An interesting conversation regarding evidence of aging in bacteria previously thought to be essentially immortal is currently taking place on the Gerontology Research Group mailing list. Biogerontologist Aubrey de Grey had this to say on the topic:
Based on the evidence so far available, I think there is indeed a truly fundamental phenomenon being demonstrated here. It seems likely to me that all unicellular organisms that have cell walls (as opposed to only cell membranes) will behave much as shown in this study, because the cell wall is a heavily cross-linked proteinaceous structure within which accumulating damage does not diffuse (as it does in lipid membranes). If E. coli grow by adding more cell wall at the centre (which I think they do), therefore, increasingly old "poles" will indeed feature an increasing amount of damage (e.g. extra cross-linking and hardening). (Note that the authors' use of "pole" is possibly misleading, as that term normally applies to the tip of the mitotic or meiotic spindle, a structire not present in bacteria.) So in hindsight (presuming that the above turns out to be true), this is (just as Pete Estep said) a case where we fell too in love with the natural hypothesis in the absence of data. Rather like the inability of one mutation to extend lifespan, which Tom Johnson had such a hard time in getting the field to accept back in 1988. But note, it doesn't apply to animal cells, as they don't have cell walls. So a crucial thing to do is to repeat this experiment with single-celled animals.
It's probably the most talked-about scientific publication of the month, in fact. It's certainly very exciting, but various details need to be kept in mind at this stage:
1) The experiment was done for eight generations and showed a linear decline in the growth rate with increasing numbers of generations in which the old pole was inherited. This is in contrast to the standard pattern in aging where the functional decline accelerates with age. It is thus very important to extend this study to 20 or 30 generations to see whether the trend eventually accelerates or levels off. Of course the entire lineage does not need to be followed -- one just needs some cells at each point in the virtual lineage.
2) The authors measured "growth rate", and they really do mean rate of increase in the size of the cells. However, they note that new-pole daughters are larger on average than old-pole ones, and additionally that the new-pole daughter tends to divide first. The latter is to be expected, since increase of size is generally limiting for generation time for bacteria in rich medium. This merits a lot more discussion (for example there is nothing about whether multi-generation-old-pole cells are especially small), because the whole result may be because smaller cells grow more slowly at first (not that that wouldn't be interesting, of course). The authors describe the above results as showing that there is no "juvenile phase" whereby the new cell needs to go through some initial maturation process before it gets going, but they forget that it may be the daughter with the old pole that is going through such a phase.
3) The authors allude in the discussion to the phenomenon constituting a 2% "cost" of the aging process at the population level. They don't elaborate, but I think the meaning must be that the colony would grow 2% faster if all cells grew as fast as the new-pole ones. But this is not the right calculation if one wants to determine cost, because if the divisions were precisely symmetrical then the old-pole cell would grow faster but the new-pole cell would grow more slowly. I haven't done the maths but I strongly suspect that asymmetrical division (and hence asymmetrical dilution of damage) can for some examples of the function linking growth rate to damage levels confer a higher colony growth rate than symmetrical division.
4) Possibly the main reason the above points matter so much, especially the last one, is because of the effect on the validity of concluding that the observed phenomenon is universal, inescapable etc. There are numerous circumstances in which an organism's optimum metabolic tactics vary non-linearly with stress: for example the increase of maintenance at the expense of reproduction in caloric restriction, or the fusion response of mitochondria to high stress, or the senescence response of cells to over-frequent double-strand breaks. I suspect that under a variety of alternative, reasonable assumptions, the colony growth rate would switch from being best with symmetrical division to being best with asymmetrical division depending on whether stress (e.g. oxidative) and hence rate of damage accumulation was above or below some threshold. If so, repeating this experiment under low oxygen might give different results.