Distribution of Brain-Derived Neurotrophic Factor in Cranialand Spinal Ganglia

X.-F. Zhou,1 E. T. Chie, and R. A. RushDepartment of Human Physiology and Centre for Neuroscience, Flinders University of South Australia,

GPO Box 2100, Adelaide 5001, Australia

Received July 10, 1997; accepted September 30, 1997

In a previous study we have shown that a subpopula-tion of primary sensory neurons contain brain-derivedneurotrophic factor immunoreactivity. In the presentstudy we investigated the distribution of brain-de-rived neurotrophic factor and its mRNA in cranial andspinal ganglia at different segmental levels, usingimmunohistochemical and quantitative reverse tran-scriptase–polymerase chain reaction techniques. Ourresults show that there is no significant difference inthe percentage of brain-derived neurotrophic factor-immunoreactive neurons in spinal ganglia of differentsegmental levels. In contrast, more brain-derived neu-rotrophic factor-immunoreactive neurons were foundin placode-derived than neural crest-derived ganglia.The percentage of brain-derived neurotrophic factor-immunoreactive neurons is consistent with the per-centage of neurons lost after deletion of brain-derivedneurotrophic factor or trkB genes. However, there isno correlation between brain-derived neurotrophicfactor mRNA levels and the number of brain-derivedneurotrophic factor immunoreactive neurons in theseganglia, suggesting that some neurons synthesizebrain-derived neurotrophic factor while others accu-mulate the factor following its retrograde transportwithin nerve fibers. In particular, the proportion ofbrain-derived neurotrophic factor that is derived fromextraganglionic sources in the placode-derived gan-glia appears greater than that in the neural crest-derived ganglia. r 1998 Academic Press

INTRODUCTION

Brain-derived neurotrophic factor (BDNF) is a mem-ber of the neurotrophin family which acts duringdevelopment to affect neuronal proliferation, differen-tiation, and survival (3). It shares about 50% aminoacid sequence homology with nerve growth factor (NGF)and forms homodimers under physiological conditions

(22). BDNF supports the survival of a subpopulation ofsensory neurons during development and protects mo-tor neurons in the spinal cord, dopaminergic neurons inthe substantia nigra (17), and hippocampal and cholin-ergic neurons in the forebrain against death induced byaxotomy (24). BDNF binds both the low-affinity recep-tor p75 and the high-affinity receptor TrkB, triggeringthe latter’s autophosphorylation and a cascade of sec-ond messengers. These reactions are necessary initialsteps toward a variety of biological responses toBDNF (2).

Gene deletion of BDNF results in death of sensoryneurons in both spinal and cranial ganglia. The percent-age of neuron loss varies between 30% in dorsal rootand 80% in vestibular ganglia (12). Recently, Achesonand co-workers showed that sensory neurons can sur-vive in a single neuron culture without addition ofneurotrophic factors since these neurons synthesizeand secrete BDNF to keep the neuron alive (1). Theseauthors proposed that BDNF-responsive neurons de-pend on BDNF provided through an autocrine orparacrine mechanism rather than through a target-derived mechanism.

In a previous study we have found about 25% of smallto medium sized sensory neurons in the lumbar dorsalroot ganglia (DRG) are immunoreactive for BDNF (34).This result is consistent with the percentage of neuronscontaining mRNAs for either BDNF (14) or TrkB (27).It is also consistent with the percentage of neuronssurviving in vitro without addition of neurotrophicfactors (1). In addition, we found that BDNF in primarysensory neurons is anterogradely transported awayfrom the DRG to both the periphery and the spinal cord.Our recent ultrastructural study showed that BDNFwas associated with releasing vesicles in presynapticterminals in laminae I and II of the spinal cord (Luo,Rush, and Zhou, submitted for publication). Theseresults, however, are consistent with either an auto-crine, paracrine, or transsynaptic role for BDNF.

Since BDNF supports the survival of more sensoryneurons derived from placodes than from neural crest,we hypothesized that more neurons from placodes

1 To whom correspondence should be addressed. Fax: 1618 82045768. E-mail: pzxfz@cc.flinders.edu.au.

EXPERIMENTAL NEUROLOGY 149, 237–242 (1998)ARTICLE NO. EN976716

237 0014-4886/98 $25.00Copyright r 1998 by Academic Press

All rights of reproduction in any form reserved.

contain or synthesize BDNF. This was tested in thepresent study by examining the distribution of BDNFimmunoreactivity and mRNA levels in cranial andspinal ganglia at different segmental levels and compar-ing these with the known BDNF survival sensitivitiesof neurons from the different ganglia.

MATERIALS AND METHODS

Animals

Adult Sprague Dawley rats, male and female, wereused under the guidelines of the National Health andMedical Research Council of Australia and approved bythe Animal Welfare Committee of Flinders Universityof South Australia.

Immunohistochemistry of BDNF

BDNF immunohistochemistry was performed as de-scribed previously. The polyclonal rabbit antibody toBDNF made against recombinant human BDNF(rhBDNF) was characterized previously and recognizesBDNF only, without crossreactivity to other neurotroph-ins (34). Five rats were perfused with modified Zambo-ni’s fixative containing 2% formaldehyde after an over-dose of pentobarbital. Trigeminal, genicular, vestibular–cochlear, spiral and nodose ganglia of cranial nerves,and spinal ganglia of different segmental levels weredissected and processed for BDNF immunohistochemis-try. After dissection, all ganglia were sectioned on aCryostat microtome at 30 µm. Freely floating sectionswere treated with 0.3% hydrogen peroxide in 50%ethanol, blocked in 20% normal horse serum, andincubated in rabbit polyclonal antibodies againstrhBDNF at a concentration of 1 µg/ml overnight. Afterextensive washing in phosphate-buffered saline contain-ing 0.1% Tween 20, the sections were incubated inbiotinylated secondary antibodies to rabbit IgG, fol-lowed by incubation with the ABC reagent (VectorLabs). Sections were developed in 0.05% diaminobenzi-dine containing 0.01% hydrogen peroxide and observedunder a light microscope. The number of BDNF-immunoreactive (ir) neurons as well as the total num-ber of neurons in 40,000 µm2 from each ganglia werecounted and the percentage of positive neurons calcu-lated. Only neurons with a visible nucleus were counted.

Quantification of BDNF mRNA

The procedure for total RNA extraction and reversetranscriptase–polymerase chain reaction (RT–PCR)have been described previously (32). Briefly, total RNAwas prepared from fresh rat ganglia (n 5 3) accordingto the acid guanidinium thiocyanate–phenol–chloro-form method (8) using an RNA isolation kit (AdvancedBiotechnologies Ltd., Leatherhead, UK). The RNAwas treated with DNase (Promega, U.S.A.) to remove

possible contaminating genomic DNA and then di-rectly subjected to first-strand cDNA synthesis byincubation with oligo(dT)15 and AMV reverse transcrip-tase. PCR primers for BDNF and GAPDH cDNAs weredesigned corresponding to the coding region of thegenes as follows: BDNF primers, sense 58-TCCCTG-GCTGACACTTTTGAG-38 and antisense 58-CTATCC-TTATGAACCGCCAGC-38; GAPDH primers, sense58-TGCTGGTGCTGAGTATGTCG-38 and antisense58-GCATGTCAGATCCACAACGG-38. PCR reaction wasperformed in a 30-µl volume containing thermostableDNA polymerase (Advanced Biotechnologies Ltd., UK)on Perkin DNA thermal cycler (Perkin Elmer, U.S.A.).All samples were heated at 95°C for 2 min and ampli-fied in cycles at 95°C, 30 s; 58°C, 30 s; 72°C, 30 s. Thelast cycle was followed by a final incubation at 72°C for10 min. The housekeeping gene GAPDH was amplifiedin parallel to serve as an internal control. PCR cyclenumbers for BDNF and GAPDH were optimized intothe linear range. The concentration of cDNA wasdetermined by a preliminary PCR and adjusted forquantitative PCR. The PCR products were electropho-resed and stained with ethidium bromide on a 1.5%agarose gel. The gel was captured as a digital imageand analyzed using a FluorImager 595 and quantita-tion software (Molecular Dynamic, U.S.A.). This analy-sis provided the same results as achieved with aprevious method using radioactive hybridization (32).The ratio of the BDNF to GADPH mRNAs was calcu-lated from the fluorescent values of the PCR products.The mRNA level for each ganglion was expressed as apercentage of the mRNA level in the nodose ganglion.

RESULTS

Distribution of Sensory NeuronsImmunoreactive for BDNF

All the ganglia examined in this study containedBDNF-ir neurons. However, the numbers of neuronsimmunoreactive for BDNF in different ganglia variedsignificantly (Figs. 1–3). More BDNF neurons werepresent in cranial than in spinal ganglia. As shown inFigs. 1–3, 94 6 4% of neurons in the vestibular–cochlear ganglion were immunoreactive for BDNF. Thegenicular ganglion contained 81 6 3% BDNF-ir neu-rons. In the nodose ganglion, there were 47 6 3% ofneurons immunoreactive for BDNF (Fig. 1). In con-trast, only 27 6 2% of sensory neurons in the trigeminalganglion contained BDNF-ir. There was no significantdifference in the number of BDNF-ir neurons in thespinal ganglia at different segmental levels despite anapparent variation from 16% in L2 to 25% in T4ganglion. As shown in Fig. 2, the staining pattern andintensity of BDNF immunoreactivity in different gan-glia were different. The intensity of the staining in theDRG and nodose ganglia was strong, but with many

238 ZHOU, CHIE, AND RUSH

neurons unstained. However, the staining in the ves-tibular–cochlear and genicular ganglia was less intensebut most neurons were stained. This staining patternmay be related to the levels of mRNA in these ganglia(see below).

Distribution of BDNF mRNA in Cranialand Spinal Ganglia

As shown in Fig. 4A, BDNF mRNA levels can bequantified by the ratio of BDNF to the housekeepergene GAPDH in different samples if their PCR productsare in the linear range. As shown in Figs. 4B and 4C, allsensory ganglia investigated contained mRNA forBDNF. However, the levels of BDNF mRNA variedsignificantly. In contrast to the number of BDNF-irsensory neurons, L5 DRG contained the highest level ofBDNF mRNA while the vestibular–cochlear gangliacontain the lowest level. The nodose and trigeminalganglia expressed a moderate level of BDNF mRNA.There was no correlation between the level of mRNAexpression and the percentage of positive neurons bylinear regression analysis (r 5 0.468, P . 0.05).

DISCUSSION

Primary sensory neurons are derived from two ori-gins: placodes and neural crest (21). All spinal gangliaand part of the trigeminal ganglion are from the neuralcrest whereas the vestibular–cochlear, genicular, spi-ral, and nodose ganglia are from placodes. Some stud-ies have indicated that neurons from different originsdepend on different neurotrophic factors for their devel-opment and survival. For example, there is someevidence to suggest that neurons from placode derivedganglia are more sensitive to BDNF than those fromthe neural crest-derived ganglia (9, 25). Genetic muta-tion experiments further confirm these early observa-tions since more neuronal loss occurs in the placodederived ganglia after BDNF or TrkB gene deletion inthe mouse (12, 18, 19). However, recent evidence sug-gests that neurons from both placodes and neural crestrequire BDNF, neurotrophin 3 (NT3), and NGF at thedifferent developmental stages (6, 10). Our presentstudy showed that BDNF distribution in neurons ofthese ganglia correlates with their BDNF sensitivity.About 94% of all neurons in the vestibular–cochlearand 81% in the genicular ganglia contain BDNF-ir,which is consistent with mutation studies indicating80% of all neurons are lost in these ganglia after genedeletion. On the other hand, we found that 16 to 25% ofall neurons were immunoreactive for BDNF in thespinal ganglia. This result is again consistent with themuch smaller neuron loss seen in these ganglia afterBDNF or TrkB gene deletion in the mouse.

It is interesting that the percentage of BDNF-irneurons in the ganglia does not correlate with thelevels of BDNF mRNA. This conflicting result probablyindicates that BDNF immunoreactivity in sensory neu-rons does not necessarily represent the levels of localsynthesis. Although we have previously found that it isusually difficult to detect internalized neurotrophinsunless the rats are perfused with acidic or basic buffers(35), the antibodies used in this study may detectBDNF which is synthesized by neurons, is retrogradelytransported from their targets, or both. The pattern ofBDNF staining in the vestibular–cochlear and genicu-lar ganglia is different from that in DRG, being lessintense but present in the majority of neurons. It isknown that most of the neurons in these placode-derived ganglia express both the trkB and low-affinityneurotrophin receptors (29, 31). It is possible thatBDNF immunoreactivity in these neurons results fromtheir targets rather than being synthesized within thecell bodies. High levels of BDNF mRNA are detectablein inner ear epithelial end organs of the vestibular–cochlear ganglion neurons and BDNF is required fortheir development and survival (5, 13). We also foundthat the BDNF mRNA level in the vestibular gangliawas less than that in the DRG. These results indicate

FIG. 1. Micrographs of BDNF-ir sensory neurons in the nodoseand T7 spinal ganglia. Bars, 250 µm.

239DISTRIBUTION OF BDNF IN CRANIAL AND SPINAL GANGLIA

that BDNF is a target-derived neurotrophic factorrather than an autocrine factor for these vestibularneurons. Nevertheless, the presence of BDNF mRNAwithin these ganglia suggests some neurons synthesizeBDNF either for local release or to be anterogradelytransported within nerve processes.

The BDNF gene is critical for the maintenance of

respiration in the neonatal mouse (11). Since about 50%of visceral sensory neurons in the nodose ganglioncontain and anterogradely transport BDNF (Zhou andRush, data not shown), it is reasonable to speculatethat BDNF not only functions as a survival factor forthese neurons but also as a neuromodulator in theregulation of visceral functions such as respiration.Fast actions of neurotrophic factors are now clearlyestablished (4). In particular, BDNF enhances acetylcho-line release at the neuromuscular junction throughtyrosine kinase receptors (26). BDNF also influencesneuronal activities in a number of brain regions includ-ing the hippocampus (20, 23), visual cortex (15), andhypothalamus (28). These fast actions of BDNF areconsistent with its localization in nerve terminals ofvarious brain regions. Our recent ultrastructural stud-ies have shown that BDNF-ir is present in releasingvesicles of nerve terminals in laminae I and II of thespinal cord (Luo, Rush, and Zhou, unpublished data),suggesting that BDNF in the nerve terminals is re-leased in response to appropriate stimuli.

No significant difference was observed in the percent-age of BDNF-ir neurons in the neural crest-derivedganglia. This result suggests that targets do not influ-ence the phenotype of BDNF expression in these gan-

FIG. 2. Higher magnification of BDNF-ir sensory neurons in cranial and dorsal root ganglia. (A) Genicular ganglia; (B) vestibular–cochlear ganglion; (C) nodose ganglion; (D) T8 spinal ganglion. Bar, 100 µm.

FIG. 3. Histogram showing the percentages of BDNF-ir neuronsin cranial and spinal ganglia of different levels.

240 ZHOU, CHIE, AND RUSH

glia, although the level of expression has been shown tobe regulated by peripherally derived NGF. This is incontrast to NT3-immunoreactive neurons in these gan-glia (33). More neurons in the cervical and lumbarganglia are dependent on NT3 for survival (16). Inaddition, we have found that there is a specific distribu-tion pattern in the percentage of NT3-ir neurons inganglia at different levels (7). The percentages ofNT3-ir neurons in cervical and lumbar ganglia aregreater than that in thoracic ganglia, reflecting theirprojections to hair follicles, muscles, and tendons butnot viscera (33). Where BDNF-ir neurons terminate inthe periphery is not known, but the absence of asegmental specific distribution of BDNF-ir in theseganglia supports the view that specific targets do notdetermine their BDNF phenotype.

In summary, we have found that a higher percentageof BDNF-ir neurons in placode-derived cranial gangliathan in neural crest-derived spinal and trigeminalganglia. The percentage of BDNF-ir neurons in theseganglia correlates with the percentage of neurons lostafter deletion of the BDNF or trkB genes. However, thenumbers of BDNF-ir neurons in these ganglia do notcorrelate with the levels of BDNF mRNA, suggestingthat a greater proportion of the BDNF in the neurons ofthe placode-derived cranial ganglia may be derivedfrom their innervated tissues. Unlike the distributionof NT3 immunoreactivity, there is no significant differ-ence in the percentage of BDNF-ir neurons betweendifferent segmental levels. Thus, our results suggestthat while some neurons synthesize BDNF, othersaccumulate the factor from distant sites via retrogradetransport. It is also possible that some BDNF is synthe-sized by nonneuronal cells (30). It is unclear whether

any single neuron transports BDNF in both the retro-grade and the anterograde directions.

ACKNOWLEDGMENTS

The authors thank Mr. Daryle Cameron for his excellent technicalassistance. This work was supported by a grant from National Healthand Medical Research Council to RAR and X-FZ.

REFERENCES

1. Acheson, A., J. C. Conover, J. P. Fandl, T. M. Dechlara, M.Russell, A. Thadani, S. P. Squinto, G. D. Yancopoulos, and R. M.Lindsay. 1995. A BDNF autocrine loop in adult sensory neuronsprevents cell death. Nature 374: 450–453.

2. Barbacid, M. 1994. The Trk family of neurotrophin receptors. J.Neurobiol. 25: 1386–1403.

3. Barde, Y. A., D. Edgar, and H. Thoenen. 1982. Purification of anew neurotrophic factor from mammalian brain. EMBO J. 1:549–553.

4. Berninger, B., and M. M. Poo. 1996. Fast actions of neurotrophicfactors. Curr. Opin. Neurobiol. 6: 324–330.

5. Bianchi, L. M., J. C. Conover, B. Fritzsch, T. Dechiara, R. M.Lindsay, and G. D. Yancopoulos. 1996. Degeneration of vestibu-lar neurons in late embryogenesis of both heterozygous andhomozygous bdnf null mutant mice. Development 122: 1965–1973.

6. Buchman, V. L., and A. M. Davies. 1993. Different neurotroph-ins are expressed and act in a developmental sequence topromote the survival of embryonic sensory neurons. Develop-ment 118: 989–1001.

7. Chen, C., X. F. Zhou, and R. A. Rush. 1996. Neurotrophin-3 andtrkc-immunoreactive neurons in rat dorsal root ganglia corre-late by distribution and morphology. Neurochem. Res. 21: 809–814.

8. Chomczynski, P., and N. Sacchi. 1987. Single-step method ofRNA isolation by acid guanidinium thiocyanate–phenol–chloroform extraction. Anal. Biochem. 162: 156–159.

FIG. 4. Relative quantitative analysis of BDNF mRNA levels in vestibular–cochlear, genicular, nodose, trigeminal, and spinal ganglia.The cDNAs derived from each ganglion were amplified for 31 cycles for BDNF and 23 cycles for GAPDH, which is within the determined linearrange, and the PCR products were quantified after electrophoresis on 1.5% agarose gel stained with ethidium bromide. The BDNF mRNAlevels are expressed as the percentage of the values in the nodose ganglion. Values are means of ganglia pairs from three rats; error barsindicate 6 standard error of mean. (A) A standard curve showing a linear relationship between levels of PCR products and the number of PCRcycles; (B) histogram of BDNF mRNA levels in different ganglia; (C) a typical photograph of PCR products from different ganglia in anelectrophoretic gel.

241DISTRIBUTION OF BDNF IN CRANIAL AND SPINAL GANGLIA

9. Davies, A. M., H. Thoenen, and Y.-A. Barde. 1986. The responseof chick sensory neurons to brain-derived neurotrophic factor. J.Neurosci. 6: 1897–1904.

10. Elshamy, W. M., and P. Ernfors. 1996. A local action of neuro-trophin-3 prevents the death of proliferating sensory neuronprecursor cells. Neuron 16: 963–972.

11. Erickson, J. T., J. C. Conover, V. Borday, J. Champagnat, M.Barbacid, G. Yancopoulos, and D. M. Katz. 1996. Mice lackingbrain-derived neurotrophic factor exhibit visceral sensory neu-ron losses distinct from mice lacking NT4 and display a severedevelopmental deficit in control of breathing. J. Neurosci. 16:5361–5371.

12. Ernfors, P., K.-F. Lee, and R. Jaenisch. 1994. Mice lackingbrain-derived neurotrophic factor develop with sensory deficits.Nature 368: 147–150.

13. Ernfors, P., T. Van De Water, J. Loring, and R. Jaenisch. 1995.Complementary roles of BDNF and NT-3 in vestibular andauditory development. Neuron 14: 1153–1164.

14. Ernfors, P., C. Wetmore, L. Olson, and H. Persson. 1990.Identification of cells in rat brain and peripheral tissues express-ing mRNA for members of the nerve growth factor family.Neuron 5: 511–526.

15. Galuske, R. A. W., D. S. Kim, E. Castren, H. Thoenen, and W.Singer. 1996. Brain-derived neurotrophic factor reverses experi-ence-dependent synaptic modifications in kitten visual cortex.Eur. J. Neurosci. 8: 1554–1559.

16. Hory-Lee, F., M. Russell, R. M. Lindsay, and E. Frank. 1993.Neurotrophin 3 supports the survival of developing musclesensory neurons in culture. Proc. Natl. Acad. Sci. USA 90:2613–2617.

17. Hyman, C., M. Hofer, Y.-A. Barde, M. Juhasz, G. D. Yancopoulos,S. P. Squinto, and R. M. Lindsay. 1991. BDNF is a neurotrophicfactor for dopaminergic neurons of the substantia nigra. Nature350: 230–232.

18. Jones, K. R., I. Farinas, C. Backus, and L. F. Reichardt. 1994.Targeted disruption of the BDNF gene perturbs brain andsensory neuron development but not motor neuron develop-ment. Cell 76: 989–999.

19. Klein, R., R. J. Smeyne, W. Wurst, L. K. Long, B. A. Auerbach,A. L. Joyner, and M. Barbacid. 1993. Targeted disruption of thetrkB neurotrophin receptor gene results in nervous systemlesions and neonatal death. Cell 75: 113–122.

20. Korte, M., P. Carroll, E. Wolf, G. Brem, H. Thoenen, and T.Bonhoeffer. 1995. Hippocampal long-term potentiation is im-paired in mice lacking brain-derived neurotrophic factor. Proc.Natl. Acad. Sci. USA 92: 8856–8860.

21. Le Douarin, N. M. 1982. ‘‘The Neural Crest.’’ Cambridge Univ.Press, Cambridge.

22. Leibrock, J., F. Lottspeich, A. Hohn, M. Hofer, B. Hengerer, P.Masiakowski, H. Thoenen, and Y.-A. Barde. 1989. Molecular

cloning and expression of brain-derived neurotrophic factor.Nature 341: 149–152.

23. Levine, E. S., C. F. Dreyfus, I. B. Black, and M. R. Plummer.1995. Brain-derived neurotrophic factor rapidly enhances synap-tic transmission in hippocampal neurons via postsynaptic tyro-sine kinase receptors. Proc. Natl. Acad. Sci. USA 92: 8074–8077.

24. Lewin, G. R., and Y. A. Barde. 1996. Physiology of the neuro-trophins. Ann. Rev. Neurosci. 19: 289–317. [Review]

25. Lindsay, R. M., H. Thoenen, and Y.-A. Barde. 1985. Placodeneural crest-derived sensory neurons are responsive at earlydevelopment stages to brain-derived neurotrophic factor. Dev.Biol. 112: 319–328.

26. Lohof, A. M., N. Y. Ip, and M. Poo. 1993. Potentiation ofdeveloping neuromuscular synapses by the neurotrophins NT-3and BDNF. Nature 363: 350–353.

27. Mu, X., I. Silos-Santiago, S. L. Carroll, and W. D. Snider. 1993.Neurotrophin receptor genes are expressed in distinct patternsin developing dorsal root ganglia. J. Neurosci. 13: 4029–4041.

28. Siuciak, J. A., C. Boylan, M. Fritsche, C. A. Altar, and R. M.Lindsay. 1996. BDNF increases monoaminergic activity in ratbrain following intracerebroventricular or intraparenchymaladministration. Brain Res. 710: 11–20.

29. Vazquez, E., T. R. Van De Water, M. Del Valle, J. A. Vega, H.Staecker, F. Giraldez, and J. Represa. 1994. Pattern of trkBprotein-like immunoreactivity in vivo and the in vitro effects ofbrain-derived neurotrophic factor (BDNF) on developing co-chlear and vestibular neurons. Anat. Embryol. Berl. 189: 157–167.

30. Wetmore, C., and L. Olson. 1995. Neuronal and nonneuronalexpression of neurotrophins and their receptors in sensory andsympathetic ganglia suggest new intercellular trophic interac-tions. J. Comp. Neurol. 353: 143–159.

31. Ylikoski, J., U. Pirvola, M. Moshnyakov, J. Palgi, U. Arumae,and M. Saarma. 1993. Expression patterns of neurotrophin andtheir receptor mRNAs in the rat inner ear. Hear. Res. 65: 69–78.

32. Zhou, X.-F., E. T. Chie, Y. S. Deng, and R. A. Rush. 1997. Maturesympathetic neurons derive neurotrophin-3 from effector tis-sues in rat. Eur. J. Neurosci., in press.

33. Zhou, X. F., and R. A. Rush. 1995. Peripheral projections of ratprimary sensory neurons immunoreactive for neurotrophin 3. J.Comp. Neurol. 363: 69–77.

34. Zhou, X. F., and R. A. Rush. 1996. Endogenous brain-derivedneurotrophic factor is anterogradely transported in primarysensory neurons. Neuroscience 74: 945–951.

35. Zhou, X. F., C. Zettler, and R. A. Rush. 1994. An improvedprocedure for the immunohistochemical localization of nervegrowth factor-like immunoreactivity. J. Neurosci. Methods 54:95–102.

242 ZHOU, CHIE, AND RUSH