A common characteristic of cancer is an increased glucose metabolism to sustain a high proliferation rate. This phenotype manifests in cultured cancer cells, in animal models, and most importantly in humans as demonstrated by the use of 18F-fluorodeoxyglucose (18F-FDG) PET imaging to diagnose cancer in clinics throughout the world. Previous studies suggest that cancer cells use the same biochemical enzymes and proteins as normal cells to consume glucose but different signaling pathways to regulate these enzymes and proteins. Targeting these cancer-specific pathways could be an effective strategy for treating cancer while sparing healthy tissue. But to do this, we would need a better understanding of these pathways and effective ways to target them.
The aim of this project was to identify proteins and pathways that regulate glucose consumption in cancer, with a particular focus on lung cancer. Basically, the question we were trying to answer was: how do we target glucose consumption in cancer but not healthy tissue? When we began this project – almost four years ago when the lab first started – we tried to approach this question using the standard 3H-2-deoxyglucose uptake assay for measuring glucose consumption. This worked well but was also really slow and laborious (two bad adjectives to describe an important assay in a young lab!). We could confirm what we already knew (for example, EGFR inhibitors block glucose consumption in EGFR-driven lung cancer) but had trouble discovering anything new. We knew we had to come up with a better way.
Our lab is two floors below the Molecular Screening Shared Resource (MSSR) high-throughput screening facility at UCLA. If you run into the MSSR director – Dr. Robert Damoiseaux – enough times (which we did), he’ll convince you to run a high-throughput screen on anything. So with Robert’s help, we developed and validated a high-throughput assay for measuring cellular glucose consumption.
Working at the MSSR, we developed a high-throughput assay based on the commercial Glucose Uptake-Glo kit from Promega that we adapted and modified to be used with automated liquid handling machinery. Cells are plated in 384-well plates, treated with small molecule inhibitor libraries, and 24 hours later both glucose consumption and cell numbers are measured, all in high-throughput automation. We validated that our assay had a high dynamic range and low variance, important properties for use of the assay in a high-throughput screen. We were ready to discover new inhibitors of cancer cell glucose consumption.
Using this assay, we screened 4 small molecule libraries containing ~3600 inhibitors, and we discovered a series of compounds that inhibit non-small cell lung cancer glucose consumption. We were really excited about the small molecule Milciclib – an inhibitor of cyclin dependent kinases – that blocked glucose consumption in one of three cell lines we studied and which suggested the possibility of further links between cell cycle regulation and metabolism. We set out to characterize how this drug reduces glucose uptake.
By combining standard biochemical assays (i.e. Western blots and qPCR) with a glucose sensor-based FRET assay, we found that Milciclib blocks glucose consumption by reducing GLUT1 protein levels and therefore activity. No change in the intracellular glucose metabolism (hexokinase activity) was identified.
We additionally identified that the Milciclib target that regulates glucose consumption is the cyclin-dependent kinase 7 (CDK7) and showed that the effect is specific to cells with activating mutations in PIK3CA. PIK3CA mutations are unique to cancer and not found in healthy tissue, and by using 18F-FDG PET imaging, we saw that Milciclib blocks glucose consumption in mouse xenograft tumors but not in normal tissue. Additional data suggests a model in which PIK3CA activates PKCι to phosphorylate CDK7 and regulate RNA transcription. Milciclib - by targeting CDK7 - reduces the amount of CDK7 that can activate RNA polymerase II and the expression of GLUT1. And notably, if you block the ability of Milciclib to inhibit glucose consumption, you block its ability to inhibit cell growth in the PIK3CA mutant cells.
We’re really excited about this work because we think we’ve identified (1) a drug that does not affect normal cell glucose consumption but that does block glucose consumption in lung cancer sufficiently to inhibit cell growth; (2) the genetic driver required for the drug to block glucose consumption; and (3) a non-invasive, clinical biomarker (18F-FDG PET) to evaluate drug efficacy.
Cancer is one of many diseases characterized by a dysregulation of glucose consumption. Will this method work to find regulators of glucose consumption in other systems or diseases? We think (we hope!) so, but that remains to be determined.
In conclusion, our results show a new pathway involved in GLUT1 regulation. Additionally, we describe a new high-throughput method to test glucose consumption in cancer cells and more generally in any cell/disease that shows a dysregulation of glucose metabolism.
Chiara Ghezzi PhD and Peter M. Clark PhD