Eur. J. Biochem. 187,425-430 (1990)

0FEBS 1990

Antigenic sites on cytochrome c2 from Rhodospirihm rubrum Bashar SAAD and Hans Rudolf BOSSHARD

Biochemisches lnstitut der Universitit Zurich, Switzerland (Received July 13/September 8, 1989) - EJB 89 0873

The antigenic determinants for three monoclonal antibodies against cytochrome c2 from Rhodospirillum vuhrum were partially characterized by differential chemical modification of free and antibody-bound cytochrome c2 and by cross-reactivity analysis with different antigens. Circular dichroism spectroscopy was used to probe the effect of antibody binding on the conformation of cytochrome c 2 . The binding of two antibodies was strongly dependent on the native folding of the antigen. The first antibody bound to a determinant around the exposed heme edge on the ‘front side’ of the molecule which is not antigenic in mitochondrial cytochrome c 2 . Binding of this antibody to cytochrome c increased the induced CD of the ferric heme in a manner similar to that observed previously when mitochondrial cytochrome-c oxidase bound to the front side of cytochrome c. This observation points to a subtle conformational adaptation of the antigen induced by the antibody. The determinant for the second antibody, which also affected the heme CD spectrum of the antigen, was on a polypeptide loop where cytochrome c2 differs from mitochondrial cytochrome c by an eight-residue insertion. The third antibody, which did not induce a change in CD, bound to a sequential determinant near the amino end of cytochrome c2. Only this antibody cross-reacted with isolated cytochrome-c-derived peptides and with apo-cytochrome c2. A preliminary analysis of the polyclonal immune response of five rats against cytochrome c2 indicates that, unlike in eukaryotic cytochrome c, antigenic determinants are distributed over the whole polypeptide chain of the prokaryotic immunogen.

The amino acid sequences of eukaryotic cytochromes c are highly conserved and the proteins maintain the same polypeptide backbone structure [I]. The cytochrome c family is therefore ideally suited for the analysis of the immune response against topographic antigenic determinants, which are the determinants due to amino acid substitutions in proteins which have the same polypeptide backbone conformations [2]. The antigenic structure of eukaryotic cytochromes c has been investigated in detail (review in 131). In general, the predominant epitopes in eukaryotic cytochromes c correspond to those few sequence positions where the protein used for immunization and the homologous protein of the immunized animal differ, although antibodies against conserved sites were observed occasionally [4, 51. Most of the antigenic sites of cytochromes c strongly depend on the conformation of the globular molecule [3],and some sites are assembled [6].The latter consist of residues remote in sequence but close in space. There seem to be very few sequential epitopes, which are defined operationally by cross-reactivity of an anti-(cytochrome c) antibody with isolated cytochrome c peptides [7, 81. We know of no report of an antibody binding to an antigenic determinant on the front side of eukaryotic cytochrome c. The front side is the area where the heme edge is accessible on the molecular surface and where electron exchange reactions with Correspondence to H. R. Bosshard, Biochemisches Institut der Universitit, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland Enzymes. cc-Chymotrypsin (EC 3.4.21 . I ) ; Sttrphylorocrus mreus V8 protease (EC 3.4.21.19).

physiological reaction partners take place (review in [9]). The front side of mitochondrial cytochrome c is a highly conserved topographical area of the molecule. The interactions of a few monoclonal anti-(horse cytochrome r ) antibodies with cytochrome c have been partially characterized by a combination of mapping techniques [I 0 151. As for polyclonal antisera, the epitopes for the monoclonal antibodies were also within the few regions of sequence variation between mouse and horse cytochrome c. The analysis of several hundred monoclonal antibodies against mitochondria] cytochrome c has shown that over 80% of the antibodies do not cross-react with cytochrome c peptides, indicating that they are directed against conformational and possibly assembled epitopes [I 61. To overcome the limited antigenicity of mitochondrial cytochrome c we have now extended the characterization of the antigenic structure of c-type cytochromes to cytochrome c 2 from the purple bacterium Rhodospirillum ruhvuz. In the present study we present a preliminary analysis of the polyclonal immune response against cytochrome c2 and the partial characterization of the epitopes recognized by three monoclonal antibodies. We used differential acetylation of lysine amino groups of free and antibody-bound cytochrome c2 to locate approximately the binding sites of the monoclonal antibodies [12]. The method rests on the expectation that a lysine residue which is located at an antibody-protected site becomes less reactive. Accordingly, lysine residues which were less acetylated in the antibody-bound cytochrome c were assigned to the antigenic sites. The results from this protein-chemical mapping were complemented by ELISA with derivatives and

426 fragments of cytochrome c2, and by CD measurements on free and antibody-bound cytochrome c. EXPERIMENTAL PROCEDURES

Antigens Bacterial cytochrome c2 was isolated from R . rubrum (carotenoid-less mutant strain G-9) and purified as before [I 71. Cytochrome c2 peptides were prepared by digesting cytochrome c2 with the Glu-specific protease from Staphylococcus aureus V8 (Miles) in 0.1 M Hepes, pH 8.0, 37"C, 6 h, at a protein/enzyme ration of 10: 1. Peptides were separated by HPLC on a reversed-phase column (Nucleosil300, Stagroma AG, Diibendorf, Schwciz), using a 1.8% min-' linear gradient of acetonitrile in 0.067% (by vol.) aqueous trifluoroacetic acid. The peptides werc identified by amino acid analysis and partial Edman degradation [17]; their concentrations were calculated from the amino acid analysis. Cross-contamination between purified peptides did not exceed 5%. Contamination Apo-cytochrome with intact cytochrome c2 was below 0.1YO. c2 and cytochrome c2 which was fully acetylated at its 16 lysine-side-chain amino groups were prepared as before [I 71.

Monoclonal antibodies Female Lovine rats were immunized with monomeric cytochrome c2 or glutaraldehyde-polymerized [I 81 cytochrome c2, applying 0.1 mg/subcutaneous injection. After the primary immunization in complete Freund's adjuvant, booster injections in incomplete ad,juvant were given after 4 and 8 weeks, and 4 days before fusion. Spleen cells from immunized rats were fused with FO myeloma cells [19]. Hybridoma cultures were screened for their ability to secrete antibodies against cytochrome c2 by ELISA on cytochrome-c,-coated microtiter plates, as described bclow. Cytochrome-c2-positive hybridomas were cloned and subcloned by limiting dilution. Ascites fluid was produced by transferring cloned hybridomas into pristane-primed irradiated (8 Gy) rats. Antibodies from ascites fluid were purified by chromatography on DEAE-cellulose. To this end, ascites was diluted with 3 vol. buffer A (0.2 g KH2P04, 1.15 g Na2HP0,, 2 H 2 0 , 8 g NaCI/I, 0.2 g KCl/ 1) and dialyzed against 18% (by mass) sodium sulfate. The precipitate was collected, dissolved in 10 mM sodium acetate, pH 5.5, and applied to a DEAE-cellulose column equilibrated and eluted in the same buffer. This column adsorbs most of the contaminating proteins while the antibodies are eluted with the buffer front. Purification was achieved by chromatography on a sccond DEAE-cellulose column, equilibrated in 10 mM potassium phosphate, pH 8, and eluted with a linear gradient in the range 20-250 mM potassium phosphate, pH 8. The purity of the antibodies was checked by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate and by isoelectric focusing. Antibody concentration was determined from = 1.4 cm Polyclonal antisera Blood was drawn from the tail vein of rats I0 days after they had received the scond booster injection (see above for immunization protocol). Serum was collected by centrifugation and stored at - 70 "C until used. The polyclonal antibody titer was defined as the serum dilution which in the ELISA yielded an absorbance value twice the background obtained on bovine-serum-albumin-cotaed wells (see below).

Dissociation constants To plastic tubes were added 25 pl each of rabbit serum (as a carrier protein), 1251-labeled[20] cytochrome c2 (40000 cpm), unlabeled cytochrome c2 (final concentration in the range 1 pM to IOnM), and 500ng antibody. Tubes were incubated at room temperature for 2 h. The antigen-antibody complex was precipitated with 1 vol. saturated ammonium sulfate, the precipitate washed and counted. Unspecific precipitation of radioactivity was measured and corrected for by replacing unlabeled cytochrome c, for bovine serum albumin. Dissociation constants were calculated from Scatchard plots.

ELISA These were performed on 96-well microtiter plates as described before in more detail [14]. Briefly, plates were coated with 1 pg cytochrome c2, 20 pM cytochrome c, peptides or apo-cytochrome c2, in 50 pl buffer A for 1 h at 37 C, or overnight at 4°C. In control experiments, it was shown that the amount of coated antigen was not limiting since a higher amount of antigen in the coating solution did not increase the color yield of the ELISA with 10 pg/ml monoclonal antibodies. To measure the cross-reactivity by different antigens, monoclonal antibodies (10 pg/ml) were preincubated with increasing amounts of competing antigen for 30 min at 4 'C, before being added to the cytochrome-c2-coated microtiter plate. Detection of plate-bound monoclonal antibodies was achieved with phosphatase-coupled anti-(rat IgG) antibodies. Adsorption was determined at 405 nm after a I-h incubation at room temperature withp-nitrophenylphosphate. The background, mostly below A = 0.2. was determined using 0.1 % (by mass) bovine serum albumin in buffer A for plate coating.

Direrentid acelylution Differential acetylation of free and antibody-bound cytochrome c2 was performed as described in detail before [12, 131. Briefly, 50 nmol cytochrome c2 in 1 ml80 mM potassium phosphate, 70 mM NaCI, pH 7.8 (experiment F) and SO nmol cytochrome c2 together with antibody (80 nmol with respect to the antibody combining site) in 1 ml of the same buffer (experiment B) were preincubated for 30 min at 25' C. [3H]Acetic anhydride (500 Ci/mol, Amersham) was then added, 1.2 mCi in experiment F and 1.5 mCi in experiment B. Both reaction mixtures contained in addition 33 nmol of the horse cytochrome c peptide 66 - 80 which served as an internal standard to monitor possible differences in reaction conditions between experiments F and B, respectively [12, 141. This internal standard peptide did not bind to the cytochrome-c2specific antibodies. Antibody, cytochrome c 2 , peptide 66 - 80 and low molecular mass reaction products were separated by gel filtration. [3H]-Acetylatedcytochrome c2 and peptide 66 80 from experiments F and B were mixed with equimolar amounts of uniformly [14C]acetylated cytochroine c2 [I71 and peptide 66 - 80 [14], respectively. The fully [14C]acetylated compounds served as a concentration standard [21]. Cytochrome c2 and peptide 66-80 mixtures from experiments F and B were then fully acetylated with nonradioactive acetic anhydride. The chemically homogeneous yet isotopically heterogeneous cytochrome c2 was digested with S. aureus V8 protease of a-chymotrypsin (Fluka), the peptides separated by HPLC, and 3H/14Cratios of labeled lysine residues determined after step-wise Edman degradation of peptides, as described in detail before [I 71.

427 A protection factor, RAP,,, was defined as the 3H/14Cratio obtained for a lysine labeled in free cytochrome c (experiment F) divided by the 3H/14Cratio of the same lsyine labeled in antibody-bound cytochrome c (experiment B). 3H/14Cratios of intact peptide 66 - 80 were analyzed in the same way to give R,. The protection factor R, corrected for small differences in reaction conditions ( R , was close to one). was calculated as

B W al 0 6 0 c m

RapplRp.

s

E

100

*

$

40

0.)

n

20

0

CD spectra were measured on a JASCO (Tokyo) spectropolarimeter (model 500 C), as before [22]. Difference CD spectra were recorded in a two-compartment cuvette. Exactly 1.00 ml 10 pM cytochrome c2 and 1.00 ml 20 pM antibody (with respect to antibody-combining site), both in 80 mM potassium phosphate, 70 mM NaCI, pH 7.8, were placed in the separate compartments. Spectra were run before and after mixing the contents of the two compartments, and difference spectra were obtained by SUbtrdcting the first from the second spectrum. RESULTS

N O

N

CD measurements

0

g

-

m

:

6

l

c

? m y my O

O

~

O

mm

m I o

r

m

w

.

O

:

N

r

r m

O o

0 m 0.

r

Fig. 1. Polyclonal immune response of five ruts ( A E ) ugainst glutaruld~~hyd~,-polymc.uized cytochrome c2 ( A - C ) and monomeric cytochrome c2 ( 0 , E ) . Antisera were diluted 1120 and tested in an ELISA against native cytochrome c2 (cyt c2), apo-cytochrome c 2 (apo cyt c2) and six peptides covering the sequenccs indicated. Antigens were coated to the testplatc and the amount of bound monoclonal antibody was measured with a second antibody that was coupled to alkaline phosphatase. 96-well plates were coated with 50 p1 20 pM antigen solution/well, as described in Experimental Procedures. The height of each bar is the average from six single determinations (standard error below f 20%). Photometer readings were normalized to the reading for native cytochrome c which always was highest ~

Polyclonal immune response against cytochrome c2 Two rats were immunized with monomeric cytochrome and three with glutaraldehyde-polymerized cytochrome c2 [18]. Polymerization of cytochrome c2 did not increase its immunogenicity in rats; the titer of all five antisera were of a similar magnitude between 1: 32 and 1 :256 against native cytochrome c,. In the case of eukaryotic cytochromes c, polymerization considerably increased the immunogenicity of cytochrome c in rabbits [18]. The antisera from the five rats were tested by ELISA for their immune reactivity with native cytochrome c 2 , apo-cytochrome c, and with the peptides 55-64,65-80,81-100and103-112. 1-8,9-37,38-50, As expected, the strongest reaction was observed with native cytochrome c2 (Fig. 1). The serum from rat D, immunized with monomeric cytochrome c2, showed considerable reactivity with apo-cytochrome c2. We assume that apocytochrome c2 is devoid of regular structure, in analogy with eukaryotic apo-cytochrome c [23, 241. In view of this it is noteworthy that rat D also showed the best immune response against single peptides. The remaining four antisera were weakly reactive with the peptides, the N-terminal nonapeptide being bound best. Spleen cells from rats A, B, and I) were used to produce monoclonal antibodies and several cytochrome c,-specific hybrids were cloned. Antibodies from three clones were selected for further characterization: antibody 30.6 from rat A ; antibody 8.11 from rat B; antibody 55.1 from rat D. All belonged to class IgG, subclass IgG,, or IgGzb(no definite assignment made). The affinity (&) of the antibodies for cytochrome c, was estimated by a Farr-type radioimmunoassay and the isoelectric points (PI) determined by isoelectric focusing. The following results were obtained: antibody 30.6, Kd approximately 80 nM, pl 6.7-6.8; antibody 8.11, Kd approximately 300 nM, PI 6.5-6.6; antibody 55.1, Kd approximately 20 nM, PI 6.2-6.5.

30

c,

Assignment qflysine residues to antigenic determinants Free and antibody-bound cytochrome c, were labeled with radioactive acetic anhydride and the amount of incorporated

K

I

1

8

b 8

6

C

.-0 5 c

4

0

L

n 2

n

m

-

~

r

m

~

l

n

c

~

~

~

m

h

~

m

n

m

~

m

w

m

m r

m

o r

m

u

sequence position of lysines

Fig. 2. Protection factors R f o r lysine residues of cytochromtz c2 hound to rnonoclonul antibodies 30.6, 8.11 and 55.1. Thc factor R indicatcs the degree by which a lysine residue is less labeled by acetic anhydride when cytochrome c2 is bound to the antibody. For residues labeled equally in frcc and antibody-bound cytochrome cz the upper limits of R at a significance level P < 0.001 were R = 1.43 for antibody 30.6, R = 1.96 for antibody 8.1 1, and R = 2.14 for antibody 55.1

label was determined for each lysine residue. A protection factor R was defined to describe the change in the degree of acetylation of lysine residues (see Experimental Procedures). R was around one if the antibody did not alter the degree of acetylation, above one if the residue was less acetylated, and below one if it was more acetylated in the antigen-antibody complex. Fig. 2 shows the results. When cytochrome c2 bound to antibody 30.6 three residues, Lys27, Lys88 and Lys90, were less labeled. The change in the degree of acetylation was high for Lys27, and much smaller, yet still significant, for Lys88 and Lys90. The location of the three protected lysines in the spatially folded polypeptide chain is shown in Fig. 3. Lys27 is to the right, Lys88 and Lys90 are to the left of the exposed heme edge. Antibody 8.11 rendered lysines 75, 81, 86 and 88 significantly less reactive in the antigen-antibody complex. while

o

l

r

c

m

428

!A

5

30.6

1.5-

v)

0

e a,

1.0-

0

c m

f

2

0.5-

r)

m

o.o! -9

.

8

-8

.

I

. ' -6

-7

.

' -5

.

I -4

8.11

Fig. 3. Model of' cytochrome c2. Van der Wads radii of the C2- and C3 atoms are shown for all the lysines which were less labeled by [3H]acetic anhydride when cytochrome cz was bound to monoclonal antibody 30.6 (lysines 27, 88 and 90), 8.11 (lysines 75, 81, 86, 88 and 90), or 55.1 (Lys9). The N-terminus (N) and the C-terminus (C), the latter hidden behind Lys27, are indicated

Lys90 had an R value which was just slightly above the upper limit of R of residues which were equally reactive in free and antibody-bound cytochrome c2 (Fig. 2). The less labeled lysines are part of a large loop to the lower left of the exposed heme edge (Fig. 3). If antibody 8.11 recognizes this loop, or part thereof, as suggested by the acetylation experiment, then the antigenic determinant of antibody 8.11may also be conformational, formation of the complex depending on the proper folding of the antigcn molecule. Only a single rcsidue, Lys9, became less reactive when cytochrome c2 bound to antibody 55.1 (Fig. 2). Thus, no information about the conformational or sequential nature of the determinant for this antibody could be obtained by the chemical modification experiment. Lys9 ends the N-terminal helix of cytochrome c2 [25].

Q 0

c

m

f 0

L

v)

a

m

0.5

-9

-8

-7

l

-6

-5

-4

log[protein]/#

Cross-reactivity analyses with diffirent untigens The conformational nature of the epitopes for antibodies 30.6 and 8.11 was substantiated by the experiments shown in the top and middle panels of Fig. 4. The figure shows the results from ELISA in which the antibodies were preincubated with different competing antigens before being added to cytochrome-c2-coated microtiter plates. Peptides 9 - 37 and 81 100 were not recognized by antibody 30.6. These peptides contain the three lysine residues which were less labeled in the antigen-antibody complex. Likewise, peptides 65 - 80 and 81 - 100 were not bound by antibody 8.11 which protected lysines 75, 81, 86 and 88. In addition, apo-cytochrome c2 was also not recognized by antibodies 30.6 (Fig. 4) and 8.11 (not shown). Antibody 30.6, but not 8.11, was slightly cross-reacting with native horse cytochrome c (Fig. 4). Neither antibody recognized a cytochrome c2 derivative which was fully acetylated at all lysine residues (Fig. 4 and data not shown). The situation was quite different for antibody 55.1. This antibody bound apo-cytochrome c2 and peptide 1-8 with about equal affinity, and more weakly peptide 9-37 (Fig. 4). Thus, the determinant for antibody 55.1 is in the N-terminal region of cytochrome c 2 , as already supposed from the acetylation experiment in which Lys9 was less labeled in antibodybound cytochrome c2. ~

Fig. 4. Competitive inhibition offhe binding ofcytochrome c2 to monoclonalantihodies30.6 jAj,h'.lI ( B ) and55.1 (C),meusuredby ELISA. The antibody and increasing amounts of competing antigen were left to eqnilibratc before being added to the microtiter plate which was covered with a fixed amount of cytochrome c2. The amount of antibody which bound to cytochrome c2 on the testplate was determined by a second antibody coupled to alkalinc phosphatase and using p-nitrophenylphosphate as a chromogen. The following antigens were tested with all three antibodies: cytochrome c2 ( x ) , apo-cytochrome cz ( A ) and horse cytochrome c ( 0 ) .Additional tests with antibody and 81-100 30.6: cytochrome c2 peptide 9-37 (g-n) (H-W). Additional tcsts with antibody 8.11 : fully acetylated cytochrome c2 (H) and bovine serum albumin (El). Additional tests with antibody 55.1 : cytochrome c2 peptide 1 - 8 (B) and 9 - 37 (0)

CD measurements with frec. and antibody-bound cytochrome c2 From a previous investigation we knew that binding of a protein near to the heme of mitochondria1 cytochrome c can perturb the induced heme CD spectrum at 400 - 420 nm [22]. Since the foregoing results provide good evidence that two of the three antibodies interact in the area of the exposed heme edge, we compared the heme CD of free and antibody-bound cytochrome c2 (Fig. 5). Indeed, antibody 30.6 induced an in-

429 10

.-

h

YE

5

f

v

0

I

I

-5 I

I

410

430

420 wavelength (nm)

Fig. 5. Diffrence CD spectra of antibody-boundminus,free cytochrome measured in the region o j the muin absorption band of the ,ferric heme group. The numbering of the spectra refers to the monoclonal antibodies. Spectra are the mean of three measurements CZ

crease of the heme CD with a maximum at 414nm and a difference extinction coefficient of 9.5 mM cm- This difference CD spectrum is conspicuously similar to that observed when horse cytochrome c binds to cytochrome-c oxidase [22], or to yeast cytochrome-c peroxidase (M. Oertle and H. R. Bosshard, unpublished). Quite a different perturbation of the heme CD spectrum was caused by antibody 8.11. With this antibody the heme CD signal decreased below 41 5 nm. Antibody 55.1 did not significantly perturb the heme C D spectrum. ~

'

'.

DISCUSSION

PoI.yclona1 immune response The immune response of five rats against prokaryotic cytochrome c2 is more varied than against eukaryotic cytochrome c. This is explained by the many sequence changes between the immunogen and the homologous cytochrome of the rat. Cytochrome c2 from R. ruhrum and cytochrome c from the rat share 43 identical sequence positions. The eukaryotic protein has 104 residues and the prokaryotic 112. Many sequence differences are clustered between residues 37 and 55, and residues 76 and 89 (cytochrome c2 numbering) [25]. There is an eight-residue insertion after position 77 of the eukaryotic sequence, and the long N-terminal helix of eukaryotic cytochrome c is shortened by a two-residue deletion in cytochrome c2 [26]. It is too early to draw a picture of the antigenic structure of cytochrome c2. Nevertheless it seems noteworthy that the five rat antisera were all reactive with the seven cytochromec,-derived peptides (Fig. 1). Unlike in the case of mammalian cytochromes c where, for example, the antigenicity of horse cytochrome c in rabbits is restricted to only three areas [27], the antigenicity of cytochrome c2 in rats seems to encompass the entire polypeptide chain. The N-terminal nonapeptide was more antigenic than all other peptides. The number of sequence changes is not particularly large in peptide 1 - 8. It is larger in peptides 38-50 (12 replacements) and 81-100 (8 replacements, 8 insertions, 1 deletion). The antigenicity of peptide 1 - 8 might originate from a high mobility of the Nterminus in native cytochrome c,. Segments of high mobility have been shown to coincide with high antigenicity [28, 291. Epitopes for three monoclonal untihodies

Eukaryotic cytochromes c often differ by only very few residues, viz. members of the vertebrate cytochrome c family.

This has made possible the precise assignment of residues to individual epitopes by comparing the cross-reactivity of closely related antigens with a monoclonal antibody or a rnonospecific serum (review in [3]). In the case of cytochrome c2 the task of defining epitopes is more difficult since cytochromes c2 are much less sequence related [30] and, hence, the sort of fine specificity analysis performed with vertebrate cytochromes c is not possible. In view of this, and keeping in mind that probably most monoclonal antibodies bind to conformational and assembled epitopes [16], we have used a combined protein chemical and spectroscopic approach for epitope mapping. The chemical procedure used differs from most other mapping techniques in that it aims at exploring the structural features of the antigen-antibody complex in a direct way. Not only does it reveal residues at antigenic sites, but it may also help to exclude residues from being part of such sites, those of equal chemical reactivity in free and antibody-bound antigen. The limit of the method is that a residue of lowered reactivity cannot always be assigned unequivocally to an antigenic site, since binding of the antibody might sometimes induce changes of reactivity outside of the antigenic site. For example, antibody 30.6 might protect only Lys27 to the right of the exposed heme edge, yet induce a conformational change to the left of the heme, thereby turning Lys88 and Lys90 less reactive. In addition, the degree of protection (the R value) is not directly related to the importance of a residue for antibody binding. These points have been discussed in detail elsewhere [12, 131. For all of these reasons it was important to substantiate the results by other experiments. The conformational nature of the epitopes for antibodies 30.6 and 8.11 was supported by competition ELISA since none of the peptides containing one or more of the lysines which were less labeled in the antigen-antibody complex was recognized by either antibody. Antibody 8.11 is directed against a structural element missing in the homologous protein of the immunized host, the eight-residue insertion which is part of a very large and complex loop from Am73 to Lys94. The distance at the 'neck 'of this large loop is only 0.54 nm between C2 atoms, similar to the distance of 0.46 nm at the neck of the 'R loop' 70-84 [31] of mitochondria1 cytochrome c. That antibody 8.11 did not react with horse cytochrome c was to be expected if the epitope is in a region where the two cytochromes strongly differ. Antibody 30.6, on the other hand, cross-reacts weakly with the homologous horse protein, which indicates binding to a more conserved area. Indeed, the front side of cytochrome c,, where the epitope for antibody 30.6 is expected, is the most conserved of all c-type cytochromes with regard to both the polypeptide folding and the amino acid sequence [I]. Based on the operational definition of a sequential determinant, namely the cross-reactivity of an anti-(cytochrome c) antibody with a short cytochrome c peptide, antibody 55.1 is directed against a sequential epitope close to the N-terminus. Native cytochrome c2 is, however, a better antigen for antibody 55.1 than peptide 1-8 or apo-cytochrome c2. Hence, to some extent antibody-binding depended on the conformation of cytochrome c2. Lys9, which was protected in the acetylation experiment, is not crucial for antibody binding since fully acetylated cytochrome c2 was as strongly bound as apocytochrome c2 (not shown). The spectroscopic properties of cytochrome c2 enabled a further mode of characterizing the antigen-antibody complex. The CD spectrum in the Soret absorption region of the ferric

430 heme is caused by the chiral polypeptide environment of the achiral chromophore. The CD spectrum is called ‘induced’ for this reason. Binding of cytochrome c to cytochrome-c oxidase elicits an increase of the heme CD spectrum [22]. We now observe a very similar CD change when cytochrome c2 binds to antibody 30.6. The difference CD spectra of the antigenantibody complex and the cytochrome-c cytochrome-c-oxidase complex both peak at 414nm, the amplitude of the former (Fig. 5 ) being about half that of the latter [22]. The origin of the spectral change induced by cytochrome-c oxidase has been traced to a structural perturbation very close to the exposed heme edge of cytochrome c [32,33]. We may therefore conclude that like the mitochondria1 oxidoreductase, antibody 30.6 induces a conformational rearrangement in the area of the heme chromophore. Conformational adaptation of the antigen to the paratope of the antibody has been documented in only a few other cases (e.g. [34- 361. One might argue that exciton coupling between the cytochrome c heme and a nearby aromatic residue of the antibody could induce a change of the heme CD spectrum, without need for a conformational change of cytochrome c 2 . Such a possibility cannot be strictly ruled out. However, the conspicuous similarity of the CD difference spectra observed with three unrelated proteins, antibody 30.6, cytochrome-c oxidase [22, 321, and cytochrome-c peroxidase (M. Oertle and H. R. Bosshard, unpublished results), support our interpretation that the antibody-induced CD change originates from a conformational adaptation in cytochrome c 2 . Antibody 8.1 1 also elicits a CD change, yet of a different shape. An observation which again complements nicely our conclusions from the acetylation experiment : antibodies 30.6 and 8.11 bind to two different antigenic sites which both, however, are sufficiently close to the heme to perturb its CD spectrum. Antibody 55.1 did not perturb the heme CD spectrum, in agreement with a binding site remote from the heme area (Fig. 3). In conclusion, it has been possible to locate three epitopes, two of which clearly are conformation-dependent. This type of epitope escapes detection by the well known mapping procedure with sequence related peptides. It has therefore been very difficult in the past to localize conformation-dependent and assembled epitopes, except in those rare cases where closely related antigens of identical polypeptide backbone conformation were available for cross-reactivity testing. ~

We thank Prof. R . Bachofen for providing R. ruhrum, Rahel Tinner for help with the Edman degradation, and Irmgard Willimann lor secretarial help. This work was supported by the Swiss National Science Foundation and the Sundoz Juhi/uumsst(ftung.

REFERENCES 1. Dickerson, R. E. & Timkovich, R. (1975) in The enzymes (Boycr, P. D., ed.) vol. 11, pp. 397 - 547. Academic Press, New York. 2. Urbanski, G. J. & Margoliash, E. (1977) J . Imnzunol. 118, 117011x0. 3. Benjamin, D. C., Berzofsky, J. A., East, I. J., Gurd, F. R . N., Hannum, C., Leach, S. J., Margoliash, E., Michael, J. G.,

Miller, A,, Prager, E. M.. Reichlin, M., Sercarz, E. E., SmithGill, S. J., Todd, P. E. & Wilson, A . C. (1984) Annu. Rrv. fmmunol. 2. 67 - 101. 4. Jcmmcrson, R. & Margoliash, E. (1979) Nature (Lond.) 282, 468-471.

5. Jcmmerson, R., Morrow, P. R., Klinman, N . R. & Paterson, Y. (1985) Proc. Nut1 Acud. Sci. US24 82, 1508 - 1512. 6. Hannum, C. H. & Margoliash, E. (1985) J. fmmunol. 135, 33033313. 7. Atassi, Z. (1981) Mol. lmmunol. 18, 1021-1025. 8. Paterson, Y. (1985) Biochemistry 24, 1048 - 1055. 9. Margoliash, E. & Bosshard, H. R. (1983) Trends Biochern. Sci. 8, 316- 320. 10. Carbone, F. R. & Paterson, Y. (1985) J. lmmunol. 135, 26092616. 11. Jemmerson, R. & Paterson, Y. (1986) Science (Wash. D C ) 232, 1001-1004. 12. Burnens, A., Demotz, S., Corradin, G., Binz, H. & Bosshard, H . R. (1987) Science (Wash. DC) 235, 780-783. 13. Ocrtle, M., Tmmergluck, K., Paterson, Y. & Bosshard, H. R . (1989) Eur. J . Biochem. 182, 699-704. 14. Saad, B., Corradin, G. & Bosshard, H. R. (1988) Eur. J . Biochem. 178, 219 -224. 15. Silvestri, I . & Taniuchi, H. (1988) J . Bid. Chem. 263, 1870218715. 16. Jcmmerson, R. (1987) Proc. Nut1 Acad. Sci. USA 84,9180 9184. 17. Bosshard, H. R., Wynn, R . M. & Knaff, D. B. (1987) Biochc~mistry 26, 7688 - 7693. 18. Rcichlin, M., Nisonoff, A. & Margoliash, E. (1970) J . Biol. (’‘hem. 245,947 - 954. 19. Fazekas de St. Groth, S. & Scheidegger, D. (1980) J . Immunol. Methods 35, 1 - 21. 20. Hunter, W. M. & Greenwood, F. C. (1962) Nutur-e (Lond.) 194, 495-496. 21. Rieder, R. & Bosshard, H. R. (1980) J. B i d . Chem. 255, 47324739. 22. Webcr, C., Michel, B. & Bosshard, H. R. (1987) Proc. Natl Acad. Sci. U S A 84,6687 - 6691. 23. Stellwagen, E., Rysavy, R. & Babul, G . (1972) J . Biol. Chem. 247, 8074 - 8077. 24. Fisher, W. R., Taniuchi, H. & Anfinsen, C. B. (1973) J . Bid. Chem. 248,3188-3195. 25. Salcmme, F. R., Kraut, J. & Kamen, M. D. (1973) J . Bid. Chenz. 248. 7701 -7716. 26. Dus, K., Stetten, K . & Kamen, M. D. (1968) .J. Bid. Chem. 243, 5507-5518. 27. Jcmmerson, R. & Margoliash, E. (1979) J . Biol. Chem. 2-54, 12706 - 12 716. 28. Westhof, E., Altschuh, D., Moras, D., Bloomer, A . C., Mondragon, A,, Klug, A. &van Regenmortel, M. €1. V. (1984) Nature (Lond.) 311, 123-126. 29. Geysen, H. M., Tainer, J. A , , Rodda, S. J., Mason, T. J., Alexander, H., Getzoff, E. D. & Lcrner, R. A. (1987) Science ( W U S ~DCj Z . 235, 1184-1190. 30. Meyer, T. E., Cusanovich, M. A. & Kamen, M. D. (1986) Proc. Nut1 Acud. Sci. USA X3, 217 - 220. 31. Leszcynski, J. F. & Rose, G. D. (1986) Science (Wash. D C ) 234, 849- 855. 32. Michel, B., Proudfoot, A. E. I., Wallace, C. J. A . & Bosshard, H. R. (1989) Biochemistry 28, 456 -462. 33. Michel, B., Mauk, A . G. & Bosshard, H. R. (1989) FEBS Lett. 243, 149 - 1 52. 34. Crumpton, M. J. (1966) Biochem. J . 100, 223-232. 35. Sheriff, S., Silverton, E. W., Padlan, E. A., Cohen, G. H., SmithGill, S. J., Finzel, B. C. & Davies, D. R. (1987) Proc. Narl Acad. Sci. USA 84, 8075 - 8079. 36. Getzoff, E. D., Geysen, H. M., Rodda, S. J . , Alexander, H., Tainer, J. A. & Lerncr, R. A . (1987) Science (Wash. D C ) 235. 1191-1196. -