How the Vertebrate Brain Regulates Behavior

by Donald Pfaff

  As a side point we also noted that aggressive behavior of gonadally intact male mice is increased by ER-β gene disruption, whereas sexual behavior remains unchanged (Nomura et al. 2006). In addition, estradiol treatment induces higher levels of aggression in βERKO mice than in WT mice. In contrast to aggression, the levels of sexual behavior induced by estradiol were not different between βERKO and WT mice. These findings support the notion that ER-β activation may exert an attenuating action on male aggression induced by estrogen through ER-α–mediated brain mechanisms, whereas its effect on male sexual behavior is relatively small.

  Moreover, Masayoshi Nomura had gone on to systematically examine genotype / age interactions in the regulation of aggressive behavior (Nomura et al. 2002). Overall, βERKO mice were significantly more aggressive than WT. These genotype differences were more pronounced in puberty and young adult age groups, and were not apparent in the adult age group, in which βERKO mice were less aggressive than those in two younger age groups. The serum testosterone levels of βERKO mice were significantly higher than those of the WT mice only in the pubertal age group, not in the young adults (when βERKO mice were still significantly more aggressive than WT mice) or adults (when no genotype differences in aggression were found). These results suggested to us that ER-β–mediated actions of gonadal steroids may be profoundly involved in the inhibitory regulation of aggressive behavior in pubertal and young adult mice.

  Incidentally, we also determined the role of ER-α activation by endogenous estrogen in the development of male-typical behaviors by the use of transgenic ER-deficient (αERKO) mice (Ogawa et al. 1997). Surprisingly, in spite of the fact that they are infertile, αERKO mice showed normal motivation to mount females but achieved fewer intromissions and virtually no ejaculations. Aggressive behaviors were dramatically reduced, and male-typical offensive attacks were rarely displayed by αERKO males. Moreover, ER gene disruption demasculinized open-field behaviors. In the brain, despite the evident loss of functional ER protein, the androgen-dependent system appears to be normally present in αERKO mice. Together, these findings indicate that ER gene expression during development also—in addition to powerful effects in the female—plays a major role in the organization of male-typical aggressive and emotional behaviors beyond simple sexual behaviors.

  As expected, when we measured the phenotype in mice that lack both ER-α and ER-β genes (α-βERKO), these α-βERKO male mice did not show any components of sexual behaviors, including simple mounting behavior (Ogawa et al. 2000). Nor did they show ultrasonic vocalizations during behavioral tests with receptive female mice. On the other hand, reduced aggressive behaviors of α-βERKO mice mimicked those of single knockout ER-α gene (αERKO) mice. They showed reduced levels of lunge and bite aggression, and rarely attempted offensive attacks. Comparing a range of behaviors in these mice not only to WT but also to the behaviors of α knockouts and β knockouts showed us that different patterns of natural behaviors require different patterns of functions by ER genes.

  In summary, extensive DNA sequencing during the last 30 years has yielded knowledge of gene and promoter structures, knowledge that has yielded a molecular neuropharmacology of behavioral sex differences. Most exquisitely, the synthesis of siRNA and the use of homologous recombination to make gene knockouts have provided tools for the analysis of female sex behavior, maternal behavior, and male sex behavior and aggression.

  Reviewing and speculating on the intermediary role for transcription of the PR, female-specific sex behavior (lordosis behavior) depends on the normal expression of the gene encoding the ligand-activated transcription factor ER-α (Ogawa, Eng, et al. 1998; Ogawa, Washburn, et al. 1998). Particularly important is expression in the VMH (Musatov et al. 2006). Speculatively, this may be, in part, because of the sexually differentiated effect of estrogens on expression of the gene for the PR (Romano et al. 1990); antisense DNA against PR messenger RNA (mRNA) microinjected there significantly reduces estrogen-stimulated lordosis behavior (Ogawa et al. 1994). In fact, disruption of the gene for ER-α effectively reversed sex roles in female mice; it caused the females to lose their normal female-typical behavior and to behave and be treated more like males (Ogawa, Taylor, et al. 1996).

  In male mice, disruption of the ER-α gene yielded surprising results (Ogawa, Lubahn, et al. 1996; Ogawa et al. 1997). In spite of the fact that they are infertile, αERKO mice showed normal motivation to mount females, but they achieved less intromissions and virtually no ejaculations. Aggressive behaviors were dramatically reduced, and male-typical offensive attacks were rarely displayed by αERKO males. Daily injection of testosterone propionate was completely ineffective in inducing aggressive behavior in gonadectomized αERKO mice, but it successfully restored aggression in WT mice (Ogawa, Washburn, et al. 1998).

  Maternal behavior in female mice also depends on the normal expression of the ER-α gene. Female αERKO mice showed markedly reduced maternal behaviors (Ogawa, Eng, et al. 1998). This was primarily due to altered gene expression in the POA, as microinjection of an AAV vector bearing siRNA among POA neurons mimicked the behavior-disrupting effect of total ER-α gene loss (Ribeiro et al. 2012).

  In females, other behaviors as well depend on ER-α: the estrogenic effect of increased running wheel activity (likely connected with female-specific courtship behaviors) (Ogawa et al. 2003) and the estrogenic effect on feeding and weight gain (Geary et al. 2001). In general, the sex-specific effects of estrogens can be conceived as maintaining a long chain of behaviors essential for successful reproduction: nutrition, courtship, copulation, and maternal behaviors.

  In dramatic contrast, both male and female sexual behaviors appeared normal when the gene expressing ER-β had been knocked out (Ogawa et al. 1999). Male mice lacking expression of the ER-β gene were significantly more aggressive than WT mice. These genotype differences were more pronounced in the puberty and young adult age groups, but were not apparent in the adult age group; in the latter, the βERKO mice were less aggressive than those in the two younger age groups (Nomura et al. 2002). Thus, there was a significant genotype / age interaction.

  In a separate study, EB treatment induced higher levels of aggression in βERKO mice than in WT mice (Nomura et al. 2006). In both sexes the effects of ER-β on behavior tend to be the opposite of ER-α and, in fact, in intact αERKO mice, the number of cells expressing ER-β was significantly decreased in the mPOA (with opposite results in other brain regions) (Nomura et al. 2003). As predicted, male-specific behaviors were abolished when both ERs were knocked out (Ogawa et al. 2000).

  Epigenetics

  In addition to the molecular pharmacological studies mentioned we have discussed, the depth and sophistication of various sequencing techniques led to a new epigenetics of nerve cells as they relate to behavior. The easiest way to begin this kind of work was with stress: acute restraint stress vastly reduced trimethylation of histone H3 lysine 27, but doubled histone H3 lysine 9 trimethylation (Hunter et al. 2009). Such work has also begun to deal with exotic epigenetic changes such as the suppression by stress of retrotransposon expression (Hunter et al. 2012; Hunter, McEwen, and Pfaff 2013). These epigenetic changes may constitute a “genomic stress response” with the effect of maintaining genomic stability in vulnerable neurons (Hunter et al. 2015).

  In contrast, the female reproductive behavior lordosis requires new mRNA synthesis in the VMH neurons. The protranscriptional histone modifications we discovered in these neurons fit the bill for explaining this female-specific behavior (Gagnidze et al. 2013). Estrogens increased protranscriptional marks on histone 3 and histone 4, as would be expected from the requirement for new mRNA expression for lordosis to occur. The opposition between stress and female-typical reproductive processes has been reviewed (Magariños and Pfaff 2016). Work with histone modifications remains an active subject of interest in our laboratory now.

  We were able to apply a molecular neuropharmacological technique specifically in the VMH to follow up Khatuna Gagnidze’s chromatin
immunoprecipitation results. That is, Zachary Weil, working with Khatuna, found that the microinjection of a histone deacetylase (HDAC) inhibitor trichostatin A—a maneuver that should increase the amount of protranscriptional histone modifications in the VMH and thus should foster estrogen-regulated gene transcription—was able to increase lordosis behavior (Figure 4.3). In a parallel study, Ken Matsuda et al. (2011) used intracerebroventricular infusion of the same HDAC inhibitor trichostatin A in newborn male rats to show that histone deacetylation during that developmental stage is essential for behavioral masculinization as measured by male sexual behavior. The complexity of histone chemistry offers the neurobiologist potential sources of finely modulated nuclear protein functions to gain further insight into questions of neuronal integration, and these results show that such chemistry can be successfully studied in my field of neuroscience.

  Figure 4.3. In the ventromedial nucleus of the hypothalamus, blocking histone deacetylases by local microinjection of trichostatic acid (TSA) raises the performance of lordosis behavior and reduces rejection behavior by estrogen-treated females. (Adapted from Gagnidze et al. 2013.)

  Manipulation of mRNA for the Progesterone Receptor

  Because estrogen-dependent reproductive behaviors of female laboratory animals can be massively amplified with estrogen-induced increases in progestin binding by hypothalamic neurons (see Chapter 3), we hypothesized that reduction of functional PR mRNA by specific PR antisense DNA sequences might decrease these behaviors (Ogawa et al. 1994). Antisense oligonucleotides (15 bases), spanning the translation start site of rabbit PR mRNA, were microinjected directly among VMH neurons, and their behavioral effects were compared with control oligonucleotides composed of the same nucleotide bases in scrambled order (Figure 4.4). To validate that PR protein was actually reduced by our molecular manipulation, antisense DNA was administered near the VMH on one side of the brain, while the other side received the scrambled control sequence or vehicle.

  The total number of PR-immunoreactive cells on the antisense side was significantly lower in the ventromedial nucleus, but not in control measurements from the mPOA. When applied 12 but not 24 hours after estradiol, the PR antisense treatment significantly reduced lordosis behavior, measured either as a reflex or in a mating behavior test. Notably, “proceptive” (courtship) behaviors, which are strongly progesterone-dependent, were greatly reduced in their occurrence (an 80 percent decrease). We concluded that interrupting gene expression for PR, a transcription factor, in hypothalamic neurons can have the predicted effects on female reproductive behaviors.

  Figure 4.4. Top: An antisense vector directed against the progesterone receptor mRNA significantly reduced lordosis behavior at 12 hours. Neurons were not damaged—at 24 hours the behavior recovered. Bottom: Similar results with courtship (proceptive) behaviors. (Adapted from Ogawa et al., 1994.)

  We were very pleased that Shaila Mani in Bert O’Malley’s laboratory at Baylor College of Medicine followed up this finding. She found, likewise, that female mice lacking a PR were not able to perform sexual behavior.

  A Different Nuclear Receptor Family Yields Different Results

  As a control series of studies, we wanted to investigate an entirely different nuclear receptor system (Vasudevan et al. 2013). Thyroid hormones influence both neuronal development and anxiety via the thyroid hormone receptors (TR). The TR are encoded by two different genes, TR-α and TR-β. The loss of TR-α1 is implicated in increased anxiety in males, possibly via a hippocampal increase in GABAergic activity. We compared both social behaviors and two underlying and related nonsocial behaviors—state anxiety and responses to acoustic and tactile startle—in gonadally intact TR-α1 knockout (α1KO) and TR-β (βKO) male mice with their WT counterparts.

  For the first time, we showed an opposing effect of the two TR isoforms, TR-α1 and TR-β, in the regulation of state anxiety. The α1KO animals showed higher levels of anxiety, and βKO males showed less anxiety compared with respective WT mice. At odds with the increased anxiety in nonsocial environments, α1KO males also showed lower levels of responsiveness to acoustic and tactile startle stimuli. Consistent with the data that T4 is inhibitory to lordosis in female mice, we showed subtly increased sex behavior in α1KO male mice.

  These behaviors support the idea that TR-α1 could be inhibitory to ER-α–driven transcription, which ultimately impacts ER-α–driven behaviors such as lordosis. The behavioral phenotypes point to novel roles for the TRs, particularly in nonsocial behaviors such as state anxiety and startle. Most of all, the phenotypes of TR knockouts have nothing to do with the ER-α results I recounted previously, thus showing the specificity of our main results.

  Genes for Neuropeptides

  Referring to the several syllogisms reported for hormone actions on genes in the brain summarized in Chapter 3, we wanted to show gene / behavior causal relations for a couple of obvious choices of neuropeptides.

  Gonadotropin-Releasing Hormone

  The most obvious place to start looking for data is in the literature on gonadotropin-releasing hormone (GnRH) and the GnRH receptor. The knockout data available deal with gene products that are intimately associated with GnRH action but not the GnRH gene itself. For example, Keith Parker at the University of Texas Southwestern in Dallas knocked out the gene for steroidogenic factor 1 and abolished the behaviorally relevant site of GnRH action in the VMH. Lordosis behavior likewise was eliminated. This agrees with the observation that intracerebroventricular treatment with antide, a GnRH receptor antagonist, prevents high levels of lordosis behavior.

  Oxytocin

  Margaret McCarthy, now chief of the Department of Pharmacology at the University of Maryland School of Medicine, came to the laboratory with a good knowledge of neurochemistry. She reminded me that exogenous administration of the neuropeptide oxytocin receptor reliably facilitates sexual behavior in the female rat, and that exposure to estrogen increases OTR binding in the VMH of the hypothalamus. So she decided to use a then-novel approach to investigate the role of hypothalamic OTR in controlling behavior by infusing antisense oligodeoxynucleotides (oligo) to the 5ʹ-region of the human OTR mRNA into the VMH of hormonally primed rats (McCarthy et al. 1994). She validated this approach and also used control infusions that consisted of a scrambled-sequence oligo that had little or no homology to known mRNAs (Figure 4.5).

  Our main result was that OTR antisense oligo infusion significantly reduced lordosis frequency and intensity in females primed with estrogen. There was also a significantly greater number of rejection behaviors exhibited by antisense-oligo-infused estrogen-treated females versus controls and no evidence of decreased locomotion with either treatment. The strength and specificity of this main result reminds me of the clarity of the hormone-action results in the OT / OTR system summarized in Chapter 3.

  Figure 4.5. Lordosis behavioral results from ovariectomized female rats given estrogen. Antisense DNA directed against mRNA for the oxytocin receptor, microinjected into the ventromedial nucleus of the hypothalamus, significantly reduced lordosis (compared with scrambled sequence controls) and increased rejections of the male by the female. (Data from McCarthy et al. 1994.)

  In dramatic contrast with the interference with OTR mRNA just noted, neuroscientists had trouble showing the effects of OT gene knockouts on female reproductive behavior. We decided we had to do something different, so we constructed a large seminatural environment to assay a large range of social behaviors under conditions a bit more like those of the wild as opposed to the standard mating behavior assay (Ragnauth et al. 2005). What had bothered us was that, compared with the results from a generation of neuropharmacological work, the phenotype of mice lacking the OT peptide gene was remarkably normal. An important component of our new experiments was to assay OTKO and WT littermate control mice living under controlled stressful conditions designed to mimic more closely the environment in which the mouse genome evolved. Furthermore, our experimental group was composed of an all-female population, in contrast to th
e previous studies that had focused on all-male populations.

  The most obvious feature of the data was the large range of aggressive behaviors initiated by OTKO females. In social rank, “alpha females” (always OTKO) were identified in each experiment. They were the most aggressive, the first to feed, and the most dominant at nesting behaviors. The aggressiveness of the OTKO females in the seminatural environment suggested that these females would fight the male rather than mate with him, but we did not test that suggestion directly.

  Enkephalin

  Another strong set of molecular data in Chapter 3 featured the neuropeptide enkephalin. There I summarized studies in female rats showing that estrogen increases preproenkephalin (PPE) mRNA levels in the ventrolateral part of the VMH, neurons crucial for female sexual behavior (Chapter 2). To assess the physiological role of hypothalamic PPE gene expression in the production lordosis behavior, we microinjected a 16-mer oligodeoxynucleotide (ODN) directed toward the PPE mRNA in the VMH of estradiol-primed rats (Nicot et al. 1997). The results were clear: antisense ODN injections near the VMH resulted in a significant reduction in lordosis quotient compared with the control (reverse sense) ODN treatment or with the antisense ODN injections targeted anterior or posterior to the VMH. The results had chemical specificity and neuroanatomical specificity.

  In contrast, the locomotor activity of these animals in the open-field test was not affected by ODN treatments. We also had validated our antisense technique. Enkephalin immunoreactive levels were determined by radioimmunoassay in the POA, a major terminal field of the VMH. Estradiol-induced enkephalin levels were greatly reduced in antisense-treated groups. We also replicated our previous findings, as summarized in Chapter 3: using the in situ hybridization technique, we saw that PPE mRNA levels in the ventromedial VMH showed 1.5- to 2-fold increase in PPE mRNA levels was observed in estradiol-treated rats compared with estrogen-free controls. This increase in PPE mRNA levels was not affected by ODN treatment, suggesting that the reduction of enkephalin expression was mainly due to physical blockade of PPE mRNA translation and not to its degradation. Taken together, these data further supported the important behavioral role of PPE gene expression in VMH neurons.