Matt Meselson describes his most famous experiment (with Frank Stahl)

July 18, 2019 • 11:00 am

In 1958 Matt Meselson, whom I knew slightly at Harvard (he was a terrific guy), performed, along with Frank Stahl, an experiment that John Cairns called “the most beautiful experiment in biology”. What he and Stahl did (see description here) was to use density-labeled components of DNA to choose among which of the three methods of DNA replication floated at the time was correct (people didn’t know how DNA replicated in 1958; this experiment settled the issue):

In “semi-conservative replication”, each strand of DNA unwinds and makes a copy of itself, so that each DNA helix in the next generation of DNA has both a parental strand and a new strand synthesized from nucleotides and sugars. “Conservative” replication involves each double strand making another whole double strand.  “Dispersive” replication involved the DNA breaking, with each break synthesizing new DNA, matched to the other strand, in bits. They’re portrayed above.

Meselson and Stahl’s genius was to use an in vivo replication in E. coli producing DNA strands labeled with heavy isotopes (15N) that, while chemically identical to non-radioactive nucleotides, would be distinguishable from the non-labeled strands because the former were heavier and could be separated by vigorous centrifugation. (They used labeled nucleic acids as the components of the original strands; those labeled nucleotides were themselves synthesized by growing the starting bacteria for a few generations on “heavy” ammonium chloride—the only source of nitrogen—a component of nucleic acids—in the bacterial medium.)

The beauty of the experiment is that the results—confirming semi-conservative replication—were visible in a single photograph (below), and were unambiguous. It was a lovely experiment, and I think deserved a Nobel Prize (sadly, one wasn’t given for this).

This nice 13-minute talk by Matt, taken from an iBiology talk website, describes this experiment. He and Stahl started by putting bacteria containing fully “heavy” DNA into medium with non-heavy ammonium chloride, so that all the new DNA synthesized would be light.

Under the semi-conservative hypothesis, the next generation of DNA would be “half heavy”, as each helix would have both an original heavy and new light strand, with the latter containing nucleic acids synthesized from the lighter nitrogen in the medium.

Under the conservative hypothesis, the next generation of DNA would consist of fully light double strands and fully non-heavy original double strands. There would only be two types of strands detectable, and those would stay, with the heavier ones eventually disappearing as their carriers died and new DNA was formed. And under the “dispersive” hypothesis, the next generation of DNA would be not fully heavy or not fully light, but a schmear of ‘partly-heavy helices”. You’d get a mess of mosaic strands in subsequent generations.

Well, listen to Matt describe this pathbreaking experiment below. I’ll give a link to their paper and the famous figure that convinced everyone below the video.

Here’s the famous figure, beginning with heavy DNA at the top from E. coli (right stripe in generation 0). The density of the centrifuge gradient increases to the right, and strands tend to settle where their density matches the density of the cesium chloride in the centrifuge tubes.

When the bacteria were put on non-labeled medium, and the tubes scanned with UV-absorption, which picks out the DNA, you see that in the first generation all the DNA is heavy (original bacterial DNA). As those bacteria replicate and form new DNA strands, the heavy helices begin to wane and we start to see half-heavy helices (lighter stripe forming in the left, lighter part of the gradient). This stripe gets darker after more “hybrid molecules” accumulate (generation time is shown on the right of the figure). After one generation of replication, you get hybrid strands which are lighter than the original ones (the bands show the position in the density gradient of the centrifuge). Then, after another generation, the hybrid stands themselves replicate, forming a double-light helix from the newly synthesized strand as well as the half-heavy strand containing the original heavy strand of DNA. By generation four, nearly all the helices are fully light (to the left), as the original strands are in a minority in the mix since their carriers have died or been outbred. In other words, the three bands predicted by the semi-conservative hypothesis were seen. The experiment ends three photos from the bottom, at generation 4.1.

The presence of the three well-demarcated strands forming in sequential order shows unambiguously that the semi-conservative model of DNA replication is correct. You don’t need statistics to get the answer here!

You can download the original paper by clicking on the screenshot:

I don’t know of a more beautiful—or unambiguous—experiment in modern molecular biology. And the stuff about Meselson and Stahl being locked in a room with food and a sleeping bag until they wrote that paper happens to be true. (For more, read The Eighth Day of Creation by Horace Freeland Judson).

39 thoughts on “Matt Meselson describes his most famous experiment (with Frank Stahl)

  1. What a wonderful and important experiment. Some layman’s questions, if I may:

    1) How do the bacteria react to the heavy nitrogen compounds? Traditionally we say that isotopy doesn’t have any chemical effect – how true is that here?
    2) “not found to be biologically significant” – do these other ways occur, just in small amounts? (I imagine they would – competing reactions seem to be endemic to organic chemistry!) If so, what biological effect would they have if somehow they occurred more? (I’m wondering if there are diseases or the like that would result from a catalyst favouring the side reactions!)

    1. If I may…

      1) The nitrogen isotope is stable, so it doesn’t have any impact on the cell. It is biochemically indistinguishable from 14N.

      2) I suppose there are exceptions (single stranded DNA viruses, for example) but I too am curious about others.

    2. Under the experimental conditions, or rather in the lead-up to the experiment, they forced the bacteria to use only a heavy form of nitrogen to build into their DNA. The bacteria had no lighter isotope to choose from. I can sort of hand-wavingly suggest that were these bacteria then made to co-exist with regular bacteria, they might somehow experience a fitness disadvantage at first in the rate that they can replicate DNA. But this disadvantage would be dispelled over a few generations as the heavy DNA is copied to make light DNA.

      In natural conditions, organisms live in an environment where the molecules they incorporate are a mixture of heavy and light isotopes. I know in particular that organisms that ‘fix carbon’ by photosynthesis will use more of the lighter isotope of carbon (called carbon-12), over the heavier stable isotope called carbon-13. So the ratio of C-12/C-13 is increased in them, above the ratio for these isotopes in the non-living surroundings. Because other organisms either directly or indirectly wind up eating photosynthetic organisms, we too contain an elevated ratio of C-12/C-13.
      Perhaps like C-12 versus C-13, organisms will prefer to incorporate the light isotope of nitrogen over the heavy isotope, if given a choice.

      1. I was pretty curious about that preference for C-12 over C-13, as it is chemically indistinguishable. Googling found this:

        “When plants photosynthesize carbon dioxide, they first capture air inside small openings in the leaves, called stomata, by a process called diffusion (diffusion is the random movement of particles from an area of higher concentration to an area with a lower concentration of that particular particle). As the air randomly enters the stomata, proportionally less heavy 13C enters a plant than the lighter and faster 12C (meaning that the isotopes fractionate according to their relative masses).”

        It is during the actual physical process of getting the carbon dioxide into the plant that the difference occurs. The C-13 moves slower, so proportionally less gets in.

        I learned something new today!

        1. The bonds to the heavier isotope take greater energy to break. This is called an “isotope effect”. Compounds with the heavier isotopes will therefore accumulate slower than compounds with lighter isotopes.

          Passive diffusion – like the drifting of CO2 into plant leaves – is independent of the carbon or oxygen isotopes (AFAIK). It’s only the enzymatic processes that exhibit the isotope effect…AFAIK…

          1. And I think this is irrelevant to the experiment at hand because it is about labeling the different strands, not relative rates of production or … quantities (I think).

  2. Damn! Just reading about this is exciting – part of the history of evolutionary theory that’s not taught in general science class, or even the Sci 105 course I took (way back when, but post-1958) where we got the black box lesson on how to think scientifically. This is a great illustration on how real scientists set up a testing experiment. Thank you, Professor.

  3. Those must have been heady days during the beginnings of what was later called molecular biology. So many brilliant minds (from disparate areas of science) were attracted to these fundamental problems, and so many classic experiments. And I note that the paper was “communicated” (as was necessary to publish in PNAS) by no less than Max Delbruck, also at Cal Tech.

  4. A wonderful experiment, to be sure. But another set of experiments rivals it, I think, both in the scope of its implications and the decisiveness with which alternative explanations are ruled out. It was also much cheaper to perform. This is the great paper on the general nature of the genetic code by Crick, Brenner, Barnett, and Watts-Tobin.

    There is a nice story about these phage genetic experiments. One night, Leslie Barnett (the technician) supposedly stopped into the lab and found Francis Crick there late, beaming over phage plaques on one of the Petri plates. According to legend, he exclaimed to Leslie: “Now you and I are the only people in the world who know the code is read in threes.”

    1. I love these science stories (even if they are sometimes apocryphal). One I heard was about how stop codon mutants were first identified was that a couple of scientists asked a colleague to lend a hand in the tedious task of replica plating bacterio-phage libraries but the colleague declined. In response the scientists said they’d name any mutants they found after him. That’s how we got “Amber” stop mutation!

  5. I wonder if there weren’t so many nitrogen nuclei in DNA bases, the difference in mass would be undetectable? In other words, the experiment works because the detection limit of their ultracentrifugation technique was low enough – I wonder how they knew that.

    1. I think they did a test run. They first ran a mixture of double stranded heavy and double stranded light DNA in the CsCl gradient and found that the DNA resolved into two distinct bands with a gap between them. They knew then that if they could get the bacteria to make DNA that was 50% heavy and 50% light it would settle out as a band in the middle.

      1. [ my Typo : “were” not “weren’t” ]

        I think this is a substantial experiment, with the amount of N15 nitrogen – but I suppose at some point they just had to try it and see if they separate. But still, I think they had to have a pretty good notion that it would…. as in, perhaps, some previous technical papers on the method…

        And if DNA did not have so much of a nitrogen fraction, this might not have had a chance – but I don’t know…

  6. It was initially unclear to me why the results don’t also support the dispersive hypothesis, which also produces half-heavy strands. Both would diminish the heavy portion by half each generation, and I didn’t understand how the centrifuge results distinguished between the heavy half being a single helix versus being randomly dispersed heavy regions comprising about half of the molecule.

    But I think I’ve got it now. The key point is that the two hypotheses should diverge after the second generation.

    For the semi-conservative model, an original heavy-heavy molecule (HH) produces 100% heavy-light molecules (HL) in the second generation, but after that begins to diverge. In the third generation, you get 50% HL and 50% LL. In the fourth you get 25% HL and 75% LL. Etc. So the centrifuge results should show one band splitting into two and diverging in weight.

    In the dispersive model, the original heavy molecule (H) produces in the second generation a heavy-light molecule of proportion H1:L1, and I think the centrifuge results would be identical. But In the third generation the H1:L1 molecules all become H1:L3, then H1:L7 in the fourth, etc. So the centrifuge results would show only a single band moving left.

    That leaves me wondering about the conservative model. The original HH molecules become diluted to 50% HH and 50% LL in the second generation, then 25% HH and 75% LL, etc. So it should also one band splitting into two, but they would diverge only in frequency and not weight.

    1. I don’t think your description of the dispersive model is quite right. As Dr PCC(e) said, if that were the case, you would wind up with smeared bands (that is not discrete bands) because at each replication the strands would have variable heavy isotope dosage.

    2. It was initially unclear to me why the results don’t also support the dispersive hypothesis, which also produces half-heavy strands

      That was my thought at first but only the semi-conservative would produce exactly half heavy strands. In order to replicate the result, the dispersive method would have to randomly split the parent strand into two exactly equal piles every time.

      And then on the second generation, where you see strands that are half heavy and strands that are fully light and nothing else, the splits would have to distribute the heavy nitrogen very carefully.

      The thing I love most about this, is that, having been educated in an era where this was already fully understood, I have always assumed that the semi-conservative process was obviously right. In fact, it never occurred to me that anybody had ever proposed anything different. Whether obvious or not, it was deemed necessary to confirm the hypothesis by experiment. This is a perfect example of Richard Feynman’s description of the scientific method.

  7. I was very impressed by the way Meselson describes how the newly discovered double helix virtually told scientists how to proceed with follow up research. The molecule itself seemed to ask them how it did all it’s functions, from splitting apart and shifting pieces around the cell, to creating and assembling the new copy. Other discoveries must do that as well, but this is a pure example.

  8. I’m puzzled, how could the conservative or dispersive alternates work? Where would the information for the others segments of DNA come from?

    1. I wonder about that too. The semi-conservative method looks like the obvious winner, although I do have some retrospectivity to help with that.

    2. I suppose in 1958 it seemed plausible that there could be molecular machinery that could read out the DNA sequence without unzipping it by inserting a probe into the groove of the helix and feeling the bumps of the different base pairs.

  9. I’ve always loved that experiment. Until you encounter it, you don’t necessarily realize that a centrifuge could make the distinction.

  10. Once in the 1990s, when I was visiting the Marine Biological Laboratory in Woods Hole to lecture in one of their courses, I was eating in their cafeteria. An older guy sat down opposite me and introduced himself as Matt Meselson. Of course I had studied his experiment with Stahl in graduate school. I suppressed the urge to reply “yes, but where’s Stahl?”.

    He wanted to grill me about whether my knowledge of phylogeny methods and population genetics could shed light on the genetic variation of the uniparental bdelloid rotifers. He worked on them for many years in his retirement. Obviously he continued to find fascinating areas of work. I agree with Jerry, he is a really nice guy too.

  11. Inspiring. It takes a remarkable mind(s) to come up with such ingenious and groundbreaking experiments.

    I imagine these photos reside at CalTech. It would be fun to see in person.

  12. Sorry, and I don’t mean to be a pain. I think you mean “nucleic acids”, not amino acids. I only mention it because of the historic significance in the Avery, McCarty, MacLeod (as well as Hershey & Chase’s confirmatory) experiment many years before when the issue was whether or not it was protein vs DNA that was the heritable material.

  13. I didn’t know of this experiment but it’s strange to see such a definitive and elegant, old school result when I’ve flash-forwarded 70 years and know exactly what this experiment did down to the atoms involved.

    It’s analogous to listening to an old phonograph record instead of the same recording via high-res audio downloaded from the cloud.

  14. Just now managed to watch it. Delightful!

    I didn’t know he thought of using deuterium first, or that he was Pauling’s student- that makes complete sense now!

    It was interesting to hear the writing process: taking a first principles approach, they wondered how anyone could know every hypothesis? This should serve as a cautionary tale for philosophy – too much philosophy and the paper never gets written. Also, publication norms seem to be quite different back then – but perhaps that’s me.

    I particularly liked the insight to the Watson and Crick paper. It’s hard to grasp how much the X-ray picture told them at that time, given how little they knew at the time, because atomic structure is so common now it is even featured on TV shows.

    1. I also liked how he said numerous accidents punctuated the story – the gradient forming before their eyes, the meetings between scientists – including Feynman!

  15. For parts like these – classics – I think it would be very interesting to read the reviewers’ comments.

Leave a Reply