Photochemical syntheses, transformations, and bioorthogonal chemistry of trans-cycloheptene and sila trans-cycloheptene Ag(i) complexes

Synthesis and transformations of AgNO3 complexes of trans-cycloheptene (TCH) and trans-1-sila-4-cycloheptene (Si-TCH) derivatives are described.

trans-Cycloheptene was rst spectroscopically characterized by Inoue via singlet sensitized photoisomerization of cis-cycloheptene at À35 C. 28,29 Unlike trans-cyclooctene, which is stable at room temperature, trans-cycloheptene undergoes rapid isomerization under ambient conditions via a proposed 'interrupted dimerization' mechanism. 30 trans-Cycloheptene has also been prepared via ligand exchange from a trans-cycloheptene$CuOTf complex. 31 1-Phenyl-trans-cycloheptene and trans-cycloheptenone derivatives are also known to be thermally unstable at ambient temperature, but can be trapped in situ. [32][33][34][35][36][37][38] While the parent trans-cycloheptene is thermally labile, it has been demonstrated in several studies that metal complexes can be isolated. CuOTf has been proposed to catalyze photodimerization reactions of cyclic olens via photoinduced cistrans isomerization, 39 with predominant formation of a cyclotrimer from cycloheptene. 40 A stable trans-cycloheptene$CuOTf complex has been prepared through irradiation of cis-cycloheptene$CuOTf, but a yield for the process was not reported. 41 A pybox-RuCl 2 complex of trans-cycloheptene has been prepared by irradiation of the corresponding ethylene complex in the presence of cis-cycloheptene under singlet sensitized conditions. 42 Jendralla described the preparation of AgOTf and AgClO 4 complexes of 3-methoxy-trans-cycloheptene and 6methoxy-(Z),4(E)-cycloheptadiene. 43,44 These compounds can be prepared through the Ag-mediated ring opening of a nitrosourea derivative of bicyclo [4.1.0]heptane. In unspecied yields, the AgClO 4 $3-methoxy-trans-cycloheptene complex was combined with a number of dienes to give the products of metal decomplexation and [4 + 2] cycloaddition. 43,44 Because C-Si bonds are long, the inclusion of silicon into the cyclic backbone can alleviate olenic strain and impart stability to trans-cycloalkenes. [45][46][47][48][49][50][51][52] In 1997, (E)-1,1,3,3,6,6hexamethyl-1-sila-4-cycloheptene was synthesized, resolved, and characterized crystallographically. 46,47 Here, the exhaustive allylic substitution imparts a high degree of stability to the trans-alkene. Recent studies by Woerpel 48-51 and Tomooka 52 have provided demonstrations of the utility of backbone heteroatom-containing trans-cycloalkene derivatives in synthesis. Woerpel has elegantly developed a general method for the preparation of trans-oxasilacycloheptenes-seven membered rings that contain trans-alkenes and siloxy bonds in the backbone. [48][49][50][51] There has been little investigation into the reaction chemistry of trans-cycloheptene derivatives. In 1980, Jendralla showed that silver complexes of 3-methoxy-trans-cycloheptene are isolable, and can reversibly dissociate and in unspecied yields undergo cycloaddition reactions. 43,44 Woerpel has described selective addition reactions and difunctionalization reactions of trans-oxasilacycloheptenes, and recently has reported the Diels-Alder reactions of trans-oxasilacycloheptenes with furan and tetrazine derivatives. [48][49][50][51] At 25 C in benzene, an oxasilacycloheptene was 7-fold more reactive toward 2,5diphenylisobenzofuran than a conformationally strained transcyclooctene ('s-TCO') derivative. 48 An oxasilacycloheptene derivative was shown to react with 3,6-diphenyl-s-tetrazine in benzene at rt in less than 10 min and in 90% NMR yield. 48 Our group has described a closed-loop ow reactor for the synthesis of trans-cyclooctene derivatives, whereby selective complexation with AgNO 3 is used to drive the formation of trans-isomer from cis-cyclooctene. 21,[53][54][55] Described herein is an approach to the synthesis of trans-cycloheptene (TCH) and trans-1-sila-4-cycloheptene (Si-TCH) derivatives via ow photochemical synthesis. The derivatives described are especially stable as their AgNO 3 metal complexes, which can be stored neat or in solution for long periods. With decomplexation of AgNO 3 in situ, metal-free TCH and Si-TCH derivatives can engage in a range of cycloaddition reactions as well as dihydroxylation reactions. Unlike the carbocycles, Si-TCH derivatives display good stability in solution and are shown to engage in the fastest bioorthogonal reaction reported to date. Decomplexation of AgNO 3 can be carried out in situ directly in cell media for bioorthogonal protein labeling in live cells.

Results and discussion
For the photochemical synthesis of trans-cyclooctene derivatives, the trans-cycloalkene is rst scavenged on AgNO 3 /SiO 2 and then liberated through treatment with aqueous or methanolic ammonia. 21,[53][54][55] However, our attempts to directly apply this procedure to the synthesis of carbocyclic TCHs was unsuccessful, most likely due to the susceptibility of carbocyclic TCHs to readily isomerize to their cis-isomers. 30 Recently, we demonstrated that the shelf-life of conformationally strained trans-cyclooctenes could be enhanced by storing the cycloalkenes as their AgNO 3 complexes, and that the free alkenes could be liberated in situ through treatment with NaCl in aqueous solution or cell media. 56 We reasoned that TCH and Si-TCH derivatives may also be isolable and better stored as their Ag-complexes, and that the corresponding free-alkenes could be liberated at later time points as required.
Photoisomerizations to form Si-TCH$AgNO 3 derivatives were carried out at rt using the previously described owphotoisomerization apparatus, 21,53,54 with the modication that Si-TCH$AgNO 3 complexes were directly isolated from SiO 2 without Ag-decomplexation. The metal complexes that were obtained were stable in neat form for >1 month in the freezer. However, it was necessary to alter our reactor design for the synthesis of carbocyclic trans-cycloheptenes due to their thermal lability (Fig. 1). As with the standard photoreactor, this system uses methyl benzoate as a sensitizer and a metering pump to pass a solution of substrate at a rate of 100 mL min À1 through a photowell and then through a column of AgNO 3 $SiO 2 , where the trans-cycloalkene is selectively captured as a AgNO 3 complex. In the reactor for carbocyclic TCH synthesis, a reservoir of solvent chilled in a cold bath (À50 C) was positioned Fig. 1 The apparatus for photoisomerization of carbocyclic transcycloheptene derivatives was designed to minimize loss of transcycloheptene due to thermal isomerization by using FEP tubing and inline cooling. For the synthesis of trans-1-sila-4-cycloheptenes, photoisomerizations could be carried out at rt using a conventional flow photoisomerization setup. before the photowell, and the photoisomerization was conducted in a coil of optically transparent FEP tubing. 57 The uoropolymer tubing provides a high surface area and minimal volume (only 30 mL for 8 m tubing) thereby minimizing the residence time before product adsorption on AgNO 3 /SiO 2 . In our standard setup, an inline thermometer was included to measure the temperature for the owing mixture either before or just aer the UV lamp. The temperature was measured as 0 C before entering the Rayonet photoreactor, and as 20 C aer exiting the photoreactor. With this apparatus, TCH$AgNO 3 complexes were eluted from the column, and isolated as semisolids that are moderately stable at rt but stable for weeks in the freezer.
The scope of TCH and Si-TCH synthesis is shown in Scheme 1. With the exception of cycloheptene itself, the carbocyclic and sila-cycloheptene precursors were prepared in 2-7 steps using olen metathesis as a key step (ESI †). Silver nitrate complexes of trans-cycloheptene (1a) and trans-5hydroxymethylcycloheptene (1b) were prepared in 53% and 64% yields, respectively. These TCH$AgNO 3 complexes are stable enough to handle on the bench for modest periods (hours), and to longer-term storage in the freezer (À18 C). NMR monitoring showed 90% delity for a CD 3 OD solution of 1a aer 10 days storage in freezer (À18 C), and 92% delity for a CD 3 OD solution of 1a aer 10 hours at rt on the bench.
The silver complexes of Si-TCH are much more stable. NMR monitoring showed 93% delity for a CD 3 OD solution of 2a aer 8 days at rt, and 96% delity for 2a aer storage for 1 month in the freezer (À18 C, neat). The photoisomerization method could be used to produce diphenyl (2a) or dialkyl (2b-g) substituted silacycles. Cyano (2b) and hydroxyl (2c-e) groups were tolerated, as were NHS ester (2f) and chloroalkane (2g) groups that could be used to enable conjugation to uorophores and HaloTag 58,59 fusion proteins, respectively. TCH$AgNO 3 and Si-TCH$AgNO 3 complexes were isolated as semisolids that contained 20-30% free AgNO 3 . The isolated yields were corrected by measuring the 1 H NMR against an internal standard.
Previously, a crystal structure showed that trans-cyclooctene coordinates with AgNO 3 as a 1 : 1 complex. 60 X-ray quality crystals of silver complex 2a were grown from ethyl acetate/ methanol. Selected bond lengths and angles are displayed in Fig. 2. The coordination environment at silver is distorted Scheme 1 Flow-photochemical synthesis of AgNO 3 complexes of trans-cycloheptenes and trans-1-sila-4-cycloheptenes. TCH$AgNO 3 and Si-TCH$AgNO 3 complexes were isolated as semisolids that contained 20-30% free AgNO 3 . Yields are the average of two runs, and were corrected by measuring the 1 H NMR against an internal standard. a Photoisomerization of carbocycles were conducted at low temperature using the flow apparatus described in Fig. 1. Photoisomerizations of silicycles were conducted at room temperature using the flow apparatus previously described for trans-cyclooctene synthesis. trigonal, with bridging coordination of nitrite in an extended polymeric structure in the solid state. For comparison, we also grew crystals of the silver(I) nitrate complex of the equatorial diastereomer of 5-hydroxy-trans-cyclooctene 3 (Fig. 2). Here, the coordination environment at silver is distorted tetrahedral due to the ability of the hydroxyl to serve as a second bridging ligand. The C-Ag bond lengths and bond angles were similar for 2a and 3. As expected, the C-C]C-C dihedral angle for 3 (136.7 ) is smaller than metal-free TCO 4 (139.1 ). 53 Similarly, the C-C]C-C dihedral angle for 2a (126.3 ) was smaller than that of metal free Si-trans-cycloheptene 5 (130.9 ), 46 but comparable to 6 (126.1 C) 50 -a compound with additional strain due to relatively short C-O and Si-O bonds in the cyclic backbone. Reecting the low level of metal backbonding that is common for Ag(I) alkene complexes, the C]C bond length for the Ag(I) complexes (1.329Å for 2a, 1.366Å for 3) was very similar to that of the metal-free complexes 4-6 (1.331-1.335Å).
Silver-free Si-TCH compounds could be prepared by treating their corresponding silver nitrate complexes with an excess of aq. NH 4 OH or aq. NaCl, followed by extraction with organic solvent. For example, silver complex 2a upon treatment with aq. NH 4 OH was extracted with C 6 D 6 to give a solution of Si-TCH 7a (98% trans isomer). Consistent with previous reports on a biomolecular mechanism for TCH isomerization, 30 variable amounts (20-30%) of the cis-isomer of 7a was observed when 7a was concentrated to dryness on the rotovap. However, 7a displayed high stability when maintained in solution, with only 8% isomerization observed for a 100 mM solution of 7a that was stored for 24 hours at room temperature, and <5% isomerization for a similar solution that was stored for 24 hours in a freezer (À20 C).
Ag(I)-complex 2d could also be freed of metal to give alkene 7d as a mixture of diastereomers (Fig. 3A). The allylic substituents of 7d protect the alkene from biomolecular chemistry, and unlike most other metal-free Si-TCH compounds, 7d is stable when stored in neat form and can be characterized by FT-IR (Fig. 3B). The weak C]C double bond stretch of 7d at 1624 cm À1 is shied to 1559 cm À1 for the Ag(I) complex 2d. This 65 cm À1 shi is consistent both in magnitude and direction for a Ag(I) alkene complex. 61 Finally, we noted that Ag(I)-complexation leads to signature shis of alkene resonances in both the 1 H and 13 C NMR spectra (Fig. 3C). For example, alkene resonances in the 1 H NMR spectrum of metal complex 2b were shied downeld relative to metal-free 7b by 0.15 ppm, while in 13 C NMR spectra alkene resonances of 2b were shied up eld by $16 ppm. In the 1 H NMR spectra, complexity arises due to the higher order effects for a ddd couplet.
TCH 1a and Si-TCH 2a were shown to engage in a range of reactions as shown in Scheme 2. Metal complex 1a was directly combined with 3,6-diphenyl-1,2,4,5-tetrazine to give pyridazine 8-the product of metal dissociation, Diels-Alder/retro-Diels-Alder, and oxidation, in 98% yield. Cyclopenta-1,3-diene was also used to trap trans-cycloheptene, delivering the [4 + 2] cycloaddition adduct 9 in 81% yield as a single diastereomer. We also investigated the vicinal dihydroxylation of 1a, and found that catalytic OsO 4 and NMO gave 10 in 82% yield as a single diastereomer. The observation that the dihydroxylation of 1a is stereospecic is in line with earlier observations by Cope with trans-cyclooctene. 2 Si-TCH$AgNO 3 complex 2a could be freed from silver by treatment with NH 4 OH (Scheme 2), and subsequently combined with cyclopentadiene, diazomethane, dichloroketene, and benzyl azide to provide the cycloadducts 11-14 in 76-96% yields (Scheme 2). In each case, a single diastereomer was obtained. Attempts to combine cis-1-diphenylsila-4cycloheptene (the cis-isomer of 7a) with dichloroketene, benzylazide or diazomethane only returned unreacted starting material. An X-ray structure was obtained for the dichloroketene adduct 13 (ESI †).
We sought to demonstrate that Si-TCH cycloadducts could be oxidized to give formal cycloadducts of 1,2,-dialkylolenswhich are recalcitrant substrates in intermolecular Diels-Alder reactions. 62 With the cycloadduct 11, we demonstrated that the diphenylsila-group could be oxidized to a diol-product. Thus 11 We further demonstrated that NHS-ester 2f could be modi-ed though coupling to a BODIPY-uorophore conjugate. As shown in Scheme 4, NHS ester 2f could be decomplexed from AgNO 3 by treatment with brine and extraction into CH 2 Cl 2 . The resulting free Si-TCH was then conjugated to an aminohexyl BODIPY 16, and the resulting conjugate was isolated and stored as a AgNO 3 complex (AgSiTCH-BODIPY). As discussed below, this uorophore conjugate nds utility for bioorthogonal labeling in live cells. We also demonstrated that the equatorial allylic alcohol 17(eq), derived from Ag-complex 2e, could be elaborated to the carbamate 18 through treatment with benzylisocyanate (Scheme 4). trans-Cyclooctenes with allylic carbamate leaving groups have been used by Robillard 64,65 and Chen 66,67 for the tetrazine-ligation initiated decaging of doxorubicin and other cargo molecules. The 7-membered analog 18 is particularly stable as the Ag-free trans-cycloalkene, and can be handled neat and stored without AgNO 3 in the freezer for long periods. Efforts to synthesize and explore the ability of analogs of 18 to function for payload release is a topic of ongoing study.
Under aqueous conditions, tetrazine ligations are accelerated due to the hydrophobic effect. We previously had studied the reactions of s-TCO, d-TCO and trans-cyclooctene with a water soluble dipyridyl-s-tetrazine derivative under stopped ow conditions in water with UV-vis monitoring. 21 In the most rapid example, a rate constant of 3.3 Â 10 6 M À1 s À1 was measured for an s-TCO derivative. 21 However, our attempts to conduct a similar measurement using Si-TCH 7c were complicated because the reaction was complete before we could collect data even with stopped ow monitoring. To enable the measurement, we synthesized a uorescent tetrazine-TAMRA conjugate 23, and used uorescence 'turn-on' 70 to monitor reaction progress. The uorogenic reaction enabled reaction monitoring at much lower concentrations (down to 3 mM in tetrazine). As shown in Scheme 6B and C, the reaction of 23 with Si-TCH 7c in 9 : 1 water : MeOH proceeds at 25 C with a second order rate constant k 2 1.14 Â 10 7 (+/À5 Â 10 5 ) M À1 s À1 . This is the fastest rate constant reported to date for a bioorthogonal reaction.
We also studied the in vitro and in vivo cycloaddition of SiTCH and a green uorescent protein with an unnatural tetrazinecontaining amino acid (sfGFP-150Tet-v.2.0, referred to as GFP-Tet), encoded via the procedure of Mehl and coworkers. 71 Thus, 4-(6-methyl-s-tetrazin-3-yl)phenylalanine was site-specically introduced into a C-terminally hexahistidine-tagged GFP (sfGFP-150TAG-His6) via orthogonal translation using the evolved aminoacyl-tRNA synthetase MjRS/tRNACUA pair. Co-expression of these components in E. coli resulted in the amino acid-dependent synthesis of full-length recombinant GFP-Tet (Scheme 7A). The reaction of GFP-Tet with dienophiles is uorogenic, and it is therefore possible to determine the reaction kinetics by monitoring the increase in GFP uorescence.
Although GFP-Tet is highly reactive, with in vitro rates as fast as 87 000 M À1 s À1 toward sTCO, GFP-Tet is not as rapid as 23 due to the less reactive nature of the tetrazine. Kinetic measurements were carried out with silver free Si-TCH 7c, which was obtained by treating Ag-complex 2c with NH 4 OH and extracting with ether. The second order rate constant of the reaction between Si-TCH 7c and GFP-Tet was determined to be 250 000 AE 15 000 M À1 s À1 in PBS at rt (Scheme 7B). The reaction was quantitative under these conditions as determined by ESI-MS (Scheme 7C), and is the fastest rate measured to date for GFP-Tet-2.9 times faster than previously measured rate constant for sTCO. 71 Si-TCH 18 was also shown to display very rapid labeling of GFP-Tet when carried out in live bacteria. The kinetics of the in vivo tetrazine ligation were monitored in a suspension (PBS) of E. coli overexpressing GFP-Tet by measuring the increase in whole-cell uorescence upon addition of 7c. At room temperature, a second-order rate constant of 155 000 AE 20 000 M À1 s À1 was measured for the in vivo reaction, which is 62% as rapid as the in vitro ligation. The modest reduction in rate is in line with previous observations with TCObased dienophiles. 21,71,72 Quantitative determination of the bioorthogonal reaction was veried by cell washing, lysis, purication by IMAC, and analysis by ESI-MS. Thus, Si-TCH 7c is capable of crossing the bacterial cell membrane and engaging in rapid, high yielding conjugation inside a living cell.
The reactivity and specicity of the SiTCH reagent with tetrazines in live mammalian cells was evaluated using the Hal-oTag platform, 56 which we have previously used to benchmark the efficiency of various bioorthogonal reactions in the cellular environment. We have previously shown that trans-cycloalkene-AgNO 3 -complexes liberate AgNO 3 immediately in cell media due to the high NaCl content, and perform identically to their metal-free analogs in cell labeling experiments. 56 Thus, we synthesized chloroalkane derivatives of SiTCH (AgSiTCH-Halo) and methyl-tetrazine (MeTz-Halo) and used these clickable HaloTag ligands to covalently label HaloTag protein expressed in HEK293T cells with the clickable tag. In a competitive pulsechase experiment, it was shown that 10 mM of these HaloTag ligands completely blocked incorporation of BODIPY-Halo substrate (Fig. S25 †), and therefore this concentration was used for subsequent experiments. Next we evaluated the tetrazine ligation of SiTCH in mammalian cells by reacting the AgSiTCH-Halo and MeTz-Halo protein conjugates with the corresponding MeTz-BODIPY or AgSiTCH-BODIPY uorescent probes (300 nM) for different times (2-90 min) (Scheme 8 and Fig. S26 †). The reaction was quenched at the various time points by chasing with excess non-uorescent tetrazine-amine (method 1) or TCO-amine (method 2) (Fig. S27 †), and in-gel uorescence was used to quantify conversion vs. time.
We found that the reaction of SiTCH-Halo with MeTz-BOD-IPY was complete within 15 minutes (Scheme 8B, method 2). Interestingly, the reaction appeared to be slower with the reverse pairing where SiTCH-BODIPY was reacted with MeTz-Halo and did not reach saturation until 90 minutes (Scheme 8B, method 1). We believe this is due to the suboptimal permeability and nonspecic protein binding of AgSiTCH-BODIPY which results in lower free concentrations available for the reaction. Compared to MeTz-BODIPY or uorescent 5-hydroxytrans-cyclooctenes, 56 AgSiTCH-BODIPY also resulted in more nonspecic protein labeling (Fig. S26 †). However, compared to in-gel uorescence with a TAMRA-labeled bicyclononyne (BCN), 56 AgSiTCH-BODIPY appears to be more selective.
To investigate in vivo stability of incorporated SiTCH probes in the intracellular environment, we compared HaloTag protein tagging by AgSiTCH-Halo to the previously described Ag-sTCO-Halo. 56 Subsequently the TAMRA-Tz uorescent probe was attached via the tetrazine ligation (Scheme 9A). As a benchmark, we also compared directly incorporated TAMRA-Halo without a second bioorthogonal step. In these experiments, HaloTagtransfected cells were treated with 10 mM HaloTag ligands for 30 min, followed by two 30 min wash periods. Aer this initial 90 min exposure to cells, time course experiments were carried out to test the intracellular stability of the sTCO-and SiTCHlabeled HaloTag proteins. Thus, cells were allowed to incubate for up to 24 additional hours, and cells initially tagged by AgSiTCH-Halo or Ag-sTCO-Halo were then treated with TAMRA-Tz at the indicated time points. Loss of uorescence intensity relative to the TAMRA-Halo benchmark indicated instability of the bioorthogonal protein tag.
Compared to the TAMRA-Halo ligand, the Ag-sTCO-Halo tagged protein displayed 84% uorescence intensity when the TAMRA-Tz probe was attached immediately aer the 30 min HaloTag-labeling and two 30 min wash periods (Scheme 9B, Fig. S28 and S29 †). By contrast, AgSiTCH-Halo showed only 50% uorescence intensity at this initial timepoint. Aer incubation for 7 h and 24 h, Ag-sTCO-Halo tagged protein showed 61% and 43% uorescence intensity-very similar to the TAMRA-Halo benchmark. However, the AgSiTCH-Halo tagged protein displayed only 10% uorescence aer 7 h, and negligible uorescence aer 24 h. The reduced cellular labeling efficiency with AgSiTCH-Halo could also be observed in live cell images according to the published protocol (Scheme 9C and Fig. S30 †). 56 Specically, HeLa cells overexpressing Halo-H2B-GFP were labeled with AgSiTCH-Halo or Ag-sTCO-Halo, and In-gel fluorescence intensity from n ¼ 3 independent replicates was normalized by the corresponding western blot signal, and was fit to a one-phase exponential equation. Data were plotted as mean AE SEM. aer two 30 min wash periods, treated with TAMRA-Tz for imaging. In both cases, specic uorescent signals were observed in the nucleus with good colocalization with GFP signals. When AgSiTCH-Halo was used, however, the uorescent intensity was only about half of that achieved with Ag-sTCO-Halo, conrming that the reduced stability of SiTCH in cells resulted in lower labeling efficiency. The lower labelling efficiency with AgSiTCH-Halo is likely due to trans-to-cis deactivation of the cycloalkane in the intracellular environment. Previously, we have studied the isomerization of trans-cyclooctenes under various conditions, and have shown that isomerization of trans-cyclooctenes likely occurs via radical mechanism that can be promoted by high thiol concentrations. Consistent with the faster in cellulo deactivation of SiTCH-Halo relative to sTCO-Halo, we observe that SiTCH 7b (30 mM) completely isomerizes in 2 h at 22 C in CD 3 OD with mercaptoethanol (30 mM), whereas as control sample without thiol was >98% stable under these conditions. At À17 C, 7b isomerizes more slowly in the presence of 30 mM mercaptoethanol, with 47% isomerization aer 24 h. By comparison, the trans-cyclooctenes d-TCO, s-TCO, and oxo-TCO are much more stable toward thiol promoted isomerization. In CD 3 OD with mercaptoethanol (30 mM) at room temperature, s-TCO (30 mM) isomerized only aer an 8 hour induction period, with complete conversion to the cis-isomer aer 4 additional hours. 54 In a similar experiment with d-TCO (30 mM), the induction period was 10 hours. Aer the induction period, there was 42% isomerization aer 4 hours, and 92% isomerization aer 14 hours. 56 5-OxoTCO (25 mM) in the presence of mercaptoethanol (25 mM) showed only 8% isomerization in CD 3 OD over a 22 hour period at room temperature. 72 Together, the labeling experiments in bacteria and HEK293T cells show that SiTCH derivatives can serve as useful probe molecules in the cellular environment, as the unprecedented speed of the bioorthogonal reactions of SiTCH are much more rapid than competing deactivation pathways. However, the utility of SiTCH derivatives as protein tagging molecules, where extended incubation in the cellular environment takes place prior to bioorthogonal reactivity, appears much more limited in utility plausibly due to alkene isomerization in the cellular environment.

Conclusions
In conclusion, AgNO 3 complexes of trans-cycloheptene and trans-1-sila-4-cycloheptene derivatives have been prepared via a ow photochemical synthesis, using a new low temperature ow photoreactor to enable the synthesis of carbocyclic TCH derivatives. TCH$AgNO 3 complexes can be handled for brief periods at rt and stored for weeks in the freezer (À18 C). Si-TCH$AgNO 3 complexes are especially stable, and can be stored on the bench stable for >1 week at rt, and for months in the freezer. X-ray crystallography was used to characterize a Si-TCH$AgNO 3 complex for the rst time. With decomplexation of AgNO 3 in situ, metal-free TCO and Si-TCH derivatives can engage in a range of cycloaddition reactions as well as dihydroxylation reactions. Computation predicted that Si-TCH Scheme 9 Live cell stability and imaging studies of SiTCH-Halo. (A) In the stability experiment, HEK293T cells expressing HaloTag were treated with 10 mM Ag-sTCO-Halo or AgSiTCH-Halo for 30 min, followed by two 30 min wash periods. Control cells were treated with 2 mM TAMRA-Halo. At 0, 7, and 24 h time points, 2 mM TAMRA-Tz was added and allowed to react for 1 h before analysis. (B) Cellular stability of Ag-sTCO-Halo and AgSiTCH-Halo. In-gel fluorescent intensities were normalized by the corresponding western blot signals, and were subsequently normalized by the value of TAMRA-Halo at time 0. Data were plotted as mean AE SEM. (C) Live cell images of HeLa cells expressing Halo-H2B-GFP. Cells were labeled with 10 mM Ag-sTCO-Halo or AgSiTCH-Halo for 30 min, followed by two 30 min wash periods, before the treatment with 1 mM TAMRA-Tz for 5 min. The reactions were quenched with 100 mM TCO-amine for 10 min, followed by two 30 min wash periods prior to addition of 8 mM Hoechst 33342 to visualize the nuclei and confocal imaging. Signals of Hoechst, GFP, and TAMRA were shown in blue, green, and red, respectively. Colocalization was demonstrated in composite images of GFP and TAMRA. Scale bar ¼ 25 mm. would display faster bioorthogonal reactions toward tetrazines than even the most reactive trans-cyclooctenes. Metal-free Si-TCH derivatives were shown to display good stability in solution, and to engage in the fastest bioorthogonal reaction reported to date (k 2 1.14 Â 10 7 M À1 s À1 in 9 : 1 H 2 O : MeOH). Utility in bioorthogonal protein labeling in live cells is described, including labeling of GFP with an unnatrual tetrazine-containing amino acid. The reactivity and specicity of the Si-TCH reagents with tetrazines in live mammalian cells was also evaluated using the HaloTag platform. The cell labeling experiments show that Si-TCH derivatives are suitable as highly reactive probe molecules in the cellular environment.

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