Photo-sensitive degron variants for tuning protein stability by light

Background Regulated proteolysis by the proteasome is one of the fundamental mechanisms used in eukaryotic cells to control cellular behavior. Efficient tools to regulate protein stability offer synthetic influence on molecular level on a selected biological process. Optogenetic control of protein stability has been achieved with the photo-sensitive degron (psd) module. This engineered tool consists of the photoreceptor domain light oxygen voltage 2 (LOV2) from Arabidopsis thaliana phototropin1 fused to a sequence that induces direct proteasomal degradation, which was derived from the carboxy-terminal degron of murine ornithine decarboxylase. The abundance of target proteins tagged with the psd module can be regulated by blue light if the degradation tag is exposed to the cytoplasm or the nucleus. Results We used the model organism Saccharomyces cerevisiae to generate psd module variants with increased and decreased stabilities in darkness or when exposed to blue light using site-specific and random mutagenesis. The variants were characterized as fusions to fluorescent reporter proteins and showed half-lives between 6 and 75 minutes in cells exposed to blue light and 14 to 187 minutes in darkness. In blue light, ten variants showed accelerated degradation and four variants increased stability compared to the original psd module. Measuring the dark/light ratio of selected constructs in yeast cells showed that two variants were obtained with ratios twice as high as in the wild type psd module. In silico modeling of photoreceptor variant characteristics suggested that for most cases alterations in behavior were induced by changes in the light-response of the LOV2 domain. Conclusions In total, the mutational analysis resulted in psd module variants, which provide tuning of protein stability over a broad range by blue light. Two variants showed characteristics that are profoundly improved compared to the original construct. The modular usage of the LOV2 domain in optogenetic tools allows the usage of the mutants in the context of other applications in synthetic and systems biology as well. Electronic supplementary material The online version of this article (doi:10.1186/s12918-014-0128-9) contains supplementary material, which is available to authorized users.


Supporting information
: Growth behavior of yeast cells in blue light compared to darkness. Figure S2: Mutational analysis of psd module stability. Figure S3: Quantification of psd module variant behavior. Figure S4: The impact of the cysteine within the cODC1 degron on psd module variant stability. Figure S5: Equations used for the in silico analysis of psd module variants. Figure S6: Parameter estimation to the experimental data of the psd module variants. Figure S7: Influence of increased k degLOV and decreased k dark on psd module behavior. Figure S8: Comparison of human parameter assumption with parameters obtained by solving the multiple experiment parameter estimation problem. Figure S9: Growth rate measurements of wild type cells and cells expressing psd module variants. Table S1: List of mutated LOV2 domain residues. Table S2: Conversion rate constants of psd module variants obtained by parameter estimation. Table S3: Starting values for parameter estimation. Table S4: Plasmids used in this study. Supplementary references.

Figure S1
Growth behavior of yeast cells in blue light compared to darkness. A) Growth rate measurements of yeast cells (ESM356-1 carrying plasmid pRS315) in darkness or under blue-light illumination (465 nm 30 µmol m -2 s -1 ) for 8 hours. The optical density at 600 nm (OD 600 ) of the cultures was measured at the indicated time points (error bars: standard deviation, n=4). B) A YAP1 deletion strain is growing slower when illuminated with blue light. Wild type and YAP1 deletion cells were streaked on YPD plates and incubated in darkness or under constant blue-light illumination (465 nm 30 µmol m -2 s -1 ) for two days at room temperature.

Figure S2
Mutational analysis of psd module stability. A) Yeast cells expressing P ADH1 -RFP-psd (plasmid based) or one of the variants (as indicated) were grown in liquid medium in darkness. After taking the first sample (t=0 hours), the translation inhibitor cycloheximide (chx) was added; cells were kept in darkness or were illuminated with blue light (LED lamp, 465 nm, 30 µmol m -2 s -1 ) for the rest of the experiment. Samples were taken at the indicated time points and subjected to alkaline lysis followed by immunoblotting. Antibodies against tRFP and Tub1 or Por1 (loading controls) were used for detection (negative control: C). B) Conditions as in A. Derivatives of plasmid pDS112 (P ADH1 -GFP-3myc-psd) were analyzed. Immunoblotting was performed with antibodies against myc and Tub1. C) Stability of the psd module and its variants exposed to blue light with an intensity of 5 µmol m -2 s -1 . Other conditions as in A.

Figure S3
Quantification of psd-module variant behavior. Curves are the means of protein amounts obtained from at least four independent measurements (error bars: s.e.m.; representative immunoblots are shown in Figure S2 A, B, and C). The half-lives (in minutes±standard error) for each condition are shown next to each curve.

Figure S4
The impact of the cysteine within the cODC1 degron on psd module variant stability. The cysteine of cODC1 is the main determinant of psd module variant degradation in all tested cases (wild type construct, K121M N128Y, K92R E132A E155G, K121M N128Y G138A, and G138A V142A R154G E155S). The variants were expressed in yeast cells (plasmid based) and subjected to cycloheximide chases (as described for Figure S2A) under blue-light illumination (30 µmol m -2 s -1 ). Left side: representative immunoblots; right side: quantification of psd module variant behavior. Four independent measurements were performed for each variant, (error bars: s.e.m.).

Figure S5
Equations used for the in silico analysis of psd module variants. Values for the conversion rate constants k dark , k degENDO , k degLOV , k leak are given in Table 2; other values can be found in the description of the simulations in the methods section. A) Equations used to simulate reaction of psd module variants towards light. B) Equations used for simulation of cycloheximide chase analysis and parameter estimations.

Figure S6
Parameter estimation to the experimental data of the psd module variants. The conversion constants k dark , k leak , k degLOV and k degENDO were adapted to reproduce the experimental data. For the psd module, k dark and k leak were taken from the literature and k degLOV as well as k degENDO were adjusted. Parameter estimation using the 5 µmol m -2 s -1 data required to adjust the k hν value slightly for the psd module.

Figure S7
Influence of increased k degLOV and decreased k dark on psd module behavior. A) Simulation of the wild type psd module using the parameters obtained by parameter estimation. B) Simulation with increased k degLOV , other parameters as in A). C) Simulations with decreased k dark , other parameters as in A). The simulations demonstrate the impact of the parameters. Qualitatively, the graph with increased k degLOV is similar to the results obtained by cycloheximide chases for the variants K92R E132A N148D E155G or K92R E132A N148D E155G, whereas the graph with decreased k dark resembles the variants V19I, K121M N128Y, K121M N128Y G138A, K92R E132A E155G or G138A V142A R154G E155S (see Figure S2).

Figure S8
Mathematical-based parameter estimation has higher accuracy than human parameter assumption. Comparison between parameters derived by TinkerCell cycloheximide chase simulations (blue curve) and parameter estimation (red curve) for the psd module variant K92R E132A E139N N148D E155G.

Figure S9
Growth rate measurements of wild type cells and cells expressing psd module variants. Yeast cells (ESM356-1) carrying either an empty plasmid (pRS315), the psd module variant K121M N128Y G138A (pDS142) or K92R E132A E155G (pDS143) were grown in darkness or under blue-light illumination (465 nm 30 µmol m -2 s -1 ) for 8 hours. The optical density at 600 nm (OD 600 ) of the cultures was measured at the indicated time points (error bars: standard deviation, n=4).   The starting values for parameter estimation were obtained from TinkerCell simulations of cycloheximide chases. For each variant, parameters were chosen that approximated the experimental data. These parameters were then used for the parameter estimation. Values in brackets indicate the variability during parameter estimation, values lacking restriction were freely estimated.