[CANCER RESEARCH 48, 6530-6534, November IS. 1988)

Neuronal Cell Differentiation of Human Neuroblastoma Cells by Retinoic Acidplus Herbimycin A1

Peter N. Preis, Hideyuki Saya, LászlóNádasdi,2Guenther Hochhaus, Victor Levin, and Wolfgang Sadée3

Brain Tumor Research Center [P. N. P.. H. S., L. N., V. L.J and Department of Pharmacy [G. H.. W. S.]. University of California. San Francisco, California 94143

ABSTRACT

We investigated the effect of retinoic acid (RA) and herbimycin A(herb-A) on cell growth, cell differentiation, and colony formation ofhuman neuroblastoma cell lines. The neuroblastoma line SK-N-SH expressed both neuroblast and nonneuronal phenotypes, whereas its sub-clone SH-SY5Y and the Kelly cell line were predominantly neuroblastic.Both herb-A and RA, given alone, moderately reduced cell growth andcolony formation of the neuroblastic cell lines. Growth curve analyseswith SK-N-SH suggested that herb-A greatly reduced the number ofinitially growing cells, whereas RA slightly enhanced initial cell growth.Morphological changes were determined with the use of rhodamine-phalloidin staining of actin. Retinoic acid caused an increase in thefraction of neuroblast cell in SK-N-SH, and conversely of nonneuronalcells in SH-SY5Y and Kelly cell lines. Both drugs also caused partialdifferentiation towards a neuronal phenotype, and herb-A induced selective lysis of nonneuronal cells of SK-N-SH. Because of their discrepanteffects, RA (10 MM)and herb-A (236 IIM)were tested in combination ata concentration that had only moderate effects when given alone. Thecombination further reduced cell growth and colony formation and dramatically enhanced differentiation towards a neuronal morphology. TheKelly cell line with amplified Vmiv genome, which correlates withclinical progression of neuroblastoma, was also sensitive to RA plus her b-A. These results recommend the combination of RA and herb-A fordifferentiation therapy of neuroblastoma.

INTRODUCTION

Neuroblastoma are tumors of early childhood that respondvariably to chemotherapeutic drugs. Because neuroblastomasoccasionally mature spontaneously to benign ganglioneuromas,agents that induce neuronal differentiation may be effective inthe treatment of neuroblastoma. Human neuroblastomas commonly express several phenotypes in vitro that may be interconvertible (1-3). The neuroblast phenotype differentiates intoneuronal cells, while the nonneuronal cell type is thought torepresent Schwann or melanocytic cells (4).

The goals of differentiation therapy may include the simultaneous differentiation of all phenotypes or the selection ofonly one mature cell type, e.g., neuronal. Several substanceshave been reported to induce differentiation in neuroblastomacell lines. Among these are RA4 and TPA (5, 6). Treatmentwith these drugs in vitro resulted in the formation of a spread-out, polar phenotype with neuritic processes and a reduction ofcell growth. Using either RA or TPA, growth inhibition wasseen. When both drugs were combined, cytostasis in additionto cell differentiation was observed. Since phorbol esters suchas TPA act as tumor promoters, the cytostatic effects of thecombination of RA plus TPA may have little therapeutic poten-

Received 7/9/87; revised 4/29/88, 8/8/88; accepted 8/18/88.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

' Supported in part by Grants CA 37655 and CA 27866 from the National

Cancer Institute, DA 04166 from the National Institute on Drug Abuse, thePreuss Foundation. Inc., and a Max Kade fellowship (P. N. P.).

J Permanent address: Eotros University, Department of Organic Chemistry,

Budapest. Hungary-' To whom requests for reprints should be addressed, at Department of

Pharmacy. School of Pharmacy. UCSF, San Francisco, CA 94143.*The abbreviations used are: RA, retinoic acid; herb-A, herbimycin A; PBS,

phosphate buffered saline; TPA, 12-O-tetradecanoylphorbol-13-acetate; DMSO,dimethyl sulfoxide; CFE, colony forming efficiency.

tial. Retinoic acid was shown to decrease the expression of theN-myc protooncogene, which is prevalent in neuroblastomas,

before induction of neuronal differentiation (7). While RA givenalone is currently under consideration for neuroblastoma therapy, it fails to drive the cells to complete the differentiationprocess and to abolish the ability to form colonies. It is, therefore, desirable to select additional agents that, either alone orin combination, bypass the differentiation defect of neuroblastoma cells. Several benzenoid antibiotics are known to haveantitumor activity (8, 9). Among these, herb-A has been reported to induce differentiation in Rous sarcoma virus transformed cells and to inhibit the protein tyrosine kinase activityof pp60v"src,the oncogene protein of the v-src gene (10). We

have observed that this drug also induces growth inhibition inseveral cell lines that do not express pp60vsrc. In the present

study, we selected the neuroblastoma SK-N-SH (1) which expresses both neuroblast and nonneuronal cell types (1-4), andits neuroblast subclone, SH-SY5Y, to study drug effects on thevarious cell types of neuroblastomas. SK-N-SH does not carryamplified N-myc DNA sequences, but it expresses N-myc andhigh levels of c-myc (11, 12). In contrast, the neuroblastomacell line, Kelly, carries highly amplified N-myc sequences (12)which may contribute to malignant progression of neuroblastomas. We show here that a combination of RA and herb-Acauses extensive differentiation of the neuroblastoma cells towards a neuronal phenotype.

MATERIALS AND METHODS

Cells and Materials. The human neuroblastoma cell line SK-N-SHand its subclone SH-5Y5Y were obtained from June Biedler (SloanKettering Memorial Institute, New York, NY), and the Kelly cell linewas from Fred Gilbert (Mount Sinai, School of Medicine, New York,NY). RA was purchased from Sigma (St. Louis, MO) and herb-A wasobtained from Mary K. Wolpert (National Cancer Institute, Bethesda,MD). For drug studies, herb-A was dissolved in DMSO (125 ¿tgherb-A/ml = 236 JIM)and RA in ethanol (3 mM RA). Rhodamine-phalloidinwas obtained from Molecular Probes, Inc. (Junction City, OR).

Growth Curves. Cells were grown in RPM1 1640 supplemented with10% fetal bovine serum, 100 ¿igstreptomycin/ml, and 100 IU penicillin/ml. For growth curves, cells (passages 30-45 for SK-N-SH and SH-SY5Y) were seeded at a cell density of 3 x 105/flask (25-cm2 flasks)

and drugs were added 24 h after seeding. Medium with and withoutherb-A and/or RA, or DMSO alone, was changed daily. Cells weretrypsinized (0.05% trypsin, 0.02% EDTA in Puck's saline, 1 g/Iiterglucose, and 0.58 g/liter NaHCOj) for 75 min at 20°Cand counted in

a Fuchs Rosenthal chamber 1-8 days after drugs were added. All cellswere maintained at 37°Cin a humidified 5% COi/air atmosphere. Up

to 0.5% DMSO and 0.3% ethanol (final concentration) in the growthmedium caused no detectable growth inhibition or differentiation inthe neuroblastoma cell lines; the highest medium concentration ofDMSO as a solvent for herb-A was 0.4%.

Analysis of Growth Curves. Cell growth was modelled by the Ver-hulst-Pearl ("logistic") equation (13, 14) in the form

C = C, + Cii + (Ci/Co - /)*<?-

where C„,is nongrowing cells, Co is growing cells at 0 time, C¡isnumber of cells at saturation, and k is growth rate. This functionprovided a quantitative description for the cell lines under investigation.

6530

on July 24, 2021. © 1988 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

NEURONAL CELL DIFFERENTIATION BY RA -t- HERB-A

Other growth models, like the Gompertz equation, Monod model, orthe piecewise-linear growth model, did not provide any improvementover the logistic equation.

An interactive computer program was written in BASIC for a Hewlett-Packard 9000 Series 300 computer to carry out a weighted nonlinear least square fitting. Approximate starting parameter estimates wereobtained with a SIMPLEX algorithm, and the final values were calculated with the Newton-Raphson iteration. Statistical analyses of the fitwere performed by (a) calculating approximate 95% confidence limitsof the parameters (with a nested iteration procedure the contours werecomputed in the multidimensional parameter space according to themethod of Draper and Smith); (b) analyzing the sign of differencesbetween the observed and computed cell number with the help of the"runs" test; and (c) graphical analysis.

The pairwise comparison of the individual growth curves were performed by fitting a "common" curve over the combined data set, and

the parameter values obtained this way were tested against the originalvalues with the F test. Accordingly, the sums of the squared residualswere computed (with the original and with the "common" values,

respectively), and, for the level of significance of the two curves beingnot different, first the F value was computed as:

55,n - p

where n is the number of data points, p is the number of fittedparameters, and the subscripts o and c refer to the sum of squaredresiduals with the original (optimal) and "common" parameter values,

respectively. Then, the corresponding statistical probabilities were calculated with degrees of freedom [p] and [n - p].

Colony Forming Efficiency Assay. Cells were preincubated with andwithout different concentrations of herb-A and/or RA or DMSO for72 and 144 h. Aliquots of 3 to 6 cell dilutions were then plated into35-mm plastic Petri dishes containing 2.5 ml medium (RPMI 1640supplemented with 10% fetal calf serum with and without herb-A) andkept for 10 days at 37°Cin a humidified 5% COi/air atmosphere. The

cells were fixed with methanol and stained with 2% Giemsa. Thenumber of colonies visible to the naked eye (~7 doubling times) wascounted, and the colony forming efficiency was determined as thepercentage of cells plated that formed colonies (15).

Actin Staining. Rhodamine-phalloidin staining was performed asdescribed previously (16). Cells were grown on sterile glass coverslips,rinsed twice in PBS, and dehydrated in absolute acetone at -20°C for

4 min. Then the fixed and air dried cells were stained for 20 min with10 n\ rhodamine-phalloidin (1 /¿g/ml),washed 3 times in PBS, andmounted for viewing. An IMT 2 Olympus fluorescence microscopeequipped with a xlOO fluorescence objective provided phase contrastand fluorescence images of cells. Cells were considered to be neuroblastwhen they displayed short neuritic processes. Neuroblast cells were alsoconsiderably smaller than the nonneuronal cells; the latter were recognized by their flat shape and a fibroblast-like actin staining pattern witha network of organized cables (11). The neuroblast cells were consideredto be differentiated towards a neuronal phenotype when they grewneuritic processes longer than 50 urn. Morphological differentiationand percentage of cellular subtypes were measured by counting 100cells on each of 5 different spots of the cultures.

Receptor Assay. SK-N-SH (SH) and SH-SY5Y (5Y) were culturedfor 7 days in the presence of herb-A and/or RA added every 24 htogether with the medium change. Cells were harvested with 0.04%EDTA in Ca2+-Mg2+free phosphate buffered saline with mechanical

agitation. Cells were subsequently washed three times with PBS, andcell homogenates prepared in 50 mM Tris-HCl buffer, pH 7.4 (18),were used without further purification. In order to detect M opioidreceptor sites, [3H][o-Ala2, MePhe4, Gly-ol5]enkephalin (1 nM finalconcentration) was incubated with the cell homogenate at 20°Cfor 1

h, and the bound radioactivity was determined by filtering the homogenate mixture through Whatman GF/B glass fiber filters (17). Tracerbinding in the presence of 10~6M diprenorphine served as the value for

nonspecific binding and was less than 15% of total binding. Proteinwas determined by the method of Lowry et al. (18).

RESULTS

Initial screening experiments with the neuroblastoma cellsindicated that a concentration of 10 UM RA was effective instimulating neuronal differentiation, as measured by the development of neuritic processes, without causing extensive inhibition of cell growth, while some differentiation effects werealready discernible at 0.1 ¿IM.For the present study we thereforeselected a concentration of 10 fiM RA throughout the experiments.

Morphology

Phenotypic characteristics of the neuroblastoma cell lineswere evaluated by microscopic inspection of overall morphologyand actin staining patterns obtained with rhodamine-phalloidin(Fig. 1).

SK-N-SH Neuroblastoma. These cells did not exhibit a uniform morphological and staining pattern. In untreated cells onecan distinguish at least two different cell types and/or growingpatterns (Table 1). It should be noted that the relative abundance of these cell types varies greatly with passage number ofthe selected cells, cell density, and incubation conditions. Underthe conditions used here, most cells had a nonneuronal appearance with intercellular contacts and large actin cables. Thesecond type of cells was similar to the neuroblast phenotype ofthe SH-SY5Y cells and had neurite-like cytoplasmic processes.Growing of SK-N-SH cells to higher cell densities changed theproportions of the two cell types. The neuroblast cells increasedfrom 14% at 24 h to 22% at 96 h of incubation. Treatment withRA resulted in partial differentiation towards the neuronalphenotype, and the proportion of neuroblast-neuronal cells wasincreased from 22% to 49% (Table 1). Herb-A treatment aloneinduced some differentiation of the neuroblast-like cells; however, the differentiation process required a longer time periodthan differentiation induced by RA. The nonneuronal cells werehighly sensitive to herb-A; they flattened, became giant cellsafter 4 days (see Fig. I//), and died after that.

The combination of both drugs had dramatic effects on SK-N-SH morphology. Differentiation towards a neuronal cell typewas apparent in 45% of the cells by 24 h. After 24 h no furthercell proliferation could be observed, the nonneuronal cells wereaffected in the same way as with herb-A alone; they grew largerand eventually disappeared from the cell cultures (Fig. IF),

SH-SY5Y Neuroblastoma Line. These cells also consisted oftwo phenotypes; however, the majority of cells were neuroblast-like, with short cytoplasmic processes, and formed densemounding aggregates, while a small percentage of cells exhibited a flat, nonneuronal phenotype (Fig. 1; Table 1). Typicallya few actin cables were seen stretching across the neuroblast-like cell clusters, but more rhodamine-phalloidin fluorescenceactivity is found on the inner surface of the cytoplasmic membrane (11). The cytoplasmic processes were accentuated by thisstaining procedure. The nonneuronal cells showed a well organized stress fiber structure. When cells were treated with RA,25% of cells grew straight cell processes longer than 50 /¿m(Table 1). In addition, RA treatment longer than 48 h resultedin a dramatic increase in the percentage of the nonneuronalphenotype and, indeed, RA treated cultures of SK-N-SH andSH-SY5Y were similar (Fig. 1, C and G). Herb-A alone did notinduce obvious cell differentiation in this cell line; however, itinduced a dose dependent selection of the neuroblast cells byeliminating the nonneuronal cells after 72 h in a manner similarto that observed for the SK-N-SH line. When herb-A and RAwere combined, differentiation of the neuroblast cells was observed in 71% of the cells after 24 h with only a small cell

6531

on July 24, 2021. © 1988 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

NEURONAL CELL DIFFERENTIATION BY RA + HERB A

IffUKA ü,%7»•¿�>Jr '

HFig. 1. Actin staining of SH-SY5Y (A-D) and SK-N-SH (E-F). Control (/4, E), 236 nM herb-A + 10 ^M RA (A, f). 10 ^M) RA (C C), and 236 nM herb-A (D.

H). Duration of drug treatment. 96 h. A. B, C, E. F, G x 100; (D, H), x 400.

Table I Percentage of cellular subtypes as determined with rhodamine-phalloidinstaining

Cells were preincubated for 24 h and then for another 96 h in the presence orabsence of the indicated drugs (herb-A. 236 nM; RA, 10 nM). The results representthe average of two independent experiments with below 5% variability. Eachexperiment involved 5 observations of 100 cells each.

Nonneuronal24

hpreincubationSK-N-SHSH-SY5YKellyAfter

96 hincubationSK-N-SHControlHerb-ARAHerb-A

-HRASH-SY5YControlHerb-ARAHerb-A

+RAKellyControlHerb-ARAHerb-A

+ RA8674782551109758764295Neuroblast149396223828159188171294914635Differentiated0000372175052581052560

fraction continuing to proliferate for another 3 days. The non-neuronal cells were almost entirely eliminated after 96 h (Fig.IB).

Opioid Receptors in SK-N-SH and SH-SY5Y. MOpioid receptors are expressed in SK-N-SH and SH-SY5Y cells, but not inthe nonneuronal subclone SH-EP (11, 17). Therefore, expression of the n receptors could serve as a neuronal marker in thiscase. In Table 2, n opioid receptor binding is shown for SK-N-SH and SH-SY5Y cells after treatment with herb-A, RA, anda combination of both agents. RA ( l OßM)and herb-A treatment(236 nM) alone increased ßopioid receptor binding in SH-SY5Y cells by 2-3- and 1.5-fold, respectively. When a combination of both agents was used, a 4-fold increase of ¿/receptorsites per mg of protein was observed. Similar results were found

Table 2 Effects of drug treatment on n opioid receptor expressionEquilibrium binding studies were carried out with crude membrane prepara

tions at 20°Cand 1 nM [3H]DAGO. Nonspecific binding was determined in thepresence of 10~*M diprenorphine and was subtracted from the results (—10%of

specific binding). All experiments were done in parallel with treated and untreatedcells. Data shown are mean ±SD (n = 3 to 5).

Treatmentdpm tracer binding/mg

protein

SH-SY5YControlRA(lOfiM)Herb-A (236 nM)RA (10 MM)+ herb-A (236 nM)

SK-N-SHControlRA(lOjiM)Herb-A (236 nM)RA (10 MM)+ herb-A (236 nM)

1826 ±845290 ±675°2965 ±96°7335 ±658°

2338 ±2332047 ±2864128 + 437°4376 ±440°

°P < 0.01, different from control.

with SK-N-SH, although RA was less effective or ineffective inenhancing n opioid receptor binding.

Kelly Neuroblastoma Line. The Kelly cells reacted to herb-Aand RA in a fashion similar to that of SH-SY5Y (Table 1).Each agent alone induced some differentiation, whereas thecombination of the two induced substantial differentiation towards the neuronal phenotype. Again, RA caused a large increase in the fraction of nonneuronal cells over a 4-day incubation period.

Growth Curves and Colony Forming Efficiency

Neuroblastoma Lines. Growth curves in the presence of herb-A (236 ¿IM),RA (10 AIM)*or both are depicted in Fig. 2, and ananalysis of the SK-N-SH results is given in Table 3. In eachcase, the drugs reduced cell densities, with the drug combinationbeing most effective. RA plus herb-A completely prevented netcell growth of SK-N-SH over 8 days, while SH-SY5Y and Kellycell number continued to increase over time. Higher concentrations of herb-A caused further reduction in cell growth, andconcentrations above 330 nM were cytotoxic to SK-N-SH andSH-SY5Y.

6532

on July 24, 2021. © 1988 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

NEURONAL CELL DIFFERENTIATION BY RA + HERB-A

C.II. .!*>

C.II. .le5

•¿�—Hfl * (W

C.II. .11»

Fig. 2. Growth curves of the neuroblastoma lines of SK-N-SH, SH-SY5Y,and Kelly. Cells were preincubated for 24 h before drug addition. O, control; O,10 ¿iMRA; A, 236 nM herb-A; »,10 mM RA + 236 nM herb-A. , computerPits to the logistic equation for growth. For SK-N-SH the initial fraction ofnongrowing cells (C,,) was >0 (see Table 3), while Cw was estimated to be zerofor SH-SY5Y and Kelly. Bars, SD.

Analysis of the growth curves with the logistic equationrevealed that the SK-N-SH curves were best fit if one assumesthe presence of a large initial fraction of nongrowing cells (Table3), whereas in the SH-SY5Y and Kelly lines, all cells wereassumed to grow at the time of drug addition. The fraction ofnongrowing cells (Cng= 71 %; Table 3) is similar to the fractionof nonneuronal cells in SK-N-SH (Cng = 78%; Table 1), suggesting that these cells contribute little to the overall growth ofthe cell culture. Treatment of SK-N-SH with RA caused anincrease in the calculated percentage of growing cells (Co) from29% to 51%, in parallel with the increase of the neuroblast +neuronal cell fraction (Table 1). In contrast, increasing concentration of herb-A dramatically reduced the initial number ofgrowing SK-N-SH cells (C0), while the drug combination fellin between (Table 3). All curves from the treatment scheduleswere significantly different from each other (Table 3).

Drug treatment also reduced CFE of the neuroblastoma celllines (Table 4). Longer incubations were more effective, and

Table 3 Analysis of growth curves of SK-N-SH treated with herb-Aand RA (10 iiMf

Estimates of nongrowing cells ((„,).growing cells at 0 time (C0). cell numberat saturation (<",)and the growth rate (A)were obtained by fitting the experimental

data to the Verhulst-Pearl equation (Experimental)

ControlRAHerb-A236

nM238ntn330

nMRA+ herb-A, 236 nMc.2.25

x1.32x3.08

x3.08x3.12x2.32

x10'10*IO5IO510»IO5C„0.90

x1.73x0.056

x0.026x0.0006

x0.69xIO5IO510»IO5IO5IO5C,43.5

x27.8x14.5

x9.25x6.05x1.34X10'IO5IO5IO5IO510'k0.78

day0.68day1.02

day1.03day1.62day0.53

day-~-~-~

Pairwise comparison of SK-NSH cell growth curves with ftest

F values and probabilities (with degrees of freedom: d.f. 1 = 4, d.f. 2 = 12) ofthe growth curves being not different. Similar results were obtained with SH-SY5Y and Kelly cells, except for the comparison of SH-SY5Y growth curves inthe presence of herb-A (236 nM) and retinoic acid (10 /IM) (F = 0.4940; P = 0.69)

ControlFPherb-A

= 236finFPRA=IO(iMFP236

nM 283 nM 330 nMIO>iM37.913

50.231 57.7195.8150.000001

0.0000002 0.00000010.0077116.7850.000074236

nM +RA67.0350.0000000436.7990.000001237.1390.000001

1" Cells were preincubated for 24 h before drug addition, and growth curves

were fitted to the logistic equation (C„= nongrowing cells; Co = growing cells atzero time; C, = number of cells at saturation: A = growth rate).

Table 4 Colony forming efficiency assay of SK-N-SH, SH-SY5Y, and Kelly cells

A. Cells were incubated with drugs for 72 and 144 h, but not during colonyformation. B. Cells were incubated with drugs for 72 and 144 h and drug treatmentwas continued during colony formation (±SD,n = 4). The plating efficiencieswere 37% for SK-N-SH, 43% for SH-SY5Y, and 48% for Kelly.

%CFE

72 h treatment

A B

144 h treatment

A B

SKN-SHControlRA236

nMherb-A330nMherb-A236nM herb-A -1-RASH-SY5YControlRA236

nMherb-A330nMherb-A236nM herb-A +RAKellyControlRA236

nMherb-A330nMherb-A236nM-A + RA100±354

±463±361±740

±2100±

294±677±354

±172±6100

±593±487

±265±337

±5100

±310±221±319±40100

±226±46±

10.86±10.57±1100

+525±513±14±20100±

326±154±153

±3I2±1100±

274±351±726±428

±7100±568

±675±545±428

±4100±33±

115±37±20100

±217±12±20.64

±10100±

518±33±20.56

±10

the drug combination reduced CFE more than either drug alone.The order of drug sensitivity was SK-N-SH > SH-SY5Y >Kelly. However, in order to achieve complete suppression ofCFE, the drug combination needed to be present continuallyduring colony formation.

DISCUSSION

Herbimycin A and RA, given alone or in combination, substantially affected the growth and morphology of the neuroblastoma lines SK-N-SH, SH-SY5Y, and Kelly. The morphology and phenotype were evaluated with the aid of fluorescencemicroscopy and phalloidin-actin staining. The correlations between cell morphology-actin staining and neuroblast or non-

6533

on July 24, 2021. © 1988 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

NEURONAL CELL DIFFERENTIATION BY RA + HERB-A

neuronal phenotype have been described earlier (11). The results suggest that RA caused differentiation towards a neuronalphenotype with long neuritic processes in a portion of the cells.Further, morphological analysis (Table 1) and detailed growthcurve analyses (Table 3; Fig. 2) suggest that retinoic acidinduced transdifferentiation in both directions, from a neuro-blast to a nonneuronal morphology, and from a nonneuronalto a neuroblast phenotype. Thus, the fraction of nonneuronalcells in the neuroblast lines SH-SY5Y and Kelly was greatlyenhanced, much beyond what one would expect only fromgrowth of the initial nonneuronal cell population. Conversely,RA rapidly increased the fraction of neuroblast and neuronalcells in the mostly nonneuronal cultures of SK-N-SH. Thegrowth curve analysis (Table 3) suggested that the initial non-neuronal cell population is identical with the calculated non-growing fraction of cells (C„g).This result supports previoussuggestions that the neuroblast cells represent a more highlytransformed phenotype than the less tumorigenic nonneuronalcells (11).

RA treatment doubled the calculated fraction of growing cells(Co), thereby initially stimulating the overall growth rate of theSK-N-SH culture, until a lower growth rate and plateau werereached. This previously undetected dual action of RA is ofconsiderable interest, because RA is currently considered forclinical use against neuroblastoma. Spontaneous transdifferentiation among the various phenotypes of SK-N-SH has beenpreviously documented by Rettig et al. (19). Furthermore, Tsun-amoto et al. (20) have also reported on a bidirectional action ofbromodeoxyuridine in neuroblastoma differentiation, with induction of either a glial/schwannian cell type or a neuronal celltype. Additional studies with differentiation markers will berequired to clarify the action of RA in neuroblastoma differentiation.

Herbimycin A effects were distinct from those of RA. In thecase of SK-N-SH, where the fraction of initially growing cells(Co) was small as estimated with the logistic function, herb-A

greatly reduced Co (whereas RA increased C0) (Table 3). Differentiation was less pronounced, but herb-A also caused enlargement of the nonneuronal cells, which eventually would lyse anddisappear from the medium (Fig. 1).

The contrasting effect of RA and herb-A prompted an investigation of their combined effect. The concentrations of herb-A(236 nM) and RA (10 pM) were selected to have only smalleffects on growth and differentiation, given alone. When combined, RA plus herb-A caused rapid and extensive differentiation of the neuroblastoma lines toward a neuronal phenotype.In parallel, combination treatment enhanced n receptor contentas a neuronal marker in SK-N-SH (11) more than either drugalone (Table 2). Moreover, no net growth was observed withSK-N-SH, the most sensitive neuroblastoma cell line tested(Fig. 2). Colony forming efficiency was also strongly suppressed(Table 4), with longer drug exposure affording greater inhibition. However, complete suppression of colony formation wasachieved only when the drugs were also present during thecolony forming incubation. It is difficult to fully dissect theresults of the drug combination into effects resulting fromcytotoxicity and differentiation. While drug exposure is clearlytoxic to the nonneuronal cells, it appears that differentiation

plays the major role in the behavior of the neuroblast cells atthe drug concentrations selected here. The Kelly cell line wasselected for its high N-myc gene amplification which correlateswith malignant progression in neuroblastoma (21). The factthat it is also responsive, albeit somewhat less sensitive thanthe SK-N-SH lines, emphasizes the potential of the RA-herb-A combination as a strategy for clinical differentiation therapy.

REFERENCES

1. Biedler, J. L., Helson, L., and Spengler, B. Morphology and growth, tumor-igenicity, and cytogenetics of human neuroblastoma cells in continuousculture. Cancer Res., 33: 2643-2652, 1973.

2. Biedler, J. L., Rozen, M. G., Osama, E. B., Meyers, M. B., Melera, P. W.,Ross, R. A., and Spengler, B. A. Growth stage-related synthesis and secretionof proteins by human neuroblastoma cells and their variants. In: A. E. Evans,G. J. D'Angio, and R. C. Seeger (eds.), Advances in Neuroblastoma Research,pp. 209-221. New York: Alan R. Liss, Inc., 1985.

3. Ross, R., Spengler, B., and Biedler, J. L. Coordinate morphological andbiochemical interconversion of human neuroblastoma cells. J. Nati. CancerInst., 71: 741-747, 1983.

4. Tsokos, M., Ross, R. A., and Triche, T. J. Neuronal, Schwannian andmelanocytic differentiation of human neuroblastoma cells in vitro. In: A. E.Evans, G. J. D'Angio, and R. C. Seeger (eds.). Advances in Neuroblastoma

Research. New York: Alan R. Liss, Inc., 1985.5. Pahlman, S., Ruusala, A. I., Abrahamson, L., Mattson, M.!•'..K...and Esscher,

T. Retinoic acid induced differentiation of cultured human neuroblastomacells: a comparison with phorbol ester induced differentiation. Cell Differ.,14:135-144, 1984.

6. Spinelli, W., Sonnenfeld, K. H., and Ishii, D. N. Effects of phorbol estertumor promoters and nerve growth factor on neurite outgrowth in culturedhuman neuroblastoma cells. Cancer Res., 42: 5067-5073, 1982.

7. Thiele, C. J., Reynolds, C. P., and Israel, M. A. Decreased expression of N-HIITprecedes retinoic acid-induced morphological differentiation of humanneuroblastoma. Nature (Lond.), 313:404-406, 1985.

8. Iwai, Y., Nakagawa, A., Sadakane, N., Omura, S., Oiwa, H., Matsumoto, S.,Takahashi, M., Ikai, T., and Ochiai, Y. Herbimycin-A, a new benzenoidansamycin with anti-TMV and herbicida! activities. J. Antibiot. (Tokyo), 33:1114-1119, 1980.

9. Muroi, M., Izawa, M., Kosai, Y., and Asai, M. Macbecins I and II, newantitumor antibiotics. II. Isolation and characterization. J. Antibiot. (Tokyo),33:205-212, 1980.

10. Uehara, Y., Hori, M., Takeuchi, T., and Umezawa, H. Screening of agentswhich convert "transformed morphology" of Rous sarcoma virus-infected ratkidney cells to "normal morphology": identification of an active agent as

herbimycin and its inhibition of intracellular sir kinase. Jpn. J. Cancer Res.,76:672-675, 1985.

11. Sadee, W., Yu, V. C., Richards, M. L., Schwab, M. R., Preis, P. N., Brodsky,F. C., and Biedler, J. L. Expression of neurotransmitter receptors and mycproto-oncogenes in subclones of a human neuroblastoma cell line. CancerRes., 47: 5207-5212, 1987.

12. Schwab, M., Alitalo, K., KJempnauer, K.-H., Varmus, H. E., Bishop, J. M.,Gilbert, F., Brodeur, G., Goldstein, M., and Trent, J. Amplified DNA withlimited homology to myc cellular oncogene is shared by human neuroblastoma cell lines and a neuroblastoma tumor. Nature (Lond.), 305: 245-248, 1983.

13. Draper, N. R., and Smith, H. Applied Regression Analysis, pp. 458-516.New York: John Wiley & Sons, 1981.

14. Kolker, Y. A piecewise-linear growth model: Comparison with competingforms in batch culture. J. Math. Biol., 25: 543-551, 1987.

15. Rosenblum, M. L., Vasquez, D. A., Hoshino, T., and Wilson, C. B. Development of a clonogenic cell assay for human brain tumors. Cancer (Phila.),47:2305-2314, 1978.

16. Wulf, E., Deboben, A., Bautz, F. A., Faulstich, H., and Wieland, T. Fluorescent phallotoxin, a tool for the visualization of cellular actin. Proc. Nati.Acad. Sci. USA, 76: 4498-4502, 1979.

17. Yu, V. C., Richards, M. L., and Sadeé,W. A human neuroblastoma cell lineexpresses n and d opioid receptor sites. J. Biol. Chem., 261: 1065-1070,1986.

18. Lowry, O. H., Rosebrough, N. J., Fair, A. L., and Randall, R. J. Proteinmeasurements with the Folin phenol reagent. J. Biol. Chem., 193: 265-275,1951.

19. Rettig, W. J., Spengler, B. A., Chesa, P. G., Old, L. J., and Biedler, J. L.Coordinate changes in neuronal phenotype and surface antigen expressionin human neuroblastoma cell variants. Cancer Res., 47: 1383-1389, 1987.

20. Tsumamoto, K., Todo, S., Imashuku, S., and Kato, K. Induction of SI00protein by 5-bromo-2'-deoxyuridine in human neuroblastoma cell lines.Cancer Res., 48: 170-174, 1988.

21. Schwab, M. Amplification of N-myc in human neuroblastoma. Trends Genet., /: 271-275, 1985.

6534

on July 24, 2021. © 1988 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1988;48:6530-6534. Cancer Res   Peter N. Preis, Hideyuki Saya, László Nádasdi, et al.   Retinoic Acid plus Herbimycin ANeuronal Cell Differentiation of Human Neuroblastoma Cells by

  Updated version

  http://cancerres.aacrjournals.org/content/48/22/6530

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/48/22/6530To request permission to re-use all or part of this article, use this link

on July 24, 2021. © 1988 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from