by Nathan H Lents, Ph.D.
Around the world, there are thousands of scientists performing experiments at any given moment. Every once in a while, an experiment is performed and published that appears so clever, so important, and so successful in its goals, that it is destined to be cheered by scientists far and wide and taught in science classrooms for decades to come. However, with the passage of time, these so-called “classic experiments” seem more dramatic, ingenious, and clear than they were in their contemporary timing as the memories of complications, contradictions, and controversy fade. Nevertheless, what often sets apart these key experiments is not that they are especially complex, but that they are elegantly simple. The power of simplicity in an experiment is that it reduces the chance of alternative explanations for the results.
As explained in our module Ideas in Science: Theories Hypotheses, and Laws, a key feature of the modern scientific method is that valid scientific hypotheses make predictions that can be tested. Thus, the testing of predictions is a major part of scientific research, and part of the historic nature of many classic experiments is that they tested the predictions of a key scientific hypothesis in a way that provided a clear answer. The 1958 experiment by Matthew Meselson and Franklin Stahl is an example of such an experiment, and is one of the most famous in all of molecular biology. With one cleverly designed experiment, they tested the predictions of three different scientific hypotheses simultaneously, and the field of DNA biology was changed forever.
Following the discovery of DNA as the genetic material (see DNA I), the new field of molecular biology focused intently on how DNA functions. One of the most important features of DNA is its ability to be copied accurately. When a cell, whether it is a yeast, a bacterium, or a human cell, divides in two, both resulting cells are genetically identical to each other and to the original parent cell. Thus, prior to division, a cell must somehow copy all of its DNA so that both resulting cells have the full complement of genetic material. Indeed, scientists such as Edwin Chargaff and others had observed that the amount of DNA in a cell doubles prior to cell division. The pool of DNA is then split equally between the two daughter cells, so that both have the same amount of DNA as the original parent cell had. But how exactly this DNA doubling takes place was at first a mystery and scientists began to propose several possible mechanisms, or “models” of DNA replication.
Following the proposal and eventual acceptance of the Watson-Crick model of DNA structure, molecular biologists believed that each strand of DNA somehow served as a copy-template for the synthesis of a new DNA molecule (see our DNA II module for more information). However, conundrums remained. Most importantly, scientists had a difficult time envisioning how two strands of DNA that are immensely long and twisted around each other could separate from each other without resulting in the breakage of the strands or them becoming hopelessly entangled. In addition, scientists wondered how the two strands could be pulled apart given the enormous number of hydrogen bonds holding them together. Some envisioned replication as proceeding in short stretches, while others imagined a continuous process much like a zipper. This paradox even caused some prominent scientists of the day to doubt the double-helical structure of DNA altogether. Nevertheless, scientists began working on possible theoretical solutions to the separation of two intertwined DNA strands and by the late 1950s, three hypothetical models for DNA synthesis were being hotly debated: the conservative model, the semi-conservative model, and the dispersive model. Figure 1 below provides a diagram of each of these mechanisms.
Figure 1: Three competing models of DNA replication. This diagram shows the three competing models of DNA replication in the 1950s and 1960s.
Briefly stated, the conservative model of DNA replication holds that when DNA is replicated prior to cell division, one of the DNA double-strands receives all newly replicated DNA in both strands, while the other receives only the two original DNA strands in the parent cell. The semi-conservative model (also called the “zipper model” by James Watson), however, holds that the original two strands of DNA are split from each other and that the two daughter molecules are each comprised of one “old” strand of DNA from the parent cell, and one newly replicated strand. Finally, the dispersive model holds that the DNA is copied in short stretches and that both daughter DNA strands will receive a mixture of the original parental DNA and newly replicated DNA. Each of these three possible models had been proposed by different scientists and each had certain advantages in explaining the separation of the intertwined parental DNA. However, evidence to disprove or support any of the models was scarce.
This changed when Matthew Meselson and Franklin Stahl, two scientists working at the California Institute of Technology (CalTech), constructed an ingenious experiment that tested all three models at the same time. To understand how this experiment worked, it is important to remember how atomic isotopes behave. Although a heavier isotope of a given atom behaves in a completely normal manner in chemical reactions, the presence of an extra neutron (or more) gives the atom a slightly higher atomic mass. As a result, molecules that contain these “heavy” isotopes are more dense. This small difference in density allows scientists to physically separate molecules with different isotopes based on the differences in their density.
For their experiment, Meselson and Stahl used a special form of nitrogen: 15N. Normally, almost all of the nitrogen in any given cell is 14N and thus contains seven neutrons in addition to its seven protons. So, 15N, with eight neutrons, is considered “heavy nitrogen” (but it is not radioactive). When growing cells are fed heavy nitrogen, the 15N isotope enters the cells’ metabolism and significant amounts of it will be incorporated into the nitrogen-rich nucleotides and DNA. Thus, the DNA of cells grown with 15N in their food source would be denser than that of normal cells. The power of having DNA of different densities is that they can be separated by centrifugation.
For this procedure, cells were first broken open; then the cellular contents (called the “crude extract”) were mixed with a solution of the heavy salt cesium chloride and placed in a centrifuge cell with clear quartz walls that allowed the solution to be photographed while spinning. The cell was then spun in a centrifuge at very high speeds for many hours and the heavy cesium ions were pulled towards the bottom of the cell by centrifugal force. Eventually, equilibrium was reached and a “density gradient” was established in the cell with the bottom containing the highest concentration of cesium and the top of the tube containing the lowest. Inside a density gradient like this, all the molecules from the cell extract, including the DNA, will “float” or “sink,” migrating to the spot in the gradient that corresponds to their density. The densest molecules will be pulled toward the bottom of the cell, while lighter molecules will settle higher in the cell, as shown in Figure 2.
Figure 2: The principle of density gradient centrifugation. When a liquid solution containing many large protein and DNA components is placed into a test tube or centrifuge cell and spun at high speed over many hours, the individual molecules separate based on their density. The most dense molecules fall to the bottom, the least dense remain at the top.
Before beginning their analysis of DNA replication, Meselson and Stahl first showed that DNA made with regular 14N could be separated from DNA containing heavy 15N. They accomplished this by growing two separate batches of Escherichia coli bacteria, feeding each batch a different nitrogen isotope. Then, they broke the bacterial cells open, mixed the extracts from both batches into one centrifuge cell, and spun it to establish the density gradient. To detect the DNA, they shined ultraviolet (UV) light on the spinning centrifuge cell because DNA absorbs UV light and thus casts a shadow during exposure of photographic film. Below, in Figure 3, is a black-and-white image of their data, and you can clearly see two bands of DNA, one lower in the cell and thus, more dense than the other.
Figure 3: Density gradient centrifugation of a mixture of 15N-DNA and 14N-DNA. Meselson and Stahl first showed that they can separate a mixture of DNA of the two different densities. The picture on the left is a UV photograph showing the banding of DNA of different densities following centrifugation. The graph on the right is a trace of the intensity of the bands in the picture.
Next, Meselson and Stahl did something interesting. They grew a large batch of bacteria in heavy nitrogen (15N) and then switched the bacteria to a diet that contained only regular nitrogen (14N). This allowed them to distinguish between pre-existing DNA from the parental cells and newly synthesized DNA, because any newly synthesized DNA strands would contain 14N and be less dense. They used this experimental set up to put the three possible models of DNA replication to the test.
Like all proper scientific hypotheses, the three models of DNA replication each make certain predictions and testing hypothetical predictions is a key part of scientific research. In the case of Meselson and Stahl’s experiment, the predictions that each of these models makes are as follows. If the conservative model of DNA replication is true, then one would predict that the bacterial cells grown for one generation (20 minutes) with 14N would have two different kinds of DNA: the original DNA would be the density of DNA grown with only 15N nitrogen, while both strands of the new DNA molecules would be the lighter 14N DNA band. However, if either the semi-conservative or the dispersive models of DNA replication are correct, the double-stranded DNA inside the bacteria after one generation would be a mixture of old and new DNA, and thus, one strand would be made of 15N and one of 14N DNA. Thus, this “hybrid” DNA would be an intermediate density halfway between the 14N and 15N bands of DNA. In Figure 4 below, you can see what the three models of DNA replication predict will happen in the Meselson and Stahl experiment, followed by what they actually observed.
Figure 4: Density gradient centrifugation of E. coli DNA after one cell division. Top panel: the three experimental predictions of three competing models of DNA replication. Bottom panel: The actual data. E. coli grown in 15N DNA were switched to 14N and then harvested at five different time points. The DNA was centrifuged resulting in the banding pattern shown here.
As you can see from their results above, after one generation of cell division, the total DNA of the growing bacterial cells had an intermediate density, halfway between that of
Figure 5: Experimental predictions of three competing models of DNA replication over three generations.
And now, here are the actual observations of Meselson and Stahl (Figure 6):
Figure 6: Density gradient centrifugation of E. coli DNA over multiple generations. E. coli grown in 15N DNA were switched to 14N and then harvested at nine different time points. The DNA was centrifuged resulting in the banding pattern shown here
As they let the bacterial cells grow and divide further, Meselson and Stahl observed that the 15N DNA band disappeared, a band of 14N DNA appeared and then got progressively darker, and a band of intermediate density appeared and persisted at about the same intensity. This strongly discredited the dispersive model of DNA replication, which predicted that only one band of DNA would exist and would get progressively less dense as the amount of 14N DNA in the dispersive mixture increased with each generation. Thus, in one simple experiment with very clear results, Meselson and Stahl solidly disproved two of the possible models of DNA replication, while strongly supporting another.
The scientific community agreed that this was powerful evidence in support of the semi-conservative model. John Cairns, one of the leading molecular biologists of the era called it, “the most beautiful experiment in biology.” To this day, the Meselson and Stahl experiment is taught around the world as a very classic example of the modern scientific method of experimentation. With one simple design, three scientific hypotheses were tested by the observation/verification of their predictions.
Scientists now have a detailed understanding of the molecular events of DNA replication (see our DNA III module). These molecular events occur just as predicted by the Watson and Crick model of DNA structure, and verify that the semi-conservative model of DNA replication is indeed correct. The Meselson and Stahl technique of labeling DNA strands with nitrogen isotopes is still employed by scientists around the world as they continue to explore the mysteries and complexities of DNA, the genetic material of life.hide
Meselson, M. and Stahl, F.W. (1958). "The Replication of DNA in Escherichia coli". PNAS 44: 671–82. doi:10.1073/pnas.44.7.671.PMID 16590258.
Holmes, Frederic Lawrence. Meselson, Stahl, and the replication of DNA [electronic resource]: a history of "the most beautiful experiment in biology." New Haven, CT: Yale University Press, 2001.
Nathan H Lents, Ph.D. "Classic Experiment: Meselson and Stahl: and the Models of DNA Replication," Visionlearning Vol. SCIRE-1 (6), 2011.