Strain-Promoted Reaction of 1,2,4-Triazines with Bicyclononynes

Strain-promoted inverse electron-demand Diels–Alder cycloaddition (SPIEDAC) reactions between 1,2,4,5-tetrazines and strained dienophiles, such as bicyclononynes, are among the fastest bioorthogonal reactions. However, the synthesis of 1,2,4,5-tetrazines is complex and can involve volatile reagents. 1,2,4-Triazines also undergo cycloaddition reactions with acyclic and unstrained dienophiles at elevated temperatures, but their reaction with strained alkynes has not been described. We postulated that 1,2,4-triazines would react with strained alkynes at low temperatures and therefore provide an alternative to the tetrazine cycloaddition reaction for use in in vitro or in vivo labelling experiments. We describe the synthesis of a 1,2,4-triazin-3-ylalanine derivative fully compatible with the fluorenylmethyloxycarbonyl (Fmoc) strategy for peptide synthesis and demonstrate its reaction with strained bicyclononynes at 37 °C with rates comparable to the reaction of azides with the same substrates. The synthetic route to triazinylalanine is readily adaptable to late-stage functionalization of other probe molecules, and the 1,2,4-triazine-SPIEDAC therefore has potential as an alternative to tetrazine cycloaddition for applications in cellular and biochemical studies.


Results and Discussion
Our initial aim was to synthesise at riazine with af unctional group handle suitable for rapid derivatisationo ft arget molecules. We initially synthesised 6-substituted3 -amino-1,2,4-triazines 5a-c (Scheme 2i)w itht he aim of generating molecules of the form 8 via amide bond coupling.H owever,t he low nucleophilicity of the exocyclica mine meantt hat we were able to generate the protected succinylamidotriazine 7 in only 1% yield over two steps. In general, the de novo synthesis of triazines by controlled condensation between aminoguanidine and glyoxal derivatives (Scheme 2i)i sl ow yielding andg ives mixtureso fi somers;h owever,t he aminotriazine derivative 5a is now commerciallya vailablei nl arge quantities, and we therefore used this as the basis for subsequentreactions (Scheme 2ii).
Chem. Eur.J.2015, 21,14376 -14381 www.chemeurj.org bamate carbonyl group to zinc, and therefore promotes belimination (by coordination to the carbonyl methyl ester) to form the organozinc reagent. [31] Zinc insertion of iodoalanine methyl ester 11 wasc omplete after two hours at room temperature( determined by loss of the iodoalanine by TLC), and the subsequentc ross-coupling with iodotriazine 9 was performed using catalytic amounts of palladium(II) acetate and 2dicyclohexylphosphino-2',6'-dimethoxybiphenyl (SPhos). This gave the desired triazinylalanine methyl ester 12 in ay ield of 69 %. Limited attempts to optimise this coupling by increased catalystl oading and the alternative palladium catalyst (tris(dibenzylideneacetone)dipalladium(0)) didn ot lead to significant increasesi ny ield. Demethylation of the methyl ester 12 to give the free acid 13 was carriedo ut using trimethyltin hydroxide [32] in ay ield of 27 %. In this case, the overall isolated yield is limited by observable degradation of the triazines 12 and 13 at the elevated temperatures required for deprotection. (Note that base-catalysed deprotection with LiOH leads to con-comitantF moc-group removal.) To demonstrate the compatibility of the protected triazinylalanine with the conditions for the Fmoc strategy for solid-phase peptides ynthesis, we generated as imple fluorescein isothiocyanate (FITC)-labelled peptide 14 using Rink amide resin and on-resin fluorescencel abelling in 37 %y ield following HPLC purification.
We next evaluated suitable cycloaddition partners to couple to our 1,2,4-triazine derivatives 12 and 13.T here are an umber of strained cyclic dienophiles reported for tetrazineS PIEDAC conjugations( with rate constants spanning five orders of magnitude). [2,3] We initially investigated the cycloaddition reaction with norbornene,w hich reacts with tetrazine with ar elatively slow rate constant (1-10 m À1 s À1 ). [2] We used the methyl ester 12 for rate determination to avoid potentiali nterference from the free acid in 13.W ef irst synthesised norbornenyl-lysine 15 accordingt ot he procedure of Lang et al. [2] ,b ut could not detect formation of reaction product between 12 and 15 following prolonged incubation up to 80 8C( below the temperature at which we had observed thermald egradation of the triazines;S cheme 3i). To ensure that the unprotected norbornenyl-lysine 15 was not interfering with our analysis,w ec onfirmed that unfunctionalised norbornene 16 also did not reactu nder these conditions. The unhindered,s trainedb icyclo[6.1.0]nonynes react more rapidly with tetrazines with approximate rate constantso f 10 2.5 -10 3.5 m À1 s À1 ; [3] only exceeded by the rates of reactiont o trans-cyclooctenes. We generated benzoyl-protected bicyclononyne 19 (Scheme 3ii) by adaptation of the synthetic route of Dommerholt et al.; [33] following rhodium-catalysed cyclopropanation of 1,5-cyclooctadiene to yield am ixture of the anti-and syn-bicyclononene ethyl esters 17 and 17 b, anti-bicyclononene ethyl ester 17 was converted to bicyclononylol 18 (Scheme 3ii) by sequential reduction, dibrominationa nd double elimination, followed by protection using benzoyl chloridetogive alkyne 19. [34] We initially assessed the reaction of the protected bicyclononyne 19 with 12 at high concentration (65 mm in dichloromethane). Incubation of an equimolar mixture of the two reaction components yieldeda1 :1 mixture of the bicyclononapyr-idyl derivatives 20 a and b in 38 %y ield after 12 ha t3 78Ct ogetherw ith unreactedt riazine 12 (Scheme 3iii). In this case, the reactionw as limited by apparent degradation of the bicyclononyne coupling partner (as was determined by NMR). Next, we assessed the reaction rate at lower concentrations by determining the rate of product formation by HPLC, using the purifiedm ixture of authentic 20 a and b as ac oncentration standard ( Figure 1). Product formation in MeCN was measured over approximately 18 hu sing 1mm 12 and increasing concentrationso ft he bicyclononyne 19.G lobal fitting of the data gave an estimate for the second-order rate constant k 2 of 2.30 AE 0.03 10 À2 m À1 min À1 .C ompounds 12 and 19 are poorly water soluble, preventing us from carrying out the analogous experiment in water;h owever,t he reaction rate in 10 %H 2 O/ MeCN (see the Supporting Information) increased slightly to 3.0 AE 0.03 10 À2 m À1 min À1 suggesting that the rate in biological media will be slightly higher, consistent with other examplesoft his class of reaction. [35,36] Comparison of this rate of reaction (0.3-0.5 10 À3 m À1 s À1 )t o other reactions in this class [37] suggestst hat although the cycloaddition of triazines with bicyclononynes is much slower than the corresponding reactiono ft etrazines, it is comparable with other known bioorthogonal reactions, such as the Staudinger ligation. With suitable reaction partners, it has thep otential to have comparabler ates to the strain-promoted addition of azides to alkynes. Very recently,K amber et al. have reported ac omplementary study of the reaction between 1,2,4triazin-6-yl derivatives and strained trans-cyclooctenes. [25] Over ar ange of triazine substrates, they observe reactionr ates of between 1a nd 710 À2 m À1 s À1 -approximately 30-fold higher than those we have determined. Thisr atio is similart ot he approximately 15-foldd ifference in rate observed for the reaction of tetrazines with bicyclononyne and trans-cyclooctene substrates by Lang et al. [3] and Kamber et al. observations are therefore fully consistent with our observed reaction rates.

Conclusion
We have defined ar oute to 1,2,4-triazin-3-yl-linked amino acids compatible withc onventional peptide-synthesis strategies using readily available and inexpensive starting materials as precursors. The alkyl triazine reacts readily with the strainedb icyclononyne dienophile at 37 8Ci ndicating that it is suitable for protein-labelling applications.T he synthetic strategy adopted can be readily adapted to generate triazine-linked scaffolds at al ate stage. The amino acidi ss imilari ns tructure to ar ange of tyrosine-based scaffolds that have been genetically incorporated into proteins in response to an amberc odon using evolvedt yrosyl-tRNA synthetases. [38,39] We hypothesise that that it will be possible to identify such systems as hasb een recently demonstrated for triazinylphenylalanine by Kamber et al. [25] But because strategies to incorporate bicyclononynecontaining amino acids into proteins are already established, [40] this is not al imiting factor for application to site-specific labelling in an in vitro or in vivo context.

Experimental Section
General chemical experimental details procedures for synthesis of compounds 5a, b, c, 6, 7, 10, 11, 14, 17 a, 18 and 19,a nd protocols for rate determination can be found in the Supporting Information.