[CANCER RESEARCH 51. 4287-4294. August 15. 1991]

Cellular Glutathione and Thiol Measurements from Surgically Resected Human

Lung Tumor and Normal Lung TissueJohn A. Cook,1 Harvey I. Pass, Susan N. lype, Norman Friedman, William DeGraff, Angelo Russo,

and James B. MitchellRadiation Oncology [J. A. C., S. N. I., N. F., W. D., A. R., J. B. M.] and Surgery [H. I. P.] Branches, National Cancer Institute, N1H, Bethesda, Maryland 20892

ABSTRACT

Cellular glutathione (GSH) levels were measured from 27 human lungtumor biopsies, enzymatically disaggregated, and compared with cellsisolated from normal lung of the same patients. GSH levels from normallung were similar among patients with a mean value of 11.20 ±0.58(SEM) niiiol GSH/mg protein (24 patients) with a range from 6.1 to 17.5nmol GSH/mg protein. GSH levels varied considerably within and acrosshistológica! tumor types with the following values: adenocarcinomas, 8.83±0.96 nmol/mg protein (8 patients); large cell carcinomas, 8.25 ±2.51nmol/mg protein (3 patients); and squamous cell carcinomas, 23.25 ±5.99 nmol/mg protein (8 patients). The cyclic GSH reducÃ-ase assay gaveonly average GSH values and could not distinguish possible GSH variation among subpopulations of cells isolated. Cell volume measurementsand microscopic evaluation of cells isolated from both tumors and normallung revealed heterogeneity with respect to cell types present. To determine the extent of thiol variation among tumor cell subpopulations, tumorcell suspensions were stained with the thiol-specific stain, monochloro-

bimane (MCB). The accuracy of MCB staining was tested by flowcytometric analysis of 12 in vitro human tumor cell lines and 3 rodentcell lines. A linear relationship was found between the bimane cellularfluorescence and the cyclic GSH reductase assay for cell lines havingless than 80 nmol GSH/mg protein ( AT' = 0.82). Above 80 nmol GSH/

mg protein the rate of change of the bimane fluorescence intensity withrespect to increasing GSH concentrations was much reduced. However,by labeling cells with MCB it was possible to distinguish between celllines with low versus high GSH content. MCB staining of tumor samplesrevealed multiple populations of cells with respect to thiol levels. Inparticular, 2 of 8 squamous cell carcinomas had a proportion of cells withelevated fluorescence intensities (from 10 to 35% of the population)suggesting the presence of cells with greatly elevated thiol levels. Thesefindings underscore the complexity of quantitating intracellular GSHlevels from tumor biopsies. The combined use of MCB with flow cytom-

etry and conventional GSH assays may help to delineate subpopulationsof cells within tumors with different thiol levels.

INTRODUCTION

Drug resistance of tumors poses a significant problem incancer treatment. The problem is further compounded by thefact that tumors may become resistant to diverse chemotherapydrugs with differing modes of cell killing. Generally, normaltissues fail to develop resistance to drug treatments therebyrendering the prospect for obtaining a therapeutic gain fortreatment of solid tumors with chemotherapy a major challenge.As more effort is focused on understanding the cellular, molecular, and physiological mechanisms underlying drug resistance,it becomes clearer that a number of factors may be involved inthe development of drug resistance. Studies at the cellular levelhave demonstrated that overexpression of membrane glycopro-teins involved in drug influx/efflux (1-3), elevated levels ofredox active molecules (4-6), and elevated activities of enzymes

Received 8/8/90; accepted 6/6/91.The costs of publication of this article were defrayed in part by the payment

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

' To whom requests for reprints should be addressed, at Radiation OncologyBranch/NCI, Bldg. 10, Room B3-B69, Bethesda, MD 20892.

involved in detoxification (7, 8) can provide significant cellularresistance to chemotherapy drugs. The extent to which thesebiochemical adaptations contribute to drug resistance in vivo isat present not clear; however, they represent major targets forfurther study and possible clinical exploitation.

GSH2 and GSH-related enzymes are known to function in

the cellular detoxification of potentially harmful xenobioticsand oxygen-related toxic species (9, 10). The importance ofGSH in altering cellular response to certain chemotherapydrugs has been demonstrated by virtue of the development ofagents that either inhibit (11) or stimulate (12) intracellularGSH synthesis. Depletion of intracellular GSH by BSO in avariety of cell types has been shown to markedly enhance thecytotoxicity of many chemotherapy drugs (13, 14) and hypoxiccell radiosensitizers (15). Conversely, elevating GSH levelsprior to drug treatment by oxothiazolidine carboxylate canafford significant protection against chemotherapy drug-mediated cytotoxicity (4, 6, 16). Further evidence linking GSHwith drug resistance has come from studies relating inherentintracellular GSH levels with drug sensitivity (17). Louie et al.(17) have shown that a human ovarian cell line made stablyresistant to either melphalan, Adriamycin, or cisplatin by longterm incubation in increasing concentrations of each respectivedrug had significantly higher GSH levels than the parent cellline. Drug resistance in each of these cell lines could be reversedby BSO-mediated GSH depletion (17). Human breast cancercell lines resistant to Adriamycin exhibit increased GSH per-oxidase and GSH transferase activity (7). Such findings haveprompted the clinical exploration of GSH modulation of tumorcell GSH by BSO treatment. Phase I trials are currently underway evaluating BSO treatment in conjunction with melphalanfor ovarian cancer.

The ongoing clinical trials centered around GSH modulationbring to attention several issues that will be important ininterpreting the success or failure of such therapeutic strategies.(a) Do tumor cells in vivo actually have higher GSH levels thannormal tissues? (b) Will BSO treatment result in sufficientGSH depletion to sensitize the tumor to the particular chemotherapy drug(s) used? (c) Will BSO treatment result in differential GSH depletion rates of tumor versus normal tissues? (d)Are there means of repleting normal tissue GSH levels aftercombination BSO/chemotherapy regimens? In order to answerthese important questions accurate GSH measurements of tumor and normal tissues are imperative.

Accurate measurement of tumor GSH may be complicatedby the presence of multiple cell populations within the tumorbiopsy (18). Currently, the most frequently used methods fordetection and quantitation of GSH include an enzymatic assayutilizing GSH reductase developed by Tietze (19) or high performance liquid Chromatographie assays using fluorescentprobes which react with GSH (20). Both assays are quite

2The abbreviations used are: GSH, glutathione; BSO, buthionine sulfoximine;MCB. monochlorobimane; PCS, fetal calf serum; HBSS, Hanks' balanced saltsolution; PBS, phosphate-buffered saline; SSA, sulfosalicylic acid; GST, GSH S-transferase; SEM, standard error of the mean.

4287

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GSH AND THIOL IN LUNG TUMORS AND TISSUE

sensitive and may be appropriate for GSH measurements ofhomogeneous, in vitro cell lines; however, each assay measuresonly the average GSH content from a general population ofcells and is less suited for specific analysis of the multiplecellular populations which may be found within a tumor. Whentwo or more populations of cells are present in a sample themean GSH value may over- or underestimate the GSH level ofa particular population of interest. To determine GSH levels insubpopulations within tumor cell suspensions, measurementson a single cell basis with the use of MCB have been proposed(21 ). MCB is a nonfluorescent molecule which when conjugatedto GSH (or other sulfhydryl containing compounds) becomeshighly fluorescent. In rodent cell lines at low MCB concentrations ( 80% as determined by trypan blue dye exclusion.

Flow Cytometer Analysis. Cells (IO6 cells/ml) in PBS were obtained

either by direct trypsinization of in vitro cultures or in the case of tumorand normal tissue specimens by the tissue disaggregation techniquedescribed above. The cells were incubated with MCB (MolecularProbes, Inc., Junction City, OR) for 60 min at room temperature. Afterstaining, the cells were washed once with PBS and then resuspended in1 ml PBS for flow cytometry analysis. Samples were analyzed using aCoulter Epics V cell sorter (Coulter Electronics, Inc., Hialeah, FL) asdescribed in a previous paper (24). For most specimens 10,000 to20,000 cells were collected for analysis.

GSH Assay (Tietze). Single cell suspensions were obtained as described above and triplicate samples of IO6 cells were resuspended in

0.6% SSA for GSH analysis. Total GSH (reduced plus oxidized) wasdetermined by the GSH cyclic reducÃ-aseassay as described by Tietze(19). Protein was measured by the method of Bradford (25). GSH isexpressed in nmol GSH per mg protein. The statistics for the Tietze's

data are reported as the mean ±SEM.Measurement of Protein Labeling with MCB. Cells (106/ml) were

labeled with 1 mM MCB for l h at room temperature. After onewashing with PBS the cells were resuspended in either 1 ml PBScontaining 0.2% Triton X-100 to measure the total fluorescence (protein plus nonprotein sulfhydryls) or resuspended in 0.6% SSA andplaced on ice for 30 min to isolate total cellular protein. The SSA-treated cells were centrifuged and the pellet was resuspended in 1 mlPBS containing 0.2% Triton X-100 to measure the protein fluorescence.

The fluorescence for each sample was measured using a SLM8000TNiC spectrofluorometer (Urbana, IL). The bimane fluorescencewas excited at 400 nm and the fluorescence was detected at 480 nm.Endogenous fluorescence from cells not treated with MCB but processed similarly to the MCB treated cells was analyzed and subtractedfrom the MCB treated cells.

Volume Analysis. Cell volumes of each cell line and of tumor/normalcell suspensions were determined by using a Elzone counter (ParticleData, Inc., Elmhurst, IL) calibrated with microspheres of knowndiameters.

RESULTS

Effect of Enzyme Digestion on GSH Levels. Because GSHwas to be measured on a single cell basis enzymatic digestionof the tumor specimens was necessary. Tissue specimens oftencontained variable amounts of blood and since RBC containGSH we thought it was necessary to remove (by lysis) the RBCto obtain more accurate tumor cell GSH values. In an effort toexamine what effect the enzyme cocktail and RBC lysis buffer

4288

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GSH AND THIOL IN LUNG TUMORS AND TISSUE

had on GSH levels in cells, human lung adenocarcinoma cells(A549) were incubated for up to l h in the enzyme cocktail andresuspended in the RBC lysis buffer for 5 min to simulate theconditions of the tumor disaggregation and RBC lysis conditions. Incubating untreated cells in the RBC lysis buffer alonehad minimal effects on the cellular GSH levels (93% of control).Incubation of cells for up to l h in the enzyme cocktail aloneor in conjunction with the RBC lysis buffer (protocol followedfor tumor samples) had essentially no effect on cellular GSHlevels (99% of untreated control cells).

GSH Measurements of Tumor and Normal Lung Tissue. TheGSH content as determined by the GSH cyclic reducÃ-aseassayfor 27 surgically resected human lung tumors and their adjacentnormal lung tissue (24 samples were obtained) is shown inTable 1. To facilitate comparisons Table 1 is separated into 4groups: group 1, adenocarcinomas; group 2, squamous cellcarcinomas; group 3, large cell and small cell carcinomas andbronchial carcinoid; and group 4, lung metastasis from varioustumors. The adenocarcinoma specimens (n = 8) had a GSHcontent of 8.86 ±0.96 nmol/mg protein while the matchingnormal lung specimens had a GSH content of 10.64 ±0.94nmol/mg protein. The squamous cell carcinoma specimens (n= 8) had a GSH content of 23.25 ±5.99 nmol/mg proteincompared to their normal lung specimens with GSH levels of13.13 ±1.23 nmol/mg protein. The small cell carcinoma, largecell carcinoma, and bronchial carcinoid specimens (n = 6) hada GSH content of 12.34 ±2.40 nmol/mg protein compared tomean GSH levels of 9.65 ±1.11 nmol/mg protein for theirnormal lung specimens (n —5). Of the specimens analyzed, 4patients' tumors were found to have GSH levels greater than

2-fold that of their corresponding normal tissue. These included2 squamous cell carcinomas (No. 1557 with 23.4 nmol GSH/mg protein and No. 1534 with 62.9 nmol GSH/mg protein), abronchial lung carcinoid (No. 1287 with 21.4 nmol GSH/mgprotein), and a metastatic osteogenic sarcoma to the lung (No.

Table 1 Human tumor and normal lung tissue GSH data

No.123456789101112131415161718192021222324252627Patient116312691270130113021349151615311169130014291439146115341557145312051235127412871316147511721177119614271492TumortypeAdenocarcinomaAdenocarcinomaAdenocarcinomaAdenocarcinomaAdenocarcinomaAdenocarcinomaAdenocarcinomaAdenocarcinomaSquamous

cellcarcinomaSquamouscellcarcinomaSquamouscellcarcinomaSquamouscellcarcinomaSquamouscellcarcinomaSquamouscellcarcinomaSquamouscellcarcinomaAdenosquamous

carcinomaSmall

cellcarcinomaLargecellcarcinomaLargecellcarcinomaBronchial

carcinoidLargecellcarcinomaBronchial

carcinoidMetastatic

soft tissuesarcomaMetastaticsoft tissuesarcomaMetastaticosteogenicsarcomaMetastaticsoft tissuesarcomaPrimary

pulmonary sarcomaTreatment

statusNoNoNoNoNoVP/CP"NoVP/CPNoRTNoNoNoNoNoNoNoVP/CPNoNoNoCyt/AdriaNoCyt/AdriaNoGSH

(nmol/mg)Tumor7.023.8010.5312.2810.447.029.2710.2323.3916.8116.2610.4410.0362.8725.2620.9410.536.4311.1121.357.2217.432.9210.6733.929.3627.36Normal11.118.197.6012.2813.7411.4911.7011.1110.2314.9113.018.7716.3217.5410.826.1412.878.779.659.369.656.4312.5714.33

TUMOR

" VP, etoposide; CP, c/i-platinum; Cyt, Cytoxan; RT, radiation therapy; Adria,

Adriamycin.

CHANNEL NUMBER(LOC INTENSITYI

Fig. 1. Volume analysis of cell suspensions from normal lung and tumorspecimens. A, normal lung cells; B, lung métastasesfrom a soft tissue sarcoma;C, lung métastasesfrom a osteogenic sarcoma. Volume analysis of humanperipheral WBC shows a peak at channel 29.

1196 with 33.9 nmol GSH/mg protein). For comparison, exponentially growing Chinese hamster V79 cells had a GSHcontent of 29-35 nmol GSH/mg protein while an establishedhuman ovarian carcinoma line (OVG-1) in exponential growthhad a GSH content of 175-234 nmol GSH/mg protein. Differences in GSH levels between tumor biopsies and in vitro tumorcell lines have been reported previously (18).

Variability of Cell Types from Tumor/Normal Lung Digests.Because Tietze's assay provides only an average measurement

of GSH content within a population of cells, the presence ofmultiple populations each having a unique GSH content couldnot be easily identified and measured. That multiple populations of cells were present in both the tumor and normal lungtissue can readily be seen by representative volume analysis ofspecimens (Fig. 1). Cells isolated from normal lung consistentlyshowed the presence of at least 3 populations while the tumorspecimens were generally much more variable (2-3 peaks). Wehave tentatively identified the peak at channel 29 as lymphocytes since WBC isolated from human blood are of similar sizeand anti-pan leukocyte antibody staining of tissue sections alsoindicates the presence of lymphocytes in these samples (datanot shown).

Use of MCB to Measure Cellular GSH Levels. Recently thedye, MCB, has been introduced as a means of specificallylabeling GSH in cells with a fluorescent probe. When MCB isused in conjunction with flow cytometry it becomes possible toexamine the GSH heterogeneity profile of a cellular population(21). Previous work has indicated that when the GSH levels inrodent cell lines were depleted or elevated by known amountsand analyzed by either the GSH cyclic reducÃ-aseassay or theMCB assay thai an excellenl quantitative correlation existedbetween ihe iwo assays (21, 25). These dala suggesled lhal by

4289

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GSH AND THIOL IN LUNG TUMORS AND TISSUE

12

esUS

Õ 3OÃ-

.-'n

12

3

50 100 150 200

nmol GSH /mg protein

250

Fig. 2. Correlation between the GSH levels determined with either Tietze's or

the MCB technique. GSH levels were determined by the GSH cyclic reducÃ-aseassay for 12 different human cells lines and 3 rodent cell lines; a ratio value(RATIO GSH) was created by taking the total GSH content for each cell line(nmol GSH) and dividing it by the total GSH content of the V79 cell. Similarly,for the MCB analysis, the mean fluorescence intensity for each cell line wasdivided by the mean fluorescence intensity of the V79 cells (RATIO MCB). Forthe MCB analysis. 10,000 cells were collected and analyzed for each of the celllines studied. The lines drawn were fitted by eye.

labeling and analyzing a cell line of known GSH content on thesame day as the tumor and normal cells it is possible toquantitate the GSH content of any population of cells found inthe specimen by simple linear extrapolation. To test this assumption the bimane fluorescence was compared to the GSHcontent measured by the GSH cyclic reducÃ-aseassay for 15different in vitro human and rodent cell lines (Fig. 2). It hasbeen shown previously that GSH in human cell lines cannot belabeled using MCB concentrations typically used to label theGSH in rodent cell lines (25); maximal GSH labeling for humancell lines requires concentrations exceeding 600 fiM MCB (25).When 1 mM MCB was used to label these 15 different celllines the relationship between the bimane fluorescence intensities and the GSH cyclic reducÃ-aseanalysis (Fig. 2) was by nomeans as good as thai determined for the rodent lines (21, 25).So thai resulls collecled on differenl days could be compared,a ralio melhod was used in which Ihe mean fluorescence fromeach cell line was compared lo ihe mean fluorescence of V79cells. Because bimane fluorescence measures Ihe lolal Ihiolconlenl and noi Ihiol concenlralion, a similar ralio analysisusing ihe lolal GSH (nmol) conlent of each cell line wasdelermined (by Ihe GSH cyclic reducÃ-aseassay), again usingihe lolal GSH (nmol) conlent in V79 cells for comparison (Fig.2). As expected, a good correlation was found between the GSHcyclic reducÃ-aseassay values correcled for prolein conlenl (nmolGSH/mg prolein) and Ihe ralio of lolal GSH conlenl (R2 =

0.84) (Fig. 2). Below 80 nmol GSH/mg prolein cellular bimanefluorescence increased linearly wilh increasing GSH conlenl(R2 = 0.82). However, above 80 nmol GSH/mg Ihe rale of

change in Ihe cellular fluorescence with respect lo increasingGSH concenlralion was much reduced.

For MCB concenlralions grealer lhan 100 ¿

GSH AND TH1OL IN LUNG TUMORS AND TISSUE

sibility of quantitative differences in thiol content among thedifferent subpopulations of cells within the tumor biopsy (Fig.3, A, E, and G). Multiple fluorescence peaks were also noted inthe normal lung specimens and the relative proportion of cellsin these peaks varied from patient to patient (Fig. 3, B, D, F,and H). (b) In a majority of specimens examined, the meanbimane fluorescence values calculated for tumor cells were lessthan the calculated mean of the corresponding normal lungcells. However, in 2 of the squamous cell carcinoma samplesstudied, a distinct population of cells was identified as havinga significantly higher fluorescence than any cells in the corresponding normal lung cell profiles (Fig. 3, E, G).

Because the MCB fluorescence profiles are a measure of totalthiol content and not thiol concentration it is not possible tostate with certainty that the peaks seen in Fig. 3 representpopulations with different thiol concentrations. However, if thevolume and thiol content of each subpopulation of cells wereknown then a more definitive statement regarding the heterogeneity of thiol concentrations between cell populations couldbe made. Therefore, in an effort to correct the fluorescence forsize differences, the fluorescences from cells with identical ornear identical light scatter properties were collected and analyzed (Fig. 4). Although not linearly related to size, the lightscatter signal collected from cells did provide a normalizationtechnique for size measurement which can be used for comparison between different cell populations (27). The first squamouscell sample (No. 1534) shown in Fig. 4 demonstrated that forcells of similar sizes (cross-hatched area) a population of cells(35%) with >5 times the fluorescence intensity of normal lungcells was identified. A second squamous cell carcinoma sample(No. 1557) was analyzed and found to be very similar to thedata shown in Fig. 4 (data not shown); however, unlike the firstsample, the proportion of cells having high fluorescence valueswas much lower (10%) (Fig. 3C7).

As shown in Fig. 3 the tumor and normal lung specimenMCB profiles showed significant variation from patient topatient. To study whether these differences were due to changesin the composition of normal lung cell populations or to aMCB staining artifact, both volume distributions (as measuredwith a particle data counter) and forward angle light scatterinformation were collected for tumor and normal cell populations. The relative proportions of cells were seen to shift frompatient to patient when analyzing either light scatter or volumemeasurements and did not appear to be due to any stainingartifact (data not shown).

DISCUSSION

Previous studies have shown that GSH can and does functionto reduce the cellular toxicity of various chemotherapy agents(13, 17). On the basis of this evidence investigators have postulated that one form of drug resistance may result from elevated levels of GSH in tumor cells (28, 29). Whether hightumor GSH levels contribute to drug resistance upon initialtreatment or at relapse is not known. Before such importantissues may be addressed, techniques must be developed toaccurately measure tumor GSH levels. In this study we haveanalyzed the GSH content of a number of lung tumors andnormal lung tissue. With the possible exception of some squamous carcinomas, in general, tumor GSH levels were found tobe not significantly higher than those for normal lung. Thepresence of subpopulations of cells having high GSH levelswithin tumor samples has been documented by the current

LIGHT SCATTER FLUORESCENCE

uca

uUEd

C TUMOR

50 100 150 200 250 50 150 200

CHANNEL NUMBER(LOG INTENSITY)

Fig. 4. Light scatter and fluorescence profiles of the squamous cell sample(No. 1534) (Table 1). The fluorescence profiles were obtained by restrictinganalysis to cells of a defined size (hatched area in light scatter profiles). A. B, cellsisolated from normal lung. C, D, cells isolated from the lung tumor. We analyzed50,000 cells to obtain the light scatter profiles and of these cells only 10.000 werecollected for the fluorescence profiles. Every 85 channels represent approximatelya 10-fold change in intensity.

work. However, several concerns and limitations regarding themeasurement of GSH from tumor biopsies and how these datamay be interpreted are raised by our findings.

Inherent in the process of tissue disaggregation to single cellsuspensions is the potential for loss, degradation, or oxidationof the molecule of interest. There are a number of techniquesavailable for tissue disaggregation involving use of differentenzymes and variable lengths of exposure (29, 31). To avoidmajor changes in GSH concentrations during the cell isolationperiod, a high activity enzyme cocktail was applied for a shorttime. The GSH levels in A549 cells incubated for l h in theenzyme digest buffer was reduced by less than 10%. Variationin the number of RBC [GSH concentrations between 1 and 3iriM(32)] posed another potential problem. Although the totalamount of GSH is small for each individual RBC, large andvarying numbers of RBC in tissue samples could have an impacton the interpretation of GSH values for the cells of interest.The simplest means of addressing the RBC problem was toselectively remove the RBC from the samples by lysis. Treatment of A549 cells with the enzyme cocktail in conjunctionwith the RBC lysis buffer also had minimal effects on thecellular GSH levels. In addition to the in vitro data, the GSHcyclic reducÃ-aseresults from the tumor and normal tissuebiopsies suggest that the perturbations was not severe because:(a) the normal lung GSH values were very uniform from patientto patient; and (b) in the majority of the lung cases studiedthere was no significant GSH differences between the tumorand normal lung tissue. Furthermore, the GSH levels for boththe tumor and normal tissues were in a range previously reported both for lung tumors and other tumor types (18, 33-34). We believe that while the absolute GSH values listed inTable 1 may have some error associated with them, the relativeGSH differences between tumor and normal lung specimensshould be reasonably accurate.

Another important variable in tumor GSH measurementswas the presence of multiple populations of cells found in bothtumor and normal lung specimens. This was immediately apparent from the analysis of cell suspension volume data (Fig.1)and from initial studies using an anti-panleukocyte monoclo-

4291

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GSH AND THIOL IN LUNG TUMORS AND TISSUE

nal antibody to establish that leukocytes (B-cells, T-cells, macrophages, and neutrophils) are present in tumor samples. Because leukocytes have low GSH levels (18), their presence intumor specimens would be expected to lower the average GSHmeasurements determined by the Tietze's assay. The composi

tion of other cell populations is unknown. Histological examination of each tumor specimen indicated that tumor cells werepresent; however, there may be different subpopulations oftumor cells present as a result of environmental and nutritionalconstraints within the tumor as well as populations of proliferating and nonproliferating cells. Additionally, normal hostfibroblasts could be present. All of these variables can have animpact on the GSH status of the population of cells measured(18,33).

Another question associated with human tumor GSH measurements is what cell types, if any, the tumor cells should becompared against. From a therapeutic view, tumor GSH levelsshould be compared with that of normal tissue(s) with thelowest levels and therefore presumably at the greatest risk fortoxicity. However, obtaining samples of normal tissue is notalways feasible. Although it was not entirely ideal, we werefortunate to have for comparison normal lung tissue obtainedfrom a location distal to the tumor site for comparison. GSHlevels measured in normal lung cells from 24 patients were verysimilar even though the size information indicated that therelative proportion of each population of cells found in normallung samples varied from patient to patient. In addition, thefluorescence profiles for the normal lung samples also showedsignificant differences from patient to patient (Fig. 3). Theseconflicting results could be explained if the cellular populationsof the normal lung had similar GSH concentrations.

For many of the tumor specimens analyzed, little or nodifference was found between the GSH content of the tumorand the surrounding normal lung cells (Table 1). However, thepossibility remains that a limited subpopulation of tumor cellswith high GSH levels could be present in tissue but, by virtueof their limited numbers, would go undetected by the averagingnature of the GSH cyclic reducÃ-aseanalysis. Multiple peaks inthe fluorescence profiles were indeed seen and the possibilitythat cells with differing thiol levels in both tumor and normalspecimens may be a frequent occurrence is supported by suchmultiplicity (Fig. 3). In a majority of the samples studied, thetumor specimens did not have populations of cells with abimane fluorescence exceeding the bimane fluorescence of cellscomposing the normal populations (Fig. 3, A-D). Thus, forthese samples, it would not appear that tumor cell subpopula-tions with higher than normal thiol levels were present. Eventhough no significant differences in GSH levels were seen inmany of the tumor specimens when compared to normal lungit is still possible that patients developing resistance to chemotherapy agents may show an increasing subpopulation of cellswith elevated levels of GSH as treatment time progresses.Another possibility is that the tumor does not have constitu-

tively higher levels of GSH but is equipped to rapidly and morereadily synthesize GSH if and when stressed by chemolhera-

peutic drugs.MCB has been shown previously to label the GSH pool in

rodent cell lines and to be an excellent indicator of the GSHstatus when GSH is modulated by either BSO (to deplete) orGSH esters (to elevate) (21, 24). However, our experience hasbeen that the concentrations used to label rodent cell lines areentirely inadequate for GSH labeling in human cell lines (24,26). In fact, we have found that concentrations up to 1 mM

MCB are required for maximal GSH labeling in human cells,a concentration some 20-100 times greater than that used for

the labeling of rodent lines (24, 26). When 1 mM MCB wasused to label various human and rodent cell lines the fluorescence intensities were correlated to GSH levels in cell lines with80 nmol GSH/mg protein or less (Fig. 2). Above 80 nmolGSH/mg protein changes in the fluorescence intensities withrespect to increasing concentrations of GSH were reduced (Fig.2). Since 1 mM MCB labels from 60 to 90% of the GSH poolsin these human tumor cell lines only a small fraction of thedifferences between the GSH cyclic reducÃ-aseresults and fluorescence resulls can be explained by inadequate GSH labeling(26). Because of the differences between Ihe fluorescence andIhe GSH cyclic reducÃ-ase resulls, we slress lhal using Ihepublished condilions for labeling GSH wilh MCB (21) andIhose we have oplimized for Ihe presenl sludy (24, 26), thetechnique cannot be used as a quantitative assay for measuringthe GSH levels in human cells.

Although not usable as a quanlilalive GSH assay the MCBassay can provide some useful qualitative cellular thiol infor-malion. Therefore, comparisons were made belween normallung cells and tumor cells which were processed, stained, andanalyzed on Ihe same day. In 2 squamous cell lung carcinomasamples (Nos. 1534 and 1557) Ihe bimane fluorescence profilesshowed the presence of a subpopulation of cells having a muchhigher fluorescence inlensily (>5 limes) lhan cells from Ihenormal lung (Fig. 3, £,G). The elevaled fluorescence intensitiescould have resulted from several possibilities. Firsl, cells wilhequal Ihiol concenlralions bul differenl volumes would havedifferenl fluorescence profiles due lo Ihe faci that the MCB-flow cytomelry assay measures Ihe Ihiol conlenl of cells. However, when cells of similar sizes from bolh normal and lumorpopulalions were analyzed (Fig. 4, B, D) Ihe subpopulalion oftumors cells wilh elevaled fluorescence inlensilies was siilipresenl, ihus supporling Ihe proposal lhal wilhin ihis lumorIhere was a subset of cells wilh iruly differenl thiolconcentralions.

Second, because Ihe MCB/GSH reaclion is catalyzed by theGSH 5-lransferases (21, 26, 35), enzyme differences (eilherquantitative or calalylic) could have resulled in Ihe elevaledfluorescence intensilies. This possibilily was considered unlikely inasmuch as the labeling limes were long (1 h) and Iheconcentration of MCB used was high. In addition, we haveshown in MCF7WT cells, which contain very low (or undetecl-able) GST aclivily, thai Ihe percenlage of GSH labeled by MCBwas similar to lhal in Ihe MCF7ADR cells which contain veryhigh GST activity (7, 26). Thus, at Ihe higher MCB concenlralions, MCB labeling of GSH occurs predominancy by nonen-zymalic means and iherefore should noi be heavily influencedby Ihe presence or absence of GST aclivily (26).

Third, because of Ihe high MCB concenlralions needed lolabel GSH in human cells il is possible lhal cellular proleinIhiols and olher nonprolein Ihiols (such as cysleine) may con-iribule subslanlially lo ihe bimane fluorescence profiles. We donoi believe lhal bimane prolein fluorescence represenls a sig-nificanl porlion of Ihe lumor or normal lung flow fluorescenceprofiles since resulls from Table 2 suggesl lhal al mosl only20-30% of the MCB-derived fluorescence is due lo proleinlabeling. Furthermore, by high performance liquid Chromatographie analysis Ihe fluorescence profile of several MCB-slainedlumor specimens revealed lhal the bimane-GSH adduci was Ihepredominanl bimane adduci presenl (>85-95% of Ihe lolal

4292

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GSH AND THIOL IN LUNG TUMORS AND TISSUE

REFERENCES

12.

nonprotein sulfhydryls labeled).3 Most importantly, in patient 4

1534 the GSH content of the tumor cells measured by the GSHcyclic reducÃ-aseassay was much higher than the normal lungcells (62.9 versus 8.8 nmol/mg protein). This suggests that the 5population of cells with increased fluorescence intensities seenin Fig. 3, E, G, was due to real GSH differences. Therefore,while additional work must be done to conclusively demonstrate 6elevated GSH levels within tumor specimens, our data provideevidence for and is consistent with the proposal that tumorscan have a subpopulation of cells with elevated GSH levels.

Since only 35% of the cells had elevated fluorescence inten- 8sities it is likely that much larger GSH differences actuallyexisted in the tumor cells (Fig. 3£). The proposal that thepresence of a subpopulation of cells with high GSH levels withina tumor could be missed by any averaging technique is illus- 9

trated by the second squamous cell case (patient 1557) in whichthe actual measured difference in GSH levels between the tumorand normal tissue was much less (25.3 versus 16.3 nmol GSH/ 10

mg protein). For this tumor only 10% of the cells had elevated nfluorescence intensities in the flow cytometry profile (Fig. 3G).These data would suggest that when utilizing averaging techniques to measure GSH it is possible to overlook significantGSH differences between tumor and normal cells.

Finally, associated with GSH measurements of tumor specimens is intratumor GSH heterogeneity. A recent report by Lee 13

et al. (18) has emphasized the fact that GSH levels can varydepending upon the location of the biopsy. Using human ovarian xenografts, differences in GSH levels from 4- to 7-fold were 14observed when individual pieces of the same tumor were compared (18). They suggested that such differences might in partresult from the growth status among cells in different regionsof the tumor and showed that plateau phase cells have much 1S

lower GSH levels than exponentially growing cells (18). GSHvariability may also be present in the data presented in thisreport; however, without exception the tumor GSH values 16

measured were either very similar to the normal lung tissueGSH values or higher. n

In conclusion, we have measured the GSH content of bothtumor and normal tissue specimens using a combination ofTietze's and MCB analyses. In many of the tumor specimensanalyzed the GSH levels as measured by the Tietze's technique is

were not different from normal lung tissue. The MCB techniqueindicated the possible presence of cells having differing thiol 19levels. Most of the tumor specimens analyzed did not have cellswith thiol levels exceeding cells in the corresponding normaltissue. The MCB and GSH cyclic reducÃ-asedata in 2 squamouscell carcinoma cases indicated that cells with elevated GSHlevels were present in the specimens. While the MCB technique 21has several problems when used to stain human cells, it mayoffer the best hope in identifying GSH heterogeneity in themultiple populations found in human tumor and normal 22specimens.

20.

23.

1. Ling, V., Kartner, N., Sudo, T., Siminovitch, L., and Riordan, J. R. Multi- 24.drug-resistance phenotype in Chinese hamster ovary cells. Cancer Treat.Rep., 67: 869-874, 1983.

2. Kartner, N., Riordan, J. R., and Ling, V. Cell surface P-glycoprotein asso- 25.ciated with multidrug resistance in mammalian cell lines. Science (Washington DC), 23: 1285-1288, 1983.

3. Kartner, N., Evernden-Porelle, D., Bradley, G., and Ling, V. Detection of P- 26.glycoprotein in multidrug-resistant cell lines by monoclonal antibodies. Nature (Lond.), 316: 820-823, 1985.

27.3 Unpublished observation.

Russo, A., DeGraff, W., Friedman, N., and Mitchell, J. B. Selective modulation of glutathione levels in human normal rerun tumor cells and subsequent differential response to chemotherapy drugs. Cancer Res., 46: 2845-2848, 1986.Teicher, B. A., Crawford, J. M., Holden, S. A., Yawares, L., Cathcart, K. N.S., Luchette, C. A., and Flatow, J. Glutathione monoethyl ester can selectively protect liver from high dose BCNU or cyclophosphamide. Cancer(Phila.), 62: 1275-1281, 1988.Mitchell, J. B. Glutathione modulation and cancer treatment. ISI Atlas ofScience: Phar., 2: 155-160, 1987.Batist, G., Tulpule, A., Sinha, B., Katki, G., Meyers, C., and Cowan, K.Overexpression of a novel anionic glutathione transferase in multidrug-resistant human breast cancer cells. J. Biol. Chem., 261:15544-15549,1986.Johnston, J. B., Israels, L. G., Goldenburg, G. J., Anhalt, C. D., Verbürg,L., Mowat, M. R. A., and Begleiter, A. Glutathione 5-transferase activity,sulfhydryl group and glutathione levels, and DNA cross-linking activity withchlorambucil in chronic lymphocytic leukemia. J. Nati. Cancer Inst., 82:776-779, 1990.Hayes, J. D., and Wolf. C. R. Role of glutathione transferase in drugresistance. In: H. Sies and B. Ketterer (eds.), Glutathione Conjugation,Mechanisms and Biological Significance, pp. 315-355. San Diego, CA:Academic Press, 1988.Ketterer, B. Protective role of glutathione and glutathione transferases inmutagenesis and carcinogenesis. Mutât.Res., 202: 343-361, 1988.Dethmers, J. K., and Meister, A. Glutathione export by human lymphoidcells: depletion of glutathione by inhibition of its synthesis decreases exportand increases sensitivity of irradiation. Proc. Nati. Acad. Sci. USA, 78:7492-7496, 1981.Anderson, M. E., Powrie, F., Puri, R. N., and Meister, A. Glutathionemonoethyl ester: Preparation, uptake by tissues, and conversion to glutathione. Arch. Biochem. Biophy., 55: 96-104, 1985.Ozols, R. F., Louie, K. G., Plowman, J., Behrens, B. C., Fine, R. L., Dykes,IX. and Hamilton, T. C. Enhanced melphalan toxicity in human ovariancancer in vitro and in tumor-bearing nude mice by buthionine sulfoximinedepletion of glutathione. Biochem. Pharmacol., 36: 147-153, 1987.Skapek, S. X., Colvin, O. M., Griffith, O. W., Groothuis, D. R., Colapinto,E. V., Lee, Y., Hilton, J., Elion, G. B., Bigner, D. D., and Friedman, H. S.Buthionine sulfoximine-mediated depletion of glutathione in intracranialhuman glioma derived xenografts. Biochem. Pharmacol., 37: 4313-4317,1988.Mitchell, J. B., Philips, T. L., DeGraff, W., Carmichael, J., Raijpal, R. K.,and Russo, A. The relationship of SR-2508 sensitizer enhancement ratio tocellular glutathione levels in human tumor cell lines. Int. J. Radiât.Oncol.Biol. Phys., /2: II43-1146, 1986.Russo, A., and Mitchell, J. B. Alteration of bleomycin cytotoxicity byglutathione depletion or elevation. Int. J. Radial. Oncol. Biol. Phys., 10:1675-1678, 1984.Louie, K. G., Behrens, B. C., Kinsella, T. J., Hamilton. T. C., Grotzinger,K. R., McKroy, W. M., Wrinker, M., and Ozols, R. F. Augmentation ofAdriamycin melphalan and cisplatin cytotoxicity in drug resistant and sensitive human ovarian cell lines by buthionine sulfoximine mediated glutathione depletion. Biochem. Pharmacol., 34: 2583-2586, 1985.Lee, F. Y. F., Vessey, A., Rofstad, E., Siemann, D. W., and Sutherland, R.M. Heterogeneity of glutathione content in human ovarian cancer. CancerRes., 49: 5244-5248, 1989.Tietze, F. Enzymic method for quantitative determination of nanogramamounts of total and oxidized glutathione. Application to mammalian bloodand other tissues. Anal. Biochem., 27: 502-522, 1969.Newton, G. L., Dorian, R., and Fahey, R. C. Derivatization of monobro-mobimane and separation by reverse-phase high performance liquid chro-matography. Anal. Biochem., 114: 383-387, 1981.Rice, G. C., Bump, E. A., Shrieve, D. C., Lee, W., and Kovacs, M. Quantitative analysis of cellular glutathione by flow cytometry utilizing monochlo-robimane: some applications to radiation and drug resistance in vitro and invivo. Cancer Res., 46: 6105-6110, 1986.Park, J. G., Oie, H. K., Sugarbaker. P. H., Henslee, J. G., Chen, T. R.,Johnson, B. E.. and Gazdar, A. Characteristics of cell lines established fromhuman colorectal carcinoma. Cancer Res., 47:6710-6718, 1987.Dickman, P. S., Liotta. L. A., and Triche. T. J. Ewing's sarcoma: character

ization in established cultures and evidence of its histogenesis. Lab. Invest.,¥7:375-382, 1982.Cook, J. A., Pass, H. I., Russo, A., lype, S. N., and Mitchell, J. B. Use ofmonochlorobimane for glutathione measurements in hamster and humantumor cell lines. Int. J. Radiât.Oncol. Biol. Phys., 16: 1321-1324, 1989.Bradford, M. M. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye binding.Anal. Biochem., 72:248-254, 1976.Cook, J. A., lype, S. N., and Mitchell, J. B. Differential specificity ofmonochlorobimane for isozymes of human and rodent glutathione S-trans-ferases. Cancer Res., 51: 1606-1612, 1991.Salzman, G. C., Mullaney, P. F., and Price, B. J. Light scatter approachesto cell characterization. In: M. R. Melaned, P. F. Mulaney, and M. L.

4293

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GSH AND THIOL IN LUNG TUMORS AND TISSUE

Mendelsohn (eds.). Flow Cytometry and Sorting, pp. 105-124. New York:John Wiley & Sons. Inc., 1979.

28. Young, R. C. Mechanisms to improve chemotherapy effectiveness. Cancer(Phila.), 65:815-822. 1990.

29. Mitchell, J. B., and Russo. A. The role of glutathione in radiation and druginduced cytotoxicity. Br. J. Cancer, 55: 96-104. 1987.

30. Ensley, J. F., Maciorowski. Z., Pietraszkiewicz, H.. Klemic, G.. KuKuruga.M., Sapareto, S.. Corbett, T.. and Crissman, J. Solid tumor preparation forflow cytometry using a standard murine model. Cytometry. 8: 479-487.1987. '

31. Allalunis-Turner, M. J., and Siemann, D. \V. Recovery of cell subpopulationsfrom human tumor xenografts following dissociation with different enzymes.Br. J. Cancer, 54: 615-622, 1986.

32. Hagenfeldt, L.. Arvidsson, A., and Larsson, A. Glutathione and 7-glutamyl-cysteine in whole blood, plasma and erythrocytes. Clin. Chim. Acta, S5:167-

173. 1978.33. Post, G. B., Keller, D. A.. Conner, D. A., and Menzel, D. B. Effect of culture

conditions on glutathione content in A459 cells. Biochem. Biophy. Res.Commun., 114: 737-742. 1983.

34. Singh, S. V., Hague, A. K., Ahmad, H., Medh, R. D., and Awasthi, Y. C.Glutathione .S'-lransferase isoenzymes in human lung tumors. Carcinogenesis

(Lond.), 9: 1681-1685, 1988.

35. Hulbert, P. B., and Yakuba, S. I. Monobromobimane, a fluorometric assayfor glutathione-5-transferase. J. Pharm. Pharmacol., 55:384-386, 1983.

4294

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

1991;51:4287-4294. Cancer Res John A. Cook, Harvey I. Pass, Susan N. Iype, et al. Resected Human Lung Tumor and Normal Lung TissueCellular Glutathione and Thiol Measurements from Surgically

Updated version

http://cancerres.aacrjournals.org/content/51/16/4287

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/51/16/4287To request permission to re-use all or part of this article, use this link

on July 9, 2021. © 1991 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/content/51/16/4287http://cancerres.aacrjournals.org/cgi/alertsmailto:pubs@aacr.orghttp://cancerres.aacrjournals.org/content/51/16/4287http://cancerres.aacrjournals.org/