Rescue of self-reactive B cells by provision of T cell help in vivo

We have previously demonstrated that antigen-specific T cell help can rescue mature Ig transgenic (Tg) hen egg lysozyme (HEL)-specific B cells from tolerance induction upon transfer into soluble HEL-expressing Tg hosts. Here we extend these findings by showing that T cell help could also rescue both immature and mature self-reactive B cells from rapid deletion in response to high-avidity membrane-bound HEL. Moreover, although short-lived anergic peripheral B cells that had matured in the presence of soluble self antigen could not be rescued by provision of T cell help, a proportion of immature anergic IgM+ IgD− CD23− B cells from the bone marrow of the same donors survived and proliferated when given help following transfer to a soluble or membrane HEL-expressing host. In other words, T cell help must be available relatively soon after the antigen signal to prevent induction of tolerance. Consistent with this interpretation, the stronger stimulus provided by membrane-bound antigen, which deletes immature B cells before they leave the bone marrow, did not afford an opportunity for T cell help to rescue tolerant immature bone marrow-derived B cells upon transfer in vivo. Nevertheless, these B cells were capable of responding to T cell help in vitro, which speaks against an immutable susceptibility of immature B cells to tolerance induction. Taken together, these data indicate that the strength of the antigen signal and availability of T cell help are the primary determinants of the fate of both immature and mature B cells, consistent with the model proposed by Bretscher and Cohn more than 25 years ago.


Introduction
Bretscher and Cohn were the first to propose that the response of an antigen-specific lymphocyte might be determined by an interaction with a second antigenspecific cell [1]. Their original model postulated that "antibody induction involves obligatory recognition of two determinants of an antigen by different antibody molecules". The second antibody, termed "carrier antibody", has now generally been replaced by T cell help in modern revisions of the theory. The Bretscher-Cohn model solved many of the difficulties of the previous models of Burnet and colleagues [2, 3] and Lederberg [4], which proposed a window of obligatory tolerance induction in the ontogeny of the animal [2, 3] or of each individual lymphocyte [4]. While the experimental evidence indicates that immunity can be induced early in ontogeny [5][6][7][8][9] and tolerance can be induced in the periphery of adult animals [10], thus formally refuting the original models, the idea that immature lymphocytes are predisposed to tolerance has remained a powerful concept in the field of self tolerance.
In the well-characterized hen egg lysozyme (HEL)specific Ig transgenic (Tg) model [11], exposure of naive B cells to self antigen by adoptive transfer into a host expressing HEL as a neo-self antigen induces tolerance which is indistinguishable from that observed in mice coexpressing both HEL and the anti-HEL B cell receptor (BCR) (double Tg mice). Both immature (IgM + IgD − CD23 -) bone marrow B cells [12] and mature (IgM + IgD + ) splenic B cells [13] are susceptible to tolerance induction and display indistinguishable sensitivity to the tolerogenic signals of different avidity provided by a range of concentrations of soluble HEL (sHEL) [13,14]. We have recently demonstrated that provision of T cell help can rescue mature HEL-specific B cells from induction of tolerance to sHEL in this model [12,15]. Thus, our exper- imental evidence is consistent with the Bretscher-Cohn hypothesis. On the other hand, attempts to rescue shortlived peripheral B cells which have matured in the presence of soluble self antigen have been unsuccessful in most cases [16,17], although in two instances, the combination of primed or alloreactive T cells and a strong antigen stimulus appeared to permit rescue [18,19]. These data raise the possibility that the requirements for rescue of B cells tolerized at an immature stage are more stringent, suggesting a developmental component in sensitivity to tolerance induction by self antigen.
To understand the factors responsible for regulating induction of self tolerance in B cells at different stages in their development, the responsiveness of immature and mature anti-HEL Tg B cells provided with antigenspecific T cell help was compared following recognition of self antigen in vivo. Immature (IgM + IgD − CD23 − ) but not mature HEL-specific B cells from sHEL donors survived and proliferated upon transfer to HEL-expressing recipients. Thus, T cell help appeared to be effective for only a limited time after antigen recognition. Since the life-span of B cells after antigen exposure in the absence of help is inversely related to the avidity of antigen binding [20,21], we postulated that the window of opportunity for T cell rescue should also be an inverse function of BCR signal strength. Consistent with this hypothesis, IgM + IgD − CD23 − immature B cells from membrane HEL (mHEL) double Tg donors could not be induced to proliferate in vivo, although they were responsive to T cell signals when co-cultured with helper T cells in vitro, presumably due to a faster interaction between T and B cells.

Experimental model for rescue of selfreactive B cells in vivo
It was shown previously that naive anti-HEL Tg B cells destined to become tolerant upon exposure to sHEL in vivo could be rescued and induced to differentiate and secrete antibody by provision of T cell help in the form of activated T cells from H-2 bk TCR Tg mice specific for moth cytochrome c peptide 87-103 (MCC 87-103 ) in the context of I-E k [12]. To control for possible antigen nonspecific effects of co-transferring activated T cells, all recipients were given activated H-2 bk T cells, together with either purified H-2 bk B cells with the capacity to present the appropriate peptide, H-2 b B cells that could not do so, or a mixture of the two (Fig. 1a). For focussing peptide to B cell MHC, one of two methods of comparable efficacy were used, i.e. in vivo administration of a fusion protein expressing both HEL and MCC 87-103 epitopes (HELcyt [15]), or pulsing of purified B cells with MCC 87-103 in vitro before transfer [12].

T cells rescue naive mature B cells from deletion by high-avidity self antigen
The  Fig. 1c). Most HELspecific B cells had divided four to six times between days 1 and 3, as indicated by their CFSE profile (Fig. 2d).
The small background of divided, HEL-binding H-2 b B cells (Fig. 2c solid box) probably represents cells which had proliferated in response to T cell help specific for B cell surface antigen/MHC complexes other than MCC 87-103 /I-E k . To rule out the possibility that bystander help stimulated by H-2 bk B cells could rescue H-2 b B cells even in the absence of direct T to B cell contact, CFSE-labeled, peptide-pulsed H-2 b B cells were transferred into sHEL recipients, with or without peptidepulsed H-2 bk B cells. Equivalent numbers of H-2 b B cells were present in both groups of recipients on day 3 (23fold and 15-fold fewer, respectively, than the number of control H-2 bk B cells, not shown).
Consistent with earlier data from sHEL recipients [21], immunohistology revealed that H-2 bk B cells had undergone arrest in the outer periarteriolar lymphoid sheath (PALS) of the spleen on day 1 (Fig. 2e), where they proliferated for several days before differentiating into proliferative foci and germinal centers (Fig. 2f) [12,21]. The response was accompanied by production of anti-HEL antibody (mean serum concentrations of 13 ? g/ml in recipients of H-2 bk B cells versus X 10 ng/ml in recipients of H-2 b B cells on day 14 after transfer). Thus, the shortlived pulse of T cell help provided by peptide-activated T cells i.v. and focussed by means of loading the B cells with peptide prior to adoptive transfer was sufficient to drive mature B cells to differentiate fully, even in the presence of high-avidity self antigen.

T cells rescue naive immature B cells from deletion by high-avidity self antigen
The fate of immature anti-HEL Tg B cells in the same experimental system was investigated using bone marrow cells depleted of mature cells on the basis of expression of IgD and CD23. Once again B cells from either H-2 b or H-2 bk Tg donors were CFSE labeled, pulsed with MCC 87-103 and transferred with primed TCR Tg T cells into mHEL Tg recipients. Both B cell populations underwent several rounds of spontaneous cell division, as expected from previous studies in which immature B cells were transferred into non-Tg or sHEL recipients in the absence on T cell help [12]. However, the H-2 b donor B cells also underwent simultaneous deletion and the very few cells remaining by day 3 (Figs. 1c; 2g, i, k) still showed down-regulation of IgM and IgD despite having acquired CD23. Immature H-2 bk B cells that could respond to T cell help survived and by day 3 had assumed a mature phenotype (IgM Hi IgD hi CD23 + ) in the presence of multivalent self antigen (Fig. 2h, j, l, solid boxes). Nevertheless, in contrast to their mature counterparts (Fig. 2f), immature B cells did not undergo differen-tiation into effector populations visible on immunohistology (not shown), and made no serum antibody.
The division-dependent increase in B cell numbers in response to T cell help was 2.5-fold between days 1 and 3, compared with 10-fold for mature B cells in the previous experiment (Fig. 1c). On the other hand, the relative difference between the numbers of H-2 bk and H-2 b B cells on day 3 was actually greater (53-fold for immature cells versus 15-fold for mature cells, Fig. 1d), principally because immature H-2 b B cells died more rapidly than mature cells when exposed to mHEL in the absence of T cell help.

Effect of T cell help on B cells previously exposed to soluble self antigen in vivo
The above data showed that naive immature as well as mature B cells were deleted upon exposure to highavidity antigen, unless rescued by simultaneous provision of T cell help. However, they left unanswered the question of whether T cell help could prevent deletion of B cells activated by previous exposure to self antigen, as may occur during normal B cell development in vivo. Initially MCC 87-103 -pulsed anti-HEL Tg B cells from the spleens of double Tg mice co-expressing sHEL were transferred together with T cell help into sHEL Tg recipients. Consistent with previously published data [16], no rescue of such mature tolerant B cells was seen ( Fig. 3a-f). We hypothesized that the time between BCR ligation in the bone marrow and provision of T cell help 3-4 days later in the periphery was too long to allow reversal of the tolerant state [20,23]. Accordingly, the impact of T help provided shortly after the BCR stimulus was investigated by substituting purified immature (IgM + IgD − CD23 − ) bone marrow B cells from sHELexpressing double Tg mice for mature B cells. When tolerant immature B cells were provided with T cell help in the presence of soluble self antigen, a significant proportion survived, proliferated and acquired a mature phenotype (Figs. 1c; 3g-l). Although the number of cell divisions in the surviving HEL-specific B cells on day 3 was similar for both naive (Fig. 2j) and tolerant (Fig. 3j) immature B cells, the number of cells on day 3 was only 53 % of the number on day 1 (Fig. 1c), indicating that rescue of tolerant immature B cells was five foldless efficient than that of naive immature cells. Moreover, the ratio of B cells from H-2 bk versus H-2 b double Tg donors was only 3.6 ( Fig. 1d), due to a combination of less efficient rescue of H-2 bk B cells and significantly increased survival of immature H-2 b B cells in sHEL versus mHEL recipients. Once again, no differentiation to germinal centers or proliferative foci was seen on immunohistology (not shown). Interestingly, provision of T cell help increased the proportion of divided and undivided B cells that had lost the capacity to bind HEL, which suggests that such cells changed their antigen specificity while in receipt of T cell help.

Prolonged T cell help fails to drive immature B cells to antibody production in vivo
Previous in vitro experiments have indicated that immature B cells are capable of differentiating into antibodyproducing cells in response to T cell help [24,25]. To formally test whether peptide-pulsing provided too brief a T cell stimulus to drive immature B cells to a stage at which they could differentiate in vivo, the experiment was repeated using HELcyt [15] (Fig. 1d, comparable to that seen when naive immature B cells were transferred to mHEL hosts in exp. 2), resulted from the reduction in survival of H-2 b B cells in mHEL compared to sHEL hosts (see above). Consistent with the findings following transfer of naive immature B cells to mHEL hosts (exp. 2, not shown), no HELbinding H-2 bk B cells were detectable on day 6 ( Fig. 4l), indicating that the fate of immature B cells in mHEL hosts was not affected by previous exposure to sHEL nor by addition of HELcyt antigen. No evidence for differentiation was seen on immunohistology (not shown), nor was anti-HEL IgM a antibody detected on day 6 or 14 after transfer.
The adoptive transfer was again repeated to provide activated T cell help over a longer period. In addition to repeated immunization with HELcyt, additional activated T cells were adoptively transferred on days 2 and 5. Once again, divided HEL-specific H-2 bk but not H-2 b B cells were seen on day 3 but had disappeared by day 6 (not shown).

Failure to rescue immature B cells previously exposed to high-avidity self antigen in vivo
The data so far suggested that immature B cells developing in the presence of the low-avidity ligand, sHEL, could be rescued from deletion if T cell help was provided sufficiently early after the antigen signal. Since the life-span of B cells is an inverse function of the strength of the BCR-mediated signal, we tested whether immature B cells exposed to a high-avidity mHEL signal in vivo could also be rescued by T cell help. The previous experiment was therefore repeated except that immature B cells derived from the bone marrow of double Tg donors expressing mHEL were transferred into mHEL Tg recipients. T cell help had little impact on B cell survival, few cells being detectable in either the periphery or bone marrow by day 3 following transfer (Figs. 1c; 5a-d). To investigate whether this reflected an intrinsic inability of tolerant B cells from mHEL mice to respond to helper T cell signals, or a failure to attract T cell help sufficiently quickly following adoptive transfer, immature B cells from the same mHEL double Tg donors were pulsed with MCC 87--103 and cultured in vitro with primed TCR Tg T cells in the presence or absence of mHEL-expressing thymocytes as a source of B cell antigen [26]. Under these circumstances, T cell help induced transient B cell expansion (Fig. 5e, f), clearly excluding an intrinsic defect in the ability of these cells to respond to T cell help as the primary reason for their failure to differentiate fully.

Discussion
The results described here and previously in the HEL Tg model indicate that the decision between deletion and survival of both immature and mature B cells is crucially dependent upon the provision of T cell signals within an appropriate period following recognition of antigen. Thus, provision of T cell help to both mature and immature naive self-reactive B cells can result in proliferation and rescue from early deletion in response to either a low- [12] or high- (Figs. 1, 2) avidity BCR-mediated signal. When a low-avidity signal has already been delivered in the bone marrow via soluble antigen, the life-span of the cells is reduced from approximately 4 weeks for naive B cells to only a few days [27].  Receipt of a higher-avidity signal in the bone marrow shortens the average life-span of tolerant B cells from 3-4 days to less than 24 h [26,27], allowing insufficient time for T cell help to act in the transfer system used here (Figs. 1, 5). Nevertheless the tolerant B cells can still be stimulated to divide in vitro, presumably because the physical proximity of T and B cells reduces the average time before T-B interaction occurs (Fig. 5).
Interestingly, provision of T cell help to naive or tolerant immature B cells increased the numbers of B cells that neither bound HEL nor expressed IgM a , suggesting that T cell signals afforded the B cells an opportunity of changing their specificity. While these data are reminiscent of the recent findings of Lang et al. [28], who showed that expression of bcl-2 in immature B cells increased the numbers of cells undergoing receptor edit-ing, the correlation between lack of HEL specificity and lack of IgM a expression was surprising, since the usual result of receptor editing [29,30] is to modify the Ig specificity by rearranging a second light rather than heavy chain. It is tempting to speculate that the mechanism of immature B cell survival and differentiation in response to T cell help is mediated by up-regulation of genes such as bcl-2 and bcl-x L .
Our data, taken together with a number of published studies, indicate that the form of T cell help is a crucial regulator of the fate of tolerant B cells which have already been exposed to self antigen. For example, in our experiments in which T cells were activated by i.v. peptide 1 day before transfer, and in those of Rathmell et al. [17,31], who used naive HEL-specific Tg T cells as a source of help, mature tolerant B cells could not be rescued upon transfer to sHEL hosts. On the other hand, when T cell help was provided by alloreactive T cells [16,19] or normal T cells primed multiple times with HEL/CFA [18], mature tolerant B cells survived and differentiated to secrete anti-HEL IgM a , provided they also received a strong antigen signal (mHEL or HEL/CFA, respectively).  [17,31,32]). Our data support the conclusion that provision of T cell help immediately after BCR ligation promotes proliferation, whereas delayed help for tolerant B cells promotes deletion. In addition, comparison of the efficiency of rescue of naive sHEL double and mHEL double immature B cells provides evidence that the window of opportunity during which Fas-and CD40mediated signals promote a combination of proliferation and survival [32] is an inverse function of the initial strength of the BCR signal.
These findings collectively underline the essentially passive role of B cells in the decision between tolerance and immunity, and suggest that the generation of self tolerance in the bone marrow may simply be a consequence of maturation in a T cell-deficient environment, rather than an intrinsic propensity of immature B cells themselves. Thus, the experimental evidence presented here is more consistent with the Bretscher-Cohn two-signal model of B cell tolerance and immunity than the Burnet-Lederberg hypothesis, which states that tolerance is the obligatory outcome of antigen recognition by immature B cells. Although immature B cells failed to differentiate fully in the experiments described here, previous data from Metcalf and Klinman [24,25] have demonstrated that these cells as well as mature B cells could be driven to antibody production in vitro. Thus, complete differentiation in our in vivo experiments may have been compromised by technical factors relating to the phenotype of T cell help provided by T cells activated by i.v. peptide immunization, which are known to produce lymphokines for a maximum of 2 days (B. Fazekas de St. Groth, M. Wikstrom, A. Smith and L. Girgis, submitted for publication). In other words, the duration of help required to stimulate an autoantibody response from mature, self-reactive B cells, is very short. Further experiments using different forms of T cell help will be required to clarify this issue.

Mice
All mice were bred and housed under specific pathogen-free conditions in the Centenary Institute Animal Facility. The following Tg mice were used: sHEL (ML5 line), in which the HEL transgene is under the control of the mouse metallothionein promoter [11]; mHEL (KLK3 line) which carries a transgene encoding HEL linked to the transmembrane region and cytoplasmic tail of MHC class I [33]; anti-HEL Ig bearing the IgH a allotype (MD4 line [11]); anti-MCC TCR (-D line [34]). To allow collaboration between T and B cells to occur via presentation of MCC 87-103 in association with I-E k , lymphocyte donors and recipients consisted of H-2 bk heterozygotes derived from crossing HEL or anti-HEL Tg mice on a C57BL/6 (IgH b allotype) background with B10.BR mice, and TCR Tg mice on a B10.BR background with C57BL/6 mice. Anti-HEL Tg mice on a C57BL/6 background served as a source of control H-2 b B cells that did not present MCC 87-103 to TCR Tg T cells.

Cell purification and adoptive transfers
Activated T cells were obtained by priming H-2 bk TCR Tg mice with an i.v. injection of 10 ? g MCC 87-103 20-24 h before isolation of lymph node cells. This resulted in activation of G 80 % of peptide-specific CD4 + T cells. H-2 bk or H-2 b B cell preparations were depleted of T cells with a mixture of rat IgM anti-CD4 (RL172.4 [35]), anti-CD8 (3.155 [36]), and anti-Thy 1 (HO13.4 [37]) followed by young rabbit complement (C-six Diagnostics, Mequon, WI). After fractionation on discontinuous Percoll gradients (Pharmacia, Uppsala, Sweden), cells at the 65/70 % interface were pooled for adoptive transfer. Bone marrow suspensions were depleted of mature (IgD + CD23 + ) B cells using a mixture of AMS15.1-biotin [38] and B3B4-biotin [39] followed by streptavidin-conjugated Dynabeads (Dynal, Oslo, Norway). The average percentage of B220 + cells in the depleted bone marrow was 8 % and the degree of residual contamination by mature B cells was less than 0.02 % in all experiments.
For pulsing with cytochrome peptide, purified B cells were resuspended at 5 × 10 7 /ml and incubated with 10 ? M MCC 87-103 for 2 h at 37°C, then washed three times before adoptive transfer. Cells were fully or quarter labeled with 5carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes, Eugene, OR) as described previously [12]. Adoptive transfers were performed by i.v. injection into the lateral tail vein of unirradiated H-2 bk HEL Tg recipients. B220 + B cells (10 6 of either mature or immature cells) were injected together with 10 6 activated T cells per recipient. In exps. 1-4 and 6, two recipients were analyzed at each of four time points, whereas twice that number were used in exp. 5.
Immunization with HELcyt, a fusion protein of HEL and MCC expressed in the baculovirus system [15], was performed by three i.v. injections of 50 ? g antigen on days 0, 3 and 6 after adoptive transfer.

Immunohistology
Fragments of spleen were snap frozen in liquid nitrogen. Frozen sections of spleen (5 ? m) were thaw-mounted onto glass slides and fixed and stained as described previously [12]. Anti-HEL Tg B cells were identified with a combination of RS3.1-FITC and RA3.6B2 followed by anti-rat Texas red (Caltag). After the final wash, slides were wet mounted and photomicrographs were obtained using a Leitz DMR BE fluorescence microscope (Wetzlar, Sweden).