Quantitative autoradiographic localization of prostaglandin E2 binding sites in monkey diencephalon

Quantitative autoradiography was performed to investigate the mapping of prostaglandin E2 binding sites in the Macaca fuscata fuscata diencephalon. Autoradiographs were prepared by incubation of 10-micron- thick serial frozen sections with 3H-prostaglandin E2 and were processed by using a rotating drum-scanner and a computer-assisted image-processing system with 3H-microscales as standards. The localization of prostaglandin E2 binding sites was remarkably discrete in the diencephalon. The highest concentrations were found in the median and medial preoptic areas, supramammilary nucleus of the hypothalamus, and centromedian nucleus of the thalamus. High density was observed in the medial and dorsal hypothalamic areas; paraventricular, anterior, dorsomedial, and infundibular nuclei of the hypothalamus; and in the anteroventral, periventricular, paraventricular, laterodorsal, and habenular nuclei of the thalamus. The distribution correlates well with the known effects of prostaglandin E2 and may also give us useful clues in unveiling the novel role of prostaglandin E2 in a variety of brain functions.


Quantitative
autoradiography was performed to investigate the mapping of prostaglandin E, binding sites in the Macaca fuscata fuscafa diencephalon.
Autoradiographs were prepared by incubation of lo-pm-thick serial frozen sections with 3H-prostaglandin E, and were processed by using a rotating drum-scanner and a computer-assisted image-processing system with 3H-microscales as standards.
The localization of prostaglandin E, binding sites was remarkably discrete in the diencephalon. The highest concentrations were found in the median and medial preoptic areas, supramammillary nucleus of the hypothalamus, and centromedian nucleus of the thalamus.
High density was observed in the medial and dorsal hypothalamic areas; paraventricular, anterior, dorsomedial, and infundibular nuclei of the hypothalamus; and in the anteroventral, periventricular, paraventricular, laterodorsal, and habenular nuclei of the thalamus. The distribution correlates well with the known effects of prostaglandin E, and may also give us useful clues in unveiling the novel role of prostaglandin E, in a variety of brain functions.
Prostaglandin (PG) E series has various neurophysiological functions, such as in the central mediation of fever (Milton and Wendlandt, 1970;Potts and East, 1972;Milton, 1976) as one of the major regulators of food intake (Scaramuzzi et al., 1970) and luteinizing hormone release (Ojeda et al., 1977;Kinoshita et al., 1982) and as an endogenous anticonvulsant (Poddubiuk, 1976;Rosenkrantz, 1978). The high-affinity and specific binding protein of 3H-PGE, was found in membrane preparations from bovine pineal gland (Cardinali et al., 1979) and from rat (Malet et al., 1982) and human (Watanabe et al., 1985a) brains. From a regional study of dissected brain areas, PGE, binding was found to be high in the hypothalamus, pituitary, and amygdala of rat brain (Malet et al., 1982) and in the amygdala, hypothalamus, and hippocampus of postmortem human brain (Watanabe et al., 1985a). The PGE, receptor protein was solubilized in our laboratory in an active form from porcine brain using 3-[(3-cholamidopropyl) dimethylammoniol-l-propane sulfonate (Yumoto et al., 1986a). Furthermore, PGE, binding to the receptor is sensitive to guanine nucleotide (GTP), and, after solubilization, the receptor and GTP-regulatory protein were separated by gel filtration using high-performance liquid chromatography (HPLC) and were reconstituted (Yumoto et al., 1986b). These results provide evidence that PGE, in the brain exerts various neurophysiological functions via specific receptor binding, and that GTP is involved in its regulation. Recently, we demonstrated the localization of PGD, binding by using in vitro autoradiography coupled with a computerized image-analysis system Yamashita et al., 1983). The specific PGD, binding was predominantly localized in the gray matter in the rat brain  and, more precisely, in specific neuronal cells, such as the Purkinje cell layer in the swine cerebellum . Preliminarily, we obtained the distinct localizations of PGD,, PGE,, and PGF,, bindings in the monkey brain (Watanabe et al., 1985b(Watanabe et al., , 1986. In this study, we developed a digital subtraction method for the quantitation of autoradiographs and demonstrated by detailed mapping the concentration of PGE, binding sites in the monkey diencephalon.

Materials and Methods
Tissue preparation. Macaca fuscata fuscata (female, 6.5-8.0 kg) was anesthetized with Ketalar (ketamine-HCl) and perfused via the left ventricle with cold 10 mM sodium phosphate-buffered saline (pH 7.4) containing 5% sucrose. The brain was rapidly removed, dissected into 3-4 mm-thick blocks (ca. 2.0 x 2.5 cm), and frozen on dry ice. Frozen serial sections of 10 brn thickness were cut in a cryostat at -14°C and mounted on gelatin-coated glass slides.
'H-PGE, binding. All procedures of 'H-PGE, binding assay were performed at 4°C as described by Yamashita et al. (1983) and Watanabe et al. (1983) with a slight modification. After preincubation in four 150 ml changes of 50 mM Tris-HCl (pH 7.4) containing 0.1 M NaCl (buffer A) for a total of 1 hr, the tissue sections were incubated with 150-200 ~1 (the volume varied depending on the size of section) of 20 nM 3H-PGE, in buffer A for 30 min with gentle shaking (60 times/min, 4 cm reciprocal). The sections were then rinsed in 4 sequential short (15 set each) dips in 50 ml of buffer A, quickly splashed with distilled water, and air-dried. Nonspecific binding was obtained with the consecutive  , 1986a). The Kd value of the 'H-PGE2 binding to that fraction was a few nanomolar at 37°C. Since the PGE, binding sites showed a temperature-dependent rise in affinity (K,, for PGEz binding to the P2 fraction prepared from monkey diencephalon and amygdala, 1.1 nM at 37°C and Relative optical density The association ofthe specific 'H-PGE? binding to the section reached a plateau within 30 min at 4°C. For the dissociation of bound 'H-PGE, from incubated sections, we have no extensive time course, but 1 or 2 further 15 set dips of the section into the buffer following the routine 4 dips resulted in little change in the total and nonspecific bindings. Autoradiography. After being dried in a desiccator, the slides were tightly juxtaposed with tritium-sensitive films (Ultrofilm, LKB) and stored at 4°C for 4 weeks. 'H-Plastic standards (3H-microscales; Amersham) were included in each cassette. After exposure, the films were developed in Kodak D19 at 20°C for 5 min and then fixed for 5 min. The nuclei were identified by toluidine blue staining on the adjacent sections, with reference to the stereotaxic atlases of Emmers and Akert (1963) and Kusama and Mabuchi (1970).
Computer-assisted image-processing system. Films were analyzed using a computerized image-analysis system, including a rotating-drum scanning densitometer (Model 2605; Kimoto, Tokyo) and a minicomputer system (ECLIPSE S/120; Data General, Westboro, MA). The rotating-drum scanning densitometer was used to convert the photometric data from the autoradiographs into the digital values of 0 to 255 (relative O.D. value). The digitized data were arranged in parallel with the optical density between 0 and 1.5. Sampling pitches were 50 pm. The array of optical density readings was processed through a minicomputer. These data were converted to pseudocolor images on an image display (1024 x 1024 elements).
To produce quantitative images of the autoradiographs of )H-PGE2 binding, we transformed the original O.D. value ofthe pixels (pictures + elements) to the binding value (BD value) that was linearly related to the quantity of ZH-PGE, bound in each pixel. This transformation was processed via the following steps: First, the mean O.D. value of each layer of 'H-microscales measured by the rotating-drum scanner was plotted against each tritium content. These data were expressed as a power function, Y = ax', where a and b were different depending on the exposure periods, the batches, and the developing conditions of the films. Then, according to this standard curve, the original O.D. value of each pixel in the image was converted to the BD value (fmol 3H-PGE,/mg tissue). This linearization procedure was performed for each sheet of autoradiographic film with a set of 'H-plastic standards.
After the conversion of the O.D. value to the BD value, further processing via subtraction between the 2 images (total and nonspecific) was performed. To superimpose the image of nonspecific binding precisely on that of total binding, we selected several characteristic points in both images. Using their locations, we transferred the image of the nonspecific binding to the position of the image of the total binding by parallel and rotational transforms (alline transforms by the least-squares method). In affine transforms, the BD value of each pixel of the image was obtained from that of the corresponding pixels of the original image by linear interpolation. After all positions of the images were put into exactly the same position, subtraction among the BD values was performed to give the image of the specific binding. To determine the average concentration of ?H-PGE2 bound in various nuclei, a cursor box was positioned over each nucleus in the image ofthe specific binding on the color display. The average BD value within the cursor box was calculated by the computer.

Results
A typical standard curve for transformation of the O.D. value to the BD value in shown in Figure 1. Both high and low microscales consisted of 8 layers. Eight points of high microscale were described as a power function, while 8 points of low microscale were fitted for a linear function. The boundary points between low and high microscales were fitted for both functions. Therefore, we employed a straight line for the standard curve that was between 0 and 1.3 nCi/mg tissue of the BD value and used a power function for the curve above 1.3 nCi/mg. In all cases in this study, the goodness of fit of the standard curve with high microscale was ascertained; the relative coefficients (r2) were greater than 0.998. According to the above-mentioned standard curve, the 'H-PGEz content bound in each pixel was calculated for the total and nonspecific binding. An example of this quantitation procedure is shown in Figures 2 and 3. Figure 2, A, B, represents the original autoradiograms of total and nonspecific binding of IH-PGE, in coronal sections of monkey brain containing the preoptic area. Figure 3, A, B, represents the corresponding pseudocolor-coded images. While the nonspecific binding (Fig. 3B) was low and rather homogeneously distributed, the total binding ( Fig. 3A) was remarkably concentrated in the medial preoptic area. The image of the specific binding (Fig. 3C) was then processed by subtracting the binding values of the image of nonspecific binding from those of total binding. The image of the specific binding exhibited a pattern very similar to that of the total binding. The proportion of specific binding to the total binding was calculated to be more than 80% in the medial preoptic area. A similar presentation of the percentage of specific binding in the specific binding image is also possible; a typical result is shown in Figure 4. The BD value of each pixel in the image of the specific binding was divided by that of the corresponding pixel in the image of the total binding to produce the image of the percentage of specific )H-PGE, binding (Fig. 4A).  Figure 4B shows the percentage of the specific binding value of each pixel along the horizontal straight line in Figure 4A. The highest specific binding in the centromedian nucleus reached 80% of the total binding.
The localization of 3H-PGE, binding was characteristic and discrete throughout the diencephalon (Fig. 5). The average 3H-PGE, concentrations in various nuclei were calculated and are summarized in Table 1. For example, the anteroventral nucleus of the thalamus is seen in Figure 5, C-G, and the centromedian nucleus of the thalamus in Figure 5, H-J. Their respective lengths were at least 1.5 and 1.9 mm in the sagittal direction, as judged from the section thickness (10 pm) and the number of sections. Since the mean value of 3H-PGE, binding in each nucleus varied anteriorly to posteriorly, we selected the highest values for presentation in Table 1.
In the limbic forebrain, a moderately high density (16 fmol/ mg tissue) of PGE, binding sites was observed in the anterior commissural nucleus (Fig. 5, A, B), and a moderate concentration (1 l-l 3 fmol/mg tissue) was observed in the lateral septum and accumbens nucleus (Fig. 5A).
In the thalamus, PGE, binding sites were the most discretely located. In addition to a very high density of binding sites in the centromedian nucleus, the nuclei surrounding the third ventricle, such as the anteroventral (Fig. 5 In the midbrain, the annulary nucleus (Fig. 5L) and oculomotor nucleus (Fig. 5P) displayed high densities (16-17 fmol/ mg tissue) of the binding sites.
We employed 2 monkey brains in the quantitative study and 4 other brains in the qualitative study. We observed little heterogeneity among the brain specimens in the pattern of localization of PGE, binding sites throughout the diencephalon.

Discussion
A number of reports have depicted the central action of PGE, mainly on the basis of pharmacological studies. However, there had been no reports concerning the precise localization of PGE, receptors in the CNS nor in the peripheral tissue before our recent studies. High-affinity PGE, binding was demonstrated by the glass filter assay using membrane preparations of bovine pineal gland (Cardinali et al., 1979) and that of rat (Malet et   al., 1982), porcine (Yumoto et al., 1986a), and human brain the monkey brain revealed the distinct localization of PGD,, (Watanabe et al., 1985a). In 198 1, we started to investigate the PGE,, and PGF,, binding sites in the diencephalon and limbic binding sites of 3H-PGD, in the rat and porcine brain by using system (Watanabe et al., 1985b(Watanabe et al., , 1986. an in vitro autoradiography technique  This paper presents the first detailed report of the quantitative Yamashita et al., 1983). Thereafter, preliminary studies using localization of PGE, binding sites in the brain. Various sites of Figure 5. Representative images of specific 3H-PGE, binding distribution in monkey diencephalon. The tissue sections (10 Km thick) were incubated with 20 nM )H-PGE2 in 50 mM Tris-HCl (pH 7.4) containing 0.1 M NaCl, as described in Materials and Methods. Nonspecific binding was obtained with the consecutive section by adding 100 PM unlabeled PGE, to the incubation mixture. Images of specific 3H-PGE2 binding were prepared by subtracting the values of the binding in the corresponding pixels of the superimposed image of nonspecific binding from those of the image of total binding. A-L, Coronal sections from one monkey, and M-P, those from another monkey, to cover almost all the diencephalic regions. A, LS, lateral septum; Act, accumbens nucleus; PM, median preoptic area. B, NST, nucleus stria terminalis; NCA, anterior commissural nucleus; APL, lateral preoptic area; APM, medial preoptic area. C, AV, anteroventral nucleus; AH, anterior hypothalamic nucleus; AMH, medial hypothalamic area; u.c., anterior commissure. Ann, annulary nu. 16 N,,,, oculomotor nu.

17
Frozen IO-rm-thick sections were labeled with 20 WI 'H-PGE,, as described in Materials and Methods, and apposed against LKB Ultrofilm for 4 weeks to generate autoradiographs.
Brain sections were selected at 300-500 pm intervals for the demonstration of typical brain structures. The concentration of PGE, binding was calculated by grid sampling of the specific 3H-PGE, binding images, and the maximum values were taken from the concentrations in several sections of the same structure.
the central action of PGE, were unveiled on the basis of the results of the nucleus-level localization. Some are closely related to the known functions of PGE,, as defined by pharmacological studies discussed below, and others may be useful clues in elucidating the novel physiological roles of PGE, in a variety of brain functions.
In the present study, the density of binding sites was calculated quantitatively in terms of fmol/mg of tissue. The standard curves for the calibration of the tissue tritium concentration were obtained from Amersham's SH-microscales. Since the standard curve in each film was strictly interpolated between 16 points of )H-microscales, the reproducibility and reliability of the data were ascertained; further, comparisons between the independent experiments could be made in a quantitative manner. Rainbow et al. (1984) pointed out a serious problem of the autoradiography technique, that is, the difference in autoabsorption between gray and white matter due to the greater density of lipids in the white matter, especially in the case of tritium-labeled ligands. Kuhar and Unnerstall (1985) and Geary et al. (1985) also reported such a difference, and Geary et al. (1985) demonstrated that the O.D. value of the gray matter paste containing a certain tritium concentration was ca. 2-fold higher than that of the white matter paste containing the equivalent amount of radioactivity. In the present case, i.e., the study of the total and nonspecific binding of 3H-PGE,, very little radioactivity was distributed in the white matter. Even if the radioactivity in the white matter were double the value obtained from the autoradiographic study, the pattern of localization of 3H-PGE, would not be significantly altered. Especially, in this study, we focused on the localization of PGE, binding sites in the gray matter regions (Table 1).
Since Milton and Wendlandt (1970) studied the febrile response following central administration of PGE,, and Vane (197 1) observed the inhibition of prostaglandin biosynthesis by antipyretic drugs, a number of studies have been done; PGE, was finally proposed to be an intermediary substance in the preoptic/ anterior hypothalamic genesis of fever (Milton, 1976;Wolfe and Coceani, 1979). However, there were no reports as to whether PGE, actually binds to the thermoregulatory neurons in the preoptic area. Recently, by use of the push-pull cannula technique, Coceani et al. (1987) assessed the enhancement of actual release of PGE, at discrete preoptic/anterior hypothalamic sites by systemic injection of exogenous (endotoxin) or endogenous (interleukin-1) pyrogens. Our finding that the high density of PGE, binding sites exists in the median and medial preoptic areas offers strong evidence for the central mediator role of PGE, in the febrile response. Moreover, we ascertained that the localization of PGE, binding is identical to that of PGE, in the monkey brain (Y. Watanabe, unpublished observations). PGE, has been postulated as being involved in the modification of algesia (Poddubiuk, 1976;Horiguchi et al., 1986). Horiguchi et al. (1986) described the biphasic effect of PGE, in regulating pain responses after its intracistemal administration to mice, and suggested that the site of PGE, action might be located in the lower portion of the CNS, i.e., brain stem and spinal cord. On the other hand, a high concentration of opiate receptors was found not only in the lower part of the CNS, but also in the centromedian and parafascicular nuclei of the thalamus in the monkey brain (Wamsley et al., 1982). The thalamus is a relay point for the fibers carrying protopathic pain. The present results show that PGE, binding is highly concentrated in the centromedian nucleus of the thalamus, although we could not distinguish the parafascicular nucleus from the centromedian nucleus. In addition, we observed the localization of PGE, binding in the central gray (midbrain periaqueductal gray matter) of monkey brain. These observations suggest a role of PGE, in the control of pain at multiple sites in the brain.
The neuroendocrine role of PGs is most evident in the hypothalamohypophysial pathway (Hedge, 1977;Behrman, 1979). Intraventricular injection of PGE, and PGD, caused stimulation