An Easy Route to Aziridine Ketones and Carbinols

N,N-Dimethylaziridine-2-carboxamides react with organolithium reagents yielding 2-aziridinylketones. The reaction with one equivalent of organolithium compound is selective to amide carbonyl at a low (−78 °C) temperature. These ketones, in reaction with organolithium reagents, give symmetrical and unsymmetrical aziridinyl carbinols. The usage of excess phenyllithium may serve as a special N-Boc-protecting group cleavage method for acid-sensitive substrates.


Introduction
The unusual bond angles and shorter than normal sp 3 carbon bond in aziridines, combined with the strain in the three-membered ring and the electronegativity of the nitrogen atom, make these molecules interesting and powerful building blocks. Aziridines are excellent substrates for the selective synthesis of unnatural amino acids, amino alcohols and different heterocycles of biomedical interest [1][2][3][4][5][6][7][8]. For this reason, easy access to 2-aziridinyl ketones and corresponding carbinols is highly necessary.
Reactions of carboxylic acid derivatives with C-nucleophiles through stable tetrahedral intermediates are common. This is an important method for the synthesis of carbonyl compounds, including ketones and carbinols [9]. It is also known that C-nucleophiles can react with aziridine-2-carboxylic acid derivatives in two ways: either by the opening of the aziridine ring or at a carbonyl C-atom on the side chain. Appropriate substrates for the latter reaction are represented by aziridine-2-carboxylic esters and lithium aziridine-2-carboxylates [10] or Weinreb amides [11][12][13]. The first equivalent of the organolithium nucleophile gives a ketone, and the second forms a carbinol [9].
As mentioned above, the corresponding Weinreb amides are useful substrates for the synthesis of 2-aziridinyl ketones and carbinols [11][12][13][14][15]. These activated substrates are successfully applied to obtain NH-aziridines [11], aziridinyl ketone and carbinol intermediates in the synthesis of optically active protected aminoalcohols [12] and 3-amino-2.3dihydrobenzopyran-4-ones [13]. Weinreb amides in the series of aziridine-2-carboxylic acid can be obtained from the glycidic ester in a four-step procedure [11,13]. This route includes: (a) saponification of ester, (b) DCC coupling with N,O-dimethylhydroxylamine to form Weinreb amides, (c) an oxirane ring opening with an azide and (d) a PPh 3 -mediated ring closure to form an aziridine ring. Alternative and much shorter routes to Weinreb amides are aminohydroxylation of cinnamamides followed by the closure of the aziridine ring through the O-mesylate-base procedure [14] and aziridination of α,β-unsaturated Weinreb amides with PhI=NTs [15]. The only direct approach to one specific aziridine Weinreb amide applied to the ester-menthyl N-(α-methylbenzyl)-aziridine-2-carboxylate by trimethylaluminium-mediated aminolysis was described by Lee [12]. Thus, the Weinreb amide 2 of 10 approach in the aziridine series is a complicated, substrate-specific and unpredictable pathway of synthesis.
Tertiary amides represent useful alternatives to Weinreb amides in carbonyl compound synthesis. Recently, the reaction of activated mono-Boc and bis-Boc amides with Grignard reagents led to high yields in the corresponding aromatic ketones [16]. Some examples of successful utilization of non-activated tertiary dimethylamides in aldehyde and ketone synthesis are also found in the literature [17][18][19]. Thus, hydride addition to dimethylamides in the presence of alkaline metal iodides forms aldehydes [17], but dimethylamide substrates with organosamariums [18] and organolithiums [19] yield the corresponding ketones. However, these methods from dimethylamides have not been applied in the aziridine series. Therefore, we describe an alternative and simple approach to 2-aziridinylketones and corresponding carbinols based on the usage of non-activated robust N,N-dimethyl aziridine-2-carboxamides. The carboxamides were obtained from commercially available methyl aziridine-2-carboxylate as the starting material in the reaction with commercially available organolithiums. This protocol may exclude the multistep synthesis of aziridine Weinreb amide substrates.

Results and Discussion
Our initial attempts to obtain aziridine Weinreb amides directly from N-Boc protected ester 1a using AlMe 3 -mediated aminolysis [12] (Scheme 1, Table 1) produced open-chain ring-opening by-products 3d and 3e (entries 1 and 2) and only traces of Weinreb amide 3b in the case of the modified aminolysis procedure (entry 3). In the case of starting with ester 1b, only an aziridine degradation by-product, namely, hydroxylamine 3f, was isolated (entries 4 and 5). With 3 eq. of hydroxylamine 3a, the starting material 1b disappeared, and only the by-product was isolated in a 62% yield. by trimethyl-aluminium-mediated aminolysis was described by Lee [12]. Thus, the Weinreb amide approach in the aziridine series is a complicated, substrate-specific and unpredictable pathway of synthesis. Tertiary amides represent useful alternatives to Weinreb amides in carbonyl compound synthesis. Recently, the reaction of activated mono-Boc and bis-Boc amides with Grignard reagents led to high yields in the corresponding aromatic ketones [16]. Some examples of successful utilization of non-activated tertiary dimethylamides in aldehyde and ketone synthesis are also found in the literature [17][18][19]. Thus, hydride addition to dimethylamides in the presence of alkaline metal iodides forms aldehydes [17], but dimethylamide substrates with organosamariums [18] and organolithiums [19] yield the corresponding ketones. However, these methods from dimethylamides have not been applied in the aziridine series. Therefore, we describe an alternative and simple approach to 2-aziridinylketones and corresponding carbinols based on the usage of non-activated robust N,N-dimethyl aziridine-2-carboxamides. The carboxamides were obtained from commercially available methyl aziridine-2-carboxylate as the starting material in the reaction with commercially available organolithiums. This protocol may exclude the multistep synthesis of aziridine Weinreb amide substrates.

Results and Discussion
Our initial attempts to obtain aziridine Weinreb amides directly from N-Boc protected ester 1a using AlMe3-mediated aminolysis [12] (Scheme 1, Table 1) produced openchain ring-opening by-products 3d and 3e (entries 1 and 2) and only traces of Weinreb amide 3b in the case of the modified aminolysis procedure (entry 3). In the case of starting with ester 1b, only an aziridine degradation by-product, namely, hydroxylamine 3f, was isolated (entries 4 and 5). With 3 eq. of hydroxylamine 3a, the starting material 1b disappeared, and only the by-product was isolated in a 62% yield.  In a recent alternative approach to the expected Weinreb amides, bis-Boc activation was used [16]. With N-Boc aziridine 2a, it formed the desired Boc-protected Weinreb  In a recent alternative approach to the expected Weinreb amides, bis-Boc activation was used [16]. With N-Boc aziridine 2a, it formed the desired Boc-protected Weinreb amide 3b in a moderate (31%) yield (entry 6). Unfortunately, N-Trt aziridine 2b yielded only traces of product 3c after 120 h reflux (entries 6 and 7). After all these failed experiments, we concluded that an alternative method with more convenient and practical starting materials for aziridinyl ketone synthesis was necessary.
From our previous work [8], we knew that dimethyl amides 4a (Scheme 2) and 4b (Scheme 4) were more readily available in comparison with Weinreb amides through direct aminolysis of corresponding methyl esters. In the synthesis of 2-aziridinylketones from N-protected N,N-dimethyl aziridine-2-carboxamides 4a,b as substrates, we have demonstrated comparable results in chemical yields and regioselectivity with the Weinreb amides described in the literature [11][12][13].
pound, ketone 5a (Table 2, entries 1-3), was obtained in better yields, but bulkier alkyllithiums gave ketones 5b-d ( Table 2, entries 4-13) in lower yields. Longer reaction times and an excess of organolithium reagents did not increase the yields of ketones in the reactions with alkyllithiums. Prolonged control experiments (Table 2, entries 3, 6, 8, 13) demonstrated that prolonged reaction time led to a lowering of product 5d yields because of side reactions. An excess of the organolithium compound did not give side products, because dimethyl amide 4a and the organolithium reagent initially formed an intermediate adduct [9], which liberated ketones 5 and 7 only in the hydrolysis step. Compared to alkyllithium compounds, phenyllithium reacted more slowly with amide 4a. The prolonging of the reaction time from 1 h to 12 h increased the yield of ketone 5e from 51% to 79% using 1 eq of phenyllithium (Table 2, entries 14 and 15). Using an excess of phenyllithium did not improve the product 5e yield ( Table 2, entries [16][17][18]. At the same time, a nucleophilic aziridine ring opening was not observed. We have found that under the explored reaction conditions (−78 °C, quenching of reaction mixture without warming), the electrophilic dimethylamide group is more reactive than an electrophilic aziridine ring toward the organolithium reagents used as carbon-nucleophiles.

Scheme 2.
Synthesis of ketones 5a-e from amide 4a. We have used organolithium reagents as carbon nucleophiles to obtain the desired 2-aziridinylketones 5 and 7 from N-tritylated and N-Boc substrates 4a and 4b. Preliminary tests with N-benzyloxycarbonyl analogues of the amide 4 in the reactions with organolithiums showed inappropriate results, most likely because of the side reactions on the active benzyl CH 2 fragment. Thus, we started with the use of protected N-tritylated amide 4a (Scheme 2, Table 2). According to our observations, the sterically less hindered compound, ketone 5a ( Table 2, entries 1-3), was obtained in better yields, but bulkier alkyllithiums gave ketones 5b-d ( Table 2, entries 4-13) in lower yields. Longer reaction times and an excess of organolithium reagents did not increase the yields of ketones in the reactions with alkyllithiums. Prolonged control experiments (Table 2, entries 3, 6, 8, 13) demonstrated that prolonged reaction time led to a lowering of product 5d yields because of side reactions. An excess of the organolithium compound did not give side products, because dimethyl amide 4a and the organolithium reagent initially formed an intermediate adduct [9], which liberated ketones 5 and 7 only in the hydrolysis step. Compared to alkyllithium compounds, phenyllithium reacted more slowly with amide 4a. The prolonging of the reaction time from 1 h to 12 h increased the yield of ketone 5e from 51% to 79% using 1 eq of phenyllithium ( Table 2, entries 14 and 15). Using an excess of phenyllithium did not improve the product 5e yield ( Table 2, entries [16][17][18]. At the same time, a nucleophilic aziridine ring opening was not observed. We have found that under the explored reaction conditions (−78 • C, quenching of reaction mixture without warming), the electrophilic dimethylamide group is more reactive than an electrophilic aziridine ring toward the organolithium reagents used as carbon-nucleophiles.
Ketone 5d reacted with MeLi unexpectedly in the N-trityl group, and instead of carbinol 6d, the hydroxylation product phenol 6d1 was isolated in a 52% yield ( Table 3, entry  14). On the other hand, from ketone 5a in reaction with tert-BuLi (Table 3, entry 13), a carbinol 6d and a phenol 6d1 as a ~1:1 mixture were obtained. Formation of the phenol 6d1 may be a result of ortho-lithiation of the phenyl group in the substrate trityl moiety and will be a matter for further investigations. The structure of phenol 6d1 was confirmed by X-ray analysis ( Figure 2).   Figure 1. Crystal X-ray structure of carbinol 6c. Figure 1. Crystal X-ray structure of carbinol 6c. Ketone 5d reacted with MeLi unexpectedly in the N-trityl group, and instead of carbinol 6d, the hydroxylation product phenol 6d1 was isolated in a 52% yield (Table 3,  entry 14). On the other hand, from ketone 5a in reaction with tert-BuLi (Table 3, entry 13), a carbinol 6d and a phenol 6d1 as a~1:1 mixture were obtained. Formation of the phenol 6d1 may be a result of ortho-lithiation of the phenyl group in the substrate trityl moiety and will be a matter for further investigations. The structure of phenol 6d1 was confirmed by X-ray analysis (Figure 2).   Reactions of organolithiums with both electron-rich Trt (substrates 4a, Scheme 2, Table 2) and electron-deficient Boc (substrate 4b, Scheme 4, Table 4) N-substituents containing dimethylamides were found to be regioselective (Table 4, entries 1-5). No side reactions at the aziridine ring system were observed. However, in the reaction of amide 4b Figure 2. Crystal X-ray structure of carbinol 6d1.
Reactions of organolithiums with both electron-rich Trt (substrates 4a, Scheme 2, Table 2) and electron-deficient Boc (substrate 4b, Scheme 4, Table 4) N-substituents containing dimethylamides were found to be regioselective (Table 4, entries 1-5). No side reactions at the aziridine ring system were observed. However, in the reaction of amide 4b with a four-fold excess of phenyllithium beside the expected ketone 7d, N-deprotected aziridine 8 was observed (Table 4, entry 6). We concluded that NH-aziridine 8 was formed by deprotection of the Boc group in basic conditions. To complete the N-Boc deprotection reaction, we have modified reaction conditions (entry 7) by warming up the reaction mixture before quenching. As a result, we have isolated NH-aziridine 8 as a sole product in a 66% yield.
Longer reaction times up to 12 h after warming the reaction mixture to room tempera with an excess of organolithium reagent only gave aziridine cleavage products as ins rable mixtures. Similarly, Weinreb amide 3b in the reaction with PhLi also resulte inseparable reaction mixtures. Thus, we have concluded that electron-deficient aziri 4b, in the reaction with one equivalent of organolithium reagent, is selective to am carbonyl only at a low (−78 °C) temperature.  To test the scope and limitations of the current reaction of N-Boc 2-carbonylazirid with PhLi, we have tested two additional substrates: ketone 7d and benzyl ester 1c. interaction of ketone 7d with 1 eq. of PhLi formed Boc-protected carbinol 9, but an ex of PhLi led to N-Boc deprotection and gave carbinol 10 (Scheme 5) in a moderate yie Scheme 5. Synthesis of carbinols 9 and 10 from ketone 7d.
The reaction of benzyl ester 1c with PhLi was more complex, and a mixture of ke 7d and carbinol 9 was obtained (Scheme 6, Table 5, entries 1 and 2). The yield of the binol 9 increased by using an excess of PhLi (entry 2). A longer reaction time decre the total yield and led to the formation of deprotected carbinol 10 (entry 4).  The structure of NH-aziridine 8 [20] was confirmed chemically by the performed reaction with Boc-anhydride, and the expected ketone 7d was isolated in a 78% yield. Longer reaction times up to 12 h after warming the reaction mixture to room temperature with an excess of organolithium reagent only gave aziridine cleavage products as inseparable mixtures. Similarly, Weinreb amide 3b in the reaction with PhLi also resulted in inseparable reaction mixtures. Thus, we have concluded that electron-deficient aziridine 4b, in the reaction with one equivalent of organolithium reagent, is selective to amide carbonyl only at a low (−78 • C) temperature.
To test the scope and limitations of the current reaction of N-Boc 2-carbonylaziridines with PhLi, we have tested two additional substrates: ketone 7d and benzyl ester 1c. The interaction of ketone 7d with 1 eq. of PhLi formed Boc-protected carbinol 9, but an excess of PhLi led to N-Boc deprotection and gave carbinol 10 (Scheme 5) in a moderate yield. reaction, we have modified reaction conditions (entry 7) by warming up the reaction ture before quenching. As a result, we have isolated NH-aziridine 8 as a sole produc 66% yield.
The structure of NH-aziridine 8 [20] was confirmed chemically by the performe action with Boc-anhydride, and the expected ketone 7d was isolated in a 78% y Longer reaction times up to 12 h after warming the reaction mixture to room tempera with an excess of organolithium reagent only gave aziridine cleavage products as ins rable mixtures. Similarly, Weinreb amide 3b in the reaction with PhLi also resulte inseparable reaction mixtures. Thus, we have concluded that electron-deficient aziri 4b, in the reaction with one equivalent of organolithium reagent, is selective to am carbonyl only at a low (−78 °C) temperature.  To test the scope and limitations of the current reaction of N-Boc 2-carbonylazirid with PhLi, we have tested two additional substrates: ketone 7d and benzyl ester 1c. interaction of ketone 7d with 1 eq. of PhLi formed Boc-protected carbinol 9, but an ex of PhLi led to N-Boc deprotection and gave carbinol 10 (Scheme 5) in a moderate yie Scheme 5. Synthesis of carbinols 9 and 10 from ketone 7d.
The reaction of benzyl ester 1c with PhLi was more complex, and a mixture of ke 7d and carbinol 9 was obtained (Scheme 6, Table 5, entries 1 and 2). The yield of the binol 9 increased by using an excess of PhLi (entry 2). A longer reaction time decre the total yield and led to the formation of deprotected carbinol 10 (entry 4). The reaction of benzyl ester 1c with PhLi was more complex, and a mixture of ketone 7d and carbinol 9 was obtained (Scheme 6, Table 5, entries 1 and 2). The yield of the carbinol 9 increased by using an excess of PhLi (entry 2). A longer reaction time decreased the total yield and led to the formation of deprotected carbinol 10 (entry 4). Scheme 6. Synthesis of ketones 7d, 11 and carbinols 9, 10 from benzyl ester 1c . * Warmed to room temperature before water addition.
During the warming of the reaction mixture to RT, the aziridine ring cleavage byproduct 11 [21] was isolated in a 9% yield (entry 5). The proposed tetrahedral intermediate 12 (Scheme 7) was formed from ester 1c in a nucleophilic attack with PhLi and, after hydrolysis, yielded the target ketone 7d. At the same time, the intermediate 12 reacted with an excess of PhLi and gave ring-opening intermediate 13, which, under hydrolysis conditions, turned to ketone 11. Scheme 7. Formation of ketones 7d and 11 from benzyl ester 1c through intermediates 12, 13.

Materials and Methods
The 1 H NMR spectra were recorded on Varian Mercury 200 (200 MHz) and Varian Mercury plus 400 (400 MHz) spectrometers. TMS or CHCl3 (δ 7.26 ppm, solvent CDCl3) were used as an internal standard. The 13 C NMR spectra were recorded on a Varian Mercury plus 400 (101 MHz) spectrometer. CHCl3 (δ 77.16 ppm) was used as an internal standard. HRMS were measured on a Waters Synapt G2-Si mass spectrometer. IR spectra were recorded on a Shimadzu FTIR IR Prestige-21 spectrometer. Melting points were determined on a Gallenkamp heating stage; uncorrected values were shown. TLC was carried out on DC Alufolien plates of Kieselgel 60. Column chromatography was carried out on Kieselgel (Acros), 0.023-0.070 mm, pore diameter ca 6 nm. Tetrahydrofuran was distilled Scheme 6. Synthesis of ketones 7d, 11 and carbinols 9, 10 from benzyl ester 1c. During the warming of the reaction mixture to RT, the aziridine ring cleavage byproduct 11 [21] was isolated in a 9% yield (entry 5). The  During the warming of the reaction mixture to RT, the aziridine ring cleavage byproduct 11 [21] was isolated in a 9% yield (entry 5). The

Materials and Methods
The 1 H NMR spectra were recorded on Varian Mercury 200 (200 MHz) and Varian Mercury plus 400 (400 MHz) spectrometers. TMS or CHCl3 (δ 7.26 ppm, solvent CDCl3) were used as an internal standard. The 13 C NMR spectra were recorded on a Varian Mercury plus 400 (101 MHz) spectrometer. CHCl3 (δ 77.16 ppm) was used as an internal standard. HRMS were measured on a Waters Synapt G2-Si mass spectrometer. IR spectra were recorded on a Shimadzu FTIR IR Prestige-21 spectrometer. Melting points were determined on a Gallenkamp heating stage; uncorrected values were shown. TLC was carried out on DC Alufolien plates of Kieselgel 60. Column chromatography was carried out on Kieselgel (Acros), 0.023-0.070 mm, pore diameter ca 6 nm. Tetrahydrofuran was distilled Scheme 7. Formation of ketones 7d and 11 from benzyl ester 1c through intermediates 12, 13.

Materials and Methods
The 1 H NMR spectra were recorded on Varian Mercury 200 (200 MHz) and Varian Mercury plus 400 (400 MHz) spectrometers. TMS or CHCl 3 (δ 7.26 ppm, solvent CDCl 3 ) were used as an internal standard. The 13 C NMR spectra were recorded on a Varian Mercury plus 400 (101 MHz) spectrometer. CHCl 3 (δ 77.16 ppm) was used as an internal standard. HRMS were measured on a Waters Synapt G2-Si mass spectrometer. IR spectra were recorded on a Shimadzu FTIR IR Prestige-21 spectrometer. Melting points were determined on a Gallenkamp heating stage; uncorrected values were shown. TLC was carried out on DC Alufolien plates of Kieselgel 60. Column chromatography was carried out on Kieselgel (Acros), 0.023-0.070 mm, pore diameter ca 6 nm. Tetrahydro-furan was distilled from Na/benzophenone. Esters 1a and 1b were prepared according to the literature from aziridine-2-carboxylate in reactions with Boc-anhydride [22] and trityl-chloride [23], respectively. Ester 1c was obtained in the reaction of benzyl aziridine-2carboxylate with Boc-anhydride [23]. Activated bis-Boc amides 2a,b were prepared by the 4-dimethylaminopyridine (DMAP)-catalyzed reaction of the corresponding amide with Boc 2 O [24]. Hydroxylamines 3b-f were synthesized from corresponding esters 1 as described in [12] or from activated bis-Boc amides 2 according to the published procedure [24]. Aziridine-2-carboxylic acid amides 4a and 4b were obtained from the aziridine-2-carboxylic acid dimethylamide in accordance with the previous reports [8]. All reactions involving organolithium compounds were carried out under an argon atmosphere. Diffraction data for structures 6d and 6d1 were collected at −90 • C on a Bruker-Nonius Kappa CCD diffractometer using graphite monochromated Mo-Kα radiation (λ = 0.71073 Å). The crystal structure was solved by direct methods [25] and refined by full-matrix least-squares [26]. All non-hydrogen atoms were refined in an anisotropic approximation. H-atoms were refined by the riding model.
The general procedure for the reaction of ester 1c, amides 4a,b, ketones 5a,d,e and 7d with organolithium reagents was the following. A round-bottomed flask was heated for 5 h at 150 • C and cooled in a stream of argon to room temperature. Then, 20 mL abs. THF was placed into the flask, and 0.5-6.5 mmol of aziridine substrate 1c, 4a,b, 5a,d,e or 7d was added. The resulting solution was cooled with dry ice and acetone in a stream of argon to −78 • C. To the cooled and stirred substrate solution, an organolithium reagent (solution in hexanes or toluene) was added dropwise over 10 min. The resulting reaction mixture was stirred at −78 • C under argon for 1, 2, 12 or 24 h, then 20 mL of a 1:1 water-THF mixture was added, and the reaction mixture was warmed to RT. The mixture was partitioned between water (50 mL) and diethyl ether (50 mL), the layers were separated and the organic layer was extracted with diethyl ether (3 × 20 mL). The combined ether extracts were washed with water (2 × 20 mL) and brine (20 mL), dried over Na 2 SO 4 , filtered off and evaporated under reduced pressure. The obtained crude product was purified with column chromatography on silica (200 mL silica for 1 g of material) with PE-EtOAc eluent.
Experimental details for compounds 2,3, 5-11 and characterization data for compounds 1-11 including 1H and 13C NMR, IR spectra, HRMS or elemental analysis and crystal data for compounds 6c and 6d1 are presented in supplementary materials.

Conclusions
In summary, we have demonstrated that both N-trityl and the more electron-deficient N-Boc N,N-dimethyl aziridine-2-carboxamides react regioselectively with organolithium reagents. No significant side reactions were observed if reaction mixtures were quenched before warming up. The reaction with one equivalent of organolithium reagent is selective to amide carbonyl at a low (−78 • C) temperature.
The best observed reaction conditions for-2-aziridinylketones are −78 • C temperature and 1-2 h reaction time for aliphatic N-trityl-2-aziridinylketones 5a-d and N-Boc ketones 7a-d. In the case of aromatic ketone 5e, a prolonged reaction time (12 h) is necessary. For both symmetrical and unsymmetrical carbinols 6a-f, a longer (12 h) reaction time is preferred.
Therefore, it has been demonstrated that this is a simple and practical route to selectively obtaining 2-aziridinylketones and both symmetrical and unsymmetrical 2aziridinylcarbinols. N,N-Dimethyl aziridine-2-carboxamides are better substrates than the corresponding esters or Weinreb amides. The N-Boc-protecting group nucleophilic cleavage with an excess of phenyllithium was demonstrated in the case of 2-aziridinylketone and the corresponding carbinol. This method may serve as a special Boc-protecting group cleavage for acid-sensitive substrates.