Psychiatry in the Postgenomic Era
Kathy L. Kopnisky, PhD
Steven E. Hyman, MD
Dr. Kopnisky is assistant to the director, and Dr. Hyman is director, both at the National Institute of Mental Health in Bethesda, MD.
A rough draft of the human genome sequence is now available in the public domain and the “finished” sequence of the 3 billion base pair genome will be available. Genes play a critical role in the risk of mental illness, but deciphering their role has proven to be extremely difficult. It is only with information from the genome project that we can succeed in finding risk conferring genes, but other challenges, including the boundaries of the illnesses themselves, still await us. Even while this research is proceeding, we should question how genomics and genetics will change our understanding of mental illness and, ultimately, the practice of psychiatry.
As we enter the 21st century, we are still uncertain about the mechanisms underlying mental disorders. Powerful imaging technologies have been developed that permit the examination of the working brain. Technologies for postmortem examination of the brain have steadily improved. Genetic information will markedly enhance such approaches, at a minimum by decreasing the heterogeneity of populations studied while providing tools for molecular and cellular studies of pathogenesis. Based on experience in other areas of medicine, it appears absolutely necessary to use genetics to identify the genes that produce risk of mental disorders if we are to ultimately understand the disease processes. Beyond insights into the pathogenesis of mental disorders, gene discovery will undoubtedly provide important insights into normal brain function as well. Moreover, discovery of risk genes should hasten the development of treatments that are not simply aimed at symptoms, but that alter the disease process.
Why is it that gene discovery is so important and why has the task been so difficult? The importance of genes as tools of investigation lies in the complexity of what it is that we are trying to understand. The brain is the most complex object in the history of human scientific inquiry and mental disorders represent problems with the highest integrated functioning of the human brain—thinking, emotion, and behavioral control. To succeed in understanding, we will need both “top down” approaches (moving from epidemiology to neurocircuitry systems on to genes) and “bottom up” approaches (moving from genes and proteins to cell biology to neurocircuitry to pathophysiology) and we will need to find ways of ultimately combining these approaches.
While the precise definition of a “gene” can be arguable, in the simplest terms a gene is considered to be a stretch of DNA that encodes one (or a family of related) RNAs and ultimately their protein
products. Genes provide not only the sequences of possible proteins that can be synthesized but also information about where, when, and under what circumstances these proteins will be made; proteins, in turn, are the building blocks of cells. Such regulatory information is far less well understood than information that codes for RNAs. We are still learning to identify stretches of DNA involved in the regulation of gene expression. Important clues are coming from the sequencing of mammalian species other than humans (eg, the recently completed mouse sequence). DNA regions that are conserved by evolution across species are likely to be functionally important, and these will include regulatory regions.
Disease risk genes can be defined as genes that contain a variation in DNA sequence that leads a change in location, timing, levels of gene expression, or the nature of the RNA or protein encoded by the gene, such that it contributes to risk of mental illness. Genetic variations may take the form of “mutations,” which are clearly deleterious errors (eg, an error that would truncate a protein or block its expression altogether) or polymorphisms, which might produce slight alterations in expression pattern or function, but which cannot be said to be clearly abnormal. The most common type of variation in the human genome is the substitution of a single nucleotide base for another; such variations are referred to as single nucleotide polymorphisms (SNPs).
Discovery of disease risk genes means that we will be able to ask fundamental pathophysiologic questions. We hope to be in a position to ask how one version of a gene (allele) confers vulnerability toward or protection against disorders such as manic depressive illness or schizophrenia, while a slightly different version of the same gene does not. Additionally, by determining at what point during brain development a relevant gene is activated, we will be able to detect the earliest moments during which normal development gets diverted to result in such illnesses as the autism spectrum disorder or perhaps schizophrenia. Most importantly, protein products of gene expression function in complex biochemical networks that subserve all cellular functions. Identification of genes that confer risk of mental illness will point us toward biochemical pathways that, in turn, may suggest entirely novel treatments or even preventive interventions for mental illness.
We are painfully aware that the majority of drug-based therapeutics currently used to treat mental disorders was discovered serendipitously while being used to treat other medical conditions. For too long we have relied on the clever exploitation of these findings which led to the use of initial reference compounds, such as chlorpromazine or imipramine, from which our current treatments evolved. Today, we sorely need pathophysiologically-based pharmaceuticals that are safer and more efficacious for the treatment of illnesses such as schizophrenia, manic depressive illness, major depression and a host of anxiety disorders. Our field needs pathophysiologically relevant protein targets for drug development. As discoveries are made regarding the identity of the genes and protein products involved in specific disorders, the development of truly novel pharmacologic compounds is increasingly likely.
Challenges to Identification of Mental Illness Genes
Why has the discovery of disease risk genes for mental illness been so difficult? Unlike the situation for diseases such cystic fibrosis or Huntington’s disease, both of which are caused by a single gene in either a Mendelian dominant or recessive pattern, family and genetic linkage studies of mental illnesses have demonstrated that the pattern of inheritance of risk for these disorders is highly complex, comprised of both genetic and environmental components. Based upon Lander and Kruglyak’s1 seminal paper establishing guidelines for determining the statistical significance of linkages associated with disease, evidence indicates that most psychiatric disorders are polygenic. One case in point, Botstein and colleagues2 excluded the possibility that bipolar illness is the result of a monogenic or even bigenic disorder based upon results from a full-genome scan of linkages associated with bipolar disorder.
Additional challenges in tackling the genetics of complex psychiatric disorders lies in the realization that no specific gene or group of genes is sufficient or necessary for producing a mental illness, clearly unlike in the cases of cystic fibrosis or Huntington’s disease. In other words, it may be that there are multiple genetic pathways leading to similar disease
phenotypes. Each disease risk gene may contribute a small increment of risk and some may have no role at all in subsets of families. Thus the “signal” provided by risk genes may be very small and technically difficult to discern. Or, it is possible that there is a single dominant gene (or group of genes) which confer most of the risk to a given men
tal illness, but because there may be several such dominant-like genes in the population, it is statistically difficult to identify them.
In addition to polygenic and multiple small-effect genes, the complexity of deciphering contributors to mental illness is further complicated by “epigenetics”—the heritable regulation of gene expression or activity that does not involve changes in DNA sequence. For instance, it is well recognized that methylation (which typically silences but occasionally activates gene expression) of specific DNA sequences is frequently based upon whether the sequence is of maternal or paternal allelic origin.3 It is not yet known what confers these heritable modifications, but they ultimately result in functional differences during development. For example, the regulation of X chromosome inactivation in mammalian females (XX) is a classic example of epigenetics.4
As prefaced earlier, multiple gene-gene and gene-environment interactions ultimately coalesce to produce genetically complex phenotypes.5,6 A clear role for the environment in eliciting one particular neuropsychiatric brain disease was brought to the foreground when individuals taking an illicit drug containing 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) acquired phenotypic symptoms and brain pathology indistinguishable from Parkinson’s disease.7 In general, it appears that in certain psychiatric illnesses, such as bipolar disorder and schizophrenia, genetics plays a greater role in pathophysiology, whereas in others, such as depression and anxiety disorders, the environment is thought to have a greater modulatory role. Regardless, the role of the environment is indisputable. For example, despite sharing 100% of their DNA, monozygotic twins do not exhibit 100% concordance for schizophrenia,8 manic-depressive illness, or any other mental disorder. Furthermore, findings from adoption studies show that the biological relatives of adopted children who are affected with mental illness are more likely to suffer from the mental illness than are the foster parents or foster family members.9,10
Gene hunting in psychiatry has been made yet more difficult as a result of a few additional factors. First, there are no laboratory tests for which a mental disorder phenotype is unambiguously determined. The diagnoses of mental disorders relies on behavioral and phenotypic diagnostic criteria as listed in the Diagnostic and Statistical Manuals of Mental Disorders, Fourth Edition11 (DSM-IV), or the International Classification of Diseases12 (ICD). Second, several illnesses can have overlapping features. For instance, a subset of patients diagnosed with manic-depressive illness or schizophrenia both present with psychotic symptoms, and the distinction between schizophrenia and schizophreniform disorders, for example, is often subtle and uncertain.
The definitive discrimination of many mental disorders may not be possible until their genetic basis is known as was the case with the spinocerebellar ataxia (SCA) disorders, many of which clinically present with similar behavioral and phenotypic features. The identification of a new mechanism of pathogenesis, via genes containing expanded triple repeats, led to the discovery of the independent etiologies responsible for each ataxic disorder and a genetic basis for differentiating between them. Finally, mental disorders are extremely common. The prevalence of major depression, for example, may exceed 10% in some populations. But, associating depression with specific alleles—even those that are retained at high frequency in the population—is not straightforward, particularly because they are likely to represent simple single base changes (polymorphisms) and not deleterious mutations such as easily recognizable deletions of large segments of DNA. Additionally, the polymorphisms, or stretches thereof, may represent silent variants on their own, but other certain infelicitous combinations may produce vulnerability to illness.
Despite these difficulties, as we are lacking a deep understanding of the pathophysiology of mental disorders, we must rely on reverse genetics or positional cloning approaches to find genes. These terms refer to approaches to gene identification which assume no prior knowledge of that disease mechanism but rather are based on relating the transmission of an illness to the presence of identified chromosomal segments. For Mendelian or single-gene disorders, such as cystic fibrosis or Huntington’s disease, the identification and cloning of disease risk genes is now relatively straightforward. For complex traits such as mental disorders, the approach is still extraordinarily difficult and there is no complex phenotype that has been fully dissected genetically.
Past Efforts to Identify Genes
Many reverse genetic studies have been performed using multiple strategies in populations with mental disorders. The most numerous of these have been for schizophrenia, manic-depressive illness and, more recently, autism. The major strategy employed is called “linkage analysis,” whereby DNA from families having a high number of affected individuals is examined in order to identify genetic markers (indicative of specific chromosomal regions) that segregate in affected members. In other words, since genes or DNA sequences at a specific location tend to be inherited together due to their proximity, the linkage of a genetic marker with a disease is indicative of a causative gene in close proximity to chromosomal marker. This linkage approach has worked extremely well for Mendelian disorders in which a given locus is fully responsible for an illness, but such linkage approaches have not had the resolving power to find loci that confer risk of complex phenotypes such as mental illness.
Attempts have been made to avoid the effort required for reverse genetic approaches by employing a candidate gene approach whereby genes that are likely to be involved in a disorder are manipulated and studied in various experimental systems. For example, a candidate gene can be knocked out or overexpressed in a transgenic mouse, its protein can be
pharmacologically manipulated up or down, and surgical “treatments” can ablate specific neuroanatomical structures and neurocircuitry connections thought to be pertinent in the disorder. Unfortunately, it is still too early in our understanding of the pathophysiology of mental illness to be in possession of many compelling candidates. Thus, almost all of the candidates that have been investigated are derived from the putative mechanism of action of therapeutic agents rather than from any real knowledge of pathophysiology.
Most efforts to date have been directed toward genes responsible for the biosynthesis, release, or reuptake of serotonin, dopamine, or other neurotransmitters. In addition, many of these association studies have not achieved adequate statistical power to be decisive. As a result, we remain in limbo regarding claims about any certain associations.
The Genomic Revolution and Major Developments
The genomic revolution should markedly aid in determining the genetic liability of risk for developing mental disorders and other complex phenotypes, ergo, the rest of this essay will focus more specifically on psychiatry in the postgenomic era. The near completion of the full sequencing of the human genome was published for the record and comparison by a large public and private effort.13,14 The sequencing of the 3 billion base pair human genome not only provides a referenced sequence, but in sequencing chromosomes from many individuals, we have also derived maps of human diversity. Here we will focus on two major developments.
First, it has been recognized that there is a most common type of variation in the genome, which, as described above is the single nucleotide polymorphism or SNP. At the time of this writing, 3 million SNPs, or one in approximately every 1,000 bases, have been identified and many o
f them have had their precise location within the genome determined.15,16 Thus, in theory, the collection of SNPs provides literally millions of markers throughout the genome. Traditionally, human
disease has been attributed to mutations in specific genes, but now, with the recognition that humans have sequence variations at specific chromosomal loci, we are provided with new insight into how specific sequence variations in a gene or genes may make us more vulnerable to, or protected from, disease and mental illness.Furthermore, other discoveries have made SNPs even more useful. As we have learned, in any human population SNPs may be inherited together representing the descent of ancestral chromosomes leading to a linkage disequilibrium (LD) at certain chromosomal locations.17 Because humanity is a young species having radiated out of east Africa perhaps only 5,000 generations ago, human genetic diversity is not so great as to have permitted time for an enormous number of new mutations and recombinations. Every extended human family has new mutations that have occurred within the last 5,000 generations (indeed, such new mutations are almost certainly responsible for most of the Mendelian disorders such as Huntington’s disease). However, the most common variations in the human genome seen in about 85% of cases to be shared across all ethnic groups and are probably a part of the human genetic heritage since the time that humans radiated out of Africa. In addition, because of population bottlenecks, some groups have relatively no levels of genetic diversity. Thus, for example, northern Europeans (perhaps as a result of the ice age), seem to have large blocks of ancestral chromosomes resulting in LD with other population samples.
The direct outcome of these and other SNP-related discoveries is the development of new research methods of finding disease-related genes. A mixture of SNPs traveling together may be described as an LD block or a haplotype block having “stretches” of polymorphisms that seem to occur in a limited number of heritable combinations. This sets the stage for haplotype mapping which is based on the discovery that some long stretches of DNA, comprising some 50,000–100,000 contiguous bases, have very few variations in SNP patterns, indicating that human genetic diversity is not only a function of individual SNPs, but also of the combination of alleles (haplotypes) at a given region.16 Therefore, rather than attending to each SNP individually to address whether or not it has a role in a disease, much larger chunks of DNA can be analyzed for variations in SNP patterns and linkage to disease. Rather than analyzing the role of each of 50 SNPs in 50,000 bases of DNA (approximately 1 per 1,000 bases), researchers would only have to analyze, perhaps, five SNP patterns that occur in those 50,000 bases and the relationship of the SNP pattern to disease.
If there should turn out to be a significant number of haplotype blocks in the human genome, each of significant length (or 50,000–100,00 contiguous bases), this approach would substantially reduce the amount of DNA needed to be “covered” in order to find disease-related genes. Maps of linkage disequilibrium among different human populations will provide us with extraordinarily powerful tools with which we can trace ancestral chromosomal segments as they are transmitted from parents to children. Such tools can be used to supplement current linkage analysis studies, or, alternatively, to perform independent whole genome association studies in which the question is asked whether a particular gene associated with a certain linkage disequilibrium block or chromosomal segment is marked by certain SNPs. Ultimately, the fruits of the genome project will give us tools of unprecedented power to identify disease genes.
New Tools for Pathophysiology
In addition to permitting gene finding, genomic tools will hasten our understanding of both pathophysiology and drug action. It is now a well-established fact that the genome does not simply set in motion processes of development and then become inert. Rather, in every cell the genome is a repository of information for appropriate responses to the environment. In the brain, these responses include not only adaptations to stresses, drugs, illness, and injury, but also they permit learning and memory.18
Specifically, current models of learning and memory suggest that memories are stored by the remodeling of synapses and related neural circuitry. This remodeling depends on the neuronal transmission of chemical and electrical signals that
ultimately activate certain genes and suppress others so that proteins are made which will appropriately remodel the nervous system. New powerful techniques based upon the completion of the human genome sequence will more readily
allow for the rapid identification of genes critical to brain remodeling in response to its environment.
More specifically, whole new fields have arisen in functional genomics and proteomics which have the power of looking in aggregate at the genes that are activated or suppressed and the proteins that are expressed or modified in response to mental stimuli, disease, or other physiologic probes. As we collect the entire complement of expressed genes—DNA segments that get transcribed into RNA in cells—we are increasingly able to create microarrays that contain these segments of DNA spotted or etched onto glass slides or other positive supports which can then be used as tools of inquiry. Rather than studying one or a few genes at a time, we can look at expression levels of thousands of genes in a given condition. For example, it will be possible to obtain a specific “DNA chip” representing a desired brain region or even individual cell type. Investigators will be able to test what genes are activated or suppressed in normal as compared with diseased tissue, in control as compared with drug-treated tissue, and in “young” as compared with “old” tissue.
While it is true that there is a long way to go before perfection of these technologies (especially in proteomics), we already have early and exciting examples relevant to psychiatric research, particularly in the area of schizophrenia. For example, Mirnics and coworkers19,20 compared the gene expression profiles from the prefrontal cortex brain region of
postmortem schizophrenic and “normal” subjects and (based on more than 7,000 partial DNA sequences) found that schizophrenia may be a disease of the synapse because transcripts that encode proteins involved in synaptic function are decreased in patients as compared with normal individuals. Follow-up experiments from studies such as this one
highlight the “bottom-up” approach to understanding mental illness whereby data from cDNA microarray analysis
leads to identification of specific disease-regulated genes, the related protein functions and its role in the nervous system, and, finally, in disease. Our previous inability to probe all the regulated genes and biochemical pathways
will no longer be a major issue. We will literally have the universe of possibilities or candidates for these physiologic responses before us to facilitate subsequent biological investigations.
Thus far, we have reviewed technological developments that are the most potentially useful in the study of mental disorders. We have also highlighted the challenges that still lie ahead due to the great difficulty of specifically identifying genetic and environmental contributors to multifactorial disorders such as mental illnesses. Thus, despite the excitement, we must recognize that progress will remain hard won and will take time. On the other hand, we are finally increasingly in possession of the tool kit that will permit true progress after a period of relative stagnation in understanding causes of mental disorders. However, this does not imply that with enough time and hard w
ork all will be well. There will be critical conceptual difficulties and none are more important than readdressing the phenotypes of mental disorders. The ability of genomic tools to find the appropriate disease-related gene(s) is limited by the “quality” or homogeneity of the phenotypic sample. For instance, the definition of someone affected by schizophrenia or manic depressive illness, is used in the collection of DNA. The current (DSM-IV) and ICD classifications, not to be belittled, reflect a substantial improvement in communication about patients, their prognosis, and their treatment. On the other hand, there has been a worrisome and highly problematic reification of these diagnostic categories that, if anything, has served to stymie scientific progress, not only in genetics but also in critical neuroscience and behavioral studies of mental disorders. Thus, the complex ideologies of these disease phenotypes remind us that these illnesses will remain fuzzy. There will be a somewhat circular process of understanding phenotype as we gain a better understanding of genotype; this, in turn, will affect our understanding of phenotype. All of this circularity may seem unsettling and unsatisfying to philosophical purists and it is difficult to see any way out of a process of constant adjustment. However, in the meantime, it is critical that we collect broad and thoughtful phenotypic information and not be handcuffed by diagnostic criterion sets that have reliability as their strong suit but were never meant to represent valid diagnostic entities.
The road to fully understanding how genes contribute to psychiatric disorders will be long and difficult. In addition, psychiatry cannot succeed unless it recruits a generation of new investigators deeply steeped in the most advanced genomic technologies. If, however, we fully avail ourselves of the tools that rest on the foundation of the genome project and proceed with open minds, we will ultimately gain information of extraordinary value for mentally ill patients.
1 Lander E, Kruglyak L. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet. 1995;11:241-247.
2. Friddle C, Koskela R, Ranade K, et al. Full-genome scan for linkage in 50 families segregating the bipolar affective disease phenotype. Am J Hum Genet. 2000;66:205-215.
3. Ferguson-Smith AC, Surani MA. Imprinting and the epigenetic asymmetry between parental genomes. Science. 2001;293:1086-1089.
4. Park Y, Kuroda MI. Epigenetic aspects of X-chromosome dosage compensation. Science. 2001;293:1083-1085.
5. Lander ES, Schork NJ. Genetic dissection of complex traits. Science. 1994;265:2037-2048.
6. Frankel WN, Schork NJ. Who’s afraid of epistasis? Nat Genet. 1996;14:371-373.
7. Langston JW, Ballard P, Tetrud JW, Irwin I. Chronic Parkinsonism in humans due to a product of meperidine-analog synthesis. Science. 1983;219:979-980.
8. Gottesman II. Schizophrenia Genesis: The Origins of Madness. New York, NY: Freeman; 1991.
9. Kallman FJ. The Genetics of Schizophrenia. Locust Valley, NY: J.J. Augustin; 1938.
10. Heston LL. The genetics of schizophrenic and schizoid disease. Science. 1970;167:249-256.
11. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.
12. The ICD-10 Classificatin of Mental and Behavioral Disorders. Geneva, Switzerland: World Health Organization; 1992.
13. Lander ES, Linton LM, Birren B, et al. Initial sequencing and analysis of the human genome. Nature. 2001;409:860-921.
14. Venter JC, Adams MD, Myers EW, et al. The sequence of the human genome. Science. 2001;291:1304-1351.
15. Gura T. Genetics. Can SNPs deliver on susceptibility genes? Science. 2001;293:593-595.
16. Sachidanandam R, Weissman D, Schmidt SC, et al. A map of human genome sequence variation containing 1.42 million single nucleotide polymorphisms. Nature. 2001;409:928-933.
17. Reich DE, Cargill M, Bolk S, et al. Linkage disequilibrium in the human genome. Nature. 2001;411:199-204.
18. Berke JD, Hyman SE. Addiction, dopamine, and the molecular mechanisms of memory. Neuron. 2000;25:515-532.
19. Mirnics K, Middleton FA, Lewis DA, Levitt P. Analysis of complex brain disorders with gene expression microarrays: schizophrenia as a disease of the synapse. Trends Neurosci. 2001;24:479-486.
20. Mirnics K, Middleton FA, Marquez A, Lewis DA, Levitt P. Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex. Neuron. 2000;28:53-67.