STRUCTURALUNITS OF HUMAN7S GAMMAGLOBULIN*

By EDWARDC. FRANKLIN WITH THE TECHNICAL ASSISTANCE OF FRANCESPRELLI

(From the Department of Medicine and the Rheumatic Diseases Study Group, New York Uni-versity College of Medicine, New York, N. Y.)

(Submitted for publication May 24, 1960; accepted August 17, 1960)

A number of observers in the past (1-3) havesuccessfully degraded rabbit and horse antibodieswith various enzymes, thus producing fragmentsretaining antibody activity. More recently Porter(4, 5) has produced split products of rabbit y-glob-ulin by treatment with papain and cysteine, two ofwhich retained antibody activity as measured bytheir ability to inhibit precipitation of the originalantiserum with homologous antigen, while the thirdappeared to retain the greatest antigenic specificity.Karush (6) and Nisonoff, Woernley and Wissler(7, 8) working with antihapten antibodies haveclearly shown that these fragments are univalent.Putnam, Hsiao and Tan (9), employing similartechniques, have been able to split human y-globu-lin and myeloma proteins into two major chro-matographically separable fractions with sedimen-tation coefficients of approximately 3.5S. Apreliminary report of some physical, chemical andimmunological properties of similar subunits fromhuman y-globulins and several human antibodieshas been presented (10). In the present report,results are presented which indicate that followingdigestion of human y-globulin there are producedat least three fragments differing from each otherin chemical properties. The major part of theantibody activity against three different antigenswas found to be associated with one of these frag-ments, while smaller amounts were present in asecond, antigenically similar fragment. The thirdfragment contained virtually no antibody activitybut had additional antigenic specificity.

MATERIALS AND METHODS

Human 'y-globulin was obtained as Cohn fraction II(Lederle, Lot no. C576). Ultracentrifugal examinationdemonstrated that this preparation consisted of approxi-mately 95 per cent 7S y-globulin, and that it containedless than 5 per cent of molecules with s rates of 9 to 11Sand no detectable amounts of 19S 'y-globulin. For stud-

* Supported by the Arthritis and Rheumatism Founda-tion, and by the United States Public Health Service,Grants A-2594 and A-1431 (C2).

ies of specific antibodies, y-globulins were also isolatedby starch zone electrophoresis (11) from the serum of apatient who developed serum sickness following the in-jection of tetanus antitoxin, and by precipitation with 34per cent saturated ammonium sulphate from the serumof a patient hyperimmunized with diphtheria toxoid.

Breakdown of 'y-globulin was carried out as describedby Porter (5) in pH 7.0 phosphate buffered saline, 0.01M cysteine and 0.002 M ethylenediamine tetraacetate(EDTA) (Versene), using 1 mg twice recrystallizedpapain per 100 mg protein.' The breakdown was allowedto proceed at 370 C for 12 to 18 hours. The material wasthen dialyzed for 4 hours against 50 to 100 ml pH 7.6,0.01 Mphosphate buffer and then against 1 L of the samebuffer for 24 hours. The amount of dialyzable materialwas determined by the modified Folin technique using'y-globulin as a standard (12). The exact nature of theseproducts was not examined further. In a number of ex-periments, 3 mg pepsin per 100 mg 'y-globulin at pH 4was used, alone or in conjuunction with subsequent treat-ment, with 0.1 M mercaptoethanol as described byNisonoff, Wissler and Woernley (8).

Ultracentrifugal analyses. Ultracentrifugal analyseswere carried out in a Spinco Model E ultracentrifuge,using cells with a double sector centerpiece. Quantita-tive measurements were carried out directly from thephotographic plates using a comparator (Gaertner) ac-cording to methods described by Trautman (13). Prepa-rative ultracentrifugation was performed in a SpincoModel L ultracentrifuge, using a swinging bucket rotor(SW-39) and a saline gradient as described previously(14).

Chromatography. Chromatography was carried outon carboxymethylcellulose (CM cellulose) or Diethyl-aminoethylcellulose (DEAE cellulose) as described byPeterson and Sober (15).2 For preparative purposesglass columns, 2.5 cm ODand 40 cm long, were employed.Ten or 20 g of adsorbent was washed with distilled waterand the starting buffer and packed into the column withapproximately 10 pounds per square inch air pressure toa height of 15 and 30 cm, respectively. The columnswere equilibrated with a continuous flow of 2,000 mlstarting buffer. Two hundred fifty to 500 mg of 'y-globu-lin was dialyzed for 4 to 24 hours against the same bufferand carefully applied to the column. In the initial experi-ments, elution was carried out by gradient elution at con-

1 Prepared by Dr. J. Potter.2 Purchased from Brown & Co., New Hampshire:

DEAEcellulose, Lot no. 1078, 1066, capacity 1 mEq perg; CMcellulose, Lot no. 1100, capacity 0.7 mEqper g.

1933

EDWARDC. FRANTKLIN

stant pH and increasing ionic strength, as described byFahey and Horbett (16). The gradient was started aftercollection of the protein that came off with the solventfront. Since only a single additional peak was obtainedfrom each column, subsequent studies were carried out bystepwise elution. Elution from the cation exchanger wascarried out with pH 7.6 phosphate buffer and from theanion exchanger at pH 8.0. Ionic strengths of 0.4 and0.3, respectively, were obtained by adding sodium chlorideto 0.01 M phosphate buffer. Ten-ml aliquots were col-lected with an automatic fraction collector, and proteinwas measured by reading the extinction in a BeckmanDU spectrophotometer at 280 m,u. An approximation ofthe amount of protein in each tube was obtained bycomparison with a standard curve prepared with Cohnfraction II y-globulins. These values can be consideredonly as an approximation, since the amino acid composi-tion of the fragments is not knowni.

Ihomumological and c-hemiiical sttudies. Immunologicalstudies were carried out using a pooled antiserum to 7S,y-globulin prepared in five rabbits over a prolonged pe-riod of immunization, and antisera to the whole unfrac-tionated digested -y-globulin prepared in rabbits withFreund's adjuvant as previously described (17). Studieswere carried out by double diffusion in agar by thetechnique of Ouchterlony (17), and by micro-immuno-electrophoresis by the techlnique of Scheidegger andBuzzi (18). Quantitative precipitin analyses were doneas described previously (17) usinig 0.05 to 0.5 ml anti-serum.

Hexose was determined by the anthrone reaction (19)using a mixture of glucose, fucose and mannlose as a stand-ard, and hexosamine by a modificationi of the Elson-Morgan reaction usinlg glucosaninle as a standard (20).To calculate the hexose protein ratio of the fractions,protein was measured by the modified Folin techniqueusing Cohn fraction II -y-globulin as a standard (12).

Anitibody assays. 1) Precipitating antibody to horseserum proteins in electrophoretically isolated y-globulinsfrom a patient wvith serum sickness was assayed by quan-

FIG. 1. ULTRACENTRIFUGALPATTFERNS OF TWO PREPA-RATIONS OF DIGESTED ly-GLOBUL IN SHOWINGIN ADDITIONTO TILE MAJOR 3.5 SPEAK, TIE PRESENCE OF VARIABLEAMOUNTSOF A MORERAPIDLY SEDIMENTING 5 To 65 com-PONENT, SPEED 52,640 RPM. Photographs taken after 64miinutes. Sedimentation procce(ls from left to rioht.

titative precipitin techniques (17). Inhibitory power ofdigested fragments was determined by quantitating theirability to decrease the amount of precipitate between theoriginal antiserum and homologous antigen, and by theircapacity to inhibit the passive cutaneous anaphylacticreaction when mixed with the antiserum injected into theskin of the guinea pig (21).3

2) Antibodies to diphtheria toxin in y-globulins iso-lated from a hyperimmunized normiial subject and in frag-ments prepared from them, were measured by their ca-pacity to neutralize the effect of the toxin in the skin ofa rabbit (22).4

3) Antibody to streptokinase was present in the prepa-ration of Cohn fraction II, Lot no. 576, employed in thesestudies and was estimated by its ability to inhibit the ac-tivity of streptokinase using the fibrinolytic assay de-scribe(l by Johnson and Tillett (23).5

RE,1SULTS

Treatment of 7S y-globulin with papain, cysteineand EDTA, as described uinder Methods, resultedin virtually complete conversion of the startingmiiaterial which had an observed s rate of 6.5 to6.8S into a major fraction with sedimentation co-efficient of 3.4 to 3.5S and smiall but variableamnounts of more rap)idly sedimenting materialwith a sedimentation coefficient of 5 to 6S. Thisse(linmented somewhat miiore slowly than the start-ing nmaterial and probably represents partially de-graded y-globulin. Approximately 80 to 90 percent of the total was not dialyzable, while the re-mainiing 10 to 20 per cent appeared as dialyzablenitrogenotus products. Two preparations oftreated y-globulin prepared from the same lot btutshlowing different amounts of the more rapidlysediimienting material are shown in Figure 1.

Wlhen the digest was subjected to chromatog-rap)hv on a CMI cellutlose columiin at pH 7.6 usinga gra(lient of 0.01 to 0.4 ALI NaCl, two major peakswere recovered. One of these came off with thestarting buffer, and the second one shortly afterthe start of the gradient. Since this second peakcould not be resolved further, subsequent experi-ments were carried out by stepwise elution usingthe starting buffer and 0.4 MNaCl, pH 7.6. Theresults of one such experiment are shown in Fig-ture 2 (top). The amounts of protein associatedwith eaclh peak were quite variable; in general,l)eak I cointaine(l soimewhat less protein than peak

3 Kindly performed by Dr. Zoltan Ovary.4 KiIIdly perfornmed by Dr. Jonathani Uhr.

Kiid(ly performed by D)r. AlaII Johlnson.

1934

STRUCTURALUNITS OF GAMMAGLOBLTLIN

2.0 - 2.0 -

6k 1.6-16 -

E ~ UB NUBRTBENM

0 0

0

1.2 -

cv j~~~~~~~~~~~~.2-..8

Oi A C D

.4 .4-

10 20 304050 60 10 203040 50

TUBE NUMBER TUBE NUMBER

FIG. 2. CHROMATOGRAPHYOF DIGEST OF HUMANly-GLOBULIN. Top: first step on

CM cellulose, pH 7.6. Arrow indicates change of eluting buffer. Bottom: rechro-matography of peak I (left) and peak II (right) on DEAEcellulose, pH 8.0. Arrowindicates change of eluting buffer.

II. In some experiments, however, the reverse

occurred. Rechromatography of each of the peaksrevealed peak I to be quite pure. However, peakII frequently contained small amounts of contami-nating protein and was therefore rechromato-graphed in a number of experiments before furtherseparation was undertaken. While this step ap-

peared to have resulted in appreciable resolution,immunological and chemical studies to be describedbelow indicated that each fraction was still hetero-geneous. Each fraction was therefore chromato-graphed further on a DEAEcellulose columin atpH 8.0, 0.01 to 0.3 u. By gradient elution eachfraction could be further resolved into a majorand one miiinor peak (IA and B) (IIC and D).Since in each instance only a single peak was re-

covered from the DEAE cellulose column afterthe starting buffer, subsequent experiments were

again performed by stepwise elution with similarresults. The peaks obtained in one such experi-ment are shown in Figure 2 (bottom). Rechro-matography of each peak under identical conditionsrevealed only a single peak in the same position,thus indicating that these represent different mo-

lecular species and not simply artifacts createdduring the fractionation procedure. The relativeamounts of each fraction were variable. In gen-

eral, fraction C was the largest single componentand made up from 40 to 50 per cent of the totalprotein, while fraction A made up from 15 to 25per cent of the total protein recovered. Aboutone-quarter to one-third was recovered with frac-tion B. Fraction D rarely contained more than 5

1 2

FIG. 3. ULTRACENTRIFUGALPATTERNS. 1: Fraction A(top) and fraction B (bottom). 2: Fraction D (top)and fraction C (bottom). Speed 42,040 rpm. Photo-graphs taken after 64 minutes. Sedimentation proceedsfrom left to right.

I I 1111 I

1935

EDWARDC. FRANKLIN

7S

FIG. 4. IMMUNOELECTROPHORETICPATTERNS COMPAR-ING 7S y-GLOBULIN ON THE BOTTOMI OF EACH PATTERN

WITH: 1) WHOLEPAPAIN DIGEST; 2) FRACTION A; 3)FRACTION B, 4) FRACTION C; AND 5) A SECONDPREPARA-

TION OF FRACTION B. Anode is to the right.

per cent of the total. Total recovery of proteinranged from 85 to 95 per cent of the materialapplied.

Examination in the analytical ultracentrifugedemonstrated A, B and C to sediment as singlerather homogeneous peaks with an observed s rateof 3.4 to 3.5S (Figure 3). D was heterogeneousand consisted of approximately equal amounts oftwo peaks with s rates of 3.5S and 5 to 6S (Figure3). Molecular weights calculated by the Archi-bald technique were obtained for A, B and C and

ranged from 40,000 to 55,000 for each. Fraction D

waS too heterogcncous for suiclh analyses and wasniot exaiiinied furtlher.

In view of the well recognized heterogeneity ofy-glohlulins and the lack of miiolar equiivalence be-tween these fractions, it wvas of interest to deter-mine if they were derived from different fractionsof y-globulin. Chromatography of whole Cohnfraction II y-globlulin on CMI cellulose at pH 7.6results in two major fractions. One of these iseluted with the starting buffer, is slightly lessbasic, and has a slightly mnore rapid electropho-retic mobility than the second one, which is elutedat higher salt concentration. When each of thesefractions was separately digested with papain andchromiiatographed in a manner similar to that usedfor whole y-globulin, both fractions yielded ap-proximately equal amounts of fraction B. How-ever, they differed in the relative amounts of thetwo other peaks. The fraction eluted first yieldedlarge amounts of fraction A an(d little C, while thesecon(l one yielded larger amounts of fraction Cand only small amounts of A. Small amounts offraction D were obtained from both. Similarlywhen y-globulin was fractionated into a slow mov-ing and rapidly moving fraction by starch zoneelectrophoresis prior to digestion, relatively largeramounts of fraction C were isolated from theformer, while the more rapidly migrating proteincontained predominantly fraction A. The relativeamount of fraction B was approximately the samein both.

Antigenic properties. Figure 4 (top) comparesthe immunoelectrophoretic patterns obtained with7S y-globulin and the whole papain digest whenthey react with rabbit antiserum to 7S y-globulin.While the native proteins gave only a single line,the papain digests formed two major lines whichappeared to be independent of each other. Oneof these migrated more rapidly and the other moreslowly than y-globulin. Occasionally, a third linewas also visible near the point of application. Thefour lower patterns in Figure 4 show similar stud-ies with each of the separated fragments. Frag-ment C corresponds to the maj or slow line of thewhole digest, B to the major fast one, and A hasan intermediate mobility but could not always bedetected in the whole digest.

In a number of experiments fraction B had a

second line near the origin (Figure 4, bottom).Fraction D was heterogeneous and contained each

1936

STRUCTURALUNITS OF GAMMAGLOBULIN

of the lines present in the whole digest and wastherefore not studied further. While fractions Aand C differed from each other in mobility, nodistinct differences could be detected between themby double diffusion according to the method ofOuchterlony. Figure 5 compares the reaction ofeach fraction and 7S y-globulin with a rabbit anti-serum to 7S y-globulin. It can be seen that thesingle major lines produced by fractions A and Cfused with each other to give a reaction of identity(Figure 5B) but were antigenically deficient whencompared to the native protein (Figure 5A). Oc-casionally a second faint outer line was visible(Figure 5A) which probably represented tracesof contaminating material. Fraction B containedtwo major components which together appearedto have all of the constituents of the native pro-tein (Figure 5A). Figure 5B demonstrates thatone of the lines in fraction B is antigenically dis-tinct from A and C and crossed through them,while the second one appeared to be similar tothem. Both lines in fraction B appear to be of the3.5S class and not due to undegraded 7S -y-globu-

140-

110

- A

wo A70

1d50

B

FIG. 5. OUCHTERLONYDOUBLEDIFFUSION PLATES COM-PARING THE FRAGMENTS: A) WITH 7S 'y-GLOBULIN; B)WVITH EACH OTHER, USING AN ANTISERUMTO 7S 'y-GLOBU-LIN IN THE CENTER WVELL.

lin, since they were recovered in the top tlhreefractions obtained by density gradient ultracentri-fugation. Only a small portion of 7S y-globulin,simultaneously separated in another tube, wasfound at this level.

Differences between these fractions and native7S -y-globulin are also demonstrated by precipitinanalyses shown in Figure 6. None of the frag-

0 20 40 60 80 100 120 140ANTIGEN (pg N)

FIG. 6. PRECIPITIN ANALYSES COMPARINGTHE REACTIONS OF FRACTION A(0), B (A), C (*) AND 7S y-GLOBULIN (A) WITH AN ANTISERUM TO 7S-y-GLOBULIN.

1937

EDWARDC. FRANKLIN

TABLE I

The effect of the papain-digested fragments from antibodiesto horse serum proteins on the precipitation of horse

serum by the untreated antiserum

Horse Humanserum- albumin-human rabbit

Fraction antiserum antiserum

mng ,g ppt ,Ag pptNonie 88.0 475Papaini digest 1.28 14.0 480A 1.25 42.5 475C 1.36 10.5 4751) 0.50 57.0 *

* Not determiiined.menlts l)recipitated as muclh antibody as did theoriginal antigen. Superinatant analyses with 7Sy-globulins in antigen excess revealedI completeneutralization of antibody by 7S y-globulin, butresidual activity against 7S y-globulin remainedin the region of antigen excess of fractions A andC.

Preliminary studies wxith antisera to the wholepapain digest offered less resolution and little ad-ditional information.

Antibody analyses. Antibody activity of threehuman antibodies was tested following breakdown.Only one, an antiserum to horse serum, was aprecipitating antibody under the conditions of thestudy. Following breakdown to 3.5S fractionsthis antibody no longer precipitated with antigen,nor did any of the separated fragments. How-ever, the whole papain (ligest inhibited the pre-cipitation of the original antiserum and of thehomologous antigen almost completely (Table I).When each of the isolated fractions was tested,fraction C had most of the inhibitory activity,while A was less active at the same concentration.This particular experiment yielded amounts of

TABLE II

Antibody activity of papain-digested fragments prepared fromdiphtheria antitoxin and antistreptokinase

Diphtheria Antistrepto-Fraction antitoxin kinase

U/mg U/mg

ey-Globulin 10 2.7Papain digest 10 2.8A 2.7

I 1.3B J 0.2C 8.2DI 6.7

D )0

fraction B too small for assay. Table I also indi-cates that none of the fractions had any effect ona heterologous system of human albumin and arabbit antiserum. Similar results to be reportedin detail separately were obtained by studying theinhibition by these fractions of passive cutaneousanaphylaxis, employing the antiserum and horseserum (24). Here, too, fraction C and the wholepapain digest were active. Fraction D did not in-hibit, while fraction A had a nonspecific hemor-rhagic and inhibitory effect on the homologous aswell as on the heterologous antigen-antibodyreaction.

Activitv of the fractions produced from the othertwo antibodies was measured by their ability toneutralize toxin. In the case of diphtheria anti-toxin, the rabbit skin assay was employed (22);for antibody to streptokinase the kinase inhibition

TABLE III

Carbohydrate composition of papain-digested fragments ofCohn fraction II -y-globulin

Per cenit Per centlhexose lhexosamnine

-y-Globulin. ... 1.2 1.0Papain (ligest. 1.2 1.1

Fractioni Prel). I Prel). 2 Prel). 1 Prel). 2

A 1.0 1.4 0.7 0.6B 3.0 3.1 1.7 2.0C 0.2 0.3 0.3 0.4

was miieastured (23). Only 15 mg of antidiph-theria y-globulin was available and fractionationiwas not carried out beyond the first step of chro-matography on CMcellulose. The restults of thesestudies are listed in Table II. In each case the pa-pain digest retained virtually all of the originalactivity. Following fractionation the bulk of ac-tivity was associated with the second peak elutedfrom the cation exchanger (peak II) in the caseof diphtheria antitoxin, and with fraction C forstreptokinase. In each instance, as in the case ofthe antibody to horse serum, about 15 to 25 percent of the total activity was also associated withfractions I and A, respectively." The significanceof this will be discussed below.

Carbohydrate analyses. In view of previous re-ports on the isolation of small glycopeptides from

6 Qualitatively similar results have recently been ob-tained with another precipitating antibody than that tothyroglobulin.

1938

STRUCTURALUNITS OF GAMMAGLOBULIN

y-globulin (25), it seemed of interest to determinethe carbohydrate content of fractions preparedfrom Cohn fraction II. Table III indicates thatdigestion did not result in significant loss of car-bohydrate. Following chromatography, fractionII and subfractions C and D contained virtuallyno hexose or hexosamine; the bulk of the carbo-hydrate was associated with fraction I. Althoughresults were variable, probably because of thepresence of small degrees of impurities in eachfraction, fraction B contained the largest amountof carbohydrate and approximately 3 per cent hex-ose and 2 per cent hexosamine, while A had some-what lower values. The results of two represen-tative experiments are listed in Table III. Thefirst preparation represents fractions that had beenrechromatographed on DEAE cellulose prior toanalysis; the second preparation had been refrac-tionated on CMcellulose. Thus, more than 75per cent of the total hexose and hexosamine re-covered in this preparation was associated withfraction B.

DISCUSSION

Normal human 7S y-globulin has a molecularweight of approximately 160,000 and has beenshown to consist of at least two peptide chains onthe basis of careful end group analyses (26, 27).Detailed structural studies have been hamperednot only by the large size and complex structureof the y-globulin molecules, but also by the hetero-geneity of y-globulin (which should be calledy-globulins). Similar difficulties have also beenencountered in the study of antibodies produced inother species. Thus, the basic question, whetherantibodies to different antigens differ in theiramino acid composition and primary structure, orwhether antibody specificity is due to changes inthe secondary structure of preformed y-globulinmolecules resulting from contact with the anti-gen, has not been answered. Previous studies, es-pecially those of Kabat (28), have indicated, how-ever, that the antibody combining sites are small,and carbohydrate analyses by Rosevear and Smith(25) indicate that the carbohydrate is associated

with certain specific areas of the molecule. The re-cent observation of Porter (5) that rabbit y-globu-lin can be reproducibly broken down into easilyseparable fragments with molecular weights of50,000 to 80,000 suggested that similar studies of

human y-globulins and antibodies might providesome answers to these basic questions and fa-cilitate further structural analyses of the y-globulinmolecule.

The results of the present study indicate thathuman y-globulin can also be broken down intofragments having one-fourth to one-third the sizeof the original molecule. However, careful analy-ses of some of these units reveal certain differ-ences which suggest a structure for human y-glob-ulin which differs from that of rabbit y-globulin.The fragments so produced can be separated bychromatography into three major components dif-fering in chemical and electrophoretic propertiesand, to some extent, antigenically. The fragmentwith the most rapid electrophoretic mobility andan additional antigenic component did not con-tain significant antibody activity, but containedmore than three-quarters of the carbohydratepresent in the original molecule. Most of theantibody activity was associated with the frac-tion with the least negative charge on immuno-electrophoresis (fraction C) which contained littlecarbohydrate; however, significant antibody ac-tivity was also associated with the fragment thatwas antigenically closely related to the latter(fraction A) but contained more carbohydrate andhad a somewhat greater negative charge. Thepresence in y-globulins of an antigenically distinctF component and of several antigenically closelyrelated S and mid-components, which contributeto the electrophoretic mobility of the y-globulins,has also recently been described by Edelman,Heremans, Heremans and Kunkel (29) in a studyof the double line phenomenon. Studies withCohn fraction II y-globulin containing antibody tostreptokinase indicate that it consists of two majorfractions which can be separated by chromatog-raphy. These fractions differ somewhat in elec-trophoretic mobility and carbohydrate content, butresemble each other in having approximately simi-lar amounts of antibody activity to streptokinase.Both fractions contain a common non-antibodyfragment (fraction B) which is rich in carbohy-drate and makes up about one-fourth to one-thirdof the molecule. Some of their differences ap-pear to be due to the presence of larger amountsof the more basic fragment C in the y-globulinwith the least negative charge and slower electro-phoretic mobility, while greater amounts of the

1939

EDWARDC. FRANKLIN

more rapidly migrating fragment A were recov-ered following digestion of the fraction of nativey-globulin with the higher electrophoretic mobility.It would thus appear that the electrophoretic mo-bility of the native protein is partially determinedby the relative amounts of those fragments whichalso have the antibody combining sites.

The chemical analyses are consistent with previ-ous reports on the association of carbohydratewith certain low molecular weight peptides withmolecular weights of less than 5,000 (25). Ofparticular interest in these studies is the demon-stration of the existence of at least two distinctantigenic fragments. While comparison of thesewith native y-globulin suggests neither loss norrelease of antigenic groups present in the nativemolecule, preliminary studies, comparing themto fragments produced by treatment with pepsinand mercaptoethanol as described by Nisonoff andassociates (8), suggest the possible loss of a fewantigenic groups present in the native proteinfollowing digestion with papain.

The exact nature of the bonds holding the frag-ments together cannot be determined with cer-tainty from these studies. However, the observa-tions that a number of closely related smaller frag-ments can be prepared from rabbit and humany-globulins by sulfhydryl reagents combined withdenaturing agents (30), papain (4, 5) or pepsin(8) suggest that the common mechanism may bethe reduction of disulfide bonds by the sulfhydrylreagent, and that the enzyme or denaturing agentmay make these accessible by unfolding the mole-cule or by breaking off a fragment. While theclose similarity of the antigenic properties of frac-tions A and C when tested with rabbit antiseraraises the possibility that they are identical, sucha view does not appear likely, since they differ insome of their chemical properties, in electropho-retic mobility, and in antigenicity when tested withantisera produced in different species (31). Fur-thermore, the differences in the composition ofy-globulins fractiolnate(d by several techlliquesprior to (ligestion, suggest that y-globulins coIn-sist of a heterogeneous group of proteins differingin the relative amounts of these fractions. It isof interest that a small fraction (about 5 per cent)of the y-globulin resists breakdown to the 3.5Sunits.

The production of 3.5S units and of nonprecipi-

tating fragments is not only of theoretical interestin elucidating the structure of y-globulin, but mayeventually shed light on the synthesis and speci-ficity of antibody structure. In addition, suchnonprecipitating fragments with antibody activitymay prove of use in neutralizing certain noxiousantigens without the danger of producing tissuedamage caused by the interaction of antigen withantibody fixed to tissues. Ovary and Karush (21)have shown that antibody-containing fragmentsof rabbit y-globulins do not fix to the skin and,therefore, presumably not to other tissues. Simi-larly, none of the human fragments appears tofix to the skin as measured by reverse passivecutaneous anaphylaxis (31). This observationwould suggest that such univalent fragments maybe of use in neutralizing circulating noxious anti-gens without the production of tissue damage be-fore they can elicit an antibody response by thehost. Similarly, the role of these fragments inthe fixation of complement and other immunemechanisms, and the relationship of these arti-ficial fragments to those produced in vivo (32)remain to be explored.

SUMMARY

1. Breakdown of human 7S y-globulin withpapain and cysteine results in three major chro-matographically separable fractions with sedimen-tation coefficients of 3.4 to 3.5S, molecular weightsof 40,000 to 55,000, and small amounts of partiallydegrade(d y-globulin.

2. Each of these is antigenically related to thenative protein. Two of these (A and C) areclosely related to each other but differ in electro-phoretic mobility, while the third one (B) differsfrom them also antigenically.

3. Most of the carbohydrate is associated witlfraction B; more than three-fourths of the anti-body activity against three different antigens isfound in fraction C, with small amounts in frac-tion A.

4. The role of these structural units in deter-mining some of the known properties of y-globu-lins is discussed.

REFERENCES

1. Petermann, M. L., and Pappenheimer, A. M., Jr.The ultracentrifugal analysis of diphtheria pro-teins. J. phys. Chem. 1941, 45, 1.

1940

STRUCTURALUNITS OF GAMMAGLOBULIN

2. Porter, R. R. The formation of a specific inhibitorby hydrolysis of rabbit antiovalbumin. Biochem.J. 1950, 46, 479.

3. Grabar, P. Sur l'action de la pepsine sur les anticorpsantipneumococciques. C. R. Acad. Sci. (Paris)1938, 207, 807.

4. Porter, R. R. Separation and isolation of fractionsof rabbit gamma-globulin containing the antibodyand antigenic combining sites. Nature (Lond.)1958, 182, 670.

5. Porter, R. R. The hydrolysis of rabbit 'y-globulinand antibodies with crystalline papain. Biochem.J. 1959, 73, 119.

6. Karush, F. Properties of papain-digested purifiedantihapten antibody. Fed. Proc. 1959, 18, 577.

7. Nisonoff, A., and Woernley, D. L. Effect of hy-drolysis by papain on the combining sites of anantibody. Nature (Lond.) 1959, 183, 1325.

8. Nisonoff, A., Wissler, F. C., and Woernley, D. L.Mechanism of formation of univalent fragments ofrabbit antibody. Biochem. biophys. Res. Com.1959, 1, 318.

9. Putnam, F. W., Hsiao, S.-h., and Tan, M. Papaincleavage of human and rabbit y-globulins. Fed.Proc. 1960, 19, 77.

10. Franklin, E. C. Some properties of the breakdownproducts of human 7S gamma globulin and anti-bodies (abstract). J. clin. Invest. 1960, 39, 986.

11. Kunkel, H. G. Zone electrophoresis in Methods ofBiochemical Analysis, D. Glick, Ed. New York,Interscience Publishers, 1954, vol. 1, p. 141.

12. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., andRandall, R. S. Protein measurement with Folinphenol reagent. J. biol. Chem. 1951, 193, 265.

13. Trautman, R. Operating and comparating proce-dures facilitating Schlieren pattern analysis inanalytical ultracentrifugation. J. phys. Chem. 1956,60, 1211.

14. Kunkel, H. G., Franklin, E. C., and Muller-Eberhard,H. J. Studies on the isolation and characteriza-tion of the "rheumatoid factor." J. clin. Invest.1959, 38, 424.

15. Peterson, E. A., and Sober, H. A. Chromatographyof proteins. I. Cellulose ion-exchange adsorbents.J. Amer. chem. Soc. 1956, 78, 751.

16. Fahey, J. L., and Horbett, A. P. Human gammaglobulin fractionation on anion exchange cellulosecolumns. J. biol. Chem. 1959, 234, 2645.

17. Franklin, E. C., and Kunkel, H. G. Immunologicdifferences between the 19S and 7S components of

nornmal human '-globulin. J. Immunol. 1957, 78,11.

18. Scheidegger, J. J., and Buzzi, C. etude immuno-electrophoretique des gamma-globulines: Identifi-cation de plusieurs determinants antigeniques graceau proteines de Bence-Jones et aux serums demyelomes. Rev. franc. At. clin. biol. 1957, 2, 895.

19. Mokrasch, L. C. Analysis of hexose phosphates andsugar mixtures with the anthrone reagent. J. biol.Chem. 1954, 208, 55.

20. Elson, L. A., and Morgan, WV. T. J. A colorimetricmethod for the determination of glucosamine andchondrosamine. Biochem. J. 1933, 27, 1824.

21. Ovary, Z., and Karush, F. Personal communication.22. Fraser, D. T. The technic of a method for the

quantitative determination of diphtheria antitoxinby a skin test in rabbits. Trans. roy. Soc. Can.1931, 25, sect. V, 175.

23. Johnson, A. J., and Tillett, W. S. The lysis in rab-bits of intravascular blood clots by the streptococcalfibrinolytic system (streptokinase). J. exp. Med.1952, 95, 449.

24. Ovary, Z. Immediate reactions in the skin of ex-perimental animals provoked by antigen antibodyinteraction in Progress in Allergy, P. Kallos, Ed.Basel, Karger, 1958, p. 459.

25. Rosevear, J. W., and Smith, E. L. Structure of gly-copeptides from human y-globulin. J. Amer. chem.Soc. 1958, 80, 250.

26. Putnam, F. W. N-terminal groups of normal humangamma globulin and of myeloma proteins. J.Amer. chem. Soc. 1953, 75, 2785.

27. McFadden, M. L., and Smith, E. C. The free aminogroups of y-globulins of different species. J. Amer.chem. Soc. 1953, 75, 2784.

28. Kabat, E. A. The upper limit for the size of thehuman antidextran combining site. J. Immunol.1960, 84, 82.

29. Edelman, G. M., Heremans, J. F., Heremans, M. T.,and Kunkel, H. G. Immunological studies of hu-man '-globulin. Relation of the precipitin linesof whole y-globulin to those of the fragments pro-duced by papain. J. exp. Med. 1960, 112, 203.

30. Edelman, G. M. Dissociation of '-globulin. J. Amer.chem. Soc. 1959, 81, 3155.

31. Ovary, Z., and Franklin, E. C. To be published.32. Franklin, E. C. Physicochemical and immunologic

studies of gammaglobulins of normal human urine.J. clin. Invest. 1959, 38, 2159.

1941