Proc. Natl. Acad. Sci. USA Vol. 81, pp. 6245-6249, October 1984 Neurobiology

Developing dorsal root ganglion neurons require trophic support from their central processes: Evidence for a role of retrogradely transported nerve growth factor from the central nervous system to the periphery (dorsal root axons/sensory neurons/retrograde transport/trophism) HENRY K. YIP AND EUGENE M. JOHNSON, JR. Department of Pharmacology, Washington University School of Medicine, 660 South Euclid Avenue, St. Louis, MO 63110 Communicated by Oliver H. Lowry, June 11, 1984

ABSTRACT Injury to the peripheral processes produces a response." The lesser role of the central relative to the pe- profound cell loss (40-50%) in the dorsal root ganglion ofnew- ripheral process could be due, in part, to the disproportion- born rats. Although division of central processes produces lit- ate loss of axoplasmic volume between the central and pe- tie or no cellular change in sensory ganglion of adult animals, ripheral axons after axotomy because the central axons are no Information has been available on the effect of dorsal root of smaller diameter (7) and the flow of axoplasm is slower section in developing dorsal root ganglion. We show that 6 than in the large peripherally directed axons (8, 9). days after dorsal rhizotomy on newborn rats, there is a 50% Populations of developing neurons come under the influ- decrease in neuronal number in L5 dorsal root ganglion. A ence of their peripheral fields and are able to make compen- combined central and peripheral lesion of the sensory process satory adjustments according to the sizes of the afferent in- results in a greater decrease in neuronal number (70%). Both put they receive (10-12) and the efferent output they project of these effects can be prevented by the concomitant treatment (13-16). The adjustments are made by way of corresponding with nerve growth factor. We also demonstrate that 12'I-la- alterations in cell number and size within the developing beled nerve growth factor is retrogradely transported with neuronal population. The sensory ganglion is unique in the high selectivity from the spinal cord to the dorsal root ganglion sense that its target areas include both central neurons in the via the dorsal roots. The results indicate that trophic support spinal cord and the innervated organs in the periphery. De- for developing sensory neurons is provided through the central spite the lack of morphological alteration reported after sec- processes. This is presumably due to the uptake and retro- tioning of the central process in adult animals, it is possible grade transport of a trophic factor by the terminals of the cen- that developing DRG neurons receive trophic support from tral processes. The data suggest that nerve growth factor may both central and peripheral target zones. be the trophic factor. Nerve growth factor (NGF) has long been thought to act as a trophic factor transferring information from the periph- Dorsal root ganglion (DRG) neurons, developing as bipolar eral target organs to the innervating sympathetic and sensory neurons from the neural crest, are known to have two very neurons. Our recent findings (17) have demonstrated that similar processes. Both the central and peripheral processes eliminating the peripheral target influence by crushing the have structural characteristics of axons (1, 2). The axons of sciatic nerve in the newborn rat resulted in a substantial loss the spinal sensory neurons bifurcate in the DRG, sending (40-50%) of neurons in L5 DRG. Treatment with NGF pre- one branch in the dorsal root centrally toward the spinal cord vented the decrease in neuronal number of the axotomized and the other in the spinal nerve toward the peripheral tis- DRG. In the present study, we examine the effects ofcentral sues. Axonal injury of the peripheral branch in adult animals process axotomy on the survival of the neonatal neurons to leads to a profound reaction in the DRG neurons. In con- evaluate the relative roles of central and peripheral contacts trast, section of the dorsal roots proximal to the DRG in on the development of the DRG neurons and to determine if adult animals does not initiate significant morphological exogenous NGF can prevent the observed deleterious ef- change in the ganglion cells (3-5). fects of central axonal injury. One mechanism postulated to initiate neuronal response to It is generally accepted that the physiological effects of injury is that neurons depend on trophic substances coming NGF on target neurons are mediated via retrograde trans- from target cells for their survival and normal maintenance. port of NGF from the target to the neuronal cell body (18). If Interruption of the supply of the trophic substances would NGF were a factor involved in providing trophic support for then initiate the response (6). A reasonable explanation for DRG neurons through the central process, it would be pre- the nonresponsiveness of DRG neurons to sectioning of dor- dicted that 125I-labeled NGF (125I-NGF) should be specifical- sal roots would be that the central process plays no role in ly retrogradely transported from the terminal fields of senso- the trophic support of DRG neurons or, alternatively, that ry neurons in the spinal cord to the DRG neurons. The data the loss of a postsynaptically derived trophic substance from presented here document that 1251-NGF is so transported. the central terminal zones can be sufficiently replaced by trophic agents from the peripheral target cells to keep the MATERIALS AND METHODS neurons adequately maintained. In the latter case it would be Experimental Animals. Litters of newborn, postnatal day assumed that the loss of supply of trophic substance by the (PND)-0 (-6.5 g), and adult (250-300 g) Sprague-Dawley centrally directed axons in the dorsal roots was quantitative- rats (Chappel, St. Louis, MO) of either sex were used in this ly relatively small and thus the cell does not show an "injury study. They were kept in controlled temperature (230C + 2°C) and in a light/dark cycle of 14/10 hr with food and water The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviations: NGF, nerve growth factor; DRG, dorsal root gangli- in accordance with 18 U.S.C. §1734 solely to indicate this fact. on; PND, postnatal day. 6245 Downloaded by guest on September 30, 2021 6246 Neurobiology: Yip and Johnson Proc. NatL Acad Sci. USA 81 (1984)

ad libitum. Litter sizes were adjusted so that each had be- with isotonic saline followed by perfusion with 10% formal- tween 8 and 10 pups. in. The L5 DRG were dissected, dehydrated with ethanol, Surgical Procedures. The two operative procedures were and embedded in paraffin. Complete serial sections of 8 pim dorsal root section and sciatic nerve crush. For either proce- thickness were cut and stained with 0.1% toluidine blue. The dure, newborn rats were anesthetized by cooling and adult contralateral L5 DRG served as control for the axotomized animals were anesthetized with an intraperitoneal injection DRG. To estimate the total number of neurons in the L5 of chloral hydrate (350 mg/kg). For dorsal root section, a DRG, cells in every 10th section of each ganglion were laminectomy was performed at the level of T13 and L1 verte- counted at a magnification of 400x on coded slides. Only brae, exposing the L4 and L5 segments of the spinal cord. cells with at least one prominent nucleolus in the plane of The 5th lumbar (L5) dorsal roots were transected at the point section were counted. Viable neurons, containing prominent of entry into the spinal cord on the right side of the animal. Nissl substance and a large nucleus with one or more nucleo- The exposed spinal cord was covered with Gelfoam; the li, were easily identified. The correction factor for split nu- muscles and the skin were sutured. The right sciatic nerve cleoli was calculated according to Konigsmark (23). The sta- was exposed in midthigh and crushed for 40 sec with a pair of tistical significance of differences obtained among treated jeweler's forceps at the tendon of obturator internus. In the and untreated animals was evaluated by using a two-way sham-operated animals, both the dorsal root and the sciatic analysis of variance with Satterwaithe's correction of de- nerve were exposed but not lesioned. grees of freedom due to unequal group size (24). Preparation and Labeling of NGF. The 2.5S mouse NGF was prepared from submaxillary glands of adult male mice as RESULTS described by Bocchini and Angeletti (19). NGF solution was Effects of Central Lesion on the Neuronal Number in Ls prepared by dissolving lyophilized NGF in 0.9% phosphate- DRG. The right L5 dorsal roots of newborn rats were tran- buffered saline (Pi/NaCl) at a concentration of 1 mg/ml. The sected on the day of birth (PND-0) and the animals were sac- solution was divided into aliquots and kept frozen at -40C rificed on PND-5. The L5 DRG were dissected out and pre- until use. Subcutaneous injections of either 20 ,ug of NGF or pared for neuronal cell count. The unoperated, contralateral an equal volume of vehicle (P,/NaCl) were given to the neo- DRG served as control. Axotomy of the dorsal roots on nates 2 hr before the operation and continued daily through PND-0 of age caused a 50% decrease of neuronal number by PND-5. NGF and cytochrome c were labeled with 1 mCi of PND-5 (Table 1). There was no decrease of cell number in Na125I (specific activity, 13-17 mCi/pg, Amersham; 1 Ci = the sham-operated animals. 37 GBq) by the lactoperoxidase method of Marchalonis (20) Effects of Combined Central and Peripheral Lesions on the with minor modifications. After iodination, the 125I-NGF Neuronal Number in Ls DRG. The effects on the neuronal was dialyzed in 3 liters of Pi/NaCl (pH 7.4) to remove any number of a single central lesion were compared to those free Na125I. The specific activities of 125I-NGF and cyto- effects from combined central and peripheral lesion pro- chrome c preparations were -80 ,Ci/,g. duced by also crushing the sciatic nerves in the newborn Microiqjection Procedures. Spinal cord injections were rats. We have previously shown (17) that such a peripheral performed according to the method of Lasek (21) as modi- lesion alone results in a 40-50% loss of neurons within 48 hr fied by Griffin et al. (22). The L4 and L5 segments of anes- of lesion. In the present study, dorsal root transection result- thetized adult rats were exposed by laminectomy of the T13 ed in a 50% decrease of L5 DRG neurons within 6 days (Ta- and L1 vertebrae. 125I-NGF was injected by using a micropi- ble 1). In animals that received double lesions, the total num- pette (tip diameter of about 20 ,m) attached to a 10-,ul Ham- ber of L5 DRG neurons decreased to 31% of the unoperated ilton glass syringe and held by a micromanipulator. The dor- sal horn of L4 or L5 lumbar enlargement where the central processes of the DRG terminate was injected after stereotac- Table 1. Effect of axotomy of central/peripheral DRG process tically positioning the micropipette tip 1 mm lateral to the and concurrent treatment with NGF on the neuronal number midline and at a depth of 1 mm. Two 1-,l injections (3 x 106 in the L5 DRG of 6-day-old rats to 4 x 106 cpm/,ul) were spaced 0.5 mm apart. The injection Lesion of DRG % of time was about 10 min and the micropipette was left in the process Treatment Neurons in untreated same position for at least another 5 min before being re- L5 moved from the spinal cord. Injection of 125I-NGF into the Central Peripheral with NGF DRG, no. control nucleus gracilis (terminal area for long ascending sensory fi- - - - 7000 ± 260* (6) lOOtt bers from the DRG) was performed stereotactically at two - - + 7490 ± 240* (4) 107t different sites close to the caudal third of the area postrema. Sham - - 7351 ± 418 (3) 105* The wound was closed by suturing the paravertebral muscles + - - 3507 ± 443 (4) 50t and using 9-mm Michel clips for the skin. At intervals (3, 4, + - + 6564 ± 616 (4) 94* 6, 8, 12, 18, 24, 36, and 48 hr) after the injection, the animals + + - 2181 ± 323 (4) 31t were killed by decapitation and the L4 or L5 DRG of the + + + 7379 ± 352 (5) 105t injected and uninjected side were removed. The radioactiv- 0 first 24 hr after rats were ity of the ganglia from each side was counted in a Beckman Day (the birth) treated with 20 Pg of 2.5S mouse NGF 2 hr prior to surgery and the treatment was contin- -y-counter. ued for 4 more days. The central process of the right L5 DRG was Autoradiography. For the autoradiographic localization of cut at the point of entry into the spinal cord. Peripherally the right 1251-NGF in the DRG neurons, the paired L5 DRG were sciatic nerve was crushed at the tendon of obturator internus. Ani- fixed in 10% formalin. The tissues were dehydrated in grad- mals were killed on the 6th day and the corresponding L5 DRG was ed ethanol and embedded in Paraplast. Eight-micron sec- dissected out for histology. Neuronal counts were determined; each tions were coated with Kodak NTB-2 emulsion and exposed value is the mean ± SEM calculated for the number of animals in for 1 week at 4°C. They were then developed with Kodak D- parentheses. The numbers of neurons in the treated and untreated 19 developer and counterstained with toluidine blue for groups were compared by two-way analysis of variance. bright- *Numbers of neurons in the contralateral control L5 DRG (unoper- and darkfield light microscopy. ated). Morphometry. For the morphometric studies, littermates tEach of these groups is significantly different from the other two; P of PND-0 rats were injected subcutaneously with 20 gg of < 0.001. NGF on 5 consecutive days. The animals were fixed 24 hr *No significant difference was found between any two of these five after the last injection on PND-5 by transcardial perfusion groups; P > 0.25. Downloaded by guest on September 30, 2021 Neurobiology: Yip and Johnson Proc. NaiL Acad Sci USA 81 (1984) 6247 control (Table 1). Thus, the additional peripheral lesion led to a further 20% decrease in cell number. Compared to the animals that received only a central lesion, the neuronal EL4 number in the double-lesioned animals showed a statistically significant (P < 0.05) decrease of 38%. A. (L4 Dorsal Root Intact) Effects ofNGF Treatment. Newborn rats were treated with 3000 4 L 20 ,ug of NGF or vehicle (P,/NaCl) daily for 5 days. The animals were killed 24 hr after the last injection. Cell counts Cr 2000 T from animals with a central lesion treated with NGF showed %I., E no loss of neurons compared to controls-i.e., NGF pre- 0. 1000 vented the cell death. In animals that had both central and peripheral lesions, NGF likewise completely prevented the cell loss (Table 1). Thus, the administration of NGF com- R L R-L R L R-L pletely prevented the cell loss resulting from axotomy either of central or of central and peripheral axonal processes of the DRG. NGF had no effect on neuronal number in unoper- ated DRG. L4 Retrograde Transport of 2'sI-NGF from the Spinal Cord to the Ls DRG Neurons. The time course of accumulation of 1251 B. (L4 Dorsal Root Cut) radioactivity in the L5 DRG after unilateral injections of 1-2 3000r L4 '"I of 125I-NGF into the corresponding spinal cord segment of adult rats is shown in Fig. 1. The first significant difference cr 2000 - accumulated in the between the amount of radioactivity L5 .0 _ DRG on the injected side and the amount on the uninjected E side (P < 0.05) was detectable as early as 4 hr after the injec- a oo tion. The accumulation increased gradually and reached a maximum after 18 hr. There was a slight concurrent increase _ EX of radioactivity in the uninjected L5 DRG. In these rats, the R L R-L R L R-L distance from the point where the dorsal roots entered the spinal cord to the L5 DRG was estimated to be 2.5-3 cm. FIG. 2. Effect of surgical transection of central process on the This represented an approximate rate ofretrograde transport accumulation of radioactivity in L4 and L5 DRG. 125I-NGF (3 x 106 that of 1251- to 4 x 106 cpm) was injected unilaterally into the corresponding spi- of 6-7 mm/hr. In the animals received injections nal cord (dorsal horn) segments. (A) Both L4 and L5 dorsal roots NGF in the nucleus gracilis, there was a significant (P < were intact; (B) L4 dorsal root was transected and L5 dorsal root 0.05) accumulation of radioactivity in the L5 DRG 24 hr after remained intact. In each case the L4 and L5 DRG of the injected and the injection, 354 ± 98 cpm in the DRG from the injected uninjected side were removed for counting 18 hr after the injection. side compared to 94 ± 30 cpm in the DRG on the uninjected Each column represents the mean ± SEM of three or four animals. side (n = 8 in each group). R, right; L, left. Accumulation of radioactivity in L4 and L5 DRG If the accumulation of 125I-NGF in ipsilateral DRG after was statistically significant (*, P < 0.001), as determined by stu- injection to the spinal cord is indeed due to retrograde trans- dent's t test (two-tailed), only when the dorsal root was intact. port via the central process, it would be predicted that surgi- cal interruption of the central process would prevent the ac- Eighteen hours after injection with 125I-NGF in the dorsal cumulation of radioactivity. As expected, surgical transec- spinal cord, autoradiographic examination of ganglia from tion of the central processes from the DRG completely injected and uninjected sides of animals strongly supported abolished the difference in the accumulation of 125I-NGF be- the observation that the 125I-NGF is retrogradely transport- tween right and left L4 ganglia normally observed 18 hr after ed to the L5 DRG neurons on the injected sides. In the DRG the spinal cord injections (Fig. 2). Thus, disruption of the from the injected side there was a clear localization of radio- usual connection between the DRG and spinal cord by tran- activity in neurons (Fig. 3A); however, no clear localization secting the dorsal root abolished the accumulation of labeled ofradioactivity was observed over specific neurons from the NGF in the DRG on the injected side. uninjected sides. The majority of the radiographic activity was concentrated over the cytoplasm ofthe labeled neurons, 5000 with relatively little labeling in the nucleus (Fig. 3B). Al- though no attempt was made to determine precisely the pro- 4000 portion of neurons that were labeled after injection of 125I- NGF in the spinal cord, we estimated that -'20-30%o of the 3000L DRG neurons were labeled under these conditions. In addi- -J tion, linear arrays of autoradiographic grains could be seen at the proximal end of the DRG (Fig. 3A), indicating the E2000 , presence of transported radioactive material in the dorsal roots. No such labeling was observed at the distal end of the 1000_ DRG (Fig. 3A). The specificity of this transport mechanism for NGF was 4 8 12 16 20 24 28 32 36 48 examined by studying the ability of 125I-labeled cytochrome TIME IN HOURS c (1251-cytochrome c) to be transported retrogradely in the central process of the DRG under the same experimental FIG. 1. Time course of accumulation of radioactivity in the L5 conditions. An equal concentration of cytochrome c was io- DRG after unilateral injections of 125I-NGF (3 x 106 to 4 x 106 cpm) of into the corresponding spinal cord (dorsal horn) segment of adult dinated in the same manner as NGF. Administration 125I1 rats. *, Ipsilateral L5 DRG; A, contralateral L5 DRG. The L5 DRG cytochrome c into the spinal cord resulted in no accumula- of the injected and uninjected sides were removed for counting dif- tion of radioactivity in the L5 DRG on the injected side (Fig. ferent time periods after the injection. Each point represents the 4). The data in Fig. 4 show that the addition of a large excess mean + SEM of three or four animals. of nonradioactive NGF was able to compete with the labeled Downloaded by guest on September 30, 2021 6248 Neurobiology: Yip and Johnson Proc. NatL Acad Sci. USA 81 (1984) DISCUSSION The present experiments demonstrate that not only the con- nection of the cell body with the peripheral target (17) but also its connection with the central target are important for the survival and normal maintenance of the developing rat DRG neuron. This conclusion is supported by the observa- tion of neuronal loss in the L5 DRG after the dorsal root was sectioned in the newborn rats. In contrast to the previous findings (3-6) in which no effect on adult DRG neurons could be observed after dorsal rhizotomy, we found that the neuronal number in neonatal rat L5 DRG was reduced by 50% 6 days after the lesion was performed on PND-0. This difference between the response of newborn and adult ani- mals could be due to a shift in the balance of trophic influ- ence between the peripheral and central target terminals dur- ing the developmental process. As the animals mature, sensory neurons may receive more trophic support from peripheral target organs. The response of DRG neurons in adult animals to peripheral axotomy could be the result of a failure of these axotomized neurons to receive sufficient trophic support from their central terminal fields to remain in steady state. The fact that the same proportion of cell loss (50%) was observed in either peripherally (17) or centrally axotomized DRG suggests that immature DRG neurons de- pend equally on their peripheral and central target cells for their survival at this stage during development. This assump- tion was supported by the results showing that in animals that received both peripheral and central lesions there was a further decrease in cell number. Therefore, it seemed rea- sonable to assume that with the elimination of both axonal processes, the DRG neurons were deprived of trophic sup- port from both the peripheral and central terminal fields. It would be interesting to speculate that either the remaining 30% of the cells in the double-lesioned DRG had a longer FIG. 3. Darkfield photomicrograph of L5 DRG from the injected latency period of or the side of an animal that received an injection of 125I-NGF 18 hr previ- degeneration sources of trophic sup- ously in the corresponding spinal cord (dorsal horn) segment. (A) port for these cells were not interrupted by the lesions of the Labeled neurons in the ganglion, with labeled dorsal root indicated central and peripheral processes performed in these experi- by the arrow. Both ventral root and spinal nerve do not have label. ments. We have found (17) that 15-20% of neurons in L5 (x20.) (B) High-power photomicrograph showing intense labeling in DRG do not project into the sciatic nerve. In addition, some the DRG neurons. (x200.) DR, dorsal root; VR, ventral root; SN, DRG neurons apparently project their central processes into spinal nerve. the spinal cord through the ventral roots (25). The signals for the initiation of cell body responses after NGF and to block its transport effectively. This indicates axonal injury are not known (5). One of the possibilities pos- that the observed transport of 125I-NGF in the central proc- tulated is loss of trophic factor from target cells. NGF has ess of the L5 DRG was a specific receptor-mediated mecha- been suggested as the natural retrograde trophic agent pro- nism. vided by the peripheral target for some neural crest-derived neurons. Several lines of evidence have supported this hy- 5000r pothesis. First, presence of NGF antiserum in immature ani- mals leads to the destruction of sympathetic (26-28) and spi- 40001 nal sensory neurons (17, 27, 28). Second, administration of exogenous NGF during the period of naturally occurring cell cr- 30001 death reduces the loss of neurons in both sympathetic (29- 0 31) and sensory ganglia (32). Third, NGF is selectively taken -j up from the periphery and retrogradely transported to the 2000F E sympathetic (33) and sensory ganglia (34, 35). Furthermore, 0. previous studies have demonstrated that the effects of pe- 0001 ripheral axotomy on developing sympathetic (36, 37) and sensory neurons (17) can be prevented by NGF treatment. These results and the fact that treatment with NGF com- pletely prevented any cell loss caused either by central le- R L RL R L sion or by both central and peripheral lesion in the present FIG. 4. Accumulation of radioactivity in the L5 DRG 18 hr after study indicate that NGF can replace the trophic agent miss- unilateral injection of 1251-cytochrome c (3 x 106 cpm) or 125I-NGF ing after the transection of either or both axonal processes. (3 x 106 to 4 x 106 cpm) in the presence of excess, unlabeled NGF in The retrograde transport of NGF occurs in the peripheral the corresponding spinal cord (dorsal horn) segment. *, Thirty neurons nanograms of 125I-NGF; a, 30 ng of 125I-NGF and 4500 ng of NGF; process of sensory throughout life (34, 38). We o, 30 ng of 1251-cytochrome c. R, right; L, left. Statistically signifi- show that 125I-NGF can also be retrogradely, axonally trans- cant (P < 0.001), as determined by student's t test (two-tailed), ac- ported in the central process of the DRG neurons. 12 5NGF cumulation of retrogradely transported radioactivity is denoted by injected in the spinal cord or nucleus gracilis leads to a pref- an asterisk. erential accumulation of radioactivity in the DRG of the in- Downloaded by guest on September 30, 2021 Neurobiology: Yip and Johnson Proc. NatL Acad Sci USA 81 (1984) 6249 jected side. This transport is specific and receptor-mediated. 12. Coustantine-Paton, M. & Ferrai-Eastman, P. (1981) J. Comp. These properties and rate of the fastest transported 125j_ Neurol. 196, 645-661. NGF are the same as that described for the retrograde trans- 13. Cowan, W. M. & Wenger, E. (1967) J. Exp. Zool. 164, 265- port via the peripheral process. Our observation that the pe- 280. 14. Prestige, M. C. (1967) J. Embryol. Exp. Morphol. 18, 359-387. ripheral roots are not labeled after injection of 125I-NGF into 15. Hollyday, M. & Hamburger, V. (1976) J. Comp. Neurol. 170, the spinal cord complements the previous findings (35) in 311-320. which the central roots are unlabeled after injection of 1251_ 16. Pilar, G. & Landmesser, L. (1976) J. Cell Biol. 68, 339-356. NGF into a crushed sciatic nerve. Thus, retrograde transport 17. Yip, H. K., Rich, K., Lampe, P. & Johnson, E. M., Jr. (1984) of NGF from either the central or peripheral terminals leads J. Neurosci., in press. to accumulation in the cell body with little or no NGF enter- 18. Thoenen, H. & Barde, Y.-A. (1980) Physiol. Rev. 60, 1284- ing the other process. 1335. In summary, these results demonstrate that the central 19. Bocchini, F. & Angeletti, P. U. (1969) Proc. Natl. Acad. Sci. process plays a critical role in providing trophic support to USA 64, 787-794. 20. Marchalonis, J. J. (1969) Biochem. J. 113, 229-305. the developing sensory neurons. Exogenous NGF can, at 21. Lasek, R. J. (1968) Exp. Neurol. 21, 41-51. least for 5 days, replace the trophic support provided by the 22. Griffin, J. W., Drachman, D. B. & Price, D. B. (1976) J. Neur- central process. 1251-NGF can be transported efficiently obiol. 7, 355-370. from the terminal fields of sensory neurons in the spinal cord 23. Konigsmark, B. W. (1970) in Contemporary Research Meth- via the central processes to the neuronal cell bodies in the ods in Neuroanatomy, eds. Nauta, W. J. H. & Ebbsson, DRG. These results strongly suggest, but do not prove, that S. 0. E. (Springer, New York), pp. 315-340. the trophic support provided by the central process is medi- 24. Snedecor, G. W. (1957) Statistical Methods (The Iowa State ated by NGF. College Press, Ames, IA), pp. 291-328. 25. Coggeshall, R. E. (1979) Neurosurgery 4, 443-448. The authors would like to thank Ms. Patricia Lampe and Ms. Pa- 26. Levi-Montalcini, R. & Booker, B. (1960) Proc. Natl. Acad. tricia Osborne for their excellent assistance. This work was support- Sci. USA 46, 384-391. ed by National Institutes of Health Grants HL20604, NS18071, and 27. Gorin, P. D. & Johnson, E. M. (1979) Proc. Natl. Acad. Sci. NS07129. USA 76, 5382-5386. 28. Johnson, E. M., Gorin, P. D., Brandeis, L. D. & Pearson, J. (1980) Science 210, 916-918. 1. Ha, H. (1970) J. Comp. Neurol. 140, 227-240. 29. Hendry, I. A. & Campbell, J. (1976) J. Neurocytol. 5, 351-360. 2. Hulsebosch, C. E. & Coggeshall, R. E. (1983) J. Comp. 30. Kessler, J. A. & Black, I. B. (1980) Brain Res. 189, 157-168. Neurol. 265, 187-197. 31. Oppenheim, R. W., Maderdrut, J. L. & Wells, D. J. (1982) J. 3. Hane, W. K. & Hinsey, J. C. (1940) J. Comp. Neurol. 73,489- Comp. Neurol. 210, 174-189. 502. J. K. & J. W. J. 4. Carmel, P. W. & Stein, B. M. (1969) J. Comp. Neurol. 135, 32. Hamburger, V., Brunso-Bechtold, Yip, (1981) 145-166. Neurosci. 1, 60-70. 5. Cragg, B. G. (1970) Brain Res. 23, 1-21. 33. Hendry, I. A., Stoeckel, K., Thoenen, H. & Iversen, L. L. 6. Lieberman, A. R. (1974) in Essay on the Nervous System, eds. (1974) Brain Res. 68, 103-121. Bellaris, R. & Gray, E. G. (Clarendon, Oxford), pp. 71-105. 34. Stoeckel, K., Schwab, M. E. & Thoenen, H. (1975) Brain Res. 7. Suh, Y. S., Chung, K. & Coggeshall, R. E. (1984) J. Comp. 89, 1-14. Neurol. 222, 473-481. 35. Yip, H. K. & Johnson, E. M., Jr. (1983) J. Neurosci. 3, 2172- 8. Komiya, Y. & Kurokawa, M. (1978) Brain Res. 139, 354-358. 2182. 9. Ochs, S., Erdman, J., Jersild, R. A. & McAdoo, V. (1978) J. 36. Hendry, I. A. (1975) Brain Res. 94, 87-97. Neurol. 9, 465-481. 37. Levi-Montalcini, R., Aloe, L., Mugnaini, E., Oesch, F. & 10. Hamburger, V. & Levi-Montalcini, R. (1949) J. Exp. Zool. 111, Thoenen, H. (1975) Proc. Natl. Acad. Sci. USA 72, 595-599. 457-502. 38. Brunso-Bechtold, J. K. & Hamburger, V. (1979) Proc. Natl. 11. Prestige, M. C. (1967) J. Embryol. Exp. Morphol. 17, 453-471. Acad. Sci. USA 76, 1494-1496. Downloaded by guest on September 30, 2021