Using a rational screening of representatives of different classes of a small molecule compound library from Specs (www.specs.net) for potential inhibitors of ABCG2-mediated drug efflux, we identified a compound, N-(4-chlorophenyl)-2-[(6-{[4,6-di(4-morpholinyl)-1,3,5-triazin-2-yl]amino}-1,3-benzothiazol-2-yl)sulfanyl]acetamide) with benzothiazole linked to triazine ring backbone, (named PZ-39 thereafter, see Fig. 1) that drastically reversed mitoxantrone accumulation in MCF7/AdVp3000 cells that over-express ABCG2. As shown in Fig. 2A, PZ-39 enhanced mitoxantrone accumulation in MCF7/AdVp3000 but not the parental sensitive MCF7 cells that do not produce ABCG2, suggesting that ABCG2-mediated drug efflux has been inhibited. Because MCF7/AdVp3000 cells also over-express other ABC transporters such as ABCC3 in addition to ABCG2 [15], PZ-39 may inhibit ABC transporters other than ABCG2, leading to increased mitoxantrone accumulation. To directly test if PZ-39 inhibits ABCG2, we performed similar studies using ABCG2-transfected stable HEK293 (HEK293/ABCG2) cells [3]. As shown in Fig. 2B, pre-incubation of cells with PZ-39 enhanced intracellular mitoxantrone accumulation in HEK293/ABCG2 but not in the vector-transfected control (HEK293/Vec) cells. Thus, PZ-39 likely inhibits ABCG2-mediated mitoxantrone efflux. Figs. 2A and 2B also show that PZ-39 at 3.3 µM achieved equivalent level of effect to the known specific ABCG2 inhibitor FTC at 10 µM, suggesting that PZ-39 may be ∼3 times more potent than FTC.

To further investigate the potency of PZ-39 for ABCG2, the dose response effect of PZ-39 on mitoxantrone accumulation in HEK293/ABCG2 cells were determined using flow cytometry. As shown in Fig. 2C, the intracellular mitoxantrone level was increased by PZ-39 in a dose-dependent manner. At 3.3 µM, PZ-39 completely restored intracellular mitoxantrone level in HEK293/ABCG2 cells. FTC, on the other hand, achieved similar level of effect only at 10 µM (Fig. 2D), supporting the argument that PZ-39 is more potent than FTC (see above).

To investigate the potential use of PZ-39 as a chemo-sensitizer of ABCG2-mediated drug resistance, the effect of PZ-39 on drug response of HEK293/ABCG2 cells was determined in the absence or presence of 0.1 µM mitoxantrone which alone produced ∼10% cell killing. As shown in Fig. 3A and 3B, PZ-39 is not cytotoxic to HEK293/ABCG2 cells and its IC₅₀ is not measurable within the concentration range used whereas the IC₅₀ of FTC is ∼24 µM. The IC₅₀ of PZ-39 and FTC required to sensitize mitoxantrone resistance is ∼15 nM and ∼387 nM, respectively. The potency index of PZ-39 is estimated to be >1600 whereas that of FTC is only ∼62 (Table 1).

To further investigate the inhibitory activity of PZ-39 on ABCG2, the effects of PZ-39 on mitoxantrone cytotoxicity in HEK293/ABCG2 cells were evaluated in the presence of three different concentrations of PZ-39 (50, 200, and 500 nM) or the vehicle control (0.1% DMSO). As shown in Fig. 3C, PZ-39 at 50 nM significantly reduced the IC₅₀ of mitoxantrone with a sensitization index of 0.4 (Table 2). At 500 nM, the sensitization index of PZ-39 is 0.04 whereas that of FTC at the same concentration is 0.26 (Fig. 3D and Table 2). These results show that PZ-39 is a very potent novel ABCG2 inhibitor.

To investigate if PZ-39 can reverse ABCG2-mediated multidrug resistance in a drug resistant cancer cell line, we used the drug-selected MCF7/AdVp3000 cells and tested two additional anticancer drug substrates of ABCG2, Adriamycin and camptothecin. As shown in Fig. 3E, PZ-39 at 200 nM drastically reduced the resistance of MCF7/AdVp3000 to Adriamycin and Camptothecin similar to mitoxantrone, whereas the control FTC showed much less sensitization effect for all three drugs compared to PZ-39 at the same concentration.

To understand the mechanism of PZ-39 action in inhibiting ABCG2-mediated drug transport, we first investigated the kinetics of PZ-39 inhibition using isolated inside-out membrane vesicles [16]. We determined the change in mitoxantrone uptake in the presence of different concentrations of PZ-39. As shown in the Lineweaver-Burk plot (Fig. 4A), the Km and Vmax of mitoxantrone transport in the absence of PZ-39 were estimated to be 2.3 µM and 455 pmol/mg protein, respectively. It appears that both the Km and Vmax of mitoxantrone transport have been decreased in the presence of PZ-39 with an estimated Ki of ∼0.52 µM, suggesting that PZ-39 behaves as a mixed-type inhibitor of mitoxantrone transport [17]. Because it was also originally speculated that some inhibitors would compete with the ATP-binding site on ABCG2 transporter, we performed another experiment using ATP as a varying substrate. As shown in Fig. 4B, both the Km and Vmax of mitoxantrone transport were also altered by PZ-39. Thus, our study showed that the process of mitoxantrone transport and ATP binding may be inhibited by PZ-39 with a mixed type of mechanism. Likely, PZ-39 binds to a different site on ABCG2 from both mitoxantrone and ATP, not competitive inhibitor of either one of them.

We next tested if PZ-39 possibly inhibits ABCG2 oligomerization since ABCG2 has been suggested to function as a homodimer or higher forms of oligomers and oligomerization may be used as target for therapeutic drug development [2], [3]. For this purpose, co-immunoprecipitation of two differentially tagged ABCG2 was performed as previously described [2] following a 6-hr treatment with PZ-39 at 3.3 µM. However, no effect of PZ-39 on ABCG2 co-immunoprecipitation was found (supplemental Fig. S1), suggesting that PZ-39 does not affect ABCG2 oligomerization.

To further examine the mechanism of PZ-39 effect on ABCG2, we performed a western blot analysis of ABCG2 expression following PZ-39 treatment. As shown in Fig. 5A, the steady state level of ABCG2 protein drastically decreased at 1 day after PZ-39 treatment. But, it had only marginal decrease at 2 hours after PZ-39 treatment. As described above, PZ-39 was able to inhibit ABCG2-mediated mitoxantrone efflux of drug resistant cells with 1 hr of incubation. These findings suggest that PZ-39 may have two modes of action by inhibiting the activity (acute effect) and expression (chronic effect) of ABCG2 and consistent with our observation that PZ-39 is about 3 times better than FTC in drug accumulation assay (acute effect) but ∼7 times more potent in drug sensitization assay (chronic effect).

To determine if the chronic effect of PZ-39 on ABCG2 expression is at the mRNA level, we performed real-time RT-PCR analysis of MCF7/AdVp3000 and HEK293/ABCG2 cells treated with PZ-39 for various times up to 3 days. No significant change was found in ABCG2 mRNA level following PZ-39 treatment in either cell line (see supplemental Fig. S2). Thus, PZ-39 unlikely affects ABCG2 expression at its mRNA level. We next hypothesized that the mechanism of PZ-39-mediated inhibition would be a posttranscriptional process and examined the possibility that PZ-39 may accelerate the degradation of ABCG2. To test this possibility, HEK293/ABCG2 cells were pre-treated with cycloheximide, which acts by inhibiting elongation during protein synthesis, followed by treatment with PZ-39 for various times to determine ABCG2 degradation rate. As shown in Fig. 5B, the loss of ABCG2 in cells treated with a combination of PZ-39 and cycloheximide was much faster than that of the control treatment without PZ-39. The half-life of ABCG2 in the presence of PZ-39 is estimated to be ∼5 hrs whereas it is stable with an estimated half-life of ∼54 hrs in the control (Fig. 5C). Thus, PZ-39 likely accelerates the degradation of ABCG2 protein.

The accelerated degradation of ABCG2 by PZ-39 may be due to that PZ-39 induces conformational change and target ABCG2 for degradation. To determine if PZ-39 potentially causes conformational changes of ABCG2, we used the monoclonal antibody 5D3 which has been reported previously to bind to ABCG2 on cell surface more readily in the presence of ABCG2 inhibitors presumably due to inhibitor-induced conformational changes [12], [18]. As shown in Fig. 5D, PZ-39 caused an increase in 5D3 staining, suggesting a possible conformational change of ABCG2 upon PZ-39 binding. FTC also increased 5D3 staining as expected. However, it did not affect the level of ABCG2 (Fig. 5E), suggesting that the conformational change induced by FTC and PZ-39 may be different. It has been reported recently that two distinct pathways exist for degradation of wild-type and mutant ABCG2 proteins [19]. While wild-type and correctly-folded protein is degraded in lysosomes, the mutant and misfolded protein is involved in ubiquitin-mediated protein degradation in proteasomes. To further determine the mechanism of PZ-39-induced ABCG2 degradation, we employed Bafilomycin A₁, an inhibitor of protein degradation in lysosomes, and MG-132, a proteosome inhibitor. As shown in Fig. 5F, co-treatment of cells with Bafilomycin A₁ and PZ-39 inhibited PZ-39-induced ABCG2 degradation whereas co-treatment with MG-132 and PZ-39 did not, indicating that PZ-39-induced ABCG2 degradation is likely lysosome-dependent. Taken together, we conclude that PZ-39 causes conformational change of ABCG2 and targets it for normal degradation in lysosomes.

To determine the specificity of PZ-39, we tested the effect of PZ-39 on the other two important ABC transporters well known in MDR, ABCB1 and ABCC1. The effect of PZ-39 on ABCB1 and ABCC1-mediated decrease in intracellular Adriamycin accumulation was tested using MCF7 cells-transfected with ABCB1 (BC19) [20] and HEK293 cells-transfected with ABCC1 (HEK293/ABCC1) [16], [21]. We found no effect of PZ-39 on the activity of ABCB1 and ABCC1 in decreasing Adriamycin accumulation (see supplemental Fig. S3A). We also found no effect of PZ-39 on the steady state protein level of ABCB1 and ABCC1 after 3 days of treatment (Fig. S3B). Thus, likely PZ-39 is specific to ABCG2 among these three well known MDR-causing ABC transporters that are believed to play important roles in clinical drug resistance.

To better understand the pharmacophore of PZ-39, we apprehended 3 analogues of PZ-39 for drug accumulation and resistance assays using HEK293/ABCG2 cells in comparison with PZ-39. These analogues (C6, C8, and E2) all have the same intact rings A, B and C as PZ-39 but with different side groups (Fig. 1). All three analogues had similar activity as PZ-39 in enhancing mitoxantrone accumulation (Fig. 6A) and sensitizing mitoxantrone resistance in HEK293/ABCG2 (Fig. 6B). To determine the chronic effect of these analogues on ABCG2 expression, we performed a western blot analysis following treatment with C6, C8, and E2 for various times up to 3 days. As shown in Fig. 6C, all three analogues caused the decreased expression of ABCG2. Furthermore, all these analogues also caused conformational changes in ABCG2 similar to PZ-39 (Fig. 6D). Thus, likely the benzothiazole linked to a triazine ring backbone may be the core structure for binding to and inhibiting ABCG2 function and stability.

Monoclonal antibody BXP-21 against ABCG2, anti-Myc and anti-HA antibodies were from ID Labs, Cell Signaling, and Roche, respectively. Monoclonal antigody against Pgp, C219, was a kind gift from Dr. Victor Ling (The British Columbia Cancer Center, Vancouver, Canada). Monoclonal antibody against MRP, MRPr1, were purchased from Kamiya Biomedical Company. Biotin-conjugated 5D3 antibody and Phycoerythrin-Streptavidin conjugates were from eBiosciences. All electrophoresis reagents, protein concentration assay kit, precast polyacrylamide gradient gels and polyvinylidene difluoride membranes were purchased from Bio-Rad. FTC, adriamycin, mitoxantrone, camptothecin, DTT, Sulforhodamine B (SRB), and Triton X-100 were from Sigma. Protein-G PLUS-Agarose and SYBR Green PCR Master Mix were from Santa Cruz Biotechnology and Applied Biosystems, respectively. LipofectAMINE Plus and G418 were from Invitrogen. Cell culture medium IMEM, DMEM, and [³H]mitoxantrone were from BioSource International, Media Tech., and Moravek Biochemical, respectively. PZ-39 and three related compounds were purchased from SPECS. All other chemicals were of molecular biology grade from Sigma or Fisher Scientific.

Human breast cancer cell line MCF7 (ATCC) and its derivative lines BC19 (a gift from Julie Horton at National Institute of Environmental Health Sciences) and MCF7/AdVp3000 (a gift from Susan Bates at National Cancer Institute), HEK293/ABCC1, HEK293/vector, HEK293/ABCG2 were cultured as previously described [2], [16], [20], [25]. Lysate preparation was performed as described previously [25]. Cell membranes were prepared in exactly the same way as previously described [16] and final membranes were resuspended in STBS (250 mM sucrose, 150 mM NaCl, 10 mM Tris/HCl, pH7.5).

Western blot, immunoprecipitation, and flow cytometry analysis of drug accumulation were performed exactly as we previously described [2], [3]. To determine the mechanism of ABCG2 degradation, HEK293/ABCG2 cells were first treated with 10 nM Bafilomycin A₁ or 2 µM MG132 for 24 hrs followed by additional treatment with 3 µM PZ-39 for various times. Cell lysates were then collected for western blot analysis of ABCG2. To determine the half-life of ABCG2, HEK293/ABCG2 cells were treated with 5 µg/ml cycloheximide, 3 µM PZ-39, or both for various times followed by collection of cell lysates for western blot analysis of ABCG2 expression. To determine the change in antibody 5D3 staining following treatment with inhibitors, HEK293/ABCG2 cells were incubated with 10 µM PZ-39, C6, C8, E2, or FTC at 37°C for 30 min before biotin-conjugated 5D3 antibody (1∶100 dilution) was added and incubated for 2 hrs. Then, the cells were washed 3 times and incubated with Phycoerythrin-Streptavidin for 30 min followed by washing for 3 times and analyzed by flow cytometry.

RNA extraction and real-time RT-PCR were performed as we described previously [25]. The sequences of ABCG2 primers are 5′-GGCTTTCTACCTGCACGAAAACCAGTTGAG-3′ (forward) and 5′-ATGGCGTTGAGACCAG-3′ (reverse). The sequences of GAPDH primers are 5′-AAGGACTCATGACCACAGTCCAT-3′ (forward) and 5′-CCATCACGCCACAGTTTCC-3′ (reverse). The relative ABCG2 RNA level (2^ΔCT) treated with inhibitors was expressed as percentage of the control (in the presence of 0.1% DMSO) where ΔCT (threshold cycle) = (CT_ABCG2-CT_GAPDH).

Cytotoxicity was determined using SRB colorimetric assay as previously described [25]. The effect of compound inhibitors on drug resistance was determined by exposing cells to a range of concentrations of anticancer drugs such as mitoxantrone in the absence or presence of different concentrations of the inhibitor. The potency and sensitization index of the inhibitors were calculated as follows:

Drug accumulation assay was performed as described previously [16], [26] with some modifications. Briefly, 10⁶ cells in culture were pre-incubated with various concentrations of PZ-39, FTC, or vehicle control (0.1% DMSO) for 1 hr at 37°C, followed by addition of 20 µM mitoxantrone and incubation for 30 min. The reaction was stopped by addition of ice-cold PBS and centrifugation, washed with ice-cold PBS, and subjected analysis flow cytometry.

Drug-uptake assay using membrane vesicles was performed as we previously described [16] using [³H]mitoxantrone as ABCG2 substrate. Kinetic analyses was performed using data generated in the presence of different concentrations of [³H]mitoxantrone, ATP, and PZ-39 to generate Lineweaver-Burk plot. Kinetic constants Km, Vmax, and Ki were calculated using procedures as previously described [27].