The Efferent Connect ions of the Suprachiasmatic Nucleus of the Hypothalamus L. W. SWANSON AND w. M. COWAN Departments of Biology and A n a t o m y , W a s h i n g t o n University, St. Louis, Missouri 631 10
ABSTRACT The efferent connections of the suprachiasmatic nucleus of the hypothalamus have been studied in the rat by the injection of ”-proline into the nucleus and the surrounding regions of the rostral hypothalamus, and by the injection of the enzyme marker, horseradish peroxidase, into the region of the ventromedial hypothalamic nucleus. After an injection of 3H-proline confined to the ventral portion of the suprachiasmatic nucleus, transported label can be followed, in the autoradiographs, dorsally and caudally in the periventricular area as far as the caudal end of the ventromedial nucleus, into the triangular area between this nucleus and the arcuate nucleus, and along the ventral aspect of the tuberal region, just lateral to the ventromedial nucleus. A small number of silver grains are also seen over the internal lamina of the median eminence. No label can be followed rostrally or immediately lateral to the nucleus. Comparable injections into adjoining regions of the hypothalamus (especially the anterior hypothalamic area, the medial preoptic area, and the retrochiasmatic region) show transported label over the same regions, but with a somewhat different pattern of grain distribution; in addition, the anterior hypothalamic area shows a n extensive projection through the medial forebrain bundle to the mammillary and supramammillary nuclei, the midbrain tegmentum, and certain of the midline thalamic nuclei. Although i t is difficult in our autoradiographs to distinguish between the course of the efferent fibers from the suprachiasmatic nucleus and the zones in which they terminate, our evidence favors a termination among the cells of the periventricular area, and upon dendrites of the cells in the ventromedial, dorsomedial and arcuate nuclei, which extend beyond the limits of the nuclei into the periventricular area and to the area beneath the ventromedial nucleus.
The discovery that the ganglion cells of the retina project bilaterally upon the suprachiasmatic nucleus of the hypothalamus in all mammals that have been studied (Moore and Lenn, ’72; Hendrickson et al., ’72; Moore, ’73; Tigges and OSteen, ’74) and that these nuclei also receive a “visual” input from the ventral nucleus of the lateral geniculate body (Swanson et al., ’74) has focused attention upon the suprachiasmatic nuclei as the probable mediators of a number of light induced behavioral and neuroendocrine activities (Stephen and Zucker, ’72a, ’72b; Hendrickson et al., ’72; Moore and Klein, ’74). While the physiological mechanisms underlying these various phenomena remain to be elucidated, it is generally assumed that the neuroendocrine effects are mediated by some pathJ. COMP. N E U R .160. 1-12
way between the suprachiasmatic nuclei and the hypophysiotrophic zone of the hypothalamus, and that those responsible for the various circadian rhythms are effected by pathways extending caudally through the lateral hypothalamus (in the medial forebrain bundle) to the brain stem and spinal cord. However, these views are based largely on indirect evidence, since at present virtually nothing is known with certainty about the efferent connections of the suprachiasmatic nuclei (Szentagothai et al., ’68; Nauta and Haymaker, ’69). From studies of normal silver impregnated material, a projection has been described into the “supraoptic system of fibers,” including 1 This work was supported in part by grants MH24604, NS-03777 and NS-10943 from the National Institutes of Health.
1
2
L. W . SWANSON AND W . M. COWAN
the so-called “fasciculus supraopticus” and the “f ascicu lu s supr achi a smatic u s” (Gurdjian, ’27; Loo, ’31; Bodian, ’40). Both Krieg (’32) and Szentagothai et al. (’68), who have examined Golgi preparations of the suprachiasmatic nucleus, report that the axons of the cells pass dorsally and caudally into the periventricular area; unfortunately they could not be traced for any significant distance so their termination remains uncertain. The only other evidence bearing on this issue, known to us, is the finding that some cells in the suprachiasmatic nucleus can be activated antidromically by electrical stimulation in the region of the median eminence (Makara et al., ’71). As part of a systematic study of the connections of the hypothalamus, using mainly the autoradiographic method for tracing pathways (Lasek et al., ’68; Cowan et al.,’72), we have recently succeeded in obtaining a number of small injections of tritium-labeled amino acids (principally %proline) in, and around, the suprachiasmatic nucleus in the brains of young adult rats. Since in one case the injection appears to be strictly confined to the suprachiasmatic nucleus (and more specifically to the ventral portion of the nucleus in which the majority of the visual afferents terminate - Hendrickson et al., ’72; Swanson et al., ’74), it seemed worth while placing our findings on record. Although the critical findings are based on only a single experiment, the results in this case seem quite unequivocal; and since it is unlikely that a more successful experiment of this kind could be performed, we feel justified in describing this case in detail.
ilton syringe. In each case the injection was made over a period of 10-20 minutes with the bevel of the needle directed towards the structure to be labeled. After a survival period of 48 hours the animals were perfused with 10% formalin, and the brains removed, dehydrated, embedded in paraffin and serially sectioned at 15 pm. From each brain one or more series of sections was mounted (at 150 pm intervals) and processed for autoradiography as described by Cowan et al. (‘72). After exposure times ranging from 2 4 weeks, the autoradiographs were developed in D19 at 18”C, fixed in Ektaflo and stained, through the emulsion, with thionin. In an attempt to identify neurons in the rostra1 hypothalamus which might project into the ventromedial and arcuate nuclei or into the median eminence, the brains of several rats were injected stereotaxically with a small amount (usually 30-50 nl) of a concentrated solution (300500 gmlpl) of horseradish peroxidase (Sigma, type VI) and processed by a modification of the Karnovsky method (’67), as described by LaVail et al. (‘73). RESULTS
Although we have several brains in which some, or most, of the cells in the suprachiasmatic nucleus have been labeled by the isotope injection, in only one case (experiment R 186) have we succeeded in getting an injection strictly confined to neurons in the nucleus. In this case the tip of the injection needle was directed into the optic chiasm at the level of the suprachiasmatic nucleus and since the bevel was directed medially towards the nucleus, there appears to MATERIALS AND METHODS have been virtually no diffusion of the The brains which will be described are isotope into the adjoining anterior hypodrawn from a large series of experiments thalamic or medial preoptic areas. A sigwith injections of either Wproline alone, nificant amount of the injected label did, or a mixture of :sH-lysine, :$H-leucineand however, extend into the caudal part of 3H-proline, into different regions of the the optic chiasm, and as a result the local hypothalamus. In the critical case (ex- glial cells are heavily labeled. periment R 186) and two other “control” Figures 1 and 2, which are photomicroexperiments which will be described, graphs of the labeled region of the hy0.4 pCi of Wproline [ L-2,3-:’H-proline, pothalamus in R 186 (at its maximum specific activity 24 Ci/Mmole, New Eng- extent), show that the heavily labeled land Nuclear) in a volume of about 20 nl, neurons in the suprachiasmatic nucleus was injected stereotaxically through a are largely confined to its ventrolateral beveled needle attached to a 1 Ham- sector. Measurements of the region con-
CONNECTIONS OF THE SUPRACHIASMATIC NUCLEUS Abbreviations
AHA, anterior hypothalamic area A&, cerebral aqueduct ARC, arcuate nucleus BST, bed nucleus of the stria terminalis CM, central medial nucleus CP, caudate-putamen DMH, dorsomedial nucleus LHA, lateral hypothalamic area ME, median eminence MeA, medial nucleus of the amygdala ML, lateral mammillary nucleus MM, medial mammillary nucleus MPO, medial preoptic area MR, mammillary recess NLOT, nucleus of the lateral olfactory tract PAG, periaqueductal gray PIR, piriform cortex PT, parataenial nucleus PVH, paraventricular nucleus PVT, periventricular nucleus (thalamus)
Re, nucleus reuniens RT, reticular nucleus (thalamus) SCh, suprachiasmatic nucleus SO, supraoptic nucleus SUM, supramammillary area VL, lateral ventricle VMH, ventromedial nucleus V3, third ventricle ZI, zona incerta cp, posterior commissure fr, fasciculus retroflexus fx, fornix ic, internal capsule ml, medial lemniscus mt, mammillothalamic tract och, optic chiasm ot, optic tract pc, cerebral peduncle sm, stria medullaris SOC,supraoptic commissures st, stria terminalis
Fig. 1 A photomicrograph of a n autoradiograph to show the site of the injection in experiment R 186. Note that the labeled zone is confined to the ventral aspect of the suprachiasmatic nucleus and to the glia in the optic chiasm (arrows). There is a small tissue defect just lateral to the suprachiasmatic nucleus due to the tip of the injection needle. This, like the following autoradiographs, was counterstained with thionin. Scale: 250 pm.
3
4
t.W. SWANSON AND W. M. COWAN
Fig. 2 A high-power photomicrograph of the injection site in the suprachiasmatic nucleus in R 186. Heavily labeled neurons are confined to the ventrolateral aspect of the nucleus (arrows). Scale: 100 pm.
taining labeled cells indicate that its dimensions are no more than 250 pm in the mediolateral dimension, 375 pm in the dorsoventral plane (including the full extent of the involvement of the glia in the optic chiasm), and that it extends for just over 500 pm in the rostro-caudal dimension (see also fig. 3 ) . Jmmediately lateral to the labeled region there is a small defect in the tissue due to the passage of the injection needle; while this appears to have limited the lateral spread of the injected isotope, it is important to point out that it may also have interrupted some of the efferent fibers from the suprachiasmatic nucleus, Although this seems unlikely from the pattern of distribution of the axonally transported label in this and other brains in which the nucleus was involved by the injection, it cannot be ignored. We have elsewhere reported an experiment in which a relatively small, but critically placed, lesion interrupted the commissural fibers to the dentate gyrus close to their origin from an injected region of the hippocampus, which completely blocked the transport of label to the dentate gyrus on the contralateral side (experiment R 12 in Gottlieb and Cowan, ’73). The distribution of the transported
label in this experiment is shown diagrammatically in figure 3. From this, several points deserving comment emerge. First, there is no evidence in this (or any of our other experiments in which the suprachiasmatic nucleus is involved) for a rostra1 projection from the nucleus. In R 186, all the silver grains outside the limits of the nucleus are found either over the periventricular area immediately medial, dorsal and caudal to the nucleus (fig. 4), or in a narrow band along the ventral aspect of the hypothalamus ventral and lateral to the ventromedial nucleus. Second, the outflow from the nuccleus appears to be strictly ipsilateral; at no point is there any indication of transported label on the opposite side of the third ventride. We wish to emphasize this in view of the finding that the “visual input” to the nucleus from the retina and the ventral lateral geniculate nucleus is bilateral (Hendrickson et al., ’72; Swanson et al., ’74) and the finding in one of the brains labeled with horseradish peroxidase which will be described below. Third, although the transported label extends into the capsular zone surrounding the ventromedial nucleus and up to the margins of the dorsomedial and arcuate nuclei, there are comparatively
CONNECTIONS OF THE SUPRACHIASMATIC NUCLEUS
m
5
6
L. W. SWANSON AND W. M. COWAN
Fig. 4 A photomicrograph to show the silver grains over the periventricular area at the level of the ventromedial nucleus in experiment R 186. This may be compared with the density of silver grains seen over the ventromedial nucleus illustrated in figure 7A. The third ventricle and ependyma are located along the extreme left of the photomicrograph. Scale: 25 pm.
few grains over the nuclei themselves. As we shall see, this does not preclude a direct connection between the cells in the ventral portion of the suprachiasmatic nucleus and the ventromedial, dorsomedia1 or arcuate nuclei since we know from Golgi studies that many of the cells in these nuclei have dendrites which extend medially beyond the nuclear boundaries into the periventricular area (especially in the region between the ventromedial and arcuate nuclei - Szentagothai et al., '68). However, our evidence suggests that there are probably few, if any, axonal projections from the suprachiasmatic nucleus to the perikarya or to the proximal dendrites of the cells in these nuclei. Fourth, the grains over the dorsal part of the periventricular area (near the upper end of the third ventricle) sharply define the medial and ventral borders of the caudal part of the periventricular nucleus. Again, it is impossible to rule out that some of the fibers from the suprachiasmatic nucleus contact processes of cells in the periventricular nucleus (since some of them are also said to have dendrites which extend medially into the periventricular area). However, from the pattern of grain distribution in this region we
are inclined to think that most of the label in the dorsal part of the periventricular area is in axons that are passing further caudally to the level of the dorsomedial nucleus. Fifth, because of its evident importance we have paid particular attention to the region of the median eminence. On the side of the injection there is a small, but unmistakable number of silver grains over the internal lamina of the eminence. The grain density over this region is substantially lower than that over, say, the neuropil between the arcuate and ventromedial nuclei, but its consistency from section to section suggests that it represents a real projection. The fibers to the median eminence appear to enter it from its dorsolateral margin beneath the caudal part of the arcuate nucleus; they then spread out over the internal lamina but again strictly on the side of the injection. Finally, some fibers from the suprachiasmatic nucleus appear to pass into the medial forebrain bundle since the grain density over the ventral portion of the lateral hypothalamic area is consistently above the background level, at least as far as the caudal end of the ventromedial nucleus. However, a much larger group of labeled fibers can be traced along the ventral aspect of the hypothalamus, at first ventral to, and further caudally, lateral to the ventromedial nucleus. The final destination of these fibers is difficult to ascertain: the number of grains over this region becomes progressively smaller as one proceeds caudally, until just behind the ventromedia1 nucleus it is reduced to background levels. From a number of other brains with injections in the vicinity of the suprachiasmatic nucleus, but which have not labeled the cells in the nucleus, we shall describe R 38 and only two experiments-rats R 166. In R 38 the area labeled by the injection is very similar in its dimensions to that in R 186, but is confined to the retrochiasmatic region immediately behind the suprachiasmatic nucleus (figs. 5, 6). As far as we can determine, none of the cells in the suprachiasmatic nucleus have been labeled by the injected isotope; we wish to emphasize this in view of the general similarity in the pattern of grain distribution in R 38 and
CONNECTIONS OF T H E SUPRACHIASMATIC NUCLEUS
0 0 M
a
7
8
L. W . SWANSON AND W. M. COWAN
R 186. Since this is adequately illustrated in figure 5, we shall not describe it in detail, except to point out that whereas after the injection of the suprachiasmatic nucleus the transported label is largely outside the limits of the ventromedial and arcuate nuclei, in R 38 the silver grains are more-or-less uniformly distributed across the extent of those nuclei (compare figs. 7A and 7B). Thus, whereas there may be some question about the relation between the suprachiasmatic nucleus and these other nuclei, there can be no doubt that the retrochiasmatic region projects directly to both of them. In experiment R 166 a substantially larger injection (0.6 pCi of :Wproline in 30 nl) was made into the medial preoptic and anterior hypothalamic areas adjacent to the suprachiasmatic nucleus (fig. 5). Again, we shall not describe in full the distribution of the transported label in
this case except to note that it is very extensive and involves a number of structures which do not receive an input from the suprachiasmatic nucleus, including the stria terminalis and its bed nucleus, certain of the midline thalamic nuclei, the mammillary and supramammillary nuclei and the periaqueductal gray (fig. 5). But what is of particular interest is that the outflow from the medial preoptic and anterior hypothalamic areas also completely overlaps that from the suprachiasmatic nucleus, and although there appears to be significantly more label transported to the dorsomedial and arcuate nuclei in R 166 than in R 186, the grain distribution around the ventral and medial aspects of the ventromedial nucleus in the two cases is rather similar. So far we have been unsuccessful in our attempts to retrogradely label cells in the suprachiasmatic nucleus after injec-
Fig. 6 A photomicrograph of the injection site in experiment R 38. Note that the labeled cells are confined to the retrochiasmatic area (arrows). In this case the injection volume was 20 nl. Scale: 250 p m .
C O N N E C T I O N S OF T H E SUPRACHIASMATIC NUCLEUS
9
Fig. 7 High-power photomicrographs to show the relative grain densities over the ventromedial nucleus following injections in the suprachiasmatic nucleus (A) and the retrochiasmatic area (B). Scale: 25 wm.
tions of horseradish peroxidase in, and around, the ventromedial and arcuate nuclei. In two cases in which the ventrolatera1 portion of the ventromedial nucleus was heavily labeled by the injection, a small number of cells containing the reaction product was seen in the suprachiasmatic nuclei o n both sides. The identity of these cells is uncertain; in several respects they resemble small vascular pericytes seen elsewhere in the vicinity of the injection, especially in and around the acuate nuclei. In view of the fact that they were found in approximately equal numbers in the suprachiasmatic nuclei of both sides (although the injection was strictly unilateral) and as they appear quite different from other neurons retrogradely labeled by the enzyme, we are inclined to think that these cells are non-neuronal and that their labeling is probably due to vascular (or perivascular) spread of the enzyme. A s we shall discuss later, our failure to label neurons in the suprachiasmatic nucleus with horseradish peroxidase is unfortunate, since a priori this approach could
offer so much for the study of hypothalamic connections. DISCUSSION
For the study of intrahypothalamic connections the autoradiographic method clearly offers several advantages over most other neuroanatomical techniques. A s we have discussed the general advantages of this approach elsewhere (Cowan et al., '72; Swanson et al., '74; Cowan and Cuenod, '75),here we need only draw attention to two features of the method which we believe to be especially apposite for studies of the hypothalamus. First and foremost is the fact that the injected isotope is only incorporated into protein for transport by nerve cell somata in or close to the site of the injection, and not by axons passing through the labeled area (Cowan et al., '72). This makes the autoradiographic method especially useful in regions such a s the hypothalamus in which large numbers of fibers either traverse, or are closely applied to, the nuclear groups whose connections one wishes to study. While the
10
L. W. SWANSON AND W . M. COWAN
involvement of fibers of passage can, to some extent, be circumvented in degeneration studies by the careful placement of control lesions, it is always difficult to completely rule out the inadvertent interruption of some unrelated fiber system (for a further discussion of this point, see Powell et al., '65). Secondly, in many systems the autoradiographic method appears to be appreciably more sensitive, and to offer better resolution than most of the available degeneration techniques. This has already been pointed out with regard to the demonstration of the afferent connections of the suprachiasmatic nucleus (Moore and Lenn, '72; Hendrickson et al., '72; Swanson et al., '74), but it has also been noted in a number of other studies (for example, Graybiel et al., '73; Hickey and Guillery, '74; Tigges and OSteen, '74). Although it is not yet known why certain pathways that are either difficult, or even impossible, to impregnate with silver methods can readily be demonstrated by the autoradiographic method, it seems possible that the diameters of the axons may be one critical factor. As so many of the fibers in the hypothalamus are extremely fine (many are unmyelinated, with diameters well below 1 pm), it seems not unlikely that they will only be clearly demonstrated by the most sensitive techniques. Unfortunately, the use of the autoradiographic method in regions like the hypothalamus is also not without its limitations. In particular, it is usually impossible to determine from the distribution of silver grains whether one is dealing with the labeling of fibers or axon terminals, not to mention possible en passage synapses. This is an especially serious limitation in regions like the periventricular area; in a light microscopic autoradiograph it is quite impossible to know whether the grains one sees are due to label within fibers passing through the area, or to label in presynaptic processes terminating upon the many dendrites which extend into this region from the neighboring hypothalamic nuclei. By the same token, even though we can be quite certain that few, if any, labeled fibers from the suprachiasmatic nucleus enter the ventromedial or arcuate nuclei, we cannot rule out the possibility that they may be in synaptic contact with neurons
in these two nuclei. Szentagothai et al. ('68) have specifically described cells in the ventromedial, dorsomedial and arcuate nuclei which have rather lengthy dendrites that extend beyond the capsule of the nucleus into the periventricular area (also Cajal, '66; Millhouse, '73). This is particularly significant in the present context since after an injection of the suprachiasmatic nucleus the greatest concentration of silver grains is found over the periventricular area, especially in the triangular zone between the ventromedial and arcuate nuclei. The second largest contingent of labeled fibers that one sees is over the ventral aspect of the hypothalamus, ventral to the ventromedial nucleus. Although this is quite far removed from the ventromedial nucleus itself, it is evident from the Golgi studies of Millhouse ('73) that a substantial proportion of the cells in the nucleus have dendrites that reach ventrally into this region, and could well be contacted by efferent fibers from the suprachiasmatic nucleus (see especially his fig. 2). A third difficulty commonly associated with the autoradiographic method is that the injected isotope may diffuse for some distance from its principal focus, and when dealing with relatively short axonal pathways it is often difficult to distinguish labeling due to axonal transport from the diffusional spread of the tracer. In our experiments this does not appear to have been a serious problem: in R 186, for example, there are fewer than 40 significantly labeled neuron somata lateral to the suprachiasmatic nucleus, and even within the nucleus only a fraction of the total cell population is heavily labeled (fig. 2). We are inclined to attribute this limited spread of the label to two factors: (i) the smallness of the injections (20-30 nl), and (ii) the relatively long period (10-20 minutes) over which they were made. With these considerations in mind we may summarize our findings on the projection of the suprachiasmatic nucleus as follows. Most of the axons leaving the nucleus appear to pass caudally and dorsally through the periventricular area where they are in close relation to the neurons which constitute the bed nucleus of the periventricular fiber system (sometimes,
CONNECTIONS OF THE SUPRACHIASMATIC NUCLEUS
referred to as the periventricular nucleus) and to the dendrites of cells in the ventromedial, dorsomedial and arcuate nuclei. A second, somewhat smaller, group of axons passes caudally in the medial forebrain bundle and ventrolaterally along the base of the hypothalamus, where they probably come into relation with some of the cells in the lateral hypothalamic area and with the terminal portions of the ventrally directed dendrites from the ventromedial nucleus which end in a pallisade-like array in this region. Finally, a small number of fibers seems to pass directly to the internal lamina of the median eminence. The absence of a localized projection from the suprachiasmatic nucleus to any well-defined cell masses in the hypothalamus makes the discussion of its role in various neuroendocrine and behavioral phenomena extremely difficult. Since the afferents to the suprachiasmatic nucleus appear to be highly localized (at least this is true of those which have been experimentally studied - including the retinal, geniculate and brain stem aminergic afferents), it is somewhat surprising to find that its efferents are distributed, not to specific nuclear groups, but to regions of neuropil which contain a variety of neurons and the processes of cells in a number of neighboring nuclei. And furthermore, it is evident from the observations in our “control” experiments, that the regions to which the suprachiasmatic nucleus projects are sites of convergence of afferents from several other structures in the rostral hypothalamus.’ Our failure to adequately label neurons in the suprachiasmatic nucleus after injections of horseradish peroxidase into the region of the ventromedial nucleus is disappointing. However, there are several possible reasons for this failure which should be considered before this approach is abandoned. The first and most obvious, is that we now know that the injections were not optimally placed. At the time these experiments were done we had assumed (from our reading of the neuroendocrine literature) that most of the efferents from the suprachiasmatic nucleus pass caudally either to the arcuate or ventromedial nuclei. It is now evident that a more appropriate focus for the horseradish per-
11
oxidase injections would have been the periventricular area, or the area ventral and lateral to the ventromedial nucleus. A second consideration which has become evident from recent studies of cell labeling with horseradish peroxidase is that for neurons to be visualized they must contain a critical amount of the enzyme, and to achieve this their axon terminals must be exposed to a sufficiently high concentration of the marker. This may be difficult to achieve if the axons of the cells branch frequently or are widely distributed. At present we do not know whether the axons of individual cells in the suprachiasmatic nucleus branch extensively, or if the relatively wide distribution of the efferents from the nucleus (which the autoradiographic method indicates) is due to the fact that different neurons project to different regions within the overall projection field of the nucleus. Lastly, the application of the retrograde transport method may be complicated in regions such as the hypothalamus by the uptake of the enzyme by axons as they pass through the injected area, as opposed to their terminals. The one situation where most of these problems should be minimal is the median eminence, and at present this seems to be the most promising area in which to apply this new experimental method. The morphological complexity of these connections is overwhelming, and should serve as a salutary warning to the too facile acceptance of hypotheses based on electrical or chemical stimulation of this region, and to the interpretation of behavioral or endocrine experiments in which they are indiscriminately ablated. It is evident that a good deal more work is required before we can provide any satisfactory account of the pathways involved in light induced behavioral or neuroendocrine phenomena. From our observations it is clear that there is no “simple route” from the visual system to the hypophysiotrophic region or to the brain stem structures which appear to Z A s we have pointed out above, it is impossible to determine from the type of material available to us, upon which cells in the critical regions the afferents from the suprachiasmatic nucleus terminate, it would be misleading therefore to imply that the afferents from the adjoining nuclei converge upon the same cellsw e simply wish to emphasize here the complexity imposed upon this system by the overlapping projections from adjacent nuclear structures.
12
L. W. SWANSON AND W. M. COWAN
be implicated in the regulation of circadian rhythms (Rusak and Zucker, '75). ACKNOWLEDGMENTS
We should like to thank Mrs. Lue Vurn Bell and Miss Lynn Rogers for their excellent histological work, Mrs. Janie Garnett for helping with several of the injections and illustrations, and Mrs. Doris Stevenson for secretarial help. LITERATURE CITED Bodian, D. 1940 Studies on the diencephalon of the Virginia opossum. 11. The fiber connections in normal and experimental material. J. Comp. Neur., 72: 207-297. Cajal, S. Ramon y 1966 Studies on the Diencephalon. Translated by E. Ramon-Moliner, C. C Thomas, Springfield. Cowan, W. M., and M. Cuenod 1975 The use of axonal transport for the study of neural connections: A retrospective survey. In: The Use of Axonal Transport for Studies of Neuronal Connectivity. Internat. Symposium, Switzerland. M. Cuenod and W. M. Cowan, eds. Elsevier, Amsterdam, in press. Cowan, W. M., D. I. Gottlieb, A . E. Hendrickson, J. L. Price and T. A. Woolsey 1972 The autoradiographic demonstration of axonal connections in the central nervous system. Brain R e search, 37: 21-51. Gottlieb, D. I., and W. M. Cowan 1973 Autoradiographic studies of the commissural and ipsilateral association connections of the hippocampus and dentate gyrus of the rat. I. The commissural connections. J. Comp. Neur., 149: 393422.
Graybiel, A. M., H. J. W. Nauta, R. J. Lasek and W. J. H. Nauta 1973 A cerebello-olivary pathway in the cat: a n experimental study using autoradiographic tracing techniques. Brain R e search, 58: 205-211. Gurdjian, E. S. 1927 T h e diencephalon of the albino rat. Studies on the brain of the rat. No. 2. J. Comp. Neur., 43: 1-114. Hendrickson, A. E., N. Wagoner and W. M. Cowan 1972 An autoradiographic and electron microscopic study of retino-hypothalamic connections. Z. Zellforsch., 1 3 5 : 1-26. Hickey, T. L., and R. W. Guillery 1974 An autoradiographic study of retino-geniculate pathways in the cat and fox. J. Comp. Neur., 1 5 6 : 239-253.
Karnovsky, J. J. 1967 T h e ultrastructural basis of capillary permeability studied with peroxidase as tracer. J. Cell Biol., 35: 213-236. Krieg, W. J. S. 1932 The hypothalamus of the albino rat. J. Comp. Neur., 5 5 : 1%89.
Lasek, R. J., B. S. Joseph and D. G. Whitlock 1968 Evaluation of a radioautographic neuroanatomical tracing method. Brain Research, 8: 31S336.
LaVail, J. M., K. R. Winston and A. Tish 1973 A method based on retrograde intraaxonal transport of protein for identification of cell bodies of axons terminating within the central nervous system. Brain Research, 5 8 : 470477. Loo, Y. T. 1931 The forebrain of the opossum, Didelphis virginiana. Part 11. Histology. J. Comp. Neur., 52: 1-148. Makara, G . B., M. C. Harris and K. M. Spyer 1972 Identification and distribution of tuberoinfundibular neurons. Brain Research, 40: 283-290. Millhouse, 0. E. 1973 The organization of the ventromedial hypothalamic nucleus. Brain R e search, 55: 71-87. Moore, R. Y. 1973 Retinohypothalamic projection i n mammals: a comparative study. Brain Research, 49: 403409. Moore, R. Y . , and D. C. Klein 1974 Visual pathways and the central neural control of a circadia n rhythm in pineal serotonin N-aceytltransferase activity. Brain Research, 71 : 17-33. Moore, R. Y . , and N. J. Lenn 1972 A retinohypothalamic projection in the rat. J. Comp. Neur., 146. 1-14. Nauta, W. J. H., and W. Haymaker 1969 Hypothalamic nuclei and fiber connections. In: The Hypothalamus. W. Haymaker, E. Anderson and W. J. H. Nauta, eds. C. C Thomas, Springfield, pp. 136209. Powell, T. P. S., W. M. Cowan and G. Raisman 1965 T h e central olfactory connexions. J. Anat., 99: 791413.
Rusak, B . , and I. Zucker 1975 Biological rhythms and animal behavior. Ann. Rev. Physiol., in press. Stephan, F. K., and I. Zucker 1972a Circadian rhythms in drinking behavior and locomotor activity of rats are eliminated by hypothalamic lesions. Proc. Nat. Acad. Sci., 69: 1583-1586. 1972b Rat drinking rhythms: Central visual pathways and endocrine factors mediating responsiveness to environmental illumination. Physiol. Behav., 8. 315326. Swanson, L. W., W. M. Cowan and E. G. Jones 1974 An autoradiographic study of the efferent connections of the ventral lateral geniculate nucleus in the albino rat and the cat. J. Comp. Neur., 156: 143-163. Szentagothai, J., B. Flerko, Z. Mess and B. Halasz 1968 The Hypothalamic Control of the Anterior Pituitary. Second ed. Akademia Kiado, Budapest. Tigges, J., and W. K. O'Steen 1974 Termination of retinofugal fibers in squirrel monkey: A r e investigation using autoradiographic methods. Brain Research, 79: 4 8 H 9 5 .
ABSTRACT The efferent connections of the suprachiasmatic nucleus of the hypothalamus have been studied in the rat by the injection of ”-proline into the nucleus and the surrounding regions of the rostral hypothalamus, and by the injection of the enzyme marker, horseradish peroxidase, into the region of the ventromedial hypothalamic nucleus. After an injection of 3H-proline confined to the ventral portion of the suprachiasmatic nucleus, transported label can be followed, in the autoradiographs, dorsally and caudally in the periventricular area as far as the caudal end of the ventromedial nucleus, into the triangular area between this nucleus and the arcuate nucleus, and along the ventral aspect of the tuberal region, just lateral to the ventromedial nucleus. A small number of silver grains are also seen over the internal lamina of the median eminence. No label can be followed rostrally or immediately lateral to the nucleus. Comparable injections into adjoining regions of the hypothalamus (especially the anterior hypothalamic area, the medial preoptic area, and the retrochiasmatic region) show transported label over the same regions, but with a somewhat different pattern of grain distribution; in addition, the anterior hypothalamic area shows a n extensive projection through the medial forebrain bundle to the mammillary and supramammillary nuclei, the midbrain tegmentum, and certain of the midline thalamic nuclei. Although i t is difficult in our autoradiographs to distinguish between the course of the efferent fibers from the suprachiasmatic nucleus and the zones in which they terminate, our evidence favors a termination among the cells of the periventricular area, and upon dendrites of the cells in the ventromedial, dorsomedial and arcuate nuclei, which extend beyond the limits of the nuclei into the periventricular area and to the area beneath the ventromedial nucleus.
The discovery that the ganglion cells of the retina project bilaterally upon the suprachiasmatic nucleus of the hypothalamus in all mammals that have been studied (Moore and Lenn, ’72; Hendrickson et al., ’72; Moore, ’73; Tigges and OSteen, ’74) and that these nuclei also receive a “visual” input from the ventral nucleus of the lateral geniculate body (Swanson et al., ’74) has focused attention upon the suprachiasmatic nuclei as the probable mediators of a number of light induced behavioral and neuroendocrine activities (Stephen and Zucker, ’72a, ’72b; Hendrickson et al., ’72; Moore and Klein, ’74). While the physiological mechanisms underlying these various phenomena remain to be elucidated, it is generally assumed that the neuroendocrine effects are mediated by some pathJ. COMP. N E U R .160. 1-12
way between the suprachiasmatic nuclei and the hypophysiotrophic zone of the hypothalamus, and that those responsible for the various circadian rhythms are effected by pathways extending caudally through the lateral hypothalamus (in the medial forebrain bundle) to the brain stem and spinal cord. However, these views are based largely on indirect evidence, since at present virtually nothing is known with certainty about the efferent connections of the suprachiasmatic nuclei (Szentagothai et al., ’68; Nauta and Haymaker, ’69). From studies of normal silver impregnated material, a projection has been described into the “supraoptic system of fibers,” including 1 This work was supported in part by grants MH24604, NS-03777 and NS-10943 from the National Institutes of Health.
1
2
L. W . SWANSON AND W . M. COWAN
the so-called “fasciculus supraopticus” and the “f ascicu lu s supr achi a smatic u s” (Gurdjian, ’27; Loo, ’31; Bodian, ’40). Both Krieg (’32) and Szentagothai et al. (’68), who have examined Golgi preparations of the suprachiasmatic nucleus, report that the axons of the cells pass dorsally and caudally into the periventricular area; unfortunately they could not be traced for any significant distance so their termination remains uncertain. The only other evidence bearing on this issue, known to us, is the finding that some cells in the suprachiasmatic nucleus can be activated antidromically by electrical stimulation in the region of the median eminence (Makara et al., ’71). As part of a systematic study of the connections of the hypothalamus, using mainly the autoradiographic method for tracing pathways (Lasek et al., ’68; Cowan et al.,’72), we have recently succeeded in obtaining a number of small injections of tritium-labeled amino acids (principally %proline) in, and around, the suprachiasmatic nucleus in the brains of young adult rats. Since in one case the injection appears to be strictly confined to the suprachiasmatic nucleus (and more specifically to the ventral portion of the nucleus in which the majority of the visual afferents terminate - Hendrickson et al., ’72; Swanson et al., ’74), it seemed worth while placing our findings on record. Although the critical findings are based on only a single experiment, the results in this case seem quite unequivocal; and since it is unlikely that a more successful experiment of this kind could be performed, we feel justified in describing this case in detail.
ilton syringe. In each case the injection was made over a period of 10-20 minutes with the bevel of the needle directed towards the structure to be labeled. After a survival period of 48 hours the animals were perfused with 10% formalin, and the brains removed, dehydrated, embedded in paraffin and serially sectioned at 15 pm. From each brain one or more series of sections was mounted (at 150 pm intervals) and processed for autoradiography as described by Cowan et al. (‘72). After exposure times ranging from 2 4 weeks, the autoradiographs were developed in D19 at 18”C, fixed in Ektaflo and stained, through the emulsion, with thionin. In an attempt to identify neurons in the rostra1 hypothalamus which might project into the ventromedial and arcuate nuclei or into the median eminence, the brains of several rats were injected stereotaxically with a small amount (usually 30-50 nl) of a concentrated solution (300500 gmlpl) of horseradish peroxidase (Sigma, type VI) and processed by a modification of the Karnovsky method (’67), as described by LaVail et al. (‘73). RESULTS
Although we have several brains in which some, or most, of the cells in the suprachiasmatic nucleus have been labeled by the isotope injection, in only one case (experiment R 186) have we succeeded in getting an injection strictly confined to neurons in the nucleus. In this case the tip of the injection needle was directed into the optic chiasm at the level of the suprachiasmatic nucleus and since the bevel was directed medially towards the nucleus, there appears to MATERIALS AND METHODS have been virtually no diffusion of the The brains which will be described are isotope into the adjoining anterior hypodrawn from a large series of experiments thalamic or medial preoptic areas. A sigwith injections of either Wproline alone, nificant amount of the injected label did, or a mixture of :sH-lysine, :$H-leucineand however, extend into the caudal part of 3H-proline, into different regions of the the optic chiasm, and as a result the local hypothalamus. In the critical case (ex- glial cells are heavily labeled. periment R 186) and two other “control” Figures 1 and 2, which are photomicroexperiments which will be described, graphs of the labeled region of the hy0.4 pCi of Wproline [ L-2,3-:’H-proline, pothalamus in R 186 (at its maximum specific activity 24 Ci/Mmole, New Eng- extent), show that the heavily labeled land Nuclear) in a volume of about 20 nl, neurons in the suprachiasmatic nucleus was injected stereotaxically through a are largely confined to its ventrolateral beveled needle attached to a 1 Ham- sector. Measurements of the region con-
CONNECTIONS OF THE SUPRACHIASMATIC NUCLEUS Abbreviations
AHA, anterior hypothalamic area A&, cerebral aqueduct ARC, arcuate nucleus BST, bed nucleus of the stria terminalis CM, central medial nucleus CP, caudate-putamen DMH, dorsomedial nucleus LHA, lateral hypothalamic area ME, median eminence MeA, medial nucleus of the amygdala ML, lateral mammillary nucleus MM, medial mammillary nucleus MPO, medial preoptic area MR, mammillary recess NLOT, nucleus of the lateral olfactory tract PAG, periaqueductal gray PIR, piriform cortex PT, parataenial nucleus PVH, paraventricular nucleus PVT, periventricular nucleus (thalamus)
Re, nucleus reuniens RT, reticular nucleus (thalamus) SCh, suprachiasmatic nucleus SO, supraoptic nucleus SUM, supramammillary area VL, lateral ventricle VMH, ventromedial nucleus V3, third ventricle ZI, zona incerta cp, posterior commissure fr, fasciculus retroflexus fx, fornix ic, internal capsule ml, medial lemniscus mt, mammillothalamic tract och, optic chiasm ot, optic tract pc, cerebral peduncle sm, stria medullaris SOC,supraoptic commissures st, stria terminalis
Fig. 1 A photomicrograph of a n autoradiograph to show the site of the injection in experiment R 186. Note that the labeled zone is confined to the ventral aspect of the suprachiasmatic nucleus and to the glia in the optic chiasm (arrows). There is a small tissue defect just lateral to the suprachiasmatic nucleus due to the tip of the injection needle. This, like the following autoradiographs, was counterstained with thionin. Scale: 250 pm.
3
4
t.W. SWANSON AND W. M. COWAN
Fig. 2 A high-power photomicrograph of the injection site in the suprachiasmatic nucleus in R 186. Heavily labeled neurons are confined to the ventrolateral aspect of the nucleus (arrows). Scale: 100 pm.
taining labeled cells indicate that its dimensions are no more than 250 pm in the mediolateral dimension, 375 pm in the dorsoventral plane (including the full extent of the involvement of the glia in the optic chiasm), and that it extends for just over 500 pm in the rostro-caudal dimension (see also fig. 3 ) . Jmmediately lateral to the labeled region there is a small defect in the tissue due to the passage of the injection needle; while this appears to have limited the lateral spread of the injected isotope, it is important to point out that it may also have interrupted some of the efferent fibers from the suprachiasmatic nucleus, Although this seems unlikely from the pattern of distribution of the axonally transported label in this and other brains in which the nucleus was involved by the injection, it cannot be ignored. We have elsewhere reported an experiment in which a relatively small, but critically placed, lesion interrupted the commissural fibers to the dentate gyrus close to their origin from an injected region of the hippocampus, which completely blocked the transport of label to the dentate gyrus on the contralateral side (experiment R 12 in Gottlieb and Cowan, ’73). The distribution of the transported
label in this experiment is shown diagrammatically in figure 3. From this, several points deserving comment emerge. First, there is no evidence in this (or any of our other experiments in which the suprachiasmatic nucleus is involved) for a rostra1 projection from the nucleus. In R 186, all the silver grains outside the limits of the nucleus are found either over the periventricular area immediately medial, dorsal and caudal to the nucleus (fig. 4), or in a narrow band along the ventral aspect of the hypothalamus ventral and lateral to the ventromedial nucleus. Second, the outflow from the nuccleus appears to be strictly ipsilateral; at no point is there any indication of transported label on the opposite side of the third ventride. We wish to emphasize this in view of the finding that the “visual input” to the nucleus from the retina and the ventral lateral geniculate nucleus is bilateral (Hendrickson et al., ’72; Swanson et al., ’74) and the finding in one of the brains labeled with horseradish peroxidase which will be described below. Third, although the transported label extends into the capsular zone surrounding the ventromedial nucleus and up to the margins of the dorsomedial and arcuate nuclei, there are comparatively
CONNECTIONS OF THE SUPRACHIASMATIC NUCLEUS
m
5
6
L. W. SWANSON AND W. M. COWAN
Fig. 4 A photomicrograph to show the silver grains over the periventricular area at the level of the ventromedial nucleus in experiment R 186. This may be compared with the density of silver grains seen over the ventromedial nucleus illustrated in figure 7A. The third ventricle and ependyma are located along the extreme left of the photomicrograph. Scale: 25 pm.
few grains over the nuclei themselves. As we shall see, this does not preclude a direct connection between the cells in the ventral portion of the suprachiasmatic nucleus and the ventromedial, dorsomedia1 or arcuate nuclei since we know from Golgi studies that many of the cells in these nuclei have dendrites which extend medially beyond the nuclear boundaries into the periventricular area (especially in the region between the ventromedial and arcuate nuclei - Szentagothai et al., '68). However, our evidence suggests that there are probably few, if any, axonal projections from the suprachiasmatic nucleus to the perikarya or to the proximal dendrites of the cells in these nuclei. Fourth, the grains over the dorsal part of the periventricular area (near the upper end of the third ventricle) sharply define the medial and ventral borders of the caudal part of the periventricular nucleus. Again, it is impossible to rule out that some of the fibers from the suprachiasmatic nucleus contact processes of cells in the periventricular nucleus (since some of them are also said to have dendrites which extend medially into the periventricular area). However, from the pattern of grain distribution in this region we
are inclined to think that most of the label in the dorsal part of the periventricular area is in axons that are passing further caudally to the level of the dorsomedial nucleus. Fifth, because of its evident importance we have paid particular attention to the region of the median eminence. On the side of the injection there is a small, but unmistakable number of silver grains over the internal lamina of the eminence. The grain density over this region is substantially lower than that over, say, the neuropil between the arcuate and ventromedial nuclei, but its consistency from section to section suggests that it represents a real projection. The fibers to the median eminence appear to enter it from its dorsolateral margin beneath the caudal part of the arcuate nucleus; they then spread out over the internal lamina but again strictly on the side of the injection. Finally, some fibers from the suprachiasmatic nucleus appear to pass into the medial forebrain bundle since the grain density over the ventral portion of the lateral hypothalamic area is consistently above the background level, at least as far as the caudal end of the ventromedial nucleus. However, a much larger group of labeled fibers can be traced along the ventral aspect of the hypothalamus, at first ventral to, and further caudally, lateral to the ventromedial nucleus. The final destination of these fibers is difficult to ascertain: the number of grains over this region becomes progressively smaller as one proceeds caudally, until just behind the ventromedia1 nucleus it is reduced to background levels. From a number of other brains with injections in the vicinity of the suprachiasmatic nucleus, but which have not labeled the cells in the nucleus, we shall describe R 38 and only two experiments-rats R 166. In R 38 the area labeled by the injection is very similar in its dimensions to that in R 186, but is confined to the retrochiasmatic region immediately behind the suprachiasmatic nucleus (figs. 5, 6). As far as we can determine, none of the cells in the suprachiasmatic nucleus have been labeled by the injected isotope; we wish to emphasize this in view of the general similarity in the pattern of grain distribution in R 38 and
CONNECTIONS OF T H E SUPRACHIASMATIC NUCLEUS
0 0 M
a
7
8
L. W . SWANSON AND W. M. COWAN
R 186. Since this is adequately illustrated in figure 5, we shall not describe it in detail, except to point out that whereas after the injection of the suprachiasmatic nucleus the transported label is largely outside the limits of the ventromedial and arcuate nuclei, in R 38 the silver grains are more-or-less uniformly distributed across the extent of those nuclei (compare figs. 7A and 7B). Thus, whereas there may be some question about the relation between the suprachiasmatic nucleus and these other nuclei, there can be no doubt that the retrochiasmatic region projects directly to both of them. In experiment R 166 a substantially larger injection (0.6 pCi of :Wproline in 30 nl) was made into the medial preoptic and anterior hypothalamic areas adjacent to the suprachiasmatic nucleus (fig. 5). Again, we shall not describe in full the distribution of the transported label in
this case except to note that it is very extensive and involves a number of structures which do not receive an input from the suprachiasmatic nucleus, including the stria terminalis and its bed nucleus, certain of the midline thalamic nuclei, the mammillary and supramammillary nuclei and the periaqueductal gray (fig. 5). But what is of particular interest is that the outflow from the medial preoptic and anterior hypothalamic areas also completely overlaps that from the suprachiasmatic nucleus, and although there appears to be significantly more label transported to the dorsomedial and arcuate nuclei in R 166 than in R 186, the grain distribution around the ventral and medial aspects of the ventromedial nucleus in the two cases is rather similar. So far we have been unsuccessful in our attempts to retrogradely label cells in the suprachiasmatic nucleus after injec-
Fig. 6 A photomicrograph of the injection site in experiment R 38. Note that the labeled cells are confined to the retrochiasmatic area (arrows). In this case the injection volume was 20 nl. Scale: 250 p m .
C O N N E C T I O N S OF T H E SUPRACHIASMATIC NUCLEUS
9
Fig. 7 High-power photomicrographs to show the relative grain densities over the ventromedial nucleus following injections in the suprachiasmatic nucleus (A) and the retrochiasmatic area (B). Scale: 25 wm.
tions of horseradish peroxidase in, and around, the ventromedial and arcuate nuclei. In two cases in which the ventrolatera1 portion of the ventromedial nucleus was heavily labeled by the injection, a small number of cells containing the reaction product was seen in the suprachiasmatic nuclei o n both sides. The identity of these cells is uncertain; in several respects they resemble small vascular pericytes seen elsewhere in the vicinity of the injection, especially in and around the acuate nuclei. In view of the fact that they were found in approximately equal numbers in the suprachiasmatic nuclei of both sides (although the injection was strictly unilateral) and as they appear quite different from other neurons retrogradely labeled by the enzyme, we are inclined to think that these cells are non-neuronal and that their labeling is probably due to vascular (or perivascular) spread of the enzyme. A s we shall discuss later, our failure to label neurons in the suprachiasmatic nucleus with horseradish peroxidase is unfortunate, since a priori this approach could
offer so much for the study of hypothalamic connections. DISCUSSION
For the study of intrahypothalamic connections the autoradiographic method clearly offers several advantages over most other neuroanatomical techniques. A s we have discussed the general advantages of this approach elsewhere (Cowan et al., '72; Swanson et al., '74; Cowan and Cuenod, '75),here we need only draw attention to two features of the method which we believe to be especially apposite for studies of the hypothalamus. First and foremost is the fact that the injected isotope is only incorporated into protein for transport by nerve cell somata in or close to the site of the injection, and not by axons passing through the labeled area (Cowan et al., '72). This makes the autoradiographic method especially useful in regions such a s the hypothalamus in which large numbers of fibers either traverse, or are closely applied to, the nuclear groups whose connections one wishes to study. While the
10
L. W. SWANSON AND W . M. COWAN
involvement of fibers of passage can, to some extent, be circumvented in degeneration studies by the careful placement of control lesions, it is always difficult to completely rule out the inadvertent interruption of some unrelated fiber system (for a further discussion of this point, see Powell et al., '65). Secondly, in many systems the autoradiographic method appears to be appreciably more sensitive, and to offer better resolution than most of the available degeneration techniques. This has already been pointed out with regard to the demonstration of the afferent connections of the suprachiasmatic nucleus (Moore and Lenn, '72; Hendrickson et al., '72; Swanson et al., '74), but it has also been noted in a number of other studies (for example, Graybiel et al., '73; Hickey and Guillery, '74; Tigges and OSteen, '74). Although it is not yet known why certain pathways that are either difficult, or even impossible, to impregnate with silver methods can readily be demonstrated by the autoradiographic method, it seems possible that the diameters of the axons may be one critical factor. As so many of the fibers in the hypothalamus are extremely fine (many are unmyelinated, with diameters well below 1 pm), it seems not unlikely that they will only be clearly demonstrated by the most sensitive techniques. Unfortunately, the use of the autoradiographic method in regions like the hypothalamus is also not without its limitations. In particular, it is usually impossible to determine from the distribution of silver grains whether one is dealing with the labeling of fibers or axon terminals, not to mention possible en passage synapses. This is an especially serious limitation in regions like the periventricular area; in a light microscopic autoradiograph it is quite impossible to know whether the grains one sees are due to label within fibers passing through the area, or to label in presynaptic processes terminating upon the many dendrites which extend into this region from the neighboring hypothalamic nuclei. By the same token, even though we can be quite certain that few, if any, labeled fibers from the suprachiasmatic nucleus enter the ventromedial or arcuate nuclei, we cannot rule out the possibility that they may be in synaptic contact with neurons
in these two nuclei. Szentagothai et al. ('68) have specifically described cells in the ventromedial, dorsomedial and arcuate nuclei which have rather lengthy dendrites that extend beyond the capsule of the nucleus into the periventricular area (also Cajal, '66; Millhouse, '73). This is particularly significant in the present context since after an injection of the suprachiasmatic nucleus the greatest concentration of silver grains is found over the periventricular area, especially in the triangular zone between the ventromedial and arcuate nuclei. The second largest contingent of labeled fibers that one sees is over the ventral aspect of the hypothalamus, ventral to the ventromedial nucleus. Although this is quite far removed from the ventromedial nucleus itself, it is evident from the Golgi studies of Millhouse ('73) that a substantial proportion of the cells in the nucleus have dendrites that reach ventrally into this region, and could well be contacted by efferent fibers from the suprachiasmatic nucleus (see especially his fig. 2). A third difficulty commonly associated with the autoradiographic method is that the injected isotope may diffuse for some distance from its principal focus, and when dealing with relatively short axonal pathways it is often difficult to distinguish labeling due to axonal transport from the diffusional spread of the tracer. In our experiments this does not appear to have been a serious problem: in R 186, for example, there are fewer than 40 significantly labeled neuron somata lateral to the suprachiasmatic nucleus, and even within the nucleus only a fraction of the total cell population is heavily labeled (fig. 2). We are inclined to attribute this limited spread of the label to two factors: (i) the smallness of the injections (20-30 nl), and (ii) the relatively long period (10-20 minutes) over which they were made. With these considerations in mind we may summarize our findings on the projection of the suprachiasmatic nucleus as follows. Most of the axons leaving the nucleus appear to pass caudally and dorsally through the periventricular area where they are in close relation to the neurons which constitute the bed nucleus of the periventricular fiber system (sometimes,
CONNECTIONS OF THE SUPRACHIASMATIC NUCLEUS
referred to as the periventricular nucleus) and to the dendrites of cells in the ventromedial, dorsomedial and arcuate nuclei. A second, somewhat smaller, group of axons passes caudally in the medial forebrain bundle and ventrolaterally along the base of the hypothalamus, where they probably come into relation with some of the cells in the lateral hypothalamic area and with the terminal portions of the ventrally directed dendrites from the ventromedial nucleus which end in a pallisade-like array in this region. Finally, a small number of fibers seems to pass directly to the internal lamina of the median eminence. The absence of a localized projection from the suprachiasmatic nucleus to any well-defined cell masses in the hypothalamus makes the discussion of its role in various neuroendocrine and behavioral phenomena extremely difficult. Since the afferents to the suprachiasmatic nucleus appear to be highly localized (at least this is true of those which have been experimentally studied - including the retinal, geniculate and brain stem aminergic afferents), it is somewhat surprising to find that its efferents are distributed, not to specific nuclear groups, but to regions of neuropil which contain a variety of neurons and the processes of cells in a number of neighboring nuclei. And furthermore, it is evident from the observations in our “control” experiments, that the regions to which the suprachiasmatic nucleus projects are sites of convergence of afferents from several other structures in the rostral hypothalamus.’ Our failure to adequately label neurons in the suprachiasmatic nucleus after injections of horseradish peroxidase into the region of the ventromedial nucleus is disappointing. However, there are several possible reasons for this failure which should be considered before this approach is abandoned. The first and most obvious, is that we now know that the injections were not optimally placed. At the time these experiments were done we had assumed (from our reading of the neuroendocrine literature) that most of the efferents from the suprachiasmatic nucleus pass caudally either to the arcuate or ventromedial nuclei. It is now evident that a more appropriate focus for the horseradish per-
11
oxidase injections would have been the periventricular area, or the area ventral and lateral to the ventromedial nucleus. A second consideration which has become evident from recent studies of cell labeling with horseradish peroxidase is that for neurons to be visualized they must contain a critical amount of the enzyme, and to achieve this their axon terminals must be exposed to a sufficiently high concentration of the marker. This may be difficult to achieve if the axons of the cells branch frequently or are widely distributed. At present we do not know whether the axons of individual cells in the suprachiasmatic nucleus branch extensively, or if the relatively wide distribution of the efferents from the nucleus (which the autoradiographic method indicates) is due to the fact that different neurons project to different regions within the overall projection field of the nucleus. Lastly, the application of the retrograde transport method may be complicated in regions such as the hypothalamus by the uptake of the enzyme by axons as they pass through the injected area, as opposed to their terminals. The one situation where most of these problems should be minimal is the median eminence, and at present this seems to be the most promising area in which to apply this new experimental method. The morphological complexity of these connections is overwhelming, and should serve as a salutary warning to the too facile acceptance of hypotheses based on electrical or chemical stimulation of this region, and to the interpretation of behavioral or endocrine experiments in which they are indiscriminately ablated. It is evident that a good deal more work is required before we can provide any satisfactory account of the pathways involved in light induced behavioral or neuroendocrine phenomena. From our observations it is clear that there is no “simple route” from the visual system to the hypophysiotrophic region or to the brain stem structures which appear to Z A s we have pointed out above, it is impossible to determine from the type of material available to us, upon which cells in the critical regions the afferents from the suprachiasmatic nucleus terminate, it would be misleading therefore to imply that the afferents from the adjoining nuclei converge upon the same cellsw e simply wish to emphasize here the complexity imposed upon this system by the overlapping projections from adjacent nuclear structures.
12
L. W. SWANSON AND W. M. COWAN
be implicated in the regulation of circadian rhythms (Rusak and Zucker, '75). ACKNOWLEDGMENTS
We should like to thank Mrs. Lue Vurn Bell and Miss Lynn Rogers for their excellent histological work, Mrs. Janie Garnett for helping with several of the injections and illustrations, and Mrs. Doris Stevenson for secretarial help. LITERATURE CITED Bodian, D. 1940 Studies on the diencephalon of the Virginia opossum. 11. The fiber connections in normal and experimental material. J. Comp. Neur., 72: 207-297. Cajal, S. Ramon y 1966 Studies on the Diencephalon. Translated by E. Ramon-Moliner, C. C Thomas, Springfield. Cowan, W. M., and M. Cuenod 1975 The use of axonal transport for the study of neural connections: A retrospective survey. In: The Use of Axonal Transport for Studies of Neuronal Connectivity. Internat. Symposium, Switzerland. M. Cuenod and W. M. Cowan, eds. Elsevier, Amsterdam, in press. Cowan, W. M., D. I. Gottlieb, A . E. Hendrickson, J. L. Price and T. A. Woolsey 1972 The autoradiographic demonstration of axonal connections in the central nervous system. Brain R e search, 37: 21-51. Gottlieb, D. I., and W. M. Cowan 1973 Autoradiographic studies of the commissural and ipsilateral association connections of the hippocampus and dentate gyrus of the rat. I. The commissural connections. J. Comp. Neur., 149: 393422.
Graybiel, A. M., H. J. W. Nauta, R. J. Lasek and W. J. H. Nauta 1973 A cerebello-olivary pathway in the cat: a n experimental study using autoradiographic tracing techniques. Brain R e search, 58: 205-211. Gurdjian, E. S. 1927 T h e diencephalon of the albino rat. Studies on the brain of the rat. No. 2. J. Comp. Neur., 43: 1-114. Hendrickson, A. E., N. Wagoner and W. M. Cowan 1972 An autoradiographic and electron microscopic study of retino-hypothalamic connections. Z. Zellforsch., 1 3 5 : 1-26. Hickey, T. L., and R. W. Guillery 1974 An autoradiographic study of retino-geniculate pathways in the cat and fox. J. Comp. Neur., 1 5 6 : 239-253.
Karnovsky, J. J. 1967 T h e ultrastructural basis of capillary permeability studied with peroxidase as tracer. J. Cell Biol., 35: 213-236. Krieg, W. J. S. 1932 The hypothalamus of the albino rat. J. Comp. Neur., 5 5 : 1%89.
Lasek, R. J., B. S. Joseph and D. G. Whitlock 1968 Evaluation of a radioautographic neuroanatomical tracing method. Brain Research, 8: 31S336.
LaVail, J. M., K. R. Winston and A. Tish 1973 A method based on retrograde intraaxonal transport of protein for identification of cell bodies of axons terminating within the central nervous system. Brain Research, 5 8 : 470477. Loo, Y. T. 1931 The forebrain of the opossum, Didelphis virginiana. Part 11. Histology. J. Comp. Neur., 52: 1-148. Makara, G . B., M. C. Harris and K. M. Spyer 1972 Identification and distribution of tuberoinfundibular neurons. Brain Research, 40: 283-290. Millhouse, 0. E. 1973 The organization of the ventromedial hypothalamic nucleus. Brain R e search, 55: 71-87. Moore, R. Y. 1973 Retinohypothalamic projection i n mammals: a comparative study. Brain Research, 49: 403409. Moore, R. Y . , and D. C. Klein 1974 Visual pathways and the central neural control of a circadia n rhythm in pineal serotonin N-aceytltransferase activity. Brain Research, 71 : 17-33. Moore, R. Y . , and N. J. Lenn 1972 A retinohypothalamic projection in the rat. J. Comp. Neur., 146. 1-14. Nauta, W. J. H., and W. Haymaker 1969 Hypothalamic nuclei and fiber connections. In: The Hypothalamus. W. Haymaker, E. Anderson and W. J. H. Nauta, eds. C. C Thomas, Springfield, pp. 136209. Powell, T. P. S., W. M. Cowan and G. Raisman 1965 T h e central olfactory connexions. J. Anat., 99: 791413.
Rusak, B . , and I. Zucker 1975 Biological rhythms and animal behavior. Ann. Rev. Physiol., in press. Stephan, F. K., and I. Zucker 1972a Circadian rhythms in drinking behavior and locomotor activity of rats are eliminated by hypothalamic lesions. Proc. Nat. Acad. Sci., 69: 1583-1586. 1972b Rat drinking rhythms: Central visual pathways and endocrine factors mediating responsiveness to environmental illumination. Physiol. Behav., 8. 315326. Swanson, L. W., W. M. Cowan and E. G. Jones 1974 An autoradiographic study of the efferent connections of the ventral lateral geniculate nucleus in the albino rat and the cat. J. Comp. Neur., 156: 143-163. Szentagothai, J., B. Flerko, Z. Mess and B. Halasz 1968 The Hypothalamic Control of the Anterior Pituitary. Second ed. Akademia Kiado, Budapest. Tigges, J., and W. K. O'Steen 1974 Termination of retinofugal fibers in squirrel monkey: A r e investigation using autoradiographic methods. Brain Research, 79: 4 8 H 9 5 .
