Click and release: bioorthogonal approaches to “on-demand” activation of prodrugs

Xingyue Ji abc, Zhixiang Pan b, Bingchen Yu b, Ladie Kimberly De La Cruz b, Yueqin Zheng b, Bowen Ke *a and Binghe Wang *b
aLaboratory of Anesthesia and Critical Care Medicine, Department of Anesthesiology, Translational Neuroscience Center, West China Hospital and State Key Laboratory of Biotherapy, Sichuan University, Chengdu, Sichuan 610041, China. E-mail: bowenke80@hotmail.com
bDepartment of Chemistry and Center for Diagnostics and Therapeutics, Georgia State University, Atlanta, Georgia 30303, USA. E-mail: wang@gsu.edu
cCollege of Pharmaceutical Sciences, Soochow University, Suzhou, Jiangsu 215021, P. R. China

Received 6th June 2018

First published on 6th February 2019


Prodrug approaches represent an excellent solution to certain pharmaceutical issues commonly encountered in the drug discovery and development process. Along this line, the chemistry needed for the bio-reversible derivatization of drug functional groups for on-demand release is critical. In recent years, “click and release” approaches have shown great promise in the design of prodrugs because of their bioorthogonality and controlled bond-cleavage, which help ensure prodrug stability during circulation and ready cleavage at the desired site of action. This review highlights recent developments of this research field and discusses issues yet to be addressed.


image file: c8cs00395e-p1.tif

Xingyue Ji

Xingyue Ji is a professor at Soochow University. He has been serving as a visiting associate professor at Sichuan University since June 2018. He got both his bachelor's and master's degrees in chemistry from the University of Science and Technology of Beijing in 2006 and 2008, respectively. Thereafter, he moved to Peking Union Medical College Institute of Medicinal Biotechnology, and received his PhD degree in medicinal chemistry in 2011. After that, he took an assistant professor position in the same institution, and he worked as a postdoctoral fellow in Dr Binghe Wang's lab from 2013 to September 2018.

image file: c8cs00395e-p2.tif

Left to right: Bingchen Yu, Ladie Kimberly De La Cruz and Zhixiang Pan

Bingchen Yu received his BS degree from Shandong University (2014) and his PhD degree in chemistry from Georgia State University in 2018. He is now working as a postdoctoral fellow in Dr Binghe Wang's lab. Ladie Kimberly De La Cruz is currently pursuing her PhD degree in chemistry under Dr Binghe Wang's supervision at Georgia State University. She obtained her BS degree from the University of the Philippines Los Baños (2011). Zhixiang Pan received his BS degree from Anhui Medical University (2014). He is currently pursuing his PhD degree in chemistry in Dr Binghe Wang's lab.

image file: c8cs00395e-p3.tif

Yueqin Zheng

Yueqin Zheng was born in 1988 in Fujian, China. He received his BS degree in materials chemistry from the University of Science and Technology of China (USTC) in 2011 and his PhD degree in medicinal chemistry in 2017 from the Chemistry Department of Georgia State University. Now he is working with Professor Daniel Kohane in Boston Children's Hospital, Harvard Medical School, and Professor Robert S. Langer at the Massachusetts Institute of Technology. His research interests include designing prodrugs for gasotransmitters, developing novel chemical reaction-based drug delivery systems, and local delivery of antibiotics and anesthetics.

image file: c8cs00395e-p4.tif

Bowen Ke

Bowen Ke received his PhD degree in 2009 from Sichuan University, followed by postdoctoral work at Georgia State University with Dr Binghe Wang. In 2013, he joined the faculty of West China Hospital of Sichuan University as an associate professor. He was awarded the Chinese Pharmaceutical Association (CPA)-Servier Young Investigator Award in Medicinal Chemistry in 2017. His current research interests lie in the general areas of medicinal chemistry and chemical biology, aiming to develop new therapeutic and diagnostic agents.

image file: c8cs00395e-p5.tif

Binghe Wang

Binghe Wang is Regents’ Professor of Chemistry and a Georgia Research Alliance Eminent Scholar in Drug Discovery at Georgia State University. He is the Founding Director of the Center for Diagnostics and Therapeutics, the Chief Editor of Medicinal Research Reviews, and the Founding Serial Editor of the “Wiley Series in Drug Discovery and Development.” He also serves on the editorial boards of numerous journals. His research interests are in the areas of drug discovery, drug delivery, and new diagnostics. Dr Wang, together with his students and postdocs, has published over 280 papers and edited several books covering the same general areas.


Introduction

Factors other than efficacy play a key role in determining the ultimate success or failure of a drug discovery and development effort. Thus, there is the concept of drug developability.1–4 Among all the factors, problems in terms of solubility, metabolic stability, membrane permeability, distribution, excretion, half-life, metabolic activation, substrate properties for transporters, and toxicity could each individually derail a drug discovery project. There are many ways to address these issues including medicinal chemistry efforts in modifying structural features responsible for undesirable properties. Another commonly used method is use of prodrugs,4 which can be employed to temporarily mask the functional group responsible for the undesirable properties of a drug. Further, a drug can be tethered to a vector molecule for targeted delivery to minimize toxicity and improve potency, such as in the case of antibody–drug conjugates (ADCs). A key feature in a prodrug approach is bioreversible derivatization of the active drug molecule. Ideally, the prodrug moiety is stable during circulation and yet readily cleavable at the desired site of action. Generally speaking, there are two types of prodrug derivatization strategies: those that release the drug upon chemical reactions and those that rely on enzymes to release the active drug. For example, there are ester prodrugs that can undergo hydrolysis and thus release the active drug. There are also disulfide linkers that are sensitive to the presence of reducing agents,5,6 and oxidation-sensitive prodrugs that can be activated by, e.g., hydrogen peroxide, which tends to accumulate in inflamed areas or in cancer cells.7–9 In the enzyme-activated prodrug area, there are esterase-, protease-, and phosphatase-sensitive prodrugs, among others.10

Recently, prodrug activations using bioorthogonal reactions have emerged as a powerful approach, affording both stability and controlled activation. When used in ADCs or with other targeting vectors, the same approach can be used for targeted delivery and activation of drugs. Below we summarize recent developments by dividing the review into two major sections: (1) prodrugs of traditional small-molecule drugs and (2) prodrugs of gasotransmitters. It should be noted that this review is mainly focused on non-transition metal-based bioorthogonal chemistry. Readers who are interested in transition metal-based decaging chemistry in living systems are referred to some very excellent reviews and research papers.11–18 In addition, this review is meant to highlight key concepts, and not meant to be comprehensive. Therefore, not all publications in the field are included. The authors encourage communications from colleagues to bring important omissions to the authors’ attention for inclusion in future updates.

Bioorthogonal triggers for prodrug activation of traditional small-molecule drugs

Much of the work in this area has been focused on delivering anticancer agents. This is because the standard first-line care for cancer still largely relies on the use of cytotoxic agents such as taxols, doxorubicin and cisplatin,19 which necessitate the work for finding ways to minimize toxicity and side effects.20 Therefore, it is highly desirable to have a pan-cytotoxic agent delivered and/or activated selectively at the site of action.

Bioorthogonal reactions refer to any chemical reactions that can be performed in a biological milieu without disrupting native biochemical processes. Such reactions are also often referred to as click reactions, and include Staudinger ligation21 and the inverse electron-demand Diels–Alder reaction (DAinv),22 among others.23 Due to its bioorthogonality, such chemistry has substantially facilitated the study at the interface of chemistry and biology, including biomolecule labeling/modifications,24,25 drug discovery26 and chemoproteomics,27 among others. Recently, bioorthogonal chemistry has also found promising applications in prodrug activation. This section discusses various chemical strategies for bioorthogonal chemistry-based prodrug activation.

Release based on using an azido group as a latent amino group

One approach of prodrug activation is to use an azido group as a latent amino group for subsequent nucleophilic addition or self-immolative elimination. Florent and co-workers described one such approach wherein Staudinger ligation followed by a self-immolation reaction led to the release of doxorubicin (Dox, Fig. 1) from its prodrug.28 The Staudinger ligation reaction is a well-known modification of the classic Staudinger reaction involving intramolecular cyclization. Thus, prodrug 1 (Drug = R1NH2) was designed to be activated by the Staudinger reaction with an azido compound (2a). The aza-ylide intermediate 3 undergoes an intramolecular cyclization to release intermediate 4. Subsequent 1,6-elimination releases the parent drug 7 with the formation of a quinomethide species, which can undergo hydration to give 6. The concept was first studied with a model compound, 4-nitroaniline (pNA) (prodrug 1a). pNA release was achieved with the addition of 10 mM of the azido compound (2a) to 5 mM of prodrug 1, leading to a half-life of 1 h. When the idea was studied using a Dox prodrug (1b), 90% drug release was achieved within 3 hours when 2.8 mM of 1b was incubated with 5.2 mM of 2a in THF/H2O (1[thin space (1/6-em)]:[thin space (1/6-em)]1) at 37 °C. Although no biological data were presented in this report, this study demonstrated the initial proof of concept. However, the concentrations of the prodrug and the azido compound used were quite high. Further improvement in the reaction rate will help the development of practical prodrug approaches based on Staudinger ligation.
image file: c8cs00395e-f1.tif
Fig. 1 Prodrug activation via Staudinger ligation.

A similar example was reported by Robillard in 2008 (Fig. 2) by using phosphine to reduce an azido to an amino group as a way to activate the prodrug.29 In particular, upon conversion of the azido group in prodrug 8 to an amino group, the intermediate (10) is seen to undergo self-immolation through 1,6-elimination to release Dox and byproduct 12. The reaction between prodrug 8 (100 μM) and phosphine reagent 9a (200 μM) in water at 37 °C was completed in 20 h as monitored by HPLC. One of the issues with the phosphine-based reducing agent is its tendency to be oxidized. In this study, the stability of the phosphine reagent was examined using LC-MS. It was found that the phosphine reagent 9a was stable in water for 20 h at 37 °C, but not in cell culture medium, in which complete oxidization was observed after 4 h of incubation. Such a stability problem necessitated a twice-daily dosing scheme in a cell proliferation assay lasting 72 h. Under those conditions, the IC50 of prodrug 8 was found to be 0.074 μM with the addition of 9a (5 × 60 μM), which was comparable to that of Dox (0.086 μM). In the absence of the phosphine agent, the prodrug was much less toxic (IC50 = 15.1 μM).


image file: c8cs00395e-f2.tif
Fig. 2 Prodrug activation via the Staudinger reaction.

Generally speaking, the examples described above established the proof of concept of using the reduction of an azido group as a trigger to activate prodrugs. However, there are some limitations that need to be addressed for future development. First, the easy oxidation of a phosphine reagent means that it will be hard to control its concentration at the desired site of action, and repeated dosing may be needed. Second, the sluggish reaction between an azido group and a phosphine agent (generally k2: ∼10−3 M−1 s−1)30 means that the concentrations of the two reactants need to be high in order to afford a reasonable half-life. Third, aryl or alkyl phosphines tend to be hydrophobic and large, which are not desirable properties for drug development.

Other bond-cleavage chemistry based on the Staudinger reaction has also been reported, though they were not intended for prodrug activation. However, replacing the releasable moiety with a drug payload would allow such chemistry to be applied to bioorthogonal prodrug design. For example, Guo and co-workers appended a fluorophore to an antibody via a cleavable linker (Fig. 3A, 8a) to label an intracellular protein of interest (POI).31 After taking the fluorescence images, treatment with 9b yielded compound 10a, which fell apart to release the fluorophore 11a. The same cell could be analyzed using 8a again with a different antibody targeting a new POI after washing away the fluorophore 11a. Another example was the one by Bertozzi and co-workers using Staudinger–Bertozzi ligation,32 wherein the click reaction between 1c and 2b afforded 3c, which underwent cyclization to form amide 6a and also triggered the release of 7a (Fig. 3B). One can easily envision applying such chemistry to the design of bioorthogonal prodrugs.


image file: c8cs00395e-f3.tif
Fig. 3 Bond-cleavage chemistry based on Staudinger ligation.

To mitigate the limitations associated with phosphine reagents used for azido group activation, Gamble's group came up with a new strategy to activate an azido group through a 1,3-dipolar cycloaddition. For this, a highly reactive trans-cyclooctene (TCO) was used to unmask an azido group under near physiological conditions.33 In particular, TCO (14) was reacted with an azido compound (13) to form an unstable 1,2,3-triazoline compound (15), which expels nitrogen to afford aldimine 16.34 Subsequent hydrolysis followed by elimination would release the payload 19 (Fig. 4). The proof-of-concept work was done using 13a and TCO-OH 14, which afforded intermediate 15a in CDCl3. Subsequent release of coumarin 19a was very fast in a protic solvent system (CDCl3/D2O = 9[thin space (1/6-em)]:[thin space (1/6-em)]1). The second-order rate constant for the initial 1,3-dipolar cycloaddition was determined to be 0.017 M−1 S−1 ± 0.003 (with the equatorial isomer of TCO-OH 14) or 0.027 M−1 s−1 ± 0.006 (with the axial isomer of 14) in CH3CN/PBS (1[thin space (1/6-em)]:[thin space (1/6-em)]1, 37 °C). The cytotoxicity (IC50 = 0.96 μM) of 13 was found to be comparable to that of the positive control Dox (0.71 μM) in the presence of 100 μM of TCO 14 after 72 h treatment. The prodrug (13) was much less toxic (IC50: 49.9 μM) in the absence of compound 14. Such results indicate complete drug release after treatment with TCO-OH. However, the azido prodrug 13 has stability issues. For example, only 95%, 68% and 56% of 13a were recovered after 4, 24 and 53 h of incubation in mouse serum/PBS (1[thin space (1/6-em)]:[thin space (1/6-em)]1), respectively. Interestingly, the drug release yield between TCO 14 (equatorial isomer) and 13c in 1[thin space (1/6-em)]:[thin space (1/6-em)]1 serum/PBS (51.4% in 4 h) was slightly higher than that in PBS alone (34% in 4 h), and the second order rate constant in serum/PBS (1[thin space (1/6-em)]:[thin space (1/6-em)]1) was determined to be 0.137 M−1 s−1 ± 0.012. This means that the first half-life in serum/PBS (1[thin space (1/6-em)]:[thin space (1/6-em)]1) would be about 20 h when the initial concentrations of both reactants were 100 μM. Such information is important for designing further bioassays and utility assessment.


image file: c8cs00395e-f4.tif
Fig. 4 Prodrug activation via 1,3-dipolar cycloaddition reaction between an azido group and TCO.

To ensure efficient payload release at low concentrations within a reasonable timescale, bioorthogonal reactions with fast kinetics are highly desirable for in vivo applications. According to Sustmann's theory,35 the click reaction between TCO and an azido compound is a type-II 1,3-dipole cycloaddition. Therefore, introduction of electron withdrawing groups on the phenyl ring of 13 should lower its LUMO energy level, and thus increase the reaction rate. Indeed, a reaction rate constant of 0.110 M−1 s−1 was achieved with the tetrafluoro-substituted analogue, as compared to 0.017 M−1 s−1 for the analog without a fluoro-substitution. However, the introduction of a fluoro group also made the subsequent decaging steps (from 15 to 19) slower, presumably due to the slowed 1,6-elimination of the stabilized aniline (18) by the electronegative fluoro group.36 Nevertheless, such 1,3-dipolar cycloaddition is an improved strategy for activating an azido group in terms of the reaction rate tunability and stability of the activating reagents as compared to the Staudinger reaction.

Notably, the same chemistry has also been applied to specifically “switch on” the functions of a POI intracellularly, wherein a lysine residue in the active site is genetically modified with an azido-masked self-immolative linker, and the functions of the POI can be easily restored by “clicking” with a secondary phosphine-based reagent or TCO under physiological conditions.37,38 This strategy is very powerful for the investigation of the functions of POI.

Release based on the reaction between an alkene and a tetrazine

The aforediscussed “click and release” strategies possess one common liability: sluggish reaction kinetics (k2 = ∼10−3 M−1 s−1). To advance such a strategy for in vivo applications, bioorthogonal reactions with fast reaction kinetics are highly desirable. The reaction between a trans-alkene and a tetrazine is among the fastest click reactions, and the reaction rate is easily tunable.39–41 Thus, there has been much work in exploiting this reaction pair for “click and release” applications in prodrug development. In 2013, Robillard's group reported this reaction for prodrug development (Fig. 5).42 After the cyclization between TCO (20) and tetrazine (21), the 1,4-dihydropyridazine intermediate 22 can undergo either direct elimination or tautomerization and 1,4-elimination to release 25 and the payload compound R1NH2. The proof of concept study was conducted using Dox as a model drug, and it was shown that incubation of the Dox–TCO conjugate (20a, 25 μM) with 10 equivalents of 21b or 21c in 25% MeCN in PBS at 37 °C led to 55% and 79% release of Dox at the 4 and 16 min points, respectively. In contrast, incubation of the prodrug alone at 37 °C in PBS (72 h) or serum (24 h) resulted in no Dox release, demonstrating the stability of the prodrug and the specificity of triggered release by a tetrazine. It is worth noting that a faster click reaction does not necessarily lead to improved drug release yield. For example, the reaction between 20a and 21a is the fastest among all the tetrazines tested. However, it only led to 7% of Dox release. The incomplete Dox release using 21a was because the majority of the cyclization product was trapped as intermediate 23/24 without further reactions. Such results indicate that the reaction rate with this system is not the only factor to consider. Subsequent cell-based cytotoxicity studies further confirmed Dox release from the click reaction between 20a and 21a–c. Dox itself showed an EC50 of 0.037 μM in A431 tumor cells, while its TCO conjugated compound 20a was about 100 times less toxic (EC50: 3.834 μM). When incubated with 10 μM 21a, 21b or 21c, the cytotoxicity increased by a factor of 14, 53 or 78, respectively (EC50: 0.280, 0.072, and 0.049 μM) as compared to 20a. However, the potency level never reached that of Dox alone, which would require 100% drug release. The same chemistry has also been successfully applied to the activation of intracellular proteins. In particular, a lysine residue in the active site of a protein was masked with TCO via a carbamate moiety, leading to the inhibition of its functions. Upon clicking with a tetrazine, the lysine residue was “unmasked,” and the functions of the protein were restored subsequently.43–45
image file: c8cs00395e-f5.tif
Fig. 5 A “click to release” strategy using an inverse-electron-demand Diels–Alder (DAinv) reaction between trans-cyclooctene (TCO) and tetrazine.

To better understand key factors that contribute to the release profiles between TCO and tetrazine, Chen's group did a systematic study of the relationship between the decaging rates and the substituents on the tetrazine ring in 2016.46 It was found that electron withdrawing groups (EWG) not only increased the initial cycloaddition rate constants, but also significantly decreased the decaging rate. In contrast, electron donating groups (EDG) helped the decaging step, but did not result in high drug release yields because of their slow cycloaddition rate. Therefore, a combination of both EWG and EDG on the same tetrazine ring achieved an optimal balance between the cycloaddition and decaging rates. Indeed, all the unsymmetrical tetrazines showed better decaging activities than their symmetric counterparts. In particular, 21d showed a four times faster decaging rate than 21c in an in vitro fluorogenic assay (Fig. 5). In a firefly luciferase (fLuc) system, where the catalytic lysine residue on fLuc was caged with TCO through a carbamate bond to abolish its activity, treatment with 21d rescued 80% of the enzyme activity in 2 min, as compared to only 10% of that of 21c.

Notably, a very recent study by Weissleder and co-workers unveiled the release mechanism of the click reaction between TCO and a tetrazine, by which idiosyncratic release yields for various tetrazines could be explained.47 Importantly, the previously reported high release yield (∼80%) of the click reaction between 21c and 20 was validated as an artifact due to the acidic LCMS conditions (MeCN/water, 0.1% formic acid) used to quantify the released payload. When a phosphate buffer (pH = 7) was used for LC analysis, the true release yield after 6 h was only around 20%. It was also further confirmed that the payload release was much enhanced under acidic conditions. Therefore, several new tetrazines with a carboxylic acid group were synthesized and studied for their ability to trigger payload release from compound 20 (Fig. 6A). Compared to 21c, tetrazines 21e–g exhibited much improved release yields (50–100%), with 21f achieving complete release in PBS (pH 7.4). Such improvement is attributed to the presence of a carboxylic acid group, which preferentially protonates the neighboring dihydropyridazine nitrogen, and thus accelerates the tautomerization to dihydropyridazine for 1,4-elimination (Fig. 6B). Additionally, it was also confirmed that the low release yield for 21c was the result of the formation of a “dead-end” product without the capacity to decage the payload. Such a dead-end product was formed by the intramolecular nucleophilic addition of the amidic nitrogen (Fig. 6C). To address this issue, an alkyl group (e.g. methyl, 22c) was introduced on the amidic nitrogen to prohibit the intramolecular cyclization. This modification led to a complete release (100%) with tetrazine 21c in PBS. It is worth noting that only a trace amount of dead-end product was observed in the cases of tetrazines 21e–g, in that the acid enhanced 1,4-elimination reaction outcompeted the formation of the dead-end product. In summary, this study provides critical insight into the (un)release mechanism of the click reaction between TCO and tetrazine, and invaluable guidance for the design of new drug delivery systems employing such bioorthogonal reactions.


image file: c8cs00395e-f6.tif
Fig. 6 The (un)release mechanism of the dihydropyridazine intermediate. (A) The chemical structures of tetrazines. (B) A proposed mechanism for intramolecular acid assisted elimination. (C) The formation of a dead-end product.

The aforementioned examples showcased the release of the payload with an amino group upon click reaction. In order to expand the scope of application, Robillard and co-workers explored the capacity of this click reaction to liberate payloads with various functional groups including amino, carboxyl and hydroxyl groups (Fig. 7). It was found that the payload was readily released with a yield of 90% after 20 h of incubation under near physiological conditions in the case of 20c and d. Surprisingly, in the case of 20e, with R being an alkyl group, only a slightly lower release yield (ca. 80–84%) was observed, despite the fact that an alkoxide is a very poor leaving group. In all cases, a biphasic release profile was observed, which showed a fast release (ca. 60–85%) in the first 30 min, followed by a sluggish release over 20 h. This work significantly expanded the release scope of the payload. Especially important is the release of alcohol via an ether linker (20e), which is superior to the others (carbamate and ester) in terms of serum stability, and hence a higher release specificity can be expected if used as a cleavable linker for ADCs.


image file: c8cs00395e-f7.tif
Fig. 7 The release scope of the click reaction between a tetrazine and TCO.

Using the same chemistry, Royzen and co-workers described fluorescently labeled magnetic iron oxide nanoparticles (MNPs) for image-guided prodrug activation (Fig. 8).48 Such MNPs were modified with tetrazine and near infrared fluorescent (NIF) dye Cy5.5. The tetrazine moiety is for payload release from prodrug 20a, and the NIF dye is employed to track MNPs both in vitro and in vivo. The fluorescence imaging and flow cytometry experiment confirmed efficient internalization of both MNPs and 20a in MDA-MB-231 cells. Cell viability studies showed that treatment with MNPs and 20a led to comparable cytotoxicity to Dox alone, and yet no obvious decrease in cell viability was observed when cells were treated with MNPs or 20a alone, demonstrating efficient Dox release from the click reaction between MNPs and 20a. The fluorescent nature of such MNPs could be employed to monitor the in vivo prodrug activation. When the specific localization of MNPs in tumors (conjugation of targeting vectors may be required for such a purpose) is confirmed by fluorescence imaging, dosing the prodrug 20a could lead to specific drug release in the tumor, and thus minimize off-target effects.


image file: c8cs00395e-f8.tif
Fig. 8 Imaging-capable nanoparticles for drug delivery.

To take the “click and release” approach further, several groups have reported targeted delivery using the TCO and tetrazine pair in vivo. Robillard's group applied this pyridazine elimination to trigger release of Dox from ADCs using a tumor pretargeting strategy (Fig. 9).49 Since this strategy is designed for extracellular release, an anti-TAG72 mAb (CC49) was employed to target the tumor and stay on the cellular surface without internalization upon binding to tumor-associated glycoprotein-72 (TAG72), which is widely present on the surfaces of many neoplastic cells. As expected, tetrazine 21c (300 μM) triggered Dox release in high yield (73%) from 27 after 1 h of incubation in PBS at 37 °C. The conjugation of TCO modified Dox to the mAb also did not perturb the in vivo behavior of the mAb in terms of half-life and biodistribution. 27 was then injected (5 mg kg−1, i.v.) into LS174T xenografted mice. To prevent off-target effects resulting from Dox release into the blood, a clearing reagent (31) was injected before the administration of activators 21b and c. The tetrazine moiety of the clearing agent was designed to click with the TCO without causing drug release, and the galactose residue was designed to result in fast clearance from the blood by the liver.50 Subsequent injection of 21b (i.v./i.p.) or 21c (i.p.) resulted in significant suppression of tumor growth. However, i.v. injection of 21c did not yield any significant blockage of tumor growth due to its fast clearance from the blood. In order to mitigate such a limitation, 21b and c were conjugated with 10 kDa dextran (28 and 29) to increase the clearance time. Although 28 and 29 led to lower Dox release yields (∼40%) in vitro, complete tumor blockage was achieved without signs of acute toxicity at a 0.336 mmol tetrazine per kg (i.v., 28) or 33.5 μmol tetrazine per kg dose (i.v., 29).


image file: c8cs00395e-f9.tif
Fig. 9 Activation of a mAb–TCO–doxorubicin conjugate with a tetrazine.

To further optimize this click and release system, Robillard's group devised a new TCO functionalized ADC (tc-ADC, Fig. 10),51 wherein a diabody without the Fc region was used to preclude the Fc-associated toxicity to normal tissue and to enhance the penetration of the ADC into tumor cells due to its decreased size. In addition, a polyethylene glycol (PEG) linker was inserted between the diabody and the TCO functionalized toxin monomethyl auristatin E (MMAE) to accelerate the clearance of the ADC from the blood (completed in 2 days post-injection), and thus obviate the need for administering a clearing reagent (i.e.31 in Fig. 9) before dosing the activator. In addition, the PEG linker also shields the TCO from binding to copper-containing serum proteins, which otherwise would isomerize and inactivate TCO. As for the activator compound 21h, tetrazine containing a tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) chelate is designed with a much slower clearance rate (t1/2 = 10 min vs. 1 min for 21a), and it can achieve almost quantitative payload release in vitro as compared to only 40% for 28/29. Studies of in vivo payload release from tc-ADC with activator 21h showed that MMAE was mainly released in an LS147T-tumor with a high expression level of TAG72. In animal model studies, a multi-dose of tc-ADC (3.75 mg kg−1) with activator 21h (36 mg kg−1) resulted in significant repression of tumor growth in LS147T-tumor bearing mice with barely detectable residual tumor masses until the end of the 4-month study. The administration of tc-ADC or activator 21h alone showed no tumor repression. Because anti-TAG72 would not initiate endocytosis upon binding to TAG72, the control vc-ADC with a protease cleavable linker only showed a slight tumor repression effect with a significantly larger tumor size as compared to the tc-ADC treated group. Such results were attributed to the inefficient extracellular protease-based cleavage. In summary, this study fully demonstrated the proof of concept, reiterated the advantages of “click and release” strategies in the context of ADCs for cancer treatment, and took one step toward the translation of this strategy into clinical applications.


image file: c8cs00395e-f10.tif
Fig. 10 The chemical structures of the compounds used in ADCs along with a relevant activator.

Following the seminal work by Robillard's group, Royzen's group also creatively solved the delivery problem of bioorthogonal prodrugs by a “catch and release” strategy.52 Briefly, biocompatible hydrogel modified tetrazine (HMT) was injected directly into the desired tumor site, followed by the systemic administration of the drug-conjugated TCO 20a. Such an approach was designed for specific Dox release at the tumor site, where the “click and release” reaction occurs (Fig. 11). The proof of principle study was conducted in nude mice bearing fibrosarcoma (HT-1080) xenografts. Initially, HMTs were injected immediately next to the tumors. The mice received either Dox treatment (14 μmol kg−1, every 4 days) or 20a (14 μmol kg−1, daily) for 10 days, and no further treatment was given after that. Two weeks after the last treatment, the median tumor size remained undetectable for both groups. However, 30 days after the last treatment, the Dox treatment group showed a median tumor size of over 2000 mm3. In contrast, the median tumor size for the 20a group remained undetectable. No sign of tumors was observed for this group even at the end of the experiment (88 days after the last treatment). Meanwhile, no tumor growth inhibition was observed for local injection of HMTs or the prodrug only, indicating that the observed tumor suppression effects were the result of the released Dox from the click reaction between 20a and HMTs. Furthermore, myelosuppression, a key dose-limiting factor of doxorubicin, was also significantly reduced by such a strategy. Compared to the “pretargeting strategy”, such a “catch and release” strategy has several advantages: (1) the need to administer only one reagent and the elimination of the need to tune the pharmacokinetics of the two components; (2) the ability to administer multiple rounds of treatments with one injection of HMTs; (3) the elimination of the need for clearing agents; and (4) the enhanced selectivity of drug release sites.


image file: c8cs00395e-f11.tif
Fig. 11 A “catch and release” approach for local drug administration.

In 2016, Bernardes’ group developed a strategy to release alcohols into cells using a vinyl ether/tetrazine pair.53 As shown in Fig. 12, the parent compounds were masked as vinyl ethers. After the cycloaddition with a tetrazine, the parent compound was designed to be released. Several vinyl ethers (100 mM) were reacted with tetrazine (21a, 200 mM) in dichloromethane at room temperature, leading to the release of the corresponding alcohol compounds in good yields (50–69%). These vinyl ether reagents were stable in PBS at 37 °C for over 8 h. The cycloaddition step was determined to be the rate-limiting step based on 1H-NMR observations. Different tetrazines (21a, 21h and i) were tested and 21i was found to be the most reactive one, presumably due to its reduced steric hindrance. It is worth noting that unlike the “click and release” between TCO and tetrazines, where the substituents on the tetrazines play a key role in determining the decaging rate and yield, the substituents on the tetrazines did not significantly impact the release yields of alcohols. A vinyl ether protected halogen-bearing duocarmycin analog (31) was chosen for evaluating the biocompatibility of this drug delivery system. It was reasoned that decaging with tetrazine should lead to intermediate 32, which would undergo a Winstein spirocyclization to form the active drug (33, Fig. 12). The intracellular prodrug activation was tested in A549 cells, and it was found that co-treatment with 10 μM 31 and 20 μM 21i resulted in the same magnitude of cytotoxicity observed with 10 μM 33 alone. Meanwhile, 10 μM 31 or 10 μM tetrazine 21i alone didn’t show significant cytotoxicity. It is worth noting that only 20 μM of tetrazine is needed for the activation of the prodrug, although the second order reaction rate constant was reported to be only around 10−4 M−1 s−1, which is roughly translated to a first half-life of over 138[thin space (1/6-em)]888 h. The possible explanations we suspect for the observed cytotoxicity within 72 h include intracellular concentration enrichment and/or cycloaddition rate acceleration by intracellular glycoproteins.54


image file: c8cs00395e-f12.tif
Fig. 12 Vinyl ether/tetrazine pairs for release of hydroxyl containing compounds.

As an expansion of the utility scope, a vinylether masked self-immolative linker was developed by Bradley and co-workers (Fig. 13).55 This method could be employed to release an amine-containing payload as well. Initially, an amine-containing fluorophore (Nile blue) was appended to such a self-immolative linker via carbamate to quench its fluorescence (34a). Upon reaction with 100 μM of tetrazine 21a, an 8-fold fluorescence increase was observed in PBS at 37 °C, indicating that such a “click and release” system worked as designed. Subsequently, this self-immolative linker was incorporated into a polymer to form nanoparticles, to which Dox was appended via a carbamate linker (34b). Dox release from such nanoparticles was tested in HEK293T and PC3 cells. When incubated with 35 μM of tetrazine 21a for 48 h, 1 μM of 34b (4 μM Dox) led to 80% cell death. In contrast, treatment with 21a or only 34b showed no toxicity. These results signified that tetrazine successfully triggered Dox release from such nanoparticles.


image file: c8cs00395e-f13.tif
Fig. 13 A tetrazine responsive self-immolative linker.

These drug delivery systems use a vinyl ether/tetrazine pair instead of a TCO/tetrazine pair. Compared with TCO, vinyl ether is more stable under near physiological conditions. However, vinyl ether has a much slower cycloaddition rate (k2 = ∼10−4 M−1 s−1) with tetrazine than TCO, which should be taken into consideration for future applications.

Release based on the reaction between benzonorbornadiene and tetrazine

A new type of tetrazine-mediated bioorthogonal “click and release” chemistry was developed in 2017 wherein benzonorbornadiene derivatives were used as payload carriers.56 In parallel to the afore-mentioned strategies, a DAinv reaction initiates the whole process. Cycloaddition between tetrazine 38 and the benzobornadiene derivative 39 coupled with nitrogen gas extrusion gives intermediate 40. Subsequent retro-Diels–Alder cycloreversion of 40 unmasks the self-immolative isobenzofuran/isoindole scaffold 42 (Fig. 14). The chemistry for this strategy was inspired by prior work in nucleic acid sensing57 wherein tweaking the isobenzofuran/isoindole scaffold 42 to include a carbamate linker attached to an amine-containing payload constituted the release portion of the design.
image file: c8cs00395e-f14.tif
Fig. 14 Tetrazine-mediated release of cargos from benzonorbornadiene derivatives.

The ability of this system to release the model payload p-nitroaniline (pNA) was demonstrated using HPLC. Incubation of tetrazine 38a (18 mM) with benzonorbornadiene analogs 39a–c (6 mM) at 37 °C in DMSO/H2O (9[thin space (1/6-em)]:[thin space (1/6-em)]1) led to complete consumption of 39a–c, a decrease in 38a concentration, and the formation of free pNA and 41. Studies of reaction kinetics were conducted monitoring the disappearance of 38a at 525 nm in the presence of excess 39a–c. In pure DMSO, the second order rate constants for 39a–c with tetrazine 38a ranged from 0.0044 to 0.015 M−1 s−1. Expectedly, a slight increase in the rate was observed in the presence of 10% H2O. Using the more soluble tetrazine 38b in 60% H2O, the rate constants increased up to 0.190 ± 0.029 M−1 s−1, which is higher than the reported release kinetics of other strategies such as those between TCO and azido compounds (k2 = 0.137 M−1 s−1) and tetrazine and vinyl ethers (k2 = 0.00021 M−1 s−1), and Staudinger ligation (k2 ≈ 0.001 M−1 s−1).

To showcase the potential application of this design as a prodrug activation strategy, 39d with doxorubicin as the cargo was studied. When 39d (0.2 mM) was incubated with 38b (1.6 mM) in DMSO[thin space (1/6-em)]:[thin space (1/6-em)]PBS (1[thin space (1/6-em)]:[thin space (1/6-em)]1) at 37 °C, it was found that prodrug 39d was completely consumed after 360 min with release of free doxorubicin. In a cytotoxicity study against A549 pulmonary adenocarcinoma cells, it was shown that the cytotoxicity of 39d (0.001 to 10 μM) after 72 h incubation increased with increasing amount of 38b (25 to 200 μM). The EC50 of prodrug 39d with 200 μM of 38b was 0.096 ± 0.022 μM, while that of doxorubicin only was 0.088 ± 0.031 μM. Control experiments with the prodrug only (39d or 38b) indicated no obvious toxicity at 10 and 200 μM, respectively. Stability studies of 39d in both DMSO/PBS and human serum indicated that the prodrug was stable for 48 h. However, longer incubation times led to 39d degradation. The decomposition of 39d was attributed to the inherent instability of doxorubicin itself rather than the benzonorbornadiene linker because 39b (analog with pNA as the cargo) was shown to be stable for one week.

Most of the “click and release” strategies in the literature use amine-containing cargos to demonstrate feasibility. In a follow-up study,58 Franzini and co-workers explored the potential of the benzobornadiene–tetrazine bioorthogonal strategy to release various leaving groups such as carboxylic acids, alcohols/phenols, and dialkylphosphates. Derivatives of 39 (6 mM) with different leaving groups were incubated with 38a (18 mM) at 37 °C in 10% H2O in DMSO, and the release was studied using proton NMR. All derivatives of 39 containing the NAc group released the leaving group in 82 to 96% yield, while derivatives of 39 containing an oxygen at the bridging position gave lower yields (in the range of 68 to 77%). Bimolecular reaction rates were determined by monitoring the decrease in absorbance at 525 nm of 38b (0.25 mM) in the presence of excess amounts of derivatives of 39 (2.5 to 6.25 mM) at 37 °C in 10% H2O in DMSO. The second order reaction rate constants in the range of k2 = 0.0092 to 0.086 M−1 s−1 were observed.

Derivatives of 39 with an oxygen bridge exhibited faster reaction rates compared to the NAc derivatives. However, cargo release from the oxygen-bridged derivatives of 39 exhibited a temperature dependence wherein a higher cargo release rate was observed at a higher temperature (50 °C), while only traces of the cargo are found at a lower temperature (22 °C). Further examination revealed that at 22 °C, instead of forming the isobenzofuran intermediate that would lead to cargo release, a 1,4-tautomer (or a decomposition product thereof) was produced. No such temperature-dependent release was observed from NAc-bridged derivatives of 39.

Relative to other strategies, the prodrug activation strategy described above is associated with several advantages. First, the synthesis of the reagents is relatively straightforward compared to that involving TCO. Furthermore, the rate of release of a model cargo is generally faster compared to other prodrug activation strategies such as that involving tetrazine-mediated unmasking of vinyl ethers and that of dissociative Staudinger ligation. It also seems that the stability issue associated with this strategy is very minimal in comparison to those plaguing other strategies such as Staudinger ligation with the use of metabolically unstable phosphines and those in TCO-mediated strategies arising from spontaneous trans/cis isomerization. However, in this study, the rate constant for Dox release from the reaction between prodrug 39d and tetrazine 38b was not reported. Moreover, whilst it was shown that 39d was indeed fully consumed after 360 min, the amount of accumulated free Dox released after the reaction was not quantified. Part of this could be due to the stability problem associated with Dox. Furthermore, the use of 10 to 20 equivalents of tetrazine 38b to achieve around the same level of cytotoxicity of 39d as that of free Dox could present an issue.

Bioorthogonal triggers for prodrugs of gasotransmitters

In the drug discovery world, there are two major classes of drugs: small organic molecules and biomacromolecules such as proteins, peptides, and nucleic acids. However, there are others such as metal complexes (cisplatin and Pepto-Bismol) and gases that also offer tremendous promise in developing therapeutic agents. Known gases approved for medical use include nitrous oxide (laughing gas as an anesthetic), nitric oxide (a vasodilator, such as INOmax) for treating neonates with hypoxic respiratory failure,59 and oxygen as a supplement for patients with breathing or cardiovascular problems.60 In recent decades, three gases, carbon monoxide (CO), hydrogen sulfide (H2S), and nitric oxide (NO), have gained much attention as endogenous signaling molecules.61 These are gaseous molecules and have unique delivery problems. While the delivery of NO has had a long history of success including the classic nitroglycerin for heart attacks, similar approaches for delivering CO and H2S have lagged far behind. The next section focuses on using bioorthogonal chemistry for developing prodrugs for these two gasotransmitters, CO and H2S. In addition, SO2 is also attracting attention and it seems more and more likely that it is also an endogenous signaling molecule: it is produced endogenously and has important physiological roles.62 Therefore, prodrugs of SO2 are included in this section as potential tools for delineating sulfur dioxide's physiological and pathological functions.

Prodrugs for carbon monoxide

Carbon monoxide (CO) is widely accepted as an endogenous signaling molecule, and is also referred to as a gasotransmitter.63 Additionally, accumulating evidence has presented CO as a potential therapeutic agent with pleiotropic pharmacological effects, including anti-bacterial,64 anti-inflammatory,65 anti-viral,66 and anti-cancer activities,67 among many others. However, efforts towards developing CO for clinical applications have been severely hampered by the lack of pharmaceutically acceptable delivery forms of CO.52,68 Inhaled CO and metal-based CO releasing molecules (CORMs) have been widely employed in basic research for CO's physiological and pathological roles with much success, but each has its limitations in clinical application, as is true with any particular class of structural scaffolds.69,70 Several elegant photo-sensitive metal-free CO prodrugs have been devised for temporal–spatial delivery of CO. Their applications are more suited for topical use or in cavities accessible by light.71–74 Therefore, the development of metal-free CO prodrugs with controllable release profiles is highly desirable to complement the existing CO delivery approaches.

A bimolecular “click and release” system for organic CO-prodrugs

CO has been found as a by-product for many chemical reactions. For example, the electron inverse demand DAinv between alkynes 45 and cyclopentadienones 44, albeit under harsh conditions (e.g. reflux in xylene), proceeds smoothly to extrude CO to yield the cycloaddition product 47 in high yield (around 90%, Fig. 14).75 Because the reaction rate for DAinv is predominantly determined by the energy gap between the HOMO of the dienophile and the LUMO of the diene, it was reasoned that the reaction between an alkyne and a cyclopentadienone could occur under mild conditions (e.g. physiological conditions) to release CO by employing an activated alkyne with high HOMO energy, such as an 8-membered strained cyclic alkyne. In 2014, Wang and coworkers conducted proof-of-concept studies using a strained alkyne (e.g. exo-BCN) and tetra-substituted cyclopentadienones (Fig. 15).76
image file: c8cs00395e-f15.tif
Fig. 15 A “click and release” strategy for CO prodrugs.

As expected, the reaction between 48a and 44a went smoothly at room temperature in MeOH with a second order reaction rate constant of 0.6 M−1 s−1. The CO release was confirmed by both the elucidation of the structure of cycloaddition product 50 and a CO-myoglobin assay. The combination of 44b and 48b was also able to suppress the lipopolysaccharide (LPS)-induced TNF secretion from Raw264.7 cells, and no similar effects were observed for either 44b or 48b alone. These results firmly established the chemical feasibility of using DAinv as a way to deliver a sufficient amount of CO for biological studies, and paved the way for further medicinal chemistry efforts for novel metal-free CO prodrugs.

It is worth noting that substituents on the dienone ring (R1–R4, Fig. 15) have to be chosen from electron withdrawing or bulky groups for this bimolecular model reaction. The reasons underlying this are two-fold. First, the LUMO energy level of the dienone can be decreased with electron-withdrawing groups to facilitate the reaction with a strained alkyne. Second, dimerization of the cyclopentadienone occurs easily, and electron withdrawing or bulky groups help keep the dienones in the monomer form.77

The properties of the cyclopentadienone can be tuned readily through conjugation chemistry or various substitutions. In studying the anti-inflammatory effect of CO prodrugs, 44a was found to decrease the viability of Raw264.7 cells after 24 h treatment with an IC50 value of 12.5 μM. Since CO is known to be highly diffusive, the click reaction for CO release does not necessarily need to occur intracellularly. Therefore, a hydrophilic mannose moiety was attached to both the dienone and the strained alkyne (44b and 48b) to decrease their membrane permeability, and hence to attenuate their cytotoxicity. As a result, both 44b and 48b presented no obvious cytotoxicity to Raw264.7 cells after 24 h incubation up to a concentration of 1 mM.

A unimolecular “click and release” system for organic CO-prodrugs

Although the aforementioned bioorthogonal prodrugs for CO demonstrated chemical feasibility, several issues remained. Due to the bimolecular nature of the bioorthogonal system in Fig. 15, the CO release rate cannot be tuned independent of the concentrations of the reactants, and this would complicate in vivo application. To synchronize the pharmacokinetic profiles of the two components in a drug delivery system would be a great challenge, especially with slow reaction kinetics (k2: <1 M−1 s−1 in DMSO/PBS at 37 °C). In order to overcome these limitations, Wang and co-workers further devised a unimolecular “click and release” system of metal-free CO prodrugs, wherein the dienone moiety and the non-activated alkyne were tethered with various linkers (Fig. 16 and 17).78 The rationale underlying such a design is the significant reaction rate acceleration afforded by entropic factors. Although the intermolecular reaction between a non-activated alkyne and a dienone requires harsh conditions (reflux in xylene), the intramolecular version with an appropriate linker occurs under near physiological conditions (DMSO/PBS = 4[thin space (1/6-em)]:[thin space (1/6-em)]1, 37 °C) to extrude CO. Additionally, by varying the tethering linker (amide or ester) and the substituent on the tethering linker, the CO release rate can be easily tuned independent of the concentration of the prodrugs, with half-lives ranging from minutes to days in a mixed aqueous solution at 37 °C. Meanwhile, with a naphthalene group attached to the 3,4-positions of the dienone ring, the inactive by-product after CO release is highly fluorescent, allowing for the real-time monitoring of CO release. These metal-free CO prodrugs were shown to deliver enough CO to recapitulate CO-associated anti-inflammatory effects in LPS challenged RAW264.7 cells. Furthermore, one representative prodrug, 51 (BW-CO-103: X = –isopropyl-NH, n = 2, R1 = H, Fig. 16), was studied for its anti-inflammatory effects in a 2,4,6-trinitrobenzenesulfonic acid (TNBS)-induced colitis model in mice. At a dosage of 15 mg kg−1, 51 showed pronounced anti-inflammatory effects as indicated by the improved survival rate (75% vs. 45% of control), injury scores, and suppression of TNF-α and MPO levels. Meanwhile, its inactive product BW-CP-103 (52 in Fig. 16) had no similar effects at the same dose. A similar study was conducted in a kidney ischemia reperfusion injury mouse model with demonstrated efficacy.79
image file: c8cs00395e-f16.tif
Fig. 16 A unimolecular “click and release” approach to CO prodrugs.

image file: c8cs00395e-f17.tif
Fig. 17 Structural modifications made to the CO prodrugs with scaffolds I and II.

Since the CO release rate plays a pivotal role in determining CO-associated biological activity, and CO prodrugs with different release profiles are needed for different applications,80 Wang and coworkers conducted intensive structural modification work to scaffolds I and II (Fig. 17), and probed effects of the tethering linker and the substituents on the CO release rate.81 Some significant structure–release rate relationships were observed; such insights can serve as invaluable guidance for the design of new CO prodrugs with specific release rates. Generally speaking, the CO release rate is primarily determined by entropic factors in the structural scaffolds described. In particular, cyclization leading to the formation of a 5-membered ring (n = 1, scaffold I) is much faster compared to that of a 6-membered ring (n = 2, scaffold I). An amide linker (X = NR, scaffold I) is more entropically favorable as compared to an ester linkage (X = O, scaffold I) due to the more rigid nature of the former. A gem-dimethyl substitution (R3, scaffold I) on the tethering linker substantially accelerates the CO release rate. An internal alkyne (R4 = methyl, scaffold I) is less favorable for the cycloaddition as compared to its terminal counterpart (R4 = H, scaffold I). Interestingly, replacement of one phenyl group on the dienone ring (Fig. 17, scaffold I, R1 = Ph) with a methylthiol group did not slow down the cycloaddition rate. This is understandable since the Hammett constant of a methylthio group is similar to that of a phenyl substituent.82 In addition, two separated phenyl ring substitutions at R2 positions (Fig. 17, scaffold I) significantly favored the cycloaddition compared to the naphthalene ring substitution. This was attributed to the lower calculated electron density of the dienone ring with two-separated phenyl ring substitutions.81 In summary, the CO release rate can be readily tuned from minutes to days by varying entropic factors, and the understanding of the structure–release rate relationships serves as a powerful guidance for the design of novel CO prodrugs with specific release half-lives. These compounds with tunable CO release rates will be powerful tools for investigating the relationships between the release rate and biological phenotypes.

An enzyme-activated “click and release” system for organic CO-prodrugs

The unimolecular CO prodrugs in the previous section overcame several limitations of the bimolecular system, such as simplifying drug delivery for in vivo studies and tuning the reaction rate independent of the concentration of the prodrug. However, one main undesirable property of such a unimolecular system is the uncontrollable CO release upon dissolution. As a result, the time it takes to administer the drug after dissolution makes a difference, especially with a prodrug that has a short half-life (min). Therefore, it is desirable to have prodrugs that are activated by a trigger. It is envisioned that the alkyne moiety could be held in a position disfavoring the cycloaddition through conformational constraints by using appropriate cleavable linkers, and hence the CO release is prevented or slowed. However, in the presence of an endogenous stimulus, the cleavable linker is broken to unleash the constraint, and thus free the alkyne for subsequent “click and release” of CO. Based on such a rationale, Wang and coworkers designed and synthesized two esterase-sensitive CO prodrugs 53a and b.83 As shown in Fig. 18, the conformation of the alkyne moiety was fixed at a position disfavoring the cycloaddition by a fused 7-membered lactone ring. Hydrolysis of the lactone ring by esterases would release this constraint and lead to spontaneous cycloaddition to release CO. With an initial concentration of 20 μM, the half-lives of CO release for 53a and 53b were 1 h and 4 h, respectively, in the presence of 10 units per mL of porcine liver esterase (PLE) in 5% of DMSO/PBS at 37 °C. In contrast, the half-lives in the absence of esterase were 17 h and 24 h, respectively. In both cases, the cycloaddition product 55 was obtained as the sole product, and the cycloaddition product 56 without lactone hydrolysis was not detected in either case, demonstrating the success of the conformational constraint strategy. Furthermore, both prodrugs showed pronounced TNF suppression effects in LPS challenged RAW264.7 cells, and their cyclized product 55 along with a negative control 53c without the alkyne group did not exert similar effects, suggesting that the observed effects were from the CO released from 53a and b. Subsequent cell imaging studies with a CO fluorescent probe, COP-1,84 further confirmed intracellular CO release from 53b.
image file: c8cs00395e-f18.tif
Fig. 18 A schematic illustration of the conformational constraint strategy for esterase-sensitive CO prodrugs.

Esterase is a ubiquitous enzyme, and is not specific to any organ or disease model. Therefore, the esterase-sensitive CO prodrugs do not offer targeted delivery. It is important to note that targeted delivery is not essential for using CO as a therapeutic agent because of its high safety margin and demonstrated efficacy after systemic delivery.4,85 However, in some application, targeted delivery of CO is desirable. Thus, Wang's lab has also prepared other CO prodrugs capable of targeting certain locations/tissues including pH-sensitive and reactive oxygen species-sensitive prodrugs.81,86,87 However, since they are technically not “click and release” prodrugs, they are not described in detail.

A cascade prodrug system for the co-delivery of CO and another drug payload

CO has been shown to exert synergistic or additive effects with anti-cancer, anti-inflammatory or anti-bacterial agents.67,88,89 Therefore, it would be highly desirable to develop a system for the co-delivery of CO and another drug payload. Actually, there have been several elegant examples for the co-delivery of CO and anti-inflammatory agents, and improved biological effects have been observed.89,90 Very recently, Wang and co-workers devised a general strategy for the co-delivery of CO and another drug payload using a single molecule, employing a cascade of bioorthogonal reactions (Fig. 19).91 In particular, the initial intramolecular DAinv led to the release of CO and an intermediate that undergoes lactonization to release a drug payload and a blue fluorescent by-product. The proof of concept study was conducted with metronidazole as the drug payload, and the release of CO and metronidazole was confirmed in vitro with a half-life of around 40 min in DMSO/PBS (4[thin space (1/6-em)]:[thin space (1/6-em)]1) at 37 °C. Most importantly, compound 56b showed a much more potent inhibitory effect against Helicobacter pylori with a MIC90 value of 0.28 μM, as compared to 14.6 μM for metronidazole, suggesting CO's ability to sensitize H. pylori to the treatment of metronidazole.
image file: c8cs00395e-f19.tif
Fig. 19 A cascade prodrug for CO and metronidazole.

A “click and release” system for sulfur dioxide prodrugs

Sulfur dioxide (SO2) has long been recognized as an air pollutant, and long-term exposure could cause severe damage to the respiratory system. However, SO2 is also produced endogenously by the metabolism of thiol-containing amino acids and hydrogen sulfide,92,93 which is another important gasotransmitter. Recently, mounting evidence has presented SO2 as another potential gasotransmitter, with known functions in the cardiovascular system.62 Additionally, it was also reported to possess a wide range of bioactivities, including anti-bacterial,94 anti-hypotensive,95 and anti-oxidative activities,96 among others. Several SO2 donors have been reported with different release mechanisms to enable the study of SO2's biological functions.97–100 This section only describes those that use “click and release” chemistry to afford controllable and tunable release rates.

A bimolecular “click and release” system for SO2 prodrugs

Borrowing the similar chemistry from bioorthogonal CO prodrugs, Wang and coworkers devised a series of prodrugs for SO2, employing the click reaction between a strained alkyne or alkene and thiophene-S-dioxide (Fig. 20).101 As expected, the “click and release” reaction occurred readily at room temperature. For example, the reaction between endo-BCN 62 and compound 60 (R1–R4 = Cl) was completed within 5 min with an initial concentration of 80 mM of 62 in MeOH at room temperature. SO2 release was confirmed by the elucidation of the cycloaddition products 64 and a well-accepted DTNB assay for the test of bisulfite derivative generation in 5% of DMSO/PBS (pH = 7.4). It is worth noting that a strained alkene (e.g. trans-cyclooctene) can also be used to trigger SO2 release from thiophene-S-dioxide. This feature is not the case in a CO prodrug system, where the initial cycloaddition product between a dienone and a strained alkene would not undergo further cheletropic reaction to extrude CO under physiological conditions. By varying the substituents on the thiophene-S-dioxide ring or using different dienophiles with different strain energies, the cycloaddition rate can be easily tuned, with the second order rate constants ranging from 0.05 to 1.5 M−1 s−1 under near physiological conditions (e.g. room temperature in MeOH, or 37 °C in DMSO/PBS). Interestingly, one SO2 prodrug 60a with a naphthalene group attached to the 3,4-positions of the thiophene-S-dioxide ring showed weak yellowish fluorescence with a quantum yield (Φ) of 0.009. Upon SO2 release after clicking with endo-BCN 62, the product (65) was fluorescent (blue, Φ = 0.14) (Fig. 21). Such a “click, release, and fluoresce” system is a very powerful tool for the real-time monitoring of the release of the payload (SO2 in this case) in a biological milieu.87 However, no biological data were reported for such SO2 prodrugs.
image file: c8cs00395e-f20.tif
Fig. 20 A biomolecular “click and release” strategy for SO2.

image file: c8cs00395e-f21.tif
Fig. 21 A “click, fluoresce and release” strategy for SO2 prodrugs.

A unimolecular “click and release” system for SO2 prodrugs

Although the bimolecular “click and release” strategy for SO2 prodrugs has the advantages of controllable and tunable release rates, such a system suffers from some inherent limitations. For example, just like bimolecular CO prodrugs, the delivery of the two click partners for in vivo study is a great challenge for such bimolecular SO2 prodrugs. Additionally, the ability to tune the reaction rate is dependent on the concentration of the reactants involved. Inspired by the unimolecular “click and release” system for CO prodrugs, Wang and co-workers designed a very similar unimolecular system for SO2 prodrugs by combining both the thiophene-S-dioxide and the alkyne into one molecule.102 In such a way, the delivery would be easier to handle for in vivo studies, and the release rate can be easily tuned by modifying entropic factors without imposing additional requirements on the reactivity of the thiophene-S-dioxide component. As expected, the unimolecular “click and release” reaction between thiophene-S-dioxide and the alkyne occurred under near physiological conditions (30% of DMSO/PBS at 37 °C, Fig. 22), despite the harsh conditions needed for the intermolecular reactions (e.g. reflux in toluene). SO2 release under near physiological conditions was confirmed by a DTNB assay, and the release rate can be easily tuned by modifying the tethering linker and the substituents on the linker, with the release half-lives ranging from minutes to days. In terms of the structure–release rate relationship, the same trends obtained in the unimolecular CO prodrug system were also observed here. For example, the amide tethering linker is more entropically favorable compared to the ester one, and the gem-dimethyl group substantially accelerates the intramolecular cycloaddition. One representative SO2 prodrug 66 (BW-SO2-106, R1 = R2 = Me, R3 = H, n = 1, X = NH, Fig. 22) demonstrated a pronounced SO2-associated DNA cleavage ability in a pBR322 supercoiled plasmid cleavage assay, and it also showed SO2 release in Raw264.7 cells by using a fluorescent SO2 probe. Altogether, such unimolecular SO2 prodrugs could serve as very powerful tools for studying the effect of the SO2 release rate on the observed biological effects.
image file: c8cs00395e-f22.tif
Fig. 22 A unimolecular “click and release” system for SO2 prodrugs.

A “click and release” system for hydrogen sulfide prodrugs

Hydrogen sulfide is recognized as one of the most important biological signaling molecules and has been shown to have great therapeutic potential in many models of diseases.103 Central to the progress in understanding the biology and clinical potential of hydrogen sulfide is the development of a wide array of hydrogen sulfide donors and prodrugs. Over the years, many H2S donors have been reported using different strategies.103–105 More recently, carbonyl sulfide (COS) has been shown to be a viable surrogate for H2S via hydrolysis by the ubiquitous carbonic anhydrase (CA).106,107 To achieve temporospatial control of COS release, Pluth and co-workers reported the first “click and release” COS/H2S donors, employing the click reaction between trans-cyclooctene and tetrazine.108 In this strategy, a benzylic thiocarbamate-functionalized trans-cyclooctene (68) is clicked with tetrazine 69 to produce a thiocarbamate-functionalized dihydropyridazine, 70, which then undergoes tautomerization (Fig. 23). Intermediate 71 is then poised to undergo self-immolative decomposition and subsequent re-aromatization to generate cyclooctylpyridazine 74, benzylamine 72, and COS. These products were confirmed with real-time mass spectrometry (DART-MS). The generation of H2S via hydrolysis of extruded COS in the presence of CA is validated using H2S-sensitive electrodes, and no H2S formation was observed in the absence of either CA or tetrazine. Dose-dependent H2S release was also observed upon mixing 68 (50 μM) with increasing concentration of tetrazine 69 (5–25 equiv.) in the presence of CA (25 μg mL−1). Furthermore, the H2S release profiles in complex biological media such as diluted whole sheep and bovine blood even without the addition of CA mirrored those of the solution behaviors. However, it was found that this “click and release” system was not compatible with cell-imaging assays using fluorescent probes for hydrogen sulfide such as HSN2, WSP-5, and SF7-Am. The reasons behind this outcome were thought to be associated with the slow release profiles of such a “click and release” system and the “H2S scavenging” properties of tetrazine, which is known to be reducible by H2S.109
image file: c8cs00395e-f23.tif
Fig. 23 Hydrogen sulfide generation triggered by a bioorthogonal click reaction.

Very recently, Taran and co-workers110 discovered several new click reaction pairs by screening the reactivity of a panel of mesoionic compounds towards strained alkynes (Fig. 24). As expected, compound 75 reacted readily with alkynes 78–80 at room temperature with concomitant extrusion of CO2. Surprisingly, compound 76 is also reactive towards alkynes 78–80, with the second reaction rate constants ranging from 0.008 to 29.2 M−1 s−1 by varying the Ar and R groups on 76 or employing different alkynes. In addition, RNH2 (for compound 76) is also readily released along with CO2 following the retro-DA reaction. The RNH2 could be the drug payload for “click and release.” Intriguingly, the click reaction between 77 and 78 led to the release of COS and acetamide in DMSO/PBS at room temperature. Such high-content release will be highly desirable for combinational therapy. The bioorthogonality of such click reactions was confirmed in vitro by a protein labelling/modifying assay.


image file: c8cs00395e-f24.tif
Fig. 24 The click and release between strained alkynes and mesoionic compounds.

In a follow-up study,111 Taran and co-workers conducted systematic studies of scaffold 76 to probe substituents’ effects on the reaction kinetics. It was found that with R1 being an amide or a sulfone group, X being a halide and R2 being an electron withdrawing group (EWG) a significantly increased reaction rate was observed between 76 and BCN 78 (Fig. 25). In addition, the positive effects of these substituents on the reaction kinetics were found to be cumulative/additive. For example, one analogue with R1, X and R2 being amide, Br and ester groups, respectively, exhibited a second reaction rate constant of 3.7 M−1 s−1 with 78 in 1% of DMSO/PBS at 25 °C. However, no real drug payload was appended to such a system and no further biological study was reported with those compounds. Nevertheless, those click reactions should be very useful for the design of bioorthogonal prodrugs.


image file: c8cs00395e-f25.tif
Fig. 25 Structure–reactivity relationships of iminosydnones with BCN.

Enrichment-triggered drug release

Bioorthogonal chemistry should be conducive to targeted delivery approaches as well by tethering to a vector. This has been successfully applied for the selective delivery of tumor imaging agents by using a pre-targeting strategy.112,113 In terms of drug delivery, Robillard and coworkers employed a similar strategy for the selective delivery of Dox to tumor cells by using the reaction between trans-cyclooctene and a tetrazine.29 Briefly, an antibody-Dox-trans-cyclooctene was prepared and administered to mice for pre-targeting. Then administration of a secondary reagent, tetrazine, initiated the release of Dox at the tumor site. There are several issues that need to be considered for such a drug delivery approach. First, such a strategy is most applicable when bioorthogonal reactions are fast (e.g. reactions between trans-cyclooctene and tetrazine). Sluggish bioorthogonal reactions (e.g. Staudinger ligation) are not considered ideal in this regard because of inefficient payload release at low concentrations. Second, after pre-targeting the tumor site with a clickable handle tethered to an antibody, there needs to be an appropriate, but sometimes long wait (e.g. several days) before dosing the secondary agent to selectively release the payload at the tumor site. Such a wait is for the drug conjugate to enrich at the desired site and yet to clear the general circulation. Alternatively, another “clearing” agent can be used to remove the unbound antibody from the general circulation. To address these issues, Wang and coworkers proposed a new concept of “enrichment-triggered prodrug activation” for the targeted delivery of bioorthogonal prodrugs.114 Such an approach allows for the simultaneous administration of the two click partners without compromising the specificity. Very importantly, such an “enrichment-triggered release” approach can also be applied to bioorthogonal reactions with moderate reaction rates. The rationale behind such a concept is to use reaction kinetics to drive the release. By conjugating both click partners with an appropriate targeting moiety, both can be enriched at the same site. Due to the slow reaction kinetics (e.g. K2 = 0.2 M−1 s−1), they would not react or would react very slowly at the circulating low concentration (e.g. low micromolar range). However, after both components are enriched at the target site, the local concentrations for both would increase dramatically. As a result, the reaction rate between these two components would increase to enable payload release at the target site (Fig. 26), and thus achieve “enrichment-triggered release.”
image file: c8cs00395e-f26.tif
Fig. 26 A schematic illustration of the concept of enrichment triggered drug release.

To establish the proof of concept, Wang and co-workers designed and synthesized compounds 81b and 82 with triphenylphosphonium (TPP) conjugation for the selective delivery of CO to mitochondria.114 The rationale underlying the choice of the TPP conjugation is two-fold. First, compounds with TPP conjugation are known to be substantially enriched in mitochondria sometimes by up to 1000-fold.115 Second, mounting evidence has firmly established mitochondria as the primary target of CO,116 and hence TPP conjugation allows for the targeted delivery of CO to its action site for improved biological outcome. Additionally, to facilitate the visualization of the intracellular click reaction, cyclopentadienone compounds with a naphthalene group attached at the 3,4-positions were employed to turn on the fluorescence upon CO release (Fig. 27). The introduction of the TPP moiety did not significantly affect the reaction rate between exo-BCN 48 and dienone 81a in 20% of DMSO/PBS at 37 °C (0.20 M−1 s−1 with TPP conjugation vs. 0.14 M−1 s−1 without TPP conjugation). Imaging studies showed that RAW264.7 cells co-treated with compounds 81b and 82 (with TPP conjugation) exhibited a dose-dependent increase in fluorescence intensity after 4 h incubation. In contrast, the co-treatment with compounds 48 and 81a (without TPP conjugation) did not present any fluorescence even at the highest concentration tested (5 μM). Assuming that both 81b and 82 were enriched in mitochondria to around 500 μM, the first half-life for the click reaction was determined to be around 2 h. However, in the case of 48 and 81a without enrichment, the first half-life was calculated to be over 100 h at an initial concentration of 5 μM. Therefore, it is not surprising that the cells treated with compounds with TPP conjugation showed strong blue florescence after 4 h of incubation, while the control group treated with 48 and 81a without TPP conjugation showed no fluorescence. Furthermore, a co-localization experiment with a mitochondrial tracker showed overlap of the fluorescence from the tracker and CO release product, consistent with CO release into mitochondria. More importantly, co-treatment with 81b and 82 showed significant TNF suppression in LPS challenged RAW264.7 cells with an IC50 value of 5 μM, which is the lowest value observed with all the organic CO prodrugs tested under the same protocols. Such results suggest that localized delivery improves potency. Meanwhile, co-treatment with 48 and 81a did not exert similar effects even at the highest concentration tested (10 μM). Altogether, these results fully demonstrated the concept of enrichment-triggered drug release and the significant benefits of targeting CO to mitochondria. In order to establish the applicability of such a concept for in vivo studies, Wang and co-workers tested these CO prodrugs in an acetaminophen (APAP)-induced acute liver injury mouse model. Co-treatment with 81b and 82 at a dosage of 0.4 mg kg−1 (tail vein i.v.) had significant protective effects on an APAP-induced liver injury as indicated by the suppression of the ALT level (58%) and recovery of inflammation injuries. However, the treatment with 48 and 81a showed no such effects under the same conditions. These results demonstrated the proof of concept in animal models. It can be envisioned that by using different targeting moieties and/or different bioorthogonal prodrugs, “on-demand” release of the payload with high specificity at the targeted disease site or organ can be readily achieved using this “enrichment triggered release” approach.


image file: c8cs00395e-f27.tif
Fig. 27 Enrichment activated bioorthogonal CO prodrugs targeting mitochondria.

Based on the same strategy, Wang's group developed a “click, cyclize and release” system for targeted drug delivery.114 Briefly, the system contains a tetrazine prodrug and a cyclooctyne trigger (Fig. 28), and the click reaction could generate intermediates 89a and 89b, where 89b is a dominant product, and could undergo further lactonization to release the parent drug. The reactivity of the click pairs and the ability to release a drug were studied by using different alkynes (83–85) to react with dansyl amino prodrugs (87) and a Dox-prodrug (88), respectively. It was found that the second order rate constants could be tuned from 0.0075 M−1 s−1 to 2 M−1 s−1 under near physiological conditions by choosing different alkynes, and more than 80% of the parent drug could be released using the system. In a cell culture study, the EC50 of the click pair (tetrazine prodrug 88 and alkyne 83) was about 1.2 μM, which is similar to that of its parent drug Dox itself (1.0 μM), and the EC50 of Dox-prodrug 88 was more than 100 μM. By using this system, the authors also demonstrated the “enrichment triggered release” strategy to address the challenge of the two component drug administration process. In particular, both the tetrazine prodrug and alkyne trigger were conjugated with a TPP moiety, which would help enrichment in mitochondria.117 Then cancer cells were treated with a TPP-containing click pair (85b and 91) or negative controls without TPP conjugation (85a and 88), respectively. The EC50 of the TPP conjugate click pair was found to be about 2 μM and that of the click pair without TPP conjugation was more than 10 μM. Such results further confirmed enrichment-triggered release in mitochondria. The same concept could be used in other enrichment-based drug delivery systems.


image file: c8cs00395e-f28.tif
Fig. 28 “Click, cyclize and release” systems.

Conclusions and perspectives

The premature release of the payload before reaching the desired site presents a major issue with prodrugs including both small molecules and ADCs. Along this line, prodrug derivatization chemistry including linker chemistry for ADCs plays a crucial role. Recent advances in the design of bioorthogonal prodrugs greatly complement traditional prodrug activation strategies, wherein the release of the payload is elicited by a click reaction between the prodrug and its click partner. Such a prodrug strategy holds great promise in ADCs, and could be superior to traditional ones in certain applications because of the bioorthogonality of the linker chemistry. In addition, “enrichment triggered release” or “activation by enrichment” affords extra safeguards against pre-mature release. However, this field of using bioorthogonal chemistry in prodrug applications is still in its infancy. Many challenges remain and there is much more work to be done collectively. Generally speaking, pharmacokinetic (PK) profiles, the stability of the “triggering” agents, and synthetic complexity are all issues to be tackled. In terms of PK profiles for approaches that use two components, the ability to synchronize or sequence with defined time delays of the two components is not a trivial issue. This is true in almost all two-component cases, including the “enrichment-triggered” release approach. Additionally, understanding the need for the optimal relative concentrations of the two components is also very important. Because of the introduction of the second component as a trigger, it is also important to understand its safety profile and off-target effect. This issue is especially prominent with TPP-conjugates because of the known issues with a large and hydrophobic TPP group. All in all, the inclusion of the secondary component as a trigger significantly complicates issues related to adsorption, distribution, metabolism, excretion, toxicity, and safety. Thus, there is a great deal of promise in using bioorthogonal chemistry in drug delivery, and yet there are many more challenges ahead. We hope that the publication of this review will stimulate more interest in this very promising field and brings new blood to collectively advance this field.

After the submission of this manuscript, several elegant bioorthogonal prodrugs with new chemistry have been reported. Such publications further enrich the literature in this area and indicate the increasing level of interest in this field.118–121

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The work in the Wang lab was partially supported by Georgia State University through internal fellowships (a Brains & Behavior fellowship to BY, a Center for Diagnostics and Therapeutics fellowship to LKDLC, and a Molecular Basis of Disease fellowship to ZXP), and a Georgia Research Alliance endowment established at Georgia State University (BW). The work in the lab of Bowen Ke was supported by the 1.3.5 Project for Disciplines of Excellence, West China Hospital, Sichuan University (No. ZYJC18032, ZY2016101 and ZY2016203).

Notes and references

  1. Pharmaceutical Profiling in Drug Discovery for Lead Selection, ed. R. T. Borchardt, E. H. Kerns, C. A. Lipinski, D. R. Thakker and B. Wang, AAPS Press, Arlington, VA, 2004 Search PubMed .
  2. Evaluation of Drug Candidates for Preclinical Development: Pharmacokinetics, Metabolism, Pharmaceutics, and Toxicology, ed. C. Han, C. David and B. Wang, John Wiley & Sons, New York, 2010 Search PubMed .
  3. C. Han and B. Wang, in Drug Delivery: Principles and Applications, ed. B. Wang, L. Hu and T. Siahaan, John Wiley & Sons, Hoboken, New Jersey, 2nd edn, 2016, pp. 1–15 Search PubMed .
  4. X. Ji, K. Damera, Y. Zheng, B. Yu, L. E. Otterbein and B. Wang, J. Pharm. Sci., 2016, 105, 406–416 CrossRef CAS PubMed .
  5. M. H. Lee, J. L. Sessler and J. S. Kim, Acc. Chem. Res., 2015, 48, 2935–2946 CrossRef CAS PubMed .
  6. J. J. Yeh and W. Y. Kim, J. Clin. Oncol., 2015, 33, 1505–1508 CrossRef CAS PubMed .
  7. X. Peng and V. Gandhi, Ther. Delivery, 2012, 3, 823–833 CrossRef CAS .
  8. G. Saravanakumar, J. Kim and W. J. Kim, Adv. Sci., 2017, 4, 1600124 CrossRef PubMed .
  9. J. Liu, Y. Pang, Z. Zhu, D. Wang, C. Li, W. Huang, X. Zhu and D. Yan, Biomacromolecules, 2013, 14, 1627–1636 CrossRef CAS PubMed .
  10. M. Rooseboom, J. N. M. Commandeur and N. P. E. Vermeulen, Pharmacol. Rev., 2004, 56, 53–102 CrossRef CAS PubMed .
  11. J. Li and P. R. Chen, Nat. Chem. Biol., 2016, 12, 129–137 CrossRef CAS PubMed .
  12. S. V. Chankeshwara, E. Indrigo and M. Bradley, Curr. Opin. Chem. Biol., 2014, 21, 128–135 CrossRef CAS PubMed .
  13. T. Völker and E. Meggers, Curr. Opin. Chem. Biol., 2015, 25, 48–54 CrossRef PubMed .
  14. A. Dumas and P. Couvreur, Chem. Sci., 2015, 6, 2153–2157 RSC .
  15. J. Li, J. Yu, J. Zhao, J. Wang, S. Zheng, S. Lin, L. Chen, M. Yang, S. Jia, X. Zhang and P. R. Chen, Nat. Chem., 2014, 6, 352–361 CrossRef CAS PubMed .
  16. J. Clavadetscher, E. Indrigo, S. V. Chankeshwara, A. Lilienkampf and M. Bradley, Angew. Chem., Int. Ed., 2017, 56, 6864–6868 CrossRef CAS PubMed .
  17. M. A. Miller, B. Askevold, H. Mikula and R. H. Kohler, Nat. Commun., 2017, 8, 15906 CrossRef CAS PubMed .
  18. A. M. Pérez-López, R. R. Belén, S. Víctor, H. Lloyd, A. Catherine, T. L. Bray, I. Silvia, P. M. Brennan, G. C. Lloyd-Jones, S. Dirk, S. Jesús and U. B. Asier, Angew. Chem., Int. Ed., 2017, 56, 12548–12552 CrossRef PubMed .
  19. E. Winquist, T. Waldron, S. Berry, D. S. Ernst, S. Hotte and H. Lukka, BMC Cancer, 2006, 6, 112 CrossRef PubMed .
  20. Y. Ichikawa, M. Ghanefar, M. Bayeva, R. Wu, A. Khechaduri, S. V. N. Prasad, R. K. Mutharasan, T. J. Naik and H. Ardehali, J. Clin. Invest., 2014, 124, 617–630 CrossRef CAS PubMed .
  21. M. Köhn and R. Breinbauer, Angew. Chem., Int. Ed., 2004, 43, 3106–3116 CrossRef PubMed .
  22. B. L. Oliveira, Z. Guo and G. J. L. Bernardes, Chem. Soc. Rev., 2017, 46, 4895–4950 RSC .
  23. D. M. Patterson, L. A. Nazarova and J. A. Prescher, ACS Chem. Biol., 2014, 9, 592–605 CrossRef CAS PubMed .
  24. M. D. Best, Biochemistry, 2009, 48, 6571–6584 CrossRef CAS PubMed .
  25. J. M. Baskin and C. R. Bertozzi, QSAR Comb. Sci., 2007, 26, 1211–1219 CrossRef CAS .
  26. P. Thirumurugan, D. Matosiuk and K. Jozwiak, Chem. Rev., 2013, 113, 4905–4979 CrossRef CAS PubMed .
  27. M. H. Wright and S. A. Sieber, Nat. Prod. Rep., 2016, 33, 681–708 RSC .
  28. M. Azoulay, G. Tuffin, W. Sallem and J. C. Florent, Bioorg. Med. Chem. Lett., 2006, 16, 3147–3149 CrossRef CAS PubMed .
  29. R. v. Brakel, R. C. M. Vulders, R. J. Bokdam, H. Grüll and M. S. Robillard, Bioconjugate Chem., 2008, 19, 714–718 CrossRef PubMed .
  30. M. B. Soellner, B. L. Nilsson and R. T. Raines, J. Am. Chem. Soc., 2006, 128, 8820–8828 CrossRef CAS PubMed .
  31. M. Mondal, R. Liao, L. Xiao, T. Eno and J. Guo, Angew. Chem., Int. Ed., 2017, 56, 2636–2639 CrossRef CAS PubMed .
  32. A. S. Cohen, E. A. Dubikovskaya, J. S. Rush and C. R. Bertozzi, J. Am. Chem. Soc., 2010, 132, 8563–8565 CrossRef CAS PubMed .
  33. S. S. Matikonda, D. L. Orsi, V. Staudacher, I. A. Jenkins, F. Fiedler, J. Chen and A. B. Gamble, Chem. Sci., 2015, 6, 1212–1218 RSC .
  34. K. J. Shea and J. S. Kim, J. Am. Chem. Soc., 1992, 114, 4846–4855 CrossRef CAS .
  35. R. Sustmann and H. Trill, Angew. Chem., Int. Ed., 1972, 11, 838–840 CrossRef CAS .
  36. S. S. Matikonda, J. M. Fairhall, F. Fiedler, S. Sanhajariya, R. A. J. Tucker, S. Hook, A. L. Garden and A. B. Gamble, Bioconjugate Chem., 2018, 29, 324–334 CrossRef CAS PubMed .
  37. J. Luo, Q. Liu, K. Morihiro and A. Deiters, Nat. Chem., 2016, 8, 1027–1034 CrossRef CAS PubMed .
  38. Y. Ge, X. Fan and P. R. Chen, Chem. Sci., 2016, 7, 7055–7060 RSC .
  39. M. L. Blackman, M. Royzen and J. M. Fox, J. Am. Chem. Soc., 2008, 130, 13518–13519 CrossRef CAS PubMed .
  40. M. T. Taylor, M. L. Blackman, O. Dmitrenko and J. M. Fox, J. Am. Chem. Soc., 2011, 133, 9646–9649 CrossRef CAS PubMed .
  41. W. Chen, D. Wang, C. Dai, D. Hamelberg and B. Wang, Chem. Commun., 2012, 48, 1736–1738 RSC .
  42. R. M. Versteegen, R. Rossin, W. ten Hoeve, H. M. Janssen and M. S. Robillard, Angew. Chem., Int. Ed., 2013, 52, 14112–14116 CrossRef CAS PubMed .
  43. J. Li, S. Jia and P. R. Chen, Nat. Chem. Biol., 2014, 10, 1003–1005 CrossRef CAS PubMed .
  44. G. Zhang, J. Li, R. Xie, X. Fan, Y. Liu, S. Zheng, Y. Ge and P. R. Chen, ACS Cent. Sci., 2016, 2, 325–331 CrossRef CAS PubMed .
  45. L. Liu, Y. Liu, G. Zhang, Y. Ge, X. Fan, F. Lin, J. Wang, H. Zheng, X. Xie, X. Zeng and P. R. Chen, Biochemistry, 2018, 57, 446–450 CrossRef CAS PubMed .
  46. X. Fan, Y. Ge, F. Lin, Y. Yang, G. Zhang, W. S. Ngai, Z. Lin, S. Zheng, J. Wang, J. Zhao, J. Li and P. R. Chen, Angew. Chem., Int. Ed., 2016, 55, 14046–14050 CrossRef CAS PubMed .
  47. J. C. T. Carlson, H. Mikula and R. Weissleder, J. Am. Chem. Soc., 2018, 140, 3603–3612 CrossRef CAS PubMed .
  48. I. Khan, P. F. Agris, M. V. Yigit and M. Royzen, Chem. Commun., 2016, 52, 6174–6177 RSC .
  49. R. Rossin, S. M. J. van Duijnhoven, W. ten Hoeve, H. M. Janssen, L. H. J. Kleijn, F. J. M. Hoeben, R. M. Versteegen and M. S. Robillard, Bioconjugate Chem., 2016, 27, 1697–1706 CrossRef CAS PubMed .
  50. R. Rossin, T. Lappchen, S. M. van den Bosch, R. Laforest and M. S. Robillard, J. Nucl. Med., 2013, 54, 1989–1995 CrossRef CAS PubMed .
  51. R. Rossin, R. M. Versteegen, J. Wu, A. Khasanov, H. J. Wessels, E. J. Steenbergen, W. Ten Hoeve, H. M. Janssen, A. van Onzen, P. J. Hudson and M. S. Robillard, Nat. Commun., 2018, 9, 1484 CrossRef PubMed .
  52. J. M. Mejia Oneto, I. Khan, L. Seebald and M. Royzen, ACS Cent. Sci., 2016, 2, 476–482 CrossRef CAS PubMed .
  53. E. Jimenez-Moreno, Z. Guo, B. L. Oliveira, I. S. Albuquerque, A. Kitowski, A. Guerreiro, O. Boutureira, T. Rodrigues, G. Jimenez-Oses and G. J. Bernardes, Angew. Chem., Int. Ed., 2016, 56, 243–247 CrossRef PubMed .
  54. J. M. Palomo, Eur. J. Org. Chem., 2010, 6303–6314 CrossRef CAS .
  55. K. Neumann, S. Jain, A. Gambardella, S. E. Walker, E. Valero, A. Lilienkampf and M. Bradley, ChemBioChem, 2016, 18, 91–95 CrossRef PubMed .
  56. M. Xu, J. Tu and R. M. Franzini, Chem. Commun., 2017, 53, 6271–6274 RSC .
  57. H. Wu, B. T. Cisneros, C. M. Cole and N. K. Devaraj, J. Am. Chem. Soc., 2014, 136, 17942–17945 CrossRef CAS PubMed .
  58. M. Xu, R. Galindo-Murillo, T. E. Cheatham and R. M. Franzini, Org. Biomol. Chem., 2017, 15, 9855–9865 RSC .
  59. P. C. Petit, D. H. Fine, G. B. Vásquez, L. Gamero, M. S. Slaughter and K. A. Dasse, ASAIO J., 2017, 63, 7–13 CrossRef CAS PubMed .
  60. N. Sepehrvand and J. A. Ezekowitz, JACC: Heart Failure, 2016, 4, 783–790 Search PubMed .
  61. R. Wang, Trends Biochem. Sci., 2014, 39, 227–232 CrossRef CAS PubMed .
  62. Y. Huang, C. Tang, J. Du and H. Jin, Oxid. Med. Cell. Longevity, 2016, 8961951 Search PubMed .
  63. R. Motterlini and L. E. Otterbein, Nat. Rev. Drug Discovery, 2010, 9, 728–743 CrossRef CAS PubMed .
  64. M. Klinger-Strobel, S. Glaser, O. Makarewicz, R. Wyrwa, J. Weisser, M. W. Pletz and A. Schiller, Antimicrob. Agents Chemother., 2016, 60, 4037–4046 CrossRef CAS PubMed .
  65. L. E. Otterbein, F. H. Bach, J. Alam, M. Soares, L. H. Tao, M. Wysk, R. J. Davis, R. A. Flavell and A. M. Choi, Nat. Med., 2000, 6, 422–428 CrossRef CAS PubMed .
  66. A. Zhang, L. Zhao, N. Li, H. Duan, H. Liu, F. Pu, G. Zhang, E. M. Zhou and S. Xiao, J. Virol., 2017, 91, e01866 CAS .
  67. B. Wegiel, D. Gallo, E. Csizmadia, C. Harris, J. Belcher, G. M. Vercellotti, N. Penacho, P. Seth, V. Sukhatme, A. Ahmed, P. P. Pandolfi, L. Helczynski, A. Bjartell, J. L. Persson and L. E. Otterbein, Cancer Res., 2013, 73, 7009–7021 CrossRef CAS PubMed .
  68. X. Ji and B. Wang, Acc. Chem. Res., 2018, 51, 1377–1385 CrossRef CAS PubMed .
  69. S. H. Heinemann, T. Hoshi, M. Westerhausen and A. Schiller, Chem. Commun., 2014, 50, 3644–3660 RSC .
  70. C. C. Romao, W. A. Blattler, J. D. Seixas and G. J. Bernardes, Chem. Soc. Rev., 2012, 41, 3571–3583 RSC .
  71. E. Palao, T. Slanina, L. Muchova, T. Solomek, L. Vitek and P. Klan, J. Am. Chem. Soc., 2016, 138, 126–133 CrossRef CAS PubMed .
  72. P. Peng, C. Wang, Z. Shi, V. K. Johns, L. Ma, J. Oyer, A. Copik, R. Igarashi and Y. Liao, Org. Biomol. Chem., 2013, 11, 6671–6674 RSC .
  73. S. N. Anderson, J. M. Richards, H. J. Esquer, A. D. Benninghoff, A. M. Arif and L. M. Berreau, ChemistryOpen, 2015, 4, 590–594 CrossRef CAS PubMed .
  74. T. Soboleva, H. J. Esquer, A. D. Benninghoff and L. M. Berreau, J. Am. Chem. Soc., 2017, 139, 9435–9438 CrossRef CAS PubMed .
  75. E. Gagnon, S. D. Halperin, V. Métivaud, K. E. Maly and J. D. Wuest, J. Org. Chem., 2010, 75, 399–406 CrossRef CAS PubMed .
  76. D. Wang, E. Viennois, K. Ji, K. Damera, A. Draganov, Y. Zheng, C. Dai, D. Merlin and B. Wang, Chem. Commun., 2014, 50, 15890–15893 RSC .
  77. M. A. Ogliaruso, M. G. Romanelli and E. I. Becker, Chem. Rev., 1965, 65, 261–367 CrossRef CAS .
  78. X. Ji, C. Zhou, K. Ji, R. E. Aghoghovbia, Z. Pan, V. Chittavong, B. Ke and B. Wang, Angew. Chem., Int. Ed., 2016, 55, 15846–15851 CrossRef CAS PubMed .
  79. M. Correa-Costa, D. Gallo, E. Csizmadia, E. Gomperts, J. L. Lieberum, C. J. Hauser, X. Ji, B. Wang, N. O. S. Camara, S. C. Robson and L. E. Otterbein, Proc. Natl. Acad. Sci. U. S. A., 2018, 115, E2302–E2310 CrossRef CAS PubMed .
  80. S. García-Gallego and G. J. L. Bernardes, Angew. Chem., Int. Ed., 2014, 53, 9712–9721 CrossRef PubMed .
  81. Z. Pan, V. Chittavong, W. Li, J. Zhang, K. Ji, M. Zhu, X. Ji and B. Wang, Chem. – Eur. J., 2017, 23, 9838–9845 CrossRef CAS PubMed .
  82. C. Hansch, A. Leo and R. W. Taft, Chem. Rev., 1991, 91, 165–195 CrossRef CAS .
  83. X. Ji, K. Ji, V. Chittavong, B. Yu, Z. Pan and B. Wang, Chem. Commun., 2017, 53, 8296–8299 RSC .
  84. B. W. Michel, A. R. Lippert and C. J. Chang, J. Am. Chem. Soc., 2012, 134, 15668–15671 CrossRef CAS PubMed .
  85. R. Motterlini and L. E. Otterbein, Nat. Rev. Drug Discovery, 2010, 9, 728–743 CrossRef CAS PubMed .
  86. Z. Pan, J. Zhang, K. Ji, V. Chittavong, X. Ji and B. Wang, Org. Lett., 2018, 20, 8–11 CrossRef CAS PubMed .
  87. X. Ji, K. Ji, V. Chittavong, R. E. Aghoghovbia, M. Zhu and B. Wang, J. Org. Chem., 2017, 82, 1471–1476 CrossRef CAS PubMed .
  88. A. F. Tavares, M. R. Parente, M. C. Justino, M. Oleastro, L. S. Nobre and L. M. Saraiva, PLoS One, 2013, 8, e83157 CrossRef PubMed .
  89. A. Nikam, A. Ollivier, M. Rivard, J. L. Wilson, K. Mebarki, T. Martens, J.-L. Dubois-Randé, R. Motterlini and R. Foresti, J. Med. Chem., 2016, 59, 756–762 CrossRef CAS PubMed .
  90. B. Bauer, A.-L. Göderz, H. Braumüller, J. M. Neudörfl, M. Röcken, T. Wieder and H.-G. Schmalz, ChemMedChem, 2017, 12, 1927–1930 CrossRef CAS PubMed .
  91. L. K. C. De La Cruz, S. Benoit, Z. Pan, B. Yu, R. J. Maier, X. Ji and B. Wang, Org. Lett., 2018, 20, 897–900 CrossRef CAS PubMed .
  92. H. Mitsuhashi, F. Ota, K. Ikeuchi, Y. Kaneko, T. Kuroiwa, K. Ueki, Y. Tsukada and Y. Nojima, Tohoku J. Exp. Med., 2002, 198, 125–132 CrossRef CAS PubMed .
  93. H. Mitsuhashi, S. Yamashita, H. Ikeuchi, T. Kuroiwa, Y. Kaneko, K. Hiromura, K. Ueki and Y. Nojima, Shock, 2005, 24, 529–534 CrossRef CAS PubMed .
  94. S. R. Malwal, D. Sriram, P. Yogeeswari, V. B. Konkimalla and H. Chakrapani, J. Med. Chem., 2012, 55, 553–557 CrossRef CAS PubMed .
  95. Y. Sun, Y. Tian, M. Prabha, D. Liu, S. Chen, R. Zhang, X. Liu, C. Tang, X. Tang, H. Jin and J. Du, Lab. Invest., 2010, 90, 68–82 CrossRef CAS PubMed .
  96. S. Chen, S. Zheng, Z. Liu, C. Tang, B. Zhao, J. Du and H. Jin, Lab. Invest., 2015, 95, 142–156 CrossRef CAS PubMed .
  97. S. R. Malwal, M. Gudem, A. Hazra and H. Chakrapani, Org. Lett., 2013, 15, 1116–1119 CrossRef CAS PubMed .
  98. S. R. Malwal and H. Chakrapani, Org. Biomol. Chem., 2015, 13, 2399–2406 RSC .
  99. J. J. Day, Z. Yang, W. Chen, A. Pacheco and M. Xian, ACS Chem. Biol., 2016, 11, 1647–1651 CrossRef CAS PubMed .
  100. K. A. Pardeshi, G. Ravikumar and H. Chakrapani, Org. Lett., 2018, 20, 4–7 CrossRef CAS PubMed .
  101. W. Wang, X. Ji, Z. Du and B. Wang, Chem. Commun., 2016, 53, 1370–1373 RSC .
  102. X. Ji, E. M. El-labbad, K. Ji, D. S. Lasheen, R. A. T. Serya, K. A. Abouzid and B. Wang, Org. Lett., 2017, 19, 818–821 CrossRef CAS PubMed .
  103. Y. Zheng, X. Ji, K. Ji and B. Wang, Acta Pharm. Sin. B, 2015, 5, 367–377 CrossRef PubMed .
  104. M. D. Hartle and M. D. Pluth, Chem. Soc. Rev., 2016, 45, 6108–6117 RSC .
  105. Y. Zheng, B. Yu, L. K. De La Cruz, M. Roy Choudhury, A. Anifowose and B. Wang, Med. Res. Rev., 2018, 38, 57–100 CrossRef CAS PubMed .
  106. Y. Zhao and M. D. Pluth, Angew. Chem., Int. Ed., 2016, 55, 14638–14642 CrossRef CAS PubMed .
  107. C. R. Powell, J. C. Foster, B. Okyere, M. H. Theus and J. B. Matson, J. Am. Chem. Soc., 2016, 138, 13477–13480 CrossRef PubMed .
  108. A. K. Steiger, Y. Yang, M. Royzen and M. D. Pluth, Chem. Commun., 2016, 53, 1378–1380 RSC .
  109. Z. Zhao, L. Cao, T. Zhang, R. Hu, S. Wang, S. Li, Y. Li and G. Yang, ChemistrySelect, 2016, 1, 2581–2585 CrossRef CAS .
  110. S. Bernard, D. Audisio, M. Riomet, S. Bregant, A. Sallustrau, L. Plougastel, E. Decuypere, S. Gabillet, R. A. Kumar, J. Elyian, M. N. Trinh, O. Koniev, A. Wagner, S. Kolodych and F. Taran, Angew. Chem., Int. Ed., 2017, 56, 15612–15616 CrossRef CAS PubMed .
  111. R. Margaux, D. Elodie, P. Karine, B. Sabrina, P. Lucie, K. Sergii, A. Davide and T. Frédéric, Chem. – Eur. J., 2018, 24, 8535–8541 CrossRef PubMed .
  112. J. C. Knight and B. Cornelissen, Am. J. Nucl. Med. Mol. Imaging, 2014, 4, 96–113 CAS .
  113. R. Rossin and M. S. Robillard, Curr. Opin. Chem. Biol., 2014, 21, 161–169 CrossRef CAS PubMed .
  114. Y. J. Zheng, X. Ji, B. Yu, K. Ji, D. Gallo, E. Csizmadia, M. Zhu, M. Roy Choudhury, L. De La Cruz, V. Chittavong, Z. Pan, Z. Yuan, L. E. Otterbein and B. Wang, Nat. Chem., 2018, 10, 787–794 CrossRef CAS PubMed .
  115. J. Zielonka, J. Joseph, A. Sikora, M. Hardy, O. Ouari, J. Vasquez-Vivar, G. Cheng, M. Lopez and B. Kalyanaraman, Chem. Rev., 2017, 117, 10043–10120 CrossRef CAS PubMed .
  116. C. S. F. Queiroga, A. S. Almeida and H. L. A. Vieira, Biochem. Res. Int., 2012, 749845 Search PubMed .
  117. R. A. Smith, C. M. Porteous, A. M. Gane and M. P. Murphy, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 5407–5412 CrossRef CAS PubMed .
  118. J. Tu, M. Xu, S. Parvez, R. T. Peterson and R. M. Franzini, J. Am. Chem. Soc., 2018, 140, 8410–8414 CrossRef CAS PubMed .
  119. K. Neumann, A. Gambardella, A. Lilienkampf and M. Bradley, Chem. Sci., 2018, 9, 7198–7203 RSC .
  120. Z. Shao, W. Liu, H. Tao, F. Liu, R. Zeng, P. A. Champagne, Y. Cao, K. N. Houk and Y. Liang, Chem. Commun., 2018, 54, 14089–14092 RSC .
  121. Q. Yao, F. Lin, X. Fan, Y. Wang, Y. Liu, Z. Liu, X. Jiang, P. Chen and Y. Gao, Nat. Commun., 2018, 9, 5032 CrossRef PubMed .

This journal is © The Royal Society of Chemistry 2019