by Donald Pfaff
Studying protein synthesis was logically required because I knew that new protein synthesis in hypothalamic tissue after estrogen treatment was required for lordosis (see Chapter 1). Following up those studies, Rockefeller graduate student Bruce Parsons showed the exact timing of the necessary protein synthetic events (Parsons et al. 1982; Parsons, McEwan, and Pfaff 1982). Not only that, but from the work of Richard Harlan in the laboratory, it was clear that such proteins had to be axonally transported out of the hypothalamus, accompanied by action potentials (see Chapter 1). Which exact hypothalamic neurons were involved? Paula Davis’s localized implants showed that estrogen administration to the ventromedial nucleus of the hypothalamus (VMH) neurons was sufficient (Davis, McEwen, and Pfaff 1979), and Bob Meisel’s work with localized antiestrogen implants demonstrated that estrogen’s impact on VMH neurons was necessary for lordosis.
Peter Blackburn, with Professor Moore’s encouragement, taught us to use liquid chromatography to separate radioactively labeled proteins after microinjections of carefully chosen labeled amino acids into the VMH (Pfaff, Rosello, and Blackburn 1984). In the experiments, we used matched pairs of female rats; in each matched pair, one had been ovariectomized (thus leaving the animal devoid of estrogen) and then given estradiol treatment for 7 days. All these rats exhibited lordosis behavior. The other of the matched pair was ovariectomized and only given the vehicle control treatment. None of these controls exhibited lordosis behavior. Then the microinjection into the VMH was followed by various survival times ranging from 2 hours to 2 days. The best results were at 4 hours and were positive in two very encouraging ways. Not only were there significantly more radioactively labeled proteins in the VMH tissue, but also some of these had been transported along axons to the midbrain central grey (MCG). One specific class of proteins in the central grey, probably a class of molecular weights based on the protein separation column we used, was more than ten times more abundant in the estrogen-treated animals than in the controls.
A second approach to estrogen-stimulated protein synthesis in the hypothalamus was initiated by Charles Mobbs. Before joining us at Rockefeller, Charles worked his way through MIT, and then got his doctorate in endocrine biochemistry at the University of Southern California. Charles tested our theory, based on some of the work summarized in Chapter 1, that systemic administration of estrogen would lead to novel proteins synthesized in the VMH. In doing so he brought the performance of two-dimensional gel analysis of radioactively labeled new proteins in the brain to a high art. This was time-consuming work, and while he was waiting for gels to run he would, in front of a huge mirror, practice his tap dancing in the marble-floored halls outside our laboratory doors.
The basic technique of two-dimensional gel analysis relies on separation in one dimension according to the molecular weight of the protein, yielding gels in the form of tubes. Then those tube gels are separated again, using a carefully chosen pair of stacking gels and resolving gels; in this latter step, proteins are separated according to their isoelectric points. Thus we produce two-dimensional gel electrophoresis. In our work, ovariectomized female rats had been prepared either with constant estrogen treatment or with the vehicle control 7 days before further experimentation, at which time they were infused with radioactive amino acids. Our best results came 14 hours after the radioactivity infusion. Consequent radioactively labeled proteins were detected with a technique known as fluorography, in which a Kodak film with high spatial resolution and high sensitivity to the S35 radioactive label is pressed against the gels produced as described. The high degree of strategic and methodological “art” that Charles Mobbs achieved (Mobbs et al. 1988) was based, at a minimum, on 1) the choice and timing of the first gel, 2) the choice and timing of the second gel, together with choice of electric field strength, 3) the choice of buffers, and 4) the exposure time of the Kodak film.
As a result of this extensive work, we discovered an estrogen-induced protein that is synthesized in the VMH and transported to the MCG. In those days, our detecting more than 250 protein “spots” of black, radioactive proteins represented an achievement due to Charles’s scientific artistry. In addition, Charles’s gels had superb resolving power when subjecting the films to computerized densitometric analysis. The big story is that we had found a specific protein with an apparent molecular mass of 70 kilodaltons and an isoelectric point of 5.9 that was clearly present in almost all the estrogen-treated females’ VMH and in only two of sixteen of the control animals. The same significant result for that exact class of protein was obtained in the MCG. In fact, disrupting axoplasmic flow from the VMH to the MCG by colchicine treatment reduced the arrival of radioactivity at the central grey by approximately 85 percent (Mobbs et al. 1988). This finding logically sent us on a hunt for estrogen-induced gene expression in the hypothalamus. We had united molecular biology and behavioral biology (Mobbs, Fink, and Pfaff 1990; Mobbs et al. 1989).
Robert Lustig, another MIT graduate, had trained as a medical endocrinologist; he worked with Charles to follow up his work. Subsequent to Robert’s laboratory work at Rockefeller, he went on to a clinical endocrine practice and has become a nationally recognized expert, warning the public about the deleterious consequences of ingesting sugar at high levels. His methodology was similar to that described here, using two-dimensional gels, and his idea was to extend Charles’s observations to include quantitative measurements of the synthesis of specific proteins in the VMH as a function of estrogen treatment (versus control) (Lustig, Pfaff, and Mobbs 1989).
His results confirmed the induction of a protein whose apparent molecular mass is 70 kilodaltons with an isoelectric point of 5.9, but his method of analysis allowed us to go much farther. Instead of focusing exclusively on that one hormone-induced protein, Robert used a flatbed laser scanner to quantify the optical density of each of the two-hundred and forty spots on the gel. Three new protein-spots were subsequently identified as quantitatively induced (72 to 96 percent increases) by estrogen treatment. Further, slight increases in the isoelectric point of two other proteins suggested estrogen-induced posttranslational modification. Again, Robert’s results reinforced our efforts to discover estrogen-influenced gene expression in the hypothalamus as causal to lordosis behavior.
Motivated by the foregoing proof that new protein synthesis is necessary for the estrogenic facilitation of lordosis behavior (Chapter 1), Rockefeller graduate student Bruce Parsons did work that justified Charles Mobbs’s and Robert Lustig’s studies (Parsons et al. 1982; Parsons, McEwen, and Pfaff 1982). Bruce used three different doses of the very effective protein synthesis inhibitor anisomycin (80 percent decrease in hypothalamus). Timing was crucial. If the anisomycin was administered just before an estrogen treatment, lordosis was abolished. If instead anisomycin was given well after estrogen treatments but hours before the behavioral assay, lordosis was at normal, control levels. This last control was important because it proved the anisomycin was not reducing lordosis simply because the animals were ill. Thus, the effect of protein synthesis reduction was specific to the estrogenic facilitation of lordosis. Again, this type of result made us want to measure levels of specific mRNAs in specific hypothalamic neurons.
In situ hybridization. Moving on to the mRNA level, we wanted cellular resolution in our data and thus were the first to work out a technique for in situ hybridization in the brain (McCabe et al. 1986). Using a precise, quantitative approach to in situ hybridization, Andrea Lauber, as mentioned in Chapter 1, not only proved the expression of the mRNA for ER-α in VMH neurons but also demonstrated that 1) estrogen administration reliably led to a decrease in expression of mRNA for ER-α, and 2) this phenomenon is sexually differentiated, with the results of females being much stronger than those of males.
Specific Transcriptional Systems Proven to Be Involved
Estrogen hormones bind to ER-α. Liganded, the receptor quickly recognizes and binds to specific DNA sequences—the consensus estrogen response element (ERE) on DNA is AGGTCAnnnTGACCT.
The n can be any nucleotide base but there must be three of them. Some substitutions in the consensus ERE are allowed. These EREs for “turning on” estrogen responsive genes in brain and other tissues can be found in the genes’ (proximal) promoters or their (distal) enhancers (see Klinge’s section in Kow et al. 2016).
Figure 3.1. Several genes have two properties: 1) estrogens facilitate their expression, and 2) their products foster lordosis behavior. (Updated from Pfaff 1999.)
Here is the logic we will follow. Several transcriptional systems have the following two properties: 1) estrogen treatment raises their mRNA levels in VMH neurons, and 2) the final gene product facilitates lordosis behavior (Figure 3.1). Thus, because we know that new mRNA synthesis and new protein synthesis are required for estrogen effects on lordosis, we take these two properties and make a logical inference in the form of a syllogism. I like to use the idea of a syllogism, that is, reasoning in the form: “If (a) John is a scientist and (b) all scientists are good, then (c) John is good.” In the case of each transcriptional system here, the logical inference will be that revving up of that transcriptional step serves to facilitate estrogen-dependent lordosis behavior.
Progesterone Receptor
Gary Romano entered the laboratory as a Rockefeller graduate student and immediately dispensed with the stereotypes of what precise scientific operations require. Gary had hands with the size and muscularity of a professional football lineman, yet he operated at the laboratory bench with a degree of order and precision that could hardly be equaled. The first breakthrough in his work was enabled by a long conversation with the great French molecular biologist Pierre Chambon. Pierre had cloned the progesterone receptor (PR) and generously gave us his reagents and advice. As a result, Gary learned our technique for in situ hybridization and compared PR mRNA levels in the VMH (with implications for lordosis) as a function of time after estrogen administration compared with control ovariectomized rats (Figure 3.2) (Romano, Krust, and Pfaff 1989). The PR mRNA levels already started up within 4 hours of estrogen administration and reached their peak at 24 hours. By 48 hours, they were headed back down.
Figure 3.2. Black dots represent neurons in which the expression of the progesterone receptor gene has been turned on by estrogen administration. Strong estrogen effects were seen in the ventromedial nucleus (VM) and the arcuate nucleus (ARC) of the hypothalamus. Expression is bilateral, plotted here on one side. (Adapted from Romano, Krust, and Pfaff 1989.)
Andrea Lauber came to Rockefeller after she got her Ph.D. from the University of California, and she worked with Gary Romano to explore the specificity of the estrogenic induction of PR mRNA (Lauber et al. 1991). She knew that the most important determinant of the ability of progesterone to facilitate reproductive behavior was the level of PR. First she replicated and extended Gary’s work by demonstrating a 3.5 times induction of PR in the VMH of the female. Then she showed that in the genetic male hardly any PR induction occurred after estrogen treatment. All these data will play into my “transcriptional formula” for behavioral regulation at the end of this chapter.
Some of my earliest studies on lordosis behavior had proven the efficacy of long-term estrogen exposure. In fact, I never considered molecular endocrine systems to be signaling systems with instantaneous time-constants because hormonal dynamics, as they affect various target tissues (including brain tissue), simply take time. Rockefeller graduate student Bruce Parsons took up the call. Crucially, Bruce demonstrated that long-term treatment with estrogens greatly facilitated the subsequent response of the brain to a later estrogen treatment as well as to a later estrogen plus progesterone treatment (Parsons et al. 1979). This was true whether that later estrogen treatment was followed by a lordosis behavior assay or a PR assay. The practical importance of Bruce’s work was that it revealed the inappropriateness of a major National Institutes of Health (NIH) clinical trial of hormone replacement therapy in which the patients had been without their estrogens for years—those subjects were lacking the long-term estrogenic support revealed as crucial by the Parsons et al. study. Under the circumstances in which the trial was terminated, molecular endocrinologists summarized “wrong trial, wrong analysis, wrong decision.” Our case was much simpler. We knew that blocking the synthesis or action of PR would block the ability of progesterone to facilitate estrogen’s actions on lordosis behavior (Ogawa et al. 1994).
To prove the parallelism between estrogenic effects on PR and estrogenic effects on lordosis behavior, Bruce made clever use of discontinuous schedules of estrogen treatment (Parsons et al. 1982; Parsons, McEwen, and Pfaff 1982). He had already shown great parallelism between the molecular end point and behavior (Parsons et al. 1980). That is, as a function of time after estrogen administration both PR and behavior rose steadily, starting at 12 hours and peaking at about 48 hours. Likewise, after removal of estrogen from the body they both decayed, reaching levels indistinguishable from controls at 36 hours after withdrawal. Then, regarding the discontinuous schedules of estrogen treatment, we analyzed the 24 hours of estrogen preceding the progesterone treatment that would finish the hormonal support for high levels of lordosis (Parsons et al. 1981, 1982). Brief estrogen exposures were accomplished by popping small silastic capsules in and out of the subcutaneous space. It turned out that two brief estrogen exposures during the 24 hours before progesterone were sufficient for lordosis if they were separated by at least 4 hours and not more than 13 hours. Well-timed use of the protein synthesis inhibitor anisomycin showed that each brief estrogen exposure needed to initiate new protein synthesis for the behavior to occur. Most important for this chapter: the same conditions that were adequate for lordosis behavior also were sufficient for PR synthesis.
Beyond the parallels between the induction of PR and the initiation of lordosis, we were able to prove that PR expression is necessary for lordosis. Sonoko Ogawa in my laboratory (now a professor of neurobiology at Tsukuba University in Japan) pioneered our use of antisense DNA technology. In this approach, we synthesize a string of nucleotide bases that are complementary to and will bind to a specific messenger RNA (in this case the mRNA for PR), thus opening that mRNA to attack by nuclease enzymes and, as a consequence, to being put totally out of action. Sonoko microinjected antisense DNA among neurons of the VMH and massively reduced the amount of lordosis behavior compared with control females given only the vehicle (Ogawa et al. 1994). Sonoko’s work was replicated by Shaila Mani in Bert O’Malley’s laboratory at Baylor College of Medicine. Our results were also complemented by findings from other laboratories that used a PR blocker called RU486 [(8S,11R,13S,14S,17S)-11-[4-(dimethylamino)phenyl]-17-hydroxy-13-methyl-17-prop-1-ynyl-1,2,6,7,8,11,12,14,15,16-decahydrocyclopenta[a]phenanthren-3-one]. Females given estrogen and progesterone but also RU486 showed only low levels of lordosis behavior. As a side point, Anne Etgen, at Albert Einstein College of Medicine showed that RU486 also blocked the ability of the enkephalin / δ-opioid system to facilitate lordosis. The powerful causal role of PR transcription in fostering lordosis behavior was thus proven beyond a doubt.
Thus, (a) estradiol, bound to the ligand-activated transcription factor ER-α, turns on the gene for the PR (another ligand-activated transcription factor). Chris Krebs in our laboratory followed up the progesterone-sensitive genes (Krebs et al. 2000). (b) In any case, the progesterone receptor is necessary for progesterone to be able to amplify estrogenic effects on lordosis. Therefore, (c) it follows that one molecular path by which estrogens facilitate lordosis is by ramping up the transcription of the PR.
Alpha Adrenergic Receptors
Having worked with the massive amplification of estrogenic effects on behavior by progesterone through the PR, it now seemed logical—and well within the purview of late-twentieth-century neurobiology—to work with neurotransmitter receptors. We dealt effectively with two: α-adrenergic receptors and muscarinic receptors.
It was clear from careful neuroanatomical studies using in situ hybridization that adrenergic-α1 mRNA is express
ed in neurons of the VMH. The main question for us was what effect estrogens might have. The rate-limiting enzyme for the receptor’s ligand, norepinephrine, is tyrosine hydroxylase. Its gene’s transcription is reduced in the brain through deprivation of estrogen by ovariectomy and is increased by estrogen treatment. As reviewed authoritatively by Anne Etgen, the ligand’s release is increased by estrogens as well. Further, she showed that estrogens increase α1 binding sites in ovariectomized rats by four to six times in the hypothalamus, and this huge binding increase is accompanied by a significant rise in α1 mRNA levels.
It follows that Lee-Ming Kow in our laboratory, using extracellular recordings of individual VMH neurons, was able to demonstrate that estrogen treatment heightens electrical responses to treatment (in the recording bath) with α1 receptor ligands (Kow and Pfaff 1985, 1995).
What ion channels are involved in the excitation of VMH neurons by α1 receptor adrenergic agonists? Our work proceeded apace with that of Anne Etgen’s molecular work at Einstein. α1B-noradrenergic transmission can facilitate estrogen-dependent lordosis behavior (Kow, Weesner, and Pfaff 1992). Estradiol treatment in vivo increases levels of mRNA encoding the α1B receptor in the hypothalamus. In the VMH, estradiol treatment potentiates α1B-adrenergic signaling by increasing the proportion of neurons that respond to stimulation of α1B-adrenergic receptors (α1B-AR). Thus, estradiol may modulate α1B-adrenergic receptors in the VMN to promote neuronal excitability and ultimately lordosis.
