Association of B-1 B Cells with Follicular Dendritic Cells in Spleen1
http://www.100md.com
免疫学杂志 2005年第11期
Abstract
Although CD5+ B-1 B cells have been recognized as an infrequent B cell subset in mice for many years, attempts to identify their histologic location in normal mouse spleen have proven difficult due to both their paucity and low level expression of CD5. In this study we have studied VH11/DH/JH gene-targeted mice, VH11t, that develop elevated numbers of CD5+ VH11/Vk9 B cells with an anti-phosphatidylcholine (anti-PtC) autoreactive specificity, allowing B-1 B cell detection by anti-PtC Id-specific Abs in spleen section staining. Using this approach we found that anti-PtC B-1 cells first appear within the white pulp in neonates, expand in association with follicular dendritic cells (FDC), and localize more centrally than other (non-B-1) IgDhigh follicular B cells in adults. Among neonatal B cells, CD5+ B-1 cells in both normal and VH11t mouse spleen and peritoneal cavity express the highest levels of CXCR5, which is important for FDC development. Injection of purified spleen or peritoneal B-1 cells into RAG knockout mice resulted in B-1 cell follicle formation in spleen, inducing FDC development and plasma cell generation. These results indicate that B-1 B cells are the first B cells to express fully mature levels of CXCR5, thereby promoting the development of FDC.
Introduction
A distinctive B cell subset in mice, B-1, expresses certain autoreactive specificities and is responsible for much of the natural serum autoantibody production (1, 2, 3). B-1 B cells were initially discovered based on CD5 expression, a surface phenotype resembling B cells increased in certain autoimmune diseases and B chronic lymphocytic leukemia (4, 5). Careful analysis by sensitive flow cytometry identified this CD5+ B cell subset as a rare population in normal adult mice, although its frequency changes depending on ontogeny and anatomical location. Whereas B-1 cell generation is frequent before birth, it declines after birth, eventually comprising only a small percentage of the B cells in adult mouse spleen. In contrast, the follicular B-2 B cell population increases over the first 2 mo of life, becoming the predominant B cell subset in adults. In comparison with this spleen pattern, B-1 cells maintain a high frequency in the peritoneal cavity throughout adulthood (2, 6).
The spleen is a critical organ for B cell maturation and differentiation, and provides a strategic site for establishment and organization of the immune system. Commencing after the appearance of the primordial follicle (7, 8), an organized white pulp architecture becomes established in spleen with mouse development, generating the marginal sinus, promoting myeloid and follicular stromal cell maturation, and eventually leading to B-T lymphocyte compartment segregation (9). Such microarchitectural development is an interactive process involving chemokines and the lymphotoxin (LT)/TNF gene family of cytokines (10, 11, 12, 13), in which B cells appear to play a critical inductive role in stromal cell maturation (14, 15, 16, 17, 18).
The initial B cell-stromal cell interaction involves CXCL13 (BLC/BCA-1)-expressing follicular dendritic cells (FDC)3 (19) binding to CXCR5 (BLR1)-expressing B cells (14, 17, 20). B-1 B cells express the highest levels of CXCR5 among B cells in adults (21, 22, 23), and considering the early appearance of B-1 B cells in mouse ontogeny, it was reasonable to expect that neonatal B-1 cells would also express CXCR5 and be able to precipitate FDC development in spleen. In exploring this process, the ability to identify their location is crucial. However, because of their paucity and low level expression of CD5 (4), detected only by sensitive flow analysis, attempts to identify the histologic location of B-1 cells in normal mouse spleen have proven difficult. For this reason, our strategy has been to use mice with a quantitatively increased B-1 cell population due to expression of the anti-phosphatidylcholine (anti-PtC) autoreactive specificity, a prototypic specificity associated with the B-1 B cell subset. Anti-PtC is an autoreactive specificity relatively highly represented in the mouse CD5+ B-1 cell subset (2, 24, 25), is naturally secreted into serum, and is predominantly encoded by either VH11/Vk9 or VH12/Vk4/5 (26, 27, 28). Considering its ability to bind to PtC exposed on senescent or bromelain-treated RBC (29, 30), anti-PtC autoantibody is presumed to play a role in the clearance of damaged or senescent erythrocytes from the circulation through the action of macrophages (31), which are abundant in red pulp in spleen. As previously shown with VH11 (32) and VH12 (33, 34) H chain transgenic and with JH locus-targeted VH12/DH/JH insertion (knockin) mice, genetic provision of rearranged VH11 (or VH12) H chain results in an increased, often overwhelming, frequency of anti-PtC CD5+ B-1 cells with the expression of appropriate Vk9 (or Vk4/5) L chains, presumed to be the result of positive selection for this autoreactive specificity by self Ag (3, 33, 35).
Interestingly, in targeted VH11/DH/JH1 insertion knockin mice, VH11t, we observed dynamic changes in the B cell repertoire and B cell populations during mouse development, resulting in the generation of multiple IgM+ B cell subsets, including CD5+ B-1, mature B-2, and marginal zone (MZ) B cells (36) in spleen, all at significant frequencies. Importantly, anti-PtC B cells are restricted to the CD5+ B-1 cell population. This allowed us to locate B-1 in comparison with other B cell compartments with the aid of anti-PtC Id-specific Abs detecting B-1 cells, instead of using anti-CD5. In this study we report a histologic analysis of such spleen cells; demonstrate the capacity of B-1 cells to form a B cell follicle and induce FDC, generate plasma cells, and conclude with a discussion of B-1 cell peritoneal-spleen dynamics.
Materials and Methods
Tissue section immunofluorescence analysis
Frozen section staining was performed as described previously (41). Briefly, spleens samples were embedded and frozen at –80°C. Frozen samples were sliced into 6-μm sections, air-dried, fixed with ice-cold ethanol for 15 min, washed with PBS, then blocked with PBS and 2% newborn bovine serum. Staining was performed at room temperature for 30 min for each step. Stained sections were washed with PBS and mounted using Fluormount G (Southern Biotechnology Associates), then viewed with a Nikon Optiphot epifluorescence microscope equipped with a Quad-Fluor (EF-1) four-cube filter holder and appropriate filter cubes. Images were recorded using the QImaging Retiga-1300 cooled CCD camera, processed with Openlab software (Improvision), and assembled with Canvas (ACD Systems of America). Unless otherwise stated, images were obtained with a x10 objective lens (original camera magnification, x264).
Cell transfer
For analysis of cell entry during the 2- to 24-h period, 5 x 107 VH11t spleen B220+CD5+ B-1 cells were injected i.p. per RAG KO mouse. For analysis of FDC development during the 7- to 10-day period, 2–3 x 107 B220+CD5+ B-1 cells were cell sorter purified from the peritoneal cavity of VH11t mice and injected i.p. into RAG KO mice. For long term self-reconstitution analysis (>2 mo), RAG KO mice were lightly irradiated (cesium, 300 rad), then injected i.p. with 3–5 x 106 cell sorter-purified B220+CD5+ peritoneal B-1 cells the following day. Similarly, spleen B-1 cells were sorted and injected i.v. into RAG KO mice for both short and long term analyses.
Results
Detection of VH11/Vk9 anti-PtC plasma cells from normal mouse spleen
In normal adult BALB/c mice, VH11/Vk9 anti-PtC B cells are rare in spleen, constituting <1% of the total spleen B cells (1–10% of CD5+ B-1 cells) as previously determined by their frequency in hybridoma panels (42) and by RNA message analysis (43). This low incidence in spleen (0.1–0.2% of the total adult spleen IgM+ B cells) has been confirmed recently by flow cytometric analysis using a combination of anti-VH11 and anti-Vk9 anti-PtC Id Abs, 3H7 and 13B5, respectively (B. G. Moon, S. A. Shinton, D. Allman, K. Hayakawa, and R. R. Hardy, manuscript in preparation). In spleen sections, it is difficult to unequivocally determine the location of such rare anti-PtC B cells by anti-Id. However, VH11/Vk9 anti-PtC plasma cells are readily detected as anti-Id brightly stained cells (Fig. 1), an indication of cells with high cytoplasmic anti-PtC content, in the red pulp outside the white pulp regions marked by the MOMA-1-stained metallophilic macrophage lining (44) (Fig. 1, A and B). They are costained with the anti-Vk9 (Vk9id, marked Vk9 in the following figures) and the anti-VH11 anti-PtC Id Abs (Fig. 1B). As seen with Vk9/B220 staining in Fig. 1A, these anti-Id brightly stained cells are B220 negative and also CD19– (which will be shown in Fig. 8), suggesting that they are at the stage of the differentiated surface IgM– Ab-secreting cells. They constitute between 10 and 50% of the total IgM plasma cells in the red pulp in average BALB/c mouse spleen, as represented in Fig. 1C (higher magnification). Thus, despite the paucity of anti-PtC B cells, B-1 cell-derived anti-PtC plasma cells are readily detectable in normal adult mouse spleen.
To characterize this B-1 localization relative to other non-PtC B cell compartments in more detail, we included IgD staining, because the majority of non-B-1 cells in adults are characteristically IgDhigh. The Vk9id/IgD staining data in Fig. 4 reveals first that most Vk9id+IgDlow/– B-1 cells (green) are clustered in the center of the follicle, whereas the Vk9id– IgDhigh (blue) cells are more densely located toward the follicle periphery. Some green Vk9id+ cells are also located outside the IgD+ region, at the zone between the red pulp (containing Vk9id bright plasma cells) and white pulp area (Fig. 4, left and right), suggesting their presence in the MZ, although none of these Vk9id+ cells are CD21high MZ B cells (Fig. 3B). Since, in this Vk9id/IgD section analysis, typical MZ B cells are not visualized (because they are Vk9id– and IgDlow/–) and some mature IgMlow B-2 cells might be missed due to their relatively lower IgD expression, we next incorporated analysis of B220 expression together with Vk9id and IgD (Fig. 5). Both IgDlow/– mature B-2 and MZ B cells express higher levels of B220 in contrast to B-1; thus, B-1 cells are recognized as green (Vk9id+IgDlow/–B220+/–), IgDhigh cells (mostly mature B-2) as purple (Vk9id–IgDhighB220+), and MZ B cells as red (Vk9id–IgDlowB220+); the potentially missed IgDlow mature B-2 cells will also be recognized as more red (Vk9id–IgDlowB220+) than purple. As Fig. 5 demonstrates, the central region of the B cell follicle shows a strong predominance of green staining, rather than yellow or orange, suggesting a relative B-1 predominance, surrounded by IgDhigh mature B-2 and IgDhighIgMhigh transitional cells (purple). In turn, red staining predominates in the MZ, indicating that typical MZ B cells cluster in this study, quantitatively more than B-1. Both B-1 (green) and mature B-2 cells (red/purple) are present in the red pulp at a similar frequency as those in peripheral blood circulation. Anti-PtC plasma cells in red pulp are recognized as the bright green-stained cells. Some of these plasma cells migrating out through the bridging channel are marked by an arrow in Fig. 5.
B-1 cells express the highest CXCR5 level among neonatal B cells
Anti-PtC cell clusters are already detectable from younger (2-wk-old) mouse spleen within the white pulp, as Fig. 2 section staining data has shown. In such neonates, all FDC networks are close to the marginal sinus boundary with the metallophilic macrophages, where Vk9id+ cell clusters are also found (Fig. 7A). As flow cytometry analysis shows (Fig. 7B), the Vk9id+ anti-PtC B cells constitute only 5–10% of the total spleen B cells, whereas the majority are not anti-PtC (Fig. 7B, spleen). Nearly all Vk9id– (non-PtC) B cells in such neonatal spleens are developing AA4+ immature B cells, IgMhighIgDlow/med (not shown), CD5–, and with a low level of the FDC CXCL13 chemokine receptor CXCR5 (Fig. 7B, blue), consistent with previous reports that immature B cells do not yet fully express CXCR5 compared with mature B cells (21, 45). Distinctively, the Vk9id+ anti-PtC B-1 cells have down-regulated AA4, are uniformly CD5+ as a result of Ag selection, and are already expressing a higher level of CXCR5 (Fig. 7B, red solid) at a level similar to that on mature B cells. In addition to the spleen, CD5+ CXCR5+ anti-PtC B-1 cells are also present in the neonatal peritoneal cavity (PerC) as an abundant AA4low/– mature PerC B cell population (Fig. 7B, red dash). CD5+ B-1 cells present in normal (BALB/C) neonatal spleen and PerC also show a surface phenotype comparable to the anti-PtC B cells in VH11t mice, including elevated CXCR5 expression (Fig. 7C). These data demonstrate the existence of B-1 cells as a CXCR5+ mature B cell source, arising before B-2 B cell maturation and consequent CXCR5 up-regulation. Both spleen and PerC B-1 cells continuously maintain a high level of CXCR5 throughout the life of the animal.
Induction of FDC, B-1 cell follicle formation, and plasma cell generation by B-1 cell transfer
CXCR5 expression by B-1 cells suggested that B-1 cells may be able to induce FDC maturation as do mature spleen B cells (16, 46, 47). To directly test this, purified anti-PtC B-1 cells from VH11t mice were injected into RAG KO mice that lack mature FDC (47), similar to newborn mice (8). In addition to i.v. or i.p. injection of spleen B-1 cells, PerC B-1 cells were injected i.p. in an attempt to determine whether the PerC B-1 cells dominating in neonatal PerC would be able to migrate to spleen. The use of B-1 cells purified from PerC washout cells has an advantage as well, by circumventing the risk of possible contamination by FDC copurified with B-1 cells from spleen. In both cell injection routes, B-1 cells entered into the white pulp via the weakly structured marginal sinus in adult RAG KO hosts, beginning to form CD19+Vk9id+ cell clusters 1 day after injection, without prior existing mature CR1/2high FDC (Fig. 8).
Induction of mature FDC and an FDC network in the white pulp became clear by 10 days after injection, as shown in this study for mice injected i.p. with PerC B-1 cells, in contrast to the uninjected RAG KO control (Fig. 9). All CR1/2high FDC were associated with CD19+ B cell follicles (Fig. 9A, left) within the marginal sinus endothelium expressing mucosal addressin cell adhesion molecule-1 (Fig. 9A, middle). Also, vast numbers of IgMhighCD19–Vk9high anti-PtC plasma cells were generated, localized to the red pulp (Fig. 9A, middle and right). Flow cytometric analysis confirmed that all CD19+ B cells, with weaker Vk9 section staining than plasma cells, were B-1 cells with retention of exclusive anti-PtC specificity and CD5+ phenotype (Fig. 9B). Some B-1 cells were also present in the MZ area (Fig. 9A, left) without a typical CD21high MZ B cell phenotype, similar to the data from intact VH11t animals shown in Fig. 4. The ability to induce FDC development together with B cell follicle formation by B-1 cells were comparable to those seen with mature B-2 cells purified from either VH11t or normal adult mouse spleen (purified as B220highCD23highAA4–CD5–CD43–VH11–Vk9– cells), except that there was greater plasma cell generation from B-1 cells than B-2 cells (data not shown).
Although such short term experiments required larger amounts of B cells for rapid induction of a readily visualizable follicular network, it has been shown previously that spleen or PerC B-1 cells from normal neonatal and adult mice are able to self-reconstitute in recipients over the long term by lower dose transfer (106 cells) (39, 40). Such B-1 self-reconstituted mice using VH11t mouse PerC anti-PtC B-1 cells revealed a similar spleen architecture profile, containing B-1 cell follicles, FDC, and plasma cells (data not shown). In addition, there was some recruitment of phagocytic ER-TR9+ macrophages in the MZ and stronger MOMA-1 staining than in uninjected RAG KO mice, suggesting either promotion of macrophage differentiation or an increase in differentiated macrophage numbers with time. Thus, B-1 cells, including those migrating from PerC, can initiate FDC development, promote microarchitecture organization, establish a B-1 cell subset, and sustain a plasma cell pool in spleen.
Discussion
In this study we observed B-1 B cell localization in the white pulp, in close association with the FDC network, and FDC induction by B-1 B cell transfer, similar to what has been found previously with mature B-2 B cells, commonly referred to as follicular B cells due to their association with FDC. Our analysis suggest that B-1 B cells are able to traverse into peripheral blood and re-enter the white pulp from the MZ, where minor vessels branching off the central arterioles deposit blood. Although a previous attempt using VH12μ H chain transgenic mice also showed B-1 B cells in the white pulp and MZ area, similar to our results (48), the near-complete predominance of B-1 B cells in this earlier analysis of adult transgenic animals made it difficult to appreciate the relationship between B-1 and other B cell compartments or maturation stages. In this regard, abundant generation of apparently diverse BCR specificities in VH11t mice together with the availability of Ab specifically detecting anti-PtC VH and VL idiotypes offer a significant advantage. In VH11t mice, ongoing generation of B cells expressing VH11μ rapidly declines after birth, possibly due to inefficient pre-BCR assembly by VH11μ H chain, which limits their progression into the pre-B cell compartment in bone marrow (32, 35). Instead, VH11– B cell generation becomes predominant, from which mature follicular and MZ B cells develop in spleen. What is particularly intriguing is the finding that the FDC network appears associated with B-1 cell clusters more than with other B cell compartments. This may lend support to the idea of precedence of the CXCR5+ B-1/FDC interaction and expansion before recruitment of other B cells subsequently generated from bone marrow.
Follicular stromal cells are specialized to produce chemokines essential for B and T cell recruitment, maturation, and B-T compartmentalization during development. Among stromal cells in the B cell and T cell areas, FDC are specifically found in association with B cells and play important roles in the generation and maintenance of high affinity immunologic memory (49, 50). FDC are presumed to be of mesenchymal origin (51, 52) and are radioresistant (53, 54), and their precursors are present in RAG KO (47) and scid mice (16, 46), although an understanding of their development remains incomplete (55). Nonetheless, the importance of B cells in FDC development is evident by their absence from B cell-deficient, but not T cell-deficient, mice, (9, 15) and from FDC generation in such mice after B cell transfer (16, 46, 47). This B cell-initiated FDC generation in spleen depends on a chemokine interaction (CXCR5/CXCL13), which, in turn, induces up-regulation of LT from B cells that binds to LT/TNFRs expressed by FDC in a positive feedback loop (17, 56). Although LT or CXCR5 is not produced by B cells alone (57, 58, 59), and FDC development appears to be dependent on secondary lymphoid organs (60, 61, 62), data from BLT knockout mice support the importance of B cell-derived LT in spleen FDC, MZ, and germinal center formation, because these deficiencies are not fully compensated by other mechanisms when B cells cannot express LT (62).
FDC development is recognized earliest by FDC-M1 expression in 3-day-old newborn mice and by the appearance of high CR1/2 complement receptor expression in 1-wk-old neonates (8). In considering the neonatal spleen, there is a paucity of B cells and a low CXCR5 level in immature B cells (21, 45), which have prompted the suggestion that initiation of FDC development may require only a few B cells (8), presumably expressing relatively few CXCR5 molecules. From our work presented in this study, we would additionally suggest that a population of B cells already up-regulating CXCR5, such as B-1 cells, may participate more effectively in this early FDC induction. CD5+ B-1 cells are generated from fetal progenitors in liver (2) or omentum (63) and disseminated to the spleen or body cavity, thus making up 80% of the B cell population in the peritoneal cavity in 7- to 10-day-old neonates. We show in this study that these B-1 cells are the only B cells expressing a mature level of CXCR5 in neonates, before B-2 or MZ B cells accumulate, and B-1 cells are able to induce FDC as effectively as other mature B cells. Our VH11t mouse analysis data appear to support such a role for early emerging B-1 B cells.
Different from VH11t mice, which show an increasing B-1 cell frequency, the B-1 cell frequency declines with age in normal mice, in which a B-1 cell role in early spleen architecture development would be readily supplanted by increasing production of other B cells. However, B-1 cells appear to be associated with FDC in normal adults as well. In contrast to VH11t mice, it has been difficult to unambiguously identify the B-1 cell location in normal adult mice without experimental manipulations, such as cell transfer. However, our recent analysis revealed the presence of small clustered B-1 cells in the white pulp with FDC, in which B-1 cells of endogenous (normal) Ig origin are distinguished by IgM allotype difference (L. Wen, S. A. Shinton, R. R. Hardy, and K. Hayakawa, unpublished observations), supporting B-1 cell association with FDC in normal adults as well as neonates. Whether a complex of self-Ag, autoantibody, and complement is involved in such intimate FDC association remains to be investigated. Nonetheless, recognizing such B-1 and FDC associations may be important in understanding B-1 cell function in autoimmune disease development and their susceptibility to dysregulated growth, particularly considering that FDC have been shown to enhance malignant B cell growth and survival (64, 65).
Why are B-1 cell-derived plasma cells disproportionately more abundant than B-1 B cells in the follicle? Whether the abundance of such anti-PtC plasma cells is due to their long life span (66), to rapid plasma cell differentiation, or to continued entry from outside the spleen remains unclear at present. Nonetheless, our peritoneal anti-PtC B-1 cell reconstitution data raise the latter possibility, that peritoneal cavity B-1 B cells may normally enter the spleen, thereafter contributing to the plasma cell pool. The ability of peritoneal B-1 cells to migrate and permanently reconstitute B-1 cells in spleen has been shown previously (39, 40), and we demonstrate in this study that such reconstitution is associated with B-1 follicle formation, induction of FDC formation, MZ microenvironment organization, and plasma cell generation in spleen. The anti-PtC B cell frequency is normally high in the PerC B-1 B cell pool (30). Thus, it is not unreasonable to posit a continual contribution by the B-1 cell pool established outside the spleen, such as in the peritoneal cavity, acting as a reserve source for plasma cell generation and incidentally serving as a source of cells for potential dysregulated growth in spleen later in life (67).
In turn, spleen B-1 cells can also migrate to the peritoneal cavity, reconstituting PerC, and a role for CXCL13+ macrophages in the omentum has been suggested in this peritoneal cavity homing (23). B-1 cells express various chemokine and homing receptors and circulate in blood, and there appears to be some level of constant B cell flux between PerC and the circulation, as previously shown by parabiosis experiments (23). Thus, considering the PerC B-1 cell FDC-inducing ability and the capacity to continuously produce natural autoantibody, the role of PerC B-1 B cells may be more significant and more integral to regulating lymphocyte population dynamics than previously considered.
Acknowledgments
We thank J. Dashoff for mouse breeding and genetic screening, D. Allman for generation of the VH11 gene knockin construct, S. Howard and K. Campbell for ES line transfection, S. Hua and D. Kappes for knockin transgenic mouse generation, P. Nakajima for anti-Id hybridoma generation, and M. Gui for help with hybridoma screening. We appreciate J. Oesterling’s help with cell sorting. T. Manser (Thomas Jefferson University, Philadelphia, PA), J. Kearney (University of Alabama, Birmingham, AL), and G. Rall offered advice on section staining. We than X. Xu for use of the cryostat. We also appreciate comments on this manuscript from T. Manser, D. Wiest, and K. Campbell.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by National Institutes of Health Grants R01AI49335 to (K.H.), R01AI26782 (to R.R.H), and R01AI40946 (to R.R.H.) and the Greenwald Fellowship of the Fox Chase Cancer Center (to L.W.).
2 Address correspondence and reprint requests to Dr. Kyoko Hayakawa, Fox Chase Cancer Center, Reimann Building, 333 Cottman Avenue, Philadelphia, PA 19111. E-mail address: k_hayakawa{at}fccc.edu
3 Abbreviations used in this paper: FDC, follicular dendritic cell; FL, fluorescein; KO, knockout; LT, lymphotoxin; MZ, marginal zone; PerC, peritoneal cavity; PtC, phosphatidylcholine; Vk9id, Vk9 anti-PtC idiotype.
Received for publication November 24, 2004. Accepted for publication March 14, 2005.
References
Herzenberg, L. A., A. M. Stall, P. A. Lalor, C. Sidman, W. A. Moore, D. R. Parks, L. A. Herzenberg. 1986. The LY-1 B cell lineage. Immunol. Rev. 93: 81-102.
Hardy, R. R., K. Hayakawa. 1994. CD5 B cells, a fetal B cell lineage. Adv. Immunol. 55: 297-339.
Hayakawa, K., M. Asano, S. A. Shinton, M. Gui, D. Allman, C. L. Stewart, J. Silver, R. R. Hardy. 1999. Positive selection of natural autoreactive B cells. Science 285: 113-116.
Hayakawa, K., R. R. Hardy, D. R. Parks, L. A. Herzenberg. 1983. The "Ly-1 B" cell subpopulation in normal immunodefective, and autoimmune mice. J. Exp. Med. 157: 202-218.
Hayakawa, K., R. R. Hardy. 1988. Normal, autoimmune, and malignant CD5+ B cells: the Ly-1 B lineage?. Annu. Rev. Immunol. 6: 197-218.
Kantor, A. B., L. A. Herzenberg. 1993. Origin of murine B cell lineages. Annu. Rev. Immunol. 11: 501-538.
Kroese, F. G., A. S. Wubbena, K. C. Kuijpers, P. Nieuwenhuis. 1987. The ontogeny of germinal centre forming capacity of neonatal rat spleen. Immunology 60: 597-602.
Balogh, P., Y. Aydar, J. G. Tew, A. K. Szakal. 2001. Ontogeny of the follicular dendritic cell phenotype and function in the postnatal murine spleen. Cell. Immunol. 214: 45-53.(Lijun Wen, Susan A. Shint)
Although CD5+ B-1 B cells have been recognized as an infrequent B cell subset in mice for many years, attempts to identify their histologic location in normal mouse spleen have proven difficult due to both their paucity and low level expression of CD5. In this study we have studied VH11/DH/JH gene-targeted mice, VH11t, that develop elevated numbers of CD5+ VH11/Vk9 B cells with an anti-phosphatidylcholine (anti-PtC) autoreactive specificity, allowing B-1 B cell detection by anti-PtC Id-specific Abs in spleen section staining. Using this approach we found that anti-PtC B-1 cells first appear within the white pulp in neonates, expand in association with follicular dendritic cells (FDC), and localize more centrally than other (non-B-1) IgDhigh follicular B cells in adults. Among neonatal B cells, CD5+ B-1 cells in both normal and VH11t mouse spleen and peritoneal cavity express the highest levels of CXCR5, which is important for FDC development. Injection of purified spleen or peritoneal B-1 cells into RAG knockout mice resulted in B-1 cell follicle formation in spleen, inducing FDC development and plasma cell generation. These results indicate that B-1 B cells are the first B cells to express fully mature levels of CXCR5, thereby promoting the development of FDC.
Introduction
A distinctive B cell subset in mice, B-1, expresses certain autoreactive specificities and is responsible for much of the natural serum autoantibody production (1, 2, 3). B-1 B cells were initially discovered based on CD5 expression, a surface phenotype resembling B cells increased in certain autoimmune diseases and B chronic lymphocytic leukemia (4, 5). Careful analysis by sensitive flow cytometry identified this CD5+ B cell subset as a rare population in normal adult mice, although its frequency changes depending on ontogeny and anatomical location. Whereas B-1 cell generation is frequent before birth, it declines after birth, eventually comprising only a small percentage of the B cells in adult mouse spleen. In contrast, the follicular B-2 B cell population increases over the first 2 mo of life, becoming the predominant B cell subset in adults. In comparison with this spleen pattern, B-1 cells maintain a high frequency in the peritoneal cavity throughout adulthood (2, 6).
The spleen is a critical organ for B cell maturation and differentiation, and provides a strategic site for establishment and organization of the immune system. Commencing after the appearance of the primordial follicle (7, 8), an organized white pulp architecture becomes established in spleen with mouse development, generating the marginal sinus, promoting myeloid and follicular stromal cell maturation, and eventually leading to B-T lymphocyte compartment segregation (9). Such microarchitectural development is an interactive process involving chemokines and the lymphotoxin (LT)/TNF gene family of cytokines (10, 11, 12, 13), in which B cells appear to play a critical inductive role in stromal cell maturation (14, 15, 16, 17, 18).
The initial B cell-stromal cell interaction involves CXCL13 (BLC/BCA-1)-expressing follicular dendritic cells (FDC)3 (19) binding to CXCR5 (BLR1)-expressing B cells (14, 17, 20). B-1 B cells express the highest levels of CXCR5 among B cells in adults (21, 22, 23), and considering the early appearance of B-1 B cells in mouse ontogeny, it was reasonable to expect that neonatal B-1 cells would also express CXCR5 and be able to precipitate FDC development in spleen. In exploring this process, the ability to identify their location is crucial. However, because of their paucity and low level expression of CD5 (4), detected only by sensitive flow analysis, attempts to identify the histologic location of B-1 cells in normal mouse spleen have proven difficult. For this reason, our strategy has been to use mice with a quantitatively increased B-1 cell population due to expression of the anti-phosphatidylcholine (anti-PtC) autoreactive specificity, a prototypic specificity associated with the B-1 B cell subset. Anti-PtC is an autoreactive specificity relatively highly represented in the mouse CD5+ B-1 cell subset (2, 24, 25), is naturally secreted into serum, and is predominantly encoded by either VH11/Vk9 or VH12/Vk4/5 (26, 27, 28). Considering its ability to bind to PtC exposed on senescent or bromelain-treated RBC (29, 30), anti-PtC autoantibody is presumed to play a role in the clearance of damaged or senescent erythrocytes from the circulation through the action of macrophages (31), which are abundant in red pulp in spleen. As previously shown with VH11 (32) and VH12 (33, 34) H chain transgenic and with JH locus-targeted VH12/DH/JH insertion (knockin) mice, genetic provision of rearranged VH11 (or VH12) H chain results in an increased, often overwhelming, frequency of anti-PtC CD5+ B-1 cells with the expression of appropriate Vk9 (or Vk4/5) L chains, presumed to be the result of positive selection for this autoreactive specificity by self Ag (3, 33, 35).
Interestingly, in targeted VH11/DH/JH1 insertion knockin mice, VH11t, we observed dynamic changes in the B cell repertoire and B cell populations during mouse development, resulting in the generation of multiple IgM+ B cell subsets, including CD5+ B-1, mature B-2, and marginal zone (MZ) B cells (36) in spleen, all at significant frequencies. Importantly, anti-PtC B cells are restricted to the CD5+ B-1 cell population. This allowed us to locate B-1 in comparison with other B cell compartments with the aid of anti-PtC Id-specific Abs detecting B-1 cells, instead of using anti-CD5. In this study we report a histologic analysis of such spleen cells; demonstrate the capacity of B-1 cells to form a B cell follicle and induce FDC, generate plasma cells, and conclude with a discussion of B-1 cell peritoneal-spleen dynamics.
Materials and Methods
Tissue section immunofluorescence analysis
Frozen section staining was performed as described previously (41). Briefly, spleens samples were embedded and frozen at –80°C. Frozen samples were sliced into 6-μm sections, air-dried, fixed with ice-cold ethanol for 15 min, washed with PBS, then blocked with PBS and 2% newborn bovine serum. Staining was performed at room temperature for 30 min for each step. Stained sections were washed with PBS and mounted using Fluormount G (Southern Biotechnology Associates), then viewed with a Nikon Optiphot epifluorescence microscope equipped with a Quad-Fluor (EF-1) four-cube filter holder and appropriate filter cubes. Images were recorded using the QImaging Retiga-1300 cooled CCD camera, processed with Openlab software (Improvision), and assembled with Canvas (ACD Systems of America). Unless otherwise stated, images were obtained with a x10 objective lens (original camera magnification, x264).
Cell transfer
For analysis of cell entry during the 2- to 24-h period, 5 x 107 VH11t spleen B220+CD5+ B-1 cells were injected i.p. per RAG KO mouse. For analysis of FDC development during the 7- to 10-day period, 2–3 x 107 B220+CD5+ B-1 cells were cell sorter purified from the peritoneal cavity of VH11t mice and injected i.p. into RAG KO mice. For long term self-reconstitution analysis (>2 mo), RAG KO mice were lightly irradiated (cesium, 300 rad), then injected i.p. with 3–5 x 106 cell sorter-purified B220+CD5+ peritoneal B-1 cells the following day. Similarly, spleen B-1 cells were sorted and injected i.v. into RAG KO mice for both short and long term analyses.
Results
Detection of VH11/Vk9 anti-PtC plasma cells from normal mouse spleen
In normal adult BALB/c mice, VH11/Vk9 anti-PtC B cells are rare in spleen, constituting <1% of the total spleen B cells (1–10% of CD5+ B-1 cells) as previously determined by their frequency in hybridoma panels (42) and by RNA message analysis (43). This low incidence in spleen (0.1–0.2% of the total adult spleen IgM+ B cells) has been confirmed recently by flow cytometric analysis using a combination of anti-VH11 and anti-Vk9 anti-PtC Id Abs, 3H7 and 13B5, respectively (B. G. Moon, S. A. Shinton, D. Allman, K. Hayakawa, and R. R. Hardy, manuscript in preparation). In spleen sections, it is difficult to unequivocally determine the location of such rare anti-PtC B cells by anti-Id. However, VH11/Vk9 anti-PtC plasma cells are readily detected as anti-Id brightly stained cells (Fig. 1), an indication of cells with high cytoplasmic anti-PtC content, in the red pulp outside the white pulp regions marked by the MOMA-1-stained metallophilic macrophage lining (44) (Fig. 1, A and B). They are costained with the anti-Vk9 (Vk9id, marked Vk9 in the following figures) and the anti-VH11 anti-PtC Id Abs (Fig. 1B). As seen with Vk9/B220 staining in Fig. 1A, these anti-Id brightly stained cells are B220 negative and also CD19– (which will be shown in Fig. 8), suggesting that they are at the stage of the differentiated surface IgM– Ab-secreting cells. They constitute between 10 and 50% of the total IgM plasma cells in the red pulp in average BALB/c mouse spleen, as represented in Fig. 1C (higher magnification). Thus, despite the paucity of anti-PtC B cells, B-1 cell-derived anti-PtC plasma cells are readily detectable in normal adult mouse spleen.
To characterize this B-1 localization relative to other non-PtC B cell compartments in more detail, we included IgD staining, because the majority of non-B-1 cells in adults are characteristically IgDhigh. The Vk9id/IgD staining data in Fig. 4 reveals first that most Vk9id+IgDlow/– B-1 cells (green) are clustered in the center of the follicle, whereas the Vk9id– IgDhigh (blue) cells are more densely located toward the follicle periphery. Some green Vk9id+ cells are also located outside the IgD+ region, at the zone between the red pulp (containing Vk9id bright plasma cells) and white pulp area (Fig. 4, left and right), suggesting their presence in the MZ, although none of these Vk9id+ cells are CD21high MZ B cells (Fig. 3B). Since, in this Vk9id/IgD section analysis, typical MZ B cells are not visualized (because they are Vk9id– and IgDlow/–) and some mature IgMlow B-2 cells might be missed due to their relatively lower IgD expression, we next incorporated analysis of B220 expression together with Vk9id and IgD (Fig. 5). Both IgDlow/– mature B-2 and MZ B cells express higher levels of B220 in contrast to B-1; thus, B-1 cells are recognized as green (Vk9id+IgDlow/–B220+/–), IgDhigh cells (mostly mature B-2) as purple (Vk9id–IgDhighB220+), and MZ B cells as red (Vk9id–IgDlowB220+); the potentially missed IgDlow mature B-2 cells will also be recognized as more red (Vk9id–IgDlowB220+) than purple. As Fig. 5 demonstrates, the central region of the B cell follicle shows a strong predominance of green staining, rather than yellow or orange, suggesting a relative B-1 predominance, surrounded by IgDhigh mature B-2 and IgDhighIgMhigh transitional cells (purple). In turn, red staining predominates in the MZ, indicating that typical MZ B cells cluster in this study, quantitatively more than B-1. Both B-1 (green) and mature B-2 cells (red/purple) are present in the red pulp at a similar frequency as those in peripheral blood circulation. Anti-PtC plasma cells in red pulp are recognized as the bright green-stained cells. Some of these plasma cells migrating out through the bridging channel are marked by an arrow in Fig. 5.
B-1 cells express the highest CXCR5 level among neonatal B cells
Anti-PtC cell clusters are already detectable from younger (2-wk-old) mouse spleen within the white pulp, as Fig. 2 section staining data has shown. In such neonates, all FDC networks are close to the marginal sinus boundary with the metallophilic macrophages, where Vk9id+ cell clusters are also found (Fig. 7A). As flow cytometry analysis shows (Fig. 7B), the Vk9id+ anti-PtC B cells constitute only 5–10% of the total spleen B cells, whereas the majority are not anti-PtC (Fig. 7B, spleen). Nearly all Vk9id– (non-PtC) B cells in such neonatal spleens are developing AA4+ immature B cells, IgMhighIgDlow/med (not shown), CD5–, and with a low level of the FDC CXCL13 chemokine receptor CXCR5 (Fig. 7B, blue), consistent with previous reports that immature B cells do not yet fully express CXCR5 compared with mature B cells (21, 45). Distinctively, the Vk9id+ anti-PtC B-1 cells have down-regulated AA4, are uniformly CD5+ as a result of Ag selection, and are already expressing a higher level of CXCR5 (Fig. 7B, red solid) at a level similar to that on mature B cells. In addition to the spleen, CD5+ CXCR5+ anti-PtC B-1 cells are also present in the neonatal peritoneal cavity (PerC) as an abundant AA4low/– mature PerC B cell population (Fig. 7B, red dash). CD5+ B-1 cells present in normal (BALB/C) neonatal spleen and PerC also show a surface phenotype comparable to the anti-PtC B cells in VH11t mice, including elevated CXCR5 expression (Fig. 7C). These data demonstrate the existence of B-1 cells as a CXCR5+ mature B cell source, arising before B-2 B cell maturation and consequent CXCR5 up-regulation. Both spleen and PerC B-1 cells continuously maintain a high level of CXCR5 throughout the life of the animal.
Induction of FDC, B-1 cell follicle formation, and plasma cell generation by B-1 cell transfer
CXCR5 expression by B-1 cells suggested that B-1 cells may be able to induce FDC maturation as do mature spleen B cells (16, 46, 47). To directly test this, purified anti-PtC B-1 cells from VH11t mice were injected into RAG KO mice that lack mature FDC (47), similar to newborn mice (8). In addition to i.v. or i.p. injection of spleen B-1 cells, PerC B-1 cells were injected i.p. in an attempt to determine whether the PerC B-1 cells dominating in neonatal PerC would be able to migrate to spleen. The use of B-1 cells purified from PerC washout cells has an advantage as well, by circumventing the risk of possible contamination by FDC copurified with B-1 cells from spleen. In both cell injection routes, B-1 cells entered into the white pulp via the weakly structured marginal sinus in adult RAG KO hosts, beginning to form CD19+Vk9id+ cell clusters 1 day after injection, without prior existing mature CR1/2high FDC (Fig. 8).
Induction of mature FDC and an FDC network in the white pulp became clear by 10 days after injection, as shown in this study for mice injected i.p. with PerC B-1 cells, in contrast to the uninjected RAG KO control (Fig. 9). All CR1/2high FDC were associated with CD19+ B cell follicles (Fig. 9A, left) within the marginal sinus endothelium expressing mucosal addressin cell adhesion molecule-1 (Fig. 9A, middle). Also, vast numbers of IgMhighCD19–Vk9high anti-PtC plasma cells were generated, localized to the red pulp (Fig. 9A, middle and right). Flow cytometric analysis confirmed that all CD19+ B cells, with weaker Vk9 section staining than plasma cells, were B-1 cells with retention of exclusive anti-PtC specificity and CD5+ phenotype (Fig. 9B). Some B-1 cells were also present in the MZ area (Fig. 9A, left) without a typical CD21high MZ B cell phenotype, similar to the data from intact VH11t animals shown in Fig. 4. The ability to induce FDC development together with B cell follicle formation by B-1 cells were comparable to those seen with mature B-2 cells purified from either VH11t or normal adult mouse spleen (purified as B220highCD23highAA4–CD5–CD43–VH11–Vk9– cells), except that there was greater plasma cell generation from B-1 cells than B-2 cells (data not shown).
Although such short term experiments required larger amounts of B cells for rapid induction of a readily visualizable follicular network, it has been shown previously that spleen or PerC B-1 cells from normal neonatal and adult mice are able to self-reconstitute in recipients over the long term by lower dose transfer (106 cells) (39, 40). Such B-1 self-reconstituted mice using VH11t mouse PerC anti-PtC B-1 cells revealed a similar spleen architecture profile, containing B-1 cell follicles, FDC, and plasma cells (data not shown). In addition, there was some recruitment of phagocytic ER-TR9+ macrophages in the MZ and stronger MOMA-1 staining than in uninjected RAG KO mice, suggesting either promotion of macrophage differentiation or an increase in differentiated macrophage numbers with time. Thus, B-1 cells, including those migrating from PerC, can initiate FDC development, promote microarchitecture organization, establish a B-1 cell subset, and sustain a plasma cell pool in spleen.
Discussion
In this study we observed B-1 B cell localization in the white pulp, in close association with the FDC network, and FDC induction by B-1 B cell transfer, similar to what has been found previously with mature B-2 B cells, commonly referred to as follicular B cells due to their association with FDC. Our analysis suggest that B-1 B cells are able to traverse into peripheral blood and re-enter the white pulp from the MZ, where minor vessels branching off the central arterioles deposit blood. Although a previous attempt using VH12μ H chain transgenic mice also showed B-1 B cells in the white pulp and MZ area, similar to our results (48), the near-complete predominance of B-1 B cells in this earlier analysis of adult transgenic animals made it difficult to appreciate the relationship between B-1 and other B cell compartments or maturation stages. In this regard, abundant generation of apparently diverse BCR specificities in VH11t mice together with the availability of Ab specifically detecting anti-PtC VH and VL idiotypes offer a significant advantage. In VH11t mice, ongoing generation of B cells expressing VH11μ rapidly declines after birth, possibly due to inefficient pre-BCR assembly by VH11μ H chain, which limits their progression into the pre-B cell compartment in bone marrow (32, 35). Instead, VH11– B cell generation becomes predominant, from which mature follicular and MZ B cells develop in spleen. What is particularly intriguing is the finding that the FDC network appears associated with B-1 cell clusters more than with other B cell compartments. This may lend support to the idea of precedence of the CXCR5+ B-1/FDC interaction and expansion before recruitment of other B cells subsequently generated from bone marrow.
Follicular stromal cells are specialized to produce chemokines essential for B and T cell recruitment, maturation, and B-T compartmentalization during development. Among stromal cells in the B cell and T cell areas, FDC are specifically found in association with B cells and play important roles in the generation and maintenance of high affinity immunologic memory (49, 50). FDC are presumed to be of mesenchymal origin (51, 52) and are radioresistant (53, 54), and their precursors are present in RAG KO (47) and scid mice (16, 46), although an understanding of their development remains incomplete (55). Nonetheless, the importance of B cells in FDC development is evident by their absence from B cell-deficient, but not T cell-deficient, mice, (9, 15) and from FDC generation in such mice after B cell transfer (16, 46, 47). This B cell-initiated FDC generation in spleen depends on a chemokine interaction (CXCR5/CXCL13), which, in turn, induces up-regulation of LT from B cells that binds to LT/TNFRs expressed by FDC in a positive feedback loop (17, 56). Although LT or CXCR5 is not produced by B cells alone (57, 58, 59), and FDC development appears to be dependent on secondary lymphoid organs (60, 61, 62), data from BLT knockout mice support the importance of B cell-derived LT in spleen FDC, MZ, and germinal center formation, because these deficiencies are not fully compensated by other mechanisms when B cells cannot express LT (62).
FDC development is recognized earliest by FDC-M1 expression in 3-day-old newborn mice and by the appearance of high CR1/2 complement receptor expression in 1-wk-old neonates (8). In considering the neonatal spleen, there is a paucity of B cells and a low CXCR5 level in immature B cells (21, 45), which have prompted the suggestion that initiation of FDC development may require only a few B cells (8), presumably expressing relatively few CXCR5 molecules. From our work presented in this study, we would additionally suggest that a population of B cells already up-regulating CXCR5, such as B-1 cells, may participate more effectively in this early FDC induction. CD5+ B-1 cells are generated from fetal progenitors in liver (2) or omentum (63) and disseminated to the spleen or body cavity, thus making up 80% of the B cell population in the peritoneal cavity in 7- to 10-day-old neonates. We show in this study that these B-1 cells are the only B cells expressing a mature level of CXCR5 in neonates, before B-2 or MZ B cells accumulate, and B-1 cells are able to induce FDC as effectively as other mature B cells. Our VH11t mouse analysis data appear to support such a role for early emerging B-1 B cells.
Different from VH11t mice, which show an increasing B-1 cell frequency, the B-1 cell frequency declines with age in normal mice, in which a B-1 cell role in early spleen architecture development would be readily supplanted by increasing production of other B cells. However, B-1 cells appear to be associated with FDC in normal adults as well. In contrast to VH11t mice, it has been difficult to unambiguously identify the B-1 cell location in normal adult mice without experimental manipulations, such as cell transfer. However, our recent analysis revealed the presence of small clustered B-1 cells in the white pulp with FDC, in which B-1 cells of endogenous (normal) Ig origin are distinguished by IgM allotype difference (L. Wen, S. A. Shinton, R. R. Hardy, and K. Hayakawa, unpublished observations), supporting B-1 cell association with FDC in normal adults as well as neonates. Whether a complex of self-Ag, autoantibody, and complement is involved in such intimate FDC association remains to be investigated. Nonetheless, recognizing such B-1 and FDC associations may be important in understanding B-1 cell function in autoimmune disease development and their susceptibility to dysregulated growth, particularly considering that FDC have been shown to enhance malignant B cell growth and survival (64, 65).
Why are B-1 cell-derived plasma cells disproportionately more abundant than B-1 B cells in the follicle? Whether the abundance of such anti-PtC plasma cells is due to their long life span (66), to rapid plasma cell differentiation, or to continued entry from outside the spleen remains unclear at present. Nonetheless, our peritoneal anti-PtC B-1 cell reconstitution data raise the latter possibility, that peritoneal cavity B-1 B cells may normally enter the spleen, thereafter contributing to the plasma cell pool. The ability of peritoneal B-1 cells to migrate and permanently reconstitute B-1 cells in spleen has been shown previously (39, 40), and we demonstrate in this study that such reconstitution is associated with B-1 follicle formation, induction of FDC formation, MZ microenvironment organization, and plasma cell generation in spleen. The anti-PtC B cell frequency is normally high in the PerC B-1 B cell pool (30). Thus, it is not unreasonable to posit a continual contribution by the B-1 cell pool established outside the spleen, such as in the peritoneal cavity, acting as a reserve source for plasma cell generation and incidentally serving as a source of cells for potential dysregulated growth in spleen later in life (67).
In turn, spleen B-1 cells can also migrate to the peritoneal cavity, reconstituting PerC, and a role for CXCL13+ macrophages in the omentum has been suggested in this peritoneal cavity homing (23). B-1 cells express various chemokine and homing receptors and circulate in blood, and there appears to be some level of constant B cell flux between PerC and the circulation, as previously shown by parabiosis experiments (23). Thus, considering the PerC B-1 cell FDC-inducing ability and the capacity to continuously produce natural autoantibody, the role of PerC B-1 B cells may be more significant and more integral to regulating lymphocyte population dynamics than previously considered.
Acknowledgments
We thank J. Dashoff for mouse breeding and genetic screening, D. Allman for generation of the VH11 gene knockin construct, S. Howard and K. Campbell for ES line transfection, S. Hua and D. Kappes for knockin transgenic mouse generation, P. Nakajima for anti-Id hybridoma generation, and M. Gui for help with hybridoma screening. We appreciate J. Oesterling’s help with cell sorting. T. Manser (Thomas Jefferson University, Philadelphia, PA), J. Kearney (University of Alabama, Birmingham, AL), and G. Rall offered advice on section staining. We than X. Xu for use of the cryostat. We also appreciate comments on this manuscript from T. Manser, D. Wiest, and K. Campbell.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by National Institutes of Health Grants R01AI49335 to (K.H.), R01AI26782 (to R.R.H), and R01AI40946 (to R.R.H.) and the Greenwald Fellowship of the Fox Chase Cancer Center (to L.W.).
2 Address correspondence and reprint requests to Dr. Kyoko Hayakawa, Fox Chase Cancer Center, Reimann Building, 333 Cottman Avenue, Philadelphia, PA 19111. E-mail address: k_hayakawa{at}fccc.edu
3 Abbreviations used in this paper: FDC, follicular dendritic cell; FL, fluorescein; KO, knockout; LT, lymphotoxin; MZ, marginal zone; PerC, peritoneal cavity; PtC, phosphatidylcholine; Vk9id, Vk9 anti-PtC idiotype.
Received for publication November 24, 2004. Accepted for publication March 14, 2005.
References
Herzenberg, L. A., A. M. Stall, P. A. Lalor, C. Sidman, W. A. Moore, D. R. Parks, L. A. Herzenberg. 1986. The LY-1 B cell lineage. Immunol. Rev. 93: 81-102.
Hardy, R. R., K. Hayakawa. 1994. CD5 B cells, a fetal B cell lineage. Adv. Immunol. 55: 297-339.
Hayakawa, K., M. Asano, S. A. Shinton, M. Gui, D. Allman, C. L. Stewart, J. Silver, R. R. Hardy. 1999. Positive selection of natural autoreactive B cells. Science 285: 113-116.
Hayakawa, K., R. R. Hardy, D. R. Parks, L. A. Herzenberg. 1983. The "Ly-1 B" cell subpopulation in normal immunodefective, and autoimmune mice. J. Exp. Med. 157: 202-218.
Hayakawa, K., R. R. Hardy. 1988. Normal, autoimmune, and malignant CD5+ B cells: the Ly-1 B lineage?. Annu. Rev. Immunol. 6: 197-218.
Kantor, A. B., L. A. Herzenberg. 1993. Origin of murine B cell lineages. Annu. Rev. Immunol. 11: 501-538.
Kroese, F. G., A. S. Wubbena, K. C. Kuijpers, P. Nieuwenhuis. 1987. The ontogeny of germinal centre forming capacity of neonatal rat spleen. Immunology 60: 597-602.
Balogh, P., Y. Aydar, J. G. Tew, A. K. Szakal. 2001. Ontogeny of the follicular dendritic cell phenotype and function in the postnatal murine spleen. Cell. Immunol. 214: 45-53.(Lijun Wen, Susan A. Shint)