travelling salesmen

A conversation with a friend about PKR has prompted me to write a bit about one way by which your body processes information and solves problems to keep you healthy. Please bear in mind that I'm not a human biologist - or rather I'm a biologist who does not study humans - so this might be a bit rough on any specialists out there. Please call me on my mistakes.

Your immune system is a miracle of rare device and here's a bridge over any chasms in your memory. When a foreign agent (such as a virus, bacteria or fungus) enters your body, if you are lucky, it will encounter a dentritic cell (DC for short). DC are highly mobile cells which guard your skin, the lining of your nose, lungs and intestines and will engulf any invader. You can see two DCs in action on wiki, the one on the left somewhat pathetically drags a fungal conidia around while the one on the right enthusiastically gorges itself on four whole ones. Ha! The unfortunate condia will then be broken down - digested - and parts of some of its proteins will be used to inform the rest of your body about the attack. The DC will migrate to a lymph node and present these little snippets of protein (an antigen) to T-cells.

cute dog from here

OK, so the faithful DC has made it to the lymph node and hopefully presents its chewed-up bit of fungal conidia protein as an antigen on its surface. Your body has population of around 25 million different T-cells which can recognise various different antigens. The DC will sit relatively still in the lymph node (having changed shape en route) and the smaller T-cells will rush by. If the antigen 'matches' a receptor on the surface of a T-cell then the two will bind quite tightly and the T-cell will pause. These T-cells will become killers; once activated by having matched an antigen to their receptor, they will go off and kill other cells - your other cells - that are presenting the same antigen on their surfaces. You see, other cells in your body will very nobly advertise that they are infected by a fungus (or whatever) and the activated T-cell will assist their suicide and prevent you from being overrun (with fungus or whatever).

But there's a problem here; very low numbers of a specific antigen are presented to very very many T-cell receptors. How can your body be sure to test all possible T-cell/antigen combinations efficiently? Remember that time is essential - once a bug gets inside you it can reproduce every 20 minutes or so and you want to nip that exponential growth early. Joost Beltman et al didn't buy the standard explanation - that T-cells would 'run and tumble' to a set program, say run for 2 microns and then tumble twice to find a new direction and hence sample lots of antigens. Even though it looks very much like that in real fluorescent-labelled cells and here in Beltman's simulation.

ah the inimitable xkcd

Running and tumbling is quite a complex thing for a cell to do. Is there an internal clock? How does a cell know how many runs to a tumble will be best? Are there any mutants out there; people with T-cells which only run or tumble insistently? And such behaviour is not appropriate at all times - once out on the prowl looking for sick cells to kill, should a T-cell be quite so giddy? No is the short answer; Beltman convincingly shows that the physical structure of the lymph node makes for efficient sorting by forcing T-cells into little turbulent streams. I think that this is a really neat solution to the problem. Take a moment to enjoy the slightly nauseous viewpoint of a T-cell being turbulently sorted. So there's no need for a complex run/tumble program. It's also noteworthy that Beltman and co. used a nice combination of real-life observations, maths and computer simulation and then went back to do some more RL observations.

To some extent (and with hindsight) the problem was a bit of an artifact created by the methods everyone used to view T-cell dynamics in the lymph node. Because fluorescence detection can easily become saturated only a small portion of T-cells could be labelled for viewing, so only a few individuals in a crowd could be tracked. But running and tumbling is not such a bad idea...plenty of cells are famous for just that sort of thing and even have some fascinating receptor dynamics and I'll talk about them in a future post.

I just love the way that nature solves these problems, getting that turbulent flow just right is a feat of engineering and fluid dynamics. Then using physical structure to tackle an NP-complete problem, well you've gotta admire it.