Proc. Nati. Acad. Sci. USAVol. 75, No. 5, pp. 2434-2438, May 1978Immunology

Mechanism of antigen-induced antibody biosynthesis from antibodyprecursors, the heavy and light immunoglobulin chains

(antigen persistence/thiol-disulfide transhydrogenation/phenotypic reaction)

FELIX HAUROWITZDepartment of Chemistry, Indiana University, Bloomington, Indiana 47401

Contributed by Felix Haurowltz, March 6, 1978

ABSTRACT The immediate precursors of antibody mole-cules, the heavy (H) and light (L) peptide chains of the immu-noglobulins, combine with each other by means of disulfidebonds formed by dehydrogenation of their cysteine residues.In the absence of an antigen this process yields the heteroge-neous mixture of normal immunoglobulins. Antigens or theirprocessed derivatives (Ag) interfere with this stochastic processby noncovalent combination with complementarily fitting Hchains. The (Ag.H). complexes thus formed, owing to the lossof rotational and translational freedom, combine preferentiallywith those L chains whose VL regions have some affinity for thedeterminants of the antigen molecule. Subsequent release ofAg from the (AgH'L). complexes yields free antigen and anti-body molecules. Each of the released Ag molecules can be usedrepeatedly for the same reaction cycle and thus can induce thebiosynthesis of a large number of antibody molecules. Anymacromolecule, natural or synthetic, that has at least a few polargroups and that can penetrate to the nascentH and L chains canthus act as an antigen. Whereas the structure of the H and Lchains is genetically determined and transmitted through thegerm. line, the process induced by the antigen is a phenotypicphenomenon. The antigen acts in this process as a stereospecificcofactor or regulator of the thiol-disuilfide transhydrogenationof the combining H and L chains of immunoglobulins.

Widely held views on the mechanism of antibody formationare based on some or all of the following claims: (i) The abilityof producing antibody of a definite specificity is transmittedgenetically through many generations of a clone of lymphoidcells; antibody production, once initiated, continues even in theabsence of antigen. (ii) The immunocompetent lymphoid cellsare unipotent; each of them produces only one type of antibodymolecule. (iii) The antigen combines selectively with homol-ogous receptors on the surface of the immunocompetent lym-phoid cells; the receptors are or contain cell-bound antibodymolecules; the combination of these receptors with the antigenstimulates the cells to divide and multiply at an increased rate,thus accelerating antibody production; the mechanism of thisstimulation is not known. Experiments performed in this andin many other laboratories during the last three decades havemade it necessary to modify these claims. In the present articlean attempt is made to formulate a concept in which the im-munogenic action of the antigen and the process of antibodybiosynthesis are described in chemical terms.

Dependence of antibody production on continuedpresence of antigenImmunity against some of the antigens frequently lasts severalyears. Long persistence of immunity against pathogenic virusesusually has been attributed to continuous reproduction of thevirus in the host in a mitigated form. Yet it was for a long time

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked"advertisement" in accordance with 18 U. S. C. §1734 solely to indicatethis fact.

not clear whether proteins or other unanimated antigens mightpersist over long periods of time in the host organism. For thisreason Burnet (1), in the original form of his clonal selectiontheory, postulated that each cell of an antibody-producing clonetransmits to its daughter cells the ability of antibody productionand that this ability is maintained through many generationsof cells. Using isotopically labeled protein antigens we were ableto show that these antigens or their fragments persist in thespleen of rabbits for long periods of time (2). The slow decreasein the antigen content of the spleen was paralleled by an anal-ogous decrease in the antibody titer of the blood serum (s). Thebulk of the radioactive antigen fragments was located in thecytoplasm of the cells; nuclei were free of the radioactive ma-terial (4).Dependence of antibody production on the persistence of

the antigen has been convincingly established in cultures ofantibody-forming cells and in syngeneic colonies of such cellsin lethally irradiated animals (5-10). Antibody production invvo and also in cell cultures strictly depends on the continuouspresence of the antigen; short exposure of the cells to the antigenis not sufficient to induce antibody production (9). If antibodyformation in vivo or in vitro stops after elimination of most ofthe antigen, it is rapidly restored after renewed addition of theantigen. North and Askonas (10) conclude that it is "memory"in the form of precursor cells rather than autonomous antibodyproduction that is transferred through many generations. Whilethe term "memory cell" is usually applied to those cells that hadbeen exposed to prior stimulation by the antigen, the termsantigen-sensitive cell or antigen-reactive cell have been usedfor cells that had no earlier contact with the antigen and yetbind a definite type of antigen. The common property of allthese cells is their content of heavy (H) and light (L) chainswhose combination yields antibodies directed against the an-tigen bound by these cells. Since the property of binding adefinite antigen is genetically determined, antibody productionis evidently a two-phase process in which the first phase, theproduction of H and L peptide chains, is directed by the ge-nome. The second phase, the mutual combination of the h andL peptide chains, is a phenotypic process typical of differen-tiation (11).H and L chains as precursors of antibodies and otherimmunoglobulinsSince all immunoglobulins, including antibodies and myelomaproteins, are formed by the combination of H and L peptidechains, these chains are indeed antibody precursors. Both H andL chains contain one NH2-terminal variable (V) region of ap-proximately 110 amino acid residues, and a COOH-terminal

Abbreviations: H chains, heavy chains; L chains, light chains; V,variable region; C, constant region; Ag, native or processed form ofantigen; Dnp, 2,4-dinitrophenyl.

2434

Proc. Natl. Acad. SC{. USA 75 (1978) 2435

constant (C) region. The latter consists of three or four domains;each of these is similar in its size and structure to the VH domain.Fig. 1 shows the structure of an IgG immunoglobulin whichconsists of only two H chains and two L chains. The domainswithin each chain are linked to each other covalently by peptidebonds, whereas the four peptide chains are cross-linked by di-sulfide bridges.

VL VH VH VL

CL-SS-CH1-SS-CH1-&S9 CL

CH2 CH2

CW3 CI3FIG. 1. Structure of rabbit IgG immunoglobulin. H, heavy chains;

L, light chains; V, variable domains; C, constant domains.

It is generally assumed that the combination of the V and Cregions within each of the H or L peptide chains occurs at theDNA level, i.e., as combination of V genes with C genes. Thisprocess seems to take place during ontogenesis; it involvesexcision, insertion, or other forms of translocation within certainDNA sequences, but not typical mutations (12). Transcriptionof the final H and L genes yields the mRNA molecules for theH and L chains; they are translated independently, theH chainson 300S polyribosomes, the L chains on 180S polyribosomes(13). Earlier suggestions that the transcription or translationprocess might be modified by the antigen (14, 15) are inval-idated by these findings. Since the amino acid sequence of theH and L chains reflects the DNA sequence of their H and Lgenes, the variety of the repertoire of H and L chains is genet-ically determined. This repertoire limits the antigen sensitivity,i.e., the antigen-binding capacity of the lymphoid cells and theirresponsiveness.

In the normal, nonimmunized organism the nascent H andL chains in the endoplasmic reticulum are rapidly convertedinto a mixture of normal immunoglobulins (16) which are thensecreted into the circulation. The ability to produce antibodiesis acquired at a certain state of maturation. In the sheep fetus,whose gestation period is 150 days, production of antibodiesagainst some of the injected antigens begins after day 40; an-tibodies against other antigens appear at later times of the in-trauterine life or after gestation (17). The production of de-tectable amounts of antibody is preceded by a phase in whichadded antigen is bound to the surface of the lymphoid cells(18).

Chemical mechanism of immunoglobulin productionin absence of antigenThe combination of H and L chains involves merely the for-mation of interchain disulfide bonds from the two SH groupsof the H chains and the single SH of the L chains. The disulfide(-SS-) bond, although usually designated as a covalent bond, iseasily cleaved by reducing agents and can be considered as

containing the isomeric sulfenium sulfide forms -S: §- and -AS:.Each of the domains of the H and L chains contains alsointrachain disulfide bonds; these are not involved in the com-bination of H and L chains.

In the following discussion the IgG molecule shown in Fig.1 will be represented by the abbreviated symbol L-H-H-L,in which the elongated horizontal lines indicate the disulfidebonds. Immunoglobulins can be formed in vitro by mild oxi-dation of aqueous solutions of H and L chains. Kinetic analyses

reveal that these solutions contain intermediates of the struc-

tures H2, HL, and H2L (16, 19,20). In some immunoglobulinsthe two H chains are linked to each other by two or more in-terchain disulfide bonds and also by other noncovalent bonds.Other immunoglobulins are polymers of the L-H-H-Lstructure. In the following discussion only the simple tetramericL-H-H-L structure will be used as a model. The three in-terchain disulfide bonds in the H2L2 structure are formed bythe mild oxidation of thiol groups of the nascent H and L chains.The biological oxidizing agent is either 02 (reaction I) or a di-sulfide of the general structure R-S-S-R (reaction II). In reac-tions I and II the H and L chains are shown by bold print inorder to distinguish the chains from the hydrogen atoms of thiolgroups.

2H(SH)2+ 2L(SH) + 1.502 -- L-H-H-L + 3H20 [I]2H(SH)2 + 2L(SH) + 3R-SS-R L-H-H-L + 6R-SH

[II]While reaction I, in the absence of reducing agents, is irre-versible, reaction II can proceed in both directions. The equi-librium of reaction II depends not only on the concentrationsof RSSR and RSH, but also on pH, temperature, other physi-cochemical conditions, and obviously also on the concentrationsof any substance that reacts with the nascent H or L chains.Among the three disulfide bonds of the H2L2 tetramer theH-H bond is more easily reduced than the H-L bonds;cleavage of the H-H bond yields two identical half-molecules;mild reoxidation restores the H2L2 tetramer (21, 22). The sta-bility of this tetramer, like that of the hemoglobin tetramera2I82, is attributed to a 2-fold axis of rotatory symmetry (23).The two identical L-H half-molecules are held together notonly by their interchain disulfide bonds, but also by noncovalentinteraction between the two H chains (24).

In vvo, hydrogen transfers from thiols to disulfides are cat-alyzed by thiol-disulfide transhydrogenases of the class EC1.8.4; most of these enzymes are flavoproteins. The best knownbiological hydrogen acceptor in these reactions is glutathione.It is not yet known whether glutathione and transhydrogenasesare involved in immunoglobulin formation in vivo. In vitro theconversion of dissolved H and L chains into immunoglobulinsproceeds in the absence of enzymes; it is probably catalyzed bytrace metal ions (19).The normal immunoglobulins of vertebrates are a highly

heterogeneous mixture of immunoglobulins which differ fromeach other chiefly in their variable VH and VL domains. Evi-dently, the nascent H and L chains, in the absence of antigens,combine with each other stochastically. If the genetically de-termined repertoire of H and L chains consists of m differenttypes of H chains and n different types of L chains, and if thesecombine randomly with each other, the maximum number ofH-L variants would be m X n. Estimates of m and n varyfrom 102 to 103, so that 104-106 different immunoglobulinsmight be produced. However, electrostatic interaction betweenpositively and negatively charged groups, dipole-dipole in-duction, formation of hydrogen bonds, and other intermolecularphenomena will render some of the combinations of H and Lchains more frequent than others. Indeed, some of the normalimmunoglobulins combine specifically with natural or syntheticantigens such as penicillin, insulin, and lysozyme (25) and alsowith synthetic haptens; thus about 1% of the normal human IgGmolecules combines with dinitrophenyl (Dnp) groups, 0.2%combine with 4-hydroxy-3-iodo-5-nitrophenyl-acetyl groups,and 0.02% combine with p-azobenzenearsonate residues (26).Thus small amounts of preformed antibodies are indeed presentin the normal organism before immunization.

Immunology: Haurowitz

Proc. Natl. Acad. Sci. USA 75 (1978)

Interference of antigens with immunoglobulinproductionImmunoglobulin production, according to reactions I and II,must be affected by all substances that have a definite affinityfor the H chains, for L chians, or for the product H2L2. It hasbeen known for several years that antigens combine specificallynot only with the homologous antibody molecules, but also withthe free H chains of these antibodies (27). The affinity of theantigen for the L chains is generally much lower than that forthe heavy chain. Consequently we must conclude that the an-tigen interfere with the normal process of immunoglobulinformation by combining first with those nascent H chains thathappen to have the highest affinity for one of the determinantgroups of the antigen molecule. In reactions III-V the samesymbols are used as in reactions I and II. Since it is well knownthat many antigens need to be "processed" in macrophagesbefore they induce antibody formation, the symbol Ag standshere for the immunogenic form of the antigen; this may be anantigen fragment containing the specific determinant groupor a complex of the determinant part of the antigen moleculewith molecules that enable the antigen to penetrate into thecytoplasm of the antibody-forming cells, where we find it.2Ag + 2H(SH)2 + R-SS-R

Ag ,Ag/

" ,

(HS)H-H(SH) + 2R-SH [III]The broken lines indicate noncovalent bonds between the Hchain dimers and the determinant groups of the two antigenmolecules. The reaction product Ag2H2 reacts then with twoL chains according to reaction IV.

Ag% ,Ag%

(HS)H-H(SH) + 2L(SH) + 2R-SS-R

Ag% , Ag

I / H-H + 4R-SH [IV]L L

Dissociation of the Ag2H2L2 complex according to reaction Vyields two molecules of the immunogenic Ag derivative andone molecule of antibody, H2L2.

Ag% , Ag

H-H 2Ag +

L L

H-H .

L L

The overall reaction of the sequence III-V is:

2H(SH)2 + 2L(SH) + 3R-SS-R -- L-H-H-L + 6R-SH.

In some animal species a dimer of the H chain is an interme-diate in reaction III; in other animals the first intermediateformed in (HS)H-L, which then is converted to (HS)H-H-L and finally to L-H-H-L (16). In each of the equa-tions the hydrogen acceptor R-SS-R can be replaced by 1/2 02;however, this would make the reactions irreversible. The H andL chains in reaction IV must be present in the reduced thiolform so that their oxidation yields the disulfide bridges presentin the antibody molecules. Indeed, it has been found thatantibody production in cell cultures takes place only in thepresence of reducing agents such as mercaptoethanol,CH2(OH)-CH2SH, or a-thioglycerol, CH2(OH)-CH(OH)-CH2SH (28, 29). In the absence of these or similar thiols, anti-body production in cell culture stops. The view that antibody

production is directed by the presence of the antigen, or of itsprocessed form, is supported by the observation that the for-mation of anti-Dnp antibody from its H and L chains is con-siderably accelerated by the presence of the homologous hap-ten, Dnp-lysine (30).

Physicochemical bases of antigen interactionAccording to reactions HI-V, the function of the antigen is thatof a specific cofactor or regulator in the formation of disulfidebridges from the thiol groups of selected nascent H and Lpeptide chains. The course of this dehydrogenation is modifiedby the affinity of the antigen for those H chains whose VH do-mains closely fit the determinant groups of the antigen. De-terminations of the affinity of protein antigens for homologousprecipitating antibodies indicate affinity constants close to106-107 (31). Much lower values, close to 105, are found whenthe macromolecular antigens are replaced by haptens of lowmolecular weight. Similarly, the affinity of the hapten for ahomologous antibody decreases by a factor of approximately102 when the antibody molecule is replaced by its H chain(32).The noncovalent bonds shown in reactions III, IV, and V by

broken lines are formed by short-range or long-range interac-tion. Short-range forces are involved in the formation of hy-drogen bonds, in dipole induction, and in other forms of mutualattraction between closely fitting, complementary structuresin the surfaces of the antigen molecules and the molecules ofthe free H chains. Electrostatic interaction between positivelyand negatively charged groups, because of its long range, is notas highly specific as the short-range interactions. The latterdecrease by a factor r6, whereas the long-range forces decreaseby a factor of only r2, where r is the distance between the de-terminant group of the antigen and the combining site of theantibody molecule. The low specificity of the long-rangeelectrostatic interaction is clearly demonstrated by hapten-specific antibodies directed against the protein-bound m-azo-benzenesulfonate group; these antibodies precipitate not onlyazoproteins containing this group, but also considerable amountsof azoproteins containing o- or p-azobenzenesulfonate,,m-azobenzenearsonate, or m-azobenzoate groups (33). The higherspecificity of the short-range forces is manifested by antibodiesdirected against the succinyl residue -GO-CH2-CH2-COOH;they combine with the residue -CO-CH=CH-COOH in itstrans form, but not in its stereoisomeric cis form (33).One of the prerequisites of strong immunogenicity is a mo-

lecular weight of at least -104. The great importance of thehigh molecular weight of the reactants for the initiation of animmune response is clearly demonstrated by the increase inimmunogenicity when polyamino acids, which are not im-munogenic, are coupled with gelatin, which also is devoid ofimmunogenicity; the macromolecular complex formed in thisreaction is immunogenic. Similarly, polynucleotides acquireimmunogenicity when they are coupled with methylated bo-vine serum albumin, which, as such, is not immunogenic. Theimmunogenicity of many of the small protein antigens can beincreased considerably by coupling them to larger macromol-ecules or by adsorbing them to an adjuvant.

Diffusion-controlled reactions such as enzyme-catalyzedprocesses are enhanced by an increase in the molecular weightof the reactants or by their embedding into a membrane (34).This enhancement is attributed to a decrease in the rotationaland translational mobility upon combination of the two reac-tants. The enhancing action of reactants of high molecularweight in catalysis has been designated by Jencks (35) as "an-

r chor principle" because the strong attractive forces of the

2436 Immunology: Haurowitz

Proc. Natl. Acad. Sci. USA 75 (1978) 2437

macromolecule keep the substrate in a site in which it can un-dergo an extraordinary transformation in form and structure.Evidently, this is also valid for the macromolecular complexAg2H2 formed in reaction III. In this complex the determinantgroup of the antigen is bound to the specific binding site of theH chain while the thiol group of the H chain, located in theremote CHI or CH2 doman, remains accessible to the thiolgroup of a suitable L chain.The earlier mentioned lower affinity of the L chains for the

antigen may be connected with their lower molecular weight,which is approximately half that of the H chains. For the samereason haptens of low molecular weight lack immunogenicity.Their combination with H chains does not significantly increasethe molecular weight of these chians. If present in a large excess,haptens compete with the homologous antigen in reaction III;they thereby prevent antibody formation and lead to tolerancefor that antigen (36, 37). It may be asked why antigens combinespecifically only with certain immunoglobulins and their Hchains and not with other proteins. One of the special propertiesof both chains of the immunoglobulins is their high content ofthe hydroxyamino acids serine and threonine. The largenumber of hydroxyl groups on their surfaces facilitates com-bination with polar groups in the antigens. We do not yet knowwhether similar highly polar protein molecules are present inthe cells of the skin, the smooth muscles, or other tissues; theymight be responsible for the phenomena observed in hyper-sensitivity.Mechanism of some antigen-antibody interactionsThe principal difference between the views presented in thepreceding paragraphs and other widely accepted views is ourclaim that the antigen combines in vivo with H and L chainsrather than with cell-bound antibody molecules. If the antigenwould combine with cell-bound antibody molecules in receptorsor inside the cell, administration of the antigen would cause arapid immune response. However, antibody production in therabbit cannot be detected before day 3 after the primary in-jection. It reaches its peak at day 8. In the secondary response,the antibody titer increases during 3-4 days to its maximum.These immune reactions are quite different from the almostimmediate skin reactions or from the rate of anaphylactic shock.The slow increase in the antibody titer indicates that the antigendoes not react with preformed antibody molecules, but that itcombines with continually formed H and L chains as shown inreactions III and IV. This view is in agreement with the finding(38) that antibody production in an immunized organism issuppressed by the injection of the homologous antibody. Thisfeedback effect indicates clearly that externally administeredantibody molecules compete with the H and L chains for theearlier injected antigen molecules and thus prevent combinationof antigen with H and L chains.The view that the administered antigen combines with H and

L chains rather than with antibody molecules allows us toreinterpret some phenomena for which hitherto there was nosatisfactory explanation. Only some of these can be discussedhere. One of them is the lack of responsiveness to an antigen insome, but not all, individuals of a species. It is usually attributedto the inability of an individual to produce the homologousantibody. According to the view presented above, lack of eitherthe specific H chains or suitable L chains would be sufficientto prevent production of the desired antibody molecules. If onlythe H chains occur in very small amounts, administration of avery small excess of antigen might cause low-dose tolerance,which otherwise is difficult to exnlain.Our views concerning the role of the antigen are difficult to

reconcile with the original form of the clonal selection theory(1) according to which the ability of producing antibodies doesnot depend on the presence and persistence of the antigen.However, reactions III-V are compatible with a modifiedclonal theory which is based on the earlier discussed findingsof Nossal (6, 7, 9), Askonas (8, 10) and their coworkers; theydefinitely proved that antibody formation can take place onlyin the presence of the antigen. It is clear from these investiga-tions that the genetic characteristic transmitted through manygenerations of a clone of lymphoid cells is the ability to producecertain genetically determined types of H and L chains which,in the presence of a suitable antigen, but not in its absence,combine to yield predominantly antibodies directed against theadministered antigen. Apparently the antigen merely modifiesthe phenotypic, stochastic combination of H and L chains, butdoes not affect the genome of the antibody-forming cells.

According to our views concerning the role of the antigen inantibody formation it is clear that almost any macromolecule,natural or synthetic, provided it contains polar groups andprovided it can penetrate to the sites of the production of nas-cent H and L chains, can combine noncovalently with some ofthese chains and thus can act as an antigen. If an antigen isimmunogenic in only some, but not all, individuals of a species,the lack of immunogenicity is usually caused genetically, bylack of the required H or L chains in the nonreactive individ-uals. This genetically caused lack of an immune response mustbe distinguished from the earlier discussed lack of an immuneresponse which can be corrected by increasing the molecularweight, the polarity, or any other factor required for the phe-notypic phase of antibody production, namely, the noncovalentinteraction of the antigen with H and L chains (reactions III-V). Antibodies, according to these views, can be defined asimmunoglobulins produced from the genetically determinedrepertoire of H and L chains in the presence of an antigen thatcombines preferentially with those H and L chains that havea significant affinity for its determinants. These H and L chainsare thus activated for formation of the homologous anti-body.Note Added in Proof. It is not yet clear whether the antibody-pro-ducing lymphoid cells are unipotent or pluripotent. The claim of un-ipotency of the cells is based on the observation that most of the cellsof an animal injected with two different antigens produce antibodiesdirected against one or the other of the two antigens; only very few cellsproduce both types of antibody (39). However, it has been shown re-cently that single lymphoid cells combine with two or more differentantigens (40) and that two or more antigens compete with each otherfor the same cells (41,42). If the cells are indeed pluripotent, the actionof the antigen involves intracellular rather than intercellular selectionof suitable H and L chains. The action of the antigen in the reactionsequence-IIIV might be characterized as instructive, a term frequentlyused in immunology. However, since the purpose of this communi-cation is to describe the biosynthesis of antibodies in terms of chemistryor molecular biology, the action of the antigen is preferably designatedas activation of closely fitting H chains for their disulfide exchangewith suitable L chains.

I am grateful to Mrs. Yvonne Everton for very careful reading ofthis manuscript and for many helpful discussions and suggestions.Research in this laboratory has been supported by Grant AI09752 ofthe National Institutes of Health, and by Grant PCM76-01866 of theNational Science Foundation. Iam deeply indebted to both agenciesfor their support through many years.

1. Burnet, M. F. (1959) The Clonal Selection Theory of AcquiredImmunity (Vanderbilt University Press, Nashville, TN).

2. Richter, M. & Haurowitz, F. (1960) J. Immunol. 84, 123-127.3. Richter, M., Zimmerman, S. & Haurowitz, F. (1965) J. Immunol.

94,938-941.

Immunology: Haurowitz

Proc. Natl. Acad. Sci. USA 75 (1978)

4. Cheng, H. F., Dicks, M., Shellhamer, R. H., Brown, E. S., Roberts,A. N. & Haurowitz, F. (1951) Proc. Soc. Exp. Biol. Med. 106,93-97.

5. Sterzl, J. (1959) Nature 183,547-548.6. Nossal, G. J. V. (1960) Immunology 3, 109-116.7. Nossal, G. J. V. (1969) in Immunological Tolerance, eds. Landy,

M. & Braun, W. (Academic Press, New York), pp. 55-70.8. Askonas, B. A., Williamson, A. R. & Wright, B. E. G. (1970) Proc.

Natl. Acad. Sci. USA 67, 1398-1403.9. Pike, B. L. & Nossal, G. J. V. (1976) J. Exp. Med. 144, 568-

572.10. North, I. R. & Askonas, B. A. (1976) Eur. J. Immunol. 6,8-15.11. Haurowitz, F. (1973) Immunochemistry 10, 775-783.12. Hozumi, N. & Tonegawa (1976) Proc. Natl. Acad. Sci. USA 73,

3628-3632.13. Askonas, B. A. & Williamson, A. R. (1972) Eur. J. Immunol. 2,

487-493.14. Haurowitz, F. (1965) Nature 205,847-851.15. Haurowitz, F. (1976) Cold Spring Harbor Symp. Quant. Biol.

32,559-567.16. Bevan, M. J., Parkhouse, R. M. E., Williamson, A. R. & Askonas,

B. A. (1969) Prog. Biophys. Mol. Biol. 25, 131-162.17. Silverstein, A. M. & Prendergast, R. A. (1970) in Developrmental

Aspects of Antibody Formation and Structure, ed. Sterzl, J.(Academic, New York), pp. 69-77.

18. Decker, J. M. & Sercarz, E. E. (1974) Nature 252,416-417.19. Sears, D. W., Mohrer, J. & Beychok, S. (1975) Proc. Natl. Acad.

Sci. USA 72,353-357.20. Sears, D. W., Kazin, A. R., Mohrer, J., Friedman, F. & Beychok,

S. (1977) Biochemistry 16,2016-2025.21. Mandy, W. J., Rivers, M. M. & Nisonoff, A. (1961) J. Biol. Chem.

236,3221-3226.22. Hochman, J., Cawisk, M., Inbar, D. & Givol, D. (1976) Bio-

chemistry 15,2706-2710.

23. PoIjak, R. J., Amzel, D. M., Avery, H. P., Chen, H. L., Phizack-erley, R. P. & Saul, F. (1972) Proc. Natl. Acad. Sci. USA 70,3305-3310.

24. Nisonoff, A. & Palmer, J. (1974) Science 143,376-379.25. Haimovich, J., Takrab, R., Sulica, A. & Sela, M. (1970) J. Im-

munol. 104, 1033-1034.26. Karjalainen, K. & Makela, 0. (1976) Eur. J. Immunol. 6, 88-

92.27. Olins, D. E. & Edelman, G. M. (1964) J. Exp. Med. 119,789-

815.28. Segal, S., Globerson, A., Feldman, M., Haimovich, J. & Sela, M.

(1970) J. Exp. Med. 131, 93-99.29. Goodman, M. G. & Weigle, W. 0. (1977) J. Exp. Med. 145,

473-489.30. Metzger, H. & Mannik, M. (1964) J. Exp. Med. 120,765-782.31. Haurowitz, F., Sowinski, R. & Cheng, H. P. (1957) J. Am. Chem.

Soc., 79, 1882-1888.32. Painter, R. G., Sage, H. C. & Tanford, C. (1972) Biochemistry

11, 1338-1345.33. Landsteiner, K. (1946) in The Specificty ofSerological Reactions

(Harvard University Press, Cambridge, MA), pp. 169 and 191.34. Richter, P. H. & Eigen, M. (1974) Biophys. Chem. 2,255-263.35. Jencks, W. B. (1975) Adv. Enzymol. 43,219-410.36. Fidler, G. M. & Golub, E. E. (1972) J. Immunol. 112, 1891-

1899.37. Grabar, P. & Escribano, M. J. (1976) C. R. Hebd. Seances Acad.

Sci. Ser. D 282, 1833-1836.38. Finkelstein, M. S. & Uhr, J. W. (1964) Science 146,67-69.39. Nossal, G. J. V. & Lederberg, J. (1958) Nature 181, 1419-

1420.40. Miller, A. (1977) Ann. Immunol. Paris 128C, 611-620.41. Liacopoulos, P., Couderc, J. & Bleux, C. (1976) Ann. Immunol.

Paris 127C,519-530.42. Panijel, J., Delpierre, M. A., Maridat, J. & Terquem, M. (1976)

Ann. Immunol. Paris 127C, 475-487.

2438 Immunology: Haurowitz