A general method for the development of multicolor biosensors with large dynamic ranges

Fluorescent biosensors enable the study of cell physiology with spatiotemporal resolution; yet, most biosensors suffer from relatively low dynamic ranges. Here, we introduce a family of designed Förster resonance energy transfer (FRET) pairs with near-quantitative FRET efficiencies based on the reversible interaction of fluorescent proteins with a fluorescently labeled HaloTag. These FRET pairs enabled the straightforward design of biosensors for calcium, ATP and NAD+ with unprecedented dynamic ranges. The color of each of these biosensors can be readily tuned by changing either the fluorescent protein or the synthetic fluorophore, which enables simultaneous monitoring of free NAD+ in different subcellular compartments following genotoxic stress. Minimal modifications of these biosensors furthermore allow their readout to be switched to fluorescence intensity, fluorescence lifetime or bioluminescence. These FRET pairs thus establish a new concept for the development of highly sensitive and tunable biosensors.


Supplementary Figures
Supplementary Figure 1| . 2f). b-c. Time course measurements of free intracellular calcium fluctuations using the synthetic calcium indicator Cal520. HeLa Kyoto cells were transiently transfected to express ChemoR-CaM. ChemoR-CaM has been chosen to not interfere spectrally with Cal520. Ten representative single cell traces from 3 biological replicates not expressing (b) and expressing (c) ChemoR-CaM were analyzed for calcium oscillations upon treatment with 10 μM histamine at the time point indicated with an arrow. Represented are the fluorescence intensity changes (ΔF/F0) of Cal520. Cells that did not express ChemoR-CaM mostly highlight calcium oscillations while the expression of ChemoR-CaM seem to repress this behavior as for YC 3.6, which is as well a calmodulin-based calcium sensor. This phenomenon was already reported in the literature and seems to occur through calcium buffering due to the sensor over-expression 21 .

Supplementary Notes Supplementary Note 1 -Development of ChemoG biosensors.
Generation of sensor variants. Certain ChemoG interface mutations increase FRET to a larger extent than others. For example, the interface mutation T225R EGFP usually leads to a stronger FRET increase than the interface mutation L271E HT7 . This feature revealed useful to fine-tune the dynamic range of ChemoGbased sensors. For the generation of new sensors (Fig. S21), we recommend to try a palette of ChemoG FRET pairs with different interface mutations (Supplementary Table 11, available on Addgene). The sensing domain can be derived from an existing biosensor as e.g. ChemoG-CaM that was derived from YC 3.6 22 or a new sensing domain, preferentially exhibiting a large conformational change. To create ChemoG sensor variants, the sensing domain should be cloned between the EGFP and HaloTag7 variants (i.e. ChemoG FRET pairs). Using ChemoG-encoding plasmids and DNA encoding the sensing domain of interest, 6 plasmids encoding sensor variants can simply be obtained through PCR and molecular cloning (e.g. by Gibson assembly 23 ). We recommend to use single GGS linkers connecting the ChemoG FRET pairs with the sensing domain but these can also be further engineered in a second step if necessary. The linkers can be created during the design of the primers used for the PCR amplification of the fragments.
We deposited plasmids encoding ChemoG variants for protein production in E. coli. In case the sensor variants should be tested in mammalian cells, the vector backbone should first be exchanged.

Testing sensor variants in vitro or in cells. Two options are available:
-produce the sensor variants in E. coli, purify them and test them in vitro, or -express and test the sensor variants in mammalian cells (require extra sub-cloning, see above).
For the first option, the purified sensor variants should be labeled with an orange/red fluorophore substrate.
We recommend SiR-halo or a rhodamine substrate with similar spectral properties to minimize direct excitation of the synthetic fluorophore. The labeled sensors should then be titrated with different concentrations of an analyte of interest (AOI). The sensor variant with the largest dynamic range can be identified from fluorescence emission spectra. An ideal sensor exhibits low FRET in absence and high FRET in presence of the AOI (or vice versa), showing a large peak inversion in each emission channel.
Some noticeable fluorescence should remain in both channels for precise measurements.
For the second option, mammalian cells should be transfected with plasmids encoding the sensor variants.
The transfected cells should be labeled with cell-permeable fluorophore substrates. As previously, we recommend SiR-halo. Labeled cells can subsequently be treated with reagents known to act on the biological activity of interest (e.g. AOI concentration change). Via fluorescence microscopy or flow cytometry, the fluorescence profile of treated and untreated cells can be compared to identify sensor variants with the largest dynamic range. Sensors presenting noticeable fluorescence signal in both channels in presence and absence of treatment should be chosen in order to ensure precise measurement.
Technical details on how to conduct the different experiments can be found in the method section of the manuscript.

Supplementary Note 2 -Tuning the spectral properties of the optimized ChemoG sensor.
The spectral properties of the optimized ChemoG sensor can be tuned by exchanging the FRET donor EGFP with other fluorescent proteins and/or by using different fluorophore substrates as FRET acceptor (Supplementary Fig. 12). For exchanging the FRET donor, EGFP is substituted with an alternative fluorescent protein (e.g. EBFP2 = ChemoB) with the same interface mutations (e.g. EGFP T225R  EBFP2 T225R ) via molecular cloning. As FRET donor, we recommend EGFP-derived fluorescent proteins such as EBFP2, mCerulean3 or Venus to ensure a good transferability of the interface mutations. We recommend to use FP constructs we deposited on Addgene to ensure that the adequate FP mutations are used. For red fluorescent proteins, we recommend using mRuby2 without additional mutations for biosensor design. The FRET acceptor can be readily chosen by simply labeling the ChemoX sensors with different rhodamine-based HaloTag substrates (e.g. JF525, CPY or JF669). The ChemoX sensors performance can be evaluated as explained in Supplementary Note 1 and in the methods.

Supplementary Note 3 -Tuning the readout mode of the optimized ChemoG sensor.
The readout of ChemoG FRET sensors can be tuned by small modifications (Supplementary Fig. 13).
Single channel fluorescence intensity and fluorescence lifetime-based ChemoD sensors are obtained by substituting EGFP with its non-fluorescent variant ShadowG 18 carrying the same interface mutation(s).
Additionally, the fluorescence quenching mutation P174W should be introduced into HaloTag7. For intensiometric sensors, we recommend labeling with JF635 while for fluorescence lifetime imaging, CPY worked best in our hands so far. The performance of the sensors can be evaluated analogously as explained in Supplementary Note 1 and in the methods.
To convert ChemoG FRET sensors into a bioluminescent ChemoL sensor, a circularly permuted variant of NanoLuc is fused to the N-terminus of EGFP. We recommend labeling the sensor with rhodamine fluorophore substrates whose spectral properties are compatible with the available equipment. In our case, CPY was the most red-shifted fluorophore compatible with our plate reader but we foresee no conceptual hurdle in using any rhodamine fluorophore substrate proven functional for FRET biosensing. The sensors performance can be evaluated analogously as explained in Supplementary Note 1 and in the methods. It should be noted, that the expression of ChemoL sensors in mammalian cells, assessed by direct excitation of EGFP, was found to be substantially lower than the corresponding ChemoG FRET sensors. Since bioluminescent signals can be detected with high sensitivity, i.e. also for sensors with low expression levels, it is possible to acquire robust emission spectra for ChemoL biosensors. Due to the dim EGFP signal, however, it is not advisable to use ChemoL sensors for FRET applications even if this is conceptually possible.