Toll-like receptor 4-activated B cells out-compete Toll-like receptor 9-activated B cells to establish peripheral immunological tolerance

Abstract

B-cell–induced peripheral T-cell tolerance is characterized by suppression of T-cell proliferation and T-cell–dependent antibody production. However, the cellular interactions that underlie tolerance induction have not been identified. Using two-photon microscopy of lymph nodes we show that tolerogenic LPS-activated membrane-bound ovalbumin (mOVA) B cells (LPS B cells) establish long-lived, highly motile conjugate pairs with responding antigen-specific OTII T cells but not with antigen-irrelevant T cells. Treatment with anti–CTLA-4 disrupts persistent B-cell–T-cell (B–T) contacts and suppresses antigen-specific tolerance. Nontolerogenic CpG-activated mOVA B cells (CpG B cells) also form prolonged, motile conjugates with responding OTII T cells when transferred separately. However, when both tolerogenic and nontolerogenic B-cell populations are present, LPS B cells suppress long-lived CpG B–OTII T-cell interactions and exhibit tolerogenic dominance. Contact of LPS B cells with previously established B–T pairs resulted in partner-swapping events in which LPS B cells preferentially migrate toward and disrupt nontolerogenic CpG mOVA B-cell–OTII T-cell pairs. Our results demonstrate that establishment of peripheral T-cell tolerance involves physical engagement of B cells with the responding T-cell population, acting in a directed and competitive manner to alter the functional outcome of B–T interactions.

B-cell presentation of antigen to CD4⁺ T cells can induce either T-cell priming or the establishment of long-lived antigen-specific tolerance characterized by reduced T-cell activation and effector responses (1–5). Although interactions of early B-cell–CD4⁺ T-cell priming events have been visualized (5), the dynamics of B-cell–T-cell (B–T) interactions that lead to long-lived peripheral tolerance remain uncharacterized. The expression of membrane-bound ovalbumin (mOVA) as a model antigen is a well-established approach for examining cell-specific immune responses and induction of tolerance to self-antigen (6–8). Here, using B cells that express mOVA (hereafter, “mOVA B cells”) as a self-antigen and ovalbumin-specific CD4⁺ OTII T cells as autoreactive T cells, we characterized tolerogenic B–T interactions and examined the dynamics of B-cell competition for antigen-specific T cells by two-photon microscopy in lymph node explants.

B cells have been investigated widely as antigen-presenting cells for the induction of therapeutic tolerance (9, 10). The induction of tolerance by B cells expressing IgG linked to a target peptide has revealed long-term protective effects in several models of autoimmune disease, including type 1 diabetes, experimental autoimmune encephalomyelitis, experimental autoimmune uveitis, and formation of hemophilia inhibitors (11–14). B-cell–mediated long-lived tolerance in vivo is generated by the presentation of endogenously expressed antigen in the context of MHC class II via an IFN-γ–inducible lysosomal thiol reductase-dependent pathway and is CTLA-4 dependent (15). Our study builds on previous investigations of tolerance induction in which adoptive transfer of LPS-activated B cells expressing antigen (hereafter, “LPS B cells”) led to tolerance and to suppression of experimental autoimmune encephalomyelitis, whereas transfer of B cells activated by unmethylated CpG oligonucleotide (hereafter “CpG B cells”) did not (10, 11, 16).

Activation of B cells by Toll-like receptor (TLR) agonists modulates B cell maturation and alters T-cell–dependent B-cell responses to cognate antigen. The modulation of immune responses by TLR-activated B cells likely results because different TLR agonists induce varying expression patterns of chemokine receptors and molecules crucial for homing, antigen presentation, and formation of the immune synapse (17–19). TLR4 activation by LPS treatment induces maturation of naive B cells and is implicated in the development of B-cell–driven lupus-like autoimmune disease (20, 21). However, LPS-induced maturation of B cells can be redirected by B-cell receptor cross-linking toward a tolerogenic phenotype, demonstrating that tolerogenic B cells can be induced by TLR4 activation (21). In contrast, the TLR9 agonist CpG is reported to play a role in the breaking of immune tolerance (22).

Previous two-photon imaging studies have revealed that exogenously supplied antigen can trigger the formation of B–T conjugate pairs within the lymph node. Introduction of soluble antigen in vivo induces naive antigen-specific B cells to migrate toward the follicle boundary, where they form long-lasting B–T conjugate pairs with antigen-specific T cells, leading to robust T-cell–dependent B-cell antibody production (5). In another study, ex vivo peptide-pulsed naive B cells transferred into a recipient induced tolerance and T-cell–associated transferable immunosuppression in a contact-dependent manner (4, 23). Here, we examined the nature of B–T interactions after TLR4 or TLR9 activation of mOVA B cells presenting antigenic peptides to naive antigen-specific T cells. We used this model to define the behavior of tolerogenic (TLR4 activated by LPS) and nontolerogenic (TLR9 activated by CpG) B cells relative to naive antigen-specific T cells and found that competition between B cells is apparent at early time points during the induction of tolerance. Moreover, we observe that LPS-activated mOVA B cells (hereafter, “LPS mOVA B cells”) physically disrupt nontolerogenic CpG B-cell–T-cell pairs, a mode of cellular interaction that may underlie the establishment of tolerogenic dominance.

Results

Naive and LPS-Treated B Cells Are Tolerogenic Antigen-Presenting Cells.

Using transgenic mOVA mice, which express ovalbumin peptides (pOVA) in both MHC class I and class II (7), we investigated the capacity of naive LPS mOVA B cells and CpG-activated mOVA B cells (hereafter “CpG mOVA B cells”) to establish tolerance. T cells recovered from recipients of naive or LPS mOVA B cells exhibited reduced CD4 T-cell proliferation upon challenge with the MHC class II OVA peptide 323–339 relative to T cells from animals that received unstimulated or LPS-treated WT B cells (Fig. 1 A and B). T cells from animals that received CpG mOVA B cells showed proliferation kinetics similar to those of T cells from recipients of WT B cells (Fig. 1C), whereas recipients of unstimulated or LPS mOVA B cells had suppressed antibody production relative to recipients of WT or CpG mOVA B cells (Fig. 1D). In addition, CD4⁺ T cells recovered from recipients of LPS mOVA B cells showed significantly reduced IFN-γ production in vitro (Fig. 1 E and F). Finally, we investigated the capacity of mOVA B cells to induce tolerance in OVA-specific CD4⁺ OTII T cells. Adoptive transfer of LPS mOVA B cells led to a significant reduction in OTII T-cell proliferation upon secondary antigen challenge in vitro (Fig. 1G). In contrast, CpG mOVA B cells failed to suppress T-cell proliferation (Fig. 1H). Therefore, naive and LPS mOVA B cells induced tolerance in both antigen-specific and polyclonal T-cell populations; however, CpG mOVA B cells failed to induce tolerance.

Fig. 1.

B cells induce peripheral tolerance in antigen-specific and polyclonal T-cell populations. (A–C) Naive, LPS or CpG WT (open symbols) or mOVA B cells (filled symbols) (10⁷) were transferred into WT C57BL/6 mice. Two weeks after OVA challenge, T cells from peripheral lymph nodes were cultured in vitro in the presence of the indicated concentrations of pOVA 323–339, and the rate of proliferation was determined by [³H] thymidine incorporation (mean Δcpm ± SE, n = 3). (A) T-cell proliferation in cells recovered from animals receiving unstimulated WT or mOVA B cells (n = 3; P < 0.03 for >3 μM OVA). (B) T-cell proliferation in cells recovered from animals receiving LPS-treated WT or mOVA B cells (n = 3; P ≤ 0.02 at all concentrations). (C) T-cell proliferation in cells recovered from animals receiving CpG-treated WT or mOVA B cells (n = 3; P values not significantly different). (D–F) Serum and whole-splenocyte cell cultures from the same animals were used to assess anti-OVA IgG titers and IFN-γ production measured by ELISpot (mean background − subtracted spots ± SE). (D) IgG titers after transfer of WT or mOVA B cells (n = 3; unstimulated B cells, P = 0.02; LPS B cells, P = 0.001; CpG B cells P values not significantly different). (E) IFN-γ production after transfer of LPS-treated WT or LPS mOVA B cells and peptide rechallenge (n = 3; P = 0.03 for 2 μM OVA; P = 0.06 for 6 μM OVA). (F) IFN-γ production after transfer of CpG-treated WT or CpG mOVA B cells and peptide rechallenge (n = 4; P values not significantly different). (G and H) LPS- or CpG mOVA B cells were transferred into a WT mouse 1 d before OTII T cells. Seven days after OTII T-cell transfer, T-cell proliferation from pooled peripheral lymph nodes and spleen was assessed. (G) T-cell proliferation in cells recovered from animals receiving LPS-treated WT or mOVA B cells (n = 3; P < 0.01). (H) T-cell proliferation in cells recovered from animals receiving CpG-treated WT or mOVA B cells (n = 4; P = no significant difference).

Visualizing Antigen-Specific B–T Interactions in Lymph Node Following LPS or CpG Activation of B Cells.

Using two-photon microscopy, we examined cell–cell interactions in the intact lymph node between OTII T cells and either LPS or CpG mOVA B cells. OTII T cells recognize the MHC class II binding peptide p323–339 and are a useful model for CD4 T-cell recognition (24), and mOVA transgenic mice express distinct pOVA in MHC class I and class II (7). LPS or CpG mOVA B cells (5–7 × 10⁶) were transferred separately into recipient mice 12 h before transfer of CD4⁺ OTII T cells (5 × 10⁶), and B–T interactions were imaged in the B-cell follicle 18 h after T-cell transfer (Fig. S1A). We tracked B cells in follicles as they migrated independently or as conjugate pairs of B and T cells (B–T pairs). LPS and CpG B cells migrated independently with average velocities of 8.2 ± 0.2 and 7.8 ± 0.2 μm/min, respectively (mean ± SE of three separate experiments, P = 0.03). Motile B–T pairs, characterized by the B cell actively leading a single attached T cell, were found within follicles but not in the T zone in both LPS and CpG mOVA B-cell experiments (Fig. 2 A–D). LPS and CpG B–T pairs migrated with instantaneous velocities that were significantly slower than those of individual B cells (4.4 ± 0.1 μm/min for LPS B–T pairs, P < 0.001 relative to LPS B cells alone; and 5.3 ± 0.1 μm/min for CpG B–T pairs, P < 0.001 relative to CpG B cells alone) (Fig. 2E). In both cases, motile B–T pairs were remarkably stable, some contacts lasting >1 h (Fig. 2 F and G). B–T pairs remained coupled an average of 20.1 ± 2.9 min and 20.8 ± 2.4 min for LPS and CpG B–T pairs, respectively. Because it was not possible to visualize all B–T pairs from their formation to separation, the measured average contact duration represents an underestimate. We conclude that, despite the differences in functional outcomes when transferred separately, both tolerogenic LPS and nontolerogenic CpG mOVA B cells are capable of forming B-cell–OTII T-cell conjugate pairs that migrate with indistinguishable dynamics.

Fig. 2.

OTII T cells form motile, stable conjugate pairs with either LPS or CpG mOVA B cells. Imaging was performed in the B-cell follicle 18 h after T-cell adoptive transfer. (A) Snapshots taken during an imaging record (Movie S1) of an OTII T-cell (T, blue) and LPS mOVA B-cell (B, green) conjugate pair 18 h after adoptive transfer of OTII T cells (imaging time indicated in min:s). White arrows indicate the direction of conjugate pair movement; no arrow indicates stoppage. (Scale bar, 5 μm.) (B) Snapshots from Movie S2 of an OTII T-cell (T, blue) and CpG mOVA B-cell (B, magenta) pair 18 h after adoptive transfer of OTII T cells. (Scale bar, 5 μm.) (C) (Upper) 3D plots of LPS mOVA B–T pairs (B cell in green, T cell in blue; n = 5 pairs), normalized to the point of initial contact. (Lower) Corresponding tracks of CpG mOVA B–T pairs (B cell in red, T cell in blue; n = 5 pairs). (D) 3D coordinates (x,y,z) of OTII T cells in conjugate pairs with LPS mOVA B cells (Upper) or CpG mOVA B cells (Lower). (E) Distributions of instantaneous velocities of unpaired T cells in the presence of LPS B cells (light gray bars, mean = 10.0 ± 0.22 μm/min, n = 517 cells) or CpG B cells (dark gray bars, 10.9 ± 0.26 μm/min, n = 304); and of paired T cells while in contact with LPS B cells (green bars, 4.4 ± 0.1 μm/min, n = 843) or CpG B cells (red bars, 5.4 ± 0.1 μm/min, n = 1083). (F) Scatter plots showing individual and mean (red horizontal lines) durations of contacts of OTII T cells with LPS mOVA B cells (mean = 20.1 ± 2.9 min, n = 69 contacts) and CpG mOVA B cells (mean = 20.8 ± 2.4 min, n = 69 contacts) in four separate experiments for each condition. Filled circles represent B–T contacts for which the initiation or termination of contact was not visualized. (G) Contact maps of interactions between OTII T cells and LPS or CpG mOVA B cells. Lengths of colored lines indicate duration of contacts between T cells and LPS mOVA B cells (green) or CpG mOVA B cells (red); times when T cells were not in contact are indicated by gray. Data are ordered from top to bottom by total duration for which T cells were tracked.

Naive mOVA B cells also induced long-lived peripheral tolerance (Fig. 1 A and B). Eighteen hours after T-cell adoptive transfer, naive mOVA B cells showed a slight but statistically significant increase in the duration of contact with cognate OTII T cells compared with the duration of contact between OTII T cells and naive WT B cells (Fig. 3A and B). After 72 h, naive mOVA B cells formed stable B–T conjugate pairs with cognate T cells that occasionally lasted >40 min (Fig. 3 C and D; and Movie S3). Thus, despite a notable delay in the formation of conjugate pairs, the behavior of naive mOVA B–OTII T-cell was indistinguishable from that of other conjugate pairs in terms of contact duration and velocity. T-cell antigen specificity is important in other models of B–T conjugate pair formation (5). As an additional control, we examined LPS and CpG WT B cells and found only short-lived interactions with WT T cells (Fig. 3 E and F). We conclude that conjugate pairs form only when both B and T cells are antigen specific.

Fig. 3.

Naive mOVA B-cell–OTII T-cell interactions and antigen specificity of conjugate pairs. Lengths of colored lines in contact maps (A, C, E) indicate duration of contacts between T cells as indicated on y axis and B cells indicated in label. (A) Contact map of OTII T-cell interaction history 18 h after adoptive transfer. (B) Duration of contact of OTII T cells with naive WT B cells averaged 1.1 ± 0.2 min. Duration of contact with naive mOVA B cells averaged 1.4 ± 0.1 min. n = 53 contacts each in three separate experiments. (C) Contact map of OTII T-cell interactions with naive mOVA B cells 72 h after adoptive transfer of T cells. (D) The mean duration of contact between OTII T cells and naive mOVA B cells was 8.3 ± 2.1 min for n = 54 contacts in three separate experiments. (E) Contact map of interactions between WT T-cell and WT B cells activated with LPS or CpG. (F) Mean contact durations = 2.2 ± 0.2 and 2.7 ± 0.6 min, and n = 60 and 99 contacts, respectively, in three separate experiments. Filled circles represent B–T contacts for which the initiation or termination of contact was not visualized.

CTLA-4 Is Required for B-Cell–Induced Tolerance and for Long-Lived B–T Conjugates.

CTLA-4 is necessary for the establishment of peripheral tolerance by B cells (25). However, it is unknown if CTLA-4 is required for the formation of long-lived tolerogenic B–T conjugate pairs. To confirm a role for CTLA-4 in tolerance induction in our model, LPS-mOVA or WT donor B cells were adoptively transferred, followed 12 h later by OTII T cells, and recipient animals received either anti–CTLA-4 or isotype control antibody by two i.p. injections, 30 min and 12 h after T-cell transfer (Fig. S1B). Measurements of recovered OTII T cells in response to challenge with pOVA demonstrated that treatment with anti–CTLA-4 at an early time point resulted in both enhanced IFN-γ production and in vivo proliferation (Fig. 4 A and B and Fig. S2). In addition, treatment with anti–CTLA-4 blocking antibody significantly reduced LPS mOVA B-cell–OTII T-cell interactions relative to treatment with control antibody (Fig. 4 C and D and Movies S4 and S5). Treatment with anti–CTLA-4 suppressed the formation of long-lived conjugate B–T pairs (Fig. 4 E and F). T-cell contact mapping further illustrates the disruption of long-lived contacts between OTII T cells and LPS mOVA B cells by anti–CTLA-4 (Fig. 4G).

Fig. 4.

Tolerogenic B–T contacts are disrupted by an anti–CTLA-4 antibody. (A) IFN-γ production by OTII T cells after in vivo treatment with anti–CTLA-4 (αCTLA4) or isotype control (ITC) antibody. Data represent mean spots after background subtraction (4–15 spots; four experiments each condition; Fig. S1A). (B) OTII T-cell proliferation in vivo measured as an activation index by carboxyfluorescein succinimidyl ester (CSFE) dilution in mice preequilibrated with mOVA B or WT B cells and treated after T-cell transfer with isotype control antibody or anti–CTLA-4 (αCTLA4) blocking antibody (three and four experiments, respectively). (C) Scatter plots of the duration of conjugate pair contacts in the presence of isotype control [21:42 (min:s), n = 64] and anti-CTLA-4 [9:54 (min:s)]. n = 40; three separate experiments each condition, P < 0.01. Filled circles represent contacts in which the start and/or finish was not visualized. (D) B–T conjugate pairs (white lines) tracked for 20 min after treatment with either isotype control antibody or anti–CTLA-4 blocking antibody. (E and F) OTII T-cell encounters with LPS mOVA B cells represented in the x, y, and z dimensions in the presence of isotype control antibody (E) or CTLA-4 blocking antibody (F). Colored tracks (green, teal, and chartreuse) represent different B cells, and bars highlight contacts with an interacting T cell (blue). (G) Mapping of OTII T-cell interactions with LPS mOVA B cells in the presence of isotype control antibody (Left) or anti–CTLA-4 blocking antibody (Right) (three experiments, n = 30 cells for each condition). Contact durations are shown in green; periods when OTII T cells were not in contact are shown in gray.

Competitive B-Cell Interactions.

Next, we assessed the functional outcome of OTII T cells adoptively transferred into a mouse preequilibrated with both LPS and CpG mOVA B cells (Fig. S1C). Coinjection of equal numbers of LPS and CpG mOVA B cells resulted in conjugate pair formation between LPS mOVA B cells and OTII T cells, but CpG mOVA B-cell encounters typically were of short duration (Fig. 5 A and B and Movie S6). 3D normalized tracks of B–T pairs (Fig. 5C) further illustrate the close association between the LPS mOVA B cells and OTII T cells within conjugate pairs and the independent motility of the CpG mOVA B cells and OTII T cells. In the presence of LPS mOVA B cells, CpG mOVA B cells and OTII T cells interacted but failed to form conjugate pairs (Fig. 5D). Adoptively transferred OTII T cells maintained normal T-cell velocities (averaging 12.4 μm/min) within the B-cell follicle, except during interaction with an LPS or CpG mOVA B cell, when velocities were reduced significantly (6.4 μm/min and 4.0 μm/min, respectively) (Fig. 5E). Contact mapping of individual T-cell–contact histories further illustrates the preferential formation of long-lived LPS mOVA B-cell–OTII T-cell conjugate pairs in contrast to the brief CpG mOVA B-cell interactions (Fig. 5F). These results indicate that, although both LPS and CpG mOVA B cells are capable of engaging OTII T cells (Fig. 2), the presence of LPS mOVA B cells significantly inhibits the stability of contacts between CpG mOVA B cells and OTII T cells. As an additional test for antigen specificity, WT or OTII T cells were adoptively transferred 12 h after LPS and CpG mOVA B cells were cotransferred. Antigen-irrelevant WT T cells predominantly made short contacts and did not form conjugate pairs with either LPS or CpG mOVA B cells (Fig. 5G). Contacts between cognate LPS mOVA B cells and OTII T cells lasted much longer [mean = 10:17 (min:s), maximum 44 min] than contacts between OTII T cells and CpG mOVA B cells [2:12 (min:s), maximum of 4:48 (min:s)] (Fig. 5H). Thus, cotransfer of LPS and CpG mOVA B cells led preferentially to the formation of B–T conjugate pairs between tolerogenic LPS B cells and antigen-specific T cells.

Fig. 5.

Predominance of LPS mOVA B-cell–OTII T-cell conjugate pairs in the presence of cotransferred CpG mOVA B cells. (A) Interactions between CD4⁺ OTII T cells (blue) and either LPS mOVA B cells (green, green tracks) or CpG mOVA B cells (red, red tracks). Tracks of conjugate pairs are indicated in the right panel (Movie S6). (B) Close-up snapshots of an LPS mOVA B-cell (B, green)–OTII T-cell (T, blue) conjugate pair (Upper) and a brief interaction between a CpG mOVA B cell (B, red) and an OTII T cell (T, blue) (Lower). (C) 3D tracks of six representative B–T encounters normalized to initial B–T contact. OTII T cells are shown in blue, LPS B cells in green, and B cells in red. (D) 3D coordinates of an OTII T cell (blue) and an LPS (green) (Left) or a CpG (red) (Right) mOVA B cell. Bars indicate duration of B–T contact. (E) Distribution of instantaneous velocities of OTII T cells imaged in the B-cell zone when not in contact (gray; arrow at mean = 12.4 ± 0.3 μm/min) and when in contact with an LPS mOVA B cell (green; arrow at mean = 6.4 ± 0.14 μm/min) or a CpG mOVA B cell (red; arrow at mean = 4.0 ± 0.4 μm/min). Data are from four separate experiments. (F and G) Contact maps of OTII T cells (F) and control WT T cells (G) with LPS (green) or CpG (red) mOVA B cells (n = 30 individual cells, three separate experiments). (H) (Left) Duration of contacts of OTII T cells with LPS (mean = 10.3 ± 1.2 min, n = 61 cells; four separate experiments) and CpG mOVA B cells (mean = 3.1 ± 0.74 min, n = 16 cells; four separate experiments). (Right) Duration of contacts of WT T cells with LPS mOVA B cells, (mean = 2.1 ± 0.3 min, n = 38) and with CpG mOVA B cells. Data are from three separate experiments; (mean = 2.2 ± 0.2 min; n = 50). Filled circles represent contacts in which the start and/or finish was not visualized.

To assess the functional outcome of the cotransfer of CpG and LPS mOVA B cells, we performed in vitro assays of T-cell responses to pOVA following cotransfer and antigen challenge in vivo. LPS mOVA B cells established tolerance to self-antigen in vivo despite the presence of CpG mOVA B cells (Fig. S3 A–C). Interestingly, at these early time points, LPS B cells were found in higher numbers than CpG B cells in the lymph node and tended to persist longer in the spleen over the course of several weeks (Fig. S3 D and E). The lower number of CpG B cells in the lymph node may result from shedding of CD62L after CpG activation of B cells (Fig. S3F), a potential mechanism for redirecting B cells to the spleen (19). LPS and CpG activation of mOVA B cells also altered the surface expression of CXCR5 and CCR7 relative to that in resting B cells (Fig. S3F), but neither treatment altered B-cell localization within the lymph node as visualized by two-photon imaging and histological sections. In summary, LPS mOVA B cells out-compete CpG B cells and direct the functional outcome of the immune response.

We also examined the expression of ligands known to play a role in tolerance and in the formation of B–T synapses (26–28). LPS B cells expressed higher levels of B7.2 and intercellular adhesion molecule 1 (ICAM-1) than CpG B cells, both by FACS and mRNA analysis (Fig. S3 A and B). In addition, we transduced LPS and CpG B cells with an IgG-tagged peptide (Eα) identifiable by FACS when presented in the context of MHC class II. Our results demonstrate that although MHC class II expression levels on CpG and LPS B cells are identical, LPS B cells present more endogenously expressed peptide in the context of MHC class II (Fig. S3 C and D). We suggest that the higher expression of B7.2, ICAM-1, and MHC class II peptide may contribute to the ability of LPS B cells to induce peripheral tolerance.

Visualizing Competition Between LPS and CpG mOVA B Cells for Antigen-Specific T Cells.

We then imaged partner-swapping interactions in which LPS mOVA B cells physically displaced CpG mOVA B cells from previously established OTII T-cell pairs (Fig. S1D). To clarify description, we refer to an existing B–T conjugate as “B1-T” and to the approaching B cell as “B2.” LPS or CpG B cells that engaged in a three-way contact with an established B1–T pair usually interacted first with the already-paired B1 cell (79%; Fig. 6 A–C and Movie S7) before making contact with the target T cell to establish a transient three-way B2–T–B1 interaction. The threesome then resolved by partner exchange, leading to the formation of a long-lived LPS mOVA B2–T contact and leaving behind the original B1 cell. T-cell exchange between B cells was highly dynamic, and reciprocal swapping events sometimes occurred in <10 min (Fig. 6B and Movie S8).

Fig. 6.

Competitive interactions between LPS mOVA B cells and pre-established CpG mOVA B-cell–OTII T-cell pairs. (A) A typical partner-swapping event initiated by an LPS mOVA B cell contacting an established CpG mOVA B-cell–OTII T-cell pair. A short-lived three-way interaction (*) is followed by displacement of the responding T cell and establishment of a new long-lived pair. (Imaging time in min:s; scale bar, 10 μm.) (B) Reciprocal partner swap by an LPS mOVA B cell with an established CpG mOVA B–OTII pair forming a three-way contact between the cell types (*) followed by a partner swap and then a reciprocal swap also preceded by a three-way contact (*). (Imaging time in min:s; scale bar, 10 μm.) (C) Percentage of contacts initiated by an approaching B cell with either the B cell (79%) or T cell (21%) in three-way contacts (n = 94, five separate experiments). (D) Duration of contacts between OTII T cells paired with LPS B cells (8.6 ± 1.9 min, n = 58) or with CpG B cells (11.3 ± 1.4 min, n = 113) or in a triplet with both B-cell types (4.7 ± 0.59 min, n = 57). Data are from seven separate experiments. Filled circles indicate contacts in which the start and/or finish was not visualized. (E) Bars show the percentages of different outcomes of three-way contacts initiated by either LPS (Left) or CpG mOVA B cells (Right); n = 94 in five separate experiments. Swarms were not used to determine partner-swapping outcomes. No significant differences were apparent in partner retention (69%), partner exchange (25% for both LPS and CpG B cells), or dissolution of all contacts (5% for LPS and 7% for CPG B cells). (F) The outcome of three-way contacts after renormalizing to account for the sixfold-higher rate of three-way contact initiation in favor of LPS over CpG mOVA B cells. (G) An LPS mOVA B-cell swarm attracted to a newly formed CpG mOVA B-cell–OTII T-cell pair [formed at 5:45 (min:s), red arrow], at the peak of the swarm [22:31 (min:s), blue arrow], and after a partner swap in favor of an LPS mOVA B cell followed by the swarm dissipating [30:06; (min:s)] (Movie S9). (H) The mean instantaneous velocity of the swarming B cells (shown in G) that were in-frame at the time of the formation of a CpG mOVA B-cell–OTII T-cell pair is plotted against time. The red bar indicates a mean overall velocity 8.0 μm/min. A fourth B cell that entered from out of frame was not included in the analysis. The red arrow indicates the initiation of CpG mOVA B-cell–OTII T-cell contact; the blue arrow indicates the time when the LPS mOVA B-cell swarm contained the greatest number of cells.

Because CpG and LPS induce different levels of B-cell ligand expression, we analyzed key ligands known to play a role in B–T interactions (B7.2, ICAM-1, and self-peptide–MHC class II) (Fig. S4). During partner exchange, three-way contacts typically were brief (4.6 ± 0.6 min) (Fig. 6D). Surprisingly, although LPS mOVA B cells have higher ICAM-1 and B7.2 expression and self-peptide presentation, we found that the outcomes of three-way contacts (partner exchange, partner retention, or complete dissolution) were independent of whether an LPS or a CpG cell initiated contact; in both cases, partner retention was approximately threefold more likely than partner exchange (Fig. 6E).

We then determined the relative number of three-way contacts initiated by LPS or CpG B cells. Because of the delayed adoptive transfer of LPS B cells, imaging frames typically showed greater numbers of individual CpG B cells and CpG–OTII conjugate pairs than LPS B cells. We thus normalized the number of contacts initiated by each B-cell type using a ratio of the total numbers of lone cells vs. target pairs within each imaging volume. When normalized in this way, LPS B cells initiated three-way contacts about six times more frequently than CpG mOVA B cells (Fig. 6F); without normalization, LPS B cells interacted with CpG B-cell–OTII T-cell pairs 2.5 times more frequently. LPS B-cell initiation of partner-swapping events is likely to contribute to the predominance of stable LPS mOVA B-cell–OTII T-cell pairs seen in cotransfer studies. On occasion, LPS mOVA B cells appeared to migrate from different points of origin as a swarm and to surround a CpG B–T pair (Fig. 6 G and H and Movie S9), suggesting the induction of directed migration by the B–T pair. This behavior suggests local chemotaxis toward the CpG B–T pair.

Directional Migration of LPS mOVA B Cells Drives Partner Exchange.

To analyze further whether directed migration of B cells occurs in the proximity of B–T pairs, we developed a quantitative measure of directional persistence by calculating the 3D displacement of a B cell from its point of origin over the total distance traveled, in a step-wise manner (Fig. 7 A and B). A value of 1 indicates straight-line motion away from the origin. We applied this analysis to control B cells migrating in the absence of OTII T cells (Fig. 7C) and to B cells 5–10 min before they contacted a B–T pair (Fig. 7D). The mean directional persistence of interacting LPS B cells was significantly greater than for the control cells (0.45 ± 0.02 vs. 0.13 + 0.01; P < 0.001 in three separate experiments) and for all other B cells imaged (Fig. 7E), indicating a directed approach of the LPS B cell toward the CpG B–T target. These results suggest that a local factor may induce B cells to migrate toward a B–T target and along straighter paths.

Fig. 7.

Analysis of LPS B-cell directional motion toward CpG mOVA B-cell–OTII T-cell pairs. (A) 3D directional persistence of cell movement relative to the origin (red dot) is assessed using the equation (D2 − D1 + D3 − D2)/(d2 + d3) where D = x, y, z displacement relative to the origin and d = x, y, z distance per step. This method of analysis is applicable to any number of sequential steps and generates a single parameter for 3D directional persistence. (B) Diagram of directional persistence value extremes. A value of −1 represents cell movement directly toward the origin, 0 represents neutral movement relative to the origin, and a value of 1 represents 3D movement in a straight line away from the origin. We chose to measure directional persistence over two sequential steps to smooth any single-step tracking anomalies that can occur if a cell does not move a significant distance from the point of origin. (C) Histogram of directional persistence of control WT B cells without treatment imaged in the absence of OTII T cells. Arrow indicates mean value of directional persistence (mean = 0.13 ± 0.01, n = 20 cells; data are from three separate experiments). (D) Directional persistence of LPS B cells tracked for 8 min before contact with CpG B-cell–T-cell pairs. Arrow indicates mean values of directional persistence. Data are from three separate experiments. mean = 0.45 ± 0.02; n = 20 cells. (E) Mean directional persistence values of control B cells from histogram in C (control B); CpG B cells that did not interact with a B–T pair (CpG non-int); CpG B cells that did interact with an LPS B–T pair (CpG int); a swarm of LPS B cells near a CpG B–T pair (swarm B); LPS B cells that did not interact with a B–T pair (LPS non-int); and LPS B cells that interacted with a CpG B–T pair (LPS int).

Discussion

Our results demonstrate that peripheral tolerance to self-antigen induced by B cells is an active, competitive process marked by the stable formation of conjugate pairs of B and T cells. Using mOVA B cells as a model for presentation of self-antigen to OTII T cells, we imaged the dynamics of antigen-specific B–T interaction and found that both LPS and CpG mOVA B cells are capable of establishing long-lived, motile conjugate pairs with responding T cells. B–T conjugate pairs migrated with B cells in the lead, similar to a previous study (5), although with slower velocities than individual B cells. B-cell tolerance was induced by both naive and LPS mOVA B cells but not by CpG mOVA B cells. Naive mOVA B cells also formed stable B–T conjugate pairs, whereas activated antigen-irrelevant B cells failed to form B–T pairs, indicating that antigen recognition by both B and T cells is required for conjugate pair formation. These studies show that both tolerogenic and nontolerogenic priming responses are characterized by the formation of long-lived antigen-specific B–T conjugate pairs.

Despite indistinguishable B–T interaction dynamics when LPS- or CpG- mOVA B cells were transferred separately, cotransfer of both cell types in equal numbers resulted in a predominance of LPS B–T conjugate pairs and dramatically reduced numbers of CpG B–T conjugate pairs. Cotransfer of both types of activated mOVA B cells also led to dominant tolerance. To visualize competitive interactions, we allowed CpG mOVA B–T pairs to form before adoptive transfer of LPS mOVA B cells. Analysis of contact and dissociation events then showed that LPS mOVA B cells physically out-compete CpG mOVA B cells for OTII T-cell partners by initiating partner-swapping events. Moreover, LPS mOVA B cells moved in a directed fashion toward CpG mOVA B–T pairs, possibly by chemotaxis. In the majority of partner-swapping events, the LPS B cell “cut in” physically by first contacting the CpG B cell to disrupt a CpG B–T conjugate pair. We conclude from our kinetic analysis that the more frequent partner swapping initiated by LPS B cells relative to CpG B cells results in the dominance of LPS B-cell interactions with antigen-specific T cells. Furthermore, we showed that LPS B cells are dominant in conferring tolerance when CpG B cells are present. Our results therefore imply that the competition of LPS B cells for antigen-specific T cells determines the outcome of naive T cell–antigen encounters. Specifically, the directional persistence exhibited by LPS mOVA B cells toward target pairs contributes to the sixfold higher number of contact-initiation events and may drive the B–T partner swapping outcomes in favor of LPS mOVA B-cell–OTII T-cell pairs.

Within the lymph node, we found that CpG and LPS B cells colocalized, suggesting that differential localization does not contribute to the contrasting functional outcomes. However, LPS B cells homed to lymph nodes more efficiently than CpG B cells and expressed significantly higher levels of CD62L, suggesting that preferential homing may serve as an additional factor to bias an immune response in favor of tolerogenic LPS B cells.

The importance of CTLA-4–based interactions for the maintenance of tolerance is evident in CTLA-4–knockout animals that develop lethal systemic autoimmune disease (29). Here, we demonstrated that blocking CTLA-4 disrupted both B–T contacts and the establishment of peripheral tolerance, indicating that CTLA-4 stabilizes B–T interactions during induction of tolerance. Collectively, these results indicate that B-cell induction of peripheral tolerance to self-antigen is CTLA-4 dependent. CTLA-4–mediated suppression of an immune response occurs via both cell-intrinsic signaling and extrinsic contact-dependent transendocytosis of costimulatory molecules B7.1 and B7.2 (30). However, it was shown previously that B7-knockout B cells fail to induce tolerance, and expression of B7 costimulatory molecules by B cells is required for both naive T-cell activation and the maintenance of self-tolerance (31–33). We found that LPS B cells expressed higher levels of B7.2 than CpG B cells but expressed similar levels of B7.1, indicating that induction of tolerance in this model is not the result of insufficient B7 costimulation. Thus, CTLA-4 engagement of B7 molecules appears necessary for the induction of tolerance and the stabilization of conjugate pairs.

In conclusion, our imaging results reveal a unique cellular choreography associated with the induction of B-cell tolerance, wherein B cells take on an active and competitive role. Although we do not rule out a role for cross-priming by dendritic cells, MHC class II expression on B cells is necessary for induction of T-cell tolerance (34), and B–T interactions in vitro in the absence of dendritic cells are sufficient to induce T-cell tolerance (4), implying a direct role for B-cell presentation of antigen. LPS B-cell–induced T-cell tolerance requires CTLA-4 and the formation of antigen-dependent B–T conjugate pairs. Importantly, LPS B cells seek responding B–T conjugates and physically out-compete nontolerogenic B cells to alter the functional course of T-cell response to self-antigen.

Materials and Methods

Mice and Antigen.

mOVA-expressing transgenic mice (Actb-OVA, stock number: 005145), OVA-specific OTII TCR transgenic mice, and C57BL/6J WT mice were purchased from the Jackson Laboratory. Animals were housed in pathogen-free microisolator cages, and all experiments were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committees of the University of California, Irvine and the University of Maryland. pOVA p323–339 was purchased from Sigma-Aldrich.

Cells, Adoptive Transfer, and Analysis of Tolerance.

T cells from OTII or WT mice and B cells from mOVA or WT mice were purified, labeled with green or red cell-tracker dyes, and transferred into WT C57BL6/J hosts. Homing, localization in lymph nodes, T-cell proliferation, T-cell–dependent antibody production, and gene expression responses were assessed as described in SI Materials and Methods.

Antibodies and Flow Cytometry.

FITC anti-mouse CD19 (1D3), phycoerythrin anti-mouse CD62L (MEL-14), CD80 (16-10A1), CD86 (GL1), CD54 (3E2), CXCR5 (2G8), CCR7 (4B12), blocking CTLA-4 (UC10-4F10-11), and isotype control mAbs were purchased from Pharmingen and eBioscience. Methods for inhibition of tolerance by anti–CTLA-4 blocking antibody treatment in vivo and for flow cytometric analysis are described in SI Materials and Methods.

Two-Photon Imaging and Analysis.

Two-photon imaging of excised peripheral lymph nodes was performed as described previously (35) and in SI Materials and Methods. Analysis was performed by a combination of automatic and manual tracking of individual cells.

Statistics.

Data are presented as mean ± SEM. Normally distributed imaging data were analyzed using a two-tailed Student’s t test. Significance of data sets containing an unmatched numbers of cells or nonnormally distributed data sets was determined using a Mann–Whitney u test. OriginPro 8 software (OrginLab) was used for all statistical analysis.