Purpose: The goal of this study was to understand the role of altered mitochondrial function in breast cancer progression and determine the potential of the molecular alteration signature in developing exosome-based biomarkers.

Experimental Design: This study was designed to characterize the critical components regulating mitochondrial function in breast tumorigenesis. Experiments were conducted to assess the potential of these molecules for exosome-based biomarker development.

Results: We observed a remarkable reduction in spontaneous metastases through the interplay in mitochondria by SH3GL2, vesicular endocytosis–associated protein and MFN2, an important regulator of mitochondrial fusion. Following its overexpression in breast cancer cells, SH3GL2 translocated to mitochondria and induced the production of superoxide and release of cytochrome C from mitochondria to the cytoplasm. These molecular changes were accompanied by decreased lung and liver metastases and primary tumor growth. SH3GL2 depletion reversed the above phenotypic and associated molecular changes in nontumorigenic and tumorigenic breast epithelial cells. Loss of SH3GL2 and MFN2 expression was evident in primary human breast cancer tissues and their positive lymph nodes, which was associated with disease progression. SH3GL2 and MFN2 expression was detected in sera exosomes of normal healthy women, but barely detectable in the majority of the women with breast cancer exhibiting SH3GL2 and MFN2 loss in their primary tumors.

Conclusions: This study identified a new mitochondria reprogramming pathway influencing breast cancer progression through SH3GL2 and MFN2. These proteins were frequently lost in breast cancer, which was traceable in the circulating exosomes. Clin Cancer Res; 22(13); 3348–60. ©2016 AACR.

This article is featured in Highlights of This Issue, p. 3119

Translational Relevance

Developing strategies for molecular early detection, monitoring, or surveillance of breast cancer may reduce disease-associated morbidity. In this study, we identified SH3GL2 and MFN2 as potential breast cancer suppressors. Frequent loss of SH3GL2 and MFN2 was observed in primary and metastatic breast cancer tissues, and their loss was found to be associated with disease progression. Functionally, their loss appeared to be associated with mitochondrial reprogramming favoring tumorigenic progression. The simultaneous loss of SH3GL2 and MFN2 was measured through the circulating exosome profiling in body fluids of women with invasive breast cancer. Measuring SH3GL2 and MFN2 expression in premalignant lesions or high-risk individuals and their timely profiling in circulating exosomes may improve current strategies for early detection, monitoring, and surveillance of breast cancer.

Breast cancer represents 14.0% of all new cancer cases and is the second most common cause of cancer-associated morbidity among the U.S. women (1, 2). In 2015, there will be an estimated 231,840 cases and 40,290 deaths (2). Being highly heterogeneous and metastatic, breast cancer poses significant challenges to clinical management (1, 3). Early breast cancer detection has a better chance of cure or prolonged disease-free survival compared with the metastatic disease (4). Although at least one progression model of normal tissue to invasive cancer has been proposed using cell morphology (5), the molecular drivers behind the initiation and stage-wise progression of breast cancer are not well characterized.

Continuous proliferation and apoptosis resistance are hallmarks of cancer cells (6). Abnormal mitochondrial function and reprogramming contribute to these hallmarks at least in part and hence are implicated in biomarker development (6, 7). Mitochondrial fusion is a process of fusion of damaged mitochondria to healthy ones (6, 8). Studies suggest that production in tumors of normal mitochondria could be tumor suppressive by promoting oxidative metabolism and enhanced reactive oxygen species (ROS) production (8). On the other hand, mitochondrial biogenesis is a process involving replication of the mitochondrial genome and coordinated expression of both nuclear and mitochondria-encoded molecules and assembly of the oxidative phosphorylation complexes (6, 8, 9). Many factors, including MFN2, PINK1, PGC-1α, and mitochondrial transcription factor A (MT-TFA) play critical role in regulating mitochondrial fusion, biogenesis, and maintaining mitochondrial integrity (6, 8, 9). The role of mitochondrial fusion and biogenesis in breast cancer development and progression remains largely unknown.

Exosomes are 50 to 200 nm, small secreted endocytic vesicles present in all cell types and body fluids (10–12). Cancer exosomes (CE) carry survival information in the form of nucleic acids and proteins, shuttle constantly between the cancer cells through the circulation, and influence growth and progression (10–12). Characterizing the CEs and deciphering the cancer-promoting information that they carry have tremendous potential for biomarker and therapeutic development.

We identified SH3GL2, a vesicular endocytosis–associated protein (13) as a potential breast cancer suppressor. When overexpressed in multiple human breast cancer cells, SH3GL2 translocated to mitochondria and became phosphorylated. This was accompanied by enhanced mitochondrial fusion and expression of mitochondrial fusion and biogenesis-associated proteins MFN2, PINK1, and PGC-1α. A concomitant increase in superoxide (O2) production and release of cytochrome C (CYTC) from mitochondria to the cytoplasm was also evident. An elevated level of growth-regulatory molecules PTEN, E-cadherin (CDH1), and ATG5 was also evident following SH3GL2 overexpression. These molecular changes were associated with reduced growth and progression of the breast cancer cells in vitro and in vivo. The orthotopically implanted SH3GL2-overexpressing xenografts exhibited appreciable epithelial features, increased cell–cell adherens, and reduced metastatic ability to the lung and liver. These cellular and molecular changes were reversed following SH3GL2 silencing in nontumorigenic and tumorigenic breast epithelial cells. We also observed colocalization and coimmunoprecipitation of SH3GL2 with MFN2, PINK1, and PGC-1α in mitochondria. Frequent SH3GL2 and MFN2 loss in primary tumor and their association with breast cancer progression was also evident. Compared with normal, the loss of the SH3GL2 and MFN2 expression was detected in the sera exosomes of breast cancer patients.

Human tissue samples and ethical statement

Formalin-fixed paraffin-embedded (FFPE)–deidentified primary breast cancer, matched lymph nodes, normal tissues along with mammoplasty tissues were collected from the Department of Pathology, The University of Texas Health Science Center at Tyler (UTHSCT; Tyler, TX). Serum samples from healthy women or women with breast cancer were also collected from the Cooperative Human Tissue Network. All tissues and serum samples were collected under an IRB-approved protocol. Relevant demographic data were collected for necessary clinical correlation analysis. The demographic data of all the patients along with the expression patterns of various altered molecules are represented in Supplementary Table S1A.

Cell culture

Authenticated MDA-MB-231, MCF-7, SUM-149, and MCF-10A cells were purchased from ATCC and other suitable vendors and cultured as recommended. The HMLE cell line was kindly provided by Dr. Guojun Wu, Wayne State University (Detroit, MI). All cell lines were periodically checked for mycoplasma contamination using a Mycoplasma Detection Kit (Sigma # MP-0025; refs. 14, 15). All tissue culture media and reagents were purchased either from ATCC or Invitrogen.

Antibodies and reagents

SH3GL2 antibody was obtained from Novus Biologicals (# NBP1-8552). The E-cadherin (#3195P), PTEN (#9559S), P-threonine (#9381S), ATG5 (#2630S), LC3B (#2775S), β-actin (#3700), and ZO-1 (#8193) antibodies were purchased from Cell Signaling Technology. The MFN2 (#ab56889), MT-TFA (#ab119684), PGC-1α (#ab54481), IMMT (#ab110329), cytochrome C (#ab90529), and anti-mitochondria (MT-CO2, #ab3298) antibodies were obtained from Abcam Inc. The PINK1 antibody (#LS-B3384) was purchased from LS Bioscience Inc. F-Actin antibody (#A12380) was obtained from Invitrogen Inc. Anti-mouse (#115-035-003) and rabbit (#111-035-003) secondary antibodies were obtained from Jackson ImmunoResearch. Anti-mouse (#A11004) or anti-rabbit (#A11011) Alexa Fluor 568 and anti-mouse (#11029) or anti-rabbit (#11008) Alexa Fluor 488 secondary antibodies were purchased from Invitrogen. MitoTracker Red (#M224250) was obtained from Invitrogen. Far-red secondary antibody (#A31573) was obtained from Thermo Fisher Scientific.

Lentiviral transduction of SH3GL2

MCF-7, SUM-149, and MDA-MB-231 cells were transduced with GFP-tagged lentivirus construct encoding SH3GL2 (#LVP303171, Applied Biological Materials). A GFP-tagged empty lentivirus construct with the same backbone (#LVP590) was used as a control. Stable clones were selected in the presence of puromycin (10 μg/mL). A single stable clone was expanded and utilized for all subsequent analyses (14, 15).

In the knockdown studies, MCF-10A and MCF-7 cells were transduced with a GFP-tagged lentivirus SH3GL2-SiRNA pool (#iV022230, Applied Biological Materials). The same lentivirus construct harboring scrambled siRNA was used as a control (#LVP015). Stable clones were selected in the presence of puromycin (1 μg/mL). A single stable clone was expanded and utilized for all subsequent analyses (14, 15). In both gain and loss-of-function studies, naïve control cells were used to examine the influence of the empty vector on SH3GL2 expression.

IHC and immunofluorescence

IHC and immunofluorescence (IF) analyses were performed as described previously (14, 15). The IF Images were captured through a Leica TCS SP8 laser-scanning confocal imaging station (Leica Microsystems). Quantitative imaging analysis to assess the colocalization of the SH3GL2, CYTC, PGC-Iα, and MTCO2 was performed using “Intensity Correlation Analysis” function in Leica confocal software. The overlap coefficients generated by Pearson correlation coefficient have values between +1 and −1 (+1 and −1 values indicate that 100% and 0% of both components of the two images overlap, respectively).

Cell proliferation, invasion, and soft agar colony formation assays

Proliferation of the transduced cells (triplicate) was determined by a BrdU incorporation assay (#B23151, Life Technologies; refs. 14, 15). Soft agar colony formation and invasion assays (#354481, Corning) were performed as described earlier (14, 15). In all cases, data were presented as mean ± SE of duplicate experiments.

Isolation of mitochondria and determination of O2 production

Mitochondria were isolated from cultured cells and tissues using kits and protocols from Pierce (#89847 and #89801). O2 production was measured using MitoSOX Red reagent (#M36008, Invitrogen).

Western blotting and coimmunoprecipitation analysis

Preparation of whole-cell or mitochondrial lysates, Western blotting analysis, and immunoprecipitation were performed following protocols described earlier (14, 15).

Docking analysis of SH3GL2, MFN2, PINK1, and PGC-1α

The complete structure of SH3GL2, MFN2, PINK1, and PGC-1α was predicted using I-TASSER server (http://zhanglab.ccmb.med.umich.edu/I-TASSER/), and appropriate models were selected to perform the docking analysis on the ZDOCK server (http://zdock.umassmed.edu/).

Mitochondrial DNA content analysis

Genomic DNA was isolated from the transduced cell lines followed by qPCR analysis using primers from mitochondria-encoded ND4 (MT-ND4) and nuclear-encoded GAPDH. The primer sequences are as follows: MT-ND4 (gene ID: 4538): forward (F)-ACTCACAACACCCTAGGCTC; reverse (R)-GCTTCGACATGGGCTTTAGG. GAPDH (gene ID: 2597): F-TCCTCCACCTTTGACGCTG; R-ACCACCCTGTTGCTGTAGCC. A ratio of mitochondrial DNA/nDNA (MT-ND-4/GAPDH) was used to determine the fold change among various groups.

In vivo xenograft and metastasis analyses

For tumor growth, 1 × 105 SH3GL2 and empty vector–transduced cells in 1:1 mixture of PBS and Matrigel were injected in the mammary fat pad of 4- to 6-week-old, female NSG mice (Charles River Laboratories; ref. 16). All experiments were performed in accordance with the IACUC guidelines. Each group consisted of 8 mice. Mice were examined every day, and mice showing any sign of morbidity were immediately sacrificed according to the University guidelines. All experiments were terminated at week 5 due to the tumor burden. After 5 weeks, mice were sacrificed, and tumor weights were taken. Lungs and livers were removed for the analysis of metastasis. Focal tumor nodules were counted in the lung and liver of all the mice from various groups. Tumors were processed for histologic, Western blotting, and immunohistochemical analyses. All histopathologic and IHC evaluations of the in vivo tumors were done per pathologic guidance (14, 15). Data were presented as mean ± SE of duplicate experiments.

Exosome preparation from human sera and established culture

Exosomes were isolated from human sera or culture supernatant using commercially available kits and protocols (#EXOQ5A-1 and # EXOTC10A-1, System Bioscience), followed by protein isolation. Exosomes were treated with proteinase K before protein isolation (12).Western blotting analysis was performed using 40 μg of total exosome protein to detect SH3GL2 and MFN2 expression. Syntenin was used as an exosome marker and loading control (14, 15).

Statistical analysis

χ2, Fisher exact, or Student t tests were used when appropriate. All P values were two sided, and all confidence intervals were at the 95% level. Computation for all the analyses was performed using the Statistical Analysis System.

SH3GL2 translocates to mitochondria, stimulates mitochondrial apoptosis, and inhibits breast cancer progression

A single study, so far, identified allelic loss of SH3GL2 in breast cancer tissues (17). However, the precise role of SH3GL2 in breast cancer has not yet been evaluated. We stably overexpressed SH3GL2 in three breast cancer cell lines MCF-7, SUM-149, and MDA-MB-231 (Supplementary Fig. S1A). The introduction of SH3GL2 reduced proliferation (P = 0.001–0.004), anchorage-independent growth (P = 0.0002–0.0016), and invasion (P = 0.0003–0.0006) of these breast cancer cells compared with the control (Supplementary Fig. S1B–S1D). The SH3GL2-overexpressing cells also produced high level of O2 (P = 0.0001–0.0003; Fig. 1A). This was accompanied by an enhanced cytoplasm/mitochondria ratio of CYTC expression (Fig. 1B). Release of CYTC from mitochondria to the cytoplasm is the central step in apoptosis (18). These results suggest for a possible interplay between SH3GL2 and mitochondria for triggering downstream apoptotic signaling. In the SH3GL2 overexpressing cells, we observed an increased distribution of mitochondrial fusion bodies (Fig. 1C), accompanied by enhanced expression of MFN2, PINK1, and MT-TFA (Fig. 1D). As MFN2, PINK1, and MT-TFA predominantly function in mitochondria (6, 8), it is likely that SH3GL2 cooperates with them in the mitochondria. We observed SH3GL2 expression and associated increase in MFN2, PINK1, and MT-TFA expression in mitochondria (Fig. 1E). We did not observe any influence of the empty vector on SH3GL2 or MFN2 expression in these cells (Supplementary Fig. S1E). The SH3GL2 construct has a GFP tag, and IF analysis of the GFP-SH3GL2–expressing MDA-MB-231 cells labeled with MitoTracker Red confirmed GFP-SH3GL2 and mitochondrial colocalization (Fig. 1F). We also observed colocalization of the endogenous SH3GL2 and mitochondria in the naïve MDA-MB-231 cells (Fig. 1G). The overlap coefficient value between SH3GL2 and mitochondrial marker CYTC (Fig. 1G) was 0.8936 (89% colocalization).

Figure 1.

SH3GL2 translocation to mitochondria induces ROS production and modulates mitochondria function. A, SH3GL2-overexpressing breast cancer cells produced significantly higher amount of superoxides (P = 0.0001–0.0003), exhibited an enhanced expression of CYTC in the cytoplasm compared with mitochondria (B), and demonstrated increased numbers of mitochondrial fusion bodies compared with the empty vector–treated cells (C). D and E, enhanced expression of MFN2, PINK1, and MT-TFA in the whole-cell or mitochondrial lysate of the SH3GL2-transduced groups compared with the empty vector–treated groups. EV, empty vector transduced; SH3GL2, SH3GL2-transduced cells. β-Actin and IMMT were used as cytosolic and mitochondrial loading controls. F, confocal imaging demonstrating colocalization of GFP-tagged SH3GL2 and MitoTracker Red–labeled mitochondria in MDA-MB-231 cells. Scale bar, 50 μm; magnification, 200× (A) and 400× (C and F). G, confocal imaging showing colocalization of endogenous SH3GL2 and mitochondria in the naïve MDA-MB-231 cells (arrowheads). Scale bar, 20 μm; magnification, 400×.

Figure 1.

SH3GL2 translocation to mitochondria induces ROS production and modulates mitochondria function. A, SH3GL2-overexpressing breast cancer cells produced significantly higher amount of superoxides (P = 0.0001–0.0003), exhibited an enhanced expression of CYTC in the cytoplasm compared with mitochondria (B), and demonstrated increased numbers of mitochondrial fusion bodies compared with the empty vector–treated cells (C). D and E, enhanced expression of MFN2, PINK1, and MT-TFA in the whole-cell or mitochondrial lysate of the SH3GL2-transduced groups compared with the empty vector–treated groups. EV, empty vector transduced; SH3GL2, SH3GL2-transduced cells. β-Actin and IMMT were used as cytosolic and mitochondrial loading controls. F, confocal imaging demonstrating colocalization of GFP-tagged SH3GL2 and MitoTracker Red–labeled mitochondria in MDA-MB-231 cells. Scale bar, 50 μm; magnification, 200× (A) and 400× (C and F). G, confocal imaging showing colocalization of endogenous SH3GL2 and mitochondria in the naïve MDA-MB-231 cells (arrowheads). Scale bar, 20 μm; magnification, 400×.

Close modal

In the SH3GL2-overexpressing cells, we also observed an enhanced expression of PINK1 regulator PTEN (Supplementary Fig. S2A; ref. 19). The breast cancer lines that we used poorly express CDH1, a regulator of cellular growth and metastasis (20, 21). We observed appreciable reversal of CDH1 expression following SH3GL2 introduction in these cells (Supplementary Fig. S2A). We also examined the expression of ATG5 and LC3B, proteins associated with mitophagy and apoptosis in these cells (22–25). An enhanced expression of ATG5, but not LC3B, except for LC3B-I in MDA-MB-231 cells, was observed (Supplementary Fig. S2B).

SH3GL2 was physically associated with various proteins in mitochondria

The above mentioned results suggest that SH3GL2 might physically be associated with MFN2 and PINK1 in mitochondria. The IF analyses confirmed colocalization of SH3GL2 and MFN2, as well as SH3GL2 and PINK1, in these cells (Supplementary Fig. S2C–S2D). On the basis of the docking analysis, the transmembrane domain and near coiled–coiled region of MFN2 appears to interact with the SH3 domain of the SH3GL2 protein (Fig. 2A and Supplementary Fig. S2E; Supplementary Table S1B). On the other hand, the amino acid residues 34–41 and 255–263 of PINK1 seem to participate in the association with SH3GL2 (Fig. 2B and Supplementary Fig. S2F; Supplementary Table S1C). We then performed coimmunoprecipitation analysis using lysates prepared from intact mitochondria, assuming that they were physically associated in mitochondria. We could pull down MFN2 or PINK1 with SH3GL2 and vice versa in mitochondria (Fig. 2C and D).

Figure 2.

SH3GL2 is physically associated with MFN2, PINK1, and PGC-1α. A and B, docking analysis predicted several interacting sites between SH3GL2, MFN2, and PINK1 (arrows). C and D, SH3GL2 was coimmunoprecipitated with MFN2 and PINK1 and vice versa using total proteins prepared from intact mitochondria. IgG was used as a control. E and F, enhanced PGC-1α expression in the whole-cell or mitochondrial lysate of the SH3GL2-overexpressing breast cancer cells compared with the control. EV, empty vector; SH3GL2, SH3GL2-transduced cells. β-Actin and IMMT were used as cytosolic and mitochondrial loading controls. G, colocalization of PGC-1α (cyan), SH3GL2 (green), and mitochondria (red) in the SH3GL2-overexpressing MDA-MB-231 cells (arrowheads). Scale bar, 20 μm; magnification, 400×. H, SH3GL2 and PGC-1α docking complex (arrows) as predicted by bioinformatic analysis. I, coimmunoprecipitation of SH3GL2 and PGC-1α and vice versa in mitochondria of the SH3GL2-overexpressing MDA-MB-231 cells. IgG was used as a control. IP, immunoprecipitation; IB, immunoblotting.

Figure 2.

SH3GL2 is physically associated with MFN2, PINK1, and PGC-1α. A and B, docking analysis predicted several interacting sites between SH3GL2, MFN2, and PINK1 (arrows). C and D, SH3GL2 was coimmunoprecipitated with MFN2 and PINK1 and vice versa using total proteins prepared from intact mitochondria. IgG was used as a control. E and F, enhanced PGC-1α expression in the whole-cell or mitochondrial lysate of the SH3GL2-overexpressing breast cancer cells compared with the control. EV, empty vector; SH3GL2, SH3GL2-transduced cells. β-Actin and IMMT were used as cytosolic and mitochondrial loading controls. G, colocalization of PGC-1α (cyan), SH3GL2 (green), and mitochondria (red) in the SH3GL2-overexpressing MDA-MB-231 cells (arrowheads). Scale bar, 20 μm; magnification, 400×. H, SH3GL2 and PGC-1α docking complex (arrows) as predicted by bioinformatic analysis. I, coimmunoprecipitation of SH3GL2 and PGC-1α and vice versa in mitochondria of the SH3GL2-overexpressing MDA-MB-231 cells. IgG was used as a control. IP, immunoprecipitation; IB, immunoblotting.

Close modal

Other than MT-TFA, PGC-1α also plays a key role in mitochondrial biogenesis (26). We observed enhanced expression of PGC-1α in the SH3GL2-overexpressing cells (Fig. 2E). Augmented expression of PGC-1α was confirmed in mitochondria (Fig. 2F). This observation suggests that SH3GL2 may interplay with PGC-1α in mitochondria as well. The confocal imaging analysis using a mitochondria-specific marker (MTCO2) further demonstrated their colocalization in the mitochondria of the SH3GL2-overexpressing MDA-MB-231 cells (Fig. 2G). The values of overlap coefficient between SH3GL2 and MTCO2 were 0.8782; PGC-Iα and MTCO2: 0.6558; and SH3GL2 and PGC-1α: 0.6603, which confirmed their colocalization. On the other hand, docking analysis predicted two interacting regions, namely amino acid 190–204 and 627–637 in PGC-1α for SH3GL2 and PGC-1α interaction (Fig. 2H; Supplementary Table S1D). Through coimmunoprecipitation and Western blotting, we could pull down SH3GL2 with PGC-1α and vice versa in the mitochondria of the SH3GL2-overexpressing MDA-MB-231 cells (Fig. 2I).

The above results raised our interest to determine whether SH3GL2 is phosphorylated in mitochondria. Bioinformatic analysis predicted several serine, threonine, and tyrosine phosphorylation sites in the BAR and SH3 domains of SH3GL2 (Supplementary Fig. S3A). The phosphorylation potential appeared to be higher for serine and threonine residues (Supplementary Fig. S3B). Several serine and threonine residues are conserved among different species (Supplementary Fig. S3C and S3D). Next, we performed Western blotting analysis with an anti-phosphothreonine antibody using SH3GL2 protein immunoprecipitated from whole cells or mitochondrial lysates of SH3GL2-overexpressing cells. Phosphothreonine-positive SH3GL2 was detected in the total and mitochondrial lysates prepared from the SH3GL2-transduced cells (Fig. 3A). To compare SH3GL2 phosphorylation specifically in mitochondria of both empty vector and SH3GL2-overexpressing cells, we performed Western blotting analysis with the same anti-phosphothreonine antibody. Phosphothreonine-positive SH3GL2 was detectable in mitochondria of the SH3GL2-overexpressing cells but barely detectable, or undetectable, in the empty vector–treated cells (Fig. 3B).

Figure 3.

SH3GL2 is phosphorylated in mitochondria, and its silencing invokes growth and progression of nontumorigenic breast epithelial cell. A, detection of phosphorylated SH3GL2 in whole-cell or mitochondrial lysate of SH3GL2-overexpressing cells using a P-threonine antibody. B, enhanced P-threonine–positive SH3GL2 signal detection in mitochondria of the SH3GL2-overexpressing cells, which was low or barely detected in the empty vector–treated groups. EV, empty vector; GL2, SH3GL2 transduced. C, silencing of SH3GL2 in nontumorigenic breast epithelial MCF-10A cells markedly reduced the expression of MFN2, PINK1, CDH1, and PTEN. D, mitochondria/cytoplasm ratio of CYTC was higher in the SH3GL2-depleted cells compared with the empty vector–treated cells. E–G, increased proliferation (P = 0.0007), anchorage-independent growth (P = 0.0003), and invasion (P = 0.0003) of the SH3GL2-depleted MCF-10A cells compared with the empty vector–treated control. H, low O2 production (P = 0.0003) by the SH3GL2-depleted cells compared with the control scrambled SiRNA-treated cells. C, control SiRNA–treated cells; KD, SH3GL2-specific SiRNA-treated cells. β-Actin and IMMT were used as cytoplasmic and mitochondrial loading controls, respectively. Scale bar, 50 μm; magnification, 200×. IP, immunoprecipitation; IB, immunoblotting; BrdU, bromodeoxyuridine.

Figure 3.

SH3GL2 is phosphorylated in mitochondria, and its silencing invokes growth and progression of nontumorigenic breast epithelial cell. A, detection of phosphorylated SH3GL2 in whole-cell or mitochondrial lysate of SH3GL2-overexpressing cells using a P-threonine antibody. B, enhanced P-threonine–positive SH3GL2 signal detection in mitochondria of the SH3GL2-overexpressing cells, which was low or barely detected in the empty vector–treated groups. EV, empty vector; GL2, SH3GL2 transduced. C, silencing of SH3GL2 in nontumorigenic breast epithelial MCF-10A cells markedly reduced the expression of MFN2, PINK1, CDH1, and PTEN. D, mitochondria/cytoplasm ratio of CYTC was higher in the SH3GL2-depleted cells compared with the empty vector–treated cells. E–G, increased proliferation (P = 0.0007), anchorage-independent growth (P = 0.0003), and invasion (P = 0.0003) of the SH3GL2-depleted MCF-10A cells compared with the empty vector–treated control. H, low O2 production (P = 0.0003) by the SH3GL2-depleted cells compared with the control scrambled SiRNA-treated cells. C, control SiRNA–treated cells; KD, SH3GL2-specific SiRNA-treated cells. β-Actin and IMMT were used as cytoplasmic and mitochondrial loading controls, respectively. Scale bar, 50 μm; magnification, 200×. IP, immunoprecipitation; IB, immunoblotting; BrdU, bromodeoxyuridine.

Close modal

Depletion of SH3GL2 promotes progression of nontumorigenic breast epithelial cells

To assess whether altered SH3GL2 expression was necessary to prevent growth and progression, we stably silenced SH3GL2 in nontumorigenic breast epithelial cells MCF-10A (Fig. 3C). Appreciable depletion of SH3GL2 in these cells markedly reduced the expression of MFN2, PINK1, PTEN, and CDH1 (Fig. 3C) molecules, which were induced following SH3GL2 overexpression in the breast cancer cell (Fig. 1 and Supplementary Fig. S2A). These cells exhibited a high mitochondria/cytoplasm ratio of CYTC expression (Fig. 3D). This was accompanied by a marked increase in cellular proliferation (P = 0.0007), anchorage-independent growth (P = 0.0003), and invasion (P = 0.0003; Fig. 3E–G). Moreover, reduction (P = 0.0003) in O2 production was also evident in the SH3GL2-depleted MCF-10A cells (Fig. 3H). We also silenced SH3GL2 in tumorigenic MCF-7 cells. Similar to MCF-10A cells, SH3GL2-silenced MCF-7 cells demonstrated a decreased expression of MFN2, PINK1, and CDH1, accompanied by a higher mitochondria/cytoplasm ratio of CYTC expression (Supplementary Fig. S4A and S4B). An increase in proliferation (P = 0.0001), invasion (P = 0.0002), and concomitant decrease in O2 production (P = 0.0001) were also noted in these cells (Supplementary Fig. S4C–S4E). The empty vector had negligible influence on SH3GL2 or MFN2 expression in MCF-10A cells (Supplementary Fig. S4F).

Mitochondrial SH3GL2 halted primary tumor growth and distant metastases in vivo

We next evaluated the impact of SH3GL2 overexpression on in vivo growth and progression using an orthotopic breast cancer model (16). SH3GL2 or empty vector–transduced MDA-MB-231 cells were implanted in the mammary fat pad of 4- to 6-week-old female NSG mice (N = 8/group). Mice were sacrificed at week 5 due to the tumor burden, and average tumor weights were taken. Mean tumor weight was lower (P = 0.001) in the SH3GL2-overexpressing group compared with the control (Fig. 4A). The GFP-SH3GL2–overexpressing breast cancer cells appeared more epithelial and differentiated with notable cell–cell adherence (Fig. 4B).The control GFP-empty vector–expressing cells appeared more mesenchymal with disrupted cell–cell contact representing the poorly differentiated features of the parental MDA-MB-231 cells (Fig 4B). The expression and distribution pattern of ZO-1, CDH1, and F-actin proteins in these tissues further support this observation (Fig. 4C). Due to the restoration of epithelial phenotype following SH3GL2 overexpression, the fluorescence signals of cortical actin were also enhanced in the SH3GL2-transduced cells (Fig. 4C). The extent of lung and liver metastases of the SH3GL2-overexpressing MDA-MB-231 cells also decreased. The number of visible lung tumor nodules was lower (P = 0.0001) in the SH3GL2-overexpressing group compared with the control (Fig. 4D). Histologic analysis of lung tissues from multiple mice revealed numerous large macroscopic tumor foci in the lung of the control group (Fig. 4E). However, in the SH3GL2-overexpressing group, small and relatively few micrometastases were noted (Fig. 4E). Macroscopic tumor nodules and histologic metastases of the MDA-MB-231 cells were noted in the liver of the control mice (Fig. 4F and G). However, no visible tumor nodule or metastatic infiltration was observed histologically in the liver of the SH3GL2-overexpressing group (Fig. 4F and G).

Figure 4.

SH3GL2 overexpression reduced primary growth and metastases of MDA-MB-231 cells in vivo. A, mean tumor weight was significantly lower (P = 0.001) in the SH3GL2-overexpressing mice group compared with the control group. Empty vector, cells transduced with the empty vector; SH3GL2, cells transduced with SH3GL2. B, epithelial-like appearance of the GFP-SH3GL2–overexpressing MDA-MB-231 mammary implants, with considerable cell–cell contact (arrowheads) compared with the GFP-empty vector–treated group exhibiting pronounced disruption of the epithelial morphology and cell–cell adhesion. C, IF analysis of the SH3GL2-overexpressing xenografts with ZO-1, CDH1, and F-actin demonstrates pronounced epithelial-like appearance and cell–cell adhesion (arrowheads) compared with the control group. Scale bar, 20 μm; magnification, 400×. D and E, the number of visible lung tumor nodules (encircled and scatter plot) and histologic macrometastases (arrows) were markedly lower (P = 0.0001) in the SH3GL2-overexpressing group compared with the empty vector–treated group. F and G, visible tumor nodules and histologic macrometastases (encircled and scatter plot) were detected only in the liver of the empty vector–treated group, but not in the SH3GL2-overexpressing group at 5 weeks. Scale bar, 50 μm; magnification, 200× (B, D–G).

Figure 4.

SH3GL2 overexpression reduced primary growth and metastases of MDA-MB-231 cells in vivo. A, mean tumor weight was significantly lower (P = 0.001) in the SH3GL2-overexpressing mice group compared with the control group. Empty vector, cells transduced with the empty vector; SH3GL2, cells transduced with SH3GL2. B, epithelial-like appearance of the GFP-SH3GL2–overexpressing MDA-MB-231 mammary implants, with considerable cell–cell contact (arrowheads) compared with the GFP-empty vector–treated group exhibiting pronounced disruption of the epithelial morphology and cell–cell adhesion. C, IF analysis of the SH3GL2-overexpressing xenografts with ZO-1, CDH1, and F-actin demonstrates pronounced epithelial-like appearance and cell–cell adhesion (arrowheads) compared with the control group. Scale bar, 20 μm; magnification, 400×. D and E, the number of visible lung tumor nodules (encircled and scatter plot) and histologic macrometastases (arrows) were markedly lower (P = 0.0001) in the SH3GL2-overexpressing group compared with the empty vector–treated group. F and G, visible tumor nodules and histologic macrometastases (encircled and scatter plot) were detected only in the liver of the empty vector–treated group, but not in the SH3GL2-overexpressing group at 5 weeks. Scale bar, 50 μm; magnification, 200× (B, D–G).

Close modal

Immunohistochemical analysis of the in vivo FFPE tissues revealed enhanced expression of SH3GL2 (P = 0.0002) and CDH1 (P = 0.01) in the SH3GL2-overexpressing MDA-MB-231 cells compared with the control (Supplementary Fig. S5A). Increased CDH1 expression was further confirmed by Western blotting analysis in SH3GL2-overexpressing tumor tissues from multiple mice (Supplementary Fig. S5B). Enhanced expression of SH3GL2, MFN2, and PINK1 was also confirmed in mitochondria of SH3GL2-overexpressing implants from multiple mice compared with the control by Western blotting analysis (Supplementary Fig. S5C).

SH3GL2 and MFN2 expression is frequently lost and associated with breast cancer progression

The expression pattern of SH3GL2 in paired (normal/tumor matched) breast cancer tissues has not yet been evaluated. We examined SH3GL2 expression pattern in 51 primary human breast cancer tissues with and without lymph node metastasis (LNM; Supplementary Table S1A). Overall, loss of SH3GL2 was detected in 61% (31/51, P = 0.001–0.003) of primary breast cancer tissues (Fig. 5A). Thirty four of these 51 cases had LNM (Table S1A). SH3GL2 loss was detected in 79% (27/34, P = 0.001–0.002) of the primary tumors positive for LNM. Corresponding positive lymph node tissues were available for 20 of 51 cases (Supplementary Table S1A). Ninety percent (18/20) of the LNM tumors had a loss (P = 0.003–0.004) of SH3GL2.

Figure 5.

SH3GL2 and MFN2 expression was frequently lost and associated with breast cancer progression. A, SH3GL2 expression was significantly low (P = 0.001–0.003) in primary invasive breast cancer and corresponding LNM tissues compared with matched normal. Representative examples from two patients were shown. B, the expression level of MFN2 was also significantly lower (P = 0.002–0.004) in primary invasive breast cancer and corresponding LNM tissues compared with matched normal. Representative examples from the same two patients were shown. N, normal tissues; IDC, invasive ductal carcinoma; T: area containing tumor cells. Magnification, 200×. C–E, loss of SH3GL2 expression was associated with stage (P = 0.003, C), grade (P = 0.009, D), and LNM (P = 0.0002, E). F, MFN2 loss was associated with lymph node metastasis (P = 0.016). G, loss of both SH3GL2 and MFN2 was associated with LNM (P = 0.007). Grade I, well differentiated; grade II, moderately differentiated; grade III, poorly differentiated; no, negative for LNM; yes, positive for LNM; NSC, no significant change; SL, significant loss.

Figure 5.

SH3GL2 and MFN2 expression was frequently lost and associated with breast cancer progression. A, SH3GL2 expression was significantly low (P = 0.001–0.003) in primary invasive breast cancer and corresponding LNM tissues compared with matched normal. Representative examples from two patients were shown. B, the expression level of MFN2 was also significantly lower (P = 0.002–0.004) in primary invasive breast cancer and corresponding LNM tissues compared with matched normal. Representative examples from the same two patients were shown. N, normal tissues; IDC, invasive ductal carcinoma; T: area containing tumor cells. Magnification, 200×. C–E, loss of SH3GL2 expression was associated with stage (P = 0.003, C), grade (P = 0.009, D), and LNM (P = 0.0002, E). F, MFN2 loss was associated with lymph node metastasis (P = 0.016). G, loss of both SH3GL2 and MFN2 was associated with LNM (P = 0.007). Grade I, well differentiated; grade II, moderately differentiated; grade III, poorly differentiated; no, negative for LNM; yes, positive for LNM; NSC, no significant change; SL, significant loss.

Close modal

To our knowledge, the expression pattern of MFN2 and its association with breast cancer progression is unknown. The same cohort of 51 patients described above was analyzed to determine MFN2 expression (Supplementary Table S1A). We detected MFN2 loss in 55% (28/51, P = 0.002–0.004) of primary breast cancer tissues (Fig. 5B). MFN2 loss was detected in 65% (22/34, P = 0.0003–0.001) of the primary tumors positive for LNM. Eighty percent (16/20) of the available LNM from these patients exhibited loss (P = 0.001–0.002) of MFN2. Notably, loss of coexpression of SH3GL2 and MFN2 was detected in 37% (19/51) cases (Supplementary Table S1A). Of the 34 LNM cases, loss of their coexpression was evident in 47% (16/34) cases (Supplementary Table S1A). Thirteen of 20 LNM tumors (65%) had a simultaneous loss of SH3GL2 and MFN2 (Supplementary Table S1A). Loss of SH3GL2 was associated with stage (P = 0.003), grade (P = 0.009), and LNM (P = 0.0002; Fig. 5C–E). Loss of MFN2 expression was associated with LNM alone (P = 0.016; Fig. 5F) or in combination with SH3GL2 (P = 0.007; Fig. 5G). No association was found between MFN2 loss and stage (P = 0.028; Supplementary Fig. S6A) or grade alone (P = 0.99; Supplementary Fig. S6B) or in combination with SH3GL2 for stage (P = 0.09; Supplementary Fig. S6C) or grade (P = 0.22; Supplementary Fig. S6D). Of note, other than the tumor adjacent normal, we detected high SH3GL2 and MFN2 expression in 100% (4/4) of normal human breast ductal epithelial tissues obtained from cancer-free normal women undergoing mammoplasty (Supplementary Fig. S6E). We also confirmed the presence of SH3GL2 in mitochondria of the above four breast tissues obtained from normal healthy women (Supplementary Fig. S6F).

An SH3GL2 and MFN2 loss was detectable in the circulating exosomes

SH3GL2 was overexpressed in three breast cancer cell lines (Supplementary Fig. S1A). To determine whether the cancer cell exosomes carry this information, we performed Western blotting analysis using the culture supernatant–derived exosomes. We could detect a low level of SH3GL2 expression in the control culture-derived exosomes (Fig. 6A). On the other hand, the enhanced expression level of SH3GL2 was readily detectable in the exosomes derived from the SH3GL2-overexpressing cultures (Fig. 6A). MFN2 expression was induced in these cell lines following SH3GL2 overexpression (Fig. 1). The induced expression level of MFN2 was also detectable in the culture-derived exosomes of the SH3GL2-overexpressing cells compared with the control (Fig. 6A).

Figure 6.

Detection of SH3GL2 and MFN2 signature in the exosomes. A, enhanced expression of SH3GL2 and MFN2 in the culture supernatant–derived exosomes of the SH3GL2-overexpressing cells. EV, empty vector transduced; SH3GL2, SH3GL2-transduced cells. Syntenin was used as an exosome marker and loading control. B, compared with the normal healthy women, all but 2 breast cancer cases (CE12 and CE21) demonstrated a low or barely detectable expression of SH3GL2 in the circulating exosomes. C, all but 4 breast cancer cases (CE1, CE2, CE18, and CE20) exhibited a low or barely detectable expression of MFN2 in the circulating exosomes, compared with the normal controls. NE, exosomes isolated from sera of cancer-free normal healthy women; CE, exosomes isolated from sera of women with invasive ductal carcinomas. D, model depicting the possible interplay between the phosphorylated SH3GL2, MFN2, and PINK1 in mitochondria, which triggered mitochondrial fusion network, O2 production, and release of CYTC, leading to apoptotic induction.

Figure 6.

Detection of SH3GL2 and MFN2 signature in the exosomes. A, enhanced expression of SH3GL2 and MFN2 in the culture supernatant–derived exosomes of the SH3GL2-overexpressing cells. EV, empty vector transduced; SH3GL2, SH3GL2-transduced cells. Syntenin was used as an exosome marker and loading control. B, compared with the normal healthy women, all but 2 breast cancer cases (CE12 and CE21) demonstrated a low or barely detectable expression of SH3GL2 in the circulating exosomes. C, all but 4 breast cancer cases (CE1, CE2, CE18, and CE20) exhibited a low or barely detectable expression of MFN2 in the circulating exosomes, compared with the normal controls. NE, exosomes isolated from sera of cancer-free normal healthy women; CE, exosomes isolated from sera of women with invasive ductal carcinomas. D, model depicting the possible interplay between the phosphorylated SH3GL2, MFN2, and PINK1 in mitochondria, which triggered mitochondrial fusion network, O2 production, and release of CYTC, leading to apoptotic induction.

Close modal

To determine SH3GL2 and MFN2 expression pattern in the circulating exosomes, we performed Western blotting analysis. Sera exosomes from 27 women with invasive breast cancer and 6 normal, healthy women were scrutinized to determine SH3GL2 and MFN2 expression. All but 2 breast cancer cases (CE12 and CE21) exhibited a low or barely detectable expression of SH3GL2 in the sera exosomes (Fig. 6B). An appreciable level of SH3GL2 expression was detectable in the circulating exosomes derived from normal sera (Fig. 6B; NE1–NE6). Similarly, all but 4 breast cancer cases (CE1, CE2, CE18, and CE20) exhibited a low or barely detectable expression of MFN2 in the circulating exosomes (Fig. 6C). An appreciable level of MFN2 expression was detectable in the normal sera–derived exosomes (Fig. 6C; NE1–NE6).

Recent reviews of the cancer genome landscape implicated that cancer mortality can be reduced to more than 70% by early detection and prevention (27). Other than mammography, MRI, and ultrasound-based screening, no other suitable method is clinically available for the early diagnosis and monitoring of breast cancer (28–30). However, these methods are limited to detect early genetic changes and predict the biologic behavior of the tumor. From the treatment standpoint, targeted cancer therapy is often dependent on overexpressed and activated oncogenes such as HER2/neu for breast cancer (31). However, in the majority of the solid tumors, tumor suppressor genes (TSG) act as the drivers of cancer development and progression (27). For example, p53 is the most frequently mutated TSG among the top 21 altered genes in breast cancer (32). Therapeutic targeting of TSGs is difficult as they are already inactivated by allelic loss, genetic mutation, or various other mechanisms. Thus, molecular characterization of the key TSGs could aid to develop strategies for early detection, monitoring, and surveillance. Our studies defined previously unidentified loss and function of SH3GL2 in breast cancer progression. SH3GL2 is expressed predominantly in the cytoplasm and altered in various malignancies (33, 34). The translocation and phosphorylation of SH3GL2 in mitochondria appears to trigger the intrinsic apoptotic pathway through induction of O2 production, mitochondrial fusion, and CYTC release. Although, there was some basal and low level of SH3GL2 expression in the breast cancer cells, its exogenous introduction induced the phenotypic and associated molecular changes. It could be due to relatively inactive endogenous SH3GL2 and its inability to augment the mitochondrial proteins and release CYTC in the cytoplasm, as evident from our studies. CYTC release from mitochondria to the cytosol is critical in the intrinsic apoptotic pathway (18), whereas production in tumors of normal mitochondria through mitochondrial fusion could be tumor suppressive by promoting oxidative metabolism and enhanced ROS production (6, 8). At the molecular level, cross-talk between SH3GL2, MFN2, and PINK1 in mitochondria appeared to be the key event in opposing progression in vitro and in vivo. Their colocalizations also indicate that the interplay occurs in mitochondria. PINK1 overexpression was shown to reduce anchorage-independent growth of breast cancer cells, and its mRNA expression was associated with better survival in adrenocortical tumors (35). A number of studies demonstrated a tumor-suppressive role of MFN2 in lung, bladder, liver, gastric, and colon carcinomas by reducing proliferation and triggering apoptosis (36–41). In lung cancer cells, enhanced MFN2 expression induced apoptosis by triggering ROS production in vitro and reduced in vivo growth by augmenting mitochondrial fusion network (36). An in vitro study also demonstrated a marked decrease in migration and invasion of breast cancer cells through enhanced MFN1 and MFN2 expression and mitochondrial fusion (42). In this context, the induction of ATG5, PTEN, and CDH1 might also have contributed to the onset of apoptosis and growth inhibition. ATG5 was shown to induce apoptosis (43), and enhanced expression of ATG5 was associated with favorable disease-free survival of breast cancer patients (44). So far, the functional role of ATG5 in breast cancer is unknown. On the other hand, PTEN is a well-defined TSG and the regulator of PINK1 (19). CDH1 is a key molecule of the cell–cell adhesion complex, and its loss promotes metastasis (20, 21). In a recent study, simultaneous loss of CDH1 and PTEN was shown to accelerate cellular invasiveness and angiogenesis in the mouse uterus (45). Thus, concurrent augmentation of these molecules following SH3GL2 overexpression might have contributed in preventing primary growth and metastasis by promoting apoptosis. The well-differentiated morphology of the SH3GL2-overexpressing in vivo implants with marked cell–cell contact as confirmed by various epithelial and mesenchymal markers could be due to the rescued expression of CDH1 and other molecules. On the other hand, SH3GL2 knockdown in nontumorigenic as well as tumorigenic breast epithelial cells confirmed the involvement of mitochondrial pathway proteins, along with PTEN–CDH1 niche for opposing tumor growth and progression.

Despite the increased expression of key mitochondrial biogenesis–associated molecules PGC-1α and MT-TFA, we did not observe an increase in mitochondrial DNA content (data not shown). However, an increase in overall mitochondrial mass was observed as demonstrated by an enhanced mitochondrial fusion network. PGC-1α is a key coordinator of mitochondrial function, including regulation of MT-TFA transcription (26). Loss of expression of PGC-1α alone, and in combination with MT-TFA, was found to be associated with highly aggressive clear-cell ovarian carcinoma progression and its resistance to chemotherapy (26). In addition, frequent truncating mutations of MT-TFA resulted in mitochondrial DNA depletion and apoptotic resistance in colon cancer (46). Thus, augmentation of both PGC-1α and MT-TFA and a physical association between PGC-1α and SH3GL2 strongly suggest for the possibility of functional cooperation among these molecules leading to normalized mitochondrial function and apoptosis.

SH3GL2 is located on human chromosome 9p22, a frequently deleted region in breast cancer (47). In addition to SH3GL2, the expression pattern of MFN2 in breast cancer progression remains unknown. Loss of coexpression of both of these potential TSGs and their association with disease progression implicate for a causative role in breast tumorigenesis. The normal expression of a TSG may be affected in cancer-adjacent normal tissues due to the “field cancerization effect”. However, high and comparable level of SH3GL2 and MFN2 expression in the mammoplasty tissues indicated that these are abundantly expressed proteins in normal breast ductal epithelial cells. Notably, the detection of SH3GL2 in mitochondria of these normal breast tissues further suggests a role of SH3GL2 in regulating mitochondrial function or biogenesis. However, due to the lack of premalignant tissue samples, we could not evaluate the expression pattern of SH3GL2 and MFN2 at this time.

Exosomes are emerging as a promising biomarker tool as they carry specific genetic information and influence tumor growth and progression (10–12). Syntenin is among the top 20 proteins most abundantly expressed in the exosomes and epithelial cell cancers (48–50). Thus, syntenin could serve as a reliable exosome marker. The detection of enhanced SH3GL2 and MFN2 expression in the SH3GL2-overexpressing cell–derived exosomes suggests that these molecules perform important regulatory functions through the exosomes. Otherwise, we would not have seen their elevated expression in the culture-derived exosomes following SH3GL2 overexpression. The tumor suppressive effect that we have seen following SH3GL2 overexpression in the breast cancer cells could partly be due the “paracrine effect” of these exosomes enriched in SH3GL2 and MFN2 proteins. As these molecules are potential TSGs and normally carried by the exosomes as evident from our studies, their early inactivation or loss could favor cancer initiation and progression and vice versa. Possibly as a reason, the expression of these molecules was barely detectable in the circulating exosomes of the majority of the women with invasive disease. To our knowledge, no studies so far reported the presence of SH3GL2 and MFN2 proteins in the circulating exosomes of normal, healthy women and their loss of expression in breast cancer patients. Taken together, our study uncovered a novel mitochondrial reprogramming pathway regulating breast cancer development and progression. In this signaling cascade, activated SH3GL2 appears to induce PTEN and interacts with MFN2 and PINK1 in mitochondria (Fig. 6D). This interplay normalizes mitochondrial function and triggers apoptosis by releasing O2 and CTYC from the mitochondria (Fig. 6D). Molecular profiling of SH3GL2 and MFN2 alterations in circulating exosomes could be a feasible approach for noninvasive early detection, monitoring, and surveillance of breast cancer.

No potential conflicts of interest were disclosed.

Conception and design: K.P. Singh, E.R. Sauter, S. Dasgupta

Development of methodology: A. Kannan, S. Dasgupta

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Kannan, S. Dasgupta

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Kannan, S. Sivakumar, S. Komatsu, K.P. Singh, B. Samten, M. Ikebe, S. Idell, S. Dasgupta

Writing, review, and/or revision of the manuscript: A. Kannan, S. Sivakumar, E.R. Sauter, M. Ikebe, S. Idell, S. Gupta, S. Dasgupta

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Kannan, J.V. Philley, S. Dasgupta

Study supervision: S. Dasgupta

Other (pathologic evaluation): R.B. Wells

This work is dedicated to the memory of Jnan Ranjan Dasgupta for his brave fight against cancer. The authors thank Zane Robertson for his generous help with the histopathology and Karen Durham, a Susan G. Komen Scholar and advocate, for critical reading of the manuscript and support of our research. The authors also thank Henry James at the UTHSCT Animal Facility for his help on the animal work.

The study was supported by the UTHSCT and Chamblee Foundation (to S. Dasgupta).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Eccles
SA
,
Aboagye
EO
,
All
S
,
Anderson
AS
,
Armes
J
,
Berditchevski
F
, et al
Critical research gaps and translational priorities for the successful prevention and treatment of breast cancer
.
Breast Cancer Res
2013
;
15
:
R92
.
2.
National Cancer Institute
.
Available from
: www.cancer.gov.
3.
Polyak
K
. 
Breast cancer: origin and evolution
.
J Clin Invest
2007
;
117
:
3155
63
.
4.
Dos Anjos Pultz
B
,
da Luz
FA
,
de Faria
PR
,
Oliveira
AP
,
de Araújo
RA
,
Silva
MJ
. 
Far beyond the usual biomarkers in breast cancer: a review
.
J Cancer
2014
;
5
:
559
71
.
5.
Bombonati
A
,
Sgori
DC
. 
The molecular pathology of breast cancer progression
.
J Pathol
2011
;
223
:
310
317
.
6.
Archer
SL
. 
Mitochondrial dynamics- mitochondrial fission and fusion in human diseases
.
New Eng J Med
2013
;
369
:
2236
51
.
7.
He
Y
,
Wu
J
,
Dressman
DC
,
Iacobuzio-Donahue
C
,
Markowitz
SD
,
Velculescu
VE
, et al
Heteroplasmic mitochondrial DNA mutations in normal and tumour cells
.
Nature
2010
;
464
:
610
14
.
8.
Boland
ML
,
Chourasia
AH
,
Macleod
KF
. 
Mitochondrial dysfunction in cancer
.
Front Oncol
2013
;
3
:
1
18
.
9.
Da silva
AF
,
Mariotti
FR
,
Maximo
V
,
Campello
S
. 
Mitochondria dynamism: of shape, transport and cell migration
.
Cell Mol Life Sci
2014
;
71
:
2313
24
.
10.
Azmi
AS
,
Bao
B
,
Sarkar
FH
. 
Exosomes in cancer development, metastasis, and drug resistance: a comprehensive review
.
Cancer Metastasis Rev
2013
;
32
:
623
42
.
11.
Kahlert
C
,
Kalluri
R
. 
Exosomes in tumor microenvironment influence cancer progression and metastasis
.
J Mol Med
2013
;
91
:
431
37
.
12.
Melo
SA
,
Sugimoto
H
,
O'Connell
JT
,
Kato
N
,
Villanueva
A
,
Vidal
A
, et al
Cancer Exosomes Perform Cell-Independent MicroRNA Biogenesis and Promote Tumorigenesis
.
Cancer Cell
2014
;
26
:
707
21
.
13.
Kjaerulff
O
,
Brodin
L
,
Jung
A
. 
The structure and function of endophilin proteins
.
Cell Biochem Biophys
2011
;
60
:
137
54
.
14.
Dasgupta
S
,
Menezes
ME
,
Das
SK
,
Emdad
L
,
Janijic
A
,
Bhatia
S
, et al
Novel role of MDA-9/syntenin in regulating urothelial cell proliferation by modulating EGFR signaling
.
Clin Cancer Res
2013
;
19
:
4621
33
.
15.
Oyesanya
RA
,
Bhatia
S
,
Menezes
ME
,
Dumur
CI
,
Singh
KP
,
Bae
S
, et al
MDA-9/Syntenin regulates differentiation and angiogenesis programs in head and neck squamous cell carcinoma
.
Oncoscience
2014
;
1
:
725
37
.
16.
Iorns
E
,
Drews-Elger
K
,
Ward
TM
,
Dean
S
,
Clarke
J
,
Berry
D
, et al
A new mouse model for the study of human breast cancer metastasis
.
PLoS One
2012
;
7
:
e4799
.
17.
Sinha
S
,
Chunder
N
,
Mukherjee
N
,
Alam
N
,
Roy
A
,
Roychoudhury
S
, et al
Frequent deletion and methylation in SH3GL2 and CDKN2A loci are associated with early- and late-onset breast carcinoma
.
Ann Surg Oncol
2008
;
15
:
1070
80
.
18.
Dasgupta
S
,
Hoque
MO
,
Upadhyay
S
,
Sidransky
D
. 
Forced Cytochrome B gene mutation expression induces mitochondrial proliferation and prevents apoptosis in human uroepithelial SV-HUC-1 cells
.
Int J Cancer
2009
;
125
:
2829
35
.
19.
O'Flanagan
CH
,
O'Neill
C
. 
PINK1 signaling in cancer biology
.
Biochim Biophys Acta
2014
;
1846
:
590
8
.
20.
Onder
TT
,
Gupta
PB
,
Mani
SA
,
Yang
J
,
Lander
ES
,
Weinberg
RA
, et al
Loss of E-cadherin promotes metastasis via multiple downstream transcriptional pathways
.
Cancer Res
2008
;
68
:
3645
54
.
21.
Berx
G
,
Van Roy
F
. 
The E-cadherin/catenin complex: an important gatekeeper in breast cancer tumorigenesis and malignant progression
.
Breast Cancer Res
2001
;
3
:
289
93
.
22.
Lu
H
,
Li
G
,
Liu
L
,
Wang
X
,
Jin
H
. 
Regulation of and function of mitophagy in development and cancer
.
Autophagy
2013
;
9
:
1720
36
.
23.
Gomes
LC
,
Scorrano
L
. 
Mitochondrial morphology in mitophagy and macroautophagy
.
Biochim Biophys Acta
2013
;
1833
:
205
12
.
24.
Dolman
NJ
,
Chambers
KM
,
Mandavilli
B
,
Batchelor
RH
,
Janes
MS
. 
Tools and techniques to measure mitophagy using fluorescence microscopy
.
Autophagy
2013
;
9
:
1653
62
.
25.
Zhu
J
,
Wang
ZQ
,
Chu
CT
. 
Mitochondrial biogenesis, mitophagy and cell survival
.
Autophagy
2013
;
9
:
1663
76
.
26.
Gabrielson
M
,
Bjorklund
M
,
Carlson
J
,
Shoshan
M
. 
Expression of mitochondrial regulators pgc-1α and tfam as putative markers of subtype and chemoresistance epithelial ovarian carcinoma
.
PLoS One
2014
;
9
:
e107109
.
27.
Vogelstein
B
,
Papadopoulos
N
,
Velculescu
VE
,
Zhou
S
,
Diaz
LA
 Jr
,
Kinzler
KW
. 
Cancer genome landscapes
.
Science
2013
;
339
:
1546
58
.
30.
Hirata
BKB
,
Oda
JMM
,
Guembarovski
RL
,
Ariza
CB
,
Coral de Oliveira
CE
,
Watanabe
MA
. 
Molecular markers for breast cancer: prediction on tumor behavior
.
Dis Markers
2014
;
2014
:
513158
.
31.
Wang
WJ
,
Lei
YY
,
Mei
JH
,
Wang
CL
. 
Recent progress in HER2 associated breast cancer
.
Asian Pac J Cancer Prev
2015
;
16
:
2591
600
.
32.
Polyak
K
,
Filho
OM
. 
Snapshots: breast cancer
.
Cancer Cell
2012
;
22
:
562
.
33.
Dasgupta
S
,
Jang
JS
,
Shao
C
,
Mukhopadhyay
ND
,
Sokhi
UK
,
Das
SK
, et al
SH3GL2 is frequently deleted in non-small cell lung cancer and downregulates tumor growth by modulating EGFR signaling
.
J Mol Med
2013
;
91
:
381
93
.
34.
Majumdar
S
,
Gong
EM
,
Di Vizio
D
,
Dreyfuss
J
,
Degraff
DJ
,
Hager
MH
, et al
Loss of Sh3gl2/endophilin A1 is a common event in urothelial carcinoma that promotes malignant behavior
.
Neoplasia
2013
;
15
:
749
60
.
35.
Gomes
LC
,
Scorrano
L
. 
Mitochondrial morphology in mitophagy and macroautophagy
.
Biochim Biophys Acta
2013
;
1833
:
205
12
.
36.
Rehman
J
,
Zhang
HJ
,
Toth
PT
,
Zhang
Y
,
Marsboom
G
,
Hong
Z
, et al
Inhibition of mitochondrial fission prevents cell cycle progression in lung cancer
.
FASEB J
2012
;
26
:
2175
86
.
37.
Jin
B
,
Fu
G
,
Pan
H
,
Cheng
X
,
Zhou
L
,
Lv
J
, et al
Anti-tumor efficacy of mitofusin-2 in urinary bladder carcinoma
.
Med Oncol
2011
;
28
Suppl 1
:
S373
S380
.
38.
Wang
W
,
Lu
J
,
Zhu
F
,
Wei
J
,
Jia
C
,
Zhang
Y
, et al
Pro-apoptotic and anti-proliferative effects of mitofusin-2 via bax signaling in hepatocellular carcinoma cells
.
Med Oncol
2012
;
29
:
70
6
.
39.
Wang
W
,
Xie
Q
,
Zhou
X
,
Yao
J
,
Zhu
X
,
Huang
P
, et al
Mitofusin-2 triggers mitochondria Ca2+ influx from the endoplasmic reticulum to induce apoptosis in hepatocellular carcinoma cells
.
Cancer Lett
2015
;
358
:
47
58
.
40.
Zhang
GE
,
Jin
HL
,
Lin
XK
,
Chen
C
,
Liu
XS
,
Zhang
Q
, et al
Anti-tumor effects of mfn2 in gastric cancer
.
Int J Mol Sci
2013
;
14
:
13005
21
.
41.
Cheng
X
,
Zhou
D
,
Wei
J
,
Lin
J
. 
Cell-cycle arrest at G2/M and proliferation inhibition by adenovirus-expressed mitofusin-2 gene in human colorectal cancer cell lines
.
Neoplasma
2013
;
60
:
620
26
.
42.
Zhao
J
,
Zhang
J
,
Yu
M
,
Xie
Y
,
Huang
Y
,
Wolff
DW
, et al
Mitochondrial dynamics regulates migration and invasion of breast cancer cells
.
Oncogene
2013
;
32
:
4814
24
.
43.
Luo
S
,
Rubinszztein
DC
. 
Atg5 and bcl-2 provide novel insight into the interplay between apoptosis and autophagy
.
Cell Death Differ
2007
;
14
:
1247
50
.
44.
Wang
L
,
Yao
L
,
Zheng
YZ
,
Xu
Q
,
Liu
XP
,
Hu
X
, et al
Expression of autophagy-related proteins atg5 and fip200 predicts favorable disease-free survival in patients with breast cancer
.
Biochem Biophys Res Commun
2015
;
458
:
816
22
.
45.
Lindberg
ME
,
Stodden
GR
,
King
ML
,
MacLean
JA
 2nd
,
Mann
JL
,
DeMayo
FJ
, et al
Loss of CDH1 and Pten accelerates cellular invasiveness and angiogenesis in the mouse uterus
.
Biol Reprod
2013
;
89
:
8
.
46.
Guo
J
,
Zheng
L
,
Liu
W
,
Wang
X
,
Wang
Z
,
Wang
Z
, et al
Frequent truncating mutation of TFAM induces mitochondrial DNA depletion and apoptotic resistance in microsatellite-unstable colorectal cancer
.
Cancer Res
2011
;
71
:
2978
87
.
47.
Brenner
AJ
,
Aldaz
CM
. 
Cancer Res. Chromosome 9p allelic loss and p16/CDKN2 in breast cancer and evidence of p16 inactivation in immortal breast epithelial cells
.
Cancer Res
1995
;
55
:
2892
95
.
48.
Baietti
MF
,
Zhang
Z
,
Mortier
E
,
Melchior
A
,
Degeest
G
,
Geeraerts
A
, et al
Syndecan-syntenin-ALIX regulates the biogenesis of exosomes
.
Nat Cell Biol
2012
;
14
:
677
85
.
49.
Tomlinson
PR
,
Zheng
Y
,
Fischer
R
,
Heidasch
R
,
Gardiner
C
,
Evetts
S
, et al
Identification of distinct circulating exosomes in Parkinson's disease
.
Ann Clin Transl Neurol
2015
;
2
:
353
61
.
50.
Philley
JV
,
Kannan
A
,
Dasgupta
S
. 
MDA-9/Syntenin control
.
J Cell Physiol
2016
;
231
:
545
50
.