Decatungstate‐Photocatalyzed Functionalization of α‐Imino Esters for the Preparation of Unnatural α‐Amino Acid Derivatives

A library of unnatural α‐amino acid derivatives has been prepared via the hydrofunctionalization of the C=N bond in a set of compounds belonging to the family of α‐imino esters. The devised methodology capitalizes on the use of tetrabutylammonium decatungstate as the photocatalyst to generate C‐centered radicals from aliphatic and aromatic aldehydes, cyclic and acyclic oxygenated compounds and even cycloalkanes, via a hydrogen atom transfer (HAT) step. The best performance was observed when α‐hydrazono esters were employed in the role of radical traps, which delivered the corresponding (protected) α‐hydrazino acids as products. The versatility of the protocol was further demonstrated by the possibility to functionalize the herbicide safener isoxadifen‐ethyl.


Introduction
α-Imino esters are versatile building blocks for diverse synthetic applications, spanning from natural products preparation to pharmaceutical research. Interestingly, these derivatives offer a straightforward access to α-amino acids and related compounds. [1] Apart from simple α-imino esters (I), α-oximino esters (II) and α-hydrazono esters (III) are important additional members of the same family, the only structural variation being the substituent attached to the N-atom (Scheme 1a). [1a] Notably, the peculiar conjugation of the C=N double bond with the ester moiety impart α-imino esters with an enhanced reactivity over imines. Thus, their fundamental reactivity mode is represented by the nucleophilic addition of organometallics or active methylene compounds, the preferred functionalization site being the C-atom of the imine functionality (Scheme 1a, right part in blue). Other elaboration strategies of this building block include: metal-catalyzed additions (allylations, alkynylations and arylations), the (asymmetric) hydrogenation of the C=N and/or C=O double bonds, as well as different pericyclic processes (cycloadditions, ene and electrocyclic reactions). [1a] The addition of radical intermediates is currently an underdeveloped pathway (Scheme 1a, right part in red), despite the involvement of a radical-mediated reaction course has been claimed in a handful of nucleophilic additions involving organometallics based on B, [2] Zn [3] and Cu. [4] A seminal example of radical functionalization methodologies described the alkylation of α-hydrazono esters with alkyl iodides in the presence of Mn 2 (CO) 10 and InCl 3 under irradiation conditions to give the corresponding α-aminoesters. [5] Along the same line, the decarboxylative functionalization of α-oximino esters with carboxylic acids under radical conditions has been reported to take place upon irradiation in the presence of equimolar amounts of 1,4-dicyanobenzene and phenanthrene. [6] Recently, photocatalysis has offered a straightforward access to radical intermediates under sustainable conditions. [7] This has led to the development of a number of strategies for the manipulation of α-imino esters under extremely mild conditions, including their use as reaction partners in transfer hydrogenation [8] and cycloaddition processes (for the preparation of butyrolactones [9] and pyrrolidines [10] ).
Concerning the photocatalytic preparation of α-amino acids [11] through the radical functionalization of α-imino esters, the chosen substrate typically undergoes a single-electron transfer (SET; either in the oxidative or reductive manifold) with the excited photocatalyst (PC * SET ) delivering the desired radical intermediate (R * ) prone to add to the C=N double bond (Scheme 1b, left part). Convenient radical precursors include alkyltrifluoroborate salts, [12] alkyl bis(catecholato)silicates, [13] and carboxylate derivatives, [14] either as free acids or N-hydroxyphthalimide esters. On the other hand, Hydrogen Atom Transfer (HAT) photocatalysis [15] (Scheme 1b, right part) has been exploited for such purpose only in a handful of instances. Here, the excited photocatalyst (PC * HAT ) cleaves homolytically a CÀ H bond in the substrate, often with high levels of chemo-and regioselectivity, while avoiding the need of any functional/ directing group otherwise required to facilitate the SET event. [15] Quite widespread is the use of aromatic ketones in the role of PC HAT . In a seminal example, 5,7,12,14-pentaceneterone (PT) enabled the stereoselective functionalization of hydrocarbons (either benzylic, allylic or unactivated alkanes) with cyclic αimino esters, which took place in the presence of a Cu-based co-catalyst and a chiral bis-oxazoline (BOX) ligand (Scheme 2a). [16] Upon exploitation of the same photocatalyst, αimino esters have been sparsely employed as coupling partners in the aminoalkylation of adamantane scaffolds as well. [17] Very recently, 4,4'-dichlorobenzophenone has been used as PC HAT to functionalize an ethyl glyoxylate-derived N-sulfonylhydrazone; the resulting hydrazide adducts were not isolated, but further treated with an organic base in a one-pot fashion to trigger a denitrogenative cleavage. [18] Likewise, aromatic ketones have been conveniently exploited as PC HAT to trigger a range of radical addition processes onto different C=N containing substrates. [19] Following our interest in the development of synthetic strategies based on HAT photocatalysis mediated by the decatungstate anion ([W 10 O 32 ] 4À ; often used as the tetrabutylammonium salt, TBADT), [20][21][22] we decided to explore the possibility to combine its peculiar radical generation capabilities with the use of α-imino esters as radical traps. As a matter of fact, this photocatalyst has been sparsely used for the functionalization of C=N multiple bonds, as demonstrated in the radical addition onto N-tosylimines, [23] iminium ions (formed in situ via the condensation of amines and aldehydes) [24] and oxime ethers. [25] Very recently, König reported the sodium decatungstate (NaDT) photocatalyzed functionalization of αimino esters with a range of C(sp 3 )À H nucleophiles, including cyclic alkanes, methylaromatics, ethers and amides. In such work, however, only α-imino esters with a p-methoxyphenyl group installed on the N-atom as a protecting group were adopted (Scheme 2b). [26] Building upon this precedent, we hereby report our study aiming at adopting different α-imino ester precursors, as well as extending the reactivity to different substrates, with particular regards to aldehydes, which offer a straightforward access to C(sp 2 )=O acyl radicals (Scheme 2c).

Results and Discussion
To prove the feasibility of our proposal, we initially tested the reactivity of heptaldehyde (1 a) in the acylation of a library of αimino esters (2) prepared from methyl phenylglyoxylate (Table 1; further details are provided in the Experimental Section and in the Supporting Information). In particular, we focused our attention on α-hydrazono and α-oximino derivatives containing different electron-withdrawing groups, with the aim to verify the effect of such moieties on the capability of 2 to act as radical acceptors.
To find the ideal reaction conditions, we extensively built upon our precedent works in the hydrofunctionalization of electron-poor olefins. [27] Thus, when a MeCN solution (5 mL, 0.5 mmol scale) of α-hydrazono ester 2 a (0.1 M; employed as a ca. 3 : 1 mixture of geometric isomers) and aldehyde 1 a (2 equiv.) was irradiated with a 390 nm LED lamp (40 W power) for 24 h in the presence of TBADT (4 mol %), the hoped for acylated product 3 was isolated in 51 % yield after purification upon column chromatography, with complete conversion of 2 a as judged by TLC analysis (entry 1). Control experiments demonstrated that both light and TBADT were mandatory for the formation of 3 (entries 2, 3; unreacted α-imino ester 2 a was recovered almost quantitatively in both cases). We next explored the reactivity of α-hydrazono esters 2 b (where the tosyl group was substituted with a benzenesulfonyl one; also in this case, a ca. 3 : 1 mixture of geometric isomers was employed) and 2 c (containing a carboxamide substituent): in the former case, the acylated adduct 4 was formed in a slightly lower yield than 3 (48 % yield of the isolated product), while a diminished performance was obtained in the latter, with 5 isolated in 38 % yield (entries 4, 5). Finally, the adoption of α-oximino ester 2 d, containing a tosyloxime moiety, in the role of radical trap was tested, however a complex mixture not containing the expected adduct 6 was obtained.
We next investigated the reaction versatility and its substrate scope, and the results are gathered in Table 2 (see Chart S1 for a complete list of the employed reaction partners). Thus, when the model reaction delivering 3 was repeated under solar-simulated irradiation conditions (a SolarBox equipped with a 1.5 kW Xe lamp was used; 500 W m À 2 light intensity), the desired product was formed in a higher yield (79 %) upon irradiation for 24 h. Similarly, when performing the process on a larger scale (2.5 mmol), a 74 % yield of isolated 3 was obtained. We also prepared 3 by implementing a flow protocol [27a, 28,29] based on a home-made reaction setup (scale: 0.5 mmol; reactor volume: 4 mL; adopted flow rate: 2 mL h À 1 ), which allowed us to isolate 3 in an excellent 92 % yield, with a 2 h residence time required.
The reactivity of α-hydrazono ester 2 a was next studied in the presence of different hydrogen donors. Thus, the hydroacylation of 2 a with aldehydes featuring aliphatic chains of different length led to products 7-9 in 50-56 % yields, the presence of aromatic substituents being well tolerated (see the case of 9). Aliphatic oxygen heterocycles are competent alkylating agents as well, despite they need to be used in a larger excess (5 equiv.). [27b] Thus, tetrahydrofuran 1 f and 2,2dimethyl-1,3-dioxolane 1 g cleanly led to adducts 10 and 11 in 53 and 58 % yield, respectively. Finally, cyclohexane 1 j Table 1. Screening of α-hydrazono and α-oximino esters in the preparation of unnatural α-amino acid derivatives. [a] [a] Reaction conditions: an Ar-bubbled MeCN solution (5 mL) containing 1 a (2 equiv.), the chosen α-imino ester derivative (2 a-d; 0.1 M, 0.5 mmol) and TBADT (4 mol %) was irradiated with a LED lamp (40 W, λ = 390 nm) for 24 h in a Pyrex vessel (see Supporting Information for further details).
[b] The reported yields are referred to the isolated products after column chromatography. n.d.: not detected; Ts: tosyl.  (10 equiv. were used, which required to adopt a MeCN/DCM 5 : 1 medium to keep the reaction mixture homogeneous) was tested as hydrogen donor, affording hydroalkylated adduct 12 in a decent 50 % yield through the activation of a strong CÀ H bond.
We then prepared α-hydrazono ester 2 e through condensation of methyl pyruvate with tosylhydrazide and subjected it to different hydrofunctionalizations under optimized conditions. Acylated products 13-17 were isolated in up to 70 % yield; notably, the aromatic aldehyde p-anisaldehyde 1 e was also tested, although it offered a modest performance, with adduct 17 formed in 41 % yield. Oxygenated substrates were next considered, including oxygen heterocycles 1 g,h and methyl tert-butyl ether 1 i, which led to functionalized products 18-20 in similar yields (51-56 % range).
Pursuing our exploration about the effect of different substituents around the α-imino ester scaffold, we next evaluated the behavior of the α-hydrazono ester 2 f, prepared from ethyl glyoxylate and tosylhydrazide (Scheme 3a). When subjected to optimized conditions in the presence of our model hydrogen donor 1 a, no functionalized α-amino acid was obtained. Instead, p-toluenesulfonamide 21 was isolated after column chromatography (71 % yield). Furthermore, when subjecting the crude reaction mixture to GC-MS analysis, significant amounts of ethyl cyanoformate 22 were detected as well. To have more insight into this unexpected result, we repeated the same reaction in the absence of 1 a; also in this case, we consistently observed the formation of 21 (although in a diminished 55 % yield), accompanied by 22 (as from GC-MS analysis). On the other hand, no 22 was observed if the photocatalyst was omitted from the reaction mixture (see Section 1.2 in the SI for further details). Taken together, these results suggest that 2 f may behave as a H-donor towards the decatungstate excited state as well. We therefore undertook a computational analysis based on density functional theory (DFT; UωB97xD/def2TZVP level of theory in MeCN; see Section 4 in the SI for further details) to get insight into the difference in terms of Bond Dissociation Energy (BDE) of the C(sp 2 )À H bonds in compounds 1 a and 2 f. Moreover, this parameter is directly connected to the relative stability of the radical intermediates formed upon cleavage of such bonds with respect to the corresponding parent compounds. Thus, we optimized the structure of the simplified derivatives propanaldehyde 23 (model for 1 a) and the α-hydrazono derivative of methyl glyoxylate 24 (model for 2 f), as well as those of the corresponding radicals (23 * and 24 * , respectively), arising via HAT by reaction with the excited photocatalyst. The equation gathered in Scheme 3b (see also Table S1 in the SI) clearly indicates that the C(sp 2 )À H bond in 24 is stronger by ca. 19 kcal mol À 1 than that present in 23.
To further test the versatility of our protocol, we embarked in the functionalization of the herbicide safener isoxadifen-ethyl 2 g, which contains an isooxazoline core substituted with an ethyl carboxylate group at the 3-position (Scheme 4). To our delight, the α-oximino ester moiety present in 2 g was smoothly functionalized to deliver products 25 and 26 in good yields when reacted under optimized conditions with hydrogen donors 1 a and 1 f, respectively.
Based on the results reported in this work, Scheme 5 gathers a mechanistic proposal for the preparation of unnatural α-amino acid derivatives through the hydrofunctionalization of α-imino esters. Thus, upon absorption of a photon in the near-UV range (simulated solar light has been demonstrated as a viable option), the decatungstate anion populates a highly reactive state, [20] responsible for the homolytic cleavage of aliphatic CÀ H bonds in substrates 1, including aldehydes, oxygenated derivatives and even cyclohexane. This step leads to the formation of a nucleophilic C-centered radical (I * ) prone to add onto the C=N double bond of the employed α-imino ester derivative 2 leading to radical adduct II * . It is worth noting that this step occurs in a highly regioselective fashion with formation of a new CÀ C linkage at the α-position with respect to the carboxylate group, in turn delivering a N-centered radical. In the last step of the catalytic cycle, the deactivated form of the photocatalyst H + [W 10 O 32 ] 5À is restored back to the original state, while leading to the formation of the final product.
On the other hand, a peculiar behavior has been observed in the attempted functionalization of 2 f, for which a diverted reactivity led to p-toluenesulfonamide 21 formation. According to control experiments (see Scheme 3a) and GC-MS analysis (see SI), we speculate that 2 f can react with the excited photocatalyst as a hydrogen donor (in competition with heptaldehyde 1 a, when present), resulting in the cleavage of the C(sp 2 )À H bond. Based on polar effects [30] and according to the computational modeling gathered in Scheme 3b, the HAT step from 2 f (to give III * ) is expected to occur less favorably than that from 1 a (to give 1a * ) due to the higher BDE and the presence of the electron-withdrawing group COOEt. [15a] The fate of the radicals photogenerated from 1 a and 2 f is also different. In the former case, the addition of 1a * on 2 f is probably slow, leaving room for a back-HAT to take place, [31] thus restoring 1 a, with no net chemical change. On the other hand, CÀ H cleavage in 2 f affords imidoyl radical III * , [32] which in turn undergoes NÀ N bond cleavage, finally delivering 21 (through IV * ) and ethyl cyanoformate 22. The slightly diminished yield of 21 obtained when 2 f alone is irradiated with TBADT in the absence of 1 a (compare the two experiments gathered in Scheme 3a) may be related to the above-mentioned disfavored HAT step. Moreover, to further complicate the issue, it is worth noting that intermediate IV * itself may act as a HAT-agent. [33] As for the α-imino esters competent in the present protocol, hydrazono derivatives offered the best performance, particularly when a sulfonyl group was attached to the distal N-atom. This is no general rule, however, as demonstrated by literature precedents [26] and by the case of 2 g, wherein a successful functionalization of the α-oximino moiety (lacking any electronwithdrawing group) has been achieved (see Scheme 4). Importantly, most of the obtained products (3-5, 7-20) are (protected) α-hydrazino acids, [34] a class of compounds that found widespread use in the synthesis of peptides. [35]

Conclusions
The preparation of unnatural α-amino acid derivatives has been realized through the hydrofunctionalization of a set of α-imino esters with different hydrogen donors, including aliphatic and aromatic aldehydes, cyclic and acyclic oxygenated compounds and even cycloalkanes. The success of the protocol is entrusted to the excellent performance of the excited decatungstate anion in the cleavage of the CÀ H bonds present in the abovementioned substrates. Additional benefits of the reported protocol include the possibility to work under (simulated) solar light irradiation and to adopt continuous flow conditions, with obvious advantages under the sustainability and scalability standpoints, respectively. Among all the derivatives tested, the best results were observed in the functionalization of αhydrazono esters containing a sulfonyl group attached to the distal N-atom, delivering the corresponding (protected) αhydrazino acids. Nevertheless, the versatility of the proposed methodology was further demonstrated by the functionalization of the herbicide safener isoxadifen-ethyl.

Experimental Section
Typical procedure for the decatungstate-photocatalyzed hydrofunctionalization of α-imino ester derivatives: A solution of αimino ester derivative 2 (0.5 mmol, 0.1 M), the chosen H-donor 1 (1-10 mmol, 0.2-1 M; 2-10 equiv.) and TBADT (4 mM; 4 mol %) in MeCN or a MeCN:DCM (5 : 1) mixture (5 mL) in a Pyrex vessel (diameter: 45 mm) was purged with Argon for 5 min. The reaction mixture was kept under continuous stirring and irradiated for 24 h by using a PR160 L Kessil lamp (40 W, full intensity; emission centered at 390 nm) positioned 5 cm away from the reaction vessel; fan cooling was maintained for the entire duration of the process. The photolyzed solution was then concentrated in vacuo and the resulting residue purified by silica gel column chromatography (SiO 2 ; eluant: cyclohexane/ethyl acetate mixtures).