High-resolution images of human metabolism have been caught on camera for the first time thanks to a new technique that can track glucose metabolism in a single cell. Tracking glucose could provide insights and even new therapies to treat cancer – a disease which often disrupts cellular metabolism.
The new technology combines fluorescence resonance energy transfer (Fret) between two light-sensitive molecules with machine learning – a form of analysis based upon the idea that computer systems can ‘learn’ from data, making the analysis more accurate. The technology has enabled Yun Fang and his team at the University of Chicago to visualise the metabolism of glucose – glycolysis. The team genetically engineered human cells to express specific Fret biosensor molecules that glow when glucose is broken down, enabling the scientists to capture the process on camera using a fluorescent microscope.
Glucose is a vital energy source for almost all cell types, but the process of glycolysis in cancer often goes wrong. This can contribute to the diseased state of a malignant cell, supporting its ability to move, grow and divide. By pairing Fret technology with a new machine learning algorithm, Fang’s team were able to obtain images of glycolysis at a previously unachievable resolution, showing exactly which parts of the cell were using glucose in real time. ‘Now we can look at and understand details within the cells, like certain areas of cells where there is an increase of glycolysis,’ Fang says. ‘This is a key technological innovation.’
The new machine learning approach to existing Fret technology has opened up a new range of experimental possibilities. Glucose metabolism can now be observed alongside other visible cell processes allowing the team to show that some human cells consume more glucose when they move and twitch – something that would not have been possible to demonstrate in a single experiment before.
Mis-regulated metabolism can result in some cancerous cells metastasising and invading new tissues. The new technique should help scientists better understand the link between these disease processes, and could help discover new therapeutic approaches. The new technology also led researchers to discover a previously unknown cell surface receptor capable of glucose uptake.
‘Coupled with existing technology, this [machine learning approach] will help further our understanding of how glucose and other key energy molecules influence the metabolic rewiring in cancer development, and could be used to develop new cancer drug targets,’ says Gemma Beasy of the Quadram Institute, who studies the effects of glucose metabolism in prostate cancer. ‘This is an amazing step forward in technology that could be used in cancer research.’
Alongside cancer research, broader applications of Fret-based machine learning technology are already being explored, including helping patients’ whose immune systems overreact to Covid-19.