Proc. Natl. Acad. Sci. USAVol. 90, pp. 3088-3092, April 1993Immunology

Soluble antigen profoundly reduces memory B-cell numbers evenwhen given after challenge immunization

(adult tolerance/variable gene hypermutation/germinal centers/carrier-specific T cells)

G. J. V. NOSSAL, MARIA KARVELAS, AND BALI PULENDRANThe Walter and Eliza Hall Institute of Medical Research, Post Office, The Royal Melbourne Hospital, Victoria 3050, Australia

Contributed by G. J. V. Nossal, December 21, 1992

ABSTRACT The splenic B-cell repertoire of unimmunizedC57BL/6 mice can be examined for anti-(4-hydroxy-3-nitrophenyl)acetyl (NP) B cells of relatively high affinity byusing a dual strategy. First, limiting numbers ofsplenocytes arepolyclonally activated by Eschericia coli lipopolysaccharideand a mixture of interleukins 2, 4, and 5 in the presence of 3T3filler cells, thus ensuring that many B-cell clones switch to IgGlantibody production. Second, an enzyme-linked immunosor-bent assay is geared to register only higher-affinity antibody by(i) detecting only bivalent IgGl antibody and ignoring IgM and(ii) using a lowly substituted NP-conjugated protein as thecapture layer. Naive spleens contain very few higher-affinityanti-NP B cells thus dermed, but thymus (T)-dependent im-munization causes the appearance of 105 per spleen within 2weeks. The development of these clonable anti-NP antibody-forming cell precursors can be virtually eliminated by a singleinjection of 1 mg of soluble, freshly deaggregated NP2-humanserum albumin (HSA). This toleragen works not only ifinjectedprior to challenge immunization, but even if given up to 6 dayslater. Soluble HSA works partially but not nearly as well asNP2-HSA, suggesting the possibility that the toleragen must acton T and B cells. NP conjugated to irrelevant carriers achievedpartial tolerance in only one of four experiments. The studiesdemonstrate the need for continuing T-cell help throughout theprocess ofmemory B-cell generation. They also show that thoserecently activated T cells involved in this process can besilenced in vivo by soluble toleragen.

An early and important feature of the immune response toT-dependent antigenic challenge is the genesis of B lympho-cytes with receptors of high affinity for the antigen inquestion (1-3). This process depends on somatic hypermu-tation of immunoglobulin variable region (V) genes (1-4),occurring mainly in lymphoid germinal centers (5-8). Astolerance can readily be induced within the primary B-cellrepertoire (9-12), we have addressed the question (13) ofwhether mechanisms exist to prevent fortuitous mutationstoward anti-self-reactivity within the secondary B-lympho-cyte repertoire-i.e., the memory B-cell compartment. Pre-viously, we used a soluble deaggregated protein antigen,human serum albumin (HSA), as a model toleragen (14, 15)and showed that it could prevent the generation of anti-HSAmemory B cells in a dose-dependent manner. Adoptivetransfer studies showed that the toleragen affected donor Tand B cells, the former effect being more profound.Another way of studying the relative contributions of T-

and B-lymphocyte populations to an antibody response is tomake use of various hapten-carrier combinations (16-18).T-dependent anti-hapten antibody formation requires thatanti-hapten B cells be "helped" by carrier-specific T cells.The anti-(4-hydroxy-3-nitrophenyl)acetyl (NP) response of

C57BL/6 mice has been extensively investigated (19-21)because of the preferential appearance of B-cell clones uti-lizing the VH 186.2 V gene and the Al light chain gene,facilitating the study of V gene hypermutation. It is also asystem in which clones making relatively high- versus rela-tively low-affinity antibody can be discriminated by the useof, respectively, relatively lowly versus relatively highlyhaptenated protein conjugates in an ELISA (22, 23). In thepresent study, we immunized mice with highly haptenatedNP-HSA, and attempted tolerance induction with lowlyhaptenated, freshly deaggregated, NP-HSA, with HSA car-rier alone, or with NP attached to irrelevant carriers. The keyfindings were that NP-HSA .Jways worked better than car-rier alone or NP attached i other carriers in preventingmemory cell appearance and that tolerance could still beinduced after immunogenic challenge, up to the time of thefirst appearance of V gene mutations. Apparently, the func-tional silencing of T and B cells is necessary for the fulltolerance effect, and T-ceil help is still required for continuedB-memory cell generation after the germinal center reactionis well under way.

MATERIALS AND METHODSAnimals, Antigens, and Immunization. Specific pathogen-

free male C57BL/6 WEHI mice, 8-10 weeks old at firstimmunization, were used. Challenge immunization consistedof 100 ,ug of alum-precipitated NP conjugated to HSA in-jected i.p. together with 109 killed Bordetella pertussis orga-nisms (Commonwealth Serum Laboratories, Melbourne,Australia). Some mice received a single i.p. dose of solubleantigen from 7 days before to 13 days after challenge immu-nization. This material had been freshly deaggregated bycentrifugation at 15,000 x g for 10 min in a Heraeus ChristBiofuge A centrifuge. Some groups ofmice were serially bledfrom the orbital venous plexus and serum samples wereprepared for antibody assay.NP conjugates were prepared as described (23). Briefly,

the succinimide ester of NP (Cambridge Research Biochem-icals, Cambridge, U.K.) was dissolved in N,N-dimethylfor-mamide (Sigma), to a final concentration of 20 mg/ml, andconjugated to the carriers bovine serum albumin (BSA)(fraction 5; Sigma), HSA (fraction 5; Sigma), horse serumalbumin (HoSA) (Sigma), and cytochrome c (Cyt-c) (frompigeon heart; Sigma). All proteins were at 5 mg/ml in 3%NaHCO3.To produce two NP epitopes per carrier molecule (NP2-

carrier), a ratio of 10-20 ,ug/mg of NP-succinimide to carrier

Abbreviations: AFCP, antibody-forming cell precursor; BSA, bo-vine serum albumin; Cyt-c, cytochrome c; FDC, follicular dendriticcell; HoSA, horse serum albumin; HSA, human serum albumin; IL,interleukin; LPS, lipopolysaccharide; MT-PBS, mouse tonicityphosphate-buffered saline; NP, (4-hydroxy-3-nitrophenyl)acetyl; r,recombinant; V, variable.

3088

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

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

6, 2

020

Proc. Natl. Acad. Sci. USA 90 (1993) 3089

was required. NP18-carrier was produced using a 800 ,g/mlratio of NP-succinimide to carrier.

After 2 hr rotating at room temperature, conjugates wereextensively dialyzed against 3% NaHCO3 and finally mousetonicity phosphate-buffered saline (MT-PBS) (pH 7.2). Con-jugation ratios were determined.

High conjugates of NP-HSA were used for all challengeimmunizations described. Low NP to carrier ratios were usedas toleragens.Lymphokines. Recombinant interleukin 4 (rIL-4; ref. 24)

and recombinant interleukin 5 (rIL-5; ref. 25) were obtainedfrom hybridoma culture supernatants of monkey COS cellstransiently transfected with a simian virus 40 murine expres-sion construct. The constructs were kindly provided by S.Gerondakis and A. B. Troutt (this institute). rIL-4 and rIL-5were added to cultures to a final concentration of 0.4%(vol/vol) and 2% (vol/vol), respectively. Highly purifiedhuman rIL-2 from Escherichia coli was the generous gift ofCetus (26). rIL-2 was added to cultures at a final concentra-tion of 10 units/ml.

Cell Preparation and Culture for Antibody-Forming CelPrecursor (AFCP) Determination. Five thousand 3T3BALB/c fibroblasts (Commonwealth Serum Laboratories)were aliquoted into flat-bottomed 96-well microtiter trays in100,4u. Culture medium (CM) was RPMI 1640 containing 100,uM 2-mercaptoethanol and 5% (vol/vol) fetal calf serum(Flow Laboratories). Spleen cell suspensions were preparedas described (27, 28), including 0.83% NH4Cl for erythrocytelysis followed by damaged and dead cell removal. Spleno-cytes were dispensed in 100 ul of CM containing E. coli0111:B4 lipopolysaccharide (LPS; Difco) at a final concen-tration of20,g/ml and rIL-2, rIL-4, and rIL-5, making a finalvolume of 200 ,u per culture. To determine IgGl anti-NPprecursor frequency, cultures were prepared at limiting di-lution with 300-10,000 splenocytes per well. A total of 48replicate cultures was studied at each cell input number.Furthermore, to determine B-cell cloning efficiency, 48 rep-licate cultures were prepared at low-input cell numbers(2.5-20 splenocytes per well). Two spleens were pooled foreach test group. Cultures were incubated at 37°C in a hu-midified atmosphere of 10% C02/90% air for 7 days.Antibody Assays of Serum Samples and Clonal Superna-

tants. An ELISA was used to detect and quantitate antibodyformation in culture supernatants of the IgM and IgGlisotype (29). U-bottomed, 96-well polyvinyl chloride plateswere coated overnight with 50 ,l of affinity-purified sheepanti-mouse immunoglobulin (Silenus Laboratories, Haw-thorn, Victoria, Australia) at 2 ,ug/ml in 0.2 M sodiumcarbonate buffer. For anti-NP specific assays, a NP2-BSA or

10,000

U2

100

-oo0

01~~~~~~~~~~~

NP18-BSA solution (20 ,tg/ml) in MT-PBS was used as theplate coat. An appropriate dilution of culture supernatantfrom each well was transferred into the precoated plates andheld overnight at room temperature (RT). The diluent for theassay was MT-PBS containing 0.05% Tween 20, 1% newbornbovine serum, and 0.1% skim milk powder. After washing,the horseradish peroxidase-coupled anti-murine immuno-globulin with specificity for either IgM or IgGl isotypes(Southern Biotechnology Associates, Birmingham, AL) wasadded for 4 hr at RT. The plates were washed before theaddition of the substrate, 2,2-azinobis(3-ethylbenzthiazolinesulfonic acid), at 0.55 mg/ml in 0.1 M citric acid with 0.1%H202 (1 hr at RT). The absorbance of the wells was measuredat 414 nm with reference wavelength at 492 nm, using aTitertek Multiskan MCC/340 [Titertek Multiskan MCC/340(MKII 347 Labsystems, Helsinki)].Anti-NP antibody serum titers were measured by using

duplicate serial dilutions of serum starting at 1:1000 and1:10,000 for anti-IgGl NP2 and anti-IgGl NP18, respectively.

Statistical Analysis of ELISA Results. Computerized anal-ysis of the ELISA was performed with software developed byA. P. Kyne (The Walter and Eliza Hall Institute). Serial 1:2dilutions of the myeloma proteins MOPC 104E (IgM) andMOPC 21 (IgGl) (Bionetics Research Institute) on each testplate were used to standardize the assays. Hyperimmuneanti-NP keyhole limpet hemocyanin (Calbiochem/BehringDiagnostics) was used for the antigen-specific ELISA. Acubic spline curve-filling technique was used to fit a curvethrough all points of the standard curves, and antibodycontents of individual wells were computed from this.

Poisson analysis (30) was used to determine the frequencyof immunoglobulin-secreting precursors and the maximumlikelihood estimator for linear regression analysis (31).

RESULTSNP2-HSA Can Lower Serum Anti-NP Antibody Titers Re-

sulting from NP18-HSA Challenge. Mice in groups offour weregiven 1 mg or 100 ,g of soluble, deaggregated NP2-HSA,NP18-HSA, or no injection, and 7 or 4 days later werechallenged with alum-adsorbed NP18-HSA plus adjuvant.They were bled at intervals and serum anti-NP IgGl antibodylevels were determined by ELISA, using plastic-bound NP2-BSA (Fig. 1). NP18-HSA was ineffective as a toleragen,perhaps because of residual aggregation or rapid clearance,and was not used in the rest of these experiments. NP2-HSAgave a dose-dependent reduction in antibody binding, whichwas about 100-fold for 1 mg. Mice were given 1 mg of solubleNP2-HSA at various times after immunogenic challenge and

Ta 7

Time after immrunization, days

FIG. 1. Serum NP2-BSA-binding IgGl values at various times after challenge immunization. Mice were preinjected 7 (A) or 4 (B) days priorto immunogenic challenge, with soluble NP-HSA. Vertical bars indicate standard deviations.

Immunology: Nossal et al.

no 17

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

6, 2

020

Proc. Natl. Acad. Sci. USA 90 (1993)

antibody levels were determined always 14 days after chal-lenge (data not shown). Toleragen given up to 6 days afterimmunization caused an up to 4-fold lowering of high-affinityantibody titers, thus a much smaller effect than toleragengiven prior to immunization. Toleragen given 8 or 13 daysafter immunization had no such effect.Appearance of Higher-Affinity Anti-NP IgGl-Forming Pre-

cursors During Primary Immunization and Reduction in TheirNumbers Through Preinjection of Soluble Antigen. Spleencells from unimmunized mice were placed in limiting dilutionmicrocultures and stimulated with LPS, IL-2, IL-4, and IL-5in the presence of 3T3 filler cells. One week later, culturesupernatants were examined for the presence ofIgGl anti-NPantibody of sufficient affinity to register on NP2- or NP18-BSA plate coats. It was found that very few B cells of thehigher affinity were present in the primary repertoire. In fourseparate experiments, the numbers were 0, 93, 126, and 240per spleen. This way of examining the repertoire is madepossible by utilizing the capacity of IL-4 to drive the isotypeswitch and by an ELISA read-out that focuses only onbivalent IgGl antibody (14, 15). Through these strategies, themany potentially anti-NP clonotypes of low affinity thatwould have registered using an IgM read-out are simplyignored.Mice were given 1 mg of NP2-HSA or no injection 4 days

before immunization with NP18-HSA and adjuvant. Theywere killed at intervals in groups of two and the numbers ofclonable higher-affinity anti-NP IgGl AFCPs were deter-mined (Fig. 2). Significant numbers of AFCPs appeared incontrols from day 5 on, peaking at day 14. The toleragenreduced these numbers, especially after day 6. At the heightof the response, the reduction was down to 2% of controlvalues. This mirrors approximately the reduction in primaryantibody titers. This tolerance among higher-affinity hapten-specific AFCPs was identical to that previously encountered(14) for protein-specific AFCPs. The tolerance effect wasdependent on the dose of soluble antigen (Fig. 3), thoughAFCP generation was still significantly affected by 10 ,g oftoleragen.

Soluble Antigen Still Causes Tolerance if Given up to 6 DaysAfter Challenge. We sought to mimic events that might occurif B cells were suddenly to hypermutate toward anti-self andtherefore injected soluble antigen at various times afterantigen as a surrogate for a soluble self-antigen. Fig. 4 showsa dose-response study where NP2-HSA was injected 6 daysafter alum-adsorbed NP18-HSA plus B. pertussis adjuvantand mice were killed a further 8 days later-i.e., 14 days afterchallenge. Both 1 mg and 100 jig caused a marked reduction

.100,000

Z 1,000

c:

100

En 10000

11

* NilO 1 mg of NP2-HSA

Ti. a1 m3 4 5 6 7 8 14 21Time after immunization, days

FIG. 2. Effect of 1 mg of soluble NP2-HSA 4 days beforeimmunogenic challenge on anti-NP2-BSA IgGl AFCPs from spleensat various times after immunization. Vertical bars indicate 95%confidence limits. The absence of a solid bar at day 3 and hatchedbars at days 7 and 8 means that no AFCPs were found in thosegroups.

LUUVUUU .

rA

OU,

10,000

1,0000 1 0.1 0.01

Soluble NP2-HSA injected4 days prior to immunization, mg

FIG. 3. Dose-response characteristics of tolerance induced bypreinjection of NP2-HSA. Vertical bars indicate 95% confidencelimits.

in the appearance of anti-NP2-AFCPs, essentially just asprofound as had preinjection been used. This is in contrast tothe much lesser effect that toleragen given at the same timehad on primary antibody production. Clearly, the AFCP cellgeneration process is particularly vulnerable to toleragenesis.AFCP values of control groups varied from experiment to

experiment. To document the consistency of the effect of 1mg of NP2-HSA given 6 days after immunogenic challenge,10 separate experiments were performed and all showed amarked reduction. The mean + SEM AFCP control valuewas 76,100 ± 44,900 for the higher-affinity (NP2-BSA-binding) and 502,000 ± 248,000 for the lower-affinity (NP18-BSA-binding) AFCPs. The tolerant groups were, respec-tively, 4.6% ± 2.5% and 37% ± 12% of controls. Thus, asexpected, the higher-affinity B cells were more profoundlyaffected.

Fig. 5 shows the effect of varying the time after immuni-zation when toleragen was given. Up to 6 days after immu-nization, the toleragen worked, and at 8 days it caused alowering of AFCP numbers in one of two experiments. Whentoleragen was given at 10 days, 4 days before sacrifice, amarked booster effect was noted-i.e., AFCP numbers roseby about 10-fold.Maximal Tolerance Is Induced After Challenge Only when

Hapten and Carrier Are Identical to the Immunogen. Wewished to determine whether soluble toleragen given afterimmunogenic challenge affected mainly the B-cell or theT-cell compartment. Accordingly, we used freshly deaggre-gated toleragens in which NP was coupled to differentcarriers and also the HSA carrier alone (Fig. 6). In each case,

a 1,000,000-a)a)

Uq 100,000-,

04(AS- 1

Z) 100,000-

10,000-

l1-

T

U I U.1 U.UISoluble NP2-HSA injected

6 days after immunization, mg

FIG. 4. Dose-response characteristics of tolerance induced byNP2-HSA injected 6 days after challenge immunization. Vertical barsindicate 95% confidence limits.

3090 Immunology: Nossal et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

6, 2

020

Proc. Natl. Acad. Sci. USA 90 (1993) 3091

8

0

0

80

0

0

oIi

* I

O

.ts

-L

a h I

3 5 6 8 10 13 0 3Time after immunization, days

S

EI

5 6 8 10 13

FIG. 5. Effect of 1 mg of NP2-HSA injected at various times after challenge immunization. All mice were killed 14 days after challenge. High-(Left) and low- (Right) affinity anti-NP-BSA IgGl AFCP numbers were normalized to results from a control uninjected group. Horizontal barsrepresent normalized geometric mean values.

the putative toleragen was injected at 1 mg 6 days afterimmunogenic challenge with alum-adsorbed NP18-HSA plusB. pertussis. Mice were killed 8 days later. In each experi-ment NP2-HSA had its usual profoundly toleragenic effectand these negative controls are not shown. HSA alone wastoleragenic in two separate experiments but not as powerfullyas NP2-HSA. NP coupled to different carriers was onlytoleragenic in one experiment of two with NP3-HoSA. Nei-ther NP8-Cyt-c nor NP3-Cyt-c was effective. Thus, the in-duction of tolerance among HSA-specific carrier T cells aslate as 6 days after immunization still substantially reducesanti-NP AFCP formation but not as completely as NP2-HSA.In the presence of continuing, nonablated T-cell help, itappears to be quite difficult to stop the further generation ofanti-NP AFCPs through the use of conjugates affecting onlyNP-specific B cells. However, with T-cell help ablated, useof a toleragen (NP2-HSA) that can also affect B cells achievesmore profound tolerance than use of one (HSA) that does notaffect the B cells in question.

1000

10

co 1

0.

< <

FIG. 6. Effect of administration of carrer alone or valious NPconjugates 6 days after challenge immunization on anti-NP AFCPnumbers. Each column gives results of a separate experiment.Values were normalized to that experiment's positive control (chal-lenge but no soluble antigen). Negative controls (NP2-HSA) wereincluded in each experiment and showed uniform tolerance as in Fig.5.

DISCUSSIONPrimary T-dependent immunization evokes two separatecascades of B-cell proliferation, which may (32) or may not(33) be related, leading, respectively, to antibody formationand germinal center formation with memory B-cell genera-tion. The response of C57BL/6 mice to NP-conjugates hasbeen particularly well studied (e.g., refs. 1, 6, 19-23, 32), withsplenic B cells demonstrating the first few V gene mutationsaround 6 or 7 days after immunization. By 1 week later,mutations are extensive, including, in the majority of NP-specific B cells, an affinity-raising tryptophan-to-leucinechange at amino acid position 33 in the first complementarity-determining region of the V heavy chain (20, 21, 34). Thus,during the second week there is extensive somatic V genehypermutation and selection for raised affinity to NP, and thechief site for these events is the germinal center (6-8). Thepresent study shows that soluble, freshly deaggregated anti-gen can reduce primary anti-NP IgGl antibody formation(Fig. 1) and the genesis of those memory B cells capable ofresponding in vitro to LPS in the presence ofIL-2, IL-4, IL-5,and 3T3 filler cells by IgGl formation. Interestiigly, theseeffects can still be achieved 6 days after immunization, a timeat which the germinal center reaction is well under way andat which somatic hypermutation is just beginning. Our ex-perimental design does not allow us to distinguish between a

"switching off" of ongoing germinal center events or areduction in memory cell generation due to an interferencewith the recruitment of new activated B cells into thegerminal center. The apparent failure of toleragen at latertime points is not surprising. First, as antibody formation issignificant by 7 days after immunization, originally solubleand freely circulating antigen would quickly be removed tofollicular dendritic cells (FDCs) and other accessory cells. Inthese sites, it may actually have an immunogenic effect, asseen by the increased number of AFCPs on day 10 in Fig. 5.Second, the events leading to memory cell formation musteventually reach a stage where it is difficult to modulate themby soluble antigen.How does soluble toleragen given after challenge achieve

its effects? The fact that carrier HSA can cause substantiallowering of AFCP numbers (Fig. 6) suggests that a majoreffect is through the abrogation ofcarrier-specific T-cell help.It is interesting that T-cell help is still required at this stageof immunization. Perhaps more B cells require recruitmentinto the germinal centers. This happens only after extrafol-licular T-dependent B-cell activation (35). Alternatively, per-

10,000

.5.v 1,000 _

0.

m 100

10m

8 _'4-4~0i

Immunology: Nossal et al.

00

.t

68

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

6, 2

020

Proc. Natl. Acad. Sci. USA 90 (1993)

haps the specialized (36) CD4+ T cells that inhabit germinalcenters are needed continuously for the B-cell mutationand/or selection process. It is clear that NP2-HSA worksmore completely than HSA as a toleragen. This could bebecause in HSA-tolerized mice some residual help is pro-vided by hapten-specific T cells. Alternatively, it could bethat a direct toleragenesis of B cells themselves augments theoverall effect. Admittedly, only one experiment of four usingNP coupled to other carriers caused tolerance in NP-specificAFCPs (Fig. 6), but this might have reflected a competitionbetween negative signaling through the toleragen and con-tinuing help from HSA-specific T cells. In most experimentson antibody formation, help overrides tolerance. The presentexperimental design thus does not critically address theconcept of a "second window" of tolerance susceptibility inrecently activated prememory B cells (13). However, it isnoteworthy that, well after the events of immunogenesis,including the induction of T-cell help, a soluble antigen canso profoundly impede the further generation of AFCPs. Onepossibility is that T-cell anergy is inducible even in recentlyactivated CD4+ T cells. Alternatively, ongoing B-cell recruit-ment may involve ongoing T-cell activation, which may benegated by soluble antigen.These results are relevant to an understanding of autoim-

munity. The clear need for continuing T-cell help in thegenesis of memory B cells within germinal centers suggeststhat if a B cell were to mutate and fortuitously acquireanti-self-reactivity, it could not continue to proliferate be-cause of an absence of T-cell help, as the T-cell populationwould presumably be tolerant of the self-antigen in question.Moreover, the greater effectiveness of NP2-HSA than HSAsuggests ancillary mechanisms working directly on the B cellwhereby a soluble self-antigen could silence the mutated Bcell. In any case, to exit from the germinal center a mutatedB cell must be positively selected, probably through inter-actions involving antigen bound to FDCs. Unless the muta-tion toward self were against some FDC surface antigen, ora self-antigen cross-reactive with the immunogen, the anti-self B cell would undergo apoptosis much as do cells withaffinity-lowering or neutral mutations. There thus appear tobe several layers of protection against mutation-derivedanti-self B cells.How, then, do autoantibodies with multiple V gene muta-

tions (37) arise? Frequently, the mechanism appears toinvolve an intracellular microparticulate entity (small nuclearribonucleoprotein particle, nucleosome, centromere, mi-crosome, mitochrondrial enzyme complex) somehow enter-ing an immunogenic processing pathway, as autoantibodiesappear to different molecules of the particle and to differentepitopes on each molecule. Frequently, it involves self-antigens against which B-cell tolerance is absent or incom-plete. This being said, autoantibodies such as rheumatoidfactor, where the antigen is the Fc piece of immunoglobulinand thus a widely available antigen, are difficult to explain.A genetic B-cell hyperreactivity and resistance to toleragen-esis (38) may be at work in some autoimmune diseases, andcross-reactive T-cell help through molecular mimicry mayalso be important. A more complete dissection of the com-plex series of processes occurring in germinal centers, whichproduce the secondary B-cell repertoire, cannot fail to illu-minate this fascinating problem.

The excellent technical assistance of Ms. Amanda Light is grate-fully acknowledged. This work was supported by the National Healthand Medical Research Council, Canberra Australia; by Grant AI-03958 from the National Institute for Allergy and Infectious Diseases,U.S. Public Health Service; and by a grant from the Human FrontierScience Program, Principal Investigator: Prof. K. Rajewsky.

1. Allen, D., Cumano, A., Dildrop, R., Kocks, C., Rajewsky, K.,

Rajewsky, N., Roes, J., Sablitzky, F. & Siekevitz, M. (1987)Immunol. Rev. 96, 5-22.

2. Berek, C. & Milstein, C. (1987) Immunol. Rev. 96, 23-41.3. McKean, D., Huppi, K., Bell, M., Standt, L., Gerhard, W. &

Weigert, M. (1984) Proc. Natl. Acad. Sci. USA 81, 3180-3184.4. Weigert, M. G., Cesari, I. M., Yonkovich, S. J. & Cohn, M.

(1970) Nature (London) 228, 1045-1047.5. MacLennan, I. C. M. & Gray, D. (1986) Immunol. Rev. 91,

61-85.6. Jacob, J., Kelsoe, G., Rajewsky, K. & Weiss, U. (1991) Nature

(London) 354, 389-392.7. Berek, C., Berger, A. & Apel, M. (1991) Cell 67, 1121-1129.8. Nossal, G. J. V. (1992) Cell 68, 1-2.9. Nossal, G. J. V. & Pike, B. L. (1980) Proc. Natl. Acad. Sci.

USA 77, 1602-1606.10. Nossal, G. J. V. (1983) Annu. Rev. Immunol. 1, 33-62.11. Goodnow, C. C., Crosbie, J., Adelstein, S., Lavoie, T. B.,

Smith-Gill, S. J., Brink, R. A., Pritchard-Briscoe, H., Wother-spoon, J. S., Loblay, R. H., Raphael, K., Trent, R. J. &Basten, A. (1988) Nature (London) 334, 676-682.

12. Nemazee, D. & Burki, K. (1989) Proc. Natl. Acad. Sci. USA86, 8039-8043.

13. Linton, P.-J., Rudie, A. & Klinman, N. R. (1991) J. Immunol.146, 4099-4104.

14. Nossal, G. J. V. & Karvelas, M. (1990) Proc. Natl. Acad. Sci.USA 87, 1615-1619.

15. Karvelas, M. & Nossal, G. J. V. (1992) Proc. Natl. Acad. Sci.USA 89, 3150-3154.

16. Mitchison, N. A. (1971) in Cell Interactions and ReceptorAntibodies in Immune Responses, eds. Makela, O., Cross, A.& Kosunen, T. U. (Academic, Orlando, FL), pp. 249-260.

17. Rajewsky, K. (1971) Proc. R. Soc. London Ser. B 176, 385-392.18. Miller, J. F. A. P. (1973) Contemp. Top. Immunobiol. 2, 151-

164.19. Makela, 0. & Karjalainen, K. (1977) Transplant. Rev. 34,

119-138.20. Cumano, A. & Rajewsky, K. (1985) Eur. J. Immunol. 15,

512-520.21. Cumano, A. & Rajewsky, K. (1986) EMBO J. 5, 2459-2468.22. Ternynck, T. & Avrameas, S. (1986) Immunol. Rev. 94, 99-112.23. Lalor, P. A., Nossal, G. J. V., Sanderson, R. D. & McHeyzer-

Williams, M. G. (1992) Eur. J. Immunol. 22, 3001-3011.24. Takebe, Y., Saki, M., Fujisawa, J., Hoy, P., Yokota, K., Arai,

K.-I., Yoshida, M. & Arai, N. (1988) Mol. Cell. Biol. 8,466-472.

25. Yokoto, T., Coffman, R. L., Hagiwara, H., Rennick, D. M.,Takabe, Y., Yokoto, K., Gemmell, L., Shrader, B., Yang, G.,Meyerson, P., Luh, J., Hoy, P., Pene, J., Briere, F., Spits, H.,Bancheread, J., de Vries, J., Lee, F. D., Arai, N. & Arai, K.-I.(1987) Proc. Natl. Acad. Sci. USA 84, 7388-7392.

26. Rosenberg, S. A., Grimm, E. A., McGroom, M., Doyle, M.,Kawasaki, E., Koths, K. & Mark, D. F. (1984) Science 223,1412.

27. Nossal, G. J. V. & Pike, B. L. (1976) Immunology 30, 189-202.28. Von Boehmer, H. & Shortman, K. D. (1973) J. Immunol.

Methods 2, 293-301.29. Nossal, G. J. V. & Riedel, C. (1989) Proc. Natl. Acad. Sci.

USA 86, 4679-4683.30. Lefkovitz, I. & Waldmann, H. (1984) Immunol. Today 5,

265-268.31. Good, M. F., Boyd, A. W. & Nossal, G. J. V. (1983) J. Im-

munol. 130, 2046-2055.32. Jacob, J. & Kelsoe, G. (1992) J. Exp. Med. 176, 679-687.33. Linton, P.-J., Lo, D., Lai, L., Thorbecke, G. J. & Klinman,

N. R. (1992) Eur. J. Immunol. 22, 1293-1297.34. Allen, D., Simon, T., Sablitzky, F., Rajewsky, K. & Cumano,

A. (1988) EMBO J. 7, 1995-2001.35. MacLennan, I. C. M., Liu, Y. J. & Johnson, G. D. (1992)

Immunol. Rev. 126, 143-161.36. Cosgrove, D., Gray, D., Dierich, A., Kaufman, J., Lemeur, M.,

Benoist, C. & Mathis, D. (1991) Cell 66, 1051-1066.37. Shlomchik, M. J., Marshak-Rothstein, A., Wolfowicz, C. B.,

Rothstein, T. L. & Weigert, M. G. (1987) Nature (London) 328,805-811.

38. Erikson, J., Radic, M. Z., Camper, S. A., Hardy, R. R., Car-mack, C. & Weigert, M. (1991) Nature (London) 349, 331-334.

3092 Immunology: Nossal et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

6, 2

020