Inducible protein switches are currently limited for use in tissues and organisms because common inducers cannot be controlled with precision in space and time in optically dense settings. Here, we introduce a protein that can be reversibly toggled with a small change in temperature, a stimulus that is both penetrant and dynamic. This protein, called Melt (Membrane localization using temperature) oligomerizes and translocates to the plasma membrane when temperature is lowered. We generated a library of Melt variants with switching temperatures ranging from 30 °C to 40 °C, including two that operate at and above 37 °C. Melt was a highly modular actuator of cell function, permitting thermal control over diverse processes including signaling, proteolysis, nuclear shuttling, cytoskeletal rearrangements and cell death. Finally, Melt permitted thermal control of cell death in a mouse model of human cancer. Melt represents a versatile thermogenetic module for straightforward, non-invasive and spatiotemporally defined control of mammalian cells with broad potential for biotechnology and biomedicine. The Melt (Membrane localization using temperature) protein translocates to the plasma membrane upon temperature shift. Melt variants with a range of switching temperatures enable straightforward thermogenetic control of diverse cellular processes.

Temperature is an important biological stimulus, yet there is a lack of approaches to modulate the temperature of biological samples in a dynamic and high-throughput manner. The thermoPlate is a device for programmable control of temperature in a 96-well plate, compatible with cell culture and microscopy. The thermoPlate maintains feedback control of temperature independently in each well, with minutes-scale heating and cooling through ΔT = 15-20°C. We first used the thermoPlate to characterize the rapid temperature-dependent phase separation of a synthetic elastin-like polypeptide (ELP53). We then examined stress granule (SG) formation in response to dynamic heat stress, revealing adaptation of SGs to persistent heat and formation of a memory of stress that prevented SG formation in response to subsequent heat shocks. The capabilities and open-source nature of the thermoPlate will empower the study and engineering of a wide range of thermoresponsive phenomena. A record of this paper's transparent peer review process is included in the Supplemental information.

Optogenetic tools respond to light through one of a small number of behaviors including allosteric changes, dimerization, clustering, or membrane translocation. Here, we describe a new class of optogenetic actuator that simultaneously clusters and translocates to the plasma membrane in response to blue light. We demonstrate that dual translocation and clustering of the BcLOV4 photoreceptor can be harnessed for novel single-component optogenetic tools, including for control of the entire family of epidermal growth factor receptor (ErbB1-4) tyrosine kinases. We further find that clustering and membrane translocation are mechanistically linked. Stronger clustering increased the magnitude of translocation and downstream signaling, increased sensitivity to light by ~threefold-to-fourfold, and decreased the expression levels needed for strong signal activation. Thus light-induced clustering of BcLOV4 provides a strategy to generate a new class of optogenetic tools and to enhance existing ones.

Protein clustering is a powerful form of optogenetic control, yet there is currently only one protein -Cry2-whose light-induced clustering has been harnessed for these purposes. Recently, the photoreceptor BcLOV4 was found to form protein clusters in mammalian cells in response to blue light, although clustering coincided with its translocation to the plasma membrane, potentially constraining its application as an optogenetic clustering module. Herein we identify key amino acids that couple clustering to membrane binding, allowing us to engineer a variant of BcLOV4 that clusters in the cytoplasm and does not associate with the membrane in response to blue light. This variant, BcLOVclust, clustered over many cycles with dramatically faster clustering and de-clustering kinetics compared to Cry2. The magnitude of BcLOVclust clustering could be strengthened by appending an intrinsically disordered region from the fused in sarcoma (FUS) protein, or by optimizing the fluorescent protein to which it was fused. BcLOVclust retained the temperature sensitivity of BcLOV4 such that light induced clustering was transient, and the rate of spontaneous declustering increased with temperature. At low temperatures, BcLOVclust and Cry2 could be multiplexed in the same cells, allowing light control of independent protein condensates. BcLOVclust could also be applied to control signaling proteins and stress granules in mammalian cells. Thus BcLOVclust provides an alternative to Cry2 for optogenetic clustering and a method for multiplexed clustering. While its usage is currently suited for organisms that can be cultured below ~30 °C, a deeper understanding of BcLOVclust thermal response will further enable its use at physiological mammalian temperatures.

Protein‐based condensates have been proposed to accelerate biochemical reactions by enriching reactants and enzymes simultaneously. Here, we engineered those condensates into a photo‐activated switch in Escherichia coli (PhASE) to regulate enzymatic reactions via tuning the spatial correlation of enzymes and substrates. In this system, scaffold proteins undergo liquid–liquid phase separation (LLPS) to form light‐responsive compartments. Tethered with a light‐responsive protein, enzymes of interest (EOIs) can be recruited by those compartments from cytosol within only a few seconds after a pulse of light induction and fully released in 15 min. Furthermore, we managed to enrich small molecular substrates simultaneously with enzymes in the compartments and achieved the acceleration of luciferin and catechol oxidation by 2.3‐ and 1.6‐folds, respectively. We also developed a quantitative model to guide the further optimization of this demixed regulatory system. Our tool can thus be used to study the rapid redistribution of proteins, and reversibly regulate enzymatic reactions in E. coli .

Enzymatic reactions in cells are well organized into different compartments, among which protein-based membraneless compartments formed through liquid-liquid phase separation (LLPS) are believed to play important roles 1,2 . Hijacking them for our own purpose has promising applications in metabolic engineering 3 . Yet, it is still hard to precisely and dynamically control target enzymatic reactions in those compartments 4 . To address those problems, we developed Photo-Activated Switch in E. coli (PhASE), based on phase separating scaffold proteins and optogenetic tools. In this system, a protein of interest (POI) can be enriched up to 15-fold by LLPS-based compartments from cytosol within only a few seconds once activated by light, and become fully dispersed again within 15 minutes. Furthermore, we explored the potentiality of the LLPS-based compartment in enriching small organic molecules directly via chemical-scaffold interaction. With enzymes and substrates co-localized under light induction, the overall reaction efficiency could be enhanced. Using luciferin and catechol oxidation as model enzymatic reactions, we found that they could accelerate 2.3-fold and 1.6-fold, respectively, when regulated by PhASE. We anticipate our system to be an extension of the synthetic biology toolkit, facilitating rapid recruitment and release of POIs, and reversible regulation of enzymatic reactions.