Abstract
Metabolic flexibility is a hallmark of cancer. Although many tumors preferentially use glycolysis in the presence of oxygen for bioenergetic purposes (Warburg effect), the effects of glycolytic metabolism on tumor metastasis have not been investigated. We have employed an extreme model of glycolytic metabolism to investigate the ability of metastatic B16 mouse melanoma cells to grow as primary subcutaneous tumors and to form lung tumors when injected intravenously into syngeneic and immunocompromised mice. Mitochondrial gene-knockout B16ρ° cells showed delayed subcutaneous tumor growth and, surprisingly, failed to form lung tumors. The results suggest that mitochondrial reactive oxygen species (ROS) may be required for tumor metastasis.
Introduction
Results
B16 melanoma cells devoid of mitochondrial DNA exhibit delayed subcutaneous tumor growth and fail to form lung tumors
B16ρ° melanoma cells were generated as described previously. 4 The absence of mitochondrial DNA was verified by the polymerase chain reaction (PCR) by failure to amplify mitochondrially encoded cytochrome b and by insensitivity of reduction of the tetrazolium salt WST-1 in the presence of 1-methoxy-5-methylphenazinium methyl sulfate (mPMS), to the uncoupler carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP). 8 These cells have been maintained in our laboratory as a stable cell line devoid of mitochondrial DNA for more that a year. B16ρ° cells have a doubling time of 36 days compared with 17.5 days for parental B16 cells when grown in serum-containing RPMI-1640 tissue culture medium under normoxic conditions. In contrast, B16ρ° cells grown under hypoxic conditions (0.5% oxygen) have a doubling time of 41.5 days compared with 48 days for parental B16 cells. When injected subcutaneously into syngeneic C57BL/6 mice, B16ρ° cells (2 × 105 cells in 0.2 mL) showed delayed growth (29 days to appearance of first palpable tumors) compared with 9 days for parental B16 melanoma cells (Table 1). B16ρ° tumors grew at about half the rate of tumors from parental B16 cells, consistent with the respective in vitro growth rates of the cells. Following intravenous injection (2 × 105 cells in 0.2 mL), B16ρ° cells failed to form lung tumors, even after a prolonged period of observation (55 days).
16ρ°
B16 and B16ρ° tumor cells (2 × 105 cells in 0.2 mL) were injected subcutaneously or intravenously into groups of 5 C57BL/6 or NOD/SCID mice. Time to tumor initiation and tumor growth was measured following subcutaneous injection, whereas mice were killed at selected time points following intravenous injection and lung weight measured or black tumor foci counted.
i.v., Intravenous; NOD/SCID, nonobese diabetic/severe combined immunodeficiency.
To investigate whether immune responses may have contributed to the delayed tumor growth observed with B16ρ° cells and the failure of these cells to form lung tumors, additional experiments were carried out with immunocompromised nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice. Similar results were obtained, except that B16ρ° tumor growth was further delayed in this model with a 40-day delay to the first palpable subcutaneous tumors. Notably, B16ρ° cells again failed to form lung tumors in NOD/SCID mice even after 64 days.
In vitro growth characteristics of B16ρ° cells
Interestingly, B16ρ° cells, showed distinct changes in tumorigenic phenotype in vitro, including loss of the ability to grow as anchorage-independent colonies in a double-layer, soft agar culture system and failure to grow as spheres in serum-free culture medium supplemented with basic fibroblast growth factor and epidermal growth factor. Under these cell culture conditions, parental B16 cells formed numerous large compact colonies in the double-layer, agar culture system and grew as small loosely packed spherical colonies in serum-free culture. Similar in vitro results were obtained with human LN18ρ° glioblastoma cells, but parental LN18 cells are poorly tumorigenic in NOD/SCID mice precluding in vivo analysis. These results indicate that B16ρ° cells are incapable of surviving under anchorage-independent growth conditions and suggest that loss of mitochondrial respiratory function and the resulting loss of ability to generate optimal levels of ROS may be associated with anchorage-independent programmed cell death, or anoikis. Recently, failure to sustain constitutive deregulated ROS production in metastatic PC3 prostate cancer cells was shown to restore cell sensitivity to anoikis, indicating a role for ROS in sustained survival of these cells and implicating ROS production as an emergent antitumor target. 9,10
ROS production by B16ρ° cells
The absence of mitochondrial DNA in B16ρ° cells is associated with the absence of mitochondrial respiratory function due to defects in respiratory complexes I, III, IV, and V. Because complexes I and III are the main sources of cellular ROS, B16ρ° cells would be expected to be severely deficient in both mitochondrial and cytosolic ROS. This was verified by staining B16 cells with MitoSox, a probe for mitochondrial superoxide production. Both B16ρ° cells and HeLaρ° cells stained poorly with MitoSox relative to parental B16 and HeLa cells, indicating that these cells were severely deficient in mitochondrial superoxide production. Similarly, B16ρ° cells and all other ρ° cells tested, including HeLa, HL60, and LN18 cells, showed intracellular ROS levels 40–60% below controls as determined by dichlorofluoroscene (DCF) fluorescence. To determine whether ROS production by B16 cells may be associated with anoikis resistance, cells were treated with the ROS scavenger, N-acetylcysteine (NAC, 10 mM) for 3 days in complete culture medium containing serum. Cells treated with NAC failed to survive this treatment, suggesting that ROS production is important for the resistance of these cells to anoikis.
Discussion
The results presented here show for the first time that mitochondrial electron transport is not mandatory for primary subcutaneous tumor formation in a metastatic B16 melanoma model. Furthermore, and contrary to expectation, B16ρ° cells failed to form lung tumors when injected intravenously into either syngeneic C57BL/6 or immunocompromised NOD/SCID recipient mice.
With syngeneic C57BL/6 mice, B16ρ° tumor formation was delayed by 20 days compared with parental tumors, but once initiated, tumor growth closely reflected the doubling time in in vitro culture. The reason for the 20-day delay in tumor appearance is not altogether clear but may relate to the time taken to adjust to local micoenvironmental conditions. The fact that an even greater delay in tumor appearance was observed in NOD/SCID mice (31 days) precludes a simple immunological explanation for the delay and suggests that it may be ROS-related and that a longer time is required for tumor cell adjustment in immunocompromised mice. These results may also address the dilemma raised recently over the use of immunocompromised NOD/SCID mice to quantitate tumor stem cells in human melanomas. 11 In our experience, syngeneic C57BL/6 and allogeneic immunocompromised NOD/SCID mice were comparable in their ability to grow B16 tumors, but differed in the time taken to establish glycolytic ρ° tumors.
B16ρ° cells injected intravenously into C57BL/6 or NOD/SCID mice failed to form lung tumors. Although it cannot be excluded that immune mechanisms may play a role in this loss of lung seeding function, our results support a mechanism involving a requirement for ROS signaling in extravasation and/or seeding and growth in the lung microenvironment. It is also possible that B16ρ° cells may have altered expression of tissue-associated cellular adhesive molecules, or that B16ρ° cells may seed in the lung but fail to engage with the local microenvironment in a manner that supports their proliferation. These alternative explanations are presently under investigation.
Finally, if the results presented here can be extended to other metastatic tumor models, it may be possible to develop antimetastatic strategies by compromising mitochondrial energy production. Mitochondrial targeting as a cancer treatment strategy has been the subject of several recent reviews, 12,13 raising the possibility it may be worthwhile investigating redox-active mitochondria-targeting molecules such as MitoQ and SkQ1 in metastatic cancer.
Footnotes
Acknowledgments
This research was supported by a development grant from the Malaghan Institute of Medical Research.