Di-2-pyridylketone 4,4-Dimethyl-3-thiosemicarbazone (Dp44mT) Overcomes Multidrug Resistance by a Novel Mechanism Involving the Hijacking of Lysosomal P-Glycoprotein (Pgp)*

Background: There is a critical need for chemotherapeutics that overcome multidrug resistance (MDR). Results: Dp44mT is transported into the lysosome by Pgp, causing lysosomal targeting of Dp44mT and resulting in enhanced cytotoxicity in vitro and in vivo. Conclusion: Dp44mT overcomes MDR via utilization of lysosomal Pgp transport activity. Significance: DpT thiosemicarbazones offer a new therapeutic strategy to overcome MDR via utilization of lysosomal Pgp transport activity. Multidrug resistance (MDR) is a major obstacle in cancer treatment. More than half of human cancers express multidrug-resistant P-glycoprotein (Pgp), which correlates with a poor prognosis. Intriguingly, through an unknown mechanism, some drugs have greater activity in drug-resistant tumor cells than their drug-sensitive counterparts. Herein, we investigate how the novel anti-tumor agent di-2-pyridylketone 4,4-dimethyl-3-thiosemicarbazone (Dp44mT) overcomes MDR. Four different cell types were utilized to evaluate the effect of Pgp-potentiated lysosomal targeting of drugs to overcome MDR. To assess the mechanism of how Dp44mT overcomes drug resistance, cellular studies utilized Pgp inhibitors, Pgp silencing, lysosomotropic agents, proliferation assays, immunoblotting, a Pgp-ATPase activity assay, radiolabeled drug uptake/efflux, a rhodamine 123 retention assay, lysosomal membrane permeability assessment, and DCF (2′,7′-dichlorofluorescin) redox studies. Anti-tumor activity and selectivity of Dp44mT in Pgp-expressing, MDR cells versus drug-sensitive cells were studied using a BALB/c nu/nu xenograft mouse model. We demonstrate that Dp44mT is transported by the lysosomal Pgp drug pump, causing lysosomal targeting of Dp44mT and resulting in enhanced cytotoxicity in MDR cells. Lysosomal Pgp and pH were shown to be crucial for increasing Dp44mT-mediated lysosomal damage and subsequent cytotoxicity in drug-resistant cells, with Dp44mT being demonstrated to be a Pgp substrate. Indeed, Pgp-dependent lysosomal damage and cytotoxicity of Dp44mT were abrogated by Pgp inhibitors, Pgp silencing, or increasing lysosomal pH using lysosomotropic bases. In vivo, Dp44mT potently targeted chemotherapy-resistant human Pgp-expressing xenografted tumors relative to non-Pgp-expressing tumors in mice. This study highlights a novel Pgp hijacking strategy of the unique dipyridylthiosemicarbazone series of thiosemicarbazones that overcome MDR via utilization of lysosomal Pgp transport activity.

resistance to chemotherapeutics by using MDR modulators have not generated useful outcomes in clinical trials (4,5). Hence, there is an increasing need to develop drugs that effectively target drug-resistant tumors.
There has been great interest in understanding the mechanism of action of agents that are more effective in drug-resistant cells than in their drug-sensitive counterparts (6,7). One of these agents, di-2-pyridylketone 4,4-dimethyl-3-thiosemicarbazone (Dp44mT; Fig. 1A) (8), has been described to overcome MDR in vitro by an unknown mechanism (9) and to be highly effective and selective against a variety of belligerent solid human tumors in vivo by the intravenous and/or oral routes (9 -12). An important aspect of the activity of these agents was shown to be due to their complexation with copper in lysosomes to form redox active complexes that caused lysosomal membrane permeabilization (LMP) and apoptosis (13).
The current study offers an unexplored approach describing how functional Pgp on the lysosomal membrane can be "hijacked" by agents, such as Dp44mT, to potentiate cytotoxicity in MDR cancers. Herein, we highlight the molecular mechanism and properties of agents required to overcome MDR. Moreover, the potentiated anti-cancer activity of Dp44mT in Pgp-expressing MDR cells, versus drug-sensitive cells, was confirmed in a human tumor in mice. Hence, this study describes a novel mechanism of action and identifies a new strategy for designing chemotherapeutics to overcome MDR by hijacking lysosomal Pgp to increase sequestration of redox active, lysosomotropic Pgp substrates into lysosomes. This effect potently enhances cytotoxicity by targeting Dp44mT to the lysosome, which is a key target of this agent, leading to LMP and death of the resistant cancer cell. This property is unique, is not found for current chemotherapeutics that are Pgp substrates, and depends on the redox activity of the Dp44mT-Cu complex. Notably, this mechanism is totally opposite to that found for standard cytotoxic drugs that are Pgp substrates, such as DOX. Indeed, in this latter case, Pgp expression results in DOX efflux and its storage in the lysosome where the organelle acts as a "safe house," preventing cytotoxicity and leading to resistance against the chemotherapeutic.
Cell Culture-Human cervical carcinoma-derived KB31 cells, the small cell lung carcinoma cell line, DMS-53, the colon adenocarcinoma cell line, HCT-15, and the MCF7 and MDA-MB-231 breast carcinoma cell lines were obtained from the American Type Culture Collection (Manassas, VA), whereas VBL-resistant KBV1 cells (grown in 1 g/ml VBL) were a gift from Dr. Maria Kavallaris (Children's Cancer Institute Australia, Sydney, Australia). The 2008 human ovarian carcinoma cell line and the paclitaxel-resistant 2008/P200 cell line (grown in 200 ng/ml paclitaxel) were from Dr. John Allen (Centenary Institute, Sydney). All cells were grown in DMEM (Life Technologies, Inc.) under standard conditions (17).
MTT Proliferation Assay-Proliferation was examined using the MTT assay and validated utilizing viable cell counts via trypan blue (13). The cells (3 ϫ 10 3 /well) were seeded in 96-well plates and preincubated with medium alone or Pgp inhibitors (1 M Val or 0.1 M Ela) or lysosomotropic agents (5 mM NH 4 Cl, 1 M CLQ, or 100 M MA) for 30 min at 37°C. This medium was then replaced, and the cells were incubated with DOX, VBL, Dp44mT, or Cu[Dp44mT] in the continued absence or presence of Pgp inhibitors or weak bases for 2, 24, or 72 h at 37°C and processed (13).
Transient Pgp Silencing Using siRNA-Pgp silencing (MDR1-siRNA; catalog no. 4123; Life Technologies, Inc.) was performed as described (17). Briefly, the siRNA-Lipofectamine mixture (50 nM MDR1-siRNA and 1:400 Lipofectamine 2000) was added to the cells (at 30% confluency) and incubated for 72 h at 37°C prior to further experiments. The effectiveness of Pgp silencing was confirmed using Western blotting and DOX and Dp44mT cytotoxicity via the MTT assay. As a relevant control, negative control siRNA with no genomic homology (Life Technologies, Inc.) was used at the same concentration as MDR1 siRNA.
Pgp-ATPase Activity Assay-The ATPase activity of Pgp was determined using Pgp-enriched membranes and a luminescent ATP detection kit (Pgp-Glo TM ATPase assay; Promega) according to the manufacturer's instructions. Briefly, Pgp-enriched membranes (0.5 mg/ml) and Mg(II)-ATP (5 mM) were incubated in the absence or presence of sodium orthovanadate (100 M) for 40 min at 37°C, and ATP levels were detected as a luciferase-generated luminescent signal. Basal Pgp-ATPase activities were calculated as the difference between the ATP hydrolysis in the presence or absence of sodium orthovanadate. As relevant negative controls, the well characterized Pgp inhibitors Ela (0.1 M) and Val (1 M) were used to inhibit Pgp-ATPase activity (18). As a positive control, verapamil-stimulated Pgp-ATPase activity was measured in the presence of the Pgp control substrate, verapamil (200 M) (19). The test substrates, Dp44mT, Bp4eT, or their Fe(III) and Cu(II) complexes all at 50 M, were dissolved in DMSO (Sigma-Aldrich). The final concentration of DMSO in the assay medium was equal to or less than 0.25%. Control experiments indicated that DMSO at this concentration had no effect on ATPase activity.
Uptake/Efflux of [ 14  Rh123 Retention Assay-Cells were preincubated with a range of concentrations (0.001-10 M) of Dp44mT, Val, or Ela for 30 min at 37°C, followed by incubation with Rh123 (1 g/ml) for 15 min at 37°C in the presence of the Pgp inhibitors or Dp44mT (17). Samples were processed and analyzed using a FACS Canto flow cytometer (BD Biosciences; 10,000 events/ sample) and FlowJo software (v.7.5.5; Tree Star Inc., Ashland, OR), respectively.
LysoTracker Red (Life Technologies, Inc.) and acridine orange (Sigma-Aldrich) were used to determine LMP (13). Cells were incubated with LysoTracker Red (50 nM) or acridine orange (20 nM) for 15 min at 37°C, washed three times with ice-cold PBS, and then incubated with Cu[Dp44mT] or Dp44mT alone (all at 25 M) for 30 min or 24 h at 37°C in the presence or absence of Val (1 M) or Ela (0.1 M). LysoTracker Red staining of cells was detected using the microscope/software above. Acridine orange staining was analyzed using the flow cytometer above.
Redox Studies: Oxidation of H 2 DCF-The effect of Val (1 M), Ela (0.1 M), or tetrathiomolybdate (5 M) on reactive oxygen species (ROS) generation by Cu[Dp44mT] (5 M) was examined using H 2 DCF (5 M) in the absence of cells over 12 min at 20°C using standard methods (13). Studies were conducted in an acetate buffer (150 mM; pH 5) containing cysteine (100 M) to mimic lysosomal conditions (13). In these experiments, H 2 O 2 (100 M) was added to initiate ROS generation. To confirm ROS production, DMSO (10% v/v) was used, because it is an effective hydroxyl radical scavenger (21).
Tumor Xenografts in Nude Mice and Dp44mT Administration-All studies involving animals were performed in accordance with ARRIVE guidelines (22) and the University of Sydney Animal Ethics Committee. In mouse studies, sample sizes were predetermined based on previous experience with Dp44mT (9) using a minimum of six mice/group, and the experiment was replicated twice to confirm findings. KBV1 or KB31 cells (3 ϫ 10 6 ) were suspended in Matrigel (BD Biosciences, San Jose, CA) in a 1:1 ratio and injected subcutaneously into the left (KBV1 cells) or right (KB31 cells) flank of each 7-week-old female BALB/c nu/nu mouse. After engraftment, tumor size was measured using Vernier calipers, and the tumor volumes were calculated (9). When tumor volumes reached 120 mm 3 , intravenous (tail vein) vehicle or Dp44mT treatment began (day 0). The mice were randomly assigned to treatment groups, and where possible, treatment groups were blinded until statistical analysis. No animals or potential outliers were excluded from the data sets presented in this study. Dp44mT (0.1 or 0.2 mg/kg) was dissolved in 30% propylene glycol in 0.9% saline and injected intravenously over 5 consecutive days/week (9). Control mice were treated with the vehicle alone. The health of mice was assessed by monitoring weight, behavior, and comprehensive biochemical/hematological analyses (9).
Pgp Immunohistochemistry-Tumor sections were deparaffinized, rehydrated through graded ethanol solutions, and brought to distilled water. They were then subjected to heat retrieval/pressure (125°C/30 s) and immersed in pH 9.0 Tris/ EDTA buffer. After washing with distilled water, endogenous peroxidase activity was inhibited (3% H 2 O 2 for 5 min), followed by a TBS/Tween wash. The primary antibody (anti-Pgp, clone EPR10363; catalog no. 170903; Abcam) was incubated at 2.5 g/ml for 30 min at room temperature, followed by TBS/ Tween washes. An IgG isotype negative control from a nonimmunized rabbit (Dako; catalog no. X0936) was matched to the concentration of the primary Pgp antibody. The complex was detected with Envision/anti-rabbit antibody (30 min at room temperature; Dako), washed with buffer, and incubated (10 min) with 3Ј,3-diaminobenzidene. Sections were counterstained in Harris hematoxylin and dehydrated in graded ethanols before being cleared in xylene and mounted. The images were taken using an Olympus BX53 microscope/DP72-3CCD camera (Olympus, Tokyo, Japan).
Statistics-The data were compared using two-tailed Student's t test. The results were expressed as means Ϯ S.D. (number of experiments) or means Ϯ S.E. (number of mice) and considered to be statistically significant when p Ͻ 0.05.
Pgp expression was marked in KBV1 and 2008/P200 cells, while being negligible in KB31 and 2008 cells (see insets in In marked contrast, the thiosemicarbazones, Dp44mT and Bp4eT, were 31-and 6.7-fold more cytotoxic (p Ͻ 0.001) in drug-resistant KBV1 cells (ϩPgp) compared with KB31 cells (ϪPgp), respectively (Fig. 1C). In an opposite manner to DOX and VBL (Fig. 1B), Pgp inhibition by the well characterized Pgp inhibitors, Val (18) or Ela (18), led to a significant (p Ͻ 0.001) decrease in cytotoxicity (increased IC 50 ) of Dp44mT and Bp4eT in KBV1 cells (ϩPgp; Fig. 1C). Notably, no significant (p Ͼ 0.05) change in DOX, VBL, or thiosemicarbazone cytotoxicity was observed in the presence of Pgp inhibitors in KB31 cells (ϪPgp; Fig. 1, B and C). These observations suggested the cytotoxicity mediated by the thiosemicarbazones was Pgp-dependent. The effect of all these agents was not cell line-specific, because similar Pgp-dependent results were also obtained using 2008/P200A cells (ϩPgp) and 2008 cells (ϪPgp; Fig. 1, D and E).
The importance of Pgp in sensitizing cells to Dp44mT were then further substantiated by Pgp silencing (see inset in Fig. 1F) that significantly (p Ͻ 0.001) sensitized KBV1 cells to DOX, relative to the negative control siRNA. In contrast, KBV1 cells with Pgp silencing became significantly (p Ͻ 0.001) more resistant to Dp44mT relative to negative control siRNA-treated cells (Fig. 1F).
Similar to the observations in highly Pgp-expressing KBV1 cells and 2008/P200A cells, the DMS-53 and HCT-15 cell lines that endogenously express low to moderate Pgp levels ( Fig. 2A) were also found to show resistance to DOX and VBL via Pgp (Fig. 2B). This was demonstrated using the Pgp inhibitors Val and Ela which decreased resistance to DOX or VBL, resulting in significantly (p Ͻ 0.001-0.01) lower IC 50 values (Fig. 2B).
Again, as observed in KBV1 cells that highly express Pgp (Fig.  1, B and C), Dp44mT and Bp4eT were significantly (p Ͻ 0.001-0.05) more cytotoxic (decreased IC 50 ) in the absence of Pgp inhibitors (Val and Ela) in both DMS-53 and HCT-15 cells (Fig.  2C). These observations suggested the enhanced cytotoxicity mediated by these thiosemicarbazones not only is relevant in MDR cells such as KBV1 and 2008/P200 (Fig. 1, C and E) but also is important in cells expressing endogenous Pgp. Collectively, these results (Figs. 1 and 2) highlight the importance of Pgp in the potentiated toxicity exerted by Dp44mT to overcome MDR in two cell types with Pgp expression induced by prior exposure to cytotoxic drugs (i.e. DOX or VBL) and two cell lines with low to moderate endogenous Pgp expression.
Pgp Expression, but Not the Expression of the ABC Transporters, MRP1 and ABCG2, Affects Dp44mT Cytotoxicity-Studies were also performed to assess whether Dp44mT could potentially be transported by other well characterized ABC transporters, namely MRP1 and ABCG2 (25)(26)(27). To assess this transport activity relative to Pgp, a number of cell types were examined in cellular cytotoxicity studies measuring the IC 50 of Dp44mT over a 24-h incubation. First, in these studies, KB31 cells were employed as a relative negative control because they do not express any of these transporters at detectable levels ( Fig. 2D). On the other hand, as positive controls for transporter expression and activity, the following cell lines were implemented, namely: 1) KBV1 and HCT-15 cells, because they are known to express functionally active Pgp (17,28); 2) MCF-7 cells, because they express functionally active MRP1 (29) and ABCG2 (30); and 3) MDA-MB-231 cells, because they express functionally active ABCG2 (30) (Fig. 2D). To assess the role of the transporters, each cell type was incubated with either the well characterized Pgp inhibitor Ela (0.2 M) (18), the MRP1 inhibitor MK-571 (20 M) (31), or the ABCG2 inhibitor KO-143 (2 M) (32). The concentrations of inhibitors utilized were characterized in control experiments to be optimal in terms of inhibiting transporter activity without inducing cytotoxicity.
Because of the lack of detectable expression of Pgp, MRP1, and ABCG2 in KB31 cells (Fig. 2D), the relative inhibitors had no significant (p Ͼ 0.05) effect on the cytotoxicity of Dp44mT versus that observed with control medium without inhibitors in this cell line (Fig. 2E). On the other hand, as shown above in Figs. 1C and 2C, the Pgp inhibitor Ela markedly and significantly (p Ͻ 0.001) increased resistance to Dp44mT (shown by increased IC 50 value) only in KBV1 and HCT-15 cells where Pgp is expressed (Fig. 2, D and E). In contrast to Ela, the cytotoxicity of Dp44mT was not significantly (p Ͼ 0.05) altered in the presence of the MRP1 inhibitor MK-571 or the ABCG2 inhibitor KO-143, in any of the cell lines (Fig. 2, D and E). Collectively, these observations suggest Dp44mT only utilizes Pgp to overcome multidrug resistance (Fig. 2, D and E).
Dp44mT and Bp4eT Are Pgp Substrates-Because the expression and function of Pgp is a prerequisite for the increased cytotoxicity of Dp44mT and Bp4eT (Figs. 1, C, E, and F, and 2), a Pgp-ATPase assay of a purified membrane preparation containing high levels of Pgp (33) was initially used to assess whether Dp44mT and Bp4eT interact with Pgp. The positive control verapamil (33) significantly (p Ͻ 0.001) increased ATPase activity ϳ4-fold, whereas the Pgp inhibitors, orthovanadate (33), Val, and Ela (18), significantly (p Ͻ 0.001) decreased the ATPase activity Ն5-fold relative to the basal level To further assess the interaction of Dp44mT with Pgp, the radiolabeled ligand, [ 14 C]Dp44mT, was implemented to determine its uptake and efflux by KBV1 (ϩPgp) and KB31 (ϪPgp) cells (Fig. 3, B-E Fig. 3, B and C). This observation was consistent with    (Fig. 3, B and C). Consistent with this finding, significantly (p Ͻ 0.001) higher efflux of [ 14 C]Dp44mT was observed in KBV1 (ϩPgp) cells relative to KB31 cells (ϪPgp; Fig. 3, D and E), whereas the Pgp inhibitors Val and Ela significantly (p Ͻ 0.01-0.001) suppressed [ 14 C]Dp44mT efflux only in Pgp-expressing KBV1 cells (Fig. 3E). Collectively, these studies indicated Dp44mT acted as a Pgp substrate. Dp44mT Competes with a Well Characterized Pgp Substrate to Inhibit Pgp Transport-Because some Pgp substrates are competitive Pgp inhibitors (33,34), we assessed whether Dp44mT competes with the fluorescent Pgp substrate Rh123 (35) for efflux from cells. As the Dp44mT concentration increased to 0.1 M or above, Rh123 retention was significantly (p Ͻ 0.001) increased in KBV1 (ϩPgp) cells but not KB31 (ϪPgp) cells (Fig. 3F). Similarly, the Pgp inhibitors Val and Ela also significantly (p Ͻ 0.001) increased Rh123 retention in KBV1 cells, but not KB31 cells (Fig. 3F). The higher potency of Ela compared with Val, in terms of inhibiting Pgp, is consistent with their affinity for Pgp (18). Taken together, results from Fig.  3 (A-F) demonstrate that Dp44mT acts as a Pgp substrate and competes for transport with other known substrates.
Pgp Sequesters Dp44mT into Lysosomes to Potentiate Its Damage-Dp44mT targets lysosomes, and after binding intralysosomal Cu (to form the Cu[Dp44mT] complex), it generates ROS to cause LMP (13). Moreover, we showed that lysosomal membrane-bound Pgp transports Pgp substrates, such as DOX (Fig. 1A), into this organelle (17). Hence, the potentiating effect of facilitating Dp44mT transport into lysosomes via Pgp was crucial to investigate.
First, in these studies, fluorescence microscopy was used to demonstrate that Pgp (green) was present not only on the cell surface, but it also co-localized (Mander's overlap coefficient (36) of 0.996) with intracellular LAMP2-stained lysosomes (red), in KBV1 cells (ϩPgp; Fig. 4A). Importantly, similar co-localization between Pgp and LAMP2 was also observed with another anti-Pgp antibody (1:100, catalog no. P7965; Sigma) and Texas Red-conjugated secondary antibody (1:1000, catalog no. sc-2781; Santa Cruz Biotechnology).
Notably, the well characterized Pgp inhibitors Val and Ela do not induce lysosomotropism in KB31 (ϪPgp) and KBV1 (ϩPgp) cells, because no change in size of the cell (forward scatter) or granularity (side scatter) can be observed with these inhibitors in our flow cytometry studies (17). Hence, under the conditions implemented in this investigation, Val and Ela inhibited Pgp without exhibiting lysosomotropic properties that could have reduced accumulation of Dp44mT in lysosomes. Therefore, these data indicate that inhibiting Pgp transport activity in KBV1 cells blocked Cu[Dp44mT]-mediated uptake and subsequent lysosomal damage, although it had no effect on non-Pgp-expressing KB31 cells (Fig. 4B).
Flow cytometric studies using another classical lysosomal marker, acridine orange (13), also demonstrated that LMP induced by Cu[Dp44mT] was Pgp-dependent because the Pgp inhibitor Val significantly (p Ͻ 0.001) prevented LMP in Pgpexpressing KBV1 cells, but not KB31 (ϪPgp) cells (Fig. 4C). Notably, Fe[Dp44mT] 2 did not affect LMP after a 30-min incubation because of its lower redox activity (13). The Pgp-potentiated LMP was also demonstrated after 24 h with the ligand alone (i.e. Dp44mT) because the Pgp inhibitor Val significantly (p Ͻ 0.001) prevented LMP in KBV1 (ϩPgp) cells, but not KB31 (ϪPgp) cells (Fig. 4D). The delayed LMP by Dp44mT compared with Cu[Dp44mT] could be explained by the facts that Cu[Dp44mT] was a better substrate than Dp44mT alone (Fig.  3A), and in contrast to the preformed Cu[Dp44mT] complex, Dp44mT would need to bind copper released from coppercontaining proteins catabolized in the lysosome before redox activity could be initiated.
The possibility that Pgp inhibitors directly prevent Cu-[Dp44mT]-induced lysosomal damage by inhibiting ROS generation was excluded through control experiments. Indeed, Val or Ela did not prevent the redox activity of Cu[Dp44mT] under lysosomal-like conditions in vitro (Fig. 4E). In contrast, the Cu chelator, tetrathiomolybdate (Fig. 4E), totally prevented Cu[Dp44mT] ROS generation (Fig. 4E). These results indicate that Val and Ela do not directly prevent ROS generation and that their ability to prevent Dp44mT or Cu[Dp44mT]-mediated LMP was due to their ability to inhibit Pgp.
Lysosomal Trapping of Dp44mT Induces LMP-To investigate the importance of lysosomal trapping of Cu[Dp44mT] in the induction of LMP, the intracellular distribution of the lysosomal enzyme, cathepsin D (13,42), and its association with lysosomal-associated membrane protein marker, LAMP2 (42), APRIL 10, 2015 • VOLUME 290 • NUMBER 15 was studied with fluorescence microscopy (Fig. 5, A-G). Control KBV1 (ϩPgp) cells stained with LAMP2 (red) and cathepsin D (green) became co-localized upon the merge, leading to yellow punctate fluorescence consistent with intact lysosomes (Fig. 5A). After a 30-min incubation with Cu[Dp44mT], the green punctate fluorescence of cathepsin D was markedly reduced in Pgp-expressing cells, because cathepsin D was released from this organelle (Fig. 5B). In contrast, LAMP2 fluorescence was unaffected, probably because LAMP2 is an integral protein of lysosomal membranes (43) (Fig. 5G), consistent with Cu[Dp44mT]-induced LMP (Fig. 4, B and C). However, cells incubated with Cu- [Dp44mT] in the presence of the lysosomotropic weak base, CLQ, demonstrated that the LAMP2 and cathepsin D markers were co-localized, leading to a yellow punctate pattern consistent with undamaged lysosomes that retained cathepsin D (Fig.  5C). Quantitation of cathepsin D intensity demonstrated that the addition of CLQ to Cu[Dp44mT] treatment completely prevented the loss in lysosomal fluorescence observed with Cu[Dp44mT] alone in KBV1 cells (Fig. 5G).

Dp44mT "Hijacks" Lysosomal Pgp to Overcome MDR
In contrast, no alteration in the punctate fluorescence of LAMP2 and cathepsin D was observed after incubation of KB31 (ϪPgp) cells with Cu[Dp44mT] or Cu[Dp44mT]/CLQ treatment (Fig. 5, D-G). Collectively, these results in Figs. 4 and 5 demonstrate that Dp44mT and Cu[Dp44mT] require Pgp for lysosomal accumulation and also an acidic lysosomal pH for trapping to enhance cytotoxicity in Pgp-expressing cells via LMP.
Dp44mT Markedly Targets Pgp-expressing Tumors in Vivo-To examine whether Dp44mT (administered intravenously (tail vein); 5 days/week) targets human Pgp-expressing tumors more effectively than their non-Pgp-expressing counterparts, we utilized BALB/c nu/nu mice subcutaneously injected with KBV1 (ϩPgp) or KB31 (ϪPgp) cells on the left and right flanks of the same mouse, respectively (Fig. 6A). Treatment began after the tumor reached 120 mm 3 (9). After 9 days of treatment, in mice receiving the vehicle control, both the KBV1 (ϩPgp) and KB31 (ϪPgp) tumors grew linearly as a function of time (r ϭ 0.93 and 0.92, respectively), reaching 632 Ϯ 63% and 894 Ϯ 104% of the initial tumor volume (day 0), respectively (Fig. 6, B and C). Because tumor size in the vehicle control group approached the maximum limit (1000 mm 3 ) prescribed by the local animal ethics committee, it was not possible to continue the study past day 9.
Notably, Dp44mT was significantly (p Ͻ 0.001) more effective at inhibiting the net growth of KBV1 (ϩPgp) tumor xenografts relative to KB31 (ϪPgp) tumor xenografts when comparisons were made after 4, 7, and 9 days of growth (Fig. 6, B and C). Dp44mT (0.1 or 0.2 mg/kg) reduced tumor growth by 9 -10fold in KBV1 (ϩPgp) xenografts relative to the vehicle control (Fig. 6B), whereas a 1.7-1.8-fold reduction in tumor growth was observed in KB31 (ϪPgp) tumors compared with the vehicle control group (Fig. 6C). Throughout the treatment time course with Dp44mT or the vehicle control, there was no significant (p Ͼ 0.05) alteration in the body weight of mice bearing KBV1 or KB31 tumors (Fig. 6D), as shown previously (9). The repre-sentative tumor size on day 9 (Fig. 6E) clearly demonstrated that Dp44mT (0.1 or 0.2 mg/kg/day) was far more efficient in inhibiting tumor growth in Pgp-expressing KBV1 tumors than in non-Pgp-expressing KB31 tumors.
Immunohistochemistry demonstrated that Pgp expression in KBV1 tumors was maintained in mice treated for 9 days, whereas no Pgp was detected in KB31 tumors (Fig. 6F). Importantly, Pgp expression in KBV1 (ϩPgp) tumors was markedly reduced upon treatment with Dp44mT at 0.1 or 0.2 mg/kg, suggesting killing of Pgp-expressing tumor cells. As a negative control, we also used an IgG isotype control antibody (IgG; Fig.  6G) from a nonimmunized rabbit to match the concentration of the primary Pgp antibody. This control demonstrated that the staining observed in KBV1 cells was specific for Pgp. In summary, this in vivo study highlights the potent activity of Dp44mT against Pgp-expressing tumors.

DISCUSSION
MDR is a major obstacle for cancer treatment (1,2), and there is a critical need for the development of alternative treatments and agents that can overcome resistance. Moreover, attempts over many decades to develop agents that overcome resistance have not been successful in the clinics (1,2,44).
Herein, we demonstrate for the first time how the new anticancer agent, Dp44mT, overcomes multidrug resistance through utilization of the lysosomal drug transporter, Pgp, resulting in enhanced cytotoxicity to the resistant tumor cell. In this case, Dp44mT is transported into the lysosome by Pgp where it induces increased anti-tumor activity (via LMP) and overcomes drug resistance. Notably, this property is not found for standard Pgp substrates and depends on the redox activity of the Dp44mT-Cu complex. This effect is in marked contrast and in fact opposite to what occurs with standard chemotherapeutics that are Pgp substrates such as DOX, where Pgp acts to prevent cytotoxicity and results in drug resistance by totally different mechanisms, namely, drug efflux out of the cell and drug storage in the lysosome where the organelle acts as a safe house and prevents the cytotoxic action of the agent (17).
Through the fundamental understanding of the molecular mechanism by which Dp44mT overcomes drug resistance via lysosomal Pgp, this study offers novel insights into the molecular and cellular interactions of these agents that can be specifically utilized to design new drugs that can overcome drug resistance. Hence, agents that target lysosomes offer an exciting new therapeutic strategy to combat MDR via utilization of lysosomal Pgp transport activity.  In the current study, it is notable that both Pgp inhibitors and siRNA-mediated Pgp silencing were utilized to demonstrate that Pgp was crucial for the potentiated cytotoxicity of Dp44mT toward MDR cells. Moreover, the interaction of Dp44mT with Pgp was also confirmed using three methods: 1) measurement of Pgp-ATPase activity of isolated Pgp-protein; 2) [ 14 C]Dp44mT uptake and efflux studies in the presence and absence of Pgp inhibitors; and 3) competition experiments where increasing Dp44mT concentrations were able to compete for Pgp transport, leading to retention of the Pgp substrate, Rh123. Collectively, these studies demonstrate that Dp44mT is a Pgp substrate (Fig. 3).
As discussed above, despite being a Pgp substrate, Dp44mT behaved oppositely to the classical Pgp substrates, DOX and VBL (3). Indeed, these latter agents were more potent in drugsensitive, non-Pgp-expressing cells (Fig. 1, B and D), whereas Dp44mT was more effective in multidrug-resistant Pgp-expressing cells and cells endogenously expressing Pgp (Figs. 1, C and E, and 2C). Furthermore, Pgp inhibitors increased sensitivity to DOX and VBL in Pgp-expressing cells (Figs. 1, B and D, and 2B), although these inhibitors significantly decreased the effect of Dp44mT (Figs. 1, C and E, and 2C). The difference in cytotoxicity between Dp44mT and these classical Pgp substrates was dependent on how these agents interact with lysosomes once transported into this organelle by Pgp. In the case of Dp44mT, it hijacks lysosomal Pgp to increase lysosomal targeting, resulting in LMP (Fig. 4, B-D). In contrast, the transport of DOX by Pgp into the lysosome leads to a protective effect, with the lysosome acting to prevent cytotoxicity and resulting in resistance (17). The difference between these Pgp substrates depends upon the interaction with lysosomal copper, with Dp44mT forming a highly redox active copper complex that leads to marked LMP (Figs. 4, B, C, and E; 5; and 7). Hence, incorporating a copper-binding pharmacophore into agents that accumulate in lysosomes offers a new strategy that can be implemented in designing anti-cancer agents for overcoming Pgp-mediated resistance.
Considering our previous studies demonstrating that Dp44mT can be charged and trapped within the acidic lysosome (13), it was hypothesized that raising lysosomal pH with lysosomotropic weak bases (45,46) could also increase the proportion of the neutral species released from lysosomes without inducing LMP (41,47). Indeed, incubation of [ 14 C]Dp44mT with three different lysosomotropic weak bases decreased [ 14 C]Dp44mT uptake (Fig. 4F), decreased cytotoxicity (Fig. 4, G  and H), and prevented LMP (Fig. 5). These observations were only found in Pgp-expressing cells and not in those without Pgp. This Pgp-mediated transport of Dp44mT increases lysosomal uptake, upon which this agent becomes charged because of the low pH (pH 5) (13), preventing it from escaping lyso-somes and enabling it to induce cytotoxicity caused by LMP (Fig. 7). Hence, although Dp44mT is effluxed out of the cell by plasma membrane-Pgp (Figs. 3E and 7), the simultaneous FIGURE 7. Schematic diagram of the mechanism of Pgp-mediated cytotoxicity by Dp44mT.
Step 1, Pgp-localized on the plasma membrane facilitates efflux of substrates such as Dp44mT out of the cell.
Step 2, as part of endocytosis, Pgp on the plasma membrane "buds" inwards to form early endosomes (60). As a consequence of endocytosis, the topology of Pgp is inverted, as demonstrated for other membrane proteins (61,62), resulting in substrate transport into the endosome.Thus,Pgpremainsfunctional,becauseitscatalyticactivesitesandATP-binding domains still remain exposed in the cytosol (61,62).
Step 3, the inversion of Pgp orientation leads to Dp44mT transport into the vesicle lumen. As the endosome matures into a lysosome, it becomes increasingly acidified.
Step 4, active Pgp-mediated transport of Dp44mT increases lysosomal uptake where it becomes positively charged and trapped because of the acidic lysosomal pH. The Dp44mT then binds copper in lysosomes (which recycle this nutrient during autophagy) to form a potent, redox active complex that generates ROS and causes LMP and apoptosis (15).

Dp44mT "Hijacks" Lysosomal Pgp to Overcome MDR
enhanced Pgp-mediated, lysosomal accumulation of Dp44mT is critical for mediating LMP and cytotoxicity.
The mechanism of action of Dp44mT involves ROS generation and potent lysosomal damage after binding Cu in Pgpexpressing cells (Figs. 4, B, C, and E, and 5). This finding is significant, because tumor cells relative to their normal counterparts have enhanced metabolism of metals including copper (48). Numerous studies have indicated higher copper levels in cancer patient serum (49) and the tumor relative to normal tissue (50 -52). It has also been suggested that because of recycling of cellular constituents via autophagy, lysosomes in tumor cells contain greater quantities of metals including copper (53, 54). Hence, the copper-related mechanism described herein involving lysosomal targeting of thiosemicarbazones via Pgp could explain their well known selectivity against cancer cells versus normal cells (8 -11, 55, 56). Therefore, ROS-generating agents such as Dp44mT (13,14) can hijack lysosomal Pgp to selectively overcome drug resistance by inducing more lysosomal damage in Pgp-expressing cells relative to their non-Pgpexpressing counterparts (Fig. 7).
Importantly, Dp44mT was found to selectively target chemotherapeutic-resistant human Pgp-expressing tumors over non-Pgp-expressing human tumors in vivo (Fig. 6, B, C, and E). Compared with vehicle-treated control tumors, the efficacy of Dp44mT at reducing tumor growth was Ͼ5-fold more in KBV1 (ϩPgp) relative to KB31 (ϪPgp) xenografts. This result is significant, because attempts to reverse resistance by using MDR modulators have not been successful in clinical trials (4,5). Hence, the ability of Dp44mT to overcome MDR offers a significant advantage over other chemotherapeutics and the unsuccessful attempts to reverse resistance with MDR modulators (4,5).
In addition to overcoming MDR, it is well described that DpT thiosemicarbazones also potently inhibit tumor growth and metastasis in vivo (9,11,12,57). This latter property is particularly significant, because metastasis is responsible for 90% of cancer deaths (58). Furthermore, the DpT class of thiosemicarbazones is well tolerated in vivo (12). These properties, combined with the ability of DpT thiosemicarbazones to overcome MDR, accredits them with unique pharmacological properties for effective cancer therapy.
In conclusion, for the first time, our studies demonstrate a key role for lysosomal Pgp in overcoming drug resistance and offer a mechanism for the potentiated cytotoxicity in MDR cells (9). Moreover, Dp44mT selectively targets chemotherapeutic resistant human Pgp-expressing tumors over non-Pgp-expressing tumors in vivo. The sensitizing action of Pgp on Dp44mT-mediated cytotoxicity was dependent on three characteristics of this agent: 1) it must be a Pgp substrate; 2) it has to become charged at acidic pH to enable accumulation in lysosomes; and 3) the agent must cause marked redox stress in the acidic lysosome leading to cytotoxic ROS that induces LMP and cell death. This latter property is not found for other classical Pgp substrates (e.g. DOX) that are sequestered in the lysosome, which leads to resistance to the agent (17). Hence, this study offers novel insights into the molecular and cellular interactions of these agents that can be specifically utilized to design new drugs, such as novel thiosemicarbazones, that can overcome drug resistance (9).