Migration Patterns of Sympathetic Preganglionic Neurons in Embryonic Rat Spinal Cord Jeffrey A. Markham* and James E. Vaughnt Division of Neurosciences, Beckman Research institute of the City of Hope, Duarte, California 91010-0269
SUMMARY The displacement of immature neurons from their place of origin in the germinal epithelium toward their adult positions in the nervous system appears to involve migratory pathways or guides. While the importance of radial glial fibers in this process has long been recognized, data from recent investigations have suggested that other mechanisms might also play a role in directing the movement of young neurons. We have labeled autonomic preganglionic cells by microinjections of horseradish peroxidase (HRP) into the sympathetic chain ganglia of embryonic rats in order to study the migration and differentiation of these spinal cord neurons. Our results, in conjunction with previous observations, suggest that the migration pattern of preganglionic neurons can be divided into three distinct phases. In the first phase, the autonomic motor neurons arise in the ventral ventricular zone and migrate radially into the ventral horn of the developing spinal cord, where, together with somatic motor neurons, they form a single, primitive motor column (Phelps P. E., Barber R.P., and Vaughn J. E. (1991). J. Comp. Neurol. 307:77-86). During the second phase, the autonomic motor neurons separate from the somatic motor neurons and are displaced dorsally toward the intermediate spinal cord. When the preganglionic neurons reach the intermediolateral (IML) region, they become progressively more multipolar, and many of them undergo a change in alignment, from a dorsoventral to a
mediolateral orientation. In the third phase of autonomic motor neuron development, some of these cells are displaced medially, and occupy sites between the IML and central canal. The primary and tertiary movements of the preganglionic neurons are in alignment with radial glial processes in the embryonic spinal cord, an arrangement that is consistent with a hypothesis that glial elements might guide autonomic motor neurons during these periods of development. In contrast, during the second phase, the dorsal translocation of preganglionic neurons occurs in an orientation perpendicular to radial glial fibers, indicating that glial elements are not involved in the secondary migration of these cells. The results of previous investigations have provided evidence that, in addition to glial processes, axonal pathways might provide a substrate for neuronal migration. Logically, therefore, it is possible that the secondary dorsolateral translocation of autonomic preganglionic neurons could be directed along early forming circumferential axons of spinal association interneurons, and this hypothesis is supported by the fact that such fibers are appropriately arrayed in both developmental time and space to guide this movement. Keywords: motor neuron development, autonomic motor neurons, somatic motor neurons, gliophilic and neurophilic migrations, H R P retrograde labeling.
INTRODUCTION
appreciated for many years. Virtually all neurons are generated in germinal regions that are some distance removed from their eventual adult loca-
The importanceof neuronal migration in the developing organization of the nervous system has been Received April 15, 1991; accepted July 15, 1991 Journal of NeurobiologY. VOl. 22, No. 8, PP. 8 1 1-822 (199 1) tC 199 I John Wiley & Sons. Inc. CCC 0022-3034/9 I/OSOCr 11-12$04.00
* Present address: Department of Radiological Sciences, School of Medicine, B2-086 CHS, University of California. Los Angeles, Los Angeles, California 90024- I 72 1. To whom correspondence should be addressed. 81I
812
Markham and Vaughn
tions, and many such cells must migrate relatively long distances to reach their ultimate sites of residence. For nearly two decades, a dominant hypothesis of neuronal migration has stressed the role of radial glia as migratory pathways or guides (Rakic, 1972; Hatten, 1990). Much support for this hypothesis has been forthcoming from investigations of cortical structures (see Rakic, 1985, 1990; McConnell, 1988 for reviews). However, the topic of neuronal migration has not been pursued in as much detail in noncortical regions of the central nervous system (CNS). In many of these regions, the orientation of radial glia is often incongruent with observed neuronal migratory routes, suggesting that alternative mechanisms might also be important for directing the movements of nerve cells (Bourrat and Sotelo, 1988,1990; Moody and Heaton, 1983a,b,c;Rakic, 1985; Ono and Kawamura, 1989). In the embryonic spinal cord, radial glial fibers extend in the transverse plane between the ventricular zone and the pial surface (Choi, 1981; Misson, Edwards, Yamamoto, and Caviness, 1988;Ram6n y Cajal, 1960). If it is assumed that radial glia are involved in the migration of spinal cord neurons toward their adult positions, one would predict that subpopulations of these nerve cells would be generated at their appropriate dorsoventral level, and that they then would migrate directly laterally from the ventricular zone or germinal epithelium into the developing intermediate zone (e.g., Altman and Bayer, 1984; Nornes and Das, 1974). However, not all neuronal populations in the mammalian spinal cord appear to follow a strictly radial migration pattern throughout their development. The sympathetic preganglionic neurons of the adult rat, for example, are located within several nuclei in the intermediate region (lamina VII) of the spinal cord (Petras and Faden, 1978;Rando, Bowers, and Zigmond, 1981), but these cells do not appear to simply migrate radially from the intermediate part of the ventricular zone. Previous studies of developing spinal cord suggest that autonomic and somatic motor neurons are generated over the same time periods (Barber, Phelps, and Vaughn, 1989), and that both subclasses of motor neurons may be clonally related (Leber, Breedlove, and Sanes, 1990). Moreover, a recent immunocytochemical study of cholinergjc neurons in developing rat spinal cord suggested that somatic and autonomic motor neurons initially form a single, primitive motor column within the developing ventral horn (Phelps, Barber, and Vaughn, 1991 ) . Earlier
studies suggested that this also might be the case in the chick (Levi-Montalcini, 1950; Oppenheim, Maderdrut, and Wells, 1982), but a more recent examination of embryonic chick spinal cord indicated that many of the preganglionic neurons (Terni cells) “maintain medial positions and do not migrate laterally to join a common motor column before initiating a dorsal migration” (Prasad and Hollyday, 1991) . In contrast, choline acetyltransferase ( ChAT) immunoreactive somata in the primitive motor column appear to be the only source of ChAT-positive fibers in the ventral roots of early rat embryonic spinal cord. These fibers innervate both somatic muscles and the sympathetic chain ganglia, and this implies that both somatic and preganglionic motor neurons are located in the primitive motor column of this species (Phelps et al., 1991 ). However, using ChAT immunocytochemistry, it was not possible to demonstrate whether these two subclasses of motor neurons were intermixed or, alternatively, whether they were segregated in different parts of the primitive motor column, because both cell types were stained by this method. Consequently, in the present investigation, we examined the development of identified preganglionic neurons using horseradish peroxidase (HRP) retrograde labeling from microinjections into the sympathetic chain ganglia of rat embryos. The results provided evidence that preganglionic autonomic neurons were initially intermixed with somatic motor neurons within the primitive motor column of the ventral spinal cord, and that these cells subsequently underwent a secondary, dorsolateral translocation into the intermediate spinal cord. This secondary translocation was followed by a tertiary movement of some autonomic motor neurons to give rise to the more medially-located preganglionic nuclei. These results confirmed the secondary and tertiary displacements of preganglionic sympathetic neurons as suggested by the independent method of ChAT immunocytochemistry (Phelps et al., 1991) .
METHODS
Experimental Animals
Female Sprague-Dawley rats were placed in the cages of males overnight. The following morning, vaginal smears were prepared and, if they were sperm positive, that day
Development of Preganglionic Neurons
813
Figure 1 Transverse sections of E13$ thoracic spinal cord. ( a ) Four retrogradely-labeled preganglionic neurons (arrowheads) are shown distributed over much of the primitive motor column (dashed outline) of the spinal cord (solid outline). The axons of these neurons project from the ventral root and then turn medially (arrows) toward the injection site (asterisk)in the sympathetic chain. The morphology of three of these neurons (arrows) is more readily apparent in a photomontage shown at higher magnification (b). The fourth neuron (arrowhead) was too far out of the primary focal plane for its focused image to be incorporated in the montage. In another specimen (c) , retrogradely-labeled preganglionic neurons (arrow) are found scattered throughout the central region ofthe primitive motor column (dashed outline), indicating that they are not yet segregated from somatic motor neurons at this age. No HRP-labeled fibers were observed coursing toward sites of developing somatic muscle in these, or any other, specimens illustrated in this study. Abbreviations: RP = roof plate; V = ventricle; FP = floor plate; VT = vertebra; DRG = dorsal root ganglia. Scale bars: a, c = 100 pm; b = 30 p m .
was designated as embryonic day 1 ( E l ) . Since some investigators designate the day of insemination as EO (e.g., Rubin, I985a,b,c), all data referred to in this report have been converted to conform with the E l staging method in order to facilitate comparisons among different studies. At appropriate ages ( E 1 3-El 6), pregnant rats were deeply anesthetized with halothane using a precision vaporizer, and the embryos were removed via cesarian section. The amniotic sacs containing the em-
bryos were kept intact and placed in chilled (4"C), oxygenated, Gey's balanced salt solution ( RSS; Freshney, 1987). All specimens were prepared for HRP injection within 3 h of delivery. Sixteen to forty-eight embryos from each age were analyzed in this study.
HRP Injections The thoracic region of each embryo, including the spinal cord and surrounding tissues, was isolated and cut into
814
Markham and Vaughn
Figure 2 Transverse sections of E l 4 thoracic spinal cord. At low magnification (a), retrogradely-labeied autonomic preganglionic neurons (arrow) are observed to be clustered in the dorsolateral region of the primitive motor column (dashed outline). The bipolar somata of the preganglionic neurons are oriented dorsoventrally, and their axons can be seen exiting the
Development of Preganglionic Neurons two transverse slices, each approximately 1 mm thick. Using a Picospritzer (General Valve), small amounts of HRP (Sigma, Type VI, 30%in distilled H 2 0containing 2% dimethylsulfoxide [ DMSO] ) were pressure injected from 10-20 pm tip diameter glass micropipettes that were lowered vertically through the region of the sympathetic chain ganglia. Subsequent to injection, the tissue was placed in oxygenated BSS and “cultured” for 4-6 h at room temperature. Following HRP uptake and transport, the specimens were fixed for 12- 18 h in a mixture of 2.5% glutaraldehyde, 1% paraformaldehyde. and 4% sucrose in 0.1 M Soretisen’s phosphate buffer (pH = 7.4), rinsed several times in phosphate buffer, and cryoprotected in 30% sucrose in buffer overnight. Coronal (transverse) sections were cut at a thickness of 40 pm with a cryostat, and stored in 24-well tissue culture plates containing phosphate buffer. Sections were serially mounted on gel-coated slides, allowed to dry, and processed for HRP histochemistry. This method is essentially similar to that used in a previous study ofthe embryonic development ofrat superior cervical ganglion (Rubin, 1985,a,b,c). Although the reliability of this method has been demonstrated, the following procedures were conducted to control against the possibility that somatic motor axons coursing lateral to the sympathetic chain might become labeled from the ganglionic HRP injections: ( 1 ) tracer was injected directly into the ventral roots, and this did not result in the consistent labeling of any neurons in the spinal cord, suggesting that the HRP was not taken up efficiently by fibers of passage; (2) HRP was injected directly into somatic muscles of E l 4 and older specimens, and these injections labeled a population of neurons distinct both in location and morphology from cells labeled by injections into the sympathetic chain ganglia; and ( 3 ) injections into the chain ganglia resulted in a different pattern and course of HRP-labeled, peripheral nerve fibers (see Figs. 1, 2 ) than did injections into somatic muscle.
Histochemical Processing Prior to HRP processing, the slide-mounted sections were incubated for 20 min in a solution of 0.3%H202 and 0.1% sodium azide in Tris-buffered saline (TBS) in an attempt to suppress endogenous peroxidase activity (Li, Ziesmer, and Lazcano-Villareal, 1987). HRP-con-
815
taining neurons were visualized by first preincubating the sections for 10 min in a solution of 0.6Y0 CoCl,, 0.4% Ni( NH4)2 6H20and 0.06% diaminobenzadine (DAB) to increase the intensity of the final reaction product (Adams, 1981). After rinsing in buffer, the specimens were transferred to a solution of 0.06% DAB/0.007% H,O, for 10-15 min, rinsed in buffer, dehydrated, and coverslipped.
RESULTS
On E 13, relatively few differentiating neurons are present in the spinal cord; virtually all of these cells are located in a single cell column in the developing ventral horn ( Altman and Bayer, 1984; Phelps et al., 1991). Moreover, the sympathetic chain ganglia are just beginning to form, and only a small number of axonal growth cones from autonomic motor neurons reach the ganglia (Rubin, 1985a,b). However, significant changes in spinal cord organization occur between E l 3 and E l 4 (e.g., see below and Phelps et al., 1991), and we therefore injected specimens late on E 13 (E 134) in order to determine if somatic and autonomic motor neurons are initially intermixed, or, altetnatively, whether they are already segregated into distinct subgroups within the primitive motor column at this age. The E13; results indicated that somatic and preganglionic autonomic motor neurons probably are not segregatedinto distinct pools during the early primitive motor column stage (Fig. I ) . By E14, most of the preganglionic axons have reached the sympathetic chain (Rubin, 1985b), and microinjections of HRP into the ganglta resulted in robust labeling of cells in the ventral intermediate zone of the spinal cord. Although it is known that preganglionic autonomic and somatic motor neurons form a single, primitive motor column at this age (Phelps et al., 1991 ), the autonomic neurons were clearly segregated from the
spinal cord (solid outline) into the ventral root (small arrowheads) on their way toward the HRP-injected sympathetic chain (large arrowhead). The ventral and ventromedial part of the primitive motor column in panel a displays no HRP-labeled cells, but in specimens where HRP was injected directly into somatic muscles this part of the column exclusively contained retrogradely-labeled neurons. In a section from another specimen ( b )where only a few cells are labeled (asterisk), the bipolar nature of the preganglionic neurons is more evident. The cells possess ventrally-directed axons (small arrowheads), and dendritic processes (e.g., arrow) that project dorsally toward the intermediate region of the spinal cord. Abbreviationsas in Figure 1. Scale bars: a = 100 p m ; b = 50 pm.
816
&larkham and Vaughn
Figure 3 Transverse sections of E 15 thoracic spinal cord. ( a ) As compared with E 14, retrogradely-labeled preganglionic neurons on E 15 are located considerably more dorsal to the somatic motor neurons of the ventral horn (VH, dashed outline), and they have reached the level of the intermediate gray matter to form the intermediolateral nucleus (IML). At this age, medially-located preganglionic neurons (e.g., arrow, a ) are seen for the first time in the region between the IML and the ventricle ( V ) . (b-d) Some of the dorsal dendrites of preganglionic neurons in the vicinity ofthe intermediate gray matter have become reoriented into the mediolateral plane (e.g., arrows in b) . In addition, labeled cell bodies (arrowhead, c) are often found in alignment with these mediolateral dendrites. The dendrites of a few neurons (e.g., arrow in d ) can sometimes be seen extending medially through the ventricular zone toward the ventncular surface ( V ). Abbreviations: DH = dorsal horn; MZ = marginal zone; VH = ventral horn. Dorsal is up, and ventral is down in a-d. Scale bars: a = 100 wm; b-d = 50 pm.
Development of Pregunglionic Neurons
somatic motor neurons, with the former being located exclusively in the dorsolateral portion of the developing motor column [Fig. 2 ( a )1. The axons of the preganglionic neurons generally coursed along the border between the intermediate and margmal zones before exiting to the ventral root, but some were also noted to course more medially amongst the somatic motor cells before reaching their ventral root exit [Fig. 2( b)] . Many somata of preganglionic neurons on El 4 were clustered at the interface between the intermediate and marginal zones; these cells had an essentially bipolar morphology that displayed a strict dorsoventral orientation [Fig. 2(b)]. Injections of somatic muscle at this developmental stage labeled cells that were restricted to the ventral and ventromedial aspects of the primitive motor column (not shown). On E 15, the position of the preganglionic neurons had shifted further dorsally, and many of the cells exhibited multiple dendritic processes. The alignment of cells located in progressively closer proximity to the intermediate gray matter changed from a dorsoventral to a predominantly mediolatera1 orientation (Fig. 3). Moreover, on E l 5 , a few labeled cells were seen for the first time in a medial location between the intermediolateral ( TML) horn and the ventricular region. In addition, a number of individual preganglionic neurons within the IML possessed dendrites that extended medially toward, and often into, the ventricular zone [ e.g., Fig. 3 (d)] . By E 16, virtually all preganglionic neurons had reached the intermediate gray matter, and the number of such cells found in medial locations was more numerous that on E l 5 (Fig. 4). Furthermore, the number of mediolaterally-directed preganglionic dendrites increased between E 15 and E16, and the first indications of dendritic bundling, a characteristic of adult preganglionic neurons (Barber et al., 1984: Vera, Ellenberger, Haselton, Haselton, and Schneiderman, 1986; Davidoff, Galabov, and Bergmann, 1989), became apparent on E 16. Some of the preganglionic neurons at E 16 also extended dendrites into the developing lateral funiculus of the marginal zone (Fig. 4). In short, most of the preganglionic neurons displayed significant morphological differentiation by E l 6, and they had many of the features of mature preganglionic cells. Taken together, the observations of preganglionic neuronal translocations made at each age with retrograde HRP-labeling in the present study substantiate previous findings reported on
81 7
the independent basis of ChAT immunocytochemistry (Phelps et al., 1991 ). DISCUSSION
In the adult mammalian spinal cord, sympathetic preganglionic neurons are located in a narrow band of the intermediate gray matter (lamina VII). Within this lamina, preganglionic neurons are found primarily in the IML nucleus; however, depending on spinal level, a significant number of cells are also located in the intercalated and central autonomic nuclei, as well as in the lateral funiculus (Petras and Faden, 1978; Rando et al., 1981). Given the distinctive mediolateral alignment of these nuclei, it might be expected that, duringdevelopment, preganglionic neurons would be generated in the intermediate part of the ventricular zone, and then migrate laterally into the various sympathetic subnuclei. It is conceivablethat such a direct migration could occur along radial glial fibers, since they are in proper alignment to guide such movements (Choi, 1981; Misson et al., 1988) . However, the results of the present study, in conjunction with those of other investigators, reveal that this hypothesis is inconsistent with the observed development of autonomic preganglionic neurons in the mammalian spinal cord. Specifically, these results suggest that the migration of autonomic preganglionic neurons can be divided into distinct stages, and that elements other that radial glia are likely to be involved in directing the movement of these cells during certain stages of their migration. Both somatic and autonomic motor neurons appear to arise in the ventral ventricular zone (Leber et al., 1990; Phelps et al., 1991); in rats, at least, these two cell populations then appear to migrate into a single primitive motor column in a pattern parallel to the alignment of radial glia (Phelps et al., 1991). The recent demonstration that somatic and autonomic motor neurons appear to be clonally related (Leber et al., 1990), in conjunction with our E13t data showing retrogradely-labeled cells distributed throughout the primitive motor column, indicate that the two populations of neurons are probably intermixed at the early primitive motor column stage. Subsequently, by E 14, the autonomic motor cells separate from the somatic motor neurons and begin a secondary, dorsal translocation toward the intermediate region of the spinal cord. The direction of this trans-
818
Markham and Vaughn
Figure 4 Transverse sections of E 16 thoracic spinal cord. (a) A11 retrogradely-labeled preganglionic neurons are located in the intermediate gray matter on E16, with the majority of
Development of Preganglianic Neurons
location is perpendicular to the alignment of radial glial fibers, indicating that such structures are not likely to be involved in the dorsal movement of preganglionic neurons. Thus, some other element( s) must be responsible for this phase of their migration. One possible substrate for the dorsal migration of preganglionic neurons is provided by the early forming circumferential axons of spinal cord association interneurons. In the rat, these axons course ventrally from the developing dorsal horn and, at appropriate developmental stages, they are located along the preganglioniccell migration route at the interface of the intermediate and marginal zones (e.g., Windle and Baxter, 1936; Vaughn and Grieshaber, 1973; Vaughn, Phelps, Yamamoto, and Barber, 1990). Several investigators have provided evidence that the movement of neurons along axonal pathways, termed “neurophilic” migration by Rakic ( 1990), occurs in certain regions of the developing CNS (Moody and Heaton, 1983a,b,c; Rakic, 1985; Bourrat and Sotelo, 1988;Ono and Kawamura, 1989). For example, Moody and Heaton (1983a,b,c) have demonstrated that trigeminal motor neurons in the brain stem originate in the medial column, become reoriented to lie parallel to the pial surface, and then migrate laterally in close association with axonal bundles. These authors also have reported the presence of numerous adherens junctions between the migrating trigeminal neurons and individual axons of the fiber pathway. Elimination of the axons prior to migration appeared to prevent the tangential movement of the motor neurons (Moody and Heaton, 1983b,c). Thus, there is precedence for the idea that axonal pathways can provide guides for neuronal migration. Although the location and orientation of association fibers suggest that they could provide a substrate for the dorsal migration of preganglionic neurons, a determination of
819
whether or not they play a role in this process awaits further study. The observation that the preganglionic autonomic neurons end their dorsal translocation in the intermediate gray matter, and that some of these cells then appear to extend medially, suggests that this region of the spinal cord may contain cues that play a role in the reorientation of preganghonic neurons from a dorsoventral to a predominantly mediolateral alignment. One such cue may involve the emergence of mediolateral dendrites from preganglionicneurons since the development of these processes temporally coincideswith the appearance of centrally-located autonomic neurons between the IML and the ventricular zone. This suggests that the medially-located cells might become displaced from the IML during the extension of dendrites, or, alternatively, that both dendrites and somata may be guided by radially-aligned glial processes during this tertiary phase of autonomic motor neuron development. In conclusion, the migration ofpreganglionicautonomic motor neurons from the ventral ventricular zone into the anlage of lamina VTI of the mature spinal cord appears to involve a series of discrete steps that are summarized in Figure 5. In addition to their relevance for the issue of cell migration, however, the present results raise several questions with respect to other aspects of sympathetic preganglionic neuronal development. For example, while it is clear from the present study that axogenesis of preganglionic neurons is well under way at least as early as the primitive motor column stage (see also Rubin, 1985c), it remains to be determined whether the initial segregation of autonomic and somatic motor neurons that occurs between E13f and El4 involves an interaction of their respective axons with peripheral targets. Moreover, apparent correlations between cell mi-
them being found in the intermediolateral nucleus (IML). Their dendrites are generally oriented in the mediolateral plane, extending laterally into the developing marginal zone (MZ), and medially toward the ventricle (V). At higher magnification ( b ) , the medially-directed dendrites (small arrows) of the preganglionic neurons are seen to be grouped together in bundles, and neuronal somata (e.g.. arrow) are sometimes observed to be enmeshed in these processes. Dendrites (e.g., arrowhead) extending into the developing lateral funiculus of the marginal zone (MZ) are seen for the first time at this age. In addition, an increase in the number of more medially-located preganglionic neurons (arrows, c) on E 16, as compared with earlier ages, is readily apparent in a section from another specimen. These medial cells probably represent the precursors of the intercalated and central autonomic neurons that are characteristic of mature animals. Abbreviations: DH = dorsal horn; VH = ventral horn; V = ventricle. Scale bars: a = 100 pm; b-d = 50 fim.
820
Markham and Vaughn
El3
El 5
El6
Figure 5 Summary of the migration patterns of autonomic preganglionic neurons in embryonic rat spinal cord. Both somatic and autonomic motor neurons are generated between E 1 1- 12 (Barber et al., 1989) , and appear to migrate radially from the ventricular zone (VZ) into the developing ventral horn of the intermediate zone (IZ), where they form a single, primitive motor column (stippled area) by El 3 (Phelps et al., 1991 ). At E13i the autonomic and somatic motor neurons appear to be intermixed within the primitive motor column. By E 14, the preganglionic autonomic neurons have segregated from the somatic motor neurons, and form a distinct population in the dorsolateral region of the primitive column (stippled area). At this age, many of the dorsoventrally-orientedpreganglionic neurons are clustered at the interface between the intermediate zone and the developing marginal zone (MZ). As the preganglionic autonomic neurons approach the intennediolateral region (stippled area) on E 15, the cells become progressively reoriented into the mediolateral plane. At this age, a few of the neurons also begin a medially-directed displacement toward the ventricular zone. By E 16. the majority of preganglionic neurons (stippled area) are oriented in the mediolateral plane, and their dendntic processes extend medially toward the receding ventricular zone, as well as laterally into the developing marginal zone. As in the mature spinal cord, the majority of cell bodies on E 16 are located within the intermediolateral region, but a significant number of cells are also present in more medial locations between the IML and the central canal. Note that the sites occupied by somatic motor neurons are indicated by the open part ofthe primitive motor column on E l 4 and by the open, outlined areas of the ventral horn on El 5 and 16.
gration and dendntic rearrangements (see also Phelps et al., 199 1; Markham, Phelps, and Vaughn, 1990) during the development of sympathetic preganglionic neurons, suggest that these two events may share common epigeneticdeterminants that remain to be discovered. This research was supported by grants NS25784 and NS 18858 from the National Institute of Neurological Disorders and Stroke. The author; thank Robert Barber for his help with various aspects of this research, Drs. Arlene Chiu, and Patricia Phelps for helpful discussions,
and Marilyn Ashley, Lynn Rrennan, Mariko Lee, and Christine Vaughn for their excellent technical assistance.
REFERENCES ADAMS,J. C. ( 1 98 1). Heavy metal intensification of DAB-based HRP reaction product. J. Histochem. Cytochem. 29:775. ALTMAN,J. and BAYER, S. A., Ed. ( 1984). TheDevelopment of the Rut Spinal Cord, Springer-Verlag, Berlin. BARBER,R. P., PHELPS,P. E., HOUSER,C. R., CRAW-
Development of Pregunglionic Neurons FORD, G. D.,
SALVATERRA, P. M., and VAUGHN, J. E.
( 1984). The morphology and distribution of neurons
containing choline acetyltransferase in the adult rat spinal cord: an immunocytochemical study. J. Comp. Neurol. 229:329-346. P. E., and VAUGHN,J. E. BARBER,R. P., PHELPS, ( 1989). Generation patterns of spinal sympathetic and parasympathetic neurons in the rat spinal cord. SOC.Neurosci. Abstr. 15588. F. and SOTELO,C. (1988). Migratory pathBOURRAT, ways and neuritic differentiation of inferior olivary neurons in the rat embryo. Axonal tracing study using the in vitro slab technique. Dev. Brain Rex 39:19-37. BOURRAT,F. and SOTELO,C. ( 1990). Migratory pathways and selective aggregation of the lateral reticular neurons in the rat embryo: a horseradish peroxidase in vitro study, with special reference to migration patterns of the precerebellar nuclei. J. Comp. Neurol. 294: 1- 13. CHOI, B. H. ( 1981). Radial glia of developing human fetal spinal cord: Golgi, immunohistochemical and electron microscopic study. Dev. Bruin Res. 1:249267. DAVIDOFF, M. S., GALABOV, P. G., and BERGMANY, M. ( 1989). The vegetative network in the thoracolumbar spinal cord of the guinea pig: a comparison of the distribution of AChE-enzyme activity and choline acetyltransferase-like immunoreactivity. J. Hirnforsch. 30:707-7 17. FRESHNEY, R. I., Ed. ( 1987). Culture ofAnimal Cells:A Manual of Basic Technique, Wiley-Liss, New Y ork. HATTEN, M. E. (1990). Riding the glial monorail: a common mechanism for glial-guided neuronal migration in different regions of the developing mammalian brain. Trends Neurosci. 13:179- 184. S. M., and SANES,J, R. LEBER,S. M., BREEDLOVE, ( 1990). Lineage, arrangement, and death of clonally related motoneurons in chick spinal cord. J. Neurosci. 10:245 1-2462. LEVI-MONTALCINI, R. (1950). The origin and development of the visceral system in the spinal cord of the chick embryo. J. Morphol. 86:258-284. LI, C.-Y., ZIESMER,S., and LAZCANO-VILLAREAL, 0. ( 1987). Use of azide and hydrogen peroxide as an inhibitor for endogenous peroxidase in the immunoperoxidase method. J. Histochem. Cytochem. 35: 14571460. MARKHAM, J. A., PHELPS,P. E., and VAUGHN,J. E. ( 1990). Development of longitudinal dendritic bundles in rat spinal cord: analysis of cholinergic sympathetic preganglionic neurons. SOC.Neurosci. Abstr. 16:644. S. K. ( 1988). Development and decisionMCCONNELL, making in the mammalian cerebral cortex. Bruin Res. Rev. 13:1-23. MISSON,J.-P., EDWARDS,M. A., YAMAMOTO, M., and V. S. JR. ( 1988). Identification of radial CAVINESS, glial cells within the developing murine central ner-
821
vous system: studies based upon a new immunohistochemical marker. Dev. Bruin Rex 44:95-108. MOODY,S. A. and HEATON,M. B. ( 1983a). Ultrastructural observations of the migration and early development of trigeminal motoneurons in chick embryos. J. Comp. Neurol. 216:20-35. MOODY,S. A. and HEATON.M. B. (1983b). Developmental relationships between trigeminal ganglia and trigeminal motoneurons in chick embryo. I. Ganglion development is necessary for motoneuron migration. J. Comp. Neurol. 213~327-343. . MOODY,S. A. and HEATON,M. B. ( 1 9 8 3 ~ )Developmental relationships between trigeminal ganglia and trigeminal motoneurons in chick embryo. 11. Ganglion axon ingrowth guides motoneuron migration. J. Comp. Neurol. 213:344-349. NORNES,H. 0. and DAS, G. D. ( 1974). Temporal pattern of neurogenesis in spinal cord of rat. I. An autoradiographic study-time and sites of origin and migration and settling patterns of neuroblasts. Brain Res. 73:12 1-1 38. K. ( 1989). Migration of imONO,K. and KAWAMURA, mature neurons along tangentially oriented fibers in the subpial part of the fetal mouse medulla oblongata. Exp. Bruin Rex 78:290-300. J. L., and WELLS, OPPENHEIM,R. W., MADERDRUT, D. J. ( 1982). Cell death of motoneurons in the chick embryo spinal cord. VI. Reduction of naturally occurring cell death in the thoracolumbar column of Terni by nerve growth factor. J . Comp. Neurol. 210: 174189. PETRAS, J. M. and FADEN,A. I. (1978). The origin of sympathetic preganglionic neurons in the dog. Bruin Rex 144:353-357. PHELPS, P. E., BARBER,R. P., and VAUGHN,J. E. ( 1991). Embryonic development of choline acetyltransferase in thoracic spinal motor neurons: somatic and autonomic neurons may be derived from a common cellular group. J. Camp. Neurol. 307:77-86. PRASAD,A. and HOLLYDAY:M. ( 1991). Development and migration ofavian sympathetic preganglionic neurons. J. Comp. Neurol. 307:237-258. RAKIC,P. ( 1972). Mode of cell migration ofthe superficial layers of fetal monkey neocortex. J. Comp. Neurol. 1456 1-84. RAKIC,P. ( 1985). Contact regulation of neuronal migration. In: The Cell in Contact, G. M. Edelman and J, P. Thiery, Eds., Wiley & Sons, New York, pp. 6791. RAIUC,P. ( 1990). Principles of neural cell migration. Experientiu 46:882-89 1. R A M ~Y N CAJAL, S. (1960). Studies in Vertebrate Neurogenesis (translated by L. Guth). Charles C Thomas, Springfield. C . W., and ZIGMOND,R. E. UNDO, T. A., BOWERS, ( 1981) . Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. J. Comp. Neurol. 196:73-83.
822
Markham and Vaughn
RUBIN,E. ( 1985a). Development of the rat superior cervical ganglion: ganglion cell maturation. J. Neurosci. 5573-684. RUBIN,E. ( 1985b).Development of the rat superior cervical ganglion: initial stages of synapse formation. J. Neurosci. 5697-704. RUBIN,E. ( 1985~).Development ofthe rat superior cervical ganglion: ingrowth of preganglionic axons. J. Neurosci. 5 6 85 -696. VAUGHN,J. E. and GRIESHABER, J. A. (1973). A morphological investigation of an early reflex pathway in developingrat spinal cord. J. Cump. Neurol. 148: 177210.
VAUGHN,J. E., PHELPS,P. E., YAMAMOTO, M., and BARBER, R. P. ( 1990). Association neurons of embryonic spinal cord express the cell surface glycoprotein TAG-1. Soc. Neurosci Abstr. 16:1005. VERA,P. L., ELLENBERGER, H. H., HASELTON, J. R., HASELTON, C. L., and SCHNEIDERMAN, N. (1986). The intermediolateral nucleus: an “open” or “closed” nucleus? Bruin Res. 386:84-92. WINDLE,W. F. and BAXTER,R. E. ( 1936). Development of reflex mechanisms in the spinal cord of albino rat embryos. Correlation between structure and function, and comparisons with the cat and the chick. J. Comp. Neurol. 63: 189-209.
SUMMARY The displacement of immature neurons from their place of origin in the germinal epithelium toward their adult positions in the nervous system appears to involve migratory pathways or guides. While the importance of radial glial fibers in this process has long been recognized, data from recent investigations have suggested that other mechanisms might also play a role in directing the movement of young neurons. We have labeled autonomic preganglionic cells by microinjections of horseradish peroxidase (HRP) into the sympathetic chain ganglia of embryonic rats in order to study the migration and differentiation of these spinal cord neurons. Our results, in conjunction with previous observations, suggest that the migration pattern of preganglionic neurons can be divided into three distinct phases. In the first phase, the autonomic motor neurons arise in the ventral ventricular zone and migrate radially into the ventral horn of the developing spinal cord, where, together with somatic motor neurons, they form a single, primitive motor column (Phelps P. E., Barber R.P., and Vaughn J. E. (1991). J. Comp. Neurol. 307:77-86). During the second phase, the autonomic motor neurons separate from the somatic motor neurons and are displaced dorsally toward the intermediate spinal cord. When the preganglionic neurons reach the intermediolateral (IML) region, they become progressively more multipolar, and many of them undergo a change in alignment, from a dorsoventral to a
mediolateral orientation. In the third phase of autonomic motor neuron development, some of these cells are displaced medially, and occupy sites between the IML and central canal. The primary and tertiary movements of the preganglionic neurons are in alignment with radial glial processes in the embryonic spinal cord, an arrangement that is consistent with a hypothesis that glial elements might guide autonomic motor neurons during these periods of development. In contrast, during the second phase, the dorsal translocation of preganglionic neurons occurs in an orientation perpendicular to radial glial fibers, indicating that glial elements are not involved in the secondary migration of these cells. The results of previous investigations have provided evidence that, in addition to glial processes, axonal pathways might provide a substrate for neuronal migration. Logically, therefore, it is possible that the secondary dorsolateral translocation of autonomic preganglionic neurons could be directed along early forming circumferential axons of spinal association interneurons, and this hypothesis is supported by the fact that such fibers are appropriately arrayed in both developmental time and space to guide this movement. Keywords: motor neuron development, autonomic motor neurons, somatic motor neurons, gliophilic and neurophilic migrations, H R P retrograde labeling.
INTRODUCTION
appreciated for many years. Virtually all neurons are generated in germinal regions that are some distance removed from their eventual adult loca-
The importanceof neuronal migration in the developing organization of the nervous system has been Received April 15, 1991; accepted July 15, 1991 Journal of NeurobiologY. VOl. 22, No. 8, PP. 8 1 1-822 (199 1) tC 199 I John Wiley & Sons. Inc. CCC 0022-3034/9 I/OSOCr 11-12$04.00
* Present address: Department of Radiological Sciences, School of Medicine, B2-086 CHS, University of California. Los Angeles, Los Angeles, California 90024- I 72 1. To whom correspondence should be addressed. 81I
812
Markham and Vaughn
tions, and many such cells must migrate relatively long distances to reach their ultimate sites of residence. For nearly two decades, a dominant hypothesis of neuronal migration has stressed the role of radial glia as migratory pathways or guides (Rakic, 1972; Hatten, 1990). Much support for this hypothesis has been forthcoming from investigations of cortical structures (see Rakic, 1985, 1990; McConnell, 1988 for reviews). However, the topic of neuronal migration has not been pursued in as much detail in noncortical regions of the central nervous system (CNS). In many of these regions, the orientation of radial glia is often incongruent with observed neuronal migratory routes, suggesting that alternative mechanisms might also be important for directing the movements of nerve cells (Bourrat and Sotelo, 1988,1990; Moody and Heaton, 1983a,b,c;Rakic, 1985; Ono and Kawamura, 1989). In the embryonic spinal cord, radial glial fibers extend in the transverse plane between the ventricular zone and the pial surface (Choi, 1981; Misson, Edwards, Yamamoto, and Caviness, 1988;Ram6n y Cajal, 1960). If it is assumed that radial glia are involved in the migration of spinal cord neurons toward their adult positions, one would predict that subpopulations of these nerve cells would be generated at their appropriate dorsoventral level, and that they then would migrate directly laterally from the ventricular zone or germinal epithelium into the developing intermediate zone (e.g., Altman and Bayer, 1984; Nornes and Das, 1974). However, not all neuronal populations in the mammalian spinal cord appear to follow a strictly radial migration pattern throughout their development. The sympathetic preganglionic neurons of the adult rat, for example, are located within several nuclei in the intermediate region (lamina VII) of the spinal cord (Petras and Faden, 1978;Rando, Bowers, and Zigmond, 1981), but these cells do not appear to simply migrate radially from the intermediate part of the ventricular zone. Previous studies of developing spinal cord suggest that autonomic and somatic motor neurons are generated over the same time periods (Barber, Phelps, and Vaughn, 1989), and that both subclasses of motor neurons may be clonally related (Leber, Breedlove, and Sanes, 1990). Moreover, a recent immunocytochemical study of cholinergjc neurons in developing rat spinal cord suggested that somatic and autonomic motor neurons initially form a single, primitive motor column within the developing ventral horn (Phelps, Barber, and Vaughn, 1991 ) . Earlier
studies suggested that this also might be the case in the chick (Levi-Montalcini, 1950; Oppenheim, Maderdrut, and Wells, 1982), but a more recent examination of embryonic chick spinal cord indicated that many of the preganglionic neurons (Terni cells) “maintain medial positions and do not migrate laterally to join a common motor column before initiating a dorsal migration” (Prasad and Hollyday, 1991) . In contrast, choline acetyltransferase ( ChAT) immunoreactive somata in the primitive motor column appear to be the only source of ChAT-positive fibers in the ventral roots of early rat embryonic spinal cord. These fibers innervate both somatic muscles and the sympathetic chain ganglia, and this implies that both somatic and preganglionic motor neurons are located in the primitive motor column of this species (Phelps et al., 1991 ). However, using ChAT immunocytochemistry, it was not possible to demonstrate whether these two subclasses of motor neurons were intermixed or, alternatively, whether they were segregated in different parts of the primitive motor column, because both cell types were stained by this method. Consequently, in the present investigation, we examined the development of identified preganglionic neurons using horseradish peroxidase (HRP) retrograde labeling from microinjections into the sympathetic chain ganglia of rat embryos. The results provided evidence that preganglionic autonomic neurons were initially intermixed with somatic motor neurons within the primitive motor column of the ventral spinal cord, and that these cells subsequently underwent a secondary, dorsolateral translocation into the intermediate spinal cord. This secondary translocation was followed by a tertiary movement of some autonomic motor neurons to give rise to the more medially-located preganglionic nuclei. These results confirmed the secondary and tertiary displacements of preganglionic sympathetic neurons as suggested by the independent method of ChAT immunocytochemistry (Phelps et al., 1991) .
METHODS
Experimental Animals
Female Sprague-Dawley rats were placed in the cages of males overnight. The following morning, vaginal smears were prepared and, if they were sperm positive, that day
Development of Preganglionic Neurons
813
Figure 1 Transverse sections of E13$ thoracic spinal cord. ( a ) Four retrogradely-labeled preganglionic neurons (arrowheads) are shown distributed over much of the primitive motor column (dashed outline) of the spinal cord (solid outline). The axons of these neurons project from the ventral root and then turn medially (arrows) toward the injection site (asterisk)in the sympathetic chain. The morphology of three of these neurons (arrows) is more readily apparent in a photomontage shown at higher magnification (b). The fourth neuron (arrowhead) was too far out of the primary focal plane for its focused image to be incorporated in the montage. In another specimen (c) , retrogradely-labeled preganglionic neurons (arrow) are found scattered throughout the central region ofthe primitive motor column (dashed outline), indicating that they are not yet segregated from somatic motor neurons at this age. No HRP-labeled fibers were observed coursing toward sites of developing somatic muscle in these, or any other, specimens illustrated in this study. Abbreviations: RP = roof plate; V = ventricle; FP = floor plate; VT = vertebra; DRG = dorsal root ganglia. Scale bars: a, c = 100 pm; b = 30 p m .
was designated as embryonic day 1 ( E l ) . Since some investigators designate the day of insemination as EO (e.g., Rubin, I985a,b,c), all data referred to in this report have been converted to conform with the E l staging method in order to facilitate comparisons among different studies. At appropriate ages ( E 1 3-El 6), pregnant rats were deeply anesthetized with halothane using a precision vaporizer, and the embryos were removed via cesarian section. The amniotic sacs containing the em-
bryos were kept intact and placed in chilled (4"C), oxygenated, Gey's balanced salt solution ( RSS; Freshney, 1987). All specimens were prepared for HRP injection within 3 h of delivery. Sixteen to forty-eight embryos from each age were analyzed in this study.
HRP Injections The thoracic region of each embryo, including the spinal cord and surrounding tissues, was isolated and cut into
814
Markham and Vaughn
Figure 2 Transverse sections of E l 4 thoracic spinal cord. At low magnification (a), retrogradely-labeied autonomic preganglionic neurons (arrow) are observed to be clustered in the dorsolateral region of the primitive motor column (dashed outline). The bipolar somata of the preganglionic neurons are oriented dorsoventrally, and their axons can be seen exiting the
Development of Preganglionic Neurons two transverse slices, each approximately 1 mm thick. Using a Picospritzer (General Valve), small amounts of HRP (Sigma, Type VI, 30%in distilled H 2 0containing 2% dimethylsulfoxide [ DMSO] ) were pressure injected from 10-20 pm tip diameter glass micropipettes that were lowered vertically through the region of the sympathetic chain ganglia. Subsequent to injection, the tissue was placed in oxygenated BSS and “cultured” for 4-6 h at room temperature. Following HRP uptake and transport, the specimens were fixed for 12- 18 h in a mixture of 2.5% glutaraldehyde, 1% paraformaldehyde. and 4% sucrose in 0.1 M Soretisen’s phosphate buffer (pH = 7.4), rinsed several times in phosphate buffer, and cryoprotected in 30% sucrose in buffer overnight. Coronal (transverse) sections were cut at a thickness of 40 pm with a cryostat, and stored in 24-well tissue culture plates containing phosphate buffer. Sections were serially mounted on gel-coated slides, allowed to dry, and processed for HRP histochemistry. This method is essentially similar to that used in a previous study ofthe embryonic development ofrat superior cervical ganglion (Rubin, 1985,a,b,c). Although the reliability of this method has been demonstrated, the following procedures were conducted to control against the possibility that somatic motor axons coursing lateral to the sympathetic chain might become labeled from the ganglionic HRP injections: ( 1 ) tracer was injected directly into the ventral roots, and this did not result in the consistent labeling of any neurons in the spinal cord, suggesting that the HRP was not taken up efficiently by fibers of passage; (2) HRP was injected directly into somatic muscles of E l 4 and older specimens, and these injections labeled a population of neurons distinct both in location and morphology from cells labeled by injections into the sympathetic chain ganglia; and ( 3 ) injections into the chain ganglia resulted in a different pattern and course of HRP-labeled, peripheral nerve fibers (see Figs. 1, 2 ) than did injections into somatic muscle.
Histochemical Processing Prior to HRP processing, the slide-mounted sections were incubated for 20 min in a solution of 0.3%H202 and 0.1% sodium azide in Tris-buffered saline (TBS) in an attempt to suppress endogenous peroxidase activity (Li, Ziesmer, and Lazcano-Villareal, 1987). HRP-con-
815
taining neurons were visualized by first preincubating the sections for 10 min in a solution of 0.6Y0 CoCl,, 0.4% Ni( NH4)2 6H20and 0.06% diaminobenzadine (DAB) to increase the intensity of the final reaction product (Adams, 1981). After rinsing in buffer, the specimens were transferred to a solution of 0.06% DAB/0.007% H,O, for 10-15 min, rinsed in buffer, dehydrated, and coverslipped.
RESULTS
On E 13, relatively few differentiating neurons are present in the spinal cord; virtually all of these cells are located in a single cell column in the developing ventral horn ( Altman and Bayer, 1984; Phelps et al., 1991). Moreover, the sympathetic chain ganglia are just beginning to form, and only a small number of axonal growth cones from autonomic motor neurons reach the ganglia (Rubin, 1985a,b). However, significant changes in spinal cord organization occur between E l 3 and E l 4 (e.g., see below and Phelps et al., 1991), and we therefore injected specimens late on E 13 (E 134) in order to determine if somatic and autonomic motor neurons are initially intermixed, or, altetnatively, whether they are already segregated into distinct subgroups within the primitive motor column at this age. The E13; results indicated that somatic and preganglionic autonomic motor neurons probably are not segregatedinto distinct pools during the early primitive motor column stage (Fig. I ) . By E14, most of the preganglionic axons have reached the sympathetic chain (Rubin, 1985b), and microinjections of HRP into the ganglta resulted in robust labeling of cells in the ventral intermediate zone of the spinal cord. Although it is known that preganglionic autonomic and somatic motor neurons form a single, primitive motor column at this age (Phelps et al., 1991 ), the autonomic neurons were clearly segregated from the
spinal cord (solid outline) into the ventral root (small arrowheads) on their way toward the HRP-injected sympathetic chain (large arrowhead). The ventral and ventromedial part of the primitive motor column in panel a displays no HRP-labeled cells, but in specimens where HRP was injected directly into somatic muscles this part of the column exclusively contained retrogradely-labeled neurons. In a section from another specimen ( b )where only a few cells are labeled (asterisk), the bipolar nature of the preganglionic neurons is more evident. The cells possess ventrally-directed axons (small arrowheads), and dendritic processes (e.g., arrow) that project dorsally toward the intermediate region of the spinal cord. Abbreviationsas in Figure 1. Scale bars: a = 100 p m ; b = 50 pm.
816
&larkham and Vaughn
Figure 3 Transverse sections of E 15 thoracic spinal cord. ( a ) As compared with E 14, retrogradely-labeled preganglionic neurons on E 15 are located considerably more dorsal to the somatic motor neurons of the ventral horn (VH, dashed outline), and they have reached the level of the intermediate gray matter to form the intermediolateral nucleus (IML). At this age, medially-located preganglionic neurons (e.g., arrow, a ) are seen for the first time in the region between the IML and the ventricle ( V ) . (b-d) Some of the dorsal dendrites of preganglionic neurons in the vicinity ofthe intermediate gray matter have become reoriented into the mediolateral plane (e.g., arrows in b) . In addition, labeled cell bodies (arrowhead, c) are often found in alignment with these mediolateral dendrites. The dendrites of a few neurons (e.g., arrow in d ) can sometimes be seen extending medially through the ventricular zone toward the ventncular surface ( V ). Abbreviations: DH = dorsal horn; MZ = marginal zone; VH = ventral horn. Dorsal is up, and ventral is down in a-d. Scale bars: a = 100 wm; b-d = 50 pm.
Development of Pregunglionic Neurons
somatic motor neurons, with the former being located exclusively in the dorsolateral portion of the developing motor column [Fig. 2 ( a )1. The axons of the preganglionic neurons generally coursed along the border between the intermediate and margmal zones before exiting to the ventral root, but some were also noted to course more medially amongst the somatic motor cells before reaching their ventral root exit [Fig. 2( b)] . Many somata of preganglionic neurons on El 4 were clustered at the interface between the intermediate and marginal zones; these cells had an essentially bipolar morphology that displayed a strict dorsoventral orientation [Fig. 2(b)]. Injections of somatic muscle at this developmental stage labeled cells that were restricted to the ventral and ventromedial aspects of the primitive motor column (not shown). On E 15, the position of the preganglionic neurons had shifted further dorsally, and many of the cells exhibited multiple dendritic processes. The alignment of cells located in progressively closer proximity to the intermediate gray matter changed from a dorsoventral to a predominantly mediolatera1 orientation (Fig. 3). Moreover, on E l 5 , a few labeled cells were seen for the first time in a medial location between the intermediolateral ( TML) horn and the ventricular region. In addition, a number of individual preganglionic neurons within the IML possessed dendrites that extended medially toward, and often into, the ventricular zone [ e.g., Fig. 3 (d)] . By E 16, virtually all preganglionic neurons had reached the intermediate gray matter, and the number of such cells found in medial locations was more numerous that on E l 5 (Fig. 4). Furthermore, the number of mediolaterally-directed preganglionic dendrites increased between E 15 and E16, and the first indications of dendritic bundling, a characteristic of adult preganglionic neurons (Barber et al., 1984: Vera, Ellenberger, Haselton, Haselton, and Schneiderman, 1986; Davidoff, Galabov, and Bergmann, 1989), became apparent on E 16. Some of the preganglionic neurons at E 16 also extended dendrites into the developing lateral funiculus of the marginal zone (Fig. 4). In short, most of the preganglionic neurons displayed significant morphological differentiation by E l 6, and they had many of the features of mature preganglionic cells. Taken together, the observations of preganglionic neuronal translocations made at each age with retrograde HRP-labeling in the present study substantiate previous findings reported on
81 7
the independent basis of ChAT immunocytochemistry (Phelps et al., 1991 ). DISCUSSION
In the adult mammalian spinal cord, sympathetic preganglionic neurons are located in a narrow band of the intermediate gray matter (lamina VII). Within this lamina, preganglionic neurons are found primarily in the IML nucleus; however, depending on spinal level, a significant number of cells are also located in the intercalated and central autonomic nuclei, as well as in the lateral funiculus (Petras and Faden, 1978; Rando et al., 1981). Given the distinctive mediolateral alignment of these nuclei, it might be expected that, duringdevelopment, preganglionic neurons would be generated in the intermediate part of the ventricular zone, and then migrate laterally into the various sympathetic subnuclei. It is conceivablethat such a direct migration could occur along radial glial fibers, since they are in proper alignment to guide such movements (Choi, 1981; Misson et al., 1988) . However, the results of the present study, in conjunction with those of other investigators, reveal that this hypothesis is inconsistent with the observed development of autonomic preganglionic neurons in the mammalian spinal cord. Specifically, these results suggest that the migration of autonomic preganglionic neurons can be divided into distinct stages, and that elements other that radial glia are likely to be involved in directing the movement of these cells during certain stages of their migration. Both somatic and autonomic motor neurons appear to arise in the ventral ventricular zone (Leber et al., 1990; Phelps et al., 1991); in rats, at least, these two cell populations then appear to migrate into a single primitive motor column in a pattern parallel to the alignment of radial glia (Phelps et al., 1991). The recent demonstration that somatic and autonomic motor neurons appear to be clonally related (Leber et al., 1990), in conjunction with our E13t data showing retrogradely-labeled cells distributed throughout the primitive motor column, indicate that the two populations of neurons are probably intermixed at the early primitive motor column stage. Subsequently, by E 14, the autonomic motor cells separate from the somatic motor neurons and begin a secondary, dorsal translocation toward the intermediate region of the spinal cord. The direction of this trans-
818
Markham and Vaughn
Figure 4 Transverse sections of E 16 thoracic spinal cord. (a) A11 retrogradely-labeled preganglionic neurons are located in the intermediate gray matter on E16, with the majority of
Development of Preganglianic Neurons
location is perpendicular to the alignment of radial glial fibers, indicating that such structures are not likely to be involved in the dorsal movement of preganglionic neurons. Thus, some other element( s) must be responsible for this phase of their migration. One possible substrate for the dorsal migration of preganglionic neurons is provided by the early forming circumferential axons of spinal cord association interneurons. In the rat, these axons course ventrally from the developing dorsal horn and, at appropriate developmental stages, they are located along the preganglioniccell migration route at the interface of the intermediate and marginal zones (e.g., Windle and Baxter, 1936; Vaughn and Grieshaber, 1973; Vaughn, Phelps, Yamamoto, and Barber, 1990). Several investigators have provided evidence that the movement of neurons along axonal pathways, termed “neurophilic” migration by Rakic ( 1990), occurs in certain regions of the developing CNS (Moody and Heaton, 1983a,b,c; Rakic, 1985; Bourrat and Sotelo, 1988;Ono and Kawamura, 1989). For example, Moody and Heaton (1983a,b,c) have demonstrated that trigeminal motor neurons in the brain stem originate in the medial column, become reoriented to lie parallel to the pial surface, and then migrate laterally in close association with axonal bundles. These authors also have reported the presence of numerous adherens junctions between the migrating trigeminal neurons and individual axons of the fiber pathway. Elimination of the axons prior to migration appeared to prevent the tangential movement of the motor neurons (Moody and Heaton, 1983b,c). Thus, there is precedence for the idea that axonal pathways can provide guides for neuronal migration. Although the location and orientation of association fibers suggest that they could provide a substrate for the dorsal migration of preganglionic neurons, a determination of
819
whether or not they play a role in this process awaits further study. The observation that the preganglionic autonomic neurons end their dorsal translocation in the intermediate gray matter, and that some of these cells then appear to extend medially, suggests that this region of the spinal cord may contain cues that play a role in the reorientation of preganghonic neurons from a dorsoventral to a predominantly mediolateral alignment. One such cue may involve the emergence of mediolateral dendrites from preganglionicneurons since the development of these processes temporally coincideswith the appearance of centrally-located autonomic neurons between the IML and the ventricular zone. This suggests that the medially-located cells might become displaced from the IML during the extension of dendrites, or, alternatively, that both dendrites and somata may be guided by radially-aligned glial processes during this tertiary phase of autonomic motor neuron development. In conclusion, the migration ofpreganglionicautonomic motor neurons from the ventral ventricular zone into the anlage of lamina VTI of the mature spinal cord appears to involve a series of discrete steps that are summarized in Figure 5. In addition to their relevance for the issue of cell migration, however, the present results raise several questions with respect to other aspects of sympathetic preganglionic neuronal development. For example, while it is clear from the present study that axogenesis of preganglionic neurons is well under way at least as early as the primitive motor column stage (see also Rubin, 1985c), it remains to be determined whether the initial segregation of autonomic and somatic motor neurons that occurs between E13f and El4 involves an interaction of their respective axons with peripheral targets. Moreover, apparent correlations between cell mi-
them being found in the intermediolateral nucleus (IML). Their dendrites are generally oriented in the mediolateral plane, extending laterally into the developing marginal zone (MZ), and medially toward the ventricle (V). At higher magnification ( b ) , the medially-directed dendrites (small arrows) of the preganglionic neurons are seen to be grouped together in bundles, and neuronal somata (e.g.. arrow) are sometimes observed to be enmeshed in these processes. Dendrites (e.g., arrowhead) extending into the developing lateral funiculus of the marginal zone (MZ) are seen for the first time at this age. In addition, an increase in the number of more medially-located preganglionic neurons (arrows, c) on E 16, as compared with earlier ages, is readily apparent in a section from another specimen. These medial cells probably represent the precursors of the intercalated and central autonomic neurons that are characteristic of mature animals. Abbreviations: DH = dorsal horn; VH = ventral horn; V = ventricle. Scale bars: a = 100 pm; b-d = 50 fim.
820
Markham and Vaughn
El3
El 5
El6
Figure 5 Summary of the migration patterns of autonomic preganglionic neurons in embryonic rat spinal cord. Both somatic and autonomic motor neurons are generated between E 1 1- 12 (Barber et al., 1989) , and appear to migrate radially from the ventricular zone (VZ) into the developing ventral horn of the intermediate zone (IZ), where they form a single, primitive motor column (stippled area) by El 3 (Phelps et al., 1991 ). At E13i the autonomic and somatic motor neurons appear to be intermixed within the primitive motor column. By E 14, the preganglionic autonomic neurons have segregated from the somatic motor neurons, and form a distinct population in the dorsolateral region of the primitive column (stippled area). At this age, many of the dorsoventrally-orientedpreganglionic neurons are clustered at the interface between the intermediate zone and the developing marginal zone (MZ). As the preganglionic autonomic neurons approach the intennediolateral region (stippled area) on E 15, the cells become progressively reoriented into the mediolateral plane. At this age, a few of the neurons also begin a medially-directed displacement toward the ventricular zone. By E 16. the majority of preganglionic neurons (stippled area) are oriented in the mediolateral plane, and their dendntic processes extend medially toward the receding ventricular zone, as well as laterally into the developing marginal zone. As in the mature spinal cord, the majority of cell bodies on E 16 are located within the intermediolateral region, but a significant number of cells are also present in more medial locations between the IML and the central canal. Note that the sites occupied by somatic motor neurons are indicated by the open part ofthe primitive motor column on E l 4 and by the open, outlined areas of the ventral horn on El 5 and 16.
gration and dendntic rearrangements (see also Phelps et al., 199 1; Markham, Phelps, and Vaughn, 1990) during the development of sympathetic preganglionic neurons, suggest that these two events may share common epigeneticdeterminants that remain to be discovered. This research was supported by grants NS25784 and NS 18858 from the National Institute of Neurological Disorders and Stroke. The author; thank Robert Barber for his help with various aspects of this research, Drs. Arlene Chiu, and Patricia Phelps for helpful discussions,
and Marilyn Ashley, Lynn Rrennan, Mariko Lee, and Christine Vaughn for their excellent technical assistance.
REFERENCES ADAMS,J. C. ( 1 98 1). Heavy metal intensification of DAB-based HRP reaction product. J. Histochem. Cytochem. 29:775. ALTMAN,J. and BAYER, S. A., Ed. ( 1984). TheDevelopment of the Rut Spinal Cord, Springer-Verlag, Berlin. BARBER,R. P., PHELPS,P. E., HOUSER,C. R., CRAW-
Development of Pregunglionic Neurons FORD, G. D.,
SALVATERRA, P. M., and VAUGHN, J. E.
( 1984). The morphology and distribution of neurons
containing choline acetyltransferase in the adult rat spinal cord: an immunocytochemical study. J. Comp. Neurol. 229:329-346. P. E., and VAUGHN,J. E. BARBER,R. P., PHELPS, ( 1989). Generation patterns of spinal sympathetic and parasympathetic neurons in the rat spinal cord. SOC.Neurosci. Abstr. 15588. F. and SOTELO,C. (1988). Migratory pathBOURRAT, ways and neuritic differentiation of inferior olivary neurons in the rat embryo. Axonal tracing study using the in vitro slab technique. Dev. Brain Rex 39:19-37. BOURRAT,F. and SOTELO,C. ( 1990). Migratory pathways and selective aggregation of the lateral reticular neurons in the rat embryo: a horseradish peroxidase in vitro study, with special reference to migration patterns of the precerebellar nuclei. J. Comp. Neurol. 294: 1- 13. CHOI, B. H. ( 1981). Radial glia of developing human fetal spinal cord: Golgi, immunohistochemical and electron microscopic study. Dev. Bruin Res. 1:249267. DAVIDOFF, M. S., GALABOV, P. G., and BERGMANY, M. ( 1989). The vegetative network in the thoracolumbar spinal cord of the guinea pig: a comparison of the distribution of AChE-enzyme activity and choline acetyltransferase-like immunoreactivity. J. Hirnforsch. 30:707-7 17. FRESHNEY, R. I., Ed. ( 1987). Culture ofAnimal Cells:A Manual of Basic Technique, Wiley-Liss, New Y ork. HATTEN, M. E. (1990). Riding the glial monorail: a common mechanism for glial-guided neuronal migration in different regions of the developing mammalian brain. Trends Neurosci. 13:179- 184. S. M., and SANES,J, R. LEBER,S. M., BREEDLOVE, ( 1990). Lineage, arrangement, and death of clonally related motoneurons in chick spinal cord. J. Neurosci. 10:245 1-2462. LEVI-MONTALCINI, R. (1950). The origin and development of the visceral system in the spinal cord of the chick embryo. J. Morphol. 86:258-284. LI, C.-Y., ZIESMER,S., and LAZCANO-VILLAREAL, 0. ( 1987). Use of azide and hydrogen peroxide as an inhibitor for endogenous peroxidase in the immunoperoxidase method. J. Histochem. Cytochem. 35: 14571460. MARKHAM, J. A., PHELPS,P. E., and VAUGHN,J. E. ( 1990). Development of longitudinal dendritic bundles in rat spinal cord: analysis of cholinergic sympathetic preganglionic neurons. SOC.Neurosci. Abstr. 16:644. S. K. ( 1988). Development and decisionMCCONNELL, making in the mammalian cerebral cortex. Bruin Res. Rev. 13:1-23. MISSON,J.-P., EDWARDS,M. A., YAMAMOTO, M., and V. S. JR. ( 1988). Identification of radial CAVINESS, glial cells within the developing murine central ner-
821
vous system: studies based upon a new immunohistochemical marker. Dev. Bruin Rex 44:95-108. MOODY,S. A. and HEATON,M. B. ( 1983a). Ultrastructural observations of the migration and early development of trigeminal motoneurons in chick embryos. J. Comp. Neurol. 216:20-35. MOODY,S. A. and HEATON.M. B. (1983b). Developmental relationships between trigeminal ganglia and trigeminal motoneurons in chick embryo. I. Ganglion development is necessary for motoneuron migration. J. Comp. Neurol. 213~327-343. . MOODY,S. A. and HEATON,M. B. ( 1 9 8 3 ~ )Developmental relationships between trigeminal ganglia and trigeminal motoneurons in chick embryo. 11. Ganglion axon ingrowth guides motoneuron migration. J. Comp. Neurol. 213:344-349. NORNES,H. 0. and DAS, G. D. ( 1974). Temporal pattern of neurogenesis in spinal cord of rat. I. An autoradiographic study-time and sites of origin and migration and settling patterns of neuroblasts. Brain Res. 73:12 1-1 38. K. ( 1989). Migration of imONO,K. and KAWAMURA, mature neurons along tangentially oriented fibers in the subpial part of the fetal mouse medulla oblongata. Exp. Bruin Rex 78:290-300. J. L., and WELLS, OPPENHEIM,R. W., MADERDRUT, D. J. ( 1982). Cell death of motoneurons in the chick embryo spinal cord. VI. Reduction of naturally occurring cell death in the thoracolumbar column of Terni by nerve growth factor. J . Comp. Neurol. 210: 174189. PETRAS, J. M. and FADEN,A. I. (1978). The origin of sympathetic preganglionic neurons in the dog. Bruin Rex 144:353-357. PHELPS, P. E., BARBER,R. P., and VAUGHN,J. E. ( 1991). Embryonic development of choline acetyltransferase in thoracic spinal motor neurons: somatic and autonomic neurons may be derived from a common cellular group. J. Camp. Neurol. 307:77-86. PRASAD,A. and HOLLYDAY:M. ( 1991). Development and migration ofavian sympathetic preganglionic neurons. J. Comp. Neurol. 307:237-258. RAKIC,P. ( 1972). Mode of cell migration ofthe superficial layers of fetal monkey neocortex. J. Comp. Neurol. 1456 1-84. RAKIC,P. ( 1985). Contact regulation of neuronal migration. In: The Cell in Contact, G. M. Edelman and J, P. Thiery, Eds., Wiley & Sons, New York, pp. 6791. RAIUC,P. ( 1990). Principles of neural cell migration. Experientiu 46:882-89 1. R A M ~Y N CAJAL, S. (1960). Studies in Vertebrate Neurogenesis (translated by L. Guth). Charles C Thomas, Springfield. C . W., and ZIGMOND,R. E. UNDO, T. A., BOWERS, ( 1981) . Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. J. Comp. Neurol. 196:73-83.
822
Markham and Vaughn
RUBIN,E. ( 1985a). Development of the rat superior cervical ganglion: ganglion cell maturation. J. Neurosci. 5573-684. RUBIN,E. ( 1985b).Development of the rat superior cervical ganglion: initial stages of synapse formation. J. Neurosci. 5697-704. RUBIN,E. ( 1985~).Development ofthe rat superior cervical ganglion: ingrowth of preganglionic axons. J. Neurosci. 5 6 85 -696. VAUGHN,J. E. and GRIESHABER, J. A. (1973). A morphological investigation of an early reflex pathway in developingrat spinal cord. J. Cump. Neurol. 148: 177210.
VAUGHN,J. E., PHELPS,P. E., YAMAMOTO, M., and BARBER, R. P. ( 1990). Association neurons of embryonic spinal cord express the cell surface glycoprotein TAG-1. Soc. Neurosci Abstr. 16:1005. VERA,P. L., ELLENBERGER, H. H., HASELTON, J. R., HASELTON, C. L., and SCHNEIDERMAN, N. (1986). The intermediolateral nucleus: an “open” or “closed” nucleus? Bruin Res. 386:84-92. WINDLE,W. F. and BAXTER,R. E. ( 1936). Development of reflex mechanisms in the spinal cord of albino rat embryos. Correlation between structure and function, and comparisons with the cat and the chick. J. Comp. Neurol. 63: 189-209.
