Figure 5. Sequence heuristics enable engineering of protein variants with tunable thermoresponsive condensation.

Nine protein variants were designed in three sets: Set 1—polar or aromatic residues mutated to achieve 50% aliphatic enrichment; Set 2—polar residues mutated to achieve 50% aliphatic enrichment; and Set 3—polar residues mutated to achieve 25% aliphatic and 25% aromatic enrichment. (a) The wild-type CAP10 sequence and its three designed variants are shown as representative examples. (b) Direct coexistence simulations using the Mpipi-T model were used to determine dense phase densities of variants at 4 °C, 22 °C, and 37 °C. The simulations predicted that Set 3 variants, enriched in both aliphatic and aromatic residues, would exhibit the strongest heat-induced condensation. Experimental high-throughput protoplast-based assays were performed at 4 °C, 22 °C, and 37 °C for each of these computationally designed variants to quantify the percentage of cells with condensates. The experimental results recapitulated the simulation predictions (Pearson r = 0.726, p = 0.02686), validating that polar-to-aliphatic and aromatic substitutions maximize condensation propensity under heat stress. Bars represent the mean across variants in a set, and individual variants are shown as points. (c) Experimental quantification of the percentage of cells forming condensates at 4 °C, 22 °C, and 37 °C for all designed variants. An average of 30 cells were imaged and quantified.

(d) Representative microscopy images show condensate formation for CAP10 protein variants under the three temperature conditions. YFP signal image (Left) and bright field image (right) for each temperature and variants. Scale bar = 10 μm.