- Email: [email protected]
A polyclonal model for B-cell tolerance
CELLULAR
138,404-4 12 ( 199 1)
IMMUNOLOGY
A Polyclonal Model for B-Cell Tolerance II. Linkage between Signaling of B-Cell Egress from GO, Class II Upregulation and Unresponsiveness’ GARVIN
L. WARNER,
ARTI GAUR, AND DAVID W.
SCOTT
Immunology Division, University of Rochester Cancer Center, 601 Elmwood Avenue, Rochester, New York 14642 Received May 22, 1991; acceptedJuly 8, 1991 Overnight exposure of adult splenic B cells to anti-Ig, a surrogate for antigen/tolerogen, can result in a hyporesponsive state in terms of antibody synthesis. Since B cells treated with either intact of F(ab’), fragments of anti-Ig will exit the Go phase of the cell cycle and enter G, or S, respectively, we examined which steps in B-cell activation were required for this form of hyporesponsiveness. We found that B-cell hyporesponsiveness could be induced under conditions leading to either abortive or productive B-cell cycle progression, depending on the immunogenic challenge employed. Thus, PMA + ionomycin, concanavalin A, PMA alone, or ionomycin alone induced hyporesponsiveness. Each of these reagents is able to drive B-cell exit from Go into Gi and cause class II hyperexpression. We next examined the effect of cyclosporin A (CSA), a reagent that blocks anti-Ig but not by PMA-induced class II hyperexpression. Interestingly, CSA only interfered with the induction of B-cell hyporesponsiveness with anti-Ig. These results suggest that upregulation of MHC class II may be coincident with a CSA-sensitive tolerance pathway in B cells stimulated by anti-Ig. Finally, IL-4 pretreatment was found to ablate hyporesponsiveness induced by either intact anti-Ig or PMA. These results parallel the Fc-dependent induction of hyporesponsiveness reported earlier (G. Warner and D. W. Scott, J. Immunol. 146,2 185, 1991). We propose that crosslinking of surface Ig, leading to cell cycle progression out of Go as well as class II hyperexpression, in the absence of a cognate T cell signal, leads to B-cell hyporesponsiveness. 0 199 I Academic
Press, Inc.
INTRODUCTION The initial steps in both B-cell activation and tolerance are assumed to involve ligation of surface immunoglobulin (sIg)’ receptors by antigen and the activation of a cascadeof biochemical eventsthat have been thoroughly investigated ( 1,2). However, the requisite stepsthat differentiate between an immunogenic signal for B-cell antibody formation versus tolerance remain obscure. We previously reported a model (3) in which crosslinking of sIg induces B-cell hyporesponsiveness (anergy) as measured by ’ This work was supported in part by research grants from the USPHS (NIH ROl CA41 363 and AI26291) and an Institutional BRSG award to D.W.S. and by NIH Training Grant T32 CA 09363 (G.L.W.). This is publication No. 62 from the Immunology Division of the University of Rochester Cancer Center. ’ Abbreviations used: FcyR, FcyII receptor; FL-AFC, anti-fluorescein antibody-forming cell; FL-BA, FLcoupled to Brucella abortus; HEM, Hepes-buffered minimal essential medium; PI, phosphatidyl inositol; PKC, protein kinase C; sIg, surface immunoglobulin. 404 0008-8749/9 1 $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproductmn in any form resewed
AN ANTI-Ig
MODEL
FOR TOLERANCE
405
antibody production in response to challenge with either LPS or FL-Bruce/la abortus (FL-&t). The induction of anergy to FL-&4 required intact anti-Ig, indicating that an interaction with the Fey receptor (FcRy) was necessary to induce hyporesponsiveness to this antigen. In contrast, the inhibitory effect of anti-Ig pretreatment on the subsequent LPS response was FcR-independent (3). This hyporesponsive state, once attained, was not affected (i.e., could not be reversed) by stimulation with IFN-7, IL-2, TNFa, or IL-4 (3-5). However, pretreatment with IL-4 caused B cells to become refractory to the induction of FcRy-dependent hyporesponsiveness, presumably by driving the B cells to a state in which they are not capable of receiving negative signals mediated by sIg and FcRy (6). The effect of IL4 in inhibiting Fc-dependent negative signaling is consistent with previous work examining the effect of the FcRy on anti-Ig-induced B-cell proliferation (7, 8). It is important to note that B cells treated with intact anti-Ig (capable of crosslinking Ig receptors and FcRy) are abortively activated. That is, they egress from Go into G,, but fail to progress into the S phase of the cell cycle (2). However, such cells continue to upregulate their expression of MHC class II molecules, as previously reported for abortively activated B cells treated with anti-Ig plus cholera toxin (CT). Interestingly, CT has no effect on the delivery of a negative signal by anti-Ig (9). In the current paper, we have more clearly defined the nature of the transmembrane and intracellular signal(s) required to deliver a negative differentiative signal in normal B cells. We have examined the ability of several agents that activate B cells to exit Go for their ability to induce B-cell hyporesponsiveness. Moreover, we have tested whether cyclosporin A (CSA), which blocks B-cell activation events, can interfere with tolerance induction in this system. Our results demonstrate that abortive activation of B cells, up to and including class II hyperexpression, can lead to subsequent inhibition of immunoglobulin synthesis stimulated by antigen (FL-BA) or LPS if not coupled with T-cell help. MATERIALS
AND METHODS
Medium and immunologic reagents. RPMI- 1640 and Hepes-buffered minimal essential medium (HEM), obtained in powder form from GIBCO (Grand Island, NY) were tested for endotoxin using the Limuhs amoebocyte assay (Sigma, St. Louis, MO) and found to contain ~0.8 ng endotoxin (LPS)/ml. Fetal bovine serum (FBS) was obtained from J. R. Scientific (lot No. 9550, Woodland, CA) and contained less than 0.025 rig/ml of LPS. Concanavalin A, phorbol myristate acetate, and ionomycin were all obtained from Sigma. Intact and F(ab’h fragments of rabbit anti-mouse F(ab’),, referred to as anti-Ig, were prepared as described earlier (10) or were obtained from Jackson Immunoresearch (West Grove, PA). Recombinant IL-4 was generously provided by Dr. Steven Gillis (Immunex, Seattle, WA). Cyclosporin A was the kind gift of Dr. Allan Hess (Johns Hopkins University, Baltimore, MD). CSA was dissolved in ethanol and stored at 4°C. When cells were pretreated with CSA or IL-4, they were incubated for 2- 18 hr with each reagent, respectively, prior to adding anti-Ig in the standard protocol described below. Preparation and incubation ofB cells. Splenocytes were obtained from adult C57Bl/ 6 X DBA/2 F1 (BDF,) (Jackson Laboratories, Bar Harbor, ME) male mice and B cells prepared as described earlier (10). Briefly, T cells were removed by incubation of splenocytes (l-4 X 10’ cells/ml in HEM, 4% FBS) with a cocktail of anti-T-cell mono-
406
WARNER,
GAUR,
AND
SCOTT
clonal antibodies followed by lysis with baby rabbit C’. These cells (routinely ~2% Thy 1+) were then centrifuged through 100% FBS and washed three times prior to initial culture. In our basic protocol, B cells (2.5 X 106/ml; 4 ml per 60 mm well) are pretreated with medium or an optimal dose (e.g., 10 pg/ml) of either F(ab’)* anti-Ig or IgG anti-Ig for up to 20 hr, followed by extensive washing and challenge in microtiter plates (1 X 106/ml; 0.2 ml per well) with FL-&t (0.002% final concentration) or LPS (10 pg/ml). The numbers of antibody-forming cells or total accumulated secreted antibody were analyzed by ELISA protocols described below. We usually measured anti-fluorescein responses (see below) to reflect the conventional B-cell repertoire, which can be stimulated by LPS or specific antigen, although similar results were obtained in terms of anti-TNP or total IgM secretion (data not shown). Proliferation in response to LPS was quantitated by pulsing with 0.5 &i [3H]thymidine on Days 2-4 and harvesting 6 hr later for scintillation counting. ELISA spot and supernatant assay for detecting antibody forming cells or accumulated antibody. This assay(10) is based on one initially described by Sedgwick and Holt (14). Briefly, polystyrene ELISA plates (Immulon, Dynatech) were coated with FL-coupled gelatin, 10 pg/ml, Tris-HCl, pH 9.5, overnight, (4”(Z), and blocked with 2% BSA prior to use. Appropriately diluted cells (in Hepes-buffered MEM, 2% FBS) were added to individual wells and incubated at 37°C for 2-4 hr. Plates were washed and 100 ~1 of a 1:500-1:lOOO dilution (in PBS, 0.25% BSA, 0.05% Tween 20, 0.01 rnA4 EDTA, 0.1% NaN3) of anti-IgM-alkaline phosphatase (Southern Biotechnology, Birmingham, AL) was added to each well. Plates were incubated overnight at 4°C and washed, and a 4: 1 mixture of substrate [ 1 pg/ml 5-bromo-4-chloro-3-indolylyphosphate p-toluidine salt (Sigma) in AMP Buffer, pH 10.251 and 3% agarose was added to each well. Plates were incubated for 30-60 min, 37°C during which time blue spots developed, each spot representing an anti-fluorescein antibody-forming cell (FL-AFC), which reflects the immune status of the B cells. In some experiments, the total accumulated antibody was measured as follows. Supematants from 4-day microcultures were collected and aliquots tested on plates coated as above with FLgelatin; standard curves with affinity-purified monoclonal IgM anti-FL validated that this assaywas linear in the antibody concentration range tested. After overnight incubation at 4°C AP-coupled heavy-chain specific anti-p or anti-y was added; agarose was omitted in the final substratestep when supematants were assayed.The absorbance was measured at 405 nm using a Titertek Multiscan ELISA reader and the data were presented either as the OD f standard deviation or as percentage control responses (control = medium followed by appropriate challenge). In the latter case,the deviations were less than 15% and for simplicity are not shown. RESULTS Can the induction of unresponsivenessby anti-Ig be mimicked by concanavalin A, PMA, or ionomycin? As described earlier (3), we have used anti-Ig as a surrogate for tolerogen in a polyclonal system to analyze the intracellular pathways leading to Bcell tolerance. Adult splenic B cells cultured overnight with anti-Ig show an Fc-dependent unresponsivenessto subsequent challenge with specific Ag, FL-Bruce/la abortus. Tolerance, which affectsdifferentiation to antibody synthesis but not B-cell clonal proliferation (3), is also observed when these B cells are challenged with LPS and measured for anti-hapten antibody and total Ig synthesis. Thus, these cells are viable,
AN ANTI-Ig
MODEL
407
FOR TOLERANCE
but anergic in terms of antibody formation (3). The kinetics of tolerance induction (reflecting a minimum 4- to 8-hr exposure to anti-Ig) suggested that early activation events up to and including class II transcription might be required for the tolerogenic process. Since B cells can also be partially activated with concanavalin A, PMA, or ionomycin (1, 2) or with anti-Ig plus CT (9) and still show nearly normal increases in class II upregulation, we asked whether these conditions also led to B-cell hyporesponsiveness. In Fig. I, we demonstrate that Con A induces hyporesponsiveness to either FL-&4 or LPS challenge, equivalent to intact anti-Ig. Figure 2 shows dose-response curves for both PMA and ionomycin in terms of their abilities to independently induce B-cell hyporesponsiveness to challenge with LPS. These data demonstrate that the activation of PKC or the elevation of intracellular Ca++ concentration, by themselves, are sufficient to induce B-cell hyporesponsiveness to subsequent challenge with the polyclonal activator LPS. Of interest is the fact that only PMA, but not ionomycin, downregulates antibody formation in response to the antigen-specific challenge with FL-&l. Thus, PMA (or Con A) appears to mimic the Fc-dependent induction of hyporesponsiveness seen with intact anti-Ig, whereas ionomycin causes a reduction in antibody formation reminiscent of the Fc-independent activity of anti-Ig on LPS responsiveness (3). Importantly, all of these reagents cause MHC class II upregulation with the following hierarchy of decreasing effectiveness: anti-Ig = Con A > PMA > ionomycin (Warner, unpublished and Ref. (10)).
Cyclosporin A prevents anti-Ig-driven class II upregulation and interferes with Bcell tolerance induction. CSA has been shown to interfere with the anti-Ig-stimulated increase in class II expression in B cells (10). To test the role of class II upregulation in the tolerance process, splenic B cells were incubated with anti-Ig + graded doses of CSA, washed, and then challenged with LPS. As shown in Fig. 3, FL-specific IgM titers show that anti-Ig induces B-cell hyporesponsiveness and that CSA prevents this
FL-BA
.L
I 1: 0’.
t
ConA
+
IgGa-lg
I i
I. 0
0.5 1.0 2.0
100
0.5
1.0 2.0
10
ConA (pg/m/) FIG. I. Concanavalin A (Con A) induces B-cell hyporesponsiveness similar to intact anti-lg. Adult splenic B cells were treated overnight with the indicated concentrations of Con A (x-axis, squares) or intact anti-Ig (10 kg/ml, triangles), washed, and challenged with either FL-E,4 or LPS. The number of IgM anti-FLproducing cells (AFCs) was determined on Day 4 using an ELISA spot assay and are expressed as the number of FL-specific AFCs per culture well. The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
408
WARNER,
GAUR,
AND
SCOTT
=I
PMA
‘“NO_MYC’N
0 0
0.1 1
0.3 3
1 10
3 30
0 0
0.1 1
0.3 3
1 10
3 lONO 30 PMA (nM)
CONCENTRA T/ON PMA//OIVOMYC/N FIG. 2. Ionomycin (squares) or PMA (circles) induce B-cell hyporesponsiveness depending on whether the B cells are subsequently challenged with either with FL-BA (left) or LPS (right). Splenic B cells were treated with the indicated amount (x-axis) of either ionomycin (0.1-3 PM) or PMA (l-30 nM) for 4 hr, washed extensively, and challenged with FL.-BA or LPS. Four days after challenge, supematants were examined for the presence of anti-FL by ELISA. Results are presented as the ODm5 “,,, reflecting IgM anti-FL. The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
in a dose-dependent manner. CSA has an inhibitory effect by itself but the ability of anti-Ig to induce hyporesponsiveness is decreased relative to the CSA alone control. Indeed, in the absence of CSA, anti-Ig caused a 92% decrease in the anti-FL response, whereas with 100 rig/ml CSA, there was only an 18% decrease relative to the CSAonly control. Independent experiments confirmed that CSA blocked the anti-Ig (but not the IL-4) stimulated increase in class II expression (Warner, unpublished), as reported by Hawrylowicz and Klaus (10). Interestingly, CSA can prevent the hyperexpression of class II induced by anti-Ig, but not that stimulated by PMA (10). We, therefore, tested the activity of CSA on tolerance induction by these (abortive) B-cell activators. The data in Fig. 4 confirm that downregulation of the LPS response can be elicited by both PMA and anti-Ig, but that CSA only prevents the induction of B-cell hyporesponsiveness by anti-Ig. This is not surprising since class II upregulation by PMA is also CSA-insensitive (see Ref. ( 10)). Effect of IL-4 on tolerance induction elicited by B-ceil activators. Previous studies have indicated that IL-4 can overcome the negative signals delivered by crosslinking sIg to Fey receptors (7, 8). Since both Con A and PMA elicit a form of hyporesponsiveness to FLBA challenge akin to intact IgG anti-Ig, we investigated the effect of IL4 on tolerance induction in our model. B cells were precultured with IL-4 at 2000 U/ml overnight and then exposed to intact anti-Ig or PMA before washing and challenge with FL-BA. The results in Fig. 5 confirm that IL-4 pretreatment does protect B cells from the induction of hyporesponsiveness by intact anti-Ig and, moreover, they demonstrate that the inhibitory effect of PMA can be overriden by IL-4. However, this cytokine has no effect on the induction of hyporesponsiveness to LPS challenge (data not shown).
AN ANTI-lg
MODEL
409
FOR TOLERANCE
2.0
aI
-
t
10
T 1II
n q
NONE 10 pg/ml a-lg
100
CYCLOSPORlN A (nglmi) FIG. 3. Cyclosporin A (CSA) inhibits tolerance induction by anti-lg. Splenic B cells were treated with the indicated concentrations of CSA for 2 hr prior to exposure to 10 pg/ml anti-Ig (a-Ig) or medium (none) for an additional 18 hr. Supematants were assayed for the presence of IgM anti-FL. When compared to the CSA-only controls (black bars), the percentage unresponsiveness observed was: 0 rig/ml CSA = 92%; I ng/ ml CSA = 52%; IO rig/ml = 32%: 100 rig/ml = 18%. Similar results were obtained regardless of whether B cells were challenged with LPS or FL-B,4 (data not shown). The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
Is (abortive) B-cell activation a negative signal,for dljjkrentiation to Ig synthesis.? Table 1 summarizes the conditions we have tested herein and elsewhere (3, 9) that lead to cell cycle entry, class II upregulation, and the induction of tolerance in this model. It is noteworthy that conditions which lead to cell cycle progression (G, > G,), whether it is productive or abortive, can lead to hyporesponsiveness. In fact, entry into S is not necessary for the induction of tolerance. Rather, abortive B-cell activation in the absence of T-cell help consistently leads to B-cell hyporesponsiveness as well as class II antigen hyperexpression. These results are summarized in Table 1. DISCUSSION In a previous paper (3) we demonstrated that the induction of B-cell hyporesponsiveness (anergy) via crosslinking of sIg occurred after a 4- to S-hr incubation period, during which time many of the early events in B-cell signaling would have occurred and the cells would have egressed out of Go and into cycle. Originally, we intended to activate B cells without crosslinking sIg by using Con A, PMA, or ionomycin, known to abortively drive cell cycle entry (I, 2) and then reinvestigate the kinetics of tolerance induction via anti-Ig. Our hypothesis was that the time required in previous experiments reflected the need to progress out of Go and into a critical phase of G, for negatively signaling (cf. Ref. (11)). However, B-cell hyporesponsiveness was induced by incubation with Con A, PMA, or in some cases, ionomycin in the absenceof an anti-Ig signal. Therefore, crosslinking sIg was not required for negative signaling; rather it appeared that the events mimicked by these reagents, which drive B-cell entry into the cell cycle, were sufficient for hyporesponsiveness. Since we previously found that
410
WARNER,
NONE
1 nM PMA
GAUR,
AND
10 nM PMA
SCOTT
10 &ml
a-lg
INDUCING AGENT FIG. 4. Cyclosporin A (CSA) fails to inhibit the induction of B-cell hyporesponsiveness induced by PMA. Splenic B cells were treated with the indicated doses of CSA for 2 hr prior to the addition of the indicated (x-axis) amount of either PMA or intact anti-&. After an additional 18 hr the cells were washed and challenged with LPS (10 &ml). Four days later supematants were analyzed for the presence of anti-FL by ELISA and are presented as ODw5 nm reflecting IgM anti-FL. The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
cholera toxin prevented positive signaling but did not prevent B-cell tolerance (9), these data imply that abortive entry of B cells into the cycle might be coincident with a tolerance pathway. It should be noted that Con A, PMA (16), and ionomycin are also able to induce hyporesponsivenessin terms of LPS-induced differentiation. However, only the former two reagents downregulate antibody formation elicited by FL-&t, which requires an interaction of FcyR crosslinked to sIg (3). Furthermore, CSA was able to block the induction of B-cell tolerance induced by anti-Ig but not hyporesponsiveness induced by PMA. Taken together, these data demonstrate that the negative signaling pathway shareswith the positive signaling pathway sensitivity to CSA and that both the positive and the negative signaling pathways can be modulated by pharmacologic or physiologic activators of PKC. These data also imply that either transcriptional factors or regulatory motifs (affected by anti-Ig) in both class II and the Ig promoter/enhancer regions may share common features including the ability to act in truns (16). Efforts to identify their common features are underway in our laboratory and in others’ (16). FcyR:sIg interactions per se can downregulate B-cell function (7,8) presumably by causing a dissociation of the sIg complex from its G protein network (12), thus leading to an abortive activation of the phosphotidyl inositol cycle with a relatively normal rise in intracellular calcium (2). Interestingly, this processis reversed if the B cells are (pre)treated with IL-4 (3). Lazslo and Dickler (6) reported that FcyR function and association may be altered by pretreating B cells with IL-4. It has been reported that IL-4 does not cause protein kinase C (PKC) translocation to the plasma membrane, whereas anti-I-A and CAMP analogues induce the translocation of PKC to the nucleus rather than the plasma membrane (13, 14). Since PKC translocation may be important in T-cell help, it will be interesting to determine the effectsof anti-I-A and modulators of CAMP activity in our model.
AN ANTI-Ig
MODEL
rz
T 1. T
NONE
1 nM PMA
lMXRX’JG
411
FOR TOLERANCE
J
r
10 nM PMA 10 pg/ml u-lg
AGENT
FIG. 5. IL-4 pretreatment prevents the induction B-cell hyporesponsiveness to FL-R4 challenge induced by either PMA or intact anti-Ig. Splenic B cells were treated with 2000 U/ml of IL-4 for I8 hr prior to the addition of the indicated amounts of PMA or intact anti-Ig. Following an additional culture period of 24 hr. the cells were washed and challenged with FL-&l. Four days later supernatants were analyzed for the presence of anti-FL by ELISA. The results are presented as OD4e5 nmreflecting IgM anti-FL. The error bars represent the standard deviation of the mean and the results are representative of three separate experiment.
It is important to note that Hunziker et al. (15) recently reported that a B-cellspecific domain of the FcyR(I1) is phosphorylated by PKC after anti-Ig crosslinking except when intact IgG antibody is employed. This suggests that the dissociation of
TABLE Relationship
I
between Productive or Abortive B-Cell Entry into Cycle. Class II Hyperexpression, and the Induction of Hyporesponsiveness B-cell hyporesponsiveness”
Stimulus F(ab’), anti-Ig PMA + ionomycin IgG anti-Ig Concanavalin A PMA Ionomycin F(ab’)* anti-Ig plus CT F(ab’h or IgG anti-Ig plus CSA
Entry in cycle
DNA synthesis
MHC class II hyperexpression
FL-BA
LPS
Abortive Abortive Abortive Abortive Abortive
Yes Yes No No No No No
Yes Yes Yes Yes Yes Yes Yes
No Yes Yes Yes Yes No -
Yes Yes Yes Yes Yes Yes Yes
Abortive
No
No
No
No
Yes Yes
a Hyporesponsiveness upon challenge with either FL-B,4 or LPS, as described in text. Data are from the current paper as well as Refs. (3, 9).
412
WARNER,
GAUR,
AND
SCOTT
the G protein:sIg complex creates physical constraints on PKC substrates, including FcyR and the Ig complex. If IL-4 modulates PKC activity, this would explain why this lymphokine is effective at preventing the negative signal via intact anti-Ig especially with regard to the FL-&4 challenge. With these results in view, one can propose that under conditions of appropriate T-cell help, positive signaling progresses because PKC might not be available to act on the sIg complex at the membrane. An alternative, but not mutually exclusive, hypothesis is that PKC activity on FcyR allows for appropriate trafficking of the sIg complex and that tyrosine phosphorylation of IgMand &D-associated molecules ( 17-20) plays a balancing role in B-cell triggering. Further work defining the role of IL-4 and the availability of PKC, tyrosine kinase(s), and their substrates in these cells should shed light on this process. REFERENCES 1. Cambier, J., and Ransom, J. T., Annu. Rev. Immunol. 5, 175, 1987. 2. Klaus, G. G. B., Bijsterbosch, M., O’Garra, A., Harnett, M., and Rigley, K., Immunol. Rev. 99, 19, 1987. 3. Warner, G., and Scott, D. W., J. Immunol. 146, 2185, 1991. 4. Chace, J., and Scott, D. W., J. Immunol. 141, 3258, 1988. 5. Nossal, G. J. V., Annu. Rev. Immunol. 1, 33, 1983. 6. Laszlo, G., and Dickler, H. B., J. Immunol. 141, 3416, 1988. 7. O’Garra, A., Rigley, K., Holman, M., McLaughlin, J., and Klaus, G. G. B., Proc. Nuti. Acud. Sci. USA 84,6254, 1987. 8. Phillips, N., Gravel, K., Tumas, K., and Parker, D., J. Immunol. 141, 4243, 1988. 9. Warner, G., and Scott, D. W., J. Immunol. 143, 458, 1989. 10. Hawrylowicz, C., and Klaus, G. G. B., Immunology 53,703, 1988. Il. Scott, D. W., and Klinman, N., Immunol. Today 8, 105, 1987. 12. Rigley, K. P., Hamett, M., and Klaus, G. G. B., Eur. J. Immunol. 19, 481, 1989. 13. Justement, L., Chen, Z. Z., Harris, L., Ransom, J. T., Sandoval, V., Smith, C., Rennick, D., Roehm, N., and Cambier, J., J. Immunol. 137, 3664, 1986. 14. Cambier, J. C., Newell, M. K., Justement, L. B., McGuire, J. C., Leach, K. L., and Chen, Z. Z., Nature 327,629, 1987. 15. Hunziker, W., Koch, T., Whitney, J. A., and Mellman, I., Nature 345, 628, 1990. 16. Martenson, I-L., Iglesias, A., and Leanderson, T., Eur. J. Immunol. 19, 1497, 1990. 17. Gold, M. R., Law, D. A., and DeFranco, A. L., Nature 345, 8 10, 1990. 18. Campbell, M-A., and Sefton, B. M., EMBO J. 9, 2 125, 1990. 19. Chen, J., Stall, A. M., Herzenberg, L. A., and Herzenberg, L. A., EMBO J. 9, 2117, 1990. 20. Ales-Martinez, J. E., Martinez-A., C. Parkhouse, R. M. E., Pezzi, L., and Scott, D. W., Immunol. Today 12,201, 1991. 21. Sedgwick, J., and Holt, P. J., J. Immunol. Meth. 57, 301, 1983.
138,404-4 12 ( 199 1)
IMMUNOLOGY
A Polyclonal Model for B-Cell Tolerance II. Linkage between Signaling of B-Cell Egress from GO, Class II Upregulation and Unresponsiveness’ GARVIN
L. WARNER,
ARTI GAUR, AND DAVID W.
SCOTT
Immunology Division, University of Rochester Cancer Center, 601 Elmwood Avenue, Rochester, New York 14642 Received May 22, 1991; acceptedJuly 8, 1991 Overnight exposure of adult splenic B cells to anti-Ig, a surrogate for antigen/tolerogen, can result in a hyporesponsive state in terms of antibody synthesis. Since B cells treated with either intact of F(ab’), fragments of anti-Ig will exit the Go phase of the cell cycle and enter G, or S, respectively, we examined which steps in B-cell activation were required for this form of hyporesponsiveness. We found that B-cell hyporesponsiveness could be induced under conditions leading to either abortive or productive B-cell cycle progression, depending on the immunogenic challenge employed. Thus, PMA + ionomycin, concanavalin A, PMA alone, or ionomycin alone induced hyporesponsiveness. Each of these reagents is able to drive B-cell exit from Go into Gi and cause class II hyperexpression. We next examined the effect of cyclosporin A (CSA), a reagent that blocks anti-Ig but not by PMA-induced class II hyperexpression. Interestingly, CSA only interfered with the induction of B-cell hyporesponsiveness with anti-Ig. These results suggest that upregulation of MHC class II may be coincident with a CSA-sensitive tolerance pathway in B cells stimulated by anti-Ig. Finally, IL-4 pretreatment was found to ablate hyporesponsiveness induced by either intact anti-Ig or PMA. These results parallel the Fc-dependent induction of hyporesponsiveness reported earlier (G. Warner and D. W. Scott, J. Immunol. 146,2 185, 1991). We propose that crosslinking of surface Ig, leading to cell cycle progression out of Go as well as class II hyperexpression, in the absence of a cognate T cell signal, leads to B-cell hyporesponsiveness. 0 199 I Academic
Press, Inc.
INTRODUCTION The initial steps in both B-cell activation and tolerance are assumed to involve ligation of surface immunoglobulin (sIg)’ receptors by antigen and the activation of a cascadeof biochemical eventsthat have been thoroughly investigated ( 1,2). However, the requisite stepsthat differentiate between an immunogenic signal for B-cell antibody formation versus tolerance remain obscure. We previously reported a model (3) in which crosslinking of sIg induces B-cell hyporesponsiveness (anergy) as measured by ’ This work was supported in part by research grants from the USPHS (NIH ROl CA41 363 and AI26291) and an Institutional BRSG award to D.W.S. and by NIH Training Grant T32 CA 09363 (G.L.W.). This is publication No. 62 from the Immunology Division of the University of Rochester Cancer Center. ’ Abbreviations used: FcyR, FcyII receptor; FL-AFC, anti-fluorescein antibody-forming cell; FL-BA, FLcoupled to Brucella abortus; HEM, Hepes-buffered minimal essential medium; PI, phosphatidyl inositol; PKC, protein kinase C; sIg, surface immunoglobulin. 404 0008-8749/9 1 $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproductmn in any form resewed
AN ANTI-Ig
MODEL
FOR TOLERANCE
405
antibody production in response to challenge with either LPS or FL-Bruce/la abortus (FL-&t). The induction of anergy to FL-&4 required intact anti-Ig, indicating that an interaction with the Fey receptor (FcRy) was necessary to induce hyporesponsiveness to this antigen. In contrast, the inhibitory effect of anti-Ig pretreatment on the subsequent LPS response was FcR-independent (3). This hyporesponsive state, once attained, was not affected (i.e., could not be reversed) by stimulation with IFN-7, IL-2, TNFa, or IL-4 (3-5). However, pretreatment with IL-4 caused B cells to become refractory to the induction of FcRy-dependent hyporesponsiveness, presumably by driving the B cells to a state in which they are not capable of receiving negative signals mediated by sIg and FcRy (6). The effect of IL4 in inhibiting Fc-dependent negative signaling is consistent with previous work examining the effect of the FcRy on anti-Ig-induced B-cell proliferation (7, 8). It is important to note that B cells treated with intact anti-Ig (capable of crosslinking Ig receptors and FcRy) are abortively activated. That is, they egress from Go into G,, but fail to progress into the S phase of the cell cycle (2). However, such cells continue to upregulate their expression of MHC class II molecules, as previously reported for abortively activated B cells treated with anti-Ig plus cholera toxin (CT). Interestingly, CT has no effect on the delivery of a negative signal by anti-Ig (9). In the current paper, we have more clearly defined the nature of the transmembrane and intracellular signal(s) required to deliver a negative differentiative signal in normal B cells. We have examined the ability of several agents that activate B cells to exit Go for their ability to induce B-cell hyporesponsiveness. Moreover, we have tested whether cyclosporin A (CSA), which blocks B-cell activation events, can interfere with tolerance induction in this system. Our results demonstrate that abortive activation of B cells, up to and including class II hyperexpression, can lead to subsequent inhibition of immunoglobulin synthesis stimulated by antigen (FL-BA) or LPS if not coupled with T-cell help. MATERIALS
AND METHODS
Medium and immunologic reagents. RPMI- 1640 and Hepes-buffered minimal essential medium (HEM), obtained in powder form from GIBCO (Grand Island, NY) were tested for endotoxin using the Limuhs amoebocyte assay (Sigma, St. Louis, MO) and found to contain ~0.8 ng endotoxin (LPS)/ml. Fetal bovine serum (FBS) was obtained from J. R. Scientific (lot No. 9550, Woodland, CA) and contained less than 0.025 rig/ml of LPS. Concanavalin A, phorbol myristate acetate, and ionomycin were all obtained from Sigma. Intact and F(ab’h fragments of rabbit anti-mouse F(ab’),, referred to as anti-Ig, were prepared as described earlier (10) or were obtained from Jackson Immunoresearch (West Grove, PA). Recombinant IL-4 was generously provided by Dr. Steven Gillis (Immunex, Seattle, WA). Cyclosporin A was the kind gift of Dr. Allan Hess (Johns Hopkins University, Baltimore, MD). CSA was dissolved in ethanol and stored at 4°C. When cells were pretreated with CSA or IL-4, they were incubated for 2- 18 hr with each reagent, respectively, prior to adding anti-Ig in the standard protocol described below. Preparation and incubation ofB cells. Splenocytes were obtained from adult C57Bl/ 6 X DBA/2 F1 (BDF,) (Jackson Laboratories, Bar Harbor, ME) male mice and B cells prepared as described earlier (10). Briefly, T cells were removed by incubation of splenocytes (l-4 X 10’ cells/ml in HEM, 4% FBS) with a cocktail of anti-T-cell mono-
406
WARNER,
GAUR,
AND
SCOTT
clonal antibodies followed by lysis with baby rabbit C’. These cells (routinely ~2% Thy 1+) were then centrifuged through 100% FBS and washed three times prior to initial culture. In our basic protocol, B cells (2.5 X 106/ml; 4 ml per 60 mm well) are pretreated with medium or an optimal dose (e.g., 10 pg/ml) of either F(ab’)* anti-Ig or IgG anti-Ig for up to 20 hr, followed by extensive washing and challenge in microtiter plates (1 X 106/ml; 0.2 ml per well) with FL-&t (0.002% final concentration) or LPS (10 pg/ml). The numbers of antibody-forming cells or total accumulated secreted antibody were analyzed by ELISA protocols described below. We usually measured anti-fluorescein responses (see below) to reflect the conventional B-cell repertoire, which can be stimulated by LPS or specific antigen, although similar results were obtained in terms of anti-TNP or total IgM secretion (data not shown). Proliferation in response to LPS was quantitated by pulsing with 0.5 &i [3H]thymidine on Days 2-4 and harvesting 6 hr later for scintillation counting. ELISA spot and supernatant assay for detecting antibody forming cells or accumulated antibody. This assay(10) is based on one initially described by Sedgwick and Holt (14). Briefly, polystyrene ELISA plates (Immulon, Dynatech) were coated with FL-coupled gelatin, 10 pg/ml, Tris-HCl, pH 9.5, overnight, (4”(Z), and blocked with 2% BSA prior to use. Appropriately diluted cells (in Hepes-buffered MEM, 2% FBS) were added to individual wells and incubated at 37°C for 2-4 hr. Plates were washed and 100 ~1 of a 1:500-1:lOOO dilution (in PBS, 0.25% BSA, 0.05% Tween 20, 0.01 rnA4 EDTA, 0.1% NaN3) of anti-IgM-alkaline phosphatase (Southern Biotechnology, Birmingham, AL) was added to each well. Plates were incubated overnight at 4°C and washed, and a 4: 1 mixture of substrate [ 1 pg/ml 5-bromo-4-chloro-3-indolylyphosphate p-toluidine salt (Sigma) in AMP Buffer, pH 10.251 and 3% agarose was added to each well. Plates were incubated for 30-60 min, 37°C during which time blue spots developed, each spot representing an anti-fluorescein antibody-forming cell (FL-AFC), which reflects the immune status of the B cells. In some experiments, the total accumulated antibody was measured as follows. Supematants from 4-day microcultures were collected and aliquots tested on plates coated as above with FLgelatin; standard curves with affinity-purified monoclonal IgM anti-FL validated that this assaywas linear in the antibody concentration range tested. After overnight incubation at 4°C AP-coupled heavy-chain specific anti-p or anti-y was added; agarose was omitted in the final substratestep when supematants were assayed.The absorbance was measured at 405 nm using a Titertek Multiscan ELISA reader and the data were presented either as the OD f standard deviation or as percentage control responses (control = medium followed by appropriate challenge). In the latter case,the deviations were less than 15% and for simplicity are not shown. RESULTS Can the induction of unresponsivenessby anti-Ig be mimicked by concanavalin A, PMA, or ionomycin? As described earlier (3), we have used anti-Ig as a surrogate for tolerogen in a polyclonal system to analyze the intracellular pathways leading to Bcell tolerance. Adult splenic B cells cultured overnight with anti-Ig show an Fc-dependent unresponsivenessto subsequent challenge with specific Ag, FL-Bruce/la abortus. Tolerance, which affectsdifferentiation to antibody synthesis but not B-cell clonal proliferation (3), is also observed when these B cells are challenged with LPS and measured for anti-hapten antibody and total Ig synthesis. Thus, these cells are viable,
AN ANTI-Ig
MODEL
407
FOR TOLERANCE
but anergic in terms of antibody formation (3). The kinetics of tolerance induction (reflecting a minimum 4- to 8-hr exposure to anti-Ig) suggested that early activation events up to and including class II transcription might be required for the tolerogenic process. Since B cells can also be partially activated with concanavalin A, PMA, or ionomycin (1, 2) or with anti-Ig plus CT (9) and still show nearly normal increases in class II upregulation, we asked whether these conditions also led to B-cell hyporesponsiveness. In Fig. I, we demonstrate that Con A induces hyporesponsiveness to either FL-&4 or LPS challenge, equivalent to intact anti-Ig. Figure 2 shows dose-response curves for both PMA and ionomycin in terms of their abilities to independently induce B-cell hyporesponsiveness to challenge with LPS. These data demonstrate that the activation of PKC or the elevation of intracellular Ca++ concentration, by themselves, are sufficient to induce B-cell hyporesponsiveness to subsequent challenge with the polyclonal activator LPS. Of interest is the fact that only PMA, but not ionomycin, downregulates antibody formation in response to the antigen-specific challenge with FL-&l. Thus, PMA (or Con A) appears to mimic the Fc-dependent induction of hyporesponsiveness seen with intact anti-Ig, whereas ionomycin causes a reduction in antibody formation reminiscent of the Fc-independent activity of anti-Ig on LPS responsiveness (3). Importantly, all of these reagents cause MHC class II upregulation with the following hierarchy of decreasing effectiveness: anti-Ig = Con A > PMA > ionomycin (Warner, unpublished and Ref. (10)).
Cyclosporin A prevents anti-Ig-driven class II upregulation and interferes with Bcell tolerance induction. CSA has been shown to interfere with the anti-Ig-stimulated increase in class II expression in B cells (10). To test the role of class II upregulation in the tolerance process, splenic B cells were incubated with anti-Ig + graded doses of CSA, washed, and then challenged with LPS. As shown in Fig. 3, FL-specific IgM titers show that anti-Ig induces B-cell hyporesponsiveness and that CSA prevents this
FL-BA
.L
I 1: 0’.
t
ConA
+
IgGa-lg
I i
I. 0
0.5 1.0 2.0
100
0.5
1.0 2.0
10
ConA (pg/m/) FIG. I. Concanavalin A (Con A) induces B-cell hyporesponsiveness similar to intact anti-lg. Adult splenic B cells were treated overnight with the indicated concentrations of Con A (x-axis, squares) or intact anti-Ig (10 kg/ml, triangles), washed, and challenged with either FL-E,4 or LPS. The number of IgM anti-FLproducing cells (AFCs) was determined on Day 4 using an ELISA spot assay and are expressed as the number of FL-specific AFCs per culture well. The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
408
WARNER,
GAUR,
AND
SCOTT
=I
PMA
‘“NO_MYC’N
0 0
0.1 1
0.3 3
1 10
3 30
0 0
0.1 1
0.3 3
1 10
3 lONO 30 PMA (nM)
CONCENTRA T/ON PMA//OIVOMYC/N FIG. 2. Ionomycin (squares) or PMA (circles) induce B-cell hyporesponsiveness depending on whether the B cells are subsequently challenged with either with FL-BA (left) or LPS (right). Splenic B cells were treated with the indicated amount (x-axis) of either ionomycin (0.1-3 PM) or PMA (l-30 nM) for 4 hr, washed extensively, and challenged with FL.-BA or LPS. Four days after challenge, supematants were examined for the presence of anti-FL by ELISA. Results are presented as the ODm5 “,,, reflecting IgM anti-FL. The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
in a dose-dependent manner. CSA has an inhibitory effect by itself but the ability of anti-Ig to induce hyporesponsiveness is decreased relative to the CSA alone control. Indeed, in the absence of CSA, anti-Ig caused a 92% decrease in the anti-FL response, whereas with 100 rig/ml CSA, there was only an 18% decrease relative to the CSAonly control. Independent experiments confirmed that CSA blocked the anti-Ig (but not the IL-4) stimulated increase in class II expression (Warner, unpublished), as reported by Hawrylowicz and Klaus (10). Interestingly, CSA can prevent the hyperexpression of class II induced by anti-Ig, but not that stimulated by PMA (10). We, therefore, tested the activity of CSA on tolerance induction by these (abortive) B-cell activators. The data in Fig. 4 confirm that downregulation of the LPS response can be elicited by both PMA and anti-Ig, but that CSA only prevents the induction of B-cell hyporesponsiveness by anti-Ig. This is not surprising since class II upregulation by PMA is also CSA-insensitive (see Ref. ( 10)). Effect of IL-4 on tolerance induction elicited by B-ceil activators. Previous studies have indicated that IL-4 can overcome the negative signals delivered by crosslinking sIg to Fey receptors (7, 8). Since both Con A and PMA elicit a form of hyporesponsiveness to FLBA challenge akin to intact IgG anti-Ig, we investigated the effect of IL4 on tolerance induction in our model. B cells were precultured with IL-4 at 2000 U/ml overnight and then exposed to intact anti-Ig or PMA before washing and challenge with FL-BA. The results in Fig. 5 confirm that IL-4 pretreatment does protect B cells from the induction of hyporesponsiveness by intact anti-Ig and, moreover, they demonstrate that the inhibitory effect of PMA can be overriden by IL-4. However, this cytokine has no effect on the induction of hyporesponsiveness to LPS challenge (data not shown).
AN ANTI-lg
MODEL
409
FOR TOLERANCE
2.0
aI
-
t
10
T 1II
n q
NONE 10 pg/ml a-lg
100
CYCLOSPORlN A (nglmi) FIG. 3. Cyclosporin A (CSA) inhibits tolerance induction by anti-lg. Splenic B cells were treated with the indicated concentrations of CSA for 2 hr prior to exposure to 10 pg/ml anti-Ig (a-Ig) or medium (none) for an additional 18 hr. Supematants were assayed for the presence of IgM anti-FL. When compared to the CSA-only controls (black bars), the percentage unresponsiveness observed was: 0 rig/ml CSA = 92%; I ng/ ml CSA = 52%; IO rig/ml = 32%: 100 rig/ml = 18%. Similar results were obtained regardless of whether B cells were challenged with LPS or FL-B,4 (data not shown). The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
Is (abortive) B-cell activation a negative signal,for dljjkrentiation to Ig synthesis.? Table 1 summarizes the conditions we have tested herein and elsewhere (3, 9) that lead to cell cycle entry, class II upregulation, and the induction of tolerance in this model. It is noteworthy that conditions which lead to cell cycle progression (G, > G,), whether it is productive or abortive, can lead to hyporesponsiveness. In fact, entry into S is not necessary for the induction of tolerance. Rather, abortive B-cell activation in the absence of T-cell help consistently leads to B-cell hyporesponsiveness as well as class II antigen hyperexpression. These results are summarized in Table 1. DISCUSSION In a previous paper (3) we demonstrated that the induction of B-cell hyporesponsiveness (anergy) via crosslinking of sIg occurred after a 4- to S-hr incubation period, during which time many of the early events in B-cell signaling would have occurred and the cells would have egressed out of Go and into cycle. Originally, we intended to activate B cells without crosslinking sIg by using Con A, PMA, or ionomycin, known to abortively drive cell cycle entry (I, 2) and then reinvestigate the kinetics of tolerance induction via anti-Ig. Our hypothesis was that the time required in previous experiments reflected the need to progress out of Go and into a critical phase of G, for negatively signaling (cf. Ref. (11)). However, B-cell hyporesponsiveness was induced by incubation with Con A, PMA, or in some cases, ionomycin in the absenceof an anti-Ig signal. Therefore, crosslinking sIg was not required for negative signaling; rather it appeared that the events mimicked by these reagents, which drive B-cell entry into the cell cycle, were sufficient for hyporesponsiveness. Since we previously found that
410
WARNER,
NONE
1 nM PMA
GAUR,
AND
10 nM PMA
SCOTT
10 &ml
a-lg
INDUCING AGENT FIG. 4. Cyclosporin A (CSA) fails to inhibit the induction of B-cell hyporesponsiveness induced by PMA. Splenic B cells were treated with the indicated doses of CSA for 2 hr prior to the addition of the indicated (x-axis) amount of either PMA or intact anti-&. After an additional 18 hr the cells were washed and challenged with LPS (10 &ml). Four days later supematants were analyzed for the presence of anti-FL by ELISA and are presented as ODw5 nm reflecting IgM anti-FL. The error bars represent the standard deviation of the mean and the results are representative of three separate experiments.
cholera toxin prevented positive signaling but did not prevent B-cell tolerance (9), these data imply that abortive entry of B cells into the cycle might be coincident with a tolerance pathway. It should be noted that Con A, PMA (16), and ionomycin are also able to induce hyporesponsivenessin terms of LPS-induced differentiation. However, only the former two reagents downregulate antibody formation elicited by FL-&t, which requires an interaction of FcyR crosslinked to sIg (3). Furthermore, CSA was able to block the induction of B-cell tolerance induced by anti-Ig but not hyporesponsiveness induced by PMA. Taken together, these data demonstrate that the negative signaling pathway shareswith the positive signaling pathway sensitivity to CSA and that both the positive and the negative signaling pathways can be modulated by pharmacologic or physiologic activators of PKC. These data also imply that either transcriptional factors or regulatory motifs (affected by anti-Ig) in both class II and the Ig promoter/enhancer regions may share common features including the ability to act in truns (16). Efforts to identify their common features are underway in our laboratory and in others’ (16). FcyR:sIg interactions per se can downregulate B-cell function (7,8) presumably by causing a dissociation of the sIg complex from its G protein network (12), thus leading to an abortive activation of the phosphotidyl inositol cycle with a relatively normal rise in intracellular calcium (2). Interestingly, this processis reversed if the B cells are (pre)treated with IL-4 (3). Lazslo and Dickler (6) reported that FcyR function and association may be altered by pretreating B cells with IL-4. It has been reported that IL-4 does not cause protein kinase C (PKC) translocation to the plasma membrane, whereas anti-I-A and CAMP analogues induce the translocation of PKC to the nucleus rather than the plasma membrane (13, 14). Since PKC translocation may be important in T-cell help, it will be interesting to determine the effectsof anti-I-A and modulators of CAMP activity in our model.
AN ANTI-Ig
MODEL
rz
T 1. T
NONE
1 nM PMA
lMXRX’JG
411
FOR TOLERANCE
J
r
10 nM PMA 10 pg/ml u-lg
AGENT
FIG. 5. IL-4 pretreatment prevents the induction B-cell hyporesponsiveness to FL-R4 challenge induced by either PMA or intact anti-Ig. Splenic B cells were treated with 2000 U/ml of IL-4 for I8 hr prior to the addition of the indicated amounts of PMA or intact anti-Ig. Following an additional culture period of 24 hr. the cells were washed and challenged with FL-&l. Four days later supernatants were analyzed for the presence of anti-FL by ELISA. The results are presented as OD4e5 nmreflecting IgM anti-FL. The error bars represent the standard deviation of the mean and the results are representative of three separate experiment.
It is important to note that Hunziker et al. (15) recently reported that a B-cellspecific domain of the FcyR(I1) is phosphorylated by PKC after anti-Ig crosslinking except when intact IgG antibody is employed. This suggests that the dissociation of
TABLE Relationship
I
between Productive or Abortive B-Cell Entry into Cycle. Class II Hyperexpression, and the Induction of Hyporesponsiveness B-cell hyporesponsiveness”
Stimulus F(ab’), anti-Ig PMA + ionomycin IgG anti-Ig Concanavalin A PMA Ionomycin F(ab’)* anti-Ig plus CT F(ab’h or IgG anti-Ig plus CSA
Entry in cycle
DNA synthesis
MHC class II hyperexpression
FL-BA
LPS
Abortive Abortive Abortive Abortive Abortive
Yes Yes No No No No No
Yes Yes Yes Yes Yes Yes Yes
No Yes Yes Yes Yes No -
Yes Yes Yes Yes Yes Yes Yes
Abortive
No
No
No
No
Yes Yes
a Hyporesponsiveness upon challenge with either FL-B,4 or LPS, as described in text. Data are from the current paper as well as Refs. (3, 9).
412
WARNER,
GAUR,
AND
SCOTT
the G protein:sIg complex creates physical constraints on PKC substrates, including FcyR and the Ig complex. If IL-4 modulates PKC activity, this would explain why this lymphokine is effective at preventing the negative signal via intact anti-Ig especially with regard to the FL-&4 challenge. With these results in view, one can propose that under conditions of appropriate T-cell help, positive signaling progresses because PKC might not be available to act on the sIg complex at the membrane. An alternative, but not mutually exclusive, hypothesis is that PKC activity on FcyR allows for appropriate trafficking of the sIg complex and that tyrosine phosphorylation of IgMand &D-associated molecules ( 17-20) plays a balancing role in B-cell triggering. Further work defining the role of IL-4 and the availability of PKC, tyrosine kinase(s), and their substrates in these cells should shed light on this process. REFERENCES 1. Cambier, J., and Ransom, J. T., Annu. Rev. Immunol. 5, 175, 1987. 2. Klaus, G. G. B., Bijsterbosch, M., O’Garra, A., Harnett, M., and Rigley, K., Immunol. Rev. 99, 19, 1987. 3. Warner, G., and Scott, D. W., J. Immunol. 146, 2185, 1991. 4. Chace, J., and Scott, D. W., J. Immunol. 141, 3258, 1988. 5. Nossal, G. J. V., Annu. Rev. Immunol. 1, 33, 1983. 6. Laszlo, G., and Dickler, H. B., J. Immunol. 141, 3416, 1988. 7. O’Garra, A., Rigley, K., Holman, M., McLaughlin, J., and Klaus, G. G. B., Proc. Nuti. Acud. Sci. USA 84,6254, 1987. 8. Phillips, N., Gravel, K., Tumas, K., and Parker, D., J. Immunol. 141, 4243, 1988. 9. Warner, G., and Scott, D. W., J. Immunol. 143, 458, 1989. 10. Hawrylowicz, C., and Klaus, G. G. B., Immunology 53,703, 1988. Il. Scott, D. W., and Klinman, N., Immunol. Today 8, 105, 1987. 12. Rigley, K. P., Hamett, M., and Klaus, G. G. B., Eur. J. Immunol. 19, 481, 1989. 13. Justement, L., Chen, Z. Z., Harris, L., Ransom, J. T., Sandoval, V., Smith, C., Rennick, D., Roehm, N., and Cambier, J., J. Immunol. 137, 3664, 1986. 14. Cambier, J. C., Newell, M. K., Justement, L. B., McGuire, J. C., Leach, K. L., and Chen, Z. Z., Nature 327,629, 1987. 15. Hunziker, W., Koch, T., Whitney, J. A., and Mellman, I., Nature 345, 628, 1990. 16. Martenson, I-L., Iglesias, A., and Leanderson, T., Eur. J. Immunol. 19, 1497, 1990. 17. Gold, M. R., Law, D. A., and DeFranco, A. L., Nature 345, 8 10, 1990. 18. Campbell, M-A., and Sefton, B. M., EMBO J. 9, 2 125, 1990. 19. Chen, J., Stall, A. M., Herzenberg, L. A., and Herzenberg, L. A., EMBO J. 9, 2117, 1990. 20. Ales-Martinez, J. E., Martinez-A., C. Parkhouse, R. M. E., Pezzi, L., and Scott, D. W., Immunol. Today 12,201, 1991. 21. Sedgwick, J., and Holt, P. J., J. Immunol. Meth. 57, 301, 1983.