|Home | About | Journals | Submit | Contact Us | Français|
Autophagy is a survival mechanism activated in response to metabolic stress. In normal tissues autophagy plays a major role in energy homeostasis through catabolic self-digestion of damaged proteins and organelles. Contrary to its survival function, autophagy defects are implicated in tumorigenesis suggesting that autophagy is a tumor suppression mechanism. Although the exact mechanism of this tumor suppressor function is not known, it likely involves mitigation of cellular damage leading to chromosomal instability. The complex role of functional autophagy in tumors calls for model systems that allow the assessment of autophagy status, stress management and the impact on oncogenesis both in vitro as well as in vivo. We developed model systems that involve generation of genetically defined, isogenic and immortal epithelial cells from different tissue types that are applicable to both wild-type and mutant mice. This permits the study of tissue- as well as gene-specific tumor promoting functions. We successfully employed this strategy to generate isogenic, immortal epithelial cell lines from wild-type and mutant mice deficient in essential autophagy genes such as beclin 1 (beclin 1+/-) and atg5 (atg 5-/-). As these cell lines are amenable to further genetic manipulation, they allowed us to generate cell lines with apoptosis defects and stable expression of the autophagy marker EGFP-LC3 that facilitate in vitro and in vivo assessment of stress-mediated autophagy induction. We applied this model system to directly monitor autophagy in cells and 3D-morphogenesis in vitro as well as in tumor allografts in vivo. Using this model system we demonstrated that autophagy is a survival response in solid tumors that co-localizes with hypoxic regions, allowing tolerance to metabolic stress. Furthermore, our studies have established that autophagy also protects tumor cells from genome damage and limits cell death and inflammation as possible means to tumor suppression. Additionally these cell lines provide an efficient way to perform biochemical analyses, and high throughput screening for modulators of autophagy for potential use in cancer therapy and prevention.
Hypoxia is a common occurrence in human solid tumors. Hypoxic regions, largely due to inefficient vasculature and rapid tumor growth, may influence tumor progression and negatively affect clinical outcome, as they are implicated in resistance to therapy (Barlogie et al., 1982; Folkman, 2003). Metabolic stress often triggers apoptotic cell death that results in cancer cell elimination, and hence mutations in the apoptotic pathways are common in human cancers (Nelson et al., 2004). For cancer cells to better survive metabolic stress, this apoptotic resistance must also be accompanied by activation of alternative pathways supporting cell survival under stress. One such mechanism is the up-regulation of the transcription factor hypoxia-inducible factor 1-α (Hif1-α) to promote metabolic adaptation and angiogenesis (Dang et al., 2008; Semenza, 2003). Another mechanism involves activation of the catabolic pathway of autophagy to facilitate cell survival (Mathew et al., 2007a). Elucidating the molecular intricacies of the pathways will hopefully lead to the development of targeted therapeutic strategies, and thus may have profound implications for cancer therapy.
Autophagy (i.e., macroautophagy) is a response to stress and starvation whereby cellular organelles and proteins are sequestered and targeted for lysosomal degradation as an alternate energy source (Levine and Kroemer, 2008). The role of autophagy as a survival mechanism under metabolic stress is well documented (Mizushima, 2007). Immortalized baby mouse kidney epithelial (iBMK) cells rendered autophagy-deficient by allelic loss of beclin 1 or atg5 deficiency display increased susceptibility to metabolic stress (Degenhardt et al., 2006; Mathew et al., 2007b). Similarly, autophagy mitigates metabolic stress in immortalized mouse mammary epithelial cells (iMMECs) (Karantza-Wadsworth et al., 2007) and promotes survival of immortalized, nontumorigenic human mammary epithelial cell lines (MCF10A) and primary human mammary cells during extracellular matrix detachment (anoikis) (Debnath, 2008; Fung et al., 2008). In apoptosis-competent cells, autophagy delays apoptotic death under metabolic stress, whereas apoptosis defects unmask autophagy-mediated cell survival (Mathew et al., 2007a). Moreover, atg5−/− mouse embryonic fibroblasts (MEFs) show signs of ATP depletion, and atg5−/− mice do not survive neonatal starvation, suggesting that autophagy promotes survival under metabolic stress during mammalian development (Kuma et al., 2004; Lum et al., 2005). Importantly, in solid tumors autophagy localizes to regions of metabolic stress, suggesting that it may be exploited by cancer cells for survival (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Mathew et al., 2007b). Thus, the functional status of autophagy is an important determinant of tumor cell response to metabolic stress.
Although autophagy induction under metabolic stress is well established as a survival strategy, interactions between metabolic stress and defective autophagy are more complex. Intuitively contradictory to the survival function of autophagy under metabolic stress, defects in autophagy are associated with increased tumorigenicity in mice as well as humans. Allelic loss of beclin 1 is frequently observed in human breast, ovarian, and prostate cancers (Aita et al., 1999; Liang et al., 1999), and beclin 1+/− and atg4C−/− mice are tumor-prone, suggesting that autophagy is a tumor suppression mechanism (Marino et al., 2007). Moreover, growth factor– and nutrient-driven oncogenic pathways, such as the PI3-kinase pathway, inhibit autophagy, whereas inhibitors of this pathway, such as the tumor suppressor protein PTEN, activate autophagy (Arico et al., 2001). Although a clear understanding of autophagy-mediated tumor suppression is only beginning to emerge, one of the likely mechanisms by which autophagy inhibits tumorigenesis is suppression of necrotic cell death (Degenhardt et al., 2006). Impairment of autophagy by monoallelic deletion of beclin 1, RNAi-mediated knockdown of beclin 1 or atg5, or constitutive activation of Akt, induces necrotic cell death when apoptosis is blocked (Degenhardt et al., 2006). In tumors in vivo, this necrosis is associated with inflammation, activation of the cytokine-responsive NF-kB pathway and tumor progression (Degenhardt et al., 2006). Remarkably, autophagy defects in mouse liver cause excessive hepatocyte cell death, steatohepatitis and hepatocellular carcinoma (HCC) suggesting that support of cell survival and suppression of inflammation may be important autophagy functions in normal tissues, as well as tumors (Komatsu et al., 2007).
Another insight into the role of autophagy in tumor suppression came from the discovery that immortalized mouse epithelial cell lines with autophagy defects show signs of genome damage, which is exacerbated under metabolic stress. iBMK and iMMEC cells rendered autophagy-defective by beclin 1 monoallelic loss or atg5 deletion display activation of the DNA damage response, gene amplification, and accelerated progression to aneuploidy (Karantza-Wadsworth et al., 2007; Mathew et al., 2007a,b). These phenotypes are accentuated in an apoptosis-defective background, together suggesting that autophagy functions to limit chromosomal instability, preferentially manifested in cells with checkpoint and apoptosis defects. Thus, autophagy-mediated housekeeping and mitigation of genome damage play important roles in the cellular response to metabolic stress and in tumorigenesis.
However, the exact mechanism by which autophagy suppresses tumorigenesis is not known. To further investigate the role of metabolic stress–induced autophagy in tumorigenesis, we developed an in vitro metabolic stress assay that combines hypoxia (defined gas mixture composed of 1% O2, 5% CO2, and 94% N2) with glucose deprivation, thus mimicking metabolic stress in the tumor microenvironment in vivo (Nelson et al., 2004). Immortalized epithelial cell model systems, such as iBMK cells (Mathew et al., 2008) and iMMECs (Karantza-Wadsworth and White, 2008), have several advantages over conventional MEF or human cancer cell lines widely used to model human cancer (Mathew et al., 2008). Being epithelial in origin, they provide a superior representation of human tumor cell physiology compared to MEFs, can be generated from different tissues from any mouse that survives to birth (for isolation of kidney, prostate, liver, and lung tissue) and to young adulthood (for mammary gland isolation), are immortalized by well-defined genetic events, are amenable to additional genetic manipulation, and can be used for the generation of tumor allografts. Immortalized epithelial cell lines derived from wild-type and mutant mice extend the utility of mouse models by enabling biochemical and cell biological analysis. Stable expression of fluorescent fusion proteins, fluorescent or luminescent reporter- and cell tracking-constructs, or RNAi-mediated knockdown of specific proteins extend the analyses to the study of the role of compound mutations in tumorigenesis. These cell models together with our in vitro metabolic stress assay have been successfully used to characterize epithelial cell response to metabolic stress in vitro and in vivo (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Karp et al., 2008; Mathew et al., 2007b; Nelson et al., 2004; Shimazu et al., 2007).
Autophagy is a highly conserved process tightly regulated by a set of essential genes such as atg5, atg7, and beclin 1, which produce a profound autophagy-defective phenotype when allelically lost (atg5−/−, atg7−/−, or beclin 1+/−). With the genetic landscape of autophagy regulation quickly emerging, several transgenic mice specifically targeting autophagy are currently available for tumorigenicity studies in vivo. Primary epithelial cells from wild-type, atg5−/−, atg7−/−, and beclin 1+/− mice can be immortalized through expression of the adenoviral protein E1A and a dominant negative p53 mutant (p53DD) to generate isogenic epithelial cell lines that are suitable for studying the role of autophagy in cancer (Mathew et al., 2008). We have generated iBMK cells and iMMECs that are wild-type or autophagy-deficient (beclin 1+/−, atg5−/−, atg7−/−) with and without a functional apoptotic pathway (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Mathew et al., 2007b, 2008). These cell lines have been successfully employed to demonstrate that their autophagy-defective phenotype is independent of the mode of autophagy impairment and the tissue of origin (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Mathew et al., 2007b).
The process of autophagy is characterized by the formation of isolation membranes (phagophores) that mature into double-membrane vesicles called autophagosomes (Levine and Kroemer, 2008). Under conditions of metabolic stress, the product of the essential autophagy gene LC3/atg8 is proteolytically cleaved, lipidated, and translocated to the forming autophagosomes, as demonstrated by the redistribution of the EGFP-LC3 fusion protein from a diffuse cytoplasmic localization under normal conditions to discrete, perinuclear puncta under metabolic stress (Fig. 4.1) (Klionsky, 2007; Mizushima, 2004). Similar induction of LC3 translocation occurs under growth factor deprivation. This process is impaired by deficiencies in essential autophagy genes, as indicated by the failure to form EGFP-LC3 puncta on monoallellic loss of beclin 1 (Fig. 4.1) or complete atg5 deficiency (Mizushima, 2004).
In cells competent for apoptosis and autophagy, the predominant phenotype under conditions of metabolic stress is apoptosis, the defects in which induce prolonged autophagy-supported cell survival. Thus, the assessment of autophagy under metabolic stress is facilitated in an apoptosis-defective background. To monitor autophagy induction under metabolic stress, we generated apoptosis-defective beclin 1+/+ and beclin 1+/− iBMK and iMMEC cells stably expressing the autophagy marker EGFP-LC3. These cell lines and their tumor allografts in nude mice allow in vitro and in vivo visualization and quantification of autophagy under a wide variety of experimental conditions (Karantza-Wadsworth and White, 2008; Mathew et al., 2008; Nelson et al., 2004).
Primary baby mouse kidney epithelial (BMK) cells are isolated from wild-type, bax−/−/bak−/−, beclin 1+/−, and atg5−/− mice and immortalized by E1A and a dominant negative p53 mutant (p53DD) to generate iBMK cells. Apoptosis-competent iBMK cells are then engineered to express either vector control (pCEP4) or Bcl-2 (pCEP-Bcl-2) as described previously (Degenhardt et al., 2006; Mathew et al., 2007b). Mouse mammary epithelial cells (MMECs) are isolated from 6–8-week-old female beclin 1+/+ and beclin 1+/− mice, immortalized by E1A and p53DD (to generate iMMECs) and rendered apoptosis-defective by stable Bcl-2 expression, as previously described (Karantza-Wadsworth et al., 2007). The detailed protocols for generating iBMK cells and iMMECs are available in an earlier volume of Methods in Enzymology (Karantza-Wadsworth and White, 2008; Mathew et al., 2008).
iBMK cells are further engineered to express EGFP-LC3 as described subsequently:
iMMECs are further engineered to express EGFP-LC3 as described subsequently:
A cell-based system that enables functional autophagy monitoring is important not only to better understand the role of autophagy in metabolic stress management and cancer progression but also to screen for autophagy modulators. Immortalized epithelial cells from wild-type and autophagy-deficient mice stably expressing EGFP-LC3 (Karantza-Wadsworth and White, 2008; Mathew et al., 2008) can be used for real-time observation of autophagy (see subsequently), as well as high-throughput screens for identifying novel autophagy inhibitors and stimulators.
Rapid tumor growth is often associated with metabolic stress, as cellular proliferation outstrips vascular supply and results in hypoxic regions within tumors. Induction of autophagy in tumors in vivo can be visualized in the first few days following subcutaneous or orthotopic implantation of apoptosis-defective iBMK cells (Fig. 4.3) and iMMECs stably expressing EGFP-LC3 in nude mice (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Mathew et al., 2007b). Tumor allografts of EGFP-LC3-expressing cells in the abdominal flank (iBMK) or the mammary fat pad (iMMECs) of nude mice provide an excellent system to monitor metabolic stress and autophagy induction during tumorigenesis (Karantza-Wadsworth et al., 2007; Mathew et al., 2007b).
Tumors generated by cells stably expressing EGFP-LC3 are excised at various times post-implantation (days 1, 3, 8 and 15) allowing spatial and temporal correlation of functional autophagy status with histological markers (e.g., hypoxia, vascularization, inflammation).
Centrosomes are cellular organelles that ensure uniform distribution of DNA during mitosis through bipolar division of chromosomes. Centrosomes themselves divide once per cell cycle during S phase, maintaining a tightly regulated centrosome number of 1 (in G1 phase) or 2 (in G2 phase) per cell. However, when this regulation is impaired, numerical abnormalities such as supernumerary centrosomes (more than 2 per cell) can result in multipolar spindles and abnormal segregation of chromosomes, leading to aneuploidy. Centrosome abnormalities are common among solid tumors and are indicative of genomic instability (Fukasawa, 2007).
Flow cytometry is a simple, but powerful, tool to study ploidy abnormalities by measuring cell DNA content. Cells are fixed and stained with propidium iodide (PI), which stains DNA by intercalating between the bases. PI also binds to RNA, necessitating treatment with ribonuclease (RNase) to minimize interference with DNA staining. Once PI is bound to nucleic acids, the fluorescence excitation and emission maxima are shifted and fluorescence emission is enhanced 20- to 30-fold and is proportional to the total amount of the DNA. Flow cytometry allows the measurement of this fluorescence per cell, thus permitting the quantification of the total amount of DNA per cell. A DNA-PI fluorescence histogram for a normal cell population is typically comprised of a 2-peak profile with fluorescence intensity on an arbitrary scale on the X-axis and frequency on the Y-axis. The first peak represents the diploid population of cells in G1 phase of the cell cycle (2N DNA content) and the second peak corresponds to cells in G2 and M phases of the cell cycle (4N DNA content). The valley connecting the two peaks represents cells with intermediate amounts of DNA (2N-4N) corresponding to cells in S phase that are undergoing DNA synthesis at the time of fixation. It may be noted that 2N and 4N notations are merely relative amounts of DNA. In samples where cell death has occurred, there can be a sub-G1 peak (less than 2N DNA content) and this is often used as a measure of apoptosis.
A major problem in the determination of ploidy abnormalities by flow cytometry is that only a relative, but not the absolute, DNA content is obtained. This is further complicated by variations in DNA staining intensity, due to cell concentration and instrument parameter variability, which can affect peak positioning and may lead to misinterpretation of the histogram. Extreme consistency in sample preparation is thus critical. These problems are circumvented by using an internal biological DNA standard with a known genome size (C-value), such as Chicken Erythrocyte Nuclei (CEN) Singlets. CEN have a C-value of 1.25 pg (2N chromosome number 18) compared to that of mouse (Mus musculus) cells (C-value = 3.25; 2N chromosome number 40) and human (Homo sapiens) cells (C-value = 3.5; 2N chromosome number 46). Therefore, when stained and analyzed together with the mouse or human cell lines under investigation, CEN provides a single reference peak to the left of the diploid (2N) peak of these cell lines and the relative position of the other peaks in reference to the CEN singlet provides a satisfactory ploidy measure. Spontaneous ploidy abnormalities due to allelic loss of beclin 1 in iBMK cells and iMMECs are analyzed as described subsequently.
Ploidy abnormalities are often caused by numerical aberrations in chromosome numbers that are telltale signs of an underlying genomic instability (Rajagopalan and Lengauer, 2004). Giemsa staining allows visualization of gross numerical and structural chromosome abnormalities (Brown and Baltimore, 2000). Prior to Giemsa staining, cells are treated with a microtubule poison, such as nocodazole, for arrest in mitosis (metaphase), where chromatin material is condensed into individual chromosomes. Cells in mitosis are spherical, and therefore loosely attached to the tissue culture plate, so they can be easily removed by shake-off. Isolated mitotic cells are then allowed to swell up by hypotonic treatment followed by methanol: acetic acid fixation before being dropped on a glass slide. A detailed protocol follows.
One of the major genomic instability phenotypes associated with autophagy defects is the random gain and losses of chromosomes (Albertson et al., 2003). Such aberrations lead to variations in DNA copy numbers and constitute a major genomic instability phenotype in cancer (Albertson, 2006). As chromosomal gains often also accompany chromosomal losses in the genome, these variations may not necessarily be reflected in the total amount of DNA per cell and therefore cannot be identified by DNA quantification by flow cytometry. Microarray-based aCGH is a powerful technique to identify DNA copy number variations by comparing the hybridization intensities between a normal (reference genome) and a test genome to detect the relative copy number variations signifying chromosomal losses and gains (Kallioniemi et al., 1992).
Gene amplification results from DNA double-strand breaks (DSBs) due to increased oxidative stress or defects in DNA repair. It is facilitated by inactivation of the p53 DNA damage checkpoint (Lin et al., 2001; Little and Chartrand, 2004; Livingstone et al., 1992; Mondello et al., 2002) and is a major mechanism of oncogene activation (Albertson, 2006; Hennessy et al., 2005; Shen et al., 1986). Therefore, iBMK and iMMEC cell lines with inactivated p53 and pRb pathways are expected to be prone to gene amplification at similar frequencies. Gene amplification is the only known mechanism of resistance to N-phosphonacetyl-l-aspartate (PALA) that prevents de novo pyrimidine biosynthesis by inhibiting the aspartate transcarbamylase activity of the carbamoyl-P synthetase/aspartate transcarbamylase/di-hydroorotase (CAD) enzyme complex (Livingstone et al., 1992). Indeed, PALA-resistant cells demonstrate amplification of the CAD gene. Thus, the frequency of clonogenic resistance to PALA is a direct measure of gene amplification, and therefore of the underlying DNA damage and genomic instability. Autophagy suppresses DNA damage and gene amplification, as monoallelic loss of beclin 1 promotes PALA resistance mediated by gene amplification (Karantza-Wadsworth et al., 2007; Mathew et al., 2007b). This function of autophagy is one of the mechanisms by which autophagy may function as a tumor suppressor pathway (Mathew et al., 2007a).
|1||Initial denaturation at 94 °C for 5 min|
|2||Denaturation at 94 °C for 30 s|
|3||Annealing at 50 °C for 30 s|
|4||Extension at 72 °C for 30 s|
|5||Repeat steps 2–4 for N times|
|6||Additional extension at 72 °C for 5 min|
Genomic instability due to defective autophagy likely plays a major role in tumorigenesis. Conventional tissue culture techniques cannot provide the spatial and temporal information that is crucial for understanding the dynamic cellular processes leading to genomic instability. Long-term observations are also restricted by the demands for physiological growth conditions and by limitations of microscopy instrumentation. CVTL, especially in combination with fluorescence protein tagging, is a powerful tool to monitor cellular processes that occur over extended periods of time. We have developed a model system comprising of a wide array of genetically defined epithelial cells that closely recapitulate in vivo tumor cell phenotypes under physiological conditions in vitro (Mathew et al., 2008). Combining the CVTL system and a panel of fluorescent probes stably expressed in these cells, we have captured and characterized cellular processes such as apoptosis, autophagy, necrosis, mitosis and cell division, wound healing response, and 3D morphogenesis that play key roles in tumorigenesis (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Karp et al., 2008; Shimazu et al., 2007). The fully automated and environmentally controlled system enables us to culture and film isogenic cell lines of varying genotype under multiple growth conditions within the same experiment. Time-lapse observations typically span up to 12 days or more and can be performed under regular and metabolic stress conditions, including drug treatments (Degenhardt et al., 2006; Karantza-Wadsworth et al., 2007; Karp et al., 2008; Shimazu et al., 2007).
Metabolic stress and genomic instability have been implicated in human cancer pathogenesis for a long time; however, the exact functional interaction between these factors has been largely elusive. By developing an in vitro assay that faithfully recapitulates metabolic stress in the tumors in vivo, we discovered that autophagy plays a major role in linking the two by mitigating the deleterious consequences of metabolic stress and the resultant DNA damage and instability (Karantza-Wadsworth et al., 2007; Mathew et al., 2007b). Therefore, activation of autophagy not only supports survival under stress but also protects the genome, and therefore promoting autophagy may be beneficial as a cancer prevention strategy.
Even more important, cancer cells overtly depend on an uninterrupted nutritional supply for meeting their proliferative needs, and this high demand in conjunction with inadequate supply is exactly what causes hypoxic regions in tumors. On one hand, metabolic stress compromises treatment efficacy, as tumor hypoxia is associated with resistance to radiation and chemotherapy. On the other hand, the greater susceptibility of cancer cells to metabolic stress can be exploited for therapeutic benefit. As a major cellular stress response, autophagy may facilitate tumor cell survival creating dormant tumor cells that cause disease recurrence. A clearer understanding of how tumor cells use autophagy to survive metabolic stress is essential for the successful use of autophagy modulation in cancer therapy. Therefore, screening for small-molecule autophagy inducers and inhibitors may prove extremely important in the development of novel preventative and therapeutic strategies. In the past few years, major strides have been made in understanding the role of autophagy in cancer progression and treatment, and there is great enthusiasm in targeting this important pathway for clinical outcome improvement. The wild-type and autophagy-defective immortalized epithelial cell lines that we have developed—in particular the cells with fluorescent readouts for autophagy monitoring—are powerful tools available to the cancer research community for this new and exciting endeavor.