CD11c-Expressing B Cells Are Located at the T Cell/B Cell Border in Spleen and Are Potent APCs

It is well known that B cells contribute to immune responses via their descendants, Ab-secreting plasma cells. In addition, B cells are also professional APCs and, therefore, can participate in immune responses by activating Ag-specific CD4⁺ T cells. Interactions between B and T cells occur mainly in secondary lymphoid organs (spleen and lymph nodes) whose architecture is favorable for such events. B cells migrate to the B cell zones in secondary lymphoid organs where they may encounter Ags and become activated (1). Migration to B cell follicles is driven by a chemokine gradient of CXCL13, which is produced by follicular stromal cells and is recognized by the CXCR5 receptor on B cells (2). In the follicles, B cells can encounter Ags from the surface of neighboring cells, such as follicular dendritic cells (3) or macrophages (4). Alternatively, small soluble Ags can be acquired by B cells directly from the lymph (5). Although B and T cells are organized into separate zones in resting lymphoid organs, upon activation they move and interact with each other at the border between these two zones (6, 7). Such directed movement is tightly regulated by interactions of multiple chemokines produced by stromal cells and chemokine receptors expressed on the surface of lymphocytes. In particular, it was demonstrated that B cells upregulate the expression of CCR7 upon encounter with Ag. CCR7 expression drives the cells to move toward the chemokines CCL19 and CCL21 that are produced by stromal cells in the T cell zone (8). In addition, the formation of stable and motile Ag-specific B cell/CD4 T cell conjugates was detected using multiphoton microscopy, indicating that B cells may indeed act as APCs during the initiation of an immune response (8). In support of this statement, it was demonstrated that B cells are able to prime naive CD4 T cells both in vitro (9) and in vivo (10, 11).

Several well-characterized subsets of B cells are found in the spleen and/or lymph nodes. These are distinguished by the expression of distinct surface markers. In addition to phenotypic differences, each B cell subset has a unique function. The ability to present Ag also differs among the different subsets of B cells. For example, it was demonstrated that marginal zone (MZ) B cells are more potent activators of naive CD4 T cells than are follicular B cells (FO cells) (12). Enhanced Ag-presenting capabilities also were demonstrated for germinal center B cells (13).

We (14, 15) and other investigators (16) recently described a novel subset of B cells (age/autoimmune-associated B cells [ABCs]) in the spleens of elderly female mice that is characterized by the expression of CD11c and the transcription factor T-bet. B cells with a similar phenotype appear in autoimmune-prone mice, at about the time that the symptoms of their disease appear, as well as in animals suffering from acute virus infections (14, 15, 17). Gene-expression analysis, as well as surface staining of these cells, indicated that the cells express high levels of the costimulatory molecules CD80 and CD86, as well as MHC class II (MHCII) (14). These characteristics led us to hypothesize that ABCs can serve as efficient APCs to prime CD4 T cells.

In this article, we demonstrate that CD11c⁺T-bet⁺ B cells acquired from aged or autoimmune female mice present Ag more efficiently than do FO cells, both in vitro and in vivo. Moreover, these cells localize in spleens at the T cell/B cell border where Ag presentation takes place. In addition, upon Ag encounter, ABCs form more stable conjugates with T cells than do FO cells. Taken together, these data indicate a previously undescribed function for ABCs as potent APCs and suggest an additional way in which they may contribute to the development of autoimmunity.

Splenic B cells were purified by negative enrichment using biotinylated TER-119, NK1.1, and anti-TCRαβ Abs, followed by anti-biotin MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). All donor mice for ABCs were 18 to 20 mo old. ABCs were purified with a MoFlo sorter (Dako Cytomation) as B220⁺CD19⁺CD11c⁺ to >95% purity. FO cells were identified as B220⁺CD19⁺CD11c⁻CD21^intCD1d^int, and MZ B cells were isolated as B220⁺CD19⁺CD11c⁻CD21^highCD1d^high. After sorting, cells were stained for CD21 and CD1d surface expression to confirm the distinction among three B cell populations: ABCs, FO cells, and MZ B cells (Supplemental Fig. 1). For analysis, events were collected on a CyAn ADP (Beckman Coulter), and data were analyzed using FlowJo version 8.8 (Tree Star).

B cell populations were isolated, as described above, and 10⁵ cells were mixed with either 10⁵ OVA-specific H-2b–restricted T cell hybridomas (BO 80.10) or CD4⁺ T cells from OT-II–transgenic mice. Cells were incubated in the presence of whole OVA protein or OVA peptide (aa 323–339) for 24 h at 37°C. The presence of IL-2 in the supernatants was determined by MTT assay using IL-2–dependent HT-2 cells, as described previously (18). IL-2 titers are expressed in U/ml.

OVA-specific CD4⁺ T cells were isolated from OT-II–transgenic mice, labeled with CFSE, and incubated with either FO cells or ABCs (isolated from 18–20 mo old mice) in the presence of whole OVA protein (10 μg/ml) or OVA peptide (aa 323–339; 10 μg/ml) for 4 h at 37°C. CFSE dilution of CD4⁺ T cells was determined by flow cytometry.

3K-specific CD4⁺ T cells were isolated using a CD4 T Cell Isolation Kit (Miltenyi Biotec) from 508 TCR-transgenic mice crossed to Rag-negative background (bred at National Jewish Health) (19). Cells were labeled with CFSE, according to the manufacturer’s protocol, and injected i.v. into C57BL/6 (B6) mice (2 × 10⁶ cells/mouse). Twenty-four hours later, ABCs or FO cells were isolated from 18–20-mo-old mice by flow cytometric sorting, as described above, incubated in the presence of 3K peptide or 3K-OVA protein for 4 h at 37°C, and injected into mice that received labeled CD4⁺ T cells. Dilution of CFSE by T cells was determined 4 d later.

ABCs or FO cells from 18–20-mo-old GFP-expressing B6 mice were isolated as described above and injected i.v. into B6 mice. Three days after injection, spleens were harvested, incubated in 4% paraformaldehyde and 10% sucrose for 2 h at room temperature, and incubated in 30% sucrose overnight at 4°C. Spleens were frozen at −80°C in OCT compound (EM Sciences). Tissues were cut into 5–7-μm sections and dried at room temperature overnight. Sections were rehydrated with PBS for 20 min and blocked for 30 min with PBS, 2% BSA, 0.05% Tween 20. Ab mixtures were added and incubated for 45 min, followed by three 5-min washes with PBS. Sections were allowed to dry and were mounted, and analyzed with a Zeiss Axiovert 200M microscope (3i Marianas System) using SlideBook 4.0 software (Intelligent Imaging Innovations).

Lymph nodes or spleens were prepared at the indicated times, and RBCs were lysed. Single-cell suspensions were stained with MHC tetramers at 37°C for 2 h. Allophycocyanin-Db/NP366–74 and PE-Db/PA224–38 were produced as described (20), and PE-IAb/NP311–25 tetramer was provided by the National Institutes of Health Tetramer Core Facility. Abs to surface proteins were added, and the cells were incubated for an additional 20 min at 4°C.

Cells were stained under saturating conditions with Abs to mouse CD4 (clone GK1.5), CD8 (clone 53-6.7), B220 (clone RA3-6B2), CD11b (clone M1/70), CD11c (clone N418), and CD19 (clone 1D3), which were purchased from eBioscience or BD Pharmingen or generated in-house.

Cells were analyzed by flow cytometry on a CyAn (Beckman Coulter) instrument, and data were analyzed using FlowJo software (TreeStar).

Recombinant murine CXCL13, CCL19, or CCL21 (PeproTech) was added to the bottom well of a 24-well plate in a total of 500 μl migration medium (Iscove’s medium supplemented with GlutaMAX, 25 mM HEPES buffer, and 1% fatty acid–free BSA). Total splenic cells from aged (18–20 mo old) or young B6 mice were depleted of RBCs and added to the upper insert of a Transwell (5 μm pore; Costar) at a concentration of 10⁶ cells/100 μl. After 3 h of incubation at 37°C, cells were removed from the upper and lower wells, pooled (two to six wells were used for each condition in each experiment), stained on ice, and analyzed by flow cytometry. For relative cell counts, a CyAn was used to acquire cells for 60 s, at which time ≥250,000–500,000 cells were collected from upper wells and 20,000–90,000 cells were collected from lower wells. The percentage of cells migrating into the bottom chamber was calculated: (number of cells in lower well)/(number of cells in lower + upper wells) × 100.

B6 mice were obtained from The Jackson Laboratory. In Ag-presentation assays, T cells were isolated from transgenic mice expressing the F508αβTCR (T cells are specific for 3K peptide presented by I-A^b) (21) or OT-II (T cells are specific for OVA peptide presented by I-A^b) (22). B6/UBI-GFP (23) mice were generated and maintained at the National Jewish Health animal facility. B6.Nba2 (24) mice were purchased from The Jackson Laboratory (B6.NZB-(D1Mit47-D1Mit209)/BkotJ). Female mice were used for all experiments. All animals were handled in strict accordance with good animal practice, as defined by the relevant national and/or local animal welfare bodies, and all animal work was approved by the National Jewish Health Animal Care and Use Committee.

3K-specific CD4⁺ T cells were isolated from 508 TCR-transgenic mice that were crossed to Rag⁻ B6 mice (bred at National Jewish Health) (19). Cells were labeled with 20 μM CMTMR, 2 μM CFSE, or 2 μM VPD; washed three times; and injected i.v. into wild-type B6 recipients. Twenty-four hours later, ABCs and FO cells were isolated from 18–20-mo-old mice by flow cytometric sorting, as described above; incubated in the presence of 3K peptide for 4 h at 37°C; labeled with 20 μM CMTMR, 2 μM CFSE, or 2 μM VPD; and injected into recipient mice that received labeled CD4⁺ T cells. Twelve hours following injection, mice were sacrificed, and their spleens were surgically removed for imaging and immobilized on coverslips. During imaging, spleens were maintained at 35–37°C in a flow chamber perfused with RPMI 1640 medium without phenol red (Life Technologies) saturated with 95% O₂/5% CO₂. Multiphoton imaging was done using an Olympus FV1000MPE microscope with an XLPLN25XWMP Super 25×/1.05 NA water-immersion objective and a Spectra Physics 10 W Mai-Tai HP DeepSee-OL laser. The 450–490, 500–550, 575–640, and 645–685-nm filters were used for blue, green, red, and far red emission channel acquisition, respectively. To avoid potential effects of the fluorescent dyes on viability and motility of ABCs and FO cells, we swapped the dyes used to label the two transferred B cell populations between experimental repeats.

For time-lapse image acquisition, each x–y plane spanned 509 × 509 μm at a resolution of 0.994 μm/pixel. Images of up to 22 x–y planes with 3-μm Z-spacing were acquired every 30 s for 30 min. Data were visualized and analyzed using Imaris (Bitplane) and MATLAB (MathWorks). To isolate each fluorophore to a single channel, linear unmixing was performed. The fluorescence intensity of a given fluorophore in its optimal channel was determined. The bleed-through fluorescence of the same fluorophore in each of the other channels was then assessed. The percentage “bleed” into each channel was calculated by dividing the fluorescence in the nonoptimal channel by the fluorescence in the optimal channel. The fluorescence in all nonoptimal channels was then subtracted out on a pixel-by-pixel basis with MATLAB and the ImarisXT “Image Arithmetic” function using the percentage bleed determined. Surfaces were made in Imaris (using the “Surface” function) to identify the T cells, FO cells, and ABCs. Based on these surface objects, individual B cells were identified and tracked by Imaris, and cellular speed and displacement were calculated from the tracks. Only cells that were tracked for ≥5 min were included in analyses of motility and displacement, and only cells that were tracked for ≥10 min were included in analyses of interactions. Using a custom automated Matlab script, T cell/B cell interactions were quantified; T cells and B cells were scored as interacting if their cellular surfaces were within 0.994 μm (1 pixel) of each other.

All statistical analyses were performed with Prism software using the Student t test.

We demonstrated previously that ABCs obtained from aged mice express high levels of MHCII, CD80, and CD86 on their surface compared with FO cells (Fig. 1A) (14), suggesting that these cells might be efficient at stimulating T cells. Because activated B cells are known to upregulate costimulation molecules and migrate toward the T cell/B cell border to present Ag to T cells (25), we decided to determine the localization of ABCs in the spleen. Unfortunately, direct immunofluorescent staining of aged spleen sections was not able to distinguish definitively between ABCs and other B cell and dendritic cell populations. Therefore, we isolated ABCs and FO cells from old B6 female mice in which all cells express GFP (UBI-GFP mice) and injected them i.v. into naive B6 mice. After 72 h, we determined the localization of GFP⁺ B cells by immunofluorescent histology. Spleen sections were examined because ABCs are found preferentially in this organ rather than in lymph nodes. Although FO cells were primarily distributed homogenously in B cell follicles, ABCs were localized either in T cell zones or at the T cell/B cell border (Fig. 1B–D). These data were quantified based on the absolute number of ABCs or FO cells localized to the T cell zone or B cell follicle (Fig. 1C) or normalized to the area of each zone (Fig. 1D). The data clearly demonstrate that, unlike conventional B cells, ABCs are primarily localized in the T cell zone.

Next, we tried to determine which factors are responsible for ABCs’ localization at the T cell/B cell border. It is commonly accepted that chemokines and chemokine receptors determine the localization and migration of lymphocytes within secondary lymphoid organs (26). FO cells express the chemokine receptor CXCR5, and thus, they respond to the chemokine CXCL13, which is produced in the B cell follicle by follicular stromal cells. In contrast, T cells generally express the CCR7 receptor and migrate toward CCL19 and CCL21 chemokines (27). To understand how ABCs are localized to the T cell zone, we first examined the expression of CXCR5 and CCR7 by these cells and compared the results with those from FO cells. Flow cytometric analysis revealed that ABCs express levels of CXCR5 that are similar to those observed on FO cells (Fig. 2A). However, they express significantly higher levels of the T cell–specific chemokine receptor CCR7 (Fig. 2A). Next, we tested whether this difference in expression levels reflects a functional difference in the ability of these cells to migrate toward specific chemokines. Using migration Transwells, we assessed the responsiveness of ABCs and FO cells to three chemokines: the B cell–specific chemokine CXCL13 and two nominally T cell–specific chemokines (CCL19 and CCL21). Although migration toward CXCL13 was comparable between ABCs and FO cells (Fig. 2B), ABCs migrated significantly better toward CCL19 and CCL21 than did FO cells (Fig. 2C, 2D). These Transwell migration data are consistent with the expression patterns of chemokine receptors by ABCs and explain their localization to the T cell/B cell border in the spleen. During their conversion from FO cells (16), ABCs retain responsiveness to CXCL13 and upregulate the expression of CCR7, thereby gaining responsiveness to the T cell zone chemokines CCL19 and CCL21 and resulting in their localization to the T cell/B cell border in the spleen.

To compare the Ag-presenting abilities of ABCs and FO cells in vitro, we isolated these cell subsets from the spleens of aged female B6 mice and incubated them with Ag-specific T cells in the presence of either whole protein Ag (OVA) or peptide Ag from the same protein. ABCs and FO cells were compared for their ability to induce IL-2 production by OVA-specific T cell hybridomas (Fig. 3A), as well as primary OVA-specific CD4⁺ T cells isolated from OT-II–transgenic mice (Fig. 3B, 3C). As shown in Fig. 3, in the presence of OVA peptide, ABCs were more efficient than FO cells at inducing secretion of IL-2 by both Ag-specific T cell hybridoma cells and primary OT-II T cells. This difference was probably due to the higher levels of MHCII and costimulatory proteins on the ABCs. Similar results were obtained when intact OVA was used as the source of Ag and detection was by OVA-specific T cell hybridomas. The results were even more dramatic when naive OT-II cells were used as detectors and intact OVA was used as Ag. In this case, ABCs were still effective APCs, whereas FO cells failed completely to stimulate the T cells. This result may have been due to differences in the levels of MHCII and costimulatory proteins, as well as to the possibility that ABCs take up and process whole OVA protein more efficiently as a result of their high levels of expression of genes involved in vesicular transport and cytoskeletal rearrangement (14). We also assessed T cell proliferation in vitro in response to Ag presentation by monitoring CFSE dilution by OT-II T cells 3 d after incubation with Ag-pulsed FO cells or ABCs (Fig. 3D). Interestingly, the highest doses of Ag (either protein or peptide) led to equal T cell stimulation by FO cells and ABCs. However, ABCs were better T cell stimulators at lower concentrations of Ag. This result contrasts with our observations of IL-2 production for which the maximum differences were observed in the presence of the highest amount of Ag (Fig. 3B, 3C). The discrepancy is probably due to consumption of IL-2 by the proliferating T cells, causing the IL-2 assays (Fig. 3B, 3C) to underestimate the amounts of IL-2 produced by the T cells in each assay.

These results demonstrate that in vitro ABCs are more efficient APCs than are FO cells.

Next we explored whether the efficient Ag-presenting activity of ABCs in vitro is also evident in vivo. To this end, we transferred CFSE-labeled OT-II T cells into B6 mice 1 d before they received Ag-preloaded FO cells or ABCs. CFSE dilution by T cells was determined 4 d after the injection of B cells; as shown in Fig. 4, ABCs were significantly more potent in stimulating T cell proliferation. ABCs were equally good T cell stimulators when given either whole protein or peptide (Fig. 4A, 4B). In contrast, FO cells presented OVA peptide quite well (Fig. 4B) but intact OVA protein poorly (Fig. 4A).

These data demonstrate that ABCs have an increased ability to take up, process, and present Ag to T cells and induce T cell proliferation both in vivo and in vitro.

There are several possible explanations for the robust Ag-presenting abilities of ABCs. These include their localization to the T cell/B cell border, their ability to process Ag, and their high expression of costimulatory and MHCII molecules. All of these properties suggest that T cell contacts with ABCs may differ in quantity or quality from their contacts with other B cells (e.g., FO cells). To test this idea, we used multiphoton microscopy to compare the interactions between Ag-pulsed B cells (FO cells or ABCs) and Ag-specific T cells in spleen explants.

Fluorescent dye–labeled T cells were transferred into B6 mice, and Ag-pulsed B cells (FO cells and ABCs, labeled with different fluorescent dyes) were injected 24 h later. B cell/T cell interactions in the spleen were tracked by time-lapse multiphoton microscopy 12 h later. We observed that ABC/T cell interactions were significantly more stable than were FO cell/T cell interactions (Fig. 5A–C, Supplemental Video 1), resulting in longer contact times between ABCs and T cells. Previous studies on interactions between APCs and T cells demonstrated that the duration of interactions is critical for the fate of T cells (28). Long APC/T cell interactions, often lasting >12 h, led to sustained T cell activation, whereas aborted or transient interactions led to tolerogenic signaling in T cells (29–31). Therefore, our results suggest that ABC/T cell interactions are more likely to lead to T cell activation than are FO cell/T cell interactions. As a control, we also injected T cells that were specific for an unrelated Ag. Neither ABCs nor FO cells interacted with these T cells, confirming the Ag-specific nature of the presenting cell/T cell interactions that we observed (Supplemental Video 2).

Overall, our multiphoton experiments demonstrated that ABCs not only localize to the area most suitable for interactions with Ag-specific T cells, they also possess some cell-intrinsic features that allow them to form more productive interactions with Ag-specific T cells, ultimately leading to more efficient T cell activation.

All of the data described above indicate that ABCs obtained from aged B6 female mice are more efficient APCs than are FO cells. This is explained by their expression of several surface markers, including chemokine receptors and costimulatory molecules, their localization in T cell zones in the spleen, and their ability to process and present protein Ags effectively.

B cells expressing T-bet, CD11b, and CD11c are also found in the spleens of autoimmune mice, appearing at about the same time as the first symptoms of autoimmunity (14, 17). These cells also have been termed ABCs; however, they may not share all of the properties of the related population in aged female animals. To determine whether ABCs from autoimmune mice have Ag-presenting capabilities similar to those of ABCs from elderly animals, ABCs were obtained from the spleens of the autoimmune-prone strain B6.Nba2. The donor animals were 8–12 mo old and already had autoantibodies in their sera. Expression of chemokine receptors on the surface of ABCs and FO cells was compared using flow cytometry. As shown in Fig. 6A, ABCs obtained from B6.Nba2 mice expressed higher levels of CCR7 and similar levels of CXCR5 compared with FO cells obtained from the same mice. A comparison of the ability of the cells to migrate in CXCL13, CCL19, and CCL21 chemokine gradients showed that ABCs and FO cells from autoimmune mice migrated equally well toward CXCL13 (a chemokine expressed in B cell follicles). However, autoimmune ABCs migrated better toward CCL19 and CCL21 (T cell zone chemokines) than did FO cells. Together, these data suggest that, like ABCs from elderly animals, the ABCs in autoimmune animals may also be localized to the T cell/B cell border (Fig. 6B).

Finally, the Ag-presenting capabilities of ABCs and FO cells from B6.Nba2 autoimmune mice were tested in vitro. Similar to the data obtained with cells from aged female mice (Fig. 3), ABCs from autoimmune mice presented Ag (peptide and whole protein) and activated primary Ag-specific T cells significantly better than did FO cells (Fig. 6C).

Taken together, these data demonstrate that ABCs from autoimmune mice possess characteristics that are similar to those of ABCs from aged wild-type mice. Thus, ABCs may influence the development and progress of autoimmunity through secretion of autoantibodies (14, 17), as well as through presentation of self-Ags and activation of autoreactive T cells.

We recently identified ABCs as cells that are able to produce self-reactive Abs and that accumulate both with age and during the onset of autoimmunity (14, 17). We also described a similar subset of B cells in autoimmune patients; however, the function of these cells in both aged and autoimmune mice remained largely unknown.

B cells can contribute to the development of autoimmune disease in various ways. The most obvious and direct way is by the secretion of autoantibodies that target and destroy the animal’s own organs. Our previously published data showed that ABCs, after activation, are capable of secreting self-reactive Abs (14, 17). In addition, we showed that ABCs are responsible for the appearance of autoantibodies and that the depletion of these cells leads to a reduction in autoreactive Abs in the serum (17).

However, there are other ways in which B cells can contribute to autoimmunity, including secretion of pro- or anti-inflammatory cytokines and presentation of self-Ags to autoreactive T cells (32). B cells express MHCII and costimulatory proteins and, therefore, may present Ag to CD4 T cells. The consequences of such presentation for the T cells depend on the state of the B cells. In some cases, B cells, particularly naive resting B cells, present Ag and tolerize the responding T cells (28). In other cases, the target T cells are productively activated. Despite the fact that B cells were demonstrated to be potent APCs, several reports suggest that the presence of B cells in mice is not required for the generation of T cell responses to Ags, such as keyhole limpet hemocyanin, alloantigens, influenza virus, and human γ globulin (33–35). Thus, it was suggested that DCs, and not B cells, are responsible for Ag presentation and activation of CD4 T cells (36). However, further studies demonstrated that this is not always the case, and B cells, as well as other APCs, can play various roles in Ag presentation depending on the nature of Ag (11, 37). Overall, it was suggested that, especially at low concentrations of an Ag, Ag-specific B cells uptake and present Ag more efficiently than do other cells (38). This might be explained by the fact that Ag-specific B cells can capture Ag via their BCR, whereas DCs and macrophages use pinocytosis, which is 1,000–10,000 times less efficient (39).

The fact that B cells can present Ags efficiently makes them good candidates to function as APCs during autoimmunity, because self-Ags are usually present at low concentrations. In support of this, it was demonstrated that B cells are the major APCs during the induction of various autoimmune diseases (40, 41). Several reports indicated that the Ag-presenting function of B cells is crucial for the development of type 1 diabetes (42–44). Moreover, NOD mice that contain B cells that can react only with hen egg lysozyme fail to develop diabetes. Conversely, disease is accelerated in NOD mice that express an IgH that is prone to react with insulin (45). These results suggest that autoreactivity of the BCRs in NOD mice is critical for presentation of Ag in this model (46). The role of B cells as APCs in autoimmunity also was suggested by an MRL^lpr model of lupus-like disease by generating MRL^lpr mice that contain no B cells or that have B cells that are unable to secrete Abs (47). In the absence of B cells, the mice do not develop lupus. In contrast, if B cells are present the animals develop disease, even if the B cells cannot give rise to soluble Abs (47).

However, the particular subset of B cells that is responsible for the presentation of the self-Ag in autoimmune settings has never been established. In this study, we investigated the role of ABCs in Ag presentation in aged wild-type or autoimmune B6.Nba2 mice. We explored the localization of ABCs in the spleen and their ability to present Ag and activate Ag-specific T cells. Our data indicate that, as a result of the expression of both T and B cell–specific chemokine receptors, ABCs are localized at the T cell/B cell border in the spleen as opposed to FO cells, which localize primarily to B cell follicles. Thus, ABCs’ localization provides a greater opportunity to form interactions with Ag-specific T cells.

Coincidental with their localization, ABCs also have a greater ability to present soluble protein Ag to Ag-specific T cells both in vitro and in vivo. The fact that this is true in in vivo assays could be due to the fact that ABCs, but not FO cells, localize to the T cell/B cell border. However, ABCs are also more potent than FO cells in activating Ag-specific T cells in vitro, indicating that ABCs possess cell-intrinsic features that allow them to be more efficient at Ag presentation. These intrinsic features could include enhanced Ag uptake and/or processing by ABCs. To test this idea, we cultured ABCs and FO cells with DQ-OVA and compared their ability to generate fluorescent Ag. We did not see any difference in the rate of Ag processing between ABCs and FO cells (data not shown). Therefore, ABCs probably present Ag to T cells in vitro more efficiently than do FO cells because ABCs express higher levels of MHCII and of the costimulatory proteins CD80 and CD86.

We showed previously that ABCs from autoimmune-prone mice can give rise to cells that secrete autoantibodies (14, 17), indicating the specificity of their BCRs for self-Ags. As such, ABCs can take up autoantigens through their AgRs and are perfect candidates for activating autoreactive T cells, leading to the onset of autoimmunity.

Moreover, multiphoton data demonstrate that ABCs form significantly more stable interactions with T cells compared with FO cells. The stability of APC/T cell contacts was shown to be critical for the fate of the T cells because more stable interactions usually lead to T cell activation, whereas less stable ones often lead to tolerance (28). Thus, interactions between ABCs and T cells have a better chance of leading to activation of the T cell than do FO/T cell interactions.

Taken together, our data strongly suggest that Ag presentation is one of the major functions of ABCs in both aged and autoimmune mice. This conclusion leads to several questions that have to be explored in the future. For example, how does the depletion of ABCs affect T cell activation during autoimmunity? We already demonstrated that depletion of ABCs leads to a reduction in the titer of autoantibodies, and we proposed that ABCs themselves were the main source of autoantibodies (17). However, in light of this study, it is possible that, in the absence of ABCs, CD4⁺ T cells do not become activated and, therefore, do not provide help to other autoreactive B cells.

We recently reported the appearance of ABCs at the peak of antiviral humoral responses (15). Our data indicated that, in the absence of ABCs, there is a reduction in antiviral IgG2a titers and inefficient viral clearance. It will be interesting to determine whether the Ag-presenting capabilities of ABCs acquired from infected animals are similar to those of aged and autoimmune ABCs. If this is the case, ABCs might play several important roles during viral infection.

In summary, the data presented in this article identify a previously unappreciated function for ABCs as potent APCs. The appearance of ABCs in autoimmune mice and autoimmune patients leads to the suggestion that their superior Ag-presenting capabilities might be critical for the development of autoimmunity.

We thank Dr. M. Phillips and L. Noges for critical review of the manuscript and Dr. P. Beemiller and B. Leavitt for programming of image analysis Matlab scripts.

The authors have no financial conflicts of interest.