Affinity maturation during the germinal center reaction

Somatic hypermutation seems to take place only during on-going immune responses, in the germinal center (GC) microenvironment. Germinal centers are specific anatomical sites in lymph nodes, spleen and other secondary lymphoid organs. They have an ephemeral existence, arising during the first week of an immune response, and lasting for approximately four weeks. They are believed to support the process of affinity maturation through somatic hypermutation () and affinity-based selection () of antigen-specific B cells. The term affinity maturation is simply used to describe the observation that the affinity of the antibodies that bind a given antigen at the end of the immune response is higher than the affinity of the antibodies that first reacted to this antigen. Thus the affinity "matures" during an immune response.

 

A small number of founder B cells (), activated outside of the GC, divide within the GC with doubling times as short as 6-7 hours (). These dividing cells, known as centroblasts, are concentrated at one pole of the GC known as the dark zone (see Fig. ) (1991). Cells exit the dark zone and rapidly move to an adjoining region of the GC, called the light zone. Kinetic studies involving labeling of dividing cells show that centroblasts pick up the label within 2 hours after it has been injected, and then move towards the light zone, where labeled cells appear only after 6-8 hours (1991). Light zone cells, centrocytes, therefore do not appear to be actively dividing, but are generated from dividing centroblasts. The antigen that the cells are selected for seems to be concentrated in the light zone of the germinal centers, on the surface of follicular dendritic cells.

This apparent separation of the proliferative compartment, the dark zone, from the compartment where the antigen is found (and selection supposedly occurs) led to a one-pass picture of the germinal center reaction (1994). That is, cells were thought to enter the germinal center, divide in the dark zone, move into the light zone, undergo selection, and the surviving cells make it into the memory compartment. The cells produced by somatic mutation and affinity selection have large numbers of point mutations in their immunoglobulin genes. Kepler and Perelson (1993) pointed out that accumulating random mutations to the extent observed in the dominating, high-affinity population, without an intervening selection event, is likely to render cells incapable of binding the antigen. They then performed an optimal control study of affinity maturation in the GC, with the mutation rate being the control variable, and concluded that cycles of mutation-less proliferation followed by mutation and selection, are the most efficient way of creating a memory population with a high average affinity. They suggested that one way to implement this optimal strategy would be to have cells proliferate and then mutate in the dark zone, undergo selection in the light zone, and then return to the dark zone and repeat the process. This model was called a cyclic re-entry or recycling model. A model based on the recycling hypothesis, which took into account the architectural and kinetic details of the germinal center was subsequently developed by Oprea and Perelson (1997). This model assumed that mutation occurs in the replicating population of the dark zone with selection occurring in the light zone. Affinity maturation was achieved with repeated movement between dark and light zones.

Testing the recycling hypothesis turns out to be difficult, as we cannot, at the moment, track the cell migration patterns in the germinal centers. In this context, one might be interested in how much affinity maturation can be achieved in a one-pass selection model. That is, the germinal center is seeded by a few B cells, which start dividing in the dark zone, move into the light zone to undergo selection, and then exit the germinal center, without being able to mutate again. This is exactly the approach that van Nimwegen, Perelson and I took, in order to give an argument "by contradiction" for why selection must be operating at multiple points during the germinal center reaction. Qualitatively, the results of such a model are completely intuitive, the number of high affinity cells has an upper bound which is given by the probability of generating such cells through mutation only. However, it turns out that if the selective agent, that is the antigen, decays during the germinal center reaction the amplification of high affinity cells is even lower than one would expect for affinity-based selection. I think that this result gives even stronger theoretical support to the recycling hypothesis, and also, that it is interesting on its own. There are probably a variety of systems in which the selective agent is decreasing with time, and thus I consider these results worth presenting.