Why what your grandmother ate while pregnant with your mother might affect
Jose M. Ordovas has been studying the role of lipoproteins in heart disease for decades. His laboratory and others have tried to tease out how these proteins factor into why some people can eat an unhealthy diet - that is, lots of dietary fat - and still have high levels of what is often referred to as good high-density lipoprotein (HDL) cholesterol. The senior scientist and director of the Nutrition and Genomics Laboratory at Tufts University in Boston honed in on APOA1, which encodes the HDL component apolipoprotein (apo) A-I. A specific SNP in its promoter region (APOA1 - 75G/A) was first identified in the early 1980s, and studies in the decade following scrutinized the association between the G and A alleles and HDL concentrations. The results varied widely. Some studies suggested that carriers of the A allele (about 25% of the population) had higher HDL levels than carriers of the more common G allele, but other studies came to the exact opposite conclusion.
In 2002, he and his colleagues tested whether dietary fat might modulate the effect of the allele. "We decided to consider whether APOA1 is regulated by nutrients, since people are eating different things," Ordovas recalls. They looked at 755 men and 822 women who were participants in the Framingham Offspring Study, a population for which there are rigorous data on HDL levels, other cardiovascular risk factors, and dietary fat intake. They paired this information with genotype data for each patient and found that the polymorphism on its own didn't have an effect on HDL level. Instead, in people heterozygous or homozygous for the A allele, "what we found is that polyunsaturated fatty acids, which are very good regulators of gene expression, happen to modulate the expression of this genotype," Ordovas says.
It was the sort of finding that laid the foundation for the nascent field of nutrigenomics. At its core, the field is the study of how genes and nutrients interact to promote health or disease. It also includes understanding how gene and protein expression are affected by the presence or absence of specific nutrients, whether and how diet-regulated genes play a role in disease, the degree to which an individual's diet affects the risk of disease given his or her genetics, and whether one's diet may be altered to maintain that balance between health and disease.
"The important thing about nutritional genomics is that it tells us that we're not slaves to our genes and that we're not victims of genetic determination," says Raymond Rodriguez, director of the Center for Excellence in Nutritional Genomics at the University of California at Davis. "We ate ourselves into a disease state, and we can eat ourselves out of that disease state."
For some of the experts in the field, a June meeting on nutrigenomics organized by the US Institute of Medicine (IOM) gave the field a much-needed boost. "It's an area that has a lot of potential, but it's also an emerging area," says Patrick J. Stover, professor and director of the Division of Nutritional Sciences at Cornell University, and a member of the planning committee for the IOM meeting, which was sponsored by the National Cancer Institute, NIH's Office of Dietary Supplements, and the USDA's Agricultural Research Service. "There's not a lot of definitive results ... but certainly an area with huge, huge potential because diet is the key component of all metabolic disease."
Some small companies, such as Genelex in Seattle and Sciona in Boulder, Colo., have taken what few definitive results are available and developed and marketed home kits in the name of nutrigenomics. Consumers send away for the kit, return a cheek swab, questionnaire, and other personal details, and the company analyzes the DNA for genes and gene mutations associated with, for example, heart or bone health or inflammation. For a cost of up to hundreds of dollars, consumers receive a report of any specific mutations related to those conditions and may also get tailored dietary recommendations.
These tests recently came under fire in a US Government Accountability Office (GAO) investigation. Officials there purchased 14 tests from four different companies and developed fictitious profiles to submit for each test. "The results we received from all the tests we purchased mislead the consumer by making health-related predictions that are medically unproven and so ambiguous that they do not provide meaningful information to consumers," said Gregory Kutz, managing director of Forensic Audits and Special Investigations at GAO, when addressing the Senate Special Committee on Aging in a July hearing on the issue.
For some researchers, including John Milner, these tests are still in their infancy. "I think it's the right approach; however, I think it may be premature at this point," says Milner, chief of the Nutritional Science Research Group at the National Cancer Institute, and longtime supporter of nutrigenomics. "What has to happen is we have to define subsets of populations, and we ought to be talking about a response that you really care about."
Even with tests that are much more thorough than the ones those companies offer, treating a complex condition such as heart disease, for example, with nutrition alone would be difficult. Salt sensitivity, which affects about 25% of the population, should be easier to identify and treat with a simple dietary recommendation, Milner says. "I think we're going to head down those types of paths before we get into disease states," he says.
Recommended intakes for specific micro- and macronutrients such as those in the USDA's food pyramid are based on what's sufficient for most people. "Those look at what's generally adequate for the majority of the population," Milner says. "That's just to prevent some classic symptom that's associated with deficiency." Researchers hope that the fruits of more intensive nutrigenomics research will allow for more specific nutrition recommendations that would actually work to promote health for smaller subgroups of the population.
"The bottom line is that we know that there's a lot of individual variability in response to specific nutrients, and to tell people that everyone is alike is pretty naìve," Milner says. "We need to understand in more detail the individual and how he or she is going to respond to foods or food components. We went down a path for a long time where nutrition was the basic food groups, and if you knew that, you didn't need to know anything else. I think we're now realizing that's a pretty simplistic approach."
In many ways, nutrigenomics is most closely related to pharmacogenomics, which aims to understand the correlation between a person's genes and his or her response to a specific drug. There are some fundamental differences: Nutrients are a life-long exposure; amounts of specific nutrients from the diet are nearly impossible to measure and control; and nutrients are rather promiscuous in their targets. "Every compound that passes through our mouths changes gene expression," notes Stephen Barnes, director of the University of Alabama-Birmingham Center for Nutrient-Gene Interaction. "There is nothing peculiar about nutrition that's different from any pharmaceutical we take. They're all chemical compounds."
In one sense, the concept of nutrigenomics is already applied in modern medicine. Consider phenylketonuria, for example. Infants with mutations in the PAH gene, which leads to impaired metabolism of phenylalanine, are fed a low- or no-phenylalanine diet for much of their childhood and usually into their adult life. Many other genes for simple metabolic disorders are tested in standard newborn screening assessments.
Most diseases, however, are much more complex. Advancing nutrigenomics for these diseases, and for overall health, will require dedicated, focused studies in genetics and epigenetics, as well as increased understanding of how genes, proteins, and epigenetic changes interact within networks and pathways. The field will also draw on knowledge and technology from some of its "omic" cousins, even though those fields are still in various stages of maturation. For nutrigenomics to be successful, though, it will have to overcome obstacles that are familiar to its related fields: a lack of dedicated funding and a need for collaboration. "Nutritionists and geneticists can [independently] solve very specific problems, but when it comes to really helping the majority of the population, they will really have to work together," Ordovas says.
An oft-cited observation is that the review articles of nutrigenomics outnumber its original research publications. But for some, that fact simply shows that there's a strong base of research that, combined with recent technological advances, provides the rationale for new (see "NuGO: A Vision for Nutrigenomic Collaboration") dedicated, coordinated efforts in the field. "We're all self-educating about what we need to do in nutrition and genetics to do better and more complete science, and that's terribly exciting," says Jim Kaput, of the Laboratory of Nutrigenomic Medicine at the University of Illinois in Chicago.
The approaches to nutrigenomics are diverse. One of the core components of nutrigenomics is nutrigenetics, which focuses on how genetic variation affects the interaction between diet and disease. For example, Steven H. Zeisel and colleagues at the University of North Carolina in Chapel Hill have studied how specific polymorphisms in genes that mediate folate metabolism can make people more or less susceptible to deficiencies in the nutrient choline. Another team of researchers from Roswell Park Cancer Institute in Buffalo, NY, has investigated the association between specific alleles in the gene for the enzyme myeloperoxidase and reduced breast cancer risk - but only among women who ate high amounts of fruits and vegetables.
Ordovas' group, which takes a classic nutrigenetics approach, has gone beyond APOA1 and has found polymorphisms in other genes that are modulated by dietary fat intake. For example, most people have a common allele in the gene for hepatic lipase, an enzyme that plays a role in regulating plasma lipids, and are able to regulate HDL and low-density lipoprotein (LDL, the "bad" cholesterol) levels in the presence of dietary fat. "Our lipid system maintains homeostasis by increasing HDL, so you maintain a kind of constant LDL to HDL ratio," Ordovas explains. But people who carry a rare allele in hepatic lipase have trouble maintaining this balance, and when they consume more fat, their HDL goes down nd the LDL still goes up. "You get into a very dangerous situation when these people consume a high-fat diet," Ordovas says.
These studies explain in part why two people with similar diets can have radically different cholesterol levels, and therefore radically different risks for cardiovascular disease. Ordovas points out that dietary fat recommendations are made to accommodate the majority of the population. However, smaller groups with specific genetic variations may benefit from increasing their polyunsaturated fat intake, or should avoid high-fat foods altogether.
Another key area of nutrigenomics will be epigenetics, which studies alterations in gene expression by chemical modifications such as methylation or acetylation. A simple and now classic experiment emphasized the importance of nutrition in epigenetic changes. George L. Wolff, of the National Center for Toxicological Research at the FDA in Jefferson, Ark., and colleagues fed sets of pregnant viable yellow agouti (Avy) mice diets that varied in methyl content. The agouti gene, which is responsible for yellow coat color and is also associated with obesity, is dysregulated in these mice, causing varying expression of the gene as seen by coat color variations, even within a single litter of pups.
Female mice that were fed the methyl donor diets, which were rich in folic acid, vitamin B12, betaine, and choline, tended to have offspring that were more often lean and brown, rather than having the obese and yellow phenotypes seen when the agouti gene is expressed. In 2003, Robert A. Waterland, now an assistant professor at Baylor College of Medicine in Houston, and Randy L. Jirtle, of Duke University in Durham, NC, repeated the experiment to look not only at coat color but also at methylation status. "We were able to show quite clearly that the effect of maternal diet on coat color of the offspring had occurred by increasing the level of methylation at the Avy locus," Waterland says.
Earlier this year, Duke colleagues Dana C. Dolinoy and Jennifer R. Wiedman, along with Waterland and Jirtle, showed that feeding pregnant Avy mice a diet supplemented with 250 mg/kg of genistein (a phytoestrogen found in soy) produced offspring that had phenotypic changes similar to those seen in previous experiments with the methyl donor diet. Waterland and his colleagues have also conducted a similar experiment with axin-fused mice, a mutation that leads to kinky tails; this phenotype can vary within a single litter of pups. Once again, they found that the methyl-donor diet produced a dramatic decrease in the kinked-tail defect.
Although it's unlikely that researchers will find a kinky-tail phenotype in humans, these experiments suggest that genes that have this interindividual variability in methylation status, called metastable epialleles, "represent really good candidates where we might find nutritional influences on epigenetic regulation in humans," Waterland explains. The gene that encodes insulin-like growth factor 2 (IGF2) is one such candidate in humans, he says. Most people express only the paternally inherited copy of IGF2; however, about 10% of normal human adults also express the maternal allele. "This IGF2 loss of imprinting, which appears to predispose individuals to colorectal cancer, reflects interindividual epigenetic variability reminiscent of a metastable epiallele," Waterland says.
The experiments also show that prenatal and early postnatal development are critical windows in which epigenetic regulation is becoming established, and nutrition during those times is critical. Once these epigenetic mechanisms are set during development, they are generally maintained throughout life, Waterland explains. "It's becoming quite clear that what your grandmother ate during her pregnancy with your mother could be affecting phenotypes that you might be exhibiting or that your children might exhibit, which is profoundly important in terms of human health and public health," says Paul Soloway, professor in the Division of Nutritional Sciences at Cornell University.
Much of the early work in nutrigenetics and epigenetics has used the so-called candidate-gene approach, interrogating known genes to search for variants that may confer a risk or protection for a specific disease. "We are taking the candidate-gene approach because that's what we've been doing for the last 25 years," says Ordovas, referring to work that he and his colleagues have underway on metabolic syndrome and its associated risk factors.
However, "by doing the candidate-gene approach, we have probably done as much as we can in terms of defining the puzzle," notes Ordovas. "So now the next stage is to question the whole genome beyond what we already know and see what information we get back." To that end, he and his colleagues are now testing 250,000 different markers to mine for risk factors for metabolic syndrome, and will increase the number of markers to 500,000 next year.
Soloway is using recombinant inbred mice to take a unique, genome-wide look for regions that control diet-dependent effects in cancer formation. He and his colleagues bred several different panels of recombinant inbred mice and fed them different, controlled diets. They then induced colon cancer in all of them, determined which strains of animals had diet-dependent responses to cancer, and mapped the regions of the genome that control those diet-dependent effects.
"With that information in hand, we can begin to explore ... what genes are within those intervals that we've mapped, and hopefully identify exactly which alleles are responsible for these diet-dependent effects," Soloway explains. So far, he and his colleagues have completed these experiments on 450 mice and are now analyzing the data.
Similar new technology will allow Waterland to expand his studies to look across the entire genome for epigenetic alterations associated with obesity in Avy mice. Mice that express the agouti gene also tend to be obese because the agouti protein binds to the melanocortin-4 receptor in the hypothalamus, impairing the satiety mechanism. "These animals just love to sit around and eat," Waterland explains.
He and his colleagues are taking wild-type offspring born to obese Avy mothers and wild-type offspring born to lean wild-type moms and comparing them across the entire epigenome to look at the methylation patterns of thousands of genes simultaneously. "That type of tool allows us to make really rapid progress in understanding how maternal obesity, for example, might cause persistent dysregulation of genes in the offspring," Waterland says.
Improvements in genomic technology and a dramatic reduction in the cost of doing whole-genome analyses will be a big help in nutrigenomics studies for comparing how subtle changes in diet affect gene expression. "Just a couple of years ago, what we're talking about doing wasn't even possible, so that shows you the rate that things are moving in this field," Waterland says.
In addition to nutrigenetics and epigenetics, nutrigenomics will also incorporate transcriptomics, the study of the effects of nutrients on gene expression. For example, the University of Alabama's Barnes and his colleagues are studying the role diet may play in gene expression and in protein abundance in the mammary gland at the time of puberty, and how those changes may affect cancer risk later in life.
To handle this abundance of data will take even further advances in bioinformatics, and so systems biology will factor into nutrigenomics as well. "By looking at variations in transcription factors and changes in expression by arrays and beginning to assemble all the 25,000 genes into networks, we'll be able to look at how each individual gene impacts another gene and then [be] able to model how the variants of that gene affect [an] entire network," says Cornell's Stover. "This is essentially putting all the parts of the engine together."
Some early efforts have begun to organize research projects. The unique nutritional genomics center at UC-Davis was established to study racial and ethnic disparities resulting from nutrient-gene interactions and to devise interventions to reduce disease in those populations. Funded by the National Center on Minority Health and Health Disparities at the NIH, the genomics center serves as a hub to organize research projects and as a funding organization. Center director Rodriguez cites two studies as examples of the center's mission: One is a nutritional intervention for African-American women who are nursing and are vitamin-D deficient; the other is a global genetic study of the phenotypic effects of caloric restriction.
In Europe, the European Commission has taken an interest in nutrigenomics and created the European Nutrigenomics Organization (NuGO), which was founded in January 2004 and is funded by the EC through 2009.
For Ordovas, who has been studying nutrigenomics for 25 years, the field still
is just beginning. The new endeavors will set the stage for coordinated, directed
efforts that will ultimately lead to a broader understanding of the intricate
balance between nutrients and genes. "In terms of the potential power
of this knowledge, we are now where we were in 1980 with computers. You could
maybe play ping-pong on the screen using primitive computers, and look where
we are now," he says. "Remember that, many times in history, we
said we had reached the maximum of what we could achieve, only to discover
that, no, that is not true. That is demonstrated every day in nutrigenomics."
2. D.C. Dolinoy et al., "Maternal genistein alters coat color and protects Avy mouse offspring from obesity by modifying the fetal epigenome," Environ Health Perspect, 114:567-72, 2006.
3. T.G. Whitsett et al., "Resveratrol, but not EGCG, in the diet suppresses DMBA-induced mammary cancer in rats," J Carcinogen, 5:15, 2006.
4. P.C. Adams et al., "Hemochromatosis and iron-overload screening in a racially diverse population," N Engl J Med, 352:1769-78, 2005.
5. T. Bersaglieri et al., "Genetic signatures of strong recent positive selection at the lactase gene," Am J Hum Genet, 74:1111-20, 2004.
6. J. Ma et al., "Methylenetetrahydrofolate reductase polymorphism, dietary interactions, and risk of colorectal cancer," Cancer Res, 57:1098-102, 1997.
7. H. Sanada et al., "Single-nucleotide polymorphisms for diagnosis of salt-sensitive hypertension," Clin Chem, 52:352-60, 2006.