A team of scientists led by systems biologist and biomedical engineer Andre Levchenko from Yale University has developed a novel mechanism for mapping the biochemical variability, or ‘noise’, in how human cells respond to chemical signals.
This research could potentially lead to tailored drug delivery to a patient’s individual cell responses and may have implications for advances in semiconductor chip design.
Levchenko’s method is founded on the recognition that every cell reacts uniquely to the body’s chemical signals, even if the cells are all from the same patient and even the same tissue — some cells may react strongly, while other cells may not react at all. A wide diversity of responses is considered a noisy response.
The new method maps noise across multiple branches of complicated biochemical networks. “Knowing how variable the activity is allows us to better target the spectra of activities in those networks,” said Levchenko, the John C. Malone Professor of Biomedical Engineering and inaugural director of the Yale Systems Biology Institute.
“For example, if a specific cell network’s spectra of response is less noisy, then a comparatively small drug dosage could target the entire spectra. Our mapping technique enables researchers and clinicians to identify those less noisy networks, which could be unique for each patient,” he said.
For this research, Levchenko’s team sought to understand the noise in the different biochemical signalling pathways activated by the cytokine tumour necrosis factor (TNF), a model system for understanding signalling heterogeneity in mammalian cells. TNF is commonly produced by cells in response to infection in order to activate the first line of the immune response.
Using a combination of experimental observations and mathematical algorithms, the team measured the effect of TNF input for a small number of target molecules, then inferred how the signal triggers by TNF propagated through the network. Because the TNF signal originated from the same point, the team could efficiently reconstruct how different branches of the cell communication networks reacted to the signal without measuring the dozens of molecules affected.
The team exposed mouse embryonic fibroblast cells to a wide range of TNF concentrations to elicit the full dynamic response of the transcription factors. For each TNF concentration, they measured the nuclear concentrations of the transcription factors in hundreds of individual cells using quantitative immunocytochemistry, where the immunofluorescence signal has a strong correlation with the protein concentration.
Many signalling networks, including that of TNF, consist of multiple levels of branching. For instance, the TNF network branches into the NF-?B and JNK pathways, and the JNK pathway subsequently branches to activate two transcription factors: ATF-2 and c-Jun. In response to a stimulus such as TNF, parallel signalling branches can have different dose dependencies leading to complex overall response characteristics, including biphasic qualities resulting in complex and highly nonlinear behaviour.
The team measured the noise in each of the pathways as relative to the trunk of the pathway or the branches. They found that the JNK branch-specific noise was higher than both the NF-?B branch-specific noise and the TNF–TNFR trunk noise. Within the JNK pathway the c-Jun branch noise was greater than the ATF-2 branch noise at higher TNF concentrations.
“Previous experiments in this field, including our own, would focus on these network responses by looking at the average cell behaviour over perhaps millions of cells at a time,” said Levchenko. “The new method is unique in that it requires relatively few targets — we observed just three target transcription factors — to reconstruct not only how responsive but also how noisy various branches of the signalling network are. Using this effective methodology, we can now embark on extensive mapping of the sources of noise across signalling networks.”
In turn, identifying which networks were noisier enabled Levchenko’s team to experimentally confirm that noise tends to increase as the communication chain gets longer, something that could be applicable to research of not only biological networks but even electrical networks.
“Despite the noise in cellular networks,” he said, “biology still allows cells and organisms to perform well. Similarly, as today’s electronic components become smaller, chip designers more often need to study and circumvent signalling noise. For this reason, the challenges of building effective computational devices and designing effective medical therapeutics are more similar than meets the eye.”