Environmental Induction of Behavioral Phenotypes

In many instances, the morphological phenotype is accompanied by a behavioral phenotype. Indeed, one might generalize (see Gilbert and Epel 2015) that while plants manifest their plasticity through morphological change, animals manifest their plasticity primarily through behavioral changes and through the morphological alterations that facilitate this change. Even those cases of profound morphological changes (such as sex determination in fish and locust crowding polyphenism) have an initial phase when a behavioral difference is manifest. This is obvious in the environmental determination of sex, where an individual’s sexual behavior generally matches the gonads and genitalia. This is also seen in the cases of butterfly wings (fliers vs. crawlers) and dung beetle horns (fighters vs. “sneakers”). Some of the reasons behavior has been poorly explored developmentally is because it is difficult to measure, often subjective, and subject to the organism’s history and context (Ballinger and Benzer 1988; Gottlieb 1992; Skuse 2000). However, behavior—the “final phenotype”—can be the major phenotype induced by the environment during development.

Adult anxiety and environmentally regulated DNA methylation

A particular organism’s behavior has both genetic and environmental components, and one of the environmental components concerns the ability of several environmental stimuli to methylate or demethylate DNA. In rats, several adult behaviors, including those involving sexuality and aggression, can be linked to changes in DNA methylation experienced soon after birth (Curley et al. 2011; Caldji et al. 2012). For example, behavioral differences in the response to stressful situations have been correlated with the number of glucorticoid receptors in the brain’s hippocampus. The more glucocorticoid receptors, the better the adult rat is able to downregulate these adrenal hormones and deal with stress. The number of glucocorticoid receptors appears to depend on the quality of grooming and licking the rat pup experiences during the first week after birth.

How is the adult phenotype regulated by these perinatal (near the time of birth) experiences? Weaver and colleagues (2004) have shown that the behavioral difference involves the methylation of a particular site in the enhancer region on the glucocorticoid receptor gene. Before birth, there is no methylation at this site; one day after birth, this site is methylated in all rat pups. However, in those pups that experience intensive grooming and licking during the first week after birth, this site loses its methylation; but methylation is retained in those rats that do not have such extensive care. Moreover, this methylation difference is not seen at other sites in or near the gene (Figure 1).

Figure 1 Differential DNA methylation due to behavioral differences in parental care. A portion of an enhancer sequence of the rat glucocorticoid receptor (GR) gene is shown, indicating the binding site for the Egr1 transcription factor. Two cytosine residues within this binding site have the potential to be methylated. The cytosine at the 5¢ end is completely methylated in the brains of pups that did not receive extensive licking and grooming from their mothers (orange bar). The Egr1 transcription factor did not bind these methylated sites, and thus the GR gene remained inactive. If the pups received proper maternal care, this same site was largely unmethylated (red bar), and the gene was transcribed in the brain. The cytosine at the 3¢ end of the enhancer (blue and purple bars) was always methylated and had no effect on Egr1 binding. (After Weaver et al. 2004.)

By switching pups and parents, Weaver and colleagues demonstrated that this methylation difference was dependent on the mother’s care and was not the result of differences in the pups themselves. When unmethylated, this enhancer site binds the Egr1 (NGF1-A) transcription factor and is associated with “active” acetylated nucleosomes. The transcription factor does not bind to the methylated site, and the chromatin in such cases is not activated. These chromatin differences, established during the first week after birth, are retained throughout the life of the rat. Thus, adult rats that received extensive perinatal grooming have more glucocorticoid receptors and are able to deal with stress better than rats that received less care.[1] Just how grooming can alter DNA methylation patterns, however, remains to be discovered.

Learning: The Developmentally Plastic Nervous System

Learning provides remarkable examples of phenotypic plasticity. Since neurons, once formed, do not divide, the “birthday” of a neuron can be identified by treating the organism with radioactive thymidine. Normally, very little radioactive thymidine is taken up into the DNA of a neuron that has already been formed. However, if a neural precursor cell divides during the treatment, it will incorporate radioactive thymidine into its DNA.

Such new neurons are seen to be generated when male songbirds first learn their songs. Juvenile zebra finches memorize a model song and then learn the pattern of muscle contractions necessary to sing a particular phrase. In this learning and repetition process, new neurons are generated in the hyperstriatum of the finch’s brain. Many of these new neurons send axons to the archistriatum, which is responsible for controlling the vocal musculature (Nordeen and Nordeen 1988; Alvarez-Buylla et al. 1990). These changes are not seen in males that are too old to learn the song, nor are they seen in juvenile females (which do not sing these phrases). In white-crowned sparrows (Zonotrichia leucophrys), whose song is regulated by photoperiod and hormones, exposing adult males to long hours of light and to testosterone induces more than 50,000 new neurons in their vocal centers (Tramontin et al. 2000). The neural circuitry of these birds’ brains shows seasonal plasticity. Testosterone is believed to increase the level of the transcription factor brain-derived neurotropic factor (BDNF, a paracrine factor associated with neuronal plasticity) in the song-producing vocal centers. If female birds are given BDNF, they also produce more neurons in the vocal centers (Rasika et al. 1999).

The cerebral cortices of young rats reared in stimulating environments are packed with more neurons, synapses, and dendrites than are found in rats reared in isolation (Turner and Greenough 1983), and mice reared in cages experienced changes in their neural circuitry when they were placed in more natural environments (Polley et al. 2004). Even the adult brain continues to develop in response to new experiences. Studies on adult rats and mice indicate that environmental stimulation can increase the number of new neurons in the dentate gyrus of the hippocampus (Kempermann et al. 1997a,b; Gould et al. 1999; van Praag et al. 1999). Similarly, when adult rats learn to keep their balance on dowels, their cerebellar Purkinje neurons develop new synapses (Black et al. 1990). The pathway underlying the formation of new synapses probably involves an activity-dependent association of histone modifying proteins with particular transcription factors (Chen et al. 2012; Sando et al. 2012).

In humans, changes in brain anatomy can be seen as a result of learning new tasks. When young adults were taught the classic three-ball cascade juggling routine (which takes months to get right), the neurons in a specific area of the temporal lobe of the brain established a new pattern—a pattern not seen in students who were not taught this skill (Draganski et al. 2004). Thus, the pattern of neuronal connections is a product of inherited patterning and patterning produced by experiences. This interplay between innate and experiential development has been detailed most dramatically in studies on mammalian vision.

Bacteria, brain, and behavior

Although it may sound like science fiction, there is now evidence that symbiotic bacteria stimulate the postnatal development of the mammalian brain (Lorber 2005; Wang and Kasper 2014). Germ-free mice have lower levels of NGF1-A and BDNF in relevant portions of their brains than do conventionally raised mice (Figure 2). This correlates with behavioral differences between groups of mice, leading Diaz Heijtz and colleagues (2011) to conclude, “During evolution, the colonization of gut microbiota has become integrated into the programming of brain development, affecting motor control and anxiety-like behavior.” In another investigation, a particular Lactobacillus strain has been reported to help regulate emotional behavior through a vagus nerve–dependent regulation of GABA receptors (Bravo et al. 2011; Forsythe et al. 2014). Thus, there may be a microbiota-gut-brain axis wherein products made by bacteria can help regulate the development of the brain (Cryan and Dinan 2012; McLean et al. 2012; Mayer et al. 2014).

Figure 2 NGF1-A expression in mice depends on symbiotic microbes. (A) In situ hybridization of NGF1-A mRNA in a section through the frontal cortex of the brain, showing high levels of NGF1-A in mice that have conventional microbes compared with those remaining germ-free. (B) Quantitation using radioactive probes shows the symbiont-containing mice had significantly higher levels of NGF1-A mRNA expression in the frontal cortex and anterior olfactory region. (After Diaz Heijtz et al. 2011.)

Remarkably, the gut microbiota were shown to be critical for normal social behavior in mice. Germ-free mice have aberrant behaviors, including excessive time spent in repetitive self-grooming, social avoidance, and very little time spent in social investigation. Desbonnet and colleagues (2014) remarked that these traits appeared to be similar to those of autistic children. Moreover, many of these behavioral traits were made normal by providing the mice with gut bacteria early in life. Within the same year, Hsaio and colleagues (2013) showed that the symbiotic bacterium Bacteroides fragilis ameliorated the aberrant communicative, anxiety-like, and stereotypic behaviors seen in another mouse model of autism (Figure 3). Moreover, it did so as it altered the composition of gut microbiota, improved the integrity of the gut epithelium, and reduced the leakage of particular gut metabolites into the blood. Several investigators had seen that a subset of autistic children had altered gut bacteria, and the germ-free mice had a similarly skewed pattern of microbial species. Once the integrity of the gut epithelium had been restored and the population of bacteria changed, different metabolites were seen in the blood. Levels of indolepyruvate and ethylphenylsulfate (a chemical that induces anxiety-like behaviors) plummeted. Microbial (probiotic) treatment has sometimes been beneficial in treating psychological distress and chronic fatigue symptoms in humans (Messaoudi et al. 2011; Rao et al. 2009). So it is possible that several of the diseases that we have come to classify as psychological have a root in bacterial metabolism. Moreover, gut microbes are responsible for controlling the permeability of the blood-brain barrier, the tight lining of the capillaries feeding the brain. This barrier shields the brain from blood-borne infections and toxins, since most large molecules cannot flow through it. However, the blood-brain barrier of germ-free mice is not impermeable to proteins, both as pups and adults. This defect can be cured by adding normal gut bacteria back to the gut of the pups (Braniste et al. 2014). It appears, then, that material from the mother’s symbiotic gut bacteria are regulating the permeability of the blood brain barrier while it is being formed in the fetal mice. It is likely that neither our brain nor our behaviors develop properly without the appropriate symbionts.

Figure 3 Effects of germ-free rearing and germ-free bacterial consortium on social behaviors in male mice. (A) Measuring social interaction by the difference between times spent in chambers with and without other mice. Germ-free mice had social avoidance. *, p < 0.05 versus controls; ##, p < 0.001 versus germ-free mice. (B) Time spent in repetitive self-grooming. Both autistic-like behaviors could be relieved by adding microbes later in life. ***, p < 0.001 versus controls. (C) Ethylphenylsulfate levels are extremely high in germ-free mice and can also be removed when epithelial integrity is restored by the addition of Bacteroides fragilis. *, p < 0.05. (A,B after Desbonnet et al. 2013; C after Hsaio et al. 2013.)

Experiential changes in mammalian visual pathways

Some of the most interesting research on mammalian neuronal patterning concerns the effects of sensory deprivation on the developing visual system in kittens and monkeys. The paths by which electric impulses pass from the retina to the brain in mammals are shown in Figure 4. Axons from the retinal ganglion cells form the two optic nerves, which meet at the optic chiasm. As in Xenopus tadpoles, some axons go to the opposite (contralateral) side of the brain, but unlike in most other vertebrates, mammalian retinal ganglion cells also send inputs into the same (ipsilateral) side of the brain (see Chapter 15). These axons end at the two lateral geniculate nuclei. Here the input from each eye is kept separate, with the uppermost and anterior layers receiving the axons from the contralateral eye, and the middle of the layers receiving input from the ipsilateral eye. The situation becomes even more complex as neurons from the lateral geniculate nuclei connect with the neurons of the visual cortex. More than 80% of the neural cells in the visual cortex receive input from both eyes. The result is binocular vision and depth perception.

Figure 4 Major pathways of the mammalian visual system. In mammals, the optic nerve from each eye branches, sending nerve fibers to a lateral geniculate nucleus on each side of the brain. On the ipsilateral side, a particular part of the retina projects to a particular part of the lateral geniculate nucleus. On the contralateral side, the lateral geniculate nucleus receives input from all parts of the retina. Neurons from each lateral geniculate nucleus innervate the visual cortex on the same side.

A remarkable finding is that the retinocortical projection pattern is the same for both eyes. If a certain cortical neuron is stimulated by light flashing across a region of the left eye 5º above and 1º to the left of the fovea,[2] it will also be stimulated by a light flashing across a region of the right eye 5º above and 1º to the left of the fovea. Moreover, the response evoked in the cortical neuron when both eyes are stimulated is greater than the response when either retina is stimulated alone.

Torsten Hubel and David Wiesel, along with their co-workers, demonstrated that the development of the nervous system depends to some degree on the experience of the individual during a critical period of development (see Hubel 1967). In other words, not all neuronal development is encoded in the genome; some is the result of learning. Experience appears to strengthen or stabilize some neuronal connections that are already present at birth and to weaken or eliminate others. These conclusions come from studies of partial sensory deprivation. Hubel and Wiesel (1962, 1963) sewed shut the right eyelids of newborn kittens and left them closed for 3 months. After this time, they unsewed the right eyelids. The cortical neurons of these kittens could not be stimulated by shining light into the right eye. Almost all the inputs into the visual cortex came from the left eye only. The behavior of the kittens revealed the inadequacy of their right eyes; when the left eyes of these kittens were covered, they became functionally blind. Because the lateral geniculate neurons appeared to be stimulated by input from both right and left eyes, the physiological defect appeared to be in the connections between the lateral geniculate nuclei and the visual cortex. Similar phenomena have been observed in rhesus monkeys, where the defect has been correlated with a lack of protein synthesis in the lateral geniculate neurons innervated by the covered eye (Kennedy et al. 1981).

Although it would be tempting to conclude that the blindness resulting from these experiments was the result of failure to form the proper visual connections, this is not the case. Rather, when a kitten or monkey is born, axons from lateral geniculate neurons receiving input from each eye overlap extensively in the visual cortex (Hubel and Wiesel 1963; Crair et al. 1998). However, when one eye is covered early in the animal’s life, its connections within the visual cortex are taken over by those of the other eye (Figure 5). The axons compete for connections, and experience plays a role in strengthening and stabilizing the connections that are made. Thus, when both eyes of a kitten are sewn shut for 3 months, most cortical neurons can still be stimulated by appropriate illumination of one eye or the other.

Figure 5 Experience alters neural connections. (A,B) Dark-field autoradiographs of monkey striate (visual) cortex 2 weeks after one eye was injected with [3H]proline in the vitreous humor. Each retinal neuron takes up the radioactive label and transfers it to the cells with which it forms synapses. (A) Normal labeling pattern. The white stripes indicate that roughly half the columns took up the label, while the other half did not—a pattern reflecting that half the cells were innervated by the labeled eye and half by the unlabeled eye. (B) Labeling pattern when the unlabeled eye was sutured shut for 18 months. Axonal projections from the normal (labeled) eye have taken over the regions that would normally have been innervated by the sutured eye. (C,D) Drawings of axons from the lateral geniculate nuclei of kittens in which one eye was occluded for 33 days. The terminal branching of axons receiving input from the occluded eye (C) was far less extensive than that of axons receiving input from the nonoccluded eye (D). (A,B from Wiesel 1982, courtesy of T. Wiesel; C,D after Antonini and Stryker 1993.)

The critical time in kitten development for this validation of neuronal connections begins between 4 and 6 weeks after birth. Monocular deprivation up to the fourth week produces little or no physiological deficit, but through the sixth week it produces all the characteristic neuronal changes. If a kitten has had normal visual experience for the first 3 months, any subsequent monocular deprivation (even for a year or more) has no effect. At that point, the synapses have been stabilized.

Two principles, then, can be seen in the patterning of the mammalian visual system. First, the neuronal connections involved in vision are present even before the animal sees. Second, experience plays an important role in determining whether or not certain connections persist.[3] Just as experience refines the original neuromuscular connections, experience plays a role in refining and improving the visual connections. It is possible, then, that adult functions such as learning and memory arise from the establishment and/or strengthening of different synapses by experience during development. As Purves and Lichtman (1985) remark:

The interaction of individual animals and their world continues to shape the nervous system throughout life in ways that could never have been programmed. Modification of the nervous system by experience is thus the last and most subtle developmental strategy.

Literature Cited

Desbonnet, L., G. Clarke, F. Shanahan, T. G. Dinan and J. F. Cryan. 2014. Microbiota is essential for social development in the mouse. Mol. Psychiatry 19: 146–148.

Forsythe, P., J. Bienenstock and W. A. Kunze. 2014. Vagal pathways for microbiome-brain-gut axis communication. Adv. Exp. Med. Biol. 817: 115–133.

Hsiao, E. Y. and 11 others. 2013. Microbiota modulate behavioral and physiological abnormalities associated with neurodevelopmental disorders. Cell 155: 1451–1463.

Lorber, B. 2005. Infection and mental illness: Do bugs make us batty? Anaerobe 11: 303–307.

Mayer, E. A., R. Knight, S. K. Mazmanian, J. F. Cryan and K. Tillisch. 2014. Gut microbes and the brain: Paradigm shift in neuroscience. J. Neurosci. 34: 15490–15496.

Wang, Y. and L. H. Kasper. 2014. The role of microbiome in central nervous system disorders. Brain Behav. Immun. 38: 1–12.

1. Does this relate to humans? Using an “extreme” set of cases, Michael Meaney’s laboratory (McGowan et al. 2009; Zhang et al. 2012) showed that the cis-regulatory region of the hippocampus-specific glucocorticoid receptor is more highly methylated in the brains of suicide victims with a history of childhood abuse than in suicide victims with no childhood abuse, or in controls. There is also the possibility that these DNA and histone modification patterns in rats (which appear to be induced by serotonin produced by the nursing pups) may be reversible (see Hellstrom et al. 2012; Zhang et al. 2012).

2. The fovea is a depression in the center of the retina where only cones are present and vision is most acute. Rods and blood vessels are absent. In this instance, it serves as a convenient landmark.

3. Studies have shown that differences in neurotransmitter release result in changes in synaptic adhesivity and cause the withdrawal of the axon providing the weaker stimulation (Colman et al. 1997). Studies in mice suggest that BDNF is crucial during the critical period (Huang et al. 1999; Katz 1999; Waterhouse and Xu 2009; Cowansage et al. 2010).