Development 101, 605-615 (1987)Printed in Great Britain © The Company of Biologists Limited 1987

605

Cell-cell interactions modulate the responsiveness of PC12 cells to

nerve growth factor

PATRICK DOHERTY, DEREK A. MANN and FRANK S. WALSH

Institute of Neurology, Queen Square, London WC1N 3BG, UK

Summary

The growth of PC 12 cells on a collagen substratum oron monolayers of several non-neuronal cell types wasstudied by measuring nerve growth factor (NGF)-dependent increases in the expression of a 150 x103 (Mr) neurofilament protein subunit and the mem-brane glycoprotein Thy-1. Both responses were foundto be greatly suppressed in cultures of fibroblasts ascompared to the C2 and G8-1 muscle cell lines and theC6 glioma cell line. This suppression was associatedwith an inhibition of NGF-dependent neuritic out-growth from PC12 cells grown on fibroblast mono-layers. There was no evidence that fibroblasts secretesoluble molecules that directly inhibit these responsesor neutralize NGF. In addition, there was no differ-ence in the neurofilament protein response from PC 12

cells that had been treated with NGF prior to cocul-ture, and the now primed PC12 cells readily extendedaxons over fibroblast monolayers. These data demon-strate that cell-cell and/or cell-matrix interactionscan modulate biochemical responses to NGF andsuggest that responsiveness of neuronal cells to en-vironmental cues is not immutable. Control of thelatter may be at the level of expression of receptormolecules for cell-surface- or matrix-associatedmacromolecules and a similar mechanism operatingduring development could play a role in growth coneguidance.

Key words: nerve growth factor, PC12 cells, neuriteoutgrowth, cell-eel] interactions.

Introduction

Although little is known about the molecular natureof the components that operate to ensure correctinnervation pattern formation during development,regulatory mechanisms have been suggested toinclude the availability of soluble neuronotrophicmolecules such as nerve growth factor (NGF)(Thoenen & Barde, 1980), as well as spatiotemporalchanges in the expression of cell surface and matrixcomponents in apposition to the growth cone (Jacob-son, 1978; Letourneau, 1982; Goodman et al. 1984;Edelman, 1984).

When bound to tissue culture plastic, relativelycommon extracellular matrix molecules such as fibro-nectin (FN) (Rogers et al. 1983), laminin (LN)(Manthorpe et al. 1983) and a LN-heparin sulphateproteoglycan complex (HeSPG) (Lander, Fujii &Reichardt, 1985) can all promote vigorous neuriteoutgrowth by both peripheral and central neurones.

Two independent McAbs, CSAT and JG22, that reactwith a set of glycoproteins exhibiting functionalproperties of a LN and FN receptor (Horwitz et al.1985) have been shown to inhibit neurite outgrowthon a variety of extracellular matrix substrata, includ-ing the LN-HeSPG complex (Tomaselli, Reichardt& Bixby, 1986). In the same study, both McAbs werereported to block neurite outgrowth on monolayercultures of intact fibroblasts, but not astrocytes,Schwann cells or skeletal muscle. Both central(Noble, Fok-Seang & Cohen, 1984) and peripheral(Fallon, 1985) neurones show preferential growth oncultured astrocyte monolayers as compared withfibroblasts, even though cell surface LN was found inassociation with the latter and not the former (Fallon,1985). These observations suggest the existence of asecond and perhaps more specialized cell-membrane-associated recognition pathway operating indepen-dently of the LN/FN receptor and the LN-HeSPGcomplex. Consistent with this contention, it has

606 P. Doherty, D. A. Mann and F. S. Walsh

recently been shown that a cocktail of antibodies tovarious cell adhesion molecules, including the puta-tive receptor for LN and FN as well as the neuralcell adhesion molecule (N-CAM), are required toinhibit neurite outgrowth from chick ciliary ganglionneurones grown on muscle monolayers (Bixby, Pratt,Lilien & Reichardt, 1987).

The rat pheochromocytoma cell line, PC12(Greene & Tischler, 1976), is useful for studies on themechanism of neurite outgrowth as these cells do notrequire exogenous trophic support for survival andmay undergo either generation (naive cells) or regen-eration (primed cells) of neurites in response to NGF(Greene, Burstein & Black, 1982; Greene, 1984). Anadditional advantage of PC12 cells is that, unlikemost primary neurones that are available for culture,they may be used to study steps whereby a classicalgrowth factor (NGF) stimulates neuritogenesis fromcells previously unexposed to detectable levels of thisfactor.

We have developed enzyme-linked immuno-adsorbent assays (ELISA) to index two independentand complementary NGF-dependent responses inPC12 cells. We have shown up to 30-fold increases inthe expression of a neurofilament protein antigen,recognized by a McAb-coded RT97 (Wood & Ander-ton, 1981), to be associated with NGF-induced mor-phological differentiation of both naive and primedPC12 cells (Doherty, Mann & Walsh, 1987). Incontrast, the NGF induction of the Thy-1 glyco-protein (Morris, 1985), which can also readily bemeasured by ELISA assay, is a response that is notdirectly associated with morphological differentiation(Doherty & Walsh, 1987a).

In the present study, we have determined theability of differing non-neuronal cellular monolayersto modulate both of these responses. The relationshipbetween the induction of both antigens and themorphological differentiation status of individualPC12 cells has been evaluated. The results obtainedsuggest that, whereas PC12 cells may constitutivelyexpress receptors for the specialized adhesion mol-ecules associated with muscle and glioma, the ex-pression of receptors for common adhesive moleculessuch as FN may be controlled by extrinsic factorsincluding target-derived growth factors. Similar con-trol during development may be important in guidingaxons to their target tissues.

Materials and methods

Cell culturePC12 cells were grown in SATO media (Bottenstein, 1985)consisting of Dulbecco's modified Eagle's medium(DMEM) supplemented with (in mgl"1) progesterone,0-062; putrescine, 16-1; thyroxine, 0-4; selenium, 0-039;

transferrin (human), 100; insulin (bovine pancreas), 10; tri-iodothyronine, 0-377. (All reagents from Sigma ChemicalCompany). The medium was further supplemented at 1 %(v/v) with Path-o-cyte 4 (bovine serum albumin solutionfrom Miles Chemical Company). For subculture of naivecells and replating of primed cells (see below), culturesgrown on collagen-coated tissue culture plates were washedonce with versene solution (Gibco) followed by incubationfor 5-7 min with 0-05% (w/v) trypsin in the same buffer.Detached cells were centrifuged and resuspended by tritu-ration with a pipette followed by a 19-gauge syringe needle.For priming, approximately 4x 10* cells were plated onto acollagen-coated 100 mm tissue culture dish in media sup-plemented with 50ngml"' NGF. The B-subunit of 7S NGFwas a generous gift from Dr A. Leon (Fidia ResearchLaboratories, Abano Terme, Italy). Fresh NGF was addedto cultures at 3-day intervals.

Monolayer cultures of G8-1 (Christian, Nelson, Peacock& Nirenberg, 1977) and Cl (Yaffe & Saxel, 1978) myotubeswere established as previously described (Walsh & Phillips,1981) in individual wells of 96-well microtitre plates. C6glioma (Benda et al. 1986) and human skin fibroblasts weregrown in microwells in DMEM supplemented with 10%FCS until confluent.

Cocultures were established by seeding 20000 PC12 cellsonto monolayers of the above cell types. For immuno-fluorescence, 50000 PC12 cells were seeded onto mono-layers grown in a 35 mm culture dish. In some experiments,PC12 cells were seeded into individual wells on a collagen-coated microtitre plate. Cocultures were maintained for upto 7 days in serum-containing media (5 % horse serum and2-5 % foetal calf serum) or alternatively for up to 3 days indefined media (see above). Monolayers were washed twicewith DMEM and a further two times with the respectivecoculture media prior to the addition of PC12 cells.

Conditioned media preparationSATO medium was conditioned over confluent cultures ofskin fibroblasts and C2 myotubes for 1 or 4 days. Theconditioned medium was collected, filtered through a0-2 ̂ m filter to remove debris and added directly to PC12cell cultures. In addition, DMEM was conditioned overG8-1 myotubes and C6 glioma monolayers for 24 h, concen-trated approximately 10-fold by ultrafiltration as describedpreviously (Doherty, Dickson, Flanigan & Walsh, 1986a)and then added to PC12 cell cultures at up to 50 % (v/v).

lmmunochemical analysis of neurofilament and Thy-1levelsCultures were fixed by a 50 % medium exchange with 4 %paraformaldehyde in DMEM for 60 min at 20°C, followedby a total medium exchange for a further 60 min. Cells werepermeabilized by treatment with methanol (—20cC) for afurther 15 min. Cultures were then washed three times withphosphate-buffered saline (PBS) and the relative level ofneurofilament protein antigen or Thy-1 protein determinedessentially as previously described (Doherty, Dickson,Flanigan & Walsh, 1984a,6; Doherty & Walsh, 1987a;Doherty et al. 1987). Briefly, following incubation with PBScontaining 0-5 % gelatin to block nonspecific protein-binding sites, cultures were incubated for 60 min at 20°C

PC12 cells responses to NGF 607

with a 1:1500 dilution of McAb RT97 ascites or alterna-tively a 1:50 dilution of McAb OX7 tissue culture super-natant. The latter was a generous gift from Dr Neil Barclay.Both reagents were in PBS containing lmgml"1 bovineserum albumin (BSA). Cultures were washed three timeswith PBS/BSA and incubated with a 1:1000 dilution ofhorseradish peroxidase-conjugated anti-mouse immuno-globulin for a further 60min. Cultures were then washedfour times with PBS, twice with distilled H2O2, and finallyincubated with 50 fA of 0-2% (w/v) o-phenylenediamine(OPD) and 002% (v/v) H2O in citrate buffer. After10-20 min the conversion of OPD to its oxidized productwas stopped by the addition of 50/xl of H2SO4 and opticaldensity (O.D.) determined at 492 nm using a Flow TitretekMultiscan apparatus.

Indirect immunofluorescenceIndirect immunofluorescence on both live and fixed cul-tures was as previously described (Moore & Walsh, 1985;Doherty et al. 1987). The OX7 McAb was used at a 1:50dilution and RT97 at a 1:1000 dilution.

Results

NGF-induced neurofilament protein responses incoculturesMcAb RT97 shows very little binding to undifferen-tiated PC12 cells, with up to 30-fold increases inbinding associated with morphological differentiationinduced by NGF (Doherty et al. 1987). Fig. 1 showslocalization of McAb RT97 immunoreactivity in atypical culture of NGF-treated PC12 cells using in-direct immunofluorescence. Staining is apparent,sparse in the cell body but intense along the fulllength of growing axons. Immunoreactivity can alsobe found associated with the growth cones. In NGF-treated PC12 cells, immunoblot analysis shows McAbRT97 to bind specifically to the 150xl0i(Mr) subunitof neurofilament protein (Doherty et al. 1987).

The effect of the cellular environment on this NGF-dependent response was determined by seeding PC12cells onto confluent monolayer cultures of either C6glioma, G8-1 skeletal myotubes or human skin fibro-blasts. Fig. 2 shows the relative level of RT97 bindingto cultures maintained in serum-containing growthmedium for 7 days. In the absence of exogenousNGF, significant time-dependent increases in McAbRT97 binding are found only in C6 glioma coculturesand not muscle or fibroblast cocultures. In serum-containing, but not defined, media, C6 has previouslybeen shown to secrete an NGF-like activity (Wester-man & Unsicker, 1986). NGF-dependent increases inMcAb RT97 binding to PC12 cells are apparent byday 5 in both muscle and glioma, but not thefibroblast cocultures. By day 7, there was a tenfoldand fivefold greater binding of RT97 to NGF-treatedPC12 cells grown on monolayers of glioma andmuscle, respectively, as compared to fibroblasts.

Morphological differentiation induced by NGFoccurs more rapidly in defined, as compared toserum-containing, media (P. Doherty, unpublisheddata). Fig. 3 shows the effect of differing concen-trations of NGF on the binding of McAb RT97 toPC12 cells grown for 3 days on a collagen-coatedsubstratum or alternatively on monolayer cultures ofC2 muscle cells or skin fibroblasts. There were 3-5-and 3-0-fold greater neurofilament protein responsesfor cells grown on collagen or C2 muscle, respect-ively, as compared to fibroblast monolayers. Anexamination of the dose-response curve clearlyshows responsiveness rather than sensitivity to NGFreduced in the fibroblast coculture.

The morphology of PC12 cells cultured on mono-layers of C2 muscle cells and skin fibroblasts washighlighted by staining live cultures with the rat-

Fig. 1. Anti-neurofilamentimmunostaining of PC12 cells. PC12cells were primed with NGF(SOngmP1) for 5 days, divested oftheir neuritic network and replatedonto a collagen-coated 35 mm tissueculture plate in the presence ofNGF (50ngml"') for a further 3-day period before being fixed andstained with RT97. Neurofilamentscan be seen to be localized toaxonal processes withimmunoreactivity present at thegrowth cone. Scale bar is 40£im.

608 P. Doherty, D. A. Mann and F. S. Wals

100 rloo r B

50

9 C6 gliomaD G8 myotubesA fibroblasts

Days in culture

Fig. 2. Time course of RT97 binding to PC12 cells grownon monolayer cultures of C6 glioma, G8-1 myotubes andskin fibroblasts in the presence (A) and absence (B) ofNGF (20ngmr') . For each data point RT97 binding wasdetermined for 8 independent cocultures with nonspecificbinding measured against sister monolayer culturessubtracted. Nonspecific binding did not vary as a functionof time or NGF concentration and the second antibodyon its own showed no significant binding to cocultures(data not shown). The results show the mean ± S.E. ofeight independent determinations and 100 arbitraryunits = 0-202O.D. units.

specific McAb OX7 reactive with Thy-1 antigen.After 3 days of culture in SATO media, there isrelatively little, if any, sign of neuritic outgrowth fromcells grown in the absence of NGF (Fig. 4A,B,E,F).In the presence of NGF, there was a definite neuriticresponse from the vast majority of PC12 cells cul-tured on muscle cells (Fig. 4G,H). PC12 cells dis-played relatively long and highly branched axonalprocesses with highly differentiated growth coneslocalized to the surface of the muscle. McAb RT97binding was again localized to neurofilaments presentwithin axonal processes (data not shown). In con-trast, PC12 cells showed no sign of morphologicaldifferentiation when cultured in the presence of NGFon fibroblast monolayers (Fig. 4C,D).

O collagenD C2 myotubes0 fibroblasts

0-2 1 5NGF(ngmr')

Fig. 3. PC12 cells were grown for 3 days on a collagensubstratum, on C2 myotube monolayers or on fibroblastmonolayers. The results show the absolute increase inRT97 binding to cocultures grown in the presence ofNGF as compared to sister cultures grown in the absenceof NGF. There was no significant difference in RT97binding to cocultures grown in the absence of NGF (datanot shown). Each value is the mean of four independentdeterminations and 100 arbitrary units = 0078 O.D. units.Standard errors of means ranged from 2-20 arbitraryunits.

Effect of 'priming' on NGF-induced responses incoculture

Cells primed by NGF over 5-7 days and subsequentlydivested of their neuritic network show a much morerapid formation of the network and accumulation ofneurofilament protein antigen on re-exposure toNGF than naive cells (Doherty et al. 1987). Fig. 5shows two examples of the neurofilament proteinresponse for primed cells grown on monolayers ofskin fibroblasts as compared with C6 glioma orfibroblasts compared with C2 muscle and collagen. Inthe former case, cultures were maintained in serum-containing media; in the latter, they were grown indefined media. In both examples, there was essen-tially no binding of McAb RT97 to PC12 cells in

PC12 cells responses to NGF 609

Fig. 4. PC12 cells were grown for 3 days on monolayer cultures of skin fibroblasts and C2 muscle. PC12 cellmorphology for the respective cultures was highlighted by staining live cultures with the OX7 McAb. The cultures werefixed with methanol and viewed under phase-contrast and epifluorescence optics. The micrographs show representativefields for PC12 cells grown in the absence of NGF on fibroblasts (A,B) and muscle (E,F), and in the presence of NGFon fibroblasts (C,D) and muscle (G,H). The scale bar represents 40^m.

610 P. Doherty, D. A. Mann and F. S. Walsh

100

EQ.

50

# C6 gliomaD fibroblasts

• collagenA C2 myotubesD fibroblasts

0-2 0-6 17 50 0-2NGF(ngmr')

25

Fig. 5. PC12 cells were primed for 5 days with NGF (50ngml ') before being subcultured onto monolayers of C6glioma and skin fibroblasts in one experiment or monolayers of C2 myotubes, skin fibroblasts and a collagen substratumin a second independent experiment. In the first experiment, cocultures were grown in serum-containing medium and inthe second in defined medium. After 48h the NGF-dependent increases in RT97 binding were determined as in Fig. 3.Each value is the mean ± S.E. of four independent determinations. In the first experiment, 100 arbitraryunits = 0-2630.D. units and in the second, 0138O.D. units.

cultures grown in the absence of exogenous NGF.However, substantial NGF-dose-dependent increasesin McAb RT97 binding were apparent after a 48 hperiod of coculture. In contrast to the results ob-tained with naive cells, the nature of the non-neuronal cellular monolayer had very little effect onthe relative increase in the expression of the neuro-filament protein antigen induced by both submaximaland maximal concentrations of NGF. Again, in con-trast to naive cells, primed cells showed an extensiveNGF-dependent neuritic outgrowth on fibroblastmonolayers. This was clearly seen in cultures of bothlive unstained cells grown in microwells (Fig. 6A,C)and also low-density cell cultures stained with theMcAb OX7 (Fig. 6B,D). On C2 monolayers, themorphological response was qualitatively similar forprimed and naive PC12 cells. There was essentially noPC12 cell differentiation in cultures grown in theabsence of NGF (Fig. 6E,F) with an extensive neuriteoutgrowth apparent from the vast majority of cellsgrown in the presence of NGF (Fig. 6G,H). The

morphological observation for cultures of naive andprimed PC12 cells further supports the contentionthat the relative expression of the neurofilamentprotein antigen recognized by RT97 is directly relatedto the degree of neuritic outgrowth induced by NGF.

NGF-induced increases in Thy-1 antigen in cocultureIn cultures of naive PC12 cells, NGF can induceaccumulation of the mRNA species encoding Thy-1(Dickson et al. 1986). NGF-induced accumulation ofthe Thy-1 glycoprotein (Richter-Landsberg, Greene& Shelanski, 1985) can be conveniently measured byELISA assay (Doherty & Walsh, 1987a). The OX7McAb reacts specifically with rat Thy-1 (Mason &Williams, 1980) and we have taken advantage of thisto quantify the relative expression of antigen incultures of PC12 cells grown for 3 days on monolayersof human skin fibroblasts and C2 muscle cells. PC12cells grown on a collagen substratum served as apositive control. Fig. 7 shows the relative level ofexpression of Thy-1 antigen as a function of NGF

PC12 cells responses to NGF 611

Fig. 6. PC12 cells were primed as in Fig. 5 and subcultured onto monolayers of fibroblasts or C2 muscle for 48 h. Phase-contrast micrographs show representative fields of live cocultures of PC12 cells grown in the absence (A) and presence(B) of NGF on fibroblast monolayers. Representative fields of these cocultures stained with OX7 McAb and fixedbefore viewing with epifluorescence optics are shown in B (—NGF) and D (+NGF). Phase-contrast and epifluorescentmicrographs of OX7-stained PC12 cells grown on C2 monolayers are shown for cells grown in the absence (E,F) andpresence (G,H) of NGF.

612 P. Doherty, D. A. Mann and F. S. Walsh

0-7

'g 0-6

Q 0-5

00.S 0-4•oc'£X 0-3

0-2 \0-05 0-25 1

NGF(ngmr ' )20

Fig. 7. Naive PC12 cells were seeded onto a collagensubstratum or fibroblast monolayer and grown for 3 daysin media supplemented with up to 20ngml~1 NGF. Foreach datum point, OX7 McAb binding was determinedfor three independent cocultures with nonspecific bindingto sister monolayer cultures subtracted. The results showthe mean ± S.E. of three independent determinations.Nonspecific binding of OX7 McAb to fibroblasts wasmeasured as 006 ± 0003 O.D. units, and this did notvary as a function of NGF concentration.

concentration for PC12 cells grown on collagen ascompared to skin fibroblasts.

As previously reported, NGF induces a two- tothreefold increase in Thy-1 antigen when PC12 cellsare grown on a collagen substratum. NGF is anti-mitotic for PC12 cells and this increased antigenexpression reflects a response at the cellular levelrather than an increase in cell number (Doherty &Walsh, 1987a). In the absence of NGF, there was nodifference in Thy-1 expression for PC12 cells grown incoculture with fibroblasts as compared to those on acollagen substratum. This is consistent with the mor-phological observations that fibroblasts are not toxicto PC12 cells. However, a total suppression of theNGF-induced increase in Thy-1 was apparent in thefibroblast cocultures.

In the absence of NGF and after the same 3-dayperiod of culture, there was a twofold increase inThy-1 expression by PC12 cells grown on C2 musclecell monolayers as compared to collagen (Fig. 8). Theaddition of a maximally active concentration of NGFto the muscle coculture increased the expression ofThy-1 by only a further 20%. The level of Thy-1antigen expression in NGF-treated cultures did notdiffer for cells grown on collagen as compared to C2muscle cell monolayers. These data suggest thatcoculture with C2 muscle cells can directly induceincreased expression of Thy-1 at the cellular level andthat this response is not additive with that induced byNGF.

EcSi

QO

0-7

0-6

0-5

0-4

•M 0-3

XO

0-2

01

D collagen• C2

o1- Ongmr'NGF

11-5 ngmlNGF

Fig. 8. Naive PC12 cells were seeded onto a collagensubstratum or C2 monolayers and cultures grown in thepresence and absence of NGF (ll-5ngml~') for 3 days.OX7 binding was determined as in Fig. 7. The resultsshow the absolute binding to PC12 cells with binding tothe monolayer subtracted. The latter was measured as006 ± 0-OO4 O.D. units. Bars show -I-1 S.E.

Effects of conditioned media from non-neuronal cellson neurofilament and Thy-1 responses

SATO medium conditioned over skin fibroblasts andC2 myotubes for 1 or 4 days was added at up to 50 %(v/v) to naive PC12 cells grown on a collagen-coatedsubstratum. After three days, the binding of RT97and Thy-1 was determined from cells grown in thepresence and absence of NGF (50ngml~'). Theconditioned medium was without effect on both thebasal and the NGF-induced increases in neurofila-ment protein and Thy-1 glycoprotein expression (datanot shown). Similarly, DMEM conditioned over G8-1myoblasts and myotubes or C6 glioma for 24 h andconcentrated up to tenfold was found to be withouteffect on the binding of RT97 to naive and primedPC12 cells grown for 3 days on collagen. The mediumconditioned with G8-1 has previously been shown tocontain an activity that increases survival of spinal

PC12 cells responses to NGF 613

neurones (Doherty, Dickson, Flanigan, Kennedy &Walsh, 19866).

To determine further if fibroblasts directly neutral-ize NGF or secrete neurotoxic components, confluentmonolayers stabilized on cellulose membranes (Milli-cell-HA, 0-45//m diameter) were carefully placedinto individual wells of a 24-well tissue culture dishcontaining naive PC12 cells grown on a collagensubstratum. Under these conditions and over a 5-dayperiod, fibroblasts failed to inhibit both the morpho-logical differentiation of the naive PC12 cells andthe increased expression of neurofilament proteininduced by exogenously added NGF.

Discussion

The polypeptide hormone, NGF, has been shown tohave wide-ranging and diverse effects throughout theperipheral and central nervous system, and has morerecently been implicated as a modulator of theimmune system (Levi-Montalcini & Calissano, 1986).In the present study, we have determined the effect ofthe cellular environment of three potentially relatedNGF responses, namely (1) the induction of neuro-filament protein, (2) Thy-1 antigen and (3) neuriteoutgrowth. To simplify interpretation, we have stud-ied differentiation of a well-characterized NGF-re-sponsive cell line cultured on monolayers derivedfrom other homogenous cell populations.

NGF-induced neurite outgrowth is closely associ-ated with the accumulation of neurofilament proteinantigen recognized by the McAb RT97. The logic,justification and advantages of primarily quantifyingthe neurofilament protein response rather than con-ventional morphological indices of neuronal differen-tiation have been outlined (Doherty et al. 1984a,6;Doherty et al. 1987; Doherty & Walsh, 1987). Themost conspicuous observation of the present studywas the relative failure of NGF to induce an increasedexpression of the 150 x 103 (Mr) neurofilament subunitwhen naive PC12 cells were cultured on a fibroblastmonolayer. This was associated with a total sup-pression of neurite outgrowth. The differentialneurofilament protein response was made relative toC6 glioma and two independent muscle cell linesunder two differing culture conditions (serum-containing and defined media). In the case of C6glioma, the release of endogenous trophic moleculesmay partially account for a tenfold greater neurofila-ment protein response from PC12 cells grown over 7days in growth medium containing NGF and serum.However, there was no evidence for the release ofsimilar molecules from G8-1 or C2 muscle cultures,with no induction of neurofilament protein or mor-phological differentiation of PC12 cells apparent incocultures maintained in the absence of exogenous

NGF. Primed PC12 cells readily extended axons overthe fibroblast monolayers and medium conditionedover fibroblasts was ineffective in modulating NGF-dependent responses from PC12 cells grown on acollagen substratum. Thus, the relative inhibition ofNGF responses from naive PC12 cells cannot beattributed to an inherent disability of fibroblasts tosupport neurite outgrowth, or to the release ofsoluble components that are toxic to PC12 cells.Similarly, the evidence is not consistent with fibro-blasts directly neutralizing NGF.

As has been suggested by others (see Introduc-tion), differential growth may simply be explained bymuscle serving as a more adhesive substratum forPC12 growth cones than fibroblasts. Candidate ad-hesive molecules present on the former, but not thelatter, have been shown to include N-CAM. Anti-bodies to N-CAM have previously been reported todisrupt adhesion of nerve to muscle (Rutishauser,Grumet & Edelman, 1983); however, others failed tofind inhibition of neurite outgrowth by the soleaddition of N-CAM antibodies (Tomaselli et al. 1986).Inhibition of chick ciliary ganglion neurone out-growth on muscle monolayers was only seen when acombination of antibodies against specialized ad-hesive molecules including N-CAM, as well as anti-bodies to the putative receptor for the commonadhesive molecules FN and LN are added to cultures(Bixby et al. 1987). One possibility is that naive PC12cells constitutively express receptors molecules forspecialized adhesion molecules, but not for commonadhesive molecules such as FN. In support, we havepreviously shown naive PC12 cells to express N-CAMat their cell surface, together with a very rapidincrease in this expression in response to NGFtreatment (Prentice et al. 1987). It has also beenobserved that naive PC12 cells show poor adhesion tofibronectin and that the latter does not support initialneurite outgrowth from these cells (P. Doherty & F.S. Walsh, unpublished data; Fujii, Massoglia, Savion& Gospodarowicz, 1982; Vlodavsky et al. 1982). Analternative possibility that we do not exclude is that,in addition to expressing adhesive molecules thatpromote neurite outgrowth, fibroblasts and possiblyother cell types may express a second class of cell-surface- or matrix-associated proteins that can inhibitneurite outgrowth.

In the present study, we have shown that cocultureof PC12 cells with C2 muscle cells can directly inducean increased expression of the Thy-1 antigen indepen-dently of overt morphological differentiation. Wehave also found that activation of protein kinase C byphorbol esters can also mimic the transcription-dependent induction of Thy-1 by NGF independentlyof morphological differentiation (Doherty & Walsh,19876). The demonstration that induction of Thy-1

614 P. Doherty, D. A. Mann and F. S. Walsh

is also suppressed in PC12-fibroblast coculturessuggests that, rather than solely altering growth conestability, the modulatory interactions between PC12and other cell types may also be operative at the muchearlier level of transduction of the NGF signal.

In contrast to naive cells, primed PC12 cells showno initial difference in their preference for sub-stratum. Extensive NGF-dependent axonal out-growth was readily apparent in fibroblast cocultures,with immunostaining used to highlight extensivedifferentiation from individual PC12 cells. Morpho-logical differentiation was again associated with aninduction of the neurofilament protein antigen andthere was little difference in the magnitude of thisresponse and its sensitivity to NGF for primed cellsgrown on fibroblasts as compared with C6 glioma, C2muscle or collagen substratum. There are severalpossibilities that could account for the differing be-haviour of naive and primed PC12 cells. The mostlikely is that NGF can induce an increased expressionof receptors for common adhesive molecules such asFN; however, we cannot yet exclude the possibilitythat NGF may induce a down-regulation of receptorsfor inhibitory molecules. Both possibilities are cur-rently under investigation. During development,neurones have to make critical decisions as to when toenter a 'common' or 'public' pathway, and similarchoices as to when to exit that pathway. Changes inthe level of the expression of receptor molecules forcommon adhesive molecules may be of crucial im-portance in determining such choices.

In conclusion, the results of the present study haveshown PC12 cell responsiveness to NGF to be criti-cally dependent on the nature of the local micro-environment. Responsiveness, in terms of neuriteoutgrowth, to the cues present in the microenviron-ment is not immutable and can in itself be modified bypre-exposure to NGF. The novel demonstration thatresponses other than overt morphological differen-tiation can also be influenced by the microenviron-ment suggests that control may be exerted at the levelof transduction of the NGF signal. Although wecannot exclude the existence of a short-range orhighly labile soluble molecule controlling the aboveresponses, it appears more probable that control ismediated via direct cell-cell and/or cell-matrix inter-actions. We are currently using the quantitativebioassays to determine the nature of the molecules onthe muscle and fibroblast membranes that modulateneurite outgrowth from PC12 cells.

This work was supported by the Medical ResearchCouncil, the Motor Neurone Disease Association and theWellcome Trust. F. S. Walsh is a Wellcome Trust SeniorLecturer.

References

BENDA, P., LIGHTBODY, J., SATO, G., LEVINE, L. &

SWEET, W. (1968). Differentiated rat glial cell strain intissue culture. Science 161, 370.

BIXBY, J. L., PRATT, R. S., LILIEN, J. & REICHARDT, L. F.

(1987). Neurite outgrowth on muscle cell surfacesinvolves extracellular matrix receptors as well as Ca + + -dependent and -independent cell adhesion molecules.Proc. natn. Acad. Sci. U.S.A. 84, 2555-2559.

BOTTENSTEIN, J. E. (1985). Growth and differentiation ofneural cells in defined media. In Cell Culture in theNeurosciences (ed. J. E. Bottenstein & G. Sato). NewYork, London: Plenum Press.

CHRISTIAN, L. N., NELSON, P. G., PEACOCK, J. &

NIRENBERG, M. (1977). Synapse formation between twoclonal cell lines. Science 196, 995-998.

DICKSON, J. G., PRENTICE, H., JULIEN, J.-P., FERRARI, G.,

LEON, A. & WALSH, F. S. (1986). NGF activates Thy-1and neurofilament gene transcription in rat PC12 cells.EMBOJ. 5,3449-3453.

DOHERTY, P., DICKSON, J. G., FLANIGAN, T. P. & WALSH,

F. S. (1984a). Quantitative evaluation of neuriteoutgrowth in cultures of human foetal brain and dorsalroot ganglion cells using an enzyme-linkedimmunoadsorbent assay for human neurofilamentprotein. J. Neurochem. 42, 1116-1123.

DOHERTY, P., DICKSON, J. G., FLANIGAN, T. P. & WALSH,

F. S. (1984b). Trie effect of nerve growth factor and itsantibodies on neurofilament protein expression inprimary cultures of sensory and spinal neurons.Neurosci. Lett. 51, 55-60.

DOHERTY, P., DICKSON, J. G., FLANIGAN, T. P. & WALSH,

F. S. (1986a). Human skeletal muscle cells synthesise aneurontrophic factor reactive with spinal neurons. / .Neurochem. 46, 133-139.

DOHERTY, P., DICKSON, J. G., FLANIGAN, T. P.,

KENNEDY, P. G. E. & WALSH, F. S. (19866). Effects ofamyotrophic lateral sclerosis serum on cultured chickspinal neurons. Neurology 36, 1330-1334.

DOHERTY, P & WALSH, F. S. (1987a). Control of Thy-1glycoprotein expression in cultures of PC12 cells. / .Neurochem. 49, 610-616.

DOHERTY, P. & WALSH, F. S. (19876). Activation ofprotein kinase C in PC12 cells mimics the NGFinduction of the cell surface TFTY-1 glycoprotein. TransAm. Soc. Neurosci. (in press).

DOHERTY, P., MANN, D. A. & WALSH, F. S. (1987).

Cholera toxin and dibutryl cyclic-AMP inhibit theexpression of neurofilament protein induced by nervegrowth factor in cultures of naive and primed PC12cells. /. Neurochem. (in press).

EDELMAN, G. M. (1984). Modulation of cell adhesionduring induction, histogenesis and perinataldevelopment of the nervous system. A. Rev. Neurosci.7, 339-377.

FALLON, J. R. (1985). Neurite guidance by non-neuronalcells in culture: preferential outgrowth of peripheralneurites on glial as compared to non-glial cell surfaces.J. Neurosci. 12, 3169-3177.

PC12 cells responses to NGF 615

FUJII, K. D., MASSOGLIA, S. L., SAVION, N. &GOSPODAROWICZ, D. (1982). Neurite outgrowth andprotein synthesis by PC12 cells as a function ofsubstratum and nerve growth factor. J. Neurosci. 8,1157-1175.

GOODMAN, C. S., BASTIANI, M. J., DOE, C. Q., DU LAC,S., HELFAND, S. L., KUWADA, J. Y. & THOMAS, J. B.(1984). Cell recognition during neuronal development.Science 225, 1271-1279.

GREENE, L. A. (1984). The importance of both early anddelayed responses in the biological actions of nervegrowth factor. Trends Neurosci. 7, 91-94.

GREENE, L. A., BURSTEIN, D. E. & BLACK, M. M. (1982).The role of transcription-dependent priming in nervegrowth factor promoted neurite outgrowth. Devi Biol.91, 305-316.

GREENE, L. A. & TISCHLER, A. S. (1976). Establishmentof noradrenergic clonal line of rat adrenalpheochromocytoma cells which respond to nervegrowth factor. Proc. natn. Acad. Sci. U.S.A. 72,2424-2428.

HORWITZ, A., DUGGAN, K., GREGGS, R., DECKER, C. &BUCK, C. (1985). The cell substrate attachment (CSAT)antigen has properties of a receptor for laminin andfibronectin. J. Cell Biol. 101, 2134-2144.

JACOBSON, M. (1978). Developmental Neurobiology. NewYork: Plenum Publ. Corp.

LANDER, A. D., FUJII, D. K. & REICHARDT, L. F. (1985).Purification of a factor that promotes neuriteoutgrowth: isolation of laminin and associatedmolecules. 7. Cell Biol. 101, 898-913.

LETOURNEAU, P. (1982). Nerve fibre growth and itsregulation by extrinsic factors. In NeuronalDevelopment (ed. N. C. Spitzer), pp. 213-254. NewYork: Plenum Publ. Corp.

LEVI-MONTALCINI, R. & CALISSANO, P. (1986). Nervegrowth factor as a paradigm for other polypeptidegrowth factors. Trends Neurosci. 9, 473-477.

MANTHORPE, M., ENGVALL, E., RUOSLAHTI, E., LONGO, F.M., DAVIS, G. E. & VARON, S. (1983). Lamininpromotes neurite regeneration from cultured peripheraland central neurons. J. Cell Biol. 97, 1882-1890.

MASON, D. W. & WILLIAMS, A. F. (1980). The kinetics ofantibody binding to membrane antigens in solution andat the cell surface. Biochem. J. 187, 1-20.

MOORE, S. E. & WALSH, F. S. (1985). Differentialexpression of cell adhesion molecule (N-CAM) inskeletal muscle. EMBOJ. 4, 626-630.

MORRIS, R. J. (1985). Thy-1 in developing nervous tissue.Dev. Neurosci. 7, 133-160.

NOBLE, M., FOK-SEANG, J. & COHEN, J. (1984). Glia are aunique substrate for the in vitro growth of centralnervous system neurons. /. Neurosci. 4, 1892-1903.

PRENTICE, H. M., MOORE, S. E., DICKSON, J. G.,DOHERTY, P. & WALSH, F. S. (1987). Nerve growthfactor induced changed in neural cell adhesionmolecule (N-CAM) in PC12 cells. EMBOJ. 6,1859-1863. ;

RICHTER-LANDSBERG, C , GREENE, L. A. & SHELANSKI,M. L. (1985). Cell surface Thy-1-cross-reactiveglycoprotein in cultured PC12 cells: modulation bynerve growth factor and association with thecytoskeleton. J. Neurosci. 5, 468-476.

ROGERS, S. L., LETOURNEAU, P. C , PALM, S. L.,MCCARTHY, J. & FURCHT, L. T. (1983). Neuriteextension by peripheral and central nervous systemneurons in response to substratum-bound fibronectinand laminin. Devi Biol. 98, 212-220.

RUTISHAUSER, U., GRUMET, M. & EDELMAN, G. M.(1983). Neural cell adhesion molecule mediates initialinteractions between spinal cord neurons and musclecells in culture. J. Cell Biol. 97, 145-152.

THOENEN, H. & BARDE, Y.-A. (1980). Physiology ofnerve growth factor. Physiol. Rev. 60, 1284-1335.

TOMASELLI, K. J., REICHARDT, L. F. & BIXBY, J. L.(1986). Distinct molecular interactions mediateneuronal process outgrowth on non-neuronal cellsurfaces and extracellular matrices. J. Cell Biol. 103,2659-2672.

VLODAVSKY, I., LEVI, A., LAX, I., FUKS, Z. &SCHLESSINGER, J. (1982). Induction of cell attachmentand morphological differentiation in apheochromocytoma cell line and embryonal sensorycells by the extracellular matrix. Devi Biol..93,285-300.

WALSH, F. S. & PHILLIPS, E. (1981). Specific changes incellular glycoproteins and surface proteins duringmyogenesis in clonal muscle cells. Devi Biol. 81,229-237.

WESTERMANN, R. & UNSICKER, K. (1986).Neuronotrophic factors released by C6 glioma cells.Soc. Neurosci. Abstr. 12, 1101.

WOOD, J. N. & ANDERTON, B. H. (1981). Monoclonalantibodies to mammalian neurofilament. Biosci. Rep. 1,263-268.

YAFFE, D. & SAXEL, O. (1978). Serial passaging anddifferentiation of myogenic cells isolated fromdystrophic mouse muscle. Nature, Lond. 270, 725-727'.

{Accepted 23 July 1987)