Conclusions

I would like now to review the questions that I addressed in my dissertation and summarize what I learned from the experiments that I constructed.

The immune system is viewed as a detection system. It detects pathogenic intrusions. In contrast to the familiar detection systems though, the immune system can recognize pathogens that the organism may have never encountered before in its life time. This ability derives, to a certain extent, from the generality of the language of biochemistry. A variety of immune receptors are generated without regard to what they may be binding. Those that react strongly with molecules that are normally present in the body are weeded out, and those that remain will, by definition, recognize "outsider" molecules. However, the immune system has only a limited number of cells that circulate at any time through the body. It is therefore crucial that the immune system makes optimal use of its resources by placing the receptors ``strategically'' in the space of all possible receptors. Having the right type of lymphocytes in the right number is crucial for the survival of the organism.

The questions that I addressed in my thesis are related to how the immune system might learn to anticipate its pathogenic environment. Based on the results that I summarize below, I argue that:

• The recognition capacity of the immune system is targeted to pathogens that it has encountered during evolution. It does not attempt to recognize as many molecular shapes as possible.
• Germline diversity does not contribute to the direct, specific recognition of pathogens, but rather realizes a coarse-grained coverage of the pathogen space.
• Immune receptor diversification during an on-going immune response is the determinant factor for the specificity and affinity of the antibodies. If the immune system fails to recognize a pathogen with high affinity, it means that:
  • the pathogen mimics the self structures too closely, or
  • the pathogen is an emergent pathogen, considerably different from those that the immune system has seen during its recent evolution, or
  • somatic hypermutation fails to produce a high affinity antibody for that pathogen, due probably to sequence peculiarities of the germline-encoded antibody that underwent somatic hypermutation.

My detailed results also bear on the construction of random antibody libraries, as well as on computational methods that may be used for mutational analysis in a variety of biological systems.

Note that I did not consider the effect of junctional diversity on the repertoire. The reasons are as follows:

• I focused on the aspects of evolutionary learning in the immune system. The rearrangement process is essentially thought to produce "random" junctions. It could thus not be the substrate for learning.
• If the rearrangement process indeed produces "random" junctions, then its contribution to the repertoire is to increase the size of the primary repertoire. However, if this repertoire cannot cover pathogens individually, then the results that I presented based on a single library are expected to hold.