Accessing Rare α -Heterocyclic Aziridines via Brønsted Acid-catalyzed Michael Addition/Annulation: Scope, Limitations, and Mechanism

We report an approach to the diastereoselective synthesis of 1,2-disubstituted heterocyclic aziridines. A Brønsted acid-catalyzed conjugate addition of anilines to trisubstituted hetero-cyclic chloroalkenes provides an intermediate 1,2-chloroamine. Diastereocontrol was found to vary significantly with solvent selection, with computational modelling confirming selective, spontaneous fragmentation in the presence of trace acids, proceeding through a pseudo-cyclic, protonated intermediate and transition state. These chloroamines can then be converted to the aziridine by treatment with LiHMDS with high stereo-chemical fidelity. This solvent-induced stereochemical enrichment thereby enables an efficient route to rare cis -aziridines with high dr. The scope, limitations, and mechanistic origins of selectivity are also presented.


Introduction
Aziridines are a privileged motif within natural products and organic synthesis.This ring system is electrophilic enabling reaction with a broad range of nucleophiles and access to functionalized amine products. [1]Aziridines have found extensive utility in the total synthesis of alkaloid natural products, as well as their downstream functionalization. [2]Moreover, aziridine natural product analogues have demonstrated potent antitumor, [3] antimicrobial, [4] and antiviral activity. [5]Such properties have made aziridines highly attractive structures for applications within the pharmaceutical industry, for example as covalent inhibitors. [6]onsidering their utility in medicinal chemistry, [7] it is interesting to note a general lack of diversity of aziridine structures within the literature (Scheme 1a).Aliphatic and aromatic substituents dominate this space; however, few αheterocyclic examples have been reported, with N-functionalized examples particularly scarce.1a] Despite extensive development, few examples of αheterocyclic aziridines have been reported using transition metal catalysis, [8] electrochemical [9] or flow electrochemical approaches, [10] organocatalysis, [11] or utilising substitution chemistry with sulfur ylides or imines. [12]revious work within our group sought to access this underexplored chemotype through a chiral phosphoric acid (CPA)-catalyzed enantioselective aza-Michael addition of aromatic amines to chlorovinyl heterocycles, followed by basic annulation (Scheme 1b). [13]This yielded chiral mono-substituted α-heterocyclic aryl aziridines with high yield and ee.A limitation of this approach was the lack of further substitution on the aziridine ring.We therefore sought to expand the scope of this CPA strategy by establishing a method for the synthesis of 1,2disubsituted α-heterocyclic aryl aziridines.Here we provide a full account of this reaction development, including scope, limitations, and mechanism.

Results and Discussion
Reaction Development: Aza-Michael Process.To investigate the potential for this acid-catalyzed synthesis of 1,2-disubstituted aziridines, our initial benchmark system was based on stereochemically pure trisubstituted chloroalkene 1 using aniline as the nucleophile and TFA as catalyst to access the desired chloroamine product (Scheme 2).We quickly found that phenyl substitution at the 2-position of the alkene was not accommodated -product 2 was not obtained under any conditions employed (see ESI for full details).We attribute this to a steric issue and not an electronic issue based on additional observations (vide infra).Changing to methyl-substituted 3, gave 64 % conversion to the desired chloroamine 4 in approx.3 : 1 d.r.under equivalent conditions.Accordingly, 3 was selected for further optimization (Table 1). [14]llowing a preliminary assessment of reaction variables (see ESI for full details), the desired conjugate addition product 4 was obtained in 61 % NMR yield and ca. 4 : 1 d.r.using chloroacetic acid as a catalyst (entry 1).Based on this, we explored the use of chiral phosphoric acid catalysts 5 and 6 to induce enantioselective bond formation (entries 2-4); however, despite similar d.r. to entry 1, levels of asymmetric induction were poor.Enantioselectivity could not be improved despite broad screening (see ESI).13a] Despite this, promising diastereoselectivity was observed using simple carboxylic acid catalysts (e. g., entry 1).We therefore sought to develop a diastereoselective approach to the rare disubstituted aziridine products.Performing the reaction in aromatic solvents, such as PhMe (entry 5) gave improved yields; however, at the expense of diastereocontrol.In contrast, excellent diastereoselectivity was obtained using polar protic solvents such as i-PrOH (entry 6), but at lower yield.Mixtures of i-PrOH and PhMe gave results in between the extremes of the pure solvents (entries 7 and 8).No further improvements to reaction conditions could be made.
The significant impact of these solvents compelled further investigation (Scheme 3).
Suspension of 4 (1.1 : 1 d.r., racemic) in mixtures of i-PrOH and PhMe led to selective degradation of syn-4 with increasing alcohol content (see ESI for full details).The same degradation of syn-4 was observed on column chromatography.
the NH protons to the acid which in turn protonates the quinoline nitrogen.This occurs through transition states like 7, computed to correspond to a barrier of 27.3 kcal/mol in toluene.The enamine 8 that forms can undergo rapid rotation to relieve the steric clashing imposed by the double bond in 3.
A second, subsequent process mediated by AcOH sees protonation of the enamine 8 concerted with deprotonation of its NH through transition states like 9 and 10.At first glance, the modelling appears to suggest a significant kinetic preference for formation of syn-4, in contradiction to the experimentally observed non-selective process.Experiment and theory are in agreement when the barriers for the back reaction are considered.The reversions to 8 by the two stereoisomers have barriers of 25.3 and 26.9 kcal/mol; these are lower than the ratelimiting barrier leading to 8 and thus reversion of 4 and equilibration of the two isomers is to be expected.The initial emergence of a modest excess of anti-4 (see ESI for details) is consistent with this isomer having the higher barrier for reversion.The calculations support a thermodynamically controlled preference, and suggest any energy difference between the syn and anti isomers is below the limit of what can be accurately computed, supportive of the non-stereoselective formation observed.
Turning to the stereoselective degradation of 4, it was assumed that this depends on trace Lewis or Brønsted acid that can complex the quinoline nitrogen of 4, hence the reaction proceeding in the presence of both i-PrOH and silica.Further computational studies used a concentration of 1×10 À 18 M of protonated i-PrOH, 4, and settings for solvation by i-PrOH to model these conditions.
Protonation of 4 promotes fragmentation with computed barriers of 26.6 kcal/mol for fragmentation of syn-4 and 28.1 kcal/mol for anti-4.This is consistent with the direction and degree of stereoselectivity observed and with catalysis by trace acid.The origin of the stereoselectivity can be understood by considering the effect of protonating the quinoline nitrogen.This creates a cationic aromatic surface (Figure 2) that makes a strong interaction with the electron rich aniline ring, rendering the system pseudo-cyclic.A further stabilizing influence is a through-space interaction between the electron-rich belt around the Cl and the nearby NH + .In the syn-11 structure (shown looking down the fragmenting bond in Figure 2), it is clear that the aniline-π cation interaction and Cl-NH interaction can be comfortably retained while permitting the methyl group to be anti-periplanar to the quinoline.Inversion to give anti-11 might require loss of the Cl-NH interaction or the methyl clashing with the quinoline.Instead, the preferred structure retains these features but in doing so requires the pseudo-cyclic structure held together by the aniline-π cation interaction to be boat-like.Thus, syn-4 is more readily protonated and fragments more rapidly than anti-4, accounting for the high diastereoselectivity observed upon exposure to i-PrOH or silica gel chromatography.
Reaction Development: Aziridination.With a greater understanding of this degradation process, we utilized this pathway in order to access 1,2-chloroamines with high d.r., followed by a base-mediated S N i to enable access to the desired 1,2-disubstituted aziridine targets (Scheme 4).
Reaction Scope Assessment.We sought to explore the scope of this process with regards to the heterocyclic chloroalkene and aniline nucleophile components (Scheme 5).  a] Yield and d.r.b] Isolated yield, d.r.determined by 1 H NMR. See ESI for full details.
For the conjugate addition process (Scheme 5a), the most practical approach was to perform the reaction in toluene to give the crude product as a mixture of diastereomers.Diastereomeric enrichment could be achieved either by suspending the crude product in i-PrOH or exposing to silica chromatography.
In most cases this delivered the anti-chloroamine product as a single diastereomer.The anti-chloroamine could then be treated with LiHMDS to deliver the desired cis-aziridine with complete stereochemical fidelity (Scheme 5b).
Regarding the generality of the two-step process, a range of heterocycles and aromatic amines were compatible.For the conjugate addition step (Scheme 5a), modification of the azaheterocycle was possible, with substituted quinoline cores achieving moderate to excellent yields (13-16, 29).In general, more electron-rich systems, such as those bearing electrondonating groups (16) or benzothiazole (17) were less reactive.Under the PhMe-based conditions, product ratio does not vary greatly from 1.2 : 1; however, except for benzothiazole 17, the enrichment process generally delivered > 20 : 1 d.r.
Scalability of this first step was demonstrated in a 4 mmol reaction of the benchmark reaction to give 0.45 g of 4 as a single diastereomer.
The chloroamine products underwent smooth conversion to the desired cis-aziridines in generally excellent yields.Confirmation of stereochemistry was obtained through single crystal Xray structures of 12 and 45.Scalability was also demonstrated in the benchmark reaction to give 0.37 g of 12 as a single diastereomer.
There are some limitations to the route outlined, as shown in Scheme 5c (see ESI for full details).
Alternative heterocycles, such as pyridine, pyrimidine, or benzimidazole systems were unreactive.The reaction was highly sensitive to amine nucleophilicity, with more nucleophilic amines (e. g., alkylamines) resulting in catalyst deactivation.In addition, anilines with strong electron-withdrawing groups, such as nitrile or ester, were also unreactive.
Derivatization of the alkene substituent was also challenging (Scheme 6).Changing from Me (3) to Ph (1), i-Pr (50), or c-Hex (52) was not accommodated.Increasing catalyst loading, temperature, aniline equivalents, or reaction time failed to enable the conjugate addition process using these chloroalkenes.
Understanding Scope Limitations via Computational Calculations.To rationalize these observations, we once again turned to DFT calculations, using the solvation parameters for toluene.Having already established that the rate-limiting step is the initial addition, the reactants and transition states for the different substituted examples were obtained.These revealed that whereas the computed barrier for 3 is 27.3, the barriers for 1, 50, and 52 are 30.7,27.5, and 29.2 kcal/mol, respectively.In general, a deactivating effect by larger groups is observed.Comparing the transition state for 3 (54) with that for 1 (55) reveals a significant difference between the two (Figure 3).In this process, the double bond shifts along one carbon atom and in the transition state there are therefore two adjacent partial double bonds that would ideally have the three atoms involved and all of their substituents co-planar (substructure highlighted in blue in Figure 3).For the reaction of 54, this can be tolerated and the C(Me)-CÀ C-Cl dihedral angle is 163°.However, co-planarity in the reaction of 1 would entail the Ph group clashing with the quinoline.To avoid this clash, the C(Ph)-CÀ C-Cl dihedral angle must reduce to 147°in transition state 55, and this species corresponds to a higher barrier than for 54.

Conclusions
In summary, an operationally simple synthesis of rare αheterocyclic aziridines has been developed via the Brønsted acid-catalyzed Michael addition of anilines to chloroalkenes.Degradation of the syn-diastereomer of the chloroamine products by trace acid allows the isolation of a single antidiastereomer, which has been rationalized by computational modelling.The anti-chloroamine products can then be readily converted to the corresponding cis-aziridines with complete stereochemical fidelity.The scope of this transformation has been demonstrated, and despite some limitations, this approach offers a solution to a considerable gap in this area of chemical space that cannot currently be accessed by alternative synthetic methodologies, which may be helpful in the generation of compound libraries.

RESEARCH ARTICLE
We report an approach to the diastereoselective synthesis of rare 1,2-disubstituted heterocyclic aziridines.Our approach utilizes an inherent degradation of 1,2-chloroamines in the presence of trace acid, followed by annulation to give a diverse array of cis-aziridines which are not captured by alternative synthetic methodologies.The scope, limitations, and mechanistic insights into the selectivity are presented.

Figure 1 .
Figure 1.Computed transition states and intermediates for the AcOHcatalyzed Michael addition of PhNH 2 to chloroalkene 3.

Figure 2 .
Figure 2. Transition states (11) for the fragmentation of 4 and the electrostatic potential of protonated quinoline, showing the beneficial throughspace interaction for syn-11.

Scheme 6 .
Scheme 6. Substitution tolerance on the chloroalkene component.Yield determined by 1 H NMR using TCE as an internal standard.