Volume-sensitive chloride currents in four epithelial cell lines are not directly correlated to the expression of the MDR-1 gene.

It has been shown recently that heterologous expression of human MDR-1 gene, which is responsible for multidrug resistance during cancer therapy, causes appearance of volume-sensitive Cl- currents, thus suggesting that the product of the MDR-1 gene (the P-glycoprotein) has a Cl- channel activity (Valverde, M. A., Diaz, M., Sepulveda, M. A., Gill, D. R., Hyde, S. C., and Higgins, C. F. (1992) Nature 355, 830-833). In the present work, we have tested four epithelial cell lines both for the expression of MDR-1 gene and for the presence of volume-sensitive Cl- currents. LoVo/H and LoVo/Dx cells derive from a human colon adenocarcinoma, the latter cell line being resistant to high concentrations of the antitumoral drug doxorubicin. 9HTEo- cells were obtained by transformation of human tracheal epithelium. The 9HTEo-/Dx cell line was established from these cells by selection in doxorubicin. As expected, higher levels of P-glycoprotein expression were detected in LoVo/Dx and 9HTEo-/Dx by means of reverse transcriptase polymerase chain reaction technique, indirect immunofluorescence, and Western immunoblot assays. In contrast with these data, the size of swelling-induced Cl- current was the same in the sensitive cell line and in its drug-resistant counterpart. Actually, the Cl- conductance of 9HTEo- and 9HTEo-/Dx was 4-fold higher than that of either LoVo/H or LoVo/Dx cells. This indicates that the amplitude of this conductance is not directly related to the expression of the MDR-1 gene.

1 Supported by grants from the National Institute of Diabetes and Institute.
cystic fibrosis gene product, which is a C1-selective channel (10,11). A similar ion channel function has been proposed recently for the P-glycoprotein itself following experiments in which a fibroblast and a lung epithelial cell line were transfected with the full-length MDR-1 cDNA (1). In that study, the lowering of extracellular osmolality caused the appearance of C1-currents only in MDR-1-expressing cells. Since volumesensitive C1-currents (Iclfvol)) represent a large fraction of the membrane conductance in many epithelial cells (12)(13)(14), it can be inferred that such cells should also display P-glycoprotein expression to some extent. In the present study, we have examined the correlation between drug resistance and volumesensitive C1-conductance in four epithelial cell lines.

MATERIALS AND METHODS
Cells and Culture Conditions-The drug-sensitive LoVo/H and the drug-resistant LoVo/Dx cells were previously obtained from a human colon adenocarcinoma (15)(16)(17). 9HTEo-are immortalized human tracheal epithelial cells (18) from which we selected the drug-resistant SHTEo-Dx cell line by continuous exposition to increasing concentrations of doxorubicin (Adriamicina, Carlo Erba, Milano, Italy) up to 0.6 The culture medium contained 45% Dulbecco's modified Eagle's medium, 45% Ham's F-12 medium, 10% fetal calf serum, 100 unitdml penicillin, 100 pg/ml streptomycin, 2 m glutamine. LoVo/Dx and 9HTEo-/Dx were continuously grown in the medium supplemented with doxorubicin (1 and 0.6 pg/ml, respectively) except for the last 5-7 days before each experiment. Clonogenic assays were carried out by plating 100 cells per 60-mm Petri dish. After 24 h, the medium was supplemented with 0.01, 0.1, or 1 pg/ml doxorubicin, and, aRer 2 weeks, cell colonies were stained with methylene blue and counted. These assays were carried out in triplicate and with a control of cells grown without doxorubicin.
Patch-clamp Analysis-Membrane currents were measured with the whole-cell configuration of the patch-clamp technique (19). Experiments were performed on nonconfluent cells plated in 35-mm Petri dishes. The extracellular (bath) solution contained (in m): 130 NaCl, 3 KCl, 2 CaC12, 2 MgC12, 10 sodium Hepes, 10 glucose (pH = 7.3). Patch pipettes were instead filled with (in m): 140 NaCl, 1 MgC12, 0.18 CaCl,, 2 EGTA, 10 sodium Hepes (pH = 7.3). This pipette (intracellular) solution included also 100 p~ ATP in part of the experiments. Osmolality was adjusted to 295-305 mosmkg with mannitol. The hypotonic medium (210 mosmkg) had the same composition of the extracellular medium but without mannitol and with only 90 m NaCl instead of 130. 1,s-Dideoxyforskolin and verapamil (Sigma) were added to this medium to test their ability to block the swelling-induced C1-current.
Series resistance was measured in each experiment and used to correct the errors in the applied membrane potential. The voltage stimulation protocol consisted of 1-s-long voltage pulses to -100 mV every 10 s, starting from a holding voltage of 0 mV. To construct currentvoltage relationships, voltage steps were applied to potentials in the range -80 to +80 mV.
RNA Analysis-Total RNAs were extracted from 10' cells with the guanidine-thiocyanate method (20), whereas poly(A+) RNAs were extracted from 5 x lo7 cells with Dynabeads "280 Oligo(dT)z5 (Dynal A.S., Oslo, Norway) as described (21). To perform reverse transcriptase pdml. Three-quarters of this cDNA were used for the detection of the MDR-1 mRNA. The remaining fraction was instead amplified to check for the presence of &-microglobulin mRNA, the positive control. The amplification was carried out in a final volume of 100 p l with 1 p concentration of primers, 2.5 units of Tuq polymerase and reaction kits (Perkin-Elmer Cetus Instruments). The primers sequence and PCR conditions were as reported by Noonan et ul. (22). The expected PCR products were of 167 and 120 base pairs (for MDR-1 and Pz-microglobulin, respectively) and were separated on 2.5% agarose gels.
Protein Analysis-For the immunofluorescent detection of P-glycoprotein, cells were cultured on glass coverslips, washed in PBS, and fixed in a cold (-20 "C) methanoVacetone (4:l) solution. They were then incubated for 30 min at room temperature with the P-glycoCHEK C219 monoclonal antibody (Centocor Inc., Malvern, PA) diluted 1:lO in PBS. M e r washing, the cells were incubated for 45 min at room temperature with a sheep anti-mouse IgG, fluorescein-linked second antibody (Amersham International, England) diluted 1:20 in PBS.
For Western blot immunoassays, cells were detached from 100-mm Petri dishes with a cell scraper, washed in PBS, lysed in an ice-cold buffer (10 IMI Tris-HC1, 1% Triton X-100, pH 8.0) and spun in a Microfuge at 13,000 rpm for 10 min. An amount of the supernatant containing 200 pg of proteins was run on a 6.5% SDS-polyacrylamide gel and transferred onto a nitrocellulose filter with a transfer buffer composed of 25 m~ Tris, 192 m~ glycine, 20% methanol, and 0.04% SDS (pH 8.3). The filter was then blocked with 5% skim milk powder in Trisbuffered saline (TBS; 200 m~ Tris, 5 M NaCl, pH 7.5), washed in ?TBS (TBS with 0.05% Tween ZO), and incubated for 4 h with 4 pg/ml mddAbl), a rabbit polyclonal antibody against P-glycoprotein (Oncogene Science), or, alternatively, with 1 pg/ml P-glycoCHEK C219 monoclonal antibody. After washing, the filter was incubated with an anti-rabbit or with an anti-mouse IgG second antibody conjugated to alkaline phosphatase. Protein detection was performed with both colorimetric and chemiluminescent techniques. Second antibodies and the colorimetric detection system were purchased from Bio-Rad (Munich, Germany). The chemiluminescent revelation kit was from New England Biolabs.
Determination of band intensities in reverse transcriptase PCR and Western blot experiments was performed with a LKB Ultroscan XL densitometer.
Statistics-Data are presented as representative experiments or as arithmetic means & S.E. Significance is calculated according to the Student's t-test.

RESULTS
To check the drug resistance of cells, we have performed clonogenic assays. As shown in Fig. 1, only LoVo/Dx and 9HTEo-/DX cells are able to form colonies in 1 pg/ml doxorubicin. A significant higher resistance of LoVo/Dx and 9HTEo-/Dx is also observed at 0.1 pg/ml ( p < 0.01 with respect to both LoVo/H and 9HTEo-cells). This suggests a higher expression of P-glycoprotein in the two drug-resistant cell lines.
To investigate the correlation between the MDR phenotype phosphate-buffered saline.
The abbreviations used are: PCR, polymerase chain reaction; PBS, and the volume-sensitive C1-currents, we used the whole-cell configuration of the patch-clamp technique. After breaking the patch to obtain the whole-cell configuration, a C1-current began to appear in isotonic conditions (Fig. 2). This spontaneous activation was previously observed by Worrell et al. (12) in T84 cells and subsequently found also in 9HTEo-and other epithelial cells (23) and attributed to an osmotic disequilibrium OCcurring in the whole-cell configuration. AS it is shown in Fig. 2, an additional conductance increase was obtained by applying a hypotonic shock (210 mosmkg). To demonstrate the C1-selectivity of this current, extracellular C1-(140 mM) was almost totally replaced (130 m~) with gluconate (Fig. 3, A and B ). This intervention shifted the reversal potential from 0 to +43.9 2 4.5 mV in 9HTEo-(n = 51, +36.7 2 1.7 in 9HTEo-/Dx (n = 31, +42.2 2 5.7 mV in LoVo/H (n = 5), and +46.5 2 5.2 mV in LoVo/Dx (n = 8) as expected for a volume-sensitive C1-selective channel (14). The swelling-induced C1-current found in all four cell lines had characteristics similar to those constitutively present in other epithelial cells (12)(13)(14)23) or induced by heterologous expression of P-glycoprotein (1). In fact, the current showed outward rectification, and a marked time-dependent inactivation was observed at positive membrane potentials ( Fig. 3C). At +80 mV, the current completely inactivated in 10 s. Another characteristic was the sensitivity to 1, and verapamil: both compounds are able to revert the multidrug resistance by blocking the ATP-dependent drug efflux through P-glycoprotein. 1,9-Dideoxyforskolin at 100 w caused in our experiments a strong and reversible current inhibition (9HTEo-: 79.4 2 9.6%, n = 3; SHTEo-iDx: 81.2 2 5.6%, n = 4; LoVo/H: 92.1 2 1.1%, n = 3; LoVo/Dx: 87.0 2 3.7%, n = 4) as also reported by Valverde et al. (1). Verapamil was instead a poor blocker of the current, the inhibition being less than 70% at 500 p~ concentration in the four cell lines (n = 3 for each one; not shown). Doxorubicin (180 p~) in the pipette solution was instead unable to inhibit the volume-sensitive C1-current in LoVo/Dx (n = 3) and 9HTEo-(n = 3) cells. Fig. 4 shows the values of the conductance activated by the hypotonic shock. The chord conductance at -100 mV was corrected for the series resistance and normalized for the membrane capacitance to compensate for the larger size of 9HTEoand 9HTEo-Dx with respect to LoVo cells. The volume-sensitive conductance of LoVoDx cells was almost identical with that of LoVo/H but 4-fold smaller than that of both 9HTEo-and 9HTEo-/DX ( p < 0.001).
Most of the experiments were performed using an ATP-free pipette (intracellular) solution. ATP (100 p~) was added in the pipette solution to investigate its possible role in the activation of the Zclcvol,. The C1-current measured under this condition was not significantly different from that obtained with ATPfree solution. For example, the mean normalized conductance in LoVo/H was 1.27 2 0.11 nS/pF (n = 7) with ATP and 1.51 2 0.19 (n = 20) without ATP ( p > 0.05). Similarly, in LoVo/Dx, the conductance was 1.49 2 0.37 nS/pF (n = 6) with ATP and 1.50 2 0.17 nS/pF ( n = 18) without it. Therefore, the results obtained under these two experimental conditions were pooled in Fig. 4.
In order to account for the smaller conductance detected in LoVo cells, we asked whether a greater swelling stimulus might be required to induce a current comparable to that seen in 9HTEo-and SHTEo-Dx. This was not the case since a decrease of extracellular osmolality from 210 to 150 mosmkg did not further activate the chloride conductance in LoVo/Dx cells (n = 3; not shown).
As it is apparent from these electrophysiological data, the hypotonic shock is able to elicit a C1-current with similar features both in doxorubicin-resistant (LoVo/Dx and 9HTEo-/ DX) and in doxorubicin-sensitive (9HTEo-and LoVo/H) cells. Unexpectedly, the size of this current was not correlated with The values of the chord conductance a t -100 mV are corrected for the series resistance and normalized for the membrane capacitance. The conductance is significantly higher in 9HTEo-(n = 12) and 9HTEo-/DX ( n = 11) than in LoVo/H (n = 27) or LoVo/Dx ( n = 24) cells ( p < 0.001 for both cases). Error bars represent S.E. the degree of doxorubicin resistance. A more subtle analysis was therefore needed both at the RNA and at the protein level to look for the presence of P-glycoprotein also in 9HTEo-and LoVo/H cells.
Due to its high sensitivity (24), we used the reverse transcriptase PCR technique to study the MDR-1 mRNA expression in our cell lines. After 46 cycles of cDNA amplification, a marked band was found in LoVo/Dx and 9HTEo-/Dx lanes (Fig.  5). A weaker band was also detected in the other two cell lines.

FIG. 5. Gel electrophoresis of t h e p r o d u c t s obtained by 46
9HTEo-/Dx (lanes 2 a n d 6), LaVofH (lanes 3 and 7). and LoVo/Dx (lanes 4 and 8). Lunes I, 2,3, and 4 show the results obtained with two primers from the P,-microglobulin gene. Lunes 5, 6, 7, and 8 show instead amplification performed with two MDR-1 primers. The marker in the middle is 4x174 HaeIII. Relative intensities of the bands were assessed by densitometry (see Table I).
After densitometric analysis, relative intensities of MDR-1 lanes normalized to those of the Pz-microglobulin were: LoVo/H, 1.00; 9HTEo-, 1.0; LoVo/Dx, 3.6; and SHTEo-Dx, 4.3. These results confirm the higher expression of the MDR-1 gene in the drug-resistant cells. We asked whether this was also seen at the protein level. Therefore, we performed both immunofluorescence and Western blot assays.
Indirect immunofluorescence was carried out using the C219 monoclonal antibody against P-glycoprotein. As shown in Fig.  6, a clear difference exists between drug-resistant and drugsensitive cells. LoVo/Dx displayed a fluorescent signal much stronger than that observed in LoVo/H. Similarly, a significant fluorescence was detected in 9HTEo-/Dx, whereas in 9HTEo-it was hardly distinguishable from the background. Fig. 7 illustrates the result of a Western blot experiment performed using the polyclonal antibody mdr(Ab-1). A diffuse band of apparent molecular weight of 141,000-172,000 was clearly detectable in the LoVo/Dx and 9HTEo-/Dx lanes. A similar band was visible in LoVo/H and 9HTEo-cells, but with a relative intensity 7-fold lower as resulted from densitometric analysis. Same results were obtained using the C219 monoclonal antibody (data not shown). The smeared shape of the band is typical of P-glycoprotein as observed by Schinkel et al. (25) and is probably due to alternative forms of the protein caused by differential glycosylation or phosphorylation (25-27).
The results obtained with the different techniques are summarized in Table I. DISCUSSION The present work describes a swelling-induced C1-conductance (Zcl(vol)) which shows similar biophysical and pharmacological features in the four cell lines that we have examined. The current-voltage relationship is outwardly rectifying, and a marked time-dependent inactivation is observed at the most positive membrane potentials (i.e. +80 mV). This cell volumedependent channel is almost completely inhibited by 100 PM l,9-dideoxyforskolin and partially inhibited by verapamil. The normalized conductance is comparable between LoVo/H and  Our data argue against a quantitative correlation between drug sensitivity, P-glycoprotein expression, and activity of volume-sensitive currents. Despite LoVo/H and LoVoDx having quite different levels of P-glycoprotein expression and drug resistance, they display almost identical amplitudes of swelling-induced C1conductance. On the other hand, the 9HTEocell line has P-glycoprotein expression comparable to that of LoVo/H but a C1conductance 4-fold higher than both LoVo cell lines. Furthermore, the generation of the 9HTEo-Dx cell line by continuous selection in doxorubicin was successful in increasing the P-glycoprotein expression to a level similar to that of LoVoDx, but did not affect the amount of the C1conductance.
A lack of correlation between swelling-induced C1secretion and transport of vinblastine has also recently been observed in T84 cells (29). To explain this discrepancy, different hypotheses can be postulated. The simplest one is that P-glycoprotein is not a C1channel. While this is not very probable due to existing experimental evidences (1,28), it must be considered in light of our data.
An alternative explanation could be that the hypotonic shock is able to activate only a fraction of the volume-sensitive channels present in the cell. LoVo cells could have a reduced ability to activate C1channels due to a less efficient regulatory mechanism linking swelling to current activation. However, this would explain the smaller conductance of LoVo cells with respect to 9HTEo-, but not the lack of difference between drugresistant and drug-sensitive cells. Furthermore, an insufficient swelling stimulus can also be excluded since our results demonstrate that a more hypotonic solution is not able to activate more channels. Also to be considered is the model proposed by Gill et al. (28).
These authors hypothesize that the ion transport and drug transport functions of P-glycoprotein exist exclusive of one another. In other words, when P-glycoprotein is "locked" in the pump mode, it is not able to function as a volume-sensitive C1channel. In accord with this model, Levo cells could have a higher fraction of P-glycoprotein locked in drug transport mode and therefore have a smaller C1-conductance. Nevertheless it is difficult to explain why the generation of the SHTEo-/Dx cell line by selection has not caused any change in the volumesensitive conductance with respect to its drug-sensitive counterpart.
Another interesting possibility is that, in addition to or as an alternative to P-glycoprotein, other membrane proteins are responsible for the volume-dependent C1-conductance in our cells. As an example, a C1-channel that shows voltage-dependent inactivation at positive potentials and outward rectification has been cloned recently in MDCK cells (30). Moreover, one of the features displayed by Zcl(vol) and P-glycoprotein, Le. the sensitivity to 1,9-dideoxyforskolin and verapamil, is not a very specific characteristic, since it is also shared by other ion channels. For instance, a voltage-dependent K+ current described in the same 9HTEo-cells is inhibited by these two compounds at even lower concentrations (31). In addition, the product of the MDR-3 gene, which shows a very high degree of sequence homology with the MDR-1 P-glycoprotein and whose function is unknown, could be responsible for a C1-conductance. Finally, a ubiquitous C1-channel termed C1C-2 is also activated by osmotic disequilibria (32, 33). Thus, volume regulation seems to be a crucial and complex mechanism controlled by different anion channels.
In conclusion, it appears from our data that the Zcl(vol) is quantitatively unrelated with the P-glycoprotein expression. This discrepancy needs further investigation since it could provide interesting information on the regulatory mechanisms and the membrane proteins responsible for the multidrug resistance and the volume-activated conductances.
technical assistance and Drs. V. Corrias and G. P. Tonini for providing