Catalytic Activation of Bioorthogonal Chemistry with Light (CABL) Enables Rapid, Spatiotemporally-controlled Labeling and No-Wash, Subcellular 3D-Patterning in Live Cells using Long Wavelength Light

Abstract

Described is the spatiotemporally controlled labeling and patterning of biomolecules in live cells through the catalytic activation of bioorthogonal chemistry with light, referred to as “CABL”. Here, an unreactive dihydrotetrazine (DHTz) is photocatalytically oxidized in the intracellular environment by ambient O₂ to produce a tetrazine that immediately reacts with a trans-cyclooctene (TCO) dienophile. 6-(2-Pyridyl)-dihydrotetrazine-3-carboxamides were developed as stable, cell permeable DHTz reagents that upon oxidation produce the most reactive tetrazines ever used in live cells with Diels-Alder kinetics exceeding k₂ 10⁶ M⁻¹s⁻¹. CABL photocatalysts are based on fluorescein or silarhodamine dyes with activation at 470 or 660 nm. Strategies for limiting extracellular production of singlet oxygen are described that increase the cytocompatibility of photocatalysis. The HaloTag self-labeling platform was used to introduce DHTz tags to proteins localized in the nucleus, mitochondria, actin or cytoplasm, and high-yielding subcellular activation and labeling with a TCO-fluorophore was demonstrated. CABL is light-dose dependent, and 2-photon excitation promotes CABL at the sub-organelle level to selectively pattern live cells under no-wash conditions. CABL was also applied to spatially resolved live-cell labeling of an endogenous protein target by using TIRF microscopy to selectively activate intracellular monoacylglycerol lipase tagged with DHTz-labeled small molecule covalent inhibitor. Beyond spatiotemporally controlled labeling, CABL also improves the efficiency of ‘ordinary’ tetrazine ligations by rescuing the reactivity of commonly used 3-aryl-6-methyltetrazine reporters that become partially reduced to DHTzs inside cells. The spatiotemporal control and fast rates of photoactivation and labeling of CABL should enable a range of biomolecular labeling applications in living systems.

Graphical Abstract

Introduction

The activation of chemical reactivity by light plays a central role in identifying and studying biological molecules in cellular context.^1–3 Prominent among the various roles of photochemistry in chemical biology are tools for tracking the dynamics of biological molecules in living cells.⁴ For example, fluorescence recovery after photobleaching (FRAP) has long been used to study the movement of mobility of cellular molecules.⁵ FRAP and related techniques operate by photobleaching a subcellular region of interest and studying the migration and exchange of fluorescently labeled molecules into that region. As a complement to FRAP, photoactivatable proteins have emerged as tools for the direct study of subcellular molecules that become fluorescent in response to light.⁶ Here, recombinant proteins can be made fluorescent in real time in response to focused light. Related techniques based on photoconvertible⁷ and photoswitchable⁸ proteins offer additional options for the study of fusion proteins in living systems. As an alternative to techniques based on fluorescent proteins, small molecule probes have been advanced as alternatives to photoactivation in the cellular environment.^9–16 Small molecule photoactivation has traditionally involved photochemical uncaging^9–14 with more recent innovations using the Wolff rearrangement¹⁵ and photoprotonation.¹⁶ An advantage to small molecule fluorophores is that they are generally brighter and more stable than fluorescent proteins.¹⁷ Challenges for current small molecules systems include increasing photoactivation rate and/or extending to longer wavelengths to further augment cytocompatibility.

Bioorthogonal chemistry is a powerful method for selective and high-yielding covalent bond formation in living cells and organisms.^18,19 Bioorthogonal chemistry is highly versatile and can be applied to essentially any biological molecule that can be adapted to incorporate small molecule chemical reporters, and as such has been used to label glycans,²⁰ lipids,²¹ nucleic acids,²² and proteins²³ engineered via enzymatic labeling,²⁴ self-labeling²⁵ or genetic code expansion.²⁶ Chemical probes bearing bioorthogonal tags have been used to probe for targets in endogenous systems,^27–29 and strategies for activity-based protein profiling³⁰ typically rely on enrichment via bioorthogonal chemistry. Because of this remarkable generality, new methodology for bioorthogonal labeling holds the potential for broad applicability in biology and medicine.

Photoactivatible bioorthogonal chemistry is an alternative method for labeling biological molecules in response to light.^31,32 Initial examples included the photolysis of tetrazole and cyclopropenones to produce cycloaddition-reactive nitrile imines and cyclooctyne derivatives, respectively.^33–35 Further examples include photochemically inducible analogs of the Staudinger³⁶ and CuAAC³⁷ reactions as well as cycloadditions involving azirines,³⁸ benzyne³⁹, diarylsydnones,^40,41 quinones,^42–45 o-napthaquinone methides⁴⁶, o-quinodimethanes^47,48 and trans-cycloheptene.⁴⁹

Variations of tetrazine ligation— the bioorthogonal Diels-Alder reaction of tetrazines— have become increasingly important due to their exceptional kinetics with trans-cyclooctenes and other strained dienophiles.^50–53 Tetrazine ligation has become an important tool for tagging bioorthogonal reporters in live cells for fluorescent and super-resolution microscopy.^54–60 Photochemically inducible variants of tetrazine ligation have been described in recent years based on methods for uncaging cyclopropene^61,62 and bicyclononyne⁶³ dienophiles. Inducible versions of tetrazine ligation can also be achieved via the oxidation of dihydrotetrazine (DHTz) precursors⁶⁴ with applications in electrochemically controlled bioconjugation at electrode surfaces,⁶⁵ in batteries,⁶⁶ and for colorimetric nitrous gas detection.⁶⁷ In preprint, an o-nitrophenylphenyl protected dihydrotetrazine has been used with 405 nm light and without catalysis to uncage tetrazines that react with TCO with rates of 10² M⁻¹s⁻¹.⁶⁸

Oxidation-induced bioorthogonal chemistry has emerged as a powerful method for on-demand bioorthogonal chemistry.⁶⁹ Previously, our group described activation of tetrazine ligation using catalytic stimuli,⁷⁰ where either visible light and a photocatalyst or the enzyme horseradish peroxidase was used to catalyze the oxidation of a DHTz to a tetrazine. (Fig 1). Oxygen serves as the terminal oxidant in this process, and several photocatalysts were found to be effective including methylene blue which catalyzes photooxidation upon excitation by 660 nm light. This initial system for photocatalytic oxidation using red light had found several in vitro applications,^70–73 but had limited use in live cell applications due to phototoxicity associated with the sensitization of singlet oxygen by methylene blue. More recently we have shown that silarhodamine (SiR) dyes, initially developed as fluorophores for biological imaging,^74–77 can be repurposed as photocatalysts for DHTz oxidation.⁷⁸ The Janelia-SiR dyes⁷⁷ were found to be especially effective even at low catalyst loadings. With SiR-photocatalysts, DHTz oxidation is more rapid than the competing sensitization of singlet oxygen, and therefore the photocatalytic activation of tetrazine ligation can be applied to protein modification while minimizing oxidative damage. In the presence of live human prostate cancer cells, SiR-red light photocatalysis was used to crosslink polymers to create hydrogels and enable their culture in 3D.⁷⁸ This photoinducible hydrogel formation could also be carried out in vivo in live mice through subcutaneous injection of a solution containing SiR photocatalyst and a hydrogel precursor, followed by brief in vivo irradiation with 660 nm light to produce a stable hydrogel material.⁷⁸ While these recent studies have demonstrated the ability to photocatalytically activate tetrazine ligation in the extracellular environment, the poor cellular permeability of the DHTz analogs used previously prevented us from exploring the potential for intracellular applications, and we had not explored spatiotemporal control of photoactivation at the cellular level.

Fig 1.

Illustration of spatiotemporally controlled subcellular labeling via catalytic activation of bioorthogonal chemistry with light, or “CABL”. (A) Directing a DHTz-functionalized ligand to a subcellular target (illustrated here for actin) does not result in recruitment of a labeled TCO until (B) illumination in the presence of light results in the oxidation of DHTz to tetrazine, enabling rapid labeling via the fastest bioorthogonal reactions observed to date in live cells.

Described herein is the spatiotemporally controlled labeling and patterning of biomolecules inside of live cells through the catalytic activation of bioorthogonal chemistry with light, referred to as “CABL”. The new method is mechanistically different from traditional methods of photoactivation involving a chromophoric change upon irradiation. In CABL, a DHTz tag is directed to a subcellular target, where it is inactive in bioorthogonal conjugation chemistry. In the presence of a photocatalyst and light, the dormant DHTz is converted into a highly reactive tetrazine that can then undergo rapid bioorthogonal chemistry. (Fig 1). Key to the success of CABL in live cells is a new class of DHTz reagents that are cell permeable and stable in the intracellular environment. The fluorescent dyes, fluorescein (FL) and Janelia Si-Rhodamine, thienyl JF₆₄₆ (SiR-tJF₆₄₆)⁷⁷ (Fig 2A), serve as photocatalysts in conjunction with brief irradiation at 470 nm and 660 nm, respectively, to produce the most reactive tetrazines ever used in live cells with Diels-Alder kinetics exceeding k₂ 10⁶ M⁻¹s⁻¹. CABL can be used to photoactivate covalent attachment via bioorthogonal chemistry on a variety of subcellular protein targets to deliver bright, stable small molecule fluorophores to illuminated regions. In conjunction with 2-photon excitation it is possible to selectively pattern subcellular structures in 3D with sub-micron spatial resolution with live imaging under no-wash conditions. Photoactivation by CABL is not limited to overexpressed proteins, and was also applied to spatially resolved live-cell labeling of the protein target monoacylglycerol lipase (MAGL) at low, endogenous cellular concentration.

Fig 2.

(A) Silarhodamine or fluorescein dyes catalyze the oxidation of dihydrotetrazine (DHTz) 1 by O₂ to produce tetrazine 2. (B) Synthesis of amine-reactive DHTz 6 and amide analogs 1a-c. (C) Relative rate of tetrazine-TCO ligation for 2a vs other tetrazine derivatives in aqueous buffer. (D) Kinetics for reaction of 2a with TCO derivatives.

Results and Discussion

Synthesis and Evaluation of DHTz’s with Improved Permeability, Stability, and Reactivity

Essential to creating a system for photocatalysis in live cells was the development of 6-(2-pyridyl)-dihydrotetrazine-3-carboxamides 1 as a new class of compounds that display improved cell permeability and stability in the reduced DHTz state and enhanced reactivity in the oxidized tetrazine state (Fig 2A). The DHTz/Tz pair was designed to contain electron-withdrawing amide and 2-pyridyl substituents that play a dual role by stabilizing the dihydrotetrazine 1 toward background oxidation and augmenting the rate of the tetrazine 2 in inverse electron demand Diels-Alder cycloadditions.

Ethyl 6-pyridyl-DHTz-3-carboxylate 6 was synthesized through the sequence outlined in Fig 2b: hydrazide 3 and ethyl chlorooxoacetate were combined to give the unsymmetrical diacylhydrazine 4, which was reacted with PCl₅ to give dichloride 5. Condensation with a stoichiometric amount of anhydrous hydrazine in ethanol gave 6 on gram scale.^79,80 DHTz ester 6 is reactive enough to directly form amides upon reaction with amines due to the electron withdrawing nature of the DHTz. Butylamine can undergo direct amidation directly with 6 to give amide 1a. For the synthesis of secondary amide 1b and HaloTag-substrate 1c, activation with trimethylaluminum⁸¹ was required to promote the amidation reaction (Figure 2B).

UV-vis and NMR spectroscopies were used to monitor the stability of DHTz derivatives in metal-free PBS-buffer, prepared by simply passing PBS through chelex resin.⁸² DHTz 1a (40 μM) was shown by UV-vis to retain 96% and 94% of the DHTz oxidation state after 24 h and 49 h, respectively (Fig S1). Similarly, DHTz 1b (25 mM) was shown by ¹H NMR to be ≥ 95% stable after 48 h in deuterated PBS (Fig S2). The stability of DHTz 1a was also studied in 10% human serum in PBS buffer; a UV-vis assay showed a 150 μM solution of 1a retained 92%, 81% and 71% DHTz oxidation state after 20 h, 2 d and 3.5 d respectively (Fig S3). For cellular experiments, FBS was replaced with serum-free Opti-MEM minimal media prior to DHTz introduction.

Photooxidation of DHTz 1a generates tetrazine 2a which participates in the fastest bioorthogonal reactions measured to date. Tetrazine 2a was also prepared independently via chemical oxidation of 1a using phenyliodonium diacetate.⁶⁴ Using stopped-flow kinetics with UV-vis monitoring, the rate of reactivity of eq-5-hydroxy-trans-cyclooctene (5-OH-TCO) toward 2a was compared to known tetrazines as shown in Fig 2C. Chosen for comparison were monoaryltetrazine 7⁸³ and dipyridyltetrazine 8,^64,84 which are commonly employed for their exceptional reactivity, and 3-methyl-6-aryltetrazine 9,⁸³ which is commonly employed for its high stability. In PBS, the rate of 2a with 5-OH-TCO was measured to be 173,000 M⁻¹s⁻¹, which is 7-times and 10-times faster than analogous reactions of 5-OH-TCO toward 7 and 8, respectively. In MeOH, tetrazine 2a is more than 300 times more reactive than 9. The rate constants for the reaction of tetrazine 2a toward other trans-cyclooctene dienophiles are displayed in Figure 2D. The high reactivity of 2a is expected due to the activating nature of the pyridyl and amide substituents, which may increase reactivity through a combination of electron withdrawing and ground-state destabilizing effects.⁸⁵ As expected, faster reactivity was observed with soluble a-TCO⁸⁶ and o-TCO⁸⁷ derivatives, which reacted with 2a in PBS with rates of 1.46 × 10⁶ M⁻¹s⁻¹ and 1.13 × 10⁶ M⁻¹s⁻¹. Under aqueous conditions, the reaction with the conformationally strained alkene s-TCO^50,88 was too rapid for us to measure using stopped flow spectroscopy (>10⁷ M⁻¹s⁻¹).

The hydrolytic stability of tetrazine 2a was studied by UV-vis spectroscopy by monitoring the disappearance of the signature tetrazine absorbance at 520 nm. Tetrazine 2a has a hydrolysis half-life of 45 min in PBS at r.t. (Figure S4). In 5 mM glutathione, the tetrazine absorbance at 520 nm decreases with a half-life of 70 seconds, but 65% of the absorbance of 2a is recovered upon photocatalytic oxidation, indicating 1a was formed as a reduction product of glutathione (Fig S7A). In HeLa cell lysate, diluted to 0.5 mg protein/mL, 2a decays with a half-life of 33 min with minor reduction to 1a, presumably due to glutathione in the lysate (Fig S7B). Any concerns about background reactivity are ameliorated by the ‘on demand’ nature of CABL as bioorthogonal reactions of 2a are much more rapid than background reactions. Thus, by activating tetrazine ligation in the presence of TCO dienophiles, CABL should not only provide spatiotemporal control but would also enable bioorthogonal reactions with exceptional rates.

Photocatalytic tetrazine ligation of 1a was initially demonstrated using BCN as the dienophile for in situ Diels-Alder reaction (Fig 3A). While BCN is less reactive than TCO in Diels-Alder chemistry, for small molecule characterization BCN conjugation has the advantage of producing a single aromatic product. As shown in Fig 3A, treatment of 1a with BCN under photocatalytic conditions using FL (2 μM) or SiR-tJF₆₄₆ (2 μM) produced conjugate 10 in 95–98% yield after irradiation at 470 nm and 660 nm, respectively (Figs S8–9). UV-vis spectroscopy was used to monitor the photocatalyzed transformation of 1a (20 μM) to tetrazine 2a in the absence of BCN as a trapping agent.

Fig 3.

(A) DHTz oxidation/Diels-Alder reaction of 1a with BCN is photocatalyzed by either SiR or fluorescein gives conjugate 10. (B) Photocatalytic DHTz oxidation/Diels-Alder reaction of Halo-DHTz in the presence of methionine, a singlet oxygen scavenger, produces conjugate 11.

Fluorescein (FL) is a particularly efficient photocatalyst for DHTz oxidation.⁷⁰ Fluorescein has been used as an in vitro photocatalyst⁸⁹ and in cellular experiments to promote arylboronate ester oxidation via singlet oxygen sensitization.⁹⁰ With 1 μM FL and 470 nm light, the formation of 2a is light dependent and yield of 2a reached a maximum of 79% yield after 40 s (Fig S5), and then decreased upon prolonged irradiation by a degradation process that is also photocatalyzed (Fig S6). Similarly, an 80% yield of 2a was observed after 400 s with photocatalysis by 1 μM SiR-tJF₆₄₆ and 660 nm light (Fig S7).

Photocatalytic tetrazine ligation was also demonstrated with a protein-DHTz conjugate with reaction monitoring by mass spectrometry as shown in Figure 3B. The self-labeling HaloTag protein was labeled with the alkylchloride 1c to produce Halo-DHTz (33,928 Da). As shown in Figure 3B, conjugation with a-TCO-TAMRA took place efficiently upon irradiation at 470 nm in the presence of fluorescein as a photocatalyst to give conjugate 11 (34,578 Da) while exposure to photocatalyst and a-TCO-TAMRA in the absence of light showed no change in mass. Under photocatalytic conditions, both FL and SiR can also sensitize the formation of singlet oxygen at a rate competitive with the photooxidation of DHTz 1. Methionine (5 mM) was added as a singlet oxygen scavenger to limit protein oxidation and mimic the intracellular mechanisms for regulating singlet oxygen. As will be discussed below, strategies for limiting extracellular singlet oxygen production were developed and shown to maximize the efficacy of CABL in experiments in live cells.

Labeling Live Cells with CABL

We next sought to demonstrate that CABL could be carried out in a variety of intracellular targets in live cells. To this end, the HaloTag self-labeling platform was used to introduce DHTz tags to a variety of mammalian subcellular targets as well as to live bacteria. As displayed in Fig 4, HeLa cells were transfected with a green fluorescent protein HaloTag-GFP construct fused to a ‘protein of interest’ (POI) that controls subcellular localization, and then labeled by the small molecule DHTz-HaloTag. Subsequently, cells are treated with a photocatalyst and either a-TCO-TAMRA or o-TCO-TAMRA (Fig 4B)—fluorescent conjugates of TCO derivatives that were chosen for their hydrophilicity and improved permeability relative to traditional TCO derivatives.⁸⁶ Upon irradiation, conjugation is expected only in those cells that express the HaloTag fusion protein, and analysis by SDS-PAGE with fluorescent imaging was used to provide a measure of conjugation efficiency. Unlike silarhodamine and fluorescein derivatives, TAMRA does not act as a photocatalyst for DHTz oxidation, and therefore functions only as a fluorescent reporter.

Fig 4.

(A) Workflow for introducing DHTz labels to Halotag fusion proteins in HeLa cells or E. coli with subsequent fluorescent labeling via CABL. (B) Structure of TAMRA-TCO conjugates. (C) Upon hydrolysis by esterases, fluoresein diacetate becomes fluorescent and an active photocatalyst in live cells. (D, E) In-gel fluorescence was used to monitor the progress of subcellular photocatalysis using (D) Fluorescein/470 nm light/60 mW/cm² and (E) SiR-tJF₆₄₆/660 nm/450 mW/cm² light.

Fluorescein diacetate (FDA) was found to be superior to the direct use of fluorescein (FL). FDA is a cell permeable derivative of the ‘closed-form’ of FL. In live mammalian cells, esterases rapidly hydrolyze FDA to produce FL, which due to limited permeability becomes ‘trapped’ intracellularly, manifesting in intracellular green fluorescence (Fig 4C). Therefore, FDA has the advantage of delivering and concentrating FL within the intracellular environment and thereby increases the efficacy of localized photocatalysis. As shown in Fig 4D, HeLa cells expressing HaloTag-H2B-GFP (nucleus) tagged with DHTz-Halo (10 μM), washed and then treated with o-TCO-TAMRA (2.5 μM) and FDA for 30 min. Cells were then washed to remove excess unbound FDA and irradiated at 470 nm at 60 mW/cm² using an LED light source. Because the nascent FL tends to leak out of the cell after hydrolysis, irradiation was performed within 30 minutes of the washing step. All labeling experiments were carried out for a total of 10 min: after irradiation for the desired amount of time (0–10 min), the cells were kept in the dark for the balance of the 10 min, at which point the TCO reagent was chased by a non-fluorescent tetrazine, and the cells were lysed and analyzed by SDS-PAGE. As shown in Fig 4D, fluorescent labeling increased with irradiation time, with 95% maximum labeling achieved after 5 minutes of illumination. As a qualitative measure of the degree of labeling efficiency, we directly labeled HeLa cells expressing HaloTag-H2B-GFP for 30 min with TAMRA-Halo (structure in Fig 5C): the degree of labeling by in-gel fluorescence was similar to that observed using CABL (Figs 4D and S19). Efficient light-dependent labeling was also observed in similar experiments where DHTz-Halo was targeted to mitochondria or cytosol and labeled with o-TCO-TAMRA under photocatalytic conditions using FDA and 470 nm light (60 mW/cm²). In all cases maximum labeling was achieved within 10 min. CABL was also efficient in E. coli expressing HaloTag that was tagged with DHTz-Halo and labeled with s-TCO-TAMRA using SiR-tJF₆₄₆ as the photocatalyst, with maximum labeling achieved within 10 min upon irradiation at 660 nm (60 mW/cm²) (Fig. S17).

Fig 5.

CABL improves the efficiency of a ‘regular’ bioorthogonal tetrazine ligation. (A) Intracellular labeling of a MeTzHalo-tagged protein proceeds with low labeling efficiency relative to direct labeling by a (C) TAMRA-halo control. (B) The low efficiency of the tetrazine ligation was hypothesized to be a result of inactivation of the tetrazine in the cellular environment. (D) UV-vis spectroscopic monitoring of Tz 9 in PBS containing 5 mM glutathione in the presence of SiR shows that tetrazine absorption slowly decreases in the dark, rapidly recovers upon illumination, and again decreases in the dark. (E) The efficiency of tetrazine ligation with MeTzHalo in live cells is improved 3.6-fold through photocatalysis and is more comparable to that observed when HeLa cells expressing Halo-Tag-H2B-GFP were directly tagged with TAMRA-Halo (Fig 5E, right).

As shown in Fig 4E, SiR-tJF₆₄₆ can be used as photocatalyst in conjunction with 660 nm light for CABL in live cells. Unlike FL, SiR-tJF₆₄₆ is membrane permeable and is not selectively delivered to the intracellular environment. Therefore, the dye can sensitize singlet oxygen in the extracellular environment where reactive oxygen species are not actively regulated by cellular mechanisms. Under the conditions used with FDA/470 nm, photocatalysis with SiR-tJF₆₄₆/660 nm led to cellular lysis, which we hypothesized was a result of oxidative stress due to the sensitization of ¹O₂ in the extracellular environment. While a range of enzymes and reductants including glutathione can mitigate intracellular oxidative stress,⁹¹ the extracellular environment lacks comparable mechanisms for resolvingoxidative damage due to ¹O₂. We found that cellular SiR-photocatalysis was successful when carried out in the presence of 2 mM sodium ascorbate– a non-toxic reductant with low membrane permeability.⁹² When irradiated by 660 nm light, SiR-tJF₆₄₆ is deactivated by ascorbate and is therefore incapable of sensitizing extracellular ¹O2. However, due to the impermeability of ascorbate, intracellular SiR-tJF₆₄₆ remains active and capable of catalyzing DHTz oxidation. As shown in Fig 4E, HeLa cells expressing HaloTag-Mito-GFP (mitochondria) were tagged with DHTz-Halo (10 μM), washed and then treated with o-TCO-TAMRA (2.5 μM), sodium ascorbate (2.0 mM) and SiR (2.5 μM) for 20 min. In time course experiments, cells were then irradiated at 660 nm and 450 mW/cm² using an LED light source. For photocatalytic oxidation by SiR dyes, the rate of photooxidation increases proportionally with increasing light power density from 18 mW/cm² to 450 mW cm². ⁷⁸ All labeling experiments were carried out for a total of 30 min: after irradiation for the desired amount of time (0–30 min), the cells were kept in the dark for the balance of the 10 min, at which point the TCO reagent was chased by a non-fluorescent tetrazine, and the cells were lysed and analyzed by SDS-PAGE. (Fig 4E). Paralleling the observations in Figure 3, CABL is less rapid with SiR/far red catalysis, with maximum labeling achieved after 30 minutes of illumination. CABL using SiR/660 nm light could also be used to activate labeling in the nucleus, actin or cytosol (Fig S20). As discussed above, the inclusion of ascorbate had a dramatic protective effect on cellular viability for SiR-photocatalysis. While high toxicity is observed in the absence of ascorbate, HeLa cells treated with 3.2 μM SiR-tJF₆₄₆ and 660 nm light (450 mW/cm²) for 30 min show no decrease in cell viability (MTT) after 24 hours when sodium ascorbate (2 mM) is included (Fig S26). By contrast, cells treated with FDA/470 nm light were initially viable but showed ~30% decrease in viability after 2 hours (Fig S26). Thus, SiR-tJF₆₄₆ photocatalysis is the preferred method where long-term cell viability is desired, whereas FDA is preferred catalyst when shorter reaction times or 2-photon excitation is desired.

Overall, CABL was successful in the majority of subcellular environments that were tested, but we did observe significant background oxidation/Diels-Alder reaction in the absence of light in experiments with a ceramide-DHTz derivative. The especially high oxidizing environment of the Golgi,⁹³ where ceramide is localized, may be responsible for the premature activation of the DHTz in this experiment. We also observed ~10% background oxidation in the activation of a DHTz-Halotag conjugate in HeLa cells expressing HalomCherry-PDGFR on the extracellular surface (Fig S23A). Here, to localize the catalyst to the extracellular environment, impermeant fluorescein (FL) was used as the catalyst instead of fluorescein diacetate.

CABL improves the efficiency of ‘regular’ tetrazine ligations

As discussed above, the efficiency of CABL in live HeLa cells with DHTz-tagged HaloTag-H2B-GFP is comparable to direct labeling by TAMRA-Halo (Fig 5C). By contrast, tagging HaloTag-H2B-GFP with a conventional tetrazine, MeTzHalo, followed by labeling with o-TCO-TAMRA is considerably less efficient (Fig 5A, 5E.). We reasoned that the low labeling efficiency may be partly a consequence of tetrazine reduction in the cellular environment,⁹⁴ as illustrated in Fig 5B. Tetrazines are reduced to DHTzs by thiols.⁹⁴ In the presence of SiR (4 μM) but in the dark, incubating 25 μM tetrazine 9 with GSH (5 mM) leads to a reduction in absorption at 262 nm, with ~12% of tetrazine 9 consumed after 110 min (Fig 5D). Irradiation at 660 nm light led to rapid and nearly complete recovery of the absorbance at 262 nm. When the light was turned off, reduction of tetrazine 9 resumed. The observation is consistent with a partial reduction of tetrazine 9 to DHTz 13 by GSH, with the recovery of the tetrazine oxidation state upon catalytic photooxidation.

The efficiency of tetrazine ligation with MeTzHalo in live cells was also improved through photocatalysis. As shown in left of Fig 5E, relatively weak fluorescence was observed in HeLa cells expressing HaloTag-H2B-GFP that were sequentially tagged with MeTzHalo, incubated for 2 h, and then labeled with o-TCO-TAMRA. However, when SiR (2.5 μM) and 660 nm light was applied during the conjugation of o-TCO-TAMRA, a 3.6-fold increase in fluorescence was observed (Fig 5E, center). The efficiency of photocatalytic conjugation is more comparable to that observed when HeLa cells expressing HaloTag-H2B-GFP were directly tagged with TAMRA-Halo (Fig 5E, right).

Fluorescence imaging of subcellular targets in live cells labeled by CABL

The ability to photocatalyze the activation of tetrazine ligation was next applied to selective fluorescent labeling and imaging of subcellular targets. Following the same workflow outlined in Fig 4, HeLa cells were transfected with a GFP-HaloTag construct fused to a POI that controls subcellular localization, and then labeled by the small molecule DHTz-Halo. The cells are treated with SiR photocatalyst (2.5 μM) and o-TCO-TAMRA (1.0 μM), irradiated with 660 nm light for 20 min, and then fixed (4% paraformaldehyde), washed, and imaged by confocal microscopy (Fig 6A).

Fig 6.

Fluorescence imaging of subcellular targets in live cells labeled by CABL. (A) DHTz-HaloTag is introduced to HeLa cells were transfected with a HaloTag-POI-GFP fusion where the protein of interest (POI) controls subcellular localization. Successful photoactivation and Diels-Alder reaction with TAMRA-TCO results in the colocalization of green fluorescence from the target and red fluorescence from TAMRA. Cells were labeled while live and fixed prior to imaging, and DAPI was added to stain cell nuclei. (B) Confocal fluorescence microscopy images of photoactivate cells with DHTz targeted to LifeAct (actin), GAP43 (cytoplasm), ActA (mitochondria) and H2B (nucleus). Scale bar = 10 μm

Because DHTz-Halo is tagged to a GFP-fusion protein, a successful experiment is expected to result in colocalization of green (GFP) and red (TAMRA) fluorescence at the subcellular target in transfected cells. As shown in Fig 6B, fluorescent colocalization was observed in CABL experiments with various subcellular targets including LifeAct (actin), GAP43 (cytoplasm), ActA (mitochondria) and H2B (nucleus). TAMRA conjugation was not observed in control experiments where light was omitted, neither by confocal microscopy nor by gel electrophoresis (see irradiation t=0 in Fig 4E and Fig S19–S20). As expected for transient transfection protocol, only a subset of cells express the fluorescent fusion protein. Fluorophore colocalization was only observed in those cells that expressed the HaloTag-GFP protein complex, and not in neighboring cells that did not display GFP fluorescence (Fig S21 A.)

Live Cell, no-wash Sub-organelle Photopatterning

Two-photon excitation can provide focused light to biological samples with spatiotemporal control. Fig 7A schematically illustrates how CABL was used in combination with two-photon excitation and confocal microscopy in order to pattern well-defined 3D structures at the subcellular level. Using CABL, two photon excitation microscopy (880 nm) is used to excite fluorescein and photocatalyze the generation of reactive tetrazines in the subcellular environment, and fluorescent reporters are concentrated in the region of illumination through Diels-Alder chemistry. The rapid kinetics of the tetrazine-TCO conjugation enables stoichiometric labeling^95,96 at very low concentrations (<1 μM) of the fluorescent reporter which is covalently attached in the illuminated area of the live cells under no-wash conditions.

Fig 7.

Live cell, no-wash photopatterning on the cell nucleus. (A) In the presence of a-TCO-SiR, two-photon excitation microscopy (880 nm) is used to activate fluorescein and photocatalyze the generation of reactive tetrazines in nuclei of HaloTag-H2B-RFP transfected cells. The whole nucleus is red-fluorescent, but only the photopatterned regions are labeled by the far-red SiR-dye. Cells are imaged live immediately after 2-photon excitation. (B-D) Illuminating for 3.3 seconds with focused, 2-photon light is used to label (B) square, (C) an ‘X’ and (D) letters in the cell nucleus. A high laser power (details) was used in (C) to demonstrate that labeling is effective even under conditions that photobleach the RFP fluorophore. (E) For the experiment in Fig 7B, the intensity of fluorescence at 633 nm due to the SiR fluorophore was monitored in the illuminated region of the nucleus as well as in a non-illuminated square region of equivalent area found directly below. (F). Fluorescence intensity timecourse for a cell nucleus of a single cell that was periodically pulsed with light from the 2-photon source at low laser power. The fluorescence threshold was set to the background fluorescence of the SiR dye, and detection above the threshold is displayed. After irradiation, the fluorescence intensity is approximately an order of magnitude higher in the irradiated nucleus relative to the nucleus of a neighboring nucleus. Scale bar = 10 μm

As depicted in Figure 7A, HeLa cells were transfected with HaloTag-H2B-RFP, where RFP is the red fluorescent protein mCherry. The nuclei of the transfected cells were then tagged with DHTz-Halo (10 μM), washed and then treated with a-TCO-SiR (100 nM) and FDA (10 μM). For this experiment, the SiR serves as fluorescent reporter and a two-photon laser provided focused light at the optimal wavelength (880 nm) for 2-photon excitation of the fluorescein photocatalyst. Because of the high intensity of the light, only brief illumination was required to promote efficient photoactivation and fluorescent labeling. As shown in Figure 7B, illuminating a square region within the cell nucleus resulted in effective labeling after irradiating for 3.3 seconds. As shown in Fig 7E and Fig S22 the intensity of fluorescence at 633 nm due to the SiR fluorophore was monitored in the illuminated region of the nucleus as well as in a non-illuminated square region of equivalent area found directly below. The fluorescence intensity was normalized to a region outside of the cell. In the irradiated area, the fluorescence intensity increased by order of magnitude over background, whereas the non-illuminated region did not change (Fig 7E). To demonstrate that CABL is light-dose dependent, a separate experiment was carried out where the entire nucleus of a HaloTag-H2B-RFP transfected cell was periodically pulsed with light from the 2-photon source at 10% laser power. A movie (supplementary movie 1) and Fig7F show that fluorescence intensity rapidly increases (t_1/2~7 sec) and then stabilizes after each of three light pulses at which point the maximum fluorescence intensity for the region is reached. During CABL, the signal-to-noise ratio increases by concentrating the fluorescent reporter in the irradiated region through Diels-Alder chemistry. Additionally, the fluorescent signal may also be enhanced by the moderate fluorogenicity of the silarhodamine reporter upon covalent attachment to a protein target.⁹⁷ As depicted in Fig 7F, in the illuminated cell nucleus the fluorescence intensity is an order of magnitude higher than in a neighboring cell nucleus that was not irradiated. The apparent bimolecular rate constant k₂(app) 1 × 10⁶ M⁻¹s⁻¹ measured for this reaction in a live cell nucleus is similar to the in vitro rates observed for a-TCO with Tz 2a (k₂ 1.46 × 10⁶ M⁻¹s⁻¹, Fig 2c). Interestingly, the rate is ~40% slower for the experiment in Figs 7B, 7E, and supplementary movie 2 where only the interior of the cell nucleus was activated k₂(app) 6 × 10⁵ M⁻¹s⁻¹, and may reflect a limitation of diffusion within the nuclear environment. As shown in Fig 7C and 7D, 2-photon irradiation could be used to introduce additional patterns in the live cell nucleus. For the “X” (Fig 7C, Supplementary movie 3), a higher laser power was intentionally used to photobleach the RFP fluorophore, and to show that SiR-fluorescence and RFP photobleaching are colocalized in the experiment. As shown in Fig 7D, CABL can also be used to create letters in the cell nucleus. Supplementary movie 4 of a rotating 3D projection generated from the z-stack of the cell demonstrates that the photopatterning exhibits a high degree of both lateral and axial resolution.

Spatiotemporally resolved labeling of endogenous MAGL

CABL was also used for the spatiotemporally resolved labeling of an endogenous protein target in live cells. We constructed a DHTz probe for the covalent modification of monoacylglycerol lipase (MAGL), a serine hydrolase from the endocannabinoid signaling pathway with broad therapeutic potential (Figure 8A).⁹⁸ DHTz probe 14 was prepared by conjugating 6 (Figure 2) to a 1-oxa-8-azaspiro[4.5]decane scaffold⁹⁹ with an electrophilic hexafluoroisopropyl (HFIP) carbamate group for covalently labeling the active site serine¹⁰⁰ (Fig 8A). Probe 14 inhibited MAGL activity with 13 nM IC50 in an in vitro assay (Fig S24).¹⁰¹ To test the labeling of endogenous MAGL in live cells, human prostate cancer PC3 cells were treated with probe 14 in Opti-Mem buffer for 1 h, washed and then treated with FDA (10 μM) for 30 min. The cells were again washed and then incubated with 100 nM of a-TCO-SiR in Opti-Mem for 5 min in live cells (Fig 8B), and then irradiated with a 470 nm LED for 1 min. 3-(p-(aminomethyl)phenyl-6-methyltetrazine 9 (125 μM) was then added to quench any unreacted a-TCO-SiR. Cells were lysed and in-gel fluorescence was used to assess MAGL-labeling (Fig 8C,D).¹⁰² Fluorescence was observed for two isoforms of MAGL with minimal non-specific labeling from a-TCO-SiR. Less prominent labeling of an additional protein at ~35 kDa was also observed, which is consistent with the reactivity of HFIP-carbamate probes toward toward α/β-hydrolase domain 6 (ABHD6) targets.^100,102 The labeling by probe 14 was dose responsive with a cellular IC₅₀ of 2.5 nM, and was competed by a MAGL inhibitor, KML29.¹⁰⁰ Consistent with the need for photoactivation, labeling was not observed in controls (1) with FDA catalyst but without irradiation, (2) with irradiation but without FDA, or (3) with neither FDA nor irradiation (Fig 8E).

Fig 8.

(A) Activity-based labeling of endogenous MAGL in live cells followed by photocatalytic oxidation (B) Structure of MAGL reactive probe 14 and competitive inhibitor KML-29. (C) Live cells were treated with probe 14 for 1 h, washed and then treated with FDA (10 μM) for 30 min, followed by 2 μM a-TCO-SiR for 30 min, and irradiation for 1 min with 470 nm light. A non-fluorescent tetrazine was added to quench unreacted a-TCO-SiR, and cells were lysed and analyzed by in-gel fluorescence. (C) In-gel fluorescence signals for a dose response of probe 14. (D) Dose response fitting of the fluorescence signals of MAGL normalized by the total protein amount indicated by Coomassie staining. Data are reported as mean ± SEM (n = 2). (E) Probe 14 (3.2 nM, 1 h) was competed by pre-treatment with MAGL inhibitor KML29 (300 nM, 1 h). Additional controls include the exclusion of FDA photocatalyst, light or both light and photocatalyst. See Fig S25 for Coomassie staining. (F-I) Confocal microscopy data for live PC3 cells that were labeled by 14 and incubated with a-TCO-SiR before and after wide field illumination with 470 nm light. The fluorescence threshold was set to the background fluorescence of the SiR dye, and detection above the threshold is displayed. (F) SiR fluorescence + DIC prior to illumination. (G) SiR fluorescence + DIC after illumination. (H) SiR fluorescence after illumination. (I) Fluorescence intensity timecourse monitoring of a region of a cell that was periodically pulsed with 488 nm light (Scale bar=10 μm).

In order to determine the kinetics of the activation of probe 14 on endogenous targets, we labeled PC3 cells, we labeled MAGL using the previously described strategy, treating them with FDA (10 μM), washing, and incubating with 50 nm a-TCO-SiR. Without washing, e fluorescence intensity was monitored via confocal microscopy while periodically irradiating short pulses of light using the 488 nm laser line (Figs 8F–I). We once again found that SiR signal only increased during photoactivation, and that the fluorescence intensity reached saturation after six pulses of light. After CABL, the cellular distribution of the fluorophore is consistent with that observed in with direct labeling with a fluorescent probe for MAGL.¹⁰²

Total internal reflection fluorescence (TIRF)¹⁰³ microscopy was used to study the ability of CABL to activate spatially resolved live-cell MAGL (Fig 9A). In TIRF, incident light is internally reflected in a glass substrate, resulting in the generation labeling of endogenous of an evanescent wave at the glass-liquid interface. As the intensity of the evanescent wave drops exponentially with distance, only chromophores near the coverslip can become efficiently excited.¹⁰³ We reasoned that TIRF would provide method to study the photocatalytic activation and labeling of DHTz-tagged MAGL at the liquid-glass surface. As DHTz-tagged protein beyond the region evanescent illumination would not be oxidized, labeling is only expected in the thin region near the glass surface (Fig 9A). A monolayer of PC3 cells were plated on piranha-cleaned, poly-L-lysine-coated glass, and sequentially treated with probe 14 in Opti-Mem buffer for 1 h, washed and then treated with FDA (10 μM) for 30 min. The cells were washed and then incubated with 100 nM of a-TCO-SiR in Opti-Mem for 5 min in live cells (Fig 7B), and then promptly illuminated with a 488 nm laser at the critical angle for TIRF irradiation. a-TCO-SiR fluorescence in the cells first was detected with TIRF using the 637 nm laser. Fig 9B displays two adjacent 200 μm² areas of cells, of which only one was irradiated. a-TCO-SiR fluorescence was higher in the cells irradiated (Fig 9B). To examine the depth of labeling, a z-stack of images through the sample was acquired with spinning disk confocal microscopy (Fig 9C–E). In the 3D projection image of an illuminated area in Fig 9C, it is evident that cells were prominently labeled only near the glass interface. An axial view orthogonal stack in Fig 9D was plotted in Fig 9E, and shows that fluorescent intensity was highest within 1 μm from the glass surface and dropped off rapidly thereafter. Together, the results illustrate spatiotemporally controlled activation of the DHTz-labeled MAGL protein in response to photocatalysis.

Fig 9.

(A) Total internal reflection fluorescence (TIRF) microscopy was used to activate spatially resolved live-cell labeling of endogenous MAGL proteins that were covalently labeled by probe 14. Only protein-DHTz conjugates in the thin region of evanescent illumination become activated and labeled by the a-TCO-SiR fluorophore. (B) Top view of illuminated and non-illuminated cells visualized by fluorescence microscopy. (C,D) Perspective and orthogonal view of illuminated cells visualized by microscopy. In the orthogonal view, the arrow points to the thin layer that becomes fluorescently labeled near the glass surface. (E) Plot of fluorescent intensity vs distance from surface for the orthogonal projection.

Conclusions

Catalytic Activation of Bioorthogonal Chemistry with Light (CABL) is a spatiotemporally controlled method for the labeling and patterning of biomolecules in live cells. Unreactive dihydrotetrazines are photocatalytically oxidized inside live cells to produce reactive tetrazines that are immediately captured by a trans-cyclooctene (TCO) dienophile with kinetics exceeding k₂ 10⁶ M⁻¹s⁻¹. Fluorescein or silarhodamine dyes are used with activation at 470 or 660 nm, and strategies for limiting the extracellular production of singlet oxygen were developed to increase the cytocompatibility.

Fusions of the HaloTag self-labeling protein were used to localize DHTz-tags and demonstrate high-yielding subcellular activation and labeling of proteins in the nucleus, mitochondria, actin or cytoplasm. 2-Photon excitation microscopy was used to demonstrate that CABL is light-dose dependent and to selectively pattern live cells at the sub-organelle level under no-wash conditions. Spatially resolved live-cell labeling of an endogenous protein target was accomplished by using TIRF microscopy to selectively activate intracellular monoacylglycerol lipase tagged with DHTz-labeled small molecule covalent inhibitor. Beyond spatiotemporally controlled labeling, photocatalysis also improves the efficiency of an ‘ordinary’ tetrazine ligations by rescuing the reactivity of commonly used 3-aryl-6-methyltetrazine reporters that become partially reduced to DHTzs inside cells. We anticipate that the spatiotemporal control and fast rates of photoactivation and labeling of CABL will enable a range of biomolecular labeling applications in living systems.