binding site. The presenceof 24 mM chloride had no effect whatsoever on the high-af-finity interaction of phosphate with cytochrome c, but highratios of chloride to cytochrome c clearly diminished bindingat the lower-affinity sites.

DISCUSSIONThe present work has mapped the binding site for carbonateon cytochrome c. It is located on the left side of the protein (Fig.4), centered between lysine residues 72, 73, 86, and 87. Al-though Phosphate Site I is located on the top left of the proteinin the area of lysine-87, it appears to be different from theCarbonate Site. Moreover, Phosphate Site I was shown to be thehigh-affinity binding site for phosphate, and it also interactswith citrate.

Biochemistry: Osheroff et al.

4442 Biochemistry: Osheroff et al.

04

._5)0

0.3

0

0 10 20 30Anion, mM

I I

B

,I __0.5 1.0 1.5

r

FIG. 5. (A) Competition by various anions for phosphate boundto ferricytochlrome c as monitored by 31P NMR, showing the effectof citrate (0) and the lack of effect of carbonate (e), chloride (-),borate (O), acetate (&), and cacodylate (-). The chemical shifts arein ppm from the position of free phosphate, pH 7.8. (B) Effect of ci-trate (12 mM) and chloride (24 mM) on the binding of phosphate (8mM) to ferricytochrome c as monitored by 3Up NMR. These data are

analyzed according to the equation r/[A] = -r/KD + n/KD, in whichr is the ratio of phosphate bound per cytochrome c, [A] is the con-

centration (mM) of free phosphate, KD is the dissociation constant,and n is the number of binding sites (29). Phosphate (0); phosphateand chloride (-); phosphate and citrate (0).

All of these anions increase the stability of the closed hemecrevice structure of cytochrome c, measured by the pKa of the695 nm absorption band (25), and inhibit noncompetitively thereaction between cytochrome c and cytochrome c oxidase, as

monitored polarographically by using the dye N,N,N',N'-te-tramethylphenylenediamine to reduce the enzyme-boundcytochrome c (refs. 10 and 12, §). Chloride, acetate, borate, andcacbdylate, on the other hand, at concentrations in the milli-molar range do not displace phosphate from Phosphate Site I(Fig. 5) or produce either of the above effects (10, 12, 25, §).Although chloride has been reported to bind to cytochrome c

(2, 3) as tightly as phosphate (1, 5-7, 20, 21, 23), clearly it mustinteract with some other area or areas of the protein.

Three other anion binding sites on cytochrome c have beenreported. One is a second phosphate site (Phosphate Site II),located on the lower right front surface of the protein in thevicinity of residues 25, 26, and 27 (8, 9, 13, 18), Lys-His-Lys,in horse cytochrome c. This site was first defined from the effectof phosphate on the 'H NMR spectrum of the protein (18). Ithas since been confirmed from the influence of phosphate onthe ion-exchange chromatography of monoCDNP-cytochromesc (8, 9) and on the kinetics of reaction of yeast iso-i and iso-2cytochromes c with beef cytochrome c oxidase (13). Whereasphosphate and nucleotide di- and triphosphates interact withthis site, citrate, chloride, maleate, acetate, sulfate, and caco-

dylate do not (8, 9, 12, 13, 18, §).Two chloride binding sites, originally observed by free-flow

electrophoresis (2, 3), have also been described. One appears

to be located on the front of the molecule in the vicinity of theexposed heme edge (20, 23). The only residue that has beenimplicated in this binding site is lysine 13 (20). The other wasdefined from the effect of anions on the 'H NMR peak corre-sponding to the e-amino protons of lysine residue 60, locatedon the back surface of horse cytochrome c (18). Only halideanions were found to interact with this site.The presence of several different anion binding sites on horse

cytochrome c may resolve apparent contradictions in the lit-erature. Whereas Stellwagen and Shulman (18) found thatchloride did not interact with Phosphate Site II and Osheroffet al. (12, §) found that chloride did not displace phosphatefrom either Phosphate Site I or Phosphate Site II, Andersson etal. (23), using MsCI NMR, reported that phosphate could dis-place bound chloride. Clearly, the different authors are mon-itoring different anion binding sites. Although chloride doesnot interact with the high-affinity Phosphate Site 1 (12, §) (Fig.5) or with Phosphate Site 1 (18), it can displace phosphate fromsome low-affinity binding sites (Fig. 5). Similarly, phosphatewill interact with chloride sites in the concentration rangestested (23).A further complication is that the reported dissociation

constants for the binding of phosphate to ferricytochrome cvary from 2.3 X 10-5 M to 5.9 X 10-2 M (4, 6, 16, 23) and toferrocytochrome c, from 1 X 10-2M to no detectable interac-tion (4, 6, 21, 23). An even wider range has been reported forthe binding of chloride, with KD values varying from 2 X 10-6M to 5.9 X 10-2 M for ferricytochrome c (4, 6, 20, 23) and from1 X 10-2 M to no detectable binding for ferrocytochrome c (4,6, 20, 23). Whether these discrepancies reflect the binding todifferent anion sites on the protein or differences in the tech-niques and conditions employed has yet to be thoroughly in-vestigated.

Although anion binding to the Carbonate Site or to PhosphateSites I or II has an influence on the activity of cytochrome c withsome of its mitochondrial electron-exchange partners, studiedby various in vitro assay systems with carbonate, citrate, ATP,ADP, and phosphate concentrations similar to those occurringin vio (10-15, §), the physiological significance of anionbinding, if any, is unclear. The fact that high-affinity anionbinding sites have been conserved during the course of evolutionwould tend to argue for an important biological role.

We gratefully acknowledge the expert help of Dr. K. A. Christensen(Analytical Services Laboratory, Department of Chemistry, North-western University) in obtaining the NMR spectra. This work wassupported by Grants GM-19121 and HL-11119 from the NationalInstitutes of Health. For part of this work, N.O. was a trainee underGrant T32-GM-07291 from the National Institutes of Health.

1. Nicholls, P. (1974) Biochim. Biophys. Acta 346,261-310.2. Barlow, G. H. & Margoliash, E. (1966) J. Biol. Chem. 241,

1473-1477.3. Margoliash, E., Barlow, G. H. & Byers, V. (1970) Nature (Lon-

don) 228, 723-726.4. Margalit, R. & Schejter, A. (1973) Eur. J. Biochem. 32, 500-

505.5. Schejter, A. & Margalit, R. (1970) FEBS Lett. 10, 179-181.6. Margalit, R. & Schejter, A. (1973) Eur. J. Blochem. 32, 492-

499.7. Stellwagen, E. & Shulman, R. G. (1973) J. Mol. Biol. 75,683-

695.8. Brautigan, D. L., Ferguson-Miller, S. & Margoliash, E. (1978) J.

Biol. Chem. 253,130-139.9. Brautigan, D. L., Ferguson-Miller, S., Tarr, G. E. & Margoliash,

E. (1978) J. Biol. Chem. 253, 140-148.10. Ferguson-Miller, S., Brautigan, D. L. & Margoliash, E. (1976) J.

Biol. Chem. 251, 1104-1115.

I , ft o_IL 0 n& rl- C- 2- a

A0.2

1.0

'O 0.75

x- 0.50

0.25

- -

Proc. Natl. Acad. Sci. USA 77 (1980)

- IIF7 " m --W A.& A.&--

Proc. Natl. Acad. Sci. USA 77 (1980) 4443

11. Ferguson-Miller, S., Brautigan, D. L. & Margoliash, E. (1978) J.Biol. Chem, 253, 149-159.

12. Osheroff, N., Koppenol, W. H. & Margoliash, E. (1978) inFrontiers of Biological Energetics, eds. Dutton, P. L., Leigh, J.S. & Scarpa, A. (Academic, New York), pp. 439449.

13. Detbmers, J. K., Ferguson-Miller, S. & Margoliash, E. (1979)1.Biol. Chem. 254, 11973-11981.

14. Kang, C. H., Ferguson-Miller, S. & Margoliash, E. (1977) J. Biol.Chem. 252,919-926.

15. Kang, C. H., Brautigan, D. L., Osheroff, N. & Margoliash, E.(1978) J. Blol. Chem. 253,6502-6510.

16. Yamazaki, L. (1962) J. Biol. Chem. 237,224-229.17. Taborsky, G. (1972) Biochemistry 11, 729-735.18. Stellwagen, E. & Shulman, R. G. (1973) J. Mol. Biol. 80,559-

573.19. Miller, W. G. & Cusanovich, M. A. (1975) Biophys. Struct.

Mechanism 1,99-111.20. Morton, R. A. & Breskvar, K. (1977) Can. J. Blochem. 55,

146-151.21. Peterman, B. F. & Morton, R. A. (1979) Can. J. Biochem. 57,

372-377.22. Taborsky, G. & McCollum, K. (1979) J. Biol. Chem. 254,

7069-7075.23. Andersson, T., Thulin, E. & Forsen, S. (1979) Biochemistry 18,

2487-2493.

24. Osheroff, N., Brautigan, D. L. & Margoliash, E. (1980) J. Biol.Chem., 255,8245-8251.

25. Osheroff, N., Borden, D., Koppenol, W. H. & Margoliash, E.(1980) J. Biol. Chem., 255,1689-1697.

26. Margoliash, E. & Walasek, 0. F. (1967) Methods Enzymol. 10,339-348.

27. Brautigan, D. L., Ferguson-Miller, S. & Margoliash, E. (1978)Methods Enzymol. 53,128-164.

28. Wada, K. & Okunuki, K. (1969) J. Biochem. (Tokyo) 66,249-262.

29. Hughes, T. R. & Klotz, I. M. (1956) Methods Biochem. Anal. 3,265-299.

30. Swanson, R., Trus, B. L., Mandel, N., Mandel, G., Kallai,0. B. & Dickerson, R. E. (1977) J. Biol. Chem. 252, 759-775.

31. Huheey, J. E. (1972) Inorganic Chemistry: Principles of Struc-ture and Reactivity (Harper & Row, New York).

32. Koppenol, W. H., Vroonland, C. A. J. & Braams, R. (1978) Bio-chim. Biophys. Acta 503,499-508.

33. Koppenol, W. H., Ferguson-Miller, S., Osheroff, N., Speck, S. H.& Margoliash, E. (1979) in Oxdases and Related Redoox Systems,eds. King, T. E., Mason, H. S. & Morrison, M. (University ParkPress, Baltimore, MD), in press.

Biochemistry: Osheroff et al.