Biomedical applications of copper-free click chemistry: in vitro, in vivo, and ex vivo

Copper-free click chemistry has resulted in a change of paradigm, showing that artificial chemical reactions can occur on cell surfaces, in cell cytosol, or within the body. It has emerged as a valuable tool in biomedical fields.


Introduction
Click chemistry has been broadly used for chemical reactions that have orthogonality, high yields, and fast kinetic second order reaction rate constants. 1 These kinds of orthogonal reactions are useful for organic synthesis containing multiple steps and various functional groups. Most chemical reactions are not biocompatible because they require toxic catalysts, organic solvents, high temperatures, or high pressures. 2 Consequently, all synthetic steps have previously been performed in asks, on benches, or under fume hoods.
The resulting compounds can only be applied to a living cell or animal aer complete purication. This means that spaces for carrying out articial chemical reactions are completely segregated from biological spaces containing cells or animals. However, click chemistry is available under aqueous conditions with orthogonality, and offers potential for articial chemical reactions Eunha Kim is an assistant professor in the Department of Molecular Science and Technology at Ajou University. His research interests center on the development of new uorescent probes to study structure and photophysical property relationships of uorescent molecules and on expanding that understanding to applications for molecular imaging, molecular diagnosis and chemical biology.
being carried out on cell surfaces, in cell cytosol, or in the body. 3 Particularly, organic chemists have attempted to remove the toxic copper catalyst from the representative click reaction, coppercatalyzed [3 + 2] azide-alkyne cycloaddition (CuAAC). Their trials resulted in 'copper-free' click chemistry which is highly attractive to biological or biomedical researchers. [4][5][6] Initially, Staudinger ligation between azide and phosphine groups was the rst chemical reaction thought to be bioorthogonal that could occur under aqueous conditions without use of a toxic catalyst. 7 A few previous studies have introduced the application of this reaction on the cell surface or in vivo. 11 However, this reaction is too slow to address researchers' requirements (the second order reaction rate constant was reported at approximately 7.7 Â 10 À3 M À1 s À1 ). 12 Therefore, this reaction could not be considered as click chemistry.
K. Barry Sharpless rst introduced the concept of click chemistry in 1998 when considering other reactions, and following this CuAAC has gained much attention. 1 The second order reaction rate constant of CuAAC has been reported to be approximately 10 M À1 s À1 with 20 mM Cu(I), which is about 1000-fold faster than Staudinger ligation (Table 1). 13 Another advantage of CuAAC is the small size of the functional groups involved in the reaction. Azide has only three nitrogen atoms while alkyne has one hydrogen and two carbons. Such a small size is helpful for minimizing perturbation of intrinsic characteristics of target molecules aer modi-cation. However, the cytotoxic copper catalyst restricts the in vitro or in vivo application of CuAAC reactions.
To reduce the cytotoxicity of the copper catalyst in CuAAC, researchers tried to use stabilizing ligands. In 2010, Amo et al. used bis(tert-butyltriazoly) ligand which is water-soluble and effective in promoting CuAAC. 14 Cells were alive aer CuAAC using the ligand and it was applied to in vivo imaging of zebrash embryo. Bevilacqua et al. and Kennedy et al. also demonstrated cell labelling by CuAAC using similar bis(tert-butyltriazoly) ligand and Cu(II)-bis-L-histidine complex, respectively. 15,16 To overcome this limitation fundamentally and increase the convenience, other chemists have increased the reactivity of alkynes using ringstrain which enables an azide-alkyne reaction without the need for a cytotoxic copper catalyst. These strain-promoted azidealkyne cycloaddition reactions (SPAAC) have favourable second order reaction rate constants (approximately 0.1 M À1 s À1 ) under aqueous conditions without a catalyst. 17 Dibenzocyclooctyne (DBCO) and bicyclo[6.1.0]nonyne (BCN) are representative of strained cyclic ring-containing alkynes for SPAAC. This copper-free click chemistry has combined chemical reactions with cell/animal studies. 18 On the other hand, some chemists are not satised with the second order reaction rate constant of SPAAC and have attempted to nd click reactions with second order reaction rate constants. They have found that inverse-electron-demand Diels-Alder (iEDDA) reactions between tetrazine (Tz) and trans-cyclooctene (TCO) could meet the requirements of click chemistry. This third-generation copperfree click chemistry provides a faster second order reaction rate constant (>10 3 M À1 s À1 ) than CuAAC and SPAAC. Thus, it is considered a powerful tool for many purposes, further widening the eld of click chemistry. 19 The disadvantage of Tz-TCO click chemistry is that both molecules are somewhat larger than an azide or a normal alkyne. This large size can limit their application, as large click molecules can affect the intrinsic properties of target molecules aer modication of the original molecules. Therefore, researchers need to consider the characteristics of click chemistry reactions carefully and select the most suitable reactions according to their intended purpose. -One small and one large structure -Moderate second order reaction rate constant -No catalyst iEDDA $1-10 6 (M À1 s À1 ) 10 -Two large structures -Very high second order reaction rate constant -No catalyst In particular, two different copper-free click chemistry reactions, SPAAC and iEDDA, allow bioorthogonal chemical conjugation in living organisms (on cell surfaces, in cell cytosol, or within the body). A number of reports have shown broad applications of click chemistry in the facile labelling of nucleotides or proteins, discovering unknown metabolic pathways, articial adhesion of cells, spatiotemporal monitoring of newly grown tissues, surface-modication of nanoparticles, nanoparticle delivery, synthesis of hydrogels, and so on. [20][21][22][23][24][25][26] These reports demonstrate that click chemistry is now an attractive technique in biomedical elds including imaging, drug delivery, or diagnostic analysis, as it is in organic chemistry. 27,28 However, there are only a few papers focusing on recent applications of click chemistry in biomedical elds. 29 In this review, we introduce recent research in which copperfree click chemistry is successfully applied in biomedical elds including imaging, drug delivery, and diagnostic analysis. We categorize this research based on the site at which the click reaction actually occurs, namely in vitro, in vivo, and ex vivo ( Table 2). We conclude the review by highlighting precautions Table 2 Summary of the copper-free click chemistry research for biomedical applications

Location Application
Click molecules used Type In vitro PARP1 imaging 35 AZD2281-TCO/Tz-Texas Red iEDDA PARP1 target identication study 36 AZD2281-TCO/Tz-cleavable linker-biotin iEDDA Tubulin protein imaging 37 Taxol-TCO/Tz-BODIPY FL iEDDA Polo-like kinase 1 protein imaging 38 BI2536-TCO/Tz-Texas Red iEDDA Aurora kinase A protein imaging 39 MLN8054-TCO/Tz-CFDA, Tz-Texas Red iEDDA MET protein imaging 40 PF04217903-TCO/Tz-CFDA iEDDA Foretinib-TCO/Tz-CFDA iEDDA Imaging binding targets of drug molecules 41 Dasatinib-TCO/Tz-CFDA iEDDA BTK protein imaging 42 Ibrutinib-TCO/Tz-Cy5 iEDDA, SPACC Ibrutinib-N 3 /DBCO-Cy5 Drug imaging 43 JQ1-TCO/Tz-Cy5 iEDDA Visualization of PTM modication 44 NAD + -DBCO/N 3 -Cy3, N 3 -Cy5 SPACC NAD + -methyl cyclopropene/Tz-Cy3 iEDDA Site-specic oligonucleotide labelling [45][46][47] RNA-norbornene/Tz-Oregon Green 488 iEDDA RNA-methyl cyclopropene/Tz-Oregon Green 488 DNA-Tz/BCN-rhodamine DNA-methyl cyclopropene/Tz-BODIPY Site-specic protein labelling 48 L-Lysine-TCO/Tz-Cy5 iEDDA No-wash STED imaging 49 Phalloidin-BCN/Tz-Cy3 iEDDA In vivo Zebrash embryo growth imaging 24 N-Acetylgalactosamine ( Unnatural amino acid containing tetrazine or norbornene/BODIPY-FL tetrazine or BODIPY-TMR-X bicyclononyne iEDDA, SPACC regarding click chemistry for biomedical purposes and the prospect of this technique in the future. Particularly, we will focus on copper-free click reactions occurring on a cell surface, in cell cytosol, in a body, or at least with proteins or nucleotides, emphasizing the advantage of copper-free click chemistry considering the environment in which the reaction occurs. Click chemistry also contributes to the development of drugs or materials for biomedical applications. However, we did not cover these kinds of applications of click chemistry outside of cells or animals in the present paper because they have been summarized well in other reviews. 30,31 In addition, 'photoclick' reactions between electron-decient olens and diaryltetrazoles are also an attractive tool with special advantages. 32 These reactions can be triggered by light of a particular wavelength, which means that two molecules are conjugated to each other under spatiotemporal control depending on the purpose. Furthermore, the nitrilimine produced by photoexcitation of diaryltetrazole is highly reactive, so that the reaction rate is very fast (k $ 60 M À1 s À1 ). 33 A recent study of the Wagenknecht group used this reaction for facile labelling of DNA using cyanine dye showing the utility of the photoclick reaction. 34 However, the wavelength of light used in photoclick reactions is generally in the ultraviolet (UV) range, which is a limitation in biomedical research. It is well-known that UV light is cytotoxic, so it is not suitable for cell studies. In addition, its short wavelength resulted in poor tissue penetration for application in vivo. For these reasons, it has been applied mainly for research outside of cells or animals. If a new photoclick reaction using visible or NIR light is proposed in the future, it will be more useful for biomedical applications.

Clickable drug surrogates
One of the most interesting applications of copper-free click chemistry might be the uorescence imaging of target of interest (TOI) proteins inside cells. The small molecular size of the tag for click chemistry means that a known ligand could be modied with the tag for labelling TOI proteins with minimized reduction of their original binding while retaining their cell permeability. Especially with iEDDA reactions, innate TOI proteins in live cells could be successfully visualized with a TCO-ligand conjugate and subsequent treatment of Tz containing uorophores (FL Tz ).
A proof of concept study for this strategy was reported by the Weissleder group in 2010. 35 In that study, they conjugated the clinical drug AZD2281 with TCO to develop a bio-probe for studying poly(ADP-ribose) polymerase-1 (PARP1) proteins which are known to be important cellular proteins for DNA repair (Fig. 1). They found that TCO modication of AZD2281 resulted in marginal IC 50 change from 5 nM to 11 nM. In addition, they found that the AZD2281-TCO conjugate was localized to the target PARP protein in the cell with Texas Red-Tz. This localization was inhibited by the original AZD2281 compound. Similar results were obtained for other clinical drugs. For example, they conjugated TCO to the well-known anticancer agent, Taxol, and successfully visualized tubulin proteins inside cells with a Taxol-TCO/Tz-BODIPY FL combination. 37 Aer successful demonstrations, multiple ligand-TCO conjugates such as BI2536, 38 MLN8054, 39 PF04217903, 40 Foretinib, 40 Dasatinib, 41 and Ibrutinib 42 have been developed for targeting various TOI proteins such as polo-like kinase 1 (PLK1), Aurora kinase A (AURKA), MET, ABL1, SRC, CSK and Bruton's tyrosine kinase (BTK) proteins.
Click chemistry is useful not only for imaging cellular proteins but also for studying drug target engagement. Many drug candidates oen fail to be considered clinically viable due to undesirable interactions with multiple different side targets. To better understand actual target proteins and pharmacokinetics in vitro, click chemistry has been highlighted as a powerful chemical tool for target identication and drugtarget engagement studies in in vitro systems. For example, the Weissleder group has reported on bioorthogonal proteomics using copper-free click chemistry. 36 They modied Olaparib with TCO for target identication of the drug. In addition, they further developed a cleavable enrichment linker containing Tz (for click chemistry), biotin (for pull down assay), and 2-(4 0hydroxy-2 0 -alkoxy phenylazo)benzoic acid (as a cleavable site). They rst conrmed the activity of Olaparib-TCO against recombinant PARP1 proteins and conrmed that Olaparib-TCO still had a nano-molar range of IC 50 . Later, MHH-ES1 Ewing's sarcoma cells and A2780 ovarian cancer cells were treated with Olaparib-TCO. Proteins labeled with TCO drugs were then pulled down by two-step bioorthogonal magnetic separation. To release proteins from magnetic beads, sodium dithionite was used to cleave the linker that allowed the specic release of small-molecule captured proteins while leaving non-specically bound proteins on the solid support. Interestingly, LC/MS-MS data revealed a list of 24 different proteins including PARP1 proteins. During a follow-up in-depth study, they identied the protein TOP2A as another binding partner of Olaparib with an estimated K d of 3.7 nM.
A recent mechanism study of BET bromodomain inhibitors and epigenetic-based therapy has highlighted the potential of click chemistry for pre-clinical assessment of various drugs (Fig. 2). 43 In that study, the authors synthesized derivatives of BET inhibitors such as JQ1 and IBET-762 compounds with propargyl and TCO for copper-catalyzed and iEDDA type click chemistry. They rst conrmed that modication of the original drug molecules with a click chemistry functional group did not inuence the functional integrity of BET inhibitors using cellbased assay, ChIP-qPCR assay, and RNA-Seq. Subsequently, they revealed, with JQ1-TCO, differences in the binding modes of BRD4 between BET inhibitor responsive and unresponsive genes. They found that bromodomain 1 (BD1) could mediate the binding of BRD4 proteins to acetyl-lysine sites and that bromodomain 2 (BD2) was the binding site of clickable drug surrogates. Furthermore, they evaluated the pharmacokinetics of the parent drug with clickable drug surrogates and conrmed the heterogeneity of drug distribution between organs as well as between different cell types in the same organ. These examples highlight the power of copper-free click chemistry for studying actual targets, pharmacokinetics (PK), and pharmacodynamics (PD) of drug molecules.
Click chemistry can be used to address long-standing mechanistic questions in the clinic. For example, the Nathan group reported the traceable mimic of the anti-cancer drug cytarabine (ara-C) by converting a single hydroxyl group to azide, giving AzC. They conrmed the equal biological prole of AzC in cell culture and in vivo as ara-C. Further studies of the AzC with classical azide-alkyne click chemistry, including the gSTED imaging experiment, allowed them to understand the contradiction of cell-type selectivity of nucleoside-based drugs. 65

Click chemistry for labelling cellular membrane lipids and proteins
To avoid issues with Tz or TCO such as metabolic stability, cross-reactivity with biological nucleophiles, and impact on protein structure, activity, or localization, tags with smaller sizes than to conventional iEDDA tags have been suggested. For example, Oliveira et al. reported that unstrained S-allyl could be used as the tag for iEDDA type bioorthogonal labelling under live cell conditions (Fig. 3). 66 They rst surveyed the reaction kinetics of S-allyl cysteine with PyTz, BnNH 2 -Tz, Tz-Rhod, and Tz-Cy3 in a PBS (pH 7.4) and methanol mixture (v/v ¼ 1 : 1) at 37 C. They found that S-allyl Cys had reasonable reaction kinetics for iEDDA reactions with screened Tzs (k 2 from 2.05 Â 10 À3 to 0.26 Â 10 À3 M À1 s À1 ), which was comparable to SPAAC reaction kinetics (10 À1 k 2 ¼ 10 À3 M À1 s À1 ). In addition, this chemistry allowed them to achieve specic recognition of apoptotic cells via [2,3]-sigmatropic rearrangement of allyl selenocyanate with phosphatidyl serine (PS) membrane lipids on the cellular surface. With Annexin V and engineered variants of the C2A domain of synaptotagmin-I (C2Am), they conrmed more efficient S-allylation of Cys via [2,3]-sigmatropic rearrangement than with direct allylation. Due to the uorogenic properties of Tz-Cy3, they successfully imaged apoptotic cells without the washing step.
To develop light-gated ion channels with longer excitation wavelengths, simple experimental setup, and efficient activation, the Xing group has incorporated NIR-sensitive-UCNs (upconversion nanocrystals) with light-gated channelrhodopsin-2 proteins (ChR2) via a copper-free click reaction between the DBCO (dibenzyl cyclooctene) moiety on UCNs and an azido tag on membrane proteins. 67 Firstly, they conrmed the azide group labelling of membrane glycans in HEK293 (human embryonic kidney 293) cells based on metabolic glycoengineering by feeding tetraacetylated N-azidoacetyl-D-mannosamine (Ac 4 ManNAz), following covalent labelling with a DBCO-Cy3 uorophore. Later, ChR2 proteins were engineered to be expressed on the cell surface and their cellular location was conrmed with GFP markers on the proteins. UCNs doped with Nd 3+ were functionalized with polyacrylic acid (PAA) while DBCO moieties were coupled with a carboxylic group in PAA.  Coating of DBCO on the surface of UCNs was conrmed with click reactions using FAM-N 3 (5-carboxyuorescein-azide). Then, they conrmed specic labelling of ChR2 proteins on the HEK293 cell membrane with UCNs using copper-free click chemistry. Inux of Ca 2+ ions into cells was triggered by 808 nm excitation and Ca 2+ -dependent apoptosis was conrmed using Cy3-tagged Annexin V. Finally, they implanted N 3 -tagged HEK292 cells expressing ChR2 proteins into the yolk sac of zebrash and incubated the larvae with DBCO/Cy5.5-UCNs. They demonstrated successful manipulation of Ca 2+ inux in vivo using copper-free click chemistry-mediated site-specic labelling and light activation of Ca 2+ channel proteins.
Non-canonical amino acids (ncAAs) approach is one of the most versatile methods for labeling in a residue specic fashion for the TOI protein and recent advances in the techniques allowed site-specic incorporation of strained alkene and alkyne to the TOI protein. In 2016, Lemke group reported in cell protein labeling utilizing cyclooctyne and trans-cyclooctene amino acid by utilizing the iEDDA reaction with 1,2,4,5-tetrazine. 48 First, they found NLS (nuclear localization signal) on M. mazei PylRS protein sequence. They found nuclear localization of PylRSAF protein in HEK293T and COS-7 cells with IF (immunouorescence) staining and FISH (uorescence in situ hybridization) study. By cytoplasmic localization reinforcement of the protein via addition of a nuclear export signal (NES), they improved the efficiency of amber suppression with the engineered protein (NESPylRSAF). With transcription factor jun-B, they conrmed that NESPylRSAF successfully incorporated ({[(E)-cyclooct-2-ene-1-yl]oxy}carbonyl)-L-lysine (TCO*a) with a better signal to noise ratio than PylRSAF. They further applied the system for super resolution microscopy (SRM) with a "Click-PAINT" approach. Basically, rst the TOITAG / TCO*a protein was equipped with Tz-ssDNA (the "docking strand"), then it was labeled with a complementary ssDNA (the "imaging strand") containing a photostable synthetic dye. They applied this technique to image Nup153 protein consisting of a nuclear pore complex (NPC) and successfully visualized the detailed structure of NPC. This study highlights that the combination of the iEDDA reaction with other biotechniques, such as DNA-PAINT, could provide more powerful uorescence imaging techniques for visualizing TOI protein to allow a more in-depth study of biological systems.
A recent study by Marx group highlights the potential of copper-free click chemistry not only for simple labeling of the protein but also for the visualization of post-translational modication (PTM) and cellular signaling in the cells. 44 In this study, they developed functionalized NAD + analogues to study poly(ADP-ribos)ylation (PARylation) in the cells as a traceable substrate for ARTD1, one of the most well studied ADP-ribosyltransferases (ARTs). By modifying adenosine with a terminal alkyne, terminal alkene, azide, DBCO and cyclopropene they developed 10 different clickable probes as substrates for ARTD1. The screening result conrmed that 2-modied NADs have good tolerance against ARTD1. Finally, they achieved bioorthogonal labeling of PAR in multiple uorescence channels via SPAAC and iEDDA reactions in the cells, incubated with two different functionalized NAD + analogues, between the analogues and clickable uorescent dye such as Cy5-N 3 or Cy3-Tz, respectively.

Cell adhesion
Recent studies on CAR T cell therapy and the effects of macrophages on tumor microenvironment have indicated that cellcell interactions are emerging as some of the most promising therapeutic strategies. 68,69 Articial cell adhesion could provide a valuable chemical tool box for therapeutic applications. Multiple attempts using nucleotides, avidin-biotin interactions, antibody dimers, and aptamers have been performed under in vitro conditions. 23,70,71 However, due to the possible degradation of the adhesive for cell-cell interactions by enzymes, such as proteases, DNases, and RNases, in vivo conditions are a drawback of complementary macromolecules. 72 Therefore, bioorthogonal small molecules could provide alternative approaches for articial cell adhesion with more stable, robust, multivalent, and economical chemical bonds.
Koo et al. have studied TCO/Tz mediated cell adhesion (Fig. 4). 73 Cells including A549 human lung cancer, Jurkat T cell lymphoma, NIH3T3 murine broblasts, and EL4 lymphoma cells were rst tagged with an azido group using metabolic glycoengineering with Ac 4 ManNAz. Subsequently, Tz and TCO were incorporated onto azido-modied sialic acid by Tz-DBCO or TCO-DBCO treatment. Aer TCO modied Jurkat T cells were added to layers of TCO modied A549 cells in a micro-uidic setting, it was conrmed that articial adhesion between A549 and Jurkat T cells was completed within 10 min. In addition, adhered cells displayed stable interactions in microuidic channels even under a ow rate of 60 mL min À1 with a calculated shear stress of 20 dyn cm 2 (2 Pa). Aer that, adhered cell pairs were introduced to wild-type mice by retroorbital injection. Adhered cells were present in murine blood vessels, they were found using two-photon microscopy. In addition, intravenously injected cell pairs were localized to lung tissues. They were trapped in lung capillary beds while maintaining their attachment, even under in vivo conditions. A recent study by the same group has demonstrated that adhered cells could be detached by chemical stimuli. 74 They rst introduced an azido group onto Jurkat and A549 cells via metabolic glycoengineering using Ac 4 ManNAz. Later, azide-modied A549 and Jurkat cells reacted with DBCO-disulde-Tz and DBCO-disulde-TCO, respectively. They conrmed that treatment with 5 mM glutathione reduced the ratio of attached Jurkat cells onto A549 cells down to 10% due to cleavage of disulde bonds. 75

Tetrazine-based turn-on probe
One of the most interesting features of Tz, especially for uorescence imaging, is the dual functionality of the compound. It can act as a bioorthogonal reaction unit and as a uorescence quencher simultaneously. 37,76-79 Therefore, uorophores conjugated with Tz could be used as bioorthogonal reaction turn-on probes for washing-free uorescent imaging. Starting with the BODIPY FL uorophore, multiple different uorophore-Tz conjugates (FL Tz ) have been developed. Most of the quenching mechanisms of FL Tz are based on energy transfer, which is basically the transfer of energy from excited uorophores to Tz units. The energy is released in a non-radiative manner. Two different energy transfer mechanisms are known. One type is based on long-range dipole-dipole interactions and is known as Förster resonance energy transfer (FRET). The other is a through-bond energy transfer (TBET) mechanism induced by making the FL Tz a conjugated, but not coplanar, pi-system (resulting in a twist in the inter-chromophore linkage). The rst usage of FL Tz for bioorthogonal turn-on probes was reported with the FRET type mechanism. 37 In that study, the authors conjugated 3-(4-benzylamino)-1,2,4,5-tetrazine with BODIPY FL, Oregon Green 488, and BODIPY TMR-X. All FL Tz s in that study exhibited turn-on effects aer reaction with TCO. The turn-on efficiency increased from 1-fold to 20-fold. Soon aer, the TBET based strategy was reported. 80 Instead of a simple coupling between Tz and uorophores, the authors directly connected Tz with the BODIPY uorophores. Using this strategy, the authors developed a bioorthogonal turn-on probe exhibiting increased uorescence up to 1600-fold. They also successfully applied these probes to washing-free uorescence imaging of EGFR proteins with TCO modied antibodies.
Hitherto, coumarin, BODIPY, uorescein, rhodamine, phenoxazine and silicon rhodamine uorophores have been utilized to develop FL Tz bioorthogonal turn-on probes. 81-85 A drawback of previous energy transfer type quenching mechanisms is the reduced quenching efficiency against more red-shied uorophores. The energy accepting efficiency of the energy acceptor Tz is approximately up to 500 nm, therefore uorophores with longer emission wavelengths exhibit a lesser turn-on effect. To overcome limitations with previous approaches, a monochromophoric design strategy was recently reported to develop bioorthogonal turn-on probes regardless of their color (Fig. 5). 86 In that study, Lee et al. introduced a new molecular design approach for full integration between Tz and a model uorophore system. They conrmed that uorophores under the new molecular design approach exhibited increased turn-on efficiency by 600-to 1000-fold, regardless of the emission wavelength of the uorophore.
The uorogenic FL Tz probe is also useful for super resolution microscopy (SRM) imaging. 79,[87][88][89][90] In 2018 the Kele group reported uorogenic cyanine-tetrazine conjugates for no-wash subsequent super-resolution microscopy (STED) imaging of an actin lament. To quench the cyanine dye via the TBET strategy along with bioorthogonality, they designed 4 different tetrazinecyanine dyes utilizing either a phenylene or a vinylene linker between tetrazine and the uorophore. They found that all four probes showed uorogenic behavior but found 14 fold uorescence enhancement by conjugating Tz on Cy3 with the vinylene linker and the other 3 different probes exhibited 3 fold uorescence enhancement in average. With phalloidin-BCN, they successfully applied the probe for STED imaging of actin laments with 168 nm resolution using a 660 nm continuous wave laser for depletion. 49   amounts of different molecules including ions, small chemicals, nucleotides, and proteins. This means that bioorthogonality is required for reactions in vivo. Furthermore, the concentration and contact time of click molecules are limited in vivo, and the second order reaction rate constant is also important.

Click chemistry in vivo
In 2008, the Bertozzi group introduced an excellent application of click chemistry in vivo for spatiotemporal imaging of zebrash embryo growth. 24 In that study, the authors treated zebrash embryos with N-acetylgalactosamine (GalNAc) to modify glycan with azide groups by metabolic glycoengineering. They then applied diuorinated cyclooctyne (DIFO)-Alexa Fluor dye that could be conjugated with azide groups by SPAAC. They repeated this process with similar dye conjugates of different colors at predetermined time points. Finally, they obtained adult zebrash labeled with multiple colors of uorescence. These labeled colors demonstrated the time points at which different sites developed during growth. These images effectively provided spatiotemporal information of zebrash growth.
In 2010, the same authors further demonstrated that SPAAC in vivo could be applied to mice. 50 They injected Ac 4 ManNAz into mice by intraperitoneal injection for azide group incorporation by metabolic glycoengineering. They then injected various molecules that could bind azide groups similarly. Based on the analysis of excised splenocytes from the mice, they determined the conjugation efficacy between azide groups and different chemicals. Their results showed that dimethoxy azacyclooctyne (DIMAC) had the highest efficacy among the ringstrained alkynes used. They also found that the hydrophobicity of chemicals induced aggregation with serum proteins resulting in non-specic binding to other tissues. Interestingly, in their paper, Staudinger ligation was efficient, suggesting that it was even better than ring-strained alkynes. The intraperitoneal space represents an environment with special conditions. It is without active in-or out-ow compared to intravenous (i.v.) injections, which is helpful for maintaining sufficient concentrations of reactive molecules while minimizing the effects of reaction rate between them.
Encouraged by these promising results regarding click chemistry in vivo, researchers have attempted to utilize click chemistry to obtain biological or biomedical information from mice. Recently, Xie et al. introduced in vivo labelling of brain sialoglycans using liposomes (Fig. 6). 51 9-Azido sialic acid, a metabolic precursor used in their paper, cannot cross the blood-brain-barrier (BBB), thus they used a liposome carrier to deliver the molecules to the brain tissue. Aer i.v. injection of 9azido sialic acid-loaded-liposomes, the molecule could reach the brain and participate in brain metabolic glycoengineering. As a result, the newly synthesized sialic acid in the brain tissue could be modied with azide groups, which could be further labeled aer i.v. injection of DBCO-Cy5.5 by SPAAC in vivo. Even though uorescence imaging has many advantages, including easy handling and high resolution, its application in medical imaging is restricted due to its short penetration depth. 91 Therefore, other imaging modalities including PET/SPECT, MRI, and ultrasound are, preferred in clinical settings, researchers have also tried to apply click chemistry in vivo with these imaging techniques. In 2010, Rossin et al. applied click chemistry in vivo to SPECT imaging for tumor imaging in mice. 52 They modied anti-TAG72 monoclonal antibody CC49 with TCO, and synthesized a Tz-conjugated chelator containing 111 In as a counterpart for SPECT imaging. They rst injected a TCOantibody to a colon cancer-bearing mouse via i.v. administration, and also injected Tz-111 In similarly the next day. They obtained 4.2% of 111 In accumulated in tumor tissues with a very high tumor-to muscle-ratio (13.1 : 1). Such two step strategies have been called 'pretargeting', which uses modied antibodies and their counterparts with imaging molecules or drugs. 28 Pretargeting has certain advantages, including that more than one click molecule can be modied to one antibody, which increases the number of potential binding chances between one receptor and introduced imaging probes. Furthermore, in the case of imaging, pretargeting can provide sufficient time for circulation and accumulation of large antibodies while small imaging probes can already distribute rapidly. 92 It means that unbound imaging molecules can be cleared faster than when they are linked to large antibodies. This can result in shortened waiting times between injection of probes and real imaging. Rossin et al. have obtained ne SPECT imaging of tumors using a fast Diels-Alder reaction in vivo, and this represents a good example of pretargeting in vivo. 52 Recently, Neves et al. have applied click chemistry in vivo to MR imaging (Fig. 7). 53 They introduced Ac 4 GalNAz to LL2 tumor-bearing mice by intraperitoneal injection, which labeled glycans in the mouse tissues with azide groups by metabolic glycoengineering. These azide groups could be further labeled with i.v. injected TMDIBO-Lys-Gd (tetramethoxydibenzocyclooctyne-lysine-gadolinium) by SPAAC in vivo. Because gadolinium can change T1 relaxation rates and generate an MR signal, increased accumulation of TMDIBO-Lys-Gd could be observed by MR. Signicant changes in T1 relaxivity were observed in most tissues including tumor tissues in the case of Ac 4 GalNAz-treated mice showing vigorous glycosylation in vivo. These results showed that the synthesized gadoliniumlabeled click probe could be used for glycosylation imaging in vivo by MR. MR imaging has superior spatial resolution and is widely used in clinical settings, these kinds of approaches for further MR imaging with click chemistry in vivo have great potential.
Zlitni et al. have introduced a targeting method based on click chemistry in vivo for ultrasound imaging. 54 Similar to the study by Rossin et al., they used TCO-modied antivascular endothelial growth factor receptor 2 antibody (TCO-anti-VEGFR2). VEGFR2 is overexpressed in tumor cells, thus TCO-antiVEGFR2 can bind SKOV-3 human adenocarcinoma tumor tissues in mice aer i.v. injection. Aer that, they injected Tz-modied microbubbles lled with gas for ultrasound contrast enhancement to mice via tail veins. These bubbles could bind TCO-antiVEGFR2 to tumor tissues by click chemistry in vivo, generating ultrasound signals in tumor tissues. Quantitative data showed that these pretargeting methods provided Fig. 6 In vivo fluorescence imaging of sialoglycans in mouse brain by copper-free click chemistry. (A) Azide group labelling of sialoglycans in mouse brain by 9-azido sialic acid-loaded liposome and metabolic glycoengineering. Liposomes were injected intravenously to mice and delivered 9-azido sialic acid to brain cells across BBB. Then, azide groups were generated in brain cells by metabolic glycoengineering. (B) SPAAC in vivo between DBCO-Cy5 and sialoglycans in mouse brain. Intravenously injected DBCO-Cy5 could label newly synthesized brain sialoglycans. (C) In vivo fluorescence images showing sialoglycans targeted by SPAAC. 9-Azido sialic acid-loaded liposomes were administered daily to mice for 7 days and DBCO-Cy5 was injected at day 8. (D) Brain signal-to-background ratio (BBR) of (C). **P < 0.01; n.s., not significant (one-way ANOVA). Reproduced from ref. 51 with permission from the National Academy of Sciences of the United States of America (NAS), copyright 2016. approximately 4-fold increased signals in tumor tissues compared to non-targeted microbubbles. It showed an approximately 40% increase in value compared to that of the control microbubbles directly modied with antibodies. This type of approach using bioorthogonally modied microbubbles and pretargeting is useful to increase targeting efficacy. Using a similar strategy, Wang et al. performed successful ultrasound imaging of acute thrombus in rats. 55 The results of the studies discussed are promising and demonstrate that click chemistry has great potential in biomedical imaging in vivo. However, strategies and the click molecules used need to be selected carefully because required doses of probes and reaction times will vary according to the types of imaging modalities and disease models.

In vivo drug delivery
Click chemistry in vivo is also attractive to researchers for therapy. In therapeutic applications, click chemistry was rst used for synthesis or modication of drug carriers. The fast second order reaction rate constant, simplicity, and orthogonality of click chemistry are useful for polymer synthesis or site-specic modication of biological ligands during development of drug carriers. 6 In 2012, Colombo et al. used SPAAC to conjugate ScFv antibodies to nanoparticles specically. The Boons group used similar reactions during development of their antibody-drug conjugate to minimize the decrease of antibody binding affinity aer modication. 22,93 However, in vivo applications began to be used later because there was not a sufficient amount of data to be condent in the potential of click chemistry in vivo. Particularly, it is more delayed in drug delivery and therapy compared to imaging, because the required amount of molecules to be delivered for therapy is generally larger than that for imaging. Therefore, research using click chemistry in vivo for drug delivery and therapy is relatively unexplored, and more information is needed.
In 2012, Koo et al. demonstrated that movement and distribution of nanoparticles could be changed by click chemistry in vivo. 25 They directly injected the metabolic precursor Ac 4 ManNAz into tumor tissues of mice. Aer the generation of azide groups on tumor cells by metabolic glycoengineering, liposomes modied with DBCO groups were injected into mice intravenously. They showed that the accumulation of DBCO-modied liposomes was signicantly increased by SPAAC in vivo between azide and DBCO. It is evident that the azide-DBCO reaction is slower than the Tz-TCO reaction. Some researchers have pointed out that it is too slow for pretargeting in vivo. However, in their paper, the amount of azide groups on tumor tissues might be larger than that of specic receptors for biological ligands, which is helpful for reactions in vivo. In addition, nanoparticles have longer circulation times than small molecules aer i.v. injection. Thus, they could provide sufficient contact time among click molecules.
Their work provides evidence, for the rst time, that click chemistry in vivo could be used for nanoparticle delivery. This opened new avenues for these kinds of approaches for further research. However, the main disadvantage of this project was that the metabolic precursor was directly injected into the tumor tissues for pretreatment. In real clinical situations, drug delivery is frequently used to kill tumor cells of unknown location and oen when intratumoral injection is impossible. 94 To avoid the need for intratumoral injection, Lee et al. (the same research group) have developed an alternative strategy with i.v. injections (Fig. 8). 27 The metabolic precursor Ac 4 ManNAz was encapsulated into glycol chitosan nanoparticles and injected intravenously into the mice. Subsequently, it was delivered to the tumor tissue by the EPR (enhanced permeability and retention) effect based on the passage of nanoparticles through fenestrated tumor vessels. The high accumulation of Ac 4 -ManNAz could induce the generation of azide groups on the tumor tissues by metabolic glycoengineering. Furthermore, second nanoparticles containing drugs and BCN were again injected intravenously. They were delivered to the tumor tissue efficiently by SPAAC in vivo compared to bare nanoparticles or without rst injection of Ac 4 ManNAz-loaded nanoparticles. Enhanced drug delivery by this two-step strategy based on metabolic glycoengineering and click chemistry in vivo resulted in effective tumor therapy in mice.
The Cheng group is also interested in using click chemistry in vivo for drug delivery. 56 They achieved selective metabolic glycoengineering of cancer cells by modifying Ac 4 ManNAz with N,N 0 -L-diacetyllysine that could be cleaved by histone deacetylase and cathepsin L (Fig. 9). These enzymes are overexpressed in cancer cells, so that the modied precursor can be selectively activated aer cellular uptake, introducing azide groups on the surface of cancer cells by metabolic glycoengineering. They showed that large amounts of azide groups were introduced by i.v. injection of the modied precursor into nude mice bearing subcutaneous LS174T tumors (primary colon tumors). In contrast with the study by Lee et al., using nanoparticles, they directly modied doxorubicin, a representative anticancer drug. The amine groups of doxorubicin were modied with DBCO and dipeptide (Val and Cit) using a self-immolative p-aminobenzylcarbamate crosslinker. Aer i.v. injection of the modied doxorubicin, DBCO groups can be conjugated with azide groups on the surface of cancer cells by SPAAC in vivo resulting in increased accumulation in the tumor tissues.
Aer cellular uptake, free form doxorubicin was generated to kill cancer cells through cleavage of the dipeptide by cathepsin B. The apoptosis index in the tumor tissue was found to be 33.5% using this strategy. It was much higher than the 18.3% found in the case without metabolic glycoengineering, thus improving tumor targeting by SPAAC in vivo. The group also showed the improved anticancer efficacy of this strategy in MDA-MB-231 triple negative breast tumor and 4T1 lung metastases models. Particularly, doxorubicin was modied and selectively activated in tumor tissues resulting in lowered toxicity to bone marrow and spleen. These overall studies contain new challenges and promising data demonstrating that click chemistry in vivo is useful for drug delivery. Better applications are expected in the future.
In 2019, Feng et al. introduced an assembly of upconversion nano-photosensitizers based on click chemistry in vivo. 57 This system is composed of upconversion NPs modied with Tz groups (UCNP-Tz) and a norborane (NB)-conjugated photosensitizer, rose bengal (RB). The UCNP-Tz was further modied with folate (FA) to target the receptors on tumor cells. They injected the UCNP-Tz-FA into MCF-7 tumor-bearing mice, and the particles accumulated in tumor tissue by the EPR effect and binding to FA receptors. Aer that, RB-NB was injected into tumor tissue directly. Because the excitation wavelength of RB is 530-560 nm, it is not reactive upon laser irradiation with wavelength in the NIR region. 95 By iEDDA between Tz and NB in tumor tissue, RB-NB can be conjugated to the surface of UCNP-Tz-FA, and it can be stimulated by energy transfer from UCNP irradiated by laser with 980 nm. The in vitro and in vivo data showed click chemistry-triggered energy transfer and the following regeneration of the photodynamic effect for therapy. The authors insisted that this 'off'-'on' system can minimize the unintended phototoxicity during photodynamic therapy. This study is rationally designed and provides new potential of the click chemistry in vivo, but intratumoral injection of RB-NB may limit wide application similar to the study of Koo et al. as mentioned above. 25 A recent study of Li et al. showed application of click chemistry in vivo to cardiac cell therapy. 58 To cure myocardiac infarction (MI) by cell therapy, large numbers of stem cells need to arrive at the injury site in the heart. For this purpose, the authors used two kinds of antibodies, CD41 and CD34 which bind platelets and endogenous stem cells, respectively. They modied CD41 antibody with PEG and DBCO, and injected it into MI mice intravenously. Then, the DBCO-PEG-CD41 binds platelets in blood, which accumulate on the MI region in the heart by their homing ability. Aer that, granulocyte colonystimulating factor was administered to the mice to stimulate and release endothelial progenitor cells (EPCs). Next, azide and PEG modied CD34 (Az-PEG-CD34) was also injected intravenously. It binds DBCO-PEG-CD41 in MI tissue area by SPACC in vivo, and recruits EPCs to the region for therapy. Injection of DBCO-PEG-CD41 and Az-PEG-CD34 increases the amount of DiR-labeled EPCs accumulated in the heart more than ve-fold. This click chemistry-based pre-targeting strategy attenuated brosis and increased CD34+ cells in the MI region showing improved therapy in vivo.

Ex vivo diagnosis
Recent studies have highlighted the potential of click chemistry ex vivo in providing valuable information for understanding tissue development, monitoring disease state, and diagnosing diseases. For instance, Rong et al. performed glycoproteomic analysis of intact and hypertrophic hearts using click chemistry with Ac 4 ManNAz for metabolic labelling of glycomes in hearts. 59 They injected Ac 4 ManNAz by intraperitoneal injection into Sprague-Dawley rats. Later sialylated glycans were labeled with alkyne-biotin using the CuAAC reaction assisted by the ligand BTTAA. With this approach, they conrmed successful metabolic incorporation of an azide tag to the glycome of the mouse model by western blotting of lysates and uorescence imaging of isolated cells using alkyne-Fluor 488. By using a Langendorff-perfusion system, they successfully performed uorescence imaging of intact hearts with a SPAAC copper-free click reaction with aza-dibenzocyclooctyne-Fluor 488 (DBCO-F488). They found an intense uorescence signal on the cell surface which was more pronounced at the cell-cell surface and on the transverse tubule (T-tubule) of the heart tissue. Furthermore, they applied this method to study cardiac glycome changes depending on pathogenesis. With SPAAC reactions, they found a signicant increase of uorescence signal in the heart of the ISO-treated cardiac hypertrophy mouse than in the heart of the salinetreated control mouse. In addition, through gel-based proteomic study assisted by the click reaction with alkyne-biotin and tandem mass spectroscopy, they found 21 and 18 proteins uniquely identied in the hypertrophic mouse and the healthy control mouse, respectively. In addition, they found that NCAM1 and a 2 M, known to be proteins relevant to hypertrophic disease, were mostly up-regulated among sialylated proteins.
Since many cancers release membrane-bound small microvesicles (MVs) into peripheral circulation, an analytical tool of MVs holds as a promising diagnosis method for diseases such as glioblastomas (GBMs). In 2012, the Lee group reported on a microuidic system combined with iEDDA type click chemistry and miniaturized micro-nuclear magnetic resonance (mNMR) for analyzing MVs from the blood of a GBM patient (Fig. 10). 60 In their study, they rst targeted protein markers of MVs with TCO modied antibodies. Later, the antibodies were coupled with Tz modied magnetic nanoparticles (MNPs) to increase the magnetic signal. With this two-step bioorthogonal approach, they successfully increased the signal >300% compared to that of direct MNP-antibody conjugation. They conrmed that the system had a low error rate (<1% instrumental error) showing excellent agreement with the uorescence ELISA method. In addition, they conrmed that the system has great detection sensitivity (from 10 4 -to 10 2 -fold), higher than other analytical methods such as western blotting, ELISA and NTA (nanoparticle tracking analysis) because bioorthogonal click chemistry allowed dense packing of MNPs on MVs.
Analysis of protein markers such as HSP90 (MV maker), CD41, and MHCII (host cell marker) positive MVs revealed that MVs indeed reected the protein proles of parent cells. Furthermore, they could monitor the efficacy of drug treatments such as temozolomide and geldanamycin for GBM disease. They conrmed that TMZ treatment did not induce signicant changes in protein markers such as CD63, EGFR, or EGFRvIII. On the other hand, geldanamycin induced a signicant decrease in EGFR and EGFRvIII expression in both cells and MVs.
For diagnostic purposes, the limit of detection (LOD) for the technology is crucial. There have been various attempts to amplify signals to lower the LOD. The Weissleder group has reported an iEDDA type signal amplication strategy for nanoparticle-based bioorthogonal diagnostic sensing assisted by a miniaturized diagnostic nuclear magnetic resonance device (DMR) (Fig. 11). 61 Briey, antibodies were labeled with TCO, and then they were coupled with Tz modied magneto-uorescent nanoparticles (MFNP-Tz). The generated signal could be ampli-ed again with TCO modied MFNP (MFNP-TCO) to form multiple MFNP layers. Since the linker between MFNP and Tz or TCO is cleavable, MFNPs are released from each other. Transverse relaxation rate (R 2 ) signals can then be measured by the DMR system. Firstly, they tested the amplication method on SK-OV-3, human ovarian carcinoma overexpressing HER2, with anti-HER2 antibodies (trastuzumab). They conrmed that the PEG linker between MFNP and TCO or Tz resulted in a higher signal to noise ratio. More importantly, alternative labelling with the MFNP-TZ/MFNP-TCO amplication method resulted in a signicantly higher signal than single antibody labelling with MFNP-TZ. Validation of the method using clinical samples from pancreatic cancer patients with EGFR (epidermal growth factor receptor), EpCAM (epithelial cell adhesion molecule), HER2, and MUC1 (mucin-1) antibodies, indicated that the method could be used for proling scarce cells from clinical samples.

Ex vivo mechanism study
Considering the importance of protein synthesis for understanding the plasticity of synaptic connections, Dieterich et al. metabolically introduced noncanonical amino acids such as azidohomoalanine (AHA) and alkyne-bearing amino acid homopropargylglycine (HPG) into newly synthesized proteins as methionine surrogates. These proteins were then chemoselectively labeled by click chemistry with Texase Red-PEGalkyne (TRA) and uorescein-PEG-azide (FKA) to study protein dynamics in neuronal systems. 62 They conrmed that newly synthesized dendritic neuronal proteins could be labeled using click reactions with AHA/TRA or HPG/FKA pairs with copper catalysts and triazole ligands, both in dissociated neurons and in a slice of brain tissue. By pulse charging HPG followed by AHA and labelling with corresponding TRA and FKA, they labeled newly synthesized proteins at two different time points. Aer demonstrating the labelling method, they used quantum dots (QDs) with copper-free click reactions between AHA-tagged proteins and DIFO (a diuorinated cyclooctyne)-biotin, followed by streptavidin-QD treatment to monitor individual newly synthesized proteins. They found that newly synthesized proteins generally have higher mobility than other neuronal membrane proteins. Furthermore, AHA-tagged proteins were slower in synapses than outside synapses.
Wang et al. have reported a method for incorporating multiple distinct noncanonical amino acids with proteins by evolving a pyrrolysyl-tRNA synthetase (PylRS/tRNA) pair with orthogonal ribosome (ribo-Q1) (Fig. 12). 63 They used saturation mutagenesis libraries consisting of 10 8 membered Pyl tRNA (N8) XXXX with eight different nucleotides in the anticodon stem loop. Desired tRNAs were selected through negative selection (to exclude Pyl tRNA(N8) XXXX as the substrate for both endogenous synthetase and decoded on ribo Q1) and positive selection (aminoacylated with an unnatural amino acid by PylRS and decoded efficiently on ribo-Q1). They conrmed that the evolved Pyl tRNA XXXX had improved quadruplet decoding efficiency and specicity which allowed efficient incorporation of unnatural amino acids into the proteins.
The study also demonstrated that unique chemical handles, including azides, alkenes, alkynes, benzophenones, and tetrazines, could be incorporated into the proteins of interest with a combination of PylRS substrates and four distinct substrates for MjTyrRS variants. They conrmed that their system had better efficiency than the MjTyrRS/tRNA CUA variant and PylRS/ tRNA UUA or PylRS/tRNA UCA . With the system, they incorporated two different noncanonical amino acids containing Tz and norbornene, into calmodulin (CaM) proteins by combination of MjTetPheRS/tRNA CUA and a NorKRS/evolved Pyl tRNA UACU pair along with the corresponding noncanonical amino acid. They conrmed that there was no intramolecular reaction between iEDDA reaction pairs on the proteins. They successfully labeled corresponding noncanonical amino acids with BODIPY-FL tetrazine or BODIPY-TMR-X bicyclononyne with a rate constant of around 1.0 M À1 s À1 . In addition, they labeled CaM protein sites specically with FRET uorophore pairs using copper-free click reactions and successfully studied conformational change of the proteins within a domain resulting from sequential ligand binding events by monitoring FRET signal changes.

Nucleotide ligation and CRISPR Cas9 system
Nucleotides such as DNA or RNA are generally linked by ligase, but their chemical ligation is also an interesting example of click chemistry. The Brown group have pioneered in this eld and developed diverse nucleotide structures by click chemistry until now. 96,97 In a recent study, this group demonstrated the utility of click chemistry-mediated nucleotide ligation in the CRISPR Cas9 system. 64 Genome editing by this system needs a single-guide (sg) RNA which enables Cas9 protein to cleave a specic DNA sequence. However, the high cost and long time involved in synthesizing sgRNA have limited wider application of CRISPR Cas9. Taemaitree et al. designed a new type of sgRNA composed of a variable 20-mer DNA-targeting region and xed the 79-mer Cas9-binding region. These two regions were synthesized separately, and the ends of 20-mer and 79-mer RNAs were modied with propargyl (alkyne) and azide groups, respectively. They were conjugated with each other by CuAAC ligation for 1-2 hour. Then, the on-target activity of the resulting clicked sgRNA was evaluated in live U2OS cells. The clicked sgRNA successfully mediated indel formation similar to control sgRNA prepared by in vitro transfection as shown in the T7E1 assay showing that the activity was not reduced aer CuAAC ligation. In addition, to prevent the potential risk of the copper catalyst, they also performed ligation using DBCO groups instead of alkynes, and showed that SPAAC between azide and DBCO groups could achieve similar results. This study demonstrates that click chemistry can be a useful tool to improve other promising technologies such as CRISPR Cas9 in biology and biomedical science.
To overcome the limitations of the broadly applied Cu(I)catalyzed azide-alkyne cycloaddition for the labeling of nucleic acids, multiple different studies for applying copper-free cycloaddition reactions to label nucleic acids have been reported. For example, the Kath-Schorr group reported site specic labeling of RNA oligonucleotides via the iEDDA reaction between norbornene and tetrazine-uorophore conjugates. 45 In this study they synthesized a clickable RNA nucleotide via norbornene-modied uridine phosphoramidite and performed a successful iEDDA reaction with multiple different tetrazine-uorophores (Oregon Green 488, and ATTO647) both in vitro and in cells. In 2016, they reported a cyclopropene-modied ribonucleotide TPT3CP TP for site specic modication of RNA nucleic acids via in vitro transcription using the (d)TPT3-dNaM system. 46 They conrmed the iEDDA reaction of cyclopropene-modied oligonucleotides with the tetrazine-uorophore conjugate via LC-MS, HPLC and PAGE analysis. Very recently, the Wagenknecht group reported copperfree dual labeling of the DNA molecule via incorporation of two different bioorthogonally reactive groups, 1-methylcyclopropenes and 1,2,4-triazines. They synthesized a 1,2,4triazine containing nucleoside via modication on the 5-position of 2 0 -deoxyuridine triphosphate and a 1-methylcyclopropene containing nucleoside via modication on the 7position of 7-deaza-2 0 -deoxyadenosine triphosphate. They demonstrated successful incorporation of the bioorthogonal nucleoside via primer extension and the resulting oligonucleotide product was labeled with two different uorescent dyes such as BCN-rhodamine and tetrazine-BODIPY. 47

Conclusions
In this review, we have introduced important researches based on copper-free click chemistry in vitro, in vivo and ex vivo for biomedical applications. Click chemistry enables chemical conjugation between two molecules with specicity and fast second order reaction rate constant under aqueous conditions without interference from other surrounding molecules. It means that articial chemical reactions are now possible on cell surfaces, in cell cytosol, and within the body. Many studies and papers have demonstrated the great potential of click chemistry not only in chemical synthesis but also for biological and biomedical applications. However, to obtain expected results in studies, we need to have an in-depth understanding of the inherent characteristics of click chemistry so as to select the optimal chemical groups for the intended purpose of the research. Particularly, for applications in vivo, there is not sufficient time for conjugation in many situations, including i.v. injections. For these cases, the importance of reaction time should be considered in more detail.
In 2012, the Weissleder group pointed out that favorable pharmacokinetics and increased circulation time of injected molecules by conjugation with polymers of high molecular weight enhanced the reaction of click chemistry in vivo. 98 Aer i.v. injection into mice, Tz groups showed a greater than 10-fold increase in the efficiency of click chemistry in vivo when they were attached to a 10 kDa dextran polymer compared to that of direct attachment to a uorescent dye. Koo et al. have also shown the effects of pharmacokinetics of injected molecules on click chemistry in vivo. 99 They conjugated Tz groups with uorescent dyes of various chemical structures. In mice, the change of charge or hydrophobicity changed their biodistribution, circulation, and secretion. When they compared the efficiency of click chemistry in vivo using these molecules, they found that the efficiency was highly dependent upon the kinetics in vivo even though all of them contained the same Tz groups. These results suggest that we need to consider all molecules and situations carefully to use click chemistry successfully. Besides the second order reaction rate constant, the stability of click molecules under special conditions is also important to consider. Generally, chemical groups with superior reactivity and fast second order reaction rate constants show relatively low stability under many conditions, which is reasonable from a chemical perspective. In 2017, the Prescher group focused on Staudinger ligation and introduced cyclopropenones with high stability and improved reactivity with phosphines. 100 They were stable aer site specic genetic incorporation and enabled sequential labeling of proteins by a bioorthogonal reaction. In 2011, Karver et al. synthesized 12 kinds of Tz derivatives and tested their stability in serum. 101 Aer 10 hours of incubation in FBS, two of them retained their reactivity almost perfectly while one molecule completely lost its reactivity. Murrey et al. also showed that azide groups were completely stable inside cells for more than one day. However, only 6% of BCN groups remained active under similar conditions, thus demonstrating different stabilities of click molecules. 102 Despite these limitations, the number of applications for click chemistry in many elds continues to grow every year. 103 This trend is supported by the development of better candidates for click chemistry. For example, to further increase second order reaction rate constants, Darko et al. recently developed a dioxolane-fused TCO (d-TCO) group which provides a second order rate constant k 2 of 366 000 M À1 s À1 with 3,6-dipyridyl-stetrazine. 10 In addition, the Devaraj group used a cyclopropene group that could participate in iEDDA reactions with Tz. 76 It may be an alternative molecule of TCO with a similar second order reaction rate constant but smaller size for specic purposes. Based on these ndings, we consider click chemistry to be emerging as a valuable technique in biomedical elds as well as organic chemistry. The scope of its application is expected to be even broader in the future with further advances in chemistry.

Conflicts of interest
There are no conicts to declare.