Behavioral Genomics

Traditionally, research efforts involving the potential impact of genes on inherited disorders have been advanced using the "candidate gene" approach, in which investigators ask whether a specific allele is more common in subjects who display a particular trait or behavior than in those subjects who do not (Fitzpatrick et al., 2005). For a gene to be considered as affecting behavior, some a priori knowledge is usually required, including either pathological mechanisms (in the case of disorders) or how a specific gene affects similar behaviors in other organisms. The latter information can sometimes be obtained from one or more of the numerous databases detailing the genomes of a variety of species, including humans, mice, chickens, fruit flies, honeybees, worms, and zebrafish.

After a gene is selected as a "candidate," its expression pattern (most often in the brain) in subjects who display the behavior of interest is compared to that in subjects who do not display the behavior. This comparison can be performed under natural conditions (i.e., without manipulation) in order to gather correlative data regarding the gene and the behavior. However, a more causal link can be established if expression of the gene is manipulated via transgenic or knock-in/knock-out techniques and the resulting behavior is observed. Indeed, these techniques have been essential in determining the genes associated with adaptive behaviors that are evolutionarily conserved in diverse species (e.g., foraging and reproduction), and they have also helped pinpoint genes associated with neurobehavioral disorders like autism.

Unfortunately, a candidate-gene-by-candidate-gene approach is often inefficient. Thus, investigators are increasingly turning to the enormous amount of data available through the genome sequencing of many organisms. Use of this information allows researchers to adopt a more unbiased approach, simultaneously examining up to thousands of genes at once. Genomic tools also allow scientists to associate genetic variants, including single nucleotide polymorphisms (SNPs), with specific behaviors or even with different gene expression patterns using microarray comparisons. Often, the genes identified through these large-scale genomic approaches are designated as candidate genes, and the effects of these candidates are then verified through more traditional approaches wherein individual genes and their functions are examined.

Male courtship in the fruit fly Drosophila melanogaster was one of the first behaviors for which researchers found strong evidence for specification by a gene—namely, the so-called "fruitless" gene (abbreviated fru). In male fruit flies, courtship is a complex, ritualistic behavior that involves many visual, olfactory, gustatory, tactile, and acoustic cues, as well as intricate motor output directed toward attracting a suitable mate (i.e., a female that has not recently mated). During courtship, the male fruit fly orients to the female, taps her with his forelegs, sings a courtship-specific song by vibrating one wing, probes the female's genitalia, and then curves his abdomen to mate (Figure 1; Hall, 1994).

The fruitless gene first emerged as a candidate gene for sexual behavior when it was discovered that male flies with fru mutations were sterile due to defective performance of courtship and copulation (Gill, 1963). Years later, the mutated gene present in these flies was cloned, and the protein that it encodes was identified as a transcription factor (Ito et al., 1996; Ryner et al., 1996). Researchers have also determined that the fru gene is spliced differentially in males and females and that this gene generates various sex-specific messenger RNAs (Demir & Dickson, 2005; Kyriacou, 2005). Interestingly, not only does the male pattern of splicing elicit male courtship behavior, but it also influences sexual orientation, as evidenced by the finding that male splicing in otherwise normal females generated male behavior (e.g., singing) that was directed toward other females (Figure 2; Demir & Dickson, 2005). Functional conservation of the fru gene has only been shown in one other species, the malaria mosquito Anopheles gambiae (Gailey et al., 2006), and a mammalian homologue for fru does not exist. Even so, fru is a good example of how genes can specify aspects of a complex inborn behavior.

Animal species vary greatly in the behaviors they use to obtain food; however, the genes that contribute to these actions have been highly conserved. In the fruit fly Drosophila melanogaster, foraging behavior is influenced by the activity of a certain enzyme called cGMP-dependant protein kinase (PKG). Fly larvae with high PKG levels typically have long food-finding paths outside of their food patch, and they are therefore called "rovers." The opposite is true of so-called "sitter" larvae, which have low PKG levels (Osborne et al., 1997). The rover and sitter traits are inherited as a single major gene called for, with a different allelic variant for each trait (for^R and for^S, respectively) that has multiple phenotypic effects and is influenced by minor genes (de Belle & Sokolowski, 1987; Sokolowski, 2001). However, both the genetic and phenotypic expression of for can be influenced by the environment. Specifically, for^R is more common in larvae from crowded environments, whereas for^S is predominant in larvae from sparsely populated surroundings (Sokolowski et al., 1997). Moreover, when food is scarce, even for a brief period, rover flies with the for^R allele can act like sitters (Sokolowski, 2001). Thus, both internal and external factors can influence foraging behavior in Drosophila.

Once the for gene was associated with foraging behavior in fruit flies, it was used as a candidate gene to study similar behaviors in other species. For example, the fruit fly for gene shows high sequence homology to the Amfor gene in honeybees. Changing Amfor expression can alter the foraging behavior of a honeybee, thereby transforming the bee from a "nurse" (a bee that cares for the young) to a "forager" (a bee that leaves the hive to find pollen and nectar). Like the rover fruit fly, the forager honeybee has higher levels of the PKG enzyme than do nurse bees (Ben-Shahar et al., 2002). Researchers have also noted that an orthologue of for, called egl-4, is associated with food-related behavior in the roundworm C. elegans (Fujiwara et al., 2002). Taken together, these findings suggest that for could be used as a candidate gene for behaviors concerning food acquisition in other species.

As previously discussed, one way to identify a candidate gene for a behavior is by comparing global expression patterns of the gene in animals that perform the behavior with patterns in animals that do not. This method led to the discovery of several genes that contribute to social behaviors, specifically affiliative, paternal, and promiscuous behaviors. Because these behaviors were linked to the neuropeptide vasopressin (AVP), genes related to AVP and/or its signaling were chosen as candidates for their control (Insel & Young, 2000).

Researchers began their inquiry into these genes using voles, which are small rodents that look like mice, because there are two particular vole species that differ greatly in terms of social behavior. Specifically, prairie voles (Microtus ochrogaster) are affiliative and monogamous (i.e., each vole has only one mate), and both parents provide care for their young, whereas montane voles (Microtus montanus) are comparatively asocial, polygamous (i.e., each vole has multiple mates), and nonpaternal. In male prairie voles, living and/or mating with a female causes AVP release in the brain, and paternal and pair-bonding behavior is triggered. However, these behaviors can be disrupted by blocking the V_1a receptor for AVP with an antagonist. Thus, comparing AVP receptor expression between these two vole species was proposed as a useful way to determine which genes contribute to social behavior.

Indeed, prairie and montane voles have markedly different patterns of binding to the V_1a receptors within their brains (Figure 3; Young et al., 1999). The polygamous montane vole has much more binding in the lateral septum of the brain than the prairie vole, whereas the monogamous prairie vole has a higher intensity of binding in the diagonal band of Broca, a brain area involved in appetitive reward (Young et al., 1999; Insel & Young, 2000). Furthermore, elevating the expression of the V_1a receptor in the brains of promiscuous voles increased the pair-bonding behavior of males to females (Lim et al., 2004). Thus, the social behavior of these two species of voles seems to be due to variations in gene expression of the V_1a receptor.

Autism is a severe neurobehavioral disorder characterized by impaired social skills and repetitive and stereotyped interests and behavior. Autism includes mental retardation in up to 70% of cases (Fombonne, 2003), and it affects many brain areas. Twin studies and examinations of family history suggest that the heritability of autism is roughly 90% (Freitag, 2007; Sutcliffe, 2008). Genetic aspects of the disorder have been substantiated by the identification of chromosomal translocations, inversions, and large deletions or duplications, which are more prevalent in individuals who have more severe cognitive problems. For example, several recent large-scale studies collectively screened almost 2,000 families with autism for recurrent copy number variations, which are small losses or gains in DNA. The consortium of researchers involved in these studies found that copy number variation in a gene called neurexin is linked to the presence of autism. The researchers also found a region on chromosome 11 that might contain sequences that are involved in the disorder (Autism Genome Project Consortium, 2007). Similarly, Weiss and colleagues (2008) found that a microdeletion and microduplication on chromosome 16p11.2 increased susceptibility for the disorder in 1% of cases. Other studies using DNA microarrays have identified many more copy number variations that can be sporadic or inherited and appear to be major genetic risk factors for autism in 10% to 20% of cases (Morrow et al., 2008; Sutcliffe, 2008).

Table 1 summarizes the genes that have been implicated in autism thus far (Sutcliffe, 2008). Although these genes are diverse, most researchers suggest that they may be linked by a common mechanism, namely dysregulation of gene expression that occurs after neural activity (Morrow et al., 2008). It is also evident that studying all of these genes, as well as the genes that they regulate and are regulated by, requires the large-scale analysis techniques employed in genomics.

Table 1: Putative and Known Autism-Related Genes

Genes Involved in Glutamatergic Synapse Function and/or Neuronal Cell Adhesion
FMR1A,B	Fragile X mental retardation 1
NLGN3B	Neuroligin 3
NLGN4B	Neuroligin 4
NRXN1B,C	Neurexin 1
SHANK3B,C	SH3 and multiple ankyrin repeat domains 3
CNTNAP2B,C,D	Contactin-associated protein-like 2
PCDH10C	Protocadherin 10
CNTN3C	Contactin 3
Genes Involved in Endosomal Trafficking
NHE9 (SLC9A9)B,C	Na+/H+ exchanger isoform 9
NHE6 (SLC9A6)B	Na+/H+ exchanger isoform 6
DIA1 (c3orf58)C	Deleted in autism 1
A2BP1C	Ataxin 2-binding protein 1
Genes Involved in Neuronal Activity Regulation
FMR1A,B	Fragile X mental retardation 1
MECP2B,C	Methyl CpG binding protein 2
DIA1 (c3orf58)C	Deleted in autism 1
PCDH10C	Protocadherin 10
NHE9 (SLC9A9)B,C	Na+/H+ exchanger isoform 9
A2BP1C	Ataxin 2-binding protein 1
UBE3AB,C	Ubiquitin protein ligase E3A
Genes Implicated in Related Disorders
FMR1A,B	Fragile X mental retardation 1
MECP2B,C	Methyl CpG binding protein 2
NHE6 (SLC9A6)C	Na+/H+ exchanger isoform 6
A2BP1C	Ataxin 2-binding protein 1
UBE3AB,C	Ubiquitin protein ligase E3A
Genes With Other Functions
EN2D	Engrailed homeobox 2
SLC6A4B,D	Serotonin transporter (SERT, 5-HTT)
METD	Met proto-oncogene (c-Met, HGFR)
SCN7AC	Na+ channel, voltage-gated, type VII
RNF8C	Ring finger protein 8

Table adapted from Sutcliffe, 2008

Schizophrenia is a neuropsychiatric illness in which individuals have impaired perceptions of reality (e.g., hallucinations, delusions, and disorganized thoughts). As with autism, genetic implications for this disorder arose from twin studies. For instance, such studies show that 45% of identical twins who are adopted by different families will both be schizophrenic. Furthermore, if one sibling within a family has been diagnosed with schizophrenia, the other siblings have a 10-time greater chance of also developing the disorder (Plomin et al., 1994).

Identifying the genetic components of schizophrenia is complicated by the difficulties associated with diagnosing the disease itself, as different doctors often use different criteria to make the same diagnosis. Thus, when they perform studies of schizophrenia, investigators must be particularly careful to look at large populations of individuals with and without the illness, to ensure that all subjects have been diagnosed using the same criteria.

Recently, about 15% of patients with schizophrenia were found to have mutations in a large set of genes, many of which are important for brain development (Walsh et al., 2008). Disturbances have also been found in the expression of genes related to synaptic function, energy metabolism, and oligodendrocyte function (Bray, 2008). These findings implicate hundreds of potential gene candidates and suggest that each person with the disorder could have a different genetic cause. Thus, despite the power of whole-genome analysis to link diseases to genes, considerable work is often required to separate the "wheat" from the "chaff" in such data sets.