Evidence of Marginal-Zone B Cell- Positive Selection in Spleen

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Lijun Wen, Joni Brill-Dashoff, Susan A. Shinton, tive selection mediated by self-ligand engagement, or Masanao Asano,1 Richard R. Hardy, instead basal signaling that occurs even in the absence and Kyoko Hayakawa* of ligand (Reth et al., 2000; Roose et al., 2003). Fox Chase Cancer Center Most mature B cells in the lymphoid organs are of 333 Cottman Avenue bone marrow (BM) origin. Immature B cells are formed Philadelphia, Pennsylvania 19111 in the BM and mature in secondary peripheral sites, predominantly in the spleen, generating two histologi- cally, phenotypically, and functionally distinct mature B Summary cell subsets: “follicular B cells” (FO B) and “marginal- zone B cells” (MZ B) (Kumararatne et al., 1981; Martin Antigen receptor-mediated signaling is critical for the and Kearney, 2002). FO B cells are small recirculating development and survival of B cells. However, it has IgMlowIgDhi lymphocytes, forming the B cell follicles not been established whether B cell development re- within the white pulp, eventually developing into the quires a signal from self-ligand engagement at the im- predominant mature B cell population in adults. In com- mature stage, a process known as “positive selec- parison, MZ B cells are identified as medium-sized tion.” Here, using a monoclonal B cell receptor (BCR) IgMhiIgDlo lymphocytes resident in the spleen marginal mouse line, specific for the self-Thy-1/CD90 glycopro- zone at the white pulp-red pulp junction (Kumararatne tein, we demonstrate that BCR crosslinking by low- et al., 1981). MZ B cells are the major constituent of the dose self-antigen promotes survival of immature B marginal zone, together with myeloid, dendritic, and cells in culture. In spleen, an increase in BCR signal- stromal cells. Although the marginal zone is a transit ing strength, induced by low-dose self-antigen, di- site for reentering B cells and various activated B cells rected naive immature B cells to mature, not into the can be found in the marginal zone (Liu et al., 1988), the default follicular B cell fate, but instead into the mar- majority of B cells in the marginal zone of unimmunized ginal-zone B cell subset. These data indicate that pos- animals appear to be “naive,” i.e., they are resident itive selection can occur in developing B cells and even in germ-free animals (Kumararatne and MacLen- that BCR signal strength is a key factor in deciding nan, 1982). Such naive B cells in the marginal zone ex- between two functionally distinct mature B cell com- hibit distinct phenotypic features, such as high expres- partments in the microenvironment of the spleen. sion of the complement receptor, CD21 (Kumararatne et al., 1981; Martin and Kearney, 2002), and the ability to respond rapidly to certain antigens or pathogens fil- Introduction tered from the blood stream (Guinamard et al., 2000; Martin and Kearney, 2002). These characteristics sug- Lymphocyte development is a dynamic, interactive pro- gest that MZ B cells play a specialized function, which cess that takes place in various microenvironments. acts in synergy with and complements adaptive immu- For developing T cells, it has been established that an- nity that is initiated by FO B cells in concert with T cells tigen receptor signaling due to self-ligand engagement (Martin and Kearney, 2002). critically affects the fate of these lymphocytes, either There are several competing hypotheses concerning positively or negatively. During T cell development in the MZ B cell developmental pathway in relation to the thymus, binding of self major histocompatibility com- FO B subset (Kumararatne and MacLennan, 1982; plex ligand by T cell receptor in a low-affinity interac- Dammers et al., 1999; Martin and Kearney, 2002), in- tion prevents immature cell death, resulting in “positive volving such issues as the role of antigen and the selection” (Kisielow et al., 1988; Jameson et al., 1995). strength of BCR signal (Martin and Kearney, 2002; Pillai Subsequent development into functionally distinct CD4 et al., 2005), leading to speculation on the role of self- or CD8 subsets appears to be determined by a balance antigen in mediating FO/MZ B cell subset determina- of strength and duration of TCR signal, influenced by tion. In addition, recent discoveries on the critical role coreceptors (Yasutomo et al., 2000; Werlen et al., 2003). of non-BCR signaling in B cell development make the In comparison, it remains uncertain whether B cell de- role of BCR signaling more difficult to incorporate into velopment also requires positive selection (Healy and a unified hypothesis. BAFF, a member of the tumor ne- Goodnow, 1998). Previous studies with BCR signaling- crosis factor family produced from a myeloid/dendritic deficient or mutant mice have clearly established the cells, which is known to activate the NF-κB pathway, dependence of B cell development on antigen receptor is essential for both FO and MZ B cell development signaling, from the early pre-B to the mature B cell (Schiemann et al., 2001; Mackay and Browning, 2002). stages (Turner et al., 1995; Benschop and Cambier, MZ B cell development further specifically involves ac- 1999; Reichlin et al., 2001; Kraus et al., 2004). However, tivation of the extracellular developmental signaling it remains unclear whether, and to what extent, such molecule, Notch 2 (Saito et al., 2003), by ligation with BCR signal dependence reflects a requirement for posi- delta-like 1(Hozumi et al., 2004), in addition to distinctive migration (Guinamard et al., 2000; Fukui et al., *Correspondence: [email protected] 2001; Girkontaite et al., 2001) and retention mecha- 1Present address: Department of Internal Medicine and Rheumatol- nisms (Lu and Cyster, 2002; Cinamon et al., 2004). ogy, Juntendo University School of Medicine, Tokyo 113, Japan. Taken together, the differences in these mechanisms Immunity 298

reveal considerable complexity in B cell maturation/dif- up to 2 weeks of age, without generation of the ferentiation, in the context of the spleen, with poten- CD24loCD21medFO B cell population during the 8–10 tially complex interactions between BCR and non-BCR- week period (data not shown). There was no CD5+ mediated signaling pathways. B-1 cell generation in all locations (data not shown). An To explore the effects of self-antigen-mediated BCR accumulation of B cells in the marginal zone became signaling, in this study we examined the development prominent by 8 weeks of age (Figure 1B). This MZ B of B cells bearing BCR immunoglobulin (Ig) for the Thy-1/ cell development by ATA-expressing cells was accom- CD90 glycoprotein (anti-Thy-1 autoantibody; ATA). In panied by the appearance of large numbers of Ig-pro- the Ig H/L transgenic mouse line 3369, essentially all ducing plasma cells in the red pulp of the spleen (Figure immature B cells express the transgenic ATA IgM BCR, 1B, arrow). There was also Thy-1-dependent serum ATA encoded by germline IgH/L genes (Hayakawa et al., antibody (Figure 1C, compare RagKO and RagKO- 2003). The ATA BCR recognizes a species- and cell ThyKO), at 70% of the level found in wt ATA mice, pro- type-specific glycoepitope on mouse Thy-1 (Thy-16C10) duced from B-1 cells (Hayakawa et al., 2003). ATA MZ B as a self-antigen, abundantly produced by thymocytes cells present in RagKO background animals are direct from all strains of mice (Gui et al., 1999). Thy-1 is a precursors for the spleen plasma cells, also contribut- 25–30 Kd GPI (glycosylphosphatidylinositol)-anchored ing to serum ATA, as shown by MZ B cell transfer to membrane glycoprotein that can be cleaved and re- nontransgenic RagKO recipients (data not shown). This leased from the cell surface (Vitetta et al., 1974; Low was a very different picture from the low amount of and Kincade, 1985; Bergman and Carlsson, 1994). Thy- spontaneous ATA production in ThyKO mice (Figure 16C10/ATA BCR signaling is able to mediate both posi- 1C). These data suggest that, similar to B-1 cells, MZ tive and negative effects by exposure to Thy-1 self-anti- B cell development could occur independently of FO B gen. The presence of Thy-1 antigen in a normal Thy-1+ cell formation and these cells could contribute to serum environment promotes B-1 B cell development as a mi- autoantibody production in the absence of T cells. nor B cell subset, producing ATA in serum, as a type of In RagKO mice, the amount of self-Thy-1 antigen is positive selection (Hayakawa et al., 1999). In contrast, expected to be significantly reduced from wild-type an arrest of maturation is observed in spleen where levels, due to a block early in T cell development, but bone marrow-derived B cell development/maturation is it will not be absolutely lacking due to the expression normally completed. Thus, as a consequence of nega- of Thy-1 antigen by residual immature thymocytes (cor- tive selection, FO B and MZ B cells are absent (Haya- responding to about 1% of wild-type numbers, see Fig- kawa et al., 2003). However, a lack of Thy-1 engage- ure S1A in the Supplemental Data available with this ment in Thy-1-deficient gene-targeted mice permits article online). Therefore, we tested whether distinctive ATA B cell maturation to proceed to the IgMlo FO B cell MZ B development in RagKO background mice was subset (Hayakawa et al., 2003). This transgenic model due to the absence of T cells or, instead, to the pres- system allowed us to assess the relative effect of dif- ence of low levels of antigen, by performing cell ferent BCR signal strength on naive immature B cells transfers of hematopoietic stem cells from 3369ThyKO by altering Thy-1 antigen quantity in mice and also in mouse bone marrow into either RagKOThyKO or RagKO culture to modulate the BCR crosslinking signal. We recipients (Figure 1D). In both types of transfers, T cells demonstrate here the critical role of BCR signaling will be generated from the stem cells and any cell types strength in determination of B cell subset fate and its arising from donor bone marrow will lack expression of consequent influence on Notch 2 susceptibility. Thy-1. Although B cells and T cells were engrafted at similar high levels in both cases (30%–40% B and 40% Results T cells), generation of the CD21hi MZ B cell phenotype occurred only in the RagKO recipients. In clear con- MZ B Cell Development Promoted by Self-Antigen trast, in the RagKOThyKO recipients, engrafted mature In 3369 ATA mice with a Thy-1+ wild-type background, B cells showed a FO B cell phenotype. A predominance exposure to self-Thy-1 antigen resulted in a block of of B cells with the distinctive MZ B cell phenotype was development of most splenic ATA B cells at the imma- seen 1–2 months after transfer into RagKO recipients, ture CD24hiCD21– stage (Figure 1A, wt; Hayakawa et when most of the B cells with MZ B cell phenotype al., 2003). In the absence of Thy-1, using Thy-1 gene- were still within the white pulp (data not shown). More targeted mice, ATA B cells progressed to populate a than 5 months after transfer, nearly all B cells in the typical mature FO B cell compartment, as demon- RagKO mice were located in the MZ, in clear contrast strated by a CD24loCD21medCD23hi phenotype (Figure to the FO B cells predominating in RagKOThyKO recipi- 1A, ThyKO; Hayakawa et al., 2003). Surprisingly, when ents (Figure 1D). MZ B cell development was also asso- the ATA BCR was expressed on a Rag-1 (recombination ciated with higher ATA serum titers than seen with FO activating gene-1) null background, there was exclusive B cell development (data not shown). Thus, the pres- development of B cells with a typical MZ B phenotype ence of self-antigen was a critical environmental factor, in the spleen (Figure 1A, RagKO). This CD24medCD21hi promoting development of ATA B cells toward a MZ B CD23lo B cell development was specific to the spleen cell fate in the spleen, rather than a FO B cell fate. and was not found in lymph nodes, peripheral blood, or the peritoneal cavity (see Figure 1A legend). These ATA MZ B Cell Development by Low MZ B cells arose from an apparently normal population Amount of Self-Antigen of immature B cells, as assessed by surface phenotype Although our data with RagKO mice demonstrated the (not shown) and histology (Figure 1B) found in animals importance of Thy-1 antigen in MZ B cell development, B Cell Subset Decision 299

Figure 1. Self-Antigen Mediates Exclusive Development of ATA B Cells to the MZ B Cell Compartment in 3369RagKO Mice (A) Spleen ATA B cells from 2-month-old 3369 mice on wild-type (wt), Thy-1 gene knockout (ThyKO), or Rag-1 gene knockout (RagKO) background. Dagger indicates short lived. Total spleen cell numbers were similar (7–10 × 107) in analysis of more than 10 mice per group. ATA B cells were gated as ATA idiotype+ B220+ (and endogenous IgMb–) cells. ATA B cells in spleen of wt, ThyKO, and RagKO background were 30%–35%, 40%–50%, and 70%–80%, respectively. In 3369RagKO mice, nearly all ATA B cells circulating in blood were at the immature AA4+CD21lo/– CD24hi stage. Lymph nodes were undetectable in RagKO, except for the mesenteric lymph nodes, where ATA B cells (30%) showed a somewhat progressed but still immature phenotype (AA4lo/– CD21lo CD24hiCD23med/–), with a lack of CD21hi cells. There was no increase in peritoneal cavity resident cell number (2 × 106 cells) in 3369RagKO mice, where ATA B cells were 60%–80% of total, with a more differentiated phenotype reminiscent of “B1b” (AA4–CD21lo+CD24medCD5–CD11blo+), either CD23+ or CD23–. Thus, CD21hi ATA B cells were restricted to the spleen. (B) Staining of spleen sections from 3369RagKO mice at indicated ages, from primordial follicle (4 days) to typical MZ B cell expansion at 8 weeks of age. MZ B cell development was accompanied by the appearance of ER-TR9+ macrophage clusters in the MZ, CR1/2hi+ follicular dendritic cells within the white pulp, and plasma cells with high cytoplasmic IgM content in the red pulp (indicated by arrow). (C) Self-antigen-dependent serum ATA production in 3369RagKO mice, similar to that from B-1 B cells in wt mice. Sera from 3369 mice on different backgrounds as indicated. Data from 5 samples per group. (D) Self-antigen-dependent ATA MZ B cell development from 3369ThyKO stem cells. Spleen ATA B cell analysis and spleen section staining 5 months after stem cell transfer. Representative of 4 mice per group. it was unclear whether this ATA MZ B cell development Thy-16C10, possibly by non-T lineage cells, in RagKO was due to an abnormally lower amount of antigen in mice. In normal mice, Thy-16C10 is a mouse- and cell RagKO than normal or else to a unique expression of type-specific glycoepitope, predominantly restricted to Immunity 300

the T cell lineage, and not (or poorly) expressed on Thy- it remained unclear whether these observations could 1 glycoprotein of brain cells, natural killer cells, or bone be interpreted as positive selection, since the effect of marrow stromal cells (Gui et al., 1999); furthermore, antigen might be indirect. Furthermore, it was uncertain there was no significant expression detectable on cells whether any positive effect by antigen fulfilled the clas- of other organs (see Figure S2D, wt). However, rare Thy- sic criteria for positive selection as used in T cell devel- 16C10+ nonlymphoid cells could conceivably play a criti- opment, i.e., rescue of immature cells from death (Kisie- cal role. Thus, to test whether the quantity of self-anti- low et al., 1988; Jameson et al., 1995). Thus, to directly gen produced from thymocytes was a critical factor, we address the effect of BCR crosslinking signal strength generated a series of Thy-1 transgenic mouse lines, on immature B cells, we carried out an ex vivo antigen with Thy-1 expression driven by the proximal lck pro- stimulation, using naive immature ATA B cells from moter and, thus, preferentially expressed by immature 3369ThyKO spleen. Though expression of surface IgM thymocytes (Chaffin et al., 1990). Lines expressing occurs abnormally early in 3369 mouse BM, due to moderately higher, moderately lower, or much lower transgene Ig H/L chain coexpression, developmental levels of Thy-16C10, compared to wild-type levels, were and maturational changes of cell surface phenotype selected, bred to a ThyKO background, and then crossed were similar to nontransgenic B cells in the absence with 3369ThyKO mice, so that the thymocyte/T cell- of Thy-1 antigen. These included changes in IgM BCR derived transgene Thy-1 was the only source of Thy-1 quantity (Hardy et al., 1982) and acquisition of apopto- self-antigen (see Figures S2C and S2D). sis resistance (Tomayko and Cancro, 1998) during mat- We found that T cell development and early bone uration in the spleen, as shown in Figure 3A. Thus, im- marrow ATA B cell development proceeded similarly in mature AA4 (CD93)+ B cells expressed high levels of all mice. Importantly, the effect of self-antigen dose IgM with poor in vitro survival, while mature AA4– FO B was readily apparent from analysis of ATA B cells in the cells expressed lower IgM levels and 10 times better spleen. Figure 2 shows analyses of surface phenotype survival in vitro. AA4+ and AA4– cells correspond to (Figure 2A), histology (Figure 2B), serum ATA level (Fig- CD24hiCD21– and CD24loCD21med cell populations, re- ure 2C), and calcium signaling (Figure 2D) in splenic spectively (Allman et al., 2001; Hayakawa et al., 2003). ATA B cells. Antigen expression at levels moderately The earliest consequence of BCR signaling was mea- higher than wild-type (lck-Thy 130) resulted in a B cell sured by calcium response of 3369 ThyKO spleen cells maturation arrest in an anergic state, as defined by ab- to stimulation with different doses of Thy-1 antigen sence of BCR-mediated calcium flux, similar to the situ- (Figure 3B). Mature ATA B cells showed a more sus- ation in Thy-1 wild-type mice (Figures 2A and 2D, and tained, higher-amplitude calcium response at the high- see Figure S1B; Hayakawa et al., 2003); thus, negative est antigen dose (inhibited by the calcium chelator, selection occurred by expression of transgene Thy-1. EGTA) than did immature B cells (Figure 3B, red). How- Lower Thy-1 levels allowed more maturation, a less an- ever, we found that immature B cells were more sensi- ergic state (Figure 2D), a progressive increase in the tive to lower antigen doses, being able to generate a hi numbers of B cells with a CD21 MZ B cell phenotype 2+ substantial [Ca ]i rise with 10-fold lower antigen levels (Figure 2A), and an expanded MZ histology (Figure 2B, 2+ (Figure 3B, blue) and a slow [Ca ]i rise at a 100-fold lck-Thy 10). Serum ATA data provided evidence of a lower antigen dose than that required by mature cells positive effect by the presence of transgene Thy-1 (Fig- (Figure 3B, green). When purified B cells were cultured ure 1C), in which lck-Thy 130 mice showed the highest for 2 days in the presence of different antigen doses + serum titer and frequency of mature CD5 B-1 cells in (Figure 3C), mature B cells showed an antigen dose- the peritoneal cavity (see Figure 1C legend), similar to dependent activation (and proliferation), with the high- 3369 wt. These effects were not observed in the Thy-1- est viability seen at the highest antigen levels. In com- negative littermate controls, marked “Thy 0” (Figures parison, immature B cells showed a low and biphasic 2A–2D), which showed a predominance of mature FO viability curve: high antigen doses induced relatively B cells. Functionally, the MZ B cells generated in lck- greater cell death, whereas rescue from cell death oc- Thy 10 mice were distinctive as compared to naive ma- curred at lower doses (with an optimum in the 0.1–1 ture B cells, exhibiting rapid and elevated calcium sig- ng/ml range; 4–40 pM). Dose-dependent CD5 induction naling when stimulated with Thy-1 antigen (Figure 2D), and MHC class II upregulation occurred similarly for resembling the calcium signaling seen in MZ B cells both mature and immature B cells (Figure 3C) and was from 3369 RagKO mice (see Figure S1B). The relative inhibited by cyclosporine A (data not shown). Thus, amount of Thy-16C10 found in the thymus correlates while a strong BCR-medited signal led to “negative se- with serum Thy-1 glycoprotein level (as will be shown lection,” a lower BCR signal induced “positive selec- below), and peripheral exposure of B cells to such anti- + gen was suggested by a dose-dependent, surface CD5 tion” in immature B cells. Though the AA4 cells in- + induction on ATA B cells (Figure 2E; Teutsch et al., 1995; cluded some more progressed CD23 cells (Hayakawa Azzam et al., 1998). In conclusion, the quantity of et al., 2003) with slightly better survival, in addition to – Thy-1 antigen is critical in determining ATA B cell fate CD23 “T1” (Carsetti et al., 1995) cells, further separa- + – + (Figure 2E). Preferential ATA MZ B cell development in tion of AA4 cells into CD23 and CD23 produced sim- 3369 RagKO mouse spleen occurred optimally with ilar dose-dependent negative and positive selection re- lower amounts of antigen than seen in Thy-1 wild-type sults. Thus, positive selection can occur at the early ATA mice. immature B cell stage.

Positive Selection at the Immature Stage Positive Selection by Serum Thy-1 Though our in vivo analysis demonstrated positive con- Positive selection may depend on the antigenic form sequences of self-antigen on MZ B cell development, (Healy and Goodnow, 1998), so that soluble Thy-1M B Cell Subset Decision 301

Figure 2. ATA MZ B Cell Development Is Fostered by Low Levels of Self-Antigen Spleen cell analysis of ATA B cells produced by crossing 3369 ThyKO.C.B17 and lck-Thy-1Tg.ThyKO.C57BL/6 mice. (A) Differences in spleen ATA B cell surface phenotype. ATA B cells in spleen are gated as in Figure 1A. CD21hi MZ B cell region is marked by a square region. WT-Thy100 = 3369WT; Thy 0 = Thy-1Tg–/–.ThyKO. Representative of 4 mice per group (see [E]). Total spleen cell numbers were all similar (7–9 × 107). ATA B cells in spleen of lck-Thy 130, wt, lck-Thy 60, lck-Thy 10, and Thy 0 mice were 30%, 30%, 50%, 50%, and 50%, respectively. In lck-Thy 10 mice, B cells in the lymph nodes and peripheral blood were similar to mature FO B cells (AA4–CD21+CD23+), but with somewhat higher CD24 expression than naive FO B cells in Thy 0 mice. Peritoneal cavity cell numbers were all similar (2–3 × 106), where ATA B cells comprised 50%–80% in all Thy-1Tg+ mice but only 20% in Thy-1Tg– mice; CD5+ B cells with typical B-1 cell phenotype were only detectable in lck-Thy 130 mice. The peritoneal cavity B cells in lck-Thy 10 mice showed a CD5– “B-1b” cell phenotype. Thus, CD21hi B cell predominance in lck-Thy-1 10 spleen was restricted to the spleen. (B) Staining of spleen sections from indicated mice. (C) Serum ATA production promoted by the presence of lck-Thy-1 transgene. Data from 4 mice per group. (D) Calcium mobilization assay of spleen B cells stimulated with Thy-1M (600 ng/ml) at4×106 spleen cells/ml. Data were gated for B220+ B cells. Anti-IgM stimulation showed similar trends, where lck-Thy 130 mouse B cells failed to respond, in contrast with Thy 0 mouse B cells (Figure S4). (E) ATA B cell fate determined by amount of self-antigen for conventional B cell development/maturation in spleen. Blue columns indicate surface CD5 levels on total ATA spleen B cells. CD5 level on mature T cells was 60 U (staining units). Black dots represent MZ B cell frequency out of total ATA B cells in individual mice.

membrane may not be equivalent to physiologic Thy-1 leased from activated endothelial cells and fibroblasts self-antigen. Therefore, we next analyzed the effect of (Saalbach et al., 1999). In mice, we found that Thy-1 is Thy-1 in serum, which is likely the antigenic form expe- detected in serum of 2-month-old wild-type mice at rienced by developing ATA B cells in the periphery. In about 1.8 ng/ml on average, with similar levels found in humans, thymocytes or T cells are not the major source mice lacking B cells (JHKO). Serum Thy-1 in RagKO of Thy-1 (Vidal et al., 1990); however, soluble Thy-1 is mice is significantly reduced, but still detectable (80 pg/ detectable in human serum by ELISA, presumably re- ml) above the background level in ThyKO mice (Figure Immunity 302

Figure 3. Immature ATA B Cell-Positive Se- lection In Vitro (A) ATA B cell maturation in 3369 ThyKO mice. Progressive maturation in the BM (first panel from left) corresponds to decreasing AA4 and increasing B220/6B2 and MHC class II levels. Right three panels show gat- ing regions of immature (AA4+) and mature (AA4–) B220+ ATA B cell fractions in 5-week- old 3369ThyKO mouse spleen, a comparison of surface IgM levels, and cell viability during 2 day culture of purified B cells. Spleen AA4+ and AA4– B cell data are shown in red and blue, respectively. In the IgM staining histo- grams, nontrangenic C.B 17 mouse spleen B cell IgM level (predominantly mature AA4– B cells) is shown in black for comparison, and the thin black line indicates unstained total B cells. (B) Differential dose dependence of antigen- mediated calcium response in immature versus mature ATA B cells (gated for the analysis as shown in [A]). For right panels, 1 mM EGTA was added together with antigen. Spleen cells were stimulated at 4 × 106 cells/ ml. Representative data from four experiments. (C) 2 day culture of purified ATA B cells shows positive selection by low-dose antigen. Dotted line indicates baseline viability without antigen. Data were based on approximately 20,000 cells analyzed by flow cytometry for most groups (i.e., 400–4000 live cells) except for activated mature B cell cultures at higher antigen doses (10 and 20 ng/ml), where 30,000 cells were analyzed. In- duction of CD5 message from immature B cells was detectable from 3 hr after stimulation and from day 1 by surface staining in high-antigen dose cultures (data not shown). Data shown are representative of eight experiments. U = staining unit.

4A, left). We could also detect lck-Thy-1 transgene- right). B cells appear to be constantly exposed to such derived Thy-1 in serum, where lck-Thy 130 mice circulating soluble Thy-1 glycoprotein, as suggested by showed the highest, followed by lck-Thy 60 (Figure 4A, a low level of anti-Thy-1 antibody staining of ATA B cells

Figure 4. Positive Selection by Serum Thy-1

(A) Serum Thy-1 levels by ELISA assay. Left: sera from 2-month-old wt (C.B17 and BALB/C), JHKO.CB17, and RagKO.BALB/C mice were compared to their ThyKO counterparts (<0.02 ng/ml). Right: sera from lck-Thy-1 Tg mouse lines and their Thy-1 Tg– counterpart with endogenous Thy-1KO background. Thy-1 content is expressed in either absolute concentration for Thy-1.2 or relative units for Thy-1.1. (B) Positive selection by serum antigen. 2 day culture of AA4+ immature B cells from 3369ThyKO (ATA B, circle) or from BALB/c (control B, + – triangle) mice with or without 1:100 diluted sera from individual JHKO (Thy-1 , blackfilled) or JHKOThyKO (Thy-1 , unfilled) mice. x fold: fold increase by serum addition relative to medium alone. B Cell Subset Decision 303

in peripheral blood, likely due to Thy-1 bound to the the absence of the Notch 2 signal, the contribution of BCR (Figure S3). BCR crosslinking could still be appreciated by observ- To test whether such serum Thy-1 bound by the ATA ing increased accumulation of CD23+ cells in cultures BCR could mediate positive selection, we analyzed di- with antigen, relative to those with BAFF alone (Figure luted JHKO mouse sera in the immature B cell culture 5A, third and fourth panels, blue), together with signifi- assay, in comparison to JHKOThyKO sera, attempting cant upregulation of CD11a integrin αL expression (Fig- to provide a low-level Thy-1 crosslinking signal compa- ure 5C; Lu and Cyster, 2002). These results support a rable to that experienced in low Thy-1-expressing MZ B-FO B subset division model, where Notch 2 sig- lo + RagKO or lck-Thy 10 mice. Use of JHKO mouse sera naling plays a role at the late CD21 CD23 immature B helps to minimize potential antigen-antibody complex- cell stage, similar to transitional “T2” (Allman et al., mediated effects from serum. As shown in Figure 4B, 2001), acting in concert with already initiated positive Thy-1-containing serum at 1:100 dilution (18 pg/ml, on selection signals (or signals emanating from moder- average) had a positive effect on cell survival in culture ately increased BCR signaling) in the presence of BAFF, and induced MHC class II, specific for ATA immature B further upregulating CD21 and facilitating differentia- cells. These effects were not seen with serum from Thy- tion to the MZ subset (Figure 5D). 1-negative mice. Thus, B cell-positive selection can occur as a consequence of exposure to naturally occur- Discussion ring Thy-1 self-antigen in serum. A key finding of this study is the demonstration that Notch 2 Signal-Dependent MZ B Cell Differentiation self-ligand-mediated positive selection can occur at from Positively Selected Cells the immature stage in B cell development. The rescue Not unexpectedly, a BCR crosslinking signal alone of immature cells from death by a BCR crosslinking sig- failed to induce MZ B cell maturation in culture. It has nal provides strong support for B cell-positive selec- been known that Notch 2 signaling is critical for MZ tion, and we show here that BCR signal strength plays B cell development, and the spleen microenvironment a role in determining immature cell maturation into provides BAFF, known to promote both FO B cell and either of the two functionally distinct FO and MZ B cell MZ B cell maturation and survival. To assess how these compartments, both in mice and in culture. The promo- non-BCR-mediated signals function in self-antigen- tion of MZ B cell development by exposure to a defined dependent ATA MZ B cell development, an activated transgenic antigen, with FO B cell generation in its ab- intracellular fragment of Notch 2 (N2IC), linked to GFP sence, indicates a BCR signal strength requirement that by an IRES, was introduced by retroviral transduction is higher for MZ B cell than for FO B cell development. into immature naive ATA B cells, isolated from the bone Our B cell culture results provided evidence that this marrow and preexpanded by culture with IL-7. effect of antigen on MZ B cell development is a direct After withdrawal of IL-7, immature B cells began to consequence of antigen-mediated BCR crosslinking on die in cultures with medium alone, resulting in 5% via- immature B cells. However, it is also clear from our bility after 5–7 days (Figure 5A). Addition of low-dose study that B cell maturation proceeds in the context of antigen or BAFF, separately or in combination, pre- the microenvironment of the organ, where non-BCR- vented immature B cell death and promoted matura- mediated signal(s) play equally important roles in regu- tion, inducing CD23 and low levels of CD21, in the ab- lating B cell development and differentiation. Appa- sence of an activated Notch 2 signal (Figure 5A and rently normal ATA B cell maturation to the FO B cell legend). The CD23+ cells generated in these cultures subset, occurring in the absence of Thy-1 antigen ex- were not fully mature, retaining an IgMhiAA4medium sur- posure, appears to be an example of such non-BCR- face phenotype (data not shown), similar to cells at the mediated effects. The ability to rescue immature cells transitional CD21lo/med “T2” stage (Allman et al., 2001). from death by either antigen-engaged BCR or BAFF- We found that only a combination of all three signals mediated signals suggests that, though a positive se- (antigen, BAFF, and Notch 2) generated CD21hi cells, lection mechanism may not be required for FO B cell including cells with a MZ B cell phenotype, CD21hi development, an optimal level of BCR ligation may pro- CD23lo/− (Figure 5A, fourth panel, red dots). Signals mote B cell maturation or survival (as FO B cells) if the from BAFF plus Notch 2 alone (Figure 5A, third panel, signal strength is sufficiently lower than the threshold red dots), or from antigen plus BAFF alone (Figure 5A, required to induce MZ B cells. Importantly, B cell devel- fourth panel, blue dots), did not generate significant opment may be more dependent on BCR signals, if numbers of CD21hi cells, indicating that both a BCR non-BCR maturation signals are deficient in the micro- signal and Notch 2 activation were required. Combina- environment. tion of these three signals led to proliferation and acti- However, understanding signal “strength” requires vation of large CD21hiCD23+ cells (L), while CD21hi caution, since clearly not all BCR engagement leads to CD23– MZ B cells were of medium-small size (MS), sug- positive selection. As predicted from previous studies gesting that MZ B cells differentiated in vitro from the (Healy and Goodnow, 1998; King and Monroe, 2000), CD23+ to CD23– cell stage (Figure 5B, middle two pan- we observed that a commonly used anti-IgM cross- els). These CD21hi cells also showed increased expres- linking stimulation approach induced mature B cell acti- sion of CD1 on both CD23+ and CD23– subsets (Figure vation but failed to rescue immature ATA B cells from 5B, right two panels), with CD21hiCD23– cells showing death at all concentrations tested (data not shown). the highest expression of CD9 (data not shown), char- Though the reasons for such ligand-dependent differ- acteristic of MZ B cells (Martin and Kearney, 2002). In ences remain unclear, one plausible factor is the quality Immunity 304

Figure 5. MZ B Cell Differentiation In Vitro (A) Generation of CD21hi CD23lo/– cells requires BAFF and Notch2 signals, in addition to antigen. Five day immature ATA B cell culture transduced with N2IC or empty vector. Cell viabilities are shown within the figure. N2IC-GFP+, vector-GFP+, and GFP– cells are presented as red, green, and blue dots, respectively. N2IC-GFP+ cells comprised approximately 2% of live cells except for the Thy-1M + BAFF group, which was higher (9.5%). Vector-GFP+ cells comprised approximately 65% of live cells and there was no surface phenotype differences between empty vector-GFP+ and GFP– cells in all culture groups analyzed. There was no CD21 expression in culture with medium alone (i.e., equal to the “no stain” control level), indicated as relative CD21 = “1.0.” Data are representative of three experiments. (B) Activation and MZ B cell differentiation in response to Notch 2 signal. Comparison of size and surface phenotype between N2IC-GFP+ (red), N2IC-GFP– (blue), and Vector-GFP+ (green) cells in Thy-1M + BAFF culture. Middle two panels: N2IC-GFP+ cells gated for large (L) and medium/small (MS) size. CD21hiCD23– cells (marked by star) were specifically enriched among N2IC-GFP+ cells of medium/small size. Right two panels: increased CD1d level in N2IC-GFP+ cell, of both large (thin red) and medium/small (thick red) size, in comparison to N2IC-GFP– cells (blue). There was no CD1 difference in vector-GFP+ (green) versus GFP– (blue) cells. (C) Increase of surface CD11a level in response to BCR crosslinking in the presence of BAFF. (D) MZ B cell differentiation scheme. Low level of self-antigen exposure increases BCR signal strength from constitutive low to intermediate level, leading to MZ B cell differentiation in the spleen microenvironment. of signal, as determined by the combination of BCR (Gui et al., 1999). Since the mouse Thy-1 peptide con- affinity and form of the ligand. Stimulation with anti- tains three glycosylation sites (Carlsson and Stigbrand, IgM, which binds to IgC␮ with high affinity, did not re- 1984), it is likely that the Thy-16C10 epitope may be reveal significant differences between immature and petitively arrayed per molecule, suited for inducing a mature B cells in their calcium responses, with a dose- sustained signaling, similar to other T cell-independent dependent decline similar to antigenic stimulation of antigens. The ATA BCR was originally isolated from the immature B cells expressing a high-affinity BCR (Cyster CD5+ B-1 subset, a population of unconventional B et al., 1996). In contrast, exposing the IgM BCR oligo- cells that is enriched for BCRs that transduce a rela- mer (Reth et al., 2000) to repetitive low-affinity antigen tively high strength of signal (Casola et al., 2004)in stimulation may be important in generating a sustained most immature B cells that precludes further conven- calcium signal, as postulated for coreceptor-dependent tional B cell development (Hayakawa et al., 2003), but T cell-positive selection (Werlen et al., 2003). The ATA that still allows B-1 cell development (Hayakawa et al., BCR used here is unmutated from the germline and rec- 1999). The use of this BCR, in concert with reducing ognizes a glycoepitope of Thy-1 with multiple charges the level of self-antigen, appears to be particularly in- B Cell Subset Decision 305

formative in examining B cell-positive selection in con- tion of Btk on this background could lower signaling ventional B cell development, based on a signal strength back into an “acceptable” range for MZ B cell selec- hierarchy. tion. In this view, FO B cells that do not experience In comparison to the monoclonal BCR mice, imma- significant BCR crosslinking are benefited from loss of ture B cells in normal mice generate a diverse repertoire Aiolos, so more enter the long-lived pool, but handi- of B cell specificities. In these normal immature B cells, capped by the loss of Btk. Another mutant where en- BCR signal strength is likely to be determined by com- hanced BCR signaling is anticipated, CD72 knockout binations of several components, including BCR affin- mice, have reduced FO B cells (Pan et al., 1999), which ity, BCR quantity, antigen amount, and antigen forms. might be interpreted as an enhancement of BCR signal- Thus, we expect B cell subset or fate decision in the ing, beyond even the level allowed for FO B cell devel- wild-type animal to be more complex than in the mo- opment. noclonal BCR mice studied here. An increase of surface Furthermore, it is unclear whether FO B cell defi- BCR quantity by transgenic Ig heavy chain expression ciency caused by lack of Btk or PLCγ2 signaling arises is often associated with increased MZ B cell frequency at the point of B cell subset fate decision or at a later (Heltemes and Manser, 2002; Martin and Kearney, 2002) maintenance phase. Since Btk and PLCγ2 are also in- as one parameter of BCR strength. Some generation of volved in non-BCR-mediated signaling pathways, such MZ B cells in our transgenic mice in the absence of as for IL-5 and BAFF signaling, respectively (Moon et Thy-1 may be explained on this basis. However, MZ B al., 2001; Hikida et al., 2003), it is possible that naive cells in unimmunized or germ-free animals exhibit more FO B cells may rely more on non-BCR signaling in com- biased BCR usage than FO B cells (Dammers et al., bination with constitutive tonic BCR signal for their 1999; Martin and Kearney, 2002; Dammers and Kroese, maintenance, compared with MZ B cells already dif- 2005), suggesting an important role of (self-) antigen- ferentiated. In this regard, B-1 cells greatly reduced in mediated selection for the MZ B cell subset. Taking all such BCR signal-deficient animals due to absolute reli- of these observations into consideration, positive se- ance by B-1 cells on BCR signal, may also be affecting lection by BCRs that have low affinity for certain self- on FO B cell generation (Wen et al., 2005), an interest- or commensal antigens appears to be a major contrib- ing issue to consider in future investigation. utor to MZ B cell development in normal animals, Although negative selection was predominant over increasing BCR signal strength relative to BCR alone B-1 cell-positive selection under exposure of normal and leading to differentiation conditions that permit ef- Thy-1 self-antigen amount, minor B-1 cells were gener- fective Notch 2 signaling. ated in our ATA BCR mice and contributed to high se- BCR signaling strength can also be modulated by ge- rum autoantibody production. We show here that MZ B netic alterations of molecules involved in BCR signaling cells also contribute to serum antibody levels, though pathway(s), changing B cell fate. A requirement for the to a lesser extent than B-1, and FO B cells showed the highest BCR signaling strength in CD5+ B-1 cell devel- lowest titer. In comparison, as expected, there seems opment is supported by numerous studies using BCR to be an upper threshold for positive impact by self- signaling-deficient/modified mice, indicating that this B antigen. A reduction of serum ATA was observed in cell subset is most sensitive to changes in BCR signal- mice overexpressing the Thy-1 transgene and generating (Hayakawa and Hardy, 2000; Cariappa and Pillai, ing a serum level 10-fold higher than wild-type, sug- 2002; Casola et al., 2004). In contrast to the B-1 subset, gesting induction of B-1 cell tolerance (data not whose development does not depend on BAFF and shown). These data suggest that the BCR crosslinking Notch non-BCR signals (Mackay and Browning, 2002; signal within appropriate range is a critical factor in nat- Tanigaki et al., 2002), FO and MZ B cells are dependent ural autoantibody secretion, rather than mitogen- or on such non-BCR signal(s) and it is not as straightfor- TLR (toll-like receptor)-mediated signaling alone, in un- ward to determine their requirement for relative BCR immunized healthy animals. Interestingly, under condi- signal strength from BCR signaling-deficient/modified tions of low Thy-1 antigen exposure (such as in mice of mouse models. The loss/mutation of Btk or PLCγ2, me- RagKO or lck-Thy-1 10 background) expanding MZ B diators of signals from the BCR, have previously been cells in spleen, ATA B cells in the peritoneal cavity ex- shown to block complete FO B cell maturation (Raw- hibited a phenotype reminiscent of CD5– “B1b” cells, a lings et al., 1993; Thomas et al., 1993; Wang et al., B cell subset reported to provide protection from path- 2000), while sparing MZ B cell development. Enhanced ogens (Alugupalli et al., 2004), suggesting that this B BCR signaling, such as by CD22 or Aiolos deficiency, cell subset could also be a product of positive selec- reduces MZ B cell frequency, whereas FO B cells re- tion, which may be contributing to serum autoantibody main normal or are even elevated, and addition of Btk in addition to MZ B cells. deficiency to the Aiolos null background restores MZ Our study highlights the important role of the amount generation (Cariappa et al., 2001; Samardzic et al., of self-antigen in B cell development, impacting both 2002). It is reasonable to interpret these data as indicat- positively and negatively. Newly formed B cells exit ing that only very weak BCR signaling allows MZ B cell from bone marrow, circulate, and enter peripheral sites entry, whereas FO B cell development proceeds with for their maturation. Since B cells directly bind native stronger signaling (Pillai et al., 2005). On the other hand, (nonprocessed) antigens, developing B cells in the an alternative hypothesis may account for these re- bloodstream will be constantly exposed to self-anti- sults: that a range of BCR signal strength, with lower gens where the fate of B cells is delicately regulated. and upper bounds, facilitates entry into the MZ B cell Though the natural autoreactive B cell repertoire is well pool, and then increased signaling would render poten- controlled, a change in the amount of self-antigen, such tial MZ precursors subject to deletion, whereas elimina- as by aging, could enhance autoreactive B cell genera- Immunity 306

tion that normally is silenced or ignored, shifting the clone was digested and ligated into the pMIG vector (MSCV-IRES- balance from tolerance to functional activity, with the GFP) (See Supplemental Experimental Procedures). N2IC-pMIG potential for pathogenesis. plasmid DNA was introduced into the Phoenix virus packaging cell line by the standard calcium-phosphate coprecipitation method. High-titer virus supernatant was added to IL-7 preexpanded (5 Experimental Procedures days) bone marrow cells from 3369 ThyKO mice (>95% surface IgM+ AA4hi+ immature B cells) for spin transfection, then incubated Mice at 37°C for an additional 3 hr. One day after transfection (5%–10% ATA Ig␮κTg mouse line 3369, Thy-1KO (“ThyKO”), and 3369Thy- N2IC-GFP+, 60%–70% vector-GFP+), IL-7 was withdrawn, cells 1KO (“3369ThyKO”), all on a C.B17 background, were described were divided, and the culture continued at 2.5 × 106 cells/ml in 24- previously (Hayakawa et al., 2003). Rag-1KO (“RagKO”) and J KO H well plate for 5–7 days in fresh medium without or with supple- mice were backcrossed to BALB/c and C.B17 mice, respectively, ments: Thy-1M 1 ng/ml, BAFF (Apotech, San Diego, CA) 0.1 ␮g/ml, for more than six generations. Studies were performed according alone or in combination. to institutional guidelines for animal use and care (IACUC).

Bone Marrow Stem Cell Transfer Generation of lck-Thy-1 Transgenic Mice Hematopoietic stem cells (HSC) were enriched from 1-month-old A mouse Thy-1.1 cDNA was inserted into the lck-hGH vector (Chaf- bone marrow of 3369ThyKO mice as c-Kithi+ IL-7R– AA4+ Lineage– fin et al., 1990) for generation of transgenic mice (see Supplemental cells by cell sorting. 3–4 × 104 HSC/mouse were injected intrave- Experimental Procedures). Selected candidate Thy-1.1 transgenic nously into lightly irradiated (300R) RagKO recipients; mice were lines were crossed with ThyKO.C57BL/6 mice to obtain an endoge- analyzed 1–6 months after transfer. nous Thy1.2 KO background. The nonallelic ATA epitope Thy-16C10 expression was restricted to the T-lineage, predominantly by thymocytes, as determined by flow cytometry analysis and tissue ab- Supplemental Data sorption (see Figure S2). Supplemental Data include four figures and Supplemental Experi- mental Procedures and can be found with this article online at Flow Cytometry Analysis and Monoclonal Antibody Reagents http://www.immunity.com/cgi/content/full/23/3/297/DC1/. Multicolor flow cytometry analysis and sorting was carried out using a FACS-VantageSE with DiVa (BD Biosciences, San Jose, Acknowledgments CA), equipped for three laser excitation; data were analyzed by FlowJo software (Tree Star, Ashland, OR). Monoclonal antibody We thank L. Tang, S. Hua, and J. Oesterling for technical assis- reagents used in this study are listed in the Supplemental Experi- tance, transgenic mice, and cell sorting. We also thank G. Koretsky mental Procedures. and D. Vignali for vectors, G. Nolan for the Phoenix line, and K. Campbell for calcium analysis. We also appreciate helpful com- Tissue Section Immunofluorescence Analysis ments on this manuscript from D. Kappes, M. Bosma, D. Wiest, Frozen spleen section staining and imaging were done as de- and K. Campbell of this institute and Drs. K. Rajewsky (Harvard scribed previously (Wen et al., 2005). Univ.) and W. Paul (NIH). This work was supported by grants from the NIH (K.H. and R.R.H.) and the Greenwald fellowship of the BCR Calcium Signaling Response FCCC (L.W.) Spleen cells were surface stained with antibodies to B220 and AA4 to allow distinguishing immature (AA4+) versus mature (AA4–) B220+ Received: March 12, 2005 B cells, then loaded with Indo-1 in buffer containing Pluronic Revised: August 12, 2005 F-127. Calcium signaling was analyzed by flow cytometry (405/485 Accepted: August 17, 2005 nm ratio) using FlowJo software (Hayakawa et al., 2003). Published: September 20, 2005

Thy-1 Antigen Stimulation In Vitro References Soluble Thy-1-containing plasma membrane was prepared from + CD5 KO.C.B17 (Thy-1.2 ) mouse thymocytes as previously de- Allman, D., Lindsley, R.C., DeMuth, W., Rudd, K., Shinton, S.A., and scribed (Hayakawa et al., 2003). The amount of Thy-1 protein in the Hardy, R.R. (2001). Resolution of three nonproliferative immature M membrane preparation was determined to be 1% of total (Thy-1 ), splenic B cell subsets reveals multiple selection points during pe- by densitometry comparison of silver-stained SDS-PAGE gels, in- ripheral B cell maturation. J. Immunol. 167, 6834–6840. cluding a Thy-1KO mouse thymocyte plasma membrane control, Alugupalli, K.R., Leong, J.M., Woodland, R.T., Muramatsu, M., and with an affinity purified Thy-1 protein standard (Gui et al., Honjo, T., and Gerstein, R.M. (2004). B1b lymphocytes confer T cell- 1999). Thy-1M stimulation for 2 day culture was carried out in a independent long-lasting immunity. Immunity 21, 379–390. U-bottom 96-well plate using8×104 purified B cells/well in 150 ␮l culture medium. Cells were pooled from duplicate cultures, Azzam, H.S., Grinberg, A., Lui, K., Shen, H., Shores, E.W., and Love, washed, and stained for flow cytometry analysis to determine via- P.E. (1998). CD5 expression is developmentally regulated by T cell bility (propidium iodide staining), surface phenotype, cell size, and receptor (TCR) signals and TCR avidity. J. Exp. Med. 188, 2301– recovered total cell numbers. 2311. Benschop, R.J., and Cambier, J.C. (1999). B cell development: sig- Serum ATA and Thy-1 Measurements by ELISA Assay nal transduction by antigen receptors and their surrogates. Curr. Serum ATA was measured by ELISA assay using 19A4 anti-ATA idi- Opin. Immunol. 11, 143–151. otype antibody as a plate coat and anti-mouse IgM coupled with Bergman, A.S., and Carlsson, S.R. (1994). Saponin-induced release alkaline phosphatase as detector (Hayakawa et al., 2003). Serum of cell-surface-anchored Thy-1 by serum glycosylphosphatidyli- Thy-1 was measured using anti-pan Thy-1 (G7) as plate coat and nositol-specific phospholipase D. Biochem. J. 298, 661–668. biotin anti-Thy-1.2 (30H12) or biotin anti-Thy-1.1 (HIS51, for lck- Cariappa, A., and Pillai, S. (2002). Antigen-dependent B-cell devel- Thy-1 transgenic mouse sera), amplified with HRP-avidin in combi- opment. Curr. Opin. Immunol. 14, 241–249. nation with TMB (tetramethyl benzidine; BioFX, MD) as detector. Thy-1M from CD5KO. C.B17 or AKR mice were used as Thy-1.2 or Cariappa, A., Tang, M., Parng, C., Nebelitskiy, E., Carroll, M., Geor- Thy-1.1 binding standards, respectively. gopoulos, K., and Pillai, S. (2001). The follicular versus marginal zone B lymphocyte cell fate decision is regulated by Aiolos, Btk, Notch 2 IC Retroviral Transduction and Culture and CD21. Immunity 14, 603–615. A cDNA clone of the mouse Notch 2 intracellular domain (N2IC) Carlsson, S.R., and Stigbrand, T. (1984). Partial characterization of was prepared by PCR amplification and inserted into pCR2.1- the oligosaccharides of mouse thymocyte Thy-1 glycoprotein. Bio- TOPO vector. After verification of sequence, the selected plasmid chem. J. 221, 379–392. B Cell Subset Decision 307

Carsetti, R., Kohler, G., and Lamers, M.C. (1995). Transitional B cells King, L.B., and Monroe, J.G. (2000). Immunobiology of the imma- are the target of negative selection in the B cell compartment. J. ture B cell: plasticity in the B-cell antigen receptor-induced re- Exp. Med. 181, 2129–2140. sponse fine tunes negative selection. Immunol. Rev. 176, 86–104. Casola, S., Otipoby, K.L., Alimzhanov, M., Humme, S., Uyttersprot, Kisielow, P., Teh, H.S., Bluthmann, H., and von Boehmer, H. (1988). N., Kutok, J.L., Carroll, M.C., and Rajewsky, K. (2004). B cell recep- Positive selection of antigen-specific T cells in thymus by re- tor signal strength determines B cell fate. Nat. Immunol. 5, 317–327. stricting MHC molecules. Nature 335, 730–733. Chaffin, K.E., Beals, C.R., Wilkie, T.M., Forbush, K.A., Simon, M.I., Kraus, M., Alimzhanov, M.B., Rajewsky, N., and Rajewsky, K. (2004). and Perlmutter, R.M. (1990). Dissection of thymocyte signaling Survival of resting mature B lymphocytes depends on BCR signal- pathways by in vivo expression of pertussis toxin ADP-ribo- ing via the Igalpha/beta heterodimer. Cell 117, 787–800. syltransferase. EMBO J. 9, 3821–3829. Kumararatne, D.S., Bazin, H., and MacLennan, I.C. (1981). Marginal Cinamon, G., Matloubian, M., Lesneski, M.J., Xu, Y., Low, C., Lu, zones: the major B cell compartment of rat spleens. Eur. J. Immu- T., Proia, R.L., and Cyster, J.G. (2004). Sphingosine 1-phosphate nol. 11, 858–864. receptor 1 promotes B cell localization in the splenic marginal zone. Kumararatne, D.S., and MacLennan, I.C. (1982). The origin of mar- Nat. Immunol. 5, 713–720. ginal-zone cells. Adv. Exp. Med. Biol. 149, 83–90. Cyster, J.G., Healy, J.I., Kishihara, K., Mak, T.W., Thomas, W.L., and Liu, Y.J., Oldfield, M., and MacLennan, I.C.M. (1988). Memory B Goodnow, C.C. (1996). Regulation of B-lymphocyte negative and cell in T cell-dependent antibody responses colonize the splenic positive selection by tyrosine phosphatase CD45. Nature 381, marginal zones. Eur. J. Immunol. 18, 355–362. 325–328. Low, M.G., and Kincade, P.W. (1985). Phosphatidylinositol is the Dammers, P.M., de Boer, N.K., Deenen, G.J., Nieuwenhuis, P., and membrane-anchoring domain of the Thy-1 glycoprotein. Nature Kroese, F.G. (1999). The origin of marginal zone B cells in the rat. 318, 62–64. Eur. J. Immunol. 29, 1522–1531. Lu, T.T., and Cyster, J.G. (2002). Integrin-mediated long-term B cell Dammers, P.M., and Kroese, F.G. (2005). Recruitment and selection retention in the splenic marginal zone. Science 297, 409–412. of marginal zone B cells is independent of exogenous antigens. Mackay, F., and Browning, J.L. (2002). BAFF: a fundamental survival Eur. J. Immunol. 35, 2089–2099. factor for B cells. Nat. Rev. Immunol. 2, 465–475. Fukui, Y., Hashimoto, O., Sanui, T., Oono, T., Koga, H., Abe, M., Martin, F., and Kearney, J.F. (2002). Marginal-zone B cells. Nat. Rev. Inayoshi, A., Noda, M., Oike, M., Shirai, T., and Sasazuki, T. (2001). Immunol. 2, 323–335. Haematopoietic cell-specific CDM family protein DOCK2 is essen- Moon, B.G., Yoshida, T., Shiiba, M., Nakao, K., Katsuki, M., Takaki, tial for lymphocyte migration. Nature 412, 826–831. S., and Takatsu, K. (2001). Functional dissection of the cytoplasmic Girkontaite, I., Missy, K., Sakk, V., Harenberg, A., Tedford, K., Pot- subregions of the interleukin-5 receptor alpha chain in growth and zel, T., Pfeffer, K., and Fischer, K.D. (2001). Lsc is required for mar- immunoglobulin G1 switch recombination of B cells. Immunology ginal zone B cells, regulation of lymphocyte motility and immune 102, 289–300. responses. Nat. Immunol. 2, 855–862. Pan, C., Baumgarth, N., and Parnes, J.R. (1999). CD72-deficient Gui, M., Wiest, D.L., Li, J., Kappes, D., Hardy, R.R., and Hayakawa, mice reveal nonredundant roles of CD72 in B cell development and + K. (1999). Peripheral CD4 T cell maturation recognized by in- activation. Immunity 11, 495–506. creased expression of Thy-1/CD90 bearing the 6C10 carbohydrate Pillai, S., Cariappa, A., and Moran, S.T. (2005). Marginal zone B epitope. J. Immunol. 163, 4796–4804. cells. Annu. Rev. Immunol. 23, 161–196. Guinamard, R., Okigaki, M., Schlessinger, J., and Ravetch, J.V. Rawlings, D.J., Saffran, D.C., Tsukada, S., Largaespada, D.A., Gri- (2000). Absence of marginal zone B cells in Pyk-2-deficient mice maldi, J.C., Cohen, L., Mohr, R.N., Bazan, J.F., Howard, M., Cope- defines their role in the humoral response. Nat. Immunol. 1, 31–36. land, N.G., et al. (1993). Mutation of unique region of Bruton's tyro- Hardy, R.R., Hayakawa, K., Haaijman, J., and Herzenberg, L.A. sine kinase in immunodeficient XID mice. Science 261, 358–361. (1982). B cell subpopulations identified by two-color fluorescence Reichlin, A., Hu, Y., Meffre, E., Nagaoka, H., Gong, S., Kraus, M., analysis. Nature 297, 589–591. Rajewsky, K., and Nussenzweig, M.C. (2001). B cell development is Hayakawa, K., and Hardy, R.R. (2000). Development and function arrested at the immature B cell stage in mice carrying a mutation of B-1 cells. Curr. Opin. Immunol. 12, 346–353. in the cytoplasmic domain of immunoglobulin beta. J. Exp. Med. Hayakawa, K., Asano, M., Shinton, S.A., Gui, M., Allman, D., Stew- 193, 13–23. art, C.L., Silver, J., and Hardy, R.R. (1999). Positive selection of nat- Reth, M., Wienands, J., and Schamel, W.W. (2000). An unsolved ural autoreactive B cells. Science 285, 113–116. problem of the clonal selection theory and the model of an oligo- Hayakawa, K., Asano, M., Shinton, S.A., Gui, M., Wen, L.J., Dashoff, meric B-cell antigen receptor. Immunol. Rev. 176, 10–18. J., and Hardy, R.R. (2003). Positive selection of anti-Thy-1 autore- Roose, J.P., Diehn, M., Tomlinson, M.G., Lin, J., Alizadeh, A.A., active B-1 cells and natural serum autoantibody production inde- Botstein, D., Brown, P.O., and Weiss, A. (2003). T cell receptor-independent from bone marrow B cell development. J. Exp. Med. 197, pendent basal signaling via Erk and Abl kinases suppresses RAG 87–99. gene expression. PLoS Biol. 1, e53. 10.1371/journal.pbio.0000053 Healy, J.I., and Goodnow, C.C. (1998). Positive versus negative sig- Saalbach, A., Wetzig, T., Haustein, U.F., and Anderegg, U. (1999). naling by lymphocyte antigen receptors. Annu. Rev. Immunol. 16, Detection of human soluble Thy-1 in serum by ELISA. Fibroblasts 645–670. and activated endothelial cells are a possible source of soluble Heltemes, L.M., and Manser, T. (2002). Level of B cell antigen re- Thy-1 in serum. Cell Tissue Res. 298, 307–315. ceptor surface expression influences both positive and negative Saito, T., Chiba, S., Ichikawa, M., Kunisato, A., Asai, T., Shimizu, K., selection of B cells during primary development. J. Immunol. 169, Yamaguchi, T., Yamamoto, G., Seo, S., Kumano, K., et al. (2003). 1283–1292. Notch2 is preferentially expressed in mature B cells and indispens- Hikida, M., Johmura, S., Hashimoto, A., Takezaki, M., and Kurosaki, able for marginal zone B lineage development. Immunity 18, 675– T. (2003). Coupling between B cell receptor and phospholipase 685. C-gamma2 is essential for mature B cell development. J. Exp. Med. Samardzic, T., Marinkovic, D., Danzer, C.P., Gerlach, J., Nitschke, 198, 581–589. L., and Wirth, T. (2002). Reduction of marginal zone B cells in CD22- Hozumi, K., Negishi, N., Suzuki, D., Abe, N., Sotomaru, Y., Tamaoki, deficient mice. Eur. J. Immunol. 32, 561–567. N., Mailhos, C., Ish-Horowicz, D., Habu, S., and Owen, M.J. (2004). Schiemann, B., Gommerman, J.L., Vora, K., Cachero, T.G., Shulga- Delta-like 1 is necessary for the generation of marginal zone B cells Morskaya, S., Dobles, M., Frew, E., and Scott, M.L. (2001). An but not T cells in vivo. Nat. Immunol. 5, 638–644. essential role for BAFF in the normal development of B cells Jameson, S.C., Hogquist, K.A., and Bevan, M.J. (1995). Positive through a BCMA-independent pathway. Science 293, 2111–2114. selection of thymocytes. Annu. Rev. Immunol. 13, 93–126. Tanigaki, K., Han, H., Yamamoto, N., Tashiro, K., Ikegawa, M., Kur- Immunity 308

oda, K., Suzuki, A., Nakano, T., and Honjo, T. (2002). Notch-RBP-J signaling is involved in cell fate determination of marginal zone B cells. Nat. Immunol. 3, 443–450. Teutsch, M., Higer, M., Wang, D., and Wortis, H.W. (1995). Induction of CD5 on B and T cells is suppressed by cyclosporin A, FK-520 and rapamycin. Int. Immunol. 7, 381–392. Thomas, J.D., Sideras, P., Smith, C.I., Vorechovsky, I., Chapman, V., and Paul, W.E. (1993). Colocalization of X-linked agammaglobuli- nemia and X-linked immunodeficiency genes. Science 261, 355– 358. Tomayko, M.M., and Cancro, M.P. (1998). Long-lived B cells are distinguished by elevated expression of A1. J. Immunol. 160, 107– 111. Turner, M., Mee, P.J., Costello, P.S., Williams, O., Price, A.A., Duddy, L.P., Furlong, M.T., Geahlen, R.L., and Tybulewicz, V.L. (1995). Peri- natal lethality and blocked B-cell development in mice lacking the tyrosine kinase Syk. Nature 378, 298–302. Vidal, M., Morris, R., Grosveld, F., and Spanopoulou, E. (1990). Tis- sue-specific control elements of the Thy-1 gene. EMBO J. 9, 833– 840. Vitetta, E.S., Uhr, J.W., and Boyse, E.A. (1974). Metabolism of H-2 and Thy-1 (theta) alloantigens in murine thymocytes. Eur. J. Immu- nol. 4, 276–282. Wang, D., Feng, J., Wen, R., Marine, J.C., Sangster, M.Y., Parganas, E., Hoffmeyer, A., Jackson, C.W., Cleveland, J.L., Murray, P.J., and Ihle, J.N. (2000). Phospholipase Cgamma2 is essential in the func- tions of B cell and several Fc receptors. Immunity 13, 25–35. Wen, L., Shinton, S.A., Hardy, R.R., and Hayakawa, K. (2005). Asso- ciation of B-1 B cells with follicular dendritic cells in spleen. J. Im- munol. 174, 6918–6926. Werlen, G., Hausmann, B., Naeher, D., and Palmer, E. (2003). Sig- naling life and death in the thymus: timing is everything. Science 299, 1859–1863. Yasutomo, K., Doyle, C., Miele, L., Fuchs, C., and Germain, R.N. (2000). The duration of antigen receptor signalling determines CD4+ versus CD8+ T-cell lineage fate. Nature 404, 506–510.