Identification and Functional Characterization of Nuclear Mortalin in Human Carcinogenesis*

Background: Mortalin/mtHsp70 is an essential stress chaperone frequently enriched in cancers. Results: Mortalin is present in the nucleus of cancer cells where it causes strong inactivation of tumor suppressor protein p53 and activation of telomerase and heterogeneous ribonucleoprotein K (hnRNP-K) proteins. Conclusion: Nuclear mortalin promotes carcinogenesis. Significance: This study is important for the development of mortalin-based anticancer treatments. The Hsp70 family protein mortalin is an essential chaperone that is frequently enriched in cancer cells and exists in various subcellular sites, including the mitochondrion, plasma membrane, endoplasmic reticulum, and cytosol. Although the molecular mechanisms underlying its multiple subcellular localizations are not yet clear, their functional significance has been revealed by several studies. In this study, we examined the nuclear fractions of human cells and found that the malignantly transformed cells have more mortalin than the normal cells. We then generated a mortalin mutant that lacked a mitochondrial targeting signal peptide. It was largely localized in the nucleus, and, hence, is called nuclear mortalin (mot-N). Functional characterization of mot-N revealed that it efficiently protects cancer cells against endogenous and exogenous oxidative stress. Furthermore, compared with the full-length mortalin overexpressing cancer cells, mot-N derivatives showed increased malignant properties, including higher proliferation rate, colony forming efficacy, motility, and tumor forming capacity both in in vitro and in vivo assays. We demonstrate that mot-N promotes carcinogenesis and cancer cell metastasis by inactivation of tumor suppressor protein p53 functions and by interaction and functional activation of telomerase and heterogeneous ribonucleoprotein K (hnRNP-K) proteins.

The Hsp70 family protein mortalin is an essential chaperone that is frequently enriched in cancer cells and exists in various subcellular sites, including the mitochondrion, plasma membrane, endoplasmic reticulum, and cytosol. Although the molecular mechanisms underlying its multiple subcellular localizations are not yet clear, their functional significance has been revealed by several studies. In this study, we examined the nuclear fractions of human cells and found that the malignantly transformed cells have more mortalin than the normal cells. We then generated a mortalin mutant that lacked a mitochondrial targeting signal peptide. It was largely localized in the nucleus, and, hence, is called nuclear mortalin (mot-N). Functional characterization of mot-N revealed that it efficiently protects cancer cells against endogenous and exogenous oxidative stress. Furthermore, compared with the full-length mortalin overexpressing cancer cells, mot-N derivatives showed increased malignant properties, including higher proliferation rate, colony forming efficacy, motility, and tumor forming capacity both in in vitro and in vivo assays. We demonstrate that mot-N promotes carcinogenesis and cancer cell metastasis by inactivation of tumor suppressor protein p53 functions and by interaction and functional activation of telomerase and heterogeneous ribonucleoprotein K (hnRNP-K) proteins.
Mortalin/mtHsp70/Grp75 is a member of the Hsp70 family of proteins (1) that plays an essential role in mitochondrial import, oxidative stress response, regulation of mitochondrial membrane potential, energy generation, intracellular transport, chaperonization, p53 functions, immune response, and protection against apoptosis and tumorigenesis (2)(3)(4)(5)(6)(7)(8). Mortalin and p53 colocalize and interact in transformed, but not in normal, human cells, resulting in the nuclear exclusion and transcriptional inactivation of p53 (7, 9 -11). Malignant transformation of NIH 3T3 cells, life span extension of MRC-5 cells, and attenuation of differentiation of HL-60 cells by overexpression of mortalin (mot-2) has been attributed, at least in part, to transcriptional inactivation of p53 (12)(13)(14). Furthermore, it has been shown to inhibit the p53-dependent suppression of centrosome duplication, leading to aneuploidy (an established hallmark of cancer cells) (6,15). Walker et al. (11) have reported that, although mortalin and p53 proteins formed complexes in the cytoplasm of leukemic clam hemocytes, normal hemocytes lacked this interaction. Treatment of leukemic clam hemocytes with MKT-077, a cationic mitochondriotropic dye that has been shown to target the mortalin-p53 interaction (16,17), resulted in the translocation and reactivation of p53 in clam cells (11). These data imply that mortalin-mediated inactivation of p53 is an evolutionarily conserved feature of cancer.
The expression profile of mortalin in normal and a variety of immortal and tumorigenic cell lines revealed its biphasic behavior: an initial elevation during immortalization (relative to a down-regulation during replicative senescence of human fibroblasts), followed by an up-regulation at a later stage that coincides with the acquisition of an invasive phenotype (18 -20). In proteomic analyses of cancer tissue arrays, mortalin has been identified as a prognostic marker of colorectal cancers (21,22). Associated with its phosphorylation, mortalin is known to show enhanced binding with FGF-1 and to be involved in the regulation of its mitogenic activity (23). It has been shown that, although cancers are frequently associated with a higher level of mortalin expression, Alzheimer and Parkinson pathologies involve the loss of mortalin and an imbalance in mitochondrial homeostasis (3, 24 -27). Overexpression of mortalin in experimental models of these diseases resulted in the improvement of disease phenotypes and protection against oxidative stress, a hallmark of these dementias (24 -26, 28, 29).
In line with the role of mortalin in carcinogenesis, anti-mortalin molecules, such as antisense, ribozyme, siRNA, p53-antagonist polypeptides, and chemicals that abrogated mortalin-p53 interaction and caused the relocation of p53 to the cell nucleus, resulted in growth arrest/apoptosis of cancer cells (2,4,6,30). Mortalin targeting adeno-oncolytic viruses caused tumor suppression in vivo by activation of p53, induction of apoptosis, and inhibition of angiogenesis (31). Furthermore, the up-regulation of mortalin correlated with an early recurrence of hepatocarcinoma in postoperative patients and liver cancer metastasis (32), suggesting that antimortalin molecules not only serve as anticancer agents but could also be potentially very important in the prevention of cancer recurrence. Together, these reports have necessitated investigations of the molecular mechanisms of the roles of mortalin in human tumorigenesis.
Mortalin has been reported to exist in multiple subcellular localizations, including the mitochondrion, endoplasmic reticulum, plasma membrane, cytosol, and centrosomes (6,15,27,33,34). Recently, Rozenberg et al. (22) have reported circulating mortalin in the serum of colorectal cancer patients, and its elevated levels (Ͼ60 ng/ml) were assigned as a risk factor for shorter survival . On the other hand, Shih et al. (35) reported that the nuclear translocation of mortalin is critically involved in neuronal cell differentiation . In light of these reports, we examined whether mortalin exists in the nucleus of human normal and transformed cells. We demonstrate that mortalin is present in the nucleus of cancer cells, where it promotes tumor aggressiveness by mechanisms involving inactivation of p53 functions and activation of telomerase, heterogeneous ribonucleoprotein K (hnRNP-K), 4 and MMPs.
Detection of Reactive Oxygen Species-Cells were cultured on glass coverslips placed in 12-well plates and stained for ROS by fluorescent staining using the Image-iT TM LIVE green ROS detection kit (Molecular Probes, Eugene, OR). Images were captured on a Zeiss Axiovert 200 M microscope and analyzed by AxioVision 4.6 software (Carl Zeiss Microimaging, Thornwood, NY).
Immunostaining-Cells were plated on coverslips placed in a 12-well culture plate. Control or etoposide-treated (10 M for 24 h) cells were incubated with primary antibodies (p53 and myc, Santa Cruz Biotechnology; Aurora A, Sigma; or mortalin (4)) at 4°C overnight, followed by extensive washing in TPBS (PBS with 0.2% Triton X-100) and incubation with the fluorochrome-conjugated secondary antibodies (Alexa Fluor 488conjugated goat anti-rabbit or anti-mouse or Alexa Fluor 594conjugated goat anti-rabbit or anti-mouse (Molecular Probes)) as described previously (4). Stained cells were examined on a Zeiss Axiovert 200 M microscope and analyzed by AxioVision 4.6 software (Carl Zeiss). To examine the presence of mortalin in the nucleus, images were acquired with a confocal laser scanning microscope (Zeiss LSM 700) confocal microscope. The files were transferred to a graphic work station and analyzed with Imaris software (Bitplane, Zurich, Switzerland).
Immunohistochemistry-Paraffin-embedded cancer and normal ovarian tissue slides were purchased from BioChain Institute, Inc. (Newark, CA). Slides were deparaffinized in xylene and incubated with anti-mortalin antibody. For detection of antibody binding, the REAL-HRP system (Dako REAL TM EnVision TM detection system, peroxidase/3,3Ј-diaminobenzidineϩ, and rabbit/mouse (Dako, Carpinteria, CA) were used. The color was developed using the liquid 3,3Ј-diaminobenzidineϩ substrate chromogen system (Dako) followed by counterstaining with Mayer's hematoxylin (Wako, Tokyo, Japan). Slides were examined using a fluorescence microscope (BZ-9000, Keyence, Tokyo, Japan), and images were obtained using BZ-II Analyzer software (Keyence).

Construction of Mortalin Mutants and Cell
Lines-Fulllength mortalin and its three deletion mutants (full-length mortalin (mot-F); mot-N, amino acid residues 42-679; mot-A, N-terminal amino acid residues 1-180; and mot-B, amino acid residues 180 -300) were generated by PCR and cloned into the HindIII site of the vector pCX4neo. Retroviruses were produced in Plat-E cells as described previously (18). The stably infected cells were maintained in 100 g/ml G418-supplemented medium. V5-tagged full-length and a deletion mutant lacking signal peptide were generated using the pcDNA3.1 plasmid vector. Transfections were performed using X-tremeGENE 9 DNA transfection reagent (Roche) following the instructions of the manufacturer.
Colony Forming Assay-A cell transformation assay was performed using a cell transformation detection kit (Chemicon). Colony forming efficacy is expressed as the mean number of colonies from three independent experiments. For colony forming assays, 1 ϫ 10 3 cells were seeded on 6-well plate. Cells were maintained until the appearance of colonies with regular change of medium. Colonies were fixed in methanol, stained with 0.5% crystal violet, photographed, and counted.
Chemotaxis and Cell Invasion Assay-Cells at 60 -70% confluency were trypsinized and resuspended in DMEM. 2 ϫ 10 4 cells were plated in 12 mm-pored Transwell inserts for the chemotaxis assay (Costar, MA). Fibronectin from human plasma (Sigma) was used as a chemoattractant. A cell invasion assay was performed using a QCM TM 24-well cell invasion assay kit (Millipore). The fluorescence of invaded cells was read using a microplate reader (Infinite 200 PRO, Tecan Group Ltd.) with a 480/520 nm filter set.
Nude Mice Tumor Assay-BALB/c nude mice (4 weeks old, female; Nihon Clea, Japan) were given subcutaneous injections of cells (1 ϫ 10 6 suspended in 0.2 ml of growth medium). Tumor formation was monitored for a month. The assay was regarded as positive if tumors appeared and grew progressively. For the metastasis assay, cells were injected intravenously into the lateral tail vein of Balb/c nude mice. Two weeks after injection, lungs were harvested and screened for tumor nodules. Five mice were used for each group, and the experiment was repeated twice.
p53-dependent Reporter Assays-Control, mot-F, and mot-N cells (1 ϫ 10 5 /well) were plated in a 6-well plate and transfected with 1 g of pGL13-luc plasmid vector containing a firefly luciferase gene and p53-responsive elements. The pRL-TK (Promega, Madison, WI) vector was used for transfection efficiency control. Cells were lysed, and luciferase activity was measured using the Dual-Luciferase reporter assay system (Promega) and microplate reader (Infinite 200 PRO, Tecan Group Ltd.).
Telomere Repeat Amplification Protocol Assay-A telomere repeat amplification protocol assay (for semiquantitative detection of telomerase activity) was performed using the Telo-TAGGG telomerase PCR ELISA kit (Roche Applied Science). Cell lysates were prepared in lysis reagent from cultured (2 ϫ 10 5 ) cells. The supernatant collected after centrifugation at 16,000 ϫ g for 20 min at 4°C was used for the telomere repeat amplification protocol assay. Telomerase activity in the sample was calculated as units of activity relative to the positive control (MRC5 cells transfected with hTERT). U2OS cells that lacked telomerase activity were used as a negative control. All assays were performed in triplicate.
Coimmunoprecipitation and Immunodepletion Assays-For coimmunoprecipitation, cell lysates (300 g of protein) in 300 l of radioimmune precipitation assay buffer were incubated at 4°C for 1-2 h with anti-myc, anti-hTERT, or anti-mortalin antibody (as indicated). Immunocomplexes were separated by incubation with protein-A/G-Sepharose (Santa Cruz Biotechnology), and Western blotting was performed with the indicated antibodies using the procedure described above. For immunodepletion of hTERT, cell lysates were incubated with anti-telomerase antibody (a gift from Dr. Reddel, Children's Medical Research Institute (CMRI), Australia). Immunoprecipitated hTERT was separated by binding of IgG with the protein-A/G-Sepharose beads. The supernatant was again incubated with the anti-hTERT antibody, and the process was repeated four times. An aliquot of the supernatant after each round of immunoprecipitation was resolved on an SDS gel and was processed for Western blotting with anti-mortalin antibody. Control immunodepletion was performed with isotypematched, irrelevant preimmune IgG.
Statistical Analysis-Data are expressed as mean Ϯ S.E., and the significance of differences between groups was determined by Mann-Whitney test (nonparametric rank sum test) using Stat View software (Abacus Concepts, Inc., Berkeley, CA). Differences were considered significant when p Ͻ 0.05.

RESULTS AND DISCUSSION
Mortalin Is Present in the Nuclei of Cancer Cells-Cytoplasmic and nuclear fractions of several human cancer cell lines were subjected to Western blotting with anti-mortalin antibody. As shown in Fig. 1A, mortalin was enriched in cancer cells and was detected in their nuclear fractions. To rule out the cross-contamination of subcellular fractions, the blots were probed with antibodies to ␣-tubulin (for cytoplasmic fractions) and lamin A/C (for nuclear fractions) (Fig. 1, A and B). Furthermore, we performed a high-resolution single-cell image analysis that confirmed the presence of mortalin in the nuclei of cancer cells (Fig. 1, C and D) and raised the proposition that nuclear mortalin may contribute functionally to the cancer properties.
Nuclear Mortalin Has a Proproliferative Function and Protects Cells against Oxidative Stress-To delineate the functional significance of nuclear mortalin in cancer cells, we generated a mutant that lacked the N-terminal 41 amino acids, the mitochondrial targeting signal peptide sequence. Retroviral expression constructs encoding mot-F (full-length protein, 1-679 residues) and mot-N (amino acid residues 42-679) tagged with a myc epitope were transduced into non-malignant MCF7 breast cancer cells. Expression of mot-F and mot-N proteins was detected by Western blotting with anti-myc and anti-mortalin antibodies. As shown in Fig. 2, A-C, Western blotting and immunostaining of the control and infected cells revealed that the mot-F and mot-N proteins were expressed equally and could be distinguished from the endogenous mortalin (Fig. 2B) and that mot-N was translocated efficiently to the nucleus (Fig.  2C). Similar results were obtained when V5-tagged mot-F and mot-N proteins were expressed in MCF7 and U2OS cells (Fig. 2,  D and E). High-resolution three-dimensional imaging confirmed the presence of mot-F and mot-N in the nucleus. There was more mot-N than mot-F. We next performed fractionation of the transduced cells. Western blotting of cell fractions with anti-myc and anti-mortalin antibodies also revealed that the mot-F localized in the cytosol and in the microsomal and mitochondrial fractions (Fig. 2F, lanes 2). On the other hand, only a fraction of the total mot-N protein in the cytosol fractions and none in the microsomal and mitochondrial fractions was detected (Fig. 2F, lanes 3 and data not shown). Examination of the nuclear fractions confirmed the nuclear localization of mot-N (Fig. 2G). Similar results were obtained using different cell lines and in independent experiments. Of note, we found that exogenously expressed full-length mortalin, mot-F, was also detected in the nuclear fractions, confirming that the nuclear translocation of mortalin is an innate phenomenon in cancer cells, as also supported by the data in Fig. 1.
We next examined the effect of mot-F (which mimics the endogenous mortalin localized at multiple subcellular sites) and mot-N (which is nuclear-enriched) on cancer cell properties. Multiple clones with stable expression of mot-F and mot-N were isolated and first examined the level of mortalin expression for two to three passages (Fig. 3A). The clones were then subjected to cell proliferation, oxidative stress, and malignant transformation assays. As shown in Fig. 3B, mortalin-overexpressing cells showed a higher proliferation rate compared with the untransfected/vector-transfected (control) cells. Furthermore, mot-N derivatives showed higher proliferation compared with the mot-F derivatives (Fig. 3, B and C, and data not shown). To further resolve the effect of mot-N on cell proliferation, we isolated low and high mot-N expression clones. As shown in Fig. 3, C and D, a dose-dependent increase in cell proliferation was detected. Cells were examined for their endogenous and exogenous oxidative stress tolerance by quantitation of ROS. As shown in Fig. 3E, mot-F-and mot-N-overexpressing cells showed significantly lower ROS levels (3-fold), suggesting a decrease in the level of oxidative stress in these cells compared with the untransfected as well as vector-transfected (control) cells. Furthermore, mot-F-and mot-N-overexpressing cells, when exposed to oxidative stress by incubation in H 2 O 2 -supplemented medium, showed better survival than the control cells, suggesting their increased tolerance to exogenous oxidative stress (Fig. 3F). In both assays, mot-N cells were more tolerant than mot-F cells. Similar data were obtained with U2OS and MCF7 cells (Fig. 3, E and F).
In vitro and in vivo cell transformation assays revealed the malignant transformation capacity of mot-F-and mot-N-overexpressing cells (Fig. 4, A-C). Overexpression of mot-F caused a malignant transformation of cancer cells as detected by subcutaneous bone and breast xenograft models of nude mice in a tumor formation assay (12, 18 and data not shown). Furthermore, we found that the mot-N derivatives were more aggressively transformed than mot-F cells (Fig. 4). The mot-N derivatives showed a rounded morphology (Fig. 4A, a), higher anchorage-independent growth capability (Fig. 4A, b), and higher colony forming efficiency (Fig. 4A, c and d) compared with the mot-F derivatives. Cell migration assays revealed higher migration (Fig. 4B, a) and invasion capacity (Fig. 4B, b and c) of mot-N cells. An in vivo tumor-forming capacity assay in nude mice revealed that mot-N derivatives were transformed aggressively. They formed tumors rapidly and aggressively compared with the mot-F derivatives (Fig. 4C, a, and D). Furthermore, when MCF7/mot-F and MCF7/mot-N cells were injected into the opposite flank of the same mouse, MCF7/ mot-N tumors appeared earlier and increased rapidly in size compared with the MCF7/mot-F tumors (Fig. 4C, b). Some mice injected with mot-N derivatives showed multiple tumors all over the injected side of the body (Fig. 4C, c). Furthermore, consistent with the in vitro migration assay data, MCF7/mot-N cells showed aggressive lung metastasis in in vivo (tail vein injection assays) models (Figs. 4E). Taken together, the data demonstrate that mot-N has a proproliferative function and that it contributes to the malignant transformation of cancer cells.
Nuclear Mortalin Contributes to Malignant Transformation by Strong Inactivation of p53-Mortalin has been shown to inactivate p53 by sequestering it in the cytoplasm and also by binding it in the nucleus at the G 1 stage of the cell cycle, where it inactivates its control of the centrosome duplication function (4,6,9,10,15). In light of the finding that mortalin exists in the nucleus of malignantly transformed cancer cells, we investigated the impact of mot-F and mot-N on p53 function by binding and activity assays. For this study, two more deletion mutants (mot-A (N-terminal amino acid residues 1-180 that lack the p53 binding region) and mot-B (amino acid residues 180 -300 containing the p53 binding region)) were generated. Similar to mot-F and mot-N, these constructs were designed to have a myc tag at the carboxyl terminus (Fig. 5A). The cells infected with mot-F, mot-N, mot-A, and mot-B were examined for nuclear translocation of p53 in response to DNA damage stress in U2OS cells that harbor wild-type p53 function. As shown in Fig. 5B, control cells showed that the etoposide caused nuclear translocation of p53 (green) as seen by its intense nuclear staining and absence of p53 (green) in the cytoplasm (Fig. 5B, compare a with b). Cells infected with mot-F, mot-N, and mot-B strongly inhibited the nuclear translocation of p53 (Fig. 5B, c, d, and f). The presence of p53 in the cytoplasm was marked by an overlay of p53 and mortalin (yellow) (Fig. 5B, c, d,  and f). Mutant mot-A-infected cells (which lacked the p53 binding region of mortalin) did not show any difference from the control etoposide-treated cells, and there was no colocalization of mortalin and p53 in the cytoplasm (Fig. 5B, b and e).
To quantitate this effect, we counted the cells with very intense p53 staining (Fig. 5C) and found that mot-F, mot-N, and mot-B caused an approximately 2-to 20-fold decrease in the etoposide-induced nuclear translocation of p53. Of note, we found that the inactivation of p53 by mot-N was significantly higher than that of mot-F. Similar results were obtained in etoposidetreated MCF7 cells overexpressing either mot-F or mot-N (data not shown).
We anticipated that the stronger inactivation of p53 by mot-N may explain the increased malignant characteristics of cancer cells. To investigate this proposition, we examined the binding of p53 to mot-F and mot-N derivatives by coimmunoprecipitation assays. As shown in Fig. 5D, the amount of p53 precipitated with mot-N was more than with mot-F, suggesting higher complex formation between p53 and mot-N (Fig. 5D). Furthermore, p53-dependent reporter assays on mot-F and mot-N derivatives of U2OS cells revealed stronger inactivation in the latter (Fig. 5E). Control of centrosome duplication by p53 has been shown to be regulated by a p53-mortalin interaction at the centrosome and by the mitotic kinase Aurora A (6,15). Hence, we examined the localization of mot-F and mot-N at the centrosome by coimmunostaining of mortalin and Aurora A. As shown in Fig. 5F, mot-N cells showed enhanced colocalization with Aurora A. Furthermore, inactivation of the control of the  SEPTEMBER 5, 2014 • VOLUME 289 • NUMBER 36

JOURNAL OF BIOLOGICAL CHEMISTRY 24839
centrosome duplication function of p53, resulting in aneuploid chromosomes as examined by ␥-tubulin staining, revealed higher aneuploidy in mot-F and mot-N derivatives compared with con-trol MCF7 cells. The aneuploidy score was higher in cells expressing mot-N compared with the ones expressing mot-F protein ( Fig.  5G and data not shown). These data suggested that nuclear mor-FIGURE 5. Nuclear mortalin causes strong inactivation of p53. A, schematic diagram of mortalin constructs. sp, signal peptide. B, overexpression of mortalin and its p53-binding mutants inhibited etoposide-induced nuclear translocation of p53. Double immunostaining with anti-p53 (green) and anti-mortalin (red) antibodies showed nuclear translocation of p53 in control, mot-A mutant-transfected etoposide-treated cells. Mot-F-, mot-N-, and mot-B-transduced cells showed cytoplasmic retention of p53, as seen by cytoplasmic staining that is colocalized with mortalin and, hence, seen as yellow. C, quantitation of cells with etoposide-induced nuclear translocation of p53. D, coimmunoprecipitation (IP) of mot-F and mot-N with p53. Immunoprecipitated mortalin-myc by anti-myc antibody (Myc-IP) was examined for the presence of p53. Control immunoprecipitation was performed with isotype-matched IgG. Quantitation of p53 precipitated with mortalin-myc is shown. E, a p53-dependent reporter assay showed a stronger inactivation of p53 activity in mot-N-transduced cells. Luc, luciferase. F, coimmunostaining of mortalin and Aurora A revealed a stronger intensity of coimmunostaining (yellow signal) of the two proteins in MCF7/mot-N than MCF/mot-F cells. G, coimmunostaining of mortalin-myc and ␥-tubulin showing the presence of aneuploid cells in mot-F and mot-N cells. The cell aneuploidy and anomalies score was higher in mot-N cells.
talin contributes to the malignant properties of cancer cells through the strong inactivation of p53 functions, including transcriptional activation and control of centrosome duplication.
Nuclear Mortalin Contributes to Malignant Transformation by Activation of Telomerase and hnRNP-K-In view of the nuclear localization of mortalin and its impact on cell proliferation, transformation, and metastasis of cancer cells, we next examined its impact on two oncogenes, telomerase and hnRNP-K, that have been shown to act as key regulators of tumorigenesis and metastasis, respectively. As shown in Fig. 6, A and B, mot-F and mot-N cells showed higher levels of telomerase protein (Fig. 6A) and activity (Fig. 6B) compared with the control. Furthermore, telomerase activity was higher in mot-N derivatives than in mot-F cells. We next investigated whether the increase in telomerase activity was due to direct interaction with mortalin in the nucleus. Coimmunoprecipitation of hTERT and mortalin from the nuclear fractions of highly malignant cells, such as MCF7/mot-N and T47D cells, showed the presence of mortalin-hTERT complexes. Non-malignant cells (MCF7, Saos-2, and YKG-1), on the other hand, lacked a mortalin-telomerase interaction ( Fig. 6C and data not shown). To quantitate these interactions, we performed serial immunodepletion assays on mot-F and mot-N cell lysates. Serial immunodepletion of hTERT using a hTERT-specific antibody caused parallel immunodepletion of mortalin (detected by antimyc and anti-mortalin antibodies) in mot-F and mot-N cells (Fig. 6D), suggesting the direct interaction of telomerase and mortalin. These data demonstrate, for the first time, that mortalin interacts with telomerase in the nucleus, causes its stabilization and activation, and contributes to the malignant transformation of cancer cells.
We next investigated the mechanism of increased metastatic properties of mot-F and mot-N cells (Figs. 7, A-D). Analyses of matrix metalloproteases (well established markers of cancer cell metastasis) in MCF7, MCF7/mot-F, and MCF7/mot-N cells revealed a statistically significant increase in the level of MMP-2, MMP-3, and MMP-9 expression (Fig. 7A). Furthermore, we investigated whether the increase in MMPs was due to activation of their upstream regulator, hnRNP-K (36 -38), by nuclear mortalin. As shown in Fig. 7A, MCF7/mot-F and mot-N cells showed a higher level of expression of hnRNP-K that matched well with their high migratory characteristics in vitro and in vivo (Fig. 4). To investigate the cross-talk between mortalin and hnRNP-K, we also generated hnRNP-K-overexpressing cells and examined the level of mortalin expression. We found that the hnRNP-K-overexpressing cells possessed highly malignant characteristics (39). However, as shown in Fig.  7B, the level of mortalin expression in these cells remained unchanged compared with the parent cells, suggesting that mortalin may work upstream of hnRNP-K in promoting the cancer metastasis. Coimmunoprecipitation of mortalin and hnRNP-K showed that the two proteins form a complex (Fig.  7C), and an increased level of expression of hnRNP-K in mor-talin-overexpressing cells might be the result of its stabilization and protection from HDM2-mediated degradation (40). In support of this, immunostaining revealed that a significantly high proportion of MCF7/mot-F and MCF7/mot-N cells have nuclear hnRNP-K, in contrast with MCF7 cells. Although only 10 -20% MCF7 cells showed nuclear hnRNP-K, more than 75% of either the MCF7/mot-F or MCF7/mot-N cells exhibited strong nuclear staining (Fig. 7D).
Telomerase is a ribonucleoprotein complex composed of the catalytic protein subunit human telomerase reverse transcriptase (hTERT) and the RNA subunit human telomerase RNA that serves as the template for the synthesis of telomeric DNA and is expressed constitutively expressed in cells. Telomerase activity is closely associated with hTERT expression. In contrast to normal somatic cells, most human tumor cells possess a high level of hTERT expression. Several studies have demonstrated that telomerase activity is required for the malignant properties of cancer cells, and, hence, that they may serve as a good target for the development of anticancer drugs (41). Stud- ies of the regulation of hTERT expression have revealed that it is regulated by trans-regulatory factors, including Sp1/Sp3, AP-2␤, Myc/Max heterodimer, HIF-2␣, TAK1, MSH2, GRHL2, p53, HER2, hnRNP-K, and hnRNP-D (42)(43)(44), that show physical interactions with the hTERT promoter. It has been shown earlier that mortalin overexpression cooperates with telomerase to extend the in vitro lifespan of normal human fibroblasts (13). Taken together with our data, we conclude that nuclear mortalin interacts with telomerase and activates its function, contributing to proliferation and malignant characteristics of cancer cells.
hnRNP-K is a member of the family of about 20 hnRNP proteins that are expressed ubiquitously, complex with heterogeneous nuclear RNA, and influence pre-mRNA processing and transport involved in transcription and posttranscriptional messenger RNA metabolism. hnRNP-K is known to regulate a multitude of gene expressions, including p53 (40), as a transcription factor or by alteration of mRNA stability and translation (45). Kang et al. (44) have shown that hnRNP-K associates physically with the hTERT promoter in vitro and in vivo. Although knockdown of hnRNP-K did not affect telomerase activity, its knockdown, along with GRHL2, resulted in an inhibition of telomerase function in oral squamous cell carcinoma (44). Consistent with its telomerase activity-enhancing function, hnRNP-K was enriched in human oral squamous cell carcinoma cells but not in normal cells (which lack telomerase activity). hnRNP proteins are known to shuttle between the nucleus and the cytoplasm and have a role during cell cycle progression. Cytoplasmic accumulation, which is required for the ability of hnRNP-K to silence mRNA translation, is dependent on phosphorylation mediated by MAPK/ERK (46). Several studies have reported up-regulation of hnRNP-K and its altered subcellular localization in tumors (47). A higher level of nuclear hnRNP-K has been reported in proliferating compared with resting cells. In contrast, the level of cytoplasmic hnRNP-K protein was either the same or lower in dividing compared with quiescent cells. States of enhanced proliferation were also associated with increased phosphorylation of hnRNP-K. On the basis of these findings, it has been suggested that the nuclear hnRNP-K is involved in cell proliferation regulatory signaling (48). It is modified in response to changes in the intracellular and extracellular environment, including cytokines, growth factors, and oxidative stress (49), and bridges signal transduction pathways to chromatin remodeling and nucleic acid-directed processes. hnRNP-K has also been shown to bind to the 3Ј UTR of VEGF mRNA and regulate its translation (50) and, in turn, would affect the expression of MMPs. Mot-F and Mot-N cells showed an increase in the level of expression of MMP-2, MMP-3, and MMP-9. MMPs are up-regulated in many tumors. They cause epithelial-mesenchymal transition, malignant transformation, and genomic instability by mechanisms involving articulation of the tumor environment, oxidative stress, and induction of transcription factors.
To obtain further insights into the clinical relevance of nuclear mortalin in cancer, we examined normal and tumor (ovary, kidney, lung, liver, and brain) tissues by mortalin immunohistochemistry. Mortalin was found to be highly expressed and enriched in the nuclei of all tumor tissues examined com-pared with the controls (Fig. 7E and data not shown). A similar increase in mortalin has also been reported in liver cancers and melanoma biopsy samples (2,10,51). In agreement with the role of nuclear mortalin in tumors, human normal lung fibroblasts lacked endogenous nuclear mortalin staining (data not shown). Furthermore, although the exogenous mot-F and mot-N were localized in the nuclei, the cells neither showed an increase in proliferation or transformation nor in malignant properties (data not shown). The unaltered phenotype of normal cells could be due to the lack of mortalin-p53 interaction (9), telomerase, and hnRNP-K proteins (39 -41).
Taken together, this study demonstrates, for the first time, a new line of action of the stress chaperone mortalin in human carcinogenesis. In addition to the cytoplasmic sequestration and inactivation of p53 described previously, it localizes in the nucleus, inactivates p53-mediated control of centrosome duplication, causing genomic instability, activates the telomere maintaining enzyme hTERT (a hallmark of most cancers) and a multifunctional chromatin-remodeling protein, hnRNP-K, involved in regulation cell migration, and, therefore, promotes carcinogenesis.