An insect’s heat limit leaves clues in its proteins

A March Nature study measured thermal tolerance in roughly 8,000 insects from Peru and Kenya. In warming tests, the endpoint was loss of mobility. The researchers also asked whether differences among groups had a molecular counterpart.[1]

A temperature prediction from a sequence

Proteins depend on their folded structures to function. Thermal proteome profiling follows what happens as samples are heated: proteins that lose stability become insoluble, and mass spectrometry measures those remaining in solution. A melting curve records how that soluble fraction changes with temperature.[2]

DeepSTABp, published in 2023, learns from these experiments. A protein language model converts the amino-acid sequence into numerical features. Further network components incorporate growth temperature and whether measurements concern intact cells or cell extracts. The output is a predicted melting temperature, rather than a three-dimensional structure.[2]

Its development dataset contained 35,112 protein sequences. In the reported held-out test, its mean absolute error was 3.22°C, compared with 4.95°C for ProTstab2. That performance describes its protein benchmark; it is not a measured error for predicting an insect’s heat limit.[2]

Comparing families

The insect researchers sampled proteins from hundreds of available genomes. They held DeepSTABp’s growth-temperature input at 22°C and its experimental setting at intact cells, making the sequence comparisons under common conditions.[1]

For the most heat-sensitive quarter of predicted proteins, mean melting temperatures ranged from 41.17°C in flies to 43.40°C in grasshoppers and their relatives. Across families, higher predicted protein stability was associated with higher measured heat tolerance.[1]

The family-level association identifies protein stability as a plausible contributor to heat tolerance. It is not a direct measurement of protein melting in the field-tested insects.[1]

Report an error or suggest a correction