Photodegradable by Yellow-Orange Light degFusionRed Optogenetic Module with Autocatalytically Formed Chromophore

Optogenetic systems driven by yellow-orange light are required for the simultaneous regulation of several cellular processes. We have engineered the red fluorescent protein FusionRed into a 26 kDa monomeric optogenetic module, called degFusionRed. Unlike other fluorescent protein-based optogenetic domains, which exhibit light-induced self-inactivation by generating reactive oxygen species, degFusionRed undergoes proteasomal degradation upon illumination with 567 nm light. Similarly to the parent protein, degFusionRed has minimal absorbance at 450 nm and above 650 nm, making it spectrally compatible with blue and near-infrared-light-controlled optogenetic tools. The autocatalytically formed chromophore provides degFusionRed with an additional advantage over most optogenetic tools that require the binding of the exogenous chromophores, the amount of which varies in different cells. The degFusionRed efficiently performed in the engineered light-controlled transcription factor and in the targeted photodegradation of the protein of interest, demonstrating its versatility as the optogenetic module of choice for spectral multiplexed interrogation of various cellular processes.


Introduction
A variety of optogenetics tools (OTs) have been developed for the temporal and spatial light regulation of biological processes in cells and organisms. Most of the biochemical pathways are regulated in several control points, and a combination of multiple OTs regulated by distinct wavelengths of light [1][2][3] is required for effective control of metabolic and signaling pathways, as well as complex physiological processes [4,5]. However, current OTs are limited by spectral ranges that activate their photosensory modules.
Despite the large success of OT applications, there are several important limitations. For example, UV light causes DNA damage, such as cyclobutane-pyrimidine dimers (CPDs) and 6-4 photoproducts (6-4PPs), leading to double-strand breaks and apoptosis induction [24]. Violet and blue light also exhibit notable cytotoxicity [25]. Additionally, various cofactors that serve as chromophores for photosensory domains may be produced in insufficient quantities or be absent in mammalian cells (for example, PCB). Moreover, some photosensors, such as CRY2 and RpBphP1, are large (MW of 80-100 kDa) and possess a multidomain structure. Because of this, some OTs engineered from them cannot be packed into adeno-associated virus (AAV) particles, thus limiting their DNA delivery in vivo.
In this regard, photosensory OT modules based on fluorescent proteins (FPs) of a GFPlike family or proteins, such as Dronpa [14] and pdDronpa [15], have important advantages, including small size (~26 kDa) and autocatalytic formation of the chromophore via the cyclization of their amino acid residues. Although the exact mechanism of photoinduced dissociation of Dronpa and its derivatives is unknown, available data suggest that photoinduced cis-trans chromophore isomerization leads to a notable rearrangement of amino acid residues in the conservative A/C dimerizing interface, which consists of β-sheets 7 and 8 of the protein β-can of the GFP-like proteins [26,27]. Since Dronpa variants are regulated by cyan light, engineering FP-based OTs sensing yellow-red light is desirable.
The regulation of OTs in cells can be carried out in various ways. Most often, OTs can only be turned "ON" and "OFF" while their presence in cells is permanent. The regulation of the level of these OTs may require additional genetic systems for transcriptional OT control or directing excess OT molecules to proteasomes [28]. Another OT mode of action is a chromophore-assisted light inactivation (CALI) [29] approach, consisting of the generation of reactive oxygen species (ROS), which inactivate proteins of interest (POIs) located nearby or fused to OT. Representative examples of these OTs are KillerRed [30], SuperNova [29] and miniSOG [31]. However, the generation of ROS in cells may negatively affect their physiology.
In the present work, we developed and characterized a novel red FP-based optogenetic module, which undergoes proteolytic degradation upon illumination with non-phototoxic 567 nm light. The performance of the engineered OT in mammalian cells was validated via the activation of the luciferase reporter, Western blot analysis and its coupling to a transactivator via the nanobody-antigen interaction. This developed OT enabled spectral multiplexing with probes activatable by blue and far-red light. Our work expands the toolbox of OTs based on the GFP-like protein family, which do not require exogenous chromophores and introduces the first photodegradable red FP-based optogenetic module.

Mutagenesis and Screening of FusionRed Mutants
To develop a small optogenetic module with autocatalytically formed chromophore for spectral multiplexing, we selected FusionRed red FP as a template because of its minimal absorption at~450 nm and above 650 nm, and the wavelengths were used for the regulation of blue and far-red/NIR light-sensing OTs [32]. We targeted via random mutagenesis a short amino acid region consisting of β-strands 7 and 8, which are close to the chromophore and comprise the A/C conservative dimerization interface of GFP-like proteins. A relatively low number of mutants due to the short length of the mutated portion allowed us to analyze them via screening in live mammalian cells.
For this, we developed a transcription regulatory system expressing a Gaussia luciferase reporter naturally having a secretion motif ( Figure 1A). Mutated FusionRed sequences were placed between transcription factor Gal4 and transactivator p65 and ex-pressed under the control of the CMV promoter. The reporter construct activated by the resulting Gal4-FusionRedmutant-p65 fusions had the minimal CMVmin promoter controlled by 5×UAS repeats upstream of the luciferase. The expression of a FusionRed mutant coupled to Gal4 and p65 should have resulted in the activation of luciferase reporter expression and secretion in the cultured medium ( Figure 1B). sequences were placed between transcription factor Gal4 and transactivator p65 and expressed under the control of the CMV promoter. The reporter construct activated by the resulting Gal4-FusionRedmutant-p65 fusions had the minimal CMVmin promoter controlled by 5×UAS repeats upstream of the luciferase. The expression of a FusionRed mutant coupled to Gal4 and p65 should have resulted in the activation of luciferase reporter expression and secretion in the cultured medium ( Figure 1B).
HEK293 cells cotransfected with the individual FusionRed mutants and reporter were cultured in multiwell plates in duplicate and one plate was illuminated with 448 nm while another, with the copies of the mutant clones, was illuminated with 567 nm light, and a difference in luciferase activities for each clone was assessed. About 10 4 mutant clones were screened.  HEK293 cells cotransfected with the individual FusionRed mutants and reporter were cultured in multiwell plates in duplicate and one plate was illuminated with 448 nm while another, with the copies of the mutant clones, was illuminated with 567 nm light, and a difference in luciferase activities for each clone was assessed. About 10 4 mutant clones were screened.
We found that upon 567 nm illumination, some FusionRed mutants promoted decreased luciferase production as compared to 448 nm illumination. Figure 1C summarizes the performance of the clones in which 567 nm light caused the decrease in the luciferase expression. Clone 291 outperformed the others and exhibited~10-fold photoinhibition of the reported production by yellow-orange light as compared to the blue light. The 291 clone was later (see below) named a degradable FusionRed (degFusionRed) and selected for further characterization.

Optimization of Transcription with degFusionRed
We found that the mass ratio between the Gal4-degFusionRed-p65 and luciferase plasmids with the total amount of DNA being constant in cell cotransfection affected the reporter production in both the 448 and 567 nm illuminated HEK293 cells (Supplementary Figure S1A). The plasmid ratio of 4:1, producing the highest luciferase activity difference between the 448 nm and 567 nm illuminated cells 48 h after the cotransfection, was used in further experiments.
When a fixed amount of reporter plasmid was used for transfection and the amount of the added Gal4-degFusionRed-p65-encoding plasmid gradually increased, the 567 nm lightinduced reporter inhibition first grew and then decreased (Supplementary Figure S2A). However, shorter 567 nm light pulses (30 s) caused a higher degree of luciferase reporter inhibition and at lower Gal4-degFusionRed-p65 concentrations than the longer ones (120 s) (Supplementary Figure S2A,B). The bell-like pattern of the reporter inhibition persisted for various illumination regimens, suggesting that higher Gal4-degFusionRed-p65 expression might lead to self-inhibition by aggregation or the molecular crowding effect.
Under the light, luciferase activity becomes detectable in the cultured medium as soon as 6-8 h after cotransfection ( Figure 1D, Supplementary Figure S1B). When HEK293 cells were first pre-incubated for 24 h under 567 nm light and then transferred to 448 nm light, the reporter was already detected 2 h under 448 nm light and then stably increased tõ 14-fold during a further 8 h of experiment ( Figure 1E). This suggests that after ending the exposure to 567 nm light, the production of its new molecules occurred de novo via protein synthesis. Likely, 567 nm light caused a substantial reduction in the Gal4-degFusionRed-p65 transcription factor in cells. Indeed, when the luciferase reporter was cotransfected with a Gal4-p65-encoding plasmid lacking the degFusionRed insert, the luciferase expression was not inhibited by 567 nm light ( Figure 1F), which excluded the possibility of a non-specific effect of light on protein synthesis.

Mode of Action of degFusionRed
We next studied HEK293 cell lysates using Western blot with polyclonal anti-tRFP antibodies that recognize both native and denatured forms of all TagRFP-derived FPs, including FusionRed. We found that 567 nm illumination for 24 h resulted in a notable decrease in degFusionRed in HEK293 cells compared to 448 nm illumination ( Figure 3A). When cells were pre-illuminated with 567 nm light for 24 h and then remained under this light for an additional 8 h or were transferred to 448 nm light with the addition of a proteasomal inhibitor bortezomib, Western blot demonstrated the abolishment of the 567 nm light-induced reduction in degFusionRed amount ( Figure 3B), indicating that 567 nm illumination induced the proteasomal degradation of our optogenetic module. However, a sufficient amount of degFusionRed was still detected after 8 h, suggesting that before being completely degraded, degFusionRed remains in the inactive unfolded state, which should still be recognized according to the antibody manufacturer (see Section 4).
To determine whether light caused the oligomerization of degFusionRed, we further applied it to a semi-native PAGE (without heating the samples). Parental FusionRed, degFusionRed and tdTomato [33] were expressed and purified from TOP10 bacteria. The non-illuminated protein denaturation control contained the proteins denatured with 8% SDS. The tdTomato dimer was used as a dimeric control. We found that both degFusionRed and parental FusionRed proteins exhibited monomeric behavior upon

Mode of Action of degFusionRed
We next studied HEK293 cell lysates using Western blot with polyclonal anti-tRFP antibodies that recognize both native and denatured forms of all TagRFP-derived FPs, including FusionRed. We found that 567 nm illumination for 24 h resulted in a notable decrease in degFusionRed in HEK293 cells compared to 448 nm illumination ( Figure 3A). When cells were pre-illuminated with 567 nm light for 24 h and then remained under this light for an additional 8 h or were transferred to 448 nm light with the addition of a proteasomal inhibitor bortezomib, Western blot demonstrated the abolishment of the 567 nm light-induced reduction in degFusionRed amount ( Figure 3B), indicating that 567 nm illumination induced the proteasomal degradation of our optogenetic module. However, a sufficient amount of degFusionRed was still detected after 8 h, suggesting that before being completely degraded, degFusionRed remains in the inactive unfolded state, which should still be recognized according to the antibody manufacturer (see Section 4).
To determine whether light caused the oligomerization of degFusionRed, we further applied it to a semi-native PAGE (without heating the samples). Parental FusionRed, degFusionRed and tdTomato [33] were expressed and purified from TOP10 bacteria. The non-illuminated protein denaturation control contained the proteins denatured with 8% SDS. The tdTomato dimer was used as a dimeric control. We found that both degFusionRed and parental FusionRed proteins exhibited monomeric behavior upon illumination with either 448 nm or 567 nm light ( Figure 3C), excluding the possibility of light-induced degFusionRed dimerization or oligomerization. difference between DHE photobleaching by degFusionRed, FusionRed or mCherry. This indicated that ROS production by degFusionRed unlikely induced CALI of fused domains, in contrast to what was observed for KillerRed [30], SuperNova [29] and miniSOG [31] ROS molecular generators.
Overall, these data suggest that the mechanism of the degFusionRed action is proteasomal degradation, likely upon its light-induced conformational changes resulting in ubiquitination.  We further studied whether reactive oxygen species (ROS) are generated by degFu-sionRed. For this, we compared degFusionRed with known ROS generator SuperNova FP [29] and commonly used red FPs, such as parental FusionRed [32] and mCherry [34]. All FPs were expressed and purified from TOP10 bacteria and applied to 530 nm light. The illumination of SuperNova induced the photobleaching of anthracene-9,10-dipropionic acid (ADPA), an indicator of singlet oxygen,~4-fold more efficiently than degFusionRed. At the same time, no differences between ADPA photobleaching caused by degFusionRed, FusionRed and mCherry were observed ( Figure 3D). The photobleaching of dihydroethidium (DHE), an indicator of superoxide ion, by SuperNova was~2.5-fold more efficient than by degFusionRed ( Figure 3E). Similarly, there was no difference between DHE photobleaching by degFusionRed, FusionRed or mCherry. This indicated that ROS production by degFusionRed unlikely induced CALI of fused domains, in contrast to what was observed for KillerRed [30], SuperNova [29] and miniSOG [31] ROS molecular generators.
Overall, these data suggest that the mechanism of the degFusionRed action is proteasomal degradation, likely upon its light-induced conformational changes resulting in ubiquitination.

Analysis of degFusionRed Spectral Compatibility
We further studied the possibility of the future spectral multiplexing of degFusionRed. In HEK293 cells, we found that 448 nm, 630 nm, 740 nm and 780 nm light did not cause the inhibition of luciferase production via our optogenetic module ( Figure 4A) as determined via bioluminescence assay. The Gaussia luciferase reporter activity at these wavelengths was not statistically different from that in the darkness. In parallel, we coexpressed in HEK293 a known RpBphP1-QPAS1 optogenetic system, which is controlled with far-red/NIR light, and the same luciferase reporter plasmid. As expected, 740 nm and 780 nm activated the luciferase production by the RpBphP1-QPAS1 system, whereas the darkness, 448 nm, 567 nm and 630 nm did not cause substantial bioluminescence signal ( Figure 4B). This indicated the spectral compatibility of degFusionRed (567 nm) with RpBphP1-QPAS1 (740-780 nm), as well as with blue (448 nm) light-activatable OTs, thus potentially enabling the spectral multiplexing of three distinct OTs in mammalian cells.

Analysis of degFusionRed Spectral Compatibility
We further studied the possibility of the future spectral multiplexing of degFusionRed. In HEK293 cells, we found that 448 nm, 630 nm, 740 nm and 780 nm light did not cause the inhibition of luciferase production via our optogenetic module ( Figure  4A) as determined via bioluminescence assay. The Gaussia luciferase reporter activity at these wavelengths was not statistically different from that in the darkness. In parallel, we coexpressed in HEK293 a known RpBphP1-QPAS1 optogenetic system, which is controlled with far-red/NIR light, and the same luciferase reporter plasmid. As expected, 740 nm and 780 nm activated the luciferase production by the RpBphP1-QPAS1 system, whereas the darkness, 448 nm, 567 nm and 630 nm did not cause substantial bioluminescence signal ( Figure 4B). This indicated the spectral compatibility of degFusionRed (567 nm) with RpBphP1-QPAS1 (740-780 nm), as well as with blue (448 nm) light-activatable OTs, thus potentially enabling the spectral multiplexing of three distinct OTs in mammalian cells.  Figure  1A (bottom).

Nanobody-Coupled degFusionRed for Protein Control
We next studied whether we could photodegrade a POI by targeting it with degFusionRed fused to a nanobody specific for this protein. Nanobodies are single variable domains on heavy chain (VHH) antibodies of camelids. Nanobodies are small (15 kDa) and fully genetically encoded, and can recognize cognate antigens with high affinity and specificity. As a model POI, we used an engineered transcription factor consisting of Gal4-ALFA-tag-VP16 fusion. In this POI, the ALFA-tag peptide [35] was placed between the Gal4 DNA-binding domain and VP16 transactivator. In turn, degFusionRed was fused with an anti-ALFA-tag nanobody (NbALFA), so that upon co-expression of both fusions, degFusionRed would be coupled with the POI via non-covalent antigen-nanobody interaction ( Figure 5A). To detect light-induced changes in the POI level, we cotransfected

Nanobody-Coupled degFusionRed for Protein Control
We next studied whether we could photodegrade a POI by targeting it with degFu-sionRed fused to a nanobody specific for this protein. Nanobodies are single variable domains on heavy chain (VHH) antibodies of camelids. Nanobodies are small (15 kDa) and fully genetically encoded, and can recognize cognate antigens with high affinity and specificity. As a model POI, we used an engineered transcription factor consisting of Gal4-ALFA-tag-VP16 fusion. In this POI, the ALFA-tag peptide [35] was placed between the Gal4 DNA-binding domain and VP16 transactivator. In turn, degFusionRed was fused with an anti-ALFA-tag nanobody (Nb ALFA ), so that upon co-expression of both fusions, degFusionRed would be coupled with the POI via non-covalent antigen-nanobody interaction ( Figure 5A). To detect light-induced changes in the POI level, we cotransfected the third plasmid, encoding the Gaussia luciferase reporter after 5×UAS repeats ( Figure 1A, bottom).
We found that under 448 nm light, the luciferase production was high, whereas 567 nm light caused a substantial decrease in the luciferase activity ( Figure 5B Because our OT can also be used as the light-regulated transcription inhibition system, it was interesting to determine the efficiency of its performance. Indeed, the maximum difference in luciferase activity readouts between the 448 nm and 567 nm illuminated cells was ~90-95-fold and was achieved 48 h after cotransfection ( Figure 5C). Notably, this difference is significantly larger than that reported for the far-red/NIR RpBphP1-QPAS1 pair (42-fold activation [2]), which is also a two-component OT. The observed overperformance of the degFusionRed-based OT can be attributed to its severalfold smaller size, resulting in its higher expression level, and to its efficiently autocatalytically formed chromophore, unlike an exogenous biliverdin chromophore of RpBphP1 to which RpBphP1 has a low binding affinity.  To characterize this OT further, we varied the transfection conditions by optimizing the mass ratio of plasmids encoding Gal4-ALFA-tag-VP16, Nb ALFA -degFusionRed, and luciferase reporter (Supplementary Figure S3). We found that the mass ratio of 10:80:10 of these cotransfected plasmids resulted in the highest decrease in the luciferase reporter production, which indicated the significant downregulation of the POI (Gal4-ALFA-tag-VP16) targeted by Nb ALFA -degFusionRed.
We next studied the time course of the luciferase reporter activity under pulsed 567 nm and 488 nm (control) illumination. The luciferase activity became apparent~12 h after co-transfection under 448 nm light and accumulated almost linearly after 36 h ( Figure 5B), indicating the continuous presence of the Gal4-ALFA-tag-VP16 protein. In contrast, the 567 nm illumination caused a significant decrease in the luciferase activity readout during all studied periods (until 72 h), which is in line with the continuous degradation of this model POI in cells.
Because our OT can also be used as the light-regulated transcription inhibition system, it was interesting to determine the efficiency of its performance. Indeed, the maximum difference in luciferase activity readouts between the 448 nm and 567 nm illuminated cells was~90-95-fold and was achieved 48 h after cotransfection ( Figure 5C). Notably, this difference is significantly larger than that reported for the far-red/NIR RpBphP1-QPAS1 pair (42-fold activation [2]), which is also a two-component OT. The observed overperformance of the degFusionRed-based OT can be attributed to its several-fold smaller size, resulting in its higher expression level, and to its efficiently autocatalytically formed chromophore, unlike an exogenous biliverdin chromophore of RpBphP1 to which RpBphP1 has a low binding affinity.

Discussion
We engineered, for the first time, the photodegradable variant of FusionRed red fluorescent protein, degFusionRed, via random mutagenesis of its 7 and 8 β-strands, which are involved in the conserved A/C dimerization interface of GFP-like proteins. The degFusionRed preserved the major characteristics of parental FusionRed including spectral characteristics, which should allow for its simultaneous use with blue and far-red/NIR light-sensing OTs for the regulation of complex molecular and physiological processes at multiple control points.
Unlike the parental protein, degFusionRed is degraded by yellow-orange light in mammalian cells, likely via the proteasomal pathway, as evidenced via Western blot analysis of samples treated with proteasomal inhibitor bortezomib. In contrast to KillerRed [30], SuperNova [29] and miniSOG [31] ROS photosensitizers, degFusionRed preserves the characteristics of parental FusionRed in terms of low ROS production.
Mutations found in degFusionRed may result in light-induced structural rearrangements of the 7 and 8 β-strands. Due to the high rigidity of proline residues [36], A142P substitution may affect the H-bond existing between the chromophore's phenolate group and neighboring Ser143 [37]. Additionally, a water-mediated H-bond between the chromophore's phenolate of FusionRed and the main chain carbonyl of Glu141 can be affected by both neighboring mutations A142P and E141V. The substitution of P149A in β-strand 7 may confer increased flexibility or destabilization of the β-strand. The hydrophobic R-group of Trp introduced instead of negatively charged Asp159 tends to burrow into the hydrophobic protein core, which can also destabilize β-strand 8.
Structural rearrangements induced by 567 nm light in degFusionRed may be similar to that of photodissociable FP Dronpa. In the latter, light induces chromophore cis-trans isomerization and the disruption of intramolecular bonds between the chromophore's phenolate and β-can, leading to the increased flexibility of the neighboring β-strands 7-10 and subsequent dimer disruption [26,27]. Ser143 and Glu141 of FusionRed correspond to Ser142 and Glu140 of Dronpa and also form H-bonds with the chromophore's phenolate. In degFusionRed, 567 nm light may enhance the flexibility of the chromophore-adjacent β-can portion, leading to an exposition of the inner hydrophobic patches, which can be recognized by ubiquitin ligases [28].
The nanobody-coupled degFusionRed can effectively degrade the transcription factor tagged with a nanobody-specific antigenic peptide, indicating that other intracellular POIs can be specifically eliminated by yellow-orange light [38].
Current approaches for the regulation of POI degradation have both advantages and limitations. The application of small exogenous molecules, such as molecular glues [39], proteolysis targeting chimeras (PROTACs) [40,41] and auxin-inducible degrons (AIDs) [42,43] is an effective chemogenetic technique; however, they have lower spatial and temporal accuracy compared to OTs. Additionally, compared to protein degradation via proteases and OTs, small molecules are not genetically encoded, which does not allow for performing non-invasive experiments. Similarly, the Trim-Away approach for targeted POI degradation [44] is used for the effective elimination of POI but requires the microinjection of these antibodies directly into cells.
The application of proteases alone may lack effective POI regulation. For instance, TEV protease-induced protein inactivation (TIPI) [45] can be used to study cell processes if both the TEV and TIPI degron are combined in a cell or intracellular compartment. Similarly, bacterial degron moieties targeted by ATP-dependent prokaryotic proteases (Clp family, Lon and AAA+) [46] can be used in bacterial cells or mitochondria but not in the cytosol of mammalian cells.
Optogenetic approaches, such as the LOV2-caging of degrons [47,48], cleaving degrons off via the LOV2-caged TEV protease site [49] and degradable CRY2 oligomers [50], provide non-invasive and efficient temporal and spatial POI regulation; however, they are regulated with blue light, which does not allow for multiplexing with other LOV-based OTs.
Compared to the other approaches, degFusionRed enables the optogenetic regulation of POIs via a previously unexplored yellow-orange spectral range, independence from cofactors, such as biliverdin or PCB, and spectral multiplexing with blue and NIR lightcontrolled OTs.
In conclusion, we developed a small versatile photodegradable optogenetic module degFusionRed with autocatalytic chromophore formation and demonstrated the effective photoinhibition of gene transcription and targeted protein photodegradation in mammalian cells, thus spectrally expanding the toolbox of genetically encoded OTs.

Random Mutagenesis
A construct for mammalian expression of Gal4-FusionRed-p65 fusion was created via the insertion of residues 2-148 of the yeast transcription activator Gal4 into the pEGFP-N1 backbone via NheI/BglII sites and subsequent substitution of EGFP-encoding sequence by AgeI/Not sites with the encoding residues 428-551 of human RELA protein (p65 domain). The resulting construct was amplified via polymerase chain reaction (PCR) using high-fidelity DNA polymerase PrimeSTAR Max (Takara Bio, Shiga, Japan), excluding the nucleotide segment corresponding to the region Val132-Asn173 of FusionRed FP, which comprised the 7 and 8 β-strands. The latter 126 bp DNA portion was amplified via errorprone PCR using a GeneMorph II random mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA) and inserted via blunt-end ligation in the amplified vector, containing the rest parts of the coding sequence. Ligated DNA mixture was purified using QIAprep 2.0 Spin Columns (Qiagen, Hilden, Germany) and used for electroporation of TOP10 E. coli cells (Invitrogen, Waltham, MA, USA). Bacteria were plated on LB agar with 100 µg/mL kanamycin. Constructs with the correct insert orientation were selected via PCR analysis.

Screening of Mutants in Mammalian Cells
The day before transfection, cells were trypsinized and transferred into wells of polystyrene 96-well culturing plates (Costar). Ten µL of a water solution containing 30 ng of the mutated Gal4-FusionRed-p65 construct, 0.1 µg of the pG5-Luc reporter (Promega), encoding the secreted Gaussia luciferase, and 0.2 µg pGl3-Basic vector was mixed with 1.5 µL of 1 mg/mL solution of PEI, pH = 7.0. The mixture was incubated for 10 min at room temperature for the efficient formation of DNA-PEI complexes and added directly to cells in plate wells containing 100 µL of the cultured medium. After transfection, cells were illuminated for 24 h with either blue pulsed light (array of 448 nm light-emitting diodes (LEDs), 30 s ON, 5 min OFF) or yellow-orange pulsed light (567 nm LED array, 30 s ON, 5 min OFF) at a light power density of 100 mW/cm 2 .
For measurements of Gaussia luciferase activity, 1 µL of cell culture medium was mixed with 100 µL of 2 µM coelenterazine (Sigma, St. Louis, MO, USA) in wells of 96-well reflective bottom plates for bioluminescent assays (Stellar Scientific, Owings Mills, MD, USA). The quantification of bioluminescence was performed using a Victor X-5 plate reader (Perkin Elmer, Waltham, MA, USA). The contrast of 448 nm/567 nm reporter expression was calculated as luciferase activity in samples under illumination with 448 nm light divided by that at 567 nm light.

Protein Expression and Purification
Nucleotide sequences of FusionRed, degFusionRed, mCherry and tdTomato proteins were inserted via BglII/EcoRI sites to a pBAD/His-B vector (Invitrogen), which was previously modified [52] to minimize a linker between N-terminal hexahistidine tag and the coding sequence (also known as pBAD/His-D). TOP10 E.coli cells transformed with corresponding constructs were cultivated to OD 600 = 0.5, and protein expression was induced by the addition of 0.002% arabinose. After incubation for 3 h at 37 • C upon vigorous shaking, bacteria were pelleted via centrifugation at 5000× g for 15 min at 4 • C (Eppendorf 5810 R Centrifuge) and used directly for the protein purification protocol or frozen at −20 • C.
The SuperNova gene in the pRSET B vector [29] was transformed in BL21(DE3) E. coli strain (Invitrogen). Bacteria were cultured at 37 • C to OD 600 = 0.5, and induction was performed by the addition of 0.05 mM IPTG. After incubation for 3 h at 30 • C upon vigorous shaking, bacteria were pelleted and used for protein purification.
After being resuspended in ice-cold phosphate-buffered saline (PBS), bacterial cells were disrupted using an ultrasound homogenizer (Model 3000 Ultrasonic Homogenizer; BioLogics). The soluble fraction of cell lysates was applied to Ni-NTA agarose (Macherey-Nagel, Dueren, Germany) columns equilibrated with a 10 mM solution of imidazole in PBS. Proteins were eluted by 400 mM imidazole in PBS and dialyzed overnight in 1000× volume of PBS.

Absorbance and Fluorescence Spectroscopy
Protein samples were diluted in PBS to the concentration of 0.1-0.5 mg/mL. Absorption spectra were recorded using the Hitachi U-2000 spectrophotometer. Fluorescence spectra were recorded using the Cary Varian Eclipse Fluorescence Spectrophotometer (Agilent Technologies, Santa Clara, CA, USA).

Western Blot Analysis
The day before transfection, cells were trypsinized and transferred into 9.5 cm 2 cell culture dishes (Costar, Washington, DC, USA). The transfection mix contained 30 µg of the plasmid encoding Gal4-degFusionRed-p65 and 150 µg PEI in 1 mL PBS, which was added to 10 mL of culturing medium in dishes. After transfection, cells were illuminated for 24 h with blue pulsed light (448 nm LED array, 30 s ON, 5 min OFF) and by pulsed light (567 nm LED array, 30 s ON, 5 min OFF) at a light power density of 100 mW/cm 2 .
Then, cells were detached using a cell scraper (Falcon) and disrupted by passing them 10 times through the 27G needle in 0.5 mL of 10 mM HEPES, 10 mM KCl, 1 mM MgCl 2 and 1 mM EDTA, pH = 7.2, with the addition of complete mini protease inhibitors (Roche Diagnostics, Mannheim, Germany). Cell homogenate was centrifuged at 700 g for 5 min to pellet the nuclei. The cytoplasmic extract was centrifuged again at 15,000 g for 5 min to remove organelles, and then combined with 4x Laemli Sample buffer and boiled for 3 min. Proteins from the resulting samples were separated via 12% SDS-PAGE, transferred to a nitrocellulose membrane and analyzed using polyclonal anti-tRFP antibodies which recognize both the native and denatured form of FusionRed (Evrogen, #AB233, Moscow, Russia) in a dilution of 1:2000, and polyclonal goat anti-rabbit IgG HRP conjugate (Thermo Fisher, #32460, Waltham, MA, USA) in a dilution of 1:2000. The enhanced chemiluminescent (ECL) approach [53] was used for the detection of protein bands.

Semi-Native PAGE Gel
Proteins were diluted to the concentration of 2 mg/mL in PBS and illuminated either with constant blue light (448 nm LED array) or yellow-orange light (567 nm LED array) for 0.5 h. Twelve µg of protein samples was diluted in 4× loading buffer (250 mM Tris-HCl pH 6.8, 0.008% bromophenol blue, 4% glycerol) or in the same loading buffer but with the addition of 8% SDS for denaturation control. The samples were then loaded on 4-20% gradient gel (BioRad, Hercules, CA, USA). The protein samples were not heated but the running buffer contained 1% SDS. After 2 h of electrophoresis, the gel was washed and stained with GelCode blue protein stain (BioRad, Hercules, CA, USA).

Photobleaching Assay
For the evaluation of singlet oxygen production via FPs, samples containing 0.5 mg/mL of respective proteins and 10 µM ADPA in PBS were illuminated for 10 min with constant 1 W/cm 2 of 530 nm light of LED array. The production of superoxide was evaluated via the illumination of samples containing 0.5 mg/mL of proteins and 0.1 mM DHE in PBS for 2 min with constant 1 W/cm 2 of 530 nm light. The registration of the fluorescence of ADPA and DHE change in fluorescence intensity (photobleaching) was performed using the Victor X5 plate reader with 350 nm fluorescence excitation and 450 nm emission filters.

Spectral Multiplexing with RpBphP1-QPAS1
For the characterization of spectral multiplexing suitability, HEK293 cells in 24-well plates were cotransfected with pG5-Luc reporter plasmid, and a plasmid encoding Gal4-degFusionRed-p65, or with pQP-T2A [2] plasmid-expressing RpBphP1-VP16 fusion and Gal4-QPAS1 fusion in a mass ratio of 1:3. Then, cells were cultivated for 24 h in the darkness or under the pulsed light (30 s ON, 3 min OFF) of different wavelengths. The light was produced by arrays of 448, 567, 630, 740 and 780 nm LEDs. After illumination, the activity of secreted luciferase in the cell supernatant was measured.

Light Regulation of Gal4-ALFA-Tag-VP16 via Nanobody-degFusionRed
HEK293 cells in 96-well plates were cotransfected with plasmids encoding Gal4-ALFAtag-VP16, nanobody to ALFA-tag [35,54] (Nb ALFA ) fused to degFusionRed and Gaussia reporter plasmid pG5-Luc. Titration from 1:89 to 80:1 was performed to determine the optimal mass ratio of plasmids encoding Gal4-ALFA-tag-VP16 and Nb ALFA -degFusionRed plasmids. Either 1 or 10 parts of the pG5-Luc reporter were used for titration. Cells were cultured for 24 h after cotransfection under either 448 nm or 567 nm light, and then luciferase activity was assessed. For the time course of luciferase production, the mass ratio of 8:1:1 for these plasmids was used. For this, the cotransfected cells were cultured for 72 h under either 448 nm or 567 nm light and luciferase activity was measured at 0, 4, 8, 12, 24, 36, 48 and 72 h after the cotransfection.