Depletion of major vault protein increases doxorubicin sensitivity and nuclear accumulation and disrupts its sequestration in lysosomes Free

Vaults are the largest cellular ribonuclear protein complexes and have been associated with the multidrug resistance phenotype (1–3). However, this association is based mainly on correlative data, and the exact mechanism of action, if any, of vaults in this process is unclear. Therefore, functional clues from the intricate structure of the vault particle have been sought. The complex consists of three protein components, namely major vault protein (MVP), v-PARP, and TEP1; the latter is present in the end structures, or the “caps,” where it associates with vault-specific structural RNA (3–5). The major body of the riboprotein complex consists of 96 copies of the MVP organized in radial symmetry to form a barrel-shaped structure (4, 6). Because MVP is only found in vaults, knockdown of MVP in cells is regarded as equivalent to vault knockdown. The first evidence of a connection between drug resistance and the vault particle was provided when Scheffer et al. (7) showed that MVP is identical to the lung resistance protein, which previously had been associated with multidrug resistance (8) in cancer cells of varied histologic origin. Subsequent reports have shown a correlation between overexpression of MVP and poor initial response to chemotherapy or intrinsic de novo drug resistance in acute myelogenous leukemia and other tumors (9–14). For example, low MVP expression levels were seen in metastatic testicular cancer (15), neuroblastoma, and childhood acute myelogenous leukemia, tumors that are often curable with conventional chemotherapy (16). In contrast, MVP expression was high in metastatic colon, renal, and pancreatic carcinomas, which are incurable with current chemotherapy (16). Furthermore, the expression of MVP was also predictive of drug response in bladder carcinomas (17), the tumor used in the present study.

In studies where a colon carcinoma cell line was chemically induced to differentiate, Kitazono et al. (18, 19) have shown a direct correlation between MVP levels and progressive resistance to doxorubicin. In this model, MVP-dependent resistance did not increase cellular efflux of doxorubicin but was associated with a subcellular redistribution of the drug away from the nucleus. Such subcellular redistribution, or sequestration, of drugs has recently emerged as an important drug resistance mechanism (20, 21). The sequestration of membrane-permeable drugs, such as doxorubicin, in low pH compartments is thought to depend on protonation (22). Protonation renders the compound membrane impermeable and its sequestration in low pH subcellular compartments, generally labeled or thought to be lysosomes, ultimately leads to its removal from the cell (23). Blockage of this process through vacuolar proton pump inhibitors reverses acquired drug resistance (24), and the drugs remain intracellular and there exert their biological activity. Recently, the ATP-binding cassette transporter family member proteins that are strongly connected to the drug resistance phenotype, such as P-glycoprotein, MRP1, and BCRP, have been localized to intracellular membranes. Such findings further highlight the relevance and importance of the sequestration mechanism in drug resistance (25, 26).

Intrinsic drug resistance is the main reason for therapeutic failure in advanced cancer treatment. Although several studies have shown a correlation between drug resistance and MVP expression, there is also a considerable amount of studies failing to show this correlation (2). It is likely that vault-dependent drug resistance is dependent on other variables, and a greater understanding of the mechanisms behind the correlative data is desirable. Here, we present a study aimed at understanding the underlying functions of vaults in de novo drug resistance by evaluating the role of MVP in doxorubicin sensitivity in a human bladder cancer cell line.

UMUC-3 is a human urothelial bladder cancer cell line obtained from the American Type Culture Collection and grown in MEM supplemented with 1 mmol/L sodium pyruvate and 10% FCS. MVP and MVP-green fluorescent protein (GFP) cDNA expression constructs were described elsewhere (27). Transfections were done as described previously (28). The MVP sequence 5′-ATCATTCGCACTGCTGTCTT-3′ was targeted by small interfering RNA (siRNA). Control transfections were carried out using siRNA-targeting luciferase. MVP-GFP that has previously been rigorously evaluated (29) was mutated in four silent positions to destroy the siRNA target site using the QuikChange Site-Directed Mutagenesis kit from Stratagene. The following primers were used: MVP3MUTF, 5′-CTCAGCCCGCATCATACGGACAGCAGTCTTTGGCTTTGAGACC-3′ and MVP3MUTR, 5′-GGTCTCAAAGCCAAAGACTGCTGTCCGTATGATGCGGGCTGAG-3′. MVP-GFP and MVP-GFPmut expression vectors were linearized with AseI before transfection with FuGENE according to the manufacturer's instructions (Roche). Cells were treated with 600 μg/mL G418 for 2 weeks before fluorescence-activated cell sorting for GFP-positive cells (Becton Dickinson FACSVantage SE Turbo).

For siRNA MVP experiments, cells prepared in parallel wells were routinely examined by Western blot to verify protein knockdown. Cells were harvested in lysis buffer [1% Triton X-100, 100 mmol/L NaCl, 1 mmol/L CaCl₂, 1 mmol/L MgCl₂, 25 mmol/L Tris-HCl (pH 7.4)] and lysed for 20 min before undissolved residues were spun down. Lysates were normalized by bicinchoninic acid before SDS-PAGE analysis. Equal loading was further controlled by anti-α-tubulin (OncoGene). MVP antibody (PharMingen) and other primary antibodies were used 1:1,000 in TBS containing 5% fat-free milk and 0.1% Tween 20 (Sigma). Protein expression was measured by digital fluorescence imaging using AlphaEaseFC and an Alpha Innotech FluorChem 8800 digital imager.

Two days after transfection with siRNA, cells were challenged with different concentrations of doxorubicin (Sigma) in complete medium. Five days after addition of doxorubicin, cells were washed once in PBS and lysed in 200 μL CyQuant lysis buffer (Molecular Probes). The lysates were homogenized and prepared for DNA measurement by CyQuant assay according to the manufacturer's instructions. Wells were analyzed for DNA content as a surrogate measure of cell numbers with a Molecular Dynamics BioLumin 960 plate reader.

Twenty-four hours after siRNA transfection, cells were incubated for 20 min with 10 μmol/L doxorubicin. Cells were then washed, fresh medium was added, and cells were left to recover for 1 or 24 h. At the end of incubation, cells were harvested with trypsin/EDTA, washed once in complete medium, and analyzed with a Becton Dickinson FACSCalibur at the indicated times. Doxorubicin was excited at 488 nm, and a 670-nm long pass emission filter was used. Nuclear concentration was estimated through measuring the fluorescent signal of doxorubicin on confocal images (acquired as described below) by ImageJ analysis.

Cells were grown on glass coverslips, washed once in PBS, and fixed for 10 min in 3.7% paraformaldehyde, blocked 20 min in 200 mmol/L glycine, permeabilized in 0.1% Triton X-100 for 10 min, and blocked in PBG (0.1% cold water fish skin gelatin, 0.5% bovine serum albumin in PBS). α-Tubulin was purchased from OncoGene. Primary antibodies were diluted in PBG, incubated for 2 h at room temperature, washed six times in PBG, and incubated with secondary antibody (goat anti-mouse Alexa Fluor 488 conjugated) 1:250 (Molecular Probes) for 30 min at 37°C. Cells were washed thrice in PBS and thrice in water before they were mounted in Elvanol [20% Mowiol in 2:1 PBS (pH 8.0)/glycerol] and analyzed as described below.

Cells were seeded on 24-well glass bottom plates (Greiner) and transfected with siRNA constructs. Two days after transfection, cells were incubated 20 min with 10 μmol/L doxorubicin, washed once, and incubated for an additional 24 h. A Zeiss Axiovert M135 microscope was used for image capture at 37°C under 5% CO₂ atmosphere. To visualize endocytic compartments, cells were incubated with 1 μmol/L LysoSensor DND-189 (Molecular Probes) for 1 min before analysis. Alternatively, the cells were loaded with dextran (10,000 molecular weight) labeled with Alexa Fluor 488 at a final concentration of 1 mg/mL and incubated for 20 min, which was followed by a 24-h chase period at normal cell culture conditions. Hoechst 33342 and dextran conjugates were purchased from Molecular Probes. For confocal experiments, cells were grown on 30-mm glass bottom plates, switched to DMEM without phenol red, and sealed for the duration of the microscopic analysis. Images were captured with a Zeiss 510 LSM.

F tests were used to compare doxorubicin fluorescent intensity in MVP knockdown cells to the intensity in luciferase-transfected controls. Regression methods were used to pool the data from two separate experiments, allowing for differences in absolute numbers across experiments. The regression analyses were done after transforming the intensities to the log scale to stabilize the variability between the MVP knockdown cells and the luciferase-transfected controls.

To study the relationship of baseline MVP expression levels to the intrinsic drug sensitivity of UMUC-3 human bladder cancer cells, we generated a siRNA construct against MVP. The construct was able to knock down protein expression to ≤15% compared with luciferase control-transfected or nontransfected cells (Fig. 1A). To determine if cells were sensitized to doxorubicin following MVP siRNA knockdown, cells were exposed to doxorubicin and analyzed for DNA content as a surrogate measure of cell number. Identical results were obtained using AlamarBlue, measuring cellular metabolic activity (data not shown). Parallel wells were harvested and analyzed by Western blotting for MVP expression to ascertain that cells examined for doxorubicin sensitivity had the expected suppression of MVP protein levels. Knockdown of MVP protein expression leads to a significantly increased sensitivity toward doxorubicin (Fig. 1B). This showed that MVP is necessary for intrinsic or de novo drug resistance in UMUC-3 bladder cancer cells.

The specificity of this result with regard to MVP knockdown was verified by introducing a mutated MVP-GFP construct. The original MVP-GFP construct was previously shown to incorporate normally into vault particles (27) and mirror endogenous protein behavior. MVP-GFP was mutated at four silent positions at the target site for the siRNA construct (MVPmut-GFP), and polyclonal pools of UMUC-3 cells expressing MVPmut-GFP were generated by transfection followed by fluorescence-activated cell sorting. These pools overexpress MVPmut-GFP up to 15-fold over endogenous MVP levels. Overexpression of MVP by MVPmut-GFP rescued the siRNA phenotype (Fig. 1C and D), indicating that the siRNA effect observed in Fig. 1B was due to lower expression levels of MVP and not due to nonspecific siRNA effects. Furthermore, despite greatly increased overall MVP expression levels, neither MVPmut-GFP (Fig. 1C) nor MVP-GFP (data not shown) transfection increased resistance of UMUC-3 cells to doxorubicin. Taken together, these results suggest that MVP is necessary but not sufficient for cellular resistance to doxorubicin in UMUC-3 cells.

To investigate whether the cellular concentration of doxorubicin is altered by MVP knockdown, we used the intrinsic fluorescence properties of doxorubicin and invoked the commonly held assumption that fluorescence level is proportional to concentration. As doxorubicin was not detectable in the nanomolar range previously used (Fig. 1B and C), a novel exposure protocol was developed. Briefly, UMUC-3 cells were transfected with either luciferase or MVP siRNA and allowed to recover for 24 h before they were exposed to 10 μmol/L doxorubicin for 20 min, after which the drug was washed off. Cell proliferation was assessed after a 48-h chase period. Although this represents a 250-fold increase in drug concentration previously used, the same drug response difference between MVP siRNA-treated cells and control was observed (Fig. 2A). To measure the cellular concentration of doxorubicin, cells were harvested at various time points after drug exposure and subsequently analyzed for fluorescence by fluorescence-activated cell sorting (Fig. 2B). These data indicate that there is no difference in initial doxorubicin uptake between MVP siRNA-transfected and control-transfected cells. More interesting, equal doxorubicin fluorescence levels was also observed after 24-h incubation in cells expressing MVP compared with MVP-depleted cells. These data suggest that MVP does not influence drug efflux but instead might function through other mechanisms, such as intracellular sequestration.

To investigate whether MVP affects intracellular doxorubicin sequestration, cells were loaded with 10 μmol/L doxorubicin, as described previously, and analyzed by microscopy. Directly after loading, doxorubicin showed a strong nuclear staining, with no visible staining in the cytoplasm. This staining persisted after overnight incubation and completely overlapped Hoechst 33342 staining, indicating a direct labeling of DNA (data not shown). Untreated and control siRNA-transfected cells showed a cluster of vesicular structures adjacent to the nucleus (Fig. 3A, arrows). Interestingly, this relocalization of the drug was not evident in MVP knockdown cells where no or few stained vesicular structures were observed (Fig. 3A, arrowheads). No apparent cellular autofluorescence was detected under these settings (data not shown). Cells were tested for MVP expression and viability after doxorubicin treatment (data not shown).

Because anthracyclines, such as doxorubicin, are sequestered in lysosomes (20, 24, 26, 30, 31) and this is in turn thought to lead to drug resistance, we hypothesized that doxorubicin is relocalized to lysosomes in the UMUC-3 bladder cancer cell line. To test this, we stained cells with a pH-dependent lysosomal marker before analysis. This staining overlapped with doxorubicin in the control-transfected cells (Fig. 3B). Surprisingly, cells in MVP-transfected cells showed a disrupted lysosomal labeling. MVP knockdown cells did not have a dense cluster of brightly labeled, highly motile vesicles adjacent to the nucleus but rather a few vesicular structures around the nucleus or with an apparently random distribution throughout the cytoplasm (Fig. 3B).

To verify that lysosomal sequestration in cells lacking MVP was perturbed, we measured the concentration of nuclear doxorubicin 24 h after drug exposure. Cells were imaged by live confocal microscopy, and the images were analyzed by ImageJ (32) for doxorubicin fluorescent intensity in the nucleus as a measure of concentration. The MVP knockdown cells showed a qualitatively higher doxorubicin concentration in the nucleus after overnight incubation (Fig. 3C). In two separate experiments, the nuclear doxorubicin concentration was measured to be 1.20-fold (shown in Fig. 3C) and 1.34-fold higher (data not shown), respectively, in cells lacking MVP compared with control-transfected cells. Pooling the data, adjusting for the differences in absolute numbers between the experiments, indicates that doxorubicin fluorescent intensity in MVP knockdown cells is significantly greater than luciferase-treated controls (P < 0.001).

Both doxorubicin and LysoSensor are depending on a low pH for their intracellular distribution. To evaluate the morphology of the lysosomal compartment in a pH-independent manner, cells were incubated at 37°C with 1 mg/mL fluorescent dextran conjugates for 20 min after which the cells were washed and placed in complete medium for 24 h (chase). Dextrans are taken up by the cells through pinocytosis and label the lysosomes on prolonged incubations (33). The lysosomal compartments labeled with dextrans and doxorubicin showed a remarkable overlap after 24-h incubation. This overlap was poor in MVP knockdown cells (Fig. 4A). To verify that this phenotype is indeed due to a lack of expressed MVP protein, cells stably expressing MVP-GFPmut were transfected with the siRNA construct. These cells express a MVP transgene that is not targeted by our siRNA and exhibit a clustered lysosomal compartment as labeled with fluorescent dextran (Fig. 4B). This also indicates that the lysosomal disruption is not doxorubicin induced in MVP knockdown cells. From these data, we conclude that doxorubicin is initially exclusively located to the nucleus but is sequestered to the lysosomal compartment in UMUC-3 cells on prolonged incubation. Redistribution of the drug from the nucleus to lysosomes is perturbed in MVP knockdown cells.

In Figs. 3 and 4, perinuclear clusters of lysosomes are readily observed in control-transfected cells. This positioning of lysosomal compartments is not evident in MVP siRNA-treated cells. However, MVP knockdown also renders the cells more susceptible to the effect of doxorubicin, which in turn could lead to the phenotype we are observing. To exclude this, and determine whether MVP knockdown can directly affect the organization of lysosomal compartments adjacent to the nucleus, we investigated the integrity of lysosomal positioning in MVP siRNA-transfected cells that were not treated with the drug. Lysosomes were labeled with fluorescent dextrans after a 24-h chase period, with LysoSensor immediately before imaging or with indirect immunofluorescence of the lysosomal-specific antigen Lamp-1 after paraformaldehyde fixation (34). We did not observe any markedly different uptake of fluorescent dextrans between control- and MVP-transfected cells (data not shown). As predicted, all the lysosomal markers showed a tight vesicular cluster at one side of the nucleus in the control-transfected cells. Most interesting, we observed a distinct staining pattern after MVP knockdown with all the used lysosomal markers. Some cells showed a concentration of vesicular staining around the nucleus, with weak or no obvious concentration to one side, whereas other cells showed no nuclear concentration of the labeled vesicular compartments at all (Fig. 5A).

Lysosomes are known to organize around the microtubule-organizing center (35) and are, like other endocytic compartments, dependent on microtubules for spatial localization (36). Because vaults are reported to localize to microtubules (37–39), it is conceivable that the effect we are observing on lysosomal positioning is due to microtubule organization. Therefore, we investigated the gross morphology of microtubules in MVP knockdown cells. This revealed a clear microtubular organization around the perinuclear localized microtubule-organizing center in cells transfected with MVP or control siRNA (Fig. 5B).

The disrupted lysosomal positioning observed in our study suggests a functional interplay between the vault complex and a vesicular compartment. The colocalization of MVP and secretory vesicles in neuron-like PC-12 cells (38) as well as the lysosomal marker CD63 in dendritic cells (40) has been published. Taken together, this motivated us to investigate the localization of MVP with respect to lysosomes and in response to doxorubicin. We used the same MVP-GFP constructs as described above and investigated MVP localization before and after doxorubicin treatment. As shown in Fig. 6A, no colocalization could be observed, and the MVP seemed in this nonconfocal, whole-cell image, to be less concentrated at the perinuclear position where most lysosomes are found. In a confocal picture, the perinuclear exclusion of MVP is less evident, and again, no colocalization between doxorubicin and MVP-GFP could be detected (Fig. 6B). No apparent overlap in distribution was observed at higher concentrations of doxorubicin or lower MVP-GFP expression levels (data not shown). The same GFP construct was evaluated before and after doxorubicin treatment. Both short-term and long-term doxorubicin treatments were evaluated. We were not able to detect any gross redistributions or local enrichments of our reporter construct in these studies (Fig. 6C).

We hence conclude that lysosomal positioning, and the lysosomes per se, seems to be disrupted in MVP-depleted cells and that this phenotype is not dependent on drug exposure. We did not observe any gross disruption of the microtubule network or colocalization of a GFP-MVP reporter construct with doxorubicin-positive lysosomes. We conclude from this that MVP is not part of the doxorubicin-loaded lysosomes but is nevertheless necessary for the lysosomal positioning in UMUC-3 cells.

In this study, we used siRNA and microscopic techniques to examine the cellular functions of the largest ribonuclear protein known to date, the vault complex. We show two consequences of MVP depletion in the bladder tumor cell line UMUC-3. First, cells become more sensitive to a commonly used antineoplastic drug in cancer therapy (i.e., doxorubicin), which is particularly relevant because this drug is used in bladder cancer patients (41). Second, we show a disruption of lysosome positioning in these cells. We hypothesize that these two phenotypes are connected and that the loss of MVP in UMUC-3 cells leads to a dysfunctional lysosomal compartment and that this in turn leads to an impaired lysosomal sequestration of the drug.

Although the sensitization of cell lines to antineoplastic drugs through MVP depletion has been shown (18, 19) and large amount of patient data point to a vital role for MVP in drug resistance (2, 3, 16), it is far from certain that this protein plays an essential role in chemosensitivity of human cancer. Indeed, it has been reported that MVP overexpression, or depletion, does not affect drug sensitivity nor intracellular sequestration (7, 42–44). There might be several explanations for these discrepancies that potentially could influence MVP-dependent drug response, including, but not limited to, cellular context and culture conditions. Ferguson et al. suggest from patient data that, in renal cell carcinomas, MVP expression might be a more important indicator of drug resistance than ATP-binding cassette transporter pumps. Surprisingly, in a panel of cell lines from the same tumor origin, the opposite is true, and the expression levels of ATP-binding cassette transporter pumps is most indicative of the sensitivity to the antineoplastic drugs (45). This not only indicates what has been reported at many separate occasions previously, that MVP protein expression is indeed important for drug response in vivo, but also indicates that this phenotype translates poorly to the in vitro situation. As we understand the vault complexes better, we might be able to control for essential factors in our systems that determine the dependence on MVP and we will be able to answer these questions. Further, as we understand the biological role of vaults, we will be able to measure end points more proximal to its molecular function than we currently do when we measure cellular viability after drug exposure. This will hopefully lead to that more viable in vitro systems are developed.

In an effort to further understand the MVP resistance mechanism, we investigated the cellular response to doxorubicin by microscopy. Directly after drug exposure, cells showed a nuclear staining, although an additional perinuclear vesicle staining was noted at later time points. We here provide data that suggest that the doxorubicin observed in the lysosomes was exported from the nucleus. Cells transfected with MVP duplexes show a higher nuclear doxorubicin concentration than those transfected with control siRNA. This argues that doxorubicin is exported from the nucleus and that MVP is necessary for this transport. Indeed, high levels of MVP expression have previously been shown to be essential for nuclear exclusion of drugs (18). Further, vaults are reported to partly reside within the nucleus (46) and are known to associate and be recruited to the nuclear envelope (39, 47). The association of vaults to the estrogen, progesterone, and glucocorticoid receptors and most recently PTEN, and specifically their nuclear localization domain, also argues for a nuclear-cytoplasmic transport function (48, 49). These data together with its hollow structure (6) have led to a theory that vaults play an active role in the cytoplasm-nucleus translocation process perhaps as a direct carrier of drugs or drug complexes (3). We therefore initially hypothesized that vaults were essential for the nuclear exclusion mechanisms of doxorubicin. However, instead of “empty,” doxorubicin-free, lysosomal compartments in MVP siRNA-treated cells, we did not see the same lysosomal distribution as in control-transfected cells. This leads us to question our hypothesis that MVP facilitates nuclear export and intracellular translocation of the drug. Instead, it does not seem to be the transport per se but rather the destination of the transport that seems to be dependent on vaults.

The endosomal network of vesicles, including lysosomes, depends largely on microtubules and microtubule-associated motor proteins for their transport (36). In UMUC-3 bladder tumor cells, there is a clear organization of the lysosomal compartments to one side of the nucleus and this level of higher organization is clearly disrupted by MVP siRNA. As mentioned earlier, there are several observations that would argue that the vault complex is associated with the microtubular network, and perhaps vesicles along it, such as colocalization of MVP and secretory vesicles in neuron-like PC-12 cells (38) and the lysosomal marker CD63 (Lamp-3) in dendritic cells (40). An interesting finding by Eichenmuller et al. (37) shows that vaults not only colocalize to microtubules but seem to directly interact with these structures via their structurally more complex cap regions. Two recent reports show the transport of MVP along microtubules (39, 50). It is, however, important to note that all these studies have in common that it is just a smaller fraction of vaults that is found in these associations, and indeed, most vaults seem to freely diffuse in the cytoplasm at a rate that is consistent with their large size (27). We fail to see any gross disruptions of the microtubular network on MVP depletion, which argues that MVP does not influence its overall integrity. We therefore hypothesize that the vault regulates the association of the vesicular cargo transported on this network, possibly through direct associations with the membrane-bound vesicle, the microtubules, or both.

In summary, we here report a novel connection between two cellular components well known to be pivotal to drug resistance (i.e., MVP and lysosomes). We hypothesize that the loss of MVP in UMUC-3 cells leads to a dysfunctional lysosomal compartment and that this in turn leads to an impaired lysosomal sequestration of the drug. Further investigation is needed to examine whether this is a direct effect of vault facilitating lysosomal transport along the microtubule or an indirect effect, such as vault-mediated regulation of cytoplasmic pH, lipid transport, or signaling through protein kinases.

We thank Jennifer Bryant for her expert technical assistance, Joanne Lannigan for assistance with the fluorescence-activated cell sorting, Erik Wiemer for the MVP and MVP-GFP constructs, and Mike Harding and all members of the Theodorescu lab for helpful discussions.