Cancer metabolism as a target for drug discovery: Agios Pharmaceuticals

In the December 10 2009 issue of Nature, researchers at Agios Pharmaceuticals (Cambridge, MA) and their academic collaborators published an article implicating mutations in a metabolic enzyme, cytosolic isocitrate dehydrogenase (IDH1) as a causative factor in a major subset of human brain cancers.

The mutated forms of IDH1 are found in around 80% of human grade II-III gliomas and secondary glioblastomas. The mutations occur in arginine 132, which is usually mutated to histidine. (In other less common mutations, arginine 132 is mutated to serine, cysteine, glycine, or leucine.) Typically, only one allele of IDH1 is mutated. These mutations appear to occur early in the process of tumorigenesis, and often appear to be the first mutation that occurs. The mutant forms of IDH1 are also found in a subset of acute myelogenous leukemia (AML).

The wild-type form of IDH1 catalyzes the NADP+-dependent oxidative decarboxylation of isocitrate to α-ketoglutarate. However, the researchers found that the mutant forms of IDH1 no longer catalyzes this reaction, but instead catalyzes the NADPH-dependent reduction of α-ketoglutarate to R(-)-2-hydroxyglutarate (2HG). This is the result of changes in the active site of the enzyme, as demonstrated by structural studies carried out by the researchers. Tumors that harbor the mutant form of IDH1 have elevated levels of 2HG. The researchers therefore hypothesize that these elevated levels of 2HG are a causative factor in tumorigenesis and/or tumor progression in human gliomas.

This hypothesis is supported by the effects of the familial metabolic disorder 2-hydroxyglutaric aciduria. This disease is caused by a deficiency of 2-hydroxyglutarate dehydrogenase, an enzyme that converts 2HG to α-ketoglutarate. Patients with this metabolic disease have elevated levels of 2HG in bodily fluids and in the brain, and an increased risk of developing brain tumors.

The mechanism by which 2HG might contribute to tumorigenesis is unknown. The authors advance several hypotheses, including increasing reactive oxygen species (ROS) levels, serving as an NMDA (N- methyl-D-aspartate) receptor agonist, and competitive inhibition of enzymes that use glutamate and/or α-ketoglutarate resulting in the induction of hypoxia-inducible factor-1α, a transcription factor that facilitates tumor growth under conditions of hypoxia.

According to the authors, these results suggest that in patients with low-grade gliomas containing mutant forms of IDH1, therapeutic inhibition of 2HG production may slow or halt progression of these tumors to lethal secondary glioblastomas. 2HG levels may also be used as a prognostic test for IDH1 mutations, since patients with these mutations tend to live longer than patients with gliomas that have other mutations.

The company that led this research, Agios Pharmaceuticals, is developing a pipeline of oncology drugs based on targeting metabolic pathways in cancer cells. Interestingly, Agios means “holy” in Greek.

Way back in 1924, Otto Warburg demonstrated a difference between cancer cells and normal adult cells in glucose metabolism. In the presence of oxygen, most normal adult cells metabolize glucose to pyruvate via the process of glycolysis, generating two molecules of ATP (the energy currency of the cell) per glucose molecule. In the mitochondria, they then utilize oxygen to catabolize pyruvate to CO2 and water, in the process generating 36 molecules of ATP per glucose molecule. Cancer cells, however, predominantly carry out aerobic glycolysis, in which they carry out glycolytic conversion of glucose to pyruvate, followed by reduction of pyruvate to lactate. Despite the presence of oxygen, cancer cells generate the bulk of their ATP from glycolysis, not mitochondrial oxidative phosphorylation, in the process consuming large amounts of glucose. The reliance of cancer cells on aerobic glycolysis for their metabolism is known as the “Warburg effect”.

Agios’ platform is based in part on the work of signal-transduction pioneer Lewis Cantley (Beth Israel Deaconess Cancer center/Harvard Medical School, Boston MA). It is Dr. Cantley’s work on the connection between growth factor-mediated signal transduction and aerobic glycolysis that is the basis for Agios’ platform. In particular, Dr. Cantley and his colleagues found that pyruvate kinase M2 (PKM2) is a link between signal transduction and aerobic glycolysis. PKM2 binds to tyrosine-phosphorylated signaling proteins, which results in the diversion of glycolytic metabolites from energy production via mitochondria oxidative phosphorylation to anabolic processes required for rapid proliferation of cancer cells.

Agios closed a $33 million Series A financing in July 2008, co-led by Third Rock Ventures, Flagship Ventures and ARCH Venture Partners. In June 2009, Fierce Biotech named Agios to the 2009 FierceBiotech “Fierce 15” list. On December 21, 2009, Agios received funding from the nonprofit organization Accelerate Brain Cancer Cure (ABC2), to supplement Agios’s research on the development of IDH1-based therapeutics and diagnostics. Agios expects to have a lead compound in the clinic some time in 2010.

The Agios website calls cancer metabolism “one of the most exciting new areas of cancer research”. But the study of cancer metabolism, and especially the Warburg effect, is not new—the Warburg effect is a classic observation going back 85 years. Moreover, biotechnologists working in such areas as production of recombinant proteins in CHO cells have been familiar with aerobic glycolysis, which is carried out by most mammalian cell lines in culture, for decades. Nevertheless, cancer metabolism has been well out of the mainstream of cancer drug discovery. It was Dr. Cantley’s work, which links the classic Warburg effect to the mainstream area of signal transduction and protein kinases, which has made Agios’ platform possible.

Similarly, it was Julian Adams’ work on the biology of the proteasome in the 1990s, through a series of biotechnology company mergers that eventually led him to Millennium Pharmaceuticals (now Millennium: The Takeda Oncology Company), which resulted in Millennium’s proteasome inhibitor Velcade (bortezomib). Velcade, the only proteasome inhibitor on the market, is now approved by the FDA for the treatment of multiple myeloma and mantle cell lymphoma. Prior to Dr. Adams’ work, proteasome biology and protein degradation were out of the mainstream of cancer drug discovery. Now Joseph Bolen, the chief scientific officer of Millennium, sees “protein homeostasis” as one of the most exciting areas of cancer research.

Finally, although the development of protein kinase inhibitors to target signaling pathways in cancer is now well within the mainstream of oncology drug discovery, prior to the discovery and development of imatinib (Novartis’ Gleevec/Glivec) (approved by the FDA in 2001), specific targeting of protein kinases was though to be unlikely, since all of these enzymes have a high degree of similarly in their ATP binding sites. Thus the field of protein kinase inhibitors did not enter the mainstream until the late 1990s-early 2000s.

The take-home lesson is that drug developers may find fertile areas for innovation in seemingly obscure or out-of-the mainstream areas of biology (or of chemistry, as we have discussed in previous blog posts). Some of these areas may be technologically premature, and not quite ready for exploitation by drug developers. However, as demonstrated by our blog post on monoclonal antibodies, even some technologically premature areas may yield to innovators who are willing and able to develop enabling technologies to move these areas up the development curve.

Small-molecule drugs for targeting an intracellular signaling pathway: research and development of new oncology drugs

In our November 27th blog post, we discussed an innovative new technology, stapled peptides, for use in targeting intracellular protein-protein interactions. In the example we gave, the target was a transcription factor complex in the Notch pathway. As we stated, protein-protein interactions are deemed to be “undruggable”, since they cannot be readily addressed with small molecule drugs.

Nevertheless, in some cases, small molecules have been discovered that do address key protein-protein interactions, and which may become clinical candidates.

Back in February 2006, Decision Resources published our report, “Protein-Protein Interactions: Are They Now Druggable Targets?" Among the case studies we discussed in that report was one in which researchers were attempting to discover small-molecule agents that targeted the Wnt pathway. The researchers discovered small-molecule agents that, as with the stapled-peptide example we discussed in our previous blog post, targeted a transcription factor complex. As of late 2009, two of these compounds are in preclinical development for treatment of various cancers.

Mutations that mediate deregulation of the Wnt pathway are causative factors in several types of cancer, most notably colorectal cancer, as well as multiple myeloma (MM), hepatocellular carcinoma (HCC), and B-cell chronic lymphocytic leukemia B-CLL). In the canonical Wnt pathway, soluble extracellular factors that are members of the Wnt family activate the pathway. A complex that includes the protein adenomatous polyplosis coli (APC) is central to the Wnt pathway. When Wnt receptors are not engaged by their ligands, kinases in the APC complex phosphorylate β-catenin, a multifunctional protein that is involved both in signal transduction and in adhesion between cells. Phosphorylation targets β-catenin for degradation.

When Wnt proteins bind to their receptors, the kinase activity of the APC complex is inactivated. This results in the accumulation of β-catenin, which moves into the nucleus. There it binds to proteins of the T cell factor (Tcf) family. β-catenin binding changes Tcf from a transcriptional repressor into a transcriptional activator. Downstream genes controlled by the β-catenin/Tcf complex include the oncogene Myc and other genes that mediate cell proliferation.

In precancerous colonic adenomas or the colorectal cancers that they may evolve into, APC is usually mutated. This results in constitutive stabilization of β-catenin and constitutive activation of Tcf and its downstream genes. In other types of cancer that involve constitutive Wnt pathway activation, β-catenin also becomes stabilized, via other means. This makes the Tcf/β-catenin a tempting target for drug discovery. However, it is a protein-protein interaction, and is thus deemed “undruggable”.

In 2004, A group led by Ramesh Shivdasani (Harvard Medical School, Dana-Farber Cancer Institute, and Brigham and Women’s Hospital, Boston MA), including researchers from the Novartis Institutes for BioMedical Research (Cambridge, MA), discovered several small-molecule inhibitors of the interaction between human Tcf4 and human β-catenin.

Dr. Shivdasani’s group, among others, had previously determined crystal structures of Tcf-β-catenin complexes. The interaction between the two proteins occurs over a large surface area. It is the large, and usually hydrophobic, interface between proteins in protein-protein interactions that forms the theoretical basis for the difficulty of addressing these interactions with small molecules. However, there is a small hydrophobic pocket that is critical for binding (as also confirmed by site-specific mutation studies), which might accommodate a small molecule inhibitor.

Therefore, the researchers screened approximately 7,000 purified natural products from public and proprietary libraries using an enzyme-linked immunosorbent (ELISA) assay involving release of a labeled Tcf4 binding fragment from its complex with a β-catenin fragment absorbed to an ELISA plate. Eight compounds were found that gave reproducible, concentration-dependent release of the Tcf4 fragment at less than 10 micromolar concentration. The structures and purity of these compounds (most of which are complex, multi-ringed planar compounds with multiple hydroxy groups) were then determined. The sources of these compounds include fungi, actinomycetes, and a marine sponge.

The researchers performed several additional biochemical assays to confirm the compounds’ specific disruption of the Tcf/β-catenin complex, and also performed cellular assays and an in vivo assay in the Xenopus (frog) embryo to study the activities of these compounds against β-catenin-mediated cellular effects. Each of the eight compounds shows different levels of potency in the different assays used in this study, and the compounds differ from each other in their activities in the different assays.

Two fungal-derived compounds, PKF115-854 and CGP04909, gave the best results in all the assays. It is those compounds that have been tested in preclinical studies as potential oncology drug candidates. In a study published in PNAS in 2007, researchers at the Dana-Farber and at Brigham and Women’s Hospital tested PKF115-584 in human MM cells in vitro and in xenograft models. The compound blocked expression of Wnt target genes, induced cytotoxicity in MM cells in vitro, and inhibited tumor growth and prolonged survival in the xenograft model. In a study in HCC at the Asian Liver Center at Stanford University School of Medicine, PKF115-584, CGP049090, and another of the Shivdasani group’s compounds, PKF118-310, also induced cytotoxicity in human HCC cell lines in vitro, and suppressed tumor growth and induced apoptosis in tumor cells in a human HCC xenograft model. Finally, in an abstract presented at the American Society of Hematology (ASH) meeting in December 2009, researchers at the Novartis Institute for Biomedical Research in Basel and their academic collaborators presented data that showed that CGP04090 and PKF115-584 potently inhibited the survival of primary human B-CLL cells in vitro and in vivo. In all three cases, the compounds showed no significant cytotoxicty against normal cells.

In the conclusion of the ASH meeting abstract, the authors stated that further investigations are warranted to determine the feasibility of testing these compounds in human clinical trials.

Many medicinal chemists remain skeptical about the ability of researchers to develop small-molecule drugs that target protein-protein interactions, which have satisfactory pharmacokinetics and can advance through clinical trials and reach the market. However, at least one nonpeptide small-molecule compound that targets a protein-protein interaction, the thrombopoietin receptor agonist eltrombopag (Ligand/GSK’s Promacta), has reached the market. (The FDA approved it in November 2008.) Several other small-molecule drugs that target protein-protein interactions are in clinical development. And Cambridge Healthtech Institute will be sponsoring a conference on this subject, which is scheduled for April 2010. This conference is in its third year. Thus, as also shown by the development of stapled peptides, there is renewed interest in discovering and developing drugs that address these “hard targets”.