Prisoners of Mendel

This inheritance business? Just about licked. We know where everything goes now, what controls what. Got a copy of the blueprint right here in my pocket—See? It folds out.

multicolored interweaving lines

"I guess you can tell I'm skeptical," Ken Weiss said, about two-thirds of the way through his Frontiers of Science lecture, "Reflections on a Golden Age." "I think we've highly oversold the idea that the genome is the blueprint for human life."

Weiss, Evan Pugh professor of anthropology and genetics at Penn State, is a lean man with a brush moustache. He speaks in torrents, often pacing or rocking on the balls of his feet. In his 20s he spent five years as an Air Force meteorologist, and he retains a certain military wryness. For the last 25 years Weiss has been immersed in genetic epidemiology, and he has come to a conclusion: If we really want to understand the role of genetics in human disease, we need to get our minds past Mendel.

Gregor Mendel and his kitchen garden full of peas. In 1865 the Austrian monk published the famous paper that launched modern genetics.

"Mendel's experiments showed that inheritance was particulate," Weiss explained, "that there were specific Ôfactors' that were passed on." His cross-breeding of green and yellow pea plants introduced the idea of dominant and recessive traits, and a corollary: the idea that a disease or trait can be the result of a single metabolic defect or variation. Mendelian inheritance is a straightforward, uniform, predictable process.

Twentieth-century science identified Mendel's "factors" as genes, protein-coding units of DNA. "We've done a terrific job learning what genes are and what they do," Weiss said. "But in many ways we are still in Mendel's conceptual world: either/or, good and bad, dominant and recessive. And I think that the biological world is only like that in exceptions."

On the screen above his head Weiss flashed a daunting block of letters: CTTGATGCTCAGAGAGGACAAGTCATTTGCCCAAGGTCA . . .

"Our genome is three billion of these," he said, "and it doesn't come with annotation. We need other biological data to understand what the sequence means to the organism." This particular stretch was the first 1,600 base pairs of APOE, a gene involved in the regulation of lipid levels in the blood. But not exactly. "If you look at 200 people," Weiss explained, "you find that not everybody's APOE genes are the same. At certain sites, some people have a C instead of a G, or an A instead of a T."

Weiss and his colleagues have found 23 of these "polymorphic" sites in APOE. Some of these sites are probably functionally neutral: A switched letter has no effect on the organism. But "each one represents the sort of variation that could make the difference, in Mendel's terms, between a green pea and a yellow pea. A change in a single nucleotide." The challenge is to find which variable sites, if any, have an important effect.

Even "simple" genetic diseases that follow inheritance patterns like those Mendel identified, Weiss concluded, have turned out not to to be so straightforward. It may be easy to point to the gene that causes a particular disease. But finding the gene isn't the end of the story.

Take phenylketonuria, or PKU, a metabolic imbalance that can result in mental retardation. Studies have linked PKU to a defect in the PAH gene, which produces the enzyme phenylalanine. But researchers looking closer at the PAH genes of PKU patients have identified different mutations in different patients: a total of about 325 individual sites. Not only that, but these variations cluster geographically: Patients from Europe have certain mutations while patients from elsewhere have others. Cystic fibrosis is another example. It is caused by a variation in the CFTR gene—one or more of the 650 or so mutations that have been identified. The champion gene for mutations is TP53, which is linked to several cancers including that of the colon. As of October 2000, Weiss reported, researchers had found 11,106 mutations. Any one of these mutations, or some combination, could be a risk factor for cancer. "And since we have two copies of each gene, the number [of genetic combinations that could be responsible] is actually much higher," Weiss said. Typically, different phenotypes—different severities of disease —;are associated with different combinations.

"When I was in graduate school, researchers wondered why some cases of PKU were responsive to dietary intervention, while others were not. Now we know that it's because different mutations in different combinations have different effects in different environments. The most severe mutations almost always give you PKU, but a lot of combinations give you something intermediate."

Even more tangled are the genetic causes of common, chronic afflictions like heart disease and diabetes. These are complex traits, the results of multiple, interacting risk factors, each one attributable to different genes interacting with environmental factors like diet. "Individually," Weiss said, "these factors heighten risk—but we can't say they cause disease by themselves."

Such traits have long been a "nightmare" for genetic researchers, Weiss said. "There's clearly something familial about them. They aggregate in families. But they don't behave in a nice Mendelian way."

Finding the "strong" signal for these traits, the genes involved, is not difficult. Over the last seven years, for example, researchers studying an inherited type of deafness have identified 75 genes that relate to the condition. "But each of these genes," Weiss said, "will be complex in ways resembling the single genes involved in simple traits." One deafness-associated gene, EYA1, contains 23 known mutations; another, PAX3, has 44 and counting. And so on down the line. "You can get non-syndromic hearing loss by inheriting any of these mutations in any of these genes."

Quantitative traits like cardiovascular disease, which are affected by levels of blood pressure, cholesterol, and numerous other factors, are even harder to pin down. Each contributing factor has its own set of responsible genes, and each gene has its own group of variables.

"Evolution generates heterogeneity," Weiss noted at last, "and evolution works by phenotype, not genotype." There are lots of those neutral mutations, in other words, that don't have a significant enough effect on the organism to be "detected" by the process of natural selection. What that means is that selection only indirectly filters variability. "It sees only the organism." And even at the organism level, natural selection is tolerant of variation. "Not just the fittest survive."

All this variation is what has allowed the human species to survive, Weiss noted, but it also makes the fine print of inheritance very difficult to read. Where we tend to get into trouble, he said, is when we try to predict complex traits in individuals from their genotypes—"especially traits that only arise after 50 or more years of successful living." There are simply too many intervening factors. "We understand ultimate cause a lot better than we do proximate cause. In other words, we know many things about how genes function, but predicting individuals' life experience from genetic variation is much more difficult. Here's a sobering thought: Even in identical twins—people we know to have the same genotype—there is sometimes only a modest correlation in disease experience."

What the existence of all that variation means in terms of public health, he argued, is that "screening windows may be preferable to screening genes." Expensive genetic approaches to disease, in other words, will probably continue to be less effective than simple environmental measures like the use of window screens to keep out disease-carrying mosquitoes, or improvements in water quality or public hygiene. Similarly, "If we really want to control cardiovascular disease, the answer is not genetic engineering. The answer is eating less."

Yes, there are exceptions: those severe genetic diseases like PKU that are as traceable as the colors of Mendel's peas. "The exceptions we do very well with," Weiss acknowledged. "But we shouldn't continue thinking that the exception is the norm. Rather, we should develop better methods for understanding the complex interactions among genes and environments that lead to disease."

Kenneth M. Weiss, Ph.D., is Evan Pugh professor of biological anthropology and genetics in the College of the Liberal Arts, 523 Carpenter Bldg., University Park, PA 16802; 814-865-0989;

Return to the "Decoding Life's Instruction Book" index page
or select another article in the series listed below.

Last Updated September 01, 2001