A turquoise fluorescence lifetime-based biosensor for quantitative imaging of intracellular calcium

The most successful genetically encoded calcium indicators (GECIs) employ an intensity or ratiometric readout. Despite a large calcium-dependent change in fluorescence intensity, the quantification of calcium concentrations with GECIs is problematic, which is further complicated by the sensitivity of all GECIs to changes in the pH in the biological range. Here, we report on a sensing strategy in which a conformational change directly modifies the fluorescence quantum yield and fluorescence lifetime of a circular permutated turquoise fluorescent protein. The fluorescence lifetime is an absolute parameter that enables straightforward quantification, eliminating intensity-related artifacts. An engineering strategy that optimizes lifetime contrast led to a biosensor that shows a 3-fold change in the calcium-dependent quantum yield and a fluorescence lifetime change of 1.3 ns. We dub the biosensor Turquoise Calcium Fluorescence LIfeTime Sensor (Tq-Ca-FLITS). The response of the calcium sensor is insensitive to pH between 6.2–9. As a result, Tq-Ca-FLITS enables robust measurements of intracellular calcium concentrations by fluorescence lifetime imaging. We demonstrate quantitative imaging of calcium concentrations with the turquoise GECI in single endothelial cells and human-derived organoids.


Supplementary note 1. Construction of the dual expression plasmid pFHL.
To allow fast and easy screening of sensor candidates, a dedicated expression vector was designed, termed pFHL, inspired by the plasmids pDuEx (pDress 1 were the mTurquoise2, large spatial linker and P2A sequences were removed from the plasmid using NheI restriction sites) and pTorPE 2 . The protein of interest, in this case a candidate sensor, is under control of a CMV promotor for mammalian expression and a rhamnose promotor for bacterial expression. This will eliminate the need of transferring the candidate sensor to a vector for mammalian expression after bacterial screening. At the N-terminus, the sensor is fused to a TorA tag, a 6xHis-tag and an Xpress-tag. The TorA tag primes transport of the sensor to the periplasm of bacteria 3 . As a result, changing the outer environment, for example adding a compound to an agar plate or to a liquid culture, will directly influence the candidate sensor. Also, easy periplasmic isolation will yield a relatively clean protein, ready for quick testing 2 . The 6xHis-tag and the Xpress-tag can be used for protein isolation.
The optimal concentration of rhamnose for expression in E. coli using the pFHL vector was determined to be 0.4% (w/v) (Supplementary Figure S3A). The performance of the dual expression vector was verified in E. coli and HeLa cells, and compared to the pTorPE plasmid. When expressed in E. coli, the R-GECO1 sensor reacted to changing calcium concentration in liquid growth medium (Supplementary Figure S3B). A bigger response was obtained when the sensor was first isolated by isolation of the periplasmid fluid with an osmotic shock. The contrast was lower using the pFHL vector compared to the TorPE vector. However, we found that the contrast was sufficient for screening. The vector also allowed expression in HeLa cells, where addition of ionomycin and extra calcium resulted in a robust intracellular calcium increase, necessary for lifetime measurements ( Figure  2B).

Supplementary note 2. Influence of residue 150 in mTurquoise2 on fluorescence lifetime.
Amino acid V150 is positioned close to the chromophore in mTurquoise2 and therefore we suspected it to affect the fluorescent lifetime of the protein. We randomly mutated this position. Fluorescent and dark colonies were picked and collected on two plates, from which the modulation lifetime was measured using frequency domain FLIM (Supplementary Figure S5) using a custom build FLIM setup and analysis as described before 4 . Lifetimes between 2.4-4.0 ns were recorded among the fluorescent colonies.
Supplementary note 3. Changing the calcium concentration in the periplasm of bacteria on agar plates. E. coli cells expressing Tq-Ca-FLITS.0 were grown on LB-agar plates. The modulation lifetime (τM) of the sensor was measured before and > 5 min after 1, 2 or 3 stimulations with a droplet of calcium or chelator EDTA (Supplementary Figure S6). The lifetime was recorded of the whole plate using a custom build FLIM setup and analysis as described before 4 . Addition of calcium increased the τM from 2.70±0.03 ns to 2.83±0.02 ns (mean±sd). Adding more calcium to a colony did not further increase the lifetime. Addition of 200 mM EDTA decreased the lifetime to 2.51±0.07 ns, 2.33±0.08 ns and 2.15±0.11 ns for 1, 2 and 3 drops respectively. The results show that the sensors in bacteria on LB-agar are primarily in the high lifetime state and that the sensors can indeed be influenced from the outer environment, as a result of expression in the periplasmic space.

Supplementary note 4. Calcium concentrations during transendothelial migration
Several reports have documented the use of calcium sensitive probes to study changes in calcium levels in endothelial cells upon their interaction with white blood cells. These studies are summarized in Supplementary Table S4.
The increase in calcium concentration that is observed in endothelial cells varies. This can be partially attributed to the different experimental conditions that are used. Here, we have tried to approach the physiological situation as close as possible by (i) studying TEM under flow at 37 °C, (ii) pretreating the endothelial monolayer with TNF to mimic inflamed conditions, (iii) activating the freshly isolated leukocytes by 20 min incubation at 37 °C and (iv) omitting any of the perturbations that are necessary for labeling cells with fluorescent dyes. Specifically, the use of a genetically encoded probe does not require pre-incubation at room temperature, does not need helper reagents (Pluronic) and omits issues with dye leakage and incomplete hydrolysis. Moreover, Tq-Ca-FLITS uses visible light for excitation (in contrast to Indo-1 and Fura-2 which require UV) and enables direct, intensity-independent quantification. Finally, we have fluorescently labeled both cell types and can therefore precisely analyze the interaction between two cell types and distinguish the different phases of TEM.
Supplementary Figure S1. Properties of current intensity-based calcium sensors. Properties of the parent FP are also included: EGFP for all green sensors, FusionRed for K-GECO1, mApple for R-GECO1, mRuby for RCaMP1h and jRCaMP1b, mTurquoise2 for Tq-Ca-FLITS. mScarlet demonstrates the theoretical possibility for improvement of the brightness of the red sensors. Horizontal color bars indicate the relative intrinsic brightness compared to sensors and FPs of the same color (green, red or cyan). Tq-Ca-FLITS published here was added for comparison, which has a notably much higher relative intrinsic brightness in the calcium free state compared to other sensors. EC indicates the extinction coefficient, QY the quantum yield.  0.564 0.108 The respective fluorescent property was measured for each sensor before (pre) and after (post) addition of 14 mM ionomycin combined with 5 mM CaCl2. For lifetime measurements both the phase and modulation lifetimes are indicated. The mean, standard deviation (sd) and coefficient of variation (CV, sd divided by mean) of a number of individually measured cells (N) are given. CV is a number for the variation, independent of the absolute value, and can therefore be used to compare the variation of different types of readouts. Data for MatryoshCaMP6s and YCaM3.60 was acquired with microscope 1 (see Figure 1). Tq-Ca-FLITS published here was added for comparison and shows the lowest CV of all sensors. Source data are provided as a Source Data file.