To study the development of B lymphocyte memory, we identified and isolated splenic B cells expressing a highly defined antibody variable region that constitutes a reproducible and predominant component of the memory antibody response to p-azophenylarsonate (Ars). Isolation was achieved during the primary immune response by surface staining and flow cytometry using a specific anti-idiotypic antibody called E4, which recognizes this canonical V region, encoded by one set of V gene segments. The isolated E4+ cells displayed all of the phenotypic characteristics of germinal center centrocytes, including a low level of surface Ig, a lack of surface IgD, a high level of receptor for peanut agglutinin, and expression of mutated antibody V genes. E4+ B cells were first detected in the spleen 7-8 d after primary immunization, reached peak numbers from days 10-13, and waned by day 16. Surprisingly, at their peak, E4+ cells comprised only 40,000 of all splenocytes, and half of these failed to bind Ars. Using this number, we estimate the total number of Ars-specific memory-lineage cells in the spleen to be no more than 50,000 (0.1%) at any one time, and presumably far fewer that are committed to the memory pool. Chromosomal copies of rearranged V genes from single E4+ cells were amplified by nested PCR, and the amplified products were sequenced directly without cloning, using standardized conditions that disclose virtually no Taq polymerase errors. V gene sequence analyses of E4+ cells isolated from single mice confirmed their canonical nature and revealed that they were derived from few precursors. In the average mouse, the E4+ pool was derived from fewer than five canonical precursors. Somatic mutations were found within the V genes of almost all cell isolates. At day 13, a significant fraction of E4+ cells had mutations known to increase antibody affinity for Ars, suggesting they were products of at least one cycle of post-mutational antigen-driven selection. However, the lack of shared mutations by clonally related cells indicated that the selective expansion of mutant subclones typical of memory responses had not yet taken place. This was supported by the observation that half of the E4+ cells failed to bind Ars. Collectively, our results indicate that the memory compartment is a highly selected entity, even at relatively early stages of the primary immune response when somatic mutation and clonal selection are still in progress. If germinal centers are the source of memory B cells, our data suggest that B cell memory may be derived from only a small fraction of all germinal centers.
Codon bias in Escherichia coli: the influence of codon context on mutation and selection