Proc. Natl. Acad. Sci. USAVol. 90, pp. 8534-8537, September 1993Cell Biology

Expression of 300-kilodalton intermediate filament-associatedprotein distinguishes human glioma cells from normal astrocytes

(gila/differentlation/astrecytomas)

HSI-YUAN YANG*, NORMAN LIESKA*, ROBERTA GLICKt, DEREN SHAO*, AND GEORGE D. PAPPAS*tDepartments of *Anatomy and Cell Biology, and tNeurosurgery, College of Medicine, University of Illinois at Chicago, Chicago, Illinois 60612

Communicated by Keith R. Porter, April 30, 1993

ABSTRACT The availability of biochemical markers todistinguish glioma cells from normal astrocytes would haveenormous diagnostic value. Such markers also may be of valuein studying the basic biology of human astrocytomas. Thevimentin-binding, 300-kDa intermediate filament (IF)-assodated protein (IFAP-300kDa) has recently been shown tobe developmentally expressed in radial glia of the centralnervous system of the rat. It is not detected in the normal orreactive astrocytes of the adult rat nor in neonatal rat brainastrocyts in primary culture. In the present study, double-label immunofluorescence microscopy using antibodies toIFAP-300kDa and gial fibrillary acidic protein (GFAP, anastrocyte-specific IF structural protein) identifies this IFAP inGFAP-containing tumor cells from examples of all three majortypes of human astrocytomas (i.e., well-differentiated, ana-plastic, and glioblastoma multiforme). Astrocytoma cells inprimary cultures prepared from all three astrocytomas alsoexpress this protein. It is not detectable in normal adult braintissue. Immunoblot analyses using the IFAP-300kDa antibodyconfirm the presence of a 300-kDa polypeptide in fresh astro-cytoma preparations enriched for IF proteins. These resultssuggest the utility of IFAP-300kDa as a marker for identifica-tion of human glioma cells both in vitro and in situ.

Little is known about the distinctive phenotypic character-istics of astrocytomas. Presently, the basis for identificationand classification of human gliomas in a diagnostic settingrests primarily upon histologic and morphologic criteria(1-5). The fundamental problem remains defining a means bywhich tumorous astrocytes can be distinguished from normalones, both in situ and in vitro. The identification ofdistinctivephenotypic markers for astrocytomas should help to alleviatethis situation. This approach furthermore would provideadditional information about the basic cell biology of thesetumors.

Proteins ofthe intermediate filament (IF) cytoskeleton thatare developmentally regulated and show tissue type-specificexpression hold great potential as cell markers for immuno-typing human astrocytomas (6). Indeed, glial fibrillary acidicprotein (GFAP), an IF structural subunit protein specificallyexpressed in astrocytes, has been widely used as a marker forastrocytic tumors (7-16). Vimentin, another IF structuralprotein present in immature and some mature astrocytes, hasalso been shown to be expressed by astrocytomas (15, 17,18). However, these proteins are poor diagnostic markersbecause they are normal constituents of both mature andreactive astrocytes (19-21).

In addition to such structural proteins, the IF cytoskeletoncomprises IF-associated proteins (IFAPs) (22, 23). We havepreviously identified and characterized a 300-kDa vimentin-binding IFAP (IFAP-300kDa) in a baby hamster kidney cell

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

w_. "..*... .. ,^., .... ................ .....*: .. .: -

*::. g..... .g.*.: .

.X. .

Se:e:-e.

:E'

.L.:^ ^:s .: :

A B C D E F

FIG. 1. The polypeptide compositions of IF preparations fromBHK-21 cells (lane A) and a human GBM specimen (lane B) weredetermined by SDS/7.5% PAGE (Coomassie blue-stained; 20 pg and50 pg of protein loaded, respectively). I, V, and G designateIFAP-300kDa, vimentin, and GFAP. IFAP-300kDa was identified onnitrocellulose transfers from lanes A and B to lanes C and D,respectively, by immunoblot analysis with anti-IFAP-300kDa mAb.Anti-IFAP-300kDa immunoblots of IF preparations from BHK-21cells (lane E) and normal adult brain (lane F) show the absence ofIFAP-300kDa in the nontumorous brain tissue (20 pg and 50 pg ofprotein loaded, respectively).

line (BHK-21) (24, 25). The ontogeny of IFAP-300kDa in thecentral nervous system of the rat demonstrates that it isexpressed in radial glia and their immediate derivatives, butits expression terminates postnatally: it is not detectable inGFAP-containing astrocytes, regardless of vimentin content,in adult rat brain. Neither is it found in rat reactive astrocytesinduced by stab wound or in astrocytes in primary culturesprepared from neonatal rat brain, although all such cells arevimentin/GFAP-containing (26).Because markers of early differentiation can be reex-

pressed during tumorigenesis, we examined a spectrum ofhuman brain astrocytoma types for IFAP-300kDa immuno-reactivity. The specific expression of this IFAP by astrocy-toma cells, both in situ and in vitro, is described.

MATERIALS AND METHODSSpecimens and Controls. Ten fresh glioma specimens ob-

tained immediately after surgical resection were frozen inliquid nitrogen and stored at -70°C. Specimens includedthree well-differentiated astrocytomas (WDA), two anaplas-tic astrocytomas (AA), and five glioblastomas multiforme(GBM). All were supratentorial astrocytomas obtained frompatients at the University of Illinois and Cook County Hos-pitals. Tumors were graded according to the WHO classifi-cation (1) by using routine light microscopy on paraffimsections of formaldehyde (3.7%)-fixed samples stained with

Abbreviations: IF, intermediate filament; IFAP, intermediate fila-ment-associated protein; GFAP, glial fibrillary acidic protein; WDA,well-differentiated astrocytomas; AA, anaplastic astrocytomas;GBM, glioblastoma(s) multiforme; mAb, monoclonal antibody.tTo whom reprint requests should be addressed.

8534

Proc. Natl. Acad. Sci. USA 90 (1993) 8535

FIG. 2. Double-label immunofluorescence microscopy of methanol-fixed cryostat sections of normal brain tissue shows the absence ofIFAP-300kDa immunoreactivity (B) in any cells in the specimens, particularly astrocytes identified here by anti-GFAP staining (A). (x120.)

hematoxylin and eosin. Control normal brain tissue wasobtained surgically from two trauma victims.

Materials. Tissue culture constituents were purchasedfrom GIBCO, and chemical reagents were from Sigma. Themonoclonal antibody (mAb) to IFAP-300kDa (anti-IFAP-300kDa) was prepared and characterized as described (24,25). Hybridoma cells were grown in DMEM supplementedwith 1 mM sodium pyruvate, 2 mM L-glutamine, 10% (vol/vol) fetal bovine serum, and penicillin/streptomycin as be-low. Culture supernatants were used for immunofluores-cence and immunoblotting studies. Rabbit anti-GFAP anti-body and mouse anti-vimentin mAb were purchased fromIncstar (Stillwater, MN) and ICN, respectively. Fluoresceinisothiocyanate (FITC)- and rhodamine-conjugated secondaryantibodies were purchased from Kirkegaard & Perry Labo-ratories.

Cell Culture. Primary cultures of astrocytoma cells wereprepared from tissue specimens of a WDA, an AA, and twoGBM according to a method modified from that ofMcCarthyand de Vellis (27). Fresh astrocytoma samples were placed inDulbecco's modified Eagle's medium (DMEM) and mincedwith a razor blade. The minced tissue was triturated with a5-ml pipette and further dissociated by passage through an18-gauge needle, a 20-gauge needle, and finally a 22-gaugeneedle. The resultant cell suspension was plated in a 1:1(vol/vol) mixture of DMEM and Ham's F-12 medium sup-plemented with 101% fetal bovine serum, 50 units ofpenicillinper ml, and 50 ,ug of streptomycin per ml. The medium waschanged on day 5 after explanation and every third daythereafter. Cell cultures were maintained at 37°C in a humid-ified atmosphere of 95% air/5% CO2. Confluent cultures ofastrocytes were transferred to other dishes after treatmentwith 0.05% trypsin/0.53 mM EDTA solution. Baby hamsterkidney cells (BHK-21) were cultured as described (25).Immunofluorescence Microscopy. Tissue and cultured cells

were prepared for indirect immunofluorescence as described(25, 28) and were used after one or two passages.IF Preparation, SDS/PAGE, and Immunoblotting. Native

IF preparations (IF-enriched Triton X-100/high salt-insolublecellular residues) were isolated from astrocytoma samplesand BHK-21 cells by the procedure ofZackroff and Goldman(29). SDS/PAGE was performed by the method of Laemmli(30) using 7.5% polyacrylamide. For immunoblot analysis,SDS/PAGE-separated specimens were electrotransferredonto nitrocellulose paper by the method ofTowbin et al. (31)and processed as described (25) with horseradish peroxidase-conjugated secondary antibody for visualization. SDS (0.1%)was included in the transfer buffer to enhance transfer ofhighmolecular weight polypeptides. Commercial protein stan-dards (Bio-Rad) were used for molecular weight calibration.

RESULTSIFAP-300kDa was originally identified and characterized bya mAb produced against IF preparations obtained from

BHK-21 cells (24, 25). As shown by immunoblot analysis, apolypeptide with a molecular weight of300 kDa is specificallyidentified (Fig. 1, lanes A, C, and E). Immunoblot analysis ofIF preparations obtained from astrocytoma specimens alsoidentified the 300-kDa polypeptide (Fig. 1, lanes B and D). Incontrast, the 300-kDa protein was not identified in IF prep-arations isolated from normal brain tissue (Fig. 1, lane F).The major polypeptides of the astrocytoma IF preparationscomprised vimentin and GFAP structural IF proteins (Fig. 1,lane B).

Methanol-fixed cryostat sections of fresh astrocytoma andnormal brain specimens were examined by immunofluores-

FIG. 3. GFAP (A), vimentin (B), and IFAP-300kDa (C) weredetectable in all human astrocytomas examined, as shown here byimmunofluorescence microscopy of a methanol-fixed cryostat sec-tion of a GBM specimen. (x200.)

-

Cell Biology: Yang et aL

Proc. Natl. Acad. Sci. USA 90 (1993)

FIG. 4. Double-label immunofluorescence microscopy with anti-IFAP-300kDa mAb (A) and rabbit anti-GFAP antibody (B) revealed thepresence of the IFAP in most cells of a methanol-fixed cryostat section of a human AA specimen. Within these cells the two IF cytoskeletalproteins were colocalized. (x85.)

cence microscopy utilizing antibodies directed against two IFstructural proteins (GFAP and vimentin) and IFAP-300kDa.Normal brain specimens demonstrated characteristic matureastrocytes via GFAP-positive staining and IFAP-300kDanegativity (Fig. 2). As expected, all astrocytoma specimens(WDA, AA, and GBM) were immunopositive for both GFAPand vimentin. Interestingly, all were also immunopositive forIFAP-300kDa (Fig. 3). This result was corroborated bydouble-label immunofluorescence microscopy, which dem-onstrated the presence of this IFAP in most of the GFAP-positive astrocytic tumor cells (Fig. 4).To determine the status of malignant astrocytes in vitro,

primary cultures ofastrocytic tumor cells were prepared fromWDA, AA, and GBM tumors, and the reactivities of IFAP-300kDa and GFAP were again examined by immunofluores-cence microscopy. Despite varied morphologies for the cul-tured astrocytoma cells, IFAP immunoreactivity was widelydistributed throughout the cultures in 80% of the GFAP-positive cells, in agreement with the results for fresh tumors.Moreover, IFAP-300kDa was shown to colocalize with theGFAP-containing glial IFs (Fig. 5), thereby supporting its IFcytoskeletal association as originally described (24, 25).

DISCUSSIONIn contrast to the other two major cytoskeletal elements(microtubules and microfilaments), IFs comprise a polydis-perse family of at least 30 biochemically distinct subunit(structural) proteins that can be classified into six subgroupson the basis of amino acid sequence and structural domainorganization (32-35). Because the expression of IF proteinsis both tissue-type-specific and developmentaJly regulated(33, 36), these proteins are widely used as markers todetermine the origin of cells in studies of differentiation andin typifying tumors (6). For example, within the macroglialcell lineage, the expression of IF structural proteins changesfrom vimentin in radial glia, to vimentin and GFAP inimmature astrocytes, to GFAP in mature astrocytes (37-44).The demonstration of both GFAP and vimentin in all astro-

cytoma specimens of the present study, in accord withprevious reports (17, 18), indicates that tissue-type-specificexpression ofGFAP is maintained by these tumors. While theexpression of vimentin suggests the relative differentiationstate of these tumor cells (namely, immature), it is importantto note that vimentin expression persists in many adultastrocytes, especially those of the white matter (19, 20).Consequently, the usefulness of these IF proteins as markersof the malignant astrocytic phenotype is limited; much reli-ance must remain upon morphologic characteristics for theidentification of astrocytomas.

In contrast, IFAPs hold great potential as indicators of themalignant phenotype. This is because these molecules can beused as differentiation markers in determining cell lineagesand the functional subtypes of mature cells. IFAPs do not ofthemselves form IFs but are physically and functionallyassociated with them. Experimentally, this association isdemonstrable by IFAPs' coisolation, colocalization, cocy-cling, and in vitro recombination with IF (24, 25). IFAPsappear to play functional roles in the supramolecular orga-nization of the cytoskeleton and in the regulation of IFinteraction with other cellular components (22, 23). Fourdifferent IFAPs have thus far been documented in glia. A48-kDa IFAP is expressed late in astroglial development butnot in radial glia or their immediate derivatives (45, 46). Incontrast, transient expression of a vimentin-associated 400-kDa IFAP (IFAPa-400) is described for chick radial glia-likecells derived from neuroectoderm (47). Recently, we havedescribed the differentiation-related expression of two vi-mentin-binding IFAPs, IFAP-300kDa and IFAP-70/280kDa,in rat radial glia and their direct derivatives (26, 28, 48).Expression of the former IFAP in rat spinal cord ceases bypostnatal day 2, while that of the latter continues untilpostnatal day 9 (26, 28). These studies clearly indicate thatthe expression of glial IFAPs is not only filament type-specific but also developmentally regulated within individualcells.Inasmuch as defining distinctive phenotypes is a funda-

mental aim of all cancer research, two findings from this

FIG. 5. GFAP-containing astrocytoma cells (A) in a primary culture prepared from a GBM specimen exhibited IFAP-300kDa reactivity (B)upon examination by double-label immunofluorescence microscopy. (x410.)

8536 Cell Biology: Yang et al.

Proc. Natl. Acad. Sci. USA 90 (1993) 8537

study are particularly significant: (i) IFAP-300kDa, a differ-entiation marker of developing astrocytes, is detectable in allastrocytic tumors but not in normal adult human astrocytesand (ii) IFAP-300kDa expression is maintained by culturedastrocytoma cells, permitting identification and selection ofsuch cells for in vitro study ofthese tumors and their proteins.The expression of an early differentiation antigen by malig-nant cells is not totally unexpected given that certain prop-erties of such cells suggest a less differentiated cellular state.Moreover, the IF cytoskeleton and the apparent regulators ofits organization, like IFAP-300kDa (24, 25), are intimatelyrelated to some ofthese properties, including cell size/shape,nuclear placement, and surface-membrane specialization forcell-cell adhesion (33, 49, 50). However, the exact biologicaladvantage conferred on astrocytoma cells by the reexpres-sion of IFAP-300kDa is not known.The identification of single glioma cells outside the central

tumor region is not possible with anti-GFAP staining alone.However, IFAP-300kDa staining has identified some GFAP-containing cells adjacent to the tumor mass (unpublishedobservations). Such cells may represent malignant astrocytesthat otherwise cannot be identified and may be responsiblefor the high rate of tumor recurrence following resection ofthe original mass. Furthermore, IFAP-300kDa has not beendetectable in reactive astrocytes induced by stab-wound andcontusion injuries of the central nervous system of the rat(26). Therefore, this IFAP may be clinically useful for as-sessing tumor cells that may reside in the surrounding glioticand largely normal tissue. This feature is under currentinvestigation.

The authors gratefully acknowledge the expertise ofDr. LawrenceW. McDonald in assessing the glioma specimens. Technical assis-tance was provided by Viginia Kriho and Diane Johnson-Seaton.This study was supported in part by National Institutes of HealthGrants NS26395, EY06804, and CA54782.

1. Zulch, K. J. (1979) Histological Typing of Tumours of theCentral Nervous System (WHO, Geneva), p. 66.

2. Nelson, J. S., Tsukada, Y., Schoenfeld, D., Fulling, K., La-marche, J. & Peress, N. (1983) Cancer 52, 550-554.

3. Daumas-Duport, C., Scheithauer, B., O'Fallon, J. & Kelly, P.(1988) Cancer 62, 2152-2165.

4. Russell, D. S. & Rubinstein, L. J. (1989) Pathology ofTumoursof the Nervous System (Arnold, London), 5th Ed., p. 1012.

5. Burger, P. C., Vogel, F. S. & Scheithauer, B. (1991) SurgicalPathology ofthe Nervous System and Its Coverings (ChurchillLivingstone, New York), 3rd Ed., p. 737.

6. Osborn, M. & Weber, K., eds. (1989) Cytoskeletal Proteins inTumor Diagnosis (Cold Spring Harbor Lab Press, Plainview,NY), p. 244.

7. Eng, L. F., Vanderhaeghen, J. J., Bignami, A. & Gerstl, B.(1971) Brain Res. 28, 351-354.

8. Bignami, A., Eng, L. F., Dahl, D. & Uyeda, C. T. (1972) BrainRes. 43, 429-435.

9. Antanitus, D. S., Choi, B. H. & Lapham, L. W. (1976) BrainRes. 103, 613-616.

10. Duffy, P. E., Graf, L. & Rapport, M. M. (1977) J. Neuro-pathol. Exp. Neurol. 36, 645-652.

11. Deck, J. H. N., Eng, L. F., Bigbee, J. & Woodcock, S. M.(1978) Acta Neuropathol. 42, 183-190.

12. Eng, L. F. & Rubinstein, L. J. (1978) J. Histochem. Cytochem.26, 513-522.

13. Goldman, J. E., Schaumburg, H. H. & Norton, W. T. (1978) J.Cell Biol. 78, 426-440.

14. Levitt, P. & Rakic, P. (1980) J. Comp. Neurol. 193, 815-840.15. Westphal, M., Nausch, H. & Herrmann, H.-D. (1990) J.

Neurocytol. 19, 466-477.16. Bongcam-Rudloff, E., Nister, M., Betsholtz, C., Wang, J.-L.,

Stenman, G., Huebner, K., Croce, C. M. & Westermark, B.(1991) Cancer Res. 51, 1553-1560.

17. Yung, W.-K. A., Luna, M. & Borit, A. (1985) J. Neuro-Oncol.3, 35-38.

18. Herpers, M. J. H. M., Ramaekers, F. C. S., Aldeweireldt, J.,Moesker, 0. & Slooff, J. (1986)Acta Neuropathol. 70, 333-339.

19. Dahl, D., Bignami, A., Weber, K. & Osborn, M. (1981a) Exp.Neurol. 73, 4%-506.

20. Schnitzer, J., Franke, W. W. & Schachner, M. (1981) J. CellBiol. 90, 435-447.

21. Schiffer, D., Giordana, M. T., Mighili, A., Giaccone, G.,Pezzotta, S. & Mauro, A. (1986) Brain Res. 374, 110-118.

22. Yang, H.-Y., Lieska, N. & Goldman, R. D. (1990) in Cellularand Molecular Biology ofIntermediate Filaments, eds. Gold-man, R. D. & Steinert, P. M. (Plenum, New York), pp. 371-391.

23. Foisner, R. & Wiche, G. (1991) Curr. Opin. Cell Biol. 3,75-81.24. Lieska, N., Yang, H.-Y. & Goldman, R. D. (1985) J. Cell Biol.

101, 802-813.25. Yang, H.-Y., Lieska, N., Goldman, A. E. & Goldman, R. D.

(1985) J. Cell Biol. 100, 620-631.26. Pappas, G. D., Glick, R. P., Shao, D., Johnson-Seaton, D. &

Yang, H.-Y. (1990) Soc. Neurosci. Abstr. 16, 350.27. McCarthy, K. D. & de Vellis, J. (1980)J. CellBiol. 85,850-902.28. Yang, H.-Y., Lieska, N., Goldman, R. D., Johnson-Seaton, D.

& Pappas, G. D. (1992b) Brain Res. 573, 161-168.29. Zackroff, R. V. & Goldman, R. D. (1979) Proc. Natl. Acad.

Sci. USA 76, 6226-6230.30. Laemmli, U. K. (1970) Nature (London) 227, 680-685.31. Towbin, H., Straehelin, T. & Gordon, J. (1979) Proc. Natl.

Acad. Sci. USA 76, 4350-4354.32. Steinert, P. M. & Roop, D. R. (1988) Annu. Rev. Biochem. 57,

593-625.33. Goldman, R. D. & Steinert, P. M., eds. (1990) Cellular and

Molecular Biology of Intermediate Filaments (Plenum, NewYork), p. 479.

34. Fliegner, K. H. & Liem, R. K. H. (1991) Int. Rev. Cytol. 131,109-167.

35. Albers, K. & Fuchs, E. (1992) Int. Rev. Cytol. 134, 243-279.36. Lazarides, E. (1980) Nature (London) 283, 249-253.37. Dahl, D. (1981) J. Neurosci. Res. 6, 741-748.38. Dahl, D., Rueger, D. C., Bignami, A., Weber, K. & Osborn, M.

(1981b) Eur. J. Cell Biol. 24, 191-1%.39. Houle, J. & Fedoroff, S. (1983) Dev. Brain Res. 9, 189-195.40. Fedoroff, S. (1986) in Astrocytes, eds. Fedoroff, S. & Ver-

nadakis, A. (Academic, New York), Vol. 1, pp. 35-74.41. Lukas, Z., Draber, P., Bucek, J., Draberova, E., Viklicky, V.

& Staskova, Z. (1989) Histochem. J. 21, 693-702.42. McDermott, K. W. G. & Lantos, P. L. (1989) Dev. Brain Res.

45, 169-177.43. Choi, B. H. (1990) in Differentiation and Functions of Glial

Cells, ed. Levi, G. (Wiley-Liss, New York), pp. 9-16.44. Landry, C. F., Ivy, G. 0. & Brown, I. R. (1990) J. Neurosci.

Res. 25, 194-203.45. Abd-El-Basset, E. M., Kalnins, V. I., Subrahmanyan, L.,

Ahmed, I. & Fedoroff, S. (1988a) J. Neurosci. Res. 19, 1-13.46. Abd-El-Basset, E. M., Kalnins, V. I. & Fedoroff, S. (1988b) J.

Neurosci. Res. 21, 226-237.47. Chabot, P. & Vincent, M. (1990) Dev. Brain Res. 54, 195-204.48. Yang, H.-Y., Lieska, N., Goldman, A. E. & Goldman, R. D.

(1992a) Cell Motil. Cytoskel. 22, 185-199.49. Duffy, P. E., Huang, Y.-Y. & Rapport, M. M. (1982) Exp. Cell

Res. 139, 145-157.50. Weinstein, D. E., Shelanski, M. L. & Liem, R. K. H. (1991)J.

Cell Biol. 112, 1205-1213.

Cell Biology: Yang et al.