Friday, October 12, 2007

The Most Beautiful Experiment in Biology

This week's citation classic is:

Meselson M, Stahl FW. 1958. The Replication of DNA in Escherichia coli
Proceedings of the National Academy of Sciences USA 44 (7): 671-682.

After Watson and Crick published their structure of DNA in Nature in 1953, the race was on to empirically demonstrate its correctness because, in fact, Watson and Crick's model had precious little going for it other than the fact that it was so damned pretty.

The beauty of the Watson-Crick model is that the structure implies that information can be coded in the sequence of bases and this sequence can be replicated by splitting the double helix in two and constructing new double helixes from the two single strands. But how to prove this?

Enter Caltech grad student Matt Meselson and postdoc Frank Stahl. The two met in a physiology course at the MBL in Wood's Hole: "I partied my way through that course," Stahl confesses. "During the partying, I met Meselson."

Meselson and Stahl decided they could determine how DNA replicated when all the big shots, such as the great Delbruck (see Delbruck, PNAS, 1954), had failed. They spent a couple of years futzing around with phages before hitting upon the idea of using E. coli and heavy nitrogen. 15N contains an extra neutron and molecules made with it would be heavier than molecules containing 14N. So Meselson and Stahl grew E. coli in 15N for many generations, resulting in cells with much heavier than normal DNA. This DNA was extracted, combined with DNA from normal E. coli, and ultracentrifuged in a cesium chloride solution. The photos of the ultracentrifuged DNA showed that the lighter DNA formed one band and the heavier DNA formed another. So far, so good.

Now the clincher. The cells grown in 15N were now transferred to regular 14N media and allowed to grow for one generation. This DNA was then extracted and ultracentrifuged. This DNA showed a band intermediate to that of the heavy DNA and the light DNA. Why? Because the original heavy DNA split in two and a new light DNA chain was constructed to match the parental heavy DNA strain. Semi-conservative replication!

“Stahl credits the beauty and success of their paper to two things. First, the "delightfully clean data" were serendipitous. The clean data peaks they observed resulted from the DNA fragmenting during handling; unfragmented DNA would not have separated as nicely. Stahl likens pipetting DNA to "throwing spaghetti over Niagara Falls." The stress of the pipetting caused tremendous shearing of the DNA, although they did not realize this at the time, nor did they realize how critical this would be to obtaining clean peaks. In addition, Meselson was a "stickler for clarity," said Stahl. "Every single word in that paper was discussed several times before being allowed to keep its position in the sentence." Tinsley Davis. PNAS | December 28, 2004 | vol. 101 | no. 52 | 17895-17896.

Animations explaining the experiment are available here and here.


  1. It's a shame that Meselson & Stahl were never honored with a Nobel Prize for this seminal work.

  2. Yeah, it is kind of weird that they didn't get a nod. Well it's not too late...

  3. One of my graduate student colleagues claimed he would pay to hear Meselson read from the phone book -- he is a wonderful speaker and explains things very well.

    At least in talking about it with Harvard grad students, Meselson claims that the goal of the experiment was to _disprove_ the Watson & Crick model.

  4. If I remember the details correctly (haven't thought about this in a while), the dirtier results with longer DNA strands would actually be due to some biology that would be discovered later: Okazaki fragments & lagging strand synthesis. If memory serves me, this does result in DNA fragments that are actually a mix of old and new DNA.

  5. I think this post over-simplifies things.

    1) The Watson Crick model had Chargaff's rules and the Franklin X-ray diffraction data going for it, as well as the correct chemical structures of the bases. So ti wasn't only pretty.

    2) Semi-conservative replication was not a necessary consequence of the structure, nor was it uniquely characteristic of the Watson-Crick structure. Several otehr structural models extant at the time could have also been replicated by a semi-conservative mechanism, and other replication mechanisms could have existed for the Watson-Crick model.

    What is true is that the combination of the structural model and the Meselson-Stahl experimental results were consistent with a model for DNA replication that turned out to be correct.

  6. One of my favorite memories of my time as an undergraduate at Harvard was when I got to play Matt Meselson in the 1st year Biochem/Cell/Molecular PhD students "play" making fun of all the faculty in their departments. This was fun in a large part because Meselson was the only professor they chose to not really crucify, because they liked him so much.