DEVELOPMENTAL

BIOLOGY

IFi&

181-l:%

(1 cjgd)

Transforming Growth Factor ,B Has Neurotrophic Actions on Sensory Neurons in Vitro and Is Synergistic with Nerve Growth Factor ALCMI~NE CHALAZONITIS,*~' JACQUELINE KALBERG,* DANIELR. RICHARDS.MORRISON,$ ANDJOHNA.KESSLER*~

TWARDZIK,~

Transforming growth factor 1,’(TGFP) influences the growth and diff’cwntiation of a wide variety of nonneuronal wlls (nnc) during embryogrnesis and in response to wounding. In the present study TGF$l and TGF$L were cbxaminvtl for their neurotrophic actions on neonatal rat dorsal root ganglion t DRG) neurons with ganglionic nnc in dissociated culturw. TGF$l and TGF/J% each increased hoth ncuronal survival and levels of the pcptidc neurotransmitter suhstancc I’ (SI’) csprcssed per neuron as well as per culture. TGFdl was maximally rffcctivc at a concentration of 40 pZr. Lvhereas TGF&! LV-;LSabout lo-fold less potent. Survival cffccts promoted by simultaneous treatment vAth both factors wcrc~ not atlditivc. TGF,U also changed the morphology and distribution of DRG nnc which resulted in closterinzof DRG nrurons on top of the nnc. Cotreatmcnt of tht cultures Cth two diffcrcnt anti-nerve growth factor (NGF) antibodies eliminated the ncurotrophic effclcts of TGF$l. However, trentmcnt with TGF/?l did not alter NGE’n1RN.A expression in the cultures nor did it chant[c t ht amount of NGF in the medium. Further, TGFdl greatly enhanced survival effects and SP stimulaLion promoted by exogenous NGF at concentrations up to 100 n&ml. The neurotrophic effects of TGF+jl were significantly attenuated by ticcreasing the proportion of the ganglionic nnc. suggestinK a role for these cells in mediating TGFdl action on thr neurons. It is hypothesized that the neurotrophic activity of TGFii’ depended upon the prcscncc of molcculcs immunolofiicall~ related to NGF and that the e!I’ects of TGF/j lvvrc synergistic with NGF. These ohserrations suggest that TGFb may play a role in the differentiation and rcgvneration of DRG neurons it/ vicv~. c l!W .\cadc~“li(

TNTRODIJ(‘TION

Transforming growth factor 13 (TGFD) is the prototype of a family of polypeptide growth factors which were originally isolated and purified from several transformed cell lines as well as nonneoplastic tissues. TGF/j is a 25-kDa homodimer differentiating factor involved in embryogenesis and in wound-healing responses (Massaguir, 1987; Must,oe et nZ., 1987; review by Roberts and Sporn, 1990a). Two of the different forms of TGFII’ which exist (TGFpl and TGFp2) differ in their binding patterns depending on the type of target cell. Some binding sites are more specific to either TGFPl or TGF[jB, or the number of each type of receptor may vary (Cheifetz ef Al., 1987; Roberts and Sporn, 1988). Most of the effects of the TGF/& have been observed ix Ctro. A wide variety of cell types respond to TGFfi including epithelial cells whose differentiation and proliferation are affected (Ellingsworth et nl., 1986).

’ To whom correspondence should be addressed at present address: Dept. of Anatomy and Ccl1 Biology, College of Physicians and Surgeons, Columbia Ilnirersity, 640 W. 168th St., New York, NY 10032.

The function of TGF/j in the nervous system is poorly understood despite the recent demonstration that TGF@ mRNA is elevated in the hippocampus in response to entorhinal lesions (Nichols f?!tAl., 1991) and that TGFpl stimulates expression of NGF in cultured rat astrocytes (Lindholm ct (I/., 1990). TGF@ mRNA has been localized in adult mouse epidermis in response to a tumor promoter (Akhurst ct (I(., 1988) as well as in developing embryos in areas of interaction between epithelium and mesenchyme (Heine et (I/., 1987; Lehnert and Akhurst, 1988). Such tissues can be potential targets of sensory neuron innervation. Furthermore, TGFfi promotes repair of incisional mounds through the skin (Roberts and Sporn, 1990b) by being released from activated macrophages at the site of lesion (Assoian cf ol., 1987). Sciatic nerve transection or local injection of an inflammatory agent into dorsal root ganglia (DRGs) can produce local regeneration of sensory neuron perikarya with the apparent involvement of macrophages and activation of satellite cells (Lu and Richardson, 1991). TGF@ also influences the growth rate of Schwann cells depending on their age in culture (Eccleston ef (&I., 1989). Taken together, these observations make it plausible that TGF$

122

DEVELOPMENTAL

BIOLOGY

influences, either directly or indirectly, the development and/or regeneration of sensory neurons. In the present study we examined the possibility that TGFfl might modulate survival and differentiation of neonatal DRG sensory neurons in culture. To determine whether possible neurotrophic actions could be mediated distinctly by one or another type of TGFP receptor, both TGF/31 and TGFP2 were tested. Both forms were neurotrophic for neonatal rat DRG neurons. Moreover, the effects of TGFPl seem to depend upon the presence of molecules immunologically related to NGF (neurotrophins), and the effects of TGFfil were synergistic with NGF. Preliminary accounts of this work have been reported (Chalazonitis et al., 1989, 1991). MATERIALS

Growth

Factor

Isolation

AND

METHODS

and PurQication

Simian recombinant TGFfil was purified following expression and secretion from CHO cells as previously reported (Gentry et al., 1987). TGFP2 was purified from bovine bone as previously described (Seyedin et ah, 1985). Both forms of TGFP exhibited identical kinetics in inhibition of DNA synthesis in CCL4 mink lung epithelial indicator cells (Ranchalis et (xl., 1987). NGF (7s form) and a rabbit anti-2.5s NGF (mouse submaxillary gland) antiserum were purchased from Collaborative Research (MA) for use in several experiments. In one experimental series another goat anti2.5s NGF antiserum was used (a generous gift from Dr. E. Johnson, Washington University Medical School). In some experiments, human recombinant, brain-derived neurotrophic factor (hrBDNF; GenenTech) was also used in combination with TGF@ or antiserum to NGF.

VOLUME

152.1992

other factors that could enhance the survival of DRG neurons (Chalazonitis and Fischbach 1980), the number of neurons that attached by 24 hr was about 10% of the number of cells plated. TGFBl (or 02) was added 3 hr after seeding the cells to assess neurotrophic effects independent of any effect on attachment to the substratum. The cultures were routinely treated with fluorodeoxyuridine (FdUr) and uridine (Ur) (Sigma) at 60 p&Z each on Days 1,4, and 8 to minimize the growth of nonneuronal cells (nnc). Under these conditions, by the second week in vitro, the proportion of nnc to neurons was estimated at 85 to 15% in control (no growth factor added) cultures vs 70% nnc to 30% neurons in TGFP-treated cultures. Alternatively, the cell suspension was preplated for up to 3 hr on a 60-mm plastic tissue culture dish and the cultures (established from the cells which did not attach during the preplating) were treated with cytosine arabinoside (AraC) (10 palm) on Days 1 and 3. Under this treatment and because AraC is a more effective antimitotic agent than FdUr the proportion of nnc decreased to 50%. In some experiments further reduction of the nnc (down to 25-1557 of the total population) was achieved by layering the ganglion cell suspension on top of a 4-ml metrizamide cushion (Sigma) made up in PBS to a density of 1.063 g/ml and centrifuging according to the protocol of Yamamoto and Gurney (1990). After collecting the pellet corresponding to the neuron-enriched fraction, the cells were resuspended in complete medium and plated. The cultures were then fed with medium containing either FdUr and Ur or AraC (highest depletion of nnc) on Days 1 and 3. Cultures were fed twice a week for up to 2 weeks and fresh TGFP was added at each feeding. Control cultures were treated with an equivalent volume of phosphate-buffered saline (PBS).

Tissue Culture One-day-postnatal rat dorsal root ganglia were collected from all segments of the neuraxis, incubated for 30 min in 0.25% trypsin (GIBCO), and dissociated in complete medium as previously described (Adler et al., 1984). The cell suspension was plated on rat tail collagen-coated (0.4-0.5 mg/ml) and laminin-coated (5 pg/ ml, Collaborative Research) 35-mm tissue culture dishes (Nunc). The mean plating cell density was 2.2 X lo5 cells per dish (ranging from 1.8 to 3 X lo5 in different experiments). The plating and maintenance medium was Eagle’s modified essential medium (GIBCO) supplemented with penicillin (50 units/ml), streptomycin (50 pg/ml), glucose (50.4 mM), glutamine (2 mM), and lo%, heat-inactivated horse serum (HyClone, UT or Hazleton, KS). Under these conditions (without added chick embryo extract or fetal calf serum which contain many

Neurons were counted at different times i?l vitro under phase-contrast optics at 100X magnification in 20 fields across both the X and Y diameters of the dish. Three dishes were used for each control or experimental group. Neurons were identified as phase-bright perikarya with a large nucleus and distinctive nucleolus and at least one process more than four times the perikaryal diameter. Corroboration that these cells were neurons was obtained by examination of sister cultures (treated with TGFfll) stained with anti-neurofilament protein (NF-M) antibody RM03 which specifically marks rat DRG neurons (Lee et al., 1987). Cultures were fixed and stained according to previously described procedures (Chalazonitis et ub, 1992). The numbers of nnc were determined by direct counts of living cultures as described above.

The assay was carried out according to the procedure of Powell et al. (1973) as modified by Kessler and Black (1980). RNA

Extrcxctiow

und Expression

of NGF

rrlRNA

by

Total RNA was isolated from DRG cultures by the guanidinium-thiocyanate-phenol-chloroform method of Chomczynski and Sacchi (1987), with minor modifications. The purity and concentration of RNA were determined spectrophotometrically. Fractionation of the denatured RNAs was performed on a 1% agarose gel with 0.66 Mformaldehyde (Sambrook et al., 1989). RNAs were transferred to Genescreen Plus Membrane (NEN) by overnight capillary blotting. NGF mRNA (1.1 kb; Edwards et ul., 1986) and mRNA for the constitutive enzyme cyclophilin (0.9 kb) were detected using cDNA constructs for NGF (Clone pSP65-NGF, generously supplied by Dr. M. Chao, Cornell University Medical College) and for cyclophilin (Clone lB15-cyclophilin, kindly supplied by Dr. James Douglass, Oregon Health Sciences University). Labeling of cRNA with 32P,either for NGF or for cyclophilin, was carried out using SP6 RNA polymerase. Northern blots were prehybridized for 30 min at 65°C in 50%. formamide, 0.1 M NaCl, 1%) SDS, and 0.1 mg/ml denatured herring sperm DNA. Blots were hybridized at 65°C overnight in prehybridization solution to which 1 x lo6 cpm/ml of one of the riboprobes had been added. Blots were washed to a final stringency of 0.1X SSC (1X SSC = 0.15 M sodium chloride, 0.015 1M sodium citrate). The blots were then exposed to Kodak XAR-5 film (2-8 days) with intensifying screens at ~70°C. Optical density of the resulting bands was volume integrated using a scanning densitometer (Molecular Dynamics ). AIeu.surenw~ t of‘ NCF Rdec~sed in Conditioned Xd iu III (CiW

C2tlfur~

Levels of NGF in CM from TGF@l-treated cultures were determined using an ELISA with an anti-recombinant human (anti-rhNGF) antibody and a secondary anti-rhNGF conjugated to HRP according to the protocol of Dr. D. V. Sinicropi, Genentech Inc. KES~JLTS

Treatment with TGFol (5-10 n&ml) significantly increased the survival of DRG neurons after l-2 weeks in culture (Fig. 1B vs Figs. 1A and 2). DRG neurons main-

tained in TGFPl (5 r&ml) also exhibited a tendency to form clusters (Figs. lB, lC, and 1D). Corroboration that the phase-bright cells were actually neurons was provided by staining with the anti-neurofilament protein NF-M antibody RM03. The number of stained fluorescent perikarya corresponded to the number of phasebright perikarya (Fig. 1D vs Fig. 1C). In addition to its effect on neuron survival, TGFPl changed the morphology and distribution of the DRG nnc. These formed islands which preferentially supported neuronal aggregation (Fig. 1B vs Fig. IA). To determine whether the two homodimeric forms (1 and 2) of TGFD have comparable neurotrophic effects and whether such effects are additive, sister cultures were treated either with each homodimeric form or with both forms simultaneously. Each factor at 10 rig/ml increased neuronal survival about twofold after 11 days irr uitm, and the effects were not additive (TGFfil, 6887 + 730 neurons, 71= 16, vs control, 3726 i 340 neurons, 7) = 16; TGFBS, 8078 f 761 neurons, ~1= 8, vs control, 4165 + 468 neurons, II = 9; simultaneous treatment with both forms, 7564 + 991 neurons, rl= 4, vs control, 3872 +- 734, II = 5). Figure 2 summarizes three independent experiments. TGF@S also induced clustering of the DRG neurons and nnc (data not shown).

Survival of DRG neurons was examined from 24 hr after plating up to 14 days i~r vitro (Fig. 3). At 24 hr the numbers of neurons in cultures treated with TGFbl were comparable to untreated controls. However, by the fifth day ir/ t1itr.o neuron numbers decreased by 75%’ in control cultures and then stabilized up to 12 days i)l r*itm. TGFPl significantly attenuated neuronal cell loss (by 50%) during this period which resulted in a sustained increase in survival (1.6-fold, compared to control cultures) for the 2-week-culture period (Fig. 3). Thus, the actions of TGFpl were not transient. The rate of increase in neuron survival in this particular experiment is lower than the mean rate of survival (1.95-fold, )I = 9 experiments) but within the range of increases (1.3- to 3.2-fold) observed to occur with TGFP treatment at this time it/ vifro. To determine the minimal and optimal concentrations at which the TGFps promoted survival, neuron counts were performed at 11 days I,r vitro with various concentrations of the growth factors (from 0.1 to 100 n&ml) (Fig. 4). Maximal survival occurred at 1 rig/ml of TGFol and at 10 n&ml of TGFp2, indicating that the 81 form was more potent than the D2 form. However, the maximal increase in survival was not significantly different for either form. No further increase in survival

124

DEVELOPMENTALBIOLOGY

VOLUME 152 , 1992 ..

was detected (Fig. 4).

TGFpl

at 100 rig/ml

for either

(xs Well as TGF@ Irmeases

type of TGFP

Substance P

The phenotype of DRG neurons rescued by TGFfl was characterized by evaluating the levels of the peptide neurotransmitter substance P (SP) after 11 days of treatment (Fig. 5). Levels of SP (in pg per dish) increased more than threefold in TGFpl-treated cultures and 2% fold in TGFBB-treated cultures. Simultaneous treatment with TGF@l and TGFp2 did not further elevate levels of the peptide (Fig. 5A). When SP was expressed in femtograms per neuron, there was still a significant increase produced by treatment with TGFPl and TGFp2 (Fig. 5B).

To determine whether the neurotrophic action of TGF@l was mediated via an interaction with ganglionic nnc rather than directly on the neurons, the effect of TGF$l was tested in neuron-enriched cultures. In routine DRG cultures treated with FdUr, nnc outnumbered (6:l) the neurons because this antimitotic inhibitor arrests cell division but does not eliminate these cells. In these cultures (11 days) the TGFpl-mediated survival was 85% higher (see data in Fig. 2; significance of difference from control at P < 0.001) and the SP level was threefold higher (see data in Fig. 5; significance of difference from control at P < 0.001) than in untreated cultures. In cultures resulting from the preplated ganglion cell suspension and treated with AraC, the proportion of nnc to neurons decreased to 1:l. By 11 days the increase in TGF@l-mediated survival was attenuated to a degree which was not significantly different (at P < 0.2) 8298 +- 1586 neurons (II = 9) from the survival in control cultures 5381 k 1184 neurons (n = 8). The SP level was attenuated to 1.6-fold higher, 190 F 17 (Z = 9) vs 118 i 17 p&dish 01 = 8), than in control cultures (difference significant at P < 0.025). Utilizing another method for preparing neuron-enriched cultures the ganglion cell suspension was plated after centrifugation us-

ing a metrizamide gradient and subsequently treated with FdUr and Ur. At 4 days in d-o there was still a significant increase in neuronal survival (at P < 0.001) in the neuron-enriched cultures treated with 5 n&ml TGF/jl, 3309 i 398 (7~ = 7), vs the untreated cultures, 957 i 100 (7) = 7). However, in similar cultures subsequently treated with AraC rather than FdUr (yielding a minority of nnc, down to 15% ), the TGFfll-mediated survival was significantly attenuated: 2111 + 220 neurons (1) = 6) vs 1336 -t 362 (i/ = 6) in untreated cultures (significance of difference at P < 0.05). These data suggest that TGFPl exerts neurotrophic effects on DRG neurons which can be enhanced by increasing the amount of nnc present in the cultures. Because effects were observed in the presence of AraC and at a late developmental stage, it is unlikely that TGF@ promotes division of neuronal precursors. However, these observations do not rule out a direct differentiating effect of TGFbl on the neurons.

Since NGF is a known survival factor for DRG neurons, it is conceivable that the ncurotrophic effects of TGFD were mediated indirectly by NGF released from ganglionic nnc (or even from the neurons themselves). To test this possibility, sister cultures were treated with TGF@l, NGF, or no growth factor, and half were simultaneously treated with a rabbit anti-mouse 2.5s NGF antiserum for 4 days. In the absence of the NGF antibody, DRG neuron survival mediated by exogenous NGF or TGF@ was 9473 t 1318 0) = 13) and 6589 +- 793 (II = 6), respectively, compared to 3357 -t 185 0, ~~ 12) in untreated cultures (Fig. 6A, white bars). In marked contrast, the presence of the NGF antibody (1:lOO) suppressed neuronal survival mediated either by NGF (3250 + 254; )t == 11) or by TGFbl (3537 t 328; )/ = 6), reducing it to the level of survival in untreated cultures (Fig. 6A, meshed bars). Treatment with NGF antibody alone did not alter significantly the level of neuron survival (3199 I 267; 71= 9) from that in untreated cultures (average of four experiments). Comparable blocking effects were achieved with an alternative goat antiserum raised against 2.5s NGF (generous gift of Dr. E. John-

FIG. 1. Phastx-contrast Ilhotomirro~raphs of rat neonatal DRG neurons in dissociated cultures gronn in the ahscncc or presence of TGFijl. The L)R(: cells xvcre prepared with FdLJr and LJr as antimitotic inhihitors as described under Materials and Methods. The cultures MYW grown either in the akencc of growth factor (AJ or in the presence of 5 n&ml of TGFiJl for X days (B-L)). Note the increase in neuronal survival (counted as :10 neurons (I31 YS ‘7 neurons (A, marked by arrows)) and the change in distribution and morphology of Schwann cells in the TGF&trcatcd cultures compared to controls. Also note the clustering of neurons on islands of nonneuronal cells (BJ compared to their srattercd and flattened appearance in control (A). (C’, I)) Phase-contrast and fluorescence illumination views of the same field from a TGFkjltreated culture stained with the monoclonal anti-ncurofilament protein antibody RMO3. Only the hirefringent neurons with fasciculatcd ncurites (counted as 1X neurons in (‘J are stained (counted as 1X nrurons in L)) in contrast to the fihrohlasts or Srhuann cells rarrows). Calibration bar. 100 pm.

DEVELOPMENTAL

,O

0

No Growth

W s

TGFI31 TGFDP

q

TGFOl+TGF02

1

BIOLOGY

VOLUME

152, 1992

Factor 6

T'

;'

U

TGFRl

-

TGFi?Z

6

I

ot

0

GROWTH

TREATMENT FIG. 2. Effects of TGFfil and TGF@ on DRG neuron survival. Cells were plated at a mean density of 2.3 x lo5 cells per dish and were maintained with no growth factor (0 &ml) or with TGFPl, TGFo2, or both at 10 n&ml. Eleven days after adding the growth factors, the numbers of neurons were determined. The vertical bars represent the mean neuron number f SEM from pooled experiments (four for TGFljl, five for TGFp2, and three for TGFpl and TGF&Z together). Note the twofold increase in neuronal survival promoted bq’ either form of TGFI.1 and the lack of additivity of this effect when both TGFDs are present in the cultures. *Differs from respective control at P < 0.001: **differs from control at P < 0.025 bg Student’s two-tailed t test.

son, Washington University, St. Louis, MO). Substance P levels were measured in one experiment of this series using the Collaborative Research NGF antiserum. The increase in SP level normally observed in TGF@l-

0-l 0

2

4

6

DAYS

8

10

12

IN VITRO

FIG. 3. Time course of DRG neuronal survival in the continued presence or absence of TGFBl. Cultures were treated as described under Materials and Methods, using FdLJr and Ur as antimitotic inhibitors. Each point represents the mean number of neurons f SEM determined from dishes in triplicate from two separate experiments. All TGFol experimental values (beyond 24 hr i?~ tlitlo) differ from control two-tailed f test. values at *P < 0.005 and at ** P < 0.05 by Student’s

I

1

1

FACTOR

10

100

nglml

FIG. 4. InHuence of TGF/?l and TGF@ concentrations on the survival of DRG neurons. DRG neuron cultures were prepared as described under Materials and Methods. The plating density ranged from 1.8 to 2.5 X lo5 cells per 35.mm dish. Three hours after plating TGFfll (open squares) or TGF/%Z (filled circles) was added to the cultures at various concentrations. Eleven days later the numbers of neurons were determined as described under Materials and Methods. Each point represents the mean neuron number f SEM per dish (U = 4-6 dishes), pooled from three (TGFBI) and two (TGFI-12) experiments. Whereas TGFU promotes maximal survival at 1 n&ml, TGFfi2 is maximally effective onlg at 10 n&ml. Note that the maximal number of surviving neurons is similar for both factors (cf. also Fig. 2).

treated cultures (103 pg/dish vs 53 pg/dish in control) was abolished (38 pg/dish) in sister cultures treated simultaneously with TGFfll and the NGF antibody. This is consistent with the data obtained for neuronal survival. The above data suggest that the neurotrophic effects of TGFP may be indirectly mediated by NGF. However, it could also implicate other members of the neurotrophin family such as BDNF (Leibrock et a,Z.,1989), NT-3 (Hohn et crl., 1990; Maisonpierre et ul., 1990), NT-4 (Hallbook et al., 1991), or NT-5 (Berkemeier et ul., 1991) that may cross-react with 2.5s NGF. In fact, in a similar experiment, a 4-day treatment with the Collaborative Research anti-NGF antiserum also prevented DRG neuronal survival promoted by hrBDNF (50 rig/ml) (BDNF, 8333 -t 570 neurons (7~= 3); BDNF + anti-NGF, 5845 & 88 neurons (71= 3); untreated controls, 5447 + 247 neurons (n = 3); and anti-NGF alone, 5487 & 595 neurons (rl = 3) (Fig. 6B). E.qression qf NGF mRNA in TGFBl-Treated and Levels of NGF in the Culture Medium

Cultures

The observations using anti-NGF antibody suggested a potential indirect role for NGF in the neurotrophic actions of TGFfi on DRG neurons. One possibility was that TGF@ stimulated the synthesis of NGF in the DRG cultures. To test this, levels of NGF mRNA were determined after treatment with TGFbl. Cultures were harvested for Northern blot analysis with clone pSP65-

1

6 0

l

30

NoTGFD

.

-I-

T

l

’

H

TGFRl

1

l

**

z !! 3

FIG. 6. Both forms of TGF$ increaw the level of substance P in DRG neuron cultures, but combined treatment is not additive. Levels of substance I’ were determind hg radioimmunoassay of 11-day-old DRG cultures (whose survival is depicted in Fig. 2) and expressed either in pg/tlish (A) or in f&neuron (B). In A the open bars (no TGFb) represent, from left to right, the mean + SEM SP immunorcactivity with respectively 50 i 5.5 (,/ = 1X), 59 ? 9 ( )I = 9), and 43 k 8 (H = C5) p&dish; the stippled bars (TGF/Ctreated) represent the mean i SEM SP actkits \vith respectively 159 z 19 (U = 15), 163 t 21 (U = 8), and 133 f 39 (t/ = 4) p&dish. *Differs from control at P < 0.001; **differs from control at I’ i 0.026. In B, note that ewn when expressed in fg per neuron, levels of SP were significantly increased (stippled bars) in TGFi_ll-treated (24.3 i 3.7, 11 15) vs control (15.4 i 2, o z 12) anti in TGF@-treated (20 t 2, II = 8) vs control (1-l k 1.5. N - 9) cultures. ***Differs from respective control at I’ i 0.05. Statistical analysis hy two-tailed Student’s f trsts.

NGF at both 48 hr and 12 days in vita. A single transcript corresponding in size (about 1 kb) to that for NGF transcripts for mouse submaxillary gland (Edwards et crl., 1986) was observed. Quantification revealed that NGF mRNA did not differ significantly between control and TGFpl-treated cultures at either time in culture. The relative level of NGF mRNA (normalized to that for lB15 (cyclophilin) mRNA) in TGFol-treated cultures was 134 F 32% (mean & SEM from 10 pooled measurements over three experiments) of that in control cultures. To determine whether TGFpl might, increase NGF release in the DRG cultures, levels of NGF in the culture medium (CM) were determined by ELISA analysis after 8 days i?l ~~itro. The cultures originally plated in the presence of HS-containing medium were switched to defined medium 18 hr later and received one change of medium at 4 days. The standard curve was established using hrNGF in defined medium. Levels were 2.0 and 1.6 ng/ ml of NGF in TGFpl-treated cultures compared to 2.3 and 2.0 n&ml in controls (duplicate measurements). Thus, there was no detectable increase of NGF protein in the CM after TGF/jl treatment.

TGF/jl did not alter levels of NGF mRNA nor increase the release of NGF protein into the medium of DRG cultures. However, the possibility remains that TGF@l enhances the responsivity of DRG neurons to low endogenous levels of NGF. To test this, DRG cul-

tures were treated with 5 n&ml TGF(31 (a saturating concentration) together with varying (0.1-750 n&ml) concentrations of 7s NGF. Neuron survival was compared in cultures treated with each growth factor separately and in sister cultures treated with both factors (Fig. 7). At 5 days in vitro, survival in untreated (control) cultures was 3624 i 244 neurons (71 == 9). At submaximal levels of NGF, neuron numbers after treatment with both factors were significantly higher than the additive effects predicted by survival due to TGF@l alone and to NGF alone. For instance, survival with TGF@l and 10 n&ml NGF (14,662 t 2015 neurons; I/ = 6) was significantly higher than survival with TGF$l alone (7344 & 709 neurons; n = 6) added to the increase in survival due to NGF alone (2416 neurons above survival in control cultures) (Fig. 7). Note t,hat the effects of TGFpl were not apparent at a saturating concentration of NGF (Fig. 7). These trends in survival effects were maintained for 12 days in [Gtro (data not shown). SP levels were also compared in cultures treated with TGFpl alone (5 r&ml), NGF alone (at various concentrations), or both factors (Fig. 8). Wit,hout NGF, TGFbl alone doubled the SP levels to reach 45 f 9 pg/dish (11 ~ 9). At saturating concentrations (100 n&ml) NGF caused a similar increase (49 + 3 p&dish (7, = 6). TGF/jl consistently enhanced the SP activity at all concentrations of NGF tested (up to 750 n&ml) with a maximal effect at lo-100 &ml NGF of 178 & 17 pg/dish (T/ r 6) (Fig. 8). Furthermore, when expressed in femtograms per neuron, SP levels were on average 2.2-fold higher in the presence of TGF@l and NGF at all concentrations

DEVELOPMENTALBIOLOGY VOLUME q q

1’

GFiOWM FACTOR GROW

FACTOR + ANWGF

T

A

L

NGF

(1

2. , f$i

::1: l-::‘\

04

B

BDNF

.I

1

10

100

1000

NGF (nglml)

GF0WTH FACTOR

NONE

-

**

NGF

FIG. 6. The survival effect of TGF@l is blocked hg an antiserum to mouse submaxillary gland 2.5s NGF; this antiserum cross-reacts with BDNF. Cultures were treated for 4 days either without growth factor (NONE) or with TGF@l (5 n&ml), hrBDNF (50 &ml), or ‘7s NGF (200 n&ml). Neuronal survival was compared in sister cultures treated as above without (white bars, A. or stippled bars, B) or with anti-NGF (anti-2.5s antiserum, dilution 1:lOO) (meshed bars). The mean plating density in the experiment in A was 1.75 X 10” cells per dish and in the experiment in B it was 3 X lo” cells per dish. (A) Each bar represents the mean neuron number per dish counted in triplicate -t SEM from two experiments (TGFol) or from four experiments (NONE and NGF). **Differs from NONE, (-t anti-NGF) at p < 0.005; *differs from respective growth factor alone at P < 0.01. (B) Each bar represents the mean neuron number per dish counted in triplicate * SD from one experiment. *Differs from BDNF and anti-NGF at P < 0.005; **differs from NGF + anti-NGF at I’ < 0.001. Statistical analgsis was performed using an analysis of variance (Anova).

tested (0.1-750 n&ml) as compared to NGF alone at the same concentrations (data not shown). TGFfil

Is Not SignQkuntly Additive to nor Synergistic BDNF ix Promoting the Sur11im1 of DRG Neurons with

To determine whether TGFPl may enhance the survival effect of BDNF on DRG neurons as it did for NGF,

FIG. 7. Enhancement by TGF@l of the NGF-mediated survival of DRG neurons as a function of increasing concentrations of NGF. Cultures were established at a mean density of 3.8 x lo5 cells per dish and treated for 5 days either with increasing concentrations of 7s NGF (O-750 n&ml) (open squares) or with TGFfll (5 n&ml) and increasing concentrations of NGF (filled squares). Each point represents the mean number of surviving neurons (+-SEM) from at least three and at most nine dishes (data are pooled from four different experiments). Note that the NGF-promoted survival in the presence of TGFpl is significantly higher than it would be if survival were strictly additive to survival promoted by the two growth factors separately (dotted line). Statistical probabilities of significance comparing survival with and without TGFB varied from I’ < 0.02 to P i 0.0001 (using two-tailed Student’s f test). a series of cultures was set up and neuron counts were carried out in triplicate dishes after 9 days of treatment with no growth factor (control), hrBDNF (100 n&ml), 200 F .$ \ 0 a a 8 fL 2

O-O 0-O

NGF NGF + TGF,Tl (5 rig/ml)

150

0 l

100 I

+,i

/ 1

/ 50

o

o/owoo

1 .o

10

-0”

0

0:1

NGF CONCENTRATION

100

1000

(rig/ml)

FIG.8. Enhancement by TGF$l of the NGF-mediated increase in substance P content of DRG cultures as a function of increasing concentrations of NGF. Levels of substance P (in pg/dish) were determined by RIA from DRG cultures after 11-12 days i?, vitro. Cultures were established at a mean cell density of 3 X lo5 cells per dish and maintained with increasing concentrations of NGF (open circles) or with TGFfil (5 n&ml) and increasing concentrations of NGF (filled circles). Each point represents the mean + SEM from at least 3 and at most 12 dishes (data pooled from five different experiments). Note that at all concentrations of NGF, the substance P content per dish in the presence of TGFDI is higher than the added increases of SP activities by NGF and TGF@l when present separately in the cultures.

TGFPl (5 &ml), or both factors together. As expected, compared to controls (4835 5 327 neurons) TGFol increased survival (9389 & 1269 neurons, 71 = 3) as did BDNF to a similar extent (9195 * 781 neurons, u = 3). Moreover, combined treatment with both factors did not produce sensory neuron survival significantly higher (12,424 k 2752 neurons, 1~= 3 at P < 0.4) than treatment with each factor alone. Furthermore, TGF01 was not synergistic with BDNF in contrast to the survival effects observed with TGFPl and NGF. DISCUSSION

The present study has demonstrated that two forms of TGFP (types 1 and 2) exert neurotrophic effects on neurons in cultures of DRGs. Increased neuronal survival as well as increased expression of the peptide neurotransmitter substance P occurred after treatment with bot,h forms of the factor. Interestingly, effects on SP levels exceeded the enhancement of neuronal survival, suggesting that these TGF& may exert differentiating effects on the DRG subpopulation of SP-containing neurons.

The concentration of TGF@s required for maximal survival differed in that TGF@l was maximally effective at 1 n&ml (i.e., 40 pIV) compared to TGFP2 with maximal effect at 10 rig/ml (400 PM). This lo-fold difference in potency suggests the existence of functional high affinity binding sites in the DRG culture system with higher affinity for TGFBl than for TGFDB, as can occur in mammalian or avian cells (Cheifetz et al., 1987). These values are consistent with the KL, for the TGF@l receptor (which ranges from 20 to 30 pM) on a variety of cells of epithelial and mesenchymal origins (review by Roberts and Sporn, 1988). The range of concentration for biological activity is similar to that of other neurotrophic factors for DRG neurons such as NGF (Greene, 1977) or BDNF (Lindsay pit ~1.. 1985). Treatment of the cultures with both TGF@l and $2 at saturating concentration (10 r&ml) did not produce additive effects over those observed for each type of TGFP separately. This suggests that both forms of TGF@ may interact either with the same receptor or with different forms of TGF@ receptors, resulting in similar intracellular transduction pathway(s).

The increases in survival as well as in SP immunoreactivity mediated by TGFPl were suppressed by simultaneous treatment with NGF antibodies from two different sources. These data could imply that NGF may play

an integral role in the neurotrophic actions of the TGF@ on DRG neurons. However, levels of mRNA for NGF in the cultures and of NGF protein in the conditioned medium of TGFpl-treated DRG cultures were not increased, suggesting that TGF@l does not stimulate either synthesis or release of NGF. Furthermore, TGF@l does not increase NGF mRNA in Schwann cell cultures (Matsuoka et ul., 1991). Recent studies have shown that NGF belongs to a family of structurally related “neurotrophins” which include BDNF, NT-3, and the recently characterized NT-4 and NT-5 Because of their considerable degrees of homology, the neurotrophins have the potential to be immunologically crossreactive. Thus, it is conceivable that TGF@l may induce the synthesis and/or release of a related neurotrophin(s) recognized by the polyclonal antibodies used in the present study. For instance, the NGF antibody from Collaborative Research did block the increase in DRG neuron survival mediated by hrBDNF. Consonant with this observation is the report that medium conditioned by cultured rat fibroblasts and Schwann cells has BDNF-like biological activity that can be blocked by antibodies against NGF (Acheson c>f rrl., 1991). Furthermore, in the present study, simultaneous treatment with maximal levels of BDNF and TGFpl resulted in DRG neuron survival not significantly higher than that obtained with treatment with each of the factors alone. Thus, it is possible that BDNF may mediate some of the actions of TGF$l. Data on synthesis or release of BDNF in the TGF$l-treated cultures are needed to demonstrate this rigorously. However, in addition to BDNF, NT-3 and no\v NT-5 have been shown to increase DRG neuron survival and NT-4 has been shown to increase DRG neurite outgrowth; thus, any one of them could be involved. Furthermore, it is likely that other members of the neurotrophin family have yet to be identified. Thus, to test which of these neurotrophins may be involved in the TGF$ response, a more complete identification of all members of the neurotrophin family and the availability of specific immunological reagents is required.

Alternatively, TGF$ could act in synergy with NGF. The enhanced neuron survival (see Fig. 7) in cultures treated simultaneously with subsaturating concentration of NGF and with TGF@ suggests that TGFP could modulate the effects of low (l-2 n&ml) endogenous levels of NGF present in the DRG cultures. Subsets of neurons among the DRG population vary in their responsivity toward NGF. At the subsaturating concentrations, NGF rescues only the subset with high expression of the high affinity receptor to NGF. At high NGF con-

130

DEVELOPMENTALBIOLOGY

centrations, saturation of receptor occupancy is reached also on the other subsets of neurons and maximal survival is reached. Also NGF can then cross-react with heterologous neurotrophin receptors (Rodriguez-T&bar et ul., 1990) and the synergistic effect of TGF/31 is no longer detectable (see convergence of the survival curves * TGF/Jl, Fig. ‘7) because survival is an all or none phenomenon. At low NGF concentrations, TGFPl is synergistic because it may directly enhance the effect of NGF at low receptor occupancy, such as increase in affinity or level of NGF receptors. For instance, in corneal endothelial cells TGFPl has been shown to increase the bioactivity of bFGF (Plouet and Gospodarowicz, 1989). It has been proposed that TGFP modulates the effects of bFGF by increasing its affinity for its receptor (Besnard et al., 1989; Baird and Biihlen, 1989) or by increasing the level of bFGF receptor mRNA, as has been shown for adult auditory neurons is/ vitro (Lefebvre et al., 1991). Alternatively, TGFfll leads to production of other neurotrophins. In this condition combination of low levels of NGF and the neurotrophin(s) can be synergistic on the subset of neurons which have receptors for both. In contrast, the synergistic action of TGFol on the level of SP is maintained at and above the saturating range of NGF concentration (Fig. 8). Unlike survival which is an all or none phenomenon, the neuropeptide continues to be upregulated in an otherwise stable population of surviving neurons (Kessler and Black, 1980). Thus, the differentiating effect of TGFPl is exerted beyond the NGF concentration needed for maximal survival and in fact TGFfil increases the level of SP per neuron.

The indication that the actions of TGF(3 in DRG neurons is indirectly mediated via neurotrophins raises a possible role for the nonneuronal cells in these cultures. The pronounced morphological effect of TGFPl on DRG nnc and the striking clustering of DRG neurons on top of islands of nnc in TGF@l-treated cultures (see Fig. 1) certainly suggests that the factor acts on these cells. Other findings indicate that fibroblasts and Schwann cells both respond to TGFP. The former exhibit increased synthesis of collagen and production of fibronectin (Ignotz et ul., 1987) and the latter are stimulated to proliferate (Ridley et ul., 1989). Furthermore, TGFb increases the cellular synthesis of protease inhibitors (Sporn et ul., 1987) and hence indirectly reinforces the accumulation of extracellular matrix proteins. Such observations raise the possibility that TGFP could influence interactions between ganglion nnc and neurons. Another possibility is that TGFP modulates the synthesis and/or release of neurotrophins from the ganglionic

VOLUME 152,1992

nnc. Each of these effects of TGFP on the nnc may contribute to enhanced survival and subsequent differentiation of DRG neurons in the present study. The data at the present stage argue in favor of an action of TGFPl mediated indirectly by the nnc, since the degree of neurotrophic effects was significantly diminished in AraC vs FdUr treatment, resulting in nnc depletion. However, since depletion was incomplete one cannot rule out a direct effect on the neurons. For instance, TGFPl promotes the i?z vitro survival of El4 rat motoneurons grown on a layer of lysed astrocytes, implicating a direct effect on the neurons (Martinou et ul., 1990). Similar to a possible role of TGFP in modulating a neurotrophin synthesis/release from nnc, it could also perform this function directly on the neurons themselves, so that the neurotrophin(s) could act in an autocrine fashion as is the case for BDNF and NT-3 in El3 mouse DRG neurons (Mark Bothwell, personal communication). TGF/3 tend DRG Neurom i?l Viw

There are several observations consistent with an in ciao role for TGFD in the differentiation and maintenance of DRG neurons. For example, TGFB2 is expressed in the developing peripheral nervous system (Roberts and Sporn, 1990b; Flanders ef al., 1991) and TGFP inhibits bFGF-induced pigmentation of E7 embryonic quail DRG cultures, suggesting a possible role on the fate of certain DRG cells early in development (Stocker et al., 1991). TGFP may also exert effects from peripheral targets since it has been localized in tissues which are innervated by sensory neurons. TGF@l protein and its mRNA (Heine ef al., 1987; Lehnert and Akhurst, 1988) have been detected during mouse embryo development in areas of mesenchyme interacting with epithelium. Furthermore, TGFP mRNA levels are increased by the tumor promoter TPA in mouse epidermis (Akhurst et nl., 1988), suggesting that this factor can be present in this sensory neuron target. Finally, a role for TGFP in sensory nerve regeneration is possible. TGFP is found in platelets (Assoian et aZ.,1983) and is secreted by macrophages (Assoian ef al., 1987). These are cell types attracted to wound areas and sites of nerve injury and are implicated in the role of TGFfi in accelerating healing (Mustoe et cd., 1987). In this context it is interesting that macrophages are suspected to play a role in peripheral sensory nerve regeneration by an indirect mechanism involving NGF (Brown et al., 1991). In summary, the present study has shown that TGFpl and TGF(32 increase the survival of cultured DRG neurons and level of the neuropeptide SP. This action appears to depend on the presence of NGF and/or crossreactive members of the neurotrophin family. TGF@l

may promote this effect by direct interaction with DRG neurons and/or indirectly via interaction with ganglionic nnc. Clearly these observations raise several interesting issues, resolution of which will in part depend on the anticipated availability of suitable antibodies that specifically recognize each member of the neurotrophin family. We thank Ms. MaryJane Dougherty for carrying out the Northern blots for mRNA NGF in these cultures. Thanks are due to Dr. Lloyd Grwne and Dr. Tom Van Dc Water for critical reading of the manuscript and for Ms. Antoinette Barnecott for its preparation. This work was supported by NIH Grants NS26766 to il.C., NS26125 to R.S.M., and NS20013 and NS20778 to J.A.K.

;VO~CJotltl~~tl !I, yo~$ While this manuscript was in review a report appeared showing that TGFdl immunorcactivitg is confined to connvctive tissue of peripheral ganglia and that TGF&‘and TGFfiS-immunoreactivc labeling was localized in DRG neurons, satellite, and Sch\vann cells (Flanders VI trl., 1991; ITnsickcr cat (I/., 1991 ).

REFERENCES r\chcson. A., Barker, P. A, Alderson, R. F., Miller, F. D.. and Murphy, R. A. ( 1991). Detection of hrain-derived neurotrophic factor-like activity in fibroblasts and Srhwann wlls: Inhibition by antihodies to iXGF. ,~v~t~otr 7, l-20. Adler. J. E., Kessler, J. A., and Black, I. B. (198-1). Development and wgulation of substance P in sensory neurons iv c>itt.r~. Z~C,,,. Viol. 102, 117-225. :\khurst, J. R., Fw, P., and Balmain, -4. (1988). Localized production of TGF-,I mKNA in tumour I)romotcr-stimulated mouse epidermis. ,VU/l~N~ 351, 33wx5. Assoian. R. Ii., Komoriya. rl., Meyers, C. A., Miller, I). M., and Sporn, M. I:. (19X3). Transforming growth factor-i,! in human platelets. J. I:ic,/. (‘I/(,/,/. 258, 715;,~7lW. ;2ssoian, R. Ii., Flcurdclys, B. E., Stevenson, H. C., Miller, P. J., Madtcs. I). K., Raines, E. R., Ross, R.. and Sporn, M. 8. (1987). Expression and secretion of type /j transforming growth factor h\ activated human macrophages. Pvw. ,“icctl. ;Ictrrl. S’ci I%‘~1 84, 6020w.24. Baird, A., and BGhlcn, P. C1989~. Fibroblast Growth Factors. It/ “Peptide Growth Factors and Their Receptors: Handbook of Expcrimrw tal Pharmacology” (M. B. Sporn and A. B. Rohrrts, Eds.), Vol. 95/l, pp. X%llX. Springer-\:erlat(, Hcidclbcrg. Bcrkvmc~ier. L. R., Winslow, J. TV., Kaplan, D. R., Nikolics, K., Goeddel, I). V., and Rosenthal, A. (1991). Neurotrophin-5: -4 novel neurotrophic factor that activates trk and trkB. ,“\:t~rrron 7, 857-%X. Besnard, F., Lawrence, D., Sensenbrenner, M., and Labourdette, G. (1989). Modulation by Transforming Growth Factor $ (TGFP), of the mitogenic effect of fihroblast growth factor (FGF) on rat oligodcndrocytcs in culture. (‘.K. S’~w~rw.s H&l. Acctrl. S’ci. 308, 287-292. Brown, M. C., Pvrry, V. H.. Lunn, E. R., Gordon, S., and Heumann, R. (1!)91). Macrophage depcndcncc of peripheral sensory nerve reyeneration: I’ossihle involvement of nerve growth factor. ,VfJ/rwr/ 6, m-:j’io. (‘halazonitis, A., and Fischbach, (:. D. (1980). Elevated potassium induces morphological differentiation of dorsal root xanglionic neurons in dissociated cell culture. DC>{%. Riol. 78, 173-183. Chalazonitis. A., Kalberg, J. A.. Twardzik, D. R., Morrison, R. S., and Kvsslvr. J. A. (1991). Translorminggrowth factor 6 modulates NGF

effects on cultured sensory neurons. 17, “Societv of Neuroscience Abstracts, New Orleans, LA,” Vol. 17, p, 731, No: X01.13. C’halazonitis, A., Kessler, J. 8.. and Morrison, R. S. (1989). Transforming growth factors o and /j in contrast to cpidcrmal growth factor stimulate survival of sensory neurons in ~li!ro. 1,~ “Society of Neuroscience Abstracts. Phoenix, AZ,” Vol. 15, p, 1X1, No. 536.2. C’halazonitis, A., Kessler, J. A., Twardzik, D. R., and Morrison, R. S. (1992). Transforming growth factor alpha hut not epidcrmal growth factor, promotes the survival of sensory neurons i,r /,ifw. ./. :V~a//r,r,~ sci. 12(L), 583-593. Cheifetz, S., TVeatherbec, J. A., Tsang, M. L.-S., Anderson, J. A., Mole, J. E., Lucas, R., and Massagu6, J. (1987). The transforminK growth factor-/j system, a complex pattern of cross-reactive ligands and receptors. ($/I 48, .109-115. Chomczynski, P., and Sacchi, N. 11987). Single-step method of RNA isolation hy acid guanidinium thiocyanate-phenol-chloroform estraction. :lrctrl. H/oc+if,ri/. 162, 156-l%). Eccleston, P. A., Jessen, K. R., and Mirsky, R. (1989). Transforming growth factor-,i and y-interftaron have dual effects on aron-th ol peripheral glia. .I. ~\T~~~~r~~sc~i.Rcs. 24, 524-530. Edwards, R. Il., Selhy, M. J.. and Rutter, R. J. (19861. Differential RNA splicing predicts two distinct nerve growth factor precursors. .vut/c,v (f,or/tlorr) 319, 7X1-787. Ellingsworth, L. R., Brennan, K.. Fok, K., Rosen, D. M., Bentz, H.. I’icz, K. A., and Seycdin, S. M. (19%). Antihodies to the N-terminal portion of cartilage-inducing factor A and transforming growth factor/j. ,I. Biol. (‘I/o,//. 261, 1%,362%1%,367. Flanders, K. C’., Ludeckc, G., Engels, S., Cissel, D. S.. Robrrts, A. I~., Kondaiah, P.. Lafyatis. R.. Sporn, M. B., and IJnsicker. K. (1991). I,ocalization and actions of transforming growth factor-/is in the txmbr\-onic ncr\ous system. l)cc,c’/o~f,rr~~,,lt 113(l), 18:3&191. Gentry. L. E., Webb, N. R., Lim, G. J., Brunner, A. M., Ranchalis, J. E., Tbvardzik. I). I~., Lioubin, M. L.. Marquardt, H., and Purchio. A. F. (19X7). Type 1 transforming growth factor beta: ,\mplifietl erprcssion and secretion of maturr and precursor polypeptides in chincsc secretion hamster ovary cells. Xlo/. (‘cJ//. Biol. 7, 3llii~3~27. Greene, L. A. (1977). Quantitative in i,itro studies on the nerve growth factor requirement of neurons. II. Sensory neurons. I)f,v. Rio/. 58, 106-11X Heinc. 11. I., Rlunoz, E. F., Flanders, K. C’., Ellingsworth, 1~. K., Lam, II-Y. P., Thompson, N. I,., Robrrts, A. B.. and Sporn. M. 13. (1987). Role of transforming growth factor-/j in the derclopmcnt of the mouse vmhryo. *J. Cdl Rid. 10.5, 286-2876. Hohn, A., Lcibrock. J., Bailey, K.. and Barde, Y-A. (l!WU). Identitication and characterization of a novel memher of the nerve growth factor/brain derived neurotrophic factor family. .\v~rt/(r~c’ UJ, X3!)& 3,ll. Hallbook, F., Ibanez, (:. F., and Prrsson, H. 11991 I. E\-olutionary studits of the nerve growth factor family reveal a novel member abundantlJ- exprcsstd in ,%,io/tvs ovary. Sf~/crnt/ 6, 813-858. Ignotz, R. A., Endo, T., and Massague, J. (19x7). Regulation of libronectin and type 1 collagen mRNA lex-vls bg transformingxro\l-th fxtor@. ,J. Bid (‘hc,n/, 262, 6443-6446. Kessler, J. A., and Black, I. 11980). Nerve growth factor stimulates the derclopment of substance P in sensory ganglia. t’rr~c. :V(irtl. :lctrtl. Sri.

IYid,l

77, 6X-651.

Lee, V. M.-Y., Carden, M. J., Schlaepfer. W. W., and Trojanowski. J. Q. (1987). Monoclonal antibodies distinguish several differentially phosphorylated states of the two largest rat nrurofilamcnt subunits (NF-H and NF-M) and demonstrate their existence in the normal ncbrvous systcam of adult rats. J. A’c~~twci 7, :317&W%. Lcfebvre, P. P., Staerker, H., Weber. T., Van De Water, T. R., Register, B., and Moonrn, G. ( 1991). TGFjfl modulates bFGF receptor mes-

132

DEVELOPMENTALBIOLOGY

sage expression in cultured adult auditory neurons. ,\‘eu~t2e~~rt 2, 305-308. Lehnert, S. A., and Akhurst, R. J. (1988). Embryonic expression pattern of TGF beta type 1 RNA suggests hoth paracrine and autocrine mechanisms of action. ~~rlo~~rt)r~f 104, 263-273. Leihrock, J., Lottspeich, F., Hohn, A., Hofcr, M., Hengerer, B., Masiakowski, P., Thoenen, H., and Barde, Y-A. (1989). Molecular cloning and expression of brain-derived neurotrophic factor. Nrtfuw 341,

149-152. Lindholm, D., Hengerer, B., Zafra, F., and Thocnen, H. (1990). Transforming growth factor+ stimulates expression of nerve growth factor in the rat CNS. iVe?~toRe~~ot? 1, 9-12. Lindsay, R. M., Thoenen, H., and Barde, Y--A. (1985). Placode and neural crest-derived sensory neurons are responsive at early developmental stages to brain-derived nrurotrophic factor. UPI.. Biol. 112,319-328. Lu, X., and Richardson, P. M. (1991). Inflammation near the nerve cell body enhances axonal regeneration. .J, Are~rrosci 11, 972-978. Maisonpierre, P. C., Belluscio, L., Squinto, S., Ip, N. Y., Furth, M. E., Lindsay, R. M., and Yancopoulos, G. D. (1990). Neurotrophin-3: A neurotrophic factor related to NGF and BDNF. Scicvcc 247, 14461451. Martinou, J. C., Le Van Thai, A., Valette, A., and Weher, M. J. (1990). Transforming growth factor $1 is a potent survival factor for rat embryo motoncurons in culture. UPC,. Brcrin Res. 52, 175-181. Massague. J. (1987). The TGF-0 family of growth and differentiation factors. Cf4 49, 437438. Matsuoka, I., Meyer, M., and Thoenen, H. (1991). Cell-type-specific regulation of nerve growth factor (NGF) synthesis in non-neuronal cells: Comparison of Schwann cells with other cell types. J. .~WW Sri. 11, 316553177. Mustoe, A., Pierce, G. F., Thomason, A., Gramates, P., Sporn, M. B., and Deuel, T. F. (1987). Accelerated healing of incisional wounds in rats induced by transforming growth factor-/j. Scie)!ce 237, 13331336. Nichols, N. R., Laping, N. J., Day, J. R., and Finch, C. E. (1991). Increases in transforming growth factor-D mRNA in hippocampus during response to entorhinal cortex lesions in intact and adrenalectomized rats. J. Xet~rosci. Rex 28, 134-139. Plouet, J., and Gospodarowicz, D. (1989). Transforming growth factor /j-l positively modulates the hioactivity of fibrohlast growth factor on cornea1 endothelial cells. J. Cell Physiol. 141, 392-399.

VOLUME 152.1992 Powell, D., Leeman, S., Tregear, G., Niall, H., and Potts, J. (19’73). Radioimmunoassay for substance P. lvuf((~c (Londorli 241, 2522.54. Ranchalis, J. E., McPherson, J., Ogawa, Y., Segedin, S. M., and Twardzik, D. R. (1987). Both forms of cartilage inducing factor inhibit the growth of tumor cells in a similar manner. Biochenc. Rioph!/s. RPS.

(5,/t/ IJ//I I/. 148(2), 783-789. Ridley, A. J., Davis, J. 8.. Stroohant, P., and Land, H. (1989). Transforming growth factors-ill and -1-12 are mitogens for rat Schwann cells. J. (‘~11 Bid. 109, 3419-3424. Roberts, A. B., and Sporn, M. B. (1988). Transforming growth factor 6. Iti “Advances in Cancer Research,” Vol. 51, pp. 107-145. dcademic Press, San Diego, (‘A. Roberts, A. B., and Sporn, M. B. (1990a). The transforming growth factors-beta. f)/ “Peptide Growth Factors and Their Receptors: Handhook of Experimental Pharmacology” (M. B. Sporn and A. B. Roherts, Eds.), T’ol. 95/l. pp. 419-472. Springer-Verlag, Heidelberg. Roherts, A. B., and Sporn, M. B. (1990b). Transforming growth factor$: Multifunctional regulator of differentiation and development. Ph ilos.

Trtr t,s. It! SK

Lot/dot/

SP,:

B 327,

145-154.

Rodriguez-T&bar, A., Dechant, G., and Barde, Y. A. (1990). Binding of hrain-derived neurotrophic factor to the nerve growth factor receptor. Xeir~i/ 4, 487-492. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Extraction, purihcation and analysis of messenger RNA from eukaryotic cells. It/ “Molecular Cloning, ” 2nd ed. Vol. 1, Chap. 7. Cold Spring Harhoi Laboratory Press, Cold Spring Harbor, NY. Seyedin, S. M., Thomas, T. C., Thompson, A. Y.. Rosen, D. M., and Piez, K. A. (1985). Purification and characterization of two cartilagc-inducing factors from bovine demineralized bone. Plot. 1vutl. ilcorl. sci.

IWA .

82 , 2267-8271.

Sporn, M. B., Roherts, A. B., Wakefeld, L. M., and de Crombrugghe, B. (1987). Some recent advances in the chemistry and biology of transforming growth factor-/j. J. Cell Bid. 105, 1039-1045. Stocker, K. M., Sherman, L., Rees, S., and Ciment, G. (1991). Basic FGF and TGF/1-I influence commitment to melanogenesis in neural crest-derired cells of avian embryos. Lk~rdopvrrnf 111, 635-645. IJnsicker, K., Flanders, K. C., Cissel, D. S., Lafyatis, R., and Sporn, M. B. (1991 I. Transforming growth factor heta isoforms in the adult rat central and peripheral nervous system. 1VeccroscicJt/ce 44, 613625. Yamamoto, H., and Gurney, M. E. (1990). Human platelets contain hrain-derived neurotrophic factor. .I. ivvrcrosci. 10, 34693478.