ui

nwi

p

scap

T

dmtir

Clinical ImmunologyVol. 96, No. 3, September, pp. 252–263, 2000doi:10.1006/clim.2000.4904, available online at http://www.idealibrary.com on

Acidic pH Inhibits Non-MHC-Restricted Killer Cell Functions

Bianca Fischer,1 Bernd Muller, Karl-Georg Fischer,* Nicole Baur, and Werner Kreutz

Institut fur Biophysik und Strahlenbiologie, Albert-Ludwigs-Universitat Freiburg, Albertstrasse 23, D-79104 Freiburg i. Br., Germany;

and *Medizinische Universitatsklinik Freiburg, Hugstetter Strasse 55, D-79106 Freiburg, Germany

ismtsdhvc

Immunotherapeutic strategies in advanced stages ofsolid tumors have generally met with little success. Var-ious mechanisms have been discussed permitting theescape of tumor cells from an effective antitumoral im-mune response. Solid tumors are known to develop re-gions with acidic interstitial pH. In a recent study per-formed in the human system, we were able todemonstrate that non-MHC-restricted cytotoxicity is in-hibited by an acidic microenvironment. To get more in-sight into the mechanisms leading to this reduced cyto-toxic activity, we have now investigated the influence ofan acidic extracellular pH (pHe) on the killing process indetail. Unstimulated PBMC and LAK cells were used aseffector cells. Both populations are able to kill tumorcells in a MHC-independent manner via perforin/gran-zymes or TNFa, whereas only IL-2-activated cells can

se the killing pathway via Fas/FasL. We studied thenfluence of a declining pHe on the different killing path-

ways against TNFa-sensitive and -resistant, as well asFas-positive and -negative, target cells. Experiments inthe absence of extracellular Ca21 were used to discrimi-

ate the Ca21-dependent perforin-mediated killing. Heree show that the release of perforin/granzyme-contain-

ng granules, the secretion of TNFa, and also the cyto-toxic action of Fas/FasL interaction or of membrane-bound TNFa were considerably inhibited by declining

He. Furthermore, the secretion of the activating cyto-kine IFNg, as well as the release of the down-regulatingcytokines IL-10 and TGF-b1, was strictly influenced byurrounding pH. As a pHe of 5.8 resulted in a nearlyomplete loss of cytotoxic effector cell functions withoutffecting their viability, we investigated the influence ofHe on basic cellular functions, e.g., mitochondrial activ-

ity and regulation of intracellular pH. We found an in-creasing inhibition of both functions with declining pHe.

herefore, an acidic pHe obviously impairs fundamentalcellular regulation, which finally prevents the killingprocess. In summary, our data show a strict pHe depen-

ence of various killer cell functions. Thus, an acidicicroenvironment within solid tumors may contribute

o the observed immunosuppression in vivo, compromis-ng antitumoral defense and immunotherapy in general,espectively. © 2000 Academic Press

1 To whom correspondence should be addressed. Fax: 0761-203-

5390.

1521-6616/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

252

Key Words: acidic tumor pH; killer cells; cytokines;respiration; pHi regulation.

INTRODUCTION

Both MHC-unrestricted NK2 cells and MHC-re-stricted cytotoxic T cells play an important role inantitumor defense in vivo. In growing solid tumors annsufficient vascularization often leads to poor oxygenupply and, as a consequence, to an acidification of theicroenvironment. Therefore, interstitial pH within

umors can be considerably lower than in normal tis-ues (1–5). In this context we recently were able toemonstrate that non-MHC-restricted cytotoxicity ofuman NK cells as well as of LAK cells against aariety of tumor cells is inhibited by an acidic extra-ellular pH (pHe) (6, 7). Recognition and concomitant

binding of effector cells to target cells by adhesionmolecules is an initial event in the multistep killingprocess (8, 9). Yet, the expression of different adhesionmolecules was not affected by acidic pHe (7). In thepresent study we focused on subsequent steps involvedin the killing process and their potential inhibition bylow pHe. At least three different mechanisms contrib-ute to the delivery of the lethal hit of activated NK orLAK cells: granule exocytosis and release of perforin/granzymes, killing mediated via Fas/FasL interaction,or soluble as well as membrane-bound TNFa (10–12).To elucidate the influence of an acidic pHe on the threedifferent killing pathways, we used the combination oftwo different effector cell populations with differenttarget cell lines. For example, in contrast to LAK cells,unstimulated NK cells are not able to kill via theFas/FasL pathway as they express FasL on the cellsurface only after activation (13). Furthermore, to dis-criminate between secretory and nonsecretory killingpathways, chelation of extracellular Ca21 by EGTA wasused, which abrogates the release of cytolytic granulesas well as the pore formation of perforin (11, 13–16).

2 Abbreviations used: E:T ratio, effector:target ratio; FL1,-2,-3,relative fluorescence intensity in channel 1, 2, 3; LAK cell, lympho-kine-activated killer cell; NK cell, natural killer cell; PBMC, periph-

eral blood mononuclear cells; pHe, pHi, extra-, intracellular pH.

t

3

n

cttc

mramfiucprladc

253pH-DEPENDENT TUMOR CELL KILLING

Whereas cell-mediated cytotoxicity via perforin orFasL can be observed within a few hours, TNFa exertsits cytotoxic function only after longer incubation peri-ods (11). Therefore, we used the TNFa-sensitive cellline U937 as a target to compare short-term with long-term cytotoxic assays. Apart from its tumoricidal ac-tivity, TNFa plays an important role for the enhance-ment of effector cell functions, especially in synergisticaction with other cytokines like IFNg or IL-12 (16, 18,19). In contrast, IL-10 or TGF-b1 are thought to exertcounteracting effects (20–22). To test the involvementof these cytokines, we analyzed their production duringthe cocultures of effector cells with target cells at de-clining pHe. As the observed suppression of the cyto-oxic response of NK and LAK cells by acidic pHe

seemed to be a general phenomenon, the question ofwhether basic cellular functions would be inhibited aswell arose. In further experiments we thus focused onthe influence of an acidic pHe on the metabolic activityof mitochondria as well as on intracellular pH regula-tion.

METHODS

Target Cells

The following human tumor cell lines were pur-chased from DSM (Braunschweig, FRG): K562 (eryth-roleukemia), Daudi (B lymphoblastoid), Raji (B lym-phoblastoid), U937 (histiocytic lymphoma). Thedifferent cell lines were cultured in Hepes-buffered,endotoxin-free RPMI 1640 (Sigma, Deisenhofen, FRG)supplemented with 2 mM L-glutamine (Biochrom, Ber-lin, FRG), 1% NEAE (Biochrom), 10% heat-inactivated,endotoxin-free fetal calf serum (Biochrom), 100 U/mlpenicillin, and 50 mg/ml streptomycin (Biochrom) at7°C and 5% CO2. Only mycoplasma-free cell cultures

were used; this was frequently tested using a specificdetection kit (Boehringer, Mannheim, FRG).

Isolation and Expansion of Effector Cells

Heparinized blood samples were obtained fromhealthy human donors. Blood was diluted 1:2 withphosphate-buffered saline (PBS); mononuclear cells(PBMC) were collected using a Ficoll–Hypaque gradi-ent (density 1.077 g/ml; Pharmacia, Freiburg, FRG).Cells were washed five times with PBS, counted,checked for viability by trypan blue exclusion, andresuspended in RPMI 1640. Depletion of monocyteswas performed for 1 h by plastic adherence in culturedishes precoated with human serum (2 3 106 cells/ml).Activated effector cells were generated by stimulationof the nonadherent cells with rIL-2 (100 U/ml; Becton–Dickinson (BD), Heidelberg, FRG) for 4 days. Unstimu-

lated PBMC were cultured in standard medium alone. e

Flow Cytometric Assay for Cell-Mediated Cytotoxicity

Target cell labeling. 3,39-Dioctadecyloxacarbocya-ine perchlorate (DIOC18 (3) (“DIO”); 2.5 mg/ml in

DMSO; Molecular Probes, Eugene, OR) was added di-rectly to K562 or U937 (1.5 3 105 cells/ml), Daudi, orRaji cells (2.5 3 105 cells/ml) in a final concentration of10 mg/ml. Labeling was performed overnight understandard culture conditions.

Coculture. The next day, target cells were washedtwice in culture medium to remove free label. Aliquotsof culture media were adjusted to pH 5.8, 6.3, 6.8, or7.2 by addition of aqueous HCl or NaOH solution (1 N)under continuous control with highly sensitive micro-electrodes. For media with pH below 7.2, RPMI 1640without bicarbonate (Sigma) was used. A total of 1 3105/ml prelabeled target cells were resuspended in me-dium with different pH, washed, and dispensed in 96-well microtiter plates (BD; 100 ml/well). Effector cellswere also resuspended in medium of the selected pHvalue (5 3 106/ml). Aliquots of 100 or 20 ml of theeffector cell suspensions were added to the target cellcultures to yield the final effector/target ratios of 50:1and 10:1. As a control, medium with different pH andwithout effector cells was added to the target cell cul-tures. Cocultures for cytotoxic assays under Ca21-freeonditions were performed with culture medium con-aining 2 mM EGTA (Sigma) at the different pH men-ioned above. All cocultures were incubated in a finalulture volume of 200 ml for 4 h. Remeasurements of

pH with a microelectrode before and after this incuba-tion period revealed no change of the initially chosenpHe. For flow cytometric analysis, cells of each wellwere resuspended and transferred into round-bottomtubes (BD).

Analysis. Analysis of cell-mediated cytotoxicitywas performed on an Epics XL-MCL flow cytometer(Coulter, Krefeld, FRG) equipped with a 15-mW argonion laser at an excitation wavelength of 488 nm aspreviously described (6, 7). In brief, before measure-ment propidium iodide (PI, final concentration: 20 mg/

l; Sigma) was added to the samples. The DIO fluo-escence was recorded in the green channel (FL1) using

525-nm bandpass filter, while PI fluorescence waseasured in the red channel (FL3; 620-nm longpass

lter). Data acquisition and analysis were performedsing the XL-System II software (Coulter). Fluores-ence was recorded on a logarithmic scale without com-ensation of interchannel crosstalk, while forward andight angle scatter characteristics were recorded on ainear scale. Calculation of specific lysis (in percent-ges) was performed by subtracting the nonspecific celleath of control samples (in percentages) from the per-entage lysis of target cells in experimental samples at

ach pH value; analysis included only DIO1 (FL11) and

254 FISCHER ET AL.

PI1 (FL31) double-positive events:

% specific lysis 5 % [DIO1PI1]exp. 2 % [DIO1PI1]control.

Analysis of CD95 (Fas/Apo-1 Antigen) orIntracellular Perforin Expression by FlowCytometry

For the analysis of CD95 expression on the cell sur-face of target cell populations the monoclonal antibodyanti-CD95 (Pharmingen, Hamburg, FRG) was used;this unlabeled antibody was detected with PE-conju-gated rabbit anti-mouse Ig (Dako, Hamburg, FRG).Unspecific IgG1 (Dako) was used as isotype control.After blocking unspecific binding of antibodies to var-ious FcR with human serum (Behring, Marburg, FRG),cells were stained according to standard protocols. Foranalysis of perforin expression in effector cells, themonoclonal antibody anti-perforin-FITC (dG9; HolzelDiagnostika, Koln, FRG) was used. Unspecific IgG2b–FITC (Sigma) served as isotype control. As perforin isan intracellular antigen, samples were pretreated with2% paraformaldehyde in PBS. Blocking of FcR andstaining of the cells was performed with standard so-lutions containing 0.1% saponin. Staining was ana-lyzed using an Epics XL-MCL flow cytometer (Coulter)as described above: FITC fluorescence was recorded influorescence channel 1 (FL1; 525-nm bandpass filter)and PE fluorescence in fluorescence channel 2 (FL2;575 nm bandpass filter) after compensation of inter-channel crosstalk.

Measurement of pHi by Flow Cytometry

To determine the influence of pHe on pHi during a4-h incubation period reflecting the conditions of thecytotoxic assays, we measured the pHi of target andeffector cells separately using the fluorescent pH indi-cator BCECF (23–25).

Target cell labeling. K562, Daudi, and Raji cells aswell as unstimulated PBMC or LAK cells were har-vested and washed twice in RPMI 1640 without bicar-bonate. Cells were resuspended (5 3 105 cells/tube) andloaded with the acetoxymethyl ester of BCECF (29,79-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein ace-toxymethyl ester; Molecular Probes; final concentra-tion 1 mM) for 30 min at 37°C in the dark. Thelipophilic BCECF-AM can penetrate the plasma mem-brane, whereafter the ester is cleaved by cytoplasmicesterases resulting in trapping of the free fluorescentdye. After being stained, one-half of the probes wasused to perform the calibration of fluorescence mea-surements, whereas the others were used to determine

the influence of surrounding pH on pHi. The probes

used for calibration were washed twice in high [K1]-buffer (30 mM NaCl, 120 mM KCl, 1 mM CaCl2, 0.5mM MgSO4, 1 mM Na2HPO4, 5 mM glucose; all fromMerck, Darmstadt, FRG; 10 mM Hepes, 10 mM Pipes;Sigma) with different pHe (5.8–7.2). Then, they wereincubated for 10 min at 37°C with the ionophore ni-gericin (Molecular Probes; final concentration 10 mM)in the same buffer samples according to the method ofCook and Fox (23). Under these conditions, pHi isequalized to pHe. For each data point, analysis of threeseparate samples has been performed. The samples fordetermination of the influence of pHe on pHi werewashed twice with RPMI 1640 without bicarbonateadjusted to different pH (5.8–7.2) and then incubatedfor another 4 h in the same medium at 37°C in thedark. For each data point three separate probes wereanalyzed.

Analysis. BCECF fluorescence of the samples usedfor calibration was analyzed immediately in an EpicsXL-MCL flow cytometer using two different emissionwavelengths with FL1 of 520 nm and FL3 of 670 nm.Each mean fluorescence was obtained by computeranalysis of the collected histograms of each probe (XL-System II software). Means of emission ratios (FL1/FL3) of the three corresponding samples were calcu-lated and calibration curves were fitted using least-squares regression analysis. Calculation of fluorescenceratios renders the measurements independent of theamount of BCECF present within the cells. Fluorescenceintensity of intracellular BCECF of the actual samples(minimum of 104 cells/probe) was measured by flow cy-tometry and pHi was calculated using the means of theemission ratios (FL1/FL3) and the cell-type-specific cali-bration curve.

Determination of Cytokine Secretion during theCytotoxic Assay

To analyze cytokine secretion during the killing pro-cess, cocultures of four different target cell lines withunstimulated PBMC or IL-2-activated LAK cells wereperformed as described above for the cytotoxic assay.Only E:T ratios of 50:1 were used. After 4 h of incuba-tion, the cell-free supernatants obtained by centrifuga-tion were stored (280°C). Analysis was performed us-ing immunoassays for the following cytokines: TNFa,IFNg, IL-2, IL-7, IL-10, IL-12 (Coulter Immunotech,Hamburg, FRG), and TGF-b1 (R&D Systems, Wiesba-den, FRG). The spectrophotometric absorbance of thesamples was measured at a wavelength of 405 or 450nm as indicated using a microtiter plate (ELISA)reader (Dynatech, Denkendorf, FRG). Data acquisitionand analysis were performed using BioLinx 2.1 soft-ware (Dynatech). All cytokine assays were calibrated

against WHO cytokine standards.

Isg

255pH-DEPENDENT TUMOR CELL KILLING

Determination of Mitochondrial Activity Using aColorimetric Assay

The nonradioactive, colorimetric assay system usingXTT (sodium 39-[1-(phenylaminocarbonyl)-3,4-tetrazo-lium]-bis(4-methoxy-6-nitro) benzene sulfonic acid hy-drate) (Boehringer) was used to evaluate the influenceof acidic pH on the mitochondrial activity of effectorcells. The mitochondrial “succinate-tetrazolium reduc-tase” system cleaves the tetrazolium salt XTT toformazan. The amount of this dye correlates directly tothe number of metabolically active cells, as the enzymesystem is active only in respirating and viable cells.For this purpose, unstimulated PBMC or IL-2-acti-vated LAK cells were harvested, washed twice in pH-adjusted culture medium (pH 5.8–7.2), and resus-pended in medium with the same pH (5.8–7.2) to yielda density of 5 3 106/ml. A total of 100 ml of the effectorcell suspensions/well was seeded into 96-well microti-ter plates and incubated for 4 h. After the addition of50 ml of XTT labeling mixture to each well, the plateswere incubated for an additional 4-h period. The spec-trophotometric absorbance of the samples was mea-sured at a wavelength of 490 nm using a microtiterplate (ELISA) reader (Dynatech) with a wavelength of630 nm as reference. Data acquisition and analysiswere performed using BioLinx 2.1 software (Dyna-tech).

RESULTS

Recently we were able to demonstrate an overallinhibition of the non-MHC-restricted cytotoxicity ofhuman unstimulated PBMC and LAK cells with de-clining pH in the surrounding microenvironment (6, 7).The observed inhibition was independent of the targetcells used. Combining different effector cell popula-tions with different target cell lines, we now were in-terested in the influence of pHe on single mechanismsof the killing process itself. Three major pathways con-tribute to cell-mediated cytotoxicity: perforin/gran-zymes via granule exocytosis, Fas/FasL interaction,and TNFa. First, we had to clarify whether both effec-tor cell populations are able to use the perforin path-way.

Expression of Perforin in Different Effector CellPopulations

Flow cytometric analysis of unstimulated PBMC andLAK cells revealed perforin expression in both effectorcell populations (Fig. 1). Upon stimulation with IL-2for 4 days only a slight increase in the number ofperforin-expressing cells could be detected (26 versus

20%). However, in a number of separate experiments a

distinct rise in the amount of perforin expressed percell could be observed (relative mean fluorescence in-tensity 1501 versus 1191, respectively).

Influence of pHe on Perforin/Granzyme- or FasL-Mediated Short-Term Cytotoxicity

To investigate the influence of extracellular pH ondifferent killing mechanisms, we performed short-termcytotoxic assays against a panel of target cell linesunder different conditions. Short-term tumor cell lysismainly results from the release of perforin-containingcytolytic granules and FasL-mediated cytotoxicity,whereas TNFa-based killing is thought to need longerincubation periods (10, 11). As shown in Fig. 2a, adeclining pHe always reduced the cytotoxic activity ofunstimulated PBMC as well as of LAK cells. Controlsamples measured by the same flow cytometric methodrevealed that the viability of both tumor and effectorcells was not affected by low pHe. Effector cell viabilitywas $97%, target cell viability was $93% independentof the respective pHe. Both effector cell populations areable to kill via the perforin/granzyme pathway, but

FIG. 1. Flow cytometric analysis of intracellular perforin in (a)unstimulated PBMC and (b) IL-2-activated LAK cells. Freshly iso-lated and monocyte-depleted PBMC from blood samples of differenthealthy donors (1 3 106/ml) were cultured in medium with or withoutL-2 (100 U/ml). After 4 days cells were harvested, washed, andtained with FITC-labeled antibodies against perforin. Each histo-ram was obtained from 2.5 3 104 cells (filled plot, isotype control;

dark line, anti-perforin mAb). Representative data of three experi-ments; for each experiment cells of one single donor were used.

only IL-2-activated effector cells could exert Fas/FasL-

E r cells of one single donor for every experiment.

256 FISCHER ET AL.

mediated target cell lysis (13). Our analysis of CD95revealed that in contrast to Daudi, Raji, or U937 cells,the K562 cell line did not express the Fas antigen(CD95) on the cell surface (Fig. 3), its lysis thus beingalmost exclusively due to perforin. Our results clearlyindicate that tumor cell killing via the secretory per-forin/granzyme pathway alone (K562) as well as incombination with the FasL-mediated pathway (othertarget cells) is strictly dependent on pHe (Fig. 2a).

Since it is known that perforin/granzyme-mediatedkilling can only operate in the presence of extracellularCa21 (10, 11), we next tested the effect of Ca21 chelationwith EGTA on the cytotoxic activity of our two effectorcell populations. As only the perforin/granzyme-medi-ated pathway should be eliminated, the discriminatedcytotoxic activity would be due to other killing path-ways, i.e., FasL-mediated cytotoxicity. Using Ca21-freeconditions, we additionally tested whether a decliningpHe would also affect this perforin-independent cyto-toxicity. As can be seen in Fig. 2b, the elimination of

FIG. 2. Influence of extracellular pH (pHe) and of extracellular Ccell lines. Human monocyte-depleted lymphocytes were cultured withtwo effector cell populations with K562, Daudi, Raji, or U937 cells fcytotoxic activity of the effector cells was determined by flow cytom

GTA (2 mM). Data represent means 6 SEM (n 5 8) using effecto

perforin-mediated killing with EGTA (2 mM) strongly

FIG. 3. Expression of Fas (Apo-1/CD95) on different tumor tar-get cell lines analyzed by flow cytometry. Each histogram was ob-tained from 2.5 3 104 cells (filled plot, isotype control; dark line,

a21 on NK and LAK cell-mediated cytotoxicity against different tumor(1) or without (2) IL-2 (100 U/ml) for 4 days. After coculturing of these

or 4 h at the indicated pH using E:T ratios of 10:1 or 50:1, the specificetry. (a) Standard conditions; (b) chelation of extracellular Ca21 with

anti-Fas mAb). Representative data of three experiments.

ki

t

v

ltw

4caPonPwcb

on

Iop

257pH-DEPENDENT TUMOR CELL KILLING

reduced the cytolytic activity of both effector cell pop-ulations. Moreover, the killing capability of unstimu-lated PBMC was totally abrogated. These data supportfurther evidence that the unstimulated PBMC popula-tion exerts the cytotoxic activity only via the perforin/granzyme pathway and that this killing mechanism isstrictly dependent on pHe. In contrast, the cytotoxicactivity of LAK cells was not completely abrogated inthe absence of extracellular Ca21. Again the remaining

illing activity against all target cell lines was inhib-ted with declining pHe (Fig. 2b). Experiments per-

formed with higher EGTA concentrations (e.g., 4 mM)did not show different results. Control samples mea-sured by the same flow cytometric analysis revealedthat the viability of all target cell lines used was notinfluenced by EGTA.

Influence of pHe on the Secretion of TNFa

Although TNFa-mediated cytotoxicity is thought tooccur mainly after longer incubation periods, the re-lease of this cytokine is also felt to play an importantrole for the induction of NK and LAK cell cytotoxicityeven in short-term experiments (10, 16, 18, 19). Forthis reason it was of interest whether extracellular pHinfluences the secretion of TNFa. The supernatants ofeffector cells cocultured with various target cells atdifferent pHe were collected after 4 h and analyzed forhe production of TNFa. As can be seen in Fig. 4a,

unstimulated PBMC as well as IL-2-activated LAKcells secreted considerable amounts of TNFa alreadywithin 4 h. The standard recombinant TNFa used inthe ELISA was also dissolved in medium with identicalpH as the sample culture supernatants. The resultsshowed that TNFa was stable in the pH range used,from 7.2 to 5.8. There were no differences in the detec-tion of TNFa due to acidic pHe. In two additional ex-periments, we observed that both effector cell popula-tions alone constitutively secreted TNFa in a pH-dependent manner without cocultured target cells(data not shown). Unstimulated PBMC secreted moreTNFa than IL-2-activated effectors, namely 257 pg/mlersus 71 pg/ml at pHe 7.2, respectively. The slightly

higher concentrations of TNFa in cultures with un-stimulated PBMC could be attributed to more remain-ing monocytes within these populations. No TNFacould be detected within culture supernatants of thedifferent target cell lines alone. Apart from cocultureswith U937 cells, we found the highest amounts ofTNFa with Daudi cells, which are known to be TNFaresistant. A strong inhibition of TNFa release couldalready be detected at pHe 6.8 in both effector cellpopulations independently of the target cell line used.Thus, an acidic pHe in the microenvironment impairsTNFa secretion dramatically. In further experiments

we looked for TNFa secretion in cocultures under Ca21-

free conditions. As can be seen in Fig. 4b, the release ofTNFa can be abrogated by chelation of extracellularCa21. As TNFa-mediated cytotoxicity seems to be slowacting, we also performed cytotoxic assays over 24 husing the highly TNFa-sensitive U937 as target cells.In addition, the secretion of TNFa in these cocultureswas analyzed. Our data depicted in Figs. 5a and 5dshow a considerable rise in the killing activity of un-stimulated PBMC and LAK cells within 24 h, althoughthe secretion of TNFa did not further increase within aprolonged incubation period (compare Figs. 5c and 5f).As there were no significant differences in TNFa re-ease of unstimulated or IL-2-activated effector cells,he increased lysis of U937 cells after 24 h by LAK cellsas not due to the action of soluble TNFa. Both the

specific lysis of U937 cells and the release of TNFa bythe effector cells remained strictly dependent on pHe.

Chelation of extracellular Ca21, which abrogates per-forin-mediated killing and TNFa secretion (see Fig.b), again led to a marked reduction of cell-mediatedytotoxicity (Figs. 5b and 5e). However, there was onlysmall effect on the killing activity of unstimulated

BMC within 24 h (Figs. 5d and 5e), indicating thatther killing mechanisms were involved. As there wereo differences in the cytotoxic activity of unstimulatedBMC and LAK cells when secretory mechanismsere inhibited by EGTA, and as only IL-2-activated

ells can kill Fas-positive U937 cells via FasL, it muste assumed that membrane-bound TNFa was respon-

sible for the observed killing of U937 cells. Neverthe-less, again an increasing suppression of the Ca21-inde-pendent killing pathways by declining pHe could bebserved. Taken together, even after 24 h it could beoted that the secretion of TNFa was strictly depen-

dent on pHe, being almost completely inhibited atpHe 6.8.

A summary of our findings concerning the inhibitionof MHC-unrestricted cytotoxicity of both NK and LAKcells by acidic pHe is given in Table 1.

Influence of pHe on the Release of Other Cytokines

Apart from TNFa, some other cytokines are sup-posed to play an important role in NK and LAK cell-mediated cytotoxicity (11, 18, 19). To address the ques-tion of whether activating cytokines like IFN-g, IL-2,L-7, and IL-12 are involved in short-term cytotoxicityr whether their production is influenced by decliningHe, we also analyzed their secretion after cocultures

of effector cells with different target cells. As can beseen in Fig. 6a, there was no IFN-g detectable in co-cultures with unstimulated PBMC. In contrast, IL-2-activated effector cells produced this cytokine in a tar-get cell type-independent manner. Yet, the observedIFN-g release was inhibited with declining pHe. As a

declining pHe always inhibited the cytotoxic process,

se

c

E ef

258 FISCHER ET AL.

we additionally analyzed the release of the down-reg-ulating cytokines IL-10 and TGF-b1 (20–22). The se-cretion of both cytokines was inhibited with decliningpHe (Figs. 6b and 6c). Thus, IL-10 and TGF-b1 couldnot be responsible for the observed suppression of cell-mediated cytotoxicity by acidic pHe. Finally, we couldnot detect any IL-2, IL-7, or IL-12 in short term cocul-tures of the two effector cell populations with differenttarget cell lines (data not shown).

Influence of pHe on Mitochondrial Activity of EffectorCells

As we observed a general suppression of effector cellfunctions by an acidic pH of the microenvironment, thequestion of whether basic cellular functions were influ-enced by declining pHe arose. Using a XTT-based as-ay, we first looked for the mitochondrial activity of ourffector cells at different pHe. As can be seen in Fig. 7,

FIG. 4. Influence of extracellular pH (pHe) and of extracellular CLAK cells. Human monocyte-depleted lymphocytes were cultured wittwo effector cell populations with K562, Daudi, Raji, or U937 cells forollected and analyzed for soluble TNFa by immunoassay. The de

cocultures under standard conditions; means 6 SEM (n 5 4). (b) SGTA (2 mM); means 6 SEM (n 5 2). Data represent means using

acidification of the microenvironment induced a

marked inhibition of the succinate-tetrazolium reduc-tase system in mitochondria of unstimulated PBMC aswell as of IL-2-activated cells already within a 4-hincubation period. Concerning this inhibition, therewere no substantial differences between the two effec-tor cell populations, both showing a decrease in enzy-matic activity of 50% at the lowest pHe of 5.8. As theenzyme system which cleaves the tetrazolium salt XTTto formazan belongs to the respiratory chain of themitochondria, our data demonstrate that the respira-tion capacity and therefore the energy supply of leuko-cytes is clearly reduced within an acidic microenviron-ment.

Influence of pHe on the Regulation of Intracellular pH

As the intracellular pH is known to be a criticaldeterminant of cellular functions, whether a decliningpHe induces considerable changes in pHi was of further

on target cell-induced TNFa secretion of unstimulated PBMC and) or without (2) IL-2 (100 U/ml) for 4 days. After coculturing of theseat the indicated pH using E:T ratios of 50:1, the supernatants were

tion limit of the test system was 10 pg/ml. (a) Secreted TNFa ineted TNFa in cocultures after chelation of extracellular Ca21 withfector cells of one single donor for every experiment.

a21

h (14 htececr

interest. For this purpose we used the pH-sensitive

uie

259pH-DEPENDENT TUMOR CELL KILLING

fluorescent dye BCECF to determine pHi in differentcell types after a 4-h incubation period in medium withdecreasing pHe. As shown in Fig. 8c, we found thatunstimulated PBMC exhibited a pHi of 7.06 6 0.06 atphysiological pHe of 7.2. Up to pHe 6.8, pHi remainednearly constant at 7.0. The corresponding LAK cellsrevealed pHi values which were shifted to slightly moreacidic pHi between 6.76 6 0.07 and 6.86 6 0.10. Yet,the stimulated cells also maintained a pHi plateau upto external pH values of 6.8. When pHe dropped tomore acidic values, an ongoing acidification of pHi inboth effector cell populations strictly depending on thepH of the microenvironment was observed. As tumorcells are thought to be adapted to an acidic microenvi-ronment, we were also interested in pHi regulation ofthe target cells. With regard to pHi homeostasis, tumorcells reacted similarly to the effector cells (Fig. 8d). Upto extracellular pH values of 6.8, we measured pHi

values between 6.9 and 7.2. With the exception of Rajicells, which seemed to regulate their pHi more effec-tively up to pHe of 6.3, at acidic pHe values below 6.8,

FIG. 5. Influence of extracellular pH (pHe) and of extracellularagainst the TNFa-sensitive U937 cell line. Human monocyte-depleteU/ml) for 4 days. After coculturing of these two effector cell populati10:1 or 50:1, the specific cytotoxic activity was determined by flow cytCa21 with EGTA (2 mM) (b and e). Influence of extracellular pH

nstimulated PBMC and LAK cells (c and f). For determination of secn (a and d) were collected and analyzed by immunoassay. Data reprvery experiment.

pHi decreased continuously in the cell lines tested.

DISCUSSION

Considerable evidence has accumulated that tumor-infiltrating leukocytes are functionally impaired, butthe mechanisms responsible for this suppression arenot entirely clear (26, 27). In comparison with normaltissues, solid tumors often possess regions with anacidic microenvironment. In a recent report we wereable to demonstrate that an acidic extracellular pHinhibits the cell-mediated cytotoxicity of human NKand LAK cells (7). Thus, interstitial pH seems to be oneimportant factor contributing to the impairment of im-mune functions in cancer. To gain insight into themechanisms of this immunosuppression, we have nowanalyzed the influence of an acidic pHe on the cytotoxicprocess in detail. The killing cascade may include threepathways: [1] release of perforin/granzyme-containinggranules, [2] Fas/FasL interaction, or [3] secretion ofTNFa.

Using K562 as target cells and unstimulated PBMCas effector cells revealed the strict pHe dependence ofthe natural occuring cytotoxic activity. After activation

a21 on short- or long-term NK and LAK cell-mediated cytotoxicitymphocytes were cultured without stimulus (2) or with (1) IL-2 (100with U937 cells for 4 or 24 h at the indicated pH and E:T ratios of

etry: standard conditions (a and d), or after chelation of extracellularHe) on short- or long-term target-cell-induced TNFa secretion ofed TNFa the supernatants of the cocultures with an E:T ratio of 50:1nt means 6 SEM (n 5 2) using effector cells of one single donor for

Cd lyonsom

(pretese

of PBMC with IL-2 for 3–4 days, more powerful effec-

obtiPopC

c

e

D

wT

not

260 FISCHER ET AL.

tor cells were generated, also being able to lyse NK-resistant target cells like Daudi, Raji, or U937 (see Fig.2a). Again, the cytotoxic activity of these stimulatedeffector cells was inhibited by declining pHe. As K562cells do not express Fas and are described to be resis-tant against soluble TNFa (28, 29), using NK cells,nly perforin/granzymes remain as killing pathway. Itecame obvious that perforin-mediated killing is con-inuously inhibited with declining pHe. Our results aren agreement with findings of Young et al. (30) orersechini et al. (31), respectively, who found a pHptimum of 7.4 for the release and the insertion oferforin into biomembranes. Chelation of extracellulara21, which abrogates perforin-mediated killing, re-

sulted in a considerable reduction of cell-mediated cy-totoxicity against all target cells (see Fig. 2b). Regard-ing K562 cells, NK cell-mediated killing could almostbe completely prevented. The remaining killing activ-ity of LAK cells could be mediated by membrane-boundTNFa, which is known to be expressed by these effectorcells (10). K562 cells were described to be sensitive forthe attack of membrane-bound, but not of soluble,TNFa already within short-term experiments (28). Inontrast to soluble TNFa, for which both the secretion

and the mechanism of action is inhibited in the absenceof Ca21 (Figs. 4 and 5; (32)), killing via mTNFa seemsto be independent of Ca21 (Figs. 2 and 5; (16)). How-ver, our results show that this mTNFa–induced cyto-

toxicity was also strictly dependent on pHe.In contrast to K562 cells, Fas could be detected on

TABInhibition of Different Killing Pathways of N

Effectorcells Killing mechanism

K562

Perf.sens.Yes

Fasexpr.No

mTNFasens.Yes

PersenYe

Ca21 in mediuma

NK Perforin with 1d pH dep. pH dwithout 2 n.e. n.e

Fas ligandb with 2 n.e.without 2 n.e.

mTNFac with 2 n.e.without 2 n.e.

LAK Perforin with 1 pH dep. pH dwithout 2 n.e. n.e

Fas ligand with 1 n.e.without 1 n.e.

mTNFa with 1 pH dep.without 1 pH dep.

Note. Perf. sens., perforin sensitive; Fas expr., Fas expression; mTNa With, physiological Ca21 concentration; without, Ca21 chelationb NK cells do not express Fas ligand (13).c Soluble TNFa does not play a cytotoxic role at pH below 7.2 (seed 1, Potential killing mechanism; 2, not present (not existent or

audi, Raji, and U937 target cells, which therefore

could also be lysed via the Fas/FasL pathway by FasL-expressing LAK cells. Unstimulated NK cells do notexpress FasL on the cell surface (13). In the absence ofextracellular Ca21, only the Fas/FasL-mediated path-

ay remains to kill Daudi cells, as they are resistant toNFa-mediated lysis even via the membrane-bound

molecule (28, 29). Our findings show that lysis of Daudicells by LAK cells via Fas/FasL interaction is alsoclearly dependent on pHe.

Unlike Daudi cells, Raji and U937 target cells areadditionally sensitive to TNFa-mediated lysis. Al-though we could show that NK cells as well as LAKcells are able to secrete considerable amounts of TNFaalready within 4 h, we did not observe a relevant lysisof Raji and U937 cells by unstimulated NK cells. Thesefindings are in agreement with other reports, whichdescribe that long-term periods are necessary for thelytic action of soluble TNFa (10, 11). Furthermore, assecretion of TNFa always is inhibited by declining pHe,soluble TNFa does not play a significant role in killingunder these experimental conditions. Thus, lysis ofRaji and U937 cells after 4-h cocultures could only bedue to perforin/granzymes, membrane-bound TNFa,and, for LAK cells, Fas/FasL interaction. In the ab-sence of extracellular Ca21, we could show that thecombination of the two latter mechanisms is alsoclearly inhibited with declining pHe.

Analysis of the secretion of several cytokines showedthat the suppression of non-MHC-restricted cytotoxic-ity was not due to an increased synthesis of down-

1-MHC-Restricted Cytotoxicity by Acidic pHe

Targets

Daudi Raji U937

Fasexpr.Yes

mTNFasens.No

Perf.sens.Yes

Fasexpr.Yes

mTNFasens.Yes

Perf.sens.Yes

Fasexpr.Yes

mTNFasens.Yes

pH dep. pH dep.n.e. n.e.

n.e. n.e. n.e.n.e. n.e. n.e.

n.e. n.e. n.e.n.e. n.e. n.e.

pH dep. pH dep.n.e. n.e.

pH dep. pH dep. pH dep.pH dep. pH dep. pH dep.

n.e. pH dep. pH dep.n.e. pH dep. pH dep.

sens, mTNFa sensitive; pH dep., pHe dependent; n.e., not existent.h EGTA (2 mM).

gs. 4 and 5).working).

LEon

f.s.s

ep..

ep..

Fawit

Fi

regulating mediators like IL-10 or TGF-b1 with declin-

scm

rI

pn

amd(1dm

ifD

261pH-DEPENDENT TUMOR CELL KILLING

ing pHe (see Fig. 6). IFN-g could only be detected incocultures with IL-2-activated effector cells. Thesedata hint at the broader lytic potential of LAK cells, asIFN-g has been described to augment their cytotoxicactivity synergistically with TNFa (18, 19). However,our data also clearly show that the secretion of cyto-kines in target/effector cell cocultures is considerablyinhibited by declining pHe, independent of their stim-ulating or down-regulating properties. We concludethat the acidic pHe within the microenvironment ofolid tumors is responsible not only for the impairedytotoxicity of effector cells, but also for the docu-

FIG. 6. Influence of extracellular pH (pHe) on target-cell-inducedcytokine secretion of unstimulated PBMC and LAK cells. Humanmonocyte-depleted lymphocytes were cultured without stimulus (2)or with (1) IL-2 (100 U/ml) for 4 days. After coculturing of these twoeffector cell populations with K562, Daudi, or Raji cells for 4 h at theindicated pH using E:T ratios of 50:1, the supernatants were col-lected and analyzed for secreted (a) IFN-g, (b) IL-10, or (c) TGF-b1 bymmunoassays. The detection limit of the test systems was 1 IU/mlor IFN-g, 5 pg/ml for IL-10, and 10 pg/ml for TGF-b1, respectively.ata represent means of two representative experiments.

ented alterations in the cytokine profiles commonly e

seen in cancer tissues. Similarly, Rabinowich et al.eported a reduced cytokine gene expression for IL-2,L-4, and IFN-g in tumor-infiltrating lymphocytes of

ovarian carcinomas (33).Finally, as we observed an increasing inhibition of

important immunologic functions of our effector cellswith declining pHe in general, we looked for an influ-ence of acidic pHe on fundamental cellular features,e.g., metabolic activity and regulation of intracellularpH. Independent of the activation state of the effectorcells, we found a dramatic reduction of mitochondrialactivity with declining pHe (see Fig. 7). The suppres-sion of energy supply correlated with a growing acidi-fication of the interior of the cells (see Fig. 8c). Stimu-lated LAK cell populations always exhibited a slightlymore acidic pHi profile than unstimulated PBMC.However, at pHe values below 6.8, which are represen-tative for the interstitial pH of solid tumor tissues(3–5), the pHi of both effector cell populations de-creased below the homeostatic range. It could be as-sumed that acidic pHi may account for the impairedimmunological functions like target cell killing andcytokine release reported in this study. Regarding solidtumor cells, there still exist conflicting results abouthow these cells are able to survive within an acidicmicroenvironment. Different mechanisms like an alka-line shift of steady-state pHi, overexpression of theNa1/H1-exchanger, the main regulator of pHi at acidicpHe, and an acidic shift for the activity of this anti-

orter have been proposed to contribute to this phe-omenon (34–38). However, looking for the pHi ho-

meostasis of the tumor target cells used in this study,

FIG. 7. Influence of extracellular pH (pHe) on the mitochondrialctivity of unstimulated PBMC and LAK cells. Freshly isolated andonocyte-depleted PBMC from blood samples of different healthy

onors (1 3 106/ml) were cultured in medium without or with IL-2100 U/ml). After 4 days cells were harvested and recultured (5 305/well) for 4 h at the indicated pHe. The mitochondrial activity wasetermined using an XTT-based colorimetric assay. Data representeans 6 SEM (n 5 3) using effector cells of one single donor for

very experiment.

7

1

1

1

1

1

1

(

(

tp

m

262 FISCHER ET AL.

we could observe a range for steady-state pHi similar tothat of effector cells (see Fig. 8d). For K562 cells, anaverage pHi of 6.9–7.2 at pHe values of 6.8–7.6 hasbeen reported (25). Our data show that if extracellularpH declined below 6.8, the pHi of K562 and Daudi cellscould not further be regulated and decreased to moreacidic values. In contrast, Raji cells kept a pHi of 6.9–.2 until pHe values below 6.3, demonstrating more

active mechanisms against acidic stress. As suggestedfor solid tumor cells, Raji cells thus may be able toregulate their pHi even at lowest pHe, providing sub-stantial protection against chronic extracellular acidi-fication.

In summary, our data support evidence that anacidic pHe exerts a clear suppression of an effectiveantitumoral immune response like cell-mediated tu-mor cell killing. We postulate that the prevention of

FIG. 8. Influence of extracellular pH (pHe) on intracellular pHpHi) of different tumor target cell lines as well as of unstimulated

PBMC or LAK cells determined with the pH-sensitive dye BCECF byflow cytometric analysis. Leukocytes were isolated and cultured asdescribed. For determination of pHi, the different cell populationswere harvested and loaded with BCECF-AM (1 mM) for 30 min at37°C. (a) and (b) show the calibration curves estimated with high[K1]-buffer and nigericin for unstimulated PBMC (n 5 10) or K562n 5 10). In (c) and (d) the pHi of leukocytes (unstimulated PBMC

(n 5 10), IL-2-activated LAK cells (n 5 5)) as well as of tumorarget cells (K562 (n 5 6), Daudi (n 5 5), Raji (n 5 5)) at differentHe is depicted. Analysis of 104 cells/sample was performed after

reculturing the cells for 4 h at the indicated pHe. Data representeans 6 SEM of separate experiments.

non-MHC-restricted cytotoxicity is the consequence of

a severe impairment of basic cellular functions of ef-fector cells by acidic pHe within the microenvironmentof solid tumors.

ACKNOWLEDGMENTS

We thank K. Neubert and I. Baumle for skillful technical assis-tance.

REFERENCES

1. Wike-Hooley, J. L., Havemann, J., and Reinhold, H. S., Therelevance of tumour pH to the treatment of malignant disease.Radiother. Oncol. 2, 343–366, 1984.

2. Thistlethwaite, A. J., Leeper, D. B., Moylan D. J., III, and Ner-linger, R. E., pH distribution in human tumors. Int. J. Radiat.Oncol. Biol. Phys. 11, 1647–1652, 1985.

3. Helmlinger, G., Yuan, F., Dellian, M., and Jain, R., InterstitialpH and pO2 gradients in solid tumors in vivo: High-resolutionmeasurements reveal a lack of correlation. Nat. Med. 3, 177–182,1997.

4. Vaupel, P., Kallinowski, F., and Okunieff, P., Blood flow, oxygenand nutrient supply, and metabolic microenvironment of humantumors: A review. Cancer Res. 49, 6449–6465, 1989.

5. Young, P. R., and Spevacek, S. M., Substratum acidification andproteinase activation by murine B16F10 melanoma cultures.Biochim. Biophys. Acta 1182, 69–74, 1993.

6. Severin, T., Muller, B., Giese, G., Uhl, B., Wolf, B., Hauschildt,S., and Kreutz, W., pH-dependent LAK cell cytotoxicity. TumorBiol. 15, 304–310, 1994.

7. Fischer, B., Muller, B., Fisch, P., and Kreutz, W., An acidicmicroenvironment inhibits antitumoral non-MHC-restricted cy-totoxicity: Implications for the immunotherapy of cancer. J. Im-munother. 23, 196–207, 2000.

8. Robertson, M. J., Caligiuri, M. A., Manley, T. J., Levine, H., andRitz, J., Human natural killer cell adhesion molecules—Differ-ential expression after activation and participation in cytolysis.J. Immunol. 145, 3194–3201, 1990.

9. Blanchard, D. K., Hall, R. E., and Djeu, J. Y., Role of CD18 inlymphokine activated killer (LAK) cell-mediated lysis of humanmonocytes: Comparison with other LAK targets. Int. J. Cancer45, 312–319, 1990.

0. Lee, R. K., Spielman, J., Zhao, D. Y., Olsen, K. J., and Podack,E. R., Perforin, Fas ligand, and tumor necrosis factor are themajor cytotoxic molecules used by lymphokine-activated killercells. J. Immunol. 157, 1919–1925, 1996.

1. Vujanovic, N. L., Nagashima, S., Herberman, R. B., and White-side, T. L., Nonsecretory apoptotic killing by human NK cells.J. Immunol. 157, 1117–1126, 1996.

2. Arase, H., Arase, N., and Saito, T., Fas-mediated cytotoxicity byfreshly isolated natural killer cells. J. Exp. Med. 181, 1235–1238,1995.

3. Eischen, C. M., Schilling, J. D., Lynch, D. H., Krammer, P. H.,and Leibson, P. J., Fc receptor-induced expression of Fas ligandon activated NK cells facilitates cell-mediated cytotoxicity andsubsequent autocrine NK cell apoptosis. J. Immunol. 156, 2693–2699, 1996.

4. Ostergaard, H. L., and Clark, W. R., Evidence for multiple lyticpathways used by cytotoxic T lymphocytes. J. Immunol. 143,2120–2126, 1989.

5. Rouvier, E., Luciani, M.-F., and Golstein, P., Fas involvement inCa21-independent T cell-mediated cytotoxicity. J. Exp. Med. 177,

195–200, 1993.

1

2

2

2

2

2

2

2

3

3

3

3

3

3

R

263pH-DEPENDENT TUMOR CELL KILLING

16. Hasegawa, Y., and Bonavida, B., Calcium-dependent pathway oftumor necrosis factor-mediated lysis of target cells. J. Immunol.142, 2670–2676, 1989.

17. Naume, B., Johnsen, A. C., Espevik, T., and Sundan, A., Geneexpression and secretion of cytokines and cytokine receptorsfrom highly purified CD561 natural killer cells stimulated withinterleukin-2, interleukin-7 and interleukin-12. Eur. J. Immu-nol. 23, 1831–1838, 1993.

8. Ostensen, M. E., Thiele, D. L., and Lipsky, P. E., Tumor necrosisfactor-a enhances cytolytic activity of human natural killer cells.J. Immunol. 138, 4185–4191, 1987.

19. Chong, A. S., Scuderi, P., Grimes, W. J., and Hersh, E. M., Tumortargets stimulate IL-2 activated killer cells to produce interfer-on-g and tumor necrosis factor. J. Immunol. 142, 2133–2139,1989.

20. D’Andrea, A., Aste-Amezaga, M., Valiante, N. M., Ma, X., Kubin,M., and Trinchieri, G., Interleukin 10 (IL-10) inhibits lympho-cyte interferon g-production by suppressing natural killer cellstimulatory factor/IL-12 synthesis in accessory cells. J. Exp.Med. 178, 1041–1048, 1993.

1. Espevik, J., Figari, I. S., Ranges, G. E., and Palladino, M. A., Jr.,Transforming growth factor-b1 (TGF-b1) and recombinant hu-man tumor necrosis factor-a reciprocally regulate the generationof lymphokine-activated killer cell activity. J. Immunol. 140,2312–2316, 1987.

2. Yamamoto, H., Hirayama, M., Genyea, C., and Kaplan, J.,TGF-b mediates natural suppressor activity of IL-2-activatedlymphocytes. J. Immunol. 152, 3842–3847, 1994.

3. Musgrove, E., Rugg, C., and Hedley, D., Flow cytometric mea-surement of cytoplasmic pH: A critical evaluation of availablefluorochromes. Cytometry 7, 347–355, 1986.

4. Galkina, S. I., Sud’Ina, G. F., and Margolis, L. B., Cell–cell contactsalter intracellular pH. Exp. Cell. Res. 200, 211–214, 1992.

5. Radosevic, K., de Grooth, B. G., and Greve, J., Changes in intra-cellular calcium concentration and pH of target cells during thecytotoxic process: A quantitative study at the single cell level.Cytometry 20, 281–289, 1995.

6. Ioannides, C. G., and Whiteside, T. L., T cell recognition ofhuman tumors: Implications for molecular immunotherapy ofcancer. Clin. Immunol. Immunopathol. 66, 91–106, 1993.

7. Whiteside, T. L., and Parmiani, G., Tumor-infiltrating lympho-cytes: Their phenotype, function and clinical use. Cancer Immu-nol. Immunother. 39, 15–21, 1994.

28. Peck, R., Brockhaus, M., and Frey, J. R., Cell surface tumor

necrosis factor (TNF) accounts for monocyte- and lymphocyte-

mediated killing of TNF-resistant target cells. Cell. Immunol.122, 1–10, 1989.

29. Yonehara, S., Ishii, A., and Yonehara, M., A cell-killing mono-clonal antibody (anti-Fas) to a cell surface antigen co-downregu-lated with the receptor of tumor necrosis factor. J. Exp. Med.169, 1747–1756, 1989.

30. Young, J. D., Damiano, A., DiNome, M. A., Leong, L. G., andCohn, Z. A., Dissociation of membrane binding and lytic activi-ties of the lymphocyte pore-forming protein (perforin). J. Exp.Med. 165, 1371–1382, 1987.

31. Persechini, P. M., Liu, C. C., Jiang, S., and Young, J. D., Thelymphocyte pore-forming protein perforin is associated withgranules by a pH-dependent mechanism. Immunol. Lett. 22,23–27, 1989.

32. Schulze-Osthoff, K., Krammer, P. H., and Droge, W., Divergentsignalling via APO-1/Fas and the TNF receptor, two homologousmolecules involved in physiological death. EMBO J. 13, 4587–4596, 1994.

3. Rabinowich, H., Suminami, Y., Reichert, T. E., Crowley-Nowick,P., Bell, M., Edwards, R., and Whiteside, T. L., Expression ofcytokine genes or proteins and signaling molecules in lympho-cytes associated with human ovarian carcinoma. Int. J. Cancer68, 276–284, 1996.

4. Bischof, G., Cosentini, E., Hamilton, G., Riegler, M., Zacherl, J.,Teleky, B., Feil, W., Schiessel, R., Machen, T. E., and Wenzl, E.,Effects of extracellular pH on intracellular pH-regulation andgrowth in a human colon carcinoma cell-line. Biochim. Biophys.Acta 1282, 131–139, 1996.

5. Roepe, P. D., Wei, L. Y., Cruz, J., and Carlson, D., Lower elec-trical membrane potential and altered pHi homeostasis in mul-tidrug-resistant (MDR) cells: Further characterization of a seriesof MDR cell lines expressing different levels of p-glycoprotein.Biochemistry 32, 11042–11056, 1993.

6. Borg, A. L., Gallice, P. M., Kovacic, H. N., Nicoara, A. E., Favre,R. G., and Crevat, A. D., Impairment of sodium pump andNa1/H1 antiport in erythrocytes isolated from cancer patients.Cancer Res. 56, 511–514, 1996.

7. Boyer, M. J., and Tannock, I. F., Regulation of intracellular pHin tumor cell lines: Influence of microenvironmental conditions.Cancer Res. 52, 4441–4447, 1992.

8. Boyer, M. J., Barnard, M., Hedley, D. W., and Tannock, I. F.,Regulation of intracellular pH in subpopulations of cells derivedfrom spheroids and solid tumours. Br. J. Cancer 68, 890–897,

1993.

eceived November 9, 1999; accepted with revision June 15, 2000