INFECTION AND IMMUNITY, JUlY 1983, p. 205-213 Vol. 41, No. 10019-9567/83/070205-09$02.00/0Copyright C 1983, American Society for Microbiology

Association of Bacterial Carbohydrate-Specific ColdAgglutinin Antibody Production with Immunization by GroupC, Group B Type III, and Streptococcus pneumoniae Type

XIV Streptococcal VaccinesRICHARD G. COLLING,t THOMAS C. PEARSONJ AND JOHN CLIFFORD BROWN*

Department of Microbiology, The University of Kansas, Lawrence, Kansas 66045

Received 24 January 1983/Accepted 18 April 1983

Rabbits immunized with group B type III, group C, and Streptococcuspneumoniae type XIV streptococcal vaccines developed autoantibodies reactivewith autologous and isologous erythrocytes and human 0-positive erythrocytes atreduced temperatures. The cold agglutinin antibodies were present in both theimmunoglobulin M (IgM) and IgG fractions of group C streptococcal antiserumand in the IgM fraction of group B type III and S. pneumoniae type XIV antisera.BALB/c, CF1, and local strains of mice immunized with group B type III and S.pneumoniae type XIV streptococcal vaccines also produced a cold agglutininantibody reactive with rabbit and human erythrocytes. The cold agglutininantibodies were reactive with saccharide compounds representative of the deter-minants present on the individual bacterial carbohydrate structures, individualvaccine preparations, and isolated polysaccharides. The group C antibodies inrabbits were reactive with sugar ligands in the following order: N-acetylgalactos-amine > melibiose > lactose > galactose > glucose. Group B type III and S.pneumoniae type XIV cold agglutinin antibodies in rabbit antisera, however,displayed reactivities different from group C antibodies and from each other.Group B type III antibodies reacted with galactose > lactose > N-acetylgalactos-amine > glucose > rhamnose; S. pneumoniae type XIV antibodies reacted withlactose > melibiose > galactose > glucose > N-acetylgalactosamine. The samerelative ligand specificity was observed for the cold agglutinin antibodies in S.pneumoniae type XIV mouse antisera. The cold agglutinin antibodies in group Btype III and S. pneumoniae type XIV antiserum reacted with erythrocytes athigher temperatures (up to 31°C) than did group C antibodies (up to 14°C). Inaddition, S. pneumoniae type XIV antibodies did not discriminate between I- or i-bearing human erythrocytes to a significant extent. The results obtained providesubstantial evidence that autoreactive cold agglutinin antibodies produced byimmunization with these vaccines represent subpopulations of bacterial carbohy-drate-specific antibodies that cross-react with mammalian carbohydrate struc-tures.

The association of autoreactive erythrocyteantibodies with infectious disease (13) and ani-mal immunization (7, 12) has been of interest formany years. The majority of these antibodiesreact with human and rabbit erythrocytes attemperatures below 37°C and are consequentlyclassified under the general category of coldagglutinin antibodies. The characteristic eryth-rocyte reactivity of these antibodies correlateswith the reactivity of certain human monoclonal(usually Waldenstrom) immunoglobulins, and

t Present address: Department of Biology, Olivet NazareneCollege, Kankakee, IL 60901.

t Present address: University of Kansas Medical School,Kansas City, KS 66103.

therefore is thought to be associated with Iiregion blood group carbohydrate structures (15).The i antigenic determinants recognized by hu-man anti-i antibodies represent structures pres-ent on a linear oligosaccharide with repeatinggalactose (P-1-4) N-acetylglucosamine (P3-1-3)units; human anti-I antibodies react, dependingupon their individual specificity, with regions ofa branched oligosaccharide structure formed byaddition of galactose (P-1-4) N-acetylglucosa-mine (P-1-6) branches to the i-active region (4).Blood group-bearing glycoproteins, in additionto the core oligosaccharide structures compris-ing the Ii regions, may also present serine orthreonine residues (or both) substituted with N-

205

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

206 COLLING, PEARSON, AND BROWN

acetylgalactosamine (GalNAc) via 0-serine or0-threonine glycosidic linkages (or both).The appearance of these antibodies during

infection or immunization (or both) may be aresult of formation of an immunogenic erythro-cyte, a nonspecific polyclonal response due toinfection or immunization conditions, or a cross-reactive antibody population generated in directresponse to the infectious or immunizing agent.Recently, it was shown that rabbits hyperim-munized with group C streptococcal vaccineproduced cold agglutinin antibodies (6). Theapparent ligand specificity of these antibodies,GalNAc, strongly suggested a bacterial cell wallcarbohydrate structure stimulus which led toproduction of an erythrocyte cross-reactive anti-body population. Given the development of bac-terial vaccines for human use, the potential forimmunization to lead to autoreactivity of thistype is of obvious concern.The present study was designed to investigate

the relationship between the appearance oferythrocyte-autoreactive antibodies and the vac-cine used for immunization. The rationale in-volved utilized the carbohydrate structural char-acteristics of the individual vaccines andcorrelation of these determinants with the rela-tive sugar ligand specificity demonstrated forany cold agglutinin antibodies that appeared as aresult of immunization of both rabbits and mice.The following vaccines were used: group C, B,and Streptococcus pneumoniae type XIV strep-tococcal heat-killed cells. The results of thisstudy revealed that group B-, C-, and S. pneu-moniae type XIV-immune rabbits each pro-duced cold agglutinin antibodies of differentrelative ligand specificity. Further, mice immu-nized with S. pneumoniae type XIV vaccineproduced cold agglutinin antibodies of similarligand specificity as those produced in rabbits.The data support the contention that certainantibodies produced as a consequence of bacte-rial challenge can cross-react with mammalianerythrocyte determinants. Such reactivity there-fore could result in an autoimmune response.

MATERIALS AND METHODS

Preparation of vaccines and immunization protocol.Group C strain C-74 vaccine was prepared as previ-ously described (6). Group B type III (typing of thegroup B organisms was kindly performed by C. J.Baker, Baylor College of Medicine, Houston, Tex.)and S. pneumoniae type XIV vaccines were preparedby growing the bacteria in Todd-Hewitt broth at 37°Cfor 18 to 24 h. The cells were pelleted by centrifugationand washed three times with sterile saline. The cellswere suspended in 30 ml of sterile saline and heatkilled by incubation at 70°C for 30 min. After the killedvaccine was washed four times with sterile saline, thecells were suspended to an optical density of 11.5 at

INFECT. IMMUN.

450 nm and then frozen until use. New Zealand Whiterabbits were injected intravenously three times perweek over a 4- to 6-week period, rested for 2 to 3months, and injected as before. The vaccine dosagewas increased by 0.25-ml increments beginning with0.25 ml the first week to 1.0 ml the last week ofimmunization. Preinoculation and weekly bleedingswere taken via the central ear artery. The serum wasroutinely prepared at 37°C and tested immediately.The serum samples were stored frozen at -20°C untilfurther use. BALB/c, CF-1, and local outbred agouti-colored mice were immunized with S. pneumoniaetype XIV streptococcal vaccine intravenously via atail vein. The immunization protocol followed thatgiven the rabbits. The bleeding schedule followed thatof the rabbits and was performed via the retroorbitalplexus.

Antibacterial response. Vaccine antibody responseswere measured by vaccine agglutination as describedby Herd and Spragg (8).

Erythrocyte preparations. Rabbits were bled fromthe central ear artery into Alsever solution. Sheep andbovine (ox) erythrocytes were obtained from ColoradoSerum Co., Denver, Colo. Whole human 0-positiveblood was provided by Lawrence Memorial Hospitalas outdated preparations. In addition, a blood typingpanel was provided by Lawrence Memorial Hospital.Each erythrocyte sample was washed three times andsuspended to 1% (vol/vol) in 0.02 M phosphate-0.15 MNaCl buffer (pH 7.2).

Hemagglutination assays. Agglutination activity inrabbit immune sera against autologous and isologousrabbit erythrocytes was assayed by adding an equalvolume of erythrocytes (0.3%, vol/vol) to twofold50-Rl dilutions of antisera in microtiter plate wells andincubating the reaction mixtures at 4°C or 37°C (orboth). The cold agglutinin antibody titer was recordedas the reciprocal of the highest dilution of antiserumthat yielded a positive reaction. No significant differ-ence in agglutination was observed when immune serawere tested against various rabbit isologous erythro-cyte preparations. Heterologous human (types A, B,OI, and Oi) sheep, and bovine (ox) erythrocytes wereexamined in the presence of whole serum and purifiedcold agglutinin antibody, either as described above orin direct slide agglutination assay. Mouse immune serawere similarly tested against rabbit, human, bovine(ox), and mouse erythrocytes.Cold hemagglutination inhibition assays. Antiserum

preparations (25 RlI) were diluted twofold in V-bottommicrotiter plate wells. To these wells was added 50 ,ulof a 0.3% (vol/vol) suspension of rabbit erythrocytesand 25 ,ul of 0.02 M Tris HCl-0.15 M NaCl-0.1%(wt/vol) gelatin buffer, pH 8.0 (Tris-gelatin buffer).After incubation for 4 h at 4°C, hemagglutination wasassessed on a scale of 4+ (highest) to + (lowest). Theantiserum dilution which yielded a 2+ reaction wassubsequently used in inhibition assays. Individual car-bohydrate solutions were prepared at 2 mg/ml in Tris-gelatin buffer. A 25-Fl (representing 50 ,ug) sample ofcarbohydrate solution was diluted twofold, and 25 ,ulof the appropriately diluted antiserum was added toeach well. The microtiter plates were incubated 1 h atroom temperature, and 50 ,ul of a 0.3% (vol/vol)suspension of rabbit erythrocytes was added. Afterincubation at 4°C for 4 h, hemagglutination was mea-sured. The lowest amount of inhibitor that completely

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

VOL. 41, 1983

inhibited agglutination of the erythrocytes was record-ed.Cold hemagglutinin antibody purification. Antibod-

ies reactive with erythrocytes at reduced temperaturewere purified as previously described (6). Cold aggluti-nin antibodies were purified from group C and S.pneumoniae type XIV antisera; however, recovery ofthese antibodies from group B antisera was poor.Consequently, subsequent examination of antibodyactivity in group B antisera utilized the exclusionvolume (void volume fraction) of Sephacryl S-200fractionated whole group B antisera. Group C and S.pneumoniae type XIV antisera were prepared at 37°C.After the addition of antisera to an equal volume ofpacked rabbit erythrocytes at 370C, the suspensionwas incubated for 1 h at 4°C and centrifuged at 800 x g

for 15 min. After centrifugation, the supematant vol-ume was removed, and an equal volume of 0.02 Mphosphate-0.15 M NaCl buffer (pH 7.2) was added.The suspension was incubated at 37°C for 1 h and at0°C for 1 h and centrifuged at 4°C, and the supernatantvolume was discarded. These procedures were repeat-ed three times. The material that eluted after the final37°C incubation step was concentrated by ultrafiltra-tion to the original serum volume and fractionated on a2.4- by 100-cm Sephacryl S-200 column equilibratedand eluted in the presence of 0.02 M Tris-hydrochlo-ride-0.5 M NaCI-0.02% (wt/vol) sodium azide buffer(pH 8.0). Individual fractions were assayed for coldagglutinin antibody activity, pooled, and concentrat-ed. Pooled fractions were examined by sodium dode-cyl sulfate-agarose-polyacrylamide gel electrophoresisand by immuhodiffusion analysis against goat anti-rabbit ,u chain- and goat anti-rabbit -y chain-specificantisera. By these criteria, group C antisera yieldedpurified cold agglutinin antibody activity in both theimmunoglobulin M (IgM) and IgG immunoglobulinclasses. No IgG antibody was detected in the IgMfraction, and no lgM monomers were detected in theIgG fraction. Only IgM immunoglobulin and coldagglutinin antibody activity was detected in the materi-al purified from S. pneumoniae type XIV antisera.Attempts to purify these antibodies from group Bantisera were unsuccessful. In further studies, only thepurified IgM cold agglutinin antibodies, or the IgM-containing serum fractions were used unless otherwisenoted.Temperature dependence of erythrocyte reactivity.

The precise temperature requirements for cold aggluti-nin antibody interaction with rabbit erythrocytes wereevaluated by analyzing twofold serial dilutions ofwhole group C and group B type III and S. pneumo-niae type XIV streptococcal antisera for cold aggluti-nin antibody activity at 4°C. The last dilution of eachantiserum which gave a 4+ agglutination reaction wasused in subsequent assays. Samples (50 ,ul) of theantiserum dilution from each of the respective antiserawere placed in duplicate microtiter wells in sevendifferent microtiter plates. After the addition of 50 ,ulof 1% (vol/vol) rabbit erythrocytes to each well, oneplate was incubated at each of the following tempera-tures; 5, 10, 15, 20, 25, 30, and 37°C for 90 min. Thewells were scored for their agglutination reaction from0 to 4+.

Hemolysis assays. Cold agglutinin antibody-depen-dent, complement-mediated hemolysis was examinedin a biphasic temperature hemolysis assay similar to

COLD AGGLUTININ ANTIBODIES 207

that of Lau and Rosse (11), with purified antibodiesand whole or fractionated antisera. Before the assay,all purified antibody preparations were equilibrated bydialysis against 0.02 M Tris-hydrochloride-0.128 MNaCl-0.5 mM Mg2+-0.15 mM Ca2+-0.1% (wt/vol)bovine serum albumin (p. = 0.15, pH 7.4). The materialwas diluted twofold, and excess guinea pig comple-ment was added, followed by the addition of rabbiterythrocytes to a final 0.25% (vol/vol) suspension.Solutions were incubated for 1 h at 0°C and 1 h at 37°C.Fourfold the lowest amount of test agent required for100%o hemolysis was subsequently determined andused for erythrocyte sensitization in complement titra-tion assays. Guinea pig complement was titrated underthese conditions to provide a 50%o lysis endpoint.From these data, the amount of complement requiredto provide 4 50%o lysis endpoint units was calculated.Subsequently, purified antibody or whole or fraction-ated antisera were diluted twofold, 4 50% lysis end-point complement units was added in the presence of a0.25% (vol/vol) suspension of rabbit erythrocytes, anda 50%o lysis endpoint was determined. In hemolyticinhibition assays, the amount of antibody or whole orfractionated antisera that yielded 50o hemolysis wasused for erythrocyte sensitization. The extent of he-molysis was determined by absorbance measurementsat 413 nm of supernatant volumes.

Hemolysis inhibition studies. Solutions of individualsaccharide compounds were prepared at initial con-centrations at 50 mM and diluted twofold. Carbohy-drate solutions were prepared at initial concentrationsof 100 ,ug/ml; these solutions were diluted twofold toapproximately 1.0 ng/ml. To inhibitor solutions wereadded an amount of purified antibody or whole orfractionated antiserum, complement, and erythrocytesto a final volume of 1.5 ml. After incubation at 0°C for1 h and 37°C for 1 h, the extent of hemolysis wasdetermined. Control solutions included those preparedin the absence of inhibitor, the absence of comple-ment, and the absence of antibody or whole or frac-tionated antiserum. In addition to these controls, themicrocomplement fixation test (14), which utilizedeither sheep or bovine (ox) erythrocytes, was used todetermine the extent of anti-complementarity associat-ed with solutions of inhibitor or antibody.

Radioactively labeled antibody studits. To assess thespecificity of purified cold agglutinin antibodies isolat-ed from S. pneumoniae type XIV antisera, theseantibodies were trace labeled by using the Enzymo-bead reagent (Bio-Rad Laboratories, Richmond, Cal-if.) to a specific activity of 100 cpm/ng. These antibod-ies were subsequently reacted with S. pneumoniaetype XIV vaccine to assess specific binding and groupC and micrococcal vaccines to assess nonspecificreactions. Approximately 110 ng of 1251I-labeled anti-body was added to bovine serum albumin-coated tubescontaining 200 p.1 of a 1:10 dilution of vaccine (absor-bance at 450 nm, 1.15). After incubation for 2 h atroom temperature, the suspensions were centrifuged,the supernatant volume was discarded, and the pelletwas suspended and washed once with 0.02 M phos-phate-0.15 M NaCl buffer made 2% (wt/vol) in bovineserum albumin (pH 7.2). After centrifugation, thesupernatant volume was discarded, and the pellet wascounted in a Packard crystal scintillation gammacounter. Binding inhibition studies were performed inthe presence of individual solutions containing 50 mM

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

208 COLLING, PEARSON, AND BROWN

lactose or glucose. Antibody labeled at higher specificactivities, i.e., 3,000 and 6,000 cpm/ng, were inactivewith respect to agglutination of erythrocytes at 4°C.

RESULTS

Cold agglutinin antibody response and erythro-cyte reactivity. Shown in Fig. 1 are the polysac-charide structures associated with group B typeIII, group C, and S. pneumoniae type XIVstreptococci. Earlier experiments that utilizedanti-group C streptococcal serum had revealedcold agglutinin antibodies that were preferential-ly reactive with GaINAc (2, 6). These datasuggested the autoreactive antibodies observedwere, in fact, bacterial carbohydrate specific,but cross-reactive with erythrocyte determi-nants. To address this hypothesis, rabbits wereindividually immunized in the present study withvaccine prepared from group C, group B typeIII, and S. pneumoniae type XIV streptococci.Sera were examined at 4°C for the ability toagglutinate rabbit erythrocytes. All rabbits im-munized produced vaccine and erythrocyte-re-active antibodies during week 1 of immuniza-tion. The vaccine agglutination titers rose duringweek 1 and plateaued at week 4. The highestvaccine titers attained were 1:1,048; 1:512; and1:4,0% for group B, S. pneumoniae type XIV,and group C antisera, respectively. The hemag-glutination titer of these antibodies in group Cand S. pneumoniae type XIV antisera rose dur-ing week 2, usually plateaued during week 3, andthereafter remained constant. Group B cold ag-glutinin antibody titers did not rise after week 1and in some instances declined. Among therabbits examined, peak cold agglutinin antibodytiters varied from 1:128 to 1:256 in group C-immune animals, 1:80 to 1:160 in S. pneumoniaetype XIV-immune animals, and 1:20 in group B-immune animals. All preinoculation sera werenegative with respect to cold agglutinin anti-body, but in some instances showed a 1:20 to1:40 group C vaccine titer.Cold agglutinin antibody activity could be

detected in only the IgM fractions of wholegroup C antisera fractionated by S-200 Sepha-cryl gel chromatography. Further, only the voidvolume fractions of group B and S. pneumoniaetype XIV antisera similarly fractionated re-vealed cold agglutinin antibody activity. Coldagglutinin antibodies purified from group C andS. pneumoniae type XIV antisera by elutionfrom rabbit erythrocytes were subsequently sub-jected to fractionation by S-200 Sephacryl gelchromatography, and 4-ml fractions were indi-vidually assayed for activity. Only the IgM peakof erythrocyte-purified S. pneumoniae type XIVmaterial (as judged with reaction with goat anti-rabbit ,u chain-specific antisera) and the exclu-

INFECT. IMMUN.

Group B Type III

-4)D -D-GIc-(1-6)-f-D-GlcNAc-(1- 3)-p-D-Gal-(l

a-D-NeuNAc-(2-6)*D-GaI-(1 -4)

Type XI V

+ 6)-O-D-GlcNAc-Q1- 3)-f-D-GaI -(-4)-D-D-Glc-(l+

frD-Gal-(1-4)

Group C

+3)-a -D-Rham-(l-2)-a -D-Rham-(1-3)-a -D-Rham-(1+

a-D-GaINAc-(l-3)

a-D-GaINAc-(I-3)FIG. 1. Structures of polysaccharides assigned to

group B type III, S. pneumoniae type XIV, and groupC streptococci.

sion volume of group B antisera contained coldagglutinin antibody activity. In contrast, bothIgM and IgG fractions of erythrocyte-purifiedcold agglutinin antibodies isolated from group Cantisera were active. Further examination ofthese antisera utilized either purified IgM coldagglutinin antibody or the exclusion volume ofS-200 Sephacryl chromatographed antisera.Cold agglutinin erythrocyte reactivity. Reactiv-

ity with erythrocytes was further examined us-ing rabbit, bovine (ox), human type 01, humantype Oi, and human type A, human type B, andsheep erythrocytes. Purified IgM cold agglutininantibody preparations isolated from group C andS. pneumoniae type XIV antisera reacted withrabbit erythrocytes at 4°C, and these antibodiesalso reacted with human type 0 erythrocytes at4°C. Furthermore, when tested by direct slideagglutination assay, group C IgM cold agglutininantibodies strongly agglutinated sheep and hu-man type A erythrocytes at 37°C. In addition,both group B fractionated serum and purified S.pneumoniae type XIV IgM cold agglutinin anti-bodies strongly agglutinated human type Berythrocytes at 37°C. In contrast, when eitherwhole serum or purified material was examined,neither preparation agglutinated human 0 orbovine (ox) erythrocytes at 37°C nor bovine (ox)erythrocytes at 4°C. S. pneumoniae type XIVwhole antisera and purified IgM antibodies werefurther examined for the ability to agglutinatehuman type 01 and human type Oi erythrocytes.Essentially no difference in reactivity was de-

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

COLD AGGLUTININ ANTIBODIES 209

tected, since a titer of 1:80 was observed whenantisera were tested against human type 01erythrocytes and a titer of 1:160 was observedwhen S. pneumoniae type XIV antisera weretested against human type Oi erythrocytes.When the 19S fraction of group B, group C,

and S. pneumoniae type XIV antisera weretested in the biphasic temperature hemolyticassay, all were capable of initiating hemolysis ofrabbit erythrocytes. Further, the antibodieswere efficient hemolysins, since usually lessthan 2 ,ug of purified material was active inproducing 50% lysis when IgM cold agglutininantibodies purified from group C and S. pneu-moniae type XIV antisera were examined.Temperature requirements of erythrocyte-cold

agglutinin antibody interaction. During thecourse of the present investigation, group B andS. pneumoniae type XIV rabbit antisera wereobserved to cause significant agglutination ofrabbit erythrocytes at room temperature, where-as most group C antisera failed to agglutinateerythrocytes until much lower temperatureswere achieved. To evaluate the precise tempera-ture requirements of cold agglutinin antibody-erythrocyte interaction, group C, group B, andS. pneumoniae type XIV whole antisera werecompared for their ability to effect agglutinationof rabbit erythrocytes at various temperatures.When a 2+ agglutination reaction was used as anarbitrary reference point, group C antisera yield-ed such a reaction at 14°C and below, whereasboth group B and S. pneumoniae type XIVantisera were reactive at 31°C and below.

Vaccine and carbohydrate reactivity. The par-allel appearance of cold agglutinin antibodyactivity and bacterial antibody activity duringimmunization with each of the vaccine prepara-tions suggested an association between the anti-body population observed and the vaccines thatelicited the response. Absorption studies wereperformed in an attempt to differentiate betweenerythrocyte-reactive and bacteria-reactive anti-body. Similar to data obtained in previous stud-ies (6), when the void volume fraction ofgroup Cantisera was absorbed only once with group Cvaccine, cold agglutinin antibody titers droppedfrom 1:256 to 1:16. When purified group C IgMcold agglutinin antibody (at approximately 1mg/ml) was absorbed once with group C vac-cine, the hemagglutination titer decreased from1:5,000 to 1:40. Absorption of the void volumefraction of group B and S. pneumoniae type XIVantisera with their respective vaccines yieldedsimilar results. The cold agglutinin antibody titerdecreased from 1:320 to 1:32 and 1:640 to 1:8,respectively. The relationship between S. pneu-moniae type XIV cold agglutinin antibodies,specificity for individual vaccines, and the dis-similarity between these antibodies and the anti-

lOOr

90

- 60c

c 402

0)a-

201

0 1 2 3 4log WM Inhibitor

FIG. 2. Effect of lactose (O), melibiose (U), galac-tose (0), glucose (0), and GalNAc (A) on S. pneumo-niae type XIV rabbit cold agglutinin antibody-sensi-tized rabbit erythrocyte hemolysis.

bodies elicited in group C antisera was furtherexamined using purified 125I-labeled antibodies.When approximately 110 ng of antibody wasadded to S. pneumoniae type XIV, group C, andmicrococcal vaccines, 97, 2, and 2 ng of anti-body, respectively, were bound. When bindingwas examined in the presence of 50 mM lactoseand 50 mM glucose, binding to S. pneumoniaetype XIV vaccine was reduced by 63% in thepresence of lactose and uninhibited in the pres-ence of glucose. Further, when the cold hemag-glutinin activity present in whole serum wasinhibited in the presence of purified carbohy-drate preparations, rabbit erythrocyte agglu-tination in the presence of group B antiserumwas completely inhibited in the presence ofapproximately 750 ng of group B carbohydrate;erythrocyte agglutination in the presence of S.pneumoniae type XIV antiserum was complete-ly inhibited in the presence of 300 ng of S.pneumoniae type XIV polysaccharide. Thesedata clearly revealed specific cold agglutininantibody reactivity associated with specific bac-terial vaccine reactivity.

Ligand reactivity. To more thoroughly investi-gate the ligand binding characteristics of thesevarious cold agglutinin antibody preparations,biphasic temperature hemolysis inhibition stud-ies were initiated. Shown in Fig. 2 are datarepresentative of purified antibody isolated fromS. pneumoniae type XIV rabbit antisera. Asshown, lactose was approximately fivefold morereactive than galactose and manyfold more reac-tive than GalNAc. In contrast (Table 1), coldagglutinin antibody purified from group C im-mune sera was preferentially inhibited in thepresence of GaINAc. Further, the data in Table1 demonstrate that galactose was significantly

VOL. 41, 1983

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

210 COLLING, PEARSON, AND BROWN

TABLE 1. Inhibition of rabbit cold agglutinin antibody hemolytic activity in biphasic temperature hemolyticinhibition assays

Amt of saccharide or polysaccharide required for 50% inhibitionCold agglutinin Saccharide (mM) Polysaccharide (ng)antibody source __________________________

Lactose Melibiose Galactose GaINAc Glucose Type XIV Group B Group C

Group C 12 3 25 1 >50 7.5Group Ba 8 2 13 >50 50Type XIV 0.04 0.2 1.3 16 6.3 15

a Sephacryl S-200 void volume fraction.

more inhibitory of cold agglutinin antibody he-molytic activity in the void volume fraction ofgroup B immune sera when compared to lactoseand GaINAc. These data revealed a clear differ-ence among the cold agglutinin antibodies' li-gand reactivity in association with the differentantipolysaccharide responses in the individuallyimmunized animals. Also, group B cold aggluti-nin antibody activity was inhibited less than 50%by 20 mM rhamnose. In addition, individualhemolytic responses were inhibited in the pres-ence of nanogram quantities of purified polysac-charides (Table 1). To further examine the asso-ciation of cold agglutinin antibody ligandspecificity with the antipolysaccharide immunergsponse, mice were immunized with S. pneu-moniae type XIV vaccine.The data in Fig. 3 represent saccharide inhibi-

tion studies of cold agglutinin antibody hemolyt-ic activity (toward rabbit erythrocytes) presentin the void volume Sephacryl S-200 fractions ofpooled BALB/c S. pneumoniae type XIV anti-sera. As shown, the relative ligand specificityobserved was identical to the data obtainedwhen rabbit S. pneumoniae type XIV antisera orpurified material was examined. As was ob-served when rabbit S. pneumoniae type XIVantisera were tested, lactose was a substantiallymore effective inhibitor of hemolysis when com-pared with galactose or GaINAc. From the datashown in Tables 2 and 3, the strain of mouseexamined did not alter the relative ligand inhibi-tion observed. However, when S. pneumoniaetype XIV polysaccharide was used as antigen incomplete Freund adjuvant, the preferred ligandwas melibiose. Mice immunized with completeFreund adjuvant only did produce a very weakcold agglutinin antibody response inhibitable bymelibiose. Therefore, it is possible that the pres-ence of adjuvant containing mycobacteria par-ticipated in the response. Importantly, however,cold agglutinin reactivity with GalNAc remainedextremely low (Table 2). Interestingly, BALB/cand CF1 group B antisera cold agglutinin anti-body reacted with rabbitverythrocytes at roomtemperature and, in some cases, at 37°C, al-though hemagglutination titers were significant-

ly lower than those observed at 4°C (data notshown).

DISCUSSION

Cold agglutinin antibodies of the IgM classhave been observed in the sera of rabbits immu-nized with Mycoplasma pneumoniae, Listeriamonocytogenes and Streptococcus MG vaccines(7, 12). These antibodies could be effectivelyremoved from sera by absorption with the vac-cine used for immunization. Further, these anti-bodies were reactive with saccharide ligandsthat contained galactoside moieties. It was diffi-cult in these studies, however, to determinewhether the autoreactive antibodies were eryth-rocyte specific, but cross-reactive, with the bac-teria, or vice versa. The present investigationapproached this problem by utilizing the recog-nized differences and similarities among the car-bohydrate structures associated with three dif-ferent microorganisms.The carbohydrate structures associated with

group C, group B type III, and S. pneumoniaetype XIV streptococcal bacteria are known.Consequently, the reactivity of cold agglutininantibodies could be more easily defined. Where-

80-

60

.40-

U 0~~~~'20

2 3 4log WM Inhibitor

FIG. 3. Effect of lactose (O), melibiose (-), galac-tose (0), GalNAc (0), and glucose (A) on S. pneumo-niae type XIV BALB/c mouse cold agglutinin anti-body-sensitized rabbit erythrocyte hemolysis.

INFECT. IMMUN.

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

COLD AGGLUTININ ANTIBODIES 211

TABLE 2. Saccharide inhibition of BALB/c cold agglutinin antibody activity in biphasic hemolytic inhibitionassay'

Amt of saccharide required for 50%o inhibition (mM)Immunizing agent Lactose Melibiose Galactose GaINAc Glu

Type XIV polysaccharide 1.1 0.6 1.9 40% at 10 mM None at 10 mMType XIV vaccine 1.7 3.5 5.3 32b 50b

a Antisera were from mice immunized with type XIVbExtrapolated to obtain 50% endpoint.

as the immunodominant structural features ofthe group C carbohydrate antigen are the a-1, 3-linked GalNAc disaccharide residues, bothgroup B type III and S. pneumoniae type XIVdeterminants are devoid of GalNAc determi-nants. Further, S. pneumoniae type XIV andgroup B type III carbohydrate antigens are es-sentially identical except for a terminal N-ace-tylneuraminic acid residue associated with thecore structure of the group B type III polysac-charide (10).

In the biphasic temperature hemolysis inhibi-tion assay, GalNAc was the most effective inhib-itor of group C cold agglutinin antibody activity.These data substantiated previously obtaineddata on this particular system (6). In contrast,the cold agglutinin antibodies induced by rabbitimmunization with either group B type III or S.pneumoniae type XIV vaccines were most reac-tive with galactose or lactose, respectively. Fur-thermore, the 50% inhibition endpoints obtainedin the presence of several other mono- anddisaccharides clearly distinguished group C coldagglutinin antibodies from both group B and typeIII and S. pneumoniae type XIV streptococcalcold agglutinin antibodies. In addition, the rela-tive inhibition profiles observed for group B typeIII and S. pneumoniae type XIV cold agglutininantibodies were similar. It is clear, therefore,that immunization with bacteria per se does notresult in the appearance of "nonspecific" coldagglutinin antibodies. Indeed, these antibodiesare unique and reflect ligand specificities whichare predicted on the basis of bacterial carbohy-drate determinants.Baker and Kasper (1) have recently shown

that the terminal sialic acid moiety of the groupB type III carbohydrate antigen is exceedinglyacid labile and is readily hydrolyzed duringculture of the microorganisms in Todd-Hewitt

polysaccharide and vaccine.

broth, due to accumulation of acid metabolites.The pH of the cultures in the present investiga-tion after 24 h of incubation had dropped to 5.0.Thus, the culture conditions employed very like-ly resulted in partial desialation of the type IIIcarbohydrate antigen. Such a loss of sialic acidwould render the type III antigen structurallyidentical to the S. pneumoniae type XIV anti-gen. Consequently, the similarity between coldagglutinin antibody ligand reactivity in bothgroup B type III and S. pneumoniae type XIVantisera would predictably be similar if the coldagglutinin antibodies produced had, indeed, aris-en in direct response to the bacterial carbohy-drate structures. By similar reasoning, the cleardifferences in the ligand specificity of group Ccold agglutinin antibodies in comparison to theantibodies generated by immunization with theother two vaccines reflects a specificity of theseantibodies for the carbohydrate structures asso-ciated with the bacterial antigens.The cold agglutinin antibodies in group B

rabbit antisera reactive with galactose and lac-tose probably represent a type III carbohydrate-specific subpopulation. There is substantial evi-dence that group B specific antibodies arepredominantly reactive with rhamnose determi-nants (3). Although this particular sugar in thepresent study was inhibitory of the group B typeIII cold agglutinin antibodies, a high (approxi-mately 20 mM) concentration was required toinhibit at the 50% level. It has also been demon-strated that galactose was an effective inhibitorof purified type III polysaccharide interactionwith type-specific antisera, although the major-ity of the reactivity with the polysaccharideappeared directed toward sialic acid determi-nants (3). Thus, although it appears that galacto-syl-containing structures are involved in the coldagglutinin activities observed in the current

TABLE 3. Saccharide inhibition of cold agglutinin antibody activity in biphasic temperature hemolytic assayaAmt of saccharide required for 50o inhibition (mM)

Mouse strain Lactose Melibiose Galactose GaINAc Glucose

Local 0.8 5.6 2.2 20% at 10 mM 20% at 10 mMCF1 0.3 0.6 1.4 20% at 10. mM 20% at 10 mM

a Antisera were from randomly bred, agouti-colored mice and CF1 mice immunized with type XIV vaccine.

VOL. 41, 1983

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

212 COLLING, PEARSON, AND BROWN

study, there is not an absolute requirement thatthe antibodies be strictly galactose specific toobserve erythrocyte reactivity. When mice wereimmunized in the present study, it was clear thatthe cold agglutinin antibodies that appearedwere strikingly similar to those produced inrabbits. It appears highly unlikely that threedifferent strains of mice would produce antibod-ies in a nonspecific manner that reflected suchsimilar carbohydrate specificity and, in turn,reflected the same characteristics as the coldagglutinin antibodies produced in rabbits. Inter-estingly, when the mice were immunized withgroup C streptococcal vaccine, no cold aggluti-nin antibodies were detected. A recent study byColigan et al. (5) provided data that clearly pointto the unlikely possibility of an immune re-sponse in mice to group C streptococcal carbo-hydrate GalNAc determinants, since such deter-minants are identical to the glycan region of theForssman glycolipid, which mice express. Thus,hyperimmunization per se and other nonspecificevents cannot easily explain the appearance ofthese autoreactive antibodies. When anti-S.pneumoniae type XIV rabbit and mouse antiserawere compared, it was interesting to note thatthe inhibition of rabbit cold agglutinin antibodyisolated from pooled antisera exhibited a bipha-sic inhibition response in the presence of lac-tose. Therefore, there appeared to be two rela-tively distinct lactose-reactive rabbit coldagglutinin antibody populations, one of loweraffinity and one of higher affinity. This type ofinhibition was not observed when mouse anti-sera were examined.The association of cold agglutinin antibodies

with overt pathological symptoms is dependentupon the temperature required for erythrocyte-antibody interaction (13). Even the amount ofsuch antibodies can become a secondary consid-eration when antibody binding to erythrocytes atnear-physiological temperatures is a characteris-tic of the response. In the present study, bothgroup B type III and S. pneumoniae type XIVcold agglutinin antibodies reacted with erythro-cytes at significantly higher temperatures incomparison with the group C antibodies. It ispossible that the structural similarities betweenS. pneumoniae type XIV structures and Ii oligo-saccharides is the explanation for the relativelybetter agglutination activity exhibited by bothrabbit and mouse antisera for rabbit erythro-cytes. However, the ability of these antibodiesto react with erythrocytes at temperatures ashigh as 31°C could have been due to the degreeof exposure on the erythrocyte surface of deter-minants with which the antibodies were reac-tive, or a generally higher affinity for saccharidedeterminants may have been present.The relative ease with which cold agglutinin

antibodies were produced in the present studysuggests that careful attention be paid to humanimmunization schedules that utilize bacterialcarbohydrates. Clearly, the potential for produc-tion of autoreactive antibodies exists and shouldbe carefully evaluated.

Indeed, the recent demonstration that certainsuppressor cell-reactive antibodies are inhibita-ble in the presence of lactose (9) suggests apotential for perturbation of normal immuneregulatory mechanisms if a sufficient quantity ofgalactose-binding cold agglutinin-like antibodiesare induced by immunization.

ACKNOWLEDGMENTS

This work was supported by Public Health Service Re-search grant Al 16220 from the National Institutes of Health.J.C.B. is a recipient of Public Health Service Research CareerDevelopment Award AI00427 from the National Institutes ofHealth.We gratefully acknowledge the technical assistance of Anne

Himmelstein and Mary Ethen.

LITERATURE CITED

1. Baker, D. J., and D. L. Kasper. 1976. Microcapsule oftype III strains of group B streptococcus: production andmorphology. Infect. Immun. 13:189-194.

2. Brown, J. C., and R. G. Colling. 1982. Properties of coldagglutinin and group carbohydrate-specific antibodies iso-lated from group C streptococcal antisera. Mol. Immunol.19:457-465.

3. Carey, R. B., T. K. Eisenstein, G. D. Shockman, T. F.Greber, and R. M. Swenson. 1980. Soluble group- andtype-specific antigens from type III group B streptococ-cus. Infect. Immun. 28:195-203.

4. ChDlds, R. A., A. Kapadia, and T. Feizi. 1980. Expressionof blood group I and i active carbohydrate sequences oncultured human and animal cell lines assessed by radioim-munoassays with monoclonal cold agglutinins. Eur. J.Immunol. 10:379-384.

5. Coligan, J. E., B. A. Fraser, and T. J. Kindt. 1977. Adisaccharide hapten from streptococcal group C carbohy-drate that cross-reacts with the Forssman glycolipid. J.Immunol. 118:6-11.

6. Colling, R. G., and J. C. Brown. 1979. The appearance ofIgM and IgG cold agglutinins in rabbits hyperimmunizedwith group C streptococcal vaccine. J. Immunol. 122:202-208.

7. Costea, N., V. J. Yakulis, and P. Heller. 1971. Themechanism of induction of cold agglutinins by Mycoplas-ma pneumoniae. J. Immunol. 106:598-604.

8. Herd, Z. L., and J. Spragg. 1972. The occurrence ofhomogeneous antibodies and antigenic selection of lightchains to streptococcal carbohydrates in Monash rabbits.Aust. J. Biol. Med. Sci. 50:225-243.

9. Honda, M., T. Sakane, A. D. Steinberg, H. Kotani, T.Tsunematsu, K. Moriyama, and M. Fukase. 1982. Studiesof immune functions of patients with systemic lupuserythematosus. J. Clin. Invest. 69:940-949.

10. Jennings, H. J., K. Rosell, and D. L. Kasper. 1980.Structural determination and serology of the native poly-saccharide antigen of Type III group B streptococcus.Can. J. Biochem. 58:112-120.

11. Lau, F. O., and W. F. Rosse. 1975. The reactivity of redblood cell membrane glycophorin with cold reacting anti-bodies. Clin. Immunol. Immunopathol. 4:1-4.

12. Lind, K. 1973. Production of cold agglutinins in rabbitsinduced by Mycoplasma pneumoniae, Listeria monocyto-genes, or streptococcus MG. Acta. Pathol. Microbiol.

INFECT. IMMUN.

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

COLD AGGLUTININ ANTIBODIES 213

Scand. Sect. B. 81:487-496.13. Pruzanskl, W., and K. H. Shumak. 1977. Biologic activity

of cold-reacting autoantibodies. N. EngI. J. Med.297:238-242.

14. Wasserman, E., and L. Levine. 1960. Quantitative micro-complement fixation and its use in the study of antigen

structure by antigen-antibody inhibition. J. Immunol.87:290-295.

15. Wood, E., J. Lecomte, R. A. Childs, and T. Feizi. 1979. Aradioimmunoassay for the measurement of blood group liactivities: its application to glycoconjugates, oligosaccha-rides and intact cells. Mol. Immunol. 16:813-819.

VOL. 41, 1983

on April 25, 2021 by guest

http://iai.asm.org/

Dow

nloaded from