2025 Volume 73 Issue 9 Pages 793-801
Antigen-binding proteins, such as nanobodies, modified with functional small molecules hold great potential for applications including imaging probes, drug conjugates, and localized catalysts. However, traditional chemical labeling methods that randomly target lysine or cysteine residues often produce heterogeneous conjugates with limited reproducibility. Conventional site-specific conjugation approaches, which typically modify only the N- or C-terminus, may also be insufficient to achieve the desired functionalities. Genetic code expansion offers a powerful alternative by enabling the site-specific incorporation of noncanonical amino acids bearing reactive handles—such as trans-cyclooctene (TCO)—thereby allowing precise bioorthogonal conjugation via click chemistry. Nevertheless, identifying suitable incorporation sites that tolerate such modifications without disrupting antigen binding remains a time- and cost-intensive process, as this process typically requires labor-intensive screening involving the expression and purification of each candidate variant. Here, using HER2 and an anti-HER2 nanobody as a model antigen–binder pair, we present a convenient mammalian cell-based screening platform for rapid, purification-free evaluation of site-specifically labeled nanobodies. The nanobody is fused to blue fluorescent protein (BFP), secreted by HEK293T cells, and labeled in situ with a tetrazine–fluorescein probe. The resulting supernatant is then applied directly to HEK293T cells stably expressing HER2–mCherry. Labeling efficiency and retention of antigen-binding activity are simultaneously assessed by fluorescence imaging in the BFP, fluorescein, and mCherry channels. This approach enables efficient identification of labeling sites that support productive click conjugation while preserving binding function. It should be broadly applicable to other antigens and binders, streamlining early-stage screening of engineered antigen–binder conjugates for diverse applications.
Bioconjugates that combine antigen-binding proteins, including genetically engineered antibody fragments like nanobodies, with small functional molecules (such as fluorophores, therapeutic agents, and catalysts) show great promise for a range of biomedical applications, including precision imaging, targeted drug delivery, and localized catalysis.1–9) However, conventional chemical labeling methods that randomly target lysine or cysteine residues produce heterogeneous conjugates with limited reproducibility, which can hinder downstream biomedical applications. Moreover, for conjugates designed to achieve specific functions (e.g., fluorogenic sensors or precisely oriented catalysts), site-specific attachment of the small molecule is essential.
In this context, genetic code expansion enables the incorporation of a reactive noncanonical amino acid (ncAA) at user-defined positions.10–14) Subsequent bioorthogonal click reactions with functional small molecules can generate highly uniform conjugates with well-defined architectures, enabling thorough quality control and functional predictability. This approach is especially valuable when conventional methods—such as sortase-mediated conjugation—are insufficient, as they typically restrict labeling to the N- or C-terminus and cannot access internal residues. In contrast, site-specific incorporation of ncAAs can precisely position small molecules at internal sites, which could be critical for constructing conjugates that exert specific functions (e.g., fluorogenic sensors or site-specific catalysts).
However, identifying residues that tolerate the insertion of a clickable ncAA without compromising antigen-binding activity or protein folding remains a significant challenge. Although one could, in principle, express (e.g., using bacterial expression systems), purify, and individually characterize each candidate conjugate, the high cost of ncAAs and the laborious nature of such protocols quickly become prohibitive.
In the present study, using HER2 and the anti-HER2 nanobody 2Rs15d15) as a model antigen/binder pair, we present a convenient mammalian cell-based screening platform to evaluate the producibility and binding capability of nanobody–small molecule conjugates generated via genetic code expansion and click chemistry. This approach bypasses laborious bacterial expression and purification by leveraging genetic code expansion in HEK293T cells. Cells were seeded into separate wells and individually transfected with plasmids encoding 2Rs15d nanobody variants, each harboring a site-specific amber codon for the incorporation of trans-cyclooctene–lysine (TCO-K) at a distinct position. Each variant is expressed and secreted as a C-terminal fusion with blue fluorescent protein (BFP). The secreted TCO-bearing 2Rs15d-BFP proteins in each well were subsequently labeled in situ with a tetrazine–fluorescein (Tz–Flu) probe. After directly applying the labeled mixture to HEK293T cells expressing HER2 fused to mCherry on their plasma membranes, live-cell observation of 3 fluorescence channels—BFP (blue), fluorescein (green), and mCherry (red)—allows rapid screening of nanobody variants to distinguish those that retain antigen-binding activity from those that either fail to expose the reactive ncAA side chain or suffer from impaired function upon chemical modification. Having validated the transferability of the mammalian cell-based screening results to a recombinant protein expressed in Escherichia coli (E. coli), we propose the platform as a practical and scalable strategy for selecting optimal conjugation sites prior to large-scale production and downstream functional characterization.
Our goal in this study was to create a direct, mammalian cell–based screening method that eliminates the need for extensive bacterial expression and purification of each candidate nanobody conjugate. For this purpose, we designed a workflow in which mammalian cells secrete nanobodies bearing an ncAA with a reactive handle for click chemistry at various positions. The secreted nanobodies are labeled directly in the culture supernatant using a fluorophore containing the complementary click handle, and the resulting conjugates are applied to antigen-expressing cells without purification, enabling a rapid, imaging-based assessment of antigen-binding properties (Fig. 1a).
(a) Schematic illustration of the cell-based screening platform. After transfecting HEK293T cells with the Igk-nanobody-BFP plasmid (bearing amber codon [TAG] at the site to incorporate TCO-K) and the Mm-PylRS-AF/tRNACUA plasmid (Addgene, #122650), the cells were cultured with 50 μM TCO-K overnight. After replacing the medium (without TCO-K), nanobody-BFP bearing TCO-K accumulates in the cell culture supernatant. Then, tetrazine–fluorescein (Tz–Flu) is added to the supernatant, reacts with TCO, and becomes fluorescent. The supernatant containing labeled nanobody is applied directly to HEK293T cells expressing HER2–mCherry. Imaging was then conducted with the CQ1 imaging plate reader. (b) Absorption spectra of 1 μM of Tz–Flu before and after the reaction with 10 μM of TCO-K. (c) Fluorescence spectra of 1 μM of Tz-Flu before and after the reaction with 10 μM of TCO-K. (d) Sequence of the nanobody-BFP fusion and the positions tested for TCO-K incorporation. Letters in green encode the Igk chain, letters in red encode 2Rs15d, and letters in blue encode the tagBFP sequence. For mutant numbering, the first Q is counted as 1, and the Final S is counted as 115. To generate amber mutants, each codon encoding one of the 115 amino acids of the nanobody was individually replaced with the amber stop codon (TAG), one by one. Detailed plasmid sequences are available in the Supplementary Materials.
Among the possible combinations of click handles to incorporate (e.g., an azide in ncAA and an alkyne in the fluorophore, or vice versa), we first selected a tetrazine-conjugated fluorophore as the labeling probe. Tetrazine not only enables copper-free click chemistry but also functions as a fluorescence quencher until reaction occurs, providing a fluorogenic labeling system suitable for evaluating fluorescence without purification.16) For the counterpart in ncAA bearing the click handle, we chose trans-cyclooct-2-en-l-lysine (axial isomer), referred to as 2′a-TCO-K, from several candidates including lysine bearing other TCO variants (e.g., 3/4-a/e-TCO) and bicyclononyne (BCN). 2′a-TCO-K was selected based on its high stability and its compatibility with the genetic code expansion system employing the publicly available evolved pyrrolysyl–tRNA synthetase/tRNACUA pair (Mm-PylRS-AF).17–20) For simplicity, we refer to 2′a-TCO-K as TCO-K in this paper.
Based on these considerations, we first synthesized fluorescein bearing H-tetrazine (a Tz derivative showing fastest click reaction) at the 6 position of the benzene moiety (Tz–Flu). We indeed confirmed that this probe exhibited a >12-fold increase in fluorescence upon reaction with TCO-K due to the increase of quantum yield of fluorescence (Figs. 1b, 1c).
Then, we constructed a plasmid encoding the anti-HER2 nanobody 2Rs15d fused to tagBFP under a cytomegalovirus (CMV) promoter, with an Igκ signal peptide at the N-terminus to ensure secretion. To enable site-specific incorporation of TCO-K, amber codons (TAG) were introduced into every position of the nanobody via site-directed mutagenesis, generating a library of 115 variants spanning the full nanobody sequence (Fig. 1d). Each mutant plasmid (1 per well in a 24-well plate) was co-transfected with the Mm-PylRS-AF system into HEK293T cells, which are known for efficient recombinant protein production. After 16 h of incubation in medium containing 50 μM TCO-K, the medium was replaced to remove unincorporated ncAA, and newly synthesized, TCO-bearing nanobody–BFP accumulated in the supernatant.
This supernatant containing TCO-bearing nanobody–BFP was then directly mixed with 100 nM Tz–Flu (determined after preliminary experiments so that the labeling efficacy saturates while background fluorescence derived from free Tz–Flu is acceptable; data not shown), enabling a rapid bioorthogonal ligation. The resulting solution—containing the labeled nanobody and residual quenched dye—was directly applied to HEK293T cells stably expressing HER2–mCherry on the plasma membrane (seeded in 96-well plates). We recorded fluorescence signals in 3 channels: BFP (blue), fluorescein (green), and mCherry (red), anticipating that this setup would enable simultaneous evaluation of (1) nanobody secretion, (2) successful labeling, and (3) retention of HER2-binding activity, indicated by colocalization of BFP and fluorescein signals with the mCherry-labeled membrane.
Screening ResultsCells from each well were analyzed using an imaging plate reader (CQ1, Yokogawa Electric, Tokyo, Japan), which enables high-throughput live-cell imaging (Fig. 2a). To assess the localization of blue or green fluorescence relative to red fluorescence derived from HER2–mCherry, we quantified the ratio of blue or green signal intensity to red signal intensity within each field of view (Figs. 2b, 2c). We note that relatively strong extracellular Tz–Flu signals were observed in some wells (e.g., 6, 16, 24, 36, and 46). This may have been caused by insufficient washing of TCO-K prior to the accumulation of the TCO-K–bearing antibody. To minimize this effect, we quantified green fluorescence only in pixels where a sufficient mCherry signal was detected.
(a) Evaluation using the CQ1 imaging plate reader. Conditioned media containing 2Rs15d–BFP fusion proteins (nanobody bearing TCO-K at the indicated position), labeled with Tz–Flu, were applied to HEK293T cells stably expressing HER2–mCherry. The numbers in the figure indicate the position of TCO-K incorporation (numbers are sequentially assigned from left to right; after every 10 numbers, the numbering continues on the next line starting from the left again). For each condition, 4 fields of view were imaged, and 10 z-stack images were acquired at 2 μm intervals. The images shown correspond to the z-slice, in which HER2–mCherry fluorescence was brightest for each condition. Blue: Ex/Em = 405/417–477 nm; green: Ex/Em = 488/500–550 nm; red: Ex/Em = 561/581–654 nm. Scale bar: 100 μm. A high-resolution version is provided in Supplementary Data 1. (b, c) Quantitative analysis of the imaging data. Membrane regions were identified based on the mCherry signal, and the corresponding BFP and fluorescein signals within these regions were quantified. For each field of view and z-section, the green-to-red and blue-to-red signal ratios were calculated. The optimal z-section for each field was selected according to the criteria described in Experimental. Box-and-whisker plots were generated using the green-to-red signal ratios (b) and blue-to-red signal ratios (c) from the best z-section in each field (n = 4–6 per group).
Notably, at mutant positions 16, 17, 26, 41, 42, 46, 57, 84, and 100, both BFP and fluorescein signals clearly colocalized with mCherry at the membrane. This pattern indicates that the nanobodies bearing TCO-K at these positions were successfully secreted, labeled with Tz–Flu, and retained HER2-binding activity. These variants therefore possess the desired combination of efficient labeling and preserved target binding.
In contrast, mutant 71 showed colocalization of BFP and mCherry at the membrane, but little or no fluorescein signal was detected. Given that Tz–Flu was applied in excess, this likely indicates that the nanobody was secreted and capable of binding HER2, but that the TCO group may have been buried or conformationally restricted, rendering it less accessible for tetrazine conjugation.
For most of the remaining mutants, colocalization of BFP and fluorescein with mCherry was weak or undetectable. These outcomes suggest either poor nanobody expression or interference with HER2 binding due to labeling. (Further experiments would be required to distinguish between these possibilities, but this lies beyond the scope of the present study.)
As a control, we also confirmed that wild-type nanobody produced a strong BFP signal that colocalized with mCherry, while no fluorescein signal overlapped with mCherry—validating the specificity and reliability of our screening approach.
Further Validation of the Screening ResultsIn the experiment shown in Fig. 2, it was practically difficult to control the number of recipient cells in each imaging field and to ensure uniform removal of excess TCO-K across all wells—factors that can affect image quality (as discussed above). To address this, we conducted a follow-up experiment using a smaller set of samples and more stringent washing procedures to remove free TCO-K. We also ensured consistent and sufficient numbers of HEK293T cells expressing HER2–mCherry. Fluorescence signals were then recorded using a Leica (Wetzlar, Germany) SP8 confocal microscope equipped with a highly sensitive HyD detector, with manual and precise focus adjustment. Among the mutants that showed strong colocalization of both BFP and fluorescein with mCherry in the initial screen, we selected mutants 17, 26, 41, 42, and 100 for further validation. The wild-type nanobody was also included as a control, expected to yield only BFP signal colocalized with mCherry.
The results obtained from this validation were consistent with the initial screen, supporting the robustness of our screening strategy (Fig. 3a). Specifically, while the wild-type nanobody showed only BFP colocalized with mCherry, all 5 selected mutants displayed clear colocalization of both BFP and fluorescein with mCherry. Furthermore, mutants 17, 26, and 41 exhibited stronger fluorescein signals than mutants 42 and 100, in agreement with the initial screen.
(a) Re-assay using the confocal microscope SP8. Experiments were conducted using a similar scheme as in Fig. 2a, ensuring a stable cell count and observation with manual focus. (b) Coarse structural prediction of each nanobody–Tz–Flu complex. Nanobody 2Rs15d is shown in blue, and HER2 extracellular domain (ECD) is shown in pale yellow. For the detailed method, see Experimental. (c) Validation of the purified conjugate of Tz–Flu and mutant 41 expressed in E. coli. One hundred nanomolar of the purified conjugate in OptiMEM was applied to HEK293T cells expressing HER2–mCherry. Note that BFP is not fused to the purified nanobody. For (a) and (c), imaging settings were as follows: blue (PMT, not obtained in (c) because BFP is not fused): excitation/emission = 405/450–510 nm; green (HyD): excitation/emission = 498/510–580 nm; red (HyD): excitation/emission = 587/600–750 nm. Scale bars: 100 μm.
To interpret these results, we performed coarse structural modeling based on the structure of the complex of HER2 extracellular domain and nanobody 2Rs15d (PDB 5MY621)) (Fig. 3b). We found that positions 17, 26, and 41 are located away from the binding interface and therefore unlikely to disrupt antigen recognition. In contrast, positions 42 and 100 are located closer to the interface, which may explain the relatively weaker fluorescein signal observed at these positions.
Nonetheless, it is noteworthy that fluorescein signals were still observed at the membrane for mutants 42 and 100. This demonstrates that our screening method can identify functional conjugates even when structural predictions alone would suggest impaired binding. This capability should be particularly useful for the design of biosensors incorporating environment-sensitive dyes, where precise positioning is critical. Indeed, the reduced fluorescence observed here might reflect a decrease in the fluorescein upon antigen binding, suggesting that the mutant 100 conjugate might be functioning as an “OFF-type” fluorescence sensor of the target antigen.
Finally, we evaluated whether the data obtained via mammalian cell–based screening are transferable to a recombinant protein expressed in E. coli. In our recent study, we demonstrated that 2Rs15d, when N-terminally fused with a bdSUMO tag, can be successfully expressed in soluble form in Shuffle T7 Express Competent E. coli cells.22) Therefore, we initially attempted to implement the genetic code expansion system based on Mm-PylRS-AF by relocating the enzyme and its corresponding tRNACUA into E. coli expression vectors (Supplementary Table 1). However, the expression level of the nanobody was too low for downstream assays, suggesting that system-specific optimization is necessary at least for the nanobody used in this study (data not shown). To overcome this limitation, we adopted a genetic code expansion system based on Methanomethylophilus alvus PylRS (MaPylRS), which has been shown by our group and others to support efficient incorporation of TCO-K into proteins.23,24) We co-transformed Shuffle T7 Express Competent E. coli cells with 2 plasmids: 1 encoding a double copy of the MaPylRS mutant (ALIP: Y126A/M129L/H227I/Y228P) and a single copy of its corresponding tRNACUA, and other encoding MBP-bdSUMO-2Rs15d (with an amber codon)-6xHis tag. (To enhance solubility, an MBP tag was additionally fused to the N-terminus of bdSUMO; see Supplementary Table 1.) We found that recombinant 2Rs15d nanobody can be successfully expressed and purified with this system (Supplementary Fig. 1).
Using this method, we prepared a recombinant TCO-K–modified 2Rs15d variant for mutant 41, labeled it with Tz–Flu, and purified the resulting conjugates (see Experimental). When the Tz–Flu conjugate of mutant 41 was applied to HEK293T cells expressing HER2–mCherry, clear cell labeling was observed, supporting the transferability of screening results obtained from mammalian cells to proteins expressed in a different host (Fig. 3c).
Thus, we have established a convenient strategy for screening nanobody variants engineered via genetic code expansion—without the cost and complexity of individually purifying each conjugate following bacterial expression. By secreting TCO-bearing nanobodies into mammalian cell culture, labeling them directly with Tz–Flu in the culture supernatant, and applying the labeled mixture to cells presenting the target antigen, we can rapidly discriminate sites that yield optimal conjugates. Because Tz–Flu remains mostly quenched until it reacts, the unbound probe contributes little background. Importantly, we have demonstrated the feasibility of a downstream assay using nanobodies expressed and purified via an E. coli–based system, indicating that screening results obtained from the mammalian cell–based platform are transferable to proteins expressed through alternative methods. While E. coli–based expression has clear advantages for large-scale production, each protein variant must be purified to evaluate it in mammalian systems, which can limit the assay throughput. In contrast, our mammalian cell–based assay enables parallel, purification-free testing of many nanobody constructs in situ, thus providing higher throughput. Used together, the 2 systems are therefore complementary: rapid, low-cost screening in mammalian cells can triage large libraries. This screening requires only 50 μM of TCO-K in a few 100 μL of culture medium per variant, which is particularly advantageous given the high cost of TCO-K. The promising hits can then be scaled up efficiently in E. coli for detailed biochemical studies. We anticipate that this combined approach could accelerate the development of functional nanobody–small molecule conjugates.
This method could be extended to other single-domain binders (e.g., scFv, DARPins25)), provided that they can be secreted from mammalian cells and incorporate a clickable ncAA—such as TCO-K or BCN-K. Furthermore, replacing Tz–Flu with alternative tetrazine-bearing probes, including rhodamine-based dyes, would be of interest, as the physicochemical properties of the dye—such as its net charge—can significantly influence the antigen-binding behavior of the nanobody–dye conjugate. Further, as previously discussed, the incorporation of environment-sensitive dyes could expand the utility of this platform for developing next-generation biosensors that report conformational or environmental changes upon antigen binding.6,7) Similarly, substituting Tz–Flu to a functional small-molecule drug could be a scope to develop homogenous antigen-binder–drug conjugate. Because the pharmacokinetic properties of such conjugates may depend on the labeling position,3) a convenient screening platform for evaluating nanobody–drug conjugates could be useful for selecting promising candidates for downstream validation. Likewise, replacing Tz–Flu with a catalytic moiety may lead to the development of protein–small molecule–catalyst hybrids for targeted catalysis in complex biological systems. Furthermore, the scope of this strategy need not be limited to small molecules. For example, replacing Tz–Flu with another protein bearing a complementary click handle could enable the assembly of nanobody-based multivalent constructs. Nanobodies assembled via click chemistry have already shown promise in improving the efficacy of nanobody-based therapeutics and biotechnological tools.26)
We also note that a similar assay pipeline could, in principle, be adapted for intracellular antigens with appropriate modifications. Specifically, a TCO-bearing nanobody–BFP fusion could be expressed intracellularly and labeled with a cell-permeable Tz dye. The target protein fused to mCherry and anchored to the inner leaflet of the plasma membrane could be co-expressed in the same cell, enabling colocalization of BFP, dye, and mCherry signals to be monitored simultaneously. As a preliminary proof of concept, we confirmed that binding of the nanobody–BFP fusion to a membrane-anchored target protein could be detected via colocalization with mCherry (Supplementary Fig. 2). This setup could potentially be adapted for pooled screening formats—e.g., by expressing different nanobody variants in different cells using lentiviral vectors, assessing binding via imaging, and identifying each variant by DNA sequencing. Such advanced approaches may form the basis for future work.
In summary, our results show that a mammalian cell-based screening approach can facilitate the identification of functional nanobody–small molecule conjugates. We envision this platform will be valuable for developing advanced biosensors, drug conjugates, and catalysts that exploit the power of site-specific labeling of antigen-binding proteins.
General chemicals were of the best grade available and were supplied by FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan), Tokyo Chemical Industries Co., Ltd. (Tokyo, Japan), KANTO CHEMICAL Co., Inc. (Tokyo, Japan), and Sigma-Aldrich Co., LLC (St. Louis, MO, U.S.A.). They were used without further purification. TCO-K (2′a-TCO-K) was purchased from SiChem (SC-8008), Bremen, Germany.
1H- and 13C-NMR spectra were recorded on an AVANCE III 400 NanoBay (Bruker, Billerica, MA, U.S.A.; 400 MHz for 1H-NMR and 101 MHz for 13C-NMR). Mass spectra (MS, electrospray ionization-time-of-flight [ESI-TOF] were measured with a MicroTOF (Bruker). High-resolution MS (HRMS) was measured using sodium formate as an external standard. Preparative reversed-phase HPLC was performed using an InertSustain C18 column (4.6 × 150 mm; GL Sciences, Tokyo, Japan) on a PU-2080 system equipped with an MD-2010 detector (JASCO, Tokyo, Japan) or a PU-2087 system equipped with an MD-2010 detector (JASCO). Solvent A: 99% H2O and 1% CH3CN (TFA 0.1%). Solvent B: 99% CH3CN and 1% H2O.
Absorbance spectrum was obtained using a UV-2450 spectrophotometer (Shimadzu, Kyoto, Japan). Fluorescence spectrum was obtained using Fluorescence Spectrophotometer F-7000 (Hitachi, Tokyo, Japan).
Synthesis of Tz–Flu
A mixture of 6-carboxyfluorescein (15 mg, 0.0399 mmol), N,N,N’,N’-tetramethyl-O-(N-succinimidyl)uronium tetrafluoroborate (TSTU) (18 mg, 0.0598 mmol), and N,N-diisopropylethylamine (DIEA) (21 μL) in 2.0 mL dried N,N-dimethylformamide (DMF) was stirred for 10 min at room temperature. Then, (4-(1,2,4,5-tetrazin-3-yl)phenyl)methanamine (11.2 mg, 0.0599 mmol) was added. The reaction mixture was stirred for 20 h at room temperature. The sample was concentrated under reduced pressure to remove DMF. The residue was purified by HPLC to afford the desired product Tz–Flu (2.0 mg, 0.00367 mmol) as a slightly yellow powder in 9.2% yield.
1H-NMR (400 MHz, methanol-d4) δ: 10.30 (s, 1H), 9.33 (t, J = 5.9 Hz, 1H), 8.53–8.51 (m, 2H), 8.21 (dd, J = 8.2, 1.4 Hz, 1H), 8.12 (d, J = 7.8 Hz, 1H), 7.70 (s, 1H), 7.56 (d, J = 8.2 Hz, 2H), 6.71 (d, J = 1.8 Hz, 2H), 6.65 (d, J = 8.2 Hz, 2H), 6.57 (dd, J = 8.5, 2.1 Hz, 2H), 4.63 (d, J = 5.5 Hz, 2H). 13C-NMR (101 MHz, chloroform-d) δ: 177.8, 172.5, 170.7, 170.1, 163.4, 157.9, 147.8, 144.1, 135.0, 133.9, 133.6, 133.1, 132.1, 131.9, 130.2, 128.9, 127.9, 117.8, 106.1, 58.8, 47.1, 43.0. ESI-HRMS m/z: [M + H]+, Calcd for C30H20N5O6+ 546.14136. Found 546.14217 (+0.81 mDa).
Plasmid PreparationThe plasmids were constructed by standard subcloning techniques, including restriction cloning, Gibson assembly, and point mutation.27) Information on the sequences of the plasmids is available in the Supplementary Information.
Cell CultureHEK293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, high glucose with l-glutamate; FUJIFILM Wako Pure Chemical Corporation) supplemented with 10% (v/v) fetal bovine serum (FBS; Biosera) and 1% (v/v) penicillin/streptomycin solution (FUJIFILM Wako Pure Chemical Corporation). Cells were cultured at 37°C in a humidified atmosphere containing 5% CO2.
Establishment of HEK293T Cells Stably Expressing HER2–mCherryThe Sleeping Beauty transposase system was used. HEK293T cells were plated at 2.5 × 105 cells/mL into wells of 12-well plates (in 1 mL of medium) and cultured for 24 h before transfection. A DNA mix containing 50 ng of transposase-encoding plasmid (Addgene, #34879, pCMV(CAT)T7-SB100) and 950 ng of transposon plasmid pNS18 (Supplementary Table 1) was combined with 4 μL of polyethyleneimine “Max” (PEI; Polyscience, Warlington, PA, U.S.A., #23765, 1 mg/mL in dH2O), briefly vortexed and incubated at room temperature for 15 min. Cell culture medium was renewed before transfection, and the DNA/PEI mixture was added dropwise to the culture. After a sufficient cultivation period (typically 16 h), the cell culture medium was renewed. At 48 h after transfection, the cell culture medium was supplemented with 0.5 μg/mL puromycin to start selection. Sufficiently expanded cells were confirmed to express HER2–mCherry by mCherry fluorescence and used for downstream assays.
Cell-Based Assay of Labeled Nanobody In Situ Labeling AssayHEK293T cells were plated at 2.5 × 105 cells/mL into wells of 24-well plates (500 μL medium per well) and cultured for 24 h before transfection. A total of 500 ng of DNA containing a 1 : 1 (weight) mixture of a nanobody-expressing plasmid and Mm-PylRS-AF/Pyl-tRNACUA plasmid (Addgene, #122650) in 50 μL of plain Opti-MEM was mixed with 2 μL of PEI “Max” (Polyscience, #24765, 1 mg/mL in dH2O), briefly vortexed, and incubated at room temperature for 15 min. Before transfection, the cell culture medium was replaced with DMEM supplemented with 10% FBS containing 50 μM TCO-K, and then the DNA/PEI mixture was added dropwise to the culture. After sufficient cultivation (typically 16 h) in medium containing 50 μM TCO-K, the medium was replaced with Advanced DMEM (Gibco), without FBS or TCO-K. The cells were then incubated for an additional 24 h to allow accumulation of TCO-K–bearing nanobodies secreted by the cells.
Subsequently, the cell culture supernatant was collected, and the cells were completely removed by centrifugation (2000 × g, 5 min). Then, Tz–Flu was added to a final concentration of 100 nM. After 1 h of incubation, the mixture—containing labeled nanobodies in the cell culture supernatant along with excess Tz–Flu—was applied to HEK293T cells stably expressing HER2–mCherry. These cells had been seeded one day prior in either an ibidi 96-well plate (for CQ1 imaging) or an ibidi μ-slide 8-well chamber. The cells were then observed using a CellVoyager CQ1 (Yokogawa Electric, Tokyo, Japan) or a Leica SP8 confocal microscope (Leica).
Assay of the Purified ConjugateOne hundred nanomolar of purified nanobody–Tz–Flu conjugate in Opti-MEM was applied to HEK293T cells stably expressing HER2–mCherry. These cells had been seeded one day prior in an ibidi μ-slide 8-well chamber.
Imaging Data Analysis for ScreeningImage analysis was performed on fluorescence images of all cells (acquired in 3 channels: red [C3], blue [C1], and green [C2], with 6 frames per well and 16 z-planes) using Python. Image loading and histogram equalization were performed using the OpenCV library. First, histogram equalization was applied to the red channel images, and pixels with brightness values of 50 or higher were defined as cell regions. Next, the blue/red ratio and green/red ratio were calculated for each pixel, and their respective averages within the cell regions were determined. Furthermore, the product of the blue/red ratio and green/red ratio was calculated, and the maximum values for each well and frame were identified. The program performing all these processes can be viewed at the following URL: https://github.com/TachibanaRyo-moroba/TCO_Nanobody.
Prediction of the Structure of Nanobody–Tz–Flu ConjugateThe crystal structure was used as a template for the protein (PDBID: 5MY6). The dye structure was prepared with GaussView (Version 6.1; Roy Dennington, Todd A. Keith, and John M. Millam, 2016, Semichem Inc., Shawnee Mission, KS, U.S.A.). For the protein structure, missing heavy atoms and hydrogen atoms were added for each amino acid. For each amino acid, except at both ends, the following operations were performed to generate the conjugate structure:
All processes were performed by Python 3. Details of the program and the structures (template protein and dye) can be found in the web repository: https://github.com/TachibanaRyo-moroba/Surface_Screening.
Throughout this paper, we present one possible product of the click reaction: a structure in which the benzene ring attached to the tetrazine and the carbamate group extending from the TCO are positioned in a trans configuration. Although multiple structural isomers are possible, this form is used for illustrative and simulation purposes.
Preparation of Recombinant bdSENP1The recombinant protein was bdSENP1, which was used for cleaving MBP-bdSUMO tag, as described by Frey and Görlich.28) Briefly, the plasmid encoding bdSENP1 (Addgene, #104962) was transformed into NEBstable (NEB), and the resultant colonies (kanamycin+) were incubated at 37°C in Terrific Broth (TB) medium supplemented with kanamycin (50 μg/mL) overnight. This preculture was 100-fold diluted into TB medium and incubated at 37°C. Once the OD reached 0.5, 0.2 mM Isopropyl β-d-1-thiogalactopyranoside (IPTG) (final) was added, and the culture was incubated at 18°C for 24 h. The culture was centrifuged, and the cell pellet was lysed with BugBuster Master Mix (Merck, Burlington, VT, U.S.A.) supplemented with 10 mM dithiothreitol (DTT). The lysate was cleared by centrifuging at 15000 × g for 15 min at 4°C and purified using His SpinTrap (Cytiva) according to the manufacturer’s instructions. After purification, the buffer was exchanged using Amicon Ultra 10K (Millipore, Billerica, MA, U.S.A.) to LS buffer (290 mM NaCl, 45 mM Tris–HCl, pH 7.5, 4.5 mM MgCl2) supplemented with 10 mM DTT. Then, this solution was treated with selecTEV (CosmoBio, Tokyo, Japan) to remove the His-tag fused with SENP1 and incubated at room temperature (r.t.) for 3 h. Then, SENP1 without the His-tag was obtained in the flow-through fraction by using His SpinTrap. The buffer was again exchanged to LS buffer supplemented with 5 mM DTT. The 100 μM recombinant SENP1 was stored at −80°C until use.
Bacterial Expression of TCO-K–Bearing Nanobody and Conjugate PreparationShuffle T7 chemical competent cells (NEB) were doubly transformed with pCDF-MaPylRS-ALIPx2-tRNAx1 (kanamycin+) and plasmids encoding MBP-bdSUMO-2Rs15d with an amber codon (ampicillin+) (position 41 [pNS218] and 84 [pNS211]). The colonies were obtained on an LB plate containing kanamycin and ampicillin. The colonies were inoculated into 2xYT media and incubated at 30°C overnight. This preculture was 100-fold diluted into 2xYT medium containing 1 mM TCO-K. Once the OD reached 0.5, 0.2% arabinose and 0.1 mM IPTG were added (final concentrations). Then, the culture was incubated at 25°C for 8 h. The culture was centrifuged, and the cell pellet was lysed with BugBuster Master Mix. The lysate was cleared by centrifugation and sequentially purified using His SpinTrap and MBP-Spin Protein Miniprep Kit (Zymo Research). Then, the elute containing MBP-bdSUMO-2Rs15d (with TCO-K) was treated with 100 nM bdSENP1 purified as described above. The resultant solution was incubated on ice for 1 h. Then, the 2Rs15d (with TCO-K) was purified by His SpinTrap. To this solution, 10 μM of Tz–Flu was added, and the solution was kept at r.t. for 1 h. The Tz–Flu–nanobody conjugate was purified by using Amicon Ultra 3K (Millipore, Billerica, MA, U.S.A.), and the concentration of the conjugate was measured with Pierce BCA Protein Assay Kits (Thermo Fisher, Waltham, MA, U.S.A.).
With this procedure, the final amount of labeled nanobody from a 1.25- mL culture scale was estimated to be approximately 0.1 nmol (1 μM ×100 μL) for both mutant 41 and 84.
We thank Mr. Sota Nagashima for the help with the synthesis of Tz–Flu, Ms. Winnie Wong for the comments on the manuscript, and the suppliers of Addgene constructs used in this study. This work was supported by Grants-in-Aid for Scientific Research (KAKENHI) (20H02874 and 24K01642 to R.K., and 24H00050 to Y.U.), Toyota Riken Scholarship (to R.K.), Nakatani Foundation Research Grant (to R.K.), Daiichi Sankyo Foundation for Life Science (grant to R.K.), Japan Science and Technology Agency (JST) Mirai Program (JPMJMI24G2 to Y.U.), and JST Moonshot Research and Development Program (JPMJMS2022 to Y.U.). N.S. was supported by the WINGS-LST program of the University of Tokyo.
N.S. conducted most of the experimental work. N.S. and R.K. designed the experiments. N.S., R.K., and R.T. analyzed the data. K.F. synthesized Tz–Flu. T.Y. and S.Y. developed the plasmid of MaPylRS/tRNACUA for TCO-K incorporation in E. coli. R.K. and N.S. co-wrote the manuscript with input from all the authors. R.K. and Y.U. supervised the project.
The authors declare no conflict of interest.
This article contains supplementary materials.
