Synthesis of 2TCOa-TriPPPro and low-background dyes

Since modifications at the C5-position of pyrimidine nucleosides are known to be well accepted by polymerases, we decided to anchor our reactive labeling site to this position.¹⁴ To synthesize a TriPPPro reporter for labeling, 5-iodo-2′-deoxycytidine (4, IdC) was used as the starting material. The reason for choosing a cytidine derivative was two-fold. First, both prokaryotes and eukaryotes, as well as some DNA viruses (e.g., herpes simplex virus 1, HSV-1) and some retroviruses (e.g., HIV-1), encode dUTPases. This enzyme lowers the intracellular bioavailability of dUTP by hydrolyzing the triphosphate to yield dUMP and pyrophosphate.¹⁵ Since most DNA polymerases misincorporate dUTP instead of TTP into DNA, which causes mutations, it can be essential for the viability of these organisms or pathogens to reduce the intracellular dUTP/TTP ratio.¹⁶ Thus, choosing a cytidine-based reporter over a 2′-deoxyuridine derivative might enhance labeling efficiency. Second, the genome of HSV-1, which serves as a model virus, is GC-rich,¹⁷ potentially resulting in a better incorporation of a cytidine-based reporter.

In the first step, IdC 4 was transferred to its 5′-O-phosphate 5 by applying the Sowa & Ouchi protocol with P(O)Cl₃, pyridine, and H₂O in acetonitrile.¹⁸ After neutralization with aqueous NH₄HCO₃ and RP₁₈ column chromatography, the counter ions were exchanged to triethylammonium (TEA) by Et₃N addition and subsequent evaporation of all volatiles. Next, the linker motif was introduced using Sonogashira cross-coupling conditions. 1-Amino-4-pentyne was selected as a building block since it provides some distance between the nucleobase and a terminal amino group for later functionalization. Hence, the C−C-bond formation between the TEA salt of IdCMP 5 and 1-amino-4-pentyne was achieved by a catalyst system formed from Pd₂(dba)₃ and tri(2-furyl)phosphine in the presence of copper(I)-iodide and Et₃N in DMSO. The precipitation of the formed nucleoside phosphate from cold acetone and subsequent RP₁₈ column chromatography of the collected precipitate with 0.05 M tetraethylammonium bicarbonate (TEAB) in acetonitrile provided the TEA salt of 5-(5-aminopent-1-yn-1-yl)-dCMP 6 in a yield of 89 %. The amino group of the linker motif allowed the introduction of the 2TCOa via N-hydroxysuccinimide (NHS) chemistry. The carbamate formation was performed in DMSO in the presence of Et₃N within 30 min at room temperature. Volatiles were removed in an oil pump vacuum at 50 °C. Subsequent RP₁₈ column chromatography with 0.05 M TEAB in acetonitrile provided the TEA salt of 2TCOa-dCMP 7.

The axial 2-ene TCO provides significantly improved stability compared to the commonly used 4-ene TCO, which rapidly isomerizes to the less reactive cis-cyclooctene in the presence of thiols under physiological conditions.¹⁹ While the bicyclic reaction product of 2TCOa and 1,2,4,5-tetrazines undergo a β-elimination reaction, resulting in the undesired dissociation of the fluorophore after successful staining, this reaction can be minimized by choosing appropriate 3,6-substituents on the tetrazine.

For example, the reaction product of 3,6-dimethyl-1,2,4,5-tetrazine (Me₂Tz) led to a much faster elimination reaction than 3,6-di(pyridin-2-yl)-1,2,4,5-tetrazine (Py2Tz), which was, as well as the 3-(pyridin-3-yl)-6-(pyrimidin-2-yl)-1,2,4,5-tetrazine (YPyTz) motif, used in this work.²⁰ Hence, a compromise was accepted between the stability of the TCO motif and the stability of the bioorthogonal reaction product.

The TriPPPro-compound γ-(bis-C9AB)-2TCOa-dCTP 1 was synthesized via the H-phosphonate route previously described.²¹ 4-(Hydroxymethyl)phenyldecanoate 8 was synthesized from 4-hydroxybenzyl alcohol and decanoyl chloride. Bis(4-decanoyloxybenzyl)-H-phosphonate 9 was prepared by displacing the phenoxy groups of diphenyl H-phosphonate (DPP) by 4-(hydroxymethyl)phenyldecanoate 8 in pyridine. Bis(4-decanoyloxybenzyl)-H-phosphonate 10 was obtained in 54 % yield after recrystallization from i-PrOH. This compound was oxidatively chlorinated with N-chlorosuccinimide (NCS) and treated with an excess of tetrabutylammonium phosphate in CH₃CN. The bis(4-decanoyloxybenzyl) pyrophosphate 7 was isolated by extraction with CH₂Cl₂ from ammonium acetate and centrifugation. The TriPPPro-synthesis was initiated by a stepwise activation of the pyrophosphate with trifluoroacetic acid anhydride (TFAA) and 1-methylimidazole in acetonitrile. After the activation, the TEA salt 2TCOa-dCMP 7 was added, dissolved in DMF, RP₁₈ column chromatography, Dowex ion-exchange chromatography (NH₄⁺ form), and a second, RP₁₈ column chromatography gave the pronucleotide reporter γ-(bis-C9AB)-2TCOa-dCTP 1 in a yield of 21 % (Scheme 1). In the following, the compound will be referred to as 2TCOa-TriPPPro 1. Moreover, the parent nucleoside 2TCOa-dC S2 was synthesized from IdC 4 over two steps as a control compound to validate the TriPPPro-approach. Furthermore, various novel and literature-known membrane-permeable fluorophore-tetrazine conjugates were synthesized (Figure 2). To lower the background fluorescence caused by non-reacted dye, a methyl-substituted SiR was attached to a methyl tetrazine (MeTz) by a trans-ethylene linkage resulting in SiR-(E)-MeTz 11. The connected π-system allows the efficient quenching of the fluorophore by the tetrazine via a through-bond energy transfer mechanism (TBET). Upon the reaction of the tetrazine with a TCO, the dye's fluorescence can “turn on”.²² To increase reactivity, a benzyl tetrazine (BzTz) was coupled to silicon rhodamine (SiR) by NHS chemistry via an amide bond (SiR-BzTz 12), as well as an azetidine-substituted SiR was coupled to the Py₂Tz motif shown above via a short diamide linker resulting in SiR-Py₂Tz 13. We also coupled the polar and membrane-impermeable AlexaFluor488 dye to the Py₂Tz motif (AF488-Py₂Tz 14) to generate a cell-impermeable, hydrophilic derivative which we used for labeling of fixed and permeabilized samples. Furthermore, we synthesized a 7-azetidinylcoumarin that we coupled to a pyrimidine(Y)-pyridine(Py) substituted tetrazine (AzC-YPyTz 2) according to the literature.²³ Finally, we improved the fluorescence properties of the coumarin by introducing fluorine atoms to the 7-azetidinyl moiety resulting in the novel derivative F₂AzC-YPyTz 3. Detailed synthesis protocols for these novel fluorescent probes are described in the Supporting Information.

Cellular and viral DNA can be marked with the 2TCOa-TriPPPro in living cells and labeled in fixed cells

First, we tested if the TriPPPro-technique enabled the incorporation of 2TCOa-modified 2′-deoxycytidine reporters into newly synthesized cellular and viral DNA in living cells. To this end, we decided to uncouple the development of cell-permeable pronucleotide reporters from the development of compatible cell-permeable dyes and use non-permeable dyes first that can be used after cell fixation and permeabilization. We chose Vero cells as a cellular model and HSV-1 as a viral model. HSV-1 is an important and well-characterized human pathogen and can be grown in Vero cells. To test 2TCOa-dCTP incorporation into the genomes of living cells as well as viral genomes, Vero cells were either left uninfected (mock) or infected with an HSV-1 mutant which codes for a fusion of the red fluorescent protein mCherry to its small capsid protein VP26 (mCherry-VP26²⁴).

We used a multiplicity of infection (MOI) of 3 plaque-forming units (PFU) per cell in 10 % FCS-containing DMEM. After 1 h, the medium was exchanged to VP-serum-free medium (VP-SFM). This media does not contain serum and thus no carboxyesterases or lipases that can cleave the TriPPPro's AB-masks prior to internalization.

At 4 hours post-infection (hpi), the cells were incubated with 5 μM of 2TCOa-TriPPPro 1 for 3–4 hours. After incorporation, the samples were washed briefly, fixed, and permeabilized. The incorporated 2TCOa motif was conjugated with the polar AF488-Py₂Tz dye 14. After the “click”-labeling, the samples were washed again, counterstained with Hoechst 33342 to label DNA, and imaged using confocal spinning disk microscopy.

As shown in Figure 3A, we found specific AF488 signal in both infected and uninfected TriPPPro-treated Vero cells. In uninfected cells, the AF488 signal colocalized with the Hoechst signal, as shown by a line plot across a representative nucleus in Figure 3C. This indicates that 2TCOa-dCTP was incorporated into the cellular genomes of dividing cells. Moreover, we compared our labeling approach with the well-established CuAAC approach utilizing a commercially available EdC click-kit with an AF488-azide conjugate and observed identical labeling patterns (Figure S4). To provide additional evidence for successful reporter incorporation into genomic DNA, we synthesized the corresponding nucleoside triphosphate 2TCOa-dCTP S5 and performed primer extension assays (Figure S5). We observed successful primer elongation for human pol β and γ, whereas pol α showed no elongation. Additionally, we tested HIV-1 RT polymerase, which also accepted 2TCOa-dCTP S5 as a substrate suggesting that the released 2TCOa-modified NTP shows good substrate properties towards a wide variety of DNA polymerases. Moreover, a cell viability assay also confirmed that up to 10 μM of 2TCOa-TriPPPro 1 did not significantly affect cell growth when incubated for 3 hours (Figure S6).

In infected cells, HSV-1 creates replication compartments (RCs) that increase the nuclear volume and displace the host chromatin to the nuclear periphery. RCs can be identified by areas without Hoechst signal as alphaherpesvirus RCs are not labeled by Hoechst²⁵). The rearrangement of the nucleus facilitates the synthesis of viral genomic DNA, transcription of viral mRNA, and assembly of new capsids.²⁶ The mCherry-VP26 fusion accumulates in nuclear RCs and can be used as a marker for cells in the late stage of virus replication. We found specific AF488 signal colocalizing with viral RCs labeled by mCherry-VP26 (Figure 3B/D), indicating that our new reporter was incorporated into viral genomes. To ensure that the viral DNA was indeed localized to viral RCs, we also stained for the infected cell protein 8 (ICP8) by immunofluorescence. ICP8 is the single-stranded DNA binding protein of HSV-1 and is routinely used as a viral RC marker.²⁷ We found strong colocalization between the signals of the metabolic label and ICP8, further supporting our conclusion that viral genomes can be labeled by our TriPPPro reporter 1 (Figure S7).

To confirm that the delivery of the triphosphate reporter into living cells is needed, we tested the parent nucleoside, which cellular kinases must still phosphorylate to form the active triphosphate. We treated uninfected and HSV-1-infected Vero cells with up to 30 μM of 2TCOa-dC S2, as described in Figure 4. As shown, we could not detect any DNA signal in the AF488 channel, stressing the importance of the TriPPPro-approach to deliver nucleoside triphosphates with bulky modification into cells. Off note, cells incubated with the lipophilic TriPPPro-nucleotides showed increased cytoplasmatic fluorescence, which might be in part due to the unspecific attachment of the lipophilic pronucleotides onto intracellular membranes as well as the incorporation of the reporter into mitochondrial DNA. Based on our experience, limited optimization of the labeling protocol is needed to fine-tune the labeling protocol to the experimental system.

Cellular and viral DNA can be visualized with 2TCOa TriPPPro-compounds in living cells using highly reactive, membrane-permeable dyes

Next, we wanted to establish highly reactive, permeable dyes which show minimal fluorescence in the unconjugated state to label and image DNA in living cells. Based on these premises, we initially developed three SiR-conjugates carrying tetrazines with different reactivities (compounds 11–13). To test them, we used uninfected and HSV-1 mCherry-VP26 infected Vero cells and incubated them with 5 μM of 2TCOa-TriPPPro 1 at 4 hpi in VP-SFM. After 6 hpi, the TriPPPro-compound-containing medium was replenished to ensure the integrity of the AB masks throughout the incorporation phase. At 8 hpi, cells were washed with VP-SFM, stained with 5 μM of the respective SiR-tetrazine conjugate, and directly imaged without further washing. As shown in Figure S8, the most reactive conjugate, SiR-Py₂Tz 13, stained both cellular and viral DNA within one hour after dye addition. In contrast, we could not detect any signal for the less reactive SiR-(E)-MeTz 11 or SiR-BzTz 12 conjugates even after prolonged incubation. Unfortunately, all SiR dyes showed unspecific staining of cytoplasmic compartments, which was even more prominent in uninfected cells (Figure S8).

We concluded that fast tetrazine probes are the key to the successful visualization of DNA in living cells and searched for potential dye-tetrazine conjugates with increased reactivity and improved background. Galeta et al. recently described a HELIOS probe consisting of a 7-azetidinylcoumarin dye directly coupled to a highly reactive tetrazine motif (YPyTz), which facilitated the staining of their cytoplasmic targets within tens of seconds.²³

The direct conjugation of the coumarin dye to the aromatic tetrazine π-system allows effective fluorescence quenching by TBET, resulting in a “turn-on” of fluorescence upon reaction with an axial 4-ene TCO.²³ In addition to the AzC-YPyTz 2, we synthesized its derivative F₂AzC−YPyTz 3, which contained a 3,3-difluoroazetidine moiety, as done previously to optimize xanthene dyes.²⁸ We determined both coumarin-tetrazine conjugates’ “turn-on” values upon reaction with the 2TCOa-dC reporter nucleoside S2. Here, the AzC−YPyTz 2 and F₂AzC−YPyTz 3 probes showed similar “turn-on” values of 150 and 175, respectively (see Figure 1 and S1).

The in vitro reaction rates of the YPyTz-based conjugates with 2TCOa-dC S2 were not improving compared to the less reactive Py₂Tz-functionalized silicon-rhodamine (although the reactivity trend is true for BCN^endo alcohol S21, Table 1 and Figure S2–S3).

However, both coumarin-based probes resulted in clear staining of cellular and viral DNA within minutes post-dye addition compared to within one hour for SiR−Py₂Tz 13, underscoring their utility in the more complex cellular context. The coumarin-YPyTz dyes were efficiently excitable at their second excitation shoulder (Figure S9) with 405 nm lasers, commonly found on contemporary confocal microscopes. The fluorinated coumarin derivative 3 was dimmer than the parent fluorogenic probe 2, but the F₂AzC−YPyTz 3 appeared to show less cytoplasmic background (Figure 5), and we proceeded with this derivate.

To optimize staining conditions, we quantified the fluorescent signal after treatment with increasing concentrations of 2TCOa-TriPPPro 1 (0, 0.5, 1.5, 5, and 10 μM) and labeling with F₂AzC−YPyTz 3 in living cells. As shown in Figures S10, S11, and Figure S12 for cellular and viral DNA, respectively, we found the best staining results at a concentration of 5 μM. The highest TriPPPro dose (10 μM) caused cell dissociation from the growth support, although cell viability was not significantly affected on this time scale (compare Figure S6).

To ensure that labeling was due to the incorporation of the reporter into cellular and viral DNA, we used aphidicolin²⁹ and phosphonoacetic acid (PAA)³⁰ as inhibitors of cellular and viral replication, respectively (Figure 6). We did not detect any DNA labeling when cellular DNA synthesis was blocked, demonstrating that the F₂AzC signals are specific and depend on cellular polymerase activity to incorporate 2TCOa-dCTP. The same was true for viral DNA labeling: PAA, an inhibitor of viral replication, abrogated 2TCOa-dCTP incorporation into viral genomes and RCs, as we did not detect any specific F₂AzC signals. Finally, we also tested DNA labeling in MDCK cells, and the results suggest that the presented approach can be generalized (Figure S13).

Future work will be targeted expand the concept to label RNA, to improve bioorthogonal partner pairing for faster reaction rates and to optimize fluorophores to minimize cytotoxicity and expand the “photon budget” available for live-cell imaging.