June 9, 2010 § 4 Comments
Roy Kishony pointed out this cool paper by Suckjoon Jun, one of the Fellows at the FAS Center for Systems Biology. (Wang P, Robert L, Pelletier J, Dang WL, Taddei F, Wright A, and Jun S. 2010. Robust growth of Escherichia coli. Current Biology 20 1-5. PMID: 20537537). There are two things that are cool here. The first is the technology: a microfluidic device called the “mother machine” that traps a single “mother” cell at the bottom of a growth channel where it is immobilized for study. As the mother grows, a chain of daughter cells get pushed up to the top of the channel and carried away by the flow of growth medium, which also provides sufficient resupply of nutrients to diffuse to the bottom of the channel and keep the mother cell happy (they checked). The whole thing looks like a tiny sausage machine.
The second thing that’s cool is that because the mother cell never moves, you can follow it (her) for hundreds of divisions (far longer than has ever been possible before) and have no doubt that you are still following the same cell. This means that you can look directly at how the growth rate of an individual cell varies with time. There is a ton of information here, but some take-away messages are: (1) the rate of growth shows short-lived fluctuations on a time scale that is similar to the cell division time scale (i.e. the factors that determine growth rate seem to be randomized at roughly the same rate as the cell divides); (2) over longer time scales, averaging out these fluctuations, growth stays remarkably stable, even tracked over 200 generations; (3) at about 50 generations, some unidentified factor builds up to the point where the SOS response is activated, and the cells begin to filament with relatively high frequency. But they don’t start to die at significant levels until after about 100 generations — unless you knock out the SOS pathway, in which case they show no filamentation and a constant death rate of 2.7% per generation. So death without SOS is random (constant from generation to generation), while death with SOS is greatly delayed and shows a kind of threshold effect. The study raises a number of questions, not least the question of why the growth rate is so stable in the face of what must be (given the results of the SOS mutants) steadily accumulating damage.
Another question I’m quite sure you’re keen to have answered is, why did they call their device a “mother machine”? After all, it isn’t producing mothers. The answer, I believe, lies in a paper published in 1963 (Helmstetter, C.E. and Cummings, D.J. 1963, Bacterial synchronization by selection of cells at division, Proc Natl Acad Sci U S A. 1963 50(4): 767–774), which described an ingenious device for synchronizing the cell cycles of bacteria without using drugs or other invasive manipulations. Basically, “mother” bacteria were stuck to a cellulose filter and medium running over the filter and into a funnel was used to collect newborn “daughter” bacteria; if you collect the eluate for a very short time, the daughters you collect were essentially born all at the same time. Helmstetter called this device the “baby machine” and he went on to adapt the basic idea to the study of yeast, and (with elaborations) to mammalian cells, and Amit Tzur and Ran Kafri in the Kirschner and Lahav labs (with the help of Valerie LeBleu from the Helmstetter lab) went on to use the mammalian system to ask how cell growth is coordinated with the cell cycle. This is impossible if you use drugs to synchronize the cells, since the drugs only stop the cells dividing, they don’t stop them from growing.
Although the “baby machine” and the “mother machine” use different strategies to hold the mother cell still, and are aimed at quite different questions, I can’t help thinking that Suckjoon found the name of the older device inspirational. Here’s hoping that the mother machine is easier to use than the baby machine; Amit still unconsciously grinds his teeth when you ask him about it.
Thank you so much for your nice comments, and also Roy for bringing attention to our paper. We know the answers, or have clues, to some of your scientific questions, but for now let me answer the last question in your blog article.
I owe the name “mother machine” to Johan Paulsson. In the last Fall, we had this fun meeting of “systems biology” folks from the Boston area as well as the UK and elsewhere. There, I described what my lab was up to and I called our device a fairly neutral name. Something like the ‘microfluidic continuous culture device,’ explaining it’s just a microfluidic version of the baby machine by Helmstetter. As soon as I finished the description, Johan exclaimed, ‘So it’s a mother machine!” and I said “It’s like a baby machine with the ability to keep and watch the mother cell.” In the mean time, we were trying various names for our device such as… the “iChannel.” Much later, I recalled Johan’s remark and liked it a lot, and the final name was decided.
So, yes, I am a great admirer of the literature in the 50s and the 60s. And, yes, by reducing the length of the growth channels, we can also collect a synchronized population, just like the baby machine and while monitoring how the mother cells are doing, if necessary. As for how easy/difficult to do an experiment with the mother machine, suffice to say that it now takes 2-3 weeks for an undergrad student to learn how to make the mother machine and set up a timelapse microscopy experiment in our lab, and the success rate is practically 100%. We plan to publish a complete blueprint + detailed protocols on the web sometime this summer so that anyone can do what we are doing (following Andrew Murray‘s suggestion to be generous, and I must say I agree with him in this case).
(writing from the Cold Spring Harbor Laboratory)
Thanks for the story, Suckjoon! Even more interesting than I thought.
So the Francois Taddei paper (PLoS Biology 2005) saying ” We show that the cell that inherits the old pole exhibits a diminished growth rate, decreased offspring production, and an increased incidence of death” is mostly wrong?
Sorry about being slow. All I should say is that our data speak for itself.