Department of Medicine and Abramson Cancer Center, University of Pennsylvania, Philadelphia, Pennsylvania, USA.
Address correspondence to: Ravi K. Amaravadi, Abramson Cancer Center, University of Pennsylvania, 12 Penn Tower, 3400 Spruce Street, Philadelphia, Pennsylvania 19104, USA. Phone: (215) 662-7402; Fax: (215) 662-7865; E-mail: email@example.com.
First published November 20, 2008 - More info
See the related article beginning on page 3917.
Autophagy — a process of “self-eating” that involves enzymatic digestion and recycling of cellular constituents in response to stress — contributes to both cancer cell death and survival. In this issue of the JCI, Lu et al. report that controlled induction of tumor suppressor gene aplasia Ras homolog member I (ARHI) results in autophagic cell death of human ovarian cancer cells in vitro (see the related article beginning on page 3917). However, within xenograft tumors in mice, multiple factors within the tumor microenvironment switched ARHI-induced autophagy to a mechanism of tumor cell survival, leading to tumor dormancy. Since ARHI expression is suppressed in the majority of breast and ovarian cancers but is high in premalignant lesions, ARHI-induced autophagy could be manipulated for therapeutic benefit.
Cells that encounter a variety of stresses undergo an evolutionarily conserved process of self-digestion termed autophagy. The importance of this intracellular damage response for pathophysiology has been established across multiple fields, including infectious disease, neurodegeneration, heart failure, and cancer (1). In cancer in particular, the debate continues to rage as to whether or not autophagy is primarily a mechanism of cell death or cell survival. This is important to understand in order to preferentially promote the clinical development of therapeutic interventions that can either inhibit or enhance autophagy in tumor cells. A number of clinically available cancer therapeutics, including DNA-damaging chemotherapy, radiation therapy, and molecularly targeted therapies have been found to induce autophagy in cell culture and animal models (2). Recent investigations have found that inhibition of therapy-induced autophagy with chloroquine derivatives can enhance cell death in established tumors, leading to enhanced tumor regression and delayed tumor regrowth (3). In this study, a combination of experiments involving RNA interference and drugs were used to demonstrate that therapy-induced autophagy in established tumors allows tumor cells to survive therapeutic stress (3). The cytoprotective role of autophagy following chemotherapy has been confirmed by other investigators (4–6). These findings have led to a growing number of early-phase clinical trials that aim to establish the safety and preliminary efficacy of combining agents that induce autophagy with hydroxychloroquine in patients with advanced malignancies refractory to standard treatments. Collectively, these actively accruing clinical trials represent the first attempt to manipulate autophagy for therapeutic benefit in cancer.
Autophagy inhibition with chloroquine has also been found to significantly delay the onset of spontaneous tumors in a transgenic mouse model of Myc-induced lymphoma (7). This study suggested that autophagy inhibition may have a role not only in combination therapies for patients with active metastatic disease, but also in the prevention of early disease. However, conducting a clinical trial of a novel autophagy inhibitor, or in the case of hydroxychloroquine, an old drug being used for new indication, in tumor-free patients is difficult, requiring large numbers of patients, and long periods of follow-up. While the known safety and tolerability of chloroquine derivatives suggest they may be ideal drugs to test as chemopreventative agents, committing the resources necessary to conduct prevention trials may not be warranted until a potential benefit of this strategy has been established in a clinical setting with short-term outcomes, such as in metastatic disease. Another clinical setting that could be more feasible to investigate autophagy inhibitors is in the setting of minimal residual disease with high risk of relapse. A common problem across multiple malignancies is that cancer therapy can initially result in a complete remission, but there is an extremely high chance of relapse. This indicates there are clinically undetectable tumor cells that are dormant and have the ability to regrow at a later date. This is an ideal setting to investigate the role of autophagy and the effects of pharmacological autophagy inhibitors.
In this issue of the JCI, Lu, Bast, and colleagues report an ovarian cancer model of minimal residual disease and identify a role for autophagy in supporting tumor dormancy (8). Additionally, their analysis of the effects of controlled expression of the autophagy-inducing tumor suppressor gene aplasia Ras homolog member I (ARHI) in human ovarian cancer cells in vitro versus in vivo sheds light on the relationship between autophagic cell death and autophagic cell survival and implicates an underappreciated role for the tumor microenvironment in deciding this fate.
Over the past decade, Bast and colleagues have extensively characterized the genetic and biochemical underpinnings of ARHI, a maternally imprinted Ras-homologous kinase that exerts growth-inhibitory effects when expressed in ovarian cancer cells (9). ARHI is expressed in normal ovarian epithelium, but through the transition to malignancy, ARHI expression is suppressed in a stepwise fashion. Examination of tumor samples from cancer patients has determined that ARHI loss of heterozygosity (LOH) occurs in 60%–70% of ovarian and breast cancers (10–12). Another gene whose expression is suppressed through the LOH in breast and ovarian cancer is the autophagy-associated gene beclin 1 (BECN1). Unlike BECN1, in which the wild-type allele is retained and therefore can be expressed (13), the retained allele of ARHI is maternally imprinted and therefore not expressed (14). Since there appears to be a selective advantage to complete loss of ARHI expression during neoplastic transformation, developing a tumor model that allows inducible expression of ARHI at physiological levels in established tumors has provided valuable insight into the ARHI-mediated mechanism of tumor suppression. The insight gained from this current study by Lu et al. (8) has the potential to launch clinical investigations aimed at changing therapeutic approaches to the treatment of ovarian and breast cancers.
Using doxycycline-inducible expression of ARHI, the authors demonstrated that stable or transient physiologic expression of ARHI in human ovarian cancer cell lines and other human cancer cell lines resulted in the induction of autophagy and not apoptosis (8). Elegant in-cell cross-linking studies demonstrated that ARHI localizes with the autophagy marker light chain 3 (LC3) on the surface of autophagic vesicles, providing evidence that ARHI may regulate autophagy though direct physical interaction with the autophagic machinery. Expression of ARHI was associated with increased expression of ATG4, which is encoded by one of the core autophagy genes (ATG4) in a group of genes denoted as ATGs. This oxidant-sensitive enzyme deconjugates LC3 from lipids on the surface of autophagic vesicles. To look for more indirect means by which ARHI could regulate autophagy, the authors used reverse-phase protein microarray analysis to look at multiple signaling events simultaneously. Expression of ARHI lead to a marked inhibition of PI3K/AKT/mammalian target of rapamycin (PI3K/AKT/mTOR) and MAPK signaling. Expression of the mutant ARHI that was missing the N-terminal domain, which distinguishes its structure from other Ras family proteins, prevented growth factor–induced membrane recruitment of AKT and limited the activation of downstream effectors mTOR and p70S6K.
To investigate the functional consequences of ARHI-induced autophagy, the authors determined that ARHI expression prevented clonogenic growth after one week and resulted in cell death after two weeks (8). In contrast to a previous study in which chronic growth factor withdrawal from apoptosis-deficient cells resulted in a reversible form of autophagy that contributed to cell survival (15), Lu et al. show that transient expression of ARHI in human ovarian cancer cells leads to a reversible form of autophagy, but that chronic ARHI expression leads to irreversible autophagic cell death (Figure 1A). When ARHI expression was induced in ovarian cancer xenografts in mice, these authors found that acute cell death ensued, leading to tumor regression. However, removal of doxycycline, which resulted in the absence of ARHI expression, led to tumor regrowth. These dormant cells were likely dependant on autophagy as a survival strategy, since systemic pretreatment of mice harboring ARHI-suppressed ovarian xenografts with the autophagy inhibitor chloroquine resulted in a significant delay of tumor regrowth once ARHI expression was downregulated.
Consequences of forced expression of the tumor suppressor gene ARHI in ovarian cancer. In their study in this issue of the JCI, Lu et al. (8) report that controlled expression of the tumor suppressor gene ARHI in human ovarian cancer cells leads to autophagy. (A) ARHI expression was associated with interrupted PI3K signaling, which resulted in reversible (under conditions of transient ARHI expression) or irreversible (under conditions of chronic ARHI expression) autophagic cell death in vitro. (B) In contrast, ARHI-induced autophagy contributed to cell survival and tumor dormancy in ovarian cancer xenografts in mice, due to partially restored PI3K signaling from elements of the tumor microenvironment.
These results suggest that, unlike the in vitro experiments, in which cells exposed to ARHI were destined for autophagic cell death, within the tumor microenvironment, ARHI-induced autophagy allows a cohort of hibernating cells to survive and eventually recur as tumors once the ARHI tumor suppressor is downregulated (8). Lu et al. investigated the contributions of growth factors, inflammatory cytokines, angiogenic factors, and extracellular matrix proteins to autophagic cell death in vitro and in vivo. They found that, by limiting the ARHI-dependent suppression of intracellular PI3K signaling, members of each of these classes of molecules within the tumor microenvironment can contribute to changing autophagy from a cell death mechanism to a cell survival mechanism. Taken together, these results suggest that the magnitude of autophagy is inversely related to PI3K signaling. ARHI suppresses intracellular PI3K signaling, and elements of the tumor microenvironment promote intracellular PI3K signaling. Autophagic cell death occurs when autophagy is maximally induced, but when levels of autophagy are elevated above basal levels, but are less than maximum levels, the result is autophagic cell survival (Figure 2).
The magnitude of autophagy may determine the fate of cancer cells in vivo. This conceptual model proposes a relationship between the magnitude of cancer cell autophagy and its consequences on cell death and cell survival. As reported by Lu et al., expression of the tumor suppressor gene ARHI suppresses signal transduction through PI3K signaling, while factors within the tumor microenvironment maintain PI3K signaling within tumor cells (8). The magnitude of autophagy may be inversely proportional to PI3K signaling, and thresholds may exist within cancer cells for the magnitude of autophagy that leads to cell death versus the magnitude of autophagy that leads to cell survival.
This microenvironment-directed switch may also shed light on why a gene whose expression can suppress multiple signaling pathways and lead to cell death is expressed constitutively in normal and premalignant epithelium. The role of autophagy in limiting DNA damage and genomic instability has been established (16). One hypothesis that can be tested in future studies is that ARHI-induced autophagy limits the rate of genomic instability and neoplastic transformation, while microenvironmental signals in normal or premalignant native tissue simultaneously limit ARHI-induced autophagic cell death. A related question that arises is whether the effects of ARHI expression are specific to ovarian and breast cancer, or whether ARHI expression is selectively suppressed in other cancers. Similarly, what is the interaction between ARHI expression and common mutations in oncogenes encountered in ovarian cancer? The SKOv3 ovarian cancer cell line has constitutively activated PI3K/AKT signaling but no known mutations in PI3K, phosphatase and tensin homolog (PTEN), RAS, or BRAF. Would the effects of ARHI expression be different in tumor cells harboring mutations in PI3K, RAS, or BRAF? Finally, ARHI can be added to a growing list of “non-ATG” genes that were not originally identified as autophagy genes but that directly regulate this process through physical engagement of the autophagic machinery. This list includes, but is not limited to, endophilin B1 (BIF1) (17), UV radiation resistance–associated gene (UVRAG) (17), p62 (18), and WD repeat domain phosphoinositide-interacting 1 (WIPI1) (19). Individually, aberrant expression of some of these genes has been demonstrated in a wide variety of malignancies. It remains to be determined whether more or all of these genes are bonafide tumor suppressor genes in human tumors. It will also be important to determine whether the genetic aberrations in multiple autophagy-associated genes can be found in the same tumor, as well as the implications of this for tumorigenesis and clinical prognosis.
Regardless of these remaining questions, the inducible ARHI tumor model reported here (8) also has practical implications in understanding the role of autophagy in the treatment of ovarian cancer patients. The authors point out that 60% of primary ovarian tumors have decreased ARHI expression associated with shortened disease-free survival after chemotherapy. This observation could be explained by a propensity for early recurrence in ovarian tumors lacking ARHI-induced autophagy, which suggests that disease-free survival could be prolonged in this subgroup of patients by combining drugs known to induce autophagy with current chemotherapy regimens. Alternatively, in patients whose tumors express ARHI, could the addition to standard chemotherapy of an autophagy inhibitor prolong disease-free survival compared with chemotherapy alone? Examining the tumor tissue of ARHI-positive and ARHI-negative tumors for markers of autophagy is a first step toward determining whether these clinical investigations should be launched. A parallel preclinical effort would be to exploit this tumor model as a platform to validate autophagy inhibitors with pharmacological properties identified in drug screens. Promising autophagy inhibitors should reproducibly delay or prevent regrowth of ARHI-supressed tumors upon removal of doxycycline. While most tumor suppressor genes are not druggable targets themselves, the current study by Lu et al. (8) demonstrates that mechanistic studies of tumor suppressor genes can uncover druggable targets that can be tested in defined subsets of cancer patients.
The author is supported by NIH grant 1K23CA120862.
Nonstandard abbreviations used: ARHI, aplasia Ras homolog member I; BECN1, beclin 1.
Conflict of interest: The author has declared that no conflict of interest exists.
Reference information: J. Clin. Invest.118:3837–3840 (2008). doi:10.1172/JCI37667.
See the related article beginning on page 3917.