Some phenomena in biology are random, and some phenomena are called random until scientists discern the patterns that underlie them.

In the Feb. 24, 2006, issue of Science, researchers from the National Cancer Institute in Bethesda, Md., published an article that moves chromosome segregation from the random into the orderly universe: They described how certain cells keep together the "originals" of DNA strands during cell division.

During the DNA synthesis phase of the cell-division cycle, the famed double helix unwinds and a new complementary strand is made from each old strand. During regular cell division, or mitosis, the DNA strands condense into chromosomes. The cell then divides, with each daughter getting one copy each of the maternal and paternal chromosome.

Theoretically, each of the new double helices is identical to the old one, and which daughter cell gets which copy of the maternal or paternal chromosome should not matter, as long as it gets one of each. Indeed, one of the original attractions of the double helix model was that it was intuitively obvious how the information in a double helix might be faithfully copied and distributed to daughter cells.

Francis Crick and James Watson ended their 1953 Nature paper proposing a double helical structure for DNA this way: "It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material." The sentence has been described as coy, which, as Watson noted in his autobiography, was not a trait normally associated with either author.

But that copying mechanism might not, in fact, turn out completely identical copies. Amar Klar, head of the developmental genetics section at the National Cancer Institute and senior author of the Science paper, has long been a proponent of the theory that the old and new strands are different and do not always randomly segregate into daughter cells during cell division.

"Everybody said it was 50-50," Klar told BioWorld Today. "But nobody checked it."

There are two copies of each chromosome in each cell; for simplicity it is best to consider only a single gene on a chromosome, as well as the centrosome - the location at which the chromosome copies are pulled apart during cell division. The cell contains two copies of the gene; one from the mother and one from the father.

In each case, the gene itself is on one DNA strand and its complementary noncoding sequence on the other. During DNA synthesis, both DNA double strands unwind and are copied, leading to four copies of the gene: two containing the old genes and new noncoding strands, and two containing new genes complementary to the old noncoding sequences.

If segregation of the chromosomes were a random event, one old and one new gene would occupy the same daughter cell around half of the time. Klar, who is himself a yeast geneticist, noticed that that was not what happened when his colleagues experimented with mitotic recombination in mice. His colleagues used the loxP-CRE system in mice doing recombination studies in mouse chromosomes 7 and 11.

The recombination itself should not affect the way chromosomes segregate during cell division; areas on the chromosome beyond the recombination point should have both copies of the maternal marker in about 50 percent of cases, because one of the paternal chromatids would have the maternal version of the marker after recombination. But Klar noticed that for chromosome 7, recombination led to daughter cells that were homozygous for the maternal (or paternal) marker in every single case, suggesting that they were not segregating at random.

Because the recombination itself theoretically could be affecting the way chromosomes segregate, Klar and his co-author Athanasios Armakolas used a combination of mitotic recombination and methylation patterns to follow each individual DNA strand through a cycle of mitotic recombination and cell division.

In their Science paper, Armakolas and Klar described how, depending on the cell type studied, the old copy of the paternal and maternal genes always were together in some cell types, sometimes together in other cell types and never together in yet other cell types.

"In theory, there can only be three patterns," Klar said. "And we got all three."

Klar said that his findings can explain many complex asymmetries. He has applied his findings to everything from hair whorls to more complex phenomena such as handedness.

"Higher-organisms people look at yeast people and say, ‘No, no, it has to be more complex than this,’" Klar said. But to him the problem of what determines handedness, or asymmetrically positioned organs in the human body, is basically no different from asymmetric division in a yeast cell.

"Left-right axis asymmetry is inherent in the DNA structure," he said. "The machinery of DNA distribution exists, if someone is willing to look. Nature made them, and nature knows that they are different, and every once in awhile, nature will use that information."