Elusive Dehydroalanine Derivatives with Enhanced Reactivity

For the first time, a simple methodology for the chemical synthesis and use of highly reactive 4‐methylenoxazol‐5(4H)‐ones from serine is presented. These dehydroalanine derivatives, which resemble the natural 4‐methylidenimidazole‐5‐one (MIO) cofactor present in lyases and aminomutases, undergo rapid reaction with carbon nucleophiles such as silyl enol ethers, as well as cycloaddition reactions with diazo compounds and reactive dienes, under very mild conditions and without any need for metal catalysts or ring‐strain activation, offering potential for bioconjugation.

For the first time, as imple methodology for the chemical synthesis and use of highly reactive 4-methylenoxazol-5(4H)-ones from serine is presented. These dehydroalanine derivatives, which resemble the natural 4-methylidenimidazole-5-one (MIO) cofactor present in lyases and aminomutases, undergo rapid reactionw ith carbon nucleophiles such as silyl enol ethers, as well as cycloadditionr eactions with diazo compounds and reactive dienes,u nder very mild conditions and without any need for metal catalysts or ring-strain activation, offering potential for bioconjugation.
Chemical modification of proteins is av ery active field of research in current chemical biology. [1] Such post-translational modification (PTM) of proteins requires site-selective reactions with high chemoselectivity. [2] In this context, a,b-unsaturated amino acids are of special interest, because they constitutea modularp latform for site-selective PTM, [3] mainly through1 ,4conjugate addition of thiols, such as those found in cysteine side chains. [4] However,c ontrolling the stereoselectivity in reactions involving a,b-dehydroamino acids and peptides still represents ac hallenge for chemists. In addressing this topic,w e have reported the synthesis of chiral dehydroalanine (Dha) and dehydrobutyrine (Dhb) buildingb locks and their applicationi n the asymmetric synthesis of lanthioninea nd b-methyllanthionine derivatives. [5] Our first-generation chiral Dha scaffold was av ersatile Michael acceptort owardsn ucleophilic thiols such as protected 1-thiocarbohydrates. [6] More recently,w ed eveloped an improved version of these chiral Dha/Dhb derivatives through lactonisation of the firstgeneration scaffolds, yieldingc hiral bicyclic structures with reduced conformational flexibility and superior reactivity and dia-stereoinducing properties with thiols as nucleophiles. [7] This methodology allowed the synthesiso fc ell-penetrating peptides containing fluorescent d-cysteine components. [8] Unfortunately,s uch Dha/Dhb scaffolds were unsuitable for introducing any other nucleophiles besidest hiols, due to their limited reactivity.
On the other hand, naturallyo ccurring Dha and Dhb have been functionalised through S-, N-and C-Michael addition, but natural reactions involving O-nucleophiles have not been discoveredy et. [9] Hence, the incorporation into peptides and proteins of ah ighly reactive a,b-dehydroamino acid derivative capable of undergoing conjugate addition with weak nucleophiles such as carbohydrates, which would directly lead to Oglycopeptides and O-glycoproteins through site-specific chemical PTM, is still am ajor challenge in chemical biology.
In our continuous search for a,b-dehydroamino acid scaffolds with improved reactivity,t he natural4 -methylidenimidazole-5-one (MIO) protein cofactord rew our attention. [10] The MIO motif is generated as aP TM from the Ala-Ser-Gly triad in ammonia lyases and aminomutases ( Figure 1A,B )a nd it is responsible for their activity towards amino acids. The amino groups of aromatic a-amino acids are N-alkylated through aza-Michael addition to the MIO structure, and this promotes beliminationt ogive cinnamic acid derivatives (in lyases) and subsequenti somerisation to b-amino acids (in aminomutases). The structures of the chromophores of the green fluorescent protein (GFP) and its relatives are very closely related to that of the MIO motif, [11] with the central serine residue in the triad being replaced by an aromatic residue, such as tyrosine in the case of GFP,leading to stable 4-arylidenimidazol-5-ones.
To the best of our knowledge, discrete 4-methylidenimidazole-5-ones have not been prepared or isolated outside the protein context of the MIO cofactor,p robably because of their very high reactivity towards nucleophiles, relative to other a,bdehydroamino acid derivatives. In an attemptt os ynthesise the MIO scaffold chemically under physiological conditions, we synthesised Ac-Ala-Ser-Gly-NH 2 ,t he minimal natural sequence leadingt oc yclisation in lyases and aminomutases ( Figure 1C). Solid-phase peptides ynthesis (SPPS) was used to obtain the linear tripeptide, with the Ca nd Ntermini capped as amides, in good yield (see the Supporting Information). The desired spontaneous cyclisation of this triad, through intramolecular aminal formation followed by two consecutive dehydrations, however,c ould not be observed either by 1 HNMR spectroscopy or by MS spectrometry after prolonged heatinga t5 08Ci n pH 8.0 PBS buffer.
With the aim of facilitating cyclisation by bringingt he reacting fragments closer together,t he stapled peptides Ac-Cys-Asp-Ser-Gly-Cys-NH 2 and Ac-Cys-Lys-Ser-Gly-Cys-NH 2 (Fig-ure 1D)w ere likewise synthesised by performing oxidative cleavage from the resin to form disulfide bonds between the two C-and N-terminal cysteiner esidues (see the Supporting Information). The native Ala residue was mutated to Asp and Lystosolubilise the resulting peptides in water.These peptides were also heated at 50 8Ci np H8.0 PBS buffer for many days, but no significant changes were detectable by 1 HNMR or MS analyses. These experiments demonstrated that MIO scaffolds are very difficult to obtain in vitro in the absence of the protein scaffold of the corresponding enzyme, which appears to promote cyclisation by imposing severe confinementc onstraints. [12] Changing the lactam nitrogena tom of the MIO scaffold to an oxygen atom-thus forming al actone-would be expected to preserve the high reactivity of the native analogue. Along these lines, 2-methyl-4-methylenoxazol-5(4H)-one has been proposed asakey Michael acceptor intermediate in the bio-mimetic synthesis of N-acetyl-4-bromotryptophane nr oute to clavicipitic acids. [13] In further studies, such ah ighly reactive type-2a lkene was found to be formed in situ from serine and acetic anhydride andc ould be detectedi ns olution by NMR spectroscopy. [14] However,a nd unlike 4-ethyliden-a nd especially 4-benzylidenoxazol-5(4H)-ones, which have been profusely synthesised and used in organics ynthesis, [15] the 4-methylidene analogues have remained very elusive to both synthesis and application,due to their very low stability.
The possibility of generating transient 4-methylenoxazol-5(4H)-ones as highly reactive Dha scaffolds for bioconjugation encouraged us to attempt their chemical synthesis. To this end, we first attempted the formationo ft he oxazol-5(4H)-one ring from O-protected N-acyl serine derivatives by using carbodiimides as coupling reagents, [16] followed by base-promoted b-elimination.T hus, racemic N,O-dibenzoylserine (1)w as readily obtained after treatmento fdl-serine with excessB zCl under basic aqueous conditions (unoptimised conditions, Supporting Information). After chromatographic purification, 5(4H)-oxazolone ring formation by treatment with N-(3-dimethylaminopropyl)-N'-ethylcarbodiimide (EDCI) in dichloromethane at 0 8C, followed by an aqueous workup, was attempted.A lthough the startingm aterialw as completely consumed, no identifiable product could be obtained.T he 1 HNMR spectrum of the obtained material showedv ery broad signals probably corresponding to apolymericmaterial.
The same results were obtained with N,N'-dicyclohexylcarbodiimide( DCC) and N,N'-diisopropylcarbodiimide (DIC) as carboxylic acid activators. Careful monitoring of the reactionb etween 1 and DIC in CDCl 3 by 1 HNMR (Figure 2) showedt he fast disappearance of the starting material signals and the appearance of two narrow doublets in the 6.00-6.20ppm region associated with methylene protons, [14] reaching itsm aximum level of conversion of 90 %i n1 5min. Thus, it was clear that 4methylene-2-phenyloxazol-5(4H)-one (MPO, 2)i sf ormed imme-Gonzalo JimØnez-OsØsreceived his PhD from the Universidad de La Rioja, before moving to the Universidad de Zaragoza-CSIC to work on heterogeneous metal catalysis and then to UCLA to work with Ken Houk on computational chemistry.H eh as been leader of the Computational Chemical Biology group at CIC bioGUNE since 2019. His research is highly collaborative and cross-disciplinary-using state-of-the art multiscale simulation methods to predict and understand complex chemical and biological processes. He has received anumber of awards such as the MBI Research Excellence Award and is involved in outreach activities and in bringing chemistry closer to society. diately through DIC-promoted cyclisation and subsequent belimination of benzoic acid, although this compound is too reactive to be isolated through ac onventionalw orkup.A s expected, the lifetime of 2 dependso nt he solventu sed in the reactiona nd the concentrationo ft he sample. In nonpolars olvents such as chloroform, compound 2 can be preserved in solution for at least 3h at concentrations up to 190 mm at 25 8C ( Figure S1 in the Supporting Information). Conversely,i na cetonitrile 2 disappears completely after 15 mina tc oncentrations around1 90 mm,a lthough it can be preserved in solution for longer times at lower concentrations ( Figure S2). Likewise, 2 is highly unstable in a1:2 mixture of acetonitrile and water,e ven at low concentrations (27 mm,F igure S3), whichr aises concerns about its potentialu se for bioconjugation in physiological media.
The influence of the serine protecting groups on the reaction outcome was then tested. With N-benzoyl-O-benzyl-dlserine, the fast formation of the oxazolone ring was also observed upon treatment with DIC. However,n otably slower b-elimination of benzyla lcoholw as observed, with mixtureso f the target compound 2 and its cyclic precursor in variable ratios being produced, because 2 decomposes over time. N-Acetyl-O-benzyl-dl-serine was tested with analogous results. Thus, N,O-dibenzoyl-dl-serine( 1)w as selected as the most convenient starting material for the rest of the study.
Generation of 2 in situ from 1 in the presence of equimolecular amounts of DIC with subsequent addition of sulfur,n itrogen and oxygen nucleophiles led to fast alkene decomposition. In no case was the desired conjugate addition reaction observed.W ith basic nucleophiles such as primary and secondary amines or with thiolates or alkoxides generated in situ in the presence of bases such as N,N-diisopropylethylamine or sodiumhydride, 2 was completely degraded, probablythrough anionic polymerisation pathways. On the other hand, protonated thiols anda lcohols weren ot reactive enough to undergo conjugate addition during the lifetimeo f2 at variousc oncentrations.
Treatmentwith ethyl diazoacetate(EDA) produced similar results, leading to am ixture of racemic cyclopropanes 5a and 5b in close to 1:1r atio. Compound 5b could also be characterised by X-ray diffraction analysis( Figure S7). MPO (2)c learly showedg reater reactivity than relateda cyclic dehydroamino acid analogues such as methyl 2-acetamidoacrylatet owards cyclopropanation with diazo compounds (Figures S4 and S5), which normally requires transition-metal catalysts such as rhodium or palladium to generate reactivem etal carbene species. [20] On the other hand, uncatalysed 1,3-dipolar cycloadditions with ethyl diazoacetate have been reported only with highly activated substrates bearing nitro (a-carbethoxy-1-nitrostyrenes and a-halo-a-nitroalkenes) [21] and nitrile groups (arylidene malononitrile and arylidene ethyl cyanoacetate), [22] althougha tv ery slow reactionr ates (2-5 days needed). Conversely, compound 2 (generated in situ) completely reacts with diazo compoundsi nafew minutes. The second-order rate constantf or the reaction between 2 and EDA in CDCl 3 at 25 8C was determinedb y 1 HNMR spectroscopy to be k 2 = 3.9 10 À3 m À1 s À1 (Figure S4), which is comparable to the rate constants found for 1,3-dipolar cycloadditions of strain-promoted alkynes. [23] Recently,R aines and co-workersh ave described selective reactions betweend iazoacetamides and dehydroalanine residues under biocompatiblec onditions, [23] as well as the manipulation of steroelectronic effects of diazo compounds to increaset heir reactivity and selectivity for bioorthogonal applications. [24] Finally,u ncatalysed Diels-Alder cycloadditions with MPO (2) were tested under the same conditions with various dienes such as cyclohexa-1,3-diene, cyclopentadiene, 2,3-dimethoxybuta-1,3-diene, 2,3-dimethylbuta-1,3-diene and 3,6-bispyridin-2-yl-1,2,4,5-tetrazine, the last of which is commonly used for protein labellingt hrough metal-free strain-promoted inverseelectronic-demand Diels-Alder cycloaddition.N otably, cyclopentadiene was reactive enough to react cleanly with freshly generated 2 to afford racemates endo-6 and exo-6 in a5 5:45 ratio. The structure of adduct exo-6 was confirmed by X-ray diffraction analysis( Figure S8). Again, this uncatalysed and nonstrain-promoted reaction with 2 proceeds much more rapidly at room temperature than that with its acyclic analogue MAA, which requires prolonged heatinga ta round1 00 8Cf or 5h or the presence of ametal catalyst such as TiCl 4 . [25] The superior reactivity (i.e.,e lectrophilicity) of cyclic MPO (2) relative to acyclic analogues such as methyl 2-acetamidoacrylate can be explainedi nt erms of the large stabilisationo ft he LUMO in the former case, by 1.2 eV,d ue to extensive conjugation over the five-membered lactone and phenyl rings.T his translatesi nto significantly lower activation free energies (DG°) from 2 than in the case of MAA, such as those calculated quantum mechanically for cycloaddition with diazomethane, methyl diazoacetatea nd cyclopentadiene (Figures 3a nd S9-S12). With regard to 1,3-dipolar cycloadditionw ith diazo com-pounds, as tepwise zwitterion-mediated cyclopropanation mechanism with spontaneous nitrogen release has recently been proposed [26] as an alternative pathway to the commonly accepted asynchronous concerted cycloaddition and subsequent nitrogen extrusion from the pyrazolinei ntermediates.I n fact, such as tepwise mechanism was calculated to be significantly favoured for compound 2,p robably due to its ability to delocalise the negative chargeg enerated upon the Michaeltype addition of diazo compounds. In view of the large activation energy required for the cleavage of the pyrazolinei ntermediates (transition structures and intermediates could be calculated only in the triplet excited state), relative to the retrocycloaddition reaction from the same intermediate, and of the fact that the energy barrierf or the stepwisez witterionic cyclopropanation reactionb etween compound 2 and methyl diazoacetate is only % 2kcal mol À1 above the concerted transition state ( Figure S10), such ap rocess is likely to take place to some extent under the assayed experimental conditions.
In summary,w eh ave developed as imple methodology for the synthesis of highly elusive 4-methylenoxazol-5(4H)-ones with aryl or alkyl groupsa tt he 2-position, depending on the amine protection of the starting serined erivative. Such cyclic dehydroalanine derivatives are highly electrophilic and can quickly react in situ with silyl enol ethers, diazo compounds and dienes under very mild conditions at room temperature, and in the absence of metal catalysts withoutt he need for ring strain activation.I nv iew of the growingi nterest in diazo groups as new chemical reportersf or bioorthogonal labelling of biomolecules [27] and versatile tools for chemical biology [28] we believe that our methodology offers potential for bioconjugation and post-translational modification of proteins under controlled physiological conditions, although the low stability of MPO derivatives in water currently limits their biological scope. This possibility,t ogether with the scope to extend our methodology to serine residues within ap eptide context to chemically installM IO-type modifications, are currently being evaluated in our laboratory.

Experimental Section
General procedure for sequential one-pot synthesis of 2a nd subsequent reaction with different reagents: N,O-Dibenzoylserine (1,0 .3 mmol) was introduced into ar ound-bottomed flask, and CHCl 3 (10 mL) was added. The heterogeneous mixture was stirred at room temperature, and DIC (0.3 mmol) was then added. The reaction mixture dissolved immediately.T he reagent of interest (0.3 mmol) was then added, and the mixture was stirred for 3-30 min at room temperature. After consumption of the starting material, the reaction mixture was transferred to an extraction funnel, and the organic layer was washed with saturated NaHCO 3 solution (2 5mL). The organic layers were combined and dried over anhydrous Na 2 SO 4 ,a nd the solvent was evaporated. The crude reaction product was purified by vacuum liquid chromatography (VLC, hexane/AcOEt 100:0 to 80:20 gradient).