NAFTA prompts health concerns across the borders.

Sublines from the small cell lung cancer (SCLC) cell lines U1285 and U1690, denoted U1285-100, U1285-250, U1690-40 and U1690-150, were adapted to grow in the continuous presence of 100, 250, 40 and 150 ng ml-' doxorubicin (Dox), respectively. The Dox resistance was accompanied by cross-resistance to vincristine (Vcr), Vp-16 and for U1285-100 also to cisplatinum. Sublines of U1690-40 and U1285-100, cultured in absence of Dox for 4 months were only partially reversed with respect to Dox resistance. Neither the parental nor the most Dox resistance sublines had detectable levels of mdr RNA but a small fraction of cells in all cell lines stained weakly positive for P-glycoprotein (P-gp). Verapamil (Ver) at 5 gsM reversed the Dox resistance completely and partly in the U1690 and U1285 sublines, respectively, but did not increase the cellular accumulation of Dox. The cytoplasmic free Ca2" concentration (Ca2+i) was close to 100 nm in both parental cell lines but elevated in the U1285-100 and U1690-40 sublines by 21 and 44%, respectively, and in U1285-250 and U1690-150 by 51 and 91%, respectively. The partly reverted sublines still showed significant but smaller elevations in Ca2+i of 10-30%. Ver was without acute or long term effects of Ca2+i in the U1285-100 and U1690-40 sublines. Selection for Dox resistance in SCLC may thus result in atypical multidrug-resistance characterised by absence of P-gp overexpression and atypical cross-resistance. Although Ver did not seem to affect Dox accumulation it may still work as a resistance modulator. There may be a role for increased Ca2+i in drug resistance in SCLC cells, but resistance reversal by Ver seems unrelated also to changes in Ca2+i.

Summary Sublines from the small cell lung cancer (SCLC) cell lines U1285 and U1690, denoted U1285-100, U1285-250, U1690-40 and U1690-150, were adapted to grow in the continuous presence of 100, 250, 40 and 150 ng ml-' doxorubicin (Dox), respectively. The Dox resistance was accompanied by cross-resistance to vincristine (Vcr), Vp-16 and for U1285-100 also to cisplatinum. Sublines of U1690-40 and U1285-100, cultured in absence of Dox for 4 months were only partially reversed with respect to Dox resistance. Neither the parental nor the most Dox resistance sublines had detectable levels of mdr RNA but a small fraction of cells in all cell lines stained weakly positive for P-glycoprotein (P-gp). Verapamil (Ver) at 5 gsM reversed the Dox resistance completely and partly in the U1690 and U1285 sublines, respectively, but did not increase the cellular accumulation of Dox. The cytoplasmic free Ca2" concentration (Ca2+i) was close to 100 nm in both parental cell lines but elevated in the U1285-100 and U1690-40 sublines by 21 and 44%, respectively, and in U1285-250 and U1690-150 by 51 and 91%, respectively. The partly reverted sublines still showed significant but smaller elevations in Ca2+i of 10-30%. Ver was without acute or long term effects of Ca2+i in the U1285-100 and U1690-40 sublines. Selection for Dox resistance in SCLC may thus result in atypical multidrug-resistance characterised by absence of P-gp overexpression and atypical cross-resistance. Although Ver did not seem to affect Dox accumulation it may still work as a resistance modulator. There may be a role for increased Ca2+i in drug resistance in SCLC cells, but resistance reversal by Ver seems unrelated also to changes in Ca2+i.
Acquired cytotoxic drug resistance is often extended also to drugs not included in the treatment regimen and the most consistent finding in vitro in such multidrug-resistance (MDR) is resistance to anthracyclines, vinca alkaloids and epipodophyllotoxins with a decreased drug accumulation compared to the sensistive cells (Beck, 1987;Bradley et al., 1988). In vivo (Goldstein et al., 1989) as well as in vitro (Beck, 1987;Bradley et al., 1988) the MDR phenotype is often characterised by expression of the 170 kDa membrane P-glycoprotein (P-gp) encoded by the mdr gene (Ueda et al., 1987). Several lines of evidence indicate that this protein mediates an energy dependent extrusion of various cytotoxic drugs (Beck, 1987;Bradley et al., 1988;Horio et al., 1988).
Several well known pharmacological agents have been shown to reverse MDR in vitro (Ford & Hait, 1990). One of the most studied of these resistance modulators is the calcium channel blocker verapamil (Ver), which modulates acquired as well as intrinsic drug resistance of various cell types (Simpson, 1985). The mechanism for resistance reversal by Ver is still not completely understood although it has been found that Ver may compete with cytotoxic drugs for binding to the P-gp, resulting in a decrease in drug efflux and thus enhanced cytotoxicity (Safa et al., 1987;Yusa & Tsuruo, 1989).
Transmembrane transport as well as many other cell functions are known to be regulated by intracellular signals such as, e.g. changes in the cytoplasmic free Ca2l concentration (Ca2+i; Rasmussen & Barrett, 1984). Based on findings of increased calcium content of MDR cells (Tsuruo et al., 1984) and resistance reversal by the calcium channel blocker Ver, it has been speculated that Ca2+i may have role in cytotoxic drug resistance and that the sensitising effect of Ver could be due to a decrease in Ca2+i (Beck, 1987).
In the present study some of these hypotheses were evaluted in two established human small cell lung cancer (SCLC) cell lines. We investigated whether development of resistance to doxorubicin (Dox) also conferred resistance to other cytotoxic drugs and was accompanied by increased expression of P-gp and changes in drug accumulation. The possible role of Ca2+i in drug resistance was evaluated by quin2 measurements in sublines showing varying degrees of drug resistance. Furthermore, the possible potentiating effect of Ver on Dox cytotoxicity was evaluated and correlated to acute and long term changes in Ca2+i.

Materials and methods
Cell lines and culture The human U1285 and U1690 SCLC were established as described previously (Bergh et al., 1982;Bergh et al., 1985a). Both  and antibiotics and were refed twice weakly. The cultures were incubated at 37°C in an atmosphere containing 5% CO2 and 95% air. In control experiments for expression of P-gp, mdrl RNA and drug accumulation the parental T-ALL cell line LO and its vincristine (Vcr) resistant subline L100, kindly provided by Dr L. Slater, Department of Medicine, University of California, Irvine, CA (Slater et al., 1986) as well as the chronic myelocytic leukaemia cell line K562 (Lozzio & Lozzio, 1975) and its subline K562/Vcr, adapted to growth in presence of 30 mm Vcr, were used. The LIOO subline is approximately 160-fold Vcr resistant, shows cross-resistance to Dox and Vp-16, expresses P-gp and is modulated by Ver (Nygren & Larsson, 1991 Meister, 1988) on Cisp sensitivity, 5 1M Ver (in dimethyl sulfoxide/PBS; Sigma) or 1-10 tLM BSO (in PBS) was added just prior to the cytotoxic drug. The volume of each added drug was always 10 l giving a final maximal dimethyl sulfoxide concentration of 0.1%, which did not affect cell growth. The cells were then cultured for 72 h under the conditions described above. No medium change was done during the culture period.

Measurement of cell survival
Cell survival after culture was estimated by using the fluorogenic substrate fluorescein diacetate (FDA; Sigma) which rapidly enters intact cells where it is hydrolysed to its fluorescent derivative fluorescein (Rotman & Papermaster, 1966). The details of this technique have recently been described (Larsson & Nygren, 1990). Briefly, after culture the plates were centrifuged (100 g, 5 min), the medium removed by flicking the plate, and the wells washed once with 200 lil of the buffer described below. To each well was then added 200 pl of assay buffer containing 0.5 mM Mg2 , 1,25 mM Ca2", 3 mM glucose, 10 mM Hepes (pH 7.4) and physiologically balanced in other cations and with Clas the sole anion and with 10l"gml-l FDA. The plates were then incubated for 60min at 37°C after which the fluorescence from each well was read in a Fluoroscan II microfluorometer (Flow) with filters set at 485 and 538 nm for excitation and emission, respectively. The fluorometer was blanked against wells containing assay medium with dye. Each plate was read in about 1 min and the fluorescence data was then imported to a Macintosh SE computer for statistical and graphical processing. Cell survival after culture is expressed as survival index (SI) defined as FDA fluorescence for treated wells/FDA fluorescence for untreated control wells expressed as per cent. IC50 was defined as the cytotoxic drug concentration resulting in a SI of 50% of control. Resistance factor (RF) was defined as IC50 for the subline/IC50 for the parental line.
Immunohistochemical staining for P-gtycoprotein P-gp staining was performed using the monoclonal anti-P-gp antibody JSB-1 (Sanbio, Uden, The Netherlands; Scheper et al., 1988) with the technique described previously (Bergh et al., 1985b). Frozen sections (4-6 tM) of human adrenal cortex and cytocentrifuge preparations of MDR L100 and K562/Vcr cells were used as positive controls whereas sensitive LO and K562 cells and preparations stained with the above technique, but without primary antibody served as negative controls. The specimens were incubated with the JSB-1 antibody diluted 1/30 in PBS for 60 min, followed by washing and application of a biotinylated rabbit-anti-mouse complex (Vector Laboratories Inc., Burlingame, CA) and then a avidin-biotinylated horseradish peroxidase complex (Vector Laboratories). The coverslips were then developed in 0.02% 3-amino-9 ethycarbazole supplemented with 0.002%
The quantities of mdrl RNa in the extracts were determined by comparison with a standard curve, generated by hybridisation with increasing amounts of the unlabelled sense (Steen et al., 1990). Samples classified as positive for mdrl RNA show at least twice the background radioactivity and a proportional increase in radioactivity with increasing amounts of added extracts. The results are presented as c.p.m. with background subtracted for three dilutions of each extract. Based on the standard curve, a molecular weight of sense RNA of 1.49 x 105, c.p.m. IAg'-DNA values for each extract and the assumption of a DNA content of 6 pg cell-', the number of mdrl RNa copies/cell may be calculated. The detection limit in a sample containing 30 fig DNA is 7.5 x I05 copies of RNA which corresponds to 0.15 RNA copies cell-'.

Measurement of cellular doxorubicin accumulation
The parental and most resistant sublines were compared with respect to Dox accumulation in absence and presence of Ver. The cells were incubated at 37°C for 1 h at a density of 1.5 x 106cells ml-' in RPMI 1640 medium supplemented with 10% newborn calf serum (Gibco, Paisley, Scotland) and containing 1 tiM Dox and with or without 5 lM Ver. The incubation was stopped by mixing 1.5 ml of the incubate with 5 ml ice cold PBS. After two washes in ice cold PBS the cell pellets were kept in -20°C until analysis. Cellular concentrations of Dox were determined by highperformance liquid chromatography as described (Baurin et al., 1978). Briefly, after sonication of each cell pellet in 0.5 ml PBS a 0.2 ml aliquit was added to 0.2 ml 0.1 M borate buffer (pH 9.8) containing 1 gM daunorubicin as internal standard.
Extraction was made by addition of 1.8 ml chloroform/ methanol (4:1 by volume) after which the drugs were separated on a Lichosorb Si-60 column (Hibar; Merck, Darmstadt, Germany) and eluted with a mixture of chloroform, methanol, glacial acetic acid and 0.3 mM MgCl2 (720:210:40:30 by volume) at a flow rate of 1.5mlmin-'.
The drugs were then quantified by fluorometry at 480 and 560 nm for excitation and emission wavelengths, respectively.
The protein content in the dissolved cell pellets was determined by the method described by Lowry (Lowry et al., 1951) and the cellular accumulation of Dox is expressed as nmol Dox mg-' protein.
Measurement of the cytoplasmic free Ca" concentration Fluorometric measurements of Ca2"i were performed as described previously (Nygren et al., 1988). Briefly, 5 x 106 quin2 (Sigma; Tsien et al., 1982) loaded cells were suspended in 1.3 ml of the buffer used for cell survival measurements and incubated with constant stirring at 37°C in a 1 cm cuvette of a Perkin-Elmer LS5 spectrofluorometer with excitation and emission wavelengths set at 339 and 492 nm, respectively. Ver was added from a 100-fold concentrated stock solution in DMSO/buffer. In the measurement of the long term effect of Ver on Ca2"i, 5 tLM Ver was present also in the physiological buffer used for washing and assay. The addition of vehicle was without effect on quin2 fluorescence. Calibration for calculation of Ca2"i was then performed as described (Nygren et al., 1988). Extracellular quin2 never exceeded 7% of Fmax as judged by the fall of fluorescence signal after addition of 3 mM EGTA and was similar in all cell lines. The increase in fluorescence upon addition of the intracellular heavy metal chelator TPEN (Calbiochem, La Jolla, CA; Arslan et al., 1985) corresponds to a rise in Ca21i of 30-40 nM in all cell lines, indicating no significant differences in quenching heavy metal content.
Student's paired t-test was used for statistical comparisons.

Results
Cytotoxic drug sensitivity and effects of resistance modulators The IC50 values for Dox were 0.4 and 0.21gml-mI for the parental U1690 and U1285 cell lines, respectively (Table I).
The resistant sublines U1690-40 and U1690-150 were approximately 6 and 7-fold resistant to Dox whereas the revertant subline U1690-(40) retained most of its Dox resistance with a RF of 4. The corresponding values for U1285-100 and U1285-250 were 18 and 16, respectively, whereas the revertant U1285-(100) showed a RF of 8. The sublines also developed cross-resistance to Vcr, which for the U1690 sublines was of even higher magnitude than for Dox with RFs of 40 and 246 for U1690-40 and U1690-150, respectively. Again the U1690-(40) revertant subline retained much of its original resistance with a RF of 34. The corresponding values for U1285-100 and U1285-250 were 5 and 17, respectively. The revertant U1285-(100) cells showed a paradoxical 2-fold increase in Vcr sensitivity compared to the parental cell line. The U1690-40 and U1285-100 sublines were also crossresistant to Vp-16 with RFs of 16 and 8, respectively. To check for cross-resistant to a cytotoxic agent not included in the typical MDR phenotype we also investigated two sublines for Cisp sensitivity. U1690-40 and U1285-100 were found 1.4 and 4-fold resistant compared to the parental lines. Ver at 5 JAM reversed the Dox resistance completely in the U1690 sublines and partly in the U1285 sublines (Table I). BSO at 1-1 LM, depending on the sensitivity of the cell lines to BSO alone, increased the Cisp sensitivity 1.3-5-fold in parental and resistant cells.
The U1690 sublines showed minor collateral sensitivity to both Ver and BSO with normalisation to the parental :q 0Z I'.   phenotype in the revertant subline (Figure 1). In the U1285 series of cell lines there was no evidence for collateral sensitivity or cross-resistance to these modulators (Figure 2).

Expression of P-glycoprotein
As expected adrenal cortex stained strongly positive for P-gp (Table II). The parental LO and K562 cells were found negative whereas the majority of the MDR L100 and K562/ Vcr cells stained positive, although weaker than adrenal cortical cells. Both the U1690 and U1285 cell lines as well as their drug resistant sublines showed faint membrane staining in a small proportion (<15%) of the cells. Qualitatively similar results were obtained in control expenrments using the monoclonal antibody C219 (not shown). Furthermore, Western blotting confirmed expression of a protein reactive with the C219 antibody in L100 cells whereas no such protein was found in the LO cells or any of the SCLC cell lines (not shown).

Expression of mdrl RNA
Data for hybridisation between labelled RNA probe and unlabelled sense RNA (standard curve) or nucleic acid extract from K562/Vcr are shown in Figure 3a. There is a proportional increase in radioactivity with increasing amounts of added sense RNA or extract. The c.p.m. for K562/Vcr corresponds to approximately 98 mdr 1 RNA copies/cell. For parental K562 cells as well as the parental and most resistant SCLC sublines the c.p.m. values were just above background and the mdr 1 RNA content of these samples was below the detection limit of the assay (Figure 3b).

Cellular doxorubicin accumulation
The U1690-150 and U1285-250 sublines accumulated somewhat more and less Dox, respectively, compared to their parental cell lines (Figure 4). Presence of 5 JM Ver during incubation did not affect the Dox accumulation in any cell line. In control experiments using the same method, K562/ Vcr cells were found to accumulate 50% less daunorubicin compared to the parental cells and presence of Ver during incubation normalised this accumulation defect (not shown).
Cytoplasmic free Ca2" concentrations Basal Ca2"i of U1690 cells was 100 nM and was increased to 144   aImmunohistochemical stainings were performed as described in Materials and methods using the monoclonal antibody JSB-1. Intensity of staining is in comparison to adrenal cortex which was judged as + + +. The fraction of cells positive was determined by examination of 100 cells in each experiment. Data from one typical experiment of 4.  Figure 3 Hybridisation between labelled mdrl RNA probe and increasing amounts of unlabelled sense RNA (standard curve, inset) or nucleic acid extract from K562/Vcr cells a. The corresponding data for the parental K562 cells as well as for the parental and most resistant SCLC cell lines indicate mdrl RNA levels below the detection limit of the assay b. Data from one typical experiment. 0 K562; U1690; * U1690-150; A U1285; A U1285-250. During the last years several human cell lines have been described showing MDR patterns deviating from the classical one. This includes atypical cross-resistance Haber et al., 1989;Baas et al., 1990), presence of P-gp without increased drug efflux (Deffie et al., 1988), absence of P-gp overexpression with (McGrath & Center, 1988;Haber et al., 1989;Reeve et al., 1990) or without  drug accumulation defects and cytotoxic drug potentiation by Ver also in absence of P-gp (Nygren & Larsson, 1990a;Baas et al., 1990) and dissociated from changes in cytotoxic drug accumulation (Chang et al., 1989). An interesting feature of these atypical MDR cell lines is that most have been selected for resistance by exposure to increasing concentrations of Dox. For SCLC, typical MDR (Morgan et al., 1989) and atypical (Cole et al., 1989) drug resistant cell lines have been described, the latter showing the classical MDR crossresistance pattern, absence of P-gp overexpression, modest potentiation by calcium channel antagonists and lack of collateral sensitivity to resistance modulators. Also the resistant SCLC cell lines in the present investigation do no adhere to the typical MDR phenotype. The cells were resistant to Dox, Vcr and Vp-16, three drugs included in the typical MDR phenotype (Ford & Hait, 1990) but for U1285-100 also to Cisp. Furthermore, the Vcr resistance was similar (U1285 series) or considerably higher (U1690 series) than for the selecting agent, a feature which is less common (Bradley et al., 1988).
There was also no evidence for P-gp overexpression in the resistant cells. This is probably not due to technical difficulties since the P-gp overexpression associated with the low grade (6-fold) Dox resistance of RPMI 8226/Dox6 myeloma cells (Dalton et al., 1989) was detected immunohistochemically (not shown) and we also found the LIOO, K562/Vcr and adrenal cortical cells to be positive. Furthermore, the findings were confirmed by Western immunoblotting using the monoclonal antibody C219 (not shown) and by the absence of mdrl RNA. The faint immunostaining in a minority of SCLC cells probably represents the background level of the method. However, more important than to discriminate between very low levels and complete absence of P-gp is that no signs of P-gp overexpression in the Dox selected sublines could be detected.
Ver reversal of cytotoxic drug resistance in classical MDR is considered to be mediated through increased cellular drug content, probably by inhibition of drug extrusion by P-gp (Ford & Hait, 1990). Despite the absence of P-gp overexpression in the resistant SCLC cell lines the Dox resistance was completely or partly reversed by Ver. Furthermore, Ver was without effect on cellular Dox accumulation. Together the present findings confirm the recent reports cited above on 'atypical' MDR also for SCLC cells and implicate the presence of other mechanisms for MDR than only P-gp mediated drug extrusion and also for Ver induced circumvention of resistance.
Despite the SCLC origin of both the U1690 and U1285 cell lines and similar procedures for establishment of resistant sublines including the selecting drug Dox, some apparent differences in the resistance phenotypes could be noted. The Dox resistance was thus completely reversed by Ver in the U1690 but only partially in the U1285 sublines. The magnitude of Vcr resistance was considerably higher than for Dox in the U1690, but not in the U1285 sublines. U1690-40 and U1690-150 showed a tendency to collateral sensitivity for Ver and BSO which was not the case for U1285-100 and U1285-250. Furthermore, the revertant U1690-(40) and U1285-(100) sublines were only partially reversed with respect to Dox sensitivity, but essentially unaffected and more than completely reversed, respectively, with respect to Vcr sensitivity.
Together with the previous findings of classical (Morgan et al., 1989) as well as atypical (Cole et al., 1989) MDR in Dox selected SCLC cell lines these findings indicate the presence of different pathways to the final resistant phenotype, also in cells of the same histological type and selected for resistance to the same cytotoxic drug. If this is true also for development of drug resistance in vivo it may have important implications for the therapy of resistant tumours. Cytotoxic drug treatment, its potentiation by resistance modulators and exploitation of collateral sensitivity could not be based on, e.g., histology, previous therapy or P-gp expression, but rather on individual in vitro testing using techniques showing good clinical correlations.
Based on the fact that Ca2"i is an important intracellular messenger regulating, e.g., cell growth, secretion and transport mechanisms (Rasmussen & Barrett, 1984), the initial finding of increased Ca2" content of drug resistant compared to sensitive cells (Tsuruo et al., 1984) as well as resistance reversal by calcium channel antagonists and calmodulin inhibitors (Ford & Hait, 1990), it has been speculated that Ca2"i may be of importance in drug resistance (Beck, 1987). A possible relationship between Ca2" and drug resistance has also been indicated by the findings that Ca2+-ionophores tend to induce resistance (Huet & Robert, 1988;Nygren & Larsson, 1990b) whereas incubation under conditions known to decrease Ca2+i has the opposite effect (Huet & Robert, 1988). However, other studies have failed to reveal a consistent relationship between drug resistance and Ca2lin pairs of sensitive and resistant cell lines (Nair et al., 1986;Vayuvegula et al., 1988) and in experiments in which Ca2"i was manipulated pharmacologically (Nygren & Larsson, 1990b).
Furthermore, indirect and direct measurements of Ca2"i have failed to show any changes induced by Ver despite its cytotoxic drug potentiation (Huet & Robert, 1988;Nygren & Larsson, 1990a).
In the SCLC cell lines included in the present study there was a fairly close correlation between the level of cytotoxic drug resitance and Ca2"i. The differences are probably not technical artifacts since the amount of extracellular quin2 was similar for all cell types and the intracellular heavy metal chelator TPEN (Arslan et al., 1985) revealed similar amounts of quenching metals in the different cell types. Although Dox had no effect on resting Ca2"i in neuroblastoma cells (Oakes et al., 1990), prolonged Dox exposure have been found to lead to cellular Ca2+ accumulation (Keyes et al., 1987) and inhibition of Na+/Ca2+ exchange (Caroni et al., 1981). One could therefore speculate that the elevated Ca2+i in the resistant SCLC cell lines is due to the direct effect of Dox exposure. However, the revertant cell lines still showed significantly elevated Ca21i, in parallel with at least some retention of drug resistance. Although previous studies are contradictory with respect to the role of Ca21i in drug resistance, the present findings indicate a possible causal relationship between Ca2+i and drug resistance in specific cell types and drug resistance phenotypes.
There was no acute change in Ca2'i after Ver addition in cells being sensitive to the potentiating effect of Ver, and long term incubation with the calcium channel blocker did not result in any further changes in Ca2'i. These findings corroborate similar findings in cells showing intrinsic (Nygren & Larsson, 1990a) and acquired (Huet & Roberts, 1988) drug resistance also in cells characterised by elevated Ca2+i. It is therefore concluded that although the exact role for Ca2+i in cytotoxic drug resistance remains to be elucidated, the potentiating effect of Ver on the effect of cytotoxic drugs is unrelated to changes in Ca21i. Since data now accumulate showing resistance modulation by Ver also in absence of P-gp overexpression and drug transport changes, altenative mechanisms for the Ver effect should be looked for. This study was supported by grants from the Swedish Cancer Society. The skilful technical assistance of Anna Johansson, Monica Ekberg and Irene Arestrom is gratefully acknowledged.