by Donald Pfaff
Figure 1.2. Black dots show locations of neurons expressing estrogen receptors in the ventral infundibular nucleus (VIN) in the hypothalamus of Xenopus. The expression is bilateral, plotted here on one side. (Adapted from Morrell, Kelley, and Pfaff, 1975.)
As expected from my work with rats, estrogen-concentrating cells are seen throughout the entire extent of the anterior preoptic area. In the preoptic area slightly more posterior they were still found throughout the dorsal-ventral extent of the preoptic area, but still more caudally they were clustered close to the ventricle, ventrally.
Considering the telencephalon (Morrell, Kelley, and Pfaff 1975), the ancient forebrain, the labeled cells were found in the septum, the striatum, and the amygdala. For example, in the ventral lateral septum, they were curved ventrally around the tip of the ventricle. Likewise, the amygdala’s labeled neurons were close to the posterior tip of the ventricle, seen most easily in horizontal section (Figure 1.2).
As is the case with rats and mice, estrogen-labeled cells extended posterior into a specific part of the midbrain. In amphibians, the structure lateral and dorsolateral to the cerebral aqueduct is called the torus semicircularis.
Comparing estrogen-binding patterns of neurons in neuroanatomical and systems terms, we found no apparent differences between the female brain and the male brain. Even the intensity of labeling (grains per cell) was about the same.
Darcy Kelley also studied testosterone binding in the brain (Kelley, Morrell, and Pfaff 1975). As with estrogens, brains were harvested two hours after systemic injection of tritiated testosterone. Again, beginning with the “hypothalamus equivalent,” she found well-labeled neurons in the ventral infundibular nucleus, especially posterior in this nucleus. Whether these are “exactly the same neurons” as can bind estrogen is open to conjecture. Such a question depends on how you define a “cell type,” a developmental biological problem we recently have argued in the light of modern epigenetics (Tabansky, Stern, and Pfaff 2015).
In the preoptic area as well, the anterior subdivision contains large numbers of labeled cells following systemic tritiated testosterone. As you move caudally in the preoptic area, the density of labeled cells declines, especially in its dorsal portion.
We were surprised by the labeled cells in the dorsal tegmental area of the medulla, and in a columnar nucleus in the hindbrain (Kelley, Morrell, and Pfaff 1975). We speculated (p. 57) that based on a comparison with the literature on androgen-modulated vocalizations, these more posterior testosterone-binding neurons are involved in the regulation of mating calls, a speculation that Kelley followed up effectively at Columbia.
Thin-layer chromatography of radioactive hormone in the brain showed that the great majority of hormones in the brain were still in the form of testosterone itself rather than a metabolite.
Another amphibian, Rana pipiens, was used for cell fractionation studies of hormone binding in the brain after systemic injections of radioactive estradiol or radioactive testosterone (Kelley et al. 1978). In the block of tissue that included preoptic area, infundibulum, and amygdala, radioactivity from the labeled estradiol was about three times that found in other brain tissue, but less than that found in the pituitary. After testosterone injections, the radioactivity was about the same as in other brain tissue, presumably due to the hindbrain groups of labeled cells discussed previously, and a substantial fraction had been converted into metabolites such as dihydrotestosterone. In fact, the hypothalamic / limbic block of tissue just mentioned had five times as much radioactive dihydrotestosterone in cell nuclei as in the rest of the brain.
Rana was also used for autoradiographic studies of hormone-binding neurons. For both estradiol and testosterone, the labeled cells were impressive in the infundibular nuclear groups, in the preoptic area, and in the ancient telencephalon around the ventromedial tip of the telencephalic ventricle. Labeled cells were seen in the septum and the amygdala. Hormone-receiving neuronal distributions were, by and large, very similar to the results for Xenopus. However, Rana lacked the hindbrain testosterone-concentrating cells seen the Xenopus brain. Thus, we referred to literature that indicated regulation of mate calling might be under the control of the preoptic area and infundibulum.
It was in this study that we extended my earlier argument (Pfaff and Keiner 1973) that sex steroid-binding nerve cells are connected in systems—that is, that they project to each other (Kelley et al. 1978). I explore the implications of this phenomenon later.
Reptiles
Joan Morrell at Rockefeller worked with reptile neuroendocrinology expert David Crews (now at the University of Texas) to study sex hormone receptors in the brain of the lizard Anolis carolinensis (Morrell et al. 1979). Females and males were injected, and the brains were harvested two hours later.
For tritiated estradiol, we concentrated first on the hypothalamus. Although few labeled cells were seen in the anterior hypothalamus, the ventromedial nucleus of the hypothalamus was well labeled. In fact, in its posterior lateral segment it was plastered with heavily labeled cells. This is reminiscent of what we originally saw in the rat brain.
The periventricular nucleus of the hypothalamus had many heavily labeled cells throughout its extent. In commonsense terms, this nucleus looks like the reptilian version of the mammalian arcuate nucleus.
The reptilian preoptic area as well showed results similar to those in mammals. A large number of heavily labeled cells was found throughout its extent. As with rat brains, estrogen-binding cells were particularly densely concentrated near the preoptic recess of the third ventricle.
The reptilian forebrain comprises an early version of the limbic system in mammals. We found estrogen-labeled neurons in the septum, particularly in the ventral and lateral septum. Amygdaloid subnuclei in the lizard brain are divided into cell groups named A1, A2, and A3. A1, the most lateral cell group, and A2 had a large number of labeled cells. A3 was surrounded by labeled cells at its dorsal and medial edges but internally was not labeled.
A smaller number of estrogen-concentrating neurons was found in the midbrain, but they were found in regions that seem to match the pattern for the mammalian brain. That is, we saw them in the midbrain central grey lateral to the cerebral aqueduct, and in a structure called the torus semicircularis, again lateral to the aqueduct.
In other animals we injected labeled testosterone or labeled dihydrotestosterone. In general, the neuroanatomical pattern of the labeled cells was very similar to that just described for estradiol. Notably, as predicted from our work with rats and from the notion of a universal vertebrate pattern of sex hormone-binding neurons, androgen hormone binding was strong in the medial hypothalamus and in the preoptic area. Also, large numbers of androgen-concentrating neurons were seen in the amygdala and the septum.
An important difference from the estrogen results, after labeled androgenic hormone administration, lay in the significant number of neurons seen in the mesencephalic tegmentum. Estrogen binding there was rare. We have not been able to link these androgen-binding cells to male sex behavior, but some investigators have suggested that these may be involved in species-specific aggressive displays.
Throughout our work with lizard brains, the patterns of hormone-binding cells were essentially identical between male and female. This was true both after labeled estradiol administration and testosterone administration.
Another reptile project used garter snakes and benefited from a collaboration with the skilled neuroanatomist Mimi Halpern, professor and later dean at the Downstate Medical Center of the State University of New York (Halpern, Morrell, and Pfaff 1982).
Brains were harvested two or three hours after labeled hormone injection. In considering the labeled estrogens, the anterior hypothalamus had a rostral and dorsally located cell group that contained large numbers of very densely labeled cells. The ventromedial hypothalamic nucleus is large and spherical and densely packed with nerve cell bodies. Labeled cells could be found there in specific subregions: a dorsal subgroup and a poste
romedial subgroup. As expected, the periventricular nucleus was well labeled, especially in its ventral aspect. Farther posterior a cell group close to the ventricle that is called the arcuate nucleus had the largest proportion of labeled neurons and the densest labeling of any area in the snake brain. Similar to other species studied, the thalamus rarely had any labeled cells.
In the snake brain, the preoptic area is similar to that in mammals, cell-dense in its medial nucleus and cell-poor laterally (containing the medial forebrain bundle). The medial preoptic neurons are extremely densely labeled, and there are a large number of them.
Regarding the limbic system, cells in the septum are lightly labeled. In the snake brain we do not talk about the amygdala; instead, in the lateral telencephalon, we have the nucleus sphericus. Cells in this nucleus are labeled, especially on the medial side, and there is excellent labeling in the bed nucleus of the stria terminalis.
In the midbrain, estrogen-binding neurons could be seen in the central gray and just lateral to it. The distribution was similar to the mammalian brain, but there are fewer cells.
Were testosterone-labeled cells significantly different than the estrogen-labeled cells? Overall, animals injected with tritiated estradiol had more densely labeled cells and a greater number of them. But the overall neuroanatomical pattern was similar.
As with the lizard brain, we saw no consistent differences in the patterns of intensities of labeling in the brains of males and females receiving a given treatment of labeled hormone.
Birds
It was natural enough for us to collaborate with our friend and Rutgers University neighbor, professor of biology Ronald Barfield, and the bird of choice was the domestic fowl—chickens (Barfield, Ronay, and Pfaff 1978). Even before the intravenous injections of radioactive testosterone I was instructed on how to hypnotize a chicken by a rapid finger movement in front of the bird’s face. This way the axial vein could be viewed directly while the bird was restrained with its wing extended. Brains were harvested one hour after injection because of the rapidity of the intravenous route.
The most densely labeled cells were, as expected, in the hypothalamus and preoptic area. The posterior portion of the medial hypothalamic nucleus displayed many testosterone-binding cells, but the lateral hypothalamus was not labeled at all. The preoptic area, especially in its medial and suprachiasmatic subdivisions, had a large number of labeled cells.
As far as the limbic system is concerned, in the bird we must look to the archistriatum, which had testosterone-binding cells ranging from lightly to quite heavily labeled. The most impressive were on its medial side in the so-called nucleus taeniae, which looks to me as though it occupies a neuroanatomical position analogous to the medial amygdala.
In the midbrain, the central grey itself had no labeled cells, but a structure lateral to it, the nucleus intercollicularis, retained as much tritiated testosterone as any neurons in the chicken brain.
Thus, results with chicken brains conformed to the vertebrate-wide pattern of limbic / hypothalamic sex steroid-concentrating cells.
Arthur Arnold, now a professor at the University of California–Los Angeles, worked with Rockefeller professor Fernando Nottebohm and me to localize testosterone or its metabolites in the brain of the zebra finch, a songbird that Fernando had made extremely popular for studying the mechanisms of song learning. Finches were double-injected 1.5 and 1.0 hours before sacrifice, and the brains, accompanied by a large number of control groups, were given autoradiographic exposures under our usual dark / dry / cold / lead box conditions for periods of 5 to 13 months (Arnold, Nottebohm, and Pfaff 1976).
I describe the results in two parts: the first part fits the usual limbic / hypothalamic pattern I began with in the rat brain. The second part is particular to the regulation of androgen-dependent singing in the zebra finch.
In the hypothalamus, an anterior field of heavily labeled cells extends from the back of the preoptic area all the way to the magnocellular periventricular nucleus. A more posterior hypothalamic field of labeled cells near the midline paraventricular organ. This heavy labeling in cells of the infundibular nucleus looks most like the ventromedial hypothalamic and arcuate nucleus labeling in rats and mice. More labeled fields were lightly labeled or not labeled at all.
Large numbers of testosterone binding cells were detected just anterior to the anterior commissure and also directly underneath it. These are exactly in the position to match our rat brain preoptic distribution of cells. The extension of labeling back into the midbrain was impressive. The most prominent labeling, in the dorsomedial region of the nucleus intercollicularis, matched what I had reported for the chicken brain.
Limbic system labeling was strong. A far-anterior cell group named the magnocellular nucleus of the anterior neostriatum was impressive, as was the labeling in the septum. The medial side of the archistriatum had testosterone-binding cells, including the nucleus taeniae (as also seen in the chicken brain). As with other species, some of the most beautiful labeling occurred near the ventral trip of the lateral ventricle. Thus, the general “limbic / hypothalamic system” formula adopted from the rodent brain can be seen here in the zebra finch brain.
The second emphasis in the results concerns what is needed for the hormonal facilitation of song in this bird. Back in the midbrain and hindbrain, in addition to labeled cells in the nucleus intercollicularis, heavily labeled cells were found in the tracheosyringeal portion of the hypoglossal nucleus in the medulla. Recognizing this as the motor nucleus of the hypoglossus gave us adequate reason to believe that it is centrally involved in the androgen-dependence of song in these animals. More than that, Art Arnold did calculations that suggested that all the neurons in this nucleus might accumulate testosterone and further proved by retrograde degeneration studies that they actually supply the syringeal musculature that produces birdsong. Moreover, a telencephalic cell group named the hyperstriatum ventrale, pars caudale (HVc), where lesions disrupt singing, also had well-labeled cells. Thus, in the song-control system, the HVc, nucleus intercollicularis, and motor nucleus of the hypoglossus all can work together, seriatum, to account for the testosterone-dependence of song production.
A third avian species we studied was the chaffinch. Richard Zigmond had begun the autoradiographic work by focusing on testosterone accumulation by neurons in an area of the midbrain known to be involved in the control of testosterone-dependent birdsong, the nucleus intercollicularis. Indeed, he found heavily labeled cells there so dramatic that he wanted to submit an article to Science. Because the format of Science was so different from that of other journals and because the journal was so exclusive, I advised him that it was too much trouble to submit there. But by then he already had submitted—and his article was published without even a request for change (Zigmond, Nottebohm, and Pfaff 1973).
In a later more comprehensive study (Zigmond, Detrick, and Pfaff 1980) that I will summarize only briefly, Richard found large numbers of testosterone-accumulating cells in the hypothalamus anteriorly in the periventricularis magnocellularis and also in the infundibular nucleus (like the mammalian arcuate nucleus). Well-labeled cells were discovered, as expected, in the medial preoptic area.
Regarding the limbic system, testosterone-binding neurons were found in the lateral septum and the nucleus magnocellularis neostriatalis anterioris. As usual, a large number of labeled cells is found in the medial part of the forebrain near the ventricle.
In addition to the limbic / hypothalamic distribution, we paid attention to cells that might have to do with song control. In addition to the nucleus intercollicularis, motor neurons, likely important for song, were labeled in the nucleus of the hypoglossal nerve. Thus, throughout our studies with bird brains we carried forward two themes: replication of the vertebrate-wide limbic hypothalamic system, and the addressing by testosterone of neurons important for testosterone-facilitated song.
Returning to mammals, we came to a carnivore, the mink Mustela vison (Morrell, Ballin, and P
faff 1977). These were scary to study because they could bite your hand off. Estradiol binding was studied in the brains of females that, on the mink ranch, were proven breeders, and that, in our laboratory, were housed in cages identical to those on the mink ranch. These are seasonal animals, and we studied them both in their estrus and anestrous conditions. Neuroanatomical results were the same for the two groups.
As expected, in the hypothalamus the arcuate nucleus and the ventrolateral section of the ventromedial nucleus were plastered with labeled cells. Estrogen-binding neurons were also seen in the anterior hypothalamic area and were scattered just outside the ventromedial nucleus. In the medial preoptic area, labeled neurons were packed especially densely in its medial and suprachiasmatic portions.
In the limbic system, starting from the anterior and working posteriorly, we found estrogen-concentrating cells in the lateral septum and in the bed nucleus of the stria terminalis. Very few labeled cells were in the medial septum. The medial nucleus of the amygdala had a substantial number of labeled neurons, and some were also seen in the cortical nucleus. The labeled neurons in the hippocampus were the large pyramidal cells in the ventral hippocampus and cells in the granular layer of the dentate gyrus. As in the other species we had studied, estrogen-binding neurons were found in the mesencephalic central grey and lateral to it in the deep layers of the superior colliculus.
Primate Brains
By collaborating with medical doctors Michel Ferin, Peter Carmel, and Earl Zimmerman from Columbia College of Physicians and Surgeons, we gained the exciting opportunity to study estrogen-binding neurons in the brain of the female rhesus monkey (Pfaff et al. 1976). The same limbic / hypothalamic system I had seen in the rat brain some years earlier showed up again in the primate brain.