香京julia种子在线播放

The MOSE cell lines represented benign (MOSE-E), slow-developing (MOSE-L), and aggressive (MOSE-L_{TIC v} ) ovarian cancer cells generated from C57BL/6 mice have been extensively characterized previously (28–33). These cells express fallopian tube markers (29) and are therefore a model for the highly aggressive serous ovarian cancer. All cells were grown in high glucose DMEM (Sigma Aldrich) supplemented with 4% fetal bovine serum (Atlanta Biological), 3.7g/l sodium bicarbonate, 10 ml/l of penicillin-streptomycin solution at 37°C with 5% CO2 under normoxic (21% O₂) or hypoxic conditions (1–2% O₂). Human TERT-immortalized benign and malignant fallopian tube (FNE1, FNLE1) and benign ovarian epithelial cells (OCE1) were obtained from the Live Tumor Culture Core at the University of Miami Sylvester Comprehensive Cancer Center and cultured in Primaria tissue culture flasks (Becton Dickinson) with FOMI media supplemented with 25ng/ml cholera toxin as described (34). SKOV3 (ascites-derived ovarian serous carcinoma) were from ATCC and cultured in DMEM supplemented with 10% fetal bovine serum. Spheroids from transformed cells were generated by seeding cells onto ultra-low adherence plates (Corning) for 48 h. Benign cells do not form viable spheroids.

Cells were grown to 80% confluency and trypsinized before seeding. Single cells were plated at a density of 2 x 10⁴ onto individual 12mm² glass coverslips and incubated for 48 h to allow for cells to adhere and begin to grow. Coverslips were stained with 50nM MitoTracker Red CMXRos (Molecular Probes) for 15 min at 37°C, fixed with paraformaldehyde with 0.5% triton-X 100 and quenched with 50mM glycine. To study mitophagy, MOSE cells were fixed in methanol and immunostained with anti-LC3B (Cell Signaling) and with a FITC-conjugated rabbit secondary antibody (Molecular Probes). Coverslips were mounted onto glass slides using Prolong gold antifade mounting medium with DAPI (Molecular Probes) to allow for visualization of the nuclei. Images were captured with a Nikon 80/fluorescent microscope equipped with DAPI, FITC, and TRITC filters using the NIS elements BR 3.0 software and were processed using Adobe Photoshop CS6. DAPI and TRITC images were merged to display localization and spread of mitochondria from the nucleus.

The MyMia algorithm measures 25 different morphological features in each cell including both cell level measurement and branch level measurements as described previously (35). In order to separate mitochondrial features from those of the nucleus, we used color channeling to identify the differentially stained features. This structure was then used to identify branches (individual mitochondrion) and branch points (mitochondrial joints) using previously described formulas (35). Each branch is used as a mask to perform measurements such as length, width, area of pixels on each individual mitochondrion. Four different parameters were quantified and observed for about 50 images per cell type; number of branches (subtracting all the intersection from the skeleton of mitochondria network and labeling all the remaining branches), mean branch length (observed length of quantified branches per image), distance from the nucleus (mean distance between centroid of the branches and the centroid of the nuclei), and circularity (determination of how many of the objects are circular.

MOSE cells were grown to 80% confluency, trypsinized, and fixed with Karnovsky’s fixative (4g PFA/50ml, 10ml 50% glutaraldehyde in 100ml 0.2M PBS) overnight. Cells were washed 3x with 0.1M PBS for 15 min. After treatment with 1% OsO₄ in 0.1M PBS for 1 h, cells were washed 2x for 10 min with PBS. Samples were then dehydrated with increasing concentrations of graded ethanol as follows: 15%, 30%, 50%, 70%, 95%, 100% for 15 min each. Dehydration was completed using propylene oxide. After incubation with a 50:50 propylene oxide: Poly/Bed812 (Polysciences Inc.) solution for 24 h, the samples were embedded in 100% Poly/Bed 812 in flat embedding molds and placed in a 60°C oven for 48 h. Mitochondrial ultrastructure was visualized at 60 and 80x magnification on a JEOL JEM 1400 scope. Images were assembled in Adobe Photoshop™.

Adherent cells and spheroids were cultured for 48 h under normoxic and hypoxic conditions. Cells were lysed in radioimmunoprecipitation buffer supplemented with protease and phosphatase inhibitors. Protein concentrations in the lysates were quantified using the Pierce Bicinchoninic acid assay (Thermo Fisher Scientific). Equal protein concentrations were loaded in a 4% stacking and 10% SDS separating gel. Proteins were transferred onto a PVDF membrane (Bio-Rad) and blocked with 5% milk in TBST. Primary antibodies were used against DRP1, OPA1 (Novus Bio), MFN1, FIS1 (Protein Tech), UCP2 (Santa Cruz), UCP3 (ThermoFisher Scientific), TOMM 20 (Millipore), and superoxide dismutase 2 (SOD2) antibody (Cell Signaling). Proteins were normalized using the ribosomal protein L19 (L19) or to total protein normalization substrate (Thermo Fisher Scientific). Blots were subsequently probed with the appropriate HRP-conjugated mouse and rabbit or IRDye 680/800cw (Licor) secondary antibodies Proteins were visualized using chemiluminescence Pico ECL (ThermoFisher Scientific) solution with an exposure of 15–30s using the transilluminator from Bio-Rad or the Licor Odyssey Clx imaging system. Proteins quantified using ImageJ software. Data presented as mean ± SEM from at least three biological replicates.

Cells were plated at 10,000 cells per well in black 96-well microplates with glass bottom (Corning) in normoxic and hypoxic media conditions. Prior to imaging, cells were treated with 10nM tetramethyl rhodamine (TMRM) for 30 min, protected from light in a 37°C/5% CO2 incubator. Fluorescent images of mitochondrial membrane potential were obtained at 542.0 nm (27.0 nm bandpass) excitation and 587nm (45nm bandpass) emission on a GE INCell Analyzer 2200 (GE Healthcare). Sequential qualitative images were taken in 10 fields of view for each channel in each well at 37°C. Images were analyzed using GE’s InCarta software version 1.6. Multiple parameters were collected from each plate by creating custom “masks” that captured the TMRM fluorescence signal, allowing for subsequent quantification. Data are expressed as intensity (the mean pixel value under the mask)—background (mean pixel value for the local background) in Arbitrary Units (A.U.) ± SEM.

MOSE cells were seeded at 2.5 x 10⁴ cells/well and incubated in normoxic and hypoxic conditions for 24 h in flat, glass bottom 96-well microplates. The cells were washed with 0.25mM sodium phosphate solution (pH to 7.4) warmed to 37°C 30 min prior to the experiment. Cells were stained with 25µM 2’7’-dichlorofluorescin diacetate (DCFDA) (Abcam) in Krebs-ringer phosphate buffer for 45 min at 37°C to measure H2O2 levels in live cells as previously described (36, 37). ROS production was quantified using a TECAN plate reader measuring the excitation fluorescence (set at 485ex/535em). MOSE-E cells treated with 1mM H2O2 served as positive control. To measure extracellular ROS production, we used Amplex Red (ThermoFisher Scientific) at 50µM in combination with 10U/ml horseradish peroxidase, according to the manufacturer’s instructions; fluorescence was read at 571ex/585em. ROS production was normalized by the protein concentration.

Confocal imaging of spheroid mitochondria was performed as described previously (34). Briefly, spheroids grown in normoxia or hypoxia were treated with 100nM of TMRM (Molecular Probes) for 30 min to ensure penetration throughout the spheroid. After incubation, the samples were washed with PBS and plated on glass coverslips in DMEM media for imaging using a confocal Leica SP8 DMi8 microscope, at excitation/emission wavelengths of 552/576nm, respectively. Images were processed using the Leica LASX software (512x512 pixel resolution) and assembled in Adobe Photoshop™.

Cultivated MOSE-L and MOSE-L_{TIC v} spheroids were treated with 25nM MTDR. Spheroids were then placed onto glass bottom 35mm petri dishes (Cellvis) to adhere for 2 h prior to fixation with 4% PFA. Stochastic Optical Reconstruction Microscopy (STORM) imaging was conducted on a Vutara SR 350 system (Bruker) using 50mM Tris-HCl, 10mM NaCl, 10% (wt/vol) glucose buffer containing 20mM mercaptoethylamine, 1% (vol/vol) 2-mercaptoethanol, 168 active units/ml glucose oxidase, and 1,404 active units/ml catalase. Five thousand frames were acquired for each dataset and cluster analysis of assigned localizations performed using Vutara SRX software. Mitochondrial fragmentation was assessed and compared between different cell types and media conditions using the image-based cluster analysis module within the Vutara SRX software. Specifically, individual mitochondria were identified as clusters through implementation of the following parameters: density map resolution: 100px/µm, density axial resolution 5 slices/µm, minimal cluster area: 0µm³, maximum cluster area: 10,000µm³, minimum particle count: 250, particle size: 100nm, opacity: 0.30, accumulation threshold: 0.0l, half alpha shape radius: 0.10µm. STORM localization datasets were acquired for 9 separate sections of each aggregate to quantify and compare the cluster area and particle count between different cell types and media conditions. The surface area and particle count were also assessed for the top, middle, and bottom portions of each spheroid.

Data are presented as mean ± SEM. Comparisons between different cell types and oxygen content were analyzed using a one-way ANOVA followed by Tukey’s multiple comparison test. Spheroid comparisons between MOSE-L and MOSE-L_{TIC v} were analyzed using a student’s two-tailed t-test. Results were considered significant at p<0.05.

The MOSE model for progressive serous ovarian cancer (29–33) was used to initiate investigations of differential mitochondrial morphology changes during tumorigenesis in culture conditions that begin to more accurately reflect conditions in the peritoneal cavity. Figure 1A illustrates qualitatively how the elongation of the mitochondria was severely reduced during cancer progression from a filamentous phenotype observed in the MOSE-E cells, to mitochondria appear aggregated and localized mostly perinuclearly in the transformed cells with no apparent difference between the MOSE-L and MOSE-L_{TIC v} . Mitochondrial morphology changes were compared to benign (FNE1) and malignant (FNLE1) human fallopian tube cells and benign (OCE1) and malignant (SKOV3) human ovarian cells using the same culture conditions ( Figure 1B ). The human cells showed a similar loss of filamentous mitochondria after transformation as seen in MOSE cells, indicating that such changes are conserved between mice and humans and occur in both ovarian and fallopian tube epithelia. Hypoxia only minimally affected the mitochondrial morphology in both the mouse and human cells ( Figures 1A, B , right panels).

In order to quantify the alterations in mitochondrial morphology, at least 40 images of each cell type were analyzed for localization of the mitochondria, their number of branches, mean branch length, and circularity as indicators of normal morphology and distribution. Both benign mouse and human cells had a significantly higher number of mitochondrial branches with a larger mean branch length in comparison to their malignant counterparts (both p<0.05) ( Figures 1C, D ). While hypoxia did not affect these parameters in the mouse cells, the benign human cells showed a significantly reduced mitochondrial branch count and area under hypoxic conditions (p<0.001) ( Figures 1C, D ). The mitochondria were distributed throughout the cells in all benign cell types but were mostly localized in the perinuclear region in the transformed cells as indicated by the significantly smaller average distance of mitochondria to the nucleus. Mitochondrial shape was significantly more circular in the transformed cells in both the mouse and human cells except the SKOV3 line; hypoxia significantly increased circularity in both mouse and human cells except in the MOSE-L and FNE1 cells (p<0.001 for all cells, p<0.01 for SKOV3 cells). These results indicate that transformation in both human and mouse ovarian and fallopian tube epithelial cells is accompanied by morphological changes and relocation of more circular mitochondria to the perinuclear region of the cells; these morphological changes are very little affected by hypoxia. While there were some differences in the metrics between the human and the mouse cells, these may be the result of the multiple transfections of the human cell lines to achieve immortalization and transformation versus the spontaneous processes in the MOSE cells; transfection reagents and empty vector backbones have been shown to affect the expression of off-targets genes (38). Further, the rounded morphology of the human cells may have contributed to the selection of cells that allowed for mitochondrial imaging. Overall, our results show the same changes in mitochondrial shape, size, and localization in both mouse and human cells. Thus, we continued our studies with the better characterized MOSE cells having validated their relevance to human disease.

We next used transmission electron microscopy (TEM) to determine if the ultrastructure of these organelles changes during malignant progression, and if the large round mitochondria localized in the perinuclear region of the cancer cells are the result of aberrant fusion. As shown in Figure 2 , the MOSE-E show well developed cristae and cristae junctions to the inner membrane. This organization is lost during malignant progression and there were fewer and disorganized cristae in the cancer cells. These cancer mitochondria were enlarged single organelles rather than fused aggregates. While the TEM images do not allow for the visualization of connections, as apparent in Figure 1 these organelles appear not to be connected in a mitochondrial network. However, even the benign MOSE-E show some swelling, suggesting that either the immortalization process or cell culture conditions impact mitochondrial morphology of cell lines. These results indicate that with increasing malignancy, mitochondria have less structural integrity; this does not impact their viability since the cancer cells grow faster than the MOSE-E (28) but may contribute to their altered metabolic phenotype (25, 26).

To determine if the morphologic changes during cancer progression are associated with autophagy or mitophagy as a measure of mitochondrial quality control, we used indirect immunofluorescence staining for LC3B to visualize autophagosomes. As shown in Figure 3A , there were no autophagosomes detectable in normoxic benign cells but low levels of LC3B-labeled autophagosomes were observed in transformed cells. This was drastically increased under hypoxic conditions ( Figure 3A , right panels). The MOSE-E cells, however, do not alter levels of autophagy during hypoxia suggesting that these cells do not respond in the same manner to lower oxygen levels in order to enhance survival.

Changes in mitochondrial membrane potential and phenotype precedes mitophagy (39). Therefore, we next investigated how the mitochondrial membrane potential is altered during cancer progression to distinguish between autophagy and mitophagy using qualitative and quantitative TMRM detection. The transformed cells have a significantly higher mitochondrial membrane potential than the benign MOSE-E cells despite their altered morphology (p<0.0001 for MOSE-L and p<0.05 MOSE-L_{TIC v} respectively) ( Figures 3B, C ). Furthermore, hypoxia increased the membrane potential in all cells (p<0.0001 for MOSE-E, p<0.01 for MOSE-L, and p<0.05 for MOSE-L_{TIC v} ). Interestingly, this increase in membrane potential coincides with the increased levels of autophagy; while a collapsed mitochondrial membrane potential identifies damaged mitochondria and often induces mitophagy, our data indicate that in the malignant MOSE cells the membrane potential is increased and enhances autophagy to promote cell survival as has been observed in breast cancer cells (40).

The shift in the balance of general ROS production and elimination has been suggested to increase ROS steady-state levels in cancer cells that is counter-balanced by a higher antioxidant capacity (41). Since ROS production is often associated with high mitochondrial membrane potential and induction of autophagy, we measured intracellular and extracellular ROS production. There was no detectable difference in intra- or extracellular ROS levels between the MOSE-E cells and the MOSE-L; however, the highly tumorigenic MOSE-L_{TIC v} showed significantly lower ROS levels (P<0.05 vs. MOSE-E and MOSE-L). Hypoxia significantly increased ROS levels in all three cell types (P<0.0001) with the MOSE-L_{TIC v} still having the lowest amounts ( Figure 3D ). To determine if hypoxic ROS accumulation is indicative of limited scavenging capacity, protein expression of SOD2 was observed by Western blotting. Neither cell line increases their SOD2 expression in hypoxia, thus explaining the increased ROS accumulation. The MOSE-L_TICv, cells, however, show consistent levels of SOD2. This indicates that either the MOSE-L_{TIC v} cells reduce ROS production or the accumulated ROS were scavenged in an alternative way to limit redox potential and prevent apoptosis. These results suggest that in contrast to the MOSE-L cells, the most aggressive MOSE-L_{TIC v} may be able to control their ROS content through selective removal of damaged mitochondria via autophagic degradation to support their viability in the hypoxic peritoneal cavity.

Aggregation is a survival signal for cancer cells and increases their metastatic potential (3, 4). To investigate how aggregation affects mitochondrial morphology, we generated spheroids from MOSE-L_{TIC v} cells since these cells are most responsive to hypoxia ( Figure 2 ) and nutrient deprivation (25). The cells were stained with the mitochondria-specific dye, TMRM that permeates throughout the spheroid ( Figure 4A and Supplementary Video 1 ). Using confocal microscopy, we show that the mitochondria are fragmented in the spheroid core; the higher fluorescence levels in the outer layer of the spheroids indicate that mitochondrial fragmentation is less frequent in areas where mitotic features can be observed (42). The fragmentation of mitochondria in the spheroid core was further enhanced by hypoxia, suggesting an adaptation of the mitochondrial morphology to differential oxygen levels.

To investigate whether the altered mitochondrial morphology was the result of changes in the expression of proteins regulating fusion and fission dynamics, we determined the expression levels of MFN1 and OPA1 (fusion) and FIS1 and DRP1 (fission). As evident by the levels of TOMM 20 (translocase of outer mitochondrial membrane 20), there was a significant increase in total mitochondrial protein content during malignant transition which was significantly reduced under hypoxic conditions in the transformed but not in the benign cells. As shown in Figures 4B, C , all observed fusion and fission proteins were lowered in the MOSE-L and MOSE-L_{TIC v} cells in comparison to the MOSE-E cells in hypoxic conditions (normalized to L19) but only MFN1 and DRP1 proteins were increased in the MOSE-E cells in normoxic conditions (0<0.05). Hypoxia had a nominal effect on protein expression in the cancer cells but significantly increased both fusion and fission proteins (MFN1 and DRP1, respectively) in the benign MOSE-E cells (p<0.05). However, with increasing malignancy, the ratio of fission to fusion proteins increased in the adherent cells, specifically the DRP1:MFN1 ratio ( Figure 4D ) which could contribute to their increasing fragmented phenotype illustrated in Figure 1 . Interestingly, aggregation did not exacerbate these ratios and in the most aggressive MOSE-L_{TIC v v} spheroids despite the changes in mitochondrial morphology, the ratios more closely mimicked those observed in the benign MOSE-E. Fission and fusion proteins were significantly lower after aggregation of MOSE-L (MFN1, DRP1, and FIS1: p<0.05 in hypoxia, MFN1: p<0.05 in normoxia), and MOSE-L_{TIC v} (MFN1, OPA1, DRP1, and FIS1: p<0.05 in hypoxia, MFN1: p<0.05 in normoxia. However, there was an increase in OPA1 proteins in comparison to DRP1 and FIS1 in the MOSE-L and MOSE-L_{TIC v} spheroids.

Uncoupling proteins regulate cellular metabolism and ROS production (43) and are overexpressed in several cancers (44). UCP2 and UCP3 protein expression increased during ovarian cancer progression and after aggregation in comparison to the MOSE-E with a significant increase in UCP3 for the adherent MOSE-L and MOSE-L_{TIC v} spheroids (p<0.05). Both proteins were further increased in hypoxic conditions especially in the MOSE-L_{TIC v} aggregates. Taken together, these data indicate that a shift in the balance of regulatory protein expression rather than an increase in fission proteins contributes to mitochondrial fragmentation in aggressive cancers. 

To quantify the fragmented mitochondrial phenotype in spheroids, MOSE-L and MOSE-L_{TIC v} cells were incubated with MitoTracker deep red (MTDR) prior to aggregation to label mitochondria in all cells. Representative confocal images were taken to confirm MTDR staining prolonged the 48hr incubation prior to fixation ( Figure 5A ). Stochastic optical reconstruction microscopy (STORM) was employed to identify mitochondrial morphology at nanoscale resolution. Individual clusters were selected in the super-resolution localizations, identifying individual mitochondria within the spheroid core ( Figure 5B ). Volume and particle count were quantified using the Vutara SRX software cluster analysis to determine fragmentation levels. As shown in Figure 5C , mitochondrial volume and particle counts are significantly lower in MOSE-L_{TIC v} after aggregation than in MOSE-L spheroids (p<0.01 in normoxic and p<0.001 in hypoxic conditions); while hypoxia reduced both parameters even further this was not statistically significant. These data suggest that the more aggressive MOSE-L_{TIC v} phenotype is associated with the ability to respond to aggregation and hypoxia with mitochondrial fragmentation.

To determine how mitochondrial morphology differs at specific regions of the cultured spheroids, and if there is a difference between the slow and fast-developing disease, STORM localization was used to image mitochondria in the top, middle, and bottom portions of MOSE-L and MOSE-L_{TIC v} spheroids in normoxic and hypoxic conditions. Individual mitochondria were identified as separate clusters using the same parameters as described above. As shown in Figures 6A–C , while there were fewer and smaller mitochondria particles in the core of MOSE-L spheroids than on the outer layer, data were variable and, thus, not statistically significant. This did not change in hypoxic conditions. However, MOSE-L_{TIC v} spheroids exhibited a lower number of mitochondrial particles with no detectable differences between the core and the top or bottom layers but a significant reduction of both particle count and surface area in hypoxic conditions (also significantly lower than MOSE-L core area, p<0.05). These results suggest that the more aggressive MOSE-L_{TIC v} spheroids reduce mitochondrial number and size in response to aggregation and reduced oxygen availability that may contribute to the more flexible metabolic phenotype we have described previously (25, 26).