H2O2 induces nuclear transport of the receptor tyrosine kinase c-MET in breast cancer cells via a membrane-bound retrograde trafficking mechanism

Reactive oxygen species (ROS) are cellular by-products produced from metabolism and also anticancer agents, such as ionizing irradiation and chemotherapy drugs. The ROS H2O2 has high rates of production in cancer cells because of their rapid proliferation. ROS oxidize DNA, protein, and lipids, causing oxidative stress in cancer cells and making them vulnerable to other stresses. Therefore, cancer cell survival relies on maintaining ROS-induced stress at tolerable levels. Hepatocyte growth factor receptor (c-MET) is a receptor tyrosine kinase overexpressed in malignant cancer types, including breast cancer. Full-length c-MET triggers a signal transduction cascade from the plasma membrane that, through downstream signaling proteins, up-regulates cell proliferation and migration. Recently, c-MET was shown to interact and phosphorylate poly(ADP-ribose) polymerase 1 in the nucleus and to induce poly(ADP-ribose) polymerase inhibitor resistance. However, it remains unclear how c-MET moves from the cell membrane to the nucleus. Here, we demonstrate that H2O2 induces retrograde transport of membrane-associated full-length c-MET into the nucleus of human MCF10A and MCF12A or primary breast cancer cells. We further show that knocking down either coatomer protein complex subunit γ1 (COPG1) or Sec61 translocon β subunit (SEC61β) attenuates the accumulation of full-length nuclear c-MET. However, a c-MET kinase inhibitor did not block nuclear c-MET transport. Moreover, nuclear c-MET interacted with KU proteins in breast cancer cells, suggesting a role of full-length nuclear c-MET in ROS-induced DNA damage repair. We conclude that a membrane-bound retrograde vesicle transport mechanism facilitates membrane-to-nucleus transport of c-MET in breast cancer cells.

nucleus via at least two main trafficking pathways: 1) RTK intracellular domains generated by ␥-secretase cleavage are released into cell cytosol for direct translocation into nucleus, and 2) RTKs with intact transmembrane domains translocate into nucleus through a clathrin-mediated endocytic mechanism and then travel in a retrograde trafficking vesicle by a membrane-bound mechanism (12,13). The retrograde membrane vesicle transport mechanism includes the trafficking from the Golgi body to endoplasmic reticulum (ER) where the coat protein complex I (COPI) vesicle is an essential carrier for MRIN transport between Golgi stacks and also carries MRINs from the Golgi to ER (14 -16). After reaching the ER, the ER-tonucleus transport of MRIN requires Sec61␤ (16,17). However, there are also two mechanisms reported for ER-to-nucleus transport of MRIN: 1) integrative nuclear FGFR-1 signaling (INFS) for FGFR-1 and 2) INTERNET for EGFR (16,18,19). For the INFS pathway, FGFR-1 is pumped out from the ER and enters the cytosol before entering the nucleus through the nuclear pore complex, whereas for the INTERNET pathway, EGFR remains membrane-bound and is transferred from the ER membrane connecting to outer nuclear membrane, through nuclear pore complex and then into the inner nuclear membrane (13,20). The exact mechanism by which the MRIN enters the nucleus from inner nuclear membrane is not yet clear. However, Sec61␤ in the inner nuclear membrane is required, and a membrane-associated vesicle is likely involved (16,20).
Various RTKs are reported to shuttle from the plasma membrane into cell nucleus where they are known to have noncanonical functions in transcriptional regulation, DNA replication, and DNA repair regulation (9,11,12,16,(21)(22)(23)(24)(25)(26). c-MET is an RTK commonly activated in EGFR inhibitor-resistant cancer patients (27)(28)(29)(30). Canonical full-length c-MET holoreceptor is a cell membrane RTK that is activated by its ligand hepatocyte growth factor (HGF), and the activation of c-MET holoreceptor is important for promoting cell proliferation, migration, and survival through the common RTK downstream messenger proteins, such as GAB1/Grb2, phosphatidylinositol 3-kinase, and STAT3 (31). Overexpression of c-MET has been observed in multiple cancer types, including triple-negative breast cancer (TNBC), and is highly correlated with poor prognosis and drug resistance in cancer patients (29,30,(32)(33)(34)(35)(36). However, the biological function of c-MET is not limited to its canonical signaling role from plasma membrane. c-MET holoreceptor also translocates into the nucleus rapidly to initiate calcium signaling in liver cells upon HGF stimulation (37). Moreover, constitutively active cytosolic fragments of c-MET (MET-CTF) are present continuously in the nuclei of TNBC cell line MDA-MB-231 and involved in transcription regulation (38). Previously, we demonstrated that c-MET can translocate into nucleus to interact with and phosphorylates PARP1 in response to H 2 O 2 stimulation, which contributes to PARP inhibitor resistance in TNBC (10).
In this study, we report that upon H 2 O 2 stimulation, c-MET holoreceptor utilizes INTERNET pathway to translocate into the nucleus via the COPI/Sec61␤-mediated retrograde vesicle trafficking. The H 2 O 2 -induced nuclear transport of c-MET occurs independently from c-MET phosphorylation. Moreover, we also show that nuclear c-MET can interact with KU proteins, which binds DNA double-strand breaks in DNA repair. Our findings suggested the involvement of H 2 O 2 -induced nuclear c-MET in DNA damage repair.

ROS-induced nuclear full-length c-MET accumulates in breast cancer cells
To investigate whether full-length c-MET holoreceptor can translocate into the nucleus in response to ROS, we treated several c-MET-expressing cell lines derived from human mammary gland (MCF10A and MCF12A) or breast cancers with 10 mM H 2 O 2 followed by cellular fraction and Western blotting to detect c-MET. We observed an accumulation of fulllength c-MET in the nucleus for all cell lines tested after H 2 O 2 treatment, whereas differential response to H 2 O 2 treatment was observed for the truncated form of c-MET (MET-CTF) (Fig. 1A). We then further stimulated MDA-MB-231 and BT-549 cells with increasing concentrations of H 2 O 2 and found that nuclear c-MET holoreceptor accumulation ocurred in a dose-dependent manner (Fig. 1B). Treatment of 1-10 mM H 2 O 2 induced c-MET nuclear translocation. The concentration of H 2 O 2 at 10 mM was used in our experimental setting to observe significant nuclear c-MET accumulation to study mechanisms of membrane receptor trafficking. This concentration has been shown to be sufficient to affect multiple physiology functions in mammalian cells including DNA damage (39,40). Time-course experiments indicated that a 30-min H 2 O 2 stimulation is sufficient for c-MET nuclear accumulation in both MDA-MB-231 and BT-549 cells (Fig. 1C). ROS inducing agents, such as doxorubicin, IR, and sodium arsenite, can also induce accumulation of full-length nuclear c-MET in breast cancer cells (Fig. 1, D and E, and Fig. S1), suggesting that c-MET nuclear translocation phenomenon is not limited to direct H 2 O 2 treatment. On the contrary, chemotherapeutic agents that do not stimulate ROS accumulation, such as cisplatin and paclitaxel, did not induce nuclear c-MET accumulation (Fig. S1). We also found that prolonged ligand treatment (100 ng/ml HGF for 1-6 h) stimulated full-length nuclear MET accumulation in both MDA-MB-231 (1 h) and BT549 cells (6 h) (Fig. S2). Quantitative Western blotting analysis showed that full-length nuclear c-MET increased by 2-fold in MDA-MB-231 cells and by up to 7-fold in BT-549 cells (Fig. S2). However, unlike ligand-stimulated nuclear EGFR accumulation, which takes only within 30 min (18), nuclear accumulation of fulllength c-MET takes at least 1 h after ligand stimulation (Fig. S2). Together, c-MET can translocate into the nucleus efficiently in response to different stimuli, but the ligand-induced translocation seems to be slower than that of EGFR.
To determine the number of cells that contained nuclear c-MET after HGF or H 2 O 2 stimulation, we performed several experiments using MDA-MB-231 and BT-549 cells. Immunostaining of c-MET indicated that it mainly located in the cytosol under normal culture conditions (control). It also aggregated at the perinuclear and nuclear regions in response to either HGF or H 2 O 2 stimulation. Quantitative confocal microscopy also showed increasing amounts of cells containing nuclear c-MET after H 2 O 2 stimulation. Consistent with the results from West-

Membrane-bound nuclear translocation mechanism of c-MET
ern blotting analysis, we observed an apparent increase in the number of c-MET-containing nuclei following H 2 O 2 treatment but only subtle increases in those treated with HGF (Fig.  1F). These results indicated that prolonged and high-concentration ligand stimulation at 100 ng/ml can induce c-MET holoreceptor nuclear translocation, but the response is less efficient than cellular stress stimulation.

Microtubules mediate full-length c-MET transport to the Golgi
As mentioned, oxidative stress elicits stronger nuclear translocation of c-MET than did ligand stimulation (Fig. 1, B and F, and Fig. S2); we further investigated the mechanisms of c-MET nuclear transport under H 2 O 2 stimulation. Based on the detection of full-length c-MET, which contains the transmembrane domain, we speculated that c-MET nuclear trafficking may occur via the membrane-bound INTERNET retrograde trafficking mechanism utilized by EGFR. To validate our hypothesis, MDA-MB-231 cells were treated with Golgi apparatus inhibitor brefeldin A (BFA) for 30 min to interrupt normal Golgi function prior to H 2 O 2 treatment. BFA treatment decreased the accumulation of nuclear c-MET holoreceptor ( Fig. 2A), suggesting that c-MET is transported to the Golgi apparatus before reaching the nucleus.
Membrane-bound cargo transport requires microtubules; therefore, we investigated whether disrupting the microtubule assembly also affects c-MET nuclear transport. We found that H 2 O 2 -induced c-MET nuclear translocation is reduced in cells pretreated with microtubule-depolarizing agent nocodazole, as well as microtubule dynamicity-decreasing agent paclitaxel (Fig. 2, B and C). Because c-MET is primarily localized in the cytosol in breast cancer cells under normal culture condition, it may be challenging to determine whether the cytoskeleton is important for c-MET nuclear transport. To this end, we investigated the effects of microtubule inhibition in membrane c-MET transport to the cytosolic compartment in HeLa cells, in which the majority of c-MET is localized in the cell membrane under normal culture condition and is widely used to demonstrate nuclear RTK transport (41)(42)(43). This microtubule-dependent c-MET transport is not limited to breast cancer cells. Membrane c-MET decreased, whereas cytosolic c-MET increased in HeLa cells and accumulated at a similar location as trans-Golgi protein, GalNac T2, after HGF stimulation. This accumulation was inhibited by nocodazole pretreatment (Fig. 2D, yellow spots marked by arrows in inset 2). Nocodazole treatment also decreased the internalization of membrane c-MET into the cytosol (Fig. 2D), suggesting that nuclear accumulation of c-MET in HeLa cells may begin as receptor endocytosis similar to that reported for EGFR nuclear transport (43,44,46). However, unlike EGFR transport in HeLa cells, knocking down dynein did not affect c-MET nuclear transport in MDA-MB-231 cells (Fig. 2E), suggesting that the microtubuledependent transport mechanism of nuclear c-MET may rely mainly on cargo transport proteins instead of dynein in breast cancer cells. We also found that both nocodazole and paclitaxel treatments increase population of nuclear pro-MET ( Fig. 2B and Fig. S1), a 170-kDa partially glycosylated single-chain precursor of c-MET synthesized in ER (47,48). These findings indicate that the nuclear localized pro-MET and c-MET may both transport from ER through retrograde mechanism.

COPI and Sec61␤ mediate c-MET ER-to-nucleus transport
To determine whether the transport of membrane-bound c-MET to the nucleus occurs via the retrograde trafficking mechanism through the Golgi apparatus and ER, we suppressed cis-Golgi-to-ER retrograde transport COPI complex vesicle by knocking down its core subunit COPG1 in MDA-MB-231 cells by shRNA targeting COPG1 to diminish vesicle trafficking from Golgi to ER. Knocking down COPG1 significantly decreased H 2 O 2 -induced c-MET nuclear accumulation (Fig. 3A). Immunofluorescent images also showed less nuclear c-METcontaining populations after H 2 O 2 treatment in COPG1 knockdown cells (Fig. 3B). These results indicated that COPI complex plays an important role in c-MET nuclear transport and supported our hypothesis that full-length c-MET is transported from cis-Golgi to ER in membrane-bound vesicles via retrograde pathway similar to transportation pathway of EGFR (16).
We then investigated the transport of c-MET from ER to nucleus by using RNAi to knockdown ER membrane protein translocator Sec61␤ in MDA-MB-231 cells. We found that H 2 O 2 -induced c-MET nuclear accumulation significantly decreased in Sec61␤ knockdown cells (Fig. 3, B and C). Because it has been reported that RTKs can travel from the ER into the nucleus through either the INFS (through cytosol) or INTER-NET (membrane bound) pathway, we performed cellular fractionation of H 2 O 2 -treated MDA-MB-231 cells followed by

Membrane-bound nuclear translocation mechanism of c-MET
Western blotting to detect c-MET. c-MET was detected in the membrane-bound fraction but not the soluble cytosolic fraction, indicating that c-MET remains membrane-bound throughout the cytosolic trafficking process before transport into the nucleus (Fig. 3D). Thus, it utilizes INTERNET but not INFS pathway.

Kinase-independent transport and kinase-dependent noncanonical function of nuclear c-MET
Traditionally, RTKs are internalized from plasma membrane into the cytosol after ligand stimulation. By using c-MET autophosphorylation sites (Tyr-1234/1235) as indicators, we found H 2 O 2 stimulation activated c-MET (Fig. 4A). Therefore, we further investigated whether this c-MET activation is required for its nuclear transport. In MDA-MB-231 and BT-549 cells, crizotinib, a c-MET kinase inhibitor inhibited H 2 O 2 -induced c-MET phosphorylation (Fig. 4A) but failed to block H 2 O 2 -induced c-MET nuclear accumulation (Fig. 4B), suggesting that nuclear transport of c-MET occurs independently of its kinase activity.
Previously, we reported that c-MET interacts with PARP1 and contributes to PARP inhibitor resistance through reduced ability of PARP1 to bind to PARP inhibitors (10). Here we showed that c-MET can also bind to KU proteins after H 2 O 2 treatment, and their interactions were decreased by pretreating cells with crizotinib ( Fig. 4C), indicating that the activated phosphorylated c-MET interacts with KU. Although PARP1 and KU are key sensors of DNA double-strand breaks (49,50), our findings suggested that nuclear c-MET may play a role in DNA damage repair. We then investigated the effects of c-MET inhibition on DNA damage repair rate. MDA-MB-231 cells were treated with crizotinib to inhibit c-MET kinase activity and hydroxyurea and AraC to accumulate DNA damage breaks by inhibiting DNA damage repair. As shown in Fig. 4D, c-MET inhibition significantly delayed repair of H 2 O 2 -induced DNA damage based on comet assay. Together, the results suggest that c-MET may interact with KU to enhance DNA damage repair efficiency.  (56). To date, the function of nuclear c-MET is not well-characterized in breast cancer cells because unlike the well-studied EGFR, in which EGF stimulates nuclear translocation of full-length EGFR in breast cancer cells, c-MET ligand cannot induce c-MET nuclear transport effectively (38). Moreover, it was not clear which types of cellular stress can stimulate c-MET nuclear transport. Here, we revealed that fulllength c-MET nuclear accumulation is induced by elevated oxidative stresses caused by H 2 O 2 and ROS-inducing anti-cancer agents in c-MET-overexpressing breast immortalized and cancer cells. Although nuclear MET-CTF has been detected in both breast and colon cancer cells by ligand induction, previous studies showed that full-length c-MET remains at plasma membrane (38,57). Consistent with those findings, we showed that unlike ligand-stimulated nuclear EGFR accumulation, the ligand-induced nuclear accumulation of full-length c-MET can only be detected in breast cancer cells under nonphysiologically relevant high concentrations of ligand and prolonged ligand treatment (Fig. S2). These results suggested that c-MET nuclear accumulation may not be affected directly through ligand-induced activation. Although Gomes et al. (37) demonstrated ligand-induced full-length c-MET accumulation in hepatocyte cells and given that c-MET ligand is hepatocyte growth factor, we speculated that c-MET nuclear transport and its functions in nucleus are different in different tissues.

Discussion
Nocodazole and paclitaxel can both decrease nuclear c-MET accumulation similar to EGFR, suggesting that the nuclear trafficking depends on cargo transport along microtubule. It is interesting that when we knocked down dynein, we did not observe any notable inhibition in c-MET nuclear accumulation under H 2 O 2 treatment in breast cancer cell, suggesting that nuclear c-MET may utilize other cargo transporting system along microtubules. Dynein and kinesin are major microtubule cargo-transporting proteins (58). Because kinesin family is overexpressed in breast cancer and plays important roles in promoting breast cancer progression (59 -61), it is conceivable, yet needs to be confirmed, that nuclear c-MET may utilize kinesin rather than dynein transport in breast cancer cells. In addition, it has been reported that cargo transport function can be rescued by activation of kinesin-14 in dynein-knockdown cells (62). We speculated that knocking down dynein triggered a

Membrane-bound nuclear translocation mechanism of c-MET
feedback regulation between dynein and kinesin and thus restored c-MET nuclear accumulation in dynein-knockdown cells. It is reported that kinesin family proteins are overexpressed in breast cancer and are related to drug resistance and poor prognosis (63,64); we speculated that kinesin may contribute to c-MET transportation in breast cancer cells. There are 45 kinesin family genes in humans (65), it may require a systematic study to clarify whether kinesin is involved in and, if so, which kinesin may play a role in c-MET nuclear transport in breast cancer cells.
We found that c-MET nuclear accumulation induction agents include doxorubicin, arsenite, and IR, indicating the function of nuclear full-length c-MET may be closely related to resistance of anti-cancer therapeutic agents. Indeed, although we previously reported that c-MET can interact with and phosphorylate PARP1 to cause resistance to PARP inhibitors (10), we further demonstrated in this study that H 2 O 2 -induced nuclear c-MET also interacts with KU proteins. Although DNA-damaging sensors PARP1 and KU proteins are known to compete with each other for DNA repair pathways (49,50,66), it would be of interest to determine the role of nuclear c-MET in DNA damage repair pathway in the future. Based on our immunoprecipitation data (Fig. 4C), crizotinib decreased the interaction between c-MET and PARP1, as well as that of c-MET and KU 70. The DNA damage repair rates were slower at the early time point (30 min after H 2 O 2 treatment) in response to crizotinib treatment (Fig. 4D). These results suggested that the interactions between c-MET, PARP1, and Ku proteins may affect DNA damage repair rates. Previous studies have shown that PARP1 and KU compete with each other and affect the choice of repair pathways (50,66). Therefore, we speculated that c-MET is involved in the early stages of DDR when PARP1 and KU proteins play key roles in pathway choices. However, more detailed investigation would be required to determine whether the decrease in repair efficiency is attributed to the different pathways utilized for repair. Other possibilities, including the reduction in the activities of DNA damage repair proteins, cannot be ruled out based on the results of the comet assay. A more complete study would be required to derive a clearer conclusion. We also demonstrated that c-MET kinase inhibitor can block H 2 O 2 -induced c-MET activation but cannot block H 2 O 2induced c-MET nuclear accumulation. Given that RTKs also have kinase-independent function (67,68), it is important to investigate the transport mechanisms of c-MET because c-MET may have kinase-independent functions in the nucleus. We specifically investigated whether c-MET and EGFR share similar nuclear transport pathways because c-MET and EGFR have been shown to function as heterodimers and contribute to each one's resistance to TKI (69,70). Although it is still unclear whether nuclear functions of EGFR can also be induced by nuclear c-MET, there is the possibility that c-MET and EGFR can assist nuclear transport of each other through heterodimer formation. Our study revealed a mechanism of c-MET nuclear transport that is similar to that of EGFR, in which full-length c-MET is transported from plasma membrane to Golgi apparatus. This process is inhibited by BFA or by knocking down cis-Golgi-to-ER cargo protein complex core subunit COPG1. After reaching the ER, c-MET requires the presence of Sec61␤ for its accumulation in the nucleus. During Sec61␤-mediated transport, c-MET remains membrane-bound because it is not detected in soluble cytosolic fraction. Therefore, we concluded that c-MET follows the INTERNET mechanism instead of INFS. In summary, because c-MET overexpression is highly correlated to poor cancer patient prognosis and the oncogenic functions of EGFR/c-MET heterodimers, in-depth studies are required in the future to explore both kinase-dependent and kinase-independent noncanonical functions of nuclear c-MET.

Materials
The antibodies used in this study are MET (C-12) antibody  and subjected to cellular fractionation using ProteoExtractா subcellular proteome extraction kit. Tubulin and Sp1 was used as markers for non-nuclear and nuclear fractions, respectively. Calnexin was used as marker for membrane-bound non-nuclear fraction. Emerin is a protein predominantly located in the Golgi and nuclear membrane and is used as a marker for membrane-bound organelles, including the nucleus.

Membrane-bound nuclear translocation mechanism of c-MET
HGF human recombinant was purchased from BioVision Inc. H 2 O 2 , BFA, nocodazole, and paclitaxel were purchased from Sigma-Aldrich.

Cell culture
All cell lines were originally obtained from ATCC (Manassas, VA). MDA-MB-231, BT549, and Hs578T cells were maintained in Dulbecco's modified Eagle's medium (DMEM)/F12 with 10% fetal bovine serum in the presence of 100 units/ml penicillin and 100 g/ml streptomycin. MCF-10A and MCF-12A cells were maintained in DMEM/F12 medium with 5% horse serum in the presence of 100 units/ml penicillin, 100 g/ml streptomycin, 10 g/ml insulin, 20 ng/ml EGF, 100 ng/ml cholera enterotoxin, 0.5 g/ml hydrocortisone, and 1 mM calcium chloride. Cell lines were validated by STR DNA fingerprinting using the AmpF_STR Identifiler kit according to manufacturer's instructions (Applied Biosystems catalog no. 4322288). The STR profiles were compared with known ATCC fingerprints and to the Cell Line Integrated Molecular Authentication database version 0.1.200808 (http://bioinformatics. hsanmartino.it/clima/) (71). 3 The STR profiles matched known DNA fingerprints.
The cells were incubated in serum-depleted medium for 18 h before HGF treatment but not before H 2 O 2 treatment. The cells were treated with 100 ng/ml HGF or 10 mM H 2 O 2 in the culture medium and incubated at 37°C water-saturated incubator supplied with 5% CO 2 for the times indicated. 50 mM H 2 O 2 is freshly prepared by adding 30% H 2 O 2 stock into cell culture medium in a ratio of 5.75 l/ml. The 50 mM H 2 O 2 were then further diluted to working concentration with cell culture medium.

Cell fractionation
Nuclear and non-nuclear cell fractionations were performed as previously described (43). Briefly, cells were harvested at 90% confluency and lysed in Nori lysis buffer (20 mM HEPES, pH 7.0, 10 mM KCl, 0.5% Nonidet P-40, 2 mM MgCl 2 , 2 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and 0.15 mg/ml aprotinin) on ice for at least 10 min. Following lysis, the cells were homogenized by Dounce homogenizer and centrifuged 1,500 ϫ g for 5 min, and the supernatants were collected as non-nuclear fractions. The pelleted nuclei were washed with the same lysis buffer and centrifuged 1,500 ϫ g for 5 min to sediment nuclear pellet. The pellets were then solubilized in NETN buffer (150 mM NaCl, 1 mM EDTA, 20 mM Tris-Cl, pH 8.0, 0.5% Nonidet P-40, and protease inhibitor mixture mixture) and incubated on ice for 10 min before 10-s sonication three times (Sonics Vibra-Cell, amplitude 30; Sonics & Materials, Newtown, CT). The extract was centrifuged at maximum speed for 10 min. The supernatant was collected as nuclear fraction. The soluble cytosolic and membrane cell fractionations were performed by using ProteoExtract subcellular proteome extraction kit (Calbiochem Merck Millipore) following the manufacturer's protocol with modifications as described (18).

Western blotting analysis
For detecting proteins in cell fractionation samples, 40 -60 g of proteins were loaded for each lane in SDS-PAGE. The same amounts of nuclear and non-nuclear fraction proteins were used in Western blotting analysis. Primary antibodies were hybridized at 4°C overnight and followed by horseradish peroxidase-conjugated secondary antibody hybridization. The images were quantified by using Image Studio TM Lite version 5.2.
For cellular fractionation experiments, nuclear envelope protein lamin B or histone H3 was used to indicate the cell nuclear fraction. Both proteins were used to insure that H 2 O 2 treatment does not cause protein degradation in either of the proteins. ER protein calregulin was used to indicate that there is no contamination of ER proteins in nuclear fraction. Tubulin was used to signify non-nuclear fractionation.

Immunoprecipitation
Immunoprecipitation is performed with MDA-MB-231 cells or MDA-MB-231 cell that stably expressing FLAG-tagged c-MET. Indicated in the experiment, the cells were incubated with or without 500 nM crizotinib overnight before stimulated with 10 mM H 2 O 2 for 30 min prior to lysate collection. In brief, the cells were lysed in radioimmune precipitation assay buffer (150 mM NaCl, 1% Triton X-100, 0.1% SDS, 50 mM Tris-HCl, pH 7.5, and protease inhibitor cacktail), and 500 g of total protein was used for immunoprecipitation by incubating with anti-FLAG M2 affinity gel beads (Sigma-Aldrich) overnight rotating at 4°C. Precipitates were washed three times with immunoprecipitation wash buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin) and eluted from magnetic beads using SDS-PAGE sample buffer.  2-4) were treated with crizotinib and H 2 O 2 alone or in combination. c-MET was then immunoprecipitated (IP) by using FLAG-tag antibody, and the precipitates were subjected to Western blotting analysis with PARP1, Ku70, Ku80, and actin antibodies. Inputs were cell lysates prior to immunoprecipitation. D, MDA-MB-231 cells were treated with hydroxyurea/AraC to inhibit DNA damage repair for overnight followed by 30-min 10 mM H 2 O 2 treatment. The H 2 O 2 -and hydroxyurea/AraC-containing media were then removed, and the cells were released for DNA damage repair in fresh culture media. The cells were then collected for comet assay analysis at different time points after release (recovery time). The amounts of DNA fragments in comet tail (% DNA in tail) were used as a parameter to calculate DNA damage. DNA damage was normalized to that in in the untreated group (0 min) as 100%. The data represent the means Ϯ S.D. (p Ͻ 0.0001, two-way analysis of variance). Representative images of the comet assay are shown. Scale bars, 20 m. DNA was stained with propidium iodide and imaged under fluorescence microscope. n.s., not significant.

RNA interference
COPG1 or Sec61␤ knockdown MDA-MB-231 stable clones were established via lentiviral gene delivery. The lentivirus particles are produced by transfecting HEK293T cells with pCMV-VSVG, pCMV-⌬ 8.91, and shRNA plasmid in a ratio of 5:1:10. Transfectants were selected by using 1% puromycin-containing DMEM/F12 medium, and stable transfectants were maintained with 0.5% puromycin-containing DMEM/F12 selective medium. The shRNA expression vectors with pLKO.1 backbone was purchased from Sigma-Aldrich, and the sequences are shown in Table 1. Scrambled shRNA (addgene plasmid 1864) was used as nontargeting shRNA controls (shCtrl) for these experiments (45). Dynein intermediate chain transient knockdown were performed with siRNA purchased from Sigma-Aldrich (Table 1). 90 nM siRNA were mixed with Lipofectamine 3000 reagent (Thermo Fisher Scientific) in Opti-MEM medium (Gibco, Thermo Fisher Scientific). The cells were H 2 O 2 treated and harvested for fractionation experiments 3 days after transfection.

Confocal microscopy
Samples for confocal microscopy were prepared as described previously (16). In brief, the cells were seeded on chamber slides for at least 18 h before treatment. After treatment, the cells were washed with PBS, fixed with 4% paraformaldehyde, permeablized with 0.25% Triton X-100/PBS, and stained with the indicated primary or fluorescent-labeled secondary antibodies. Primary antibodies were diluted in 3% normal goat serum/ PBST at the following ratios: c-MET (C-12) antibody, 1:100, HGF antibody, 1:100, and GalNac T2, 1:100. DNA was counterstained with DAPI-containing mounting solution (Vector Laboratories Inc.). Immunostained cells were examined and imaged with Zeiss LSM 710 laser microscope.

Comet assay
After DNA damage induced by H 2 O 2 treatment, cell culture medium was refreshed to allow DNA damage repair process. The base layer of comet slides was prepared with 1.4% high melting temperature agarose and solidified on ice. The cells were harvested at the times indicated, mixed with 1.2% low melting temperature agarose at 1:1 (v/v) ratio, and put onto the base layer of comet slides. The slides were then covered with 1.2% low melting agarose gel and incubated overnight in lysis buffer (2.5 M NaCl, 100 mM EDTA, 10 mM Tris, pH 10, 1% N-laurylsarcosine, 1% Triton, and 10% DMSO). The slides were washed in di-deionized water five times before DNA were denatured in alkaline comet electrophoresis buffer (0.3 N NaOH, 1 mM EDTA) for 25 min. Electrophoresis was performed with alkaline comet electrophoresis buffer at 0.3 A, 25 V for 25 min. Each slide was washed with di-deionized water and transferred to 0.4 M Tris-Cl (pH 7.5) until DNA was stained by applying 40 l of propidium iodide (50 g/ml) onto the slides. A coverslip was applied, and the slides were subjected to examination with fluorescence microscope (Zeiss). Images of at least 50 cells/treatment were recorded with close-circuit display camera (CoolSNAP). DNA damage is quantified by using CometScore (tritekcorp) with a percentages of DNA in tail parameter.

Statistical analysis
For Western blotting signal quantifications, signal intensities were analyzed by using Image Studio Lite (version 5.2) (LI-COR Biosciences, Lincoln, NE). MET signal intensities were first normalized to that of loading control proteins such as lamin B or calregulin before the signals from treatment groups were normalized again to the control group in the experiments. Unless specified, 140-kDa mature MET signal is quantified in Western blotting. Every independent experiment repeats was quantified individually. Fold changes (ϫ) in Western blotting signals were analyzed by a nonparametric Friedman test using the GraphPad Prism 8.0 software. The percentage of MET-containing cell nuclei was calculated based on the immunofluorescent images obtained by confocal microscopy, and at least 160 cell nuclei were counted for each experiment. The differences in colocalization coefficient were analyzed by a nonparametric Kruskal-Wallis test using the GraphPad Prism 8.0 software. A p value less of 0.05 was considered statistically significant (*, p Ͻ 0.05; **, p Ͻ 0.002; ***, p Ͻ 0.001).  Membrane-bound nuclear translocation mechanism of c-MET