Proposed model for affinity-based selection in the germinal center. I. B cells compete for a limiting number of T cells at the T:B border; only the cells with the highest affinity are allowed to enter the germinal center reaction. II. GC cells in the DZ proliferate and turn on the somatic hypermutation machinery, which includes the enzymes AID and Polη. These cells are maintained in the DZ by high expression of chemokine receptor CXCR4. Cells that introduce mutations that impair BCR expression at this point (∅ BCR) die by lack of a BCR signal. After one (or potentially more) division/mutation cycles, surviving DZ B cells migrate towards the LZ, in a process that involves upregulation of chemokine receptor CXCR5. III. B cells interact with antigen in immune complexes on FDCs. B cells with very low or no affinity for antigen (∅ Aff.) may also die from lack of a BCR signal at this step. Antigen signals are not limiting at this point, and all B cells transiting to the LZ upregulate CD83 and CD86. BCR affinity determines the level of pMHC and perhaps co-stimulation on the B cell surface. IV. B cells present antigen to a limiting number of T helper cells. B cells with higher pMHC/co-stimulation interact preferentially with T cells, preventing their interaction with lower-affinity B cells, in a competitive process. B cells that successfully interact with T helper cells can then follow three potential fates: recycling, with upregulation of CXCR4 and re-entry into the DZ for further proliferation/mutation; exit from the GC into the plasma cell fate (likely under conditions of high affinity/pMCH density); or exit into the memory B cell fate. The latter possibility has not been shown to be enhanced by interaction with T helper cells. In different points during this process, B cells that develop strongly auto-reactive BCRs are eliminated by multiple checkpoints, which may be B cell-intrinsic or T cell-dependent (not shown). LZ, light zone; DZ, dark zone; FDC, follicular dendritic cell; Aff., affinity; Arrow width represents the proportion of cells thought to follow a given path. Dotted arrows represent likely events for which the available data is inconclusive.