Molecular epitope identification by limited proteolysis of an immobilized antigen-antibody complex and mass spectrometric peptide mapping

Proc.Natl. Acad. Sci. USA

Vol. 87,pp.9848-9852, December 1990 Biochemistry

Molecular epitope identification by limited proteolysis of an

immobilized antigen-antibody complex and mass

spectrometric peptide mapping

(epitopemapping/molecularweightdetermination/plasmadesorption massspectrometry/complementcomponentC3a/ monoclonal antibody)

DETLEV SUCKAU*, JORG KOHLt, GABRIELE KARWATHt,

KLAUS

SCHNEIDER*,

MONIKA

CASARETTOt,

DIETER

BITTER-SUERMANNt,

AND

MICHAEL

PRZYBYLSKI*§

*Fakultat furChemie,UniversitatKonstanz, D-7750 Konstanz, Federal Republic of Germany;tInstitut furMedizinischeMikrobiologie,Medizinische Hochschule Hannover, D-3000 Hannover, Federal Republic of Germany; andtDeutschesWollforschungsinstitutAachen,D-5100Aachen,

FederalRepublicofGermany

Communicatedby FredW. McLafferty, September 10, 1990 (received for review July 9, 1990) ABSTRACT Sequences of antigenic determinants were

identified by limited proteolysis of peptide antigens bound to an immobilized monoclonal antibody and direct molecular weight determination of the monoclonalantibody-bound peptide frag-ments by

252Cf

plasma desorption mass spectrometry. The epitope peptides to the monoclonal antibody h453

[Burger,

R., Zilow, G., Bader, A., Friedlein, A. & Naser, W. (1988) J. Immunol. 141, 553-558] wereisolated from immobilized

anti-gen-antibody complexes by partial trypsin digestion. A syn-theticeicosapeptide comprised of the C-terminal sequence of the humancomplement componentpolypeptide

des-Arg77-C3a

as well as guinea pigdes-Arg78-C3a was used as an antigen. Conditions were developed under which trypsin specifically degraded theantigenswithoutinactivation of the immobilized antibody. After proteolysis, epitope peptides were dissociated from theantibody with 4 MMgCl2. The antigenic peptides were purified by HPLC and identified by

252Cf

plasmadesorption massspectro! retry. The epitope recognized by h453 resides on the C-terminaltryptic peptides of human(residues70-76)and guineapig (residues70-77)C3a.Asanestimationof accuracy this method is able to provide, trypsin digestion of immune complexes caused cleavage of theantigenwithin adistance of two amino acid residues upstream from theepitope.

A variety of methods have been applied to the study of monoclonal antibody

(mAb)-antigen

interactions and the characterization of their

respective

epitopes.

Two

major

approaches that have been

widely employed

for

epitope

characterization are

competitive

binding analysis using

syn-thetic peptides and

fine

specificity

studies with

panels

of

evolutionary

variant orrecombinant

proteins

(1).

Although

wellestablished, these methodshave

major limitations;

e.g.,

discontinuousor

conformationally

defined

epitopes

maynot

be detectable by

using peptide

probes

(2). Site-directed mutagenesisexperimentsin epitopestudies could begreatly facilitated if some information abouttheputative epitope is

availablein advance. Adirectapproach ofepitope mapping, which seems promising in this respect, has been more

re-cently introduced basedonthe

finding

that(i) mAbsexhibit remarkable resistance towards

proteolytic

enzymes,

(ii)

in

immunecomplexes,

antigenic

determinantscanbeprotected

fromproteolytic degradation, and

(iii)

proteolysis does not lead to dissociation of immune complexes (3-6). Limited

proteolytic cleavage ofimmunecomplexeshas been usedfor epitopecharacterizationbymeansofPAGE(7)and

by

HPLC

(6)of the respective

peptide digests.

However, HPLC

sep-aration of complex digest mixtures followed by amino acid analysisorpeptide sequencingmay not enableunambiguous epitope identification duetounresolvedpeptides.

Thefeasibility of fastatom bombardment mass spectrom-etry(FABMS) and 252Cf plasma desorptionmassspectrometry

(PDMS) foraccuratemolecularweight determinationof

poly-peptides has been established in several bioanalytical appli-cations (8).Particularly, abundantmolecularions of polypep-tides up to small proteins have been obtained with high sensitivity by PDMS(9). Apromisingapproach inrecent work

hasbeentheapplicationtomulticomponentpeptidemixtures,

suchasproteolytic digests (peptide mapping; ref.8). Thehigh molecularspecificity provided bymassspectrometricpeptide mapping has been successfully used in protein structural studies, such as the characterization of cDNA-derived se-quences,identification ofposttranslationalmodifications, and differentiation of isoenzymestructures (10-12).

Inthisstudy the combination of limited enzymatic

prote-olysis

and PDMS has beenapplied to themolecularepitope analysis of complementcomponentC3a,which isrecognized

as a potent mediator of inflammation (13). Amino acid

sequences(14, 15) andtertiarystructure(16, 17)of C3a from severalspecies including human (77-amino acid residues) and guinea pig (78

residues)

have been reported. The three-residue C-terminal sequenceof C3arepresentstheessential receptor-binding site (18, 19). However, biological

activity

is

lostby C-terminal desargination (13), concomitant with the exposureofaneoantigenic determinant that is

recognized

by

mAb h453 (20). h453 coupled to

tresyl-activated

Sepharose

(h453-TAS) was used for the preparation of immune

com-plexes with (i)aneicosapeptide comprising human(h) C-ter-minalC3a, and (ii) guinea

pig

(gp)

des-Arg78-C3a

(des-Arg78-gpC3a). For both

antigens,

mass

spectrometric

peptide

map-ping of tryptic peptides dissociated from the truncated

immunecomplex byMgCl2 directly establishedtheepitope,

whereas the nonepitopepeptides were identified after

sepa-ration fromtheimmunecomplex,as

schematically

illustrated

inFig.1.Moreover, thesynthetic peptide (hC21)

comprising

the sequence[Gln

65,Arg66]hC3a-(57-77)

was

designed

to

pro-videanestimation ofthestericrequirements forthe ternary

trypsin-antigen-antibody complex, by

introducing

three

equidistant tryptic cleavage sites.

Abbreviations: E;S, enzyme-to-substrate ratio (wt/wt); gpC3a, guinea pigC3a; hC3a, human C3a;h453-TAS,mAbh453coupledto

tresyl-activated Sepharose; hC21, [Gln65,Arg66]hC3a-(57-77); hC21dR,des-Arg21-hC21; mAb,monoclonalantibody;PDMS,252Cf plasmadesorptionmassspectrometry;FABMS,fastatom bombard-mentmass spectrometry.

§Towhomreprintrequestsshould be addressed. 9848

Thepublicationcostsofthis articleweredefrayedinpartbypagecharge payment.Thisarticle musttherefore beherebymarked"advertisement" in accordance with18U.S.C. §1734solelytoindicatethisfact.

Proc. Nati. Acad. Sci. USA 87 (1990) 9849 Immune-Complex

I

0.2 ml of TS. Pooled supernatant and remaining immune complexes after dissociation were subjected to HPLC

puri-fication on a0.4 x 25 cm Nucleosil 300-7-C18 column

(Ma-cherey & Nagel). Elution was carried out with a linear

gradient of 0-52% acetonitrile in water containing 0.04% trifluoroacetic acid over 25 minat aflow rate of 1 ml/min.

Peptide fractions were detected with an M490 multiwave-length detector (Waters),lyophilized, andredissolvedin 5

,ul

of 0.1% trifluoroacetic acid for mass spectrometric analysis. Isolation of Epitope Peptides fromdes-Arg78-gpC3a. Sixty microgramsofdes-Arg78-gpC3awasallowedto bind to 500,g of immobilized h453 for1 hr at20'C.Thegel was rinsed with

TS, and samples equivalentto6.4

gg

of bound antigenwere incubatedfor 30minat370C with 50

gl

of TScontaining 0,5,

or 45

/ig

of trypsin, After removal of supernatant and treatmentthreetimes with 1 ml ofTS, dissociation and HPLC analysis were performed as describedabove.

Mass Spectrometry. Nitrocellulose surfaces for sample adsorptioninPDMS werepreparedasdescribed (24). Peptide

solutionswereallowedtoadsorb for 2-3minfollowedbyspin

drying (25). Spectra wereobtained on atime-of-flight spec-trometer(Bio-Ion 20 K,Uppsala, Sweden) at a 15-kV accel-erating voltage. FABMS was performed on a Finnigan (San

Jose, CA) MAT-312/AMD-5000 double-focusing spectrom-eter, with a 20-kV cesiumprimaryion source(AMD,

Beck-eln, F.R.G.); glycerol was used as a matrix for the sample.

RESULTS AND

DISCUSSION

FIG. 1. Scheme of the mass spectrometric epitope mapping method ofan antigen-antibody complex as compared to peptide mapping of the free antigen. The use ofan immobilized antibody

(mAb) allows the separation ofnonepitope and epitope peptides after limited proteolysis of the immune complex. Molecular ions of nonepitope and epitope proteolytic peptidesareillustrated by solid

and openbars, respectively.

MATERIALS AND METHODS

Preparation of Antigens and Immobilized Antibody. The peptide hC21, which has the sequence

[Gln65,Arg661hC3a-(57-77)was prepared by solid-phase synthesis (21) and

con-tained a Cys'-S-acetamidomethyl protecting group. gpC3a wasisolated and purifiedasdescribed (22). The desarginated

forms,des-Arg21-hC21 (hC2ldR) and des-Arg78-gpC3a,were

prepared by carboxy-peptidase B treatment (23) and were

purified by HPLC, which yielded -95% purity for

des-Arg-C3a. mAb h453 (20) was affinity purified using protein A-Sepharose (Pharmacia) and was coupled to tresyl-activatedSepharose (Pharmacia) accordingtothe supplier's procedures. The immobilized mAb (h453-TAS)wasstored in

phosphate-buffered salinecontaining 0.03% sodium azideat

a concentration of 800 ,ug of bound mAb per ml of gel

suspension.

Preparation and Dissociation of ImmuneComplexes. h453-TAS(1 ml)wasequilibrated with TS buffer (50 mM TrisHCI,

pH 7.5/150 mM NaCI), and 50 ,g of hC21dR in 0.5 ml of TS

was allowed to bind for 2 hr at 20°C. Free antigen was

removed by three washes each with 1 ml of TS, and the gel volumewas adjustedto 1 ml. Aliquots wereused either for

proteolysisorfordirect dissociationby addition of 0.4 ml of

4M MgCl2. Dissociation wasallowedtoproceed for 30min at37°C. The gel wasthen washedtwice with 0.15 ml of 4 M

MgCI2, and thepooled supernatantwassubjected toHPLC analysis.

Proteolytic Digestion and Purification of hC21dR Frag-ments. The immunecomplex of hC21dR and h453-TAS was

digested at37°C with trypsin(Sigma; seeTable 1). The gel waschilled onice, separated from supernatant by centrifu-gation (2000 x g, 4°C, 3min),and washed three times with

ProteolyticCharacterization of Free Antigen and the Anti-body. The structure and purity of hC21dRwas verified by

FABMS of the intact peptide and by direct analysis of the mixtureoftrypticpeptides, which yielded abundant proton-ated molecular ions [M + H]+ for peptides T1, T3, and T4 (Fig. 2). Relative cleavagerateswithtrypsinweredetermined

atconditions oflimitedproteolysis and showed rapid hydrol-ysis of peptide bonds at Arg'0 and

Arg13

and a somewhat

slower cleavage at Arg8 (data not shown). However, this reaction scheme issignificantly differentuponbinding of the

antibodyasdescribed below.

Incontrast,theantibody revealedaremarkable resistance

towards proteolytic degradation. No cleavage of mAb h453

I

~~~~~~~~~~~~~~..

CNYITELR

QR HAR::

... ...

ASUG

...

I -T1 1 ST3j T4 1 Cu la ._ en O T4 669 -T---T3/4 Ti 1083 T3 383 i)1 825 I ..IL 1-I . I. 1 I11 400 600 800 1000 m/z

FIG. 2. Amino acid sequence and FABMS analysis oftryptic peptidesof thesynthetic antigen hC21dR.Labeledpeaksdenote[M

+H]'ions ofpeptide fragmentsT1, T3,and T4. Thesequenceshown represents[GIn65,Arg66]hC3a-(57-76).Thestippled partialsequence

representsthesynthetic octapeptideused forproductionof the mAb

h453.AAM,S-acetamidomethyl. FreeAntigen

NO

Mass SpectrometricPeptideMapping Antigen Non-Epitope EpitopePeptide

4

%

h

Peptides

m/z m/ m/z

Biochemistry:

Suckauetal.

Proc. Natl.Acad. Sci. USA 87 (1990) Table 1. Quantification of tryptic peptides isolated from h453-TAS-bound hC21dR at different

proteolysis conditions

mAb-bound Reaction %tryptic peptide Ratio

Trypsin,

hC21dR,

:S (wt/wt) time,

rycpepe

of

lug ,ug hC21dR* h453-TAS min Tlt T4+T3/T4§ T4 to

T3/T49

0.05 1.67 1:33 1:4400 20 80 ND -0.05 1.67 1:3.3 1:440 20 90 68 1:5 5 1.67 3:1 1:44 20 90 68 2:1 20 3.25 6:1 1:18 10 97 65 13:1 50 3.25 15:1 1:7 120 -11 -11 ND, notdetermined.

*E:S

formAb-bound antigen.

tExpressed

asthe molar percentage of mAb-bound-hC21dR as determined by HPLC.

tPeptide

T1 from supernatant.

§Sum

ofpeptidesT4andT3/T4fromMgCI2dissociation.

1Estimatedbyratio of [M + H]+ionabundances in PDMS. "Notdetermined due to extensivedigestion of mAb. was detectable by

SDS/PAGE

at conditions that provided complete proteolysis of the antigen [enzyme-to-substrate ratio (E:S) = 1:100 (wt/wt)]. The native mAb was not

degraded,even athighproteaseconcentrations (E:S = 1:2),

whereas heatdenaturation

led

torapid proteolytic digestion.

A comparable stability of the mAb was found towards a-chymotrypsin and Staphylococcus aureus V8 protease, as

previously reported for different mAbs (6,

7),

suggesting a

similar

utility

of theseenzymesfor

peptide

mappinganalysis ofimmune complexes.

Molecular

weight

determination of intact

gpC3a

and

des-Arg78-gpC3a

by PDMS and mass spectrometric analysis of trypticpeptide mixtures

provided

structural

information

con-cerning

the entire

gpC3a

sequence (26).

Particularly,

[M +

H]+

ions of the N-terminal

tryptic peptides

and peptides

containing

residues 66-70 or residues 71-77 from the C terminus of

des-Arg78-gpC3a

were identified in

high

abun-dances,

indicating

complete

cleavage

at

Arg65

and Arg70, which is incontrast totheepitope

mapping discussed

below (seeFig. 5). At

high

E:S

(1:20),

additional molecular ions of large polypeptide fragments could be

assigned

to partial

digestion

in the central part of the C3a sequence

(data

not

shown).

Formation and Dissociation of Immune Complexes. The

binding capacity

and

specificity

of

h453-TAS,

the dissocia-tion

procedure,

andrecoveryof intact

antigen

wereevaluated with the aid of HPLC andmassspectralanalysis.In atypical

experiment,

from 4 ,gofhC21dR bound toh453-TAS

con-taining 75,gofmAb, 1.4 ,ugofantigenwasrecovered with TS

buffer,

andnofree

hC21dR

wasdetectable

by

HPLC in thesupernatantafter additionalTStreatment.Dissociation of theimmune

complex

with

MgCI2

ledtothe release of -1

jig

of

antigen

(i.e.,

40% of the theoretical

binding

capacity

of the

mAb).

After HPLC

purification,

the

antigen

was

identified

as

intact hC21dR

by

the [M +

H]+

ion

(m/z

=

2382)

and the doubly charged ion intheplasma

desorption

massspectrum

(see

Fig.

4c). In acontrol

experiment

with

Sepharose

con-taining

tresyl groups blocked

by

treatment with 0.1 M

Tris-HCl/0.5

MNaCI,2.4

,ug

ofhC21dRwasrecovered in the

TS

supernatant, andnopeptidewasdetectable in the

MgCl2

eluate. In

addition,

the

specificity

of h453-TAS was tested with a mixture of partial tryptic peptides of hC21dR and exclusively yielded peptides in the

MgCl2

eluate that con-tainedtheantigenic C terminus

(data

not

shown).

The

binding

and dissociation procedure could be

performed repeatedly

with the same batch of h453-TAS without loss of

binding

capacity,

indicating

the

feasibility

of

the

lyotropic

agent

MgCI2

(27) for

efficient

dissociation of the immobilized immunecomplex without

affecting

themAb's function.

IdentificationofTrypticEpitope Peptidesfrom mAb-Bound

hC21dR.

Samples

of the

purified

immune

complex

of

hC21dR

and h453-TAS were subjected to trypsin digestion using a wide range ofenzyme-to-hC21dR ratios and reaction times (Table 1). Under variousdigestion conditions, HPLC anal-yses ofsupernatants andMgC12eluatesconsistentlyyielded

major peptidefractions with retention times of 26.2 and 25.1

min,

respectively (Fig. 3). In addition, nonpeptide contami-nantcomponents were presentin thechromatogramsat21, 23.3, and 24.1 min. PDMS analysis of thepurified peptides

contained in the supernatant (i.e., nonepitope fraction)

yielded a predominant [M + H]+ ion (m/z = 1083) of fragment T1 due tocleavage atArg8(Fig. 4b). By contrast,

thespectrum of thefraction afterMgC12dissociation (Fig.4a) showed a most abundant [M + H]+ ion of the C-terminal

peptide T4 (m/z, = 669), together with a small ion ofthe

coeluting peptide T3/T4 (m/z, = 1033), which was unre-solved by HPLC. An=90% release of T1from mAb-bound

hC21dR in thesupernatantand65% release ofT4andT3/T4 from the truncatedimmune complexwerefound with E:Sas

-I I 8 q .0 0 DA .0 CU 6) 30 4.) 0 20 10 20 22 24 mm

FIG. 3. HPLC analysisof supernatant (tracesd-f) and

MgCI2-dissociatedpeptidefragments(tracesa-c)aftertrypsindigestionof h453-TAS-boundhC21dR, atdifferent protease concentrations and reaction times(see Table1). HPLC conditionswere asdescribedin Materials and Methods. Trace a, 20 ,ug oftrypsinfor10min;traces

b andd,5,ugoftrypsinfor 20min;tracescand e, 0.5jgoftrypsin

for 20min;tracef,0.05,ug oftrypsinfor 20 min. Unlabeledpeaksin thechromatogramsareduetononpeptidecontamination.

8 I a T4 669

.1]

a T3/4 1033 r-i b I I1 1083 1 (M+H) c 2382

M2+

I

,I.,

.

J

600 1600 2600 m/z

FIG. 4. PDMS analysis ofHPLC-purified hC21dR andtryptic peptides from the immune complex with h453-TAS. (a) Epitope fraction at 25.1 min in trace aofFig. 3. (b) Nonepitope peptide

fraction at 26.2min in trace dofFig. 3c. (c) IntacthC21dR after MgCI2 dissociation from the immune complex. M2' denotes the doubly protonatedmolecular ion(m/z=1191);the ionatm/z=794

is duetoanonpeptide contaminant.

high as6:1 (see Table 1). At veryhigh protease concentra-tions, specific peptide fragments were no longer detectable

because ofprogressive destruction of the mAb. Low E:S led to theformation ofincreasing amountsofT3/T4 relative to T4asestimatedby their [M+H]+ ion abundances, indicating

areduced tryptic cleavage rateatArg'3 relative to Arg8 due to steric hindranceby the mAb boundtoits epitope.

How-ever, atanyproteolytic conditions amenabletoPDMS

pep-tide mapping analysis and irrespective of nonpeptide

con-taminations, the epitope peptides T4 and T3/T4weretheonly

peptide fragments in the MgCI2 eluate. In contrast, some

contamination of thenonepitope fraction by epitope peptides

was observed, due to release of fragments from hC2ldR nonspecifically bound to the polypropylene cup surface.

Only the epitope fractionwas,therefore, used for the epitope identification ofgpC3a.

Epitope Identification from gpC3a. Proteolytic procedures and PDMS analysis were applied in the same manner to epitope mapping of des-Arg-gpC3a bound toh453-TAS. On the basis of the results obtained with theimmunecomplex of hC21dR with h453-TAS, digestion was performed at two differenttrypsin concentrations that didnotdegrade the mAb while fragmenting C3a (enzyme-to-C3a ratios, 6.5:1 and 1:1.5). In all experiments, HPLC analysis yielded a single majorpeptide fractionat23.8min uponMgCI2 dissociation.

PDMSanalysis (Fig. 5) showedapredominant [M +H]+ion (m/z = 901) of the C-terminal fragment containing residues

8

a

4) 0_

1000 2000 m/z

FIG. 5. PDMS analysis oftryptic epitope peptidesdissociated fromthe truncated immune complexofdes-Arg78-gpC3a with the

mAb h453 (ES = 6.5:1). Numbers in parentheses denote partial

sequencesofgpC3a;the ionatm/z= 1057(70-78) originatesfrom

traces ofgpC3a present. The C-terminal sequence of des-Arg78-gpC3ais shown. The C-terminal tryptic cleavage sites in the free

antigen (Arg70,openarrowhead)and in the immunecomplex (solid arrowhead)areindicated.

70-77, associated with minor ions of the peptides containing residues 66-77 and residues 70-78 ofgpC3a, due to incom-plete cleavage at Arg69 and a small amount of

non-desarginatedgpC3a,respectively. These peptides clearlycan

be named epitope peptides since they all contain the se-quencemotifLeu-Gly-Leu-Ala, which ispartoftheepitope

sequence(see Fig.2and thesequenceinFig. 5).In contrast to the cleavage at Arg70 found by tryptic digestion of free gpC3a (26), cleavage ofthe mAb-bound antigen occurred exclusivelyatArg69. Asnotedabove, this effect isin

accord-ancewiththe stericrequirements of trypsin binding (28)and withinaccessibility of Arg70 duetobinding of the mAbtoits epitope approximately three amino acid residues

down-stream.

Moreover, ELISA data yielded a30-fold loweraffinityof

h453 for des-Arg78-gpC3a as compared to des-Arg77-hC3a (data not shown), which strongly indicatessequencevariation within the epitope region. In addition to thesequence motif

Leu-Gly-Leu-Ala, which is conserved in both species andis responsible for the immunological cross-reactivity of h453,the epitopeondes-Arg77-hC3amusttherefore containone orboth of theevolutionary variant residues, His72 and Ser71. On the other hand, the weak exopeptidase activity of trypsin and crystal structure data of trypsin inhibitor complexes (28) indicatebinding approximatelytworesidues downstreamasa

prerequisite to tryptic cleavage at Arg69 in the gpC3a se-quence. Supported by these additional data, the mass

spec-trometric results provide evidence forHis-Leu-Gly-Leu-Ala [i.e., hC3a-(72-76)] being the epitope recognized by h453,with Ser71as apossiblepartof theepitopesequence.

CONCLUSIONS

The approach described in this study presents several ad-vantages to the molecular characterization ofepitopes by combining the selectivity of partial proteolysis of immune complexes with the molecular specificity of accurate mass

65 70 75 1

-L-R-Q-Q-H-R-R-E-Q-N-L-G-L-A|

(70-in 901

(70

-

78)

(66-

77)

1057 1450 m lb

r

Proc. Natl. Acad. Sci. USA 87 (1990) spectrometric molecular weight determination. Irreversible

immobilization of the mAb and analysis of the bound peptides after proteolytic digestion provide the basis to name the selectively dissociated fragments epitope peptides.

Important prerequisites for accurately defining epitope structures are(i) high affinity of the proteolytic peptides (i.e., truncated antigen) to the mAb and (ii) the availability and

accessibility of suitable proteolytic cleavage sites. Despite somesuppression by the mAb, tryptic cleavage was shown to occurapproximately two amino acid residues away from the epitope. At least two other endoproteases, a-chymotrypsin and S. aureus V8 protease, appeartobe well suitedand could besupplemented by exopeptidases to further define epitope sequences. Epitope extraction and direct identification of peptide fragment mixtures by PDMS was shown to be applicable to the analysis of a sequential epitope and may

enable the characterization ofevenconformationally defined

and assembled topographic epitopes, which are not detect-ableby synthetic peptides as antigenic probes. In the course of an exactdescription of an epitope (e.g., using site-directed mutagenesis), its application will greatly reduce the number ofpossible antigenic residues to be tested. Thus, epitope

extraction and mass spectrometric peptide mapping of a single immune complex represents a sensitive and rapid

methodofhighmolecularspecificityin theanalysis of protein antigens.

Wethank R. Burgerfor the generous gift of the mAb h453 and R. Gerardy-Schahn for stimulating discussions. This work was sup-ported by grants from the DeutscheForschungsgemeinschaft,Bonn, F.R.G. (Pr 175/2), by the Bundesministerium fur Forschung und Technologie, Bonn (01VM89014), and by the University of Kon-stanz.

1. 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., Smith-Gill, S. J., Todd, P. E. & Wilson, A.C. (1984) Annu. Rev. Immunol. 2, 67-101.

2. Berzofsky,J. A. (1985) Science 229, 932-940.

3. Schwyzer, M., Weil,R.,Frank,G. &Zuber,H.(1980) J. Biol. Chem. 255, 5627-5634.

4. Moelling, K., Scott, A., Dittmann, K. E. J. & Owada, M. (1980) J. Virol. 33, 680-688.

5. Eisenberg, R. J., Long, D.,Pereira,L., Hampar, B.,Zweig,M. &Cohen, G. H.(1982)J. Virol.41, 478-488.

6. Jemmerson, D. &Paterson, Y.(1986)Science232, 1001-1004.

7. Sheshberadaran, H. & Payne, L. G. (1988) Proc. Natl. Acad. Sci. USA85, 1-5.

8. Biemann, K. & Martin, S. A. (1987) Mass Spectrom. Rev. 6, 1-75.

9. Cotter, R. J. (1988) Anal. Chem. 60, 781A-793A.

10. Suckau, D., Manz, I., Schneider, K., Przybylski, M., Thomas, H., Milbert, U., Klein, J., Post, K. & Oesch, F. (1989) Adv. Mass Spectrom. 11, 492-493.

11. Gauss, C., Klein, J., Post, K., Suckau, D., Schneider, K., Thomas, H., Oesch, F. & Przybylski, M. (1990) Environ. Health Perspect. 88, 57-62.

12. Svoboda, M., Przybylski, M., Schreurs, J., Miyajima, A., Hogeland, K.&Deinzer, M. (1990) J. Chromatogr.,in press. 13. Hugh, T. E. (1986) Complement 3, 111-127.

14. Hugh,T. E. (1975)J. Biol. Chem.250, 8293-8301.

15. Gerard, N. P., Lively, M. 0. & Gerard, C. (1988) Protein Sequences Data Anal. 1, 473-478.

16. Huber, R., Scholze, H., Paques, E. P. & Deisenhofer, J. (1980) Hoppe-Seyler Z. Physiol. Chem. 361, 1389-1399.

17. Chazin, W. J., Hugh,T. E.& Wright, P. E. (1988) Biochem-istry27, 9139-9148.

18. Gerardy-Schahn, R., Ambrosius, D., Saunders, D., Casaretto, M.,Mittler, C., Karwarth, G., Gorgen, S. & Bitter-Suermann, D.(1989) Eur. J. Immunol. 19, 1095-1102.

19. Kohl, J., Casaretto, M., Gier, M., Karwath, G., Gietz, C., Bautsch, W.,Saunders, D. & Bitter-Suermann, D. (1990) Eur. J. Immunol. 20,1463-1468.

20. Burger, R.,Zilow, G., Bader, A., Friedlein,A. & Naser, W. (1988) J.Immunol. 141, 553-558.

21. Ambrosius, D., Casaretto,M.,Gerardy-Schahn, R., Saunders, D., Brandenburg, D. & Zahn, H. (1989)Biol. Chem. Hoppe-Seyler 370, 217-227.

22. Hoffmann, T., Bottger, E. C., Baum, H. P., Messner, M., Hadding, U. & Bitter-Suermann, D. (1988) Clin. Exp. Immunol. 71, 487-492.

23. Gerardy-Schahn, R., Ambrosius, D., Casaretto, M., Grotz-inger, J., Saunders, D., Wollmer, A., Brandenburg, D. & Bitter-Suermann, D. (1988)Biochem. J. 255, 209-216. 24. Jonsson, G. P., Hedin, A. B., Hakansson, P. L., Sundqvist,

B. U. R., Save, B.G. S., Nielsen, P.F., Roepstorff, P., Jo-hansson, K. E.,Kamensky, 1. & Lindberg, M. S. L.(1986) J. Anal. Chem. 58, 1084-1087.

25. Nielsen, P. F., Klarskov, K., Hojrup, P. & Roepstorff, P. (1988) Biomed.Environ. MassSpectrom. 17,355-362. 26. Schneider, K., Manz, 1., Messner, M., Gerardy-Schahn, R.,

Bitter-Suermann,D.&Przybylski,M.(1988) Adv. Mass Spec-trom. 11, 1338-1339.

27. Kessler, S. W. (1975) J. Immunol. 115, 1617-1624.

28. Walter, J. &Bode,W.(1983)Hoppe-SeylerZ.Physiol. Chem. 364,949-959.