Proc. Natl. Acad. Sci. USA Vol. 76, No. 3, pp, 1485-1488, March 1979 Neurobiology Immunocytochemical localization of cyclic GMP: Light and electron microscope evidence for involvement of neuroglia (cerebellum/cyclic AMP/y-aminobutyric acid/harmaline/) V. CHAN-PALAY* AND S. L. PALAYt Departments of *Neurobiology and tAnatomy, Harvard Medical School, Boston, Massachusetts 02115 Contributed by Sanford L. Palay, December 7, 1978

ABSTRACT Guanosine 3',5'-cyclic monophosphate (cGMP) presents cytopharmacological evidence from immunocyto- immunoreactivity in the rat's cerebellum was studied with light chemistry for the cellular and subcellular location of cGMP in and electron microscopy by the indirect fluorescence method some cerebellar neurons and particularly in neuroglial cells. and the peroxidase-antiperoxidase method. Labeled cells in- cluded neuroglial cells in the cerebellar cortex, white matter, and deep nuclei; some stellate and basket cells in the cortex; and MATERIALS AND METHODS some large neurons in the deep nuclei. No evidence was found Adult 200-300 g Sprague-Dawley rats, the indirect immu- for sagittal microzonation in the cGMP distribution. In the la- nofluorescence method (11), and the peroxidase-antiperoxidase beled cells, cGMP immunoreactive sites were localized to sur- face membranes, organelles, and the cytoplasmic matrix. (PAP) method (12) for light and electron microscopy were used. Specificity was indicated by the same pattern of labeling after Rabbit antisera were raised against the 2'-O-succinyl derivative treatment with cGMP immunoglobulin that had been adsorbed of cGMP or cAMP conjugated to limpet hemocyanin (13). More with adenosine 3',5'-cyclic monophosphate (cAMP) and by the than 50% of 2'-O-succinyl-cGMP ([125I]iodotyrosine methyl failure to label after treatment with normal rabbit sera or with ester derivative) or 2'-O-succinyl-cAMP ([1251]iodotyrosine cGMP immunoglobulin that had been adsorbed with 1 mM methyl ester derivative) (<0.01 pmol) was bound at a serum cGMP. Cerebella treated with cAMP antisera, however, showed immunoreactivity in Purkinje cells, granule cells, and Golgi cells dilution of 1:2000. Specificity was determined by radioim- in addition to neuroglia in cortex and deep nuclei. Sequential munoassay. The antisera raised against cGMP gave 20-30% norepinephrine and glutamate superfusions generally intensi- displacement of bound '25I-labeled cGMP with 5 fmol of fied cGMP immunoreactivity, not only in neuroglial cells but acetylated cGMP. Acetylated cAMP did not produce significant also in the background. Under these conditions some Purkinje displacement up to 1000 fmol. Corresponding experiments with cells and some granule cells were also labeled. Increased cGMP antisera raised against cAMP produced the equivalent appro- immunoreactivity was also obtained by treatment with har- maline, y-aminobutyric acidand aminooxyacetic acid, musci- priate result, indicating its specificity. The immunoglobulin mol, y-aminobutyric acid, or apomorphine in order of de- (IgG) fractions of these antisera were used. The basic cyto- creasing effectiveness. Serotonin and colchicine produced no chemical methods have been described (14). Cryostat sections detectable increase of cGMP immunoreactivity above normal, (10,um thick) of cerebella from formaldehyde-perfused brains and diazepam and sodium pentobarbital decreased it. In these and unfixed cerebella frozen in liquid nitrogen 1-2 min after experiments, was preferable to sodium pento- decapitation were treated with the IgG and stained by the barbital for anesthesia on account of the depressive action of the latter on cGMP immunoreactivity. Thus, drugs that increase immunofluorescence method, using a fluorescein isothiocya- cerebellar activity enhance cGMP levels, whereas those that nate-conjugated goat IgG. Vibratome sections (20-30 Am thick, decrease cerebellar activity decrease cGMP levels. However, cut in cold 0.05 M Tris-HCI buffer, pH 7.6) of cerebella from it is not clear whether these fluctuations in cGMP levels are a perfused brains were treated sequentially with cGMP IgG (18 direct consequence of neurotransmitter function or are sequelae hr at 40C in a humid environment, dilution 1:100 in 0.5% Triton to other related events. The present study suggests that some X-100/0.05 M Tris-HCI buffer, pH 7.6), goat-anti-rabbit IgG, neurons and many neuroglial cells are the major sites of cGMP PAP antiserum, and diaminobenzidine with hydrogen peroxide, in the cerebellum. prior to postfixation in 2% osmium tetroxide, dehydration with The mammalian cerebellum contains unusually high concen- methanol, and embedding in epoxy resin. These sections were trations of endogenous guanosine 3',5'-cyclic monophosphate then examined with the light microscope and subsequently (cGMP) (1). Experiments on brain slices, cultures, and in vitro thin-sectioned in serial order for electron microscopy. No preparations suggest that cGMP may be associated with a va- counterstains were used for electron microscopy. Two forms riety of possible neurotransmitter substances in mechanisms that of anesthesia were employed and their results were compared: are not understood. Selective changes of cerebellar cGMP in- 2% sodium pentobarbital (0.1 ml/100 g of body weight) injected dependent of adenosine 3',5'-cyclic monophosphate (cAMP) intraperitoneally and diethyl ether by inhalation. Two fixatives can be produced by a number of agents and conditions such as were used in the perfusions, either 4% (wt/vol) formaldehyde (2), excitatory and inhibitory transmitter substances (3, in Na phosphate buffer, pH 7.4, or 4% formaldehyde and 0.25% 4), decapitation (5), and (6). Considerable glutaraldehyde in phosphate buffer, pH 7.4, both at 4°C; each speculation marks the interpretation of the correlations among was preceded by a wash with cold Ca2+_free Tyrode's buf- cGMP levels, putative transmitter substances, and their possible fer. localization in cerebellar neurons, particularly Purkinje cells Nine separate sets of experiments were performed, on groups (refs. 7 and 8; reviewed in ref. 9, p. 229). The presence of cGMP of three rats each, using drugs to alter cGMP levels detectable in Purkinje cells has not been demonstrated (10). This study by immunocytochemistry. All animals were perfused with fixative at the stated intervals after drug administration: (i) The publication costs of this article were defrayed in part by page charge payment. This article must therefore by hereby marked "ad- Abbreviations: cGMP, guanosine 3',5'-cyclic monophosphate; cAMP, vertisement" in accordance with 18 U. S. C. §1734 solely to indicate adenosine 3',5'-cyclic monophosphate, PAP, peroxidase-antiperoxidase; this fact. GABA, y-aminobutyric acid. 1485 1486 Neurobiology: Chan-Palay and Palay Proc. Natl. Acad. Sci. USA 76 (1979) Diazepam, 25 mg/kg of body weight, intraperitoneal, 30 min in all experiments using a scale of 1+ to 4+ for increasing prior to perfusion; (ii) , 10 mg/kg of body weight, amounts and intensities of reactions. Electron microscopy was intravenous, 30 min prior to perfusion; (iii) apomorphine, 5 conducted on serial sections of six specimens obtained from mg/kg of body weight, intraperitoneal, 5 min prior to perfu- normal, untreated, anesthetized animals. The sections were sion; (iv) harmaline, 40 mg/kg of body weight, intraperitoneal, treated with cGMP IgG, and cGMP IgG adsorbed with cGMP 30 min prior to perfusion; (v) y-aminobutyric acid (GABA) 10 was used for controls. ,uM, 2-,l intracerebellar injections, 5-7 min prior to perfusion; (vi) GABA, 10 AM, 2-,A intracerebellar injections, 5-7 min prior RESULTS to perfusion with aminooxyacetic acid, 10 mg/250 g of body cGMP immunoreactivity was greater: (i) in the cerebella from weight, intraperitoneal, 30 min prior; (vii) norepinephrine, 300 animals anesthetized with ether compared to sodium pento- JAM, and sodium L-glutamate, 3 mM in sterile saline at 37°C, barbital; (ii) in the more sensitive PAP method compared to superfused in sequence for 10 min each by using a push-pull immunofluorescence; (iii) in tissues fixed in formaldehyde cannula and agar superfusion chamber over the cerebellar without glutaraldehyde compared to unfixed frozen material. cortical surface; (viii) serotonin, 0.1 ,uM (250,l), intraventric- The presence of glutaraldehyde in the primary fixative en- ular infusion over 3 hr with monoamine oxidase inhibition hanced morphological preservation for electron microscopy (clorgyline, 10 mg/100 g of body weight); (ix) colchicine (3 but reduced immunoreactivity. Colchicine treatment did not ,ug/,ul), intracerebellar injection, 2 ,ul per electrode track, and enhance cGMP immunoreactivity. In the molecular layer, some 25 ,ul intraventricularly 24 hr prior to perfusion. stellate and basket somata (approximately 30%) were cGMP Parasagittal and transverse sections from three cerebellar immunoreactive. The Golgi epithelial neuroglial somata sur- regions (vermis, hemisphere, and paravermis including deep rounding Purkinje cells and their radial fibers (15) displayed nuclei) in each animal were processed. Tissues from the separate the most intense immunoreactivity. Purkinje cell somata re- experiments and controls were handled at the same time in mained unreactive with all modes of fixation and drug ma- order to reduce technical differences. Controls for specificity nipulations attempted except for the superfusion with norepi- included: cAMP antisera; cyclic GMP IgG adsorbed overnight nephrine and glutamate. Neuroglial cells in the granular layer with 1 mM cGMP; cGMP IgG adsorbed overnight with 1 mM and their processes (15, 16) were immunoreactive, whereas cAMP; and normal rabbit serum. The resulting material was Golgi neurons and granule cell somata were not. In the white studied with the light microscope; neurons and neuroglia with matter, some oligodendroglial cells and some axons were im- cGMP immunoreactivity were identified in the cerebellar munoreactive. In the cerebellar nuclei, neuroglial cells were cortex, white matter, and deep nuclei. The distribution and generally reactive, whereas the neurons were not and the total intensity of immunoreactivity were determined for specimens amount of cGMP immunoreactivity was considerably lower

441

.. I , 't -is - T '.. .4 -, '.IYIZ,'?-.lA .1to! 1 a V I '4 -'. .' --:.le' 14 '. ... t; .._.,'.. -.itS f- .-, '%. - .. W#4. '00 4 -U FIG. 1. Light micrographs of immunoreactive neuroglial and neuronal elements of the cerebellar cortex visualized by cGMP IgG (a-d, f-i) and cAMP antisera (j) by using the PAP method. (a and b) Tissue from normal, untreated, ether-anesthetized animal showing cGMP immu- noreactivity in glial cells surrounding Purkinje cells (PC) and in the granular layer (arrows). (a, X120; b, X480.) (c and d) Ether-anesthetized animal treated with apomorphine. (c, X120; d, X480.) (e) Control tissue reacted with cGMP IgG preadsorbed with 1mM cGMP. (X480.) (f and g) Normal untreated animal anesthetized with sodium pentobarbital. (f, X120; g, X480.) (h and i) cGMP immunoreactivity in tissue from animal anesthetized with ether and treated with harmaline. (h, X120; i, X480.) (j) Normal, untreated, ether-anesthetized animal treated with cAMP antisera. (X120.) Neurobiology: Chan-Palay and Palay Proc. Natl. Acad. Sci. USA 76 (1979) 1487

@.Figs:ts...... at <'$.X ¢.s-..¢' ...... 94'...... am rI' At wi!.X' , -' .,''t" r .. s go. A<... w i,> '" )' tkg . ': ..COW,''' ' b; .!pi,.,8.x.:,;,;.X*,.8W...r...... e:.,.,.,.z.^...... ''.';.'..''.9',g:.Pt,; %bto.t.tfirs

st s t ;9 f; '4 .*i$....s..v,*tv.*,::,,,tA>,,^..,,7Ad_.99\1...... An,.,.->+:...... -Nno' s-.', .t'..--,,,;:.-.,-,.a-.I!:s-HX.-Newt8

FIG. 2. Electron micrographs of cerebellar cortex from normal, untreated, ether-anesthetized animals, treated with cGMP IgG, and visualized after the PAP reaction; no counterstains. (a) cGMP immunoreactivity is found in a Golgi epithelial (Go ep) neuroglial cell and its processes lying adjacent to and surrounding an unreactive Purkinje cell (PC) and other neural elements in the neuropil. (X9000.) (b and c) Two examples of unreactive synapses of thorns (t) with axons (Ax) surrounded by reactive neuroglial processes. (b, X35,000; c, X46,000.) (d) Unreactive mossy fiber rosettes (MF) and a granule cell (gr c) with intercalated processes of a labeled glial cell (gl). (X11,000.) (e) A labeled glial process (gl) abutting the endothelial lining of blood capillary (bl cap) and surrounded by unreactive neural elements. (X34,000.) in the cerebellar nuclei than in the cortex. Intracerebellar in- and no indication of sagittal microzonation was observed in jections of GABA with or without aminooxyacetic acid induced cGMP distribution (17). cGMP immunoreactivity in some large neurons. Labeled cells Selective changes in the location and intensity of cGMP and their processes were randomly scattered in the cerebellum immunoreactivity were detectable at a cellular level on ad- 1488 Neurobiology: Chan-Palay and Palay Proc. Natl'. Acad. Sci. USA 76 (1979) ministration of various drugs. Superfusion with norepinephrine immunoreactivity and substances that decrease cerebellar ac- and glutamate caused the greatest increase (18). Under these tivity decrease cGMP immunoreactivity. Because the majority conditions superficial cortical folia directly perfused showed, of cGMP is in neuroglial cells, it would be expected that their in addition to the reactive neuronal and glia elements usually activity would fluctuate in response to activity of adjacent or found, reactive Purkinje cell somata (up to 80% of the popula- related neurons. Although the data indicate that these sub- tion) and dendrites, reactive granule cell somata (up to 50%), stances can alter levels of cGMP detectable by immunocyto- and Golgi neurons (30%). Because these reactions all occurred chemical methods, their mechanisms of action remain obscure. in an unusually high background, it is difficult to assess their Interpretation of the pharmacological data is impeded by the significance. In the cerebella of animals treated with harmaline, reduction in cerebellar activity imposed by the necessary an- GABA with aminooxyacetic acid, muscimol, GABA, or apo- esthesia of the experimental animals. Moreover, it is not clear, , cyclic GMP immunoreactivity in neuroglial cells for example, whether the increase in cGMP immunoreactivity generally was increased above that in normal animals, but it was after drug treatment is the direct result of drug action or se- lower in the neuroglial cells of the Purkinje cell layer. Stellate lective neuronal activity or is secondary to other effects, such and basket neurons were more reactive after GABA and har- as interference with calcium influx or GABA release and up- maline treatment than in untreated animals. Intraventricular take. The present study provides evidence for the presence of infusions of serotonin produced cGMP immunoreactivity equal membranous (particulate) and cytoplasmic (soluble) cGMP, in intensity and distribution to that of the normal untreated particularly in neuroglial cells, and indicates that cGMP is in animal, and diazepam and sodium pentobarbital treatment a position to interact with externally applied substances, in- reduced cGMP immunoreactivity below normal levels (see Fig. cluding putative neurotransmitters, as well as participate in the 1 a-d and f-i). No-specific staining was obtained when normal regulation intracellular processes. This focuses our attention on rabbit serum and cGMP IgG adsorbed with cGMP were used. the possible nonneuronal, nonsynaptic functions of cGMP and cGMP IgG adsorbed with cAMP showed a labeling pattern no forces us to recognize the importance of the neuroglia in neu- different from that obtained with unadsorbed cGMP IgG. rotransmitter-related neuronal events. Comparative studies with cAMP antibody repeatedly indicated We thank Dr. A. Guidotti for the gift of the characterized cGMP and immunoreactivity in Purkinje cells (Fig. 1j), granule cells (19), cAMP antisera and for his critical discussion of the manuscript, and stellate and basket cells, Golgi cells, numerous Golgi epithelial Mr. H. Cook for photographic assistance. This work was supported in neuroglial cells and their processes, and some large and small part by U.S. Public Health Service Grant NS 03659; Parkinson's Disease neurons in the deep cerebellar nuclei. Because these results Project, Massachusetts General Hospital, Boston; and an Alfred P. Sloan differ significantly from the sites of cGMP immunoreactivity, Foundation Fellowship (to V.C.-P.). the data indicate that the cytochemical activities of cGMP and cAMP antisera are distinguishable. 1. Ferrendelli, J. A., Steiner, A. L., McDougal, D. B. & Kipnis, D. mi- M. (1970) Biochem. Biophys. Res. Commun. 41,1061-1067. Electron microscopy confirmed the findings of light 2. Redos, J. D., Catravas, G. N. & Hunt, W. A. (1976) Science 193, croscopy that the majority of labeled cells and processes were 58-59. neuroglial: Golgi epithelial cells in between Purkinje cells, their 3. Ferrendelli, J. A., Chang, M. M. & Kinscherf, D. A. (1974) J. processes the Bergmann fibers in the molecular layer, and other Neurochem. 22, 535-540. neuroglial cells in the granular layer and around blood vessels 4. Mao, C. C., Guidotti, A. & Costa, E. (1974) Brain Res. 79, (Fig. 2 a-e). Rarely were labeled stellate or basket cells ob- 510-514. served. There were two components of immunoreactivity in 5. Kimura, H., Thomas, E. & Murad, F. (1974) Biochim. Biophys. labeled cells-membranous and cytoplasmic. The membranous Acta 343, 519-528. reactive material is detectable on the plasma membranes, on 6. Mao, C. C., Guidotti, A. & Costa, E. (1975) Naunyn-Schmiede- outer nuclear membranes but not in the nucleus on outer bergs Arch. Pharmacol. 289,369-378. itself, 7. Biggio, G., Costa, E. & Guidotti, A. (1977) J. Pharmacol. Exp. mitochondrial membranes but not within mitochondria or on Ther. 200,207-215. the cristae, on outer membranes of the granular and smooth 8. Biggio, G. & Guidotti, A. (1976) Brain Res. 107,365-373. endoplasmic reticulum but not within the lumen, and on mi- 9. Nathanson, J. A. (1977) Physiol. Rev. 57, 157-256. crotubules and neurofilaments. The cytoplasmic matrix appears 10. Cumming, R., Eccleston, D. & Steiner, A. (1977) J. Cyclic Nu- as a dark reactive flocculent material between the cellular or- cleotide Res. 3, 275-282. ganelles. Where the label occurs in glial cells, the unlabeled 11. Coons, A. G. (1958) in Advanced Cytochemical Methods, ed. neuronal elements are encircled by reactive processes, for ex- Danielli, J. F. (Academic, New York), pp. 399-422. ample, around Purkinje cells (Fig. 2a), around synapses between 12. Sternberger, L. (1974) in Immunocytochemistry, Foundations Purkinje cell thorns and axonal boutons (Fig. 2 b and c), around of Immunology Series, eds. Osler, A. & Weiss, L. (Prentice Hall, Englewood Cliffs, NJ), p. 171. mossy fiber rosettes and granule cells (Fig. 2d), and surrounding 13. Steiner, A. C., Parker, C. W. & Kipnis, D. M. (1972) J. Biol. Chem. a blood capillary (Fig. 2e). No specifically labeled glial or 247, 1106-1124. neuronal elements were found in control tissue treated with 14. Chan-Palay, V. & Palay, S. L. (1977) Proc. Natl. Acad. Sci. USA cGMP IgG adsorbed with cGMP. 74, 4050-4054. 15. Palay, S. L. & Chan-Palay, V. (1974) Cerebellar Cortex, Cytology DISCUSSION and Organization (Springer, Berlin). Immunocytochemical methods can be used to demonstrate the 16. Chan-Palay, V. & Palay, S. L. (1972) Z. Anat. Entwicklungsgesch. cellular and subcellular locations of cGMP immunoreactivity. 138, 1-19. These are to a large extent in neuroglial cells throughout the 17. Chan-Palay, V., Palay, S. L., Brown, J. T. & Van Itallie, C. (1977) but in those to and Exp. Brain Res. 30,561-576. cerebellum particular adjacent surrounding 18. Ferrendelli, J. A., Kinscherf, D. A. & Chang, M. M. (1975) Brain Purkinje cells. The intensity and distribution of cGMP immu- Res. 84, 63-73. noreactivity are affected by pharmacological manipulation and 19. Bloom, F. E., Hoffer, B. J., Battenberg, E. F., Siggins, G. R., the cellular locations and levels of such changes can be studied. Steiner, A. L., Parker, C. W. & Wedner, H. J. (1972) Science 177, Substances that increase cerebellar activity increase cGMP 436-438.