The Journal of Neuroscience, April 1989, g(4): 13751388

A Morphometric Analysis of Functionally Defined Subpopulations of in the Paraventricular of the Rat with Observations on the Effects of Colchicine

J.-H. Rho’ and L. W. Swanson”* ‘Neural Systems Laboratory, The Salk Institute for Biological Studies, La Jolla, California 92037, and 2Howard Hughes Medical Institute, Bethesda, Maryland 20817

Two populations of neurons in the paraventricular nucleus The paraventricular nucleus of the (PVH) con- of the hypothalamus that have different efferent projections tains a heterogenouspopulation of neuronsthat have been dif- and physiological roles in the regulation of visceral re- ferentiated on the basis of their content and on sponses were characterized morphologically with a com- their patterns of afferent and efferent connections (seeSwanson bined intracellular filling, retrograde tracer, and immunohis- and Sawchenko, 1983; Swansonet al., 1986, 1987, for reviews). tochemical method. Neuroendocrine cells were retrogradely Two classesof neuroendocrine cells are found in the nucleus. labeled by an intravenous injection of Fast blue, and distin- One classconsists of magnocellularcells that releaseoxytocin guished from descending neurons that were retrogradely or vasopressininto the general circulation at the level of the labeled by an injection of fluorogold into the spinal cord. , while the other consistsof parvicellular neu- Retrogradely labeled neurons were selectively penetrated rons that secretehormones into the hypophyseal portal system and filled intracellularly with Lucifer yellow to visualize de- to influence in turn the secretionof . tailed features of their morphology. Corticotropin-releasing These neurons together provide an interface between the CNS (CRH)-containing neurons were distinguished from and the , translating neural information into other neuroendocrine cells by immunostaining the tissue with a physiological code that consistsof blood-borne factors. Parvi- an antiserum to rat CRH. Morphometric features of defined cellular neurons that contain corticotropin-releasing hormone populations of neurons were then quantified and recon- (CRH), for example, have a specializedfunction sincethey play structed graphically to generate multicellular montage draw- a major role in regulating the secretion of ACTH from the ings that demonstrate their spatial organization. Descending anterior pituitary, and are therefore involved in mediating ad- neurons were further separated into dorsal and ventral me- renal glucocorticoid responsesto a wide variety of stressors. dial parvicellular components, while the neuroendocrine These CRH neurons are highly complex in the sensethat they population was differentiated into parvicellular and mag- may also express many other neuoactive , including nocellular groups. The mean somal areas, total dendritic , angiotensin, enkephalin, neurotensin, and chole- lengths, and spine densities were compared between groups cystokinin, and the ratios of these peptides are altered by cir- of neurons, and these showed significant differences across culating levels of glucocorticoids. Descending neurons in the cell types. These measures were also dramatically affected PVH, on the other hand, send to the brain stem and by colchicine, which appears to induce the formation of new spinal cord that are presumably involved in coordinating both dendritic appendages, swelling of the , and reduction sympathetic and parasympathetic, as well as somatomotor re- of dendritic length. Whether colchicine is acting directly upon sponses,with endocrine activity. cytoskeletal structures or indirectly by altering the physiol- The PVH is, however, anatomically compartmentalized, with ogy of the cell is unclear. However, the precise effects of descendingand neuroendocrine cells segregatedlargely, though colchicine on mean somal area, total dendritic length, and not completely, within separateparts of the nucleus. Most de- spine density appear to be dependent upon individual cell scendingneurons, for example, are distributed in distinct clus- type. Colchicine may therefore act in a nonspecific, but none- ters that have been calledthe dorsal (dp), ventral medial (mpv), theless highly selective, manner in disrupting an endoge- and lateral (lp) parvicellular parts of the PVH (Swansonand nous mechanism regulating the number, morphology, and Kuypers, 1980). Magnocellular neuronsare clusteredin 3 dense- location of spines. ly packed nuclear groups that can be distinguished in Nissl- stained material by their density and the size of their somata. Parvicellular neurosecretory neurons, on the other hand, are centered in the dorsal medial parvicellular (mpd) and periven- tricular (pv) parts of the PVH, which contain a more hetero- Received June 6, 1988; revised Aug. 15, 1988; accepted Aug. 25, 1988. This work was supported in part by NIH Grant RlNS16686C and by the Rice geneousprofile of morphologically and immunohistochemically Foundation. We thank Dr. D. A. Brittain for help with the computer system, Mr. defined cell types, including those synthesizing CRH, thyrotro- K. Trulock for photographic assistance, and Mr. D. Warren for preparing the pin-releasing hormone (TRH), growth hormone-releasinghor- manuscript. Correspondence should be addressed to L. W. Swanson, The Neural Systems mone (GRH), and dopamine. Laboratory, The Salk Institute, P.O. Box 85800, San Diego, CA 92138-9216. Despite this anatomical compartmentalization, neuroendo- Copyright 0 1989 Society for Neuroscience 0270-6474/89/041375-14$02.00/O crine, autonomic and somatomotor responsesto particular stim- 1376 Rho and Swanson - Morphometrics of Paraventricular Neurons uli require integrated neurological activity that must be coor- dinated between and within these neuronal populations. In the Results present study, we have analyzed the spatial relationships and Following the intravenous injection of Fast blue, magnocellular morphological features of several neuroanatomically distin- and parvicellular neurosecretory neurons in the PVH appeared guishable cell types within the PVH using immunohistochem- to be selectively labeled regrogradely since their terminals ical and retrograde labeling methods combined with an intra- end in regions that lack a blood-brain barrier (the posterior cellular filling technique (Rho and Swanson, 1987). In 100-Frn- pituitary and neurohemal zone of the ), and thick paraformaldehyde-fixed tissue sections through the PVH, connectional studies indicate that the PVH does not project to rctrogradely labeled descending or neuroendocrine cells were any other circumventricular organ (see Rho and Swanson, 1987). selectively penetrated and intracellularly filled with Lucifer yel- On the other hand, injections of fluorogold into the spinal cord low (LY). Tissue sections were then processed for immuno- led to extensive retrograde labeling of neurons in the dorsal, histochemistry to distinguish CRH-immunoreactive cells from ventral medial, and lateral parvicellular parts of the nucleus (see other neuroendocrine cells. Finally, with computer graphics Fig. 3, inset). The neuroendocrine population of cells was further methods, their morphological features were quantified and com- divided on 2 grounds. First, magnocellular neurons were local- pared to characterize differences between defined groups of neu- ized on topological grounds within the lateral subdivision of the rons. posterior magnocellular part of the nucleus (where the vast ma- jority of neurons are of this class and are vasopressinergic; see Swanson and Sawchenko, 1983). And second, one class of parvi- Materials and Methods cellular neurosecretory neurons was positively identified im- Neuroanatomical staining. To differentiatedescending from neuroen- munohistochemically with an antiserum to CRH. To enhance docrinecell types in the PVH, 0.8 ml of a 0.25% aqueous solution of CRH immunostaining one group of animalswas pretreated with Fast blue (Dr. Illig Gml3H and Co., Grob-Umstadt, Germany) was intraventricular colchicine 3 d prior to death. Both descending injected into the tail vein of 38 adult male rats (Sprague-Dawley), and and neuroendocrine cell types were filled intracellularly in this pellets of fluorogold (Fluorochrome Inc., Inglewood, CO) were placed in the second thoracic segment of the spinal cord 7-14 d prior to death. material, and the resulting morphology was compared with that To enhance the level of CRH immunostaining, a solution of 80 Kg from untreated material. To simplify the analysis,one charac- cholchicine in 20 ~1 of 0.9% saline was injected into the lateral cerebral teristic level through the PVH was chosen for intensive study ventricle of 18 rats 3 d prior to death. To prepare tissue slices for (Figs. l-5). This plane of section lies through midrostrocaudal intracellular filling, animals were first anesthetized with 35% chloral levels of the medial and dorsal parvicellular parts and the lateral hydrate and then perfused transcardially and sequentially with solutions of 4% paraformaldehyde in acetate buffer (pH 6.5) with 0.025% glutar- subdivision of the posterior magnocellularpart of the nucleus. aldehyde, 4% paraformaldehyde in borate buffer (pH 9.5) with 0.025% glutaraldehyde, and 0.1 M PBS. The brains were then removed and cut Neuronal morphology into 100~pm-thick frontal sections through the PVH with a Vibratome. From 38 animals, 128 descendingneurons and 2 14 neuroen- Individual sections were placed on subbed slides and kept cold in a pool of phosphate buffer in a humidified chamber until they were placed on docrine cells were filled intracellularly with LY and digitally the intracellular injection apparatus. In each tissue slice, 5-10 Fast blue- reconstructed with the aid of computer graphics. Forty-four of or fluorogold-labeled neurons were selectively penetrated and filled in- the neuroendocrine neuronsstained intensely in a characteristic tracellulaily with pulled 20 MO resistance glass micropipettes. The dif- punctate manner with anti-CRH (seeFig. Sk, while 44 others fusion of LY, pulsed from the microelectrode tip, was observed with in the posterior magnocellularpart of the PVH had distinctive long focal-length water-immersion fluorescence optics. LY diffuses amorphously until the advancing microelectrode penetrates a cell body features of magnocellularneurons, including a large soma and or process, when the dye becomes confined intracellularly. Finally, after characteristic . Digital outlines of individual neurons several neurons were filled intracellularly, the tissue sections were pro- were combined graphically into multicellular montagedrawings cessed for immunohistochemistry. A rabbit primary antiserum against (Figs. 1-5) consisting of 50-60 representative neurons distrib- rat CRH (r70. eift of W. Vale). diluted l/2000. wasused for 48 hr at 4”C, and goat’aiti-rabbit, rho&mine-conjugateh secondary antibodies uted within particular cytoarchitectonic parts of the nucleusand (Tago Inc., Burlingame, CA), diluted l/50, were used for 1 hr at room resembling camera lucida drawings of Golgi-stained neurons. temperature. Tissue sections were then air-dried overnight, cleared in However, the morphological features of each in these graded ethanols, methyl salicylate, and xylene, and mounted with D.P.X. montagedrawings were analyzed separately(see Figs. 6 and 7), Morphometric quant$cation. To characterize the morphology of iden- and the presumedfunctional roles of theseneurons were char- tified cell types in the PVH, cleared and mounted tissue sections con- taining LY-stained neurons were placed on the motor-driven stage of acterized on the basisof their location, retrograde labeling, and a fluorescencecompound microscope coupled to a Joyce-LoeblMag- immunohistochemical staining patterns. iscan Image Analysis system. Under low magnification, binary outlines The descending cell population was found to be relatively were first drawn and superimposed with a lightpen over SITC video homogeneous,with many neurons sharing similar morpholog- images of the cytoarchitectonic boundaries that subdivide and circum- scribe the PVH. Then, under higher magnification (Leitz 100 x oil ob- ical features. In sectionsfrom animals that had not been treated jective with a 10 x phototube), binary outlines of the 5-10 LY-stained with colchicine (Fig. l), they typically gave rise to 3 major neurons in each tissue section were drawn in registration with cytoar- dendrites that extend from medium to large cell bodies in a chitectonic boundaries and with each other, not&g detailed features of stereotypical tripolar arrangement, and the axon often arises axons and dendrites, and counting dendritic spines and axon collaterals. from one of the dendrites.Their dendritesare long and thin and These binary outlines were then combined graphically into multicellular montagedrawings, consisting of 50-60representative neurons that were more highly branched than those of neuroendocrine cells. In evenlv distributed over the extent of their cvtoarchitectonic subdivision. addition, they tend to courseparallel to the long axis of the part The n;mber of spines, the total dendritic length, and the cross-sectional of the nucleis they lie within (as viewed in frontal sections). area of the somata were quantified for all neurons that were recon- The descendingneurons also have a sparsecomplement of spines strutted graphically. Finally, in addition to proprietary software, several commercially available software packages, including Symphony (Lotus (as usually defined at the light microscopic level), which usually Development Co.) and SPSS/PC+ (Microsoft Corp.), were also used have a short, thin stalk with a bulb on the end. Spinesmay be for data manipulation and statistical analysis. found on any part of the , and occasionally on the cell The Journal of Neuroscience, April 1989, 9(4) 1377

Figure 1. Drawing of the normal mor- phology of neurons in the dorsal and ventral medial parvicellular parts of the PVH that project to the spinalcord in the adult male rat. In this, and the fol- lowing 5 figures,LY-filled cellsfrom a \ number of experiments with tissue sec- tions from approximately the same level have been combined to provide an overview of the total dendritic domain of particular subpopulations. Note that axons indicated by arrowheads give rise to a collateral that reenters the PVH. Comparewith Figure2. Abbreviations: mdp, dorsal medial parvicellular; pm, posteriormagnocellular; pv, periven- tricular. body as well. Although descending neurons in the dorsal and branch reenteredthe part of the PVH containing the parent cell ventral medial parvicellular parts of the PVH are separated by body, although its precise site(s) of termination could usually the dorsal medial parvicellular part of the nucleus, the dendrites not be determined with certainty (seeFig. 1). of some descending neurons span this intervening region and Parvicellular CRH neuroendocrinecells were localized mostly display spines. It should also be pointed out that while the in the dorsal medial parvicellular part of the PVH. These neu- majority of dendrites from descending neurons in the dorsal and rons typically have 2 relatively thick primary dendrites that ventral medial parvicellular parts ofthe nucleus stay within their extend from opposite sidesof the cell body in a bipolar arrange- respective cytoarchitectonic borders, a “halo” of secondary den- ment and branch once (Fig. 3). In general,the dendritesof these drites surrounds each region. Thus, some dendrites of descend- neuronsare alignedin such a way that they coursemore or less ing neurons extend into the medial rim of the lateral posterior parallel to the borders of the dorsal medial parvicellular part of magnocellular (vasopressinergic) part of the nucleus and into the PVH. When both dendrites wrap around and extend in the the periventricular part of the nucleus, as well as into the dorsal same direction to conform to the borders of this part of the medial parvicellular part. PVH, a thin varicose axon typically emergesfrom the opposite A distinct axon could be distinguished from dendrites in 22 side and extends laterally toward the fornix. The presenceof of the 61 descending neurons examined in normal material. terminal boutons arising from the axon in the zone surrounding These axons were thin and varicose, and often arose from a the PVH has been described previously (Rho and Swanson, primary dendrite, where an axon hillock 5-10 pm long was 1987). These cells have several distinct types of spineson their observed. In a few cases the axon could be traced for several somataand dendrites: large appendage-likeextensions are often hundred microns, and when this was possible a bifurcation was located on the cell body, long hair-like spinesare distributed on usually observed. The major branch continued laterally or ven- the cell body and proximal dendrites, and smaller bulbous-type trolaterally toward the lateral hypothalamic area, while the other spinesare typical on more distal processes. 1378 Rho and Swanson * Morphometrics of Paraventricular Neurons

Figure 2. As in Figure 1, neurons that project to the spinal cord are presented, except in this figure the animals had been pretreated for 3 d with colchicine. Abbreviations as in Figure 1.

Magnocellular neurosecretoryneurons in normal materialwere the sakeof convenience. While other large neuroendocrine cells distinguished from parvicellular neurosecretory neuronsin the resemblingthose in the posterior magnocellularregion were also present analysisby the size and shapeof their somata, which sparsely scattered in other parts of the PVH, they were not were larger and more rounded, and by their location within the included in the sampleof magnocellularneurosecretory neurons lateral areaof the posterior magnocellularpart of the PVH (Fig. usedfor quantitative analysis (below). 4). Perhaps due to their dense clustering or to their intimate associationwith a very high density of capillaries(see Zambrano Quantitative comparisons and de Robertis, 1966) the somata of magnocellular neurons In untreated animals, the mean cross-sectionalarea of descend- are often highly irregular in shape, with deep invaginations. ing somata is similar to that for the total neuroendocrine pop- Surprisingly, however, while their somatawere distinctly larger ulation. However, becausethe dendrites of descendingneurons than those of parvicellular neurosecretory neurons, their den- are more highly branched, their total length is significantly great- drites and accompanying spinesappeared remarkably similar. er than that for the neuroendocrinecells. If the descendingpop- Thesedendrites appear to extend throughout the posterior mag- ulation is differentiated into dorsal and ventral groups,the mean nocellular part of the PVH and, in many cases,extend medially somal area and the total dendritic length of neurons in the beyond its cytoarchitectonic boundaries to almost reach the ventral medial group is larger than that for neuronsin the dorsal third ventricle. In general, however, the dendrites of theseneu- group. Similarly, if the total neuroendocrinepopulation (Fig. 4) rons stay within the borders of the posterior magnocellularpart is separatedinto parvicellular and magnocellular groups, the of the PVH. The medially directed dendrites tend to course mean somal area of magnocellularneurons is larger. However, through the medial (oxytocinergic) subdivision of the posterior despite this size difference, the total dendritic length and the magnocellularpart of the PVH, although they are drawn in the spine density of the 2 neuroendocrine groups were not signifi- ventral medial parvicellular part of the nucleus in Figure 4 for cantly different. The Journal of Neuroscience, April 1989, 9(4) 1379

Figure 3. Representative sample of LY-filled fluorescent cells that had been retrogradely labeled with Fast blue after the tracer was injected intravenously. Both magnocellular and parvicellular neurosecretory neurons are illustrated, and were taken from brain tissue that had not been pretreated with colchicine or subjected to immunohistochemical staining (compare with Figs. 4 and 5). Abbreviations: dp, dorsal parvicellular; mpv, ventral medial parvicellular; pv, periventricular.

After treatment with colchicine for 72 hr, the mean somal bulbous spines;however, after treatment with colchicine, nu- area of both the descending (Fig. 2) and the neuroendocrine merous hair-like spines(more typical of neuroendocrine cells) (Fig. 5) populations increased significantly (320/o), although the becameapparent on their somataand proximal dendrites. The increase was larger in the descending group (52%) than in the density of appendages,as measuredby taking the ratio of the neuroendocrine group (2 lo/o).Even though the dorsal descending number of appendagesto the total dendritic length, also in- neurons were distinguishable from the ventral descending neu- creased dramatically, from an average density of 1.7 to 12.9 rons in both mean somal area and total dendritic length in spines/100 Wm.The density of spineson dorsal descendingneu- untreated animals, after treatment with colchicine, both groups rons was not significantly different from that for ventral medial demonstrated similar mean somal areas and total dendritic descendingneurons in untreated animals, and, after colchicine lengths. After treatment with colchicine, the mean somal area treatment, both types were affected to a similar extent. The of both the parvicellular and magnocellular neuroendocrine density of spineson neuroendocrine cells was also changedby groups also increased significantly; here, the magnocellular neu- colchicine treatment, increasing from an averageof 8.7 to 19.3 rons showed an average increase of 49%, whereas the parvicel- spines/100pm. Before treatment with colchicine, the average lular neurons showed a 39% increase in mean somal area. The spine densitiesof magnocellular and parvicellular neurosecre- mean total dendritic length of parvicellular neurons was indis- tory neurons were similar, and after colchicine, the density of tinguishable from that for magnocellular neurons in untreated spinesincreased somewhat more for the magnocellularsubpop- animals, and after colchicine treatment, they were both reduced ulation. to a similar extent. Treatment with colchicine also appeared to influence the Discussion number and types of appendages on descending and neuroen- The technique usedhere has allowed us to characterize the total docrine cells in quite selective ways (Figs. 6-8). In untreated dendritic field of large populations of specific cell types within animals, descending neurons have a sparse distribution of small a cytoarchitectonically defined nucleus, as well as to provide 1380 Rho and Swanson - Morphometrics of Paraventricular Neurons

dp

Figure 4. As in Figure 3, drawings of representative neurosecretory neurons in the PVH are illustrated, except that here the animals were pretreated with colchicine. Abbreviations as in Fig- ure 3. quantitative information about the morphological features of ventral medial descending groups studied here. In the end, how- individual cell types within these populations. There are 3 major ever, the method described here cannot provide definitive evi- advantages to the approach used here. First, cells that have been dence about the total extent of dendrites and the axon of in- retrogradely labeled from a specific terminal field can be filled dividual neurons; only estimates based on populations can be with LY under visual inspection in a single tissue section, which provided. The full morphology of individual cells can be pro- can then be subjected to immunostaining for additional infor- vided only by the Golgi method or by filling cells in intact mation about filled cells. Second, the method is convenient for animals and subsequently reconstructing them in serial sections. gathering information about large populations, since many cells With these approaches, however, it is difficult to characterize can be filled in each tissue section. And third, quantitative mor- cells immunohistochemically or to sample large numbers of phological comparisons can be made relatively easily between cells. control and experimental conditions for a particular cell type. It is well established that most neurons in the PVH with The major disadvantage of the method is that dendrites and descending projections to the brain stem and spinal cord are axons may be cut off because the filled cells lie within a lOO- found in the dorsal, ventral medial, and lateral parvicellular pm-thick section. For a particular plane of section, this problem parts of the nucleus (Hosoya and Matsushita, 1979; Swanson can be approached in a statistical way, since many cells can be and Kuypers, 1980). Van Den Pol (1982) has illustrated the filled and many apparently intact dendrites can be observed. appearance of Golgi-impregnated cells in the lateral parvicel- However, a more complete characterization of a population of lular part of the nucleus, and while it cannot be stated with cells would require its analysis in the 3 standard planes of sec- certainty which of these cells gives rise to descending projec- tion. This is particularly relevant when attempting to determine tions, the morphology of several of them is quite similar to the significance of morphological differences between topo- those we have filled in the dorsal and ventral medial parvicel- graphically separate populations of cells, such as the dorsal and lular parts of the nucleus. Our analysis of retrogradely labeled The Journal of Neuroscience, April 1989, 9(4) 1381

Figure 5. This drawing illustrates the morphology of CRH-immunostained neurons (red) in the dorsal medial parvicellular part of the PVH that had been retrogradely labeled from the median eminence and several days later treated with colchicine. For comparison, neurons that project to the spinal cord are drawn in black. Inset, Photomicrograph showing the appearance of retrograde labeling in the PVH following a fluorogold injection in the spinal cord and an intravenous injection of Fast blue. x 50. 1382 Rho and Swanson * Morphometrics of Paraventricular Neurons

0 100 200 300

microns

Figure 6. These drawings show in more detail the morphology of individual neurons in the PVH that project to the spinal cord. u and C, Cells in the ventral medial parvicellular part of the nucleus; no colchicine. b and d, Cells in the ventral medial parvicellular part; colchicine pretreatment. e and g, Dorsal parvicellular part; no colchicine. fand h, Dorsal parvicellular part; colchicine pretreatment neurons in these 2 regions indicates that a majority of their We did not examine the dendritic fields of magnocellular dendrites lie within the cytoarchitectonically defined borders of neuroendocrine neurons in the lateral posterior magnocellular these regions. However, it is also clear that their secondary part of the PVH in as much detail as the populations of de- dendrites extend to form a halo that is on the order of 20-50 scending and CRH neuroendocrine cells. However, it is clear pm wide around each region and that the dendrites of some that some dendrites of magnocellular neurosecretory neurons neurons in one region extend through the area (the dorsal medial extend beyond the limits of the lateral posterior magnocellular parvicellular part) between them to reach the other region. One part of the PVH, particularly medially, where many of them such neuron in what is probably the dorsal parvicellular part of course in or near the medial area (predominantly oxytocinergic) the nucleus was illustrated by Van Den Pol (1982, his figure of the posterior magnocellular part of the nucleus toward the 4B), although he identified the cell as “magnocellular,” appar- third ventricle. In addition, many dendrites curve to stay within ently on the basis of its size alone. the lateral borders of the cell group. This general pattern con- These results indicate that a majority, but certainly not all, firms earlier work in the rat PVH by Armstrong et al. (1980) of the neural inputs to descending neurons in the dorsal and and Van Den Pol (1982), and the evidence suggests that the ventral medial groups form within the cytoarchitec- dendrites of magnocellular neurons also stay largely within the tonic boundaries of these groups and within the zone imme- limits of the supraoptic nucleus (Leontovich 1969/ 1970; Arm- diately surrounding them. A similar arrangement was found for strong et al., 1982; Dyball and Kemplay, 1982; Randle et al., parvicellular CRH neuroendocrine cells in the dorsal medial 1986). parvicellular part of the PVH: most of their dendrites stay within It should also be pointed out that the dendrites of neurons in (and course parallel to the long axis of) this region, although the relatively cell-sparse zone surrounding the PVH tend to run some clearly extend into adjacent areas. parallel to the border of the nucleus, which they rarely cross The Journal of Neuroscience, April 1989, 9(4) 1383

300

microns Figure 7. Drawings of individual neuroendocrine neurons in the PVH. a, c, and e, Cells in the dorsal medial parvicellular part of the nucleus; no colchicine. b and d, CRH-immunostained neurons in the dorsal medial parvicellular part of the nucleus; colchicine pretreatment. A CRH-immu- nonegative cells in the dorsal medial parvicellular part of the PVH; colchicine. g and i, Large neurons in the lateral subdivision of the posterior magnocellular part of the PVH; no colchicine. h and j, Magnocellular neurosecretory neurons in the same region as in g and i; colchicine pretreatment. All of the neurons shown here were retrogradely labeled following intravenous injections of Fast blue.

(seefigure 2 in Van Den Pol, 1982). The view that emergesfrom neuronsin the posterior magnocellularand dorsal medial parvi- the evidence taken as a whole is that the dendrites of neurons cellular parts of the nucleus are neurosecretory(see Sherlock et in the PVH show a strong tendency to stay within the nucleus al., 1975; Armstrong et al., 1980; Swansonet al., 1987). How- itself, while the distal parts of dendrites from cells within a ever, a complete understandingof the synaptology of PVH cir- specificpart ofthe nucleusform a relatively thin halo in adjacent cuitry will require ultrastructural evidence that includes a regions of the nucleus. In addition, the dendrites of some de- knowledge of the origin and neurotransmitter content of the scendingand magnocellularneurosecretory neurons extend well presynaptic element, as well as the projections and neurotrans- into the medial parvicellular part of the nucleus, and even into mitter content of the postsynaptic element. Furthermore, it will its periventricular zone. be important to determine the extent to which individual axons The method usedhere is particularly suited for characterizing of known origin innervate one or more cell types within the the total dendritic field of a defined population of neurons, an PVH, as this has a direct bearing on mechanismsunderlying important considerationwhen mapping the distribution of neu- the coordination of responsesmediated by the nucleus. The ral inputs. Earlier work along theselines had to rely solely upon results of the present study support the view that a projection cytoarchitectonic boundaries,and the generalizationthat emerged ending heavily and selectively within one of the parts of the was that most projections end within specificparts of the PVH PVH examined here preferentially innervates neurons in that (seeSawchenko and Swanson, 1982a; Swansonet al., 1986, for part of the PVH. However, no part of the PVH contains a reviews). This approach doesnot take into account the orien- homogenouspopulation of neurons,and similarly the dendrites tation and extent of dendrites, or the fact that each part of the of neurons in a particular part of the nucleus are not strictly PVH harbors a mixture of cell types, even though one class confined to that part. usually predominates.For example, at least 88% of the neurons The morphology of individual cell types in the PVH has only in the dorsal parvicellular part of the PVH project to the spinal beenclarified in the last severalyears. Early workers commented cord (Sawchenkoand Swanson, 198l), and the vast majority of on the difficulty of impregnating what we now know are mag- 1384 Rho and Swanson * Morphometrics of Paraventricular Neurons

Figure 8. Photomicrographs to illustrate the appearance of dendritic appendages on various cell types in the VPH. A and B, Untreated (A) and colchicine-treated (B) descending neurons in the dorsal parvicellular part. C and D, Untreated (C) and colchicine-treated (0) parvicellular neu- rosecretory neurons in the dorsal medial parvicellular part. E and F, Untreated (IT) and colchicine-treated Q magnocellular neurosecretory neurons in the lateral posterior magnocellular part. G and H, LY-filled (G) and CRH-immunostained (If) parvicellular neurosecretory neurons in the dorsal medial parvicellular part. All micrographs x 1250. The Journal of Neuroscience, April 1989, 9(4) 1355 nocellular neurosecretory cells (Cajal, 1894; Szentagothai et al., POPULATIONS SUBPOPULATIONS 1968), although several large neurons with 2 or 3 simple den- ““treated Colchicine “fltk%t~d colchicine , drites in the supraoptic (Gurdjian, 1927; Krieg, 1932) and PVH (Krieg, 1932) of the rat were illustrated rather early on, and Lefranc and Courtier (1966) demonstrated similar neurons with sparse dendritic spines in the kitten and neonatal guinea pig. The first detailed study of this problem was published by Lcon- tovich (1969/ 1970), who showed that most neurons in the puppy supraoptic nucleus are bipolar with short “sticklike” spines on the dcndritcs and somata. The dendrites are simple and may branch once, and they generally stay within the borders of the nucleus. Similar neurons were observed in the PVH, although their dendrites were said to ramify somewhat more broadly. This general description of neurons in the supraoptic nucleus has since been confirmed with the Golgi method in the adult rat (Ifft and McCarthy, 1974; Armstrong et al., 1982; Dyball and Kemplay, 1982), monkey (LuQui and Fox, 1976), and rab- bit (Felten and Cashner, 1979), and with the intracellular LY filling ofneurons in rat hypothalamic slices (Andrew et al., 198 1) or explants (Randle et al., 1986). Large cells in the PVH were also described in Golgi studies of the mouse (Barry, 1975), and Van Den Pol (1982) showed that such neurons in the rat display somatic and dendritic spines. This description is similar to that for LY-filled neurons that project to the posterior pituitary in the cat (Reaves and Hay- ward, 198 1) and to vasopressin- and -stained neurons in the rat (Sofroniew and Glasmann, 198 1; Kayser et al., 1982) although the latter authors did not comment on the presence of spines. Ultrastructural studies in the mouse and rat supraoptic nu- cleus have confirmed the existence of synapses on somatic and 260 dendritic spines, as well as on the soma itself and on the dendritic shafts (Polenov and Senchik, 1966; Enestrom, 1967; Rechardt, 200 1969; Leranth et al., 1971, 1975; Ifft and McCarthy, 1974). In i the rat, LCranth et al. (1975) have even estimated that, on av- erage, each magnocellular neurosecretory neuron receives on the order of 600 synapses, with about 500 equally distributed on dendritic shafts and spines, and the rest on the soma and axon. In the rat PVH, synapses have also been reported on the soma and dendrites, on somatic and dendritic spines, and on the axon of magnocellular neurosecretory neurons (Van Den Pol, 1982; Piekut, 1983; Silverman et al., 1983). Thus, our rather limited examination of magnocellular neurosecretory neurons is fully consistent with the substantial earlier literature on such neurons in the PVH and supraoptic nucleus in a variety of mammalian species. The issue of the extent to which the axon of magnocellular neurosecretory neurons gives rise to collaterals is somewhat Figure 9. Quantitative analyses of the effects of colchicine on neuron controversial. There appears to be general agreement (see Ene- morphology in the PVH. With computer graphics, the mean (*SEM) Strom, 1967; Leontovich, 1969/1970; Mason et al., 1984; Hat- somal cross-sectional area, total dendritic length (within the 100~urn- ton et al., 1985; Randle et al., 1986) that collaterals do not arise, thick tissue section), and spine density of individual cells were deter- mined, and comparisons between different populations and subpopu- or are very infrequent, within the PVH and supraoptic nuclei lations were made. The total neuroendocrine population (left) included themselves, which is in agreement with our results. On the other all retrogradely labeled neurons in the PVH following an intravenous hand, at least some terminal boutons (Randle et al., 1986) or injection of Fast blue; parvicellular (right) refers just to cells in the dorsal longer collaterals (Mason et al., 1984; Hatton et al., 1985) appear medial parvicellular part, and magnocellular refers just to cells in the lateral posterior magnocellular part of the PVH. The numbers (N) of to arise outside the nuclei, in the lateral hypothalamic area, cells in each group are listed below each column. before the parent axon enters the . We observed that it was sometimes difficult to identify unequivocally the axon of neurosecretory cells in our material, particularly when it arose reticulum and ribosomes (Enestrom, 1967). Nevertheless, the from a primary dendrite. It has been noted that the axon hillock anatomical evidence as a whole does not support the existence of magnocellular neurosecretory neurons can be quite large (e.g., of recurrent collaterals for magnocellular neurosecretory neu- Carrato Ibafiez, 1952) and often contains rough endoplasmic rons, and it has been suggested that electrophysiological evi- 1386 Rho and Swanson - Morphometrics of Paraventricular Neurons dence for recurrent inhibition may have been misinterpreted, pendagesvary widely in morphology, from large,complex struc- since these cells often show a 30-100 msec silent period follow- tures, to typical bulbous spines, to thin, hair-like processes. ing the generation of an (see Leng, 198 1). While many of these processesmay be dendritic or somatic Much less is known about the morphology of parvicellular spines,it is possible that at least some of them arc cilia. Thcsc neurosecretory neurons in the PVH. Van Den Po! (I 982) has structures~~~~ -~~--Ll. which may be 0.2-X.3 pm in diameter and up to S described G&i-impregnated neurons similar in appearance ta pm long, were first &scribed in magnncellular ncurosecretory those illustrated here, although he could not further classify cells of the goldfish (Palay, 196l)* and subsequentlyin neonatal them. This is a problem becausethe PVH contains neurons (Lafarga et al., 1980), mature (En&r&n, 1967), and aged (%a= prcojecting to the median eminence that contain CRH (Bloom rez et a!., 1985) rnagnoccllular neurosecretory cells ofrnanmials. et a!., ! 987; Antoni ct al., ! 9X3), TRH (Aizawa and C;rccr, 198 1; It has been known for some time that colchicinc and rclatcd Brownstcin ct al., 19821,somatastatin (Sawchenko and Swan- drugs increase the number of cilia on a variety of cell types, son, 1982b; lshikawa et al., 19X7), GRH (Sawchenko et al., including fibroblasts(Stubbleheld and Brinkley, 1966) and brain 1985; Bruhn et al., 1987), and dopamine (Swansonet al., 1981; glial cells (Milhaud and Pappas, 1968). Interestingly, it has been Kawano and Daikoku, 1987). Liposits et al. (1983, 1985) ap- reported that increasedactivity of the magnocellularneurose- proached this problem using immunohistochemical staining at cretory system stimulates proliferation of pituicytes in the pos- the ultrastructural level and demonstratedthat CRH cellsin the terior lobe (Selye and Hall, 1943), and that this is accqmpanied dorsal medial parvicellular part of the PVH display somaticand by an increased number of cilia, which is accentuated by the dendritic spinesand that synapsesare formed with these ap- administration of microtubule poisons such as colchicine (see pendages,as well as directly on the somatic membrane.Ltrdnth Dustin et al., 1979). Although it is not clear whether these changes et al. (1983) have also presentedultrastructural evidence that are due to the well-known disruption of microtubules by col- as the axons of magnocellularneurosecretory cells passthrough chicine, the results suggestthat the drug may influence an in- the dorsal medial parvicellular part of the PVH, they may form trinsic mechanism that regulates the number, morphology, and synapses with the somata of CRH neurons. Our results confirm location of somaticand dendritic appendages.There is evidence the presenceof relatively infrequent somaticand dendritic spines to suggest that during early stages of development, many neu- on parvicellular neurosecretory CRH neurons. They also indi- rons display far more spines, of varying shapes and sizes, than cate not only that the cells are smaller than magnocellularneu- in the fully differentiated state (seeMorest, 1969). rosecretory cells,but also that they tend to have a more elaborate dendritic tree, although the density of spinesin the 2 classesof neurons is similar. It has not been possible to follow the axons References of these CRH neurons for any great distance in our lOO-pm- Aizawa, T., and M. A. Greer (198 1) Delineation of the hypothalamic thick sections. However, as with magnocellular neurosecretory area controlling thyrotropin secretion in the rat. Endocrinology 109: neurons, axon collaterals and terminal boutons appear to be 1731-1737. Andrew, R. D., B. A. MacVicar, F. E. Dudek, and G. I. Hatton (198 1) very infrequent within the PVH itself, although they have been Dye transfer through gap iunctions between neuroendocrine cells of detected within the zone immediately surrounding the nucleus rat hypothalamus. &i&e 211: 1187-l 189. (Rho and Swanson, 1987). The experimental approach usedin Antoni. F. A.. M. Palkovits. G. B. Makara. E. A. Linton. P. J. Lowrv. the present work to identify parvicellular neurosecretory CRH and J. 2. K&,s (1983) Immunoreactive corticotropin-releasing ho;: mone in the hypothalamoinfundibular tract. 36: cells is important becausesome neurons in the PVH that express 415-423. CRH are magnocellularneurosecretory neurons, which also ex- Armstrong, W. E., S. Warach, G. I. Hatton, and T. H. McNeil1 (1980) press oxytocin (Sawchenko et al., 1984) or are neurons with Subnuclei in the rat hypothalamic paraventricular nucleus: A cy- descendingprojections (Swansonet al., 1986). toarchitectural, horseradish peroxidase and immunocytochemical This is the first report on the detailed morphology of descend- analysis. Neuroscience 5: 193 l-l 958. Armstrong,W. E., J. SchGler,and T. H. McNeil1 (1982) Immunocy- ing neurons in the PVH, although Golgi-impregnated cells with tochemical, Golgi and electron microscopic characterization of pu- a similar morphology and location were describedby Van Den tative dendrites in the ventral glial lamina of the rat supraoptic nu- Pol(1982). Interestingly, thesecells have longer dendrites, and cleus. Neuroscience 7: 679-694. considerably fewer spines,than neurosecretorycells in the PVH, Barry, J. (1975) Essai de classification, en technique de Golgi, des diverses cattgories de neurones du noyau paraventriculaire chez la although their somata are, on the average, intermediate in size Souris. C.R. Seances Sot. Biol. 169: 978-980. betweenthe parvicellular and magnocellularneurosecretory cells. Bloom, F. E., E. L. F. Battenberg, J. Rivier, and W. Vale (1982) Cor- It is worth pointing out that neuronsin the ventral descending ticotropin releasing factor (CRF): Immunoreactive neurones and fi- group tend to be larger than those in the dorsal group, when bers in rat hypothalamus. Regulat. Peptides 4: 43-48. viewed in the frontal plane.This is of interestin view of evidence Brownstein, M. J., R. L. Eskay, and M. Palkovits (1982) Thyrotropin releasing hormone in the median eminence is in processes of paraven- that neurons in the ventral group project densely to both the tricular nucleus neurons. 2: 197-20 1. dorsal vagal complex and the spinal cord, whereas neurons in Bruhn, T. O., E. L. P. Anthony, P. Wu, and I. M. D. Jackson (1987) the dorsal group project heavily to the spinal cord but much GRF immunoreactive neurons in the paraventricular nucleus of the lessheavily to the dorsal medulla (Swansonand Kuypers, 1980). rat: An immunohistochemicalstudy with monoclonaland polyclonal However, an analysisof thesetwo cell groups in all 3 planesof antibodies. Brain Res. 424: 290-298. Carrato IbBiiez, A. (1952) Aportaciones a la citologia de1 sistema hi- section will be necessary to determine whether the differences potalamo-hipofisario. Trab. Inst. Cajal Invest. Biol. 44: 159-2 10. observed are due to changesin cell orientation or morphology. Dustin, P., A. Anjo, and J. Flament-Durand (1979) Centriologenesis Finally, the present results indicate that colchicine treatment and ciliogenesis in rat pituicytes. Biol. Cellulaire 34: 227-236. leads to the swelling of neuronal somata, the swelling and short- Dvball. R. E.. and S. K. Kemnlav (19821 Dendritic trees of neurones ;n thk rat supraoptic nucleus. Ne&osciknce 7: 223-230. ening of dendrites, and an increasein the number and length of EnestrGm, S. (1967) Nucleus supraopticus. A morphological and ex- somatic and dendritic appendages;and that the extent of these perimental study in the rat. Acta Pathol. Microbial. Stand. 186: l- changesdepends upon cell type. The dendritic and somaticap- 101. The Journal of Neuroscience, April 1989, 9(4) 1387

Felten, D. L., and K. A. Cashner (1979) Cytoarchitecture of the su- Polenov, A. L., and J. I. Senchik (1966) Synapses on neurosecretory praoptic nucleus. Neuroendocrinology 29: 221-230. cells of the supra-optic nucleus in white mice. Nature 211: 1423- Gurdjian, E. S. (1927) The of the albino rat. J. Comp. 1424. Neurol. 43: l-l 14. Ramon y Cajal, S. (1894) Algunas contribuciones al concimento de Hatton, G. I., P. Cobbett, and A. K. Salm (1985) Extranuclear axon 10s ganglios de1 encefalo. III. Hipofisis. An. Sot. Esp. Hist. Rat. 2 Ser. collaterals ofparaventricu!ar neurons in the rat hypothalamus: Intra- 23: 195-233, cellular afaininy, immunocyfachemisfry and e!ectrophysialagy. Brain Randle, J. C. R,, C. W, Bourque, and L. P. Renaud (1986) Serial Rrs. Bull. 14: 123-132. reconstruction of lucifer yrllo~w=labeled supraoptic nucieus neuf~ns Hosoya, Y., and M. Matsushita (lY7Y) Tdcntification and distribution in ectfuscd rat hvaathalamic cxalants. Neuroscience 67: 453=46f. of the spinal and hypophysial projection neurons in the paraventric- Keav&, T. A., Jr., &id J. N. Hayward (lYX1) Dye=marked paraven= ular nucicus of the rat. A light and cicctran microscopic study with tricuiar ncurocndacrinc cciisin vivn in cat hypothalamus. Neurosci. the horseradish peroxidaxc method. Exp. Brain Res. 35: 315-331. bat. 25: 263-267. Ifft, J. D., and L. McCarthy (1974) Somatic spines in the supraoptic Rechardt, L. (1 Y6Y) Ultrastructure ofthe supraoptic nucleus ofnormal nucleus of the rat hypothalamus. Cell Tissue Res. 148: 203-2 11. and dehydrated rats. Acta Physiol. Stand. 329: l-79. Ishikawa, K., Y. Taniguchi, K. Kurosumi, M. Suzuki, and M. Shinoda Rho, J.-H., and L. W. Swanson (1987) Neuroendocrine CRF moto- (1987) Immunohistochemical identification of somatostatin-con- neurons: Intrahypothalamic axon terminals shown with a new retro- taining neurons projecting to the median eminence of the rat. En- grade-Lucifer-immuno method. Brain Res. 436: 143-147. docrinology 121: 94-97. Sawchenko, P. E., and L. W. Swanson (198 1) A method for tracing Kawano, H., and S. Daikoku (1987) Functional topography of the rat biochemically defined pathways in the central nervous system using hypothalamic dopamine neuron systems: Retrograde tracing and im- combined fluorescence retrograde transport and immuncrhistochem- munohistochemical study. J. Comp. Neurol. 265: 242-253. ical techniques. Brain Res. 210: 3 l-5 1. Kayser, B. E. J., M. Mtihlethaler, and J. J. Dreifuss (1982) Paraven- Sawchenko, P. E., and L. W. Swanson (1982a) The organization of tricular neurones in the rat hypothalamic slice: Lucifer yellow injec- noradrenergic pathways from the to the paraventricular tion and immunocytochemical identification. Experientia 38: 39 l- and supraoptic nuclei in the rat. Brain Res. Rev. 4: 275-325. 393. Sawchenko, P. E., and L. W. Swanson (1982b) Immunohistochemical Krieg, W. J. S. (1932) The hypothalamus of the albino rat. J. Comp. identification of neurons in the paraventricular nucleus of the hy- Neurol. 55: 19-89. pothalamus that project to the medulla or to the spinal cord in the Lafarga, M., J.-P. Hervas, D. Crespo, and J. Villegas (1980) Ciliated rat. J. Comp. Neurol. 205: 260-272. neurons in supraoptic nucleus of rat hypothalamus during neonatal Sawchenko, P. E., L. W. Swanson, and W. W. Vale (1984) Cortico- period. Anat. Embryol. 160: 29-3X. -- tropin releasing factor: Co-expression within distinct subsets of oxyto- Lefranc. M. G.. and M. R. Courrier (1966) Etude neurohistoloaiaue tin-, vasopressin-, and neurotensin-immunoreactive neurons in the des nbyaux supraoptique et paraventriculaire chez le Cobayekt le hypothalamus of the male rat. Neuroscience 4: 111 X-l 129. Chat par la technique de triple impregnation de Golgi. CR. Acad. Sawchenko, P. E., L. W. Swanson, J. Rivier, and W. W. Vale (1985) Sci. Paris 263: 976-97X. The distribution of growth hormone-releasing factor (GRF)-immu- Leng, G. (198 1) The effects of neural stalk stimulation upon firing noreactivity in the central nervous system of the rat: An immunohis- pattern in rat supraoptic neurones. Exp. Brain Res. 41: 135-145. tochemical study using antisera directed against rat hypothalamic Leontovich, T. A. (1969/ 1970) The neurons of the magnocellular neu- GRF. J. Comp. Neurol. 237: 100-l 15. rosecretory nuclei of the dog’s hypothalamus. A Golgi study. J. Him- Selye, H., and C. E. Hall (1943) Further studies concerning the action forsch. 11: 499-5 17. of sodium chloride on the pituitary. Anat. Rec. 86: 579-583. Leranth, C., L. Zaborszky, G. Makara, and M. Palkovits (1971) De- Sherlock, D. A., P. M. Field, and G. Raisman (1975) Retrograde generation von Synapsen im Nucleus supraopticus nach Unterbre- transport of horseradish peroxidase in the magnocellular neurosecre- chung der afferenten Systeme des Hypothalamus. 66. Vers Anat. Ges. tory system of the rat. Brain Res. 88: 403-414. (Zagreb) Anat. Anz. Erg. N. Bd. l?& 5X7-593. Silverman, A. J., A. Hou-You, and E. A. Zimmerman (1983) Ultra- Leranth, C., L. Zaborszky, J. Marton, and M. Palkovits (1975) Quan- structural studies of vasopressin neurons of the paraventricular nu- titative studies on the supraoptic nucleus in the rat. I. Synaptic or- cleus of the hypothalamus using a monoclonal antibody to vaso- ganization. Exp. Brain Res. 22: 509-523. pressin: Analysis of synaptic input. Neuroscience 9: 141-l 55. Ltranth. C.. F. A. Antoni. and M. Palkovits (1983) Ultrastructural Sofroniew, M. V., and W. Glasmann (1981) Golgi-like immunoper- demonstration of ovine CRF-like immunoreactivity(oCRF-LI) in the oxidase staining of hypothalamic magnocellular neurons that contain rat hypothalamus: Processes of magnocellular neurons establish mem- vasopressin, oxytocin or neurophysin in the rat. Neuroscience 6: 6 19- brane specializations with parvocellular neurons containing oCRF- 643. LI. Regulat. Peptides 6: 179-18X. Stubblefield, E., and B. R. Brinkley (1966) Cilia formation in Chinese Liposits, Z., T. GBrcs, G. S&lo, I. Lengv&ri, B. Flerko, S. Vigh, and hamster fibroblasts in vitro as a response to colcemid treatment. J. A. V. Schally (1983) Ultrastructural characteristics of immunola- Cell Biol. 30: 645-652. belled, corticotropin releasing factor (CRF)-synthesizing neurons in Suarez, I., B. Fernandez, M. A. Perez-Batista, and I. Azcoitia (1985) the rat brain. Cell Tissue Res. 229: 191-196. Ciliated neurons in the paraventricular nuclei in old hamsters. J. Liposits, Z., W. K. Paull, G. Set&lo, and S. Vigh (1985) Evidence for Submicrosc. Cytol. 17: 351-356. local corticotropin releasing factor (CRF)-immunoreactive neuronal Swanson, L. W., and H. G. J. M. Kuypers (1980) The paraventricular circuits in the paraventricular nucleus of the rat hypothalamus. His- nucleus of the hypothalamus: Cytoarchitectonic subdivisions and the tochemistry 83: 5-l 6. organization of projections to the pituitary, dorsal vagal complex and LuQui, I. J., and C. A. Fox (1976) The supraoptic nucleus and the spinal cord as demonstrated by retrograde fluorescence double-la- supraopticoneurohypophyseal tract in the monkey (Mucucu muluttu). beling. methods. J. Comu. Neurol. 194: 555-570. J. Comp. Neurol. 168: 7-40. Swanson, L. W., and P. E.. Sawchenko (1983) Hypothalamic integra- Mason, W. T., Y. W. Ho, and G. I. Hatton (1984) Axon collaterals tion: Organization of the paraventricular and supraoptic nuclei. Annu. of supraoptic neurones: Anatomical and electrophysiological evidence Rev. Neurosci. 6: 275-325. for their existence in the . Neuroscience 11: 169- Swanson, L. W., P. E. Sawchenko, A. BCrod, B. K. Hartman, K. B. 182. Helle, and D. E. Van Orden (198 1) An immunohistochemical study Milhaud, M., and G. D. Pappas (196X) Cilia formation in the adult of the organization of catecholaminergic cells and terminal fields in cat brain after pargyline treatment. J. Cell Biol. 37: 599-609. the paraventricular and supraoptic nuclei of the hypothalamus. J. Morest, D. K. (1969) The growth of dendrites in the mammalian brain. Comp. Neurol. 196: 271-285. Z. Anat. Entwickl. (Gesch.) 128: 290-317. Swanson, L. W., P. E. Sawchenko, J. Rivier, and W. W. Vale (1983) Palay, S. L. (196 1) Structural peculiarities of the neurosecretory cells Organization of ovine corticotropin-releasing factor immunoreactive in the preoptic nucleus of the goldfish, Carassius auratus. Anat. Rec. cells and fibers in the rat brain: An immunohistochemical study. 139: 262. Neuroendocrinology 36: 16 l-l 78. Piekut, D. T. (1983) Ultrastructural characteristics of vasopressin- Swanson, L. W., P. E. Sawchenko, and R. W. Lind (1986) Regulation containing neurons in the paraventricular nucleus of the hypothala- of multiple peptides in CRF parvocellular neurosecretory neurons: mus. Cell Tissue Res. 234: 125-134. Implications for the stress response. Prog. Brain Res., 68: 169-l 90. 1388 Rho and Swanson l Morohometrics of Paraventricular Neurons

Swanson. L. W., P. E, Sawchenko, R. W. Lind, and J.-H. Rho (1987) Van Den Pal, A. N. (1982) The magnocellular and parvoce!!u!ar para- The CRf motoneuron: Differential regulation in neurons with ventricular nucleus of the rat: Intrinsic organization. J. Camp. Neural. possible synaptic, paracrine, and endocrine outputs. Ann. NY Acad. 206: 3 1T-345. Sci. 512: 12=23. Zambrano, D.. and E. de Robertis (1966) The secretory cycle of su- Szenthgothai, J., B. Flerk6, B. Mess, and B. Hal&z (1968) Hypotha= praoptic’ n&rons in the rat. A st;uct&al=functional cbrr~lation. Z. lamic Control of tha Anterior Pituitary. An Ex~~ri~~nt~l~~~r~h~- Zellforsch. 73: 4 1443 1. logical Study. Akademiai Kiad6, Budapest.