γ-Amino Alcohols via Energy Transfer Enabled Brook Rearrangement

In the long-standing quest to synthesize fundamental building blocks with key functional group motifs, photochemistry in the recent past has comprehensively established its attractiveness. Amino alcohols are not only functionally diverse but are ubiquitous in the biologically active realm of compounds. We developed bench-stable bifunctional reagents that could then access the sparsely reported γ-amino alcohols directly from feedstock alkenes through energy transfer (EnT) photocatalysis. A designed 1,3-linkage across alkenes is made possible by the intervention of a radical Brook rearrangement that takes place downstream to the EnT-mediated homolysis of our reagent(s). A combination of experimental mechanistic investigations and detailed computational studies (DFT) indicates a radical chain propagated reaction pathway.


■ INTRODUCTION
Amino alcohols�one of the most sought-after chemical motifs�are equipped with the potent amine (NH 2 ) and hydroxyl (OH) functional groups within their core motif.These entities are known to bind with Lewis acids 1 and transition metals 2 making them flexible building blocks in synthetic, material, as well as in polymer chemistry. 3As per the 2019 FDA-approved report, out of 24 small-molecule drugs, 13 were found to have either amino acid (AA) or amino alcohol residues, outlining the success of amino alcohols in modern drug design. 4Updating on these statistics, the following year the FDA report established that over 30% of small-molecule drugs contain residues of AAs or are derived from amino alcohols. 5oreover, amino alcohols are highly water-soluble and, hence, can be handled robustly in water-sensitive reaction conditions. 6o be competent in generating 1,3-amino alcohols in a one-pot procedure from alkene feedstocks through visible-light energy would facilitate easy access to these handy chemical frameworks. 7While most synthetic efforts focused on 1,2-amino alcohols and their reactivity, relatively fewer studies have explored γ-amino alcohols�the higher analogues for that matter. 81,3-Amino alcohols, the rarer homologues of their generic 1,2-counterparts, prominently feature as pharmacologically relevant molecules (for example, anti-HIV, 9 antitumor 10 drugs), natural products (alkaloids, envelope glycoprotein 11 ), and also as chiral auxiliaries 12 (Figure 1A).
So far, the de novo construction of 1,3-amino alcohols has been fairly limited, in terms of starting materials and methodologies. 13,14The synthetic progression of γ-amino alcohols is outlined by conventional methods in literature, such as aza-aldol condensation, ring-opening of aziridines and azetidines followed by carbonyl additions, and metal-mediated C−H functionalizations, to name a few (Figure 1B). 15Reports, however, on synthesis of the titular motifs through photocatalysis remains peculiarly underexplored. 16isible light induced energy transfer (EnT) processes are well-known for their ability to induce excited state reactivity under mild conditions. 17,18Based on our group's expertise in this field, we chose to investigate an innovative approach in this stride, as to date, there exist no reports leading to 1,3-amino alcohols by using this strategy.−22 This neat strategy enables the installation of two fragments of interest across the alkenes for difunctionalization reactions (Figure 1C).The primary barrier to the ease of such simultaneous addition reactions would be a nondesired radical−radical homocoupling, which may be intercepted by a kinetic phenomenon known as the persistent radical effect (PRE). 23The differentiating lifetimes of the two radical species formed�persistent radical A and transient radical B�escalate selectivity. 24The major motivations at the commencement in this domain are bifaceted�(i) designing and conceptualizing reagents with unexplored persistent/transient radical motifs and (ii) synthesizing industrially relevant and biologically active product motifs that engage in pharmaceutical drug space. 25n this scenario, the challenge comprises the construction of a 1,3-linkage across a double bond to achieve the targeted γ-amino alcohols, since previous reports have illustrated only 1,2 additions and 1,4 additions in cases of two distinct alkenes.There is a recent stand-alone report of a 1,2,5-trifunctionalization across two alkenes; however, this is a special case with a 1,2boron shift within an allylic boronic species, which stands of course as a prerequisite. 21While it is established that bifunctional reagents 26 containing a photolabile N−X bond form chemically distinct entities that stay intact during the addition across an olefin, we desired for an exception.We postulated an in situ conversion of the transient radical B to a more stable radical C prior to the addition to olefin (Figure 1C).Inspired by the classical Brook rearrangement, 27 we contemplated whether the radical version 28 could meet the specified requirements (Figure 1D).In this regard, we proposed the design of our novel oxime-carbonate bifunctional reagent 2. 20 The introduction of a Si group on this reagent was expected to trigger the radical Brook rearrangement (RBR) downstream of the homolysis of reagent 2 (Figure 1E).This would lead to the transfer of the silyl group to the adjacent oxygen center, forming a strong covalent O−Si bond (120−130 kcal•mol −1 , vs C−Si bond energy of 75−85 kcal•mol −1 ). 29Leading from our hypothesis, this would cause the transient alkoxy radical 2' from species 2 to generate the more stabilized carbon-centered radical 2'', which could then undergo 1,2-addition to alkenes, while formally producing 1,3-amino alcohols (Figure 1E).The achievement of targeted 1,3-amino alcohols pointed to the success of our hypothesis and confirmed a kinetic triumph of the rapid RBR, which takes place before the alkoxy radical can directly add to the alkene (no 1,2-oxyimination products were observed).With these first results in the success of constructing protected γ-amino alcohols starting from feedstock alkenes, we strove for a more substantial overview of this radical Brook rearrangement-assisted energy transfer pathway.

■ RESULTS AND DISCUSSIONS
Reagent Design and Screening Conditions.At the outset of designing bifunctional reagents, it is imperative to decide on a retrosynthetic route that guides the choice of accessible starting materials.To synthesize our Si-headed oxime carbonates, we identified the two major factions to be diaryloximes S5 and aliphatic alcohols S3 with a Si head on its farthest end (Figure 2A).Following a two-step synthesis from the silyl alcohols S3, with a final nucleophilic substitution on the oximes, the reagent motifs were assembled from commercially available starting materials (S1 or S3, depending on the specific structures).
With the triplet-state energy for 2a calculated (47.6 kcal• mol −1 ), we started our screen with thioxanthone (TXT) as the preliminary choice for photocatalyst (E T = 65.5 kcal•mol −1 ). 30ratifyingly, we observed the formation of our targeted product in 65% 1 H NMR yield with acrylonitrile (1a) as the substrate.Instigated by the initial hit, we started screening through a series of parameters such as solvent, concentration, equivalents of the reagent, and a list of different photocatalysts (refer to Supporting Information, section 2.3).Hence, the reaction yield could be improved to 78% ( 1 H NMR yield) by employing Ir[dF(CF 3 )ppy] 2 (dtbbpy))PF 6 (Ir−F, E T = 61.6 kcal•mol −1 ) as the photocatalyst with 1a as the standard alkene.
Reaction Scope.At this point, we planned to span our scope studies in two broad divisions�(i) with the synthesized reagents and (ii) with alkenes.Reagents 2b, 2d, and 2e offered the desired protected γ-amino alcohol product in synthetically useful yields of 78%, 25%, and 55%, respectively.However, 2c and 2f gave only trace amounts of product.To this end, deviation of the reaction outcomes may involve multiple parameters such as triplet-state energies, reagent solubility, as well as steric and electronic properties.
Styrenic alkenes lead to the formation of the corresponding 1,3-amino alcohol in decent to excellent yields.Varied substitutions on the aromatic ring were tolerated; notable examples would be products which have an electron-donating (3h) and a withdrawing group (3j) at the para-position, halogen substitution at the meta-position (3l), and pentafluoro substitution (3o) (Figure 3).The presence of pyridine (3m)  and thiazole (3n) rings delivered excellent yields of the difunctionalized products of 77% and 75%, wherein the success of heteroarenes in the scope indicates broad applicability due to their prominence in pharmaceutical drugs. 31Competently, we managed to obtain the targeted protecting-group-free γ-amino alcohols (3g', 3k, and 3l) in good yields after acidic hydrolysis.These were procured as their respective hydrochloride salts.Aliphatic terminal alkenes added another set of functional group tolerance to the scheme of the reaction.Ester (3p), phosphate (3q), terminal alkyne (3r), sulfones (3s, 3t), amide (3u), and trifluoromethyl (3v) groups are well-tolerated, affording the corresponding products in moderate to good yields.
The reaction also supports the use of disubstituted alkenes with both 1,1-and 1,2-substitution patterns.Both cyclic alkenes like the nonsymmetric cyclohexenone (3w), symmetric phthalimidic alkene (3x), and acyclic alkenes (3y) are compatible substrates, affording the diastereomers as expected through their generated chiral centers.For 3w, we note a high diastereomeric ratio (d.r.> 95:5) yielding majorly a cyclic transγ-amino alcohol owing to the addition of the generated radicals to the alkene in a sterically controlled manner.
1,1-Disubstituted alkenes also displayed exceptional suitability for the current protocol, exemplified by diphenylethylene with the best yielding entry 3ad (deprotected product, yield 89%).The product 3ae was cyclized on silica gel during the standard column chromatography isolation.More sterically hindered alkenes could be incorporated in the reaction protocol, showcased for both tri-(3ag) and tetra-substituted alkenes (3ah).However, no product formation was observed for unactivated or electron-rich alkenes.This is in line with the electronic nature of the C-centered transient radical 2a'' which is seminucleophilic in nature. 32In 2a'', the presence of vacant 3d orbitals of Si makes the reaction possible for remote electrondonating groups such as 3h.This also accounts for the relative stabilization of the transient radical 2a'' (elaborated in Computational studies).
Product Diversifications.The amine and hydroxy functionalities of amino alcohols are two key starting points in terms of designing downstream reactions.In this regard, we have already mentioned the complete deprotection of some scope entries as 1,3-amino alcohol hydrochloride salts (entries 3g', 3k, 3l, 3z, 3ad).We managed to completely reduce the imine and deprotect the silyl group using LiAlH 4 , generating γ-amino alcohol 4a in good conversion (Figure 4A).On the hydrolyzed product 3ad, a cyclization with 1,1′-carbonyldiimidazole (CDI), often used in the peptide industry to link amino acids, 33 was conducted to form carbamate 4b, 34 an entity with excellent proteolytic stability and a favorite class of modern drug discovery compounds (Figure 4B).Second, a single-step orthogonal protection of the amine functionality of 3ad led to phthalimide 4c, another class of medicinally significant products (Figure 4B). 35Inspired by the rising prevalence of peptidyl building blocks in various industries and the drug market, 36 we aimed to synthesize an amino acid with an inherent 1,3-amino alcohol handle.The catalytic product 3a with its nitrile group seemed to be the perfect substrate for this.Upon complete hydrolysis of 3a under strong acidic conditions (6.0 M HCl) and elevated temperatures at 90 °C, we successfully synthesized racemic homoserine 4d, a nonessential amino acid as well as an intermediate for the synthesis of three different essential amino acids (Figure 4C).
Mechanistic Investigations.We conducted a series of mechanistic studies to further elucidate the reaction pathway.
Cyclic voltammetry measurements showed reagent 2a to undergo one single reduction event at −2.29 V vs the SCE (Figure 5A).This is in line with the assumption that the photocatalyst Ir−F {E 1/2 (M*/M + ) = −0.89V vs SCE)} 37 does not facilitate the reduction process, thereby discarding a plausible photoredox cycle in this case.The UV/visible absorption spectrum indicates a prominent absorption of Ir−F around the operational wavelength, which indicates that the photocatalyst is the primary absorbing species (Figure 5B).In addition, 2a is also seen to absorb around this wavelength, which made it worth investigating the photocatalytic reaction under direct excitation conditions.This led us to perform the standard reaction at 365 nm in the absence of an Ir−F (Figure 5C).As expected, it delivered the anticipated product in a 20% 1 H NMR yield.The independent addition of three distinct radical initiators in the dark did not deliver any catalytic product, thereby demonstrating the pivotal role of the photocatalyst (Figure 5D).Stern Volmer quenching studies indicated 2a as the sole species quenching the photocatalyst, as 1a showed no signs of quenching Ir−F (Figure 5E).To identify the radicals and estimate their reactivities, a trapping experiment with TEMPO was conducted.The formation of the desired catalytic product was completely ceased, while a transient radical adduct (5) and an adduct of α-oxy radical added to the alkene 1a (6) were observed with TEMPO (Figure 5F).In our best guess, the persistent radical undergoes homocoupling 38 in this time frame, as is detected in the ESI spectra.Additionally, we performed the quantum yield experiments using 395 nm wavelength to achieve a quantum yield, Φ = 4.48, for our present protocol (Figure 5G).This indicates a plausible radical chain propagation in addition to the radical recombination pathway. 22omputational Studies (DFT).Primarily, to gain insights into the radical-chain mechanism, we turned to dispersioncorrected density functional theory (DFT) calculations (see Supporting Information for additional details).As shown in Figure 6A, N  Then, we hypothesized that the long lifetime of diphenyliminyl radical C allows the transient radical 2a'' to add exclusively to the terminal position of the alkene, generating the more stable radical E (Figure 6B).As expected, DFT calculations showed that the addition of 2a'' radical onto acrylonitrile 1a via TS3 proceeds regioselectively and irreversibly via a low energy barrier of 7.3 kcal•mol −1 to deliver intermediate E (downhill in energy by 18.6 kcal•mol −1 ).In addition, we explored the addition of alternative radicals over 1a (e.g., the ambiphilic radical C, nucleophilic alkoxy radical D, etc.).However, the energy barriers were found to be significantly higher (>16 kcal•mol −1 , Figure F10, see Supporting Information) and thus were not The radical D then undergoes Brook rearrangement resulting in the formation of 2a'' which selectively adds to another alkene, thus restarting the cycle.In addition, we also considered an alternative path for the addition of radical E onto oxime ester (2a) through TS4′ but ruled out this pathway based on a higher energy barrier compared to TS4 (19.2 vs 15.7 kcal•mol −1 ).

■ CONCLUSION
In a one-pot protocol, alkenes were photocatalytically converted into 1,3-amino alcohols by employing bifunctional reagents that could undergo N−O bond homolysis, followed by radical Brook rearrangement.A diverse substrate scope and notable product diversifications contribute to the broad significance of this study.Mechanistic experiments, carried out to decipher the reaction pathway, point toward an EnT-based mechanism through radical chain propagation, which is in line with our computational studies (DFT).We hope that the herein-exhibited mechanistic combination of bifunctional reagent homolysis along with rearrangement reactions stands as a gateway, unfolding substantial room in the realm of energy transfer photocatalysis.

All experimental data and characterization of compounds synthesized (PDF)
Accession Codes CCDC 2312966−2312967 contain the supplementary crystallographic data for this paper.These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

Figure 3 .
Figure 3. Scope table and sensitivity screen.Standard conditions: 0.2 mmol of 1, 0.3 mmol of 2, Ir−F (1.0 mol %), and CH 2 Cl 2 (0.05 M).Crude 1 H NMR yields are given in parentheses unless otherwise mentioned.a Yield after isolation of deprotected salts (0.2 mmol)�deprotection with 2.5 mL of MeOH and 2.5 mL of 1.0 M HCl solution under air for 1 h.b1 H NMR yields are reported.
−O and C−O bonds cleavage takes place in the triplet state of 2a (47.6 kcal•mol −1 uphill in energy which can be
considered further.Next, a selective radical−radical crosscoupling of E and C to form 3a would be feasible based on the PRE, but given the experimental evidence (ϕ = 4.48) and the relative concentration of the oxime ester 2a versus the iminyl radical (C), intermediate radical E could undergo addition to 2a via TS4 (barrier of 15.7 kcal•mol −1 ) along the radical chain pathway, 22 leading to the formation of radical intermediate F.Then, F can rapidly fragment and form product 3a via TS5 (barrier of 5.6 kcal•mol −1 ), simultaneously releasing G. Finally, after facile decarboxylation, radical D is formed along with CO 2 .