A Bioorthogonal Double Fluorogenic Probe to Visualize Protein–DNA Interaction

.S.); Abstract: Two sets of bioorthogonally applicable, double ﬂuorogenic probes, capable of sensing DNA–protein interactions, were prepared by installing an azide or tetrazine motif onto structurally ﬂuorogenic, DNA sensitive frames. Installation of these bioorthogonal functions onto DNA intercalating dyes furnished these scaffolds with reactivity based ﬂuorogenicity, rendering these probes double-ﬂuorogenic, AND-type logic switches that require the simultaneous occurrence of a bioorthogonal reaction and interaction with DNA to trigger high intensity ﬂuorescence. The probes were evaluated for double ﬂuorogenic behavior in the presence/absence of DNA and a complementary bioorthogonal function. Our studies revealed that azide and tetrazine appending thiazole orange frames show remarkable double ﬂuorogenic features. One of these probes, a membrane permeable tetrazine modiﬁed thiazole orange derivative was further tested in live cell labeling studies. Cells expressing bioorthogonalized DNA-binding proteins showed intensive ﬂuorescence characteristics of the localization of the proteins upon treatment with our double ﬂuorogenic probe. On the contrary, labeling similarly bioorthogonalized cytosolic proteins did not result in the appearance of the ﬂuorescence signal. These studies suggest that such double-ﬂuorogenic probes are indeed capable of sensing DNA–protein interactions in cells.


Introduction
DNA-protein interactions are essential to many of the basic mechanisms of life, such as transcription, replication, gene activation/silencing, DNA packaging, or gene editing. The central role of these processes in cell functioning, maturation, differentiation, or in tumorigenesis calls for methods that allow us to obtain improved insight into these events at the biomolecular level. These needs have placed the in cellulo sensing of specific DNAprotein interactions into the focus of chemical biology research. To date, several methods have emerged to study nucleic acid-protein interactions. Chromatin immunoprecipitation (ChIP) [1] based methods are suitable for capturing snapshots of DNA-protein interactions in live cells with in vitro work-up and detection. Others, i.e., electrophoretic mobility shift assays (EMSA) [2] or pull-down assays [3], provide the means to detect low abundance proteins in lysates. Further instances have applied the intercalating dye, thiazole orange (TO), which was covalently attached to DNA binding proteins (a zinc finger containing glucocorticoid receptor [4] or histone protein [5]) for the fluorescent detection of DNA-protein interactions in vitro. Reporter assays [6] provide an indirect way to track down protein (e.g., transcription factor)-DNA interaction via real-time monitoring of translational activity of a promoter DNA sequence coding for the target promoter protein fused to a reporter gene that encodes a protein with detectable properties (e.g., fluorescent protein). Other approaches employ oligonucleotides tagged with a probe sensitive to protein binding [7,8]. Such tagging can be carried out post-synthetically or by means of non-canonical building blocks modified with an environmentally (e.g., polarity, viscosity) sensitive motif [9]. Non-canonical nucleotides can be incorporated into DNA synthetically (e.g., on solid phase) or by means of metabolic incorporation or primer extension (PEX) [10]. The changes in fluorescence due to the altered microenvironment upon protein binding are then correlated with the DNA-protein interaction. The pioneering work of Hocek and co-workers [11,12] has resulted in several probe-appending nucleosides capable of sensing DNA-protein interactions. In one of their latest reports, they describe aminophenyl-or aminonaphthyl-methoxychromone modified nucleosides that show a moderate increase in fluorescence and a remarkable change in emission color at the same time [13]. In an earlier report, Seio et al. [14] reported on a benzofurane appending nucleoside that allowed a ca. 100-fold increase upon its interaction with a single-strand binding protein. Although the above examples are suitable to detect DNA-protein interactions mostly in vitro, there are no viable methodologies with the ability to visualize the ever-changing relationship between protein and DNA in live cells.
The methods that utilize environment sensitive fluorescent probes described above generally depend on the covalent modification of DNA building blocks. The incorporation of non-canonical nucleotides into DNA is possible through solid phase synthesis or PEX, but these modified DNAs are still considered extrachromosomal when delivered into cells. Chromosomal incorporation of non-canonical nucleotides can occur when the desired base is tolerable to enzymatic machinery; however, the insertion of these is non-specific and general. Neither of these cases allow for specific targeting of protein-DNA interaction at focused places. When there is a specific interaction with a protein of interest (POI), manipulation can be achieved site-specifically at both termini, e.g., with small self-labeling enzyme tags, or virtually at any positions with minimally perturbing non-canonical amino acids (ncAAs) via the extension of the genetic code [15][16][17][18][19][20]. Such genetic incorporation of, e.g., bioorthogonalized ncAAs, enable site-specific fluorescent tagging. Our continuing efforts to develop bioorthogonally applicable, fluorogenic probes highlighted the viability of an azide-or tetrazine-based modulation of fluorescence properties. Several examples proved that both bioorthogonal functions are suitable for the efficient quenching of the fluorescence of probes [21][22][23][24].
The above considerations and challenges prompted us to design and synthesize a set of bioorthogonally functionalized environment sensitive probes that combine the reactivitybased fluorogenicity of an azide or tetrazine bearing scaffold with the structural fluorogenicity of DNA probes. We hypothesized that such double-fluorogenic probes require the simultaneous events of a covalent conjugation to a site specifically bioorthogonalized POI and interaction with DNA to fully restore fluorescence (Scheme 1). Such probes are anticipated to be suitable to indicate DNA-protein interaction in a cellular environment using confocal microscopy imaging.

Materials and Methods
Experimental details, synthetic procedures, spectroscopic characterization, physical data determination, details on the cell imaging experiments, further images, and the viability assessment can be found in the Supplementary Materials.

Synthesis and Fluorogenic Evaluation of Probes
We chose four structurally fluorogenic DNA probe frames [25][26][27][28] to be modified with an azide motif. The selection of the dye-frames was based on their spectral properties, synthetic accessibility, and modifiability ( Figure 1). The plain probes, together with their azide modified congeners, were accessed through convergent synthetic routes (Scheme S1).

Materials and Methods
Experimental details, synthetic procedures, spectroscopic characterization, physical data determination, details on the cell imaging experiments, further images, and the viability assessment can be found in the Supplementary Materials.

Synthesis and Fluorogenic Evaluation of Probes
We chose four structurally fluorogenic DNA probe frames [25][26][27][28] to be modified with an azide motif. The selection of the dye-frames was based on their spectral properties, synthetic accessibility, and modifiability ( Figure 1). The plain probes, together with their azide modified congeners, were accessed through convergent synthetic routes (Scheme S1).
With the probes in hand, we next assessed their fluorogenic features by measuring the fluorescence intensities of the probes in the absence or presence of a complementary bioorthogonal motif (i.e., a cyclooctyne, (1R,8S,9s)-bicyclo [6.1.0] non-4-yn-9-ylmethanol, BCN) (Scheme S4), and DNA in a buffered medium (Tris-PBS 1:4). For comparison, we also measured the fluorescence of non-bioorthogonalized parent frames under similar conditions. In general, most of the non-bioorthogonalized probes exhibited negligible fluorescence in the absence of DNA and an enhanced signal in the presence of nucleic acids. TO*, on the other hand, showed a negligible effect in the presence of nucleic acids, pointing out the importance of the substitution pattern on intercalation or structural constriction.
Installation of an azide function onto the structurally fluorogenic frameworks generally resulted in quenched fluorescence even in the presence of DNA ( Figure 2, Tables 1, S1 and S2). The combined addition of BCN and DNA, on the other hand, resulted in increased emission intensities. Evaluation of the results revealed that TOA possesses remarkable double fluorogenic features. Although synthetic accessibility of organic azides is quite appealing, the copper-free, strain-promoted cycloaddition reactions of azides and cyclooctynes suitable for live cell applications have limited reaction rates, requiring either extended reaction times or higher probe loading [29]. Indeed, when we measured the conversion percentage of the reaction of TOA and BCN, a 90% conversion could be detected after 10 h ( Figure S1). With the probes in hand, we next assessed their fluorogenic featur the fluorescence intensities of the probes in the absence or presence of a bioorthogonal motif (i.e., a cyclooctyne, (1R,8S,9s)-bicyclo [6.1.0] non-4-y BCN) (Scheme S4), and DNA in a buffered medium (Tris-PBS 1:4). For also measured the fluorescence of non-bioorthogonalized parent fram conditions. In general, most of the non-bioorthogonalized probes exhibit orescence in the absence of DNA and an enhanced signal in the presence TO*, on the other hand, showed a negligible effect in the presence of nuc ing out the importance of the substitution pattern on intercalation o striction.
Installation of an azide function onto the structurally fluorogenic fra ally resulted in quenched fluorescence even in the presence of DNA (Figu Table S1, Table S2). The combined addition of BCN and DNA, on the oth in increased emission intensities. Evaluation of the results revealed tha remarkable double fluorogenic features. Although synthetic accessibility is quite appealing, the copper-free, strain-promoted cycloaddition reactio cyclooctynes suitable for live cell applications have limited reaction rates extended reaction times or higher probe loading [29]. Indeed, when we m version percentage of the reaction of TOA and BCN, a 90% conversion c after 10 h ( Figure S1).
In order to avoid extended labeling times with the live cells, we ex    In order to avoid extended labeling times with the live cells, we explored the possibility of replacing the azide with the synthetically less accessible tetrazine as a quencher and bioorthogonal handle. At the same time, we involved further frames with red-shifted excitation/emission spectra. To this end, we have designed TO-derived (TOT) and cyaninestyrene-based scaffolds (StyT1, StyT2), as presented in Figure 3. These tetrazine modified environment sensitive scaffolds were accessed via routine synthetic steps as presented in Schemes S2 and S3. In the same manner as for the azide motifs, we assessed the fluorogenicity of the probes and their parent frames in the presence/absence of BCN, and DNA ( Figure 4, Table 2 and Table S1, Table S2).
Similarly to its azide derivative, TOT displayed significant double fluorogenicity. The Sty scaffold showed excellent fluorogenicity in good accordance with its original report [34]. Tetrazine modification via direct conjugation at the indole-ring (StyT1) or through a flexible linker appending at the quinolinium unit (StyT2), however, resulted in considerably lower fluorogenic performance, compared to the parent Sty. As expected, the qualitative assessment of the reaction rate showed much faster kinetics, as the cycloaddition of TOT with BCN (Scheme S4) went to completion within 2 h ( Figure S2). TOT was then selected for cellular labeling studies. Tetrazines can exert their quenching through various pathways. Through-bond energy transfer (TBET) [30] is believed to be the major mechanism, where the tetrazine is connected to the fluorescent frame via a conjugatable but electronically decoupled (twisted) linker. Förster resonance energy transfer is reported to be the main path of quenching, where the tetrazine and the fluorophore are connected through a flexible linker [31]. Finally, tetrazines that are directly conjugated to a fluorophore are reported to quench fluorescence by opening non-radiative relaxation pathways via internal conversion [32,33].
TOT and Sty1 links the tetrazine via a conjugating vynilene motif, while StyT2 carries the tetrazine through a flexible linker. We assumed that the quenching will be effected via internal conversion in the case of TOT and StyT1, while FRET is expected to be the main cause of quenching in the case of StyT2.
These tetrazine modified environment sensitive scaffolds were accessed via routine synthetic steps as presented in Schemes S2 and S3. In the same manner as for the azide motifs, we assessed the fluorogenicity of the probes and their parent frames in the presence/absence of BCN, and DNA ( Figure 4, Tables 2, S1 and S2). port [34]. Tetrazine modification via direct conjugation at the indole-ring (StyT1) or through a flexible linker appending at the quinolinium unit (StyT2), however, resulted in considerably lower fluorogenic performance, compared to the parent Sty. As expected, the qualitative assessment of the reaction rate showed much faster kinetics, as the cycloaddition of TOT with BCN (Scheme S4) went to completion within 2 h ( Figure S2). TOT was then selected for cellular labeling studies.  Similarly to its azide derivative, TOT displayed significant double fluorogenicity. The Sty scaffold showed excellent fluorogenicity in good accordance with its original report [34]. Tetrazine modification via direct conjugation at the indole-ring (StyT1) or through a flexible linker appending at the quinolinium unit (StyT2), however, resulted in considerably lower fluorogenic performance, compared to the parent Sty. As expected, the qualitative assessment of the reaction rate showed much faster kinetics, as the cycloaddition of TOT with BCN (Scheme S4) went to completion within 2 h ( Figure S2). TOT was then selected for cellular labeling studies.

Cellular Studies
To explore the DNA-protein sensory potential of TOT in the more challenging environment of live cells, we assessed the cytotoxicity of our bifunctional probe TOT on a HEK293T cell culture. The cells were treated with TO and TOT in the concentration range of 1-10 µM. Following a 24-h incubation period we performed an MTT test. TO exhibited a significant cytotoxic effect due to its strong intercalation activity. On the contrary, TOT did not affect cell viability substantially ( Figure S3). This difference likely arises from the decreased intercalation activity of TOT due to the structural modification with the tetrazine unit.
Next, we wanted to demonstrate the ability of TOT to specifically sense protein-DNA interaction in cells. Therefore, we selected LaminA [35,36] and H2B [37,38] nuclear proteins as model POIs for the live cell labeling and imaging experiments. Intermediate filaments of LaminA polymerize into fibers to provide a complex meshwork in the inner surface of the nuclear membrane. Based on its structure and localization, it provides a platform for anchoring proteins and gives mechanical stability to the nucleus; furthermore, it has a pivotal role in higher order genome organization, chromatin regulation, the replication and repair of DNA and even transcription. Histone protein, H2B, plays an active role in the formation of nucleosomes, the basic units of DNA coiling. Their central character in chromatin organization, and transcription regulation is unquestionable. Their posttranslational modification, methylation/demethylation, etc., are involved in epigenetic modulations as well. To install a complementary bioorthogonal function onto DNAbinding proteins, the C-termini of LaminA and H2B were fused with a self-labeling enzyme tag, HaloTag.
Cells expressing either LaminA-HaloTag or H2B-HaloTag proteins were treated with a cyclooctyne bearing the HaloTag substrate, HaloBCN (Scheme 2) [39]. The bioorthogonalized proteins were then treated with TOT. Following the incubation time, the cells were washed and fixed, then subjected to confocal microscopy. The microscopy images showed bright structured fluorescence, characteristics of the LaminA, and H2B localization ( Figure 5). These experiments proved that (i) TOT is membrane permeable and (ii) it is able to sense protein-DNA interaction. In order to unquestionably prove that we sense DNA-protein interaction, non-DNA-binding cytoplasmic proteins were studied as well. To this end, cytosolic intermediate filament Vimentin [40] and an outer membrane protein of mitochondria (TOMM20) [41,42] were also fused to HaloTag at their C-termini (the C-terminus of TOMM20 is located in the cytosolic region, preventing interference with mitochondrial DNA). Similarly to the previous labeling experiments, we treated Vimentin-HaloTag or TOMM20-HaloTag expressing cells with HaloBCN first, then washed them and allowed them to react with TOT. Following washing and fixation, the cells were subjected to confocal microscopy imaging. To our delight, no fluorescence signal could be detected at the location of the proteins, presumably due to the lack of nucleic acid interaction ( Figure 5). and allowed them to react with TOT. Following washing and fixation, the cells were subjected to confocal microscopy imaging. To our delight, no fluorescence signal could be detected at the location of the proteins, presumably due to the lack of nucleic acid interaction ( Figure 5). These results indicate that the bioorthogonal reaction with BCN alone is not sufficient to trigger intense fluorescence and the presence of DNA in close proximity is also necessary. Control labeling of Vimentin-HaloTag or TOMM20-HaloTag with a non-DNA binding, fluorogenic tetrazine-probe (SiR-Tet, Figure S4) [43] showed that both proteins express the self-labeling tag ( Figure 6). These results indicate that the bioorthogonal reaction with BCN alone is not sufficient to trigger intense fluorescence and the presence of DNA in close proximity is also necessary. Control labeling of Vimentin-HaloTag or TOMM20-HaloTag with a non-DNA binding, fluorogenic tetrazine-probe (SiR-Tet, Figure S4) [43] showed that both proteins express the self-labeling tag ( Figure 6).

Conclusions
In this study, we have explored the viability of bimodal chemosensors in order to address the great challenge of cellular sensingof DNA-protein interactions. Toward these aims, we have designed and synthesized two sets of bioorthogonally modified DNA sensitive, structurally fluorogenic frameworks in order to create double-fluorogeni AND-type logic switches. The installation of an azide or tetrazine function to these struc turally fluorogenic frames furnished them with reactivity based fluorogenicity, thus re quiring the simultaneous occurrence of a bioorthogonal reaction and interaction with DNA to trigger high intensity fluorescence. We have elaborated the two sets of such azide or tetrazine based probes for double fluorogenic behavior and found that, in both sets, the thiazole orange frame gives the best performance in terms of fluorescence increase. We then applied the membrane permeable tetrazine appending thiazole orange (TOT) to live cells expressing proteins bearing the complementary bioorthogonal function (BCN). Con focal microscopy studies showed that in the case of the DNA-interacting proteins, LaminA or H2B, an intense fluorescence appeared, in accordance with the localization of the pro teins. This finding is further supported by the lack of detectable fluorescence of the two control cytosolic proteins, Vimentin and TOMM20.
These studies indicate that our double fluorogenic (i.e., reactivity-based-and struc tural-fluorogenicity) probes are indeed suitable for cellular sensing of DNA-protein in teractions. Moreover, unlike other attempts to sense DNA-protein interactions using flu orogenic DNA building blocks, our approach allows protein-specific and site-specific in corporation of the probes. Such protein modifications are possible via fusion tags, as used in this work, or via genetic incorporation of non-canonical amino acids, leading to mini mally perturbed modifications. Although this proof-of-concept study confirms the viabil ity of such probes, their use, e.g., in gene-expression or epigenetic studies, requires furthe improvements, such as screening for more photostable frames, or conformational studie of structures to achieve optimized binding between the POI and DNA.

Conclusions
In this study, we have explored the viability of bimodal chemosensors in order to address the great challenge of cellular sensingof DNA-protein interactions. Towards these aims, we have designed and synthesized two sets of bioorthogonally modified DNA sensitive, structurally fluorogenic frameworks in order to create double-fluorogenic ANDtype logic switches. The installation of an azide or tetrazine function to these structurally fluorogenic frames furnished them with reactivity based fluorogenicity, thus requiring the simultaneous occurrence of a bioorthogonal reaction and interaction with DNA to trigger high intensity fluorescence. We have elaborated the two sets of such azide or tetrazine based probes for double fluorogenic behavior and found that, in both sets, the thiazole orange frame gives the best performance in terms of fluorescence increase. We then applied the membrane permeable tetrazine appending thiazole orange (TOT) to live cells expressing proteins bearing the complementary bioorthogonal function (BCN). Confocal microscopy studies showed that in the case of the DNA-interacting proteins, LaminA or H2B, an intense fluorescence appeared, in accordance with the localization of the proteins. This finding is further supported by the lack of detectable fluorescence of the two control cytosolic proteins, Vimentin and TOMM20.
These studies indicate that our double fluorogenic (i.e., reactivity-based-and structuralfluorogenicity) probes are indeed suitable for cellular sensing of DNA-protein interactions. Moreover, unlike other attempts to sense DNA-protein interactions using fluorogenic DNA building blocks, our approach allows protein-specific and site-specific incorporation of the probes. Such protein modifications are possible via fusion tags, as used in this work, or via genetic incorporation of non-canonical amino acids, leading to minimally perturbed modifications. Although this proof-of-concept study confirms the viability of such probes, their use, e.g., in gene-expression or epigenetic studies, requires further improvements, such as screening for more photostable frames, or conformational studies of structures to achieve optimized binding between the POI and DNA.