Proc. Natl. Acad. Sci. USAVol. 92, pp. 3348-3352, April 1995Immunology

B-cell activation by crosslinking of surface IgM or ligation ofCD40 involves alternative signal pathways and results indifferent B-cell phenotypesHENRY H. WORTIS*t, MARK TEUTSCH*, MINDY HIGER*, JENNY ZHENG*, AND DAVID C. PARKERt§*Department of Pathology, Tufts University School of Medicine, and Graduate Program in Immunology, Sackler School of Graduate Biomedical Sciences, Boston,MA 02111; and tDepartment of Molecular Genetics and Microbiology, University of Massachusetts Medical School, Worcester, MA 01655

Communicated by Salome G. Waelsch, Albert Einstein College of Medicine of Yeshiva University, Bronx, NY December 7, 1994

ABSTRACT Treatment of small resting B cells with sol-uble F(ab')2 fragments of anti-IgM, an analogue of T-inde-pendent type 2 antigens, induced activation characterized byproliferation and the expression of surface CD5. In contrast,B cells induced to proliferate in response to thymus-dependentinductive signals provided by either fixed activated T-helper 2cells or soluble CD40 ligand-CD8 (CD40L) recombinant pro-tein displayed elevated levels of CD23 (FcJII receptor) and nosurface CD5. Treatment with anti-IgM and CD40L inducedhigher levels of proliferation and generated a single popula-tion ofB cells coexpressing minimal amounts ofCD5 and onlya slight elevation ofCD23. Anti-IgM- but not CD40L-mediatedactivation was highly sensitive to inhibition by cyclosporin Aand FK520. Sp-cAMPS, an analogue of cAMP, augmentedCD40L and suppressed surface IgM-mediated activation.Taken together these results are interpreted to mean thatthere is a single population of small resting B cells that canrespond"to either T-independent type 2 (surface IgM)- orT-dependent (CD40)-mediated activation. In response to dif-ferent intracellular signals these cells are induced to enteralternative differentiation pathways.

Though it has been suggested that T-cell-independent (TI) andthymus-dependent (TD) responses (1) might be generatedfrom different B-cell precursors (2), the weight of evidencesuggests that a given B cell can enter either the TI or TDpathway (3-5). We wanted to directly determine if B cells existas two precursor populations or if there is a single populationthat can be induced to enter alternative differentiation path-ways. If the latter proved to be the case, we wanted todetermine if the minimal inductive events were sufficient toinduce phenotypic differences in responding cells. To studythis we chose to use anti-IgM antibody as an experimentalanalogue of one type of TI antigen. TI antigens are divided intotype 1, which are intrinsically mitogenic [such as lipopolysac-charide (LPS) in the mouse], and type 2, which are nonmito-genic repeating unit molecules such as polysaccharides or theirhaptenated derivatives (6). TI-2 activation is initiated by thecrosslinking of surface immunoglobulin (slg) followed by acascade of events that includes cytoplasmic tyrosine kinases, aCa2+ flux, and the catalytic activity of protein kinase C (PKC)(reviewed in ref. 7). The F(ab')2 fragments of antibody tomouse 1gM can bind to B-cell sIgM so as to effect itscrosslinking and thereby can provide a polyclonal B-cell acti-vating signal equivalent to a TI-2 antigen (8).

Unlike TI-2 responses, a TD response does not require sIgcrosslinking, as even monomeric antigen can be bound, endo-cytosed, processed, and presented in the context of majorhistocompatibility complex class II for T-helper cell recogni-tion and activation followed ultimately by T-cell-mediatedactivation of B cells (9). The crucial TD signal is provided by

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

direct contact with an activated T-helper cell such that thesurface molecule gp39 [CD40 ligand (CD40L)] of the T cellligates the CD40 of the B cell (10).Our strategy to determine if minimal TD and TI-2 signals

were sufficient to induce phenotypic differences in B cells wasto use an antigen that could be modified so as to initiate eithera TD or a TI-2 response. To produce a TD response we usedmonovalent Fab of rabbit anti-IgM and provided help in theform of the clone CDC35, a T-helper 2 line known to providehelp to B cells that have bound, processed, and presentedrabbit Fab in the context of IAd (9, 11). To provide theequivalent of a TI-2 signal we used the divalent F(ab')2 formof the same antibody. We then compared the surface pheno-types of the responding B cells.We followed up these initial experiments with others using

fixed, preactivated helper T cells as B-cell activators (12). Thisallowed us to examine the phenotype of B cells activated byT-helper cells in the absence of interleukins. Finally, weexamined the phenotype of B cells stimulated by a solublefusion protein containing the CD40L (13) so as to directlycompare the phenotypes and the signaling pathways involvedin minimal TI-2 [F(ab')2 anti-IgM] and TD (CD40L) re-sponses. Our experiments led us to the conclusion that a singlepopulation of B cells can respond to either sIgM or CD40ligation using alternative signal pathways to generate activatedB cells of different phenotypes.

MATERIALS AND METHODS-Mice. Experiments involving CDC35 cells were done with

cells from BALB/c/J mice purchased from The JacksonLaboratory. Others were done using cells from CBA/HHW orCBA/HHW.Ighb mice (14).

Cells. Cells of the T-helper 2 line CDC35 were used 2-3weeks after their last stimulation. B cells were partially purifiedas described (15). Activated CDC35 T cells were generated byincubating cells at 1 x 106 per ml for 24 hr on plates precoatedwith anti-CD3 (145-2C11). The activated cells were thenwashed three times in phosphate-buffered saline and fixed with0.4% fresh paraformaldehyde for 5 min (12). Fixed cells werekept in tissue culture medium at 4°C. The supernate generatedby these activated T cells was harvested and stored frozen.

B-Cell Culture and Stimulation. Cells were cultured in RPMI1640 medium supplemented as described (15). A polyclonalrabbit anti-mouse IgM was prepared as Fab and F(ab')2 frag-ments as described (9). For experiments involving immunosup-pressants we prepared F(ab')2 fragments from goat anti-(MOPC-

Abbreviations: CsA, cyclosporin A; LPS, lipopolysaccharide; FKBP,FK506 binding protein; CD40L, CD40 ligand; TD, thymus-dependent;TI, T-cell-independent; sIg, surface Ig; IL, interleukin; PK, proteinkinase.tTo whom reprint requests should be addressed.§Present address: Department of Molecular Microbiology and Immu-nology, Oregon Health Science University, 3181 S.W. Sam JacksonPark Road, L220-Portland, OR 97201.

3348

Proc. Natl. Acad. Sci. USA 92 (1995) 3349

Table 1. Proliferation of small resting splenic B cells

Fixed CD40LFab F(ab')2 T T cells Fixed T cells +

Exp. Medium anti-IgM anti-IgM cells + Fab SN T cells + SN CD40L LPS F(ab')2DS-4 0.3* 1.2 17.7 42.0DS-9 1.2 5.8 45.2J180 [2.3] 2.7 12.7 96.5 [4.6] 22.9 42.9 96.6J185 8.1 12.5 14.9 83.1 129.6J181 0.2 0.2 1.6 25.1 15.1 85.6117t 3.2 78.0 42.5 108.6123t 1.6 13.8 37.6 85.8SA 3.1 101.3 52.0 120.8 200.2

SN, supernate.*[3H]Thymidine at 60 hr (42 hr in experiments marked t) expressed as the geometric mean of triplicate values of cpm x 10-3.SEM were -x/1.2 for all values -2 x 103 cpm except for the values in experiment J180 (shown in brackets), in which theSEM values were x/1.5.

104E, TEPC-183). By a Limulus amebocyte lysate method (Asso-ciates of Cape Cod) the levels of endotoxin were -10 ng per 30,ug of F(ab')2 for goat anti-mouse and rabbit anti-mouse prepa-rations. Cells producing the fusion protein CD40L-CD8 (13) werefrom Peter Lane (Basel Institute for Immunology). Supernatecontaining recombinant CD40L produced optimal proliferationwhen used at a final dilution of 1:2. For thymidine incorporationcells were cultured in 96-microwell plates in 200 ,ul of medium.For cytometric analysis the cells were cultured in bulk. In bothcases they were established at a density of 1 x 106 per ml. Freshor fixed T cells were added at 1 x 106 cells per ml. Fresh T cellswere irradiated to a total dose of3600 rads (1 rad = 0.01 Gy) froma cesium source before being placed in culture with the B cells.K235 LPS was used at a concentration of 5 ,ug/ml. FK520(L-683,590), rapamycin (L-657,862), and L-685,818 were kindgifts from Merck that were stored as 1 mg/ml stock solutions inethanol. Cyclosporin A (CsA) (Sandoz Pharmaceutical) wasstored at 2 mg/ml in 71.5% ethanol/28.5% Tween 80. Sp-cAMPSwas purchased from Research Biochemicals (Natick, MA).

Flow Cytometry. Cultures containing CDC35 T cells weretreated with anti-CD4 plus complement before staining. Insome experiments living cells were enriched by centrifugationover a Ficoll gradient (Lympholyte-M, Cedarlane Laborato-ries) prior to staining. Cells were stained as described (15).Flow cytometry data were obtained on a FACScan using LYSISacquisition and analytic software (Becton Dickinson).

RESULTSWe first asked whether the surface phenotype of B cellsactivated by the same nominal antigen in TI-2 and TD forms

SSC/FSC CD44

Medium

Anti 1gM

T Help

-- _ ..4y ..

would express identical surface phenotypes. Small restingBALB/c splenic B cells were placed in culture with either 10,ug of F(ab')2 of rabbit anti-IgM per ml or 10 ng of Fabfragments of the same antibody per ml together with CDC35T-helper 2 cells specific for Fab. Both treatments inducedB-cell proliferation (experiments DS-4 and DS-9, Table 1), cellenlargement, and increased levels of surface CD44 (Fig. 1),events previously described in association with lymphocyteactivation (15). Culture with Fab or T cells alone did notinduce proliferation (Table 1), nor was any change in B-cellsurface CD5 or CD23 seen (data not shown). F(ab')2-treatedcells became CD5+ and showed no change in CD23 levels (Fig.1). The T cell plus Fab-stimulated cells neither became CD5+nor had increased CD23 (Fig. 1); in fact, CD23 expression wentdown. Therefore, in response to TI-2 and TD antigens, B cellsacquire different surface phenotypes.

Previous experiments indicated that the CD23 levels ofactivated B cells could be lowered in response to interleukin 6(IL-6) or increased by IL-4 treatment (15). To determine if thechanges in B cells activated by T-helper cells resulted only fromdirect contact with T-cell membranes or were dependent inwhole or in part on antigen ligation and/or soluble ILs, wecultured B cells with fixed activated CDC35 cells. In theabsence of any additional antigen they also induced prolifer-ation (Table 1, experiments J-180 and J-181), enlargement,and increased CD44 (Fig. 2). As before, B cells activated bycontact with activated T-helper cells did not express CD5 butthey now showed increased levels of CD23 (Fig. 2). Thesupernate from activated T cells did not induce proliferation

CD5 CD23

FIG. 1. Flow cytometry analysisof B cells. Top row, small restingsplenic B cells at time 0; middlerow, cells 3.5 days after coculturewith F(ab')2 fragments of anti-IgM; bottom row, cells after stim-ulation with Fab fragments of thesame antibody plus irradiated T-helper cells from the line CDC35.The lefthand column shows for-ward (FSC) and side (SSC) lightscatter. Smaller, less complex cellsare in the far left corner; the largerand more complex cells are in thefar right corner. CD44, CD5,CD23, and negative control stain-ing is also shown.

Immunology: Wortis et al

Proc. Natl. Acad. Sci. USA 92 (1995)

or increased levels of CD23 or CD5 but, when combined withfixed T cells, augmented proliferation (Table 1) and loweredCD23 levels (Fig. 2).Next we asked whether ligation ofCD40 by gp39, the CD40L

on activated T-helper cell membranes (10), was sufficient toproduce any of these changes in B cells. We consistently foundthat CD40L in the form of a soluble fusion molecule (CD40L-CD8) induced proliferation and an increase in cell size andincreased surface CD44 and CD23 expression without alteringCD5 levels (Table 1 and Fig. 3). Therefore the signals gener-ated by crosslinking sIgM on the one hand and ligation ofCD40 on the other are sufficient to induce different activatedB-cell phenotypes.We considered the possibility that the CD5+ CD23 medium

cells on the one hand and the CD5- CD23 high B cells on theother might represent two populations arising from differentprecursors. To test this possibility we cultured B cells with acombination of CD40L and F(ab')2. The cells responded byincorporating even higher levels of thymidine (Table 1) thanwith either reagent alone. Flow cytometry revealed a singlepopulation of cells with little increase in either CD5 or CD23surface expression (Fig. 4), consistent with there being a singlepopulation of multipotential precursors. If some B cells couldrespond only to slg crosslinking, while others could respondonly to CD40 ligation, then two populations of activated cellsshould have been generated: CD5+ CD23 medium and CD5-CD23hi.

In a similar experiment, we cultured B cells with F(ab')2(rather than Fab) together with T cells. In this situation only11% of the B cells became CD5+ while cells cultured withF(ab')2 in the absence of T-helper cells were 56% CD5+. Thissupports the notion that simultaneous TD- and TI-2-typestimulation causes modulation of the induction of CD5. CD23

SSC/FSC CD5

Medium

T - SN

T fixed

T fixed + SN

CD23

I p1 2%

L2zz .. ^...i

i

8% <1%

SSC/FSC CD44

Medium

CD40L

CD5 CD23

Anti IgM

LPS

FIG. 3. Flow cytometry of small resting B cells after 42 hr ofstimulation. Top row, small resting splenic B cells after culture inmedium only; second row, cells after-coculture with CD40L; third row,cells after stimulation with F(ab')2 fragments of anti-IgM; fourth row,cells after coculture with LPS. The lefthand column shows forward(FSC) and side (SSC) light scatter. The second column shows CD44levels; the third, CD5; the fourth, CD23 levels. Negative controlstaining is also shown.

levels were low on all B cells in cultures with T cells, butbecause activated T-cell supernate is sufficient to down-modulate CD23 induction, this observation does not provideinformation on the possible effects of sIg ligation on theinduction of CD23.We wished to determine if CD40L and sIg-induced B-cell

signaling utilized precisely the same elements (7). To test thiswe took advantage of the fact that anti-IgM activation can beblocked by the immunosuppressants CsA and FK520 (16).These agents have many effects, but their ability to immuno-suppress T cells is contingent on their binding to one of severalproteins known collectively as immunophilins (16, 17). Theresulting complex then inhibits the Ser/Thr phosphatase cal-

Anti IgM

SSC/FSC

LOc]

FIG. 2. Flow cytometry analysis of B cells. Top row, small restingsplenic B cells after coculture for 2.5 days without stimulation; secondrow, B cells in activated T-cell supernate (SN) alone; third row, cellsafter culture with activated and fixed T-helper 2 CDC35 cells; fourthrow, cells in the combination of supemate and fixed T cells. Thelefthand column shows forward (FSC) and side (SSC) light scatter.The second column shows levels of CD5; the third, CD23.

CD40L Anti IgM+ CD40L

13% 7%

CD23FIG. 4. Effects of stimulation of B cells with a combination of

CD40L and F(ab')2 for 40 hr. Two-color flow cytometry shown ascontour diagrams. First column, cells stimulated with F(ab')2 ofanti-IgM; second column, B cells after culture with CD40L; thirdcolumn, cytographs of B cells cultured with both stimuli. The top rowshows forward (FSC) and side (SSC) scatter. In the second row the xaxis shows CD23 staining (fluorescein isothiocyanate) and the y axisshows CD5 (phycoerythrin). The quadrants were drawn on control(unstained) cells. The percentage of cells in each quadrant is shown.In this experiment 20% of the untreated B cells were CD5+ afterculture in medium alone. One of two similar experiments is shown.

3350 Immunology: Wortis et aL

k

44%I

Proc. Natl. Acad. Sci. USA 92 (1995) 3351

cineurin (18). The phosphatase activity of calcineurin is re-quired for the transport into the nucleus of several transcrip-tion factors, most particularly NFAT, required for expressionof IL-2 and IL-4 (19, 20). Although not characterized in suchdetail, in B cells the actions of these immunosuppressantsappear to involve similar mechanisms. Fig. 5 shows that CsAand FK520 blocked anti-IgM-induced thymidine incorporationbut that CD40L-induced proliferation was relatively insensi-tive to these drugs.To determine if this immunosuppression required binding to

an immunophilin we performed an experiment using theFK520 analogue L685,818. This molecule behaves as an an-tagonist of FK520 by competing for binding sites on FK506binding protein (FKBP) forming a complex that does notinhibit calcineurin (21). Unlike FK520, CsA binds to cyclo-philin molecules, immunophilins that do not bind L685,818(22). Thus, if activation is blocked by binding of drug toimmunophilins, then CsA, but not FK520, should suppress Bcells in the presence of L685,818. We observed that in thepresence of L685,818 CsA blocked the thymidine uptakeinduced by anti-IgM but FK520 did not (Fig. 5).As a second test of possible differences in the signal

pathways activated by slg and CD40 we asked whether additionof Sp-cAMPS, a potent membrane-permeablogue (23), would differentially affect activ;Concentrations of Sp-cAMPS that inhibited ation had the effect of augmenting CD40L-med(Fig. 6).

A

E

0

0

B

E

._

0

0

100_80-

60-

40-

20- ----- ---- -OA- m

0-Medium 1 3 1CsA ng/ml

Medium 0.1 0.3 1FK520 ng/mi

Anti Ig >< CD40L Anti Ig + L685 -4

FIG. 5. (A) Effect of CsA on proliferation of B ccstimulation for 42 hr. Also shown are the effects of Csof 300 ng of L685,818 per ml. Control thymidine incX 10-3) values: anti-IgM, 55.5; CD40L, 31.5; anti-Ig:47.7; CD40L plus L615,818, 24.8; medium (no smedium plus L615,818 (no stimulation), 1.2. Oneexperiments is shown. (B) Effect of FK520 on prolifiassessed after stimulation for 42 hr. Also shown aFKS20 in the presence of 300 ng of L685,818 per ml. I

expressed as a percentage of the values obtained iFK520. Control thymidine incorporation (cpm x 1(

IgM, 58.9; CD40L, 35.8; anti-IgM plus L615,818, 4L615,818, 30.1; medium (no stimulation), 1.4; mediu(no stimulation), 1.4. One of three similar experime

E.0a)214-

140-

120-

100=

80-

60-

40- \ _20

0-Medium 3 10 30

Sp-cAMPS uM100 200

Anti Ig-1 Anti Ig-2 s Anti Ig-3I CD40L-1 CD40L-2 X- CD40L-3

FIG. 6. Effect of the cAMP agonist Sp-cAMPS on B-cell incor-poration of thymidine as measured after 42 hr of stimulation. Resultsshown are from three separate experiments, each expressed as apercent of the values obtained in the absence of Sp-cAMPS. Loga-rithmic mean control (x 10-3) values for these three experiments:anti-IgM, 99, 25, 49; CD40L, 44, 43, 49; medium (no stimulation), 3,3, 3.

DISCUSSION

Ple cAMP ana- Klaus (2) showed that CsA inhibits B-cell production Qfation (24, 25). antibody in response to TI-2 but not TI-1 or TD antigens. Thenti-IgM activa- existence of alternative B-cell signal pathways was proposediated activation but did not exclude the possibility of two progenitors. Our

observations on the combined effects of minimal TI-2 and TDstimuli are also most simply explained as due to individual cellsresponding via different signal pathways and argue against theidea of separate precursors for TI-2 and TD responses. How-ever, we cannot rule out the formal possibility that there aretwo responsive B-cell populations and that activated TI-2responders cause a down-regulation of CD23 on the TDresponders while at the same time the activated TD responderscause a down-regulation of CD5 on the TI-2 responders.

Considerable evidence has accumulated indicating that thesignal pathways activated by minimal TI-2 and TD stimuli aredifferent. The Klaus experiments mentioned above are im-portant, as are other more recent direct tests. First, it was

30 100 established that T-cell-dependent activation did not require sIgcrosslinking (9), although in itself this did not show that thesignal pathways were different. Then it was shown that theactivation through membrane immunoglobulin was more sen-sitive to the depletion of PKC by phorbol esters than thatinduced by fixed, activated T-helper cells (12, 25). Experimentsto demonstrate the relative dependency of anti-Ig and T-helpermembranes on PKC and PKA enzymes have generated con-tradictory results (12, 25-27). The sIg response appears to bePKC dependent and PKA independent, while the CD40 re-sponse is PKC independent. In any case, since these inhibitors

1 are only relatively specific, the interpretation of these results3 10 should be viewed with caution. Overall, the evidence generally

favors a dichotomy between the TD and TI induction pathwaysCD40L + L685 and our own experiments provide additional support for this

zlls assessed after view.A in the presence Agonists of cAMP down-modulate sIgM-mediated activa-:orporation (cpm tion (12) and we found that Sp-cAMPs had a similar effect.M plus L615,818, Two groups reported an increase in intracellular cAMP levelsstimulation), 1.3; in response to activated T-helper cells (24, 25), but this is notof three similar a uniform finding (27). IfcAMP is increased by CD40 ligation,eration of B cells and if cAMP dampens the proliferation induced by anti-IgM,Ire the effects of then the total proliferation induced by CD40L and anti-IgM inResults shown are combination might be expected to be no greater than that

0-3) values: anti- induced by either agent alone. Therefore, the augmented16.4; CD40L plus proliferation following combined sIgM and CD40 stimulationim plus L615,818 suggests crosstalk between the sIgM and CD40 signal path--nts is shown. ways.

Immunology: Wortis et at

Proc. Natl. Acad Sci USA 92 (1995)

Previous reports indicate that ligation of CD40 in mice (28)and humans (29) does not generate an immediate Ca2+ flux asdoes slg ligation. Importantly, because L685,818 blocks therotamase activity of FKBP, but does not inhibit B-cell activa-tion, rotamase activity is not required for B-cell activation.Although CsA and FK520 only partially blocked CD40-mediated activation, this partial suppression is likely to be ofphysiological significance because the effects of FK520 but notCsA were reversed by L685,818 treatment. Our experimentswith CsA and FK520, drugs that suppress Ca2+-dependentsignaling (17), indicate that signaling via CD40 may not requireutilization of such a pathway. Possible counter evidence for thisconclusion is a report that CD40-mediated induction of NF-ATp expression is completely blocked by CsA treatment (30).However, there is as yet no evidence that NF-ATp is requiredfor the activation of B cells. It is also reasonable to considerthat an overlap or redundancy of signaling intermediates isutilized after stimulation of CD40 or IgM (31).Our results establish that the minimum activating inducers

are sufficient to induce differences in B-cell phenotypes. Thesignificance of the induction of CD5 expression after TI-2activation of B cells is not known. This molecule is found onall mature T cells and a limited population of B cells, theso-called B-la subset (32), and appears to be associated withthe T-cell receptor (TCR) (33) and B-cell receptor (34). Yet,a recent report suggests that mice expressing a CD5 deletionmutant had no defect in immune responses (35), although theauthors raised the possibility that lack of CD5 expression mightincrease thymocyte responses to TCR stimulation. The originof CD5+ B cells is a matter of controversy; some investigatorsbelieve that they derive from a precommitted pool of fetal/neonatal precursors (32). Others, including ourselves, believethat any naive [newly differentiated, "B-0" (36)] B cell canbecome CD5+ as a consequence of sIgM ligation (37, 38).CD23, the low-affinity receptor for the Fc of IgE (Fc£RII),

is a 45-kDa surface glycoprotein found on hematopoietic andfollicular dendritic cells (39) but limited to the B-2 populationof B cells (40). Up-regulation of CD23 occurs in mice andhumans following CD40 ligation by specific antibody orligand (28, 41-43). Although CD23 is up-regulated onpreviously activated B cells by IL-4, it is not expressed onrecently activated B cells within germinal centers (39),perhaps as a consequence of IL-6 activity (15). Interestingly,IL-4, an inducer of CD23, suppresses CD5 (44-47), and, asshown here, CD40 ligation, another inducer of high levels ofCD23, also down-modulates the induction of CD5 by anti-IgM. Thus the reciprocal induction of CD5 and CD23 on Bcells appears to be a general characteristic of normal B-cellsignaling.

Previously it was suggested that a generally held view ofB-cell activation, that antigen-mediated slg crosslinking is anecessary first step for TD activation (48), may not be true (9).We now propose that the quality of the initial activation eventmay commit a B cell to a particular differentiation pathway anda distinct set of functions.

We acknowledge with thanks the experimental contributions ofSimon Anderson, Dana Schmidt, Robert Seward, Damien Slater, andCamellia Symonowicz. This research was supported, in part, byNational Institutes of Health Grants P01 A133180, R01 A115803,lF32AI0918301, and R01 AI24303.

1. Sharon, R., McMaster, P. R. B., Kask, A. M., Owens, J. D. & Paul,W. E. (1975) J. Immunol. 114, 1585-1589.

2. Klaus, G. G. B. (1988) Transplantation 46, Suppl., 11-14.3. Mongini, P. K. A., Paul, W. E. & Metcalf, E. S. (1982) J. Exp. Med.

155, 884-902.4. Umetsu, D. T., Chapman-Alexander, J. M. & Thorbecke, G. J. (1979)

J. Immunol. 123, 396-404.5. Forum, M. S. & Pure, E. (1991) J. Exp. Med. 173, 687-697.

6. Mosier, D. E., Mond, J. J. & Goldings, E. A. (1977) J. Immunol. 119,1874-1878.

7. Cambier, J. C., Pleiman, C. M. & Clark, M. R. (1994) Annu. Rev.Immunol. 12, 457-486.

8. Parker, D. C., Wadsworth, D. C. & Schneider, G. B. (1980) J. Exp.Med. 152, 138-151.

9. Tony, H. P., Phillips, N. E. & Parker, D. C. (1985) J. Exp. Med. 162,1695-1708.

10. Armitage, R. J., Fanslow, W. C., Strockbine, L., Sato, T. A., Clifford,K. N., Macduff, B. M., Anderson, D. M., Gimpel, S. D., Davis-Smith,T., Maliszewski, C. R., Clark, E. A., Smith, C. A., Grabstein, K. H.,Cosman, D. & Spriggs, M. K. (1992) Nature (London) 357, 80-82.

11. Tony, H. P. & Parker, D. C. (1985) J. Exp. Med. 161, 223-241.12. Kawakami, K. & Parker, D. C. (1993) Eur. J. Immunol. 23, 77-84.13. Lane, P., Brocker, T., Hubele, S., Padovan, E., Lanzavecchia, A. &

McConnell, F. (1993) J. Exp. Med. 177, 1209-1213.14. Burkly, L. C. & Wortis, H. H. (1985) J. Immunol. 135, 1577-1581.15. Cong, Y.-Z., Rabin, E. & Wortis, H. H. (1991) Int. Immunol. 3,

467-476.16. Sigal, N. H. & Dumont, F. J. (1992)Annu. Rev. Immunol. 10, 519-560.17. Schreiber, S. L. & Crabtree, G. R. (1992) Immunol. Today 13, 136-142.18. Liu, J., Farmer, J. D., Jr., Lane, W. S., Friedman, J., Weissman, I. &

Schreiber, S. L. (1991) Cell 66, 807-815.19. Clipstone, N. A. & Crabtree, G. A. (1992) Nature (London) 357,

695-697.20. Jain, J., McCaffrey, P. G., Miner, Z., Kerppola, T. K., Lambert, J. N.,

Verdine, G. L., Curran, T. & Rao, A. (1993) Nature (London) 365,352-355.

21. Dumont, F. J., Staruch, M. J., Koprak, S. L., Siekierka, J. J., Lin, C. S.,Harrison, R., Sewell, T., Kindt, V. M., Beattie, T. R., Wyvratt, M. &Sigal, N. H. (1992) J. Exp. Med. 176, 751-760.

22. Bram, R. J., Hung, D. T., Martin, P. K., Schreiber, S. L. & Crabtree,G. R. (1993) Mol. Cell. Biol. 13, 4760-4769.

23. Van Haastert, P. J., Van Driel, R., Jastorff, B., Baraniak, J., Stec,W. J. & De Wit, R. J. (1984) J. Biol. Chem. 259, 10020-10024.

24. Pollok, K. E., O'Brien, V. O., Marshall, L., Olson, J. W., Noelle, R. J.& Snow, E. C. (1991) J. Immunol. 146, 1633-1641.

25. Kato, T., Kokuho, T., Tamura, T. & Nariuchi, H. (1994) J. Immunol.152, 2130-2138.

26. Lalmanach-Girard, A. C., Chiles, T. C., Parker, D. C. & Rothstein,T. L. (1993) J. Exp. Med. 177, 1215-1219.

27. Marshall, L. S., Shepherd, D. M., Ledbetter, J. A., Aruffo, A. &Noelle, R. J. (1994) J. Immunol. 152, 4816-4825.

28. Hasbold, J. & Klaus, G. G. (1994) Eur. J. Immunol. 24, 152-158.29. Knox, K. A. & Gordon, J. (1993) Eur. J. Immunol. 23, 2578-2584.30. Choi, M. S. K., Brines, R. D., Holman, M. J. & Klaus, G. G. B. (1994)

Immunity 1, 179-187.31. Klaus, G. G. B., Choi, M. S. K. & Holman, M. (1994) Eur. J. Immunol.

24, 3229-3232.32. Herzenberg, L. A. & Kantor, A. B. (1993) Immunol. Today 14,79-83.33. Beyers, A. D., Spruyt, L. L. & Williams, A. F. (1992) Proc. Natl. Acad.

Sci. USA 89, 2945-2949.34. Lankester, A. C., van Schijndel, G. M. W., Cordell, J. L., van Noesel,

C. J. M. & van Lier, R. A. W. (1994) Eur. J. Immunol. 24, 812-816.35. Tarakhovsky, A., Muller, W. & Rajewsky, K (1994) Eur. J. Immunol.

24, 1678-1684.36. Haughton, G., Arnold, L. W., Whitmore, A. C. & Clarke, S. H. (1993)

Immunol. Today 14, 84.37. Arnold, L. W., Pennell, C. A., McCray, S. K. & Clarke, S. H. (1994)

J. Eap. Med. 179,1585-1596.38. Teutsch, M., Higer, M., Wang, D. & Wortis, H. H. (1995) Int.

Immunol. 7, 381-392.39. Conrad, D. H. (1990) Annu. Rev. Immunol. 8, 623-645.40. Waldschmidt, T. J., Kroese, F. G. M., Tygrett, L. T., Conrad, D. H. &

Lynch, R. G. (1991) Int. Immunol. 3, 305-315.41. Saeland, S., Duvert, V., Moreau, I. & Banchereau, J. (1993)J. Exp.

Med. 178, 113-120.42. Katira, A., Knox, K. A., Finney, M., Michell, R. H., Wakelam, M. &

Gordon, J. (1993) Clin. Exp. Immunol. 92, 347-352.43. Maliszewski, C. R., Grabstein, K., Fanslow, W. C., Armitage, R.,

Spriggs, M. K. & Sato, T. A. (1993) Eur. J. Immunol. 23, 1044-1049.44. Hidaka, T., Kitani, A., Hara, M., Harigai, M., Suzuki, K., Kawaguchi,

Y., Ishizuka, T., Kawagoe, M. & Nakamura, H. (1992) Clin. Exp.Immunol. 89, 223-229.

45. Jyonouchi, H., Voss, R. M. & Good, R. A. (1990) Cell. Immunol. 130,66-78.

46. Freedman, A. S., Freeman, G., Whitman, J., Segil, J., Daley, J.,Levine, H. & Nadler, L. M. (1989) Eur. J. Immunol. 19, 849-855.

47. Zupo, S., Dono, M., Massara, R., Taborelli, G., Chiorazzi, N. &Ferrarini, M. (1994) Eur. 1. Immunol. 24, 1426-1433.

48. Abbas, A. K., Lichtman, A. H. & Pober, J. S. (1994) Cellular andMolecular Immunology (Saunders, Philadelphia).

3352 Immunology: Wortis et at