Proc. Nat. Acad. Sci. USAVol. 69, No. 1, pp. 124-128, January 1972

Characterization of the J Chain from Polymeric Immunoglobulins(IgA/IgM/immunological specificity/primary structure)

SHERIE L. MORRISON* AND MARIAN ELLIOTT KOSHLAND

Department of Bacteriology and Immunology, and the Department of Molecular BiologyUniversity of California, Berkeley, Calif. 94720

Communicated by William Z. Hassid, October 26, 1971

ABSTRACT The J chains isolated from polymeric IgAand IgM were shown to be distinct from the other im-munoglobulin subunits, including the secretory com-ponent, both on the basis of amino-acid content and anti-genic determinants. By the same criteria, the J chains werefound to be indistinguishable from each other whetherthey were derived from pentamer IgM, polydispersemyeloma IgA, or dimer colostral IgA. The only differenceswere observed in (a) the extent of disulfide bond cleavagerequired for J chain release from the different polymers,and (b) the three-dimensional arrangement of J chains inthe different polymers. These data support the hypothesisthat the same J chain joins the monomeric units of IgAand IgM to form their respective polymeric molecules.

Two classes of immunoglobulins are characterized by poly-meric structures. In both classes the polymers are assembledfrom IgG-like monomeric units consisting of two heavy andtwo light chains, linked by disulfide bonds. However, thedegree of polymerization varies with the class and is a func-tion of the distinct sequences in the respective heavy chains.Thus, IgM is predominantly pentameric, while IgA is foundin the secretions as a dimer and in the serum in both mono-meric and various polydisperse forms.

Although many of the structural features of the polymershave been determined (1, 2), several critical questions remainconcerning their biosynthesis. First, the linkage between themonomeric units has never been clearly established. Disulfidebonds are known to be involved, but the exact location of thebonds and the mechanism for controlled polymerization arenot understood. Second, the biological significance of thepolymers has not been clearly defined, but there is increasingevidence that the properties acquired through polymeriza-tion may be important in the specialized functions of IgMand IgA. For example, recent studies (3) have shown that theincreased affinity achieved by the multi-attachment of IgMto the same antigen may provide effective protection duringthe early stages of the immune response. In the case of IgA,studies (4) have shown that the dimer is transported acrossepithelial barriers because of its ability to interact with thetransport protein, the secretory component.A clue to these problems was obtained when a new compo-

nent, designated J chain, was found to be part of the covalentstructure of human and rabbit polymeric IgA (5). A similarcomponent was subsequently described in human IgM (6).Although J chain is approximately of the same size as the lightchain, it was shown to be distinctly different on the basis of itsamino-acid composition, peptide mapping, and antigenicproperties (5, 6). Furthermore, the J chain was found to be

linked by disulfide bonds to the carboxy-terminal halves ofthe a and u chains, while the light chains are joined to theamino-terminal portions (7, 8). The presence of J chain in thepolymeric immunoglobulins and its absence from monomericIgA and monomeric IgG suggested that one of its functionsmight be to join the monomeric units.To ascertain its linkage role, it was necessary to show that

the J chain was not an incidental fragment derived from theheavy chains or secretory component. Moreover, it was crit-ical to establish the identity of Ja and JA and clarify the link-age mechanism in the different size polymers. In the presentstudy, these questions were approached by isolating J chainin as native a form as possible from human IgM and IgApolymers. After purification by affinity chromatography, theJ chains were analyzed for their amino-acid contents and theextent of disulfide bond cleavage. They were also used for thepreparation of specific antibody to determine the immuno-logical relationships between the J chain and the other chains,and to explore the three-dimensional arrangement of the Jchain in the intact polymers.

MATERIALS AND METHODS

Antiserum. Antiserum was prepared by injection of rabbitswith a total of 1 mg of antigen (light, heavy, or J chain) incomplete Freund's adjuvant. The dose was administered byfour subcutaneous injections, two in the foot pads and two inthe back. Precipitating antibody appeared 3-4 weeks later.

Immunoglobulin Purification. Human colostral IgA was iso-lated by the method of Tomasi and Bienenstock (9), humanmyeloma IgA was isolated by the method of Vyas and Fuden-berg (10), and human Waldenstrom's macroglobulin was pre-pared by the method of Morris and Inman (11).

Subunit Purification. The component chains of the poly-meric immunoglobulins were prepared by (1) preliminaryseparation on a sizing column of Sephadex G-100 after mildreductive cleavage of the parent protein, and (2) purificationon immunoadsorbent columns to which were conjugated theappropriate monospecific antisera. For the preliminary separa-tion, myeloma IgA and IgM were reduced with 0.1 M 2-mercaptoethanol and colostral IgA was reduced with 0.2 M2-mercaptoethanol; the resulting SH groups were alkylatedwith a concentration of iodoacetamide in 5% excess of thereducing agent. The heavy chains and secretory componentwere then separated from light and J chains by chromatog-raphy of the reduced and alkylated mixture on a column ofSephadex G-100 equilibrated with 1 M acetic acid (12).

This procedure was also used to prepare antigens for im-

124

* Present address: Department of Cell Biology, Albert EinsteinCollege of Medicine, Bronx, N.Y. 10461

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

Characterization of the J Chain 125

munization and for coupling to immunoadsorbents. PurifiedK and X chains were obtained from IgG K and X myelomas;purified a chain was obtained from monomer IgA myelomathat lacked J chain and secretory component.For the immunoadsorbent columns, Sepharose beads were

activated by cyanogen bromide according to the method ofCuatrecasas et al. (13). 5-10 mg of K, X, or a chain were thencoupled to the beads at a ratio of 2 mg of protein to 1 ml ofpacked resin. After thorough washing with PBSA [0.01 Mphosphate-0.15 M NaCl-0.02% azide (pH 7.5) 1, these prep-arations were used to isolate their respective homologousantibodies. Rabbit antiserum was passed through the Sepha-rose beads until the antigen binding sites were saturated.The column was washed free of nonspecifically absorbed pro-tein with PBSA and the specific antibody was eluted with0.1 M glycine (pH 2.5). The yield of antibody ranged from16 to 30 mg. After dialysis against carbonate buffer (pH 9.0),each antibody was conjugated to Sepharose under the condi-tions described above. The bound antibody retained 70-90%of its combining capacity.

Before the separation of J and light chains, the fractionsobtained from the acetate chromatography were dialyzed toneutrality against 0.01 M Tris buffer (pH 8.0) and concen-trated by lyophilization. Depending on the capacity of the im-munoadsorbent column, aliquots of 1.5-3 mg were passedthrough a column containing the appropriate antibody to thelight chain. Purified J chain eluted at one column volume andthe bound light chains were removed with 0.1 M glycine(pH 2.5). The immunoadsorbent was then washed back toneutrality with PBSA and used again. a chain and secretorycomponent were isolated by applying the same procedures toan anti-a chain immunoadsorbent.

Immune Diffusion. 5 ml of a 1% Ionagar solution contain-ing 0.15 M NaCl were pipetted into plastic Petri dishes,5 cm in diameter. After the agar solidified, wells 0.6 cm indiameter were made in the agar. Into each well was pipetted40 ul of either antigen or antibody at the appropriate con-centration.

Amino-Acid Analyses were performed as described (14).

Inhibition Tests. Anti-J antiserum was combined with intactIgM or IgA or their isolated subunits and brought to a finaldilution of 1:4. When necessary, nonspecific immunoglobulinwas added so that the concentration of added protein wasmaintained at 3.5 mg/ml. After incubation overnight at 4VC,the mixture was centrifuged to remove nonspecific precipitate.An aliquot of 1.2 ml was then removed, pipetted into acuvette, and the amount of J chain known to give precipita-tion at equivalence was added. The precipitation was followedas a function of time by reading the absorbance at 460 nm ina Zeiss spectrophotometer.

RESULTS

Previous work (5) indicated that the disulfide bonds linkingJ to the a or u chain are relatively resistant to reductive cleav-age. To isolate J chain in as native a form as possible, it wasnecessary to use the mildest reducing conditions that wouldgive a good yield of J chains. Treatment with 100 mM 2-mer-captoethanol followed by alkylation with 105 mM iodoacet-amide was found to be satisfactory for the myeloma IgA andIgM, but twice these reagent concentrations were required to

mers were reduced under these conditions and then chromato-graphed on a Sephadex G-100 column equilibrated with 1 Macetic acid, at least 95% of the light-chain and 20-30% of theJ chain were released. The bulk of the J chain eluted with theascending limb of the light chain-peak. The difference in theelution positions of J and light chain was dependent on twofactors: first, the slightly larger size of J, 24,500 +

1,000 daltons (15), compared to 23,000 for light chain, and sec-

ond, the extent of reductive modification of the light chain. Themaximum difference was achieved when only the light-heavyinterchain bond was broken. As an increasing number of itsintrachain bonds were reduced, the light chain eluted pro-

gressively earlier, until its peak coincided with that of J chain.The J was separated from the light-chain mixture by use of

a Sepharose column to which was conjugated the appropriateantibody to human light chain, either anti-K, anti-X, or a mix-ture of both. With the proper choice of sample size, all thelight chain was bound to the resin and the J chain eluted atone column volume. In some preparations from myeloma andsecretory IgA, small amounts of an a-chain fragment were

found to cochromatograph with J chain. In these cases, theJ chain was purified by an additional passage through an im-munoadsorbent prepared with anti-a chain antibody.For each polymeric serum immunoglobulin studied, amino-

acid analyses were performed on the isolated J chain, the lightchain eluted from the immunoadsorbent, and the monomericheavy chain that was obtained in the acetate chromatography.In the case of colostral IgA, the heavy-chain peak containedsecretory component as well as a chain. These were separatedby use of an anti-a chain immunoadsorbent and then sub-jected to amino-acid analysis. The data from the componentsof a colostral IgA are presented in Table 1 to illustrate theresults obtained (columns 1-4). The average composition of Jchain was clearly distinct from those of the other subunits.The differences were so great, particularly in the arginine,aspartic acid, glycine, isoleucine, and half-cystine contents,that J could not be explained as a fragment from heavy chainor secretory component, or as a special type of light chain.Furthermore, the composition of the J chain was identical,within experimental error, whether the J chain was preparedfrom colostral IgA, myeloma IgA, or a Waldenstrom's IgM(columns 4-6 of Table 1).Although the various J chain preparations had the same

total half-cystine content, differences were observed in thenumber of half-cystines alkylated under the mildest conditionsof reductive cleavage. After treatment with 100 mM 2-mercap-toethanol and 105mM iodoacetamide, the J chain released frommyeloma IgA contained an average of 8 residues of amido-carboxymethylcysteine, while the J chain released from pen-

tamer IgM contained an average of 10 residues. No free sulf-hydryl groups could be detected in the J chain by alkylationof the intact immunoglobulins, so that the difference in theamidocarboxymethylcysteine yields appeared to reflect a

difference in the disulfide structure of the J chain. Thus, thereare either fewer J-a chain bonds than J-,u chain bonds or theJa intrachain bonds are more resistant to reductive cleavage.When the concentrations of 2-mercaptoethanol and iodoacet-amide were increased to 0.2 and 0.21 M, respectively, all 12 ofthe half-cystines were alkylated in both Ja and J,.To study the immunological relationships between J and

the other immunoglobulin subunits, antisera were preparedfree any J chain from colostral IgA. When the various poly-

Proc. Nat. Acad. Sci. USA 69 (1972)

in rabbits against the partially alkylated. J chains isolated

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

126 Immunology: Morrison and Koshland

Jo

._-* '4

a

anti-a

anti -i

K

anti-K

x. .

'I

A

B CFIG. 1. Immunological specificity of J chain. (A) Reaction of anti-J antiserum with the isolated immunoglobulin chains. See accom-

panying diagram for explanation. (B) Reaction of anti-J antiserum with secretory component isolated from colostral IgA. Wells 1, 2, 4,and 6 contain secretory component at concentrations of 0.4, 0.1, 0.05, and 0.02 mg/ml, respectively. Wells 3 and 5 contain J chain fromcolostral IgA at a concentration of 0.02 mg/ml. (C) Reaction of anti-J antiserum (center well) with J chain isolated from (1) IgM, (2)colostral IgA, (3) myeloma IgA-K, (4) myeloma IgA-X and with intact myeloma IgA (5) and intact IgM (6). The J chain solutions con-tained 0.02 mg/ml; the intact polymers were tested at a concentration of 18 mg/ml.

from an IgA-K and an IgA-X myeloma. Although the re-sponse was low, about 0.1 mg of antibody per ml of sera, theantibody had a high affinity for J and was precipitated by theaddition of as little as 5 uAg of antigen per ml of sera. In pre-liminary precipitin tests, all the antisera were found also togive a reaction with a chain or with proteins containing achain. This reactivity could be removed by absorption of theantisera with mildly reduced and alkylated heavy chain pre-pared from J-free monomer IgA. The results of Ouchterlonyanalyses performed with the absorbed antisera are given inFig. 1. The antisera were still capable of combining specificallywith Ja and Jp, as indicated by the strong precipitin lines ob-tained. On the other hand, the antisera showed no reactionwith 'c or X chains over a 100-fold range in antigen concentra-tion (Fig. 1A), nor with secretory component over an80-fold antigen dilution (Fig. 1B). The absence of crossreactionwith light chain was expected, since the method of purifying Jchain was based on the observation that the J chain was not

bound by any antilight-chain antibody. These results pro-vided strong evidence that the J chain is antigenically dis-tinct from both light chain and secretory component.The question remained whether the observed anti-a activity

reflected a crossreaction between J and a chain. To test thispossibility, the fraction of the anti-J antiserum that was boundto the anti-a immunoadsorbent was examined for its speci-ficity. After elution, the fraction retained its capacity toprecipitate a chain but exhibited no reaction with J chain. Ina second test, 420 nmol of J chain were passed through a10 nmol column of Sepharose beads to which were conjugated210 nmol of anti-a antibody. Although J was exposed to a 10-fold excess of anti-a chain sites, over 97%, 41 nmol, was re-covered in the eluate. In contrast, when 380 nmol of mildlyreduced and alkylated a chain were applied to the same immu-noabsorbent, 360 nmol were bound to the resin. The results ofthese two experiments indicated that J chain has no determi-nants in common with a chain and that the observed anti-a

Proc. Nat. Acad. Sci. USA 69 (1972)

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

Characterization of the J Chain 127

activity represented noncrossreacting antibody producedagainst a minor a chain contaminant in the J preparationsused for immunization.The Ouchterlony analyses also revealed that Ja and J, are

antigenically indistinguishable from each other. As shown inFig. 1C, a single line of indentity was obtained among the Jchains isolated from a Waldenstrom's IgM, from K and XIgA myelomas, and from normal colostral IgA. The specificitywas not appreciably affected by the extent of disulfide bondcleavage. In one Ja preparation, 8 half-cystines were alkylated,while in the other all 12 half-cystines were modified. Yet bothpreparations gave a reaction of identity with each other andwith JA. Furthermore, a reaction of near identity was ob-tained between completely reduced and alkylated J,, and re-oxidized, reformed Ja. The reoxidized, reformed J,, prepara-tion was isolated by chromatography of IgA on a columnequilibrated with reducing agent, and the S-S bonds were thenallowed to reform during dialysis against buffered saline.No precipitation was observed when the anti-J sera were

tested against intact IgA or IgM that contained J chain. Theresults were negative even at immunoglobulin concentrationsof 10-20 mg/ml (Fig. 1C). However, the myeloma IgA orIgM was found to inhibit the reaction between J chain and itsantisera. Large quantities were required because the J chaincomprised such a small percentage of the whole molecule.When the J-chain concentration in the inhibitor was equal to

TABLE 1. Comparison of the composition* of J chainisolated from colostral IgA with (1) the compositions of theother subunits present and (2) the compositions of J chains

from other Ig polymers

Aminoacid

LysHisArgAsxThrSerGlxProGlyAlaValMetIleLeuTyrPheCMCys-CRAtCMCys-MRA§

at

18.09.319.234.044.843.844.040.535.832.237.03.87.6

46.014.215.015.5

Colostral IgA

SCt Lightt

30.4 11.15.0 2.6

28.3 5.464.0 14.630.9 18.141.7 27.363.49 23.032.6 14.258.1 15.332.0 15.651.1 15.90 0.4

20.9 5.355.7 14.824.5 9.018.8 6.617.7 5.113.0 1.0

Jt

9.21.5

15.436.019.511.124.012.93.49.3

16.31.2

14.413.09.02.1

12.3

My-

elomaIgA

J

9.11.4

15.235.919.411.224.012.93.69.2

16.11.0

14.313.09.12.1

12.18.3

IgM

J

9.41.8

15.234.918.911.324.012.93.99.015.81.1

14.113.39.02.511.610.1

* After 20-hr hydrolysis. Average of three different prepara-

tions.t Moles of amino acid/peptide molecular weight. a chain =

53,000, secretory component (SC) = 67,500, light chain (L) =

23,500, and J chain = 24,000.1 After complete reduction and alkylation; CMCys, represents

amidocarboxymethylcyteine.§ After reduction with 100 mM 2-mercaptoethanol and alkyla-

tion with 105 mM iodoacetamide.

0 20 40 60 80 100 120 0 20 40 60 80 100 120

minutes

FIG. 2. Inhibition by polymeric IgA and IgM of the pre-cipitin reaction between free J chain and its specific antibody.See text for details. A, IgM; *, IgA; *, control.

that of the test antigen, only the rate of precipitation wasaffected. At higher inhibitor concentrations, both the rateand the amount precipitated at equilibrium were significantlydecreased. The kinetics of the inhibition determined with thesame preparations of IgA and IgM, but with two differentanti-J sera, are illustrated in Fig. 2. In these experiments thetotal protein concentration was kept constant to avoid non-specific solvent effects. The control precipitation was per-formed in the presence of 3.5 mg of bovine gamma globulinand the substitution of an equal quantity of an IgG K myeloma,and IgG X myeloma, a chain, or secretory component werefound to have no effect on the reaction. The amounts of my-eloma IgA and IgM were chosen so that the ratio of inhibitorJ to antigen J was 30:1 and the total protein concentrationwas adjusted to 3.5 mg by the addition of the appropriatequantities of nonspecific immunoglobulin. The myeloma IgAand IgM blocked the reaction of J with one antiserum to anequivalent extent (Fig. 2A). For example, the precipitationwas inhibited by 50% after 60 min of reaction time and by20% at equilibrium. With the second antiserum (Fig. 2B), theIgM was a poorer inhibitor than the IgA; at 60 min theamount of precipitate was decreased by 30% in the presenceof IgM, compared to 60% inhibition in the presence of IgA;at equilibrium, the values were 15 and 25%, respectively.When secretory IgA was tested under the same experimentalconditions, no significant inhibition was observed. These re-sults indicated that most of the J chain was buried within thepolymeric structure of myeloma IgA or IgM. Only one or, atmost, a very limited number of determinants was accessiblefor reaction, sufficient to bind to the anti-J serum but not tobuild up insoluble complexes. Furthermore, the independentpatterns of inhibition obtained for IgA and IgM suggestedthat a different J determinant was exposed in the dimer thanin the pentamer. In contrast, the J chain appeared to be com-pletely inaccessible in secretory IgA, perhaps because of theadditional blocking action of the secretory component.

DISCUSSIONWhen the J chain was first described, it was hypothesized to bea novel polypeptide chain because of its characteristic anodalmobility in alkaline-urea disc electrophoresis, its differentamino-acid composition, and its presence only in the poly-meric forms of IgA and IgM (5). The studies presented here

Proc. Nat. Acad. Sci. USA 69 (1972)

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

128 Immunology: Morrison and Koshland

confirm and extend this hypothesis. First, the amino-acid com-position of purified J chain was found to be so distinctive thatit could not have been derived directly from any otherimmunoglobulin subunit. For example, J chain contains moreisoleucine residues than either a or 1A chains, which are twiceits size; it contains at least twice the half-cystine contentof light chains and more methionines than the secretory com-ponent. Second, the J chains isolated from polymeric IgA andIgM were identical by the criteria of composition and anti-genicity. Third, the J chain was shown to be antigenicallydistinct from the other subunits of IgA and IgM. J chain wasnot bound by antilight-or antiheavy-chain antibody, nor didantisera to the J chain exhibit any crossreaction with lightchains, heavy chains, or secretory component. The argumentmight be made that the antisera were prepared against mildlyreduced and alkylated J chain and thus did not recognize Jor crossreacting determinants in their native conformation.This argument was invalidated when the antisera were shownto combine with the available J determinants in intact serumIgA and IgM molecules and to precipitate reoxidized, re-formed J chain almost as well as the homologous, alkylated Jantigen.

Because of its high content of half-cystine and chargedamino-acid residues, it seems unlikely that the J chain wasevolved from the same common precursor as light and heavychains. Moreover, evidence has been obtained that the Jchain arose very early in the development of the immunesystem, before the divergence of y chains. Studies on the singleIgM-like class synthesized by the elasmobranch, the leopardshark, have shown that J is present in the 19S, but absentfrom the 7S form (16). Thus, it would appear that in theevolution of the a and , chains, structures were selected thatcould form covalent S-S links with the J chain. It would be ofconsiderable interest to determine whether these structureswere lost in the evolution of the -y chain or whether the switch-over from IgM or IgA to IgG synthesis involves the coordinateturning off of the gene responsible for J chain production.The proposal that the J chain has a joining function has been

supported by quantitative determinations of its stoichiometrythat showed one J chain per polymer independent of thesize of the polymer (15). Additional support was obtained inthe present studies. In the isolation of J chain, it was foundnecessary to modify 8 of the 12 half-cystines in Ja and 10 ofthe 12 half-cystines in J., before appreciable amounts werereleased from the respective polymers. Since interchain S-Sbonds are more susceptible to reduction under the conditionsused, these results indicated that J was joined to the heavychains by multiple disulfide bridges and that the number ofbonds was sufficient to allow J to be joined to each heavy chainin the polymer. Moreover, the difference in the number ofhalf-cystines modified in Ja and J,1 suggested that the disulfidebonds can be alternately arranged so that the same J chaincan function in the formation of various size polymers. Theconcept of a J chain bridging each heavy chain would requireJ to be a linear molecule. Evidence for such a structure wasobtained when the antigenic specificity of the J chain wasfound to be quite independent of the extent of half-cystinemodification. Since the S-S bonds do not contribute signifi-

cantly to the conformation of the antigenic determinants, theJ molecule must be sufficiently stretched out to separate thedisulfide bridges from the residues that determine the anti-genic sites. While the data from these and other studies sup-port the proposed linkage model, its rigorous proof will de-pend on the reconstitution of the proper polymer from J andmonomer units, and the demonstration of a link between J andeach heavy chain.During the assembly process, the Jchain appears to be folded

inside the polymers. The inhibition experiments showed thatonly one determinant in IgM and myeloma IgA and none insecretory IgA was available for reaction with antisera to theJ chain. These data explain the usual failure to elicit antibodiesto J chain after immunization with polymeric IgA or IgM andthey suggest that an effective J antigen might be obtained bythe partial denaturation or reduction of the polymers to in-crease the exposure of J determinants. The inhibition experi-ments also indicated that a different J determinant is exposedin IgM than in IgA. This result would be expected from thedifferences both in the structure of the pentamer and thedimer, and in the J linkage. However, since the same J chainis present in both polymers, the exposure of different deter-minants might also be important in maintaining the specificbiological roles of the IgM and IgA classes.The evidence that a specialized polypeptide chain induces

the polymerization of other peptide chains is at presentlimited to the immunoglobulins. However, the use of such achain would seem to offer an opportunity for controlled poly-merization and added biological function.

This investigation was supported by U.S. Public Health ServiceResearch Grant Al 07079 from the Institute of Allergy and In-fectious Diseases.

1. Metzger, H. (1970) in Advances in Immunology, ed. I)ixon,F. J. & Kunkel, H. G. (Academic Press, New York), Vol.12, pp. 57-116.

2. Cebra, J. J. & Small, P. A. (1967) Biochemistry 6, 503-512.3. Hornick, C. L. & Karush, F. Immunochemistry, in press.

--4. Tourville, D. R. Adler, R. H., Bienenstock, J. & Tomasi,T. B. (1969) J. Exp. Med. 129, 411-430.

5. Halpern, M. S. & Koshland, M. E. (1970) Nature 228, 1276-1278.

6. Mestecky, J., Zikan, J. & Butler, W. T. (1971) Science 171,1163-1165.

7. Meinke, G. C. & Spiegelberg, H. L. (1971) Fed. Proc. 30,468.

8. Mestecky, J., Kulhavy, R. & Kraus, F. W. (1971) Fed.Proc. 30, 468.

9. Tomasi, T. B. & Bienenstock, J. (1968) in Advances inImmunology, ed. Dixon, F. J. & Kunkel, H. G. (AcademicPress, New York), Vol. 9, pp. 1-96.

10. Vyas, E. & Fudenberg, H. (1969) Clin. Res. 17, 469.11. Morris, J. E. & Inman, F. P. (1967) Biochemistry 7, 2851-

2857.12. Fleischman, J. B., Pain, R. H. & Porter, R. R. (1962) Arch.

Biochem. Biophys. Suppl. 1, 174-180.13. Cuatrecasas, P., Wilchek, M. & Anfinsen, C. B. (1968)

Proc. Nat. Acad. Sci. USA 61, 636-643.14. Koshland, M. E., Englberger, F. M. & Shapanka, R. (1966)

Biochemistry 5, 641-651.15. Halpern, M. S. & Koshland, M. E. submitted to Biochem-

istry.16. Klaus, G. G. B., Halpern, M. S., Koshland, M. E. & Good-

man, J. W. submitted to J. Immunol.

Proc. Nat. Acad. Sci. USA 69 (1972)

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021