P-glycoprotein-mediated acquired multidrug resistance of human lung cancer cells in vivo.

We examined whether the increased expression of P-glycoprotein (P-gp) encoded by the human multidrug resistance gene MDR1 is related to the acquired multidrug resistance of lung cancer in vivo. We estimated the chemosensitivity of lung cancer xenografts (LC-6, adenocarcinoma; Lu-24, small-cell cancer) by calculation of relative tumour growth (T/C%, treated/control) in vivo, based on statistical significance determined by the Mann-Whitney U test (P < 0.01, one-sided). MDR1 gene expression levels were evaluated by reverse transcription-polymerase chain reaction (RT-PCR) assay. P-gp production and P-gp localisation were examined by Western blotting and by immunohistochemical analysis respectively. LC-6 and Lu-24 were initially sensitive to both vincristine (VCR, 1.6 mg kg-1: LC-6, 45%; Lu-24, 39%) and doxorubicin (DOX, 12 mg kg-1: LC-6, 26%; Lu-24, 27%) in vivo. VCR-resistant variants (LC-6R, 66% and Lu-24R, 68%) selected with VCR (0.4 mg kg-1, x 9) significantly acquired cross-resistance to DOX (LC-6R, 55% and Lu-24R, 55% respectively). RT-PCR assay showed increased levels of MDR1 expression in LC-6R and Lu-24R with stable MDR1 expression levels. P-gp expression levels were elevated, and the percentage of P-gp-positive tumour cells increased in both LC-6R and Lu-24R. These results suggest that P-gp/MDR1 overexpression is related to acquired multidrug resistance in lung cancer in vivo.

Lung cancer is generally treated by a combination of therapeutic protocols using cisplatin, vinca alkaloids, doxorubicin (DOX) and etoposide (VP-16) (Britran et al., 1988;Williams, 1989;Hansen, 1992). However, the failure of chemotherapy as a result of cellular drug resistance is still a major problem in the treatment of lung cancer. Especially, development of acquired drug resistance in tumours initially sensitive to chemotherapy is a major issue in the treatment of lung cancer patients.
Mechansims of multidrug resistance were analysed in various human neoplastic cell lines resistant to anti-cancer agents in vitro (Chen et al., 1986(Chen et al., , 1990Ueda et al., 1987). Selection of cells resistant to lipophilic compounds (DOX, vinca alkaloids, podophyllotoxins and colchicine) results in the development of cross-resistance to other related drugs (Fojo et al., 1985). This classical multidrug resistance phenomenon is known to be related to the overexpression of P-glycoprotein (P-gp) encoded by the human multidrug resistance gene (MDR1) (Gros et al., 1986). Recently, atypical multidrug resistance induced by overexpression of multidrug resistance-associated protein (MRP) has been reported in lung cancer cells in vitro (Cole et al., 1992;Versantvoort et al., 1992;Zaman et al., 1993).
Our previous clinicopathological studies have not shown intrinsic multidrug resistance in non-small-cell lung cancer (NSCLC) to be related to P-gp/MDRJ (Abe et al., 1994a). However, certain pulmonary adenocarcinomas revealed significantly increased MDR1 expression. Many authors have also reported drug resistance mechanisms associated with P-gp in lung cancer Volm et al., 1991;Holzmayer et al., 1992). We did not find multidrug resistance to be intrinsically related to MDR] overexpression in lung cancer xenografts (including  in vivo (Abe et al., 1994b). However, it has not been clarified whether acquired multidrug resistance is related to increased levels of MDR1 expression in human cancer cells in vivo.
In this study, we selected VCR-resistant variants from human NSCLC (LC-6R) and small-cell lung cancer (SCLC, Lu-24R) xenografts in vivo, and evaluated whether these VCR-resistant xenografts showed cross-resistance to DOX in chemosensitivity tests in vivo. The expression levels of P-gp/ MDR] were also analysed before and after selection in these xenografts. We also examined the gene expression levels of miscellaneous factors associated with multidrug resistance including MRP, topoisomerase IIcx (Topo Ila) and glutathione-S-transferase-ir (GST-7t) in the xenografts. We discuss here the hypothesis that acquired multidrug resistance is induced by the increased expression of P-gp/ MDR1 in lung cancer in vivo.

Materials and methods
Human lung cancer xenografts Two human xenografts (LC-6, NSCLC, adenocarcinoma; Lu-24, SCLC, oat-cell type) were originally established at the Central Institute for Experimental Animals (Kanagawa, Japan) from primary lung cancer materials from patients who had received no anti-cancer chemotherapy. The tumour xenografts were maintained by serial subcutaneous transplantation in nude mice (BALB/c-nu/nu, Clea Japan, Tokyo), and used at 10-20 passages in this study. Xenograft specimens obtained from mice sacrificed under deep anaesthesia were frozen and stored at -80°C until analysed. Tumour xenografts were also prepared for routine histopathological values.
Correspondence: M Nakamura, Department of Pathology, Tokai University School of Medicine, Bohseidai, Isehara-shi, Kanagawa 259-11, Japan Received 22 March 1996; revised 12 July 1996; accepted 16 July 1996 Establishment of VCR-resistant xenografts in vivo The human NSCLC (LC-6) and SCLC (Lu-24) xenografts were sensitive to the maximum tolerated doses (MTDs) of both VCR and DOX in vivo. We selected VCR-resistant xenografts, LC-6R and Lu-24R, from LC-6 and Lu-24, respectively, by serial passage in mice and by administration of VCR (0.4 x 9 mg kg-') in vivo, according to our previous report (Abe et al., 1993). No significant morphological differences were noted between parental and VCR-resistant xenografts.
We performed in vivo chemosensitivity tests on the lung cancer xenografts (LC-6, LC-6R, Lu-24 and Lu-24R) according to the procedures reported previously (Inaba et al., 1988(Inaba et al., , 1989. Six female mice (BALB/c-nu/nu, 6-15 weeks old) bearing xenografts (tumour volume: 100 -300 mm3) were given the MTD of VCR (1.6 mg kg-1) or DOX (12 mg kg-'). The tumour volume (V) was calculated by the equation, V= 1/2 x A x B2, in which A and B are the experimental measurements in mm of length and width respectively. Growth of the tumour xenografts was measured by the relative tumour volume (RV), which was expressed as RV= V14/ VO, in which V14 iS the tumour volume at day 14 and V0 is the initial tumour volume when the treatment was started (day 0). The effects of the drugs were represented by RV of the xenografts, and the T/C% values were defined as the ratio of the RV of the treated tumour xenografts to controls after drug administration. Animal experiments were carried out in accordance with the guidelines established by the Central Institute for Experimental Animals.
We examined the effects of prior inoculation with the P-gp inhibitor CysA on the sensitivity to anti-cancer agents in the xenografts in vivo, according to our previous report (Abe et al., 1996). Nude mice bearing tumour xenografts were treated with VCR (0.4 mg kg-') or DOX (12 mg kg-') 3 h after intravenous administration of CysA (50 mg kg-'). VCR was used at the low concentration of 0.4 mg kg in this study because we had certified in advance that co-administration of high doses of VCR (1.6 mg kg-') was fatal for the mice with CysA.
Reverse transcription-polymerase chain reaction (RT-PCR) assay Total cellular RNA specimens were prepared from frozen materials (Sambrook et al., 1989). The expression levels of MDR] transcripts were determined by the modified RT-PCR procedure described previously (Noonan et al., 1990), using the following primers: sense, AAGCTTAGTACCAAA-GAGGCTCTG, nucleotides 2041-2046; antisense, GGCTA-GAAACATAGTGAAAAACAA, nucleotides 2260 -2283 (Abe et al., 1994a). The primer sequences were derived from exon 16 and exon 18, respectively, separated by introns to prevent amplification of contaminating genomic DNA. We avoided amplification of contaminating murine MDR gene transcripts in the tumour xenografts by using the above primers specific for the human MDR] gene (Abe et al., 1994b). RT-PCR with these primers amplified a 243 bp fragment of MDR] cDNA. We estimated MDRJ expression level in comparison with that of the housekeeping gene J2microglobulin (,B2m).
Northern blot analyses Total cellular RNA samples (20 ,ug) fractionated through agarose gels were blotted onto nitrocellulose membranes (Gene Screen Plus, New England Nuclear), and the blots were hybridised with 32P-labelled MDR1 cDNA probe (kindly supplied by Dr I Pastan, National Cancer Institute, Bethesda, MD, USA). The level of MDRJ-specific transcript expression (4.2 kb) was evaluated in comparison with that of the housekeeping gene ,B-actin (2.2 kb). Levels of expression of Topo Ila, GST-i and MRP genes were also evaluated in the xenografts by Northern blot analysis. Complementary DNAs (Topo Ila, Dr T Ando; rat GST-i cDNA, Dr A Sugioka through the Japanese Cancer Research Resources Bank) were used. A human MRP cDNA was prepared by PCR amplification of the fragment corresponding to nucleotides 240-503 from KB8-5 cells (Ota et al., 1995). We evaluated expression of each gene-specific transcript (Topo Ila, 4.6 kb; GST-n, 0.7 kb; MRP, 6.5 kb).

Results
In vivo drug sensitivity The growth rates of human lung cancer xenografts in the chemosensitivity tests are shown with relative tumour volume (Figures 1-3). Evaluation as 'sensitive' was defined based on statistical significance determined by the Mann-Whitney Utest (P<0.01, one-sided) (Abe et al., 1994b).
LC-6 was sensitive to the MTD of both VCR and DOX (Figure la), and LC-6R selected in vivo by VCR was resistant to VCR and acquired cross-resistance to DOX (Figure la). Lu-24 was also initially sensitive to the MTD of both VCR and DOX (Figure lb), and Lu-24R selected by VCR was resistant to VCR and acquired cross-resistance to DOX (Figure lb). Table I shows T/C% values of each xenograft in vivo on day 14 after drug administration. The T/C% values of LC-6 exposed to the MTD of VCR (45%) and DOX (26%) were significantly lower than those (66% and 55%) of LC-6R. The T/C% values of Lu-24 to the MTD of VCR (39%) and DOX (27%) were also significantly lower than those of Lu-24R (68% and 55%).
This acquired drug resistance in LC-6R was circumvented by co-administration of CysA (Figure 2). The acquired drug resistance to VCR of LC-6R was reversed by co-administration of CysA (T/C%: 90% to 38%), which when administered alone showed no anti-cancer effect. The acquired cross-resistance of LC-6R to DOX was also circumvented by co-administration of CysA (T/C%: 55% to 15%). CysA did not apparently affect the growth of LC-6, when it was administered with or without anti-cancer drugs (data not shown).
The changes in responsiveness to non-P-gp-mediated anticancer agents (cisplatin and mitomycin) were not significantly different between LC-6 and LC-6R, while LC-6R showed a 3fold greater susceptibility to mitomycin C (Table I). RT-PCR assay showed no MDRJ expression in LC-6 or Lu-24 xenografts (Abe et al., 1994b). LC-6R and Lu-24R with acquired cross-resistance, however, showed increased levels of MDRI expression compared with the sensitive parent xenografts LC-6 and Lu-24 (Figure 3). The xenografts LC-6R and Lu-24R serially transplanted into nude mice (four generations) without VCR showed no marked fluctuations in the levels of MDRJ expression. P-gp production The VCR-resistant xenografts, LC-6R and Lu-24R, showed increased production of P-gp protein by Western blotting in comparison with the respective parent xenografts (LC-6 and Lu-24 respectively; Figure 4). Immunohistochemical anlaysis with anti-P-gp polyclonal antibody (Ab-1) also revealed marked increases in the number of P-gp-positive tumour cells in LC-6R and Lu-24R compared with their respective parental xenografts (Figure 5), whereas LC-6 and Lu-24 xenografts showed no P-gp-positive tumour cells.   Relative tumour volume (RV) = V14/ Vo, where VI4 iS tumour volume at day 14 and V0 is the initial value at the beginning of treatment (day 0). T/C%, growth ratio of the relative volume of the treated xenografts to controls (untreated) on day 14 of treatment (VCR, 1.6 mg kg 1, *0.4 mg kg-1; DOX, 12 mg kg-1; CDDP, 7 mg kg-'; MMC, 1.7 mg kg '). U-test, significance of differences were estimated by the Mann-Whitney U-test (P< 0.01, one-sided; +, significant; -, not significant).

Discussion
Many studies using human tumour cell lines have revealed that multidrug resistance mechanisms are correlated to the overexpression of P-gp/MDRI in vitro (Chen et al., 1990;Roninson, 1991). It has, however, not been clearly demonstrated whether acquired multidrug resistance is influenced by P-gp/MDRI overexpression in human cancers in vivo (Starling et al., 1990). The VCR-resistant variants (LC-6R and Lu-24R) selected in vivo from drug-sensitive xenografts (LC-6 and Lu-24) showed cross-resistance to DOX, and the drug resistance to VCR and DOX of LC-6R was overcome by co-administration of the P-gp inhibitor, CysA. RT-PCR assay showed increased levels of MDR] expression in LC-6R and Lu-24R, whereas no marked changes were seen in the expression of other miscellaneous drug resistance-related factors (Topo IIa, 1933 GST-7 and MRP) (Zwelling et al., 1990;Nakagawa et al., 1990;Cole et al., 1992). In LC-6R and Lu-24R, P-gp expression levels were elevated and P-gp-positive tumour cells increased. These results supported the hypothesis that acquired multidrug resistance is induced by increased P-gp protein/MDRJ gene expression in human lung cancer xenografts.
Western blotting showed small amounts of P-gp in LC-6 and Lu-24, whereas a highly sensitive RT -PCR assay revealed no MDRJ expression in these sensitive xenografts. In this RT -PCR assay, we selected primers which were specific for human MDR] and did not amplify the murine mdr gene. Immunohistochemical analysis showed no P-gppositive tumour cells in these sensitive xenografts. Therefore, the signals seen in LC-6 and Lu-24 might have included nonspecific reactions to murine P-gp-related molecules probably in the stromal elements by Western blotting with murine monoclonal anti-P-gp antibody, C219.
Several studies have shown that NSCLC with neuroendocrine properties expresses high levels of P-gp/MDRI , while some authors revealed that the expression levels of P-gp/MDRI in lung cancer were not so high (Fojo et al., 1987;Goldstein et al., 1989). On the other hand, we reported enhanced MDRJ expression in a limited number of pulmonary adenocarcinomas (Abe et al., 1994a). However, it has not been demonstrated conclusively whether acquired multidrug resistance in lung cancer is related to P-gp/MDRI overexpression in vivo. The results presented here strongly support the hypothesis that acquired multidrug resistance is related to the increased expression of P-gp/MDRI in pulmonary adenocarcinoma and small-cell lung carcinoma in vivo.
The multidrug-resistant xenografts expressed MDR] at lower levels than the in vitro multidrug-resistant carcinoma line, KB8-5. It is very important to determine how MDR] expression levels can induce the multidrug resistance of tumour cells in lung cancer in vivo. Previously, we suggested that in vivo sensitivity assays more accurately reflect drug resistance as a result of low-level MDR] overexpression in the human epidermoid carcinoma KB line (Abe et al., 1996). Reduced levels of MDR] expression might be related to P-gpmediated multidrug resistance in vivo compared with that in vitro.
It is difficult to determine whether the observed multidrug resistance phenotype was caused by the clonal selection of intrinsically P-gp-positive cancer cells or the activated production of P-gp in resistant cancer cells (Chaudrey et al., 1993;Chen et al., 1994;Brock et al., 1995). The multidrug-resistant xenografts, used in this study showed stable MDRI expression during four serial passages without exposure to VCR. Immunohistochemical analysis revealed definite P-gp-positive cancer cells in multidrug-resistant xenografts, whereas no P-gp-positive tumour cells were detected in the parental xenografts. It is impossible to conclude from these results whether the observed multidrug resistance was owing to clonal selection of P-gp-expressing cells or the activated production of P-gp.
Recently, the mechanism of atypical multidrug resistance in lung cancer by MRP has been discussed (Cole et al., 1992). Previously we demonstrated the clinical relevance of MRP overexpression in the intrinsic multidrug resistance of NSCLC, especially in pulmonary squamous cell carcinoma (Ota et al., 1995). We are also currently engaged in studies to determine the relevance of MRP in the acquired multidrug resistance phenotype in pulmonary squamous cell carcinoma xenografts.