MHC Disparate Resting B Cells Are Tolerogenic in the Absence of Alloantigen-Expressing Dendritic Cells

Resting B cell (rB) populations have been shown to tolerize to soluble proteins and to minor-H but not to MHC alloantigens. We speculated that the reason for failing to tolerize to MHC alloantigen is that the few remaining dendritic cells (DCs) contaminating puri�ed rB cell populations e�ciently activate MHC allospeci�c T cells which are present at a higher frequency than T cells speci�c for minor-H alloantigen and soluble proteins. We established that MHC disparate rB cells are indeed tolerogenic when devoid of DC populations, as parental strain mice showed delayed D d skin gra rejection when infused with rB cells from mice in which MHC class I D d alloantigen was speci�cally targeted to T and B cells (CD2-D d transgenic mice). In contrast, treatment of parental strain mice with allogeneic rB cells puri�ed from MHC-D d transgenic mice, in which D d is ubiquitously expressed, including DCs, induced accelerated D d gra rejection. We also showed that adding only 5,000 D d expressing DCs to CD2-D d rB cells abrogated the tolerogenic eﬀect. Surprisingly, allogeneic rB cells prolonged gra survival in D d -primed mice. us, MHC disparate rB cells are tolerogenic and their failure to delay gra rejection can be explained by contaminating allogeneic DCs.


Introduction
Resting B cells have been suggested to be potent inducers of speci�c unresponsiveness both in vitro and in vivo [1,2] and have been thought to mediate such effects by directly presenting Ag (signal 1) in the absence of productive costimulatory interactions, principally mediated by B7-CD28 and CD40-CD40L interactions (signal 2) [3][4][5][6]. is hypothesis, however, has been challenged by studies demonstrating that rB cells not only express CD40, but also upregulate expression of B7 family members in a timely fashion in vivo [7]. Moreover, it has become increasingly appreciated that indirect antigen presentation, in which antigen shed from donor cells is presented via immature host DC, may be crucial for tolerance induction [8][9][10][11]. Regardless of the mechanism, rB cells clearly have tolerogenic effects on T cells speci�c for some types of antigen, but not others. For instance, rB cells induce tolerance to soluble proteins [2,12] and to minor-H mismatched transplants [1], but not to transplants discrepant in MHC alloantigens [7]. at MHC disparate gras present a higher bar for tolerance induction in general is emphasized by studies showing that MHC-expressing allogras are rejected even in the absence of the "danger" signals that appear requisite in generation of immune responses to other antigens, presumably by enhancement of APC activity [13]. One explanation for the ability of rB cells to induce tolerance to minor-H but not MHC disparate gras is that the few DCs associated with rB cells puri�ed from normal mice effectively activate the far more frequent and high a�nity MHC allospeci�c T cells. To investigate this possibility, we assessed the ability of allogeneic rB cells, in which alloantigen expression is genetically restricted to T and B cells, to thwart allogra rejection.  (Figure 1). rB cells recovered from the spleens of the two transgenic lines aer extensive depletion of T cells, macrophages, and DC populations were comparable with respect to DC content (<1%), as well as expression of B lineage and costimulatory markers ( Figure  2). Naive and D d -primed FVB mice were pretreated with 10 7 rB cells from MHC-D d or CD2-D d mice. ough B cells puri�ed from these two transgenics were highly comparable in terms of DC content and other markers ( . Surprisingly, infusing either B cell population into primed mice consistently abrogated accelerated rejection (MST 12 days versus 8.5 days for control) and converted response kinetics to a primary rejection rate (Figure 3( that rejection did not depend on the activity of lymphocytes newly emigrated from the thymus that had not been initially exposed to alloantigen. Furthermore, we do not think that the difference in response is attributable to higher D d expression on MHC-D d versus CD2-D d B cells because in our CD2-D d transgenic mouse lines, with line-dependent D d expression levels, we have observed that tolerance, as assessed by skin allogra rejection, correlated directly with the D d expression level [14], suggesting that extensive TCR cross-linking by the rB cells is of importance in tolerance induction [15]. ese data suggest that if D d expression was as high on CD2-D d B cell as on MHC-D d rB cells, the CD2-D d cells would be able to induce more prolonged hyporesponsiveness, or even tolerance.

Primary and Secondary Gra Rejection Responses Mediated by CD8 + T Cells Are Abrogated by Genetically Targeted D d -Expressing rB Cells.
To better explore the effects of rB cells on the T cells mediating rejection, we utilized an adoptive transfer model in which T-depleted FVB host mice (thymectomized and irradiated with 500 cGy [16]) were reconstituted with limiting numbers of CD8 + T responder cells. CD8 + T cells are both necessary and sufficient to reject MHC class I disparate D d skin allogras in naïve mice and for accelerated rejection in antigen primed-mice, whereas CD4 + T cells are neither necessary nor sufficient for allogra rejection [17,18]. D d skin allogra rejection was substantially delayed in naïve CD8 + T cell-reconstituted mice that were infused with CD2-D d rB cells (MST 35.5 days versus 18.5 days) ( Figure  4). Even more striking was the substantially prolonged gra survival in mice reconstituted with D d -primed CD8 + T cells and infused with CD2-D d B cells (MST 17.5 versus 7.5 days) ( Figure 4). ese data con�rm that both naïve and memory CD8 + allospeci�c T cells are rendered hyporesponsive in vivo by alloantigen expressed on nonprofessional APCs, when contaminating allogeneic DCs are eliminated.

Effects of Number of Dendritic Cells on Priming versus
Hyporesponsiveness. Since contaminating DC in puri�ed populations of rB cells are likely responsible for priming naïve class I allospeci�c lymphocytes, we wished to quantitate the minimum number required. Accordingly, we added graded numbers of MHC-D d DCs (positively selected by cell sorting with anti-CD11c mAb and identi�ed as 97% pure) to 10 7 puri�ed CD2-D d B cells and infused the cell mixture into naive FVB mice. Remarkably, as few as 5,000 DCs (0.005%) abrogated the tolerogenic effect of the rB cells, and partially primed naive allospeci�c T cells with full priming were achieved with as few as 10,000 DCs (Table 1) Figure 5). While DC from the CD2-D d mice failed to delay gra rejection at any dose tested remarkably, rejection was accelerated at the highest (10 ) DC dose. us, the infused DCs from CD2-D d mice were not responsible for the induction of hyporesponsiveness to allogeneic skin gras. Moreover, even in genetic targeting strategies in which alloantigen is not directly expressed by DC, but rather may be acquired from neighboring cells, high doses of DC may prime rather than tolerize, an important caveat for this tolerance strategy.

Discussion
Our studies indicate that MHC disparate rB cells are tolerogenic, when alloantigen-expressing DCs are completely eliminated, implicating contaminating DC as a source of allostimulation in B lymphocyte populations isolated from normal mice. us, in an ideal world, where all potential DC can be depleted, one could anticipate that puri�ed populations of rB cells would be tolerogenic, even for MHC alloantigenic disparities. e likelihood that current technologies can reduce the DC content of a cellular population to 0.005%, the percentage of DCs that gave rise to partial priming for a single class I MHC mismatch is low. A gene therapy approach [20] with B cell-speci�c transduction of alloantigen may hold more promise in facilitating alloengrament in human clinical trials. However, our data indicate that the success of this approach requires limitations in the numbers of nontargeted DC in the infusate, as expression of allogeneic peptides on nontargeted DC, which acquire allopeptide via cross-presentation mechanisms, provokes immunity. �uite perplexing was the �nding that the memory (accelerated) rejection response, mediated by CD8 + T memory cells, was eliminated by the same population of rB cells that primed naïve CD8 + T cells (i.e., by MHC-D d rB cells), suggesting that memory CD8 + T cells are even more sensitive than naïve T cells to tolerogenic effects of quiescently presented antigen and require either larger numbers of DC than were provided in puri�ed MHC-D d B cell populations, or require more mature and activated DCs, such as those provided by skin gras. Indeed, skin-graing D d -primed mice have never failed to elicit a memory response. However perplexing, the �nding that rB cell populations delayed the memory response in vivo is consistent with recent studies demonstrating that CD8 + memory T cells can be tolerized to the same extent as naïve populations [21], and with earlier studies demonstrating that CD8 + T cell clones, representative of memory populations, exhibit split anergy, that is, mediate CTL activity, but fail to proliferate and clonally expand unless supplied with exogenous IL-2 [22,23] when presented with antigen by quiescent APC.
Although our �ndings appear to be at odds with other reports in the literature showing that memory T cells can be activated to alloantigen expressed on rB cells [1], such is not the case. As with memory responses to minor-H alloantigen, we too saw preservation of memory CTL responses in mice exposed to alloantigen on rB cells in vivo (data not shown). However, skin gra rejection in minor-H primed mice exposed to minor-H disparate rB cells was not tested [1], so it is possible that exposure of minor-H primed mice to minor-H-expressing rB cells, prior to skin graing, would also have abrogated accelerated rejection.
In conclusion, we have demonstrated that MHC disparate rB cell populations are tolerogenic and that the reason that MHC disparate rB cells derived from normal mice are immunogenic is the presence of small numbers of contaminating dendritic cells. Lastly, we have con�rmed the susceptibility of memory CD8 + T cell responses to disruption via exposure to tolerogenic lymphocyte populations.

Transgenic
Mice. CD2-D d transgenic mice were generated by injecting FVB embryos linking the CD2 promoter with D d cDNA construct (a kind gi of Dr. Randy Ribaudo) coding sequence in pD d SELFIX.34 and an appropriate enhancer expression cassette p29Δ2(sal-) [14]. e CD2-D d mice used were from the 4906 line, which showed the highest level of D d expression [14]. MHC-D d transgenic mice were generated by injecting FVB embryos with the genomic D d gene, including the MHC class I promoter isolated from pDd1 (a kind gi of Dr. Gilbert Jay) [14]. Mouse studies were performed in accordance with Institutional Animal Care and Use Committee guidelines.

Isolation of DC by FACS.
CD11c + cells from MHC-D d mice were isolated by cell sorting, using positive selection, as described previously [24]. Brie�y, low-density spleen cells from MHC-D d were incubated with PE-CD11c mAb (PharMingen

T Cell Adoptive
Transfer. Naive or D d -primed T cells were prepared by negative selection from spleen cells of naive or primed FVB as described in [25]. Brie�y, spleen cell suspensions from naive or primed FVB mice were incubated for 30 minutes on ice with a cocktail of mAbs directed against B220, I-A q , CD11b, and CD11c mAbs, then were washed, incubated with streptavidin-coated microbeads for 15 min, and depleted by MACS. 3 × 10 7 of these cells were infused into naive ATX FVB mice that had been irradiated with 500 cGy 1 week previously. For puri�cation of CD8 + T cells by negative selection with MACS, spleen cells from naive or primed FVB mice were depleted of CD4-, CD11b-, CD11c-, B220-, and I-A q -positive cells using similar techniques, with a resulting CD8 + cell purity of >91.9%. 2 × 10 7 of these CD8 + cells were infused into naive ATX FVB mice that had been irradiated with 500 cGy 1 week previously.

Transplant Recipient Treatment.
In vivo depletion of CD4 + and/or CD8 + T cells was performed by mAb injection.
ATX FVB mice were injected i.p. with GK1.5 (150 g/mouse) mAb and/or 53.6.7 (200 g/mouse) mAb on days −6, −5, −1 and +5, with respect to transplantation. Mice were checked for depletion on the day prior to skin graing [14,26]. Depletion of CD4 + T cells or CD8 + T cells, or both, from the peripheral blood was >98% in all mice, as assessed by �ow cytometry (FACScan, Becton Dickinson). In studies evaluating the role of CD40-CD40L interactions, MR1 (anti-CD40L mAb, 250 g/mouse) was injected into ATX FVB mice on days -14, −10, −7, −3, 0, 4, and 7 and twice weekly thereaer, with respect to skin transplantation [15]. Anti-D d sera were prepared by injecting MHC-D d splenocytes into naive FVB mice i.p. (5 × 10 6 cells per mouse). e mice received three injections of cells at weekly intervals and 7 days aer the last injection they were bled for the purpose of preparing a single pool of antisera. Mice were injected i.v. with 0.2 mL of anti-D d sera 7, 9, or 12 days aer receiving MHC-D d skin gra.

Skin
Graing. Tail skin from donor mice was engraed onto the recipient�s �ank as described previously [27]. Gras were scored daily or every other day until they were rejected (>80% loss of gra tissue) .

Abbreviations
MST: Median survival time ATX: ymectomized rB cells: Resting B cells.