Correspondence: Professor ICM MacLennan, Professor Emeritus of Immunology at the University of Birmingham, MRC Centre for Immune Regulation, Room 435 the IBR, Birmingham B15 2TT, UK. E-mail: I.C.M.MACLENNAN@bham.ac.uk

The splenic marginal zone (MZ) contains B cells that are phenotypically and functionally distinct from recirculating B cells and B1 cells. Unlike recirculating B cells, MZ B cells do not migrate via blood and lymph between the follicles of secondary lymphoid organs. Consequently, selective depletion of recirculating cells does not obviously affect splenic MZ B-cell numbers, while it deprives lymph node follicles of over 90% of their small B lymphocytes.¹ By contrast, a sizable population of B cells that like MZ B cells are IgM⁺, IgD^low/-, CD21^high CD23^low/-, CD1d^+,2 remain in the splenic follicles after recirculating cell depletion.¹ In a recent article in Nature Immunology, Cinamon et al.³ provide evidence that many of the non-recirculating B cells in splenic follicles are an integral part of the MZ B cell pool. They go on to show that these cells constitutively oscillate between the MZ and splenic follicles (Figure 1).

Figure 1.

Follicular trafficking of marginal zone (MZ) B cells. The MZ has an open (sinusoidal) blood supply. S1P in MZ blood sinusoids attracts follicular B cells with a MZ phenotype, through their S1P₁ receptors (purple). These cells cross the marginal sinus and enter the MZ. There they can engage Ag in the blood through their BCR, or immune complex through their CR1/2. Functional expression of CXCR5 (green) by B cells within the MZ enables them to be attracted to adjacent splenic follicles by CXCL13 produced by follicular dendritic cell (FDC). Immune complex on the migrating cells is transferred to the CR1/2 of FDC. The follicular capillary blood supply allows little S1P, antigen or immune complex to access follicular B cells or FDC.

Full figure and legend (61K)

The splenic MZ lies between the B-cell follicles and red pulp. In rodents and rabbits, it is entirely perfused by a blood sinusoidal system that is directly fed from terminal branches of the splenic artery.⁴ This enables MZ B cells to bind blood-borne antigen via their B cell receptors for antigen (BCR), or immune complexes by Fc or complement receptors (CRs). MZ B cells that engage thymus-dependent⁵ or thymus-independent⁶ antigen through their BCR, often with crosslinking to their CR2, are induced to leave the MZ and migrate to the outer T zone as the first stage in a splenic antibody response.^{5, 6} Conversely, MZ B cells that bind antigen–antibody complexes via their CR1/2 alone actively transport the complex to follicular dendritic cells (FDC).^{7, 8} Recirculating B cells, by contrast seem unable to transport antigen to splenic FDC.⁹ It should be noted that recirculating B cells appear to have a role in antigen transport to FDC in lymph node follicles.^{10, 11} In addition to immune complex transport by MZ B cells to FDC, killed Gram-negative bacteria can induce mass exodus of B cells from the MZ,¹² causing a commensurate increase in the number of B cells with a MZ phenotype in the splenic follicles.⁷ This migration is reversed after a few hours. Any movement between the MZ and follicles must be local, for prolonged -irradiation to one-half of the spleen depletes recirculating B cells, but does not affect the number of cells with MZ phenotype in the MZ or follicles in the other half of the spleen.¹

MZ B cells require the sphingosine 1-phosphate receptor (S1P₁)^{3, 13} and to some extent S1P₃³ to bind to and migrate to the MZ stroma. In their elegant study, Cinamon et al.³ show that blocking S1P₁ receptor, or inactivating Edg1, which encodes it, results in an increase in B cells with MZ phenotype in follicles, but a relative absence of B cells in the MZ. Further, cells with a MZ B-cell phenotype are confined to the MZ³ in chimeras constructed with functionally normal recirculating B cells, but MZ B cells that lack CXCR5—a chemokine receptor associated with the attraction of B cells to CXCL13 produced by FDC.¹⁴ Consistent with failure of the CXCR5-deficient MZ B cells to enter follicles, they do not transport immune complex to FDC even though they bind C3b/d-containing immune complex.

More positive evidence for MZ B cells shuttling to and from follicles was obtained using pulse-chase labeling. The labels were selectively targeted against surface molecules of B cells resident in the MZ by using intravenous injection, follicular B cells having limited access to label because of the their capillary blood supply. At 5 min after intravenous injection of fluorescent anti-CD21, only a proportion of B cells with a MZ phenotype (CD19⁺, CD23^low/- an CD1b⁺) were labeled, while by 1 h almost all of the cells were marked. It was postulated that the increase in cells labeled at 1 h was due to follicular-based B cells with MZ phenotype migrating to the MZ where they gain access to anti-CD21 in the blood. Further evidence for this was provided by finding that there is no uptake of anti-CD21 from the blood, even after 1 h in S1P₁-deficient mice, where the B cells with a MZ phenotype are confined to follicles. Fluorescent anti-CD35 (anti-CR1) given intravenously to wild-type mice caused similar labeling patterns with time. Importantly, when a pulse of fluorescent anti-CD21 was given 20 min after injecting anti-CD35 only some of the CD35-tagged cells took up CD21, suggesting the others had already left the MZ for splenic follicles.

Although CR1/2 engagement by MZ B cells is required for the transport of immune complexes to FDC,^{7, 8} this engagement does not seem to induce migration of MZ B cells to follicles. This follows from the finding that cells with a MZ B-cell phenotype are found in both the splenic follicles and MZ in CR1/2-deficient mice. Despite the absence of CR1/2 they still move, at least from follicles to MZ, as tested by pulse-chase studies using intravenous fluorescent anti-B220.³ Cinamon et al. talk about constitutive (non-triggered) shuttling of MZ B cells between MZ and follicles. It is a matter of speculation as to how these B cells change their phenotype to migrate at one time toward FDC and at others to the MZ. These changes might be intrinsic to the system, but as described above we know they can be influenced by extrinsic signals including Gram-negative bacteria.^{7, 12}

The studies discussed advance our understanding of the molecular basis for MZ B-cell transfer of blood-borne immune complexes to splenic FDC. They emphasize that there is a follicular dimension to the MZ B-cell pool with regular exchange of cells between MZ and adjacent follicles. This mimics the way recirculating B cells first migrate to and then move out of follicles. It remains to be determined if there is further significance to this migration. For example, can MZ B cells, or for that matter recirculating B cells and B1 cells, be triggered to respond to antigen on FDC when they visit splenic follicles?