Psychophysiology, 51 (2014), 1203–1204. Wiley Periodicals, Inc. Printed in the USA. Copyright © 2014 Society for Psychophysiological Research DOI: 10.1111/psyp.12341
PERSPECTIVE
Molecular genetic psychophysiology: A perspective on the Minnesota contribution
ECO J. C. DE GEUSa,b a
Department of Biological Psychology, VU University, Amsterdam, The Netherlands Neuroscience Campus Amsterdam & EMGO+ Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
b
Abstract Progress in molecular genetic psychophysiology, on display in this special issue, shows that the affective and cognitive processes tagged by psychophysiological endophenotypes are highly polygenic. Identifying overlap in the genetic variants that influence these endophenotypes as well as behavioral traits can help us understand where in the brain, in which stage, and during what type of information processing these variants play a role in normal and abnormal behavior. The Minnesota Twin Family Study has demonstrated that the assessment of genetic markers and extensive psychophysiological experimentation can be done at a genetic epidemiological scale, that is, in thousands of subjects. A genome-wide association meta-analysis consortium consisting of psychophysiological research teams following the Minnesotan example is the obvious next step. Descriptors: Endophenotypes, Genetic association, Meta-analysis, Functional annotation
polygenic inheritance for their “constituent” endophenotypes like body mass index, cholesterol, and glucose levels, heart rate, and blood pressure, or IL6 levels (Visscher, Brown, McCarthy, & Yang, 2012; Welter et al., 2014). Thousands of genetic variants with very tiny effect sizes contribute to the heritability of these endophenotypes. I suspected a similar polygenic inheritance was likely to apply to the brain-related traits populating the pages of the Psychophysiology journal. Sample sizes would have to be ramped up to tens of thousands of subjects. These numbers led me to some skepticism about the usefulness of psychophysiological measures in the discovery phase of gene finding, as they require prolonged presence of subjects in controlled laboratory experiments with sophisticated and rather expensive equipment often followed by labor-intensive data inspection and analytic efforts (de Geus, 2010). After reading the papers in this special issue, I gladly stand corrected. The Minnesota Twin Family Study (MTFS) provides astute proof of the principle that measuring skin conductance, electroencephalogram (EEG), and eye and eyelid movement can be done in thousands of subjects in typical psychophysiological experimental settings, while at the same time enriching the sample with tens of millions of genetic markers that allow the full breadth of current molecular genetic strategies, including single nucleotide polymorphism (SNP) and gene-based association analysis and genome-wide complex trait analysis (GCTA). Admittedly, the net yield of genome-wide significant findings in the MTFS appears modest. With one or two exceptions, SNPs did not make it to genome-wide significance for any of the 17 endophenotypes, and gene-based tests fared only slightly better. This is not alarming. Null findings are in fact the norm rather than the exception in
About two decades ago, we made a plea for genetic psychophysiology—a new branch of behavioral genetics aiming to understand the heritability of normal and deviant behavior (Boomsma, Anokhin, & de Geus, 1997). Various twin registries, prominently including the Minnesota Center for Twin Family Research (MCTFR), had started revealing that a large part of the stable individual differences in psychophysiological traits derived from genetic sources (Iacono, 1983; Katsanis, Iacono, McGue, & Carlson, 1997). Furthermore, a reasonable assumption was that the genetic architecture of the affective and cognitive processes tagged by psychophysiological measures would be less complex than that of the highly complex behavioral traits, so that a smaller number of genetic variants of large effect might be involved in these measures. If true, psychophysiology could be used to boost the statistical power of the gene-hunting efforts ushered by the completion of the Human Genome Project. A specific hope was that, using wellpicked endophenotypes, samples of hundreds to thousands would suffice to detect genetic variants through linkage or association mapping. Much larger samples would be needed when using the actual behavioral outcomes or disease end points themselves. As a longstanding advocate for psychophysiological endophenotypes, my faith had started to waver over the past years. In particular, I struggled with the assumption of a simple genetic architecture of the endophenotypes. Genome-wide association (GWA) showed almost all complex disease traits to be of a highly polygenic nature, but more importantly also showed highly
bs_bs_banner
Address correspondence to: Eco de Geus, PhD, Department of Biological Psychology, VU University Amsterdam, van der Boechorststraat 1, 1081 BT, Amsterdam, The Netherlands. E-mail: [email protected] 1203
1204
E.J.C. de Geus
individual GWA studies, and the authors prudently refrained from interpreting findings as meaningful if they did not survive very stringent levels of correction for multiple testing. When GCTA was used to quantify the sum of these thousands of very small effects, as tagged by the SNPs on the Illumina 660-W Quad platform, more than half of the twin heritability for occipital alpha power, alpha frequency, all four electrodermal measures, and overall startle could be explained by the SNPs hiding in the underside and middle parts of the “buildings” constituting the Manhattan plots. All of this simply reinforces the need to form psychophysiological GWA meta-analysis consortia. A meta-analysis of perhaps even as few as 5 or 10 cohorts like the MTFS could already aspire to genome-wide significance. GWA meta-analysis really works. By now, nearly 9,000 SNPs have been identified by GWA meta-analysis consortia for a variety of traits and clinical end points (Welter et al., 2014). The glory and sacrifice in such GWA metaanalysis consortia cannot be emphasized sufficiently. It is a testimony to science and civilization that so many researchers worldwide are willing to share their data, even when they see the work of months or years rewarded with position 567 on a 700+ authored paper, often compounded by collegial ridicule for being on papers with “more authors than words.” The latter is particularly unfair when coming from colleagues who favor genetics with biological candidates. In rather sharp contrast to many candidate gene findings, variants identified by GWA meta-analysis are virtually always real, survive replication, and lead to meaningful results in follow-up experimentation. It is this follow-up biological experimentation and annotation that give GWA meta-analysis findings their true value. The aim of gene hunting remains to generate testable hypotheses about the biological pathways shaping individual differences in behavioral traits and risk for somatic and mental disorders. This first requires functional annotation of lead SNPs in terms of their effects on protein function, the level of its expression, or its posttranslational
fate. However, the next crucial steps are to detect how these changes in protein function/abundance influence cognitive and affective brain processes underlying behavioral traits. This is exactly where progress in the genetics of psychophysiological endophenotypes can play a unique role. Overlap between the risk alleles detected using GWA on psychiatric symptoms and diagnoses with those influencing EEG and event-related potential endophenotypes can help us to understand where in the brain, in which stage, and during what type of information processing the genetic variant has a role. Such value can already be harvested from the resource created by the MCTFR who will deposit genetic and phenotypic data at the National Institutes of Health repositories (e.g., www .ncbi.nlm.nih.gov/gap). When genetic variants are associated in meta-analyses of GWAs on schizophrenia, attention deficit hyperactivity disorder (ADHD), or depression, it can be tested whether and in what direction they influence the 17 endophenotypes tested in the MCTFR. Moreover, in existing genotyped samples, results from the MCTFR can be used to compute polygenic risk scores for the endophenotypes that can be correlated to psychiatric end points. For instance, without measuring EEG themselves, researchers with genetic data and ADHD diagnoses could compute a polygenic risk score for low theta power based on the MCTFR findings and test whether it is associated with ADHD, which would suggest a shared genetic etiology for theta power and ADHD. In summary, the important message to be extracted from this special issue is that psychophysiological endophenotypes are amenable to modern molecular genetic investigation. The Minnesotans have set the standards, and created an exemplary resource by investing in enriching their psychophysiology with state-of-the-art genetic data. Now it is up to all of us who have collected the same or comparable psychophysiological endophenotypes to follow suit. Let there be strength in numbers.
References Boomsma, D., Anokhin, A., & de Geus, E. (1997). Genetics of electrophysiology: Linking genes, brain, and behavior. Current Directions in Psychological Science, 6, 106–110. de Geus, E. J. (2010). From genotype to EEG endophenotype: A route for post-genomic understanding of complex psychiatric disease? Genome Medicine, 2, 63. doi: 10.1186/gm184 Iacono, W. G. (1983). Young psychophysiologist award address, 1982. Psychophysiology and genetics: A key to psychopathology research. Psychophysiology, 20, 371–383.
Katsanis, J., Iacono, W. G., McGue, M. K., & Carlson, S. R. (1997). P300 event-related potential heritability in monozygotic and dizygotic twins. Psychophysiology, 34, 47–58. Visscher, P. M., Brown, M. A., McCarthy, M. I., & Yang, J. (2012). Five years of GWAS discovery. American Journal of Human Genetics, 90, 7–24. Welter, D., MacArthur, J., Morales, J., Burdett, T., Hall, P., Junkins, H. . . . Parkinson, H. (2014). The NHGRI GWAS catalog, a curated resource of SNP-trait associations. Nucleic Acids Research, 42, D1001–D1006.
PERSPECTIVE
Molecular genetic psychophysiology: A perspective on the Minnesota contribution
ECO J. C. DE GEUSa,b a
Department of Biological Psychology, VU University, Amsterdam, The Netherlands Neuroscience Campus Amsterdam & EMGO+ Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands
b
Abstract Progress in molecular genetic psychophysiology, on display in this special issue, shows that the affective and cognitive processes tagged by psychophysiological endophenotypes are highly polygenic. Identifying overlap in the genetic variants that influence these endophenotypes as well as behavioral traits can help us understand where in the brain, in which stage, and during what type of information processing these variants play a role in normal and abnormal behavior. The Minnesota Twin Family Study has demonstrated that the assessment of genetic markers and extensive psychophysiological experimentation can be done at a genetic epidemiological scale, that is, in thousands of subjects. A genome-wide association meta-analysis consortium consisting of psychophysiological research teams following the Minnesotan example is the obvious next step. Descriptors: Endophenotypes, Genetic association, Meta-analysis, Functional annotation
polygenic inheritance for their “constituent” endophenotypes like body mass index, cholesterol, and glucose levels, heart rate, and blood pressure, or IL6 levels (Visscher, Brown, McCarthy, & Yang, 2012; Welter et al., 2014). Thousands of genetic variants with very tiny effect sizes contribute to the heritability of these endophenotypes. I suspected a similar polygenic inheritance was likely to apply to the brain-related traits populating the pages of the Psychophysiology journal. Sample sizes would have to be ramped up to tens of thousands of subjects. These numbers led me to some skepticism about the usefulness of psychophysiological measures in the discovery phase of gene finding, as they require prolonged presence of subjects in controlled laboratory experiments with sophisticated and rather expensive equipment often followed by labor-intensive data inspection and analytic efforts (de Geus, 2010). After reading the papers in this special issue, I gladly stand corrected. The Minnesota Twin Family Study (MTFS) provides astute proof of the principle that measuring skin conductance, electroencephalogram (EEG), and eye and eyelid movement can be done in thousands of subjects in typical psychophysiological experimental settings, while at the same time enriching the sample with tens of millions of genetic markers that allow the full breadth of current molecular genetic strategies, including single nucleotide polymorphism (SNP) and gene-based association analysis and genome-wide complex trait analysis (GCTA). Admittedly, the net yield of genome-wide significant findings in the MTFS appears modest. With one or two exceptions, SNPs did not make it to genome-wide significance for any of the 17 endophenotypes, and gene-based tests fared only slightly better. This is not alarming. Null findings are in fact the norm rather than the exception in
About two decades ago, we made a plea for genetic psychophysiology—a new branch of behavioral genetics aiming to understand the heritability of normal and deviant behavior (Boomsma, Anokhin, & de Geus, 1997). Various twin registries, prominently including the Minnesota Center for Twin Family Research (MCTFR), had started revealing that a large part of the stable individual differences in psychophysiological traits derived from genetic sources (Iacono, 1983; Katsanis, Iacono, McGue, & Carlson, 1997). Furthermore, a reasonable assumption was that the genetic architecture of the affective and cognitive processes tagged by psychophysiological measures would be less complex than that of the highly complex behavioral traits, so that a smaller number of genetic variants of large effect might be involved in these measures. If true, psychophysiology could be used to boost the statistical power of the gene-hunting efforts ushered by the completion of the Human Genome Project. A specific hope was that, using wellpicked endophenotypes, samples of hundreds to thousands would suffice to detect genetic variants through linkage or association mapping. Much larger samples would be needed when using the actual behavioral outcomes or disease end points themselves. As a longstanding advocate for psychophysiological endophenotypes, my faith had started to waver over the past years. In particular, I struggled with the assumption of a simple genetic architecture of the endophenotypes. Genome-wide association (GWA) showed almost all complex disease traits to be of a highly polygenic nature, but more importantly also showed highly
bs_bs_banner
Address correspondence to: Eco de Geus, PhD, Department of Biological Psychology, VU University Amsterdam, van der Boechorststraat 1, 1081 BT, Amsterdam, The Netherlands. E-mail: [email protected] 1203
1204
E.J.C. de Geus
individual GWA studies, and the authors prudently refrained from interpreting findings as meaningful if they did not survive very stringent levels of correction for multiple testing. When GCTA was used to quantify the sum of these thousands of very small effects, as tagged by the SNPs on the Illumina 660-W Quad platform, more than half of the twin heritability for occipital alpha power, alpha frequency, all four electrodermal measures, and overall startle could be explained by the SNPs hiding in the underside and middle parts of the “buildings” constituting the Manhattan plots. All of this simply reinforces the need to form psychophysiological GWA meta-analysis consortia. A meta-analysis of perhaps even as few as 5 or 10 cohorts like the MTFS could already aspire to genome-wide significance. GWA meta-analysis really works. By now, nearly 9,000 SNPs have been identified by GWA meta-analysis consortia for a variety of traits and clinical end points (Welter et al., 2014). The glory and sacrifice in such GWA metaanalysis consortia cannot be emphasized sufficiently. It is a testimony to science and civilization that so many researchers worldwide are willing to share their data, even when they see the work of months or years rewarded with position 567 on a 700+ authored paper, often compounded by collegial ridicule for being on papers with “more authors than words.” The latter is particularly unfair when coming from colleagues who favor genetics with biological candidates. In rather sharp contrast to many candidate gene findings, variants identified by GWA meta-analysis are virtually always real, survive replication, and lead to meaningful results in follow-up experimentation. It is this follow-up biological experimentation and annotation that give GWA meta-analysis findings their true value. The aim of gene hunting remains to generate testable hypotheses about the biological pathways shaping individual differences in behavioral traits and risk for somatic and mental disorders. This first requires functional annotation of lead SNPs in terms of their effects on protein function, the level of its expression, or its posttranslational
fate. However, the next crucial steps are to detect how these changes in protein function/abundance influence cognitive and affective brain processes underlying behavioral traits. This is exactly where progress in the genetics of psychophysiological endophenotypes can play a unique role. Overlap between the risk alleles detected using GWA on psychiatric symptoms and diagnoses with those influencing EEG and event-related potential endophenotypes can help us to understand where in the brain, in which stage, and during what type of information processing the genetic variant has a role. Such value can already be harvested from the resource created by the MCTFR who will deposit genetic and phenotypic data at the National Institutes of Health repositories (e.g., www .ncbi.nlm.nih.gov/gap). When genetic variants are associated in meta-analyses of GWAs on schizophrenia, attention deficit hyperactivity disorder (ADHD), or depression, it can be tested whether and in what direction they influence the 17 endophenotypes tested in the MCTFR. Moreover, in existing genotyped samples, results from the MCTFR can be used to compute polygenic risk scores for the endophenotypes that can be correlated to psychiatric end points. For instance, without measuring EEG themselves, researchers with genetic data and ADHD diagnoses could compute a polygenic risk score for low theta power based on the MCTFR findings and test whether it is associated with ADHD, which would suggest a shared genetic etiology for theta power and ADHD. In summary, the important message to be extracted from this special issue is that psychophysiological endophenotypes are amenable to modern molecular genetic investigation. The Minnesotans have set the standards, and created an exemplary resource by investing in enriching their psychophysiology with state-of-the-art genetic data. Now it is up to all of us who have collected the same or comparable psychophysiological endophenotypes to follow suit. Let there be strength in numbers.
References Boomsma, D., Anokhin, A., & de Geus, E. (1997). Genetics of electrophysiology: Linking genes, brain, and behavior. Current Directions in Psychological Science, 6, 106–110. de Geus, E. J. (2010). From genotype to EEG endophenotype: A route for post-genomic understanding of complex psychiatric disease? Genome Medicine, 2, 63. doi: 10.1186/gm184 Iacono, W. G. (1983). Young psychophysiologist award address, 1982. Psychophysiology and genetics: A key to psychopathology research. Psychophysiology, 20, 371–383.
Katsanis, J., Iacono, W. G., McGue, M. K., & Carlson, S. R. (1997). P300 event-related potential heritability in monozygotic and dizygotic twins. Psychophysiology, 34, 47–58. Visscher, P. M., Brown, M. A., McCarthy, M. I., & Yang, J. (2012). Five years of GWAS discovery. American Journal of Human Genetics, 90, 7–24. Welter, D., MacArthur, J., Morales, J., Burdett, T., Hall, P., Junkins, H. . . . Parkinson, H. (2014). The NHGRI GWAS catalog, a curated resource of SNP-trait associations. Nucleic Acids Research, 42, D1001–D1006.
