Neuroscience Vol. 36, No. 2, pp. 377-391, 1990 Printed in Great Britain
0306-4522/90 $3.00 + 0.00 Pergamon Press plc 1990 IBRO
U L T R A S T R U C T U R A L LOCALIZATION OF [125I]NEUROTENSIN B I N D I N G SITES TO CHOLINERGIC NEURONS OF THE RAT NUCLEUS BASALIS MAGNOCELLULARIS E. SZIGETHY,K. LEONARD and A. BEAUDET* Laboratory of Neuroanatomy, Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada H3A 2B4 Abstract--The distribution of specifically-labeled neurotensin binding sites was examined in relation to that of cholinergic neurons in the rat nucleus basalis magnocellularis at both light and electron microscopic levels. Lightly prefixed forebrain slices were either labeled with [t25l](Tyr3) neurotensin alone or processed for combined [~25I]neurotensin radioautography and acetylcholinesterase histochemistry. In light microscopic radioautographs from l-#m-thick sections taken from the surface of single-labeled slices, silver grains were found to be preferentially localized over perikarya and proximal processes of nucleus basalis cells. The label was distributed both throughout the cytoplasm and along the plasma membrane of magnocellular neurons all of which were found to be cholinesterase-positive in double-labeled material. Probability circle analysis of silver grain distribution in electron microscopic radioautographs confirmed that the major fraction (80--89%) of specifically-labeled binding sites associated with cholinesterasereactive cell bodies and dendrites was intraneuronal. These intraneuronal sites were mainly dispersed throughout the cytoplasm and are thus likely to represent receptors undergoing synthesis, transport and/or recycling. A proportion of the specific label was also localized over the nucleus, suggesting that neurotensin could modulate the expression of acetylcholine-related enzymes in the nucleus basalis. The remainder of the grains (11-20%) were classified as shared, i.e. overlied the plasma membrane of acetylcholinesterase-positive neuronal perikarya and dendrites. Extrapolation from light microscopic data, combined with the observation that shared grains were detected at several contact points along the plasma membrane of cells which also exhibited exclusive grains, made it possible to ascribe these membrane-associated receptors to the cholinergic neurons themselves rather than to abutting cellular profiles. Comparison of grain distribution with the frequency of occurrence of elements directly abutting the plasma membrane of neurotensin-labeled/cholinesterase-positive perikarya indicated that labeled cell surface receptors were more or less evenly distributed along the membrane as opposed to being concentrated opposite abutting axon terminals endowed or not with a visible junctional specialization. The low incidence of labeled binding sites found in close association with abutting axons makes it unlikely that only this sub-population of sites corresponds to functional receptors. On the contrary, the dispersion of labeled receptors seen here along the plasma membrane of cholinergic neurons suggests that neurotensin acts primarily in a paracrine mode to influence the magnocellular cholinergic system in the nucleus basalis.
Neurotensin is a tridecapeptide, originally isolated from bovine hypothalamus, ~2and subsequently found to be widely and heterogeneously distributed in the CNS of all vertebrate classes, including mammals, using both radioimmunoassay and immunohistochemical t e c h n i q u e s J 3'I6'24"38'44"59 In keeping with its postulated neurotransmitter role, neurotensin has been reported to be preferentially localized in the synaptosomal fraction of brain homogenates, 88 readily released from brain slices by depolarizing agents 37'5s and rapidly degraded by endogenous peptidases of C N S origin. ~4'~5 Furthermore, specific, high affinity neurotensin binding sites 7z89'97'98 have been found to be co-distributed with neuro-
*To whom correspondence should be addressed. Abbreviations: ACh, acetylcholine; ACHE, acetylcholinesterase; CHAT, choline acetyltransferase; DFP, bis-lmethyl-ethyl phosphorofluoridate; NBM, nucleus basalis magnocellularis.
tensin-immunoreactive terminals in many brain regions, 38'73'96 including those where exogenously applied neurotensin has been shown to produce altered electrical activity. 2'79 Activation of these sites is known to increase intracellular levels of c G M P 28 and rates of inositol phosphate turnover 3° in brain tissue and is likely to mediate the diverse physiological and behavioral effects documented after central neurotensin administration (for review, see Ref. 23). A high concentration of both 3H-labeled and 125Ilabeled neurotensin binding sites has been localized in the basal forebrain of the rat using light microscopic or film radioautography. 67~u'98 In addition, radioautographic investigations using emulsion-coated sections have shown [~25I]neurotensin binding sites to be predominantly concentrated over neurons in certain magnocellular basal forebrain nuclei of the rat, including the horizontal and vertical limbs of the diagonal band of Broca, the substantia innominata
377
378
E. SZIGETHY et al.
and the ventral pallidum 67 [the latter two regions being often referred to collectively as the nucleus basalis magnocellularis (NBM)56]. Adjacent section radioautography/pharmacohistochemistry subsequently revealed that these cells c o n t a i n e d acetylcholinesterase (ACHE), the m a j o r acetylcholine (ACh) degrading enzyme. 83 Since it has been shown that the vast majority o f AChE-positive n e u r o n s in these forebrain regions also contain choline acetyltransferase (CHAT), an A C h synthesizing enzyme, it can be assumed t h a t they are in fact cholinergicfl 2'53 The N B M is a m a j o r source o f cholinergic input to the entire neocortex 39'52'62 a n d has been implicated in the m e d i a t i o n o f n u m e r o u s complex behaviorsfl °'4~ The selective a c c u m u l a t i o n of [~25I]neurotensin binding sites over cholinergic n e u r o n s in this nucleus suggests t h a t neurotensin may m o d u l a t e the activity of these diffusely projecting cells. In s u p p o r t of such a m o d u l a t o r y role is the d e m o n s t r a t e d presence o f n e u r o t e n s i n - i m m u n o r e a c t i v e fibers in the basal forebrain providing a potential source of endogenous ligand to act at these sites. 5°'73'95 In order to further d o c u m e n t the association of [L25I]neurotensin binding sites with cholinergic neurons in the rat N B M a n d to assess the distribution of these sites in relation to cellular plasma membranes, we c o m b i n e d the techniques of 125I-radioa u t o g r a p h y and A C h E p h a r m a c o h i s t o c h e m i s t r y on the same sections b o t h at the light and electron microscopic levels.
EXPERIMENTAL PROCEDURES
Five adult male Sprague-Dawley rats (200-225 g) were anesthetized with chloral hydrate (350mg/kg, i.p.) and perfused through the ascending aorta with an ice-cold mixture of 1% tannic acid, 0.75% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M phosphate buffer) ~ The perfusion flow rate was set at 150 ml/min for the first 2.5 min and then slowed to 75 ml/min for an additional 8 min. This pre-fixation was shown to affect neither the distribution nor the kinetics of [~25I]neurotensin binding to rat brain slices.~8 To promote selective visualization of AChE-positive perikarya and dendrites, an additional three animals were given bis-l-methyl-ethyl phosphorofluoridate (DFP; Sigma, 2 mg/kg, i.m.) along with atropine sulphate (0.2 mg/kg, i.p.) 8-10 h prior to being killed. The perfused brains were dissected from the skull and the caudal NBM, comprising sub-lenticular, peri-pallidal and intracapsular corticopetal cell groups, was blocked on ice. Tissue blocks were immersed in ice-cold 0.12 M phosphate buffer and cut at 75 pm thickness on a vibrating microtome (Vibratome). Vibratome-cut slices were immediately collected in porcelain test spot plates and individually incubated with 0.3nM monoiodo [125I]Tyr3 neurotensin (2000 Ci/mmol; see Ref. 75 for details on iodination and purification) in 50 mM ice-cold Tris-HC1 buffer (pH 7.4) containing 5 mM MgC12, 0.2% bovine serum albumin, 2 x t0 5 M bacitracin, and 0.25 M sucrose for I h at 4°C. Every third slice was incubated with [125I]neurotensin as above, but in the presence of 0.1/~ M non-radioactive neurotensin for determination of non-specific binding. At the end of the incubation, slices were transferred through two consecutive ice-cold buffer baths (5 min/bath) and fixed for 30min by immersion in a solution containing 2.5 3% glutaraldehyde, 2% paraformaldehyde, and 15% picric
acid in 50mM phosphate buffer. At this point, slices were divided into two alternate series. Slices from the first series were processed for [~25I]neurotensin radioautography alone and slices from the second series for combined [~251]neurotensin radioautography and AChE histochemistry. Slices from series 1 were post-fixed for l h in a 2% phosphate-buffered OsO4 solution containing 7% dextrose, dehydrated in graded ethanols, and fiat-embedded in Epon between two plastic coverslips after measuring their radioactivity content in a Packard Multiprias gamma counter. Only those slices containing more than 1000 cpm/slice (total binding) were further processed for radioautography together with their corresponding blank slices (non-specific binding). Slices from series 2 were first processed for AChE histochemistry using modifications of procedures by Karnovsky and Roots, 42 Geneser-Jensen and Blackstad, 27 Hanker et al., 32"33 Tsuji 85 and Tsuji and Fournier. 87 In brief, slices were rinsed in a cold aqueous 25% sodium sulphate solution (3 x 5 s) and transferred to a filtered pre-incubation medium containing 2 m M copper sulphate, 10mM glycine and 0.2 mM ethopropazine in 50 mM cold acetate buffer (pH 5) for 15 min. They were then transferred to the same medium with the addition of 4raM acetylthiocholine iodide and incubated for 2 h at 4°C. For specificity controls (1) acetylthiocholine iodide was omitted from the incubation medium, (2) butyrylthiocholine iodide was substituted for acetylthiocholine iodide to give an indication of the endogenous butyrylcholinesterase activity, and (3) ethopropazine, a potent butyrylcholinesterase inhibitor was added to the incubation medium. The incubation was followed by 4 x l-rain rinses in cold 0.1 M acetate buffer, 15 min in cold 3.5% potassium ferricyanide, and repeated acetate buffer rinses. Slices were then rinsed for 15 min in cold 0.1 M Tris (pH 7.2), dipped in distilled H20 and incubated in 50 mg 3,3'-diaminobenzidine in 100ml 0,05M acetate buffer (pH 5.6) at 4 ° C . 33 Osmification and subsequent embedding was performed as described above for series 1. Only those 75-#m-thick slices from series 2 having visible AChE reaction product in large magnocellular-type neurons at the light microscopic level and with a radioactive content above 1000 cpm/slice (total binding) were processed for radioautography. All blank slides (non-specific binding) with visible AChE-positive neurons were also included for radioautographic processing. Slices from both series 1 and 2 were then re-embedded in Beem capsules, polymerized, trimmed and sectioned for either light or electron microscopy. For light microscopy, semi-thin (l-pm-thick) sections were cut from the surface of the blocks on a Reichert-Jung ultra-microtome, collected on glass slides, dipped in Kodak NTB-2 nuclear emulsion diluted I:1 with distilled water and exposed for eight to 12 weeks at 4°C. Sections from series 2 were carbon-coated prior to emulsion dipping to protect against possible chemography due to the presence of metallic AChE reaction product. These light microscopic radioautographs were developed in freshly prepared D-19 (Kodak; 4rain at 17°C). Sections were counterstained (series 1) or not (series 2) with Toluidine Blue and examined with a Leitz Orthoplan microscope. Grains were counted over [125I]neurotensin-labeled perikarya and proximal processes in sections from both series to determine nonspecific versus total labeling density (n grains/p m 2) for each condition. For electron microscopic radioautography, 80-nm-thin sections were cut from a second group of blocks, collected on parlodion-subbed slides, stained with lead citrate, sprayed with vaporized carbon and radioautographed by dipping into Ilford L4 emulsion diluted 1:4. After 12 weeks of exposure, these radioautograms were developed in D-19 diluted 1 : 5 with distilled water (1 min at 20°C), collected on 150-mesh copper grids, and examined with a JEOL 1200EX
Neurotensin receptors on cholinergic cells electron microscope after thinning the parlodion membrane in isoamyl acetate. Quantitative analysis of electron microscopic radioautographs was carried out in three slices (two totals/one blank) from each series (single-labeled/double-labeled) in three DFP-pretreated and three untreated rats. All sections were systematically scanned with a JEOL 1200EX electron microscope and labeled sites associated with large magnocellulartype neuronal soma and dendrites (series 1) or with AChE-positive perikarya and dendrites (series 2) were photographed at a magnification of 10,000 x. The distribution of silver grains over these cells was then analysed using a probability circle method modified from Williams?4 Briefly, a 50% resolution circle [diameter: 3.4 × half distance (HD); HD = 90 nM; see Refs 9 and 76], drawn on a transparent overlay, was centered over each grain (series 1: n = 457; series 2: n = 1021) and the structure (exclusive grains) or pair of structures (shared grains) included within this circle recorded and tabulated. Grains overlying an interface between a soma or dendrite and an unidentified profile were recorded as such. Grains associated with more than two structures (e.g. dendrite-axon-glia) were ascribed to the pair (e.g. dendrite/axon or dendrite/glia) closest to the center of the grain (defined as the geometric center of the resolution circle). To determine the distribution of specific binding, the distribution of grains in sections incubated with an excess of non-radioactive neurotensin was subtracted from that observed in sections incubated with [~25I]neurotensin alone after multiplying values from the former by the appropriate non-specific over total binding ratio (derived from grain counts over individual neurons in semi-thin sections) to proportionally adjust the number of silver grains within each tissue compartment. Finally, the resulting distribution (specific binding) was normalized to 100. The distribution of grains (total, specific and non-specific) in each series was then compared using a X 2 analysis. The criterion for statistical significance was set at P < 0.05. Sections from DFP-pretreated and DFP-untreated animals were pooled together for statistical analysis since it was previously shown that a single pretreatment with DFP did not produce a detectable difference in the density or distribution of [t251]neurotensin-labeling (unpublished observations). In order to determine the frequency of occurrence of elements directly abutting the plasma membrane of [125I]neurotensin-labeled (series 1) or AChE-reactive nerve cell bodies (series 2), five randomly-oriented lines were drawn on a transparent overlay and placed over 250 randomly selected electron micrographs with a visible perikaryal plasma membrane. The profile immediately adjoining the plasma membrane at the points of intersection of the lines with the membrane were recorded and the resulting distribution profile was compared with that of shared grains associated with somatic interfaces, by X 2 analysis.
379
and radiating processes. Only low levels of [~25I]neurotensin binding were detected in the surrounding neuropil, except over the endothelium of intraparenchymal capillaries and associated pericytes which were both heavily labeled. In sections incubated with an excess of non-radioactive neurotensin, silver grains were scarce and mainly associated with capillary walls. In fact, endothelial cells and pericytes appeared as heavily labeled in these sections as in sections incubated with [~25I]neurotensin alone. In semi-thin sections from double-labeled material (series 2), dark brown, flocular AChE-reaction product was heavily concentrated throughout the cytoplasm and proximal processes of magnocellular NBM neurons (Fig, lc,d). In DFP-pretreated animals, AChE staining was less intense than in sections from normal animals and mainly confined to cell bodies (Fig. ld). In both pretreated and untreated rats, the nucleus was devoid of AChE reactivity (Fig. lc,d). No detectable staining was observed on control slides. The vast majority (95%) of ACHEpositive cells in the NBM were labeled with [~25I]neurotensin (Fig. lc,d). These large, doublelabeled neurons were easily distinguished from the smaller, less intensely AChE-positive neurons of the neighboring pallidal regions, over which no [~25I]neurotensin labeling was apparent. The distribution of silver grains both within and outside NBM magnocellular neurons was identical to that observed in single-labeled sections (compare a, b with c, d in Fig. 1). None of the cells lacking AChE reaction product exhibited [~2SI]neurotensin labeling in either DFP-treated or DFP-untreated animals. However, occasional large AChE-positive cells were devoid of [J-'SI]neurotensin labeling. Sections incubated in the presence of non-radioactive neurotensin exhibited the same labeling pattern as their single-labeled counterparts. Grain counts performed over individual [~2~I]neurotensin-labeled neurons revealed that 85% of the grains in series 1 and 83% in series 2 could be attributed to specific binding. Of the total number of grains representing non-specific binding, 80% in series 1 and 87% in series 2 were associated with the nucleus.
RESULTS
Electron microscopy Light microscopy In light microscopic radioautographs of semi-thin sections taken from the surface of single-labeled slices (series 1), silver grains were mainly concentrated over the nerve cell bodies and proximal processes of NBM neurons (Fig. la,b). These neurons were characteristically large (long axis 24-38 pm) and distributed in clusters alongside or within the internal capsule. Within the cells, silver grains were distributed throughout the cytoplasm and occasionally over the nucleus (Fig. la,b). A proportion was also detected along the plasma membrane of both the soma NSC 36,'2 I)
In electron microscopic radioautographs from both single-labeled (series 1), and double-labeled (series 2) sections, [~25I]neurotensin-labeled binding sites were detected in the form of individual silver grains. The distribution of these silver grains, as assessed by probability circle analysis, was significantly different in sections incubated with [~25I]neurotensin alone (total binding) as compared with sections incubated with an excess of nonradioactive neurotensin ( P < 0 . 0 0 1 ) . After subtraction of non-specific from total binding, 80% of the grains detected in single-labeled sections were
380
E. SZIGETI4Yet al.
Fig. 1. Radioautographic distribution of [t25I]neurotensin-labeled binding sites in 1-#m-thick semi-thin sections from rat NBM stained with Toluidine Blue (a, b) or processed for AChE histochemistry (c, d). Note the selective accumulation of silver grains over neuronal perikarya and proximal dendrites of magnocellular type neurons and the relative sparing of the surrounding neuropil (a, b). In c and d, [~2~I]neurotensin-labeled cells are AChE-positive. AChE-reactivity is selectively concentrated in the neuronal cytoplasm while [125I]neurotensin-labeling is found over the entire cell, including even the nucleus. Scale bar = 10/~m.
ascribed to individual soma and dendrites (exclusive grains) as opposed to 20% to cellular interfaces (shared grains). In double-labeled material, the proportion was 89% for exclusive grains versus 11% for shared (Table l).
The cytological features of [J25I]neurotensinlabeled neurons were best appreciated in singlelabeled sections and conformed to those previously reported for basal forebrain cholinergic magnoc e l l u l a r n e u r o n s . 3'35'36'61'99 Briefly, these neurons were
Neurotensin receptors on cholinergic cells
381
Table I. Distribution frequency of specifically labeled neurotensin binding sites associated with cholinergic neurons in rat nucleus basalis magnocellularis*
Exclusive grains Soma Cytoplasm Nucleus Dendrite Shared grains Somatic Somato-unidentified Somatoglial Somatoaxonal Somatodendritic Somatosomatic Dendritic Dendroaxonal Dendro-unidentified Dendroglial
[125I]Neurotensin (series 1)
[125l]Neurotensin + AChE (series 2)
80
89
48 22 10 20 11 7 2 1
0306-4522/90 $3.00 + 0.00 Pergamon Press plc 1990 IBRO
U L T R A S T R U C T U R A L LOCALIZATION OF [125I]NEUROTENSIN B I N D I N G SITES TO CHOLINERGIC NEURONS OF THE RAT NUCLEUS BASALIS MAGNOCELLULARIS E. SZIGETHY,K. LEONARD and A. BEAUDET* Laboratory of Neuroanatomy, Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada H3A 2B4 Abstract--The distribution of specifically-labeled neurotensin binding sites was examined in relation to that of cholinergic neurons in the rat nucleus basalis magnocellularis at both light and electron microscopic levels. Lightly prefixed forebrain slices were either labeled with [t25l](Tyr3) neurotensin alone or processed for combined [~25I]neurotensin radioautography and acetylcholinesterase histochemistry. In light microscopic radioautographs from l-#m-thick sections taken from the surface of single-labeled slices, silver grains were found to be preferentially localized over perikarya and proximal processes of nucleus basalis cells. The label was distributed both throughout the cytoplasm and along the plasma membrane of magnocellular neurons all of which were found to be cholinesterase-positive in double-labeled material. Probability circle analysis of silver grain distribution in electron microscopic radioautographs confirmed that the major fraction (80--89%) of specifically-labeled binding sites associated with cholinesterasereactive cell bodies and dendrites was intraneuronal. These intraneuronal sites were mainly dispersed throughout the cytoplasm and are thus likely to represent receptors undergoing synthesis, transport and/or recycling. A proportion of the specific label was also localized over the nucleus, suggesting that neurotensin could modulate the expression of acetylcholine-related enzymes in the nucleus basalis. The remainder of the grains (11-20%) were classified as shared, i.e. overlied the plasma membrane of acetylcholinesterase-positive neuronal perikarya and dendrites. Extrapolation from light microscopic data, combined with the observation that shared grains were detected at several contact points along the plasma membrane of cells which also exhibited exclusive grains, made it possible to ascribe these membrane-associated receptors to the cholinergic neurons themselves rather than to abutting cellular profiles. Comparison of grain distribution with the frequency of occurrence of elements directly abutting the plasma membrane of neurotensin-labeled/cholinesterase-positive perikarya indicated that labeled cell surface receptors were more or less evenly distributed along the membrane as opposed to being concentrated opposite abutting axon terminals endowed or not with a visible junctional specialization. The low incidence of labeled binding sites found in close association with abutting axons makes it unlikely that only this sub-population of sites corresponds to functional receptors. On the contrary, the dispersion of labeled receptors seen here along the plasma membrane of cholinergic neurons suggests that neurotensin acts primarily in a paracrine mode to influence the magnocellular cholinergic system in the nucleus basalis.
Neurotensin is a tridecapeptide, originally isolated from bovine hypothalamus, ~2and subsequently found to be widely and heterogeneously distributed in the CNS of all vertebrate classes, including mammals, using both radioimmunoassay and immunohistochemical t e c h n i q u e s J 3'I6'24"38'44"59 In keeping with its postulated neurotransmitter role, neurotensin has been reported to be preferentially localized in the synaptosomal fraction of brain homogenates, 88 readily released from brain slices by depolarizing agents 37'5s and rapidly degraded by endogenous peptidases of C N S origin. ~4'~5 Furthermore, specific, high affinity neurotensin binding sites 7z89'97'98 have been found to be co-distributed with neuro-
*To whom correspondence should be addressed. Abbreviations: ACh, acetylcholine; ACHE, acetylcholinesterase; CHAT, choline acetyltransferase; DFP, bis-lmethyl-ethyl phosphorofluoridate; NBM, nucleus basalis magnocellularis.
tensin-immunoreactive terminals in many brain regions, 38'73'96 including those where exogenously applied neurotensin has been shown to produce altered electrical activity. 2'79 Activation of these sites is known to increase intracellular levels of c G M P 28 and rates of inositol phosphate turnover 3° in brain tissue and is likely to mediate the diverse physiological and behavioral effects documented after central neurotensin administration (for review, see Ref. 23). A high concentration of both 3H-labeled and 125Ilabeled neurotensin binding sites has been localized in the basal forebrain of the rat using light microscopic or film radioautography. 67~u'98 In addition, radioautographic investigations using emulsion-coated sections have shown [~25I]neurotensin binding sites to be predominantly concentrated over neurons in certain magnocellular basal forebrain nuclei of the rat, including the horizontal and vertical limbs of the diagonal band of Broca, the substantia innominata
377
378
E. SZIGETHY et al.
and the ventral pallidum 67 [the latter two regions being often referred to collectively as the nucleus basalis magnocellularis (NBM)56]. Adjacent section radioautography/pharmacohistochemistry subsequently revealed that these cells c o n t a i n e d acetylcholinesterase (ACHE), the m a j o r acetylcholine (ACh) degrading enzyme. 83 Since it has been shown that the vast majority o f AChE-positive n e u r o n s in these forebrain regions also contain choline acetyltransferase (CHAT), an A C h synthesizing enzyme, it can be assumed t h a t they are in fact cholinergicfl 2'53 The N B M is a m a j o r source o f cholinergic input to the entire neocortex 39'52'62 a n d has been implicated in the m e d i a t i o n o f n u m e r o u s complex behaviorsfl °'4~ The selective a c c u m u l a t i o n of [~25I]neurotensin binding sites over cholinergic n e u r o n s in this nucleus suggests t h a t neurotensin may m o d u l a t e the activity of these diffusely projecting cells. In s u p p o r t of such a m o d u l a t o r y role is the d e m o n s t r a t e d presence o f n e u r o t e n s i n - i m m u n o r e a c t i v e fibers in the basal forebrain providing a potential source of endogenous ligand to act at these sites. 5°'73'95 In order to further d o c u m e n t the association of [L25I]neurotensin binding sites with cholinergic neurons in the rat N B M a n d to assess the distribution of these sites in relation to cellular plasma membranes, we c o m b i n e d the techniques of 125I-radioa u t o g r a p h y and A C h E p h a r m a c o h i s t o c h e m i s t r y on the same sections b o t h at the light and electron microscopic levels.
EXPERIMENTAL PROCEDURES
Five adult male Sprague-Dawley rats (200-225 g) were anesthetized with chloral hydrate (350mg/kg, i.p.) and perfused through the ascending aorta with an ice-cold mixture of 1% tannic acid, 0.75% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M phosphate buffer) ~ The perfusion flow rate was set at 150 ml/min for the first 2.5 min and then slowed to 75 ml/min for an additional 8 min. This pre-fixation was shown to affect neither the distribution nor the kinetics of [~25I]neurotensin binding to rat brain slices.~8 To promote selective visualization of AChE-positive perikarya and dendrites, an additional three animals were given bis-l-methyl-ethyl phosphorofluoridate (DFP; Sigma, 2 mg/kg, i.m.) along with atropine sulphate (0.2 mg/kg, i.p.) 8-10 h prior to being killed. The perfused brains were dissected from the skull and the caudal NBM, comprising sub-lenticular, peri-pallidal and intracapsular corticopetal cell groups, was blocked on ice. Tissue blocks were immersed in ice-cold 0.12 M phosphate buffer and cut at 75 pm thickness on a vibrating microtome (Vibratome). Vibratome-cut slices were immediately collected in porcelain test spot plates and individually incubated with 0.3nM monoiodo [125I]Tyr3 neurotensin (2000 Ci/mmol; see Ref. 75 for details on iodination and purification) in 50 mM ice-cold Tris-HC1 buffer (pH 7.4) containing 5 mM MgC12, 0.2% bovine serum albumin, 2 x t0 5 M bacitracin, and 0.25 M sucrose for I h at 4°C. Every third slice was incubated with [125I]neurotensin as above, but in the presence of 0.1/~ M non-radioactive neurotensin for determination of non-specific binding. At the end of the incubation, slices were transferred through two consecutive ice-cold buffer baths (5 min/bath) and fixed for 30min by immersion in a solution containing 2.5 3% glutaraldehyde, 2% paraformaldehyde, and 15% picric
acid in 50mM phosphate buffer. At this point, slices were divided into two alternate series. Slices from the first series were processed for [~25I]neurotensin radioautography alone and slices from the second series for combined [~251]neurotensin radioautography and AChE histochemistry. Slices from series 1 were post-fixed for l h in a 2% phosphate-buffered OsO4 solution containing 7% dextrose, dehydrated in graded ethanols, and fiat-embedded in Epon between two plastic coverslips after measuring their radioactivity content in a Packard Multiprias gamma counter. Only those slices containing more than 1000 cpm/slice (total binding) were further processed for radioautography together with their corresponding blank slices (non-specific binding). Slices from series 2 were first processed for AChE histochemistry using modifications of procedures by Karnovsky and Roots, 42 Geneser-Jensen and Blackstad, 27 Hanker et al., 32"33 Tsuji 85 and Tsuji and Fournier. 87 In brief, slices were rinsed in a cold aqueous 25% sodium sulphate solution (3 x 5 s) and transferred to a filtered pre-incubation medium containing 2 m M copper sulphate, 10mM glycine and 0.2 mM ethopropazine in 50 mM cold acetate buffer (pH 5) for 15 min. They were then transferred to the same medium with the addition of 4raM acetylthiocholine iodide and incubated for 2 h at 4°C. For specificity controls (1) acetylthiocholine iodide was omitted from the incubation medium, (2) butyrylthiocholine iodide was substituted for acetylthiocholine iodide to give an indication of the endogenous butyrylcholinesterase activity, and (3) ethopropazine, a potent butyrylcholinesterase inhibitor was added to the incubation medium. The incubation was followed by 4 x l-rain rinses in cold 0.1 M acetate buffer, 15 min in cold 3.5% potassium ferricyanide, and repeated acetate buffer rinses. Slices were then rinsed for 15 min in cold 0.1 M Tris (pH 7.2), dipped in distilled H20 and incubated in 50 mg 3,3'-diaminobenzidine in 100ml 0,05M acetate buffer (pH 5.6) at 4 ° C . 33 Osmification and subsequent embedding was performed as described above for series 1. Only those 75-#m-thick slices from series 2 having visible AChE reaction product in large magnocellular-type neurons at the light microscopic level and with a radioactive content above 1000 cpm/slice (total binding) were processed for radioautography. All blank slides (non-specific binding) with visible AChE-positive neurons were also included for radioautographic processing. Slices from both series 1 and 2 were then re-embedded in Beem capsules, polymerized, trimmed and sectioned for either light or electron microscopy. For light microscopy, semi-thin (l-pm-thick) sections were cut from the surface of the blocks on a Reichert-Jung ultra-microtome, collected on glass slides, dipped in Kodak NTB-2 nuclear emulsion diluted I:1 with distilled water and exposed for eight to 12 weeks at 4°C. Sections from series 2 were carbon-coated prior to emulsion dipping to protect against possible chemography due to the presence of metallic AChE reaction product. These light microscopic radioautographs were developed in freshly prepared D-19 (Kodak; 4rain at 17°C). Sections were counterstained (series 1) or not (series 2) with Toluidine Blue and examined with a Leitz Orthoplan microscope. Grains were counted over [125I]neurotensin-labeled perikarya and proximal processes in sections from both series to determine nonspecific versus total labeling density (n grains/p m 2) for each condition. For electron microscopic radioautography, 80-nm-thin sections were cut from a second group of blocks, collected on parlodion-subbed slides, stained with lead citrate, sprayed with vaporized carbon and radioautographed by dipping into Ilford L4 emulsion diluted 1:4. After 12 weeks of exposure, these radioautograms were developed in D-19 diluted 1 : 5 with distilled water (1 min at 20°C), collected on 150-mesh copper grids, and examined with a JEOL 1200EX
Neurotensin receptors on cholinergic cells electron microscope after thinning the parlodion membrane in isoamyl acetate. Quantitative analysis of electron microscopic radioautographs was carried out in three slices (two totals/one blank) from each series (single-labeled/double-labeled) in three DFP-pretreated and three untreated rats. All sections were systematically scanned with a JEOL 1200EX electron microscope and labeled sites associated with large magnocellulartype neuronal soma and dendrites (series 1) or with AChE-positive perikarya and dendrites (series 2) were photographed at a magnification of 10,000 x. The distribution of silver grains over these cells was then analysed using a probability circle method modified from Williams?4 Briefly, a 50% resolution circle [diameter: 3.4 × half distance (HD); HD = 90 nM; see Refs 9 and 76], drawn on a transparent overlay, was centered over each grain (series 1: n = 457; series 2: n = 1021) and the structure (exclusive grains) or pair of structures (shared grains) included within this circle recorded and tabulated. Grains overlying an interface between a soma or dendrite and an unidentified profile were recorded as such. Grains associated with more than two structures (e.g. dendrite-axon-glia) were ascribed to the pair (e.g. dendrite/axon or dendrite/glia) closest to the center of the grain (defined as the geometric center of the resolution circle). To determine the distribution of specific binding, the distribution of grains in sections incubated with an excess of non-radioactive neurotensin was subtracted from that observed in sections incubated with [~25I]neurotensin alone after multiplying values from the former by the appropriate non-specific over total binding ratio (derived from grain counts over individual neurons in semi-thin sections) to proportionally adjust the number of silver grains within each tissue compartment. Finally, the resulting distribution (specific binding) was normalized to 100. The distribution of grains (total, specific and non-specific) in each series was then compared using a X 2 analysis. The criterion for statistical significance was set at P < 0.05. Sections from DFP-pretreated and DFP-untreated animals were pooled together for statistical analysis since it was previously shown that a single pretreatment with DFP did not produce a detectable difference in the density or distribution of [t251]neurotensin-labeling (unpublished observations). In order to determine the frequency of occurrence of elements directly abutting the plasma membrane of [125I]neurotensin-labeled (series 1) or AChE-reactive nerve cell bodies (series 2), five randomly-oriented lines were drawn on a transparent overlay and placed over 250 randomly selected electron micrographs with a visible perikaryal plasma membrane. The profile immediately adjoining the plasma membrane at the points of intersection of the lines with the membrane were recorded and the resulting distribution profile was compared with that of shared grains associated with somatic interfaces, by X 2 analysis.
379
and radiating processes. Only low levels of [~25I]neurotensin binding were detected in the surrounding neuropil, except over the endothelium of intraparenchymal capillaries and associated pericytes which were both heavily labeled. In sections incubated with an excess of non-radioactive neurotensin, silver grains were scarce and mainly associated with capillary walls. In fact, endothelial cells and pericytes appeared as heavily labeled in these sections as in sections incubated with [~25I]neurotensin alone. In semi-thin sections from double-labeled material (series 2), dark brown, flocular AChE-reaction product was heavily concentrated throughout the cytoplasm and proximal processes of magnocellular NBM neurons (Fig, lc,d). In DFP-pretreated animals, AChE staining was less intense than in sections from normal animals and mainly confined to cell bodies (Fig. ld). In both pretreated and untreated rats, the nucleus was devoid of AChE reactivity (Fig. lc,d). No detectable staining was observed on control slides. The vast majority (95%) of ACHEpositive cells in the NBM were labeled with [~25I]neurotensin (Fig. lc,d). These large, doublelabeled neurons were easily distinguished from the smaller, less intensely AChE-positive neurons of the neighboring pallidal regions, over which no [~25I]neurotensin labeling was apparent. The distribution of silver grains both within and outside NBM magnocellular neurons was identical to that observed in single-labeled sections (compare a, b with c, d in Fig. 1). None of the cells lacking AChE reaction product exhibited [~2SI]neurotensin labeling in either DFP-treated or DFP-untreated animals. However, occasional large AChE-positive cells were devoid of [J-'SI]neurotensin labeling. Sections incubated in the presence of non-radioactive neurotensin exhibited the same labeling pattern as their single-labeled counterparts. Grain counts performed over individual [~2~I]neurotensin-labeled neurons revealed that 85% of the grains in series 1 and 83% in series 2 could be attributed to specific binding. Of the total number of grains representing non-specific binding, 80% in series 1 and 87% in series 2 were associated with the nucleus.
RESULTS
Electron microscopy Light microscopy In light microscopic radioautographs of semi-thin sections taken from the surface of single-labeled slices (series 1), silver grains were mainly concentrated over the nerve cell bodies and proximal processes of NBM neurons (Fig. la,b). These neurons were characteristically large (long axis 24-38 pm) and distributed in clusters alongside or within the internal capsule. Within the cells, silver grains were distributed throughout the cytoplasm and occasionally over the nucleus (Fig. la,b). A proportion was also detected along the plasma membrane of both the soma NSC 36,'2 I)
In electron microscopic radioautographs from both single-labeled (series 1), and double-labeled (series 2) sections, [~25I]neurotensin-labeled binding sites were detected in the form of individual silver grains. The distribution of these silver grains, as assessed by probability circle analysis, was significantly different in sections incubated with [~25I]neurotensin alone (total binding) as compared with sections incubated with an excess of nonradioactive neurotensin ( P < 0 . 0 0 1 ) . After subtraction of non-specific from total binding, 80% of the grains detected in single-labeled sections were
380
E. SZIGETI4Yet al.
Fig. 1. Radioautographic distribution of [t25I]neurotensin-labeled binding sites in 1-#m-thick semi-thin sections from rat NBM stained with Toluidine Blue (a, b) or processed for AChE histochemistry (c, d). Note the selective accumulation of silver grains over neuronal perikarya and proximal dendrites of magnocellular type neurons and the relative sparing of the surrounding neuropil (a, b). In c and d, [~2~I]neurotensin-labeled cells are AChE-positive. AChE-reactivity is selectively concentrated in the neuronal cytoplasm while [125I]neurotensin-labeling is found over the entire cell, including even the nucleus. Scale bar = 10/~m.
ascribed to individual soma and dendrites (exclusive grains) as opposed to 20% to cellular interfaces (shared grains). In double-labeled material, the proportion was 89% for exclusive grains versus 11% for shared (Table l).
The cytological features of [J25I]neurotensinlabeled neurons were best appreciated in singlelabeled sections and conformed to those previously reported for basal forebrain cholinergic magnoc e l l u l a r n e u r o n s . 3'35'36'61'99 Briefly, these neurons were
Neurotensin receptors on cholinergic cells
381
Table I. Distribution frequency of specifically labeled neurotensin binding sites associated with cholinergic neurons in rat nucleus basalis magnocellularis*
Exclusive grains Soma Cytoplasm Nucleus Dendrite Shared grains Somatic Somato-unidentified Somatoglial Somatoaxonal Somatodendritic Somatosomatic Dendritic Dendroaxonal Dendro-unidentified Dendroglial
[125I]Neurotensin (series 1)
[125l]Neurotensin + AChE (series 2)
80
89
48 22 10 20 11 7 2 1
