About 40% of the B cells in human peripheral blood are CD27+ and have hypermutated variable regions in their immunoglobulin receptors, and as such, they are classified as memory B cells. However, a proportion of these CD27+ cells are IgM+ and have not undergone class switching. Weller and colleagues previously suggested that these cells could be a subset that is distinct from the classical germinal-centre-derived memory B-cell subset. The present study shows that the peripheral IgM+IgD+CD27+ B cells correspond to splenic marginal zone (MZ) B cells, which control T-cell-independent responses.

First, the authors used phenotypic analysis to show that the memory IgM+ B cells in the blood were similar to those in the splenic MZ; both cell types were found to be IgMhiIgDlowCD1chiCD21hi. Second, gene-expression profiling showed that the two cell types expressed a similar pattern of genes, which differed from the genes expressed by naive or class-switched memory B cells. Third, the authors used spectratyping analysis of the CDR3 (complementarity-determining region 3) of the immunoglobulin receptor to establish a relationship between the cell types. This technique enabled the authors to track the development of cells with specific V(D)J rearrangements after immunization with a T-cell-independent antigen. The results showed that a particular B-cell clone was detected in both the blood and the splenic B-cell population after vaccination with pneumococcal and meningococcal polysaccharides, indicating that IgM+ B cells recirculate between the blood and the splenic MZ. When the authors studied children, they found IgM+ B cells with hypermutated receptors in children of a very young age, before the splenic MZ has matured, indicating that these cells can develop and mutate before being able to respond to T-cell-independent antigens. Moreover, asplenic children have mutated IgM+ B cells, indicating that this population can mature in extra-splenic sites.

Previous work from the same authors showed that patients with hyper-IgM syndrome who have a defect in CD40 ligand do not have germinal centres and cannot generate class-switched memory B cells, but they can generate IgM+ memory B cells. On the basis of these studies, the authors propose that there might be two different programmes for B-cell development — one that depends on germinal-centre formation and leads to 'true' hypermutated memory B cells in response to antigen stimulation, and another in which a diverse set of B cells is generated by hypermutation early in life. It seems that these IgM+ 'memory' B cells are not really memory cells at all but, instead, function as innate-like cells, producing IgM at a time when the adaptive immune response is still developing.