The Application of Bio-orthogonality for In Vivo Animal Imaging

Abstract

The application of bio-orthogonality has greatly facilitated numerous aspects of biological studies in recent years. In particular, bio-orthogonal chemistry has transformed biological research, including in vitro conjugate chemistry, target identification, and biomedical imaging. In this review, we highlighted examples of bio-orthogonal in vivo imaging published in recent years. We grouped the references into two major categories: bio-orthogonal chemistry-related imaging and in vivo imaging with bio-orthogonal nonconjugated pairing. Lastly, we discussed the challenges and opportunities of bio-orthogonality for in vivo imaging.

1. Introduction

Orthogonality is originally a mathematical term referring to generalizing the geometric notion of perpendicularity. The simplest orthogonal example is two crossed straight lines forming a 90° angle to maximally separate the two crossed lines, and the only shared point of the lines is the perpendicular point (Figure 1). The unique property of orthogonality is that two individuals are maximally independent of each other while converging into a shared point only under very strict conditions. The orthogonality concept has been widely used in various scientific research and technology development, including mathematics, physics, chemistry, computer science, telecommunications, statistics, econometrics, and economics.

Figure 1.

Diagram of bio-orthogonality. The individuals of the A–B pair are independent of each other without bio-orthogonal events, such as bio-orthogonal/click reactions, enzyme–substrate pairing, and a platform that can converge A and B. Upon the orthogonal event, A and B become dependent on each other and produce a new identity, including a reaction product and a new event, such as light production and energy transfer.

In chemistry, orthogonal experimental design has been widely used for optimizing reaction conditions,¹ and the orthogonal design of protecting groups is also a very important strategy for the organic synthesis of complex molecules.^2,3 Bio-orthogonal chemistry refers to an orthogonal chemical event that occurs in a biological system. The term was proposed by Dr. Carolyn R. Bertozzi in 2003.^4,5 Bio-orthogonal chemistry has been widely applied in biological research, and the 2022 Nobel Prize in chemistry has been awarded to scientists, including Dr. Carolyn R. Bertozzi, in this field.⁶ It is obvious that a chemical reaction between pairs is needed for the occurrence of orthogonality. Ideally, a bio-orthogonal chemical reaction is rapid and high-yield, while the orthogonal pair have minimal side reactions toward endogenous functional groups/substrates to provide high selectivity/specificity.^7,8 The most widely applied bio-orthogonal chemical reactions include the Staudinger ligation, azide–alkyne cycloaddition, strain-promoted [3 + 2] reactions, tetrazine ligation, metal-catalyzed coupling reactions, oxime and hydrazone ligations, photoinducible bio-orthogonal reactions, and native chemical ligation involving the ligation of a peptide bearing a C-terminal thioester with a second peptide with an N-terminal cysteine^7,9−16 (Figure 2). Bio-orthogonal chemistry has a significant overlap with click chemistry, particularly if click chemistry is applied in a biological system.¹⁷

Figure 2.

Most used bio-orthogonal chemistry (click chemistry) in biological studies (ref (7)).

2. In Vivo Imaging with Bio-orthogonal Ligation

Bio-orthogonal chemistry has been widely applied for in vivo biomedical imaging, and it has created new opportunities for interrogating biomolecules in cells and organisms.^8,18−20 The most used strategies include strain-promoted [3 + 2] reactions, particularly the trans-cyclooctene (TCO)–azide reaction and TCO–tetrazine ligation.²¹ The popularity of the ligation reactions for in vivo imaging is largely due to rapid and tunable kinetics and the high selectivity of reactions.²⁰ For example, the TCO–tetrazine reaction is extremely rapid with a second order rate constant of 2000 M^–1 s^–1, which allows modifications of biomolecules at extremely low concentrations.²¹ The fastest bio-orthogonal reaction was reported by Darko and Fox et al., in which the ligation rate of TCO–tetrazine could reach 3 × 10⁶ M^–1 s^–1.²² In comparison, a typical Staudinger ligation provides a rate of about 10^–3 M^–1 s^–1.^23,24 For in vitro bio-orthogonal chemistry, reaction rates are preferred to be rapid; however, if the reaction is slow, high concentrations and high temperatures can be used to overcome the problem. Unlike bio-orthogonal chemistry for in vitro studies and cell imaging, in vivo imaging requires the orthogonal pair to be assembled in situ through the click/ligation reaction. In this regard, only rapid reactions can meet this demand, because it is difficult to maintain high concentrations of the pair at the target of interest.

2.1. Near-Infrared Fluorescence (NIRF) Imaging

NIRF imaging is one of the most used technologies for preclinical animal studies and fluorescence imaging-guided surgery (FIGS), due to its acceptable tissue penetration, low cost, and simple imaging acquisition and data analysis.²⁵ The recent development of NIR-II imaging probes and methods further enhances the capacity of NIRF imaging.²⁶⁻²⁸ Bio-orthogonal chemistry provides enormous potential for in vivo imaging with NIRF probes, since nonspecific fluorescence from the excess probe, which can be washed away, has minimal interference. In addition, the tissue penetrance of NIRF is considerably high. Bio-orthogonal chemistry has been used for NIRF imaging via several strategies:²⁹⁻³¹ (1) incorporation of noncanonical amino acids (ncAAs) with alkyne or azide tags that can be conjugated with a fluorophore via bio-orthogonal reactions; (2) pretargeting that uses a molecule with a clickable tag to bind to the target of interest; (3) incorporation of clickable tags into glycans and lipids. The ncAA strategy has been widely used for cell imaging, and the pretargeting and glycan/lipid tagging strategies have been used for in vivo animal imaging.

For the majority of ncAA bio-orthogonal chemistry studies, one tag was usually incorporated into one amino acid. Remarkably, Wang et al. recently demonstrated that it was feasible to have two orthogonal tags in one ncAA. In this study, they demonstrated that 4-(6-(3-azidopropyl)-s-tetrazin-3-yl) phenylalanine (pTAF) and 3-(6-(3-azidopropyl)-s-tetrazin-3-yl) phenylalanine (mTAF) could be incorporated into recombinant proteins and antibody fragments.³⁰ This method could enable combinations of commercially available fluorophores, radioisotopes, poly(ethylene glycol) (PEG), and drugs in a plug-and-play manner in one pot. They showed that this method could be utilized for tumor imaging, image-guided surgery, and targeted therapy in mouse models. However, it is unclear whether this method is feasible for in situ bio-orthogonal reactions for in vivo animal imaging.

NIRF lanthanide complexes are promising compounds for developing NIR fluorophores, due to their attractive features of the sharp and characteristic emission peak and a long fluorescence lifetime.³² In 2022, Jin et al. reported that bio-orthogonal chemistry could be harnessed to conjugate lanthanide (Ln) complexes to biomolecules, such as ncAAs, N-azidoacetylgalactosamine (GalNAz), azido sialic acid (SiaNAz) and nucleotides, that were metabolically incorporated into cells with azido- or alkyne- handles, and the functionalized Ln-complex showed a strong increase of NIR fluorescence upon conjugation with biomolecules. They further showed that multicolor imaging was feasible for immunostaining by combining click-labeling with the Ln complexes.³³

In 2007, Hsu et al. demonstrated that metabolic oligosaccharide engineering could be used to insert sugar-reporting groups, such as alkyne or azido, into cellular glycoconjugates, which can be tagged with fluorophores via Cu(I) catalyzed [3 + 2] azide–alkyne cycloaddition to visualize glycoconjugates at the cell surface and in intracellular organelles.³¹ In 2014, a similar strategy was used to visualize bacterial peptidoglycan with NIRF azide probes via non-copper-catalyzed TCO–azido click reaction. They showed that the optimization of NIRF Si-rhodamine azide probes could provide a fluorescence enhancement of up to 48-fold upon reaction with terminal or strained alkynes from the glycans.³⁴ In 2015, Agarwal et al. reported that the metabolic incorporation of a cyclooctyne-functionalized sialic acid derivative could rapidly react with a fluorogenic tetrazine. This method enabled the imaging of sialylated glycoconjugates within live zebrafish embryos, but this report did not show the feasibility of in vivo mouse imaging.³⁵

In a typical case of bio-orthogonal chemistry for imaging, the reaction was utilized to conjugate a fluorophore tag to a warhead for the target of interest. Interestingly, in 2016, Wu et al. reported that the bio-orthogonal reaction of vinyl and tetrazine could be utilized to uncage a vinyl-masked fluorophore, and this process could be used to detect mRNA in cells and tissues³⁶ (Figure 3). In this study, vinyl-ether-caged fluorophores and tetrazine partners were conjugated to high-affinity antisense nucleic acid sequences. When the pair was bound to their respective target RNA sequences, the individuals of the pair were positioned close enough to allow the orthogonal reaction to occur, and consequently a 70-fold fluorescence intensity increase was observed from the cyanine dye. However, this study did not validate whether this strategy could be used for in vivo animal studies. One obvious hurdle is the delivery of the pair to the target of interest in live animals. Similarly, Xie et al. used the vinyl ether mask strategy to turn on the fluorescence of a tetrazine tethered fluorescent NIRF probe of a Nile red derivative, while the unmasking bio-orthogonal reaction was utilized to release a prodrug that was caged with the vinyl ester. They demonstrated that the turn-on NIRF probe could be used to image xenografted tumors, and the released drug camptothecin could inhibit tumor growth.³⁷

Figure 3.

(A) Strategy of turn-on (uncaging) fluorescence of cyanine dye via bio-orthogonal chemistry. (B) Diagram of detecting mRNA with antisense RNA probes bearing the bio-orthogonal pair. (C) Live cell detection by fluorogenic tetrazine uncaging oligomeric probes, and a cartoon depicting transfection and imaging protocol for live cell imaging. (D) CHO cells treated with 25 nM fluorogenic NIR antisense probes in the presence (top row) or absence (bottom row) of prior transfection with sfGFP-3′ BT plasmid containing the complementary target sequence in the 3′-UTR. Reprinted with permission from ref (36). Copyright 2016 American Chemical Society.

Although bio-orthogonal chemistry has been widely applied for cell, tissue, and zebrafish imaging, in vivo NIRF imaging via bio-orthogonal reactions is still challenging. Interestingly, in 2022, Zhang et al.⁴⁰ reported a new and unique strategy to utilize bio-orthogonal chemistry for in vivo mouse imaging. It is well-documented that small molecule NIRF probes tend to aggregate, which results in quenching of the fluorescence and narrowing of the Stokes shift.²⁵ Moreover, the tetrazine tag in a NIRF probe (CyP7T) also causes a quenching effect.^38,39 In this study, they discovered a “torsion-induced disaggregation (TIDA)” phenomenon in the design of a NIRF tetrazine–cyanine probe. Upon the bio-orthogonal reaction, the double-quenching effects were turned off via conformational changes of the double bonds and the conversion of tetrazine into dihydropyridazine. They showed that this approach could be used to sensitively delineate the tumor in living mice as early as 5 min post intravenous injection.⁴⁰

In recent years, one very active research direction has been NIR-II imaging due to the excellent tissue penetration and minimal tissue scattering of NIR-II probes. In 2023, Li et al. used an extracellular vesicle (EV)-mediated targeting method and the combination of bio-orthogonal reaction between the glycan-azide of the EV particle and the dibenzocyclooctyne (DBCO) handle in the NIR-II rare-earth doped nanoparticles to demonstrate the feasibility of in vivo bio-orthogonal labeling for improved NIR-II tumor imaging.⁴¹

2.2. Bioluminescence Imaging (BLI)

BLI has been used in daily imaging practice in thousands of laboratories around the world, and it has significantly promoted the progress of drug development in preclinical stages for various diseases, particularly for cancer research. Over the past three decades, BLI has profoundly changed the daily practice for cancer research.^42,43 Bio-orthogonal chemistry has also been applied for BLI.⁴⁴ In 2010, Cohen et al. designed a caged luciferin via blocking the hydroxyl group of the firefly luciferin with a moiety bearing phosphines, the ligand for Staudinger ligation. Their purpose was to use bioluminescence imaging to visualize glycans in live cells in a real-time manner. To achieve this goal, they metabolically incorporated azidoglycans into cell membranes. Upon the addition of phosphine-caged luciferin, the Staudinger reaction could remove the phosphine group to uncage the luciferin into its active form.⁴⁵ They revealed that cell-surface glycans could be imaged with probe concentrations as low as a few nanomolar and times as short as 5 min, suggesting that this method could provide excellent specificity and high sensitivity.⁴⁵ In 2020, Goun’s group used a similar strategy for monitoring the changes in mitochondrial membrane potential. They utilized the Staudinger ligation for “mitoclick” reaction of two mitochondrial-targeting precursors in vivo (in cells and in mice). By the “mitoclick” reaction, firefly luciferin can be released in situ.⁴⁶ With such a strategy, mitochondrial membrane potential could be monitored in different organs in animal models, such as old vs. young mice, brown adipose tissue, and tumors (Figure 4).⁴⁶

Figure 4.

Utilizing Staudinger ligation reaction for designing mitoclick pairs to report mitochondrial membrane potential. (A) MAL probe consisting of two components, a triphenylphosphine-caged luciferin probe (TPP-CL, pink structures) and an azido-triphenylphosphine reagent (azido-TPP, red structures), both of which are targeted to mitochondria by the triphenylphosphonium (TPP) group; bio-orthogonal reaction between the pair generates an active luciferin derivative (blue structure). (B) Linkers of caged luciferin and azido-TPPs. (C) Produced active luciferin derivative reacts with luciferase to generate photon flux.⁴⁶

The reaction between cyanobenzothiazole (CBT) and cysteine is not a classical bio-orthogonal reaction because cysteine and its analogues are not strictly exogenous; however, the CBT–cysteine reaction has been widely applied in biological studies.^44,47−51 One interesting feature of this reaction is its product, which is an analogue of firefly luciferin. Based on this feature, Godinat et al. utilized this reaction to generate firefly luciferin in situ via uncaging Asp-Glu-Val-Asp (DEVD)-caged d-cysteine. They demonstrated that this was practical for monitoring caspase-3 activity (DEVD is the substrate for caspase-3) in live mice.⁴⁷ In addition, this CBT–cysteine ligation reaction was applied for NIRF and positron emission tomography (PET) imaging via in situ assembly to amplify imaging signals.⁴⁸⁻⁵¹

2.3. PET Imaging

The pretargeting strategy is an obvious approach to take advantage of rapid bio-orthogonal reaction; however, this approach has not been widely applied in NIRF imaging. Nonetheless, the pretargeting strategy has been widely applied in PET imaging and multimodal imaging.^52,53 The most used strategy for pretargeting PET imaging is antibody-based staggered injection. In this approach, the antibody is conjugated with a clickable handle and injected prior to the radiolabeled PET probe being injected with the clickable partner. The benefits of this antibody pretargeting strategy are 3-fold: (1) it enables high selectivity to the target, because the circulating (unbound) antibody can be eliminated before imaging; (2) it can significantly improve signal-to-noise ratio (SNR) because the background signal from the free (unreacted) probe can be rapidly washed away; and (3) the side effects of radiation can be significantly reduced.

In 2012, Devaraj et al. used the pretargeting strategy for PET imaging via in vivo bio-orthogonal reactions.⁵⁴ In this study, anti-CD45 monoclonal antibodies were labeled with TCO and intravenously (iv) injected in mice 24 h before the in vivo click chemistry. For in vivo click chemistry, reaction rates are critical. To enhance the reaction rate, they used polymer-modified tetrazine (PMT) as a key enabler to artificially increase the local concentrations, thus accelerating the rate of in vivo click reaction of TCO–tetrazine (Figure 5). They demonstrated that fluorescent PMT could be used for cell imaging with cellular resolution, and the ¹⁸F labeled PMT could be used for whole animal imaging.⁵⁴ In 2013, Zeglis used a similar pretargeting approach via bio-orthogonal Diels–Alder reaction to achieve in vivo PET imaging.⁵⁵ In this study, the A33 antibody was labeled with TCO to target SW1222 colorectal cancer cells and a NOTA-conjugated tetrazine probe was radiolabeled with ⁶⁴Cu. For in vivo experiments, mice with SW1222 xenografted tumors were injected with TCO-labeled A33 antibody, and the ⁶⁴Cu-NOTA-labeled tetrazine was injected 24 h after the administration of the pretargeted A33 antibody. Results showed that, at 12 h after injection (tetrazine probe), the retention of uptake in the tumor was about 4.1% id/cc, which was excellent accumulation for in vivo PET imaging.⁵⁵

Figure 5.

In vivo pretargeting PET imaging via bio-orthogonal reactions. (A) Tetrazine cycloaddition with trans-cyclooctene forming a dihydropyrazine. (B) Schematic of PMT used in this study. The scaffold consists of dextran that has been aminated to allow attachment of tetrazine reactive groups as well as imaging agents such as near-infrared fluorophores and radioisotopes. (C) In vivo multistep delivery of imaging agent. A slow clearing targeting agent is administered first (green) and is given 24 h for localization and background clearance. Next, a lower molecular weight secondary agent (red) is delivered that rapidly reacts and is cleared from the background tissue much faster than the primary agent. (D) Kinetic parameters of consideration for in vivo clicking. The secondary tetrazine agent reacts with transcyclooctene antibodies at a given rate (k_reaction). This rate is in competition with other rates including the clearance of the secondary agent from the body (k_clearance) and internalization of the antibody (k_endocytocis). (E) PET and autoradiography using ¹⁸F tetrazine agents. PET/CT fusion of LS174T tumor xenograft labeled using either trans-cyclooctene (TCO) monoclonal antibodies (mAb TCO) or control unlabeled antibodies (mAb) followed by ¹⁸F-PMT10. Arrows indicate the location of the tumor xenograft. (F) Ex vivo imaging using autoradiography (left side) and NIRF imaging (right side) of 1 mm LS174T tumor slices after targeting with fluorescent TCO monoclonal antibody and ¹⁸F-PMT10. Reprinted with permission from ref (54). Copyright 2012 National Academy of Science.

The above examples are excellent for showing the feasibility of the pretargeting method for antibodies to cell surface epitopes; however, it is more challenging to pretarget epitopes inside cells. In 2017, Keinänen et al. used the well-documented and clinically relevant monoclonal antibodies cetuximab and trastuzumab to demonstrate that it was practical to target intracellular epitopes. In this study, the antibodies were conjugated with the TCO tag and injected into tumor-bearing mice at least 24 h before the injection of the ¹⁸F-labeled tetrazine probe. For both antibodies, the tumors could be clearly visualized in the PET images, and the final uptake of the PET tracer in tumors was up to 3.7 ± 0.1% ID/g for cetuximab and 1.5 ± 0.1% ID/g for trastuzumab.⁵⁶ Similarly, Shi et al. showed that epidermal growth factor receptor (EGFR)-specific monoclonal antibodies cetuximab and panitumumab could be used for imaging EGFR-positive colorectal cancers in preclinical animal studies.⁵⁷

2.4. Magnetic Resonance Imaging (MRI)

Although bio-orthogonal chemistry has been widely used for in vivo optical and PET imaging, it has been less applied in MRI imaging. This is likely due to the low sensitivity of MRI and the large quantities of probes needed, which are difficult to provide by in vivo click chemistry in situ. While this is very challenging, several groups have successfully demonstrated the feasibility. In 2015, Neves et al. used a metabolic labeling strategy with N-azidoacetylgalactosamine and a gadolinium (Gd)-based TCO probe to facilitate in vivo bio-orthogonal conjugation reactions. Through such a conjugation, the MRI-responsive Gd-chelator moiety could significantly accumulate on the surface membrane. In vivo MRI showed that the signals (R1 relaxation) in tumors were increased, accompanied by signals increasing in multiple organs. The increases in R1 relaxation in vivo were consistent with the ex vivo Gd concentrations measured by inductively coupled plasma mass spectrometry (ICP-MS),⁵⁸ suggesting the in vivo imaging results are reliable. In 2022, Lin and Gao’s group adopted a similar bio-orthogonal strategy with tetra-acetylated N-azidoacetylgalactosamine (Ac₄ManAz) to investigate whether ¹⁹F MRI is feasible for in vivo imaging.⁵⁹ Through intratumoral injection of the bio-orthogonal pair (Ac₄ManAz/DBCO-¹⁹F) into xenografted tumors, they showed that in vivo bioorthogonal ¹⁹F-MRI could be feasible (Figure 6). For validation, they used tris(2-carboxyethyl) phosphine (TCEP), a chemical that can reduce the azido in Ac₄ManAz to an amine, to abolish the bio-orthogonal reaction. As expected, the signals were significantly reduced with TCEP treatment.⁵⁹

Figure 6.

Bio-orthogonal chemistry for in vivo ¹⁹F-MRI. In this figure, the mechanism of BOMFLA for tumor cells is shown. Ac₄ManAz is selectively taken up by tumor cells and undergoes a series of metabolic transformations, resulting in the incorporation of azido groups into surface glycans. Specific and efficient ligation between these surface glycans and fluorinated molecules (DBCO-F) is achieved via SPAAC between the azido and cyclooctynyl moieties, which enables selective ¹⁹F labeling of tumor cells for hot spot ¹⁹F-MRI. Reprinted with permission from ref (59). Copyright 2022 American Chemical Society.

To overcome the problem of low concentrations of bio-orthogonal products in the target-of-interest, one approach is to aim for proteins that are abundant in the targets. In recent years, Ning and Caravan et al. approved that liver fibrogenesis could be an excellent target.^60,61 Extracellular matrix proteins are abundant in many tissues, such as the liver, while lysyl oxidase enzymes can catalyze the oxidation of lysine ε-amino groups on the extracellular matrix proteins to form the aldehyde containing the amino acid allysine (Lys^Ald). Notably, liver fibrogenesis is normally accompanied by the upregulation of lysyl oxidase. This suggests that the fibrogenic liver contains high concentrations of allysines, which are excellent substrates for the aldehyde-oxime/hydrazine bio-orthogonal chemistry.^8,62 Through rational design, a series of stable hydrazine-equipped manganese MRI probes were synthesized and tested. Remarkably, the probe with two hydrazine moieties (Mn-2CHyd) provided the highest affinity and turn-on relaxivity (4-fold) upon reaction with Lys^Ald. In vivo MRI imaging showed that Mn-2CHyd could be used to in vivo detect liver fibrogenesis in CCl₄-injured mice.⁶¹ Furthermore, they showed that the probe had the capacity for early detection of liver fibrogenesis and response to treatment.⁶¹

Hyperpolarized magnetic resonance (HP-MR), a powerful, sensitive, and noninvasive imaging approach, has been broadly applied in recent years in both preclinical and clinical imaging studies, particularly for metabolic imaging.^63,64 One key feature of HP-MR is utilizing the rapid metabolic reactions in vivo that result in different chemical shifts of the same nucleus, such as ¹³C, to track metabolic activities. Given that the TCO–tetrazine bio-orthogonal reaction is also a very rapid reaction, Bae et al. used ¹⁵N₄-1,2,4,5-tetrazines as potential molecular tags.⁶⁵ In this study, ¹⁵N₄-1,2,4,5-tetrazines were hyperpolarized and underwent rapid and selective cycloaddition with TCO to generate hyperpolarized ¹⁵N₂-containing cycloaddition products and hyperpolarized ¹⁵N₂ gas. The chemical shifts of ¹⁵N in tetrazines and the cyclized products are different, and can be harnessed to provide multiplex imaging information. Interestingly, this work also supports the production of hyperpolarized para-¹⁵N₂ gas, which is a biologically and medically inert gas with great potential for HP-MRI, due to its long T1 lifetime.⁶⁵ However, it seems that this strategy is still a few steps away from in vivo imaging with animals.

2.5. Bio-orthogonal Förster Resonance Energy Transfer (FRET) Imaging

In principle, bio-orthogonal chemistry is an ideal tool for constructing a FRET pair. FRET is based on the transfer of energy between two fluorophores. When one fluorophore (donor) is excited by a light source, it can transfer its energy to the other fluorophore (acceptor) through a nonradiative energy transfer process. Upon receiving the energy, the acceptor then emits light at a wavelength longer than that of the donor. The FRET efficiency depends on the distance and orientation of the pair, and the efficiency decreases rapidly as the distance increases. Normally, the distance should be less than 10 nm.⁶⁶ Ideally, the individuals of the FRET pair are totally independent, and the pairing occurs only upon the click reaction at the target of interest. By doing this, the nonspecific signals or background signals from the nonpaired individuals can be significantly reduced. Thus, the SNR can be notably improved. The parameters, including the overlap of the donor emission with the acceptor excitation/absorption, the donor and acceptor quantum yield, and the distance between the donor and acceptor, are essential to achieve high efficiency, strong intensity, and high SNR.⁶⁷ For FRET imaging, both the intensity and lifetime of the pair and individuals have been regularly applied. Although bio-orthogonal FRET is feasible for microscopic imaging with cells and tissues, it has been less explored for in vivo animal imaging.

In 2020, Wu et al. proposed a novel scheme for imaging based on the lifetime changes of a FRET pair, and they termed it as bio-orthogonal “labeling after recognition” (LaR), and monitoring the cleavage activity of caspase-3 was used as a proof-of-concept study⁶⁸ (Figure 7). In FRET imaging, the lifetime of the donor usually is significantly shortened upon FRET. In this study, they utilized this change for cell imaging. In their scheme, the typical substrate of caspase-3, DEVD, was conjugated with an azide and a Gly-Lys (GK)-norbornylene tag at the amino and carboxyl termini, respectively. Then the modified DEVD peptide was incubated with caspase-3 for recognition with cleavage releasing the tagged fragments that could be labeled via the bio-orthogonal reaction to install the fluorophores. As expected, the lifetime of the donor iridium(III) complex was significantly longer (408 ns) than that of the iridium(III) complex (57 ns) in the FRET pair, in which an iridium(III) complex and a rhodamine derivative were used as the FRET pair. The additional benefit of this LaR approach is enabling more efficient bio-orthogonal reaction, due to the reduced hindrance after cleavage.⁶⁸ However, it is unclear whether this strategy is feasible for in vivo animal imaging.

Figure 7.

“Labeling after recognition” strategy for detecting caspase 3 activity. This approach can result in lifetime changes and avoid hindrance for click-reactions. (A) Chemical structures of N3-DEVDGK-norbornylene and the bio-orthogonal labeling reaction with 1 and 2. Scheme showing the caspase-3 detection in the “labeling before recognition” approach (B) and the “labeling after recognition” approach (C). (D) Phosphorescence lifetime traces upon caspase-3–DEVD recognition for 0.5–2 h in two sensing approaches. Error bars represent the standard deviations of three independent measurements. (E) Photoluminescence lifetime confocal microscopy images of HeLa cells stimulated with cisplatin, and zoomed-in view of the marked area, and (F) photoluminescence decay curves of the circled area. Scale bar: 20 μm. Reprinted with permission from ref (68). Copyright 2019 American Chemical Society.

In 2022, Albitz et al. demonstrated that bio-orthogonal ligation could be harnessed to activate FRET dyads via manipulating the quenching effect of tetrazine in the FRET pair.⁶⁹ In the preactivated FRET pair, coumarin–tetrazine and Cy5 were conjugated into one molecule. Upon the bio-orthogonal reaction, the fluorescence of coumarin–tetrazine was turned on because of the conversion of tetrazine into dihydropyridazine, and consequently, the energy transfer to Cy5 was feasible. They demonstrated that this FRET mechanism could result in improved cyanine fluorogenicity together with increased photostability and large apparent Stokes shift, which can enable lower background fluorescence in live cell imaging.⁶⁹ However, it is unclear whether this strategy is feasible for in vivo animal imaging.

In 2019, Beliu et al. systemically investigated the Diels–Alder reaction between TCO-modified ncAAs and 22 known and novel 1,2,4,5-tetrazine–dye conjugates spanning the entire visible wavelength range. They discovered that photoinduced electron transfer (PeT) from the excited dye to tetrazine is the main quenching mechanism in red-absorbing oxazine and rhodamine derivatives. They further demonstrated that efficient and specific labeling of all tetrazine dyes investigated could be used for super-resolution microscopy with high SNR, including single-molecule imaging.⁷⁰ In the line of high-resolution microscopic imaging, Bessa-Neto et al. showed that bio-orthogonal labeling of transmembrane proteins with ncAAs allows access to masked epitopes in live neurons of both dissociated culture and organotypic brain slices. They showed that this method could reveal the distinct dendritic surface distribution of γ2 S44* and γ8 S72* in neurons.⁷¹

2.6. Bio-orthogonal Chemiluminescence Resonance Energy Transfer (CRET) Imaging

Bio-orthogonal chemistry has not been widely applied to construct CRET pairs for in vivo imaging. In 2017, Degirmenci et al. synthesized a CRET pair through the alkyne–azide click reaction to conjugate luminol, the most used chemiluminescence reagent for biological assays, with BODIPY scaffolds.²⁰⁰ However, this CRET pair could not be used for in vivo animal imaging because of the need for alkaline and H₂O₂.

3. In Vivo Imaging with Bio-orthogonal Nonconjugated Pairing

In principle, any pairing events can be orthogonal. Orthogonality commonly refers to two independent individuals converging into a single entity or event that occurs only under a unique condition. The formed single identity can be a reaction product, such as a click-chemistry reaction product,⁷ and the event also can be FRET, CRET, and protein–ligand pairing.⁷² For in vivo imaging, the most frequent pairing events can be found in FRET, CRET, and bioluminescence imaging (luciferin/luciferase pairing). In the above sections, we have discussed these imaging methods with the help of bio-orthogonal chemistry. In fact, numerous examples could be found for these imaging modalities without bio-orthogonal chemistry. In the following section, bio-orthogonal examples of bioluminescence imaging are presented. Nonconjugated FRET and CRET imaging are also discussed.

3.1. Bio-orthogonal BLI via Optimizing Luciferin/Luciferase Pairs

Given that bioluminescence imaging is a pairing event between luciferin and luciferase, great efforts have been devoted to optimizing the orthogonality between the pair for multiplex imaging in cells and animals. In this regard, the most used strategy is to re-engineer luciferases through mutation selection. With this optimization, enhanced orthogonality could enable multiplex imaging with several luciferin/luciferase pairs. In 2017, Jones and Prescher et al. reported such orthogonal luciferase–luciferin pairs.⁷³ In this study, a series of sterically modified luciferins, which were poor emitters of native firefly luciferase (Fluc) but intrinsically capable of robust light production, were synthesized. These luciferins were then screened against libraries of mutant luciferases to identify the pair that offered the best orthogonality. Interestingly, high orthogonality was observed with luciferins that were modified at 4′- and 7′-positions and mutation B and C, respectively.⁷³ To further generalize the screening method, the same group used parallel screening of 12 luciferin analogues with panels of 159 mutant enzymes to generate 1908 (12 substrates × 159 enzymes) individual data points. They used data mining to rank the orthogonality, and over 100 pairs were validated in vitro. Remarkably, three pairs were applied in cell and animal models and showed excellent orthogonality (Figure 8).⁷⁴

Figure 8.

Parallel screening against mutated firefly luciferases and a luciferin library for rapid identification of orthogonal Luc–luciferin pairs. Reprinted with permission from ref (74). Copyright 2017 American Chemical Society.

Although the above parallel screening method is generalizable and rapid, it still needs a point mutation of the native luciferase. Ideally, to achieve high orthogonality, the catalytic enzyme should be totally different from its native enzyme. To achieve this goal, in 2023, Yeh et al. used a deep-learning-based “family wide hallucination” approach to design artificial luciferases, with which the oxidative chemiluminescence of the synthetic luciferin substrates diphenylterazine and 2-deoxycoelenterazine can be selectively catalyzed. The designed luciferases are totally not sequence-related to the original luciferases (such as Renilla luciferase) for coelenterazine analogues. In addition, the artificial luciferases have a much higher substrate specificity than the native luciferases (Figure 9).⁷⁵ This new artificial intelligence (AI) approach provided enormous potential to design substrate-specific luciferases, which can be harnessed for multiplex and orthogonal bioluminescence imaging.⁷⁴

Figure 9.

Machine learning assisted de novo design of artificial luciferases for diphenylterazine (DTZ). This strategy can provide excellent bio-orthogonality for Luc–luciferin pairs. (A) Family wide hallucination. Sequences encoding proteins with the desired topology are optimized by Markov chain Monte Carlo (MCMC) sampling with a multicomponent loss function. Structurally conserved regions (peach) are evaluated on the basis of consistency with input residue–residue distance and orientation distributions obtained from 85 experimental structures of NTF2-like proteins, whereas variable nonideal regions (teal) are evaluated on the basis of the confidence of predicted inter-residue geometries calculated as the KL divergence between network predictions and the background distribution. The sequence-space MCMC sampling incorporates both sequence changes and insertions and deletions to guide the hallucinated sequence toward encoding structures with the desired folds. Hydrogen-bonding networks are incorporated into the designed structures to increase structural specificity. (B–D) Design of luciferase active sites: (B) generation of luciferase substrate (DTZ) conformers; (C) generation of a rotamer interaction field (RIF) to stabilize anionic DTZ and form hydrophobic packing interactions; (D) docking of the RIF into the hallucinated scaffolds and optimization of substrate–scaffold interactions using position-specific score matrix (PSSM)-biased sequence design. Reprinted with permission from ref (75). Copyright 2023 Springer Nature.

3.2. Bio-orthogonal FRET via Nonconjugated Pairing

Although bio-orthogonal chemistry has been widely used to assemble FRET pairs in vitro and in cell and tissue imaging, it is still far less applied to in vivo animal studies. The possible reasons include the following: (1) the efficiency of the click chemistry is not high enough to in situ conjugate the donor and the acceptor when their concentrations are critically low at the target of interest, (2) since the individuals of the FRET pair for in vivo imaging need to be in the NIR range, the molecular weight and hydrophobicity are usually high, and this may lead to inadequate accumulation of them in the target of interest, and (3) the amount of the click product is not adequate to provide strong enough signals in vivo, because NIRF–FRET imaging is not as sensitive as PET imaging. In contrast, pretargeting bio-orthogonal chemistry has been successfully used for PET imaging. This is likely due to the highly sensitive detection of PET imaging.

One strategy to overcome the above problem is to use nonconjugated FRET. In 2011, our group proposed to utilize nonconjugated small molecule FRET for differentiating amyloid β (Aβ) species.⁷⁶ From the cryo-electron microscopy (cryo-EM) structure, each Aβ fibril contains numerous ligand/probe binding cavities, and each neighboring cavity is very close (<1 nm).⁷⁷⁻⁷⁹ If the individuals of the FRET pair can bind to the fibrils, there are great chances for the pair to be positioned close enough (<10 nm) to allow FRET to occur. In other words, the Aβ fibrils can be considered as the orthogonal converging points to assemble the nonconjugated donor and acceptor into a FRET pair (Figure 10). In this study, we showed evidence that Aβ fibrils could be used as platforms for FRET pair assembly and that this technique could be used to differentiate Aβ fibrils, oligomers, and monomers. Upon mixing two structurally similar curcumin derivatives (CRANAD-2 and CRANAD-5) that served as the FRET pair with Aβ40 aggregates, a FRET signal could be observed via spectroscopic recording. In contrast, no FRET signal was detected when using Aβ40 monomer solutions. Lastly, this FRET technique enabled us to quantify the concentrations of Aβ monomers and high molecular weight species in solutions.⁷⁶

Figure 10.

(A) Diagram of bio-orthogonal CRET, in which the nonconjugated ADLumin-1/CRANAD-3 pair insert into Aβ fibrils concurrently to enable energy transfer. No CRETing between the pair can be observed before interacting with Aβ fibrils. CRETing can be observed when two nonconjugated molecule probes bind to fibrils to bring the ADLumin-1 (donor) close enough to the CRANAD-3 (acceptor). (B) Spectrum of the CRET pair with Aβ40 fibrils in PBS (red line); the peak was consistent with the emission of CRANAD-3 in the presence of Aβ40 fibrils. Chemiluminescence spectrum of ADLumin-1 with Aβ40 fibrils (black line), and spectrum of the mixture of ADLumin-1 and CRANAD-3 without Aβ40 fibrils (pink line). The CRETing efficiency was very high, evident by the low intensity in the 500–560 nm range. (C–E) Observation of CRET under in vivo mimic conditions: (C) mixture of ADLumin-1 and CRANAD-3 with Aβ40 aggregates (CRET+Aβ) injected subcutaneously into the right inner thigh of a female nude mouse, and the control group of ADLumin-1 + Aβ injected into the left side; (D) in vivo spectra of the unmixed contributors, ADLumin-1 + Aβ (green), and CRET + Aβ (red); (E) spectral unmixing imaging was conducted with sequence imaging at 15 min postinjection to separate signals from ADLumin-1 (left, green) and CRET pair (middle, red), and a composite image of the two unmixed components (right). Reprinted with permission from ref (81). Copyright 2020 Springer Nature.

3.3. In Vivo Bio-orthogonal CRET Imaging via Nonconjugated Pairing

Although the above-discussed nonconjugated FRET was feasible in solution, this FRET method was nearly impossible to apply in vivo due to its short excitation of the donor probe. Given that chemiluminescence imaging does not need excitation, the short excitation problem can be solved with CRET. The fundamental principle of CRET is similar to that of FRET. However, the major difference is that the external light source in FRET is no longer needed. The energy transfer is between an acceptor and a high-energy intermediate (HEI) that is generated through a chemical reaction (instead of excitation by an external light source in FRET). Pairing a short emission chemiluminescence probe with a NIRF probe can extend the pair’s emission into the NIR window, greatly benefiting in vivo brain imaging. Similar to FRET with Aβ fibrils, we reasoned that bio-orthogonality could be achieved with the assistance of Aβ fibrils to assemble a CRET pair (Figure 10). In 2020, we used the nonconjugated pair of ADLumin-1 and CRANAD-3,⁸⁰ a smart NIRF probe for Aβs, to demonstrate that bio-orthogonal CRET was achievable with Aβs in solutions, in brain homogenates, and for in vivo whole-brain imaging.⁸¹

We first validated the bio-orthogonal CRET pair solutions via mixing ADLumin-1, CRANAD-3, and Aβ aggregates in PBS buffer. We found that the CRET between ADLumin-1 and CRANAD-3 was obvious in the presence of Aβs, while there is no observable CRET without Aβs (Figure 10). Remarkably, compared to the CRET pair without Aβs, the CRET signal had about a 133-fold increase with Aβs at 660 nm (Figure 10). The high CRET efficiency was likely due to the very close proximity of the pair when they randomly inserted into β-sheets of Aβ aggregates. In addition, the well-overlapped spectra of ADLumin-1 emission and CRANAD-3 excitation also contributed to the highly efficient CRET. We further verified the CRET feasibility in a brain-like environment with mouse brain homogenate and a mimic in vivo environment via subcutaneous injection of the mixture of the CRET pair with Aβs into a nude mouse at the ventral hind limb.⁸¹

For in vivo brain CRET imaging, we injected (iv) a solution containing both ADLumin-1 and CRANAD-3 into 5xFAD transgenic mice and wild-type (WT) mice. Given that FRET/CRET imaging is always accompanied by spectral analysis, we collected images with an open filter and 18 filters from 500 to 840 nm. With the open filter setting, we observed 2.04-fold differences between the 5xFAD group and the WT group 15 min after the injection of the CRET pair solution. After spectral unmixing, the margin between AD and WT was significantly increased and reached 3.22-fold from the CRET pair. In addition, we performed ocular CRET imaging and found that the CRET pair provided a considerably large margin of 2.11-fold between 5xFAD and WT mice. This large margin indicated that ocular CRET could be a very useful tool for monitoring the changes in Aβ concentrations via ocular imaging.⁸¹

We believe this strategy can also be extended to detect other misfolding-prone proteins such as misfolded α-syn, tau, and TDP-43. Indeed, our preliminary results suggested that bio-orthogonal CRET imaging with the ADLumin-1/CRANAD-14 pair was feasible for imaging α-syn deposits in solutions and in vivo transgenic mouse models (unpublished). However, as with any molecular brain imaging, one of the significant challenges is penetration of the blood–brain barrier (BBB). With this nonconjugated bio-orthogonal strategy, both of the individuals of the pair need to meet the Lipinski rule for BBB.⁸²

4. Perspectives and Conclusions

Bio-orthogonal chemistry has transformed the landscapes of numerous biological research areas, and its comprehensive significance is beyond the scope of this review.^6,7 In this review, we have focused on its application for in vivo imaging, particularly for animal imaging. Although a large volume of literature can be found for in vitro and cell imaging, its application for in vivo animal imaging still requires more effort to expand its scope. From our point of view, several factors have limited its application in this field. The key issue is to increase the amount of the orthogonal reaction product, which will provide adequate signals for detectors, particularly for NIRF and MRI imaging. Until now, the most used strategy is utilizing antibodies for pretargeting; however, this approach may not be efficient enough to increase the local concentration, because of the antibody’s large size and poor penetration into targets, such as solid tumors and brain targets. The potential alternatives include affibodies and nanobodies, which can be much smaller and have better tissue penetration. Although nanoparticles have been widely used for in vivo animal imaging, they may have the same limitations as antibodies, due to their low uptake in solid targets.^83,84 Another strategy to increase the concentration of the orthogonal product is seeking high-affinity ligands for the target of interest. This could be particularly useful for a target with a large Bmax. Learning from the successful pretargeting PET imaging, we can have more sensitive readouts to overcome the problem. For NIRF imaging, selecting a high-quantum-yield fluorophore is one feasible option. Bio-orthogonal NIR-II fluorescence imaging is also a highly attractive approach; however, the quantum yields of NIR-II probes, particularly small molecule organic probes, need significant improvements. In recent years, chemiluminescence in vivo imaging has been a very active research direction, due to the high SNRs that chemiluminescence probes can provide, because the excellent SNRs enable sensitive detection and deep target imaging.^85,86 In this regard, bio-orthogonal chemiluminescence imaging, which is rarely explored,⁸¹ is very promising for in vivo animal imaging.

Bio-orthogonal chemistry for bioluminescence imaging represents an appealing approach because of its high SNRs. However, only a few examples can be found in the literature. We expect novel strategies to expand this scope. Bio-orthogonal chemistry for MRI imaging remains highly challenging. However, it is still possible to use this chemistry for targets with high abundances such as collagens in fibrogenesis.

In vivo imaging with bio-orthogonal nonconjugated pairing represents a new direction. The advantages of non-click/conjugated bio-orthogonal imaging include the following: (1) Probe design is simpler and probes are easier to synthesize because no clickable tags are needed for the bio-orthogonal pair. (2) The bio-orthogonal event is not dependent on the click/ligation reaction rates. (3) It is possible that much lower amounts of probes are needed. On the other hand, the disadvantages include the following: (1) Its application scope could be limited. By far, the most applicable cases are for misfolded proteins. However, theoretically, this method could be extended to other targets. In this regard, more studies are needed. (2) It requires both probes to have strong binding to the target. Ideally, the two probes do not compete with each other for binding sites.

For bioluminescence imaging, deep learning for de novo design of highly orthogonal artificial luciferase is highly attractive,⁷⁵ which has great potential to transform multiplex bioluminescence imaging in the coming years. Given the unique features of misfolded proteins, such as Aβ, tau, and α-syn, we demonstrated that bio-orthogonal CRET was feasible with nonconjugated pairs for in vivo imaging. We believe that there is still plentiful room to be explored for imaging the misfolded proteins.

Glossary

Vocabulary Section

Molecular imaging

An imaging technology that utilizes the signals associated with molecule probes to generate images. The associated signals can be magnetic resonance signals, fluorescence, bioluminescence, radioactive emission, and acoustic signal.

Bio-orthogonal chemical reactions

Chemical reactions that can occur in biological systems without interfering with endogenous biochemical processes.

Click chemistry

Chemical reactions featuring quick and reliable conversion and completion for producing various molecules through splicing of the small units.

Bioluminescence imaging

An imaging modality that is widely utilized in biomedical imaging by using the reaction between luciferin and its enzyme luciferase to emit light in the 300–800 nm range.

Förster resonance energy transfer

An energy transfer process between a matched pair with a spectral overlap of a donor’s emission and an acceptor’s absorption (excitation). Additionally, the pair is required to be positioned within a 10 nm range.

Imaging with bio-orthogonal nonconjugated pairing

An imaging method that involves two bio-orthogonal entities for producing imaging signals without covalent conjugation.

The authors declare no competing financial interest.