Tetrazine‐Triggered Release of Carboxylic‐Acid‐Containing Molecules for Activation of an Anti‐inflammatory Drug

In addition to its use for the study of biomolecules in living systems, bioorthogonal chemistry has emerged as a promising strategy to enable protein or drug activation in a spatially and temporally controlled manner. This study demonstrates the application of a bioorthogonal inverse electron‐demand Diels–Alder (iEDDA) reaction to cleave trans‐cyclooctene (TCO) and vinyl protecting groups from carboxylic acid‐containing molecules. The tetrazine‐mediated decaging reaction proceeded under biocompatible conditions with fast reaction kinetics (<2 min). The anti‐inflammatory activity of ketoprofen was successfully reinstated after decaging of the nontoxic TCOprodrug in live macrophages. Overall, this work expands the scope of functional groups and the application of decaging reactions to a new class of drugs.

In addition to its use for the study of biomolecules in living systems, bioorthogonal chemistry has emerged as ap romising strategyt oe nable protein or drug activation in as patially and temporally controlled manner.T his study demonstrates the application of ab ioorthogonal inverse electron-demand Diels-Alder (iEDDA) reaction to cleave trans-cyclooctene (TCO) and vinyl protecting groups from carboxylic acid-containing molecules. The tetrazine-mediated decaging reactionp roceeded under biocompatiblec onditions with fast reaction kinetics (< 2min). The anti-inflammatory activity of ketoprofen was successfully reinstated after decaging of the nontoxic TCOprodrug in live macrophages. Overall, this work expandst he scope of functional groups and the application of decaging reactions to an ew class of drugs.
Early research in the fieldo fb ioorthogonal chemistry focused on ligation reactions such as the Staudinger reaction, [1] coppercatalysed azide-alkyne 1,3-dipolar cycloaddition (CuAAC), [2,3] palladium-catalysed cross-couplings, [4] ruthenium-catalysed olefin metatheses, [5] strain-promoted azide-alkyne cycloaddition (SPAAC), [6] tetrazole photoinduced 1,3-dipolar cycloadditions, [7,8] and inverse electron-demandD iels-Alder (iEDDA) tetrazine ligation. [9,10] Of these, the iEDDA reactionb etween transcyclooctene (TCO) and at etrazine is one of the more selective and fastestb ioorthogonal reactions to date. [11] Since it was first introduced by Fox et al., [9] this reactionh as been used in numerous biological applicationss uch as cell and in vivo pretargeting imaging. [12][13][14] Recently,b ioorthogonal cleavage reactions have emerged as promising strategies to control the activation of caged proteins,f luorophores, and small-molecule drugs in living systems. [15] The TCO-tetrazine iEDDA ligation can be re-engineered into ac leavage reaction by placing a leaving group at the allylic positiono fT CO. After the initial cycloaddition and elimination of nitrogen, the 4,5-dihydropyridazine now contains an appropriatelyp laced substituent that eliminates upon tautomerisation. [10] Robillard'sg roup reported the first use of the TCO-tetrazine reaction for bioorthogonal decaging to release amine-containing drugs ( Figure 1A), in which they demonstrated the release of doxorubicin (Dox) from aT CO carbamate prodrug in vitro. [16] They then applied this "click-to-release" strategy to successfullytrigger the release of Dox and monomethyl auristatin E( MMAE) from an antibody-drug conjugate (ADC). [17,18] Mejia Oneto and co-workers also reported targeted in vivo activation of aD ox-TCO carbamate prodrug by injecting an alginate hydrogel modified with tetrazines near the tumour site. [19] Al imitation of the click-toreleases trategy is the need for delivery,a nd therefore optimisation of the pharmacokinetic properties, of both the prodrug and the tetrazine. [20,21] However,t he previously mentioned approaches demonstrate the potential of bioorthogonal decaging reactions for targeted drug activation in vivo.
Bioorthogonal chemistry has also been appliedf or the release of alcohols. Our group [22] and the groups of Bradley [23] and Devaraj [24] independently reported using the vinyl ether protecting group, which could be cleaved by reaction with tetrazines to release alcohols ( Figure 1B). This is, however,s ignificantly slower than the TCO reaction for the release of amines. In addition, Robillard recently reported bioorthogonal cleavage of ethers, carbonates and esters from TCO to release alcohols ( Figure 1C)o rc arboxylic acids ( Figure 1D), respectively.H owever,t he reported TCO-protected carboxylic acids proved highlyu nstable ( % 90 %f ragmentation in 50 %m ouse serum at 37 8C). In addition, the ether linker was only successfully used to deprotect tyrosine and control cell growth in tyrosine-free medium. [25] Previous examples of drug releaseh ave so far been limited to the release of amine-( Figure 1A)o ra lcohol-containing (Figure 1B,C )a nticancer drugs.T hese groups, although often found in small-molecule drugs, are not alwaysp resent and might notb ev ital for the function of the drug;t his means that chemical modification at this site to form ap rodrug does not lead to reduced activity.F or this reason, it is important to [ expand the scope of bioorthogonal decaging reactions to include other functional groups. For example, nonsteroidal antiinflammatory drugs (NSAIDs) are an important class of drugs that contain ac arboxylic acidg roup essential for their function. [26] In our work, we expanded bioorthogonal cleavage reactions to the carboxylic acid functional group. As table TCO-protectedN SAID was successfully decaged in the presence of tetrazinew ithin 2min ( Figure 1E), thus reinstating the anti-inflammatory activity in living macrophages. Initially,f ollowing on from previous work in our group on the vinyl ether handleo na lcohols, [22] we investigated the protection of carboxylic acids with av inyl group. Computational studies on the reaction between vinyl acetate and different tetrazines (1-5;F igure 2A)p redicted that the first cycloaddition step was rate-determining ( Figure S20 in the Supporting Information), and that all tetrazines should have similar reactivity except dimethyltetrazine (5), which wasp redicted to be the least reactive (Tables S1 and S3). Thek inetics of the cycloaddition were experimentally determined with these tetrazines and the test substrate vinyl propionate 6 (Figures S1 and 2B). The fastest rate occurredw ith tetrazine 4,amonosubstituted tetrazine bearingamoderately electron-withdrawing group (benzoic acid, Figure 2B). It has previously been shown that tetrazines bearings trong electron-withdrawing substituents have faster rates for cycloadditions, whereas as mall, non-bulky group increasest he rate of the elimination step. [27] Next, the stability of tetrazines 1-5 in 50 %D MSO/H 2 Ow as assessed by monitoring the UV absorbance at 530 nm ( Figure S2). Te trazine 1 showed moderate stability (t 1/2 = 15.8 h), and tetrazine 2 was the most unstable( t 1/2 = 5.7 h). Te trazines 3-5 provedh ighly stable ( % 85-90% intact after 24 h, FigureS2). The biological stability of tetrazine 4 was then assessed, and it proved to be stable in cell culture medium, phosphate-buffered saline (PBS, pH 7.4),   Figure S3). Therefore, we decided to use tetrazine 4 in furtherstudies.
Ketoprofen (8)i saNSAID with ac hiral centre. Although it is used as ar acemate, the anti-inflammatory activity of ketoprofen mainly resides with the S enantiomer.A lthough the R enantiomer is approximately 100 to 1000 times less potent than the S enantiomer as ac yclooxygenasei nhibitor,r esearch has shown that the R enantiomer is still important in that it contributest ot he analgesic effect of ketoprofen. [28] Using palladium coupling, [29] we converted ketoprofen into the vinyl ester 9,w hich proved stable in PBS (pH 7.4;F igure S4). However,l imited stability (as assessed by HPLC) was observed in 10 % plasma (t 1/2 = 12 min) and cell culture medium( t 1/2 = 4h)p robably because the vinyl group does not sterically protect the ester group from nucleophilic attack and subsequenth ydrolysis. This fact, along with the slow rate of reaction( ca. 20 %o f free drug observed after 24 hw ith 30 equiv of tetrazine, Figure S5) resulted in the vinyl handle being abandoned, as its use for in vivoapplications would be rather limited.
Next, we decided to investigate TCO as ac aging group for ketoprofen. Am ore reactive alkene was necessary to make this ar apid, useful bioorthogonal cleavage reaction. Quantum mechanical calculations (Table S2) suggested that the initial cycloaddition between TCO esters and different tetrazines is much faster than with vinyl esters, thus causing the rate-limiting step to depend on the tetrazine substituent. Hence,w hereas the initial cycloaddition step is rate-limiting for 5 ( Figure S22), for 1, the allylic elimination step (decaging) is the rate-limiting step ( Figure S21). Irrespective of which step determines the reaction rate, all reactions involving TCO acetate were calculated to be much faster than those with vinyl acetate (Table S2). Note, our calculations reproduced the experimentally observed trend of axial-TCO being slightly more reactive than its equatorial isomer ( Figures S21a nd S22). Concerning the isomerisation of the dihydropyridazine intermediate necessary for carboxylate release( i.e.,d ecaging), the proposed water-mediated shift (in which one water molecule adds to one of the imine moieties and is subsequently b-eliminated)a ppears to be more favourable than direct water-assisted imine-enamine tautomerisation ( Figure S23).
We startedb ya ssessing the stability of the TCO ester bond. For this, we used cis-cycloocten-1-ol (10)t ot est both the synthetic feasibility and stability of the ester bond. The cis-protected ketoprofen drug 11 was synthesised in 75 %y ield (see the Supporting Information) and, unlike the vinyl ester,p roved to be stable in cell medium, PBS (pH 7.4), and 10 %p lasma (only 5% free drug after 24 h; Figure S6). We propose that this increase in stabilityi sd ue to the increased steric hindrance at the ester bond caused by the TCO handle compared with the vinyl handle. It appears that significant steric hindrance is required on both sides of the ester bond, and the ester proves unstablei fe ither the protectingg roup (vinyl handle) or active molecule (in the case of the TCO esters reported by Robillard) [25] are not sterically bulky.I nt he case of 12,astereocentre a to the ester bond provides steric protection on one side as does TCO on the other.
With these results in hand, we decidedt oe valuatet he TCO ester for bioorthogonal iEDDA decaging. Repeating the synthesis with trans-cycloocten-1-ol resulted in the desired product 12 with approximately 50 %o ft he TCO isomerising to the cis form ( Figure 3A). This highlights ac ommon problem with synthesising TCO-functionalised molecules. The isomerisationo f cis to trans under UV light is very low yieldinga nd is not always suitable for the final step in syntheses that need al arge amount of valuable drug. However,T CO's highly reactive double bond is not compatible with severalr eaction conditions, such as the halide ions used in the formationo ft he acyl chloridei ntermediate. In addition to the modes of chirality on the cis and trans isomers of the TCO protecting group, ketoprofen also has ac hiral centre. Indeed, using thionyl chloride as an activating agent,w eo bserved by HPLC the formation of eight diastereomers ( Figure 3B)f rom the cis and trans isomers of TCO andt he chiral centre of the protected ketoprofen. Using chiral HPLC, we were able to separate each diastereomer,a nalyse them by NMR spectroscopy and characterise the four trans isomersa se nantiomeric pairs of either the axial 12 ax or equatorial 12 eq isomer ( Figure 3C). As the axial TCO isomer has previously been shown to have different reaction rates from the equatorial isomer,e ach enantiomeric pair of axial and equatorial isomersw as combined. We demonstrated am ethod of separating isomers of TCO and successfully overcame the isomerisation problem commonly experienced in synthesis with TCO, even in the challenging case of having an additional chiral centre on the drug. Now,a lthoughalow yield might be obtained, it is possible to subjectT CO to reaction conditions that readily cause isomerisation and still obtain the pure trans isomer at the end.
The reactiono fT CO-ketoprofen (12;a xial and equatorial isomers) with tetrazine 4 (Figures 4A and S7) was then studied by HPLC overt ime. Considering the fast kinetics observed for the decaging, an excess of free TCO was added to quench the re- showeds imilarr eaction profiles and decaging yields for both isomers, therefore further tetrazines were tested with only the axial isomer.T he reaction of TCO-ketoprofen was then studied with tetrazines 1, 3,a nd 5 (tetrazine 2 was not chosen due to its high instability compared to other tetrazines, Figure S2). For tetrazines 1, 3,and 4,TCO-ketoprofen is consumed within 30 s, and after 2min the change in the amount of ketoprofen is negligible (Figures 4B and S8);t his is in good agreement with the low activation barriers predicted computationally.T he ob-served accumulationo fd ihydropyridazine intermediate(s) A/B ( Figure 4B)d emonstrates our prediction (for tetrazine 1,F igure S21) that elimination of the carboxylate after iEDDA is the rate-limiting step. In agreement with computational predictions, tetrazine 5 had ad ifferent reactionp rofile ( Figure S8). In this case, no significant amount of long-lived intermediate was observed; this indicated that the eliminationi sm uch quicker and therefore, for this tetrazine, it is the cycloaddition step that is rate-limiting. This is also confirmed by the much slower disappearance of TCO-ketoprofen and the corresponding formation of ketoprofen (incomplete after 2min). It is also important to note that the three tetrazines with the same rate-determining step all show comparable decaging yields ( % 25 %). Interestingly,t etrazine 5,w hich showedn or eactionw ith vinyl ketoprofen, gives ad ecaging yield double that of the other tetrazines (54 %, Figure4C). This highlights the fact that different tetrazines are optimal for different decaging reactions.
Next, the effect of water content and pH on decaging yield were assessed ( Figures 4D and S9-S12). Te trazine 3 was chosen as ar epresentativee xample of tetrazines 1, 3 and 4.I t was shown that the reactiony ield increased from 26 (no water) to 33 %( 75 %w ater);h owever,n oi ncrease in yield was observed when 1% formic acid was added. Conversely,t etrazine 5 showedn oi ncrease in yield upon increasing water concentration. However, the yield wasi ncreased to 65 %b yt he addition of 1% formic acid ( Figure 4D). This study highlights the importance of optimising the tetrazinef or the decaging reaction, as changing the tetrazine substituents can alter the rate-limiting step of the reaction, resultingi nd ifferent kinetics and yields of decaging.
The promising stabilitya nd decaging results prompted us to furthere valuatet he application of this strategy in live-cell studies. Using the macrophage cell line RAW264.7 (ATCC T1B-71), we established the nontoxicc oncentrations of each compound ( Figures S13 and S14). Although tetrazine 5 results in a higher decaging yield by LC-MS,t his tetrazine provedt oxic to macrophages even at low concentrations( % 70 %v iability at 5 mm). Furthermore, the volatility of this tetrazine made it impractical for use in cell experiments.T etrazine 4 was chosen for furthers tudies, as it provedt ob en ontoxic at high concentrations ( % 90 %v iabilitya t1 48 mm). Surprisingly, the anti-inflammatorye ffect of tetrazines and their reactivity with nitric oxide was observed, as previously described ( Figure S15). [30] However, ac oncentration of tetrazine (50 mm)w as chosen such that no anti-inflammatory activity was observed.
Inflammationwas induced on macrophages by using lipopolysaccharide (LPS, see Figure S16 for optimisation of concentration),a nd the anti-inflammatory effect of the bioorthogonal pair (TCO-ester 12 and tetrazine 4)w as assessed by using the Griess assay ( Figure 5A). By monitoring the levels of nitric oxide (NO), we verified that when 12 was treated with 4 on LPS-stimulated macrophages as ignificantly enhanced anti-inflammatory effect was observed after 11 h( Figure 5B; p < 0.001 for equatorial, p < 0.01 for axial). This reduction in inflammationc orresponds to the successful cleavage of the TCOester bond from the caged drug leading to the releaseo fk etoprofen.D uring our studies, we also observed that ketoprofen itself failed to reduce the NO levels,w hereas 12 has am oderate effect on reducing NO levels( Figure 5B). This is likely to be due to the poor membrane permeability of ketoprofen when compared to the caged drug. When ketoprofen was protected as the TCO-ester,i ts cell permeability greatlyi mproved, leading to ah igher concentration of free drug in the cell, as assessed by HPLC( Figure 5C). Briefly,t his study involved incubating cells for 24 hw ith either the prodrug or free drug. Subsequent HPLC analysis of the extracellular mediumr evealed almost no prodrug, although as ignificant amount of ketoprofen was still observed( Figure5C). After 24 h, the level of NO from the prodrug alone was the same as that from the bioorthogonalp air ( Figure S17). Althoughweexpected the caged drug to showvery little anti-inflammatory activity,t his results uggested that activation might also happenw ithout the tetrazine trigger.I ti sw orth mentioning once again that the ester bond wass hown to be fully stable in completecell culture mediumfor 24 hat3 78C. Therefore, this observation might be due to the hydrolytic enzymes inside the cell, as was confirmed by ar eaction carriedo ut with TCO-ketoprofen and esterase from porcine liver( see the synthesis section). After 4h,as mall amount of ketoprofen had already been released, and the amount increased over the next 20 ht oayield of 71 %f or the axial isomer and 14 %f or the equatorial. The release of carboxylic acids from ester prodrugs through the action of intracellular esterases has been widely reported. [31][32][33] Although enzyme-mediated hydrolytic activation of TCO-ketoprofen was observed, the nearly spontaneousr e-lease of the active drug after tetrazine reactions uggestst hat this approach mightp lay an importantr ole in biological applications.F or example, we anticipate that an ADC could be used to target the ester prodrug to extracellular receptors expressed on macrophages,w hich would allow the fast and local delivery of ketoprofen at sites of inflammation.
The levelso fi nflammationw ere then assessed by using an enzyme-linked immunosorbent assay (ELISA,R &D Systems) to monitort he levels of prostaglandin E 2 (PGE 2 ;F igure 5D), which has been shown to be overexpressed in this cell line when inflammation is stimulated by LPS. [34] Briefly,t his assay uses a monoclonal antibodyt hat competitively binds both PGE 2 in the samples and PGE 2 -alkaline phosphatase molecules. The alkaline phosphatasep roduces ac hromogenic signal upon the addition of p-nitrophenyl phosphate. Therefore, the concentration of PGE 2 present in as ample is inversely proportionalt o the absorbance produced by the bound enzyme( Figure S18). Cells with only LPS showedt he highest level of PGE 2 (4060 pg mL À1 ), which confirmed that inflammationw as successfully stimulated. As imilarly high concentration (4010 pg mL À1 )w as observed with the tetrazine, thus confirming that the tetrazinea lone does not have an anti-inflammatory effect. Despite the poor cell permeability of ketoprofen, cells incubated with ketoprofen showed the lowest level of PGE 2 (215 pg mL À1 ). The anti-inflammatory effect of ketoprofenc an be seen by using this assaya sE LISA has ah igher sensitivity than the Griess assay (detection limit 0.5 mm). TCO-ketoprofen also showedl ower levels of PGE 2 than the LPS control (12 ax : www.chembiochem.org 2019 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim 686 pg mL À1 , 12 eq :4 86 pg mL À1 ). The bioorthogonal pair resulted in as tatistically significant reduction in PGE 2 concentration compared to the prodrug alone (12 ax + 4:1 93 pg mL À1 , p < 0.001; 12 eq + 4:2 00 pg mL À1 , p < 0.01). It was observed that the concentration of PGE 2 was the same for the bioorthogonalp air as for free ketoprofen, therefore confirming thatt he anti-inflammatory activityw as successfully reinstated up decaging in live macrophages.
In summary we have described the bioorthogonal tetrazinetriggered releaseo fc arboxylic acid-containing molecules and demonstrated their application on an ew class of anti-inflammatoryd rugs.I nd oing so, we have expanded the bioorthogonal drug activation strategy to encompass aw ider range of drugs andd iseases. Finally,t his approach might also find use for protein activation where carboxylic acid side chains can be caged/decaged.