How the Vertebrate Brain Regulates Behavior

by Donald Pfaff

  We had the courage to study the migration mechanisms only because of the NCAM expertise of the Edelman laboratory. Since then, many laboratories have begun to piece together the mechanisms underlying GnRH neuronal migration. Much to our surprise, GnRH neurons have been found to have cilia. Novel chemicals on the migration route have been discovered, such as an embryonic luteinizing hormone-releasing factor. The gene product from sirtuin-1 (Sirt1) encodes a catalytic function that induces GnRH neuronal migration via binding and deacetylating the protein cortactin in GnRH neurons, so this gene is required for normal migration. Mutations in the Kall gene itself interact with mutations in mitochondrial genes to disrupt mutation. The myosin superfamily has been called into play. And all of this, as discovered by Stuart Tobet and Margaret Weirman at the University of Colorado Medical School, can be influenced by the epigenetic influences of interactions among histone deacetylases.

  Further, a Belgian group found that Kallmann’s syndrome is associated with mutations in the KAL1, FGFR1 / FGF8, FGF17, IL17RD, PROK2 / PROKR2, NELF, CHD7, HS6ST1, FLRT3, SPRY4, DUSP6, SEMA3A, NELF, and WDR11 genes that are related to defects in neuronal migration. Susan Wray herself got evidence for the importance of P75 nerve growth factor receptors for the developing GnRH system. Still another group showed that the FEZF1 gene product enables axons of olfactory receptor neurons to penetrate the central nervous system basal lamina in mice. Because a subset of axons in these tracks is the migratory pathway for GnRH neurons, in FEZF1 deficiency GnRH neurons also fail to enter the human brain.

  It is indeed becoming clear that a complex and heterogeneous set of gene products is required for normal GnRH neuronal migration. From Marlene’s and my point of view, the greatest pleasure was that, again, the discovery turned out to be true for all vertebrates studied, from fish to philosopher. Ishwar Parhar’s original work with fish brains and all these studies of Kallmann’s syndrome attest to a finding universal among vertebrates.

  Figure 5.3. The decapeptide gonadotropin-releasing hormone (GnRH, formerly known as luteinizing hormone-releasing hormone), famous for the stimulation of the ovulatory luteinizing hormone surge release from the pituitary gland, was shown by our laboratory to foster female reproductive behavior.

  Brain and Behavior

  The decapeptide GnRH gave me the opportunity to link chemistry (Figure 5.3) to physiology to behavior in a series of causal links. My work depended on the chemistry worked out by two Nobel Prize–winners, Roger Guillemin, who grew up in Dijon, France, and worked at the Salk Institute, and Andrew Schally, who grew up in Poland and worked at Tulane University. I mention this because everyone in my field knew that these two men could not stand each other and competed bitterly to be the first to describe the chemistry of hypothalamic neuropeptides that act on the anterior pituitary. Roger Guillemin often seemed quite aristocratic and a bit scary, but I benefited from his demonstration of the chemistry of GnRH and his gift of some GnRH to me.

  In 1972 it was easier to do research on the brain and behavior than it is now. The bureaucracy, most of it the product of federal regulations, has increased exponentially. Thus, I got the idea of looking for direct effects of GnRH on behavior about 9:00 PM on a Tuesday night, and I ordered the animals from Charles River Laboratories about 9:00 AM the next Wednesday morning. The entire experiment was done in the time it takes now to get an animal use protocol approved.

  What I did (1973) was to order, from Charles River, female rats in which both the ovaries and the pituitary had been removed surgically. After they were acclimatized to my laboratory’s light cycle, they were given a low dose of estradiol. I wanted to make it as easy as possible to demonstrate a GnRH effect on behavior by giving enough estrogen such that lordosis would not be impossible, but to keep the estrogen dose at a subthreshold concentration—just below the amount that would increase reproductive behavior with estrogen alone. Matched triplets of animals were then given the vehicle control, a subcutaneous injection of a lower or higher dose of GnRH at the beginning of the lights-off phase of the daily light cycle. That is, I was trying to get the GnRH effect on behavior just about the time that ovulation would occur in a female’s normal estrous cycle. The higher dose (4 micrograms, subcutaneous) gave significant results 90, 180, and 360 minutes after injection (Pfaff 1973) (Figure 5.4).

  Figure 5.4. Subcutaneous injection of the neuropeptide gonadotropin-releasing hormone (GnRH) significantly elevated lordosis behavior according to dose (0.4 or 4.0 micrograms). Thus, the neuropeptide that activates the endocrine physiology of reproduction also stimulates the behavior required for reproduction. (Adapted from Pfaff 1973.)

  In terms of controls, I not only had vehicle controls, but I also used animals in which the pituitary gland had been removed surgically before the experiment. That is, my GnRH effect on behavior was direct; it could not have been due to effects on the pituitary because the pituitary was not there. Many years before I already had shown that pituitary hormones are not necessary for estrogen-induced lordosis behavior (Pfaff 1970).

  We knew that GnRH could increase lordosis by acting directly on the VMH. In fact, mutations of the gene for steroidogenic factor-1 that interfere with the development of the VMH abolish the GnRH effect on behavior. In addition, we used current electrophysiological techniques to show neuromodulatory effects of GnRH on norepinephrine signaling in the hypothalamus (Pan et al. 1988).

  James Pfaus brought a new dimension to the work. He had already studied Fos-like immunoreactivity that, within forebrain regions, was specific to afferent sexual sensory stimulation and did not require treatment with estrogen and progesterone (Pfaus et al. 1994). Now he primed the female rats with estrogen and progesterone and found Fos-like immunoreactivity within a significant number of GnRH-positive neurons in the anterior preoptic area caudal to the organum vasculosum after copulation with intromission or cervicovaginal stimulation as compared with no stimulation. For the first time, this use of the afferent, sensory input brought the GnRH neurons into the behavioral picture.

  Meanwhile, much to my surprise, Robert Moss and Don McCann in Dallas were doing similar experiments. They also showed that GnRH injected subcutaneously could elevate lordosis behavior. The timing of their experiment was slightly different, but most importantly they used different controls than I did. They showed that luteinizing hormone, follicle-stimulating hormone, or thyrotropin-releasing factor did not induce lordosis. That is, while competing with my laboratory, they actually strengthened the conclusion because I had controls that they did not and vice versa. Together, the two reports led to an entire field of work on neuropeptides and behavior.

  My findings were replicated many times. Ann Etgen, a professor at the Albert Einstein College of Medicine, in cooperation with Oscar González-Flores showed our GnRH effect on behavior interacted with progesterone and its metabolites and also with leptin effects on lordosis behavior. They also connected the GnRH effect to nitric oxide (see Chapter 3) and to the cyclic AMP / protein kinase A pathway. In fact, the Argentinian physiologist Viviane Rettori thought that the oxytocin effect on lordosis (discussed in Chapter 6) depended on the nitric oxide / GnRH connection. Moreover, later work showed that the behavioral effects of GnRH are blocked by a GnRH receptor antagonist, thus demonstrating that my effects of GnRH on lordosis are direct rather than indirect.

  Yasuo Sakuma, who had joined us from the Yokohama University School of Medicine, took the next logical step (Sakuma and Pfaff 1980, 1983). Because the VMH projects to the midbrain central grey (MCG) at the top of the lordosis behavior circuit (Chapter 2) and since Richard Harlan, Brenda Shivers, and I had shown that GnRH neuronal axons project through a periventricular route to the MCG, Yasuo examined the behavioral effects of GnRH applied directly to the MCG. He prepared female rats with bilateral intracerebral cannulae directed to the dorsal portion of the MCG, where the majority of the synapses from VMH axons are located. Four days before behavioral assays, the animals were given a preparatory dose of estradiol, high enough to prime lordosis be
havioral circuitry but not high enough to cause a high level of lordosis all by itself. On the day of the behavioral assay, GnRH was microinjected bilaterally into the MCG, and within an hour lordosis behavior was elevated (Figure 5.5). The control condition, physiological saline, had no effect. By contrast, when an antibody against GnRH was microinjected into the MCG, lordosis was abolished within 90 minutes (see Figure 5.5), and normal serum globulin had no effect. We speculated that the behavioral effects of GnRH were due to its neuromodulatory actions, which we recorded years later (Ogawa, Kow, and Pfaff 1992). Three years later we replicated these results (Sakuma and Pfaff 1983) and also showed that pretreatment with estrogen was necessary for behavioral facilitation by GnRH in the MCG.

  Figure 5.5. Microinjection of gonadotropin-releasing hormone (GnRH) to the portion of the midbrain central grey that receives inputs from the ventromedial hypothalamus potentiated lordosis behavior (a); whereas an anti-GnRH antibody injected there decreased lordosis (b). Control placements in the superior colliculus were not effective. (Adapted from Sakuma and Pfaff 1980.)

  Thus, GnRH signaling to the MCG is sufficient to amplify the estrogen effect on reproductive behavior and is also necessary. The GnRH effect helps to synchronize brain and behavior with the endocrine preparations for ovulation. I think that our success in this series of experiments led, as I mentioned, to an entire field of work on neuropeptides and behavior.

  Chemistry

  As an example of this, I delve briefly into the chemistry of GnRH. One good effort was to see how large a fragment of the C-terminus would be necessary for a behavioral effect. An excellent team, Carol Dudley and Bob Moss in Dallas, ovariectomized female rats and implanted cannulae directed bilaterally toward the VMH. Of course the animals had to be pretreated with a low estrogen dose. The nonacetylated LHRH5–10 fragment and the GnRH5–9OH fragment were effective in enhancing lordosis. Dudley and Moss’s data suggested that, overall, amino acids in positions 6 through 9 may be important for this effect.

  Andrea Gore, now the editor of Endocrinology, became interested in the colocalization of GnRH neurons with the NMDA-R1 subunit. Her studies suggested that an increase in glutamatergic input to GnRH neurons plays a role in the increase in GnRH release and gene expression that occurs at the initiation of puberty. Enzymologic studies by Johnny Wu, who was working with Jimmy Roberts, showed that the decapeptide GnRH is processed by a zinc metalloendopeptidase EC 3.4.24.15 (EP24.15) that cleaves GnRH at the Tyr(5)-Gly(6) bond. Then they examined the role of the metabolite GnRH(1–5) in mediating peptide-facilitated reproductive behavior. Intracerebroventricular administration of GnRH(1–5) facilitated sexual behavior responses, similar to those facilitated by the decapeptide GnRH, in ovariectomized estradiol-primed female rats. Surprisingly, immunoneutralization of EP24.15 resulted in the inhibition of the GnRH-facilitated lordosis but had no inhibitory effects on lordosis facilitated by GnRH(1–5). These and other studies suggested to Wu and Roberts a nontrivial relation between GnRH fragments and glutamatergic signaling important for behavior.

  Although Antonio Donoso’s results supported the idea that glutamate, through NMDA receptors, participates in the regulation of lordosis behavior, glutamate seems to exert its actions in the behavioral and endocrine patterns through different mechanisms; the first seems not to be mediated by GnRH. So it is clear from all this work that GnRH peptide fragments can foster lordosis behavior, but the relation to NMDA signaling remains controversial.

  I was more interested in a different property of the decapeptide. The late protein chemist Tom Kaiser, who had just come to Rockefeller from the department of chemistry at the University of Chicago, had gotten a lot of attention for demonstrating the “amphiphilic” structures in the alpha-helices of other physiologically important peptide hormones. Amphilic means that one face of a peptide hormone would be lipophilic (hydrophobic), to facilitate binding to a lipid membrane, while the other face would be hydrophilic. That is, alternating side chains would project below and above the peptide backbone. A visiting electrophysiologist from Taiwan, Jenn-Tser Pan, worked with Lee-Ming Kow and me (Pan et al. 1986) to collaborate with Tom Kaiser using models, analogs of GnRH designed to test the importance of the amphiphilic structure. We used micropipettes to record electrical activity from hypothalamic and preoptic neurons after the application of vehicle, GnRH itself, or altered peptides that possess amphiphilic characteristics. The answer was clear. About 70 percent of the neurons responsive to GnRH were also responsive to an analog named GnRH3. Further, if GnRH excited the neuron recorded, so did the analog; if GnRH inhibited, so did the analog. We concluded that the amphiphilic property of the natural GnRH decapeptide accounts for an important part of its mechanism of action as a neuromodulator.

  Intellectual Background

  This entire field of work, both my discoveries of the GnRH neuronal migration and GnRH effects on behavior, must be set against the background of Sir Geoffrey Harris’s accomplishments. Many felt Harris would have won the Nobel Prize if he had not suffered an early death. As summarized by his most accomplished protégé, George Fink (2015), the main force of Harris’s work was to demonstrate control over the anterior pituitary by the hypothalamus. The presence of GnRH in the portal blood running from the hypothalamus to the pituitary was shown both by bioassays and by radioimmunoassays. Then they inferred that the massive ovulatory surge in luteinizing hormone was caused not only by a surge in GnRH levels but also by an increase in pituitary sensitivity to GnRH. Most exciting, Roger Guillemin won the Nobel Prize (together with his bitter competitor Andrew Schally) for delineating the chemistry of GnRH (see Figure 5.3). Of course, what I did was to use the relatively simple chemistry of GnRH and to incorporate it into my brain / behavioral system.

  As summarized by the authoritative Allan Herbison in Plant and Zeleznik (see Chapter 7), many GnRH neurons have a bipolar appearance: a cell body, usually long and narrow, with two simple dendrites, often one dorsal and the other ventral. Descending axonal projections tend to be periventricular. Neuroendocrinologists are most interested in those going to the median eminence, but I have paid special attention, regarding the regulation of lordosis behavior, to projections to the VMH. Then, of course, Brenda Shivers, Richard Harlan, and I followed immunocytochemically identified axons in a periventricular projection to the MCG, a projection exploited by Yasuo Sakuma, as described earlier (see Figure 5.5). Interestingly, GnRH immunoreactive terminals comprise inputs to GnRH neurons. Their effects on GnRH neurons’ electrical excitability are still under debate. Glutamatergic afferents clearly excite GnRH neurons, and noradrenergic inputs clearly suppress.

  GnRH neurons turned out to have many properties that we did not expect, quite beyond the migration from the nose into the brain. Brenda Shivers, Richard Harlan, and I firmly predicted that GnRH neurons would bind radioactive estradiol (Shivers et al. 1983a,b). After all, the neuroanatomical distributions of GnRH neurons and neurons expressing estrogen receptor overlap, and there are physiological feedback effects on GnRH that must be accounted for. We worked long and hard to develop immunocytochemical methods for localizing GnRH neurons that would be optimized and made compatible with autoradiographic methods for detecting estrogen-concentrating neurons. To our surprise, double-labeled cells were very rarely seen. Only about 0.2 percent of the GnRH neurons studied had radioactive estrogen bound in the cell nucleus; a far smaller percentage of the estrogen-binding neurons in the preoptic area produced GnRH. We are now convinced that GnRH neurons do not express the gene for estrogen receptor-α. But how, then, do estrogens increase the production of GnRH mRNA (Rothfeld et al. 1989; see Chapter 3)? Brenda Shivers was forced to the idea that estrogenic effects on GnRH production and release must be mediated by another cell type. That turned out to be right. Gaby Flugge and Wilhelm Wuttke, working in Göttingen, published convincing evidence that GABA (γ-aminobutyric acid) neurons close to GnRH neurons in the preoptic area bind estrogens and account for the hormonal effect. Further, researchers in Homburg, Germany, h
ave developed a mouse strain in which GnRHR neurons express a fluorescent marker, and have used these mice to show that the electrical excitability of these neurons alternates in synchrony with the estrous cycle. Further, they have demonstrated that GnRH stimulation elevates bursts of firing that in turn are reversed by a GnRH receptor antagonist. Their work was most interesting to me because if GnRH acts on GnRH neurons to rev up bursts of firing one could see how the positive feedback–amplified signal would become strong enough to promote lordosis behavior.

  Question Answered

  How are the behavioral preparations for reproduction harmonized with all the other changes in a female’s body that are necessary for successful fertilization by the male? The answer lies in one simple compound, a peptide with only 10 amino acids. GnRH does the job. Consonant with the idea of “unity of the body,” GnRH not only causes the ovulatory release of luteinizing hormone from the anterior pituitary gland but also facilitates female reproductive behavior. The combination of behavioral, electrophysiological, and peptide-chemical studies revealed a set of mechanisms that holds true from fish to laboratory animals to humans and likely has import for a wide range of mammals: all these adult behavioral phenomena follow a startling migration of GnRH neurons during development from the olfactory epithelium into the brain.

  Further Reading

  Dellovade, T. L., M. Schwanzel-Fukuda, J. Gordan, and D. W. Pfaff. 1998. “Aspects of GnRH Neurobiology Conserved across Vertebrate Forms.” General and Comparative Endocrinology 112: 276–282.