Abstract
Genomics has revolutionized and personalized our approach to cancer therapy, with clinical trials now frequently involving patient stratification based on tumor genotype. Rational drug design specifically targeting the most common genetic events and aberrantly regulated pathways in human cancers makes this approach possible. However, our understanding of the wiring of oncogenic signaling networks and the key downstream effectors driving human cancers is incomplete, limiting our ability to predict clinical responses or identify mechanisms of resistance to targeted therapeutics. Recent studies in independent cancer lineages driven by distinct oncogenic signaling events point to a common downstream target, the mammalian (or mechanistic) target of rapamycin complex 1 (mTORC1), which dictates the cellular and clinical response to pathway-specific inhibitors. mTORC1 is a highly integrated signaling node that promotes anabolic cell growth and proliferation and lies downstream of multiple oncogenes and tumor suppressors, including those influencing the PI3K-Akt and RAS-RAF-MEK-ERK pathways. Studies are now suggesting that to effectively target the major oncogenic signaling pathway in a given tumor, mTORC1 must be inhibited, and that its sustained activation is a major mechanism of resistance to such targeted therapies.
The notion that single oncogenic events can render cancer cells addicted to specific molecular pathways has led to a movement toward genotype-based personalized approaches to cancer treatment using targeted therapeutics. Drugs that target the most common oncogenic signaling pathways, such as the PI3K-Akt (PI3K, phosphatidylinositol 3 kinase) and RAS-RAF-MEK-ERK (MEK, mitogen-activated protein kinase kinase; ERK, extracellular signal–regulated kinase) pathways, are at the forefront of this revolution. Clinical trials using tumor genetics to stratify patients for these treatments will surely increase response rates, an idea supported by clinical data from such trials [for example, (1)]. However, it is also evident that there will be some nonresponders and that resistance can develop in initial responders. Therefore, although genetic analyses of tumor tissue to determine the primary molecular drivers of the cancer will be important for dictating first-line therapies, there will continue to be tumors that resist these precision treatments. As such, there is a need for defining resistance mechanisms to targeted therapies and for identifying biomarkers to predict and monitor tumor response. As our understanding of cell signaling networks improves, it becomes evident that a major mechanism of resistance to pathway inhibitors will be the activation of alternative signaling events converging on a few common downstream targets that are critical for cancer progression.
Two groups have recently reported in Science Translational Medicine the identification of a common biomarker for cancer resistance to targeted therapeutics that also provides a rationale for specific combination therapies in distinct cancer lineages. Elkabets et al., studying the effects of a PI3K p110α-specific inhibitor in PIK3CA-mutant breast cancer (2), and Corcoran et al., studying the effects of RAF and MEK inhibitors in BRAF-mutant melanoma (3), found that the status of signaling downstream of mTORC1 (mammalian target of rapamycin complex 1) after treatment predicts the response to these pathway inhibitors in genetically defined cancer cell lines and a small number of cancer patients. mTORC1 is an important promoter of cell growth and proliferation that is frequently activated in human cancers across nearly all lineages because of its convergent regulation by multiple upstream signaling pathways that contribute to the development of cancer (4) (Fig. 1). Whereas mTORC1 normally senses cellular growth conditions through these pathways to properly regulate anabolic processes, its persistent activation in cancer cells facilitates the metabolic changes that promote uncontrolled growth (5). These two current studies indicate that mTORC1 inhibition is required for the sensitivity of tumor cells to drugs targeting upstream oncogenic pathways.
Growth factor and nutrient regulation of mTORC1 through a signaling network composed of oncogenes and tumor suppressors. Secreted mitogens, cytokines, and hormones (collectively referred to as growth factors) stimulate mTORC1 activation through receptor tyrosine kinases (RTKs), as well as G protein–coupled receptors (not shown). Two parallel pathways dominate this regulation: the PI3K-Akt and Ras-Raf-MEK-ERK pathways, which both inhibit a complex composed of the tuberous sclerosis complex (TSC) tumor suppressors. This relieves the TSC complex–mediated inhibition of Rheb (Ras homolog enriched in brain), a small GTPase that is an essential and direct activator of mTORC1. Glucose, in part through the control of cellular energy levels [adenosine triphosphate (ATP)], can influence mTORC1 through effects on 5′ AMP (adenosine monophosphate)–activated protein kinase (AMPK). When glucose availability is low, or when cells are under other states of cellular energy stress, liver kinase B1 (LKB1) activates AMPK, which has activating effects on the TSC complex to inhibit the Rheb-mediated activation of mTORC1, as well as directly inhibiting mTORC1. Amino acids activate mTORC1 through the Rag GTPases and poorly understood sensing mechanisms involving the Ragulator and GATOR1 protein complexes, which activate or inhibit, respectively, Rag proteins and mTORC1. In this manner, cellular growth conditions are sensed by mTORC1, which in turn controls the anabolic processes needed for cell growth and proliferation. Within the upstream signaling network are numerous oncogenes (teal) and tumor suppressors (orange), leading to multiple parallel paths that aberrantly activate mTORC1 in human cancers, which will differentially influence the effectiveness of pathway-specific inhibitors for cancer therapy.
CREDIT: V. ALTOUNIAN/SCIENCE SIGNALING
Elkabets et al. investigated the response of a panel of 20 breast cancer cell lines with mutations in the gene PIK3CA, encoding the p110α subunit of PI3K, to the p110α-selective PI3K inhibitor BYL719 (2). Whether sensitive to this compound or not, all of the cell lines showed strong inhibition of Akt, the major oncogenic target activated by PI3K signaling. The PI3K-Akt pathway regulates many downstream effectors, including the stimulation of mTORC1 (6, 7). However, resistant cell lines displayed sustained mTORC1 signaling in the presence of the PI3K inhibitor, as indicated by phosphorylation of a downstream target, ribosomal protein S6. Furthermore, cells selected for acquired resistance exhibited reactivation of mTORC1. Through genetic manipulation of mTORC1 signaling, it was found that sustained mTORC1 activation was sufficient to cause resistance to BYL719. Importantly, there was an inverse correlation between phospho-S6 levels and the clinical response to BYL719 in paired biopsies taken before and after treatment of patients with confirmed PIK3CA mutations in their tumors. Furthermore, acquired resistance in patients that initially responded and later progressed was characterized by the restoration of phospho-S6 in tumor samples. Finally, treatment with RAD001 (everolimus), an mTORC1-specific inhibitor derived from rapamycin (sirolimus), restored sensitivity to BYL719 in resistant cell lines and xenograft tumors. In addition to supporting mTORC1 activation as a major resistance mechanism, these findings indicate that the addition of an mTORC1 inhibitor to anti-PI3K therapy could enhance or prolong the clinical response of PIK3CA-mutant breast cancers.
In a genetically and pathologically distinct setting, Corcoran et al. arrived at similar conclusions in BRAF-mutant melanoma treated with either the RAF inhibitor vemurafenib or the selective MEK inhibitor selumetinib (3). The activation status of ERK after treatment of a panel of BRAF-mutant melanoma cell lines with these drugs did not predict sensitivity. Like the PI3K-Akt pathway, the RAS-RAF-MEK-ERK pathway has many downstream targets and potently activates mTORC1 (8, 9). Melanoma cell lines that were sensitive to either vemurafenib or selumetinib displayed inhibition of mTORC1 signaling after treatment, whereas resistant cell lines were all characterized by sustained mTORC1 signaling, assessed by phospho-S6 detection. The strong correlation between sustained mTORC1 signaling and resistance to these RAF or MEK inhibitors was also observed in mouse xenograft models and in paired biopsies from melanoma patients receiving RAF inhibitor therapy.
The results from these two studies advance an approach to cancer therapy that entails not only genetic diagnosis of the major tumor-promoting events to choose an initial targeted therapeutic, but also early monitoring of established biomarkers that are predictive of therapeutic response. The studies also suggest that monitoring the effects of a pathway-specific drug on the phosphorylation of ribosomal protein S6 might be predictive of sensitivity or resistance. It is worth noting that this marker of mTORC1 signaling is more than just an indicator of cell proliferation, in contrast to Ki67 staining, for example. It reflects the complexity of the signaling network upstream of mTORC1, where there are numerous partially redundant pathways that can influence the activation state of mTORC1 (10) (Fig. 1). The finding that PI3K inhibitor–resistant breast cancer cells are sensitized by co-treatment with an mTORC1 inhibitor demonstrates that network redundancy leading to PI3K p110α-independent mTORC1 activation is a major cause of resistance to such agents (2). In the case of PIK3CA-mutant breast cancers, mechanisms of resistance to BYL719 contributing to sustained mTORC1 activation might involve receptor tyrosine kinase or G protein–coupled receptor-mediated activation of other PI3K isoforms or other upstream pathways, such as those activating ERK. Likewise, in BRAF-mutant melanoma, alternative mechanisms of ERK activation or concomitant activation of the PI3K-Akt pathway are likely to drive sustained mTORC1 signaling and resistance to RAF and MEK inhibitors. Regardless of the mechanism, these studies suggest that detection of sustained phosphorylation of ribosomal protein S6 and other downstream markers of mTORC1 signaling [such as phosphorylation of 4E-BP1 (eukaryotic initiation factor 4E–binding protein 1) or CAD (carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase)] in a biopsy taken shortly after treatment initiation and again if responsive tumors begin to progress might warrant combining the PI3K or RAF/MEK inhibitors with an mTORC1 inhibitor. The requirement for repeated biopsies is a limitation of this approach in breast cancer but would be more attainable in melanoma.
Given that mTORC1 is activated in well over 50% of human cancers (4), there has been much interest in using rapamycin and its analogs (rapalogs) to treat tumors. However, rapalogs have had limited success as single-agent cancer therapies in hundreds of clinical trials to date. Although the interpretation of these clinical findings is complicated by the more recent realization that rapalogs only partially inhibit mTORC1 activity (11), the modest efficacy of these inhibitors indicates that, in most cancers, one must target the upstream oncogenic pathways that have many downstream effectors in addition to mTORC1. However, as might have been predicted by the sheer number of common cancer pathways that influence mTORC1 signaling, the Elkabets et al. (2) and Corcoran et al. (3) studies demonstrate that mTORC1 inhibition is essential to effectively target these upstream pathways. These findings, along with the numerous ongoing trials using rapalogs in combination with other agents, restore mTORC1 signaling to prominence when considering a targeted approach to personalized cancer therapy, and they also demonstrate the importance of defining the molecular mechanisms influencing the activation state of mTORC1.
