Myc Regulation of a Mitochondrial Trafficking Network Mediates Tumor Cell Invasion and Metastasis

The Myc gene is a universal oncogene that promotes aggressive cancer, but its role in metastasis has remained elusive. Here, we show that Myc transcriptionally controls a gene network of subcellular mitochondrial trafficking that includes the atypical mitochondrial GTPases RHOT1 and RHOT2, the adapter protein TRAK2, the anterograde motor Kif5B, and an effector of mitochondrial fission, Drp1.

ABSTRACT

The Myc gene is a universal oncogene that promotes aggressive cancer, but its role in metastasis has remained elusive. Here, we show that Myc transcriptionally controls a gene network of subcellular mitochondrial trafficking that includes the atypical mitochondrial GTPases RHOT1 and RHOT2, the adapter protein TRAK2, the anterograde motor Kif5B, and an effector of mitochondrial fission, Drp1. Interference with this pathway deregulates mitochondrial dynamics, shuts off subcellular organelle movements, and prevents the recruitment of mitochondria to the cortical cytoskeleton of tumor cells. In turn, this inhibits tumor chemotaxis, blocks cell invasion, and prevents metastatic spreading in preclinical models. Therefore, Myc regulation of mitochondrial trafficking enables tumor cell motility and metastasis.

INTRODUCTION

Myc proteins are products of a family of potent, transforming oncogenes (1) that are amplified, deregulated, or translocated in most human cancers (2). As a member of the basic helix-loop-helix group of transcription factors (3), oncogenic Myc drives a panoply of transcriptional (4) and nontranscriptional (5) responses that promote tumor growth and proliferation. The identity of Myc target genes remains to be fully elucidated (6), and controversy still exists about the requirements of Myc-directed transcription (7), but there is evidence that this pathway affects multiple functions in mitochondria of transformed cells (1).

Accordingly, Myc-directed transcription has been associated with the expression of oxidative phosphorylation genes (8), exploitation of multiple bioenergetics pathways (9), modulation of biogenesis (10), and changes in mitochondrial dynamics (11), an adaptive process that controls the size, shape, and subcellular position of mitochondria (12). How these functions contribute to Myc-directed tumorigenesis is still largely elusive, and their impact on advanced disease traits, such as tumor cell dissemination to distant organs, or metastasis (13), has not been elucidated.

In this context, approximately 90% of all cancer deaths are due to metastasis (13). This is a multistep process that integrates disparate cellular responses, including chemotaxis (14), epithelial-mesenchymal transition (15), plasticity of circulating tumor cells (16), signaling from immune-inflammatory cells (17), and cues from the microenvironment (18). With a better understanding of the role of mitochondria in cancer (19), we now know that mitochondrial reprogramming is important for metastasis (20). Mechanistically, this may involve oxidative bioenergetics (21), signaling by reactive oxygen species (ROS) (22), and active redistribution of mitochondria to the cortical cytoskeleton of tumor cells, fueling membrane dynamics of cell motility and tumor cell invasion (23–25).

In this study, we investigated a novel interface between Myc-directed transcription and subcellular mitochondrial trafficking in regulating tumor cell movements and metastasis.

RESULTS

Myc regulation of mitochondrial dynamics.

We began this study by mapping by RNA sequencing (RNA-Seq) the gene expression profile of Burkitt lymphoma P493 cells after doxycycline (Dox)-mediated conditional silencing of Myc (Dox-off system; t = 24 h) (26). We found that Myc loss affected the expression of 6,860 genes (false discovery rate [FDR] of <5%; at least a 2-fold change) and that regulators of mitochondrial functions comprised the most overrepresented class of modulated genes (n = 515; FDR = 4 × 10⁻²⁷) (Fig. 1A). Other Myc-regulated gene networks in these experiments included cell cycle transitions (n = 273; FDR = 5 × 10⁻⁶), DNA replication (n = 62; FDR = 4 × 10⁻⁸), ribonucleoproteins (n = 142; FDR = 2 × 10⁻⁷), transit peptides (n = 271; FDR = 4 × 10⁻⁸), the immune response (n = 206; FDR = 5 × 10⁻¹⁵), innate immunity (n = 119; FDR = 5 × 10⁻⁶), and mitochondrial inner membrane functions (n = 128; FDR = 6 × 10⁻⁶) (Fig. 1A). Bioinformatics analysis of this data set identified mitochondrial dynamics (27) as a major target of Myc-directed transcription, affecting organelle shape and size (↓ fusion, ↓ fission, ↓ mass, ↑ fragmentation, and ↑ aggregation), quality control (↑ degradation, ↑ damage, and ↑ autophagy), and subcellular trafficking (↓ distribution) (down arrows mean reduction and up arrows mean increase in the corresponding pathways) (Fig. 1B).

FIG 1.

Oncogenic Myc regulation of mitochondrial dynamics. (A) UniProt classification categories of RNA-Seq genes most significantly affected by conditional Myc knockdown in Burkitt lymphoma P493 cells. Enr, enrichment (fold number of genes from each individual category over values expected by chance); FDR, false-discovery rate (Bonferroni-corrected P value); IM, inner membrane. (B) Ingenuity Pathway Analysis of 24 mitochondrial functions with at least 10% of genes significantly modulated by conditional Myc knockdown in P493 cells. The majority of identified functions shared at least 50% of the affected genes with other functions. Unique genes in selected mitochondrial functions are indicated. KO, knockout. (C and D) PC3 cells were transfected with control nontargeting siRNA (siCtrl) or Myc-directed siRNA (siMyc), and mitochondrial accumulation at the cortical cytoskeleton was imaged by confocal microscopy (C) and quantified (D). 3D isosurface renderings of representative cells are shown. Symbols correspond to an individual determination (n = 45 to 48). Scale bars, 5 μm. ***, P < 0.0001. (E) PC3 cells were transfected as described above for panel C, and the speed of individual MitoTracker-labeled mitochondria was quantified by time-lapse videomicroscopy. Representative tracings from two independent experiments (Exp.) are shown. (F) Experiments were performed under the conditions described above for panel C, and transfected PC3 cells expressing RFP-labeled mitochondria were imaged for mitochondrial motility by time-lapse videomicroscopy, expressed in 2D plots. Each tracing corresponds to the movement of an individual mitochondrion. The cutoff velocities for slow-moving (blue) or fast-moving (magenta) mitochondria are indicated. Data are from a representative experiment (n = 3). (G) Experiments were performed under the conditions described above for panel C, and the speed of mitochondrial movements (top) (n = 52 to 55) and total distance traveled by individual mitochondria (bottom) (n = 56) were quantified. Each symbol corresponds to an individual determination. ***, P < 0.0001.

Based on these data, we next studied mitochondrial dynamics in a model of prostate cancer, where deregulated Myc drives lethal disease (28). We found that Myc silencing in prostate adenocarcinoma PC3 cells caused mitochondria to remain densely clustered in the perinuclear region, with virtually no subcellular distribution to the cortical cytoskeleton (Fig. 1C and D). In contrast, mitochondria infiltrated the cortical cytoskeleton in control small interfering RNA (siRNA) transfectants (Fig. 1C and D). Consistent with these findings, Myc silencing suppressed mitochondrial movements in PC3 cells (Fig. 1E), converting rapidly moving mitochondria into slow-moving or stationary organelles, by time-lapse videomicroscopy (Fig. 1F). Quantitative analysis of these data demonstrated that Myc loss inhibited both the speed of mitochondrial movements and the total distance traveled by individual mitochondria, compared to control siRNA transfectants (Fig. 1G).

Next, we searched for the effector(s) of mitochondrial motility in Myc-targeted cells, and we focused on dynamin-related protein 1 (Drp1), a mediator of mitochondrial fission (27) and subcellular mitochondrial trafficking (29). Chromatin immunoprecipitation sequencing (ChIP-Seq) tracks showed that Myc occupied the Drp1 promoter in a time-dependent manner after Dox removal in P493 (t = 0, 1, and 24 h) as well as neuroblastoma BE2C, Kelly, and NGP cells (Fig. 2A). Consistent with these findings, Myc bound to the Drp1 promoter of PC3 cells by ChIP, in a reaction abolished by siRNA knockdown of Myc (Fig. 2B). Transfection with a control siRNA had no effect (Fig. 2B). Loss of Myc in these settings was associated with reduced Drp1 mRNA levels (Fig. 2C) as well as protein expression (Fig. 2D) by quantitative PCR (qPCR) and Western blotting, respectively.

FIG 2.

Myc regulation of Drp1. (A) Chromatin immunoprecipitation sequencing (ChIP-Seq) tracks of Myc occupancy of the Drp1 promoter in Burkitt lymphoma P493 cells at three time points (t = 0, 1, and 24 h) after Dox removal and in neuroblastoma BE2C, Kelly, or NGP cells. (B) PC3 cells transfected with siCtrl or siMyc were analyzed for Myc accumulation at the Drp1 promoter by ChIP. IgG, nonbinding IgG. Means ± SD are shown (n = 3). (C and D) Experiments were performed under the conditions described above for panel B, and transfected PC3 cells were analyzed for changes in Drp1 mRNA by quantitative PCR (qPCR) (C) or protein expression by Western blotting (D). Means ± SD are shown (n = 3). ***, P < 0.0001. MTE, mitochondrial extracts; p, phosphorylated. (E and F) PC3 cells transfected as described above for panel B were labeled as indicated and analyzed for colocalization of Ser616-phosphorylated Drp1 (pDrp1) and mitochondria (TOM20) by confocal fluorescence microscopy (E), with quantification of the Pearson correlation coefficient (PCC) of Drp1-mitochondrion colocalization (F). Representative images and higher-magnification insets of selected areas are shown. Each symbol corresponds to an individual determination (n = 2). ***, P = 0.0003. (G) PC3 cells were transfected as described above for panel B, and isolated mitochondrial extracts were analyzed by Western blotting. (H) PC3 cells transfected as described above for panel B were analyzed by Western blotting. (I) PC3 cells transfected with siCtrl or siMyc were reconstituted with Drp1 cDNA and analyzed by Western blotting. (J and K) Experiments were performed under the reconstitution conditions described above for panel I, and transfected PC3 cells were imaged for individual mitochondrial motility by time-lapse videomicroscopy expressed in 2D plots (J), and the speed of mitochondrial movements (K, top) (n = 97 to 99) or total distance traveled by individual mitochondria (K, bottom) (n = 63 to 67) was quantified. The cutoff velocities for slow-moving (blue) or fast-moving (magenta) mitochondria are indicated. Each symbol corresponds to an individual determination. Data from a representative experiment are shown (n = 3). ***, P < 0.0001; ns, not significant. (L) DU145 cells were transfected as described above for panel B and analyzed for mitochondrial ROS generation by MitoSox red staining and flow cytometry. Representative tracings are shown.

Based on these data, we next looked at the function of Drp1 in mitochondria. Consistent with the overall decrease in protein expression (Fig. 2D), Myc silencing reduced the recruitment of Drp1 to mitochondria (Fig. 2E and F) and lowered the fraction of Drp1 phosphorylated on Ser616, i.e., active, compared to control transfectants (Fig. 2G). The expression of other regulators of mitochondrial dynamics, including mitofusin-1 (MFN1) or Opa1, was not affected, whereas MFN2 levels increased after Myc silencing in PC3 cells (Fig. 2H). To test the specificity of these findings, we next reconstituted Myc-silenced PC3 cells with non-siRNA-inhibitable Drp1 cDNA (Fig. 2I). Reexpression of Drp1 in these settings rescued mitochondrial motility by time-lapse videomicroscopy (Fig. 2J) and restored the speed of mitochondrial movements and the total distance traveled by individual mitochondria to levels observed in control siRNA transfectants (Fig. 2K). Conversely, Drp1 expression in control or Myc-silenced cells did not affect mitochondrial ROS production (Fig. 2L), which was recently implicated as a second messenger in mitochondrial trafficking (30).

Myc modulation of a subcellular mitochondrial trafficking network in tumor cells.

As elucidated in neurons (31), subcellular mitochondrial trafficking relies on an integrated protein network that comprises the mitochondrial atypical GTPases RHOT1 and RHOT2, the adapter molecules TRAK1 and TRAK2, and the anterograde motor Kif5B. Analysis of ChIP-Seq tracks in P493 or neuroblastoma BE2C, Kelly, and NGP cells revealed strong accumulation of Myc at promoter regions of RHOT1, RHOT2, TRAK2, and Kif5B (Fig. 3A). Accordingly, Myc bound to all these promoters by ChIP in control siRNA transfectants but not after transfection of Myc-directed siRNA (siMyc) (Fig. 3B). Consistent with these data, Myc silencing in PC3 cells reduced mRNA (Fig. 3C) and protein (Fig. 3D) expression of RHOT1, RHOT2, TRAK2, and Kif5B, compared to control cultures.

FIG 3.

Myc transcriptional control of a mitochondrial trafficking network. (A) ChIP-Seq tracks of Myc accumulation at the RHOT1, RHOT2, TRAK2, or Kif5B promoter in P493 cells at three time points (t = 0 h, 1 h, and 24 h) after Dox removal or in neuroblastoma BE2C, Kelly, and NGP cells. (B) PC3 cells were transfected with siCtrl or siMyc and analyzed for Myc accumulation at the RHOT1, RHOT2, TRAK2, or Kif5B promoter by ChIP. IgG, nonbinding IgG. Means ± SD are shown (n = 3). (C and D) PC3 cells transfected as described above for panel B were analyzed for changes in RHOT1, RHOT2, TRAK2, or Kif5B mRNA levels by quantitative PCR (C) or protein expression by Western blotting (D). Means ± SD are shown (n = 3). ***, P < 0.0001.

To test the functional implication(s) of this response, we next individually silenced the expression of RHOT1, RHOT2, TRAK2, or Kif5B in PC3 cells (Fig. 4A) and looked at changes in mitochondrial trafficking. Knockdown of each of these molecules all independently inhibited mitochondrial accumulation at the cortical cytoskeleton of PC3 cells, compared to control siRNA transfectants (Fig. 4B and C). Reciprocally, reconstitution of Myc knockdown cells with RHOT1, RHOT2, or Kif5B cDNA (Fig. 4D) restored the accumulation of mitochondria at the cortical cytoskeleton, indistinguishably from control transfectants (Fig. 4E). Quantitative analysis of mitochondrial motility under these conditions demonstrated that reexpression of RHOT1, RHOT2, or Kif5B normalized the speed of mitochondrial movements as well as the distance traveled by each mitochondrion in Myc knockdown cells (Fig. 4F and G).

FIG 4.

Myc regulation of mitochondrial trafficking. (A) PC3 cells transfected with siCtrl or individual siRNA sequences targeting RHOT1, RHOT2, TRAK2, or Kif5B were analyzed by Western blotting. (B and C) PC3 cells transfected with siCtrl or individual siRNAs targeting RHOT1, RHOT2, or Kif5B were labeled as indicated, and mitochondrial accumulation at the cortical cytoskeleton was imaged by confocal fluorescence microscopy (B) and quantified (C). 3D isosurface renderings of representative images are shown. Scale bars, 20 μm. Each symbol corresponds to an individual determination. (siCtrl, n = 43; siRHOT1, n = 43; siRHOT2, n = 39; siKif5B, n = 32). ***, P < 0.0001. (D) PC3 cells transfected with siCtrl or siMyc were reconstituted with individual cDNAs encoding RHOT1, RHOT2, or Kif5B and analyzed by quantitative PCR. Means ± SD are shown (n = 3). (E) Experiments were performed under the reconstitution conditions described above for panel D, and transfected PC3 cells were quantified for mitochondrial accumulation at the cortical cytoskeleton by confocal fluorescence microscopy. Each symbol corresponds to an individual determination (siCtrl, n = 27; siMyc, n = 30; RHOT1, n = 32; RHOT2, n = 33; Kif5B, n = 23). ***, P < 0.0001; *, P = 0.02; ns, not significant. (F and G) Experiments were performed under the reconstitution conditions described above for panel E, and transfected PC3 cells were analyzed for speed of mitochondrial movements (F) and total distance traveled by individual mitochondria (G). Each symbol corresponds to an individual determination (Ctrl, n = 148; Myc, n = 132; RHOT1, n = 32; RHOT2, n = 35; Kif5B, n = 35; Drp1, n = 97). ***, P < 0.0001.

Myc-mitochondrial signaling enables tumor cell motility.

Recent studies have shown that mitochondrial trafficking to the cortical cytoskeleton fuels tumor cell movements (20). Consistent with this possibility, Myc silencing suppressed focal adhesion (FA) complex dynamics in PC3 cells (Fig. 5A), a requisite of cellular motility (14), increasing the fraction of stable FA complexes and reducing the formation of new FA complexes, compared to control transfectants (Fig. 5B). Quantitative analysis of FA parameters in these experiments demonstrated that Myc silencing decreased the velocity of FA assembly/disassembly as well as the length of FA complexes (Fig. 5C). siRNA silencing of the Myc-regulated mitochondrial trafficking molecule TRAK2 or Kif5B gave similar results, decreasing the formation of new and decayed FA complexes and significantly increasing the fraction of stable FA complexes (Fig. 5D), consistent with a requirement of mitochondrial trafficking to fuel membrane FA dynamics (23).

FIG 5.

Myc control of tumor chemotaxis. (A) PC3 cells transfected with siCtrl or siMyc were labeled with talin-RFP and analyzed for focal adhesion (FA) complex dynamics by time-lapse videomicroscopy. Representative merged frames at 0 h and 2 h are shown (n = 2). (B) PC3 cells transfected as described above for panel A were analyzed for FA complex dynamics, and the percentage of new, stable, or decayed FA complexes was quantified. Means ± SD are shown (n = 10 to 15). **, P = 0.006; ***, P = 0.0005 to <0.0001. (C) Experiments were performed under the conditions described above for panel A, and the speed (top) and length (bottom) of FA complexes in siRNA-transfected PC3 cells was quantified (siCtrl, n = 891; siMyc, n = 380). ***, P < 0.0001. (D) PC3 cells transfected with siCtrl or siTRAK2 (top) or siKif5B (bottom) were labeled with talin-RFP and analyzed for FA complex dynamics by time-lapse videomicroscopy as described above for panel A. Means ± SD are shown (siTRAK2, n = 5 to 9; siKif5B, n = 9 to 19). *, P = 0.02; ***, P < 0.0001. (E) PC3 cells were transfected with siCtrl or two independent siRNA sequences targeting Myc (siMyc #1 and siMyc #2) and analyzed by Western blotting. (F and G) PC3 cells were transfected as described above for panel A and analyzed for 2D cell motility by time-lapse videomicroscopy (F), with quantification of the speed of cell movements and total distance traveled by individual cells (G). Each tracing corresponds to the movements of an individual cell (n = 19 or 20). The cutoff velocities for slow-moving (blue) or fast-moving (magenta) cells are indicated. Data are from a representative experiment (n = 3). ***, P < 0.0001. (H) PC3 cells were transfected with siCtrl or siMyc and analyzed for chemotactic cell motility with quantification of the forward migration index. Each symbol corresponds to an individual cell. Data are from a representative experiment. Arrows, direction of the chemotactic gradient. (I) PC3 cells transfected as described above for panel A were analyzed by Western blotting. p, phosphorylated.

To determine whether this pathway was important for tumor cell movements, we next silenced Myc expression in PC3 cells using two independent siRNA sequences (Fig. 5E). In these experiments, Myc knockdown suppressed the two-dimensional (2D) motility of PC3 cells (Fig. 5F), with nearly complete inhibition of the speed of cell movements and distance traveled by individual cells (Fig. 5G). To test whether the loss of Myc affected directional cell movements, additional studies were carried out in a chemotaxis chamber. In these experiments, PC3 cells transfected with control siRNA (siCtrl) migrated along a chemotactic gradient (Fig. 5H). Conversely, Myc-silenced cells exhibited random cell migration in the same settings (Fig. 5H). Biochemically, Myc knockdown was accompanied by decreased phosphorylation of cell motility kinases, focal adhesion kinase (FAK) (Tyr925 and Tyr397) and Src (Tyr416), compared to control siRNA transfectants (Fig. 5I).

Consistent with these results, Myc silencing inhibited the directional migration of PC3 cells, as quantified in a wound closure assay (Fig. 6A and B). To rule out potential effects of Myc silencing on tumor cell proliferation, similar experiments were carried out in the presence of the cell cycle inhibitor mitomycin C. Myc silencing in these settings still inhibited directional tumor cell motility in a wound closure assay, compared to mitomycin C-treated control transfectants (Fig. 6C). In addition, Myc knockdown suppressed tumor cell invasion across Matrigel-coated inserts (Fig. 6D and E) as well as in three-dimensional (3D) spheroids embedded in a collagen matrix (Fig. 6F and G). Conditional modulation of N-Myc expression in neuroblastoma cells gave similar results, as Dox-mediated silencing of N-Myc in Shep21 cells inhibited Matrigel invasion (Fig. 6H), whereas N-Myc induction by 4-hydroxytamoxifen (4OHT) enhanced Shep cell invasion (Fig. 6I).

FIG 6.

Myc regulation of tumor cell invasion. (A to C) PC3 cells transfected with siCtrl or siMyc were analyzed for directional cell migration in the presence or absence of mitomycin C in a wound closure assay (A), and the area covered by cell migration in the absence (−Mito C) (B) or presence (+Mito C) (C) of mitomycin C was quantified. Representative images at the indicated time point are shown. BAF, binary area fraction (n = 3). (D and E) PC3 cells transfected with siCtrl or two independent Myc-directed siRNA sequences (siMyc #1 or siMyc #2) were analyzed for invasion across Matrigel-coated Transwell inserts (D) and quantified (E). Representative images of DAPI-stained nuclei of invaded cells are shown (n = 3). Each symbol corresponds to an individual determination (siCtrl, n = 18; siMyc #1, n = 16; siMyc #2, n = 14). ***, P < 0.0001. (F and G) PC3 cells transfected with siCtrl or siMyc were embedded in 3D organotypic spheroids in a collagen matrix (F), and cell invasion was quantified after 1 to 3 days by phase-contrast microscopy (G). Representative images are shown. Red contour, mask-inverted images used to quantify the length and area between the core and the invasive edge; AU, arbitrary units. ***, P < 0.0001. (H) Neuroblastoma Shep21 cells containing a doxycycline (Dox)-regulated Myc transgene (Dox-off) were analyzed for Matrigel invasion in the presence (+) or absence (−) of Dox. Each symbol corresponds to an individual determination (n = 5). *, P = 0.02. (I) Neuroblastoma SHEP-NMycER cells expressing a 4OHT-inducible Myc transgene were analyzed for Matrigel invasion in the presence (+) or absence (−) of 4OHT. Each symbol corresponds to an individual determination (n = 11). **, P = 0.008.

Myc-regulated mitochondrial trafficking controls tumor cell invasion and metastasis.

Based on these data, we next asked if the Myc-regulated network of mitochondrial trafficking was important for tumor cell movements in vitro and in vivo. In a first series of experiments, we found that reconstitution of Myc knockdown PC3, DU145, or Shep21 cells with Drp1 cDNA was sufficient to restore Matrigel invasion to the levels of control transfectants (Fig. 7A and B). Similarly, reexpression of RHOT1, RHOT2, or Kif5B under the same conditions corrected the defect of 2D cell motility of Myc-silenced cultures by time-lapse videomicroscopy (Fig. 7C), restoring the speed of cell movements and the total distance traveled by individual cells (Fig. 7D). Conversely, reexpression of the mitochondrial trafficking molecule RHOT1 or Kif5B did not restore chemotactic cell migration in Myc knockdown cells (Fig. 7E), consistent with previously reported data showing that exaggerated stimulation of mitochondrial trafficking promotes random tumor cell motility even in the presence of a chemotactic gradient (23). Finally, these reconstituted cells exhibited normalized Matrigel invasion, quantitatively indistinguishable from that of control siRNA transfectants (Fig. 7F). In control experiments, siRNA silencing of RHOT1, RHOT2, TRAK2, or Kif5B independently inhibited Matrigel invasion in PC3 cells (Fig. 7G), whereas forced expression of RHOT2 or Kif5B further increased tumor cell invasion (Fig. 7H).

FIG 7.

Myc-mitochondrial trafficking regulation of tumor cell invasion. (A and B) PC3 or DU145 cells transfected with siCtrl or siMyc or neuroblastoma Shep21 cells containing a Dox-regulated Myc transgene (Dox-off) were reconstituted with Drp1 cDNA and analyzed for invasion across Matrigel-coated inserts (A) and quantified (B). Representative images of DAPI-stained nuclei of invaded cells are shown (n = 3). Each symbol corresponds to an individual determination (PC3, n = 30 [siCtrl], n = 19 [siMyc], and n = 16 [siMyc plus Drp1]; DU145, n = 13 [siCtrl], n = 16 [siMyc], and n = 18 [siMyc plus Drp1]; Shep21, n = 6 [siCtrl], n = 7 [siMyc], and n = 7 [siMyc plus Drp1]). ***, P < 0.0001; ns, not significant. (C and D) PC3 cells transfected with siCtrl or siMyc were reconstituted with cDNA encoding RHOT1, RHOT2, or Kif5B and analyzed for 2D cell motility by time-lapse videomicroscopy (C), with quantification of the speed of cell movements and total distance traveled by individual cells (D) (siCtrl, n = 21; siMyc, n = 21; RHOT1, n = 19; RHOT2, n = 22; Kif5B, n = 22). ***, P < 0.0001. (E) PC3 cells transfected with siMyc were reconstituted with the vector (top), RHOT1 (middle), or Kif5B (bottom) cDNA and analyzed for chemotactic cell motility with quantification of the forward migration index in rose plots. Arrows, direction of the chemotactic gradient. (F) Experiments were performed under the reconstitution conditions described above for panel C, and transfected PC3 cells were analyzed for Matrigel invasion. Each symbol corresponds to an individual determination (Ctrl, n = 21; Myc, n = 31; RHOT1, n = 11; RHOT2, n = 31; Kif5B, n = 19). ***, P = 0.0004 to <0.0001. (G) PC3 cells were transfected with siCtrl or siRNA targeting RHOT1, RHOT2, TRAK2, or Kif5B and analyzed for Matrigel invasion. Each symbol corresponds to an individual determination (siCtrl, n = 84; siRHOT1, n = 25; siRHOT2, n = 21; siTRAK2, n = 28; siKif5B, n = 23). ***, P = 0.0001 to 0.0002. (H) PC3 cells transfected with the vector or cDNA encoding RHOT2 or Kif5B were analyzed for Matrigel invasion. Each symbol corresponds to an individual determination (vector, n = 8; RHOT2, n = 12; Kif5B, n = 12). ***, P = 0.0001 to 0.0004.

Finally, we asked whether Myc regulation of mitochondrial trafficking and tumor cell motility was important for metastasis in vivo. For these experiments, we first established clones of melanoma Yumm 1.7 or PC3 cells carrying stable short hairpin RNA (shRNA) knockdown of Kif5B by Western blotting (Fig. 8A). Functionally, these Kif5B shRNA transfectants were nearly completely impaired in Matrigel invasion (Fig. 8B and C), consistent with the data described above (Fig. 7G). When analyzed in a syngeneic model of lung metastasis, Yumm 1.7 cells expressing control shRNA gave rise to mCherry-positive (mCherry⁺) foci of disseminated tumor cells (DTC) in the lungs by immunohistochemistry (IHC) (Fig. 8D and E). Conversely, stable Kif5B knockdown significantly inhibited the appearance of lung DTC (Fig. 8D and E). To independently validate these results, we examined a second model of localized and metastatic prostate cancer in the liver. When engrafted in immunocompromised animals, PC3 cells stably expressing Kif5B shRNA (shKif5B) formed superficial flank tumors, mostly comparable in size to those for control transfectants (Fig. 8F). Consistent with these data and the specificity of this pathway for tumor cell movements, loss of Kif5B did not significantly affect cell cycle transitions, compared to control shRNA transfectants (Fig. 8G). Intrasplenic injection of PC3 cells transfected with control shRNA produced extensive metastatic foci in the liver, as assessed histologically 11 days after engraftment (Fig. 8H to J). Conversely, PC3 cells stably expressing shKif5B showed reduced metastatic potential (Fig. 8H). Histological quantification of these data demonstrated that loss of Kif5B did not significantly affect the surface area of PC3 metastases in this model (Fig. 8I) but strongly reduced the number of metastatic foci in the liver (Fig. 8J).

FIG 8.

Myc-mitochondrial trafficking regulation of metastasis. (A) Yumm 1.7 or PC3 cells stably transfected with control nontargeting shRNA (shCtrl) or five independent shRNA sequences targeting Kif5B were analyzed by Western blotting. (B and C) Yumm 1.7 shRNA transfectants as described above for panel A were analyzed for Matrigel invasion (B) and quantified (C). Representative images of DAPI-stained nuclei of invaded cells are shown (n = 3). Each symbol corresponds to an individual determination (Ctrl, n = 27; shKif5B 1, n = 17; shKif5B 2, n = 15; shKif5B 3, n = 24). ***, P = 0.0004 to <0.0001. (D and E) Yumm 1.7 cells expressing mCherry and stably transfected as described above for panel A were engrafted subcutaneously (s.c.) on the flanks of syngeneic C57BL/6NCr mice, and lungs harvested after 1 to 3 weeks were analyzed by immunohistochemistry (D) with quantification of mCherry⁺ disseminated tumor cells (DTC) (E). Representative images are shown (n = 23 to 26). ***, P < 0.0001. (F) PC3 cells expressing shCtrl or Kif5B-directed shRNA (shKif5B) were engrafted s.c. in athymic nude mice, and superficial tumor growth was measured at the indicated time intervals with a caliper. Each line corresponds to data for an individual tumor. (G) PC3 cells as described above for panel F were analyzed for cell cycle transitions by propidium iodide staining and flow cytometry. The percentage of cells in each cell cycle phase is indicated. Means ± SD are shown. Data are from a representative experiment. (H) PC3 cells expressing shCtrl or shKif5B were injected into the spleen of immunocompromised mice, and livers were analyzed by hematoxylin and eosin staining and light microscopy after 11 days. White circles, metastatic foci. Representative images are shown. (I and J) Experiments were performed under the conditions described above for panel H, and the surface area (n = 49 to 60) (I) and number (n = 8 to 11) (J) of liver metastatic foci per microscopy field were quantified. **, P = 0.001; ns, not significant. (K) Schematic model for Myc transcriptional regulation of a mitochondrial trafficking network in tumor cell motility and invasion.

DISCUSSION

In this study, we have shown that oncogenic Myc transcriptionally controls a gene network of subcellular mitochondrial trafficking that comprises the atypical mitochondrial GTPases RHOT1 and RHOT2, the adapter protein TRAK2, the anterograde motor Kif5B, and an effector of mitochondrial fission, Drp1 (Fig. 8K). This Myc-regulated network enables the accumulation of mitochondria at the cortical cytoskeleton of tumor cells, fueling focal adhesion dynamics, chemotaxis and tumor cell invasion, and metastatic spreading in preclinical models in vivo.

Against the backdrop of a ubiquitous “Warburg effect” (32), where tumors shift their metabolism to glycolysis even if oxygen is present, the role of mitochondria in cancer has remained controversial (19), and whether a potential exploitation of mitochondrial biology reflects a general trait of malignancy or a tumor- or context-specific response(s) has not been clearly delineated. Here, the finding that remodeling of mitochondrial functions is the most common alteration downstream of a ubiquitous oncogene, i.e., Myc (2), demonstrates that mitochondria play a general role in cancer and constitute a hallmark of Myc-driven tumors. This conclusion is in line with an expanding role of Myc in affecting mitochondrial biology for tumor maintenance, including regulation of oxidative phosphorylation gene expression (8), reprogramming of glutamine metabolism (33), and fatty acid oxidation (34). However, the importance of mitochondrial dynamics (11), in particular subcellular mitochondrial trafficking (23), in Myc regulation of tumor cell movements has not been previously reported, and in fact, changes in mitochondrial biogenesis (35) and fusion/fission balance (36) downstream of Myc have rather been linked to inhibition of YAP/TAZ oncogenes (11).

Despite the importance of metastasis in the prognosis of cancer, a detailed molecular understanding of what fuels the ability of tumor cells to invade across basement membranes and disseminate to distant organs is still lacking (13). A role of Myc in this process has remained controversial in the literature (37), as oncogenic Myc has been linked to inhibition of tumor cell invasion via suppression of Jun N-terminal protein kinase (JNK) activity (38) or sensitization to transforming growth factor β (TGF-β) signaling (39). Conversely, the data presented here showing that Myc transcriptionally controls an integrated cellular machinery for mitochondrial movements, which in turn fuels cell motility and invasion (20), points to this pathway as a ubiquitous aspect of Myc-driven tumors, important for metastatic competence, in vivo. In this context, mitochondrial trafficking to the cortical cytoskeleton at the leading edge of migration has been proposed as an efficient, “spatiotemporal” energy source to fuel membrane lamellipodium dynamics, persistent phosphorylation of cell motility kinases, and heightened tumor chemotaxis and invasion (23–25).

The machinery of subcellular mitochondrial trafficking has been characterized in detail in neurons (31), where repositioning of mitochondria to specialized microdomains sustains axonal fitness and fuels highly energy-intensive processes, such as synaptic function, active growth cones, and branches (40). Although long considered “neuron specific” (40), there is now evidence that the same protein network is broadly exploited outside the central nervous system (CNS) to propel normal (41) and tumor (42) cell movements (43) (Fig. 8K). Accordingly, Myc-regulated RHOT2 (this study) is frequently upregulated in cancer (43), whereas syntaphilin (SNPH), an endogenous inhibitor of mitochondrial trafficking (40), is downregulated during tumor progression and may function as a “metastasis suppressor” in vivo (30).

How mitochondrial trafficking is regulated in tumors remains to be fully elucidated. In addition to the new mechanism of Myc-dependent transcriptional control (this study), posttranslational modifications, such as RHOT phosphorylation by PINK kinase (44) or a nondegradative step of SNPH ubiquitination by the E3 ligase CHIP (29), have been implicated in mitochondrial movements, including in tumors. In addition, stress stimuli of the tumor microenvironment, such as hypoxia (30), ROS production (30), or exposure to molecular therapy (23), are important drivers of subcellular mitochondrial trafficking, enhancing tumor cell motility to potentially “escape” a noxious, unfavorable ecosystem (45).

As a pivotal component of this pathway, Drp1 (46) was identified here as a novel transcriptional target of Myc in cancer (Fig. 8K). Structural studies have shown that Drp1 is the main effector of mitochondrial fragmentation, or fission through the assembly of ringlike structures at the organelle outer membrane (47). Recent time-lapse videomicroscopy studies have suggested that this pathway is important for mitochondrial motility, as smaller, fragmented mitochondria travel faster than elongated organelles along polymerized microtubules (43). Consistent with this model, Drp1-dependent mitochondrial fission has been associated with increased tumor chemotaxis and heightened metastatic dissemination in vivo (29, 48). In addition, Myc-dependent transcription may explain the overexpression of Drp1 frequently seen in cancer and its oncogenic role in mitogen-activated protein kinase (MAPK) (49)- and Ras (50)-dependent tumorigenesis, modulation of cell cycle transitions (51), and cancer stemness (52, 53).

In sum, we have shown that exploitation of mitochondrial trafficking is a novel hallmark of Myc-driven oncogenesis, enabling advanced disease traits of tumor cell invasion and metastasis. Building on a better understanding of the role of mitochondria in tumors (19), there has been renewed interest in targeting mitochondrial pathways for novel cancer therapeutics (54), and the mitochondrial trafficking network described here may expose new therapeutic vulnerabilities to antagonize metastatic spreading in Myc-driven tumors.

MATERIALS AND METHODS

Cells and cell culture.

Prostate adenocarcinoma PC3 and DU145 cells were obtained from the American Type Culture Collection (ATCC) (Manassas, VA) and maintained in culture according to the supplier’s specifications. The human Burkitt lymphoma P493-6 cell line was described previously (55). Clones of P493 cells containing a doxycycline (Dox)-regulated Myc transgene induced after Dox removal (Dox-off system) were described previously (26). These cells were maintained in RPMI 1640 medium with 10% fetal bovine serum (FBS) plus 1% streptomycin and penicillin. Treatment of P493-6 cells with 0.1 μg/ml Dox for 48 to 72 h led to a significant reduction of Myc expression. Neuroblastoma Shep21N and Shep21-NMycER cells containing a conditionally regulated N-Myc transgene were described previously (56). In these cells, treatment with 50 ng/ml Dox for 48 h suppresses N-Myc expression, whereas the addition of 4-hydroxytamoxifen (4OHT) (0.5 μg/ml) results in strong N-Myc induction. In addition, the neuroblastoma cell lines Kelly, NLF, and IMR5 were used (57). Yale University mouse melanoma (Yumm 1.7) cells isolated from a genetically engineered mouse model of melanoma with the genotype Braf^V600E; Cdkn2a^−/−; Pten^−/− were used (30). Conditioned media used for cell invasion assays was prepared from exponentially growing cultures of NIH 3T3 cells (ADCC) in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 4.5 g/liter d-glucose, sodium pyruvate, 10 mM HEPES, and 10% FBS for 48 h. Cell passaging was limited to <40 passages from receipt, and cell lines were authenticated by short tandem repeat (STR) profiling with an AmpFlSTR Identifiler PCR amplification kit (Life Technologies) at The Wistar Institute’s genomics facility. Mycoplasma-free cultures were confirmed at the beginning of the studies and every 2 months afterwards by PCR amplification of cultures using Bioo Scientific mycoplasma primer sets (catalog number 375501) and Hot Start polymerase (Qiagen).

Antibodies and reagents.

A rabbit monoclonal antibody to Myc was purchased from Abcam. Antibodies to Src, Tyr416-phosphorylated Src, focal adhesion kinase (FAK), and Tyr925- or Tyr397-phosphorylated FAK were obtained from Cell Signaling. Antibodies to β-actin, RHOT1, RHOT2, and TRAK2 were obtained from Santa Cruz and Sigma. An antibody to Kif5B was obtained from Abcam. Secondary antibodies for immunofluorescence analysis were obtained from Molecular Probes. Dox, β-estradiol, and 4OHT were purchased from Sigma.

Plasmid and siRNA transfection.

Gene knockdown experiments with small interfering RNA (siRNA) were carried out as described previously (58). The following siRNA sequences were used: a control ON-TARGETplus nontargeting siRNA pool (catalog number D-001810; Dharmacon), human Myc-directed siRNA (catalog number L-020010; Dharmacon), Kif5B siRNA (catalog number sc-36777; Santa Cruz), RHOT1 siRNA (catalog number sc-93809; Santa Cruz), and RHOT2 siRNA (catalog number sc-93029; Santa Cruz). The various tumor cell types were transfected with the individual siRNA pools at 40 nM in Lipofectamine RNAiMAX (Invitrogen) at a 1:1 ratio (20 μM siRNA–Lipofectamine RNAiMAX [vol/vol]). After 72 h, transfected cells were validated for target protein knockdown by Western blotting and processed for functional experiments.

Chromatin immunoprecipitation assay.

P493 cells treated with β-estradiol plus Dox for 48 h or PC3 cells were transfected with control nontargeting siRNA or Myc-directed siRNA for 72 h and used for ChIP experiments. Cells were cross-linked with formaldehyde, and the fragmented chromatin was immunoprecipitated using a rabbit monoclonal antibody to Myc (Ab32072). Nonbinding rabbit IgG was used as a control. The total input was the supernatant of each incubation reaction mixture without antibody treatment. Real-time PCR amplification of the precipitated chromatin fragments was performed using SYBR green master mix (Applied Biosystems) on an ABI7500 sequence detection system according to the manufacturer’s instructions.

Protein analysis.

Protein lysates were prepared in radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris [pH 8.0]) in the presence of an EDTA-free protease inhibitor cocktail (Roche) and a phosphatase inhibitor cocktail (Roche). Equal amounts of protein lysates were separated by SDS gel electrophoresis, transferred to polyvinylidene difluoride (PVDF) membranes, and incubated with primary antibodies of various specificities. Protein bands were visualized by chemiluminescence.

Immunofluorescence.

Cells were fixed in formalin–phosphate-buffered saline (PBS) (4% final concentration) (pH 7.4) for 15 min at 22°C, permeabilized in 0.1% Triton X-100–PBS for 5 min, washed, and incubated in 5% normal goat serum (NGS; Vector Laboratories) diluted in 0.3 M glycine–PBS for 60 min. Primary antibodies against β-tubulin (diluted 1:200), mitochondrial cytochrome c oxidase subunit II (MT-CO2) (diluted 1:500), or TOM20 (1:100) were added in 5% NGS–0.3 M glycine–PBS and incubated for 18 h at 4°C. After 3 washes in PBS, secondary antibodies conjugated to tetramethyl rhodamine isothiocyanate (TRITC) were diluted 1:500 in 5% NGS–0.3 M glycine–PBS and added to cells for 1 h at 22°C. Where indicated, F-actin was stained with phalloidin-Alexa 488 (1:200 dilution) for 30 min at 22°C. Slides were washed and mounted in 4′6-diamidino-2-phenylindole (DAPI)-containing Prolong Gold mounting medium (Invitrogen). To determine the positioning of mitochondria at the cortical cytoskeleton, about 45 to 48 cells were selected per sample.

Mitochondrial time-lapse videomicroscopy.

Cells (2 × 10⁴) growing on high-optical-quality glass-bottom 35-mm plates (MatTek Corporation) were incubated with 100 nM MitoTracker deep red FM dye for 1 h and imaged on a Leica TCS SP8 X inverted confocal laser scanning microscope using a 63×, 1.40-numerical-aperture (NA) oil objective. Short-duration time-lapse sequences were carried out using a Tokai Hit incubation chamber equilibrated to bidirectional scanning at 8,000 Hz at 37°C with 5% CO₂ using a resonant scanner. Time-lapse microscopy was performed for 1,000 s (10 s per frame). Individual 12-bit images were acquired using a white-light supercontinuum laser (2% at 645 nm) and hybrid detectors at a 2× digital zoom with a pixel size of 90 nm by 90 nm. A pinhole setting of 1 Airy unit provided a section thickness of 0.896 μm. Each time point was captured as a stack of approximately 11 overlapping sections with a step size of 0.5 μm. At least 5 single cells under each condition were collected for analysis. Initial postprocessing of the 3D sequences was carried out with Leica LAS X software to create an isosurface visualization. Images were imported into ImageJ Fiji, and individual mitochondria were manually tracked using the Manual Tracking plugin. Mitochondria (approximately 10 mitochondria per cell) were tracked along the stacks until a fusion event prevented continued tracking. The speed and distance for each time interval were used to calculate the mean speed and cumulative distance traveled by each individual mitochondrion.

Cortical mitochondrion quantification.

Mitochondrion/F-actin composite images were analyzed in ImageJ. The F-actin channel was used to manually label the cell boundary, and a belt extending from the boundary toward the inside of the cell was marked as the “cortical mask.” This cortical mask was subsequently applied to the mitochondrial channel to measure the intensity at the cortical region and normalized to the total mitochondrial intensity per cell and cell area.

Time-lapse videomicroscopy analysis of focal adhesion complex dynamics.

PC3 cells were transfected with control nontargeting siRNA, Myc-directed siRNA, or, alternatively, siRNA against TRAK2 or Kif5B for 24 h. The transfected cells were plated on high-optical-quality 35-mm glass-bottom plates and transduced with talin-red fluorescent protein (RFP) BacMam virus for 18 h. Time-lapse videomicroscopy was carried out using a Leica TCS SP8 confocal laser scanning microscope system with an HCX PL APO CS 63×, 1.40-NA oil UV objective. Acquisition of live cells using integrated Leica LAS software was performed every 3 min per frame for a total interval of 2 h. Sequences were imported into ImageJ for further analysis. The initial and final frames were duplicated and assembled as composite images. FA complexes were manually counted and classified (according to the presence in some or all the time frames) into three groups: decaying, newly formed, and stable (merged areas). The analysis was carried out with 7 cells (about 150 FA complexes) under each condition in 2 independent time-lapse experiments using the LASX software package.

mRNA quantification.

mRNA levels under the various conditions tested were determined by quantitative PCR (qPCR). For these experiments, RNA was extracted with a PureLink RNA minikit (Life Technologies) according to an in-column DNA digestion protocol. One microgram of RNA was reverse transcribed using a ThermoScript reverse transcription-PCR (RT-PCR) system (Life Technologies). One microliter of cDNA (1:5 dilution) was used as the template for qPCRs using TaqMan gene expression assays. Predesigned TaqMan assays were as follows: HS00153408 for human Myc, 4352930E for eukaryotic 18S rRNA, HS00232074 for N-Myc, HS00430256 for RHOT1, HS00969181 for RHOT2, HS00209308 for TRAK2, and HS01037194 for Kif5B. The ΔΔC_T method (where C_T is threshold cycle) was used to determine the fold changes in levels of expression of individual mRNAs.

Tumor cell motility and 2D chemotaxis.

For 2D motility studies, PC3 cells (2 × 10⁴) were seeded in 4-well Ph⁺ chambers (Ibidi) in complete medium and allowed to attach for 16 h at 37°C. For 2D chemotaxis, PC3 cells (5 × 10⁴) were placed in 8-μm chemotaxis slides (Ibidi) and allowed to attach for 16 h at 37°C. Time-lapse videomicroscopy was performed over 10 h, with a time-lapse interval of 10 min, as described previously (58). Stacks were imported into ImageJ Fiji software for analysis, and at least 10 to 20 cells under each condition were tracked using the Manual Tracking plugin for ImageJ Fiji. Tracking data were exported into Chemotaxis and Migration Tool v. 2.0 (Ibidi) for graphing and calculation of means and standard deviations (SD) of speed, accumulated distance of movement, and the forward migration index (FMI). For analysis of directional cell migration using a wound closure assay, PC3 cells were transfected with control nontargeting siRNA or Myc-directed siRNA in the presence or absence of mitomycin C (16 h), “wounds” in the cell monolayer were made with a 10-μl pipette tip, cell debris was washed off, and cultures were maintained in complete medium containing 10% FBS at 37°C with 5% CO₂ for 24 h. Time-lapse imaging of migrating cells was performed using a TE300 inverted microscope (Nikon) equipped with an incubator set at 37°C with 5% CO₂ and 95% relative humidity. Each image was acquired using a 10× objective of the same fields at each 10-min interval for a total of 24 h.

Tumor cell invasion.

Tumor cell invasion experiments were carried out essentially as described previously (43), using growth factor-reduced Matrigel-coated 8-μm PET Transwell chambers (Corning). Tumor cell types were seeded in duplicates onto the coated Transwell filters (1 × 10⁵ cells/well) in medium containing 0.1% bovine serum albumin (BSA), and conditioned medium from NIH 3T3 fibroblasts was placed in the lower chamber as a chemoattractant. Cells were allowed to invade for 16 h, noninvading cells were scraped off the topside of the membranes, and invasive cells on the Transwell insert were fixed in methanol. Membranes were mounted in medium containing DAPI (Vector Laboratories) and analyzed by fluorescence microscopy. Five random fields at a ×10 magnification were collected for each membrane. Digital images were batch imported into ImageJ Fiji, thresholded, and analyzed with the Analyze Particles function.

For analysis of tumor cell invasion in 3D organotypic spheroids, PC3 cells (5 × 10⁴) were transfected with control nontargeting siRNA or Myc-directed siRNA and seeded onto 96-well plates coated with 1.5% agar (Difco Noble agar) in PBS (pH 7.4). PC3 spheroids were allowed to form over a 72-h period and then embedded in 600 μl of bovine collagen type I (Organogenesis) in 24-well plates. Spheroids were overlaid with 1 ml of growth medium, and samples were imaged using a T2000 inverted microscope and analyzed using NIS elements software. The analysis was performed on 10 to 12 spheroids per sample in two independent experiments.

Animal studies.

Studies involving vertebrate animals (rodents) were carried out in accordance with the Guide for the Care and Use of Laboratory Animals (59). Protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of The Wistar Institute (protocol numbers 112625 and 112610). For a syngeneic model of metastasis, Yumm 1.7 cells expressing mCherry (30) were stably transfected with control nontargeting shRNA or Kif5B-directed shRNA and selected in puromycin-containing growth medium (2 μg/ml) for 15 days at 37°C. Transfected cells (3.5 × 10⁵) were injected into the flanks of syngeneic 8-week-old male C57BL/6NCr mice (NCI inbred mice; Charles River strain code 556). Ten days later, disseminated tumor cells (DTC) in the lungs (five lungs per group) were identified and quantified based on the expression of the mCherry transgene by immunohistochemistry (IHC) using a Nikon i80 upright microscope. For each animal, the average number of mCherry⁺ cells per lung was calculated and represented. For a liver metastasis model, 6- to 8-week-old severe combined immunodeficient (SCID)/beige mice (three mice under each experimental condition) were anesthetized with ketamine hydrochloride, the abdominal cavity was exposed by laparotomy, and PC3 cells (1 × 10⁶) transfected with control nontargeting shRNA or Kif5B shRNA were injected into the spleen. Spleens were removed the first day after injection to minimize potentially confounding effects on metastasis due to variable growth of primary tumors. Animals were sacrificed after 11 days, and their livers were resected, fixed in formalin, and paraffin embedded. Serial liver sections 500 μm apart (n = 15 under each condition) were stained with hematoxylin and eosin and analyzed using NIS elements software. Metastatic foci were counted manually, and the surface area of the foci was determined using NIS elements.

Immunohistochemistry.

Lungs were fixed in neutral formalin (catalog number SF93-4; Fisher Scientific) for 36 h, transferred to 70% ethanol for 3 days, and paraffin embedded. Five-micrometer-thick tissue slides were warmed at 50°C for 30 min, deparaffinized in xylene for 20 min and then xylene-ethanol (1:1) for 5 min, and rehydrated in an alcohol series (100%, 95%, 90%, 70%, 50%, and 30% ethanol and distilled water [dH₂O], for 5 min each). Antigen retrieval was done in a citrate-based solution (catalog number H-3300; Vector Laboratories) in a pressure cooker for 5 min (pH 6.0), followed by cooling to room temperature (RT). After washes, slides were treated with 3% hydrogen peroxide for 20 min, washed in PBS (pH 7.4), for 5 min, and blocked in 10% normal goat serum–PBS for 1 h at RT. A primary rabbit polyclonal antibody to mCherry (catalog number NBP2-25157; Novus) was diluted 1:400 in 10% normal goat serum–PBS and incubated in a humidified chamber for 16 h. After three washes in PBS (pH 7.4) for 5 min each, slides were incubated with an anti-rabbit horseradish peroxidase (HRP)-labeled polymer (catalog number K4002; Dako) for 30 min at RT, washed, and developed with a DAB⁺ substrate chromogen system (catalog number K3467; Dako) for 30 min, followed by staining with Mayer’s hematoxylin solution (catalog number MHS16; Sigma) for 10 s. Slides were sequentially dehydrated in graded alcohol, immersed in xylene for 15 min, and mounted with Permount mounting medium (catalog number SP15-100; Fisher Scientific).

Bioinformatics analysis.

RNA-Seq data were aligned using the STAR algorithm (60), and RSEM v1.2.12 software (61) was used to estimate read counts and reads per kilobase per million (RPKM) values on the gene level using the hg19 genome version with Ensemble transcriptome information. Deseq2 (62) was used to estimate the significance of differential expression differences between two experimental groups, and genes that passed an FDR of <10% and a >2-fold threshold were called significant. Enrichment of UniProt classification keywords (“UP_KEYWORDS” category) was done using DAVID software (63), and results with Bonferroni-corrected P values of <10⁻⁵ are reported. Pathway enrichment analysis was done using Qiagen’s Ingenuity Pathway Analysis (IPA) software (Qiagen, Redwood City, CA), using the “canonical pathways” option. Enriched pathways that passed an FDR of <5% and had a significantly predicted inhibited state (Z of less than −3) were considered. A list of all affected mitochondrial functions was derived from IPA. Only functions with at least 10% significantly affected genes were shown on a combined diagram. As most functions shared at least 50% of genes, genes unique to the functions were indicated on the diagram. ChIP-Seq data for cMyc in P493 Burkitt lymphoma cells (26) at three independent time points (t = 0 h, t = 1 h, and t = 24 h) after Dox removal were derived from the data set under NCBI GEO accession number GSE36354. ChIP-Seq analysis of the N-Myc distribution in neuroblastoma BE2C, Kelly, and NGP cells (64) was performed using the data set under GEO accession number GSE36354. Data were aligned using bowtie (65) against the hg19 genome version and analyzed using the HOMER algorithm (66) to generate bigwig files used for visualizing ChIP-Seq tracks in the University of California—Santa Cruz (UCSC) browser.

Statistical analysis.

Data are expressed as means ± SD of results from multiple independent experiments or replicates of representative experiments out of a minimum of two or three independent determinations. Two-tailed Student’s t test or a Wilcoxon rank sum test was used for two-group comparative analyses. For multiple-group comparisons, analysis of variance (ANOVA) or a Kruskal-Wallis test with Bonferroni’s post hoc procedure was applied. All statistical analyses were performed using the GraphPad software package (Prism 6.0) for Windows. A P value of <0.05 was considered statistically significant.