Cell-surface Labeling via Bioorthogonal Host–Guest Chemistry

Abstract

The widespread adoption of the bioorthogonal chemical reporter strategy revolutionized chemical biology. However, its translation to living mammals has been challenging, due to the size/stability properties of the chemical reporter group and/or the reaction kinetics of the labeling step. While developing new bioorthogonal reactions has been the traditional approach to optimizing the bioorthogonal chemical reporter strategy, here we present a different avenue, leveraging intermolecular interactions, to create bioorthogonal host–guest pairs. This approach, deemed “bioorthogonal complexation, does not rely on activated functional groups or second-order rate constants. We utilize the cucurbit[7]uril (CB[7]) scaffold to showcase bioorthogonal complexation and determine that medium-affinity (K_a ≈ 10⁸–10⁹ M^–1) guests efficiently label cell surfaces and outperform the strain-promoted azidealkyne cycloaddition. Finally, we implement bioorthogonal complexation in the chemical reporter strategy through the metabolic incorporation of ortho-carborane into cell-surface glycans and detection with a CB[7]-fluorescein conjugate.

Graphical Abstract

Advances in chemical biology have resulted in unique methods to probe and manipulate biomolecules in cells and animals.^1,2 A classic chemical biology method is the bioorthogonal chemical reporter strategy (Figure 1A)—a two-step approach to biomolecule labeling involving the incorporation of an unnatural metabolite, followed by a selective covalent chemical reaction.^3,4 The bioorthogonal chemical reporter strategy expanded the scope of biomolecules analyzed in their native environments by introducing unnatural functionality through metabolic incorporation.^5–9 A challenge that has arisen with this method surrounds its reliance on second-order reaction kinetics that often require high concentrations of a labeling reagent (Y), primed for reactivity with the chemical reporter group (X) (see Figure 1B, as well as Figure S1 in the Supporting Information). The high concentration of activated functionality leads to high background, especially in animal models.^10–12 To overcome these challenges, the field has developed new bioorthogonal chemistries focusing on increased second-order rate constants (k).^13–15 Significant success resulted in the tetrazine ligation with reaction kinetics 7–10 orders of magnitude faster than early bioorthogonal chemistries.^16,17 However, the size and stability^17–19 of the chemical reporters/labeling reagents in the fastest bioorthogonal chemistries limit opportunities for metabolic incorporation.

Figure 1.

A complementary method to biomolecular tagging in living systems. (A) The bioorthogonal chemical reporter strategy. (B, C) Advantages, challenges, and examples of bioorthogonal chemistry (panel (B)) and complexation (panel (C)). See Figure S1 for further details.

A complementary approach to detect chemical reporter groups is to leverage host–guest chemistry.^20–23 We give this approach using bioorthogonal noncovalent pairs the name bioorthogonal complexation (Figure 1C) to clearly differentiate it from bioorthogonal chemistry, which relies on covalent chemistry. The advantages of bioorthogonal complexation include (1) high on-rates (often approaching diffusion control in water),²⁴ (2) absence of activated functional groups mitigating side reactivity, and (3) reversibility allowing undesired labeling to be corrected. We envision that these advantages will be particularly relevant for detection of chemical reporter groups in dilute conditions and complex environments.^25,26 The widespread utility of the biotin–(strept)avidin pair showcase the power of noncovalent chemistries; however, their endogenous presence and large size is limiting.²⁶

The challenge of bioorthogonal complexation is the development of selective host–guest pairs with small guests that can be metabolically incorporated yet have high enough K_a values for efficient labeling (Figure 1C). With the long-term goal of designing bioorthogonal host–guest pairs that fulfill these requirements, we sought to determine the minimum K_a necessary to efficiently label cell surfaces. We utilized the cucurbit[7]uril (CB[7]) scaffold, a synthetic biotin–avidin mimic, due to the impressive binding affinities that can be achieved.^27–29 CB[7] and high-affinity guests adamantylamine and ferrocene amine have been used to isolate proteins^30,31 and label proteins in live^32,33 and fixed cells.³⁴ While this work suggested that CB[7] would successfully detect high-affinity (K_a ≥ 10¹² M^–1) guests presented on the surface, we were particularly intrigued by CB[7]’s low to medium-affinity guests (K_a = 10⁵–10¹⁰ M^–1). Their smaller size offers more opportunities for metabolic incorporation, an advantage of the bioorthogonal chemical reporter strategy. Furthermore, medium-affinity complexes are more accessible, setting the stage for custom design of bioorthogonal host–guest complexes.

To explore the limits of bioorthogonal complexation, we designed a flow cytometry-based assay where guests were introduced on Jurkat cell surfaces via oxime ligation, followed by tagging with a direct CB[7]-fluorescein conjugate (CB[7]-FITC) (Figure 2A). The oxime ligation was necessary to ensure compatibility with the amine functionality present in CB[7] guests, and Jurkat cells were chosen for their high sialic acid content.³⁵ We chose CB[7] guests spanning a K_a range of 10⁶–10¹⁴ M^–1 to determine the minimum K_a value and lowest concentration required to achieve statistically significant labeling using CB[7]-FITC (Figure 2B). CB[7]-FITC was prepared via functionalizing CB[7]-N₃ with 1 through strain-promoted azide–alkyne cycloaddition (SPAAC; see Figure 2B, as well as Scheme S1 in the Supporting Information).³⁴ The requisite amino-oxy functionalized guests 2–6 were prepared in four steps (Figure 2C, as well as Scheme S2A in the Supporting Information).³⁶ After optimizing the cell-surface oxidation and oxime ligation conditions with an amino-oxy fluorophore (Figure S2, as well as Scheme S3 in the Supporting Information), we presented guests 2–6 on the cell surface.

Figure 2.

Determination of the minimum K_a required for cell-surface bioorthogonal complexation. (A) Bioorthogonal complexation assay developed to assess the minimum K_a requirement of the strategy (B) CB[7]-FITC conjugate prepared from CB[7]-N₃ and 1. (C) Amino-oxy-guest compounds (2–6) with varying K_a values used in the assay. (D) Histogram and (E) bar graph of flow cytometry analysis of bioorthogonal complexation with CB[7]-FITC and different guests. Jurkat cells were oxidized using NaIO₄ (1 mM) for 10 min at 0 °C, followed by incubation with 2–6 (500 μM), and aniline (10 mM) for 1 h at 0 °C. Cells were then treated with CB[7]-FITC (1 μM) for 30 min at 0 °C and analyzed by flow cytometry. See Figures S3 and S4 in the Supporting Information for replicates. Error bars represent the standard deviation of the mean of triplicates in a single experiment.

Excitingly, treatment of guest-functionalized cells with 1 μM CB[7]-FITC (Figures 2D and 2E, as well as Figures S3 and S4 in the Supporting Information) displayed excellent cell-surface labeling for four of the five guests tested with signal-to-noise (S/N) values ranging from 8.4 ± 0.2 to 122 ± 8 (3–6, K_a = 10⁸–10¹⁴ M^–1). The lowest K_a guest tested (2, K_a = 10⁶ M^–1)²⁷ did not display statistically significant labeling at any concentration of CB[7] tested (up to 20 μM; see Figure S5 in the Supporting Information), establishing 10⁸ M^–1 as a benchmark K_a requirement for noncovalent labeling approaches that are not enhanced via avidity. It is possible that this K_a threshold could be lowered if alternative methods to increase guest concentrations on the cell surface or the use of hosts with even lower background binding were available. We did observe variability in the high-binding affinity guests (5 and 6), suggesting that K_a values of >10¹² M^–1 may not be necessary.

We further explored bioorthogonal complexation requirements with dose- and time-dependent experiments. Interestingly, labeling efficiency did not improve significantly at higher CB[7]-FITC concentrations (2–10 μM) (Figure S6 in the Supporting Information). Probing the lower concentration limit of guests 4–6 demonstrated that statistically significant labeling at low nanomolar concentrations could be obtained (10 nM for 4, 1 nM for 5 and 6; see Figures S6–S8 in the Supporting Information). These values are 2–3 orders of magnitude lower than concentrations used for labeling cell surface chemical reporter groups with bioorthogonal chemistry and detection by flow cytometry.^13,37–40 Time-dependent experiments with guest 6 and 1 μM CB[7]-FITC showed immediate statistically significant labeling with a signal-to-noise (S/N) ratio of 9.1 ± 0.2, which increased to 22 ± 3 after 30 min. Interestingly, the flow cytometry histograms showed two clear populations (Figure S9 in the Supporting Information), suggesting that there is heterogeneity in the guest accessibility on the cell surface. Further studies are necessary to fully understand the biomodal distributions observed, which we also believe will provide insight on the differences in the high binding (K_a > 10¹² M⁻¹) guests.

The low concentration limits of bioorthogonal complexation prompted us to perform a comparative study with bioorthogonal chemistry.¹⁷ Besides higher concentrations needed for bioorthogonal chemistry with flow cytometry detection, a two-step labeling where the bioorthogonal chemistry introduces a cell-impermeable biotin that is then detected by fluorescently labeled (strept)avidin is commonly used. This secondary labeling step amplifies the signal by delivering many fluorophores per biotin. To compare bioorthogonal chemistry and complexation labeling, we chose to compare the widely used SPAAC between an azide and bicyclononyne (BCN)³⁷ to the medium-affinity host–guest complex of (trimethylsilyl)methylamine and CB[7]-FITC.²⁷ We introduced a consistent amount of chemical reporters onto cell surfaces using the oxime ligation with 7 (Scheme S2B in the Supporting Information) and 4, respectively. The cells were then treated with a fluorescein-BCN conjugate (1) or CB[7]-FITC at 10 μM (Figure 3A). While labeling with 4 and CB[7]-FITC resulted in a robust S/N ratio of 6 ± 2, treatment with 10 μM 1 did not produce statistically significant results (see Figures 3B and 3C, as well as Figure S10 in the Supporting Information). Applying the traditionally employed signal amplification step through treatment with an α-fluorescein IgG-CF640R conjugate, resulted in statistically significant labeling (see Figure 3D, as well as S9 in the Supporting Information); however, the S/N remained below bioorthogonal complexation (2.2 ± 0.5). Increasing the concentration of 1 (100 μM) (Figure S10 in the Supporting Information) to match previous reports^37,40 improved the S/N to 3.5 ± 0.2. Note that the conditions in this comparison experiment were optimized for bioorthogonal chemistry, resulting in lower S/N ratios than previously observed with bioorthogonal complexation (Figure S6 in the Supporting Information). Similar experiments with 5 demonstrated even larger differences between bioorthogonal chemistry and complexation (see Figures S11–S13 in the Supporting Information).

Figure 3.

Comparative study between bioorthogonal complexation and bioorthogonal chemistry. (A) Schematic of experiment comparing bioorthogonal chemistry (SPAAC) and bioorthogonal complexations (K_a = 10⁹ M^–1). (B–D) Jurkat cells were oxidized using NaIO₄ (1 mM) for 10 min at 0 °C followed by incubation with 4, 7 (500 μM), and aniline (10 mM) for 1 h at 0 °C. Fluorescence labeling occurred with 1 or CB[7]-FITC (10 μM) for 45 min at 0 °C. Cells treated with 7 and 1 were labeled with α-Fluorescein IgG-CF640R (0.002 mg mL^–1) twice for 15 min at 0 °C and analyzed by flow cytometry. Data are represented as a bar graph of mean fluorescence intensities (MFI in arbitrary units, au). (B) Representative histograms for (C) fluorescein (FL1) and (D) CF640R (FL4) emission. See Figure S10 in the Supporting Information for replicate experiments.

Here, we have presented one point of comparison and we note that exchanging bioorthogonal reactions and/or fluorescent payloads^14,15 could yield improved labeling. However, the demonstration of the superior labeling provided by bioorthogonal complexation at micromolar concentrations and the finding that it can be successful at nanomolar concentrations with a direct fluorophore conjugate showcase the potential for labeling chemical reporter groups with high S/N ratios in living systems. Future generations of host–guest pairs with smaller size, higher cell permeability and faster diffusion would translate bioorthogonal complexation to intracellular targets that can currently be labeled with bioorthogonal chemistries.

While the oxime ligation allowed direct comparison of bioorthogonal complexation and chemistry, ultimately for bioorthogonal complexation to be employed in the bioorthogonal chemical reporter strategy, guests must be metabolically incorporated. Metabolic oligosaccharide engineering allows guests to be appended to sialic acid residues on the cell surface when modified mannosamine or sialic acid monosaccharides are introduced. Given that a majority of known CB[7] guests are bulky molecules and cationic under physiological conditions, the incorporation of most guests in Figure 2 would be challenging.^41,42 Indeed, our efforts to incorporate the smallest guest, (trimethylsilyl)methylamine, as a sialic acid derivative were unsuccessful (see Scheme S3 and Figures S14 and S15 in the Supporting Information). Considering the large chemical reporters previously incorporated in the 9-position of sialic acid,⁴² we attributed the lack of incorporation to the cationic nature of the (trimethylsilyl)methylamine guest.

Thus, to apply bioorthogonal complexation to the chemical reporter strategy, we turned to a recently reported medium-affinity neutral guest for CB[7]: ortho-carborane.⁴³ Orthocarborane ethanoic acid was appended to 9-aminosialic acid (Figure 4A, as well as Scheme S4 in the Supporting Information) to produce 10 that was found to bind CB[7] with K_a = 2 × 10⁹ M^–1 in 50 mM NaOAc buffer (Figure S16 in the Supporting Information). Successful incorporation in engineered, hyposialylated BJAB K20 cells was determined using a flow cytometry-based lectin assay with 11 (the peracetylated derivative of 10) (see Figures S17 and S18 in the Supporting Information). Upon incorporation of 11, CB[7]-FITC was used for cell-surface labeling via bioorthogonal complexation (see Figures 4B–D, as well as Figures S19 and S20 in the Supporting Information). The higher CB[7]-FITC concentrations (20 μM) required were attributed to higher background due to the hyposialylated cell line, although a lower K_a value of 10 in cellulo could also be responsible. This confirms previous work on the suitability of carboranes as chemical reporters for cell-surface processes.⁴⁴ We envision that the unique properties of carborane will allow for expanded applications of bioorthogonal complexation.⁴³ Further guest optimization strives to allow the incorporation of smaller, neutral guests that would allow use of wild-type cell lines, such as Jurkat.

Figure 4.

Metabolic incorporation of a CB[7] guest and labeling via bioorthogonal complexation. (A) Preparation of 11 from sialic acid precursor 8 and activated 9. (B–D) BJAB K20 cells were grown in 11 (0–50 μM) for 3 days in 90% nutridoma/FBS RPMI media. After washing, the cells were incubated with CB[7]-FITC (20 μM) for 30 min at 0 °C and analyzed by flow cytometry. A representative histogram is shown in panel (C) and an averaged MFI bar graph is shown in panel (D). Error bars represent the standard deviation of the mean of triplicates in a single experiment. [See Figures S19 and S20 for replicate experiments.]

In this report, we introduce bioorthogonal complexation as a complementary strategy to bioorthogonal chemistry for labeling chemical reporter groups. Bioorthogonal complexation efficiency relies on thermodynamics rather than kinetics, and we determined that a K_a value of ≥ 10⁸ M^–1 is necessary for robust detection of cell-surface chemical reporter groups. This is a threshold that is achievable through molecular design, providing opportunities for the development of custom bioorthogonal host–guest pairs. Concentration-dependent studies and a direct comparison to SPAAC confirmed that lower concentrations of reagent were necessary to achieve high S/N labeling with bioorthogonal complexation than bioorthogonal chemistry. The nM concentrations necessary and the decreased propensity for covalent background labeling overcome long-standing challenges for bioorthogonal chemistry.⁴⁵ Finally, the metabolic incorporation of an ortho-carborane sialic acid derivative and its detection using CB[7]-FITC achieved the use of bioorthogonal complexation in the chemical reporter strategy.

These findings expand the chemical toolbox for labeling unnatural functionality in biological systems. We expect bioorthogonal complexation to be particularly relevant for direct imaging experiments, where a covalent bond is not required, such as flow cytometry and microscopy experiments. Notably, in some instances, such as activity-based protein profiling, a covalent bond is essential to the experiment and bioorthogonal chemistry will still be the preferred choice. Other applications where noncovalent chemistries are suitable, such as in pull-down experiments, may have different K_a requirements, which remains an area of exploration. Overall, we believe bioorthogonal complexation will be an advantageous approach for translating the bioorthogonal chemical reporter strategy to complex organisms.

METHODS

Details of experimental material and methods are provided in the Supporting Information.