Culture System - Europe PMC

AmericanJournal of Pathology, Vol. 134, No. 1, January 1989 Copyright © American Association ofPathologists

Neuroblastic Differentiation Potential of the Human Retinoblastoma Cell Lines Y-79 and WERI-Rbl Maintained in an Organ Culture System An Immunohistochemical, Electron Microscopic, and Biochemical Study

Mary M. Herman,* Elias Perentes,* Christos D. Katsetos,* Francoise Darcel,* Anthony Frankfurter,t V. Peter Collins, Larry A. Donoso,§ Lawrence F. Eng," Paul J. Marangos,¶ Allan F. Wiechmann,** Estelle E. May,* Christine B. Thomas,* and Lucien J. Rubinstein* From the Departments ofPathology (Neuropathology*)

and Biology,t University of Virginia School ofMedicine, Charlottesville, Virginia, the LudwigInstitutefor Cancer Research, Stockholm, Sweden;* the Wills Eye Hospital, Philadelphia, Pennsylvania,§ the Laboratory Service, VeteransAdministration Medical Center, Palo Alto,11 Gensia Pharmaceutical, San Diego, California, and the Bowman Gray School ofMedicine, Winston Salem, North Carolina* *

The differentiation potential of the human retinoblastoma cell lines Y- 79 and WERI-Rbl was evaluated in vitro for up to 120 days in a matrix system and in rotary suspension for 30 days. Matrix cultures were grown with 10% fetal calf serum (FCS), with and without differentiation-promoting agents. The latter were appliedfor a total of 5-45 days (usually 30 days) and included 7S nerve growth factor, dibutyryl cyclic AMP, sodium butyrate, retinoic acid, hydrocortisone, and ascorbic acid. Fully defined, serum-free medium and medium containing 5 or 15% FCS were also used for matrix cultures, and medium with 5 or 10% FCS for suspension cultures. By immunoperoxidase (performed on matrix cultures, both untreated and treated for 30 days with differentiation-promoting agents), the cells ofboth lines were positive for neuron-specific enolase (NSE), microtubule-as-

sociatedprotein 2 (MAP2), class III4-tubulin (human hb4) isotype, and synaptophysin. In addition, the WERI-Rb1 cells expressed 200 kd neurofilament protein (NFP-H) and retinal S-antigen. Both lines were invariably negative for glial fibrillary acidic (GFA) protein, myelin-associated glycoprotein, myelin basic protein, the epitope recognized by the Leu- 7 monoclonal antibody, opsin, and hydroxyindole-0-methyltransferase. In the Y-79 line the presence of NSE and the absence ofNFproteins-H, M and -L, of GFA protein, and of retinal S-antigen were confirmed biochemically. No differentiated features were found by electron microscopy in either line. Thus, in the matrix system employed, both lines exhibited solely a potential for neuroblastic differentiation, which was more advanced in the WERI-Rbl line, as reflected by the antigenic expression of NFP-H and of retinal S-antigen. (Am JPathol 1989, 134:115-132)

The cytogenesis of retinoblastoma, the most common intraocular malignancy of childhood, is controversial. The Supported by Research Grants CA 31271 from the National Cancer Institute and EY 5095 from the National Eye Institute, and by Graduate Neuropathology Training Grant T32 NS 7236 from the National Institute of Neurological and Communicative Diseases and Stroke, U.S. Department of Health and Human Services. EP was supported by a grant from the Fondation Suisse de Bourses de M6decine et Biologie. FD was supported by a grant from the Fondation pour la Recherche M6dicale (France). EEM and CT were partly supported by Clinical Fellowships of the American Cancer Society (85-43 and 6064, respectively). Accepted for publication September 8, 1988. Address reprint requests to Mary M. Herman, MD, Department of Pathology (Neuropathology), University of Virginia School of Medicine, Charlottesville, VA 22908.

115

116

Herman et al

AJPJanuary 1989, Vol. 134, No. I

majority of electron microscopic, tissue culture and, more recently, immunohistochemical studies performed on the tumors obtained at operation or on primary culture explants favor an origin from embryonal cells capable solely of neuroblastic or, more precisely, photosensory differentiation.1-14 The presence of glial elements has long been recognized to be a common feature traceable to a reactive gliosis originating in the disorganized retina,15 an interpretation supported in some of the immunohistochemical investigations.9'13'16 A minority of electron microscopic, in vitro, and immunohistochemical studies have interpreted the presence of glial cells as evidence of astrocytic tumor differentiation, however,17-21 thus suggesting that the cell of origin of retinoblastoma is a primitive neuroepithelial stem cell capable of bipotential neuronal and glial differentiation. The basis of that evidence is further examined in the discussion of this paper. A second approach to the problem of cytogenesis and differentiation potential is the investigation of cell lines derived from human retinoblastomas. Tools for the experimental study of the tumor's differentiating capacity became available with the development of two such lines, Y7922 and WERI-Rbl .23 The cells of both lines in monolayer culture have been reported by Jiang and co-workers24 to show immunopositivity for both glial fibrillary acidic (GFA) protein, an astrocytic cell marker, and for dopamine-f3-hydroxylase and tetanus toxin, both of which are neuronal markers. The Y-79 line, again in monolayer, has been stated by Kyritsis et a125 to be composed of undifferentiated cells, all of which contain both GFA protein and neuron-specific enolase (NSE), but exhibiting only one of these markers when cultured in serum-free medium and after the addition of the differentiating agent dibutyryl cyclic AMP. With either dibutyryl cyclic AMP or sodium butyrate, the cultured cells also have been reported by the same group as forming processes claimed to be neuritelike, 26,27 as expressing myelin basic protein (MBP) and catecholamines or catecholamine precursor substances,28 and as demonstrating electron microscopic features indicative of early photoreceptor cell differentiation.27'28 The development of cells containing melanosomes and therefore some of the characteristics of pigment-containing epithelial cells have also been reported by the same workers.27 Collectively, these studies have led their authors to the conclusion that the Y-79 retinoblastoma line is a primitive neuroectodermal cell of multipotential character. The present paper describes the results obtained on culturing the Y-79 and WERI-Rbl retinoblastoma lines in an in vitro matrix system that has been shown to favor both morphologic and biochemical differentiation in a variety of human and experimental gliomas.29-33 With primary explants of a human cerebellar medulloblastoma, this sys-

tem has resulted in divergent glial and neuroblastic differ-

entiation.34 The effects of different media and of supplementing the latter with agents that have been reported to favor differentiation in neuroblastoma cell lines35-6 and in RAO 188 cells (a retinoblastomalike cell line derived from a murine intraocular tumor produced by the intravitreous inoculation of the human adenovirus type 12)37 also were investigated. The results indicate that in the matrix system the retinoblastoma lines display solely a potential towards neuroblastic differentiation, which was more advanced in the WERI-Rbl line, as shown by the antigenic expression of the high molecular weight subunit of neurofilament protein (NFP-H) and of the retinal S-antigen.

Methods Tissue Culture Cell Lines and Overview of Experiments The Y-79 human retinoblastoma cell line was obtained from Dr. Timothy J. Triche of the National Cancer Institute and was carried in our laboratory at the University of Virginia for 2 years. Three experiments using these cells in matrix cultures were performed for morphology over the first 18-month period, and cultures collected for biochemical analysis over the next 6 months. A total of 200 dishes were evaluated by morphology and peroxidase immunohistochemistry in the 3 experiments (2-3 dishes for each growth condition per experiment). Light microscopy was performed on paraffin-embedded tissues from all three experiments, immunohistochemistry chiefly on the last two experiments, and electron microscopy on the first and third. Following termination of those cultures, 20-30 dishes were set up and collected for each data point for the biochemical studies. The WERI-Rbl line was provided by LAD and was carried in our laboratory for 12 months. Two duplicate sets of experiments in matrix cultures were performed, with a total of about 115 cultures over a period of 8 months. Cultures from both sets were studied immunohistochemically and by electron microscopy. Two to four dishes were used for each growth condition per experiment. Biochemical studies on this cell line were not performed.

Mycoplasma Testing Both lines were monitored at monthly or bimonthly intervals for the presence of mycoplasma contamination (mycoplasma detection test, Gen-Probe, San Diego, CA) on the spent medium. In the Y-79 line, mycoplasma testing was first performed 3 months after initiating the cul-

Neuroblastic Potential of Retinoblastoma Cell Lines

117

AjPJanuary 1989, Vol. 134, No. 1

tures in our laboratory, and was detected at this testing. Before the continuation of experiments, the cells were successfully treated with tetracycline and heat using the method of Bigner et al,3 and remained free of mycoplasma thereafter. Of the three sets of experiments performed for morphology, one was completed before treatment for mycoplasma and two thereafter; the results were similar. The WERI-Rbl line was monitored initially and thereafter every 2 months for mycoplasma contamination (GenProbe) and was found to be consistently negative. Aside from the initial treatment of the Y-79 cell line, antibiotics were never used.

Growth Conditions for Matrix Cultures Cells used for explantation into organ culture dishes (Falcon Labware, Oxnard, CA) were collected by gravity sedimentation from small floating clumps maintained in 60 X 15 mm plastic dishes (Falcon Labware). A drop of cells was put on each Gelfoam (Upjohn, Kalamazoo, Ml) matrix after the foam had been saturated with nutrient medium. The inner well had been previously filled with nutrient medium and the outer well with Hanks BSS 1x (GIBCO, Grand Island, NY), the latter to help maintain internal humidity of the dish. The cultures were maintained in a humidified CO2 incubator (Model 3156 or 3157, Forma Scientific, Marietta, OH) at 4% CO2 and 35.5-36 C. The medium was replaced daily during the experiments and daily or every other day in the stock cultures, depending on the degree of pH shift in the latter.

Table 1. Agents Usedfor Y- 79 Retinoblastoma Cell Line in Matrix Culture

Agent Nerve growth factor (7S) Dibutyryl cyclic AMP (97% cAMP-free) Sodium butyrate Retinoic acid in 95% ethanol Ethanol, 95% Hydrocortisone (cortisol: 17hydroxycorticosterone) Vitamin C (L-ascorbic acid, sodium salt or free acid form)

Final concentration

Range, total days in vitro

120 or 240 ng/mI

10-32

1 or2mM 1 or 2 mM 50 or 100 AM 0.01 or 0.02% (vol/vol)

10-68 7-46 11-32 11-32

0.32 or 0.70 ug/ml

7-32

100 or 200 jg/ml

8-80

fully-defined medium,3940 sometimes supplemented with similar 10% FCS. With the WERI-Rbl line, matrix cultures were grown in DMEM-F1 0 medium with 5 or 10% FCS (HyClone Laboratories) supplemented as above (level of glucose added was 500 mg % and of HEPES was 10 mM). Rotary Suspension Cultures Rotary suspension cultures were maintained at 36 C at 40-50 rpm on a gyratory shaker (New Brunswick Scientific, New Brunswick, NJ) for 30 days. The experiments were performed twice with each line. The cultures were fed with DMEM-F1O medium containing 5 or 10% FCS and carried in 50 ml flasks using 30 ml medium per flask. They were aerated daily with replacement of over 90% of the medium either daily or every other day, depending on the pH shift.

Details of Nutrient Medium The medium used most frequently for the Y-79 line was a 50:50 mixture of Dulbecco's modified Eagle medium (DMEM) (low glucose) and Ham's F10 medium (both reconstituted from powder and obtained from GIBCO) supplemented with 200 (1st experiment) or 500 (last 2 experiments) mg % glucose (Sigma Chemical Co., St. Louis, MO), 28 mM sodium bicarbonate (GIBCO), 0.09 U/ml insulin (Sigma), 1-10 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) (Sigma), and 10% fetal calf serum (FCS). The medium was frozen at -20 C after making, and aliquots were thawed every 3-7 days for use. The serum was a 50:50 mixture of filtered serum from Reheis Chemical Co. (Phoenix, AZ) and Hyclone Laboratories, Inc. (Logan, UT). This complete medium will hereafter be referred to as DMEM-F1 0 medium. Some cells in each experiment were also grown in Waymouth's MB 752/1 medium (GIBCO) supplemented with 100 mg % glucose, 0.09 U/ml insulin, 10 mM HEPES, and 10% fetal calf serum (similar to the above) or with serum-free,

Differentiation-Promoting Agents Used in Matrix Cultures and Evaluated by Immunomorphology and Electron Microscopy Y-79 Line Table 1 indicates the differentiation agents that were individually added to the matrix cultures, their final concentrations, and the range of days in vitro of the cultures. Agents were administered daily, usually beginning 2 days after explanting but sometimes also at longer intervals; all but ethanol were purchased from Sigma Chemical Co., St. Louis, MO. Three experiments were performed over 18 months. Additional cultures were used for biochemical study, as indicated below.

WERI-Rbl Line Differentiation-enhancing agents were added daily for 30 days (±2 days) to the matrix cultures with 10% FCS, beginning 2 days after explantation (see Table 2).

118

Herman et al

AJPJanuary 1989, Vol. 134, No. 1

Table 2. Agents and Growth Conditions Used for WERI-Rbl Retinoblastoma Cell Line Range, total Agents used and growth conditions

days in vitro

Untreated, 10% FCS, matrix Untreated, 5% FCS, matrix Untreated, 10% FCS, rotary suspension Untreated, 5% FCS, rotary suspension Nerve growth factor (7S), 120 ng/ml,

30-60 30 30-34 30-34

matrix Dibutyryl cyclic AMP, 1 mM, matrix Sodium butyrate, 1 mM, matrix Retinoic acid in 95% ethanol, 50 or 100 juM, matrix Ethanol, 95%, 0.01% (vol/vol), matrix

32 ± 2 days 32 ± 2 days 32 ± 2 days 32 ± 2 days 32 ± 2 days

FCS, fetal calf serum; matrix, organ culture system with Gelfoam matrix; rotary suspension culture growth conditions, 40-50 rpm, medium and temperature as for matrix cultures; concentrations of agents are final concentrations.

The levels of agents used were approximate because they were added as 1 or 2 drops of concentrated stock solution to the medium with each daily feeding and mixed by gently swirling the dish. Stocks were freshly made up every 7-14 days and stored at -20 C (retinoic acid and 95% ethanol) or at 4 C when not in use. Retinoic acid was shielded from light and was made up in a solution of 95% ethanol. The other agents were made up in Hanks Balanced Salt Solution (HBBS) (1 X). Untreated cultures and explants treated with 1 or 2 drops of 95% ethanol served as controls.

Untreated Matrix Cultures for Immunomorphologic and Electron Microscopic Studies Y-79 Cell Line

Explants were grown in 5, 10, or 15% FCS for 5-120 days (about 50 dishes) (either DMEM-F1 0 or Waymouth's); serum-free fully defined medium (6 dishes)3940 for 11-67 days; and for 30-59 days in the last medium which had been supplemented with 10% FCS (10 dishes).

WERI-Rbl Line Explants were grown in DMEM-F1 0 medium with 5 or 10% FCS for 30-60 days (see Table 2). Two to four dishes were obtained for each growth condition per experiment.

Light and Electron Microscopy Methods For standard light microscopy, 5-6 ,u sections of formalin-fixed, paraffin-embedded Gelfoam explants and of rotary suspension cultures (cells pelleted by gravity sedi-

mentation) were stained with hematoxylin and eosin (H & E) or occasionally with the Fontana silver method for melanin, or were used for immunohistochemistry. One culture from each pooled group of dishes used for biochemical analysis of the Y-79 line was observed by light microscopy to monitor growth. For electron microscopy, explants on the matrices were processed as described previously34 and examined with an Hitachi HU-12A or a JEOL 100 S electron microscope. For the Y-79 cells, a total of about 700 electron micrographs were analyzed from all groups of cultures in the first and third experiments, including matrix cultures treated with various agents and untreated cultures maintained for similar time periods or longer (see section on untreated matrix cultures above). Suspension cultures were not examined with this technique. For the WERI-Rbl cells, about 300 electron micrographs were studied from all groups of cultures, including suspension cultures.

Immunohistochemical Methods Sections of matrix cultures were immunostained for a number of markers, the majority of which have been applied previously to the study of human nervous system tumors in situ.41 The peroxidase-antiperoxidase method was used42 with a series of rabbit polyvalent antisera and of mouse and human monoclonal antibodies (MAbs, Table 3). All sections were deparaffinized in xylene for 10 minutes and the sections studied with MAb HNK-1 were delipidized previously in chloroform for 24 hours.43 After rehydration through graded methanol (for MAb HNK-1) or ethanol (for all other primary antibodies) and blockage of the endogenous peroxidase activity with hydrogen peroxide (0.5% in methanol for 30 minutes), three-step and fourstep peroxidase-antiperoxidase reactions were performed using polyvalent antisera and MAbs respectively. Before the application of the primary antibodies, specimens were saturated in 10% normal goat (Dako Corp.) (for polyvalent antisera) or 10% normal swine sera (Dako Corp.) (for MAbs) for 30 minutes. Sections were incubated at 4 C for 48 hours (dilution 1:30,000) or 18 hours (dilution 1:2000) with the anti-MAG MAb, and at a similar temperature for 18 or 24 hours with all the other primary antibodies. Goat anti-rabbit immunoglobulins (Cooper Biomedical, Malvern, PA) (dilution, 1:50), and rabbit peroxidase-antiperoxidase-complex (Dako) (dilution, 1:200) were applied at room temperature for 10 and 20 minutes, respectively, after polyvalent primary antisera, except in the case of MBP, where swine antirabbit immunoglobulins were applied at room temperature for 20 minutes as supplied (Dako PAP Kit, System 20, K531). Rabbit anti-mouse im-


119


Table 3. Primary Antibodies Used in Immunohistochemical Characterization of Y- 79 and WERI-Rbl Retinoblastoma Lines Molecule Dilution Primary antibody Source Reference Monoclonal Mouse and Human Antibodies NFP-H VP Collins Tp-NFP1 A3 1:100 54,55 MAP 2 A Frankfurter AP18 1:100 60 A Frankfurter TUJ1 1:100 h,34 70 1:100 SY38 Synaptophysin Boehringer Mannheim 71; 74-76 Retinal S-antigen LA Donoso MAbA9-C6 1:2,000* 56 M Vandevelde (Univ. Berne, Anti-MAGt MAG 1:30,000 or 88 1:2,000 Switzerland) HNK-1 Leu-7 1 :300t Becton Dickinson 43 Polyvalent Rabbit Antisera Anti-rat NSE NSE 1:1000 PJ Marangos 59 (E.C. 4.2.1.11) Anti-bovine opsin H-W Korf, A Ho (NIH, 1:400 Opsin 89 Bethesda, MD) Anti-bovine HIOMT (E.C. 2.1.1.4) HIOMT 1:200 AF Wiechmann 84 Anti-human GFA protein GFA protein 1:1400 LF Eng 13, 58 MBP Anti-human MBP (Dako PAP kit, 1:3 (prediluted Dako Corporation 90 System 20 k531) working solution) NFP-H, 200 kD subunit of neurofilament protein; MAP 2, microtubule associated protein 2; h#4, human homologous gene of class IlIl ,-tubulin isotype; MAG, myelin-associated glycoprotein; NSE, neuron-specific enolase; HIOMT, hydroxyindole-0-methyltransferase; GFA protein, glial fibnllary acidic protein; MBP, myelin basic protein. * Original concentration in ascites fluid: 2.84 mg/ml.

t Clone designation not available.

1 0.671gg of purfied immunoglobulin/ml. Note: When 2 dilutions are given, the first is for Y-79 and the second for WERI-Rbl line.

munoglobulins (Dako) (dilution, 1:25 for MAbs TpNFP1A3, AP18, TUJ1, and SY38, and 1:50 for the MAb HNK-1 and MAbA9-C6), swine anti-rabbit immunoglobulins (Dako) (dilution, 1:25 for MAbs Tp-NFP1A3, AP18, TUJ1, and SY38, and 1:50 for MAb HNK-1 and MAbA9C6) and rabbit peroxidase-antiperoxidase-complex (dilution, 1:1 00 for MAbs Tp-NFP1 A3, AP1 8, TUJ1, and SY38; and 1:200 for MAb HNK-1 and MAbA9-C6) were sequentially applied at room temperature for 20, 20, and 30 minutes, respectively after the mouse MAb. Rabbit anti-human IgM ,u-chain (Dako) (dilution, 1:500, for 30 minutes), swine anti-rabbit immunoglobulins (dilution, 1:50, for 20 minutes) and rabbit peroxidase-antiperoxidase-complex (dilution 1:200, for 20 minutes) were sequentially used at room temperature after the human anti-MAG MAb. All reagents were dissolved in 0.05 M Tris-buffered saline (pH 7.6) containing 1% normal goat (for polyvalent antisera) or 1 % normal swine (for MAb) sera. The immunohistochemical reactions were developed at room temperature in freshly prepared 3,3'-diaminobenzidine tetrahydrochloride (Sigma), 4.0 or 5.0 mg in 6.5 ml of 0.05 M Trisbuffered saline containing 0.015% of hydrogen peroxide. As positive controls, surgical specimens of human gangliogliomas (for anti-GFA protein and anti-NSE polyvalent antisera, and for Tp-NFP1A3 MAb); of human neuroblastoma and human fetal and adult cerebrum, brainstem and cerebellum (for MAbs AP18 and TUJ1); of human enucleated eyes (for MAbA9-C6, anti-MAG MAb,

MBP, opsin, and HIOMT); and autopsy specimens of spinal nerve roots (for HNK-1 and anti-MAG MAbs, and MBP) and of human acoustic schwannoma (for MAb HNK-1) were stained in parallel. Negative controls were obtained by applying rabbit serum instead of the polyvalent antisera, and the following nonspecific mouse myeloma and human purified immunoglobulins instead of the MAbs: mouse IgM kappa/lambda-1 (Miles Scientific, Naperville, IL) instead of MAb HNK-1; mouse IgGl-kappa (Sigma) instead of Tp-NFP1A3 and SY38; mouse IgG2a (Sigma) instead of MAbA9-C6, and MAbs AP18 and TUJ1; and chromatographically purified human IgM immunoglobulin (Cooper Biomedical) instead of anti-MAG MAb. Mouse and human nonspecific immunoglobulins were applied at the same dilutions as the MAbs. For GFA protein immunoperoxidase staining, trypsinization of additional sections of a Y-79 explant exposed to 2 mM dcAMP for 30 days was carried out using poly-Llysine coated glass slides as follows (procedure adopted after LFE): 1) 5 mg of poly-L-lysine (Sigma) were dissolved in 100 ml triple distilled H20; 2) 1 ml of (1) was placed per slide (using surface tension) and left at room temperature for 30 minutes; 3) the solution was in turn removed by rinsing three times in triple distilled H20; slides were subsequently air-dried and placed in a 60 C oven for 1 hour; 4) tissue sections were then placed on coated slides and subjected to deparaffinization and rehydration steps as above; 5) tissue sections were incubated for 5 minutes

120

Herman et al


(1-10 minutes) with trypsin (0.5 mg/ml acetylated trypsin) (type V-S, Sigma) in 0.05 M Tris-buffered saline, pH 8.0, containing 11 mM of CaCI2, immediately before incubation with blocking antibody (10% normal goat serum-Trisbuffered saline). The final sections were counterstained lightly with Harris's hematoxylin; occasional sections were not counterstained.

Biochemical Methods For all biochemical assays the following materials were evaluated: 1) Treated (1 mM butyrate) and untreated Y79 cells in organ culture for 30-32 days. 2) KB epidermoid carcinoma cells, obtained from the American Type Culture Collection (Rockville, MD) and carried for a total of 6 months in DMEM-F10 medium with 10% FCS without antibiotics. The line was mycoplasma-free initially and continued to be so on retesting every 2 months. Stock cultures in plastic dishes and Gelfoam cultures were set up and maintained as for the untreated retinoblastoma lines, and 20-30 dishes of matrix cultures were terminated at 30-32 DIV and pooled. 3) Gelfoam matrix. 4) DMEM-F1O nutrient medium with 10% FCS. 5) Hanks BSS (1 X). All samples to be assayed (except Gelfoam) were dipped several times in R.T. BSS 1 X, blotted, rapidly frozen on dry ice, and stored in liquid nitrogen vapors (cultures with their matrices) or at -80 C (others). Gelfoam was not frozen but was stored and shipped at ambient temperature. The other materials were shipped on dry ice to the appropriate laboratories for biochemical analyses. Twenty to thirty Gelfoam cultures were pooled for each experimental assay and 1 set of the 5 samples (see above) was evaluated by each laboratory, except in the case of GFA protein where in addition to the foam, medium and BSS, 7 different experimental groups of cells were assayed, including untreated Y-79 cells (3 experiments), 1 mM dibutyryl cyclic AMP-treated Y-79 cells (1 experiment), 1 mM sodium butyrate-treated Y-79 cells (1 experiment) and KB cells (2 experiments). In all experiments the treated Y-79 cells received agents daily for 30 days, beginning 2 days after explantation. For all assays the cells were homogenized and the extracts evaluated. The results were quantitated on a total protein or DNA basis. The method of Burton"445 with or without modifications46-48 was used for DNA assay, and the method of Lowry et al49 was used for protein determinations.

NSE

Using a double antibody radioimmunoassay procedure as described previously,50 pure human NSE was io-

dinated with the Bolton-Hunter procedure.51 Standard curves ranged from 50 pg to 5 ng. Anti-NSE serum was raised in rabbits and displayed no crossreactivity with liver or muscle enolase. Cell culture extracts were prepared by homogenization in 10 mM Tris-phosphate buffer, pH 7.5, containing 1 mM MgSO4 using a Brinkman polytron (setting 6 for 20 seconds). The resulting homogenate was centrifuged at 100,000g for 1 hour and the supernatant saved for assay. All assays were performed in triplicate. NF Protein The three neurofilament protein subunits (68, 160, and 200 kD) were quantified by a double enzyme immune assay (EIA) (Bio-Rad Labs, Richmond, CA) in frozen Gelfoam cultures as follows. The weighed cultures were homogenized (4 C) in isolation buffer,52 agitated for 30 minutes, and centrifuged (1 0,000g, 20 minutes, 0 C). Mouse brain homogenate supernatant was used as a positive control, while nutrient medium and the supernatant of homogenized Gelfoam were used as negative controls. Pellets were collected for DNA determinations45 and the supernatant applied in serial dilutions to nitrocellulose membranes (Bio-dot microfiltration apparatus, Bio-Rad Labs). The EIA was performed as recommended (Bio-Rad instruction manual) with horseradish peroxidase and 4chloro-1 -naphthol at a sensitivity of 50 pg protein.53 Four MAbs were used to identify the NF protein epitopes as follows: 2 for the 200 kd subunit (Tp-NFP1A35455 and RPN. 1103, Amersham International plc, Stockholm); 1 for the 160 kd subunit (RPN.1 104, Amersham); and 1 for the 68 kd subunit (RPN. 1105, Amersham). The second antibody was a peroxidase conjugated rabbit anti-mouse antiserum (Dako) diluted 1:500 (2% bovine serum albumin, Tris-buffered saline). All assays were repeated 3 times for all MAbs and gave consistent results. Homogenate concentrations greater than an equivalent of approximately 22 ,ug protein could not be used.

Retinal S-Antigen Foam cultures of Y-79 cells, both untreated and treated with 1 mM butyrate, KB cells, and a sample of the Gelfoam matrix were homogenized in HBBS and centrifuged briefly (10 minutes at 2000g). The supernatant was removed and total protein was determined. Retinal S-antigen was quantitated by an ELISA method as described previously56 and by developing the plate with the addition of ascites purified anti-S-antigen MAbA9-C6 (2.84 mg/ml; 1:1000 dilution). Absorbance was read at 490 nm using a Bio-Tex El 310 automated microtiter plate reader. All assays were performed in triplicate.


121

AJPJanuary 1989, VoL 134, No. 1

A: All treated and unFigure 1. Immunohistochemical characterization of the Y- 79 retinoblastoma cell line in matrix culture. treated matrix cultures were immunopositive for anti-NSE polyvalent antiserum. Thirty-day culture grown in a supplemented B: ConfluDMEM-F10 medium with 5%fetal calfserum. Peroxidase-antiperoxidase method; bematoxylin counterstain. (X400) ent clusters of cellsstrongly immunopositiveforMAb TUJi to class III -tubulin isotype (human: b4)following treatment with dcAMP C: Higher (1 mM) for 30 days. Thirty-two days in vitro. Peroxidase-antiperoxidase method; hematoxylin counterstain. 0X400) magnification of B. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X600) D: Enhanced immunoreactivityfor MAb AP18 to MAP2 following exposure to dcAMP (1 mM)for 29 days. Thirty-one days in vitro. Peroxidase-antiperoxidase method; E: Immunopositive cells for MAb SY38 to synaptophysin, treated with Na+ butyrate (1 mM) hematoxylin counterstain. (X600) F: Low-power view of for 30 days. Thirty-two days in-vitro. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X 600) culture in E showing a pattern ofgeneralizedMAb SY38 immunoreactivity. Peroxidase-antiperoxidase method without hematoxylin counterstain. (X400)

122

Herman et al

AJPJanuarj' 1989, Vol. 134, No. 1

Neuroblastic Potential of Retinoblastoma Cell Lines 123 AJPJanuary 1989, Vol. 134, No. 1 Figure 2. Immunohistochemical characterization of the WERI-Rbl retinoblastoma cell line in matrix culture. A: Widespread immunopositivity for anti-NSE polyvalent antiserum was present in matrix cultures irrespective of treatment. Thirty-day culture cultivated in a supplemented DMEM-FlO medium with 5% fetal calfserum. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X400) B: Focally immunopositive cells for MAb SY38 to synaptophysin. Thirty-two-day culture after treatment with NGF, 7S (120 ng/ml)for 30 days. Peroxidase-antiperoxidase method without hematoxylin counterstain. (X400) C: Occasional positive cells for MAb Tp-NFPlA3 to NFP-H. Thirty-three-day culture after treatment with Na+ butyrate (1 mM)for 31 days. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X400) D: Focallypronounced immunoreactivityfor MAb Tp-NFPlA3 following exposure to dcAMP (1 mM) for 28 days. Thirty-day matrix culture. Peroxidase-antiperoxidase method without hematoxylin counterstain. (X 600) E: Groups of cells positive for AAb TUJl to type HI II-tubulin isotype (human: b*4). Thirty-one-day culture treated with Na+ butyrate (1 mM)for 29 days. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X400) F: Accentuated immunoreactivityforMAMb TUJl in 32-day explant, exposed to dcAMP (1 mM)for30 days. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X600) G: MAb TUJl reacts with perikarya and neurites ofpolar neurons. Human fetal entorhinal cortex, 20 weeks of gestation. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X400) H: Immunopositive cell clusters for MAb AP18 to MAP2. Thirty-two-day culture treated with 1 mM dcAMP for 30 days. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X400) I: Clusters of cells positive for MAbA9-C6 to retinal S-antigen. Thirty-two day culture treated with 1 mM dcAMPfor30days. Peroxidase-antiperoxidase method; hematoxylin counterstain. (X400) _9

GFA Protein The ELISA,57 SDS-PAGE,

analyses with anti-GFA protein- were employed. All says were performed in triplicate.

as-

Results

Light Microscopy and Immunohistochemistry of Matrix Cultures Light Microscopy Both lines grew in sheets of tightly aligned, generally round, uniform cells with scanty cytoplasm and large round-to-oval nuclei with small, prominent nucleoli. Mitotic figures (5 per low-power field) were readily seen. There were no differences between cells superficial or deep within the explants, or between those cells adjacent to the foam or at a distance, nor were differences found between cells cultured in Waymouths' or DMEM-F10 medium with 5, 10, or 15% FCS for periods up to 120 days, or in the fully-defined serum-free medium. Although cells sometimes clustered in a tubulelike arrangement, no true rosettes were seen, nor were ganglionic or astrocyticlike processes or melanin pigment ever found. The variously treated cultures appeared similarly devoid of differentiated features in routinely stained preparations. Foci of necrosis were noted in most matrix cultures, but they generally occupied no more than a quarter of the deeper parts of the explants. After the addition of 100 uM retinoic acid or of 2 mM sodium butyrate, however, necrosis was generally increased, although it still occupied less than half of the total area of the explants. By immunohistochemistry the cells of both lines were widely positive for NSE (Figures 1 A and 2A). Clusters of cells were positive for h#4 tubulin (Figures 1 B, 1C, and 2E), MAP2 (Figures 1 D and 2H), and synaptophysin (Figures 1 E, 1 F, and 2B). In both lines, immunopositivity for

hf34 tubulin and MAP2 was accentuated by the addition of 1 mM dcAMP, as shown in Figures 2F and 2H for the WERI-Rbl cells. In addition, the latter were immunopositive for NFP-H (Figure 2C), the reaction being more pronounced after the addition of dcAMP (Figure 2D). A number of both treated and untreated WERI-Rbl cells were positive for the retinal S-antigen (Figure 21), but none of the Y-79 cells were positive for either NFP-H or S-antigen. The cells of both lines were consistently negative for GFA protein, MAG, MBP, opsin, HIOMT, and for the epitope recognized by the Leu-7 antibody. Prior trypsinization did not alter the negative results for GFA protein. A comparison of the immunohistochemical results in both retinoblastoma cell lines is summarized in Table 4. Light Microscopy of Rotary Suspension Cultures

Cell aggregates obtained from suspension cultures were also devoid of differentiated features. No rosettes were found. Central necrosis was sometimes present in the cell spheroids. Table 4. Comparison ofImmunohistochemical Results on WERI-Rbl and Y- 79 Retinoblastoma Cell Lines in Matrix Culture Results Markers WERI-Rbl Y-79 + + Neuron-specific enolase NFP-H (MAb Tp-NFP1 A3) + 0 MAP 2 (MAb AP18) + + + + h,4 (MAb TUJ1) + + Synaptophysin (MAb SY38) Retinal S-antigen (MAbA9-C6) + 0 0 0 Opsin HIOMT 0 0 Glial fibrillary acidic protein 0 0 0 0 Myelin-associated glycoprotein (MAb) 0 0 Myelin basic protein Leu-7 (MAb HNK-1) 0 0 NFP-H, 200 kd subunit of neurofilament protein; MAP 2, microtubuleassociated protein 2; hf#4, designation of human homologous gene of class Il ,B-tubulin isotype; HIOMT, hydroxyindole-0-methyltransferase.

124

Herman et al


Electron Microscopy of all Cultures No differences were found between treated and untreated, or between short- and long-term matrix cultures. In general, the cells grew in sheets and were closely packed (Figure 3A). The cell surfaces sometimes showed a few short microvilli, and occasionally the apices of a cluster of cells were compactly arranged and oriented toward a central point. No rosettes were seen in the Y-79 cultures, whereas occasional polarization of processes towards a lumen in a rosettelike arrangement was found in the WERI-Rbl cultures (Figure 4A). In the suspension cultures of WERI-Rbl, there were sparse cell contact specializations; no cell polarization was noted. No electron microscopy was performed on the Y-79 cells in suspension culture. The cells had large round-to-oval nuclei, with prominent nucleoli, and were sometimes lobulated. The cytoplasm contained abundant ribosomes, sparse smooth and granular reticulum, occasional microtubules, and Golgi organelles. Pools of glycogen rosettes were sometimes found, and clusters of annulate lamellae (almost always only one per cytoplasmic profile) were frequent (Figure 4B). Occasional centrioles and basal bodies were noted. Adhesion points, sometimes in series of 2-3, and tight junctionlike structures were seen between some cells in most low-magnification fields. The cytoplasm of the Y-79 cells frequently contained large mitochondria with stacked or disoriented cristae, most prominent in the ethanol and retinoic-acid treated cells, but also in other treated and untreated cells (Figures 3B and 3C). These abnormal mitochondria were rare in the WERI-Rbl line. Foci of necrosis were sometimes observed, especially with the higher levels of added butyrate or retinoic acid, as noted by light microscopy. Intermediate filaments, synapses, synaptic ribbons, clear-centered or clustered dense-core vesicles, processes resembling neurites (numerous microtubules arranged in parallel) or myelinlike concentric lamellae were never found.

Biochemical Estimations In Y-79 cells grown with and without 1 mM sodium butyrate, 160 ng NSE/ml/mg protein was present, while in the KB epidermoid cells the NSE level was 84.8 ng NSE/ ml/mg protein. All other samples were negative for NSE. No GFA protein was detected in any of the samples, either as quantitated on a total DNA or on a total protein basis. No retinal S-antigen was present in either the Y-79 or KB cell line, whereas the retinal S-antigen controls were positive; the Gelfoam extracts, nutrient medium, and BSS showed no color development. All homogenates of both the butyrate-treated and un-

treated Gelfoam cultures of the Y-79 line as well as those of the KB cells were consistently negative for the 200, 160 and 68 kd NF proteins, even when amounts of homogenate equivalent to 11.3 ,ug DNA were applied to the nitrocellulose and studied with the three MAbs which gave results. This is an amount of homogenate with an equivalent DNA concentration more than 2000 times greater than that required to detect the 200 kd protein in the brain homogenate. The 200 kd NF protein could be consistently detected when amounts of the homogenate from normal mouse brain, equivalent to 4.75 ng DNA, were applied to the nitrocellulose, and when the MAb Tp-NFP1 A3 was used. The MAb RPN.1 103 gave almost identical results at double the antigen concentration. The MAb RPN.1 104 directed against the 160 kd protein gave no results. The MAb RPN.1 105 directed against the 68 kd protein required an amount of homogenate equivalent to 39 ng DNA for detection of the protein.

Discussion The aim of this study was to assess the differentiation potential of the human Y-79 and WERI-Rbl retinoblastoma cell lines when maintained in an organ culture system and after exposure to various differentiation-promoting agents. In our in vitro system, the cells of both lines demonstrated immunopositivity for three lineage-related neuronal markers: NSE, MAP2, and the class IlIl ,-tubulin isotype (human designation: h34), and for synaptophysin. In addition, the cells of the WERI-Rbl line were immunopositive for NFP-H, an unequivocal neuronal marker, and for the retinal S-antigen, a marker of photoreceptor differentiation. Thus, while the Y-79 demonstrated the acquisition of antigenic determinants associated with neuronal differentiation to a limited extent, the WERI-Rbl line exhibited this potential to a more advanced degree. In both lines, immunopositivity for MAP2 and h,34 was accentuated following exposure of the explants to dcAMP, and that of NFP-H in the WERI-Rbl was similarly accentuated. In the Y-79 cells, the direction of differentiation conformed to the observations by two other groups of workers,24,25,28 who reported that the cells express NSE,2425 catecholamines,28 tetanus toxin, dopamine-3-hydroxylase,24 and choline acetyltransferase.27 In addition to our negative histochemical and biochemical results for the presence of antigenic markers characteristic of the retina, however, we were unable to demonstrate any morphologic differentiation towards photoreceptor cells as reported by others with the electron microscope.2728 But our immunopositive findings with the WERI-Rbl line confirm and extend previous observations7 on the capacity of this line to express photoreceptor differentiation.


125


A: Cell line maintained on spongefoamfor2 months Figure 3. Electron microscopy of Y- 79 retinoblastoma cells in matrix culture. B: Another field from a sister culture in DMEM-F10 medium + 10% FCS. The cells lack features of differentiation. (X6000) C: Another matrix culture similarly maintained showing a giant atypical mitochondrion with redundant cristae. (X 18,000) maintainedfor one month in Waymouth's medium +5% FCS. A different type of giant mitochondrion with markedly convoluted cristae. (X15,000)

126 Herman et al AJPJanuary 1989, Vol. 134, No. 1

;.74

.*klii