How the Vertebrate Brain Regulates Behavior

by Donald Pfaff

  Elsewhere in the laboratory, Kathy Jones, who was cognizant of Rochelle’s and Sookja’s work, undertook a more comprehensive ultrastructural study of VMH neuronal nuclei in the presence or absence of estrogens (Figure 3.3) (Jones, Pfaff, and McEwen 1985). Several major findings emerged: a replication of the discovery of estrogen-associated protuberances on the surfaces of VMH nucleoli, significant alterations in nuclear size and appearance, and another replication of the appearance of massive “stacked” rough endoplasmic reticulum in the VMH neurons after estrogen treatment (increased more than three times by estrogen). Regarding nuclear structure, under the influence of estrogens, VMH neuronal nuclei 1) increased in size, 2) changed toward a spherical shape, 3) reduced the appearance of chromatin clumps adjacent to the nuclear envelope, and 4) the appearance of a homogeneous nucleoplasm. Most amazing, we discovered marked invaginations of the nuclear envelope in the ovariectomized controls that simply disappeared as a result of estrogen treatment. Overall, this ultrastructural study gave evidence of a cascade of nuclear changes that support higher rates of VMH synthetic activity after estrogen treatment, consistent with estrogenic effects on nucleolar ultrastructure, rRNA, increases in several transcriptional systems, and estrogen-induced protein synthesis.

  We were excited that the ultrastructural results of estrogen treatment extended back to the MCG, consonant with ER and estrogen nuclear binding there, at the next lower step of our lordosis behavior circuit. We had already reported transsynaptic degeneration in the MCG after VMH lesions, thus revealing that link in the lordosis circuit (Chung, Pfaff, and Cohen 1990b). Now we wanted to look at estrogen effects on synaptic morphology in the central grey (Chung, Pfaff, and Cohen 1988). Estrogen treatment for 20 days (compared with estrogen-free ovariectomized control animals) revealed 1) an estrogen-dependent increase in the number of dense-cored vesicles and a corresponding increase in the number of terminals containing dense-cored vesicles, 2) an increase in the lengths of postsynaptic densities, 3) an increase in the number of densities showing perforations, 4) an increase in the absolute number of synapses per unit area, and 5) an increase in the number of synapses with positive synaptic curvature. It seems likely that the dense cored vesicles contain neurotransmitters or neuromodulators important for central grey neuronal excitation. Likewise, cell biologists had shown the dynamism of postsynaptic densities during the imposition of various physiological changes, even including the suggestion that their perforations could lead to an increase in the number of functional synapses.

  Summarizing these several demonstrations in logical order, in the service of transcriptional systems important for lordosis behavior, estrogens acting in the VMH 1) increase nucleolus-associated chromatin, and, as a result, 2) increase rRNA synthesis, associated with 3) marked enlargement and structural changes in the cell nucleus and nucleoplasm. Assuming that all of these are associated with higher mRNA and protein synthetic rates, it fits that estrogens in the VMH also 4) increase the amount and stacking of the rough endoplasmic reticulum. Thus, in addition to increases in specific transcriptional systems, an overall greater synthetic capacity is part of the program after hormonal treatment.

  It seems likely that because estrogens heighten rRNA synthesis and VMH neuronal growth, and new mRNA and protein synthesis in the hypothalamus are required for lordosis, these hormone-dependent neuronal growth processes play a role in estrogen-facilitated lordosis behavior. A stronger hormone-dependent signal from the VMH to the central grey in the lordosis circuit (Chapter 2) should result.

  Transcription Factor Competition

  When she was in the laboratory, Maria Morgan, a skilled behavioral neuroscientist, had shown that rendering female mice hyperthyroid significantly reduced lordosis behavior in estrogen-primed female mice (Morgan, Dellovade, and Pfaff 2000). Neuroanatomist Tammy Dellovade followed that and showed the same result for female rats (Dellovade et al. 1996): high thyroid hormone results inhibit lordosis. We would use the relative simplicity of nuclear receptor signaling to make a point, as I tell here.

  My first thought about how to explain the ability of thyroxine to down-regulate female reproductive behavior centered on the overlapping nature of the DNA response elements for T3 and estradiol. As previously mentioned, a consensus ERE, beginning from the 5ʹ end, is AGGTCAnnnTGACCT. The n can be any nucleotide base, but there must be three of them in order to get the spacing right for ERs to bind as dimers. A consensus thyroid hormone response element, starting from the 5ʹ end, is AGGTCA, which essentially is one-half of an ERE. So I thought that high levels of thyroxine would bind to TR and compete with ER for DNA binding and thus hormone-regulated transcription. This thinking, plus the collaboration with a TR expert, the Harvard molecular endocrinologist William Chin, led to the three experiments summarized here, and subsequently to the epigenetic (histone modification) work covered in the next section.

  Tammy Dellovade followed up her previously discussed behavioral work by considering that expression of the enkephalin gene in the VMH of the female rat had been correlated with the performance of lordosis behavior (Dellovade et al. 1999b). By antisense DNA evidence, it has been drawn into a causal role as well. So we explored whether, parallel to these earlier molecular and behavioral results, thyroid hormone coadministration could disrupt the estrogenic induction of PPE mRNA. As expected, estradiol benzoate treatment to ovariectomized rats led to a large and significant increase in PPE gene expression in the VMH. The main point was that this increase was inhibited by coadministration of thyroid hormone. The thyroid hormone interference in PPE gene expression was specific to the VMH, as there were no significant effects in the central nucleus of the amygdala or in the caudate / putamen.

  These in situ hybridization histochemical results formed a direct parallel both to previous transcriptional measurements and to the reproductive behavior assays in which thyroid hormones were able to oppose estrogenic facilitation. Further, Tammy Dellovade showed that treatment with exogenous thyroid hormones significantly reduced estrogen effects on the expression of the OT gene (Dellovade et al. 1999a). Previous evidence supported the notion of competitive DNA binding and protein–protein interactions, providing mechanisms for nuclear TR to affect ER function; but whatever the mechanisms, these results with PPE mRNA and OT mRNA reinforced the parallelisms between estrogen-induced gene expression and lordosis behavior.

  Nandini Vasudevan, fresh in New York from a molecular endocrinology laboratory in a national Indian research center in Bangalore, replicated and extended Tammy’s work. She examined the effect of multiple ligand-binding TR isoforms on the ER-mediated induction of the PPE gene in transient transfection assays in CV-1 cells (Vasudevan, Zhu, et al. 2001). On a natural PPE gene promoter fragment containing two putative EREs, both ER-α and ER-β isoforms mediate a four to five times greater induction by estrogen. Cotransfection of TR-α1 along with ER-α inhibited the ER-α transactivation of PPE by approximately 50 percent. However, cotransfection with either TR-β1 or TR-β2 expression plasmids produced no effect on the induction of PPE mediated by ER-α or ER-β. Therefore, under these experimental conditions, interactions with a single ER isoform are specific to an individual TR isoform.

  By the way, transfection with a TR-α1 DNA-binding mutant could also inhibit ER-α transactivation, suggesting that my idea of competition for binding on the ERE may not be the exclusive mechanism for inhibition. In fact, data with the coactivator SRC-1 suggested that coactivator squelching may participate in the inhibition. In dramatic contrast, when ER-β is cotransfected, TR-α1 stimulated ER-β–mediated transactivation of PPE by approximately eight times over control levels.

  While not directly related to lordosis behavior, this was the first study revealing specific interactions among nuclear receptor isoforms on a neuroendocrine promoter. These data also suggested that the combinatorics of ER and TR isoforms allow multiple forms of flexible gene regulations in the service of neuroendocrine integration (Vasudevan, Davidkova, et al. 2001; Vasudevan, Koibuchi, et al. 2001; revi
ewed in Vasudevan, Ogawa, and Pfaff 2002).

  Y. S. Zhu, a medical doctor with expertise in molecular biology, took up this phenomenon of TR / ER interactions for the purpose of a more thorough molecular analysis (Zhu et al. 2001). Knowing that ER and TR are ligand-dependent nuclear transcription factors and that estrogen-induced PPE gene expression in the hypothalamus is directly related to estrogen-induced lordosis behavior in the rat, Zhu wanted more mechanistic detail. Using transient transfection and electrophoretic mobility shift assays (EMSA), functional ERE were identified between −437 and −145 base pairs of the rat PPE gene promoter region. Two ERE-like elements are present between −405 and −364 of the rat PPE gene promoter, which bind ER-α as demonstrated by EMSA. Estrogen produced a dose-dependent increase in chloramphenicol acetyl transferase (CAT) activity in cotransfection assays with ER-α expression vector and a 437PPE-CAT reporter construct containing 437 base pairs of the rat PPE gene promoter and the CAT reporter gene. Crucially, this estrogen-induced PPE promoter activity was inhibited by liganded TR in transient cotransfection assays. The analysis of DNA–protein interactions by EMSA revealed that both ER-α and TR could bind to the ERE in the rat PPE gene promoter.

  Furthermore, we replicated that estrogen induction of PPE mRNA in the VMH of the ovariectomized female rat was significantly attenuated by concomitant administration of triiodothyronine. Thus, we showed that estrogen regulation of the hypothalamic PPE gene expression is mediated through an ER complex directly interacting with the functional ERE in its promoter region and that this estrogen effect can be modified by thyroid hormones.

  How these TR / ER interactions come to occur is explored by histone chemical work from Larissa Faustino in the next section.

  In summary, it is easy to see how the relative simplicity of nuclear hormone receptor signaling to transcription regulating mechanisms could be used to advantage to push forward the field of molecular neuroendocrinology. How do these phenomena come about?

  How Transcription Is Altered—Epigenetic Methodology

  At this point we had proven that several transcriptional systems operating in specific sets of hypothalamic neurons work to produce lordosis behavior. Luckily we were just entering an era in which the regulation of transcription was benefiting from new insights based on protein chemistry. We took advantage of those insights as follows.

  How do those transcriptional events I documented above come about? How are they regulated? A few years ago, molecular chemical methodology presented me with three choices of paths to follow: DNA methylation, histone N-terminus modifications, and noncoding RNAs. DNA methylation seemed altogether too simple to participate in dynamic, subtle epigenetic alterations during the female’s estrous cycle. At the other extreme, noncoding RNAs were being reported in such large numbers and were so obscure that at that time I could not handle the complexity. Thus, I chose histone modifications. In sum, there were two reasons for me to encourage biochemist Khatuna Gagnidze in my laboratory to begin to study histone protein chemistry as a function of estrogen treatment. First, the chemistry did not seem overwhelmingly complex, yet it has the variety and subtlety that you might expect for dynamic adjustments in the central nervous system regulation of behavior. Second, and just as important, my laboratory was blessed with the proximity of the Rockefeller University laboratory of C. David Allis, a histone protein chemist of brilliant clarity, who is generous besides.

  I briefly noted earlier that if and only if ER proteins are liganded by an estrogen molecule will they join together as dimers (two ERs joined) and bind to a consensus ERE (AGGTCAnnnTGACCT) or certain permitted variants of the consensus ERE. But remember that in the usual case there is no such thing as naked DNA. DNA is covered with proteins collectively called chromatin (because of their classically recognized staining properties). The most important elements of chromatin are basic proteins called histones, the chemistry of which has been explicated in yeast cells. Segments of approximately 146 nucleotide bases of inactive DNA are wound around histone protein assemblies in structures called nucleosomes. The saving grace, for the purpose of freeing up DNA to initiate gene transcription is that one end of some of those histone proteins in nucleosomes (the N-terminus) sticks out of the nucleosome and is susceptible to chemical modification. One kind of modification, acetylation, fosters transcription. Another kind of modification, methylation, will either foster transcription or repress transcription, depending on exactly where—on exactly what—amino acid the methyl group is added. As Dave Allis might say, every amino acid is important in this game. There are other types of modifications, too, but I will not mention them because we have not studied them.

  So for us, the question was whether we could discover estrogen-dependent modifications in the VMH that would help to shed light on exactly how estrogens facilitate the transcriptional increases I documented earlier in this chapter. The answer was yes. Biochemist Khatuna Gagnidze, whose calm and unfailingly pleasant demeanor conceals her scientific zeal, entered the laboratory and began the study of estrogenic effects on histone chemistry in the VMH. The minimal latency from estradiol administration to lordosis is 18 hours. During that time, ligand-bound ER, members of a nuclear receptor superfamily, recruit transcriptional coregulators, which induce covalent modifications of histone proteins, thus leading to transcriptional activation or repression of target genes. Khatuna’s study investigated the early molecular epigenetic events underlying estrogen-regulated transcriptional activation of the Pgr gene in the VMH of female mice (Gagnidze et al. 2013). Estrogen administration induced rapid and transient global histone modifications in the VMH of ovariectomized female mice. Histone H3 N-terminus phosphorylation (H3S10phK14Ac), acetylation (H3Ac), and methylation (H3K4me3) exhibited distinct temporal patterns facilitative to the induction of transcription.

  These particular histone modifications create a permissive environment for the transcriptional activity necessary for lordosis, within 3 to 6 hours after estradiol treatment. In the VMH, transcription-promoting changes in the H3Ac and H3K4me3 levels of histone H3 were also detected at the promoter region of the PR gene within the same time window. Moreover, examination of histone modifications associated with the promoter of another ER-target gene, whose product facilitates lordosis, the OT receptor (Oxtr), revealed gene- and brain-region specific effects of estrogens. In sum, Khatuna showed that, early responses to estradiol treatment involve highly specific changes in chromatin structure and clearly show how histone modifications in the VMH can promote estrogen-dependent transcription. In fact, raising VMH acetylation levels by inhibiting the enzyme that would break it down (histone deacetylase) can increase lordosis behavior (as illustrated in the next chapter; see Figure 4.3).

  When Brazilian molecular biology student Larissa Faustino came to the laboratory, Khatuna and I wanted to continue studies of histone modifications in the VMH (Faustino et al. 2015). I wanted to make the experiments simpler, but with Larissa’s insistence we made them more complicated. That is, Larissa referred to Tammy Dellovade’s and Maria Morgan’s data, discussed earlier, that showed inhibiting effects of high thyroid hormone levels on lordosis behavior. And Y.-S. Zhu, also discussed earlier, had shown the binding of TR to a consensus ERE, as I had predicted. Therefore, Larissa hypothesized that high levels of thyroid hormones would inhibit protranscriptional estrogen-dependent histone modifications in the VMH and the preoptic area. Larissa started by replicating our earlier results: as expected, estrogen treatment increased the PR mRNA levels by more than two times and the OT receptor mRNA levels by about two times.

  Indeed, as we hypothesized, making the animals hyperthyroid abolished these transcriptional increases (just as hyperthyroidism reduces lordosis). But we were surprised to find that the opposite thyroid condition, hypothyroidism, could also reduce the estrogen effect. Epigenetic studies yield the same mixture of predicted and unpredicted results. Two protranscriptional histone modifications were studied on the promoters of the PR and OT receptor genes: histone H3 acetylation and histo
ne H3 lysine 4 trimethylation. First, Larissa replicated Khatuna’s results: for instance, in the VMH, estrogen treatment could increase acetylation about four times (Faustino et al. 2015). Yes, rendering animals hyperthyroid could interfere with estrogen effects; but, against prediction, rendering animals hypothyroid could as well. And when we went farther, to study gene expression levels for an ER coactivator—SRC-1, a nuclear protein that enhances ER effects on estrogen-dependent genes—again in the VMH estrogen treatment increased SRC-1 mRNA levels by almost two times. Yes, parallel to lordosis, high thyroxine levels blocked the estrogen effect; but a hypothyroid condition did as well. In summary, in this complicated study we got many parallelisms between lordosis behavior and epigenetic changes on the PR and OT receptor promoters, but the hypothyroid results continue to puzzle me.