Advances in 2009 cast light on the mystery that enshrouds the genetics of mental illness. A convergence of scientific insight and increasingly sophisticated technology is allowing scientists to peer deeper into DNA, finding mutations that may help uncover the underpinnings of brain dysfunction in psychiatric disorders.
Several studies during the year revealed genetic mutations that may lead to new ways to approach schizophrenia. Others may help explain why autism is a central feature of many syndromes that in other respects differ greatly. Yet another study showed that adverse experiences in childhood can lead to “epigenetic” changes—actual changes to one’s genetic makeup that creep in during the process of converting DNA code into a functional protein—resulting in the seeming paradox of an environmentally induced genetic condition.
“The power of genetics is now so advanced that we can check many pieces of the genome to find rare mutations and variants that, together, may lead to new understanding,” said Scharahm Akbarian of the University of Massachusetts. Akbarian compared the advances in recent years to the evolution of Google Earth, which began as a series of satellite images of the planet. “Now you can use Google Earth to find an intersection in a town. In the same way, we can scan larger pieces of DNA faster than ever, finding lots of candidate regions involved in brain disorders. And a very encouraging development is that some of these regions are being replicated in different groups of subjects.”
Setting the Stage for Genetic Breakthrough
New understanding is sorely needed in a field where no new therapies have emerged for several decades. The medications used to treat psychiatric illness still attempt to adjust the balance of brain chemicals such as dopamine and serotonin. But most take weeks to have an effect and come with severe side effects, not to mention that, always, a significant number of patients fail to find any benefit.
The 1990s saw an intensified search for “culprit genes” involved in brain disorders and many other illnesses. The approach was to identify susceptibility genes among patients with a particular disease, use the genes to develop animal models and identify the molecular pathways involved, thus leading to possible therapies such as developing a custom-designed or “monoclonal” antibody or other therapeutic compound to block the effects of the malfunctioning gene’s protein product. In Alzheimer’s disease, for example, the discovery of the amyloid precursor and presenilin genes have pointed to molecular pathways involved in the disease, as well as to several potential drug targets.
In general, though, the paradigm has not held up for neuropsychiatric disorders, due to what Harvard University’s Steven Hyman calls the “fiendish complexity” of the underlying genetics.1 Similar symptoms can result from different genetic risk factors; or, conversely, individuals with the same genetic variant can have different DSM-IV diagnoses or show no symptoms at all. In one large family, for example, a disrupted gene on chromosome 1 can lead to schizophrenia, the comparatively milder form known as schizoaffective disorder, bipolar disorder or major depression.
According to Douglas Levinson of Stanford University, some of the first hunts for “culprit genes” in mental illness looked for genes involved with the chemical messengers that were targeted by the available drugs—schizophrenia, for example, is still treated with dopamine blockers—yet none of these genetic dragnets yielded candidate genes of any statistical significance. Now, however, new insights are appearing at levels of precision that would not have been possible without the convergence of knowledge, procedures and technology that enabled researchers to find needles in the haystack of the genome.
Each colored spot on a microarray is associated with a different gene. The different colors represent either healthy (control) or diseased (sample) tissue. The location and intensity of a color shows whether the gene, or mutation, is present in either the control and/or sample DNA, and its level of expression. (National Medical Library)
In 2001, the entire human genome (the complete sequence of human DNA) was published in a worldwide effort appropriately named the Human Genome Project.
“The Human Genome Project sparked an intense wave of competition among biotechnology companies to develop ‘microarrays’ on which a million or more genetic variants could be tested or assayed,” said Levinson. “Meanwhile, clinicians had to find ways to recruit not just hundreds but thousands of people with schizophrenia. Finally, computers had to be sophisticated enough to handle the data. When the first search for schizophrenia genes began, back in the 1990s, the average hard drive was 32 megabytes. Today one of our files wouldn’t even fit on a hard drive that size, never mind the software needed to analyze all the data.”
Misspellings in the Code
Some of the first clues to result from the Human Genome Project were minute changes called single-nucleotide polymorphisms, abbreviated as SNPs and pronounced “snips.” A further research push, the International HapMap Project, cataloged about three million SNPs between 2005 and 2007.
SNPs are genetic differences between individuals at the level of one “letter” in the genetic code. This code consists of long chains of four “bases,” or building blocks, called nucleotides—adenine, guanine, cytosine and thymine, each abbreviated by its first letter—held together in pairs by chemical bonds that twist the chain of DNA into its famous double-helix shape. A SNP is a substitution of one of these four letters with another. These differences are thought to account for genetic diversity as well as susceptibility to disease.
Though other discoveries in recent years show that SNPs are not the whole story, they remain a promising line of research for understanding how a disease works. Three papers published in the July 1, 2009, issue of Nature uncovered a trove of SNPs that may help explain the development of schizophrenia.
Each study was a genome-wide association scan—a systematic search for common SNPs that influence a disease or trait—led by an international consortium of scientists. Because the three groups shared their results, cross-checking their findings with the other groups’ patient samples to make each study a large “meta-analysis,” the number of subjects ultimately totaled more than 8,000 patients with schizophrenia and 19,000 controls (individuals without the disease). The pooled results of the three studies turned up seven SNPs in an area on chromosome 6 that contains many genes involved in infection and immunity. Chromosome 6 hosts genes involved in the major histocompatibility complex (MHC), a set of proteins found on all cells, which signal to the immune system whether the cell is “self” or “non-self.” If the MHC binds to a non-self entity, such as a virus, the immune system launches its attack.
Single-nucleotide polymorphisms are genetic differences between individuals at the level of one “letter” in the genetic code. Because of base pairing, both nucleotides must change. In frame 2, the original “CTA” (frame 1) becomes “TTA” and “GAT” becomes “AAT.”(Copyright David Hall, used according to the GNU Free Documentation License.)
All three studies, individually and collectively, implicated the MHC—an intriguing finding since infection during pregnancy has long been suspected as one aspect of the prenatal environment that can increase risk for schizophrenia.
In a study by the Molecular Genetics of Schizophrenia (MGS) consortium, the SNPs identified were near a cluster of histone protein genes, which form a structure for DNA molecules and can be chemically modified to alter the expression of other genes.2 (Histones have another, lesser-known role in antibacterial defense.)
The MHC was implicated in another study, published by the Sgene Consortium with Kari Stefansson of deCODE genetics, a Reykjavíc, Iceland–based pharmaceutical company, as lead author. A genomewide scan of their own 13,000 patients and more than 2,500 controls, plus meta-analysis of patients from all three studies, detected an even stronger “signal” closest to a specific gene, called PRSS16, which is located in a cluster of histone genes and is involved in immunity. The Sgene study also identified sites on chromosomes 11 and 18 that are involved with brain development and memory.3
“We have to keep an open mind in schizophrenia research,” Levinson, a member of the MGS and study coauthor, said. “What if we eventually find that an abnormal response to an infection increases one’s risk of schizophrenia, as some researchers have suggested? Maybe that will lead to strategies for preventing schizophrenia in some people by preventing certain infections. But other findings suggested abnormalities in the development of brain cells. We still have a lot to learn.”
The third study, presented by the International Schizophrenia Consortium, also pointed to the MHC, finding significant overlap in gene variants that increase the risk for schizophrenia and bipolar disorder, but it found no overlaps with a host of non-psychiatric disorders, including hypertension and type 2 diabetes—indicating that these SNPs were specific signals for vulnerability to mental illness. This study also described a statistical model for deducing, on the basis of SNPs already identified, that a large set of common SNPs—most of them unknown at present—could account for at least 33 percent of risk.
The authors wrote that their model “suggests that genetically influenced individual differences across domains of brain development and function may form a [predisposition] for major psychiatric illness, perhaps as multiple growth and metabolic pathways influence human height."4
SNPs are providing other clues as well. In the August Molecular Psychiatry, other teams led by Levinson published a genome-wide linkage scan, plus a meta-analysis of a larger population, pointing to several chromosomal regions that might contain genes that play a role in schizophrenia.5,6 Unlike a genome-wide association scan, which compares the genomes of individuals with and without a given disorder, a linkage scan concentrates on families in which two or more people have the disease, looking for “marker” locations that are near disease-causing variants. When a marker is found near another variant in many families affected by a disease, more often than might be expected by chance, this “linkage” is thought to signal a nearby disease-related gene.
Levinson explained that while association studies cast a wider net in terms of finding more of the “common” SNPs (those affecting more than 5 percent of patients), linkage studies might do a better job of finding regions with many different, rarer mutations (affecting fewer than 1 percent). Since rare SNPs often confer a higher risk of disease, the linkage scan remains a powerful tool. The SNPs uncovered in the Molecular Psychiatry paper include a suspect region on chromosome 8, where a gene for neuregulin 1 is also found. The finding supports studies in mice showing that mutations in this gene lead to poorly developed neurons and a schizophrenia-like condition in what would be the mouse’s adolescence, tracking with the disorder’s time course in humans. Though the linkage study by Levinson and coworkers did not directly implicate the neuregulin 1 gene, the finding hinted at multiple problematic sites in that region, possibly providing a rationale for future studies to re-sequence this area, the authors wrote.
From SNP to Brain Imaging
The larger meta-analysis also turned up a possibility on chromosome 2, in an area associated with bipolar disorder and psychosis. A gene on this chromosome, dubbed ZNF8044, was tentatively implicated in schizophrenia in a 2008 Nature Genetics study.7 In 2009, a team of researchers reported in the May 1 issue of Science that they used a SNP in this gene as the basis of an imaging study to determine the SNP’s role.8
Schizophrenia is increasingly seen as a disruption of the synchronicity between brain areas, not simply as a deficit in one area or neurotransmitter. To investigate whether the affected gene plays a role, Andreas Meyer-Lindenberg of the University of Heidelberg, along with colleagues at Heidelberg and the University of Bonn, studied 115 otherwise healthy subjects carrying the SNP in question. The researchers asked the subjects to perform tasks designed to challenge two brain areas, the dorsolateral prefrontal cortex and hippocampus—both essential for complex thinking and memory, which are impaired in schizophrenia.
Using functional magnetic resonance imaging to monitor brain activity while the subjects performed the tasks, the researchers found that patients with the SNP showed reduced connectivity between these areas: the brain structures were working, just not coordinating. The authors suggested that future research should examine the role of ZNF8044 in the development of axons (projections along which neurons communicate) and in plasticity (the fine-tuning of connections between axons and neurons).
Copy Number Variations
Another discovery has opened up even more possibilities in understanding the genetics of psychiatric illness. This is a mutation called a copy number variation (CNV). If a SNP is a misspelling of a single letter of the genetic code, CNVs are analogous to whole groups of paragraphs being deleted, duplicated or shuffled.
Many brain disorders are known to result from deletions or duplications of chromosomes: Children born with Down’s syndrome, for example, have an extra copy of part or all of chromosome 21. Chromosomes—and extra chromosomes—can be seen with an ordinary microscope. CNVs, however, are deletions or duplications of stretches of nucleotides—often quite long—within a given chromosome. The first genome-wide description of CNVs was reported in 2004 by Stephen Scherer at the Hospital for Sick Children, Toronto, and his colleague Charles Lee at Harvard University, and concurrently by Michael Wigler at Cold Spring Harbor Laboratories, in New York State. Then a team led by Scherer published a “map” of CNVs in worldwide populations in the November 23, 2006, issue of Nature.9
CNVs have been observed between identical twins who otherwise have the same genome. This fact may explain why, in many diseases such as schizophrenia, the twin of an afflicted individual has only a 50 percent chance of having the disease—presumably, with identical DNA, the odds should be 100 percent. Though rare, CNVs are powerful, conferring a high risk of disease. They are already proving informative when it comes to mental illness, Scherer said. “CNVs seem to have a propensity for neuropsychiatric disorders—they’re involved in much higher proportions, and early data suggests they play a role in almost all psychiatric illnesses. You don’t yet see the same significance with other types of disease, although it is early days.”
Cellular Suspects in Autism
Among researchers, hopes are high that CNVs will shed new light on autism, which is not so much a distinct disorder as a common factor in many conditions and syndromes—some of which differ widely in other respects. Autism is increasingly described as any of a number of “autism spectrum disorders,” which can include impairments in social interaction and communication, repetitive behaviors, hypersensitivity to stimulation and onset before age 3.
Like SNPs, CNVs help to illuminate the process through which a disease unfolds. In the February 2008 issue of the American Journal of Human Genetics, Scherer’s group reported more than 200 CNVs in families with autism, some of them encompassing half a dozen genes involved in neural development but never thought to play a role in autism—as well as further implicating several genes already suspected.10
A study in the April 2009 Molecular Psychiatry used both SNP and CNV clues to pinpoint several genes that warrant further study.11 All are thought to be involved in the formation of synapses, the points of contact between neurons.
A child is born with far more neurons than he or she will end up with as an adult. Circuits in the brain are sculpted on the basis of the child’s experiences, with the density of synapses increasing in areas that are used—music, foreign language, sports, for example—and decreasing in those that are not. Synaptic formation and, eventually, “pruning” are normal stages for the developing brain. Autism, which usually appears between 18 months and 2 years of age, is thought to result from a disruption in this process. Too many or too few synapses are features in several syndromes that include autism.
Anthony Monaco of the Wellcome Trust Centre for Human Genetics at the University of Oxford joined with colleagues at 11 centers in the United States and Europe to examine areas on chromosomes 2 and 7, implicated in previous research. Drawing on techniques from both association and linkage studies, the researchers checked for SNPs and CNVs of approximately 250 families, matched against 188 controls. The hot spots that turned up were compared with a European sample representing 300 afflicted families.
Inês Sousa selects a patient DNA sample from the -20 degrees Celsius freezer at the Wellcome Trust Centre for Human Genetics. (Courtesy of Anthony Monaco / Wellcome Trust Centre for Human Genetics)
One gene of interest on chromosome 2, containing a SNP, was ZNF533. This gene encodes for several proteins that attach to DNA molecules and play a role in turning the gene on or off. Deletions in this gene have been found in patients with mental retardation.
The team found CNVs (deletions) spanning two genes on chromosome 7, called IMMP2L and DOCK4, which are prevalent in the fetal brain and active during neural development. A SNP, of the too few synapse commonality mentioned above, also turned up in DOCK4, which is thought to be involved in the growth of dendrites (the “receiving” points on neurons with which axons communicate).
“Taken together, these findings and others point to the synapse as a possible site for genetic effects of autism to occur,” said Monaco. He also believes that CNVs will eventually be diagnostic, adding that as microarrays become more sophisticated they are also getting cheaper.
Back-to-back studies in the April 28 Nature also pointed to pathways of brain development. Both were led by Hakon Hakonarson of the Children’s Hospital of Philadelphia, turning up SNPs and CNVs encompassing genes involved in brain development. Some of the affected genes coded for cell adhesion molecules, which are important in order for developing neurons to reach their proper location in the brain.12,13 Others coded for ubiquitin, a neuronal protein that “tags” other proteins to be degraded and disposed of (a necessary step in brain development).
“The findings support imaging studies that suggest a lack of connectivity between brain regions involved in higher-order activities,” said coauthor Daniel Geschwind of the University of California, Los Angeles. “The variants seem to be in pathways involved in the brain’s wiring during development, but it’s not a done deal yet.”
Like Monaco, Geschwind is excited about the diagnostic possibilities suggested by both SNPs and CNVs. He added that between 5 and 10 percent of variants are “de novo” mutations, meaning they were not passed on by either parent. “If you can say for sure that your child has a de novo mutation, then subsequent children are at no higher risk than anyone else.”
CNVs and Schizophrenia
Some copy number variations now emerging are common to both schizophrenia and autism. A CNV on chromosome 15, for example, increases the risk of schizophrenia, autism, mental retardation and epilepsy.
At the National Institute of Mental Health, Anjene Addington and Judith Rapoport have discovered CNVs involved in childhoodonset schizophrenia, a rare form of the disorder in which symptoms appear before age twelve (onset in late adolescence is typical). Like other early-onset diseases, schizophrenia of this type is more severe. Working with a group of 150 patients with childhood-onset schizophrenia, followed by the NIMH for more than twenty years, the researchers replicated findings identifying four susceptibility genes in adult-onset schizophrenia and showed that the same four suspects were found in early-onset disease as well. Reasoning that these might confer especially high risk, the researchers looked for copy number variations in these genes.14 Four patients had a deletion on chromosome 22 that has previously been associated with autism, mental retardation and facial dysmorphy. Two patients showed a duplication in a region of chromosome 16 also found to be disrupted in adult schizophrenia in previous studies. Because two out of one hundred (2 percent) is a far greater rate than was found in adults, the researchers surmise that this CNV may confer exceptionally high risk.
The researchers found CNVs in three other genes involved in neural development and implicated in autism, adult schizophrenia or both. The authors wrote that if all the CNVs impacting the genes identified in their 105-patient group were disease-causing, they could explain the origins of schizophrenia in almost 40 percent of the patients. “This is a huge leap from where we stood just one or two years ago,” the authors wrote, adding that studies to re-sequence the target areas are under way.
miRNAs Suppress the Code
Some genes malfunction not because of any flaw in their construction, but because the information they encode is never brought into reality. Such alterations in a gene’s “expression,” rather than its DNA sequence, are increasingly described as “epigenetic” and represent a new way of understanding many kinds of illness, including psychiatric disorders.
A gene’s protein is produced in two steps. The first, transcription, occurs when the DNA code is synthesized into an intermediate molecule called messenger RNA (mRNA). The second step, translation, converts the mRNA into the sequence of amino acids (chemical building blocks) that make up the final protein “product.” In recent years, small molecules called microRNAs (miRNAs) have been shown to influence translation, in normal processes as well as in disease. A better understanding of their actions may lead to bettertargeted “gene silencing” therapies that prevent faulty genes from being translated.
miRNAs do not become translated into protein but attach themselves to other stretches of RNA that do—regulating the production of the gene’s protein and sometimes even “silencing” the gene. When acting normally, miRNAs elegantly help to control cellular processes. But miRNAs have also been implicated in many diseases, including heart disease and some cancers. Though they do not act on all genes everywhere in the body and brain, miRNAs are prevalent in the prefrontal cortex (the seat of “higher” functions such as reasoning and analysis).
A diagram of microRNA pathways in the vertebrate cell. Within the nucleus, the initial expression, or primary miRNA (pri-miRNA), is digested by a microprocessor, an enzyme called Dicer because it “dices” the pri-miRNA, and becomes precursor miRNA (pre-miRNA). The pre-miRNA is exported to the cytoplasm, where the Dicer protein cleaves it into mature miRNA, which then loads onto RNA-induced silencing complex (RISC). Carried by RISC, the miRNA then binds to mRNA and either inhibits expression or speeds degradation of the mRNA. (Courtesy of Schahram Akbarian / University of Massachusetts Medical School)
Schahram Akbarian, Nikolaos Mellios and colleagues at the University of Massachusetts Medical School have shown that in schizophrenia, a specific miRNA, designated miRNA 195, may upset the balance of brain chemicals. In a postmortem study of the prefrontal cortex of twenty subjects with schizophrenia and twenty controls, the researchers found that higher amounts of miRNA 195 led to reduced levels of two important messenger chemicals: gamma-aminobutyric acid (GABA) and a neuron-nourishing compound called brain-derived neurotrophic factor (BDNF). GABA is an “inhibitory” neurotransmitter that signals neurons to slow down their firing rate. Previous research had shown that this messenger is insufficient in schizophrenia.
“Many researchers believe that GABA acts as a kind of orchestra conductor to coordinate activity among brain areas,” Akbarian said. “For this neurotransmitter to be disrupted may be like the conductor becoming distracted and the musicians playing chaotically.”
The finding by Akbarian and Mellios, reported in the June 15 Biological Psychiatry, revealed yet another level of complexity but also of specificity, suggesting that miRNA 195 contributes to the disease by reducing these two key chemical messengers.15 The study also points to miRNA 195 as a possible target for gene-silencing therapies still to be developed in the future, perhaps ushering in a day when schizophrenia could be treated by preventing certain genes from being translated into protein.
Epigenetic Changes in Suicide
Many psychiatric illnesses are assumed to result from one or more “susceptibility genes” that are set in motion by some trigger from the environment. But a striking study from McGill University, Montreal, showed just the opposite: environmental trauma in the form of child abuse can actually cause genetic changes—in ways that can lead to suicide.
Reporting in the March Nature Neuroscience,16Michael Meaney and colleagues examined brain tissue taken postmortem from suicide victims with and without histories of childhood abuse, as well as control samples from individuals who had died suddenly of other causes. The researchers focused on the hippocampus, a part of the brain that plays a role in stress, emotion and memory. The hippocampus is studded with receptors for the stress hormones known as glucocorticoids, which play many roles in the stress response.
Glucocorticoid receptors have their own shutoff mechanism, keeping in balance the amount of stress hormones that enter the hippocampus. Some conditions like major depression and posttraumatic stress disorder involve a loss of neurons containing the receptors—which, paradoxically, can lead to high levels of stress hormones in the brain and a host of stress-related disorders. When Meaney and colleagues examined the brain tissue of abused subjects who had committed suicide, the brains of child-abuse victims showed several differences. There were signs that glucocorticoid receptors had decreased in number; the genes encoding the receptors showed alterations in the “promoter” region where the process of gene expression begins; and activation of the promoter region occurred through a different type of chemical process. None of these changes were seen either in the brains of suicides who were not abused or in the controls.
The finding suggests that suicide can be considered a developmental disorder, in the sense that trauma in childhood causes biochemical changes and changes in gene expression that ultimately lead to the tragic event. “Our data are consistent with the hypothesis that early life events can alter the epigenetic state of relevant genomic regions, the expression of which may contribute to individual differences in the risk for psychopathology,” the authors wrote.
The explosion of research in the last few years will make it possible to scan the entire genomes of thousands, even tens of thousands of individuals. “The challenge will be to analyze the data and figure out what it all means,” said UCLA’s Geschwind. The genetic variants now showing up in multiple genes may map out different pathways of disease development that converge at one target site for future medications. This scenario is possible, though unlikely, Geschwind said. More probable is that subsets of psychiatric illnesses will be classified and understood according to their genetic basis and development.
Geschwind added that even though known genetic variants remain rare, still they add up. In autism, for example, the best-known variants affect only 0.5 to 1 percent of patients. “But if you can develop tests for several SNPs or CNVs at once, you might be able to explain 5 or even 10 percent of cases,” Geschwind said. “Since schizophrenia affects about one in 100 people, and autism about one in 150, the total number of patients and families is considerable.” A genome-wide scan might cost about $1,000, but Geschwind notes that routine tests such as CT scans and MRIs cost as much—with a similarly “low yield” of finding disease in 1 to 5 percent of patients. The same rationale holds true for neuropsychiatric illness, Geschwind said. The power of genetics will be of great benefit for the families coping with these disorders.