by Donald Pfaff
Early ideas about how steroid hormones could influence cells began with the knowledge that steroids are flat, planar, rigid molecules. Arriving at the cell, they could influence the fluidity of the cell membrane. Adding to that idea, biochemists such as Claude Villee at Harvard Medical School studied the effect of steroid sex hormones on the activity of cytoplasmic enzymes, especially those that had to do with steroid metabolism. That was the situation when I entered the field.
Those lines of research were shelved when organic chemist Elwood Jensen synthesized the first carrier-free radioactively labeled steroid hormone: tritiated estradiol. He used biochemical techniques to show that after a systemic injection of labeled estradiol, the hormone at first flooded uterine cells then quickly subsided (within minutes); however, in uterine cell nuclei the hormone was retained much longer—for hours. I obtained tritiated estradiol from the first batch produced by New England Nuclear and took Jensen’s developments to the brain.
Hormone Receptors in the Brain
As mentioned in the introduction, as a first-year graduate student at MIT in 1961, I was helping Joseph Altman (the discoverer of postnatal neurogenesis) trace the migration of newly divided neurons around the developing nervous system. But as a young researcher who wanted to contribute to the science underlying psychiatry and neurology, I said, “Joe, I’d like to do something more closely related to brain function.” His seven-word response, “I’d like to do something with hormones,” set up my discovery of hormone receptors in the brain. The limbic-hypothalamic system of sex hormone receptors that I found in the rat brain turned out to be universal among vertebrates—“from the fish to the philosopher.” As well, since these are nuclear receptors, they offered the unique advantage of studying hormone-dependent transcription factors. Behavioral and neural biology were linked in the early days to molecular biology (see Chapter 3).
Joe Altman was a true artist at the laboratory bench. His feeling for beautiful histochemistry emerged most dramatically when he responded to the illustrations published by Tomas Hokfelt and Kjell Fuxe, their histochemistry demonstrating aminergic systems in the brain. Joe was a really tall and muscular guy, and his sharp intake of breath when he saw those illustrations, and then explained them to me, said it all. So he was a great guy from whom to learn histochemistry.
For almost four years I worked to optimize histochemical techniques for discovering hormone-binding neurons in rat brain tissue. Repeated attempts to optimize the methodology were necessary for three reasons: the low specific activity of labeled steroids in those days, the difficulty of keeping the bound hormone exactly in place in the target neurons, and the necessity of avoiding histochemical artifacts due to the pressure sensitivity of the Kodak NTB-3 nuclear emulsion, which was best for detecting the β particle that emanated from the bound tritiated hormone. The tritium was great for my purpose because the β particle has such low energy that it can only travel into the nuclear emulsion directly over the target neuron, thus offering good spatial resolution at the light microscope level.
Eventually I succeeded, and I published my preliminary results (Pfaff 1965). But a tremendous amount of quantification remained to be done: counting grains over labeled hormone-binding neurons. The highest numbers of estradiol-binding neurons were found in a subset of limbic / telencephalic and hypothalamic / diencephalic cell groups (Pfaff 1968) that formed a limbic / hypothalamic system with an extension into the central grey.
For the purpose of facilitating female mating behavior, the most important concentration of estradiol-binding neurons was in the ventrolateral corner of the ventromedial nucleus of the hypothalamus (Figure 1.1), because this is where estrogens act to facilitate lordosis (as I’ll discuss more later). In general, the grouping of estrogen-receptive neurons did not always respect classic neuroanatomical boundaries; for example, a few such neurons were just outside the ventromedial nucleus.
Elsewhere in the hypothalamus, tritiated estradiol retention was strong in the arcuate nucleus (of importance for controls over the pituitary) and the ventral premammillary nucleus (of importance for pheromonal effects on reproduction). I found labeled cells as well in the anterior hypothalamic area and at the lateral caudal tip of the magnocellular portion of the paraventricular nucleus of the hypothalamus.
Forward of the hypothalamus, large numbers of estrogen-binding neurons were seen in the medial preoptic nucleus—heavily labeled—especially near the midline at levels under the anterior commissure and in the suprachiasmatic portion of the preoptic area. The former turned out to be important for the estrogen dependence of maternal behaviors (see Chapters 4 and 7).
Figure 1.1. Locations of estrogen receptor-expressing cells in the female rat brain (symbolized by black dots). Top: Drawing of a sagittal section through the rat brain, looking from the left side. Each black dot represents clusters of neurons expressing estrogen receptors. The vertical black line approximates the coronal section. Bottom: Looking at the cross section indicated at the top, the nerve cells expressing the estrogen receptors crucial for lordosis behavior are in the blackened area (very densely packed cells) in the ventrolateral corner of the ventromedial nucleus of the hypothalamus (vm, arrow). (Adapted from Pfaff and Keiner 1973.)
In the limbic forebrain I found impressive groups of estrogen receptive neurons in the medial nucleus of the amygdala, particularly in its dorsal posterior subdivision, and in a cell group named the bed nucleus of the stria terminalis (which fosters communication between the amygdala and the diencephalon). Certain cells in the lateral septum were well labeled. Following up my brief 1965 report, I found labeled cells at all levels of the hippocampus, both among the small granule cells of the dentate gyrus and the large pyramidal cells of Ammon’s horn.
Also important for lordosis behavior regulation, because it fits into the neural circuit for lordosis (see Chapter 2), the central grey of the mesencephalon contains labeled cells throughout its extent, from the diencephalic / mesencephalic junction all the way back to the transition from the cerebral aqueduct and the fourth ventricle. I found labeled cells most easily directly lateral or ventrolateral to the aqueduct. The number of such cells was not as great as in the most densely labeled hypothalamic nuclei.
After I moved from MIT to Rockefeller, and because I had explored a number of histochemical approaches, I was able to follow up the original report with a different procedure involving frozen sections and direct mounting onto the nuclear emulsion (Pfaff and Keiner 1973). All the major features of the original paper were replicated in the atlas I published.
Later, using biochemical techniques and working with Bruce McEwen and Richard Zigmond, we confirmed my histochemical / neuroanatomical results. We exploited my cellular findings to design dissection schemes that could give us clear results. The major features of our findings, besides confirming my findings, were that prior injections of unlabeled estradiol significantly reduced tritiated estradiol uptake in brain regions—indicating limited capacity binding as is true for the uterus—and that hypophysectomy reduced uptake in all brain structures. Similar results were found for tritiated testosterone uptake in the rat brain. For the estrogen findings, the results from my laboratory and Richard Whalen’s (at the University of California–Riverside) were confirmed by James Clark at Baylor using biochemical methods, by John Cidlowski at the University of North Carolina, and by Fred Naftolin working with Jack Fishman at Yale.
A whole new field of work was initiated when a second estrogen receptor, ER-β, was cloned by Jan-Ake Gustafsson and his team at the Karolinska Institutet. The behavioral consequences of ER-β expression are interesting and are significantly different from the ER-α we have concentrated on (see Chapter 4).
Strong evidence for the specificity of the estrogen-binding neurons I have reviewed here comes from other results of Bruce McEwen and his team at Rockefeller. By far the most impressive binding of corticosterone, a glucocorticoid hormone, was found in the hippocampus, not in the hypothalamic and preoptic neurons just emphasized
here, so important for mating and maternal behaviors.
Likewise, the thyroid hormones that are so crucial for the regulation of metabolism are bound in the brain by the gene product from thyroid hormone receptor α1 (TR-α 1), as discovered by Karolinska professor Björn Vennström. TR-α1 is found expressed in an extremely wide neuroanatomical pattern, giving further indication of the specificity of our results with labeled estrogens (Wallis et al. 2010). It was known that thyroid hormone is essential for brain development, but the potential for the hormone to act in adult neurons needed molecular markers in order to be proven. In Vennström’s results in the adult brain, TR-α1 expression was detected in essentially all neurons. So a new field of work on thyroid hormones in the brain was opened up.
Always we have tried for accurate quantification of our results, with the desire that precision in our area of neuroscience would bring it one step closer to the precision typical of the physical sciences. This underlying desire has animated all my work, from neuroanatomy to electrophysiology, molecular biology, and ultimately behavior. With this in mind, Joan Morrell and Monica Krieger in my laboratory used 2-micron sections and cell-by-cell grain counting in four areas of the forebrain that feature high levels of estrogen binding: the ventromedial nucleus of the hypothalamus, preoptic area, arcuate nucleus of the hypothalamus, and the medial nucleus of the amygdala (Krieger, Morrell, and Pfaff 1986). The null hypothesis was that the frequency distribution of grains-per-cell would fit a random process, a simple Poisson distribution. Impressively, the frequency distribution in all four brain regions deviated markedly from the Poisson. Then, when the labeled estrogen was accompanied in other animals by a tenfold excess of nonradioactive estradiol to wipe out the specific binding sites in the brain, the results in all four brain regions, indeed, collapsed onto frequency distributions indistinguishable from the random, Poisson curve.
Later, considerable work was started to identify the neurochemical contents of neurons in the hypothalamus that bind high levels of radioactive estrogen. For example, Andrea Gore, now at the University of Texas, reported that expression of ER-α is colocalized with the obligatory glutamatergic NMDAR (N-methyl-D-aspartate receptor) subunit NR1 in neurons of the anterior hypothalamus and medial preoptic nucleus, showing how hormonal and glutamate neurotransmitter influences might synergize. Another example: our student Harker Rhodes, working with professor Joan Morrell, found paraventricular hypothalamic neurons that would produce oxytocin or vasopressin, which retained radioactive estradiol (Rhodes, Morrell, and Pfaff 1981). This type of work, characterizing ER-α receiving neurons in neurochemical terms, is still ongoing.
As we began to put together the entire circuit for lordosis behavior (Chapter 2), we needed to know where estrogen-sensitive signals were going after the uptake of estradiol by ventromedial hypothalamic neurons. Joan Morrell at Rockefeller invented a technique by which she could characterize estrogen-concentrating hypothalamic neurons according to their axonal projections (Morrell and Pfaff 1982). Most importantly, retrograde-transported fluorescent dye primulin, injected into the dorsal midbrain by the central gray, identified estrogen-binding neurons in the ventrolateral portion of the ventromedial nucleus of the hypothalamus. This discovery provided part of the evidence for an estrogen-sensitive link at the top of the lordosis behavior circuit, later verified by other techniques.
Thus, with neuroanatomical precision and biochemical sophistication we established a hormone–brain link that would provide one crucial entry point to working out the female reproductive behavior circuit (Chapter 2) and to introducing the tools of molecular biology (Chapter 3) to our program of research.
Hormone Binding in a System Universal among Vertebrates
I doubted that all the data discovered in rat brains were particular to that one species. The overall neuroanatomical pattern seemed too solid, too integral for that, and made too much physiological sense to be so biologically restricted. So we branched out toward other species. Of course, at first we simply used other mammals to extend the discoveries from rats to other species. But the long-term aim was always to establish an endocrine-neural link from “fish to philosopher”—across all vertebrate species.
Here, after summarizing results from other rodent species, I will review our hormone-binding findings from other classes of vertebrate species: fish, amphibian, reptiles, and birds. Then things will get very interesting with results from the monkey brain and human brain. The bottom line of the hundreds of studies from a lot of laboratories, mine included, is that the limbic / hypothalamic system of sex hormone-binding neurons I discovered in rat brain is universal among vertebrates.
Hamster Brain
The experimental skill of postdoctoral researcher Monica Krieger came to the fore in her work at Rockefeller with Professor Joan Morrell in my laboratory. Female hamster brains were a natural for study because of their striking lordosis displays upon mounting by the male. The autoradiographic results mirrored my results with the rat brain. There were heavily labeled cells in the medial preoptic area posterior to the lamina terminalis, especially in the medial portion of the preoptic area. The anterior hypothalamus had large numbers of estrogen-binding neurons, particularly in the distribution of the stria terminalis.
Important for the female hamster’s lordosis, the ventrolateral subdivision of the ventromedial hypothalamus showed estrogen-binding neurons while the dorsomedial nucleus was virtually unlabeled. As expected, because of the neuroendocrine functions of the arcuate nucleus of rat hypothalamus, it had heavily labeled neurons.
In the limbic system, the medial nucleus of the amygdala showed some of the heaviest labeling in the brain whereas other amygdaloid subregions were almost unlabeled. Interestingly, the bed nucleus of the stria terminalis had strongly estrogen-retaining neurons. Throughout the hippocampus we saw some labeled pyramidal neurons, plus some well-labeled cells in the subiculum lateral to the tip of the ventral hippocampus.
As expected from my rat brain findings, Krieger showed well-labeled neurons in the lateral portion of the midbrain central gray.
Mouse Brain
It was natural to extend my findings in the rat brain to mice because of the importance of mouse studies for behavior genetics. The most comprehensive study was published by Marie Warembourg and Eduard Milgrom working in the INSERM unit in Lille, France. Their immunocytochemical work replicated my findings in the preoptic area and medial hypothalamus of the mouse brain. These results were confirmed by James Clark at Baylor using biochemical methods, and by endocrine chemist John Cidlowski at the University of North Carolina.
A Wide Range of Vertebrate Species
Fish
Not everyone understands how much of the vertebrate body plan has been conserved from fish to humans. We studied sex steroid-binding neurons in paradise fish with the leadership of visiting University of Michigan professor Richard Davis (Davis, Morrell, and Pfaff 1977).
Starting our summary with the binding of tritiated estradiol in the hypothalamus, we saw excellent hormone retention in the medial hypothalamus, with the greatest intensity of binding in a cell group called nucleus lateral tuberis. Farther posterior we saw estrogen binding in the nucleus recessus lateralis, and some binding in the nucleus posterioris periventricularis. No labeled cells were found in the mesencephalon, rhombencephalon, or spinal cord.
I am not trying to match the fish brain with mammalian brain nuclear group for nuclear group, but the principle that there are a high number of estrogen-binding cells in the hypothalamus remains true in the fish. We were especially pleased about the tightly backed estrogen-binding cells in the preoptic area, nucleus preopticus parvocellularis—most of these are very small cells, and they are found dorsal and posterior to nucleus preopticus periventricularis, also labeled.
What can we say about the limbic system in fish? Considering the limbic system as phylogenetically ancient telencephalon, it was gratifying that we could find estrogen-labeled cells in the area ventralis telencephali pars ventralis. For ma
ny of these cells, just rostral to the anterior commissure and extending dorsally, one could imagine a homology to the mammalian bed nucleus of the hypothalamus.
For testosterone-labeled cells, locations in the brain of the paradise fish were, in general, similar to the estrogen-labeled cells. The “limbic” (telencephali pars ventralis) neurons were essentially identical to the estrogen story, as was the case in the hypothalamus. Overall, there tended to be fewer testosterone-binding neurons than estrogen-binding neurons.
We also confirmed these findings with the green sunfish, a second teleost species, and other laboratories got convergent data from the brain and pituitary of the brown trout.
Amphibians
Rockefeller graduate student Darcy Kelley, a brilliant young woman who went on to become the first female head of the department of biology at Columbia University, worked with Joan Morrell in my laboratory on two species of amphibians, Xenopus laevis and Rana pipiens.
Xenopus are amazing. Their copulatory response lasts longer than I had ever seen. Darcy injected them systemically with small doses of tritiated estradiol (0.1 millicurie) and sacrificed them two hours later to prepare the brain tissue for the steroid autoradiographic procedures I had used for the rat brain.
Starting the description of her results with the diencephalic nerve cell groups that “stand in for” the hypothalamus, Darcy discovered estrogen-binding cells in the ventral infundibular nucleus (Morrell, Kelley, and Pfaff 1975). Especially impressive are the compacted labeled cells along the border of the ventricle.