Far Side "Just gotta be me"At the peak of an immune response, hundreds of thousands of identical T cells are scampering about, searching out the pathogen and doing their own special T cell things to try to get rid of it. We know that these hundreds of thousands of cells weren’t there at the onset of infection; the whole T cell schtick involves rapid expansion of very, very rare cells. Only a very few T cells are able to recognize any particular antigen; but within a few days, the progeny of those rare cells are now common, and all retain their ability to recognize the same antigen. 1

In the past few years, we’ve learned a little more, quantitatively, just how dramatic this expansion phase is. Delicate work has established that there are maybe 20 to 1000 potentially-reactive T cells in a mouse, before infection (see my discussion here and here). Those few cells are the precursors of the huge numbers of T cells a week or so after infection.

If you think about it, it’s pretty remarkable that these few T cells, hidden within millions upon millions of other, irrelevant, T cells, ever get the signal to divide. That signal is carried by specialist antigen-presenting cells, most often dendritic cells. A simplified overview goes something like this: A dendritic cell is hanging out somewhere in the body — let’s say in the skin. It’s constantly filter-feeding, sampling the surrounding environment. Mostly, this surrounding environment is innocuous; there are only normal self antigens in it. If the DC finds evidence of an infection, like, say, viral RNA, then it grinds into action; it migrates to a local lymph node and shows the local T cells everything it (the DC) has been exposed to over the past 24 hours or so. Early in an infection, there usually aren’t a lot of pathogens present, and so there aren’t very many of these activated DC; maybe a few hundred or a thousand.

Meanwhile, the naive T cells are also roaming around, from lymph node to lymph node, scanning as many DC as they can. If there’s nothing they recognize in the DCs of one lymph node, off they go to the next for more scanning.

So for a T cell to get the go signal, a tiny number of specific T cells (behaving identically to the vast number of irrelevant T cells) have to make contact with a tiny number of specific DC (hidden within the vast expanse of all the body’s lymphoid tissue), just by randomly bumping into each other.

Now, here’s a question: When you get that large pool of activated T cells a week after infection, are these the product of all (let’s say) 500 precursors, or are they all the progeny of one or two lucky precursors that managed to bump into the right DC? In other words, how efficient is the bumping-into-DC process? It’s remarkable enough that it works at all; could it, even more remarkably, be so efficient that all the possible precursors manage to find a partner in the first day?

Amazingly enough, the answer is yes, all the precursors do find a partner. A paper from Ton Schumacher’s group2 has showed that there are virtually no wallflowers. Recruitment into the activated state is close to perfect (they demonstrate 95%). They go on to calculate how many random interactions you need, to get this level of recruitment, and suggest that there have to be over 50 million interactions to get this level of recruitment — but this is still in the right ballpark from what we know about rate of interactions:

Assuming that naïve antigen-specific CD8+ T cells are present at a frequency of ~1:100,000 within a CD8+ T cell pool of ~20 x 106 cells, it would require around 59 x 106 T-DC interactions to achieve 95% recruitment, a number that is largely independent of variations in precursor frequency within the physiological range. It has been estimated that DCs are able to interact with at least 500 different T cells/hour; thus, a pool of <2000 antigen-presenting DCs could suffice to achieve this near-complete recruitment. 2

A couple of groups have made movies of T cell/DC interactions, and when you look at those the figures seem more plausible.  Here3 is a movie from the Cahalan lab.4  (Go to their lab page for more fascinating immunology movies, by the way.) This is two-photon microscopy, taken in the lymph node of a live mouse.  DC are green, T cells are, hmm, sort of an orangey red.  The left panel shows T cells interacting with DC that don’t have the appropriate antigen; the T cells charge in, take a quick look at the DC, bounce off, and move on to the next one.  (This movie is time-compressed, of course; we’re looking at a couple of hours here.)  On the right, we see DC that do have the appropriate target for the T cells.  Here the T cells have bumped into the DC and immediately stopped looking further; they just sort of hang with the DC, soaking up the information, and preparing to move off into the activated T cell program.

The frenetic action on the left gives a sense of the rate of interaction, and 500 T cells per hour doesn’t seem too far off.


  1. And then, as part of the same program that triggered their massive expansion, the cast majority of the expanded T cells die off again, restoring the immune system almost to the original status quo.[]
  2. van Heijst, J., Gerlach, C., Swart, E., Sie, D., Nunes-Alves, C., Kerkhoven, R., Arens, R., Correia-Neves, M., Schepers, K., & Schumacher, T. (2009). Recruitment of Antigen-Specific CD8+ T Cells in Response to Infection Is Markedly Efficient Science, 325 (5945), 1265-1269 DOI: 10.1126/science.1175455[][]
  3. If I’ve embedded this properly[]
  4. Miller, M. (2002). Two-Photon Imaging of Lymphocyte Motility and Antigen Response in Intact Lymph Node Science, 296 (5574), 1869-1873 DOI: 10.1126/science.1070051[]