Functional Genomics of Neural and Behavioral Plasticity Hans A. Hofmann Harvard University, Bauer Center for Genomics Research, 7 Divinity Ave., Cambridge, Massachusetts 02138 ABSTRACT: How does the environment, particu- accessible at the physiological and molecular level. In larly the social environment, influence brain and behav- this article, I explore how the new field of functional ior and what are the underlying physiologic, molecular, genomics can contribute to an understanding of the and genetic mechanisms? Adaptations of brain and be- complex relationship between genome and environment havior to changes in the social or physical environment that results in highly plastic phenotypes. This approach are common in the animal world, either as short-term will lead to the discovery of genes under environmental (i.e., modulatory) or as long-term modifications (e.g., via control and provide the basis for the study of the inter- gene expression changes) in behavioral or physiologic relationship between an individual’s gene expression properties. The study of the mechanisms and constraints profile and its social phenotype in a given environmental underlying these dynamic changes requires model sys- context. © 2003 Wiley Periodicals, Inc. J Neurobiol 54: 272–282, 2003 tems that offer plastic phenotypes as well as a sufficient Keywords: neural and behavioral plasticity; genomics; level of quantifiable behavioral complexity while being brain and behavior INTRODUCTION common diseases) requires the dissection and analysis of complex phenotypes or traits. Complex traits are How nature and nurture contribute to who we are, determined by many factors (genetic, epigenetic, and what we do, and why we do it, has long been debated environmental) whose interactions are nonlinear and by scientists and philosophers alike. We now know often unpredictable. As a consequence, the genetic that there are biologic roots to our behavior. In fact, architecture of a complex trait cannot be derived from behavioral genetics has provided us with a wealth of the individual effects of each of the component factors information regarding the roles of individual genes in alone, however well studied they are. It is in principle the implementation of behavior. Yet behavior (just possible to define the genetic components of a com- like any other phenotypic trait) is commonly influ- plex trait in terms of Mendelian segregation and lo- enced by both environmental and epigenetic factors cation along a genetic map. It is, however, crucial to (see Pigliucci, 2001). Because of this intricate rela- recognize that the genetic architecture is not so much tionship between genes and environment, we can dis- a fundamental biological attribute of a trait as it is a tinguish four major challenges facing the genetic characteristic of a trait in a particular population de- study of behavior that have to be addressed before we pendent on gene and genotype frequencies, the distri- will gain a better understanding of the genetic mech- butions of environmental factors, age and sex, etc. anisms underlying behavior. For example, in typical, late-onset Parkinson Dis- (1) First, the genetics of behavior (and of most ease, a neurodegenerative disorder that is associated with a reduction of dopaminergic activity in the sub- stantia nigra, tracking down disease-causing genes Correspondence to: H. A. Hofmann ([email protected]). Contract grant sponsor: the Bauer Center for Genomics Re- has been elusive. However, the existence of “suscep- search. tibility” genes has been suggested (Martin et al., Contract grant sponsor: NSF; contract grant number: 0217915. 2001), particularly in the context of the alleged dis- © 2003 Wiley Periodicals, Inc. Published online in Wiley InterScience (www.interscience.wiley.com). ease-causing role of environmental toxins (Checko- DOI 10.1002/neu.10172 way and Nelson, 1999). Several twin studies have so 272 Genomics and Plasticity 273 far been unable to identify genetic factors (Ward et gion of the monoamine oxidase A gene is correlated al., 1983; Tanner et al., 1999). Surprisingly, patients with violent behavior in adult males, but only in men with Parkinson often have peculiar premorbid person- who were maltreated as children. ality traits (e.g., industriousness, punctuality, lack of In a different study, Sillaber et al. (2002) showed novelty seeking, low life-time risk for cigarette smok- that mice lacking the corticotropin-releasing hormone ing) whose relevance for the disease onset is entirely receptor (CRH1-R) show enhanced and persistent al- unclear (Menza, 2000). cohol consumption (compared with wild-type con- Clearly, if many genes contribute to a given com- trols) only after stressful experiences such as social plex phenotype, it can become difficult to dissect its defeat. Also in mice, social stress exacerbates stroke genetic basis, even when a few candidate genes have outcome by suppressing expression of Bcl-2, a proto- been identified (Tabor et al., 2002). Although one oncogene that promotes cell survival and protects particular gene may contribute significantly to the trait against cell-death (De Vries et al., 2001). (within a given genetic and environmental back- Finally, the increase in mortality (mostly through ground), one often can only estimate how many other coronary heart disease) in Eastern Europe after the loci are involved. To solve this problem, researchers dissolution of the Soviet Union is a sobering example have developed a range of strategies, the most impor- of G ϫ E interactions. Psychosocial stressors as a tant of which is the mapping of quantitative trait loci consequence of societal uncertainty are thought to be (QTL) to defined chromosomal locations along the the primary cause of this unexpected change (Stone, genome (see Flint, this issue). However, it is often 2000). challenging to make the connection between a given Many studies have correlated differences in behav- marker (that represents a QTL on the genomic map) ioral phenotypes with differences in the nervous sys- and the actual gene (or group of genes) that participate tem. Often, the conclusion has been that the neural in the quantitative trait under study. differences are causally responsible for the behavioral (2) Another problem facing the genetics of com- differences observed. For example, LeVay (1991) re- plex traits arises due to the fact that the contribution ported in a widely cited article that the volume of a of genes may be epistatic rather than additive nucleus in the anterior hypothalamus was larger in (Wade, 2001; Wade et al., 2001). Behavioral ge- heterosexual than homosexual men. Many interpreted neticists encounter epistasis (gene– gene interac- this result as evidence that differences in the brain tions) most often as differences in phenotype that caused the observed differences in sexual behavior, relate to the genetic background or “modifier possibly reflecting the influence of genetic factors genes” of the animal strain used (Gerlai, 2001; during development. However, until cause and effect Nadeau, 2001). These gene interactions are nonlin- have been identified, the alternative explanation that ear and intricate, which greatly complicates the these differences were a consequence of years of identification and mapping of genes underlying a differential sexual behavior is just as valid. In this trait. Similarly, many gene-targeting studies (where context it is interesting to note that in adult male rats one particular gene was “knocked out” in a trans- differences in sexual experience lead to differences in genic animal) have shown that phenotypes can dif- motor neuron size (Breedlove, 1997). It is important fer greatly depending on genetic background and to keep in mind that both genotype and environment “compensatory mechanisms” (Gerlai, 2001). contribute to the phenotype. (3) The relationship between genotype and envi- (4) Finally, means of inheritance exist that are not ronment (G ϫ E) presents a third problem confound- dependent on DNA. During gametogenesis, epige- ing the study of complex behavior. In most behavioral netic modifications of the genome occur (i.e., genetic experiments environmental conditions are genomic imprinting), which can lead to profound be- kept constant (for a discussion of standardization see havioral differences in the offspring (Li et al., 1999; Wu¨rbel, 2002; van der Staay and Steckler, 2002). Keverne, 2001). Furthermore, maternal factors in the However, it has become increasingly clear that com- egg (Davdison, 1986), maternal care (Meaney, 2001), plex and nonlinear G ϫ E interactions exist (Pigliucci, as well as traditions (mediated by social learning; 2001; Sokolowski & Wahlsten, 2001; Wade, 2001). A Avital and Jablonka, 2000) can contribute to the trans- striking example of the complex interplay between mission of neural and behavioral phenotypes. As a genotype and social environment has been presented consequence, a purely gene-based approach to the recently. Caspi et al. (2002) showed in a human dissection of complex phenotypes may overlook im- population that a polymorphism in the promoter re- portant factors of inheritance for a given trait. 274 Hofmann Figure 1 A concept map of phenotypic plasticity as it applies to brain and behavior. See explanation in text. FUNCTIONAL GENOMICS OF NEURAL processes and time scales where plasticity can be AND BEHAVIORAL PLASTICITY adaptive. In cases where the frequency of the envi- ronmental change (the stimulus) is high or when im- In this essay I will introduce a conceptual framework