[CANCERRESEARCH55, 420-427, January 15, 19951

phosphocreatine, Cho and NAA can be detected routinely by ‘H-NMR spectroscopy in vivo in animals and humans (2, 3). Additionalsignals in normal brain, including those of glutamate, inositol, andglucose, can be detected in animal studies or in humans with the latestgeneration of clinical NMR scanners (4). Glutamine, alanine, lactate,and GABA are present in lower amounts in normal brain and aredifficult to detect due to sensitivity and resolution problems but can bedetected with the aid of spectral editing techniques or in pathologicalstates when their concentration is elevated (5, 6). NMR spectra ofextracts obtained at high magnetic field strengths offer much improved sensitivity and resolution compared to spectra recorded in vivaand enable a wide range of hydrogen-containing metabolites to beidentified and quantified. Thus, such extracts can be used to detectfine biochemical differences between various cell lines or types ofcells.

In examining purified populations of different cell types derivedfrom a variety of cellular lineages of the central nervous system, we(7) have found that it is possible to distinguish unambiguously cere

bellar granule neurones, meningeal cells, cortical astrocytes, oligodendrocyte-type-2 astrocyte progenitor cells, and oligodendrocytes byanalysis of their ‘H-NMRspectra. Each cell population studied cxpressed a characteristic distribution and amount of metabolites, including free amino acids and derivatives, compounds involved inmembrane biosynthesis and catabolism, and compounds related toenergy metabolism. Thus, different cell types could be recognized bythe application of ‘H-NMRspectroscopy.

An important question that follows from our observations that

different cell types express characteristic ‘H-NMRspectra relates tothe applicability of this technique to tumor diagnosis, i.e., whetherrelated tumor cells express similar spectra. To investigate this possibiity, we compared ‘H-NMRspectra from six human meningiomacell lines. We found that these spectra were not only similar to eachother but also shared characteristics with spectra obtained from primary rat meningeal cells. Thus, our results suggest that it mighteventually be possible to use NMR for noninvasive diagnosis ofmeningiomas and that cell type-specific metabolites are so integral tothe identity of a cell type that they are conserved across species andthrough transformation.

MATERIALS AND METhODS

Preparation and Analysis of Cell CUltUres

All cells were cultured in DMEM containing 1 gfliter glucose (GIBCOBRL, Paisley, United Kingdom), supplemented with 10% heat-inactivated

FCS (Imperial Laboratories), 2 mM glutamine (Sigma Chemical Co., Poole,United Kingdom), and 25 @g/m1gentamicin (Flow Laboratories, Rickmansworth, United Kingdom; DMEM-FCS). Cells were grown in NUNC tissueculture flasks (GIBCO.BRL) coated with 13 pg/ed poly-L-lysine (Sigma) andmaintained in a 37°Cincubator with humidified atmosphere containing 7.5%

@2The immunocytochemical characterization of all cell cultures was per

formed at each passage by plating cells in parallel onto poly-L-lysine-coatedglass coverslips and growing them in the same conditions as above. Antibodies

and staining methods described in the literature (8, 9) were used with minormodifications. The specificity of immunochemical labeling was tested for allantibodies on mixed cultures of cerebral cortices.

420

Cell Type-specific Fingerprinting of Meningioma and Meningeal Cells by ProtonNuclear Magnetic Resonance Spectroscopy'

Catarina L. Florian,2 Nicholas E. Preece, Kishore K. Bhakoo, Stephen R. Williams, and Mark D. Noble

The Royal College of Surgeons Unit of Biophysics, Institute of Child Health, 30 Guilford Street, London WCIN JEH (C. L F., N. E. P., K. K. B., S. R. W.J; Department ofNeurobiology, Ludwig institute for Cancer Research. Courtauld Building, 91 Riding House Street, London WJP 8BT; and Departments of Biochemistry and Molecular Biology,Anatomy and Developmental Biology, University College London. London (M. D. NJ, United Kingdom

ABSTRACT

We compared the properties of six human meningiomas with normalrat meningeal cells using cell culture techniques, high resolution in vitro‘H-NMR (nuclear magnetic resonance) spectroscopy, and chromatographic analysis. Cell cultures were immunocytochemicauy characterizedat all stages with specific antibodles Quantitative and qualitative metabolite assessments in cell extracts were obtained from ‘H-NMRspectra and

chromatographic analysis. Human meningioma cells expressed a characteristic spectrum of metabolltes including free amino acids, compoundsrelated to membrane phospholipid metabolism, energy metabolites, andother intermediary products. These spectral characteristics, although different in some respects, were strikingly similar to the ones of rat meningeal cells. Particularly, several metabolites that allow discrimination between meningeal cells and other cell types of the central nervous systemwere preserved in meningiomas. These similarities suggest that the regislation of intracellular levels of such metaboiltes is so intrinsic to theidentity of cell type as to be conserved across species and through tramformation. Additionally, human meningioma cultures expressed somespectroscopic characteristics that enabled them to be clearly distinguishedfrom primary rat meningeal cultures. Thus, human meninglomas may beboth specifically recognizable by ‘H-NMRspectroscopy and also distinguishable from normal rat meningeal tissue. Our results raise the eventualpossibility of using NMR In the nonlnvaslve diagnosis of brain tumorsin vivo.

INTRODUCTION

The evolution of tumor classification has been closely correlatedwith the development of methods for identifying the normal cells fromwhich tumors arise. For example, as the cellular specificity of labelingwith various antibodies has become more clearly defined, the use ofhistological criteria for tumor classification has been gradually supplemented and, in some instances, replaced by the use of immunocytochemical analysis. In addition, the increase in our understanding ofthe cellular lineages involved in normal development has led to therecognition of categories of tumors that were previously not appreciated.

One of the tools for cell type recognition that has lately proved ofpotential interest is ‘H-NMR3spectroscopy. NMR analysis of extractsobtained from cultured cells provides a reliable method to assess themetabolites present in these cells (1). A major advantage of NMRspectroscopy is that it is a noninvasive technique that can be appliedin vivo, enabling information to be obtained about the metaboliccomposition of living tissue. Cerebral metabolites such as creatine and

Received 8/11/94; accepted 11/11/94.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 in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

I This work was supported by the Cancer Research Campaign U.K. and the Welcome

Trust.2 To whom requests for reprints should be addressed.

3 The abbreviations used are: NMR, nuclear magentic resonance; Cho, choline-con

taming compounds; NAA, N-acetyl-aspartate; GABA, @y-aminobutyricacid; GFAP, glialfibrillary acidic protein; GalC, galactocerebroside; PCA, perchloric acid; TSP, 3-trimethylsilyl-tetradeuterosodium propionate; 2D-COSY, homonuclear proton two-dimensionalcorrelated spectroscopy; PFG, pulsed field gradients; NAAG, N-acetyl-aspartyl-glutamate; PC, phosphoryl choline; GPC, glycerophosphoryl choline; OPA, o-phthaldialdehyde; BCA, bicinchoninic acid; PK, pyruvate kinase.

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

METABOUC PROFILES OF MENINGES AND MENINGIOMA CELLS

Meningioma Cells. Six human meningioma cell lines, of which one wasderived from a recurrent meningioma, were cultured. One of the cell linesoriginated from a spinal meningioma. The composition of cultures was evaluated by immunolabeling cells on coverslips with antibodies at 48—72h afterplating onto coverslips (initial density of 500 cells/coverslip). Cells were testedfor the presence of cytoplasmic intermediate filaments, such as glial fibrillaryacidic protein (jrotein specific to astrocytes in mature brain; Refs. 10—12) and

vimentin (ii, 13), and also for fibronectin (12). Additionally, contamination byother cell types of the nervous system was assessed by testing with monoclonalantibody A2B5 (a marker characteristic of neurones and cells of the oligodendrocyte-type-2 astrocyte lineage (14); hybridoma supernatant, diluted 1:2), oranti-GalC(a specific surface marker for oligodendrocytes; Ref. 15); hybridoma

supernatant, diluted 1:10).

Primary Meningeal Cells. Cell cultures were prepared as described previously (9). Leptomeninges of 1-day-old Wistar rats were removed and dissociated with collagenase and trypsin and then plated onto 175-cm2flasks andgrown in DMEM-FCS. The cells formed a confluent monolayer within 5—6days after plating, when they were passaged and replated at 1:4 dilution into175-cm2flasks. Cells were plated in parallel onto coverslips (density of 3000cells/coverslip). When meningeal cells formed a confluent monolayer, theywere harvestedfromcultureflasks. The meningealcultureswere assessed forpurity by immunocytochemical labeling on coverslips with anti-Ran-2 (amonoclonal antibody which recognizes an antigen present among others on thesurface of astrocytes and leptomeningeal cells; Ref. 16). All antibodies used inthe characterization of human meningioma cells were also used for ratmeningeal cells.

Cell Harvesting, Extraction, and Preparation of Samples

All cells were continuously cultured until the amount necessary for high

resolution NMR spectroscopy was obtained (typically iO@-i0' cells for asample). They were harvested at 95—100% confluence and always 24 h after a

final medium change. Cells were detached from culture flasks by a 3-min rinsewith Ca2@/Mg@@-free DMEM (CMF-DMEM) containing 034 mM EDTA(Sigma), followed by 3—5mm incubation with a solution of 300 units/mItrypsin (Sigma; bovine pancreas, type III) in CMF-DMEM. One ml soybean

trypsin inhibitor-DNase (5200 units/mI soybean trypsin inhibitor, 74 units/mi

bovine pancreas DNase I, and 3 mglml BSA; Sigma) was added per 10 ml cellsuspension to prevent the cells from clumping and to stop the trypsinizationprocess. Cell suspensions were centrifuged at 500 X g for 5 mm, followed bya 30-s spin at 700 x g. Cells were washed three times with 30 ml PBS, and thepellet finally was extracted with 2 ml ice-cold PCA (12% v/v; Merck, Ltd.).PCAextractswere sonicated(2 times30 s, 4°C;Soniprep150, MSEScientificInstruments, Crawley, United Kingdom) and then centrifuged at 3000 X g at

4°Cfor 20 mm.The pelletswere keptat —20°Cforproteinquantification,andthe supernatantswere furthertreated.At least two PCAextractswere preparedfor each humanmeningiomacell line, and four extractswere obtainedfromprimary cultures of rat meningeal cells (replicate samples).

The pH of the supernatantwas adjustedto slightly alkaline(pH > 7.5) andthen mixed with 5 mg Chelex-100 (Bio-Rad Laboratories, Richmond, CA) per0.2 ml sample in order to remove multivalent ion contaminants. A subsequentadjustment of pH to 8.5—8.9with 3 M KOH was followed by removal of the

precipitated KC104by centrifugation at 3000 X g for 5 mm. The sample waslyophilized and redissolved in 0.5—0.7ml D2O (Goss Scientific Instruments,Ltd., Ingatestone, United Kingdom). As internal concentration and chemicalshift standard TSP (Goss Scientific Instruments, Ltd.) was added (10 @lof 10mM TSP in D20). Finally, the pH of the sample was adjusted to 8.9 with DC1or NaOD (Aldrich, Gihingham, United Kingdom).

‘H-NMRSpectroscopy

All spectra were obtained at 26°C-30°Con a Varian Unity-plus NMRspectrometer (Varian Associates, Inc., NMR instruments, Palo Alto, CA)

operating at a proton frequency ofSOOMHz. Single-pulse spectra (approachingfull relaxation) were acquired with 45°pulses applied every 5 s. The samplespinning rate was 20 Hz, spectral width was 5000 Hz, and data size was 32Kpoints. A second radiofrequency pulse was applied at the frequency of the

water peak in the delay between acquisitions in order to suppress the residualwater (HOD) signal. For a satisfactory signal:noise ratio, 512 or 1024 scans

were accumulated and Fourier transformed. A line broadening of 0.2—0.3Hzwas applied for spectral processing.

2D-COSY was also used for identification of coupled resonances. Absolute-value and, where possible, phase-sensitive conventional COSY anddouble quantum filtered COSY spectra were acquired, either using standardphase cycling or with the application of PFG (20 0/cm; duration 2 ms, riseand fall time of 100 @i.s)to select coherence pathways (17). Typically, 192transients were collected for each of 256 T, increments (repetition time, 1.2s; spectral width, 5000 Hz). Water suppression was achieved by presatu

ration of the HOD peak between acquisition periods as in the one-dimensional spectra. Data were collected into 2K data points in F2, and 1K in Fl,which was then zero-filled to 2K. Sine-bell and Gaussian weighting functions were applied during the double Fourier transform, and the COSYplots were symmetrized after initial analysis.

Metaboilte Identification and Quantification. Metabolite peaks found inthe one-dimensional spectra were identified by: (a) their chemical shift andcoupling pattern as described in the literature (1, 18); (b) comparison withspectra of metabolites in known concentrations obtained at the same pH andspectroscopic conditions; and (c) two-dimensional spectroscopic methods.

The following metabolites were identified from their strongest and bestresolved resonances: g3-hydroxybutyrate.@yCH@1.19 ppm (doublet); threonine--yCH31.30ppm (doublet); lactate .CH@1.34 ppm (doublet); alanine @CH@1.47ppm (doublet); acetate -CH3 1.92 ppm (singlet); NAA -NCOCH3 2.02 ppm(singlet); NAAG -NCOCH3 2.05 ppm (singlet); glutamate -yCH2 2.34 ppm(triplet); succinate -CH2 2.41 ppm (singlet); glutamine --yCH2 2.44 ppm(triplet); aspartate -CH2 2.56 and 2.75 ppm (double doublet); creatine -NCH33.04 ppm (singlet); taurine -SCH2 3.08 ppm (triplet) and -NCH2 3.42 ppm(triplet); choline -N(CH3)33.21 ppm (singlet); PC -N(CH3)33.22 ppm (singlet); GPC -N(CH3)3 3.23 ppm (singlet); glycine -CH2 3.56 ppm (singlet); andinositol(H2) 4.05 ppm (triplet). Metabolite amounts were calculated from theirintensities in 1H-NMRspectra by reference to the internal standard TSP afterbaseline correction. The intensity of a given signal in the proton spectrum isproportional to the amount of compound and number of protons contributingto that signal.

HPLC Analysis of Metabolites

Quantitative determination of amino acids, NAA, and NAAG was performed by HPLC to complement the spectroscopic data. Amino acids wereanalyzed by derivatization with OPA (19), with the concentrations beingcalculated from external standards. A LiChroGraph HPLC system (Merck

Hitachi, Darmstadt, Germany) equipped with a LiChrospher 100HP-18 separation column (5 @m;4 mm internal diameter, 25-cm length) was used. TheOPA derivatives were detected with a UV/VIS fluorometer(LiChroGraphF1050), characterized by wavelengths of 340 nm for excitation and 450 nm foremission. Aliquots taken from NMR samples were diluted 5—200-foldwithultrapure water, and a 150—250-idsample was mixed with 350 p.1reagentcontaining 0.8 mg/ml OPA (Sigma). The compounds were allowed to react for3 mm and then 150 pi mixture was injected onto the column. Elution was

achieved at 40°Cwith a three-step linear gradient of methanol in phosphatebuffer (50 mM in ultrapure water, pH 6) starting at 20% methanol, increasingto 35% methanol in 5 min, then decreasing to 30% methanol in 30 min, andfinishing at 35 mm with 20% methanol.

NAA and NAAG were detected with an HPLC System Gold (BeckmanInstruments, San Ramon, CA) equipped with a UV detector (wavelength, 210nn@),and their concentrations were calculated from external standards. NMRsamples were diluted 1:10 with 5 mMH2SO4,and 50-id aliquots were injectedonto the column. NAA was analyzed by a method described previously (7)using the same conditions. Detection for NAAG was carried out according toKoller et al. (20) using an Machem SAX (46 mm internal diameter, 25-cmlength) column fitted with a 1-cm guard column.

Protein Quantification

Protein quantification was performed on the final PCA precipitate bycomplexing with BCA (21). The pellet was dissolved in 1 ml of 1 mM KOHby incubation at 37°Cfor 30 min. BCA reagent (BCA kit, Pierce, Rockford,CA) was furtheradded, and the absorbancewas detected after incubation(wavelength, 256 sum; UV/VIS Spectrophotometer Ultrospecll), and protein

was calculated by reference to BSA standards (BCA kit, Pierce).421

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 1 Immisnocytochemicalcharacteristics of the human meningioma cell linesResults are calculated as a percentage of positive cells from the total number of cells. At least 200 cells were counted on each coverslip. At every passage (P), at least two coverslips

were prepared for each antibody, or combination of antibodies, for each cell line. Tests were carried out in the following combinations: A2B5 and aGalC; aFN; aGFAP; aVim. Figuresare given for the start and for the end passage number, where these results differ for the two passagenumbers.Passage

numberHuman

meningioma cell line Start End A2B5 (%) aGalC' (%) aFN (%) aGFAP (%) aVim(%)IN1O67

P6 P9 0 0 29.0±8.7 (P6) 0 10082.0 ±8.9(P9)lNlO69

P3 P6 0 0 26.7±3.4 (P3) 0 10058.6 ±2.6(P6)IN1O81

(recurrent) P3 P6 0 0 26.0 ±4.2 (P3) 0 10078.3±3.3(P6)IN11I4

PS P8 0 0 43.1±5.1(P5) 0 10080.1±8.5(P8)IN1123

P3 P7 0 0 32.2±6.4 (P3) 0.5 (P3) 10067.0±5.6 (P7) 2 ±03(P7)IN1239

(spinal) P3 P7 0 0 31.5 ±6.4 (P3) 0 10079.5 ±14.9(P7)a

aGalC, anti-galactocerebroside; aFN, anti-fibronectin; aGFAP, anti-glial fibrillary acidic protein; aVim, anti-vimentin.

METABOLIC PROFILES OF MENJNGES AND MENINGIOMACELLS

Statistics

Results are presented as mean ±SD. Immunocytochemical results areexpressed as a percentage of positive cells from the total number of cells. Theamounts of metabolites were referenced to the amount of protein in thesamples, as nmol/mg protein. The results from replicate samples of 2—6extracts for each of the six human meningioma cell lines were averaged per

each cell line. Data obtained per cell line were finally averaged per meningioma cells. Data from each of the four independent observations for meningealcells were also averaged. Two sample Student's t tests were performed todetermine significant differences among group means, and Ps were quoted

without correction for multiple comparisons. The critical values for assessingsignificance levels were obtained after performing a Bonferroni correction formultiple comparisons (22) independently for NMR and HPLC data sets (12and 13 comparisons, respectively).These critical values were P < 0.005 for 5%significance level (i.e., 0.05/12 or 0.05/13), and P < 0.001 for 1% significancelevel (analogous).

RESULTS

Immunocytochemical and Cell Morphologj Characterization.All four preparations of primary rat meningeal cells contained flat,bipolar or tripolar A2B5 and Ga1C cells. Positive immunocytochemical labeling was obtained for fibronectin and vimentin. Therewas over 95% positive labeling with Ran-2 antibody, showing thepurity of meningeal cultures relative to contamination with cells of thenervous system. There were some GFAP@ astrocytes (4 ± 1%)scattered within the meningeal cultures.

Consistent and reproducible results in the expression of specificantibodies were obtained in the six human meningioma cell lines(Table 1). The specificity of immunocytochemical labeling was confirmed by negative results in all controls. Neither anti-Ga1C nor themonoclonal antibody A2B5 (which recognizes specific surface gangliosides) labeled any of the cells of the six human meningioma linesat either early or late passage. Fibronectin, an extracellular matrixglycoprotein with adhesive and opsonic properties (12), was widelyexpressed by all six meningioma cell lines in all passages, the proportion of fibronectin@ cells in culture increasing with the passagenumber. The expression of GFAP was either negligible in one cell lineor absent (in the other five cell lines). Vimentin was expressed by allsix meningiomas (100% vimentin@ cells), irrespective of passagenumber or stage in culture.

Metabolites Identified in ‘H-NMR Spectra and Comparisonwith HPLC Results. Representative high-resolution 1H-NMRspectra obtained from extracts of acid-soluble metabolites from

cultured rat meningeal and human meningioma cells are shown inFig. 1. These high-field regions of spectra (0.5—4.5ppm) containsignals from a variety of metabolites and present qualitative similarities and also qualitative and quantitative differences. Metabolites detected and quantified include amino acids and related compounds; substances involved in membrane biosynthesis such ascholine, Cho, and inositol; and intermediary metabolites such assuccinate and lactate. Metabolite contents (nmol/mg protein) of thecultures examined are given in Tables 2 and 3. In Table 4, selectedmetabolite ratios are presented for rat meningeal and human meningioma cells. These ratios were chosen because of their relevance either to NMR in vivo (involving creatine, Cho, and inositol)or to previous work (7) identifying metabolite ratios which coulddistinguish between preparations of rat meningeal cells and otherprimary cultures from rat brain (involving Cho, glutamate, aspartate, and creatine). Not all identifiable compounds in the ‘H-NMRspectra could be accurately quantified because of incomplete resolution in certain zones of the spectra (e.g., isoleucine, leucine, andvaline with signals between 0.9—1.05 ppm). In addition, signalsfrom some of the identifiable compounds were often too low foraccurate quantification.

Within each category of cells (rat normal and human tumor),spectra were qualitatively identical (i.e., all identifiable peaks werepresent in all spectra from that category), with the exception ofcreatine, which was not detectable in spectra from one-half of themeningioma cell lines (see below). Quantitatively, the spectra fromthe meningioma cells lines displayed more variability between different cell lines (Table 2), but excellent reproducibility was achieved inreplicate samples (i.e., preparations obtained from one cell line).There was very good reproducibility in replicate preparations fromnormal rat meninges. Good agreement was obtained between quantification from NMR spectra and by HPLC analysis in all samples. Thetaurine content of human meningioma cell lines could not be reliablyquantified in the NMR spectra, although it was present at levels

detectable by HPLC. The greatest variability in absolute metaboliteamounts was displayed by glutamate in human meningiomas, but the

data for glutamate from NMR spectra and by HPLC analysis were ingood agreement for individual cell lines.

The identification of coupled metabolites in single-pulse spectrawas confirmed and supplemented by identification of crosspeaks in2D-COSY spectra. Fig. 2 displays a 2D-PFG-COSY spectrumfromone of the meningioma cell lines.

422

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table2 Comparativecompositionof metabolitesquantsfledfrom‘H-NMRspectraofratmeningeal and human meningiomacellsMetabolite

concentrations(nmol/mgprotein)are expressedas mean±SD.Spectraobtainedfrom replicate preparations for each of the independent observations forhumanmeningioma

cell lines and rat meningeal cells, respectively, were analyzed byreferencetoTSP.P obtainedfromtwosampleStudent'st testaregivenwithoutcorrectionforthenumber

ofcomparisons. The results for normal cells were considered statisticallydifferentfromthe results for tumor cells if P < 0.05/12 (5% level) or P < 0.01/12 (1%level).Rat

HumanMetabolitesmeningeal cells meningioma celllines(nmol/mg

protein) (n = 4) (n = 6)PAlanine

15.6 ±3.8 16.3 ±3.60.77313.9±2.827.1±5.10.004Choline

2.0 ±0.4 6.2 ±4.80.124GPC7.5 ±2.2 9.1 ±4.50.526PC4.4 ±0.7 14.0 ±3.6<0.001Creatine

17.6 ±4.8 7.3 ±5.80.003Glutamate114.1 ±8.0 298.4 ±43.9<0.001Glycine

12.9 ±5.8 27.5 ±10.10.033@3-Hydroxybutyrste6.6 ±1.4 13.4 ±8.00.140Inositol5.4 ±1.9 18.5 ±14.20.116Succinate

12.1 ±5.2 13.1 ±5.40.771Threonine12.7 ±3.5 23.8 ±5.70.009Taurine33.6 ±5.5 Not measurable <0.001

Table 3 Comparative composition of metabolites determined by HPLC analyses inratmeningea!and hwnan meningiomacellsMetabolite

concentrations (nmol/mg protein) are expressed as mean ±SD oftheindependentobservations for the rat meningeal cells preparations and for human menin

gioma cell lines. P obtained from two sample Student's I test are given withoutcorrectionforthe number of comparisons. The results for normal cells were consideredstatisticallydifferent

from the results for tumor cells if P < 0.05/13 (5% level) or P < 0.01/13(1%level).Metabolites@'

i@t meningeal cells Human meningioma celllines(nmol/mgprotein) (n = 4) (n = 6) P

METABOUC PROFILES OF MENINGES AND MENINGIOMA CELLS

RMm@iIn@@ (a)

Human msnlnØoma (a)

4.1 4.0 3.5 3.0 2.5 2.0 1.8 1.0 0.5ppm

4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5ppm

Fig. 1. Representative single-pulse high-field high resolution ‘H-NMRspectra fromrat meningeal and human meningioma cells. Spectra were obtained from PCA extractsof primary cultures of rat meningeal cells (a) and cultures of human meningioma celllines (b and c) NMR spectroscopic analysis was performed at pH 8.9, with 512—1024scans recorded on a spectrometer operating at the proton frequency of 500 MHz.Typically, iO7@iO8cells were obtained for one extract. The content in protein of samplesranged from 5.1 to 8.2 mg for rat meningeal cell preparations and from 0.92 to 5.3 mgfor human meningioma cell extracts. Spectra referenced to TSP (0 ppm) from thefollowing cell lines are displayed: b = IN1067; c = 1N1114.The peak marked () wasanexogeneouscontaminant,probablyfromtheplasticwareusedinsamplepreparationsIn spectra a and b, the height of the [ac doublet was truncated to 2/3 of the originalheight in the spectrum. tie, isoleucine; Leu, leucine; Val, valine; (3-HB. @-hydroxybutyrate; Thr, threonine; Lac, lactate; Ala, alanine; Ace, acetate; Glu, glutamate; Suc,succinate; GIn, glutamine; Asp, aspartate; Cr. creatine; Thu. taurine; 0,, choline; PC,phosphorylcholine;GPC,glycerophosphorylcholine;Cho,choline-containingcompounds; Gly, glycine; mo, inositol.

cho

Ac,

ac

Lac

4.5 4.0 3@5 3.0@ 2.0 1.5 1.0 pp@

Cho

aThedataforcholine-containingcompounds(Qio,3.2ppm)arethesumofthreeindividual peaks: choline (3.21 ppm), PC (3.22 ppm), and GPC (3.23 ppm).

Alanine 17.8 ±3.8 25.2 ±7.1 0.264Arginine 2.0 ±0.9 6.1 ±2.3 0.026Asparagine 1.4 ±1.0 4.9 ±1.9 0.018Aspartate 20.3 ±0.4 27.0 ±14.6 0.404GABA 5.9 ±0.8 0.6 ±0.2 <0.001Glutamine 43.6 ±5.6 72.9 ±32.6 0.145Glutamate 105.4 ±2.6 305.5 ±47.4 <0.001Hypotaurine 19.4 ±2.9 5.0 ±3.0 <0.001NAA Not detected Not detected―NAAG 3.0 ±1.5 4.5 ±3.9 0.474Serine 18.0 ±1.9 28.6 ±5.9 0.018Taurine 38.4 ±4.2 18.0 ±2.8 <0.001Tyrosine 2.3 ±0.6 10.3 ±2.1 <0.001

a Theamountof NAAwastoo lowto be detectedby HPLCinvestigationsin fiveofthe six meningioma cell lines we examined, but one meningioma cell line (IN1239)contained a very low amount of NAA (2.9 ±1.7 nmol/mg protein).

Common Features of the Spectra of Rat Meningeal and HumanMeningioma Cell EXtraCts. In the light of existing data on metabolites detectable by 1H-NMR in whole brain extracts and cellpreparations from either rat or human brain (5, 7, 23), we note thatspectra from both rat meningeal cells and human meningioma celllines were characterized by either very low or not detectablesignals for NAA (2.02 ppm, singlet), NAAG (2.05 ppm, singlet),aspartate (2.56 and 2.75 ppm, double doublet), and the neuroactive

amino acid GABA (3.0 ppm, triplet). NAA was not detectable byHPLC (lower detection limit, 0.1 nmol) in any of the samplesexamined, with the exception of one meningioma cell line thatcontained a small amount of NAA. Low NAAG concentrationswere also determined by HPLC in preparations from both ratmeninges and human meningiomas, with no significant differencesbetween the two categories. Compounds such as tyrosine (3.05,3.15, and 3.93 ppm, all double doublets) and hypotaurine (2.65 and3.30 ppm, both triplets) were present in too low an amount to bedetected by NMR but were detectable by chromatographic analysis. Complementary determinations by HPLC showed statistically

5@.mmmsn0,@a (C)

v@LaoI.

as,

423

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 4 Metabolite rajios in rat meningeal and human meningioma cellsMetabolite ratios (mean ±SD) were calculated from metabolite concentrations quasi

rifled from the ‘H-NMRspectra of acid extracts of rat meningeal and human meningiomacells. P values obtained from Two-sample Student's t test are given without correction forthe number of comparisons. The results for normal cells were considered statisticallydifferent from the results for tumour cells if P < 0.05/12 (5% level) or P < 0.01/12 (1%level) Abbreviations Cho choline-containing compounds.

Metabolite ratiosRatmeningeal cells

(n = 4)Humanmeningioma

cell lines (n =6)PCho/creatine1.12

±0.185.5 ±2.90.019creatine/inositol3.2±0.90.63 ±0.43<0.001aspartate/creatine1.2±0.35.6 ±3.30.024Cho/aspartate0.7±0.11.3 ±0.70.120Cho/glutamate0.12±0.020.1 ±0.050.217Cho/glycine1.2±0.41.7 ±1.60.277glycine/creatine0.78±0.235.9 ±4.60.063

METABOUC PROFILES OF MENINGES AND MENINGIOMA CELLS

of these metabolites between the preparations of rat meningeal andhuman meningioma cells. There was a large variation in the glutamineconcentration of tumor cells, ranging from 29.3 nmol/mg protein forone cell line up to 105.0 nmollmg protein for another, but the averaged glutamine content in human meningiomas was not statisticallysignificantly different from the value for rat meningeal cells.

The methine region of the spectrum (3.2—4.0ppm) had somecommon general characteristics for both types of cells examined,including a relatively high signal for glycine and relatively low signalfor inositol. No significant difference in the inositol content determined from the 1H-NMR spectra was obtained between normal andtransformed cells. We did not have a reliable independent assessmentfor glycine and threonine content in samples because their separationon the HPLC column was poor. Allowing for correction of the Ps forthe number of comparisons, the difference in threonine and glycinecontents of rat meningeal and human meningioma cells was notsignificant at the 5% level.

Differential Characteristics in Spectra of Normal Rat Menhigesand Human Meningiomas. There were also differences in the1H-NMR spectra obtained from cell extracts of cultured rat meningesand human meningiomas. For example, a prominent compound inpreparations of meningeal cells obtained from rat was taurine, ametabolite which was either not detected by NMR in meningiomaextracts, or which gave very low signals in spectra. Quantification oftaurine in spectra obtained from extracts of human transformed cellswas not possible due to overlap of the creatine signal at 3.04 ppm withone of the taurine peaks (3.08 ppm, triplet) and because of thecomplexity of the ‘Hspectrum where the second taurine signal lies(3.42 ppm, triplet). Taurine was among the most abundant free aminoacid detected by HPLC (Table 3) in rat meningeal cells and waspresent in amounts significantly different (P < 0.001) from thoseexistent in human meningiomas.

Among other signals in the high field region, the singlets at 3.04ppm and 3.94 ppm, which are assigned to creatine, were more prom

significant differences at the 1% level in the concentration ofamino acids such as GABA, hypotaurine, and tyrosine in extractsof rat meningeal cells compared to human meningioma cells.

Spectra of preparations from rat meninges and human meningiomacell lines had common features in the 1.0—1.5ppm region, displayingthe doublet signals of f3-hydroxybutyrate, threonine, lactate, andalanine. The concentrations of none of these metabolites were significantly different between rat normal and human tumor cells. Relativeto ‘H-NMRspectra in vivo, lactate signals were elevated in the spectraobtained from cultured cells, probably due to anaerobic glycolysistaking place in cells during the extraction process. Lactate content washighly variable in the cell preparations (from 25 to 230 nmol/mgprotein), depending on the rapidity of cell extraction and on thedimension of the cell pellet. The signals for alanine were relativelyhigh in spectra from preparations of both categories of cells comparedto the typical levels of alanine detected by spectroscopy in vivo innormal human brain (5, 23).

In the methylene region of the spectra (2—3ppm), common signalsfor both types of cells included those of glutamine and succinate. Nostatistically significant differences were obtained in the concentrations

Fig. 2. Two-dimensional (pulsed-field gradient) correlated ‘H-NMRspectrum of a human memngioma cell line.The 2D-PFG-COSY spectrum was obtained from a PCAextract of a human meningioma cell line (1N1067) on thesame spectrometer as above but using the application ofpulsed field gradients to select coherence pathways. Foreach of the 256 T, increments 192 transients were collected.The inset shows the whole 2D ‘H-NMRspectrum (0—l0ppm), while the main figure represents an expansion ofthe spectral region of interest (0.5—4.5ppm). The coupledpeaks with chemical shifts of 1.7 and 3.05 ppm were tentatively assigned to polyamines (PA) by running singlepluse ‘H-NMRspectra of putrescine and spermidine and bycomparison of HPLC profiles ofsingle polyamines with theHPLC profiles obtained from human meningioma cell lineextracts. Abbreviations same as for Fig 1; @-A1a,f3-alanine.

n

424

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

METABOLIC PROFILES OF MENINGES AND MENINGIOMA CELLS

inent in ‘H-NMRspectra from normal rat meninges than in spectrafrom human meningioma cell lines. The creatine peak in vitro represents the sum of creatine and phosphocreatine present in the livingcells, since hydrolysis of phosphocreatine to creatine occurs duringthe cell harvesting and extraction process. Spectra from preparationsof three meningioma cell lines were characterized by the absence ofcreatine peaks; in two other cell lines, the creatine levels were belowthe lowest amount in any of the meningeal preparations; in one cellline, the creatine level appeared relatively “normal―(16.9 nmol/mgprotein). However, overall there was a significant difference(P = 0.002) in the creatine content of rat normal and human transformed cells. The creatine:inositol metabolite ratio was also significantly different at the 1% level in meningeal compared with meningioma cells (Tables 2 and 4).

The signals for glutamate were relatively higher in spectra ofhuman tumor cells than in those of normal rat cells. This was alsoreflected, by HPLC, in the significantly elevated (P < 0.001) contentof glutamate in human meningioma lines (305.5 ±47.4 nmol/mgprotein) versus normal rat meningeal cells (105.4 ±2.6 nmol/mgprotein), and related to this, in a significantly different (P < 0.001)glutamate:alanine ratio.

Peaks from Cho were prominent in spectra of both normal andtumor cells. The signals in the 3.2—3.3ppm region are assigned tocholine itself (3.21 ppm, singlet), PC (3.23 ppm, singlet), and GPC(3.24 ppm, singlet); these metabolites were identifiable in thespectra of both categories of cells. Other metabolites, includingcarnitine, ethanolamine, and phosphoethanolamine might also havecontributed to signals in this region. A significant difference at the5% level (P = 0.004) was detected in total Cho (sum of the threepeaks) in rat meningeal compared to human meningioma cellextracts (Table 2).

The relative heights of individual peaks contributing to the Chosignal in the two categories of cells were also different. There was asignificant increase (P < 0.001) in the amount of PC in humanmeningiomas compared to rat meninges. A shift was observed in thePC:GPCratio from below one in rat meningeal cells to greaterthanone in the human meningioma cell lines. PC is synthesized in vivo in thefirst step of phospholipid biosynthesis, whereas GPC is a phosphodiester

breakdown product of phospholipids. The predominance or otherwise ofone kind of phosphoesters in the composition of the Cho peak in thesespectra may indicate differences in phospholipid metabolism in meningioma cells of human origin versus meningeal cells from rat.

DISCUSSION

We used cell culture techniques and investigations by ‘H-NMRspectroscopy complemented with HPLC analysis in order to comparemetabolite profiles of normal rat meningeal cells and human meningioma.

Meningioma Cell Lines in the Light of Their Immunocytochemlcal Characteristics and Cell Types of the Meninges. All cell typesof meningeal tissues covering the brain and spinal cord originate inthe primitive meninx derived from the layer of mesenchymal tissuethat comes to surround the neural tube in the early development of theembryo (24). Meningiomas, as primary tumors of the meninges composed of or differentiating towards arachnoidal cells (25), can involvedura mater, leptomeninges, or both and account for about 15% of allhuman brain tumors. Tumors of the meninges display a high phenotypical variability in cell products, shape, pattern, and stroma, butequally, human meningiomas of various origins have similarities intheir immunocytochemical characteristics, histological features, metabolism, behavior, etc. (11, 13, 25). The results we obtained were in

agreement with these latter observations of commonalities amongmeningiomas.

Our findings related to fibronectin and vimentin expression ofmeningioma cell lines were in concordance with previous studies on

human intracranial meningiomas. It has been shown (12) that neoplastic cells retain the intermediate filaments of their tissue of originand that virtually all meningeal neoplasms have a uniform distributionof fibronectin filaments. Additionally, coexpression of vimentin anddesmosomal proteins and the association of this type of intermediatefilaments to desmosomal plaque proteins are a combination unique tomeningioma cells (13). The negative GFAP labeling obtained in themeningioma cells was in agreement with their cell of origin and their

usual localization outside brain tissue (10, 24).Interpretation of HPLC ReSUlts and NMR Findings. Although

there were quantitative variations for certain metabolites between thehuman meningioma cell lines, qualitatively identical ‘H-NMRspectrawere obtained from all six cell lines we examined. This fact suggestsa preservation of the characteristic metabolite profile within the typeof tumor, irrespective of its origin.

Spectra from preparations of rat meningeal cells displayed obvioussimilarities to those we have reported previously from extracts ofprimary cultures of rat meningeal cells (7). These results confirm thereproducibility of cell culture techniques, metabolite extraction pro

cedures, and NMR spectroscopy investigations. Although cell cultureprocedures and methods of extract preparation could be a source ofvariability (1,26), in studies such as ours, these potential problems can

be readily overcome if factors including composition of the culturemedium, degree of cell culture confluence, time of harvesting, andmethod of extraction are carefully controlled.

We have shown previously that rat meningeal cells can be distinguished from other primary cultures of central nervous system ratcells (neurones, astrocytes, oligodendrocyte-type-2-astrocyte progenitors, and oligodendrocytes) on the basis of ‘H-NMRspectra ofextracts (7). The distinguishing features were the presence of highamounts of succinate and @-hydroxybutyrate, the absence of NAA andhypotaurine, and relatively low amounts (compared to oligodendrocytes) of creatine. These characteristics were shared by all the humanmeningiomas we examined, with no significant differences in thesuccinate and @3-hydroxybutyrate concentrations between rat meningeal cells and human meningioma cell lines, and an even lowercreatine content in the meningioma preparations. In addition, themetabolite ratios Cho:aspartate and Cho:glutamate, which were usedpreviously to discriminate rat meningeal cells from rat astrocytes, donot differ significantly between rat meningeal and human meningiomacells (Table 4). The aspartate:creatine ratio, which was more than10-fold higher in rat meningeal cells compared to cortical rat astrocytes, was even higher in human meningioma cell lines, reflectingtheir lower creatine content.

Although the key features described above clearly relate humanmeningioma cells to their tissue of origin even when consideredacross these widely separated species, there were also a number ofcharacteristics which discriminated the transformed from normalcells. These statistically significant features of the human meningiomametabolite profiles were reductions in creatine and taurine and increases in Cho (attributable mainly to increased PC) and glutamate.

Taurine is known to be much lower in human brain than in rat brain(5, 27); therefore, it is possible that the reduced taurine in the meningioma cell lines reflected a species difference rather than an effect oftransformation. Although it is not yet known whether the dissimilarities in other signals reflect differences across species or differencesbetween normal and transformed cells, arguments can be raised that at

least some of these changes may be associated with transformation.A feature of human meningioma cell lines that we examined was

425

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

METABOUC PROFILES OF MENINGES AND MENINGIOMA CELLS

their high content of glutamate, consistently 2.5- to 3-fold higher thanrat meningeal cells. As previous studies (24) have demonstrated thathuman meningiomas in culture express a relatively low uptake ofglutamate (lower even than normal human fibroblast cultures), thequestion arises as to how the high glutamate content in meningiomacells might be achieved. A variety of observations raise the possibilitythat such an elevation would be a predicted consequence, both ofenzymatic alterations characteristic of brain tumors and of the highalanine content of primary meningeal cells (7). We propose thefollowing hypothesis to explain the high glutamate levels observed inour human meningioma cultures.

One of the enzymes involved in the glycolytic pathway in eukaryotic cells is PK, which catalyses the transfer of phosphate fromphosphoenolpyruvate to ADP, yielding free pyruvate and ATP. PK is

“turnedoff―when Al?, or other fuels including alanine, are presentin high concentration in the cell, and it “turnson―when there is abuildup of the preceding glycolytic intermediates, especially phosphoenolpyruvate (28). It has been shown (24) that this enzyme is

present both in normal brain and brain tumors mostly as the muscletype (M), in the form of two subgroups: Ml, which is not inhibited byalanine in its action; and M2, which is inhibited by alanine. The Mlform is present in normal brain, while the M2 form is predominant intumors. These observations, considered in the light of the results ofour analyses of amino acids, especially alanine and glutamate, in ratnormal and human tumor meningeal cells, raise the possibility thatthere might be a relationship between the relatively high alaninecontent of both meninges and meningiomas and PK in these cells. Itmight be that in normal meningeal cells, PK (the Ml form which isnot inhibited by alanine) will be active, while in meningioma cells, theM2 form of PK which is inhibited by alanine is “turnedoff.―Such aninhibition of PK would lead to a lack of production of pyruvate in the

glycolytic process and accumulation of phosphoenolpyruvate. Alternatively, pyruvate could be generated in the tumor cells via thetransamination reaction from alanine and a-ketoglutarate, yielding

glutamate in addition to pyruvate. The presence of this reaction would

account for the buildup of glutamate in meningioma cells withoutuptake from the culture medium. This reversible transamination process could also replenish alanine. Our results showing that human

meningiomas display a constant glutamate:alanine ratio, which is also

significantly different from the ratio for rat meningeal cells, areconsistent with this suggestion. Subsequently, when phosphoenolpyruvate has accumulated in meningioma cells in high amounts, it will

“turnon―the PK (M2 form), and glycolysis will follow its normalpath. This hypothesis is also consistent with observations that normalallosteric factors regulating the glycolytic rate to that of the tricarboxylic acid cycle are defective or altered in transformed cells (28).

Our results showed that the PC peak was predominant amongstsignals contributing to Cho in spectra from all six human meningiomacell lines we examined, as opposed to a predominant GPC peak in the3.2 ppm region of the ‘H-NMRspectra obtained from rat meningealcells. This finding was consistent with the idea that increased levels ofphosphomonoesters (e.g., PC and phosphoethanolamine) may be as

sociated with intensified cell membrane synthesis and increased rateof cell replication and have been, therefore, proposed as an indicatorof cell proliferation (29).

Our findings showing raised amounts of Cho in preparations fromhuman meningiomas compared to extracts from rat meningeal cellswere also in agreement with the outcome of studies performed onhuman meningiomas in vivo and on human biopsy extracts. Clinicalstudies by ‘H-NMRspectroscopy on humans have detected an increase of Cho levels in meningiomas compared to normal contralateral brain (30), but the differences in distribution of individual peakscontributing to the Cho signal (i.e., phosphomonoesters and phos

phodiesters) cannot be detected due to the decreased spectral resolution achievable in NMR spectroscopy in vivo. 31P-NMR spectroscopyhas showed elevated phosphomonoester signals in untreated meningiomas of human patients in vivo (31, 32), as well as in spectra fromPCA extractsof resected humanmeningiomas(32).

In the comparison of NMR spectra from extracts to those obtained

in vivo, it is important to appreciate that metabolite levels may differbecause certain metabolite pools may be “NMR-invisible―in vivo.There is little evidence to suggest that most of the major metabolitesdetected by ‘H-NMRspectroscopy are to any extent “invisible,―incontrast to the situation for 31P-NMR spectra in vivo in which ADPand inorganic phosphate certainly contain NMR-invisible pools (33).An exception to this is glutamate, which is about 80% NMR-visible in

rat brain (34) in vivo or in guinea pig brain slices (35). It has beensuggested that the invisible pool is neurotransmitter glutamate (36), inwhich case interpretation of our data from nonneuronal preparationswould be unaffected.

The ‘H-NMRprofiles we obtained from cell populations derivedfrom human meningiomas were very similar to the ‘H-NMRspectrathat have been reported (30) for extracts of meningiomas obtainedfrom human biopsy samples. These similarities included relativelyhigh signals for alanine and Cho, a reduced signal for creatine, andlow or not detectable signals for neuroactive amino acids and relatedcompounds such as GABA, NAA, and NAAG. The low or notdetectable levels of such compounds (which are present in analyses ofhuman central neural tissue) were consistent with the embryonicderivation of the meninges and the location of meninges andmeningiomas outside of the central nervous system.

Metabolic regulation in cultured cells may diverge compared to thesituation in vivo. Furthermore, such changes may be different innormal cells and in transformed cells. However, all of the features thatwe observed in spectra from preparations of cultured meningiomacells in vitro (see above) were in agreement with data reported (23,30)by noninvasive NMR spectroscopy on meningiomas in vivo. Such aconsistency lends support to the ideas that: (a) the metabolic patternin vivo is generally preserved by cells in culture when culture conditions are carefully controlled; and (b) that meningioma cell lines mayoffer a suitable model for studying at least some of the properties oftheir parental tumors.

In conclusion, our investigations showed that human meningiomacell lines derived from excised primary meningiomas expressed similar characteristic metabolic profiles, even when isolated from differ

ent patients. Furthermore, both rat meninges and human meningiomacell lines shared a number of features, which makes them distinguishable from other cells of the central nervous system. It is interesting tonote the conservation of metabolites such as @3-hydroxybutyrate,alanine, glutamine, succinate, inositol, threonine, and maybe glycineacross species and across transformation from normal to tumor tissue.Additionally, we identified certain characteristics such as Cho, PCIGPC,creatine,glutamate,andtaurinewhich enabledthe spectrafromthe human transformed cells to be distinguished from those of normalrat meningeal cells. If these differences prove to be due to metabolicalterations in transformed cells rather than due to species differences,

they might enable neoplastic and normal tissues to be distinguishednoninvasively by ‘H-NMRspectroscopy. Together, our findings raisethe possibility of identifying meningiomas in vivo by performingnoninvasive ‘H-NMRspectroscopy on patients.

ACKNOWLEDGMENTS

We areindebtedto Dr. Sian Davies for helpwith the HPLCdeterminationsfor NAA and NAAG and to Sati Sahota for assistance in the detection of aminoacids by HPLC. Drs. Albert Busza and Martin King are gratefully acknowl

426

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

METABOLIC PROFILES OF MENINGES AND MENINGIOMA CELLS

edged for their advice on software packages and statistical methods,respectively.

REFERENCES

1. Sze, D. Y., and Jardetzky, 0. Determination of metabolite and nucleotide concentrations in proliferating lymphocytes by ‘H-NMRof acid extracts. Biochim. Biophys.Acta, 1054: 181—197,1990.

2. Behar, K. L, Den Hollander, J. A., Stromski, M. E., Ogino, T., Shulman, R. 0.,Petroff, 0. A. C., and Prichard, J. W. High resolution ‘HNMR study of cerebralhypoxia in vivo. Proc. Natl. Acad. Sci. USA, 80: 4945—4948,1983.

3. Gadian, D. 0., Proctor, E., williams, S. R., Cady, E. B., and Gardiner, R. M.Neurometabolic effects of an inborn error of amino acid metabolism demonstrated invivo by ‘HNMR. Magn. Reson.Med., 3: 150—156,1986.

4. Williams, S. R. In vivo proton spectroscopy: experimental aspects and potential. In:M. Rudin and J. Seeig (eds.), NMR—BasicPrinciples and Progress: ln VivoMagnetic Resonance Spectroscopy, pp. 55—72.Heidelberg: Springer Verlag, 1992.

5. Rothman,D. L, Behar,K. L, Hetherington,H. P.,andShulman,R. 0. Homonuclear‘Hdouble-resonance difference spectroscopy of the rat brain in vivo. Proc. Nail.Acad. Sci. USA, 81: 6330—6334,1984.

6. Gadian,D.0. ProtonNMRstudiesofbrainmetabolism.Philos.Trans.R.Soc.Land.A333: 561—570,1990.

7. Urenjak, J., Williams, S. R., Gadian, D. 0., and Noble, M. Proton nuclear magneticresonance spectroscopy unambiguously identifies different neural cell types.J. Neurosci., 13: 981—989,1993.

8. Wolswijk, 0., and Noble, M. Identification of an adult-specific glial progenitor cell.Development (Carob.), 105: 387—400, 1989.

9. Noble, M., Fok-Seang, J., and Cohen, J. Ohs are a unique substrate for the in vitrogrowth of central nervous system neurons. J. Neurosci., 4: 1892—1903,1984.

10. Bignami, A., Eng@L F., Dahi, D., and Uyeda, C. T. Lacalizationof the glial fibrillaryacidic protein in astrocytes by immunofluorescence. Brain Res., 43: 429-435, 1972.

11. Halliday, W. C., Yeger, H., Duwe, 0. F., and Phillips, M. J. Intermediate filamentsin meningiomas. J. Neuropathol. Exp. Neurol., 44: 617—623,1985.

12. Hitchcock, E., and Morris, C. Immunocytochemistry of intracranial meningiomas.J. Neuro-Oncol., 5: 357—368,1987.

13. Schwechheimer, K., Kartenbeck, J., Moll, R., and Franke, W. W. Vimentin filamentdesmosome cytoskeleton of diverse types of human meningiomas. Lab. Invest., 51:548—591, 1984.

14. Drago, J., Reid, K. L, and Bartlett, P. F. Induction of the ganglioside marker A2B5on cultured cerebellar neural cells by growth factors. Neurosci. Len., 107: 245—250,1989.

15. Raff, M. C., Mirsky, R., Fields, K. L, Lisak, R. P., Dorfman, S. H., Silberberg, D. H.,Gregson, N. A., Liebowitz, S., and Kennedy, M. C. Galactocerebroside is a specificcell surface antigenic marker for oligodendrocytes in culture. Nature (Land.), 274:813—816,1978.

16. Mirsky, R. The use of antibodies to define and study major cell types in the centraland peripheral nervous system. In: J. Brockes (ed), Neuroimmunology, pp. 141—177.New York: Plenum Publishing Corp., 1982.

17. Hurd, R. E. Gradient-enhanced spectroscopy. J. Magn. Reson., 87: 422-428, 199018. Preece, N. B., Baker, D., Butter, C., Gadian, D. 0., and Urenjak, J. Experimental

allergic encephalomyelitis raises betaine levels in the spinal cord of strain 1 guineapigs. NMR Biomed., 6: 194-200., 1993.

19. Lindroth, P., and Mopper, K. High perfonnance liquid chromatographic determina

tion of subpicomole amounts of amino acids by precolumn fluorescence derivationwith o-phthaldialdehyde. Anal. Chem., 51: 1667—1674,1979.

20. Koller, K. J., Zaczek, R., and Coyle, J. T. N-Acetyl-aspartyl-glutainate: regionallevels in rat brain and the effects of brain lesions as determined by a new HPLCmethod. J. Neurochem., 43: 1136—1142, 1984.

21. Smith, P. K.. Krohn, R. I., Hermanson, 0. T., Mallia, A. K., Gartner, F. H.,Provenzano, M. D., Fujimoto, E. K.. Goeke, N. M., Olson, B. J., and Kienk, D. C.Measurement ofprotein using bicinchoninic acid. Anal. Biochem., 150: 76—85,1985.

22. Altman, D., Three or more independent groups of observations. In: Chapman and Hall(eds.), Practical Statistics for Medical Research, pp. 205-218. Padstow, Cornwall: T.J. Press(Padstow),Ltd.,1991.

23. Bruhn, H., Frahm, J., Gyngell, M. L., Merboldt, K. D., Hanicke, w., Sauter, R., andHamburger, C. Non-invasive differentiation of tumours with use of localised ‘H-MRspectroscopy in vivo: initial experience in patients with cerebral tumours. Radiology,172: 541—548,1989.

24. Kepes, J. J., Meningiomas-biology, pathology and differential diagnosis. In: MassonMonographs in Diagnostic Pathology, pp. 3—20.New York: Masson Publishing,1982.

25. Scheithauer, B. W. Tumors of the meninges: proposed modifications of the WorldHealth Organization classification. Acts Neuropathol., 80: 343—354,1990.

26. Patel, A. J., and Hunt, A. Concentration of free amino acids in primary cultures ofneurons and astrocytes. J. Neurochem., 44: 1816—1921,1985.

27. Frahm, J., Michaelis, T., Merboldt, K-D., Hanicke, W., Gyngell, M. L., and Bruhn, H.On the N-acetyl methyl resonance in localized ‘HNMR spectra of human brain invivo. NMR Biomed., 4: 201—204,1991.

28. Lehninger, A. L. Glycolysis. In: Biochemistry, pp. 417—441.New York: WorthPublishers, Inc., 1976.

29. Ruiz-Cabello, J., and Cohen, J. Phospholipid metabolites as indicators of cancer cellfunction. NMR Biomed., 5: 226-233, 1992.

30. Gill, S. S. Thomas, D. 0. T., van Bruggen, N., Gadian, D. 0., Peden, C. J., Bell, J. D.,Cox, I. J., Menon, D. K., lies, R. A., Bryant, D. I., and Courts, 0. A. Proton NMRspectroscopy of intracranial tumours: in vivo and in vitro studies. J. Comput. Assist.Tomogr., 14: 497—504,1990.

31. Blankenhorn, M., Bachert, P., Ende, 0., Knopp, M. V., Hess, T., Tromier, V.,Forsting, M., Kolem, H., Lorenz, W. J., and van Kaick, 0. Clinical phosphorus-CSIin preoperative embolization therapy ofmeningioma. Proc. Soc. Magn. Reson. Med.,1026, 1992.

32. Kalra, R., Wade, K. E., Cadoux-Hudson, T. A. D., Parkes, H. 0., McDonald, B., Esiri,M.,andRadda,0. K.NMRstudyofphospholipidmetabolismofhumanmeningioma.Proc. Soc. Magn. Reson. Med., 628, 1993.

33. lIes, R. A., Stevens, A. N., Griffiths, J. R., and Moths, P. 0. Phosphorylation statusby 31P-NMR spectroscopy and its implications for metabolic control. Biochem. 3.,229:141—151,1985.

34. de Graaf, A. A., and Bovee, W. M. M. 3. Improved quantification of in vivo ‘H-NMRspectra by optimization of signal acquisition and incorporation of prior knowledgeinto spectral fitting. Magn. Reson. Med., 15: 305—319,1990.

35. Kauppinen, R. A., and Williams, S. R. Nondestructive detection of glutamate by‘H-NMRspectroscopy in cortical brain slices from the guinea pig: evidence forchanges in detectability during severe anoxic insults. J. Neurochem., 57: 1136—1144,1991.

36. Kauppinen, R. A., Pirttila, T-R. M., Auriola, S. 0. K., and Williams, S. R. Compartmentation of cerebral glutamate in situ as detected by ‘H/'3C-NMR.Biochem.J.,298:121—127,1994.

427

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

1995;55:420-427. Cancer Res   Catarina L. Florian, Nicholas E. Preece, Kishore K. Bhakoo, et al.   Cells by Proton Nuclear Magnetic Resonance SpectroscopyCell Type-specific Fingerprinting of Meningioma and Meningeal

  Updated version

  http://cancerres.aacrjournals.org/content/55/2/420

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/55/2/420To request permission to re-use all or part of this article, use this link

Research. on January 6, 2021. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from