NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division; Food and Nutrition Board; Food Forum. Nutrigenomics and the Future of Nutrition: Proceedings of a Workshop. Washington (DC): National Academies Press (US); 2018 Jul 25.

Cover of Nutrigenomics and the Future of Nutrition

Nutrigenomics and the Future of Nutrition: Proceedings of a Workshop.

Show details

2Nutrigenomics and Chronic Disease Endpoints

In session 1, moderated by Naomi Fukagawa, U.S. Department of Agriculture, speakers discussed the interrelationships among diet, genomics, and health or disease prevention. This chapter summarizes the first portion of the session, which included presentations by José Ordovás, Tufts University, and Douglas Wallace, Perelman Medical School, University of Pennsylvania. (The remainder of the session is summarized in Chapter 3.) Highlights from the presentations of Ordovás and Wallace are provided in Box 2-1.

Box Icon

BOX 2-1

Overview of Points Presented by Individual Speakers.


The genome contains more than 3 billion base pairs, Ordovás observed, in contrast to the epigenome's 30 million CpG dinucleotides1 in various states of methylation. Although the smaller size of the epigenome may make it appear easier to work with than the genome, he stated that it in fact poses a greater challenge. According to Ordovás, this is the case because unlike single nucleotide polymorphisms (SNPs), which either do or do not exist across all cells in an organism, the epigenome changes over time and across organs and cell types. “So we have something much more difficult to deal with in terms of using the epigenome as part of this task of nutrigenomics, or nutrigenetics,” he said.

Moreover, Ordovás continued, scientists have studied the genome more than they have the epigenome. The SNP database2 now contains more than 300 million of what he called “needles in this haystack.” And there has been enough research in this area, he added, to know that the genome can indicate what people can eat, as well as what people want to eat. But there has also been enough research to know the complexity of the road ahead, he cautioned. Thus, to provide some context for his discussion of the epigenome in relation to nutrition, he began by speaking briefly of the genome in relation to nutrition.

Nutrition and the Genome

The root of personalized medicine and nutrition, Ordovás said, is newborn screening, which he considers the simplest example of genomic screening for specific personalized treatments. In the United States, he reported, about 12,000 of the 4.2 million babies born each year are born with a monogenic disease—a disease that if not detected in time, usually at birth, can mean death or a life with severe disabilities. One of the most common such diseases is congenital hypothyroidism (CH), which is detected in more than 1,000 infants annually in the United States. According to Ordovás, the approximate cost of screening for CH is $20 million, compared with $400 million in benefits (i.e., later costs avoided by having diagnosed and treated the disease).3 “So the benefit [of genetic screening] is obvious—it's 20 times the cost,” he said. Regardless of the economic costs and benefits, he added, “what we know is that, based on genotype, individuals need to receive specific therapies and either can eat or cannot eat certain foods.”

Ordovás acknowledged that CH is an extreme case. As a broader example of how the genome and health influence one another, he pointed to past positive selection as a significant driver of nutrition-related genetic variation. He cited the lactase persistent gene as the classic example of this phenomenon, with different ethnic groups having more or less prevalence of the gene depending on past access to dairy products. The same is true of the fat-related APOE gene, he noted, with respect to its prevalence among groups with a hunter-gatherer versus agricultural history, as well as of alcohol-related ADH1B. All three examples, he observed, illustrate that what individuals from different cultures eat is determined partly by how their genomes have been responding to selective pressures in their nutritional environments.

Ordovás went on to explain that past exposure to different nutrition-related environments has impacted genetic variation in taste and food preference as well (e.g., Chmurzynska and Mlodzike, 2017). To illustrate this point, he described what scientists have been learning about APOA2, a gene that is expressed primarily in the liver and that produces APOA2, a protein present in high-density lipoprotein (HDL). He noted that, although scientists have known about this protein and its abundance in plasma for some 40 years, only when the genetic work began did they start to see some of what it actually does. He pointed to one of the initial findings from the Genetics of Lipid Lowering Drugs and Diet Network (GOLDN) nutri-pharmacogenomic study, which enrolled more than 1,000 people. The researchers found that APOA2 has a common polymorphism in the promoter region of the gene (APOA2 m265T>C), and individuals who are homozygous for the C allele, which is the less common of the two alleles, eat more and weigh more relative to individuals with the TT and TC genotypes (Corella et al., 2007). He described the polymorphism as “an example of a genetic variant that predisposes you to obesity because it drives you to eat more of certain foods.”

In later GOLDN and other studies (Correla et al., 2009, 2011; Delgado et al., 2007; Smith et al., 2012), Ordovás and his team replicated a gene–diet interaction between the APOA2 polymorphism and saturated fat and found that under low saturated fat conditions, one's genotype does not matter; body mass index (BMI) is the same. It is only when the physiology, or genome, is stressed by a diet high in saturated fat that individuals with the CC genotype have higher BMIs, whereas for individuals with the TT or TC genotype, the amount of saturated fat in the diet does not matter for their BMI. In the past, Ordovás added, lack of validation, or replication, has been a problem with gene–environment interaction studies. The fact that this same pattern has been seen in six independent populations and five ethnicities across the world indicates, he said, that “this is a polymorphism that may have a significant impact in terms of personalized recommendations.” However, he again cautioned that this is only one piece of the complex nutrigenomics puzzle, which he predicted probably will not be completed any time soon. “But at least we have a better idea of where the pieces fit together,” he said.

Epigenetics in Nutrition Research

Ordovás reiterated that, compared with genetics, epigenetics is more difficult to study with respect to its importance in nutrition because it is what he described as a “moving target,” with different organs, different tissues, and different cells having different epigenetic patterns. Moreover, he explained, because the study of epigenetics is restricted primarily to those cells researchers can access, much of the progress in understanding epigenetics in relation to nutrition has been made in experimental animal models. That said, he added, evidence in humans is beginning to emerge.

As an example of this evidence, Ordovás highlighted what is being learned about the famous Dutch Hunger Winter of 1944–1945 in the Netherlands, when people had to survive on rations as low as 400–800 calories per day. An estimated 22,000 people died. But those deaths were not the end of the story, Ordovás noted. He explained that investigators studied individuals who were still in the fetal state during that time and found, for example, that their birthweight had been different relative to those not exposed to the Hunger Winter. Moreover, individuals exposed to the famine, particularly those exposed during middle and late gestation, had impaired glucose tolerance even at age 60 years (e.g., de Rooij et al., 2006a,b; Ravelli et al., 1998). According to Ordovás, researchers have found other differences as well, such as problems with obesity and neurological disorders among those exposed to the Hunger Winter during fetal development, and highly significant differences in the epigenetic profiles of individuals who were and were not born during the Hunger Winter (e.g., Tobi et al., 2014).

Ordovás then turned from the Dutch Hunger Winter, a set point in history, to parts of the world that experience yearly seasonalities, reporting that investigators have found similar changes in the DNA methylation patterns depending on the season of conception. In The Gambia, for example, there are two seasons—dry and wet. Waterland and colleagues (2010) found that individuals born during the wet, or harvest, season have different DNA methylation patterns from those of individuals born during the dry, or hungry, season. Ordovás remarked that these differences may play a significant role in survival and disease.

As another example of nutrition-related epigenetic differences, Ordovás cited whole fruits versus fruit juices. He observed that whole fruits are typically presented as being rich in fiber and low in energy and containing intact vitamins and minerals, whereas fruit juices are portrayed as being deficient in fiber and high in energy because of added sugar, and lacking in most vitamins and minerals found in whole fruits (although in some cases, he noted, the lost vitamins and minerals are replenished). But, he asked, how true are these characterizations? Using data from the Framingham study, Nicodemus-Johnson and Sinnott (2017) examined the epigenomes of individuals who consumed whole fruits and those who consumed fruit juices and found significant differences in the epigenetic signatures associated with pathways involved in the immune system. More specifically, the whole fruit–specific epigenetic signature was enriched for adaptive immune system genes but not innate immune system genes, and for genes involved in telomere maintenance and other aging-related pathways. In contrast, the juice-specific epigenetic signature was enriched for both adaptive and immune system genes, so for genes involved in proinflammatory pathways. Based on these findings, Ordovás said, “it's better to eat fruits, thanks to the epigenome.”

The Integration of Genetics and Epigenetics in Nutrition Research

Ordovás observed that methylation, on which he had been focusing during his presentation to this point, is only a small fraction of the complex epigenome. That said, he continued, methylation is dependent on two factors within the context of nutrition: (1) what one eats and (2) what one's genome is. Regarding the latter, he explained that any time a gene variant removes either the C or the G from a CpG site, that site will no longer be methylated. Conversely, if a variant introduces a new CpG site, methylation will emerge in a place where it did not exist before. So it is not only the environment, or diet, that affects methylation, Ordovás stated, but also the genome itself (Zhi et al., 2013).

To illustrate, Ordovás described how ABCA1, which is involved with HDL metabolism, is a common genetic variant that sits atop a differentially methylated 5′ CpG region of the genome. By sitting there, it can not only change the methylation of that CpG but also affect the methylation of all the other neighboring CpGs. Ordovás described the results of a meta-analysis of methylation data from more than 10,000 individuals from different cohorts showing that a change in EPA4 in the diet resulted in either decreased methylation and increased HDL cholesterol or increased methylation and decreased HDL, depending on the ABCA1 genotype (Ma et al., 2016). More specifically, with a 1 percent increase in EPA, individuals with the CC genotype experienced decreased methylation, increased gene expression, and increased HDL cholesterol, in contrast to the increased methylation, reduced gene expression, and lowered HDL cholesterol experienced by individuals with the GG genotype under the same conditions. (Ordovás noted that decreased methylation is usually associated with increased gene expression.) Thus, he said, this is a case in which both genetics and epigenetics need to be considered. Otherwise, he stated, a general recommendation to increase one's EPA levels would have a positive effect in some individuals with respect to HDL cholesterol but a negative effect in others.

As another example of a nutrition-related gene–epigene interaction, Ordovás cited perilipin, a protein that surrounds lipid droplets in adipocytes. Many studies have shown, he reported, that variation in the perilipin genes affects obesity risk. His research team found that one of these genes, PERILIPIN4 (PLIN4), is not in the promoter region of the gene, as are the other polymorphisms, but in the 3′ region of the gene (Richardson et al., 2011). It turns out, he explained, that microRNAs (miRNAs), which are another type of epigenetic phenomenon, can bind to the 3′ region and decrease expression. This particular polymorphism is at a site where there normally is no bound miRNA, but when a G is replaced with an A, miRNA suddenly binds and decreases the expression of PLIN4. The question, Ordovás suggested, is whether something can be done about this. In fact, his team has observed that when individuals with the less common allele—that is, with bound miRNA at that site—are placed on a diet rich in omega-3s, their phenotype reverts from that of the less common allele to that of the more common allele. So again, he said, this is an example of gene–diet interaction that works through a combination of genomics and epigenomics. He noted that there are thousands of miRNAs, many of which scientists know very little about, although they are beginning to make progress, especially in terms of understanding miRNAs in relation to cancer.

The Future

In closing, Ordovás remarked that the microbiome plays a role in nutrition and that personalized nutrition will require not just combining genomics and epigenomics but also considering the microbiome, as well as the metabolome. “I don't know that we'll ever get to perfect,” he observed, meaning perfect personalized nutrition. However, quoting Voltaire,5 who in turn borrowed from an old Italian proverb, Ordovás said, “perfection is the enemy of good.” There is enough evidence now, in his opinion, to begin putting the pieces of the puzzle together and to control some of what he described as the “snake oil” being sold by some consumer ventures.


“Why can't we understand or cure the common metabolic and degenerative diseases?” Wallace asked, noting that there is an enormous number of such diseases, including neurodegenerative and neuropsychiatric disorders (e.g., autism, Alzheimer's disease), heart and muscle diseases (e.g., cardiomyopathy, chronic fatigue), visceral diseases (e.g., renal and hepatic diseases), metabolic diseases (e.g., diabetes, obesity, cardiovascular disease), cancer, and aging. He suspects that perhaps the problem is not the effort expended, given the trillions of dollars that have been spent trying to understand human disease, but the basic assumptions the scientific community is making in addressing the problem.

According to Wallace, the first basic assumption upon which Western medicine is premised is that disease is organ based, an assumption that goes back to Andreas Vesalius.6 For example, if one has a headache, one is referred to a neurologist because the assumption is that a symptom in the head means there is something wrong with the head. “I call that the anatomical paradigm of disease,” Wallace said. The second basic assumption, he continued, goes back to Gregor Mendel: if a trait is inherited according to “the laws of Mendel,” then it is genetic, and if it is not inherited, then it must be environmental. He remarked that these ideas are the framework with which all medical students are taught and all basic scientists design their experiments.

According to Wallace, however, these ideas may not be sufficient. While it is true that anatomy is encoded by Mendelian genes and that there are specific disease mutations that gave rise to the newborn screening program, he observed, this knowledge does not appear to be helping with the common complex diseases. “It is not enough to have an anatomy,” he said. “One has to be animated.” He mentioned Newton's work and the fact that mass does not change unless it is acted upon by energy. “Therefore,” he asserted, “if we are going to think about being alive, we have to think not only about anatomy, but also energy.” Furthermore, he argued, one must think not only about the information for anatomy, but also the information for energy, and when one begins to think this way, medicine becomes not just about anatomy but also about energetics. He believes bioenergetic dysfunction lies at the nexus of the genetic and environmental “causes” of the common complex diseases.

The Dichotomy Between Anatomy and Energy

Wallace then turned to how the dichotomy between anatomy and energy arose from the origin of the very cells that make up the human body, that is, the eukaryotic cells, which in turn arose from the symbiosis between Archaebacteria and the oxidative Alphaproteobacteria. It was the Archaebacteria that ultimately gave rise to the nuclear cytosol and the Alphaproteobacteria that gave rise to the mitochondria.

These different bacteria, Wallace continued, each had—and still have—their own information storage and retrieval systems. The nucleus has DNA, which is transcribed into RNA, which in turn is translated in the cytosolic ribosomes, producing about 20,000 to 25,000 proteins. An estimated 1,000 to 2,000 of these proteins make up the anatomy of the mitochondria. But the mitochondria have retained their own DNA as well, Wallace emphasized. Their DNA is replicated and transcribed in the mitochondria, and the messenger RNA is translated on mitochondria ribosomes in a bacteria-like way. Not only is mitochondrial translation initiated by N-formyl methionine, just as bacterial proteins are, but mitochondria ribosomes also are sensitive to chloramphenicol and aminoglycosides, just as bacterial ribosomes are.

Over evolutionary time, Wallace explained, the two originally co-equal genomes began to specialize, with the nuclear genome specializing in creating the anatomy not only of the cell but also of the mitochondria, and the mitochondrial genome specializing in energy and developing what he described as “the wiring diagram for the power plant.” More specifically, the mitochondrial DNA evolved to code for critical proteins involved in the energy process called oxidative phosphorylation, including 7 of the 45 respiratory complex I proteins, 1 of the 11 complex III proteins, 3 of the 13 complex IV proteins, and 2 of the 15 complex V proteins. Wallace explained that each of these complexes is a system by which the cell takes the nutrients in the diet and the oxygen that is being breathed and converts them into potential energy. That energy, he said, “you then use for everything that you want to do.”

With respect to how these complexes are involved in oxidative phosphorylation, Wallace explained that glucose goes through glycolysis to produce pyruvate, which can then be either reduced to produce lactate or amino-grouped to produce alanine. Alternatively, the pyruvate can be transported into the mitochondria through the pyruvate transporter, where it is converted by pyruvate dehydrogenase to make acetyl coA, driving the tricarboxylic acid (TCA) cycle. The purpose of the TCA cycle, Wallace continued, is to strip hydrogens off hydrocarbons and put them on the carrier nicotinamide adenine dinucleotide (NAD), creating NADH. The two hydrogens of NADH are then burned by what is called the electron transport chain, which is made up of complexes I, III, and IV, as well as coenzyme Q and cytochrome C. As they flow down this chain, the two electrons (of the two hydrogens) react with oxygen to produce water. The energy that is released is not just dissipated, Wallace clarified, but is used to create what is essentially a capacitor. As the electrons flow through the complexes (I, III, and IV), they pump protons from inside the mitochondrial matrix, across the mitochondrial inner membrane, and into the intermembrane space, thereby creating a positive, acid exterior and a negative, alkaline interior. This membrane potential can be used for many functions, Wallace added, one of which is to make adenosine triphosphate (ATP): protons flow through a proton channel in complex V (i.e., ATP synthase) to condense adenosine diphosphate (ADP) and phosphate and make ATP. This coupling of oxidation and phosphorylation is oxidative phosphorylation. The membrane potential can also be used in other ways, Wallace noted—for example, to regulate a positive cation, such as calcium, by electrophoresing into the negatively charged mitochondrial matrix.

“So sitting in your chair right now,” Wallace said, there are 100 trillion cells, each cell with about 1,000 mitochondrial bacteria—about 30 percent of one's body weight—and each mitochondrion having a potential across its membrane of about 0.2 volts. “So the total energetics in your body right now is the equivalent of a lightning bolt,” he said, “and that is the energy for everything that you do every day of your life . . . so this flow of energy is absolutely critical.”

However, like any furnace, Wallace continued, mitochondria also make smoke—the reactive oxidant species that form when not fully oxidized electrons bind with oxygen (O2) to produce hydrogen peroxide. If the hydrogen peroxide is not reduced to water (i.e., by nicotinamide nucleoside transhydrogenase), then another electron, provided by a transition metal, will bind with it to produce a hydroxyl radical, a reactive oxygen species and a potent damaging agent. Wallace noted that some people take vitamin C, vitamin E, beta carotene, or coenzyme Q (CoQ) to prevent this from happening.

The mitochondria also have a self-destruct system (i.e., apoptosis). According to Wallace, there is active debate around what the structure of this system is, but its job normally is to maintain a closed door. He explained that when the membrane potential becomes low, high-energy phosphates decline, oxidative stress becomes excessive, or calcium load occurs, the self-destruct system senses these changes and ultimately pops into an open channel that short-circuits the membrane potential. As a result, fluids flow in, the inner membrane swells, and pro-apoptotic proteins flow out and degrade the cell from the inside out. Without enough energy, this self-destruct system fails. If bacteria are released into the bloodstream with all of their bacterial antigens, the result will be inflammation, which is believed to accompany all the metabolic and degenerative diseases.

In summary, Wallace said, the mitochondria generate most of the body's energy; regulate the redox state of the cells; make reactive oxygen species, which are signaling molecules but at high levels are toxic; regulate calcium; regulate apoptosis; and generate all the intermediates for regulating the epigenome (e.g., ATP, acetyl CoA).

Wallace added that different tissues have different energetic demands. For example, the brain is about 3 percent of body weight but uses about 20 percent of all mitochondrial energy. So a 10 percent reduction in mitochondrial energy, Wallace said, “is going to give you a very bad headache.” The headache occurs not because the brain has altered, he clarified, but because there is a systemic defect, and the brain is specifically affected. The hierarchy of energetics, he explained, is brain, heart, muscle, renal, endocrine, and liver, which are the organs affected in all the common, complex diseases.

Mitochondrial Inheritance

The 13 proteins retained by the mitochondria that make up what Wallace described as the “electron and proton wires of the wiring diagram” must co-evolve, he argued, because if any one were to become leaky for protons, it would “short the capacitor.” This is the case, he elaborated, because the wiring diagram is an integrated circuit, with all of the enzymes in the system relying on the same substrate, that is, the membrane potential. But how could such co-evolution occur, he asked, since nuclear genes undergo recombination? Thus, for example, if there were a polymorphism in complex I but not in complexes III and IV, recombination would be mixing and matching coupled and uncoupled variation.

To explain coupling, Wallace described someone who burns the least number of calories for the maximum amount of ATP as someone who is very good at taking in calories, burning them, making that membrane potential, and then converting the membrane potential to ATP. In other words, this person's coupling system is tight. Because a calorie is a unit of heat, Wallace continued, that person also generates less heat. But, he added, someone who is less efficient at pumping protons out or converting protons into ATP must burn more calories for the same amount of ATP, plus that person generates more heat. In other words, this person's coupling system is loose. “Your mitochondria is regulating your thermal and your energy balance based on the coupling efficiency,” Wallace said.

The wiring diagram of the mitochondria is inherited only from women, Wallace continued, thus ensuring that there will never be recombination. That is, mitochondrial DNA is transmitted from a mother to all of her children and from her daughters to the daughters' children, and so on, while when a male's mitochondria enter the egg, they are perceived as foreign and selectively destroyed.

Mitochondrial Mutations and Heteroplasmy

Wallace went on to explain that mitochondria are constantly replicating inside the cell. They are also constantly being eaten by mycophagy, and thus are in what he described as a colony in steady state. Nonetheless, as they replicate, mutations accumulate. Wallace observed that, because mitochondria have very high mutation rates—one or two orders of magnitude greater than that of nuclear DNA—they are characterized by a tremendous amount of genetic variation. Moreover, he added, as mutations accumulate, they create mixed populations of mutant and normal mitochondria within single cells. If one of these mixed, so-called heteroplasmic cells were to divide down the middle, both new cells would have some mutant and some normal mitochondria, so they would both be heteroplasmic. But a cell could also divide such that one new cell would contain only normal mitochondria, the other containing twice as many mutant mitochondria. Thus, Wallace said, “the tissues in our bodies are a mosaic of different mitochondrial genotypes, with different tissues having different percentages of mutant and normal mitochondrial DNAs.” And as the number of mutant mitochondrial DNAs increases, he noted, energy output declines, eventually falling below the minimum energy for that organ and reaching what he characterized as the equivalent of an energetic disease.

Three Types of Mitochondrial Mutations

Wallace next described three categories of mitochondrial mutations. First are mutations that arise along the maternal lineage and give rise to maternally inherited diseases. For example, if an individual inherits a mutation in the ND4 gene at nucleotide position 11778, he or she will be fine throughout midlife, but then will suddenly go blind in one eye and then in the other because of what is known as Leber's hereditary optic neuropathy. A mutation in the ND6 gene at 14484 causes the same Leber's blindness, and a mutation in ND6 at 14459 causes a more severe Leber's blindness when heteroplasmic and generalized dystonia when homoplamic (i.e., pure mutant). As another example, Wallace cited a mutation in the ATPase6 gene at 8993, which causes retinitis pigmentosa at 70 percent mutant, olivopontocerebellar atrophy at 85 percent mutant, and death as an infant with Leigh's syndrome at 90 percent mutant.

According to Wallace, there are also several (maternally) inherited protein synthesis mutations, including a mutation in the tRNA leucine gene at 3243, which causes diabetes at 20 percent mutant, cardiovascular disease at 50 percent, and lethality in childhood at 100 percent. A mutation in the tRNA glutamine gene accounts for about 3–5 percent of late-onset Alzheimer's and Parkinson's disease. And a mutation in the tRNA lysine gene causes myoclonic epilepsy. Added to these are many kinds of cancer mutations.

Wallace then described a second class of mitochondrial mutations—the ancient polymorphisms. For example, one variant in the ND1 gene is found in three-quarters of sub-Saharan Africans (“macrohaplogroup L”); another variant (H) is found in half of Europeans; and four variants (A, B, C, and D) arose in central Asia and then crossed the Bering land bridge, allowing people to colonize the Americas. “We believe these are markers for mitochondrial lineages that adapted human energy metabolism to live in different environments,” Wallace said.

Finally, Wallace explained, as people age, they accumulate somatic mutations. This, he said, is “the aging clock.”

The Mitochondrial Etiology of Complex Diseases: An Energetic Approach to Medicine

Wallace reiterated, “Once we begin to think energetically, then all the common diseases have the same etiology: a bioenergetic defect due to oxidative phosphorylation.” Nuclear mutations can affect this process, he noted, as can changes in the epigenome, as well as both ancient adaptive polymorphisms and recent deleterious mutations. Finally, he added, the calories and types of nutrients one ingests, or how one exercises and uses those nutrients, as well as whether one smokes or is exposed to other toxins, all impinge on energetics.

If energetics is affected, Wallace continued, then mitochondrial DNA damage accumulates over time, which leads to age-related decline and the delayed onset and progressive course of all the common diseases. Additionally, he explained, as the furnace is impaired, substrates (glucose, fatty acids, and cholesterol) build up, and that is what creates metabolic syndrome. When apoptosis fails, he noted, bacteria are released into the bloodstream, initiating all of the inflammatory processes that accompany the complex diseases. And finally, turning to cancer, he characterized it as “all about adjusting energy based on nutrients and oxygen. In fact, a bioenergetic way of looking at the disease takes us away from the anatomical approach and into an energetic approach to medicine.”

Studies of Mitochondrial Mutations

Wallace went on to describe results from studies of mitochondrial mutations, beginning with work he and his team did in the 1980s on a family in which the mother had lactic acidosis and growth retardation, and many of her children had lactic acidosis, growth retardation, progressive dementia, stroke-like episodes, hypertrophic cardiomyopathy, and cardiac conduction defects. Additionally, mitochondrial oxidative fibers degenerated in the muscle, although the glycolate muscle fibers were fine. All of these individuals had the same mutation in the tRNA dileucine gene. However, Wallace noted, different percentages of heteroplasmy in this same mutation can cause different phenotypes. Thus, he explained, individuals with greater than about 70 percent of the mutant mitochondria have myopathy, cardiomyopathy, and stroke-like episodes; individuals with 10–30 percent have autism and type 1 or type 2 diabetes; and individuals with 100 percent die as infants from Leigh syndrome.

Wallace and his team wanted to know whether a change in heteroplasmy in a cell could affect these different phenotypes, so starting with normal cells, or cells with zero percent heteroplasmy, they made what are known as cytoplasmic hybrids, or cybrids—cells with the same nucleus but different percentages of mutant mitochondrial DNA. They characterized the cybrids physiologically, then performed RNA sequencing and examined all the transcription factors that were regulated in the different cell lines. The patterns they observed indicated to Wallace that phenotype (e.g., the diabetes and autism associated with 10–30 percent heteroplasmy) is in fact determined by the mitochondrial signaling to the nucleus, which in turn determines gene expression and phenotype. “So, in fact,” he said, “the mitochondria is determining what the environmental challenges are and telling the nucleus how to respond, and that is telling the physician what he or she will see in the clinic.”

Wallace described another pedigree, which had a mutation in the ND6 gene—one of the complex I genes. A woman had 50 percent mutant mitochondria in her white blood cells, and she had optic atrophy and cerebellar ataxia (Malfatti et al., 2007). Her sister had 5 percent mutant mitochondria in her white blood cells and was perfectly normal, although every one of the sister's children was 100 percent mutant and died. Wallace interpreted this case as an indication of the rapidity of segregation that occurs along the germline and an explanation for why this class of disease was never understood with a Mendelian conceptual framework. This particular mutation, he noted, changes proline at codon 25 to a leucine (P25L).

Wallace described how he and his team took cultured mouse cells, mutagenized the mitochondrial DNA, then sequenced all the mutants that were respiratory deficient. They found an array of different mitochondrial DNA mutations, one of which was exactly the same mutation as that observed in the human phenotype described above. Another was a cytochrome oxidase mutation that changes valine at codon 421 to alanine (V421A). The researchers then created pluripotent cytoplasmic hybrids by, first, taking a cell with mutant mitochondrial DNA, removing its nuclei, and keeping the cytoplasmic fragment with the mutations and, second, making a female embryonic stem cell, killing its mitochondria, and then using cell fusion to substitute the mutant mitochondria (from the cytoplasmic fragment) for the mitochondria. Next, they put these hybrids into blastocysts and then into a foster mother to create chimeric females. They then bred the females for transmission of the dominant agouti locus through the oocytes to pick up the mutant mitochondrial DNA.

According to Wallace, the researchers found that mice with the P25L mutation had increased axonal swelling, demyelination, and abnormal mitochondria. But what he found interesting was that when they looked biochemically at the synaptosomes of the brains of these mice, they saw that the problem was not with ATP—the ATP in the mice with and without the mutation was the same. They observed a 30 percent reduction in respiration, but, Wallace noted, if energy demand is increased, respiration will rise to near normal. Rather than an effect on ATP, he explained, the mice with the mutation were experiencing reactive oxygen species production “through the roof.” Wallace characterized this as a “disease of oxidative stress,” which kills the neurons.

Wallace added that mice with the cytochrome oxidase mutation had a 50 percent reduction in cytochrome oxidase in all of their tissues. Additionally, they had what he described as the “ragged red fibers” that were seen in the original human family with abnormal mitochondria, progressive cardiomyopathy with fibrosis, and type 2 diabetes with aging. So again, he pointed out, the mice showed the same phenotypes as those seen in humans with a single-point mutation.

The Effect of Heteroplasmy on Phenotype

Among all of these murine studies, Wallace cited as most interesting that in which the researchers took two perfectly normal mitochondrial DNAs, 129 and NZB, mixed them together, and then segregated the heteroplasmic animals with this same nucleus back into homoplasmy and heteroplasmy groups. They then examined activity in the mice, which are normally active at night but not during the day. They found that both the 129 and NZB animals were active at night but not during the day, while the heteroplasmic animals “just sat there,” Wallace said. He characterized them as depressed.

Wallace went on to explain that if the researchers subjected the mice to a color-cued task—providing them with an open field with differently colored holes around the outside of the field, one of which contained a black box where a mouse could hide—the homoplastic animals learned over successive days where the black box was. Mice do not like to be in open, exposed areas, he noted. The heteroplasmic mice learned where the black box was as well, he reported, but it took them longer. Moreover, after four trials, when the animals were removed from the field for 2 days and then returned, the homoplasmic mice immediately ran and jumped into the black box, having remembered where it was, but the heteroplasmic mice did not, having, Wallace noted, learned nothing because they had no long-term memory. “So simply converting a homoplasmic cell to a heteroplasmic cell, of perfectly normal mitochondrial DNAs,” he said, “severely altered the neuropsychiatric pattern of those animals.”

Geographic Constraints of Human Mitochondrial DNA

Wallace pointed out that because mitochondrial DNA can be transmitted only from mother to daughter, the only way it can change is by sequential mutations. So, he explained, if he were to sequence the mitochondrial DNA of any two individuals, the number of changes would be equivalent to the generations since those two individuals shared a common mother. Thus, he said, by sequencing the mitochondrial DNA from people around the world, one can reconstruct their genetic relationships and migration patterns.

According to Wallace, in humans, mitochondrial DNA arose about 200,000 years ago in sub-Saharan Africa. Today, he explained, the Khoisan people in that region have the most ancient mitochondrial DNA lineage. Two lineages that arose in Ethiopia—M and N—left Africa and colonized the rest of the world. N went to the temperate zone and gave rise to a number of European lineages (I, J, T, U, K, W, and Z), and also went to the temperate zone of Asia, whereas M stayed in the tropics, south to Australia, and later acquired new mutations to live in the temperate zone of Asia as well. Lineages C and D from M and lineage A from N crossed the Bering land bridge, from Chukotka, and colonized the Americas.

For Wallace, this geographic pattern is astonishing because nuclear polymorphisms, in contrast, are found panmictically (occurring through random mating) throughout all populations. The fact that mitochondrial DNA variation is highly geographically constrained based on the geographic origin of people's ancestors, he explained, is why 23andMe is able to analyze customers' mitochondrial DNA and provide them with information about their relations in other parts of the world.

Wallace's explanation for why geography constrains mitochondrial variation is that people's human ancestors evolved different adaptive mutations that allowed them to live in different environments and cope with different problems. In Africa, for example, he imagines the need to run away from lions, which would have required a great deal of ATP and thus a tightly coupled mitochondrial system. In the north, by contrast, the problem was not predation, he said, but freezing temperatures. Individuals there accumulated mutations that decreased the efficiency of the mitochondria, so that they were eating more calories for the same amount of ATP and generating more heat. According to Wallace, this is why people in the north still consume a high-fat marine mammal diet. “That's their niche,” he said.

As an example of adaptive mitochondrial variation in humans, Wallace mentioned lineage J, which he described as a tiny part of the European lineage founded by two cytochrome B mutations, 15257 and 14798. The latter mutation is conserved in all mesosomic animals, but is polymorphic otherwise. In contrast, 15257 is conserved across evolution, yet Wallace estimated that 5 percent of the workshop participants had a variant of this gene. “That's unheard of,” he said. “That is antithetical to what we think about evolutionary biology. If something is conserved across evolution, it should be homogeneous within a population. Not so for mitochondrial variation.” He characterized mitochondrial variation as “our adaptive engine. It allows us to adapt our energy to environmental changes.”

Wallace went on to talk about an A-to-G mutation that arose 10,000 years ago in Europe in the tRNA glutamine gene (Hutchin and Cortopassi, 1995). This mutation is found in only 0.4 percent of modern Europeans, but in about 3 percent of individuals with Alzheimer's disease, 5 percent of those with Parkinson's disease, and 7 percent of those with both diseases. Wallace noted that other mutations are much more deleterious than even this one—for example, ND1 methionine 31 valine (M31V), which causes both Alzheimer's and Parkinson's diseases. He reported further that in a study of European individuals with autism, mitochondrial variation was shown to be correlated dramatically with that condition, with mitochondrial haplogroups accounting for 55 percent of the risk. He added that mitochondrial DNA variation has been associated with a range of other common neurodegenerative, neurological, metabolic, and inflammatory diseases, as well as with altitude adaptation, cancer, aging, and athletic performance.

Nuclear–Mitochondrial DNA Interaction

As an example of a nuclear–mitochondrial DNA interaction, Wallace described a nuclear mutation, Ant1, in a Mennonite pedigree that affects the adenine nucleotide translocator isoform-1. The mutation arose in Switzerland about 500 years ago and was carried to North America. It was originally a recessive mutation, but then heterozygous individuals married each other and gave rise to the affected homozygous mutants. What is interesting about this mutation, in Wallace's view, is that some people with the affected mutants have very mild cardiomyopathy, while others die of fulminant, dilated cardiomyopathy. Compared with the heartbeat of cultured cardiomyoctes derived from embryonic stem cells in a healthy individual, the heartbeat in a mutant individual is highly dysrhythmic. According to Wallace, the different phenotypes are explained by the fact that people with severe cardiomyopathy have European mitochondrial lineage U, while those with mild cardiomyopathy have European mitochondrial lineage H. “So it is the mitochondrial DNA that is determining the severity of the disease,” he said, “not the nuclear mutation.”

In mice with the same mutation, Wallace continued, it has been shown that wild type mice (Ant1+/+) will continue to exercise until giving up, while mutant animals (Ant1–/–) start running but fall down because of the progressive accumulation of abnormal mitochondria. The mutant mice also have been shown to exhibit highly dysrhythmic cardiomyopathy, as opposed to the wild type. When the Ant1–/– mutation is combined with a mitochondrial COI mutation, there is no significant change in phenotype, but when it is combined with a mitochondrial ND6 mutation, mice develop severe cardiomyopathy with half the life span. Wallace interpreted this to mean, again, that mitochondrial DNA, not nuclear DNA, is determining the phenotype.

In terms of behavior, Wallace continued, when these same mice were put into a restraint to see how they would respond, the Ant1–/– mice with the mitochondrial ND6 mutation had a much stronger corticosterone response relative to the wild type mice. Thus, he noted, this same mitochondrial variant also affects brain development.

Closing Thoughts on the Mitochondrial View of Disease

Wallace closed by describing mitochondria as “the environmental sensors.” That is, when the environment changes, the epigenome changes, which in turn reconfigures the mitochondrial genotype to maintain homeostasis. But if the environmental change is too great or if there are mutations in the nuclear cytoplasmic system, Wallace explained, the mitochondria are unable to adjust, leading to energetic deficiency and disease. Based on this mitochondrial view of disease, he and his colleagues are now looking at traditional Chinese therapeutics. For 5,000 years, he said, Chinese medicine has been based on the idea of Qi (Chi), which he described as “vital force.” He speculated that eastern therapeutics may act through mitochondrial biology.



CpG is a coupling of a cytosine and guanine nucleotide in linear sequence; the cytosines in CpG dinucleotides can be methylated, unmethylated, or hemimethylated, with methylation status affecting gene expression.


The SNP database (dbSNP) is a public domain archive of SNP and other small-scale genetic variations, not just in humans but in all species. See https://www​ (accessed February 20, 2018).


Another one of these diseases with an important nutritional implication, Ordovás noted, is phenylketonuria (PKU), which is detected in about 400 infants born each year in the United States. The approximate cost of screening per child with PKU detected is $2,500, and the cost of dietary treatment for 10 years is approximately $8,000. In comparison, the expected cost of institutionalization over a 30-year period is estimated to be $162,000 (Grosse, 2015).


EPA is eicosapentaenoic acid, an omega-3 fatty acid.


Voltaire was an 18th-century writer, historian, and philosopher.


Vesalius was a 16th-century anatomist and physician.

Copyright 2018 by the National Academy of Sciences. All rights reserved.
Bookshelf ID: NBK518613


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (3.8M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...