by Donald Pfaff
In the hypothalamus, we saw heavy labeling throughout the extent of the arcuate (infundibular) nucleus (Figure 1.3) and by the lateral-placed neurons within the ventromedial nucleus of the hypothalamus. Estrogen-binding cells were also found in the anterior hypothalamic area and posteriorly near the borders of the medial mammillary nucleus.
Large numbers of well-labeled cells were in the medial preoptic area. The estrogen concentration was especially strong on its medial side and in the suprachiasmatic portion of the preoptic area.
In the limbic system, we could see estrogen binding in the ventrolateral septum and, to a much lesser extent, in the medial septum. The bed nucleus of the stria terminalis was well labeled, as were neurons in the medial nucleus of the amygdala. All the tissue from the basal ganglia was not labeled.
In the anterior mesencephalon, lightly labeled cells were seen behind the mammillary bodies in the periventricular stratum leading to the central grey. There was a small number of labeled cells in the central grey near the dorsolateral and ventrolateral borders. Other mesencephalic tissue was unlabeled.
The neuroanatomical pattern of cell nuclear estrogen binding in this limbic / hypothalamic system was seen to be specific because the corticosterone binding was by far highest in the hippocampus, as Bruce McEwen had found for the rat brain, and some corticosterone-labeled cells were even seen among the cerebellar granule cells.
Figure 1.3. Black dots show the locations of neurons expressing estrogen receptors in the ventromedial hypothalamus (VM) of the female rhesus monkey. The expression is bilateral, plotted here on one side. (Adapted from Pfaff et al. 1976).
Human Brain
Finally, of course, we approached the most important species for medical approaches to the brain: humans. Dick Swaab, a leader at the National Institute for Brain Research in Amsterdam, used an antibody against ER-α to study nuclear ER in material from the Netherlands brain bank. His distribution of limbic and hypothalamic neurons showing ER immunoreactivity matches what I originally reported in the rat brain. Likewise, Jeffrey Blaustein, a professor at the University of Massachusetts, not only extended my findings in the diencephalon of the rat to the human brain but also documented statistical sex differences: men’s brains showed somewhat more ER immunoreactivity in the medial preoptic area whereas women’s brains showed somewhat more in the ventromedial hypothalamus.
Overall, this was a long saga. During the first couple of years after my articles were published I had to fight off a challenge from the late Walter Stumpf, whose technique led to artifacts and whose neuroanatomical charting was oversimplified, which confused things for a while. But my work enjoyed a massive amount of replication, as I’ll discuss later.
In summary, from fish through amphibians through reptiles, through birds and several mammalian brains including monkeys and humans, we discovered a common neuroanatomical pattern of sex hormone-binding neurons. Hormone binding in limbic structures is strong. Medial preoptic neurons and medial hypothalamic neurons show high levels of sex hormone concentration in the cell nucleus, especially in the arcuate nucleus and the ventromedial nucleus (as will be crucial for lordosis behavior). A frequent posterior extension of this system is the mesencephalic central grey.
Replication at the mRNA Level
The canonical sequence of biochemical and physiological events that we addressed includes 1) gene expression for ER, 2) translation to the ER protein, 3) binding of estrogens by the ER protein, and 4) consequent estrogenic facilitation of female reproductive behavior. This last step I established in the work I discussed at the beginning of this chapter, and I will continue that discussion here. First, let’s address step 1.
By studying hormone binding in neurons, I was addressing the actual function of an ER gene and its protein products. Nevertheless, it was comforting to see in situ hybridization results from my own and other laboratories demonstrating ER gene expression in the same nerve cell groups I have emphasized previously. The earliest confirmation came from the in situ work of the always upbeat Richard Simerly, who had shown his brilliance early by choosing to work with neuroendocrine pioneer Roger Gorski (at the University of California–Los Angeles), and then with Larry Swanson (now at the University of Southern California), one of the premier neuroanatomists of his generation. Simerly’s results matched our binding results so well that we decided to follow them up.
Andrea Lauber not only showed ER messenger RNA (mRNA) expression in the ventrolateral corner of the ventromedial nucleus of the hypothalamus (the most important for lordosis behavior regulation) and arcuate nucleus of the hypothalamus, but she also demonstrated that estrogen administration down-regulates the mRNA for its own ER receptor there (Lauber et al. 1990). As measured by grain counting over individual cells in those hypothalamic structures and via in situ hybridization, the decline in ER mRNA reached its lowest point 18 hours after a subcutaneous injection of estradiol, having knocked down gene expression to less than half its initial rate. After that, by 24 hours the mRNA levels were recovering.
Further, this endocrine feedback dynamic was sexually differentiated (Lauber et al. 1991). To begin with, there is a clear sex difference in initial ER mRNA levels in these hypothalamic neuronal groups, notably the ventromedial nucleus. Males exhibited 52 percent less ER message than females. On top of that finding, the decline in ER mRNA levels in males was small and not statistically significant in the same experiment where the large (>50 percent) decline was replicated in the female hypothalamus. Interestingly, these endocrine dynamics and sex differences did not show up in the medial amygdala. Although there was a trend toward a decline in the female amygdala and a trend toward a sex difference, those trends were not statistically significant.
Molecular endocrinologists were surprised when the laboratory of Jan-Ake Gustafsson, at the Karolinska Institutet, cloned a second ER, now called ER-β. As a result of that, the ER highly expressed in the hypothalamus and important for lordosis is now called ER-α. ER-β binding is not crucial for lordosis; its import in the brain, highly significant, is sometimes the opposite of ER-α. Its impact on estrogen-dependent cancers is of great interest.
Replication at the Protein Level
The most popular way to study ER-α protein in the brain has been to use specific antibodies for immunocytochemistry. Mapping studies doing exactly this have repeated the neuroanatomical distribution I reported for estrogen binding.
Jeffrey Blaustein at the University of Massachusetts was the first to show ER-α protein in the same preoptic, hypothalamic, and limbic areas as I had observed estradiol binding. In addition, he reported that progesterone receptor (PR) immunoreactivity tended to appear only in ER-expressing neurons. This is important because estradiol working through ER-α strongly induces expression of the progesterone receptor gene (Chapter 3). Blaustein’s work in the guinea pig was replicated by Marie Warembourg and Eduard Milgrom in the INSERM Unit in Lille, France, and also by Joan Morrell, now at Rutgers. The same limbic / hypothalamic distribution of ER protein was reported for the rat brain, with an emphasis on high levels of immunocytochemical product in the medial preoptic area and ventromedial hypothalamus. And for our later work on behavior genetics, it was important that these ICC results were also repeated in the mouse brain.
Mitsuhiro Kawata, in Kyoto, did an immunocytochemical experiment that replicated with an amazing degree of precision Andrea Lauber’s previously mentioned work at Rockefeller. He not only reported the distribution of ER identically to the Lauber’s studies, but he also looked at the negative feedback effects of circulating estrogen on ER protein in the brain. In his results, there were more ER-immunoreactivity cells with higher immunoreactivities in ovariectomized rats than in estradiol-treated rats. The number of ER-immunoreactivity cells in estradiol-treated rats compared with ovariectomized rats was reduced by 43 percent. In particular, the number of ER-immunoreactivity cells in the central part of the MPN was largely decreased. These data quantitatively match Lauber’s results with in situ hybridization.
Results with other species fell into line. Carol Jacobson, at Iowa State University, saw the identical limbic / hypothalamic distribution of ER protein in the Brazilian opossum. In the brain of the Japanese quail, Jacques Balthazart, at the University of Liege in Belgium, reported a high percentage of labeled cells was observed in the lateral septum, the nucleus accumbens, the preoptic medial nucleus, the supraoptic nuclei, the anterior medial hypothalamus, the paraventricular magnocellular nucleus, the caudal parts of the lateral hypothalamus, and the whole tuberal and infundibular area. A small number of labeled cells was also observed in the ventromedial nucleus of the hypothalamus
Primate brain. Most importantly, Alan Herbison, then at the University of Cambridge, studied monkey brains and showed that ER-immunoreactive cells in the monkey hypothalamus are distributed in a manner similar to that observed in other mammalian species. His double-labeling experiments provide further evidence that luteinizing hormone-releasing hormone (aka gonadotropin-releasing hormone) neurons do not possess ERs, thus replicating a somewhat controversial finding I had reported for rat brain some years before.
Summary to This Point
Studies of ER-α gene expression, protein levels, and actual estrogen binding all tell us of high levels of hormone binding in the medial hypothalamus, the medial preoptic area, and specific portions of the limbic system, the phylogenetically ancient portions of the forebrain. Expression in the ventromedial nucleus of the hypothalamus is most important for the estrogenic facilitation of lordosis behavior.
Importance for Behavior
Paula Davis, in my laboratory, used an extremely discriminating approach to hormone implantation into the brain to demonstrate that ventromedial hypothalamic neurons are sensitive to estrogens in a manner that facilitates lordosis behavior. First, she implanted extremely small amounts of tritiated estradiol directly into the ventromedial hypothalamus and showed that it could facilitate lordosis behavior of female rats (Davis, McEwen, and Pfaff 1979). Although significant amounts of radioactivity were concentrated in the nuclei of ventromedial hypothalamic neurons, no leakage to other sites—such as the lateral hypothalamus, the preoptic area, the amygdala, or the cortex, or even high-affinity sites such as the pituitary and the uterus—could be detected. Thus, estrogenic binding in ventromedial hypothalamic neurons is sufficient for facilitating lordosis behavior.
Davis followed up those experiments by using autoradiography to prove, in quantitative terms (grains per unit area), the absence of significant diffusion of estrogen from the ventromedial hypothalamus implantation site, and further to prove that estrogen exposure in the central and lateral subdivisions of the ventromedial hypothalamus (not the anterior pole) acts to promote lordosis (Davis et al. 1982).
Estrogen binding to ER-α in the ventromedial hypothalamus is also necessary for lordosis. Implantation of an antiestrogen, an ER blocker, in the ventromedial hypothalamus but not in other parts of the brain could significantly reduce the effects of systemic estrogen administration on female reproductive behavior (Meisel et al. 1987).
Necessary and Sufficient
Thus, estrogen binding in the nuclei of ventromedial hypothalamic neurons is necessary and sufficient for elevated levels of female reproductive behavior. Estradiol in the ventromedial hypothalamus can permit lordosis responses to the male’s mounting even when there is no estrogen elsewhere in the body. Conversely, estrogens can be flooding tissues elsewhere in the body but if an antiestrogen implantation in the ventromedial hypothalamus prevents its nuclear binding there, then lordosis behavior will not occur.
Paula Davis’s and Bob Meisel’s work in my laboratory was complemented by elegant results from Ron Barfield’s laboratory at Rutgers. He found that in the ventromedial hypothalamus dilute estrogen implants facilitated lordosis and an ER antagonist implant blocked lordosis.
Because ER-α is a ligand-activated transcription factor (see Chapter 3) we can expect to see a train of events that leads from nuclear binding of estradiol in the ventromedial hypothalamus to the promotion of lordosis behavior: new mRNA synthesis, followed by new protein synthesis, followed by axoplasmic flow from the ventromedial hypothalamic cell body toward the midbrain (Chapter 2), accompanied by action potentials. All these are necessary for estrogen-stimulated sex behavior.
In the 1970s David Quadagno working with Roger Gorski at the University of California–Los Angeles showed that intrahypothalamic infusion of the RNA synthesis inhibitor actinomycin D blocked hormone-induced lordosis behavior. Brenda Shivers and Richard Harlan, working at the University of Texas before they came to my laboratory, as well as Tom Rainbow who was working with Bruce McEwen at Rockefeller, replicated the Quadagno finding. In turn, Quadagno infused the protein synthesis inhibitor cycloheximide and blocked lordosis.
Our own work (Harlan et al. 1982), led by Harlan and Shivers, concentrated on the axoplasmic flow of newly synthesized protein from the hypothalamus toward the midbrain (see the circuitry information in Chapter 2). Colchicine is a drug well-known to disrupt axoplasmic transport by inhibiting microtubule polymerization by binding to tubulin, one of the main constituents of microtubules. It was infused into or near the ventromedial hypothalamus, and its effect on lordosis was compared with vehicle controls. In a dose-dependent manner, colchicine rapidly and significantly reduced behavioral sensitivity to estrogen treatment.
In later experiments, Harlan and Shivers studied the requirement for hypothalamic action potentials in the maintenance of estrogen-supported action potentials (Harlan et al. 1983). Local anesthetics such as procaine or Marcaine infused into the hypothalamus were not very effective. However, infusion of tetrodotoxin for blocking voltage-dependent sodium currents was very effective, reducing estrogen-dependent lordosis in a dose-dependent manner for eight hours. Electrophysiological recordings confirmed the expected abolition of electrical activity by the affected hypothalamic neurons, the basis of the reproductive behavioral effect. Joel Rothfeld in our laboratory extended this work by microinfusing tetrodotoxin into the midbrain central grey, where ventromedial hypothalamic axons that are relevant for lordosis terminate (see Chapter 2). For a time period from 10 minutes to four hours, tetrodotoxin in the central grey virtually abolished lordosis (Rothfeld et al. 1986).
Taking this all together we can summarize that after estrogen binding to ER in the ventromedial hypothalamus, estrogen-responsive gene expression is necessary for lordosis behavioral regulation, followed canonically by protein synthesis and flow down the axon, accompanied by estrogen-sensitive electrical activity.
Hormone-Binding Neurons Projecting to Form Networks
This chapter is not an appropriate venue for a comprehensive review of limbic system, preoptic, or hypothalamic neuroanatomy. Examples of where such reviews can be found include an overview by Swanson (2013), with behavioral interpretations by Kringelbach (2016). The oldest ideas about close limbic / hypothalamic relations began with the Papez circuit, which related hippocampal outputs to preoptic and hypothalamic function (with a return through the anterior thalamic nuclei). Swanson and Cowan detailed hippocampal efferents to the hypothalamus. Later they described dense and multiple interconnections between the medial and lateral septum, on the one hand, and the preoptic area and hypothalamus, on the other. David Amaral, when he was at the Salk Institute, extended our knowledge of hypothalamo-amygdaloid relations to the monkey brain. In his words, all these projections connect the amygdala to the hypothalamus by way of the so-called ventral amygdalofugal pathway, but at least some of the fibers that arise in the ventromedial nucleus run in the stria terminalis. These are some of the leading neuroanatomical scholars and findings that gave me the courage to theorize as well.
Starting about 1971, it became obvious to me that sex hormone-concentrating nerve cell groups tend to project to other sex hormone concentrating nerve cell groups, and our first chart was published as figure 4 in Pfaff and Keiner (1973). Here are just a few examples: 1) strong medial preoptic projections to the medial amygdala, 2)
medial preoptic projections to the ventromedial hypothalamus, 3) medial preoptic efferents to the septum, 4) anterior hypothalamic projections to the preoptic area, 5) anterior hypothalamic efferents to the septum, 6) anterior hypothalamic connections to the ventromedial hypothalamus, 7) anterior hypothalamic projections to the medial amygdala, 8) amygdala projections to the preoptic area, 9) amygdala projections to the ventromedial hypothalamus, 10) septal projections to the hippocampus, 11) hippocampal projections to the preoptic area, 12) hippocampal projections to the ventromedial hypothalamus, 13) ventromedial hypothalamic projections to the preoptic area, 14) ventromedial hypothalamic projections to the septum, 15) ventromedial hypothalamic projections to the amygdala, and 16) ventromedial hypothalamic projections to the central grey.
By no means does this dense web of connections follow a pattern of laminar flow, analogous to some parts of the visual system. However, Rockefeller graduate student Lily Conrad and I noticed a greater trend toward spatial orderliness than is usually claimed for the hypothalamus and basal forebrain (Pfaff and Conrad 1978, see figures 8 and 9). That is, more medially placed nerve cell bodies gave rise to axons that tended to descend medially—and laterally placed neurons, to axons more laterally, and dorsally placed neurons, to axons more dorsally. The implications of this neuroanatomical trend for reproductive behavior are not clear, but the interconnectedness of sex hormone-binding nerve cell groups does have a theoretical implication.
We derived the theory that networks of hormone-sensitive neurons offer possibilities of systems that effectively multiply hormone effects. One reason for pointing out the neuroanatomical interconnectedness of sex hormone-binding neuronal cell groups is that it offers the possibility of hypothesizing interesting hormone-sensitive mathematical functions that could result from such interconnectedness. These interactions could be important for the regulation of behavior.
