Julien Ablain, Laura Attardi, Alberto Bardelli, Glenn Dranoff, Mikala Egeblad, Jeffrey A. (Jeff) Engelman, Gerard Evan, Shridar Ganesan, Joseph (Jos) Jonkers, Anthony (Tony) Letai, Richard Marais, Frank McCormick, Benjamin (Ben) Neel, Scott Powers, Neal Rosen (organisateur), Louis M. Staudt, Hugues de Thé, David Tuveson (organisateur), Matthew G. (Matt) Vander Heiden, Valerie (Val) Weaver.
by David Tuveson
27 June – 2 July 2014
Keywords: Cancer drug resistance, adaptive resistance, immune therapies, cancer cell death, synthetic lethality, organoids, mitochondrial priming, cancer metabolism, Myc, Ras, P53, phosphatases, tumor tensile forces, drug delivery, cancer dormancy, clonal heterogeneity and evolution.
The 20 participants represented expertise in clinical oncology, genetics, drug development, biochemistry, cancer modeling, and immunology. The participants engaged in animated, unrestrained conversations on the mysteries and controversies in our field. Collaborations were formed, and unsupported areas were discarded.
Anthony Letai proposed that the response of patients to therapy reflected the apoptotic threshold of their cancer cells. Furthermore, that this can be measured as a fundamental property of the mitochondria contained in their cancer cells. Accordingly, Tony developed BH3 profiling for the neoplastic cell mitochondria, a technique that measures the ability of BH3 pro-death peptides to induce mitochondrial outer membrane permeabilization. BH3 profiling revealed that patients who respond to chemotherapy had highly primed mitochondria prior to treatment, as patients who responded poorly had poorly primed mitochondria. Most normal primary human cells were not primed, while bone marrow-derived cells were, perhaps explaining the special sensitivity of the bone marrow to cytotoxic therapy. Tony is now applying this technique to short term drug treatments of primary neoplastic cells harvested from patients, and finds that he can predict active drugs using “Dynamic BH3 Profiling. Dr. Letai posed the following questions that need to be addressed: “For which drugs will Dynamic BH3 profiling be most predicitive of clinical response? For which tumors? Also, will there be a limit to the number of treatments that can be simultaneously tested?”
David Tuveson discussed pancreatic cancer, a lethal malignancy due to its late diagnosis and limited response to treatment. Pancreatic cancer is predicted to become the second most common cause of cancer death in the USA within a decade. Therefore, tractable methods that enable the development of new diagnostic and therapeutic approaches for pancreatic cancer patients was developed by Dave in collaboration with Hans Clevers (Utrecht, NE). Robust organoid models were prepared from normal and neoplastic adult murine and human pancreas tissues. Pancreatic organoids can be rapidly generated from small amounts of tissue, proliferate indefinitely with ductal characteristics when passaged in semisolid media, survive freezing and thawing, and maintain a stable genome. Following orthotopic transplantation, the neoplastic organoids recapitulate the postulated stages of tumor development by forming intraepithelial and cystic neoplasms prior to progressing to invasive carcinomas. In culture, the organoids correctly predicted the resistance of Kras mutant cells to combined MEK and AKT inhibitors. Also, they suggested that combinations with pan-Her inhibitors would circumvent this adaptive resistance. Pancreatic organoids provide new molecular and cellular insights into cancer pathogenesis, and enable translational approaches that may reasonably accelerate the care of pancreatic cancer patients. Dr. Tuveson posed two challenges based upon this work: “Will organoid models of cancer accelerate individualized approaches to cancer genetics and cancer therapy?” Also, will the organoid approaches be useful methods to model adaptive resistance to therapies?”
Ben Neel discussed new insights into the role of PTP1B in breast cancer. Previous studies by his group and that of Michel Tremblay had found that PTP1B-deleted mice are resistant to HER2+ breast cancer. In addition, the Tremblay group had shown that mammary specific over-expression of PTP1B alone could be oncogenic. Neel’s group tested the effects of PTP1B depletion (by shRNA) in a set of HER2+ human breast cancer cell lines. Surprisingly, there was no effect on proliferation of this lines in normal 2D culture, in low serum, at low density or in low glutamine or glucose. Likewise, colony formation in soft agare or Matrigel was unaffected. When these lines were injected into nude mice, however, tumorigenicity was impaired. Inducible depletion of PTP1B also arrested tumor growth. Remarkably, in both settings, PTP1B depletion resulted in marked hypoxia and necrosis. Furtheremore, the PTP depleted lines also were much more sensitive to hypoxia in vitro. Reconstitution and inhibitor studies showed that these effects were PTP1B-specific and required PTP1B activity. Subsequent investigations showed that none of the known hypoxia pathways were abberant in PTP1B-deficient cells. Instead, PTP1B appears to regulate a large (600kd) protein, RNF213, recently implicated in Moyamoya disease. RNF213, in turn, controls the activity of one or more a-ketoglutarate dependent dioxygenase enzymes, as the death of PTP1B-deficient cells can be rescued by specific oxygenase inhibitors. Dr. Neel feels that future studies need to be aimed at elucidating the details of this novel PTP1B/RNF213/a-kg dioxygenase pathway, and defining its roles in other malignancies.
Glenn Dranoff presented evidence that the NKG2D pathway may be involved in the therapeutic effects of cancer vaccines and immune checkpoint blockade. NKG2D is an activating receptor expressed on NK cells and CD8+ cytotoxic T lymphocytes, which contributes to immune mediated tumor control. Although NKG2D ligands show minimal expression in healthy tissues, oncogenic stress triggers upregulation of these ligands on the surface of cancer cells. Tumors escape from NKG2D mediated cytotoxicity, however, through shedding surface ligands, particularly MICA, which leads to NKG2D downregulation and dysfunction. Some patients who respond to immunotherapy generate antibody responses to MICA that overcome this immune escape through binding shed MICA and promoting MICA specific cytotoxicity. In collaboration with Kai Wucherpfennig, fully human anti-MICA monoclonal antibodies have been isolated from responding cancer patients, and these enhance NK and CD8+ T cell function through several MICA-specific mechanisms. The anti-MICA monoclonal antibodies as well as a novel NKG2D-Fc fusion protein that targets all NKG2D ligands might prove useful in cancer immunotherapy. Dr. Dranoff challenged the community to determine whether restoration of NKG2D function will be sufficient to promote immune mediated tumor destruction in cancer patients.
Frank McCormick presented a cell system of Ras-less MEFs as a platform to investigate the specific effects of Ras oncogenes. K-Ras plays a direct causal role in many human cancers, most notably in pancreatic cancers, which are almost always driven by K-Ras mutations. We have discovered that K-Ras, initiates a stem-like program that increases drug resistance, enables K-Ras cancer cells to initiate tumors and metastasize efficiently and produce the cytokine LIF that maintains cells in a stem-like state. Blocking LIF with neutralizing antibodies could represent a new therapeutic strategy for K-Ras cancers. The stem-like program that K-Ras initiates is mediated by direct binding to Calmodulin, resulting in loss of CaM kinase and inhibition of non-canonical wnt signaling. We have been able to disrupt K-Ras/Calmodulin binding by activating PKC, using an orally available natural compound related to phorbol esters. This compound prevents tumor formation in mouse models of pancreatic cancer.
Laura Attardi discovered that P53 tumor suppressor mechanisms are separable from the canonical acute DNA damage induced cell cycle arrest and cell death pathways. By leveraging a p53 knock-in mouse strain expressing a transcriptional activation domain mutant that activates only a subset of p53 target genes yet is completely active in tumor suppression in a variety of different tumor types, she has been able to discover a set of novel p53 tumor suppression-associated target genes, which she is currently analyzing. She discussed the approaches she is taking to functionally identify the key targets mediating p53 tumor suppression.
Gerard Evan discussed the importance of considering nodes of tumor cells as opposed to isolated oncogenes, including the Mitogen kinase/Ras cloud, the Myc cloud, and the E2F cloud. Gerard also discussed cooperation of the Myc and Ras oncogenes in mouse models of lung and pancreatic cancer. He demonstrated that Myc/Ras oncogenic cooperation, known for some 30 years but without an underlying mechanism, is very multifaceted and includes both tumor cell intrinsic and extrinsic (microenvironmental) processes. For example, he demonstrated how acute activation of Myc can rapidly convert indolent PanIN lesions to PDAC, together with the dramatic signature desmoplasia that characterizes this particular cancer: subsequent de-activation of Myc triggers rapid regression back to a PanIN state, accompanied by regression of desmoplasia. These studies reveal the complex way that nodal oncogenes cooperate with each other to drive and maintain cancers,
Matt Vander Heiden argued that metabolic changes in the tumor can be predictive of therapeutic response, and might be targets for therapies, but success in both areas will require a better understanding of tumor metabolism. He presented data to suggest that the metabolism of tumors in mice is not the same as the metabolism of cell lines in culture. Lung and pancreatic tumors oxidize more glucose than cell lines derived from those tumors grown in culture, and also use amino acids as a nutrient source. At least some of the amino acids tumors use can be derived from the breakdown of normal tissue proteins. The increased turnover of tissue proteins in pancreatic cancer results in an increased in branch chain amino acids in both human patients and in mouse models of pancreatic cancer. One explanation for the differences in metabolism between tumors and cultured cells is that tumors are comprised primarily of non-proliferating cells and thus are likely to dominate the metabolic signal measured in the tumor. Cell culture is a reasonable model of proliferative cancer cell metabolism, and data was also presented to suggest that regulation of glucose metabolism is important in part to supply the nucleotides for DNA replication. Serine synthesis and metabolism via the mitochondria to generate folates for nucleotide synthesis are important for this process as well.
Neal Rosen discussed the mechanisms of drug response to targeted therapies, and resistance in human tumors. Neal discussed the pathway modulation that occurs in cancer cells when treated with therapies that target oncogene dependent targets, including oncogenic BRAF in melanoma and Her2 in TNBC. Neal coined this process “adaptive resistance”, and it appears to be a general phenomenon in neoplastic cells, including ER and AR blockade in breast and prostate cancer, respectively. To circumvent adaptive resistance, the pathway of resistance must be elaborated such that suitable additional antagonists are employed. Dr. Rosen hypothesized that strong blockage of cancer cell dependencies would leave less opportunity for adaptive resistance to occur in the first place.
Mikala Egeblad has devised intravital microscopy methods to visualize drug penetration, cancer cell death, and stromal responses in tumors after administration of cytotoxic chemotherapy. She has observed that heterogeneity in drug penetration into tumors can be a barrier against tumor response to therapy. She has also documented an influx of inflammatory cells into tumors after cytotoxic therapy, and she and many other groups have found that this influx of reactive inflammatory cell reduces the response to therapy. Currently, Mikala Egeblad’s group is studying whether regional variation in the microenvironment influence cancer cells ability to acquire genomic changes conferring resistance. They have also started to implement imaging in lungs during dissemination of cancer cells and found that neutrophils promote cancer cell invasion and metastatic seeding. Major unanswered questions from this work are 1) how do inflammatory cells alter cancer cells to reduce therapy responses? And 2) to what extent do inflammatory cells contribute to drug resistance of metastatic cancer?
Julien Ablain discussed a new zebrafish system of cancer modeling. He adapted the CRISPR/cas9 technology of genome editing to develop a vector system for tissue-specific gene inactivation in zebrafish. This tool allows cell type-restricted, high-throughput loss-of-function studies. Knock-out of p53 specifically in melanocytes of melanoma-prone fish expressing mutant Braf resulted in rapid tumor formation, thus providing a proof of principle for an in vivo screen of candidate melanoma-suppressor genes. Dr. Ablain proposed that this zebrafish system will allow the identification of novel tumor-suppressor genes (other than cdkn2a, pten and p53) that cooperate with activated Braf to drive melanomagenesis.
Lou Staudt discussed the development of rational drug combinations to improve the therapy of the most aggressive subtype of diffuse large B cell lymphoma (DLBCL), termed activated B cell-like DLBCL. The B cell receptor (BCR) signaling pathway is key to the pathogenesis of ABC DLBCL, which turns on multiple downstream survival pathways, including NF-B and PI(3) kinase. This “chronic active” BCR signaling can be targeted by the Bruton’s tyrosine kinase (BTK) inhibitor ibrutinib, which has produced a high response rate in ABC DLBCL in early phase clinical trials. While some patients have prolonged remissions, the majority eventually relapse, necessitating the development of combination therapies. Ibrutinib synergizes well with drugs targeting the other major survival mechanisms in ABC DLBLC including the PI(3) kinase pathway and BCL2. The drug lenalidomide inhibits IRF4, the master transcription factor for ABC DLBCL, and synergizes strongly with ibrutinib in killing ABC DLBCL cells. It is evident that there are too many combinations of drugs to be tested sequentially in clinical trials. Therefore, it will be imperative to derive rational combinations of 3 or more drugs that inhibit all parallel survival mechanisms as well as common resistance mechanisms to achieve a high rate of long-lasting remissions.
Valerie Weaver emphasized the importance of biophysical and biochemical properties of the tissue microenvironment in regulating cancer progression and treatment. She discussed the role of biophysical properties of the cells and the extracellular matrix microenvironment and presented data to illustrate how tissue mechanics is able to disrupt normal tissue homeostasis and contribute to the malignant behavior of cells and tissues. In her presentation she stressed that all tumors are mechanically corrupted and she presented evidence to demonstrate how high tissue levels force, derived from elevated cell contractility (intrinsic force), and the stiffness of the extracellular matrix (extrinsic force), collaborate to stimulate tumor cell growth, survival and motility/invasion in culture and in vivo. Her findings indicate that many oncogenes could promote the malignant behavior of tissues because they alter cell contractility and induce tissue fibrosis. Mechanistically her data argue that at the cellular level high force promotes integrin adhesions to alter growth factor receptor signaling and at the tissue level high force alters the vasculature to promote hypoxia and impede drug delivery and stimulates chemokine expression to modify the immune response. In support of this hypothesis she presented work relating to pancreatic fibrosis and a JAK-Stat3-integrin-tissue tension circuit that fosters pancreatic tumor progression/aggression. Understanding how intrinsic and extrinsic force regulates tumor behavior and treatment response may reveal novel targetable pathways towards which new therapies can be developed. Dr. Weaver’s work will address the fundamental issues of how abnormal cell and tissue mechanics of tumors contribute to treatment resistance, and the role of oncogenes in this process.
Scott Powers proposed a more focused approach to developing therapeutic strategies from the extensive genomic characterizations of DNA copy number alterations across many tumor types. Most copy number alterations are large and likely involve the action of tens perhaps hundreds of dosage-sensitive oncogenic drivers and tumor suppressors. In contrast, the number of focal amplicons like HER2 that can be considered as a distinct genetic event, independent of other nearby changes, is limited. Based on analysis of expression and protein data in TCGA, showed that our current understanding of how the HER2 amplicon works is backed by human data but that our current understanding of how the CCND1 amplicon works is not supported at all. Scott proposed that new genetic tools and models will be needed to develop a better understanding of recurrent focal amplicons and hopefully lead to new targeted therapeutic strategies.
Jeff Engelman described the response of lung cancer patients to drugs targeting EGFR or ALK. Transient responses are usually followed by resistance in most patients, and re-biopsy of the lung lesions revealed that on-target gatekeeper mutations occurs in many cases, whereas parallel pathways are activated in others. The parallel pathways include amplification of c-MET, and therefore strategies that include chemical kinase inhibitors that resist gatekeeper mutations, and those that also target c-MET and other bypass kinases, are under way. Finally, the surprising observation that adenocarcinoma cells display altered differentiation to neuroendocrine cancers resembling small cell carcinoma has been made in a minority of patients. These neuroendocrine-like cells still harbor the mutant EGFR alleles, and respond to classical SCLC regimens such as etoposide and cisplatin. Dr. Engelman hypothesized that altered cellular differentiation states may be another important resistance mechanism for human cancer, as appears to be the case in prostate cancer following blockade with new AR antagonists.
Alberto Bardelli presented models of colorectal cancer progression based on genetic and pharmacological data. He proposed that CRC develops in at least three major subtypes all of which start by acquiring an APC/beta cat mutation. At this stage all 3 subtypes are highly dependent on EGFR GFs provided by the environment. Upon selective forces imposed by yet unknown pressures the first groups survives by mutational activation of KRAS/NRAS and BRAF. The second subtype evolves through acquisition of deregulated RTKs (HER2, ALK, NTRK, IGFR1). The third subgroup progresses by overcoming the selective pressure through increased production of EGFR ligands (autocrine loop). Dr Bardelli also showed that EGFR blockade with the monoclonal antibodies cetuximab and panitumumab is effective in cells ad patients only in the third subgroup. Dr Bardelli went on to identify the mechanisms of acquired resistance to EGFR blockade and described an approach based on individual cell barcoding to visualize clonal evolution under drug selection. Using this approach he argued that in addition to the major clones that emerge during drug selection a large number of minority populations also survive indicating the presence of additional mechanisms of resistance or persistence in the population. The polyclonal nature of therapeutic resistance can also be observed in patients by measuring the concomitant emergence of multiple mutations in KRAS, NRAS and BRAF in the blood of individual treated with anti EGFR antibodies. Finally he presented data suggesting that the concomitant blockade of EGFR and MEK intercepts multiple mechanisms of acquired resistance to EGFR blockade. These findings represent the basis for clinical trials that are ongoing in Dr Bardelli’s institution.
Richard Marais showed that ultraviolet (UV) light accelerates BRAF-driven melanoma by targeting TP53. Sunscreen slows the development of melanoma, but offers incomplete protection, so it should be used in conjunction with other strategies to protect the skin from the damaging effects of UV light. He described the induction of paradoxical activation of ERK signaling by BRAF inhibitors in tumors when RAS is active in cells. He described use of biochemical and genomic approaches to identify mechanisms of resistance to BRAF inhibitors in BRAF mutant melanoma patients and commented that many mechanisms of resistance exist, although not all mechanisms have been validated in patients. Paradox breaking inhibitors appear to be active in resistant tumors. He also showed that reactivation of the ERK pathway in resistant cells drives invasion and metastasis through upregulation of IL8 and expression of extracellular proteases. Thus, resistance may be accompanied by increased metastasis in the presence of the drug. Also, he described that resistant cells appear to switch from glucose to glutamine as a major carbon source and this appears to contribute to the resistant phenotype. He concluded that it is important to develop precision medicine approaches to improve outcomes for melanoma patients who have developed resistance to targeted therapies.
Hugues de The discussed how his laboratory has been interested in the molecular basis for therapy response in the field of cancer. I first presented data on breast cancer linking P53 status to initial response to dose-dense chemotherapy. In the neo-adjuvant setting, we demonstrated that pathologically complete remissions in P53-mutant tumours were followed by long-term survival, likely pointing to the cure of the majority of these patients. Thus, P53-mutant breast cancers are exquisitely sensitive to dose-dense alkylation. I then presented data on the acute promyelocytic leukemia, a disease that is exquisitely sensitive to retinoic acid and arsenic. We showed evidence that differentiation is not the primary basis for APL cure. Animal models argue that PML/RARA degradation by these two drugs is the basis for irreversible loss of leukemia self-renewal. Loss of the driving oncogene triggers a PML/P53 senescence program that is ultimately responsible for APL eradication. Mutations that abolish degradation by arsenic or retinoic acid were associated to therapy resistance. Resistance to arsenic maps to the PML part of PML/RARA, while resistance to retinoic acid is due to mutations in the RARA part. Other models were presented in which therapy-triggered degradation of the driving oncogene is associated to remission.
Jos Jonkers discussed mouse models of TNBC, and therapeutic approaches using GEMMS and PDX models. Dr. Jonkers demonstrated that strong alkylating agents could cure mouse models of Brca1/Brca2/p53 mutant TNBC, and old clinical data support the use of these agents. Furthermore, mechanisms of escape for PARP inhibitors and Adriamycin could also be demonstrated using these TNBC allograft models, including drug efflux pumps. Dr. Jonkers suggests that methods to cure people of TNBC can be developed by combining cytotoxic drugs, targeted drugs, drugs that block resistance mechanisms, and immunotherapies.
Shridar Ganesan analyzed published clinical data to develop a model of how adjuvant chemotherapy works to prevent breast cancer recurrence. In this model a stochastic transition of disseminated metastatic cells from a dormant/G0 state to active growth is postulated to drive recurrence. Different breast cancer classes are characterized by different transition rates, that leads to different dynamics of relapse. Analysis data from adjuvant treatment trials suggest that adjuvant therapies can only eliminate metastatic foci that are in active growth during the time in which adjuvant therapy is delivered. One implication is that we need to reconsider how to optimize scheduling and dosing of chemotherapy to improve outcomes in early stage breast cancer. This work poses the following clinically relevant question. First, what is the optimal dosing schedule of chemotherapy to deliver maximum efficacy and minimal toxicity? Second, is maximizing time of exposure to minimum effective dose more important than achieving maximum tolerated dose? Finally, how can non-cycle/G0 cancer cells best be killed?