Photocatalytic Umpolung of N- and O-substituted alkenes for the synthesis of 1,2-amino alcohols and diols†‡

We report an organophotocatalytic 1,2-oxyalkynylation of ene-carbamates and enol ethers using Ethynyl BenziodoXolones (EBXs). 1-Alkynyl-1,2-amino alcohols and diols were obtained in up to 89% yield. Photocatalytic formation of radical cations led to Umpolung of the innate reactivity of the alkenes, enabling addition of a nucleophilic benzoate followed by radical alkynylation.


Introduction
Accessing 1,2-amino-alcohols and 1,2-diols has been a longstanding target in synthetic methodology. These scaffolds have found multiple applications in pharmaceutical, material and agrochemical sciences. 1,2 Alkynes are highly useful building blocks for synthesis, as starting points for product diversication. 3 Combining both functionalities, 1-alkynyl-1,2-amino alcohols can be found as intermediates in the synthesis of insecticidal 4-alkynyloxazolines 4 and b-erythroidine, 5,6 as well as essential structural elements in bioactive antitumoral enediynes. 7,8 1-Alkynyl-1,2-diols can be found, for example, in the Petrosiol family of neurotrophic diyne tetraols. 9,10 Enamides and ene-carbamates are versatile starting materials for the generation of complex aminated building blocks. [11][12][13][14][15][16] In particular, they have been used extensively in atom transfer radical addition (ATRA) reactions. 17 Due to their innate nucleophilicity, they are excellent traps for electrophilic radicals, leading to the formation of a nucleophilic aamino radical I (Scheme 1A, a). The latter can then react with a radical trap, 18 undergo oxidation to the a-amino cation, [19][20][21] reduction to the a-amino anion, 22 or addition to an organometallic species followed by reductive elimination. 23 Despite the efficiency associated to such transformations, all enamide difunctionalizations reported so far are based on the initial addition of a highly reactive electrophilic radical, limiting functional group tolerance and the structural diversity of the obtained products.
Nicewicz and co-workers developed a different approach towards alkene difunctionalization based on oxidation under photoredox conditions for the generation of radical cations. [24][25][26] Scheme 1 (A) Difunctionalization of enamides: innate reactivity vs. Umpolung. (B) Atom-economical enamide difunctionalization with benziodoxolone reagents. (C) This work: photocatalytic Umpolung enabling the synthesis of 1-alkynyl-1,2-amino alcohols and diols. This highly electrophilic species can then react with various nucleophiles, enabling new types of hydrofunctionalizations. [27][28][29][30][31][32] In the case of enamides or ene-carbamates, such a strategy would result in a neat Umpolung of the reactivity (Scheme 1A, b). It is important to stress that such an approach would completely change the type of transformations accessible, as the rst step would involve reaction with a nucleophile, in opposition to the electrophilic radical already intensively investigated. [17][18][19][20][21][22][23] Although this strategy appears highly attractive to answer current limitations in enamide functionalization, only one example of ene-carbamate hydroacetoxylation has been reported by Nicewicz and co-workers. 29 When considering the importance of nitrogen-containing compounds, a difunctionalization of enamides via photocatalytic Umpolung would be highly desirable.
In order to develop such a process, we turned to Ethynyl BenziodoXolone (EBX) hypervalent iodine reagents, which have been identied as efficient traps for radicals. [33][34][35][36][37] Their application in radical-mediated olen alkynylation has also been explored. [38][39][40][41][42] Recently, our group has exploited the nucleophilicity of the carboxylate group of EBX reagents in atomeconomical reactions such as the 1,1-oxyalkynylation of diazo compounds and the ring-opening/oxyalkynylation of thiiranes. [43][44][45][46] Therefore, EBX reagents appear ideally suited for the functionalization of radical cations due to their dual nucleophilic/somophilic nature. For what concerns atom economical enamide 1,2-difunctionalization with benziodoxole reagents, the Gillaizeau group has reported an iron-catalyzed enamide oxyazidation (Scheme 1B, a: X ¼ N 3 ) with Zhdankin's reagent. 47,48 This reaction was proposed to occur via a classical ATRA mechanism. The Patil group reported a gold catalysed 1,2oxyalkynylation of allenenamides with EBXs (Scheme 1B, b, X ¼ alkynyl), involving both redox and p-activation by the gold catalyst. 49 Consequently, 1,2-oxyalkynylation remains limited to allenenamides as substrates and photocatalytically generated radical cations have never been intercepted with EBX reagents. 50 Herein, we show that ene-carbamate radical cations can be generated under oxidative photoredox conditions using 4-CzIPN-derived organic dyes. [51][52][53] The formed intermediates react with Umpolung of the reactivity in an atom-economical fashion with EBX reagents acting as both O-nucleophile and alkynylating radical trap sources (Scheme 1C). This methodology could then be extended to commercially available enamides and enol ethers. The mild oxidative conditions allowed selective reaction of electron-rich alkenes in presence of non-activated ones. This procedure provides easy access to orthogonally protected 1alkynyl-1,2-amino alcohols and diols, setting the foundations for the development of further difunctionalizations of electronrich olens via radical cation intermediates.
With the optimized reaction conditions in hand, we explored the scope of the reaction (Scheme 2). Acyclic ene-carbamates were tolerated affording Boc and Cbz protected amines 3b and 3c in 63% and 86% yield. Although N-H vinyl carbamates degraded under the reaction conditions, the orthogonally diprotected ene-carbamate 1d was converted to 3d in 73% yield. A methyl amine worked well under our reaction conditions (3e, 83% yield). An allyl amine was also tolerated affording 3f in 54% yield, 60 demonstrating that selective functionalization of enecarbamates over alkenes was possible. Substrates bearing a silylated alcohol and an ethyl ester yielded the desired compounds 3g and 3h in 52% and 69% yield. The procedure also worked with secondary amines (3i, 65% yield). Commercial N-vinylpyrrolidinone gave compound 3j in 85% yield. a-Substitution of the alkene was not tolerated (1k no conversion), but b-substituted (E)-ene-carbamate afforded 3l as a mixture of diastereoisomers in 80% yield (3 : 1 dr.).
Finally, we examined enolethers, which have comparable oxidation potentials (e.g. dihydropyran (DHP, 9i), 1.51 V vs. SCE). Aliphatic (10a), benzylic (10b) and allylic (10c) ethers were obtained in 89%, 58% and 45% yield. Chlorinated product 10d was obtained in 77% yield. A secondary enol ether afforded 10e in 62% yield. Although a-substituted ene-carbamates were not tolerated (1k, no conv.), tertiary ether 10f was obtained from propen-2-yl enol ether in 82% yield. b-Substitution afforded a mixture of compounds, from which acetal 10g corresponding to Markovnikov addition could be isolated in 13% yield. 61 Finally, DHP 9h 62 afforded two regioisomers: anti-Markovnikov product 10ha (58% yield) and Markovnikov product 10hb (15% yield). 63 Diverse EBX reagents were then examined (Scheme 3). Both electron-poor and electron-rich arenes on the alkyne provided the desired compounds 12a-12e in up to 76% yield. The transfer of a silyl protected alkyne was less efficient and product 12f was obtained in 9% yield only. 64 EBXs bearing sensitive functionalities such as an alkyl bromide or a terminal alkene gave the corresponding products 12g and 12h in 51% and 37% yield. Functionalized EBXs could also be used with enol ether 9a affording 12i and 12j in 62% and 52% yield. We then investigated the mechanism of the reaction. Without the photocatalyst and/or a light source, no product was detected (Scheme 4A, eqn (a)). We then considered the possibility of acyloxyl radical Ia (Scheme 4B) adding to the electronrich olen 1a. Previously, Ia (a resonance structure of the iodanyl radical Ib) has been reported predominantly as a Hatom abstractor. [65][66][67] To the best of our knowledge, the only proposed report of Ia adding to an alkene is that of the Gillaizeau group. 47 Chen and co-workers reported the generation of Ia through the reduction of BIOAc (7) by excited Ru(bpy) 3 2+ . 57 With this photocatalyst, we observed no conversion or product formation (eqn (b)). This suggests that the generation of Ia alone does not lead to product formation. In addition, no conversion was observed with substrate 1k, which is tougher to oxidize (1kc + /1k z 1.86 V) (Scheme 2). Some conversion would have been expected if the reaction proceeded through the addition of oxygen-centred radical Ia. b-Substituted alkene 1l gave product 3l in 80% yield with the same efficiency as for terminal enamide 1a. An ATRA process would have been more signicantly impaired by the substituent. The observed anti-Markovnikov selectivity is in agreement with the reactivity reported for radical cations. 26 In order to test our hypothesis, we performed the reaction under the standard conditions in presence of sodium acetate and acetic acid (eqn (c)). The acetate could indeed be introduced (compound 13), but EBX-addition product 3j remains the main product (2 : 1 ratio). When the reaction was performed with a 2 : 1 ratio of Z and E isomers of 1l instead of pure E compound, no change in yield and diastereoselectivity was observed, supporting the presence of a radical intermediate (eqn (d)). With the strongly oxidizing catalyst 5, low yields were observed even in presence of BIOAc (7), despite almost full conversion of 1a (eqn (e)). Based on these results, we propose a tentative mechanism for the oxyalkynylation (Scheme 4B). First, the excited photocatalyst 4b* oxidizes 1a generating radical cation II and reduced catalyst 4bc À . As support for this step, quenching of uorescence of catalyst 4b by 1a was observed in a Stern-Volmer experiment (see ESI ‡ for details). 68 This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 11274-11279 | 11277 Then II is trapped by carboxylate III. This results in the formation of radical IV, which can add to 2 affording the product and iodanyl radical Ib. The latter can close the catalytic cycle by oxidizing 4bc À to regenerate catalyst 4b. 69 We suspect that BIOAc (7) serves as initiator for the reaction by generating Ia via reduction of BIOAc (7) with 4b*. The resulting oxidized catalyst 4bc + would be also competent to oxidize 1a. This pathway would also help sustaining the catalytic cycle by ensuring a sufficient concentration of Ia. A nal control experiment corroborates this hypothesis: the reaction was performed with no additive (eqn (f)). The desired compound was obtained in 65% 1 H NMR yield aer 5 days of reaction time with 20% residual PhEBX (2). We then performed the transformation on gram scale (Scheme 5, eqn (a)), affording 3j in 76% yield (0.998 g). Selective hydrolysis of the ester group from 3j gave 14 in 96% yield (eqn (b)). Hydrolysis of 10a provided 15 in 91% yield (eqn (c)). Finally, 3b underwent Boc deprotection to give amino ester 16 in 74% yield (eqn (d)).

Conclusions
In conclusion, we have developed a photocatalytic 1,2-oxyalkynylation of ene-carbamates based on Umpolung of the reactivity. The transformation proceeds in an atom-economical fashion with EBXs acting both as alkynylating and carboxylating reagents. The reaction occurs at room temperature under blue LED irradiation using 4-ClCzIPN (4b) as an organic photocatalyst and does not require the use of highly reactive electrophilic radicals. The methodology could be extended to enamides and enolethers. The method shows good chemoselectivity for nitrogen or oxygen-substituted olens over aliphatic alkenes. Based on preliminary mechanistic studies, we propose that an ene-carbamate radical cation is the key intermediate that ensures the anti-Markovnikov regioselectivity initiated by nucleophile addition, contrasting with the classical ATRA mechanism usually invoked for the functionalization of alkenes with hypervalent iodine reagents. This reaction allows quick access to protected 1-alkynyl-1,2-amino alcohols and 1alkynyl-1,2-diols, which are important building blocks in agrochemical, pharmaceutical and material sciences.

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