Superfast Tetrazole–BCN Cycloaddition Reaction for Bioorthogonal Protein Labeling on Live Cells

Abstract

Here we report the design of a superfast bioorthogonal ligation reactant pair comprising a sterically shielded, sulfonated tetrazole and bicyclo[6.1.0]non-4-yn-9-ylmethanol (BCN). The design involves placing a pair of water-soluble N-sulfonylpyrrole substituents at the C-phenyl ring of diphenyltetrazoles to favor the photoinduced cycloaddition reaction over the competing nucleophilic additions. First-principle computations provide vital insights into the origin of the tetrazole-BCN cycloaddition’s superior kinetics compared to the tetrazole–spirohexene cycloaddition. The tetrazole–BCN cycloaddition also enabled rapid bioorthogonal labeling of glucagon receptors on live cells in as little as 15 seconds.

Graphical Abstract

Bioorthogonal ligation reactions offer a powerful tool to visualize and perturb biomolecules including lipids, glycans, proteins, and nucleic acids in their native cellular environment.^1–3 In a typical experiment, a pair of reactants are used; one serves as a chemical reporter for specific incorporation into a biomolecule, and the other acts as the reaction partner to modify the tagged biomolecule to probe its function. Given the immense complexity of biological systems, the success of this reaction-based approach depends critically on the reaction rate and selectivity and the physicochemical properties of the reactant pairs, including aqueous solubility and lipophilicity, and bioavailability.^4–5 Like other chemical systems, there are no bioorthogonal reactions that possess all desired properties, as improving one property often comes at the expense of the other. Indeed, selecting appropriate bioorthogonal chemical probes for studying a specific biological problem remains an enticing yet elusive endeavor.

In optimizing bioorthogonal chemical reporters, ring strain for substrate activation represents one of the most fruitful approaches.^6–7 A large number of strained alkenes and alkynes have been developed, including norbornene,⁸ trans-cyclooctene^9–10 and its water-solubile variant,¹¹ cyclopropene,^12–14 spiroalkenes,^15–17 and cyclooctynes.^18–23 Among them, bicyclo[6.1.0]non-4-yn-9-ylmethanol (BCN) stood out due to its remarkable stability, fast kinetics, commercial availability, and facile incorporation into small and large molecules.^24–31 For protein studies, it was noted that the BCN-derived lysine could be readily removed from intracellular compartments by gently rinsing cells with a culture medium. As a result, it offers substantially reduced background during bioorthogonal labeling.³² Because of BCN’s exceptional stability and accessibility, several BCN-mediated bioorthogonal reactions have been reported, including strain-promoted alkyne–azide cycloaddition (SPAAC),^{21, 33} tetrazine–BCN ligation,^{30, 34} strain-promoted alkyne–nitrone cycloaddition (SPANC),^{27, 35} and sydnone–BCN cycloaddition^{24, 29} (Figure 1). In particular, BCN–tetrazine ligation received intense interests because of its superior kinetics and excellent biocompatibility. Indeed, tetrazine–BCN ligation has been successfully employed to introduce anti-fouling coating at the surface of supramolecular materials,³⁶ generate octa-arginine cell-penetrating peptides from constitutive tetra-arginine halves,³⁷ prepare hydrogels in situ from clickable polymers for cell encapsulation,³⁸ and activate the photocages for conditional release of a fluorogenic probe in the subcellular environment.³⁹ However, the rapid-reacting tetrazines tend to be less stable in cells,^40–41 impeding its broader adoption in biological studies.

Figure 1.

Select examples of BCN-mediated bioorthogonal ligation reactions and the corresponding rate constants.

Over the last decade, we have developed a photoinduced tetrazole-alkene cycloaddition-based “photoclick” chemistry for imaging proteins in cellular systems.^42–44 Compared with tetrazine ligation, photoclick chemistry offers a spatiotemporal control desirable in many biological investigations. However, alkynes are rarely used in photoclick chemistry because they exhibit poor reactivity owing to their high LUMO energies. Recently, Yu and coworkers reported that BCN could serve as a dipolarophile in the photoclick reaction with diaryltetrazoles containing a pair of ortho-CF₃ substituents on the C-aryl ring.⁴⁵ While the reaction proceeded selectively, the CF₃ substituents greatly reduce the aqueous solubility and restrict its potential utility in biological systems.

Herein, we report the design of a new class of water-soluble, sterically shielded tetrazoles that react with BCN with second-order rate constants up to ~ 39,200 M⁻¹ s⁻¹ in mixed acetonitrile–phosphate buffer, providing one of the fastest bioorthogonal ligation reactions involving BCN. Our strategy involves placing a pair of N-sulfonylpyrrole at ortho-positions of the C-phenyl ring of diphenyltetrazoles to favor the cycloaddition over the competing nucleophilic additions. Density functional theory (DFT) studies indicate that the strong reactivity stems from the lower activation barrier for the nitrile imine (NI)–BCN pair compared to the NI–Sph pair. Using an optimized sulfonated tetrazole, we showed rapid labeling of class B GPCRs^46–47 carrying the BCN moiety within seconds on live cells.

In designing water-soluble tetrazoles for selective ligation with BCN, we chose N-sulfonylpyrrole as the steric shielding group as it displays improved water solubility than N-Boc-pyrrole⁴⁴ and the CF₃ group based on both calculations⁴⁸ and HPLC results (Figures S1–S2, Supporting Information). Accordingly, the sterically shielded tetrazoles 1–4 were prepared through N-sulfonation of tetrazole S3 using suitable sulfonyl chlorides under the optimized reaction conditions (Table S1). The X-ray crystal structure of tetrazole 1 confirmed that the diphenyltetrazole core retains the coplanar structure (Scheme 1; Table S2). Separately, tetrazoles 5–6 were synthesized in two steps: (i) treatment of dipyrrole-tetrazole S6 with methyl triflate gave the activated imidazolium intermediate; (ii) nucleophilic substitution by morpholine or N-Boc-piperazine afforded tetrazole 5 or 6, respectively, in good yields (Scheme 1).

Scheme 1.

Synthesis of the sterically shielded, sulfonated tetrazoles 1-6. The X-ray crystal structure of tetrazole 1 is shown.

With tetrazoles 1–6 in hand, we examined the selectivity for the cycloaddition over the competing nucleophilic additions in 1:1 (v/v) acetonitrile–phosphate buffer mixture and observed excellent selectivity for all tetrazoles (Table S3). Furthermore, the photo-generated nitrile imine from tetrazole 1 displayed a very long half-life (t_1/2 = 138 sec; Figure S3). To our surprise, in addition to acrylamide, dimethyl furamate, and spirohexene (Sph), BCN also served as an outstanding reaction partner for tetrazole 1 (Figure S4). We then compared the reaction kinetics of the BCN-mediated photoclick reaction to that of Sph. The second-order rate constant for the reaction between tetrazole 1 and Sph was determined to be 2,321 ± 76 M⁻¹ s⁻¹ using a pseudo first-order kinetic approach.⁴⁹ The alkyl groups on N-sulfonylpyrrole appear to affect on the reaction rate in a size-dependent manner as tetrazoles 4-6 carrying the bulkier substituents gave 2–3-fold lower rates (Table 1; Figures S5–S10). To compare the reactivity of BCN to Sph, we performed a series of fluorescence-based competition experiments and found that BCN was about 15~17 times more reactive than Sph (Table 1; Figures S11–S16). The second-order rate constants for the tetrazole-BCN ligation reactions range from 11,400 M⁻¹ s⁻¹ to 39,200 M⁻¹ s⁻¹, with tetrazole 1 giving the fastest reaction. Notably, while the tetrazole-BCN ligation is faster than other BCN-mediated bioorthogonal reactions, it is slower than many tetrazine-TCO ligation reactions.^50–51

Table 1.

Kinetic characterization of the photoinduced cycloaddition of tetrazoles 1-6 with Sph or BCN^a

	k2 (M−1s−1)
tetrazole	Sph	BCN	kBCN/kSph
1	2,321 ± 76	39,200 ± 4,600	17
2	1,398 ± 98	20,800 ± 2,600	15
3	1,247 ± 25	21,000 ± 3,200	17
4	687 ± 37	11,400 ± 1,400	17
5	1,023 ± 59	17,700 ± 2,100	17
6	925 ± 79	14,500 ± 990	16

^aReactions were set up by mixing 5 μM tetrazole and 25 μM Sph or BCN in 0.5 mL PB/ACN (1:1). The mixture was photoirradiated with a handheld UV lamp (UVM-57; 302 nm, 0.16 AMPS) for a specified time under ambient conditions. The pyrazoline product was quantified based on its fluorescence. λ_ex = 405 nm. Measurements were repeated three times to derive the mean values and the standard deviations.

To probe the origin of the difference in reactivity between Sph and BCN, we used DFT to compute both the optimized geometries of the reactant complexes and transition states of the cycloaddition reactions. We considered the exo geometry of BCN in the reactant complex and the transition state because its free energy at 298 K is 5.0 kcal/mol lower than the endo. Calculations of the NI:Sph and NI:BCN reactant complexes (R) and their corresponding transition states (TS) revealed two discrete reaction paths. In principle, both Sph and BCN can approach the NI either end-on or side-on (Figure 2). However, the end-on path is not productive for Sph because its frontier π orbital is perpendicular to the cyclopropene ring (Figure 2). Moreover, the end-on approach is 3.1 kcal/mol less favorable than the side-on for Sph (Figures S27–S28). In contrast, despite its larger size, BCN preferentially approaches the NI end-on, allowing a closer interaction without causing any steric repulsions (Figure 2). The differences can be viewed by the torsional angle between the sulfonylpyrrole ring and the benzene ring in the transition states; this angle is −71.8° for BCN and −96.0° for Sph, suggesting that the sulfonylpyrrole ring rotates away during the side-on approach of Sph (Figure 2). The geometrical distortion effect on the TS complexes was quantified using the activation strain analysis⁵² (Table S4). This analysis confirmed that the TS for Sph experiences a larger strain. Furthermore, the transition state appears to be reached earlier with BCN (TS2) than Sph (TS1), as indicated by the longer bond distances in TS2 (Figure 2). In accordance with Hammond-Leffler postulate,^53–54 the transition state involving BCN and NI is structurally more similar to the reactant complex, which favors the product formation. Together, these geometrical and mechanistic differences result in a lower activation barrier (ΔG^ǂ) for BCN than for Sph by 1.2 kcal/mol (Figure 2) and thus eight times faster reaction for BCN, matching well with the experimental data.

Figure 2.

Comparison of the different approaches (top) and the transition state structures and energies (bottom) of the cycloaddition reactions with Sph and BCN. Green checkmarks depict favorable paths with their size proportional to the degree of preference, whereas the red X depicts the unfavorable path. DFT calculations were performed at the ωB97x-D/6–311++G(d,p)/SMD (water) level of theory using geometries optimized at the B3LYP-D3/6–31+G(d)/SMD(water) level of theory. Relative ΔG^ǂ for the transition states are reported in kcal/mol at 298 K and shown on the right, while distances in the reactant complexes (R) and the transition states are reported in Å. The methylsulfonyl groups were omitted for clarity. See Figures S25 and S29 in the Supporting Information for complete TS structures.

To assess the efficiency and orthogonality of the tetrazole–BCN ligation, we chemically modified lysozyme with BCN by treating lysozyme with para-nitrophenol-BCN carbamate in basic buffer for 8 hours (Figure 3a). QTOF-LC/MS analysis revealed mono-BCN modification to be around 64% with less than 6% double-modification based on the total ion counts of the various protein species (Figure 3b; Figure S17). The BCN-modified lysozyme (Lyso-BCN, 5 μM) was then photoirradiated with a handheld 302-nm UV lamp for 2 seconds in the presence of 10 equivalents of tetrazole 1. Quantitative conversion of Lyso-BCN to Lyso-pyrazole was observed by QTOF–LC/MS analysis (Figure 3c; Figure S18a). Importantly, no modification was detected with the unmodified lysozyme in the mixture (Figure 3c) or the pure lysozyme in solution (Figure S18b), indicating that the tetrazole–BCN ligation is orthogonal to the native functional groups present on the protein surface.

Figure 3.

Bioorthogonal labeling of chicken lysozyme via tetrazole-BCN ligation. (a) Scheme for sequential, selective modification of lysozyme using the BCN–tetrazole 1 pair. (b, c) Deconvoluted mass spectra for the BCN- and pyrazole-modified lysozymes. A small amount of doubly modified lysozymes was detected: Lyso-BCN₂, calcd 14658 Da, found 14,657.65 Da; Lyso-Pyrazole₂, calcd 15676 Da, found 15,678.73 Da.

To examine whether the tetrazole–BCN ligation is suitable for labeling low-abundance membrane proteins on live cells, we treated HEK 293T cells expressing glucagon receptor (GCGR) encoding BCNK at position-372 with 200 nM Cy5-conjugated tetrazole 7. Cells were photo-illuminated with a 302-nm handheld UV lamp for 15–60 sec (Figure 4a). After removing the excess probe, the cells were examined by a confocal microscope. The Cy5 channels showed strong fluorescence signals at the cell membrane after only 15-sec photoirradiation, which co-localized with fluorescence signals in GFP channel, indicating high labeling specificity (Figure 4b and Figure S19). As a control, cells expressing GCGR-H372BocK did not show any specific signals in the Cy5 channel (Figure S20). Compared to the genetically encoded Sph-lysine reporter, the BCN-lysine reporter served as a more robust reaction partner toward tetrazole 7 in fluorescent labeling of GCGR (Figure S21), consistent with BCN’s superior kinetics as well as efficient ring rupture of tetrazole 1 with a high quantum yield (Φ = 0.13, Figure S22).

Figure 4.

Bioorthogonal fluorescent labeling of GCGR in live cells via tetrazole-BCN ligation. (a) Scheme for bioorthogonal labeling of BCNK-encoded GCGR-GFP receptors by a Cy5-functionalized tetrazole. (b) Confocal micrographs of HEK 293T cells expressing GCGR-H372BCNK-GFP after photoirradiation with a handheld 302 nm UV lamp for 15, 30, or 60 sec in DMEM supplemented with 200 nM tetrazole-Cy5 (7). Scale bar = 10 μm.

In summary, we have designed and synthesized a new class of sterically shielded, sulfonated tetrazoles that react with BCN with exceedingly fast kinetics (k₂ = 11,400 ~ 39,200 M⁻¹ s⁻¹) in mixed acetonitrile–phosphate buffer (1:1), representing one of the fastest bioorthogonal ligation reactions involving BCN. Notably, the new bioorthogonal reagents offer improved water solubility while maintaining high reaction selectivity. Compared to the tetrazole–spiroalkene ligation, the tetrazole–BCN ligation reaction is ~17 times faster and employs a readily available chemical reporter. In addition, DFT calculations suggest that the increased rate is attributed to the unique ability of BCN to interact with the nitrile imine in an end-on fashion in the transition state. The power of the new tetrazole–BCN pair was demonstrated through a rapid (~2 sec) and selective modification of a BCN-modified protein in solution and a class B GPCR expressed on the mammalian cell surface (in 15 sec). The discovery of this robust bioorthogonal reactant pair should enhance the utility of BCN as a chemical reporter in diverse biological applications, particularly those involving spatiotemporal manipulation of the low-abundance biomolecules in living systems.