Autocrine Secretion of Interferon γ Negatively Regulates Homing of Immature B Cells

B cells develop in the bone marrow and, through a process that involves rearrangement and expression of Ig genes, produce an antigen-specific receptor, which is first manifested in the immature stage. Immature B cells emerge from the bone marrow to the periphery and migrate into the spleen for their final maturation step. This migration proceeds through the terminal branches of central arterioles to blood sinusoids of the marginal zone 1,2. Newly made B cells then penetrate the marginal zone sinus to end up in the outer zone of the periarteriolar lymphocytic sheath 3. At this site in the spleen, B cells are still immature and can be distinguished from their mature counterparts 1,2,4,5.

One of the most fundamental and important aspects of the immune system is its ability to recognize and respond to virtually any foreign antigen while maintaining strict unresponsiveness to self-antigens. For B cells, antigens encountered during the immature stage, at anatomic sites populated with these cells, are likely to be self-antigens, and therefore, reactive clones are eliminated. The selection process occurring within the immature B cell compartments is also extended to the periphery, allowing deletion of all B cells recognizing peripheral self-antigens outside of the bone marrow 1,6,7.

Immature B cells leave the bone marrow and migrate to the spleen where they can mature before antigen encounter. Like other naive lymphocytes, before their arrival to the spleen immature B cells might recirculate to nonsplenic secondary lymphoid organs, which are specialized tissues for collecting antigens 8, or to sites of infection and inflammation. In these secondary lymphoid organs, where differentiation to mature cells does not occur, antigen encounter would lead to the death of the immature B cells and elimination of effective clones due to the negative selection process. Although recently it was suggested that CD40 signaling can induce immature B cell responsiveness to multiple members of the γ chain–common cytokine family 9, physiological stimulation of immature cells is known to usually lead to an arrest of cell differentiation and clonal deletion 1,6,7. In this work, we postulated the existence of biological mechanisms to prevent immature B cells from reaching antigen-enriched sites before maturation.

Herein we show that by autocrine secretion of IFN-γ, B cells developmentally regulate their recruitment to LNs.

Cells were obtained from control C57BL/6, invariant chain (Ii)^−/− 10, or IFN-γ–deficient mice on C57BL/6 background (The Jackson Laboratory) at 6–8 wk of age. Spleen and LN cells were purified as described previously 10.

B cells were purified by treating the splenocyte suspension with anti–Thy 1.2, CD4, and CD8 (Southern Biotechnology Associates, Inc.), followed by low Tox-M complement (Cedarlane). To separate between the control IgD⁻ cells from the IgD⁺ population, we used the MACS system (Miltenyi Biotec). The IgD⁻ population was incubated with anti-CD45RA (B220) magnetic beads and reseparated using the MACS system. For comparison of populations, in those experiments in which the control immature B cell population was separated, the Ii^−/− B cells were purified by the MACS system using the CD45RA beads. T cells were obtained by collecting the B220⁻ population. To separate between the IgD⁻CD21⁺ and IgD⁻CD21⁻ populations, IgD⁻ cells were divided according to their CD21 expression using the MACS system. The IgD⁻CD21⁻ cells were incubated with anti-CD45RA magnetic beads and reseparated. The purity of each cell population was determined by flow cytometry.

The following antibodies used in the experiments were purchased from BD PharMingen: 14.8 anti-CD45RA, B3b4 anti-CD23, 7G6 anti-CD21/CD35, R6-60.2 anti-IgM, AMS 9.1 anti-IgD, 7G6 anti-CD21/CD35, MEL-14 anti–L-selectin, R1-2 anti–very late antigen (VLA)-4, 5H10-27 anti–VLA-5, 2D7 anti–LFA-1, and anti-α4.

Adhesion assays were performed as described previously 11. 96-well plates were coated with fibronectin (100 μg/ml). ⁵¹Cr-labeled cells (2 × 10⁵/well) were suspended in RPMI supplemented with 0.1% FCS or conditioned medium collected from control or Ii^−/− cells. The cells were plated in each coated well, in the presence or absence of LPS (10 μg/ml), PMA (0.25 μg/ml), IL-2 (0.2 U/μl), stromal cell–derived factor (SDF)-1 (1 μg/ml), EDTA (5 mM), or LDV- and RGD-containing peptides (800 μg/ml). After 30 min, plates were washed. The adherent cells were lysed and the radioactivity was determined. Results were expressed as the mean percentage (± SD) of bound cells from quadruplicate wells. The percentage of cells that adhered was calculated as follows: (total counts of residual cells in the well)/(total counts of cells added to the well) × 100.

Total RNA was isolated from cells using the TRI reagent kit (Molecular Research Center). Reverse transcription was carried out using Superscript II RT (GIBCO BRL). The primers and PCR conditions were as described previously 12.

Cells were lysed as described previously 10. IFN-γ was immunoprecipitated by overnight incubation with the R4-6A2 monoclonal antibody to IFN-γ (BD PharMingen) followed by protein G–Sepharose beads (Amersham Pharmacia Biotech). Lysates or immunoprecipitated proteins were separated by 12% (wt/vol) SDS-PAGE. The proteins were transferred onto nitrocellulose and probed with the R4-6A2 anti–IFN-γ antibody (BD PharMingen), followed by horseradish peroxidase–conjugated goat anti–rat IgG (Jackson ImmunoResearch Laboratories).

Cells were incubated for 30 min with various treatments. The cells were then washed and labeled for 30 min with the fluorescent marker BCECF-AM (10 μg/ml; Molecular Probes; ∼97% of cells were stained). 5 × 10⁷ washed cells were injected intravenously into C57BL/6 mice in duplicates. After 3.5 h, spleen and LNs were removed, and the percentage of fluorescent cells in both organs was analyzed by FACS^®. Inhibition of cell accumulation was calculated as: 100% − (the amount of cells present after treatment/the cells present in the control mice). The results presented are the average of three different experiments.

Polystyrene plates were coated with fibronectin (30 μg/ml; Sigma-Aldrich) in the presence or absence of SDF-1 (0.5 μg/ml). These plastic plates were assembled to form the lower wall in a parallel wall flow chamber as described previously 13. Cells were allowed to settle in the coated capillary tube for 2 min before being subjected to increasing laminar flow. Cell images were videotaped and manually quantitated by analysis of images directly from the monitor screen.

Cytokines (GIBCO BRL) were added to the adhesion assay at the following concentrations: IL-4 (10³ U/ml), IL-5 (16 U/ml), IL-6 (16 U/ml), IL-12 (3.5 ng/ml), and IFN-γ (0.1 U/ml).

To determine whether immature B cells are indeed excluded from secondary lymphoid organs while recirculating through the body, we first compared the B cell distribution in spleen and LNs in control and in Ii–deficient (Ii^−/−) mice. B cells from Ii^−/− mice are arrested at an immature stage and have an immature phenotype expressing low levels of IgD and CD23 and partial CD21 expression (14,15,16; Fig. 1B and Fig. C, and Table). Moreover, these Ii^−/− B cells proliferate in response to LPS stimulation, but anti-IgM causes their death (data not shown). However, control and Ii^−/−-derived B cells found in the LNs were enriched for cells expressing a more mature phenotype, higher levels of IgD. Moreover, as was described previously, in Ii^−/− mice there was a striking decrease in the proportion of B cells populating the LNs (15,16,17; Fig. 1 and Table). These observations confirm that immature B cells are depleted from the LNs, although the Ii^−/− population expresses significant levels of L-selectin on their cell surface (Fig. 1 D).

To migrate to the LNs, leukocytes, including B cells, must interact with adhesive components of the extracellular matrix (ECM; 18). We therefore compared the adhesion response of mature (control) and immature (Ii^−/−) B splenocytes by studying their ability to adhere, upon activation, to fibronectin, a major ECM-based cell adhesive integrin ligand. ⁵¹Cr-labeled B cells were plated for 30 min on fibronectin-coated microtiter wells in the presence of PMA, a potent agonist of integrin-mediated adhesion 19,20, LPS, a B cell mitogen, or SDF-1, a potent B cell chemoattractant 21. After stimulation, control B cells dramatically increased their adhesion to fibronectin. In contrast, the adhesion of Ii^−/− immature B cells remained unchanged after activation (Fig. 2 A). As expected, B cells did not respond to stimulation with IL-2. The adhesion of B lymphocyte to fibronectin in response to stimulation was abrogated in the presence of the integrin inhibitor, EDTA, as well as by the LDV and RGD peptides, selective blockers of VLA-4 and VLA-5 (Fig. 2 B). Thus, the adhesion response observed was integrin mediated.

The inability of Ii^−/− B cells to increase their adhesion to fibronectin could be ascribed to the developmental stage of the cells, or to Ii deficiency. Therefore, we compared the adherence of purified control immature IgD⁻ and Ii^−/− B cells in response to SDF-1 or PMA stimulation. Like immature B cells derived from Ii^−/− mice, IgD⁻ immature B cells from control mice were unable to respond to stimulation and did not increase their adhesion to fibronectin (Fig. 2 C). Thus, the stage of maturity, and not the lack of Ii, determined the response. Similar cell surface expression levels of the integrins VLA-4, VLA-5, LFA-1, and α4, the major fibronectin receptors, were detected on both Ii^−/− and control B cells (Fig. 2 D). Thus, immature B cells possess the adhesive molecules required for stimulation and adherence to the ECM. To verify that intrinsic integrin function is not impaired, we tested both control and Ii^−/− populations in a short-term adhesion assay on fibronectin-coated surfaces assembled in a flow chamber apparatus. In the absence of stimulation, Ii^−/− B cells exhibited a slightly stronger spontaneous adhesion to fibronectin than that of the control cells. Surprisingly, rapid adhesion stimulation to fibronectin was induced in both cell types by 2-min treatment with SDF-1 (Fig. 2 E) or PMA (not shown). These results indicate that both the expression of integrin receptors for fibronectin and their ability to respond to rapid stimulation are normal in immature B cells. Thus, the lack of adhesion by immature cells during the 30-min adhesion assay appeared to result from inhibitory events subsequent to normal initial integrin-mediated adhesion.

To test whether the inhibition of adhesion may be transferred to other cells, we determined the ability of purified wild-type IgD⁻ immature or Ii^−/− B cells to inhibit the adhesion of purified mature IgD⁺ B cells when comixed. Surprisingly, both the control IgD⁻ and Ii^−/− immature B cells had an inhibitory effect on the adhesion of mature B lymphocytes to fibronectin (Fig. 3 A); the inhibition could be abrogated by fivefold dilution of the immature cells within the mature population.

The inhibitory effect mediated by immature B cells could be explained by direct cell–cell interaction, or by secretion of a soluble inhibitory factor. To determine whether a factor secreted from the cells could affect B cell adhesion, we analyzed the activity of conditioned medium collected from the cells. Wild-type or Ii^−/− immature B cells were plated on fibronectin-coated microwells and 30 min later their conditioned medium was collected. The ability of stimulated control B cells to adhere to fibronectin in the presence of conditioned medium derived from control or Ii^−/− immature B cells was studied. Whereas conditioned medium collected from control B cells did not affect cell adhesion, cells incubated with conditioned medium collected from Ii^−/− B cells (Fig. 3B and Fig. c) or control IgD⁻ immature B cells (Fig. 3 C) maintained a low adhesion response after stimulation. Thus, the conditioned medium derived from immature B cells contained a factor(s) that inhibited the adhesion of stimulated B cells. The factor was heat sensitive (Fig. 3 C).

To determine the identity of this factor, various cytokines were screened for their ability to suppress agonist-stimulated B cell adhesion to fibronectin. Of the chemokines tested, only IFN-γ could mimic the effect of the conditioned medium derived from immature B cells (Fig. 4 A), and acted in a concentration-dependent manner (Fig. 4 B). Antibodies against IFN-γ partially blocked (∼50%) the inhibitory effect of immature B cell–conditioned medium. The partial nature of the antibody blocking may indicate involvement of an additional factor(s) in this process (Fig. 4 C). Reverse transcription PCR analysis revealed that IFN-γ expression by immature B cells is developmentally regulated at the transcriptional level. Although mRNA was detected in both control and Ii^−/−IgD⁻ immature B cells (15,16,17; Fig. 1), this transcript was not found in IgD⁺ B cells from control mice (Fig. 4 D). Furthermore, the low partially mature IgD⁺ B cell population, which is present in Ii^−/− mice, downregulated its IFN-γ mRNA levels as well (data not shown). Newly emigrant immature B cells can be separated from the marginal zone cells by the expression of the CD21 marker. To determine which population secretes IFN-γ, we analyzed IFN-γ message in the CD21⁺ and CD21⁻ IgD⁻ populations. As can be seen in Fig. 4 D, the IFN-γ mRNA can be detected in the IgD⁻CD21⁻ cells from both control and Ii^−/− mice. After acquisition of CD21, the cells downregulate their IFN-γ expression. Thus, IFN-γ is transcribed mostly in newly arrived immature B cells. The Western blot analysis of IFN-γ expression in B cells during maturation revealed maturation-dependent regulation of IFN-γ production and secretion. Whereas immature (IgD⁻) B cells from control or Ii^−/− mice express high levels of IFN-γ, mature (IgD⁺) cells produce lower IFN-γ levels. Importantly, conditioned medium derived from Ii^−/− cells contained significant levels of IFN-γ, indicating that the cytokine is both produced and secreted by immature B cells (Fig. 4 E).

We further studied the effect of the conditioned medium and IFN-γ on the entry into LNs in vivo (Fig. 5 A). Control or Ii^−/− total splenocytes were preincubated in the presence of fresh medium (Fig. 5 A, bars 1 and 2), IFN- γ (Fig. 5 A, bar 6), conditioned medium collected from control B cells (Fig. 5 A, bar 3) or from Ii^−/− B cells (Fig. 5 A, bar 4), or conditioned medium collected from Ii^−/− immature B cells treated with anti–IFN-γ (Fig. 5 A, bar 5). The cells were then washed, labeled with the fluorescent dye BCECF-AM, and an equal number of live cells was injected intravenously into control mice. The proportion of labeled cells recovered in the spleen and LNs was determined 3.5 h after injection. The extent of labeled cell accumulation in the spleen was unaffected by the various pretreatments. In contrast, migration of Ii^−/− and of control cells incubated with Ii^−/− conditioned medium or IFN-γ to the LNs was significantly decreased, whereas cells incubated with conditioned medium derived from control B cells or control medium exhibited a degree of migration similar to untreated cells. When control cells were incubated in Ii^−/− conditioned medium treated with antibodies against IFN-γ, the ability of this conditioned medium to inhibit homing of the cells to the LNs was partially blocked (∼50%). In agreement with previous studies, our results suggest that the entry of lymphocytes from the blood into the spleen does not depend on integrin–fibronectin interactions, whereas migration to peripheral LNs is critically regulated by these adhesive interactions 22.

To determine the biological significance of our studies, we analyzed the B cell profile in spleen and LNs of genetically IFN-γ–deficient mice (IFN-γ^−/−). Although spleens derived from IFN-γ^−/− exhibited wild-type levels of immature and mature B cells, there was a striking difference in the profile of the B cell population in the LNs. LNs of IFN-γ^−/− were populated with significantly greater levels of immature B cells compared with their control counterparts (Fig. 5 B). These results demonstrate the physiological role of IFN-γ in regulating the arrival of immature B cells to the LNs.

En route to the spleen, for their final maturation, immature B cells recycle through the blood and can be recruited either to LNs or to sites of inflammation. Immature B cells prevent premature encounter with antigen by downregulating their own migration into the LNs. This inhibition is mediated by IFN-γ, which is transcribed and secreted by immature B cells but dramatically downregulated in mature B cells. Our in vivo results directly demonstrate the physiological significance of IFN-γ secretion in preventing immature B cell arrival to sites of foreign antigen presentation. In addition, low levels of IFN-γ have the ability to regulate the adhesion of mature B cells (23; Fig. 4). As immature B cells normally migrate to specific microenvironments, and the proportion of immature B cells in the circulation is low, it is more likely that immature B cells use IFN-γ to locally influence their integrin function rather than to exert global effects on mature B cells in vivo. The powerful inhibitory effect of IFN-γ on adhesion of B lymphocytes both in vitro and in vivo suggests that this protein if properly targeted may have therapeutic clinical application as a B cell–specific antiinflammatory agent in addition to its many other functions.

The authors wish to thank Amnon Peled for helpful discussions and ideas, and the Shachar lab for review of this manuscript.

This research was supported by Minerva Foundation, Germany, and the Israel Science Foundation.