INSIGHT REVIEW NATURE Vol 447 24 May 2007 Epigenetics and common complex disease environment, a form of loss of phenotypic plasticity. This loss of pheno- The next great frontier in the epigenetics of human disease is to estab- typic plasticity could be mediated epigenetically if loss of the normal lish its potential role in common non-neoplastic human diseases. At balance between gene-promoting and gene-silencing factors occurred the moment, the most appealing candidates are disorders affecting across the genome( Fig 3). This idea is supported by a study showing behaviour, on the basis of clues from Rett syndrome and Prader-Willi greater variance of total DNA methylation and histone H3K9 acetylatic syndrome, as well as the intriguing story of Turner syndrome in girls in older monozygotic twins than in younger twins, although that study with only one X chromosome. Girls lacking the paternal X chromo- did not measure epigenetic changes over time in the same individual some exhibit behavioural socialization problems more frequently than The CDGE hypothesis is also supported by two compelling lines of girls lacking the maternal X chromosome", and a candidate imprinted evidence in model organisms. First, inhibition of the chaperone protein gene region that could be responsible has been identified". Autism and Hsp90 in Drosophila leads to the expression of previously latent herit bipolar disorder are two common complex traits that have defied gene able mutations within a single meiotic cell division, which then become identification, and both show surprisingly high frequencies of pheno- independent of Hsp90(ref. 92). This mutational suppression is chro- typic discordance in monozygotic twins. Only 60% concordance was matin mediated and can be reversed by mutations in several trithorax reported in autism using strict criteria", and neuroanatomical differ- group proteins".Second, a screen for genes that cooperate in disrupting ences have been found in cerebellar grey- and white-matter volumes Caenorhabditis elegans phenotypes revealed six hub genes that inter monozygotic twins are discordant, and the disease itself is episodic, nents of chromatin-modifying complexes".Thus, common diseases with patients being seriously ill at some times and perfectly normal at may involve phenotypic variants with both genetic variation and envi- others, often for long stretches of time, for no apparent reason. Some ronmentally triggered epigenetic change that modulates the effects of studies of both autism and bipolar disorder have also shown parent- DNA sequence variation(Fig 4). These epigenetic modifiers are, in turn, of-origin-specific linkage, with excess transmission of paternal alleles affected by variation in the genes that encode them, and environmental to autism cases", and excess transmission of maternal alleles to bipolar factors(hormones, growth factors, toxins and dietary methyl donors) disorder cases? influence both the genome and epigenome( Fig 4). This idea can be A third candidate common disease with an epigenetic compor tested by incorporating an assessment of the epigenome into popula is systemic autoimmune disease. Aberrant hypomethylation is found tion epidemiological studies(see ref. 95 for a review), rather than simply T cells of patients with systemic lupus erythematosis, including in stratifying risk for environmental exposures as is done currently genes such as lymphocyte function-associated antigen-1, which is overexpressed in lupus T cells". Treatment of viable T cells with 5-aza- Prospects for epigenetic therapy 2-deoxycytidine induces a syndrome in mice similar to systemiclupus As epigenetic mechanisms for human disease are identified, epigenetic rythematosis. Procainamide and hydralazine both cause hypomethyl- therapies are being developed or rediscovered. Some drugs are used ation and can cause lupus, and treatment of T cells with these drugs specifically because of their known effects on the epigenome. For exam elicits a lupus-like syndrome in mice ple, two classes of epigenome-modifying agent are currently in clinical trials for cancer, for example, for the treatment of myelodysplasia Epigenetics and the environment DNA methyltransferase inhibitors such as decitabine, and histone he epigenome is an important target of environmental modification. deacetylase inhibitors such as SAHA(suberoylanilide hydroxamic Environmental toxins such as heavy metals disrupt DNA methylation acid). SAHA is being used for cancer treatment, although its in vivo and chromatin" Oestrogenic and anti-androgenic toxins that decrease targets are still unknown. The overall response rate with decitabine in a male fertility alter DNA methylation, and these changes are inherited phase Ill study showed a small but statistically significant difference for by subsequent generations. Dietary modification also can have a myelodysplasia(9% complete response and 8% partial response, com- profound effect on DNA methylation and genomic imprinting Defi- pared with no response in controls), and half of the clinically responsive ciency in folate and methionine, necessary for normal biosynthesis of patients showed a cytogenetic response One cautionary note about S-adenosylmethionine, the methyl donor for methylcytosine, leads to the use of nonselective agents that inhibit DNA methylation is that these aberrant imprinting of IGF2 (ref. 84), and methylation supplementa- drugs may activate as many genes as they silence An effect opposite to tion can cause methylation and silencing of a retroposon in mice with that of methyltransferase and histone deacetylase inhibitors is achieved silencing of the nearby agouti coat-colour gene. Colorectal cancer risk through rational drug design of histone acetyltransferase inhibitors, for is linked to both dietary folate deficiency and variants in methylene- example, bisubstrate analogues such as Lys-CoA, a selective P300/CBP tetrahydrofolate reductase, which has a critical role in directing the inhibitor. Such drugs may be useful in cancer treatment, because P300- folate pool toward remethylation of homocysteine to methionine .A negative cells undergo increased apoptosis after chemotherapy. emarkable example of an environmental effect on the epigenome is Some drugs that have an effect on the epigenome are already in wide the modification of glucocorticoid receptor methylation seen in the spread use, but their epigenetic effect has only recently been discovered. hippocampus of rat pups in response to maternal grooming. A sur- For example, valproic acid is used to treat various disorders, including prising environmental modulator of the epigenome is assisted repro- seizures, bipolar disorder and cancer, and valproic acid was recently ductive technology (ART), which has been shown to be the method found to be a potent histone deacetylase inhibitor. A relatively simple of conception at higher than expected frequency in Beckwith-Wiede- drug strategy could be to target rationally designed small compounds mann syndrome and Angelman syndrome. Intriguingly, all but 1 of to a epigenetically altered pathway, rather than attempting to repair an the 14 reported cases of Beckwith-Wiedemann syndrome associated epigenetic lesion. For example, in patients with LOI of IG F2, the IGF2 with ART involved hypomethylation of LITI (ref. 89), although this signalling receptor, IGFIR tyrosine kinase, or the downstream Akt abnormality is normally present in only about one-third of patients or ERK signalling pathways could be targeted with existing drugs or with Beckwith-Wiedemann syndrome compounds under development rather than attempting to reverse the The common disease genetic and epigenetic(CDGE) hypothesis epigenetic lesion itself. argues that in addition to genetic variation, epigenetics provides an added Given that epigenetics is at the heart of phenotypic variation in health layer of variation that might mediate the relationship between genotype and disease, it seems likely that understanding and manipulating the and internal and external environmental factors". This epigenetic com- epigenome holds enormous promise for preventing and treating com nent could help to explain the marked increase in common diseases mon human illness. Epigenetics also offers an important window to ith age, as well as the frequent discordance of diseases such as bipolar understanding the role of the environment's interactions with the disorder between monozygotic twins". A common characteristic of age- genome in causing disease, and in modulating those interactions to ing is a time-dependent decline in responsiveness or adaptation to the improve human health @2007 Nature Publishing GroupEpigenetics and common complex disease The next great frontier in the epigenetics of human disease is to estab￾lish its potential role in common non-neoplastic human diseases. At the moment, the most appealing candidates are disorders affecting behaviour, on the basis of clues from Rett syndrome and Prader–Willi syndrome, as well as the intriguing story of Turner syndrome in girls with only one X chromosome. Girls lacking the paternal X chromo￾some exhibit behavioural socialization problems more frequently than girls lacking the maternal X chromosome72, and a candidate imprinted gene region that could be responsible has been identified73. Autism and bipolar disorder are two common complex traits that have defied gene identification, and both show surprisingly high frequencies of pheno￾typic discordance in monozygotic twins. Only 60% concordance was reported in autism using strict criteria74, and neuroanatomical differ￾ences have been found in cerebellar grey- and white-matter volumes between discordant monozygotic twins75. In bipolar disorder, 30% of monozygotic twins are discordant76, and the disease itself is episodic, with patients being seriously ill at some times and perfectly normal at others, often for long stretches of time, for no apparent reason. Some studies of both autism and bipolar disorder have also shown parent￾of-origin-specific linkage, with excess transmission of paternal alleles to autism cases77, and excess transmission of maternal alleles to bipolar disorder cases78. A third candidate common disease with an epigenetic component is systemic autoimmune disease. Aberrant hypomethylation is found in T cells of patients with systemic lupus erythematosis, including in genes such as lymphocyte function-associated antigen-1, which is overexpressed in lupus T cells79. Treatment of viable T cells with 5-aza- 2ʹ-deoxycytidine induces a syndrome in mice similar to systemic lupus erythematosis80. Procainamide and hydralazine both cause hypomethyl￾ation and can cause lupus, and treatment of T cells with these drugs elicits a lupus-like syndrome in mice81. Epigenetics and the environment The epigenome is an important target of environmental modification. Environmental toxins such as heavy metals disrupt DNA methylation and chromatin82. Oestrogenic and anti-androgenic toxins that decrease male fertility alter DNA methylation, and these changes are inherited by subsequent generations83. Dietary modification also can have a profound effect on DNA methylation and genomic imprinting. Defi￾ciency in folate and methionine, necessary for normal biosynthesis of S-adenosylmethionine, the methyl donor for methylcytosine, leads to aberrant imprinting of IGF2 (ref. 84), and methylation supplementa￾tion can cause methylation and silencing of a retroposon in mice with silencing of the nearby agouti coat-colour gene85. Colorectal cancer risk is linked to both dietary folate deficiency and variants in methylene￾tetrahydrofolate reductase, which has a critical role in directing the folate pool toward remethylation of homocysteine to methionine86. A remarkable example of an environmental effect on the epigenome is the modification of glucocorticoid receptor methylation seen in the hippocampus of rat pups in response to maternal grooming87. A sur￾prising environmental modulator of the epigenome is assisted repro￾ductive technology (ART), which has been shown to be the method of conception at higher than expected frequency in Beckwith–Wiede￾mann syndrome and Angelman syndrome88. Intriguingly, all but 1 of the 14 reported cases of Beckwith–Wiedemann syndrome associated with ART involved hypomethylation of LIT1 (ref. 89), although this abnormality is normally present in only about one-third of patients with Beckwith–Wiedemann syndrome. The common disease genetic and epigenetic (CDGE) hypothesis argues that in addition to genetic variation, epigenetics provides an added layer of variation that might mediate the relationship between genotype and internal and external environmental factors90. This epigenetic com￾ponent could help to explain the marked increase in common diseases with age, as well as the frequent discordance of diseases such as bipolar disorder between monozygotic twins76. A common characteristic of age￾ing is a time-dependent decline in responsiveness or adaptation to the environment, a form of loss of phenotypic plasticity. This loss of pheno￾typic plasticity could be mediated epigenetically if loss of the normal balance between gene-promoting and gene-silencing factors occurred across the genome (Fig. 3). This idea is supported by a study showing greater variance of total DNA methylation and histone H3K9 acetylation in older monozygotic twins than in younger twins, although that study did not measure epigenetic changes over time in the same individual91. The CDGE hypothesis is also supported by two compelling lines of evidence in model organisms. First, inhibition of the chaperone protein Hsp90 in Drosophila leads to the expression of previously latent herit￾able mutations within a single meiotic cell division, which then become independent of Hsp90 (ref. 92). This mutational suppression is chro￾matin mediated and can be reversed by mutations in several trithorax group proteins93. Second, a screen for genes that cooperate in disrupting Caenorhabditis elegans phenotypes revealed six ‘hub’ genes that inter￾acted with as many as one-quarter of all genes tested. All were compo￾nents of chromatin-modifying complexes94. Thus, common diseases may involve phenotypic variants with both genetic variation and envi￾ronmentally triggered epigenetic change that modulates the effects of DNA sequence variation (Fig. 4). These epigenetic modifiers are, in turn, affected by variation in the genes that encode them, and environmental factors (hormones, growth factors, toxins and dietary methyl donors) influence both the genome and epigenome (Fig. 4). This idea can be tested by incorporating an assessment of the epigenome into popula￾tion epidemiological studies (see ref. 95 for a review), rather than simply stratifying risk for environmental exposures as is done currently. Prospects for epigenetic therapy As epigenetic mechanisms for human disease are identified, epigenetic therapies are being developed or rediscovered. Some drugs are used specifically because of their known effects on the epigenome. For exam￾ple, two classes of epigenome-modifying agent are currently in clinical trials for cancer, for example, for the treatment of myelodysplasia96: DNA methyltransferase inhibitors such as decitabine, and histone deacetylase inhibitors such as SAHA (suberoylanilide hydroxamic acid). SAHA is being used for cancer treatment, although its in vivo targets are still unknown. The overall response rate with decitabine in a phase III study showed a small but statistically significant difference for myelodysplasia (9% complete response and 8% partial response, com￾pared with no response in controls), and half of the clinically responsive patients showed a cytogenetic response97. One cautionary note about the use of nonselective agents that inhibit DNA methylation is that these drugs may activate as many genes as they silence27. An effect opposite to that of methyltransferase and histone deacetylase inhibitors is achieved through rational drug design of histone acetyltransferase inhibitors, for example, bisubstrate analogues such as Lys-CoA, a selective p300/CBP inhibitor98. Such drugs may be useful in cancer treatment, because p300- negative cells undergo increased apoptosis after chemotherapy99. Some drugs that have an effect on the epigenome are already in wide￾spread use, but their epigenetic effect has only recently been discovered. For example, valproic acid is used to treat various disorders, including seizures, bipolar disorder and cancer, and valproic acid was recently found to be a potent histone deacetylase inhibitor100. A relatively simple drug strategy could be to target rationally designed small compounds to a epigenetically altered pathway, rather than attempting to repair an epigenetic lesion. For example, in patients with LOI of IGF2, the IGF2 signalling receptor, IGF1R tyrosine kinase, or the downstream Akt or ERK signalling pathways could be targeted with existing drugs or compounds under development rather than attempting to reverse the epigenetic lesion itself. Given that epigenetics is at the heart of phenotypic variation in health and disease, it seems likely that understanding and manipulating the epigenome holds enormous promise for preventing and treating com￾mon human illness. Epigenetics also offers an important window to understanding the role of the environment’s interactions with the genome in causing disease, and in modulating those interactions to improve human health. ■ 438 INSIGHT REVIEW NATURE|Vol 447|24 May 2007 ￾