Extensive Contribution of the Multidrug Transporters P-Glycoprotein and Mrp1 to Basal Drug Resistance

Fibroblast Cell Lines.

Most of the cell lines used in this study were derived by us from MEFs by 3T3-like procedures (12) . The source embryos were of mixed genetic background (F₅ or later crosses between 129/Ola and FVB strains). Lines lacking functional P-gp were obtained from Mdr1a^−/−1b^−/− embryos (13) , and lines lacking both P-gp and Mrp1 were obtained from Mdr1a^−/−1b^−/−Mrp1^−/− embryos (14) . The MEF cultures went through crises at variable passage numbers and were cloned between passages 25 and 40, as allowed by their growth characteristics. NIH/3T3 cells (15) , passage 126, were obtained from the American Type Culture Collection (Manassas, VA; CRL 1658). Note that this passage is no longer strongly contact inhibited, growing to approximately six times the density originally reported. All lines and MEF cultures were maintained in DMEM plus 10% FCS (Life Technologies, Breda, The Netherlands) and were passaged by treatment with trypsin-EDTA. Mean volumes of log phase cells were measured with a pore resistance counter (Casy; Schärfe System, Reutlingen, Germany). Single-cell cloning efficiencies were determined by 7-day colony assays stained with crystal violet.

P-gp and Mrp1 Protein Levels.

Total tissue and cell lysates for immunoblots were prepared by homogenizing and sonicating cells at 10⁷/ml or tissues at 50 mg wet weight/ml in 10 mm Tris-HCl (pH 7.4), 10 mm KCl, 1.5 mm MgCl₂, 1% w/v SDS, 1% v/v 2-mercaptoethanol, 1 mm phenylmethylsulfonyl fluoride and Complete protease inhibitor mixture (Boehringer Mannheim, Mannheim, Germany). Samples were mixed with SDS-PAGE loading buffer and heated to 65°C for 5 min, this being adequate to dissociate immunoglobulins in the tissue samples without causing significant aggregation of either P-gp or Mrp1. Samples were analyzed on 8% SDS-PAGE gels and electroblotted to nitrocellulose. Blots were divided at the 97-kDa marker. The high molecular weight halves were probed for either P-gp or Mrp1 using the C219 (Centocor, Leiden, The Netherlands) or MRPr1 (16) monoclonal antibodies, respectively. These antibodies were raised against the human and Chinese hamster proteins but cross-react with the mouse homologues. The low molecular weight halves of the blots were probed with the anti-α-tubulin monoclonal YL 1/2 (17) as a loading control. Target proteins were visualized with horseradish peroxidase-conjugated secondary antibodies and enhanced chemiluminescence (ECL). Replicate blots of total RNA from the cell lines and mouse tissues were hybridized with ³²P-labeled antisense RNA probes for Bcrp1 (18) or Mdr2 (19) or random-primed DNA probes for Spgp (IMAGE clone 313236, GenBank accession number W11894) or Cyp3a16 (a complete coding sequence amplified by PCR from liver cDNA, kindly provided by J. W. Smit, The Netherlands Cancer Institute). Hybridization was performed in Ultrahyb buffer (Ambion, Austin, TX) under the recommended conditions, except for Cyp3a16, for which the hybridization and wash temperatures were reduced by 5°C to ensure detection of other homologous Cyp3a mRNAs. One blot was rehybridized with a cDNA probe for the mouse 18S rRNA as a loading control.

Drugs.

Paclitaxel, vinblastine sulfate, doxorubicin (Adriamycin), cytarabine, melphalan, and sodium arsenite were obtained from Sigma-Aldrich (Zwijndrecht, The Netherlands). Formulations of vincristine sulfate, cisplatin, daunorubicin, mitoxantrone, etoposide, and teniposide were obtained from local pharmaceutical suppliers; the topotecan formulation was from SmithKline Beecham (Brentford, Middlesex, England); and Taxotere (docetaxel) and SN-38 were from Rhône-Poulenc Rorer (Vitry-Alfortville, France).

Drug Sensitivity Assays.

Growth inhibition (IC₅₀) assays were performed by seeding 200 cells per well in 96-well plates in complete medium and, after cell attachment, applying drugs in a dilution series of 2-fold concentration steps, with each concentration in quadruplicate wells. After 4–4.5 days, when control wells were still subconfluent, cells were lysed in situ, and nucleic acids were stained with a proprietary dye (Cyquant; Molecular Probes, Eugene, OR) and quantified by fluorescence (485 nm excitation, 530 nm emission).

Descriptive and Inferential Statistics.

The SDs shown in Table 2 ⇓ represent variability in absolute IC₅₀ determinations across independent experiments. Sensitivity factors in Table 3 ⇓ are computed directly from the means shown in Table 2 ⇓ . Because distributions of drug resistance data are, by nature, positively skewed, IC₅₀ values were transformed to logarithms for the purpose of statistical comparisons (20) . All tests were two-tailed t tests assuming unequal variances.

Derivation and Characteristics of Fibroblast Cell Lines.

Mouse cell lines of three genotypes were generated and compared: wild-type, P-gp knockout (Mdr1a^−/−1b^−/−), and combined P-gp-Mrp1 knockout (Mdr1a^−/−1b^−/−Mrp1^−/−). The lines were obtained by spontaneous immortalization of MEFs (see“ Materials and Methods” for details). Their genetic background is a mix of 129/Ola and FVB strains, except for the NIH/3T3 line included for comparison, which was derived from an NIH/Swiss strain embryo (15) . The MEF lines to be analyzed were chosen from a larger set on the basis of favorable growth characteristics—stability, vigorous growth to high density, and single-cell cloning efficiency—but differ somewhat in these respects (Table 1) ⇓ . Although growth properties might be expected to affect drug sensitivity, in practice no systematic effects from this source were observed.

Expression of Multidrug Transporters in the Cell Lines and Mouse Tissues.

The wild-type cell lines are useful as models of innate drug sensitivity provided their expression of P-gp and Mrp1 is not substantially greater than that seen in normal tissues. P-gp and Mrp1 levels in wild-type and knockout cell lines and a panel of mouse whole tissue homogenates were therefore compared by immunoblot analysis (Fig. 1, A and B) ⇓ . P-gp levels in the wild-type cell lines were comparable with or lower than those seen in normal mouse tissues. The C219 antibody cross-reacts with Mdr2 P-gp (homologue of human MDR3) and Spgp, but their expression is mostly confined to liver (21 , 22) . Two observations exclude significant levels of either of these proteins in the fibroblast cell lines: Mdr2 and Spgp mRNAs were undetectable by Northern analysis of the cell lines but readily detectable in mouse liver (Fig. 2,A and B) ⇓ , and the C219 antibody detected no Mdr2 or Spgp protein in the six fibroblast lines null for Mdr1a/1b (Fig. 1A) ⇓ . Thus, the C219 signal in the wild-type cell lines reliably reflects their expression of P-gp.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Immunoblot analysis of P-glycoprotein (A) and Mrp1 (B) levels in mouse embryonic fibroblast lines versus normal mouse tissues. Fibroblast lines were grouped by genotype as shown. Each lane contained 30μ g of protein from total cell or tissue lysates. Pairs of PAGE gels were run together, blotted, probed, washed, treated with enhanced chemiluminescence reagents, and exposed together. The NIH/3T3 sample was included on both blots to ensure equivalence, although the lane was partially truncated on the right blot during transfer. The high molecular mass halves of the blots were probed with the anti-P-gp monoclonal antibody C219 or the anti-MRP1 monoclonal antibody MRPr1. C219 also detects mouse Mdr2 and Spgp proteins, present in some tissues but not in the cell lines (see “Results”). The higher molecular mass band (∼200 kDa) observed in some tissues in A is aggregated P-gp. The variable migration of P-gp and multiple Mrp1 bands arises from differences in glycosylation. The bottom halves of the blots were probed with a monoclonal antibody directed against α-tubulin (C) as a loading control; this is reliable for the fibroblast lines, but some variation in relative expression of tubulins should be expected for the tissues. Markers on the left are in kilodaltons.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Northern analysis of Mdr2 (A), Spgp (B), Cyp3a (C), and Bcrp1 (D) expression in the fibroblast cell lines. Probes for these mRNAs were hybridized to replicate blots containing 10 μg of total RNA/lane. One blot was subsequently rehybridized with an 18S rRNA cDNA probe as a loading control (E).

The substrate specificity of the drug-metabolizing cytochrome P450–3A family of enzymes overlaps that of P-gp; therefore, differences in their activity between different cell lines could, in principle, affect drug sensitivity. However, Northern analysis of the fibroblast lines with a Cyp3a16 probe at reduced stringency to recognize the major Cyp3a mRNAs did not reveal detectable expression in any of the lines, whereas a strong signal was observed in liver and intestine, as expected (Fig. 2C) ⇓ .

Mrp1 levels in the cell lines were somewhat higher than in most tissues examined. This result is adequate for the present purpose, given that Mrp1 tends to be concentrated in epithelial layers having substantially higher protein levels than the tissue taken as a whole (23, 24, 25) . This also holds true for P-gp (26) .

Sensitivity of the Lines to Antineoplastic Drugs.

The relative sensitivities of the cell lines to an extensive panel of antineoplastic drugs were determined by growth inhibition assay (Table 2) ⇓ . The drug panel contained representatives from the major families of clinically important antineoplastic agents that are known substrates for P-gp or Mrp1, including taxanes, Vinca alkaloids, anthracyclines, epipodophyllotoxins, and camptothecin derivatives, as well as an arsenical. Control drugs were also included from several classes that are poor or insignificant substrates of these proteins but for which acquired resistance is also a significant clinical problem: cisplatin, cytarabine (a nucleoside analogue), and melphalan (an alkylating agent). Mitoxantrone was included because it is an excellent substrate for the recently identified multidrug transporter BCRP/Bcrp1 (5 , 6 , 18) .

The data show that basal expression of P-gp and Mrp1 can have a major impact on innate resistance to substrate drugs. Table 2 ⇓ presents the mean IC₅₀ values, and Table 3 ⇓ , to aid comparisons, shows the sensitivity factors plus results of statistical tests of the differences between genotypes. The Mdr1a^−/−1b^−/− cell lines were substantially more sensitive than the wild-type lines to all drugs tested that are good P-gp substrates—the taxanes, anthracyclines, and Vinca alkaloids. The differences were surprisingly large given the relatively low levels of P-gp in the wild-type cell lines; the averages were 16-fold for paclitaxel, 4–5-fold for the anthracyclines, and ∼3-fold for Vinca alkaloids. It may be noted that the cell lines lacking P-gp expressed somewhat higher levels of Mrp1 than the wild-type lines (Fig. 1) ⇓ . It is not clear whether this is a coincidental or a compensatory effect. Either way, it might partially mask the effect of inactivating P-gp for drugs that are substrates of both transporters, Vinca alkaloids, epipodophyllotoxins, and possibly anthracyclines (see below).

The Mdr1a^−/−1b^−/−Mrp1^−/− cell lines were markedly more sensitive than the Mdr1a^−/−1b^−/− cell lines to several Mrp1 substrate drugs, most notably vincristine (>10-fold), sodium arsenite (∼7-fold), and the epipodophyllotoxins etoposide (7-fold) and teniposide (4-fold). Qualitatively, these differences in drug sensitivity are in agreement with those seen in previous studies of mouse embryonic stem cell lines and bone marrow mast cells in which only the Mrp1 gene had been inactivated by gene targeting (27 , 28) . However, the differences we observed are generally greater due, at least in part, to the absence of masking P-gp activity on drugs that are common substrates. For such drugs, the combined effect of eliminating the function of two transporters can sometimes be very striking, as exemplified by vincristine, for which the difference in sensitivity between Mdr1a^−/−1b^−/−Mrp1^−/− and wild-type lines averaged 28-fold.

Human MRP1 transports anthracyclines effectively and can confer resistance to these drugs, but it is unclear to what extent mouse Mrp1 does likewise (28, 29, 30, 31, 32) . We found that mouse fibroblast lines lacking both Mrp1 and P-gp were only slightly more sensitive to doxorubicin and daunorubicin than lines lacking only P-gp (∼1.5-fold), suggesting that these drugs are relatively poor substrates for mouse Mrp1.

It is interesting that the Mdr1a^−/−1b^−/−Mrp1^−/− cell lines were 2–3-fold more sensitive to the camptothecin derivatives topotecan and SN-38 than the Mdr1a^−/−1b^−/− lines. This result suggests that mouse Mrp1 can transport these drugs, in line with a recent report (33) that elevated expression of human MRP1 in drug-selected or MRP1-transfected cells mediates resistance to SN-38 and CPT-11 (another camptothecin derivative), which is associated with ATP-dependent drug efflux and reversed by MRP1 inhibitors.

Overexpression of P-gp is usually associated with only modest resistance to mitoxantrone (e.g., Refs. 34 ,, 35 ), implying that this drug is a relatively poor substrate. Increases in resistance to mitoxantrone have also been observed in MRP1-transfected cells (11) . In line with these results, we did observe moderately increased sensitivity to mitoxantrone in Mdr1a^−/−1b^−/− and Mdr1a^−/−1b^−/−Mrp1^−/− fibroblast lines of 2.5- and 4.3-fold, respectively, compared with wild-type lines, but only the latter difference was statistically significant (P = 0.041, two tailed). This is in part due to the considerable variability in resistance to mitoxantrone between different cell lines within each genotype. Mitoxantrone is a good substrate of the BCRP/Bcrp1 transporter (5 , 6 , 18) , and there is indeed a qualitative correlation between Bcrp1 mRNA levels (Fig. 2D) ⇓ and resistance to mitoxantrone. However, the relationship is not so good that Bcrp1 expression might account for all of the differences in sensitivity to this drug.

Within genotypes, substantial systematic differences in drug sensitivity were evident only in the wild-type group of cell lines; the WT1.2 and especially the NIH/3T3 lines were frequently more sensitive than the 2ac.1 and 2ac.2 lines. This pattern was most pronounced for P-gp substrate drugs and was not observed consistently for drugs that are neither P-gp nor Mrp1 substrates (cisplatin, cytarabine, and melphalan). The systematic differences are thus probably at least partly the result of the lower P-gp expression observed in the WT1.2 and NIH/3T3 lines (Fig. 1) ⇓ . The different mouse strain origin of the NIH/3T3 line might also contribute to the differences observed.