Initial+Thoughts+about+Idea

Rebecca has looked at entropy with the immunosignatures before. In her case, she was looking at the complexities of the peptide sequences bound by antibody. She found that antibodies generally seem to prefer to bind more complex sequences.

I'm interested in looking at the entropy of the distrubtion of peptides bound by antibody by one sera sample as a general measure of the health/order of the immune system.

Muskan looked at the immunosignatures of children and found that there was overall very high intensity. This could have been due to the fact that children often receive many vaccines.

E-mail Conversation with Stephen, Kathy, and Phil about this concept

My e-mail I have an idea that involves entropy and our immunosignatures. I hypothesize that sicker or older people have higher entropy in their immune system, and I think that this can be detected with our immunosignature technology. It is already stated in textbooks that as people get older they produce lower affinity antibodies with broader reactivity. This fact would correspond to a higher entropy in the distribution of peptides bound by antibodies from older individuals.

One could just compare old (let's say 80 or over) or young (let's say around 20; I know Muskan looked at children's immunosignatures but observed unusually high intensity possibly due to vaccine times) immunosignatures. However, it might be easier to tell the difference between the two groups if one were to vaccinate both groups with some protein. The protein doesn't matter; it could just be GST. The point is that you would raise the immune response of both groups to an observable level. Then once you had done this, I predict that the younger or healthier person would have a narrow range of peptides bound (lower entropy in the peptide distribution) than the older or unhealthier person.

This can be quantitated. The formula for Shannon's Entropy is -Sum (from i=0 to n) p(Xi)*log(Xi). So for example, let's say the young person after immunizations has this distribution (extreme case but this is just an example): 1,000 peptides with 30,000-40,000 intensity 8,000 peptides with 40,000-50,000 intensity 1,000 peptides with 50,000-60,000 intensity Then the entropy would be -(1000/10000*log(1000/10000)+8000/10000*log(8000/10000)+1000/10000*log(1000/10000))=0.639

Let's say the old person after immunization has this distribution: 1,000 peptide with 10,000-20,000 intensity 2,000 peptide with 20,000-30,000 intensity 4,000 peptide with 30,000-40,000 intensity 3,000 peptide with 50,000-60,000 intensity Then the entropy would be -(1000/10000*log(1000/10000)+2000/10000*log(2000/10000)+4000/10000*log(4000/10000)+3000/10000*log(3000/10000))=1.28

So essentially the healthy or young immune system would be more efficient and accurate at recognizing the immunized antigen than the older or sick immune system and this behavior could be quantified. Although one might not be able to use traditional classifying algorithms to separate young and old, it might be possible to distinguish them with entropy.

I think the same behavior could be observed if one compared a healthy person and a sick person of the same age (let's say someone with asthma or something)

So why should we care about this? 1. If this is true, it would help us to understand the immune system better. Although it is widely known that entropy accumulates in any system with time, we would now have a new way of actually quantifying this in the immune system. 2. If this is true, methods for preventing or slowing the increase in entropy in the immune system could be tested. For example, perhaps constantly slightly stimulating/challenging the immune system overtime could help remind it to maintain itself and stay more efficient. In other words, perhaps stimulating the immune system with something like B2L or any other adjuvant or yogurt etc. would make it produce narrower peptide distributions when challenged than an immune system that did not receive this stimulation.

Just some thoughts.

-Kurt

Kathy's Response

I like your idea Kurt, and think you might be right about the ‘entropy”. However, I can’t decide yet what the fix would be. I can envision that the lower intensity and broader reactivity profiles might be a result of too much stimulation over the course of their long lives. In this case, further stimulation would not be productive. On the other hand, it may reflect too much chronic antigen exposure—which is seen in some infectious and non-infectious disease. In this case, an acute exposure to a directed immunomodulator may work to refocus and re-invigorate the response. This may also require modifying immune cell microenvironments. There’s one simple way to find out………….

Phil's Response You could do the same experiment with mice actually.

AAT invokes a very strong humoral response. Very old but naive mice should produce a detectable response; if you are correct the number of peptides lighting up would be different from 6 week old immunized mice. You would expect high correlation within age groups, low across age groups.

Granted it would be difficult to get old mice, but if you found someone with a BALB colony who inadvertently kept naives around for 2 years or more, you could do the immunization and collect a terminal bleed. You would need about 10 mice per group to see a common effect.

Phil

Stephen's Response Interesting thoughts. The basic questions about immune and age on the signatures is good. Whether this would reflect "entropy" in the immune system would need more careful definition. I am willing to take the counter bet on the experiment you propose, with the caveat that we normalize to total binding. I agree with Phil that the experiments could be done nicely in mice. You can buy old mice. saj