Synthesis of Cyclic N-Acyl Amidines by [3 + 2] Cycloaddition of N-Silyl Enamines and Activated Acyl Azides

In this study, we describe the synthesis of cyclic N-acyl amidines from readily available N-heteroarenes. The synthetic methodology utilized the versatile N-silyl enamine intermediates from the hydrosilylation of N-heteroarenes for the [3 + 2] cycloaddition reaction step. We evaluated various acyl azides and selected an electronically activated acyl azide, thereby achieving a reasonable yield of cyclic N-acyl amidines. We analyzed the relationship between the reactivity of each step and the electronic nature of substrates using in situ nuclear magnetic resonance spectroscopy. In addition, we demonstrated gram-scale synthesis using the proposed methodology.

Versatile triazole intermediates are typically formed during the [3 + 2] cycloaddition reactions of enamine derivatives and organic azides. These intermediates have been utilized in various synthetic methodologies, such as amidine synthesis with a rearrangement involving nitrogen extrusion [21]. However, reports on the reactivity of N-silyl enamine are relatively rare, which is most likely due to the intrinsic instability of its N-Si bond at ambient conditions [22,23]. Nevertheless, the in situ use of N-silyl enamine is still an attractive strategy, in which the silyl group is used as the transient protecting group for the fragile but useful free enamine for organic synthesis [14,15].
Our previous works demonstrated [14,15] the wide substrate scope of N-heteroarenes; however, the scope of organic azides has been mainly limited to sulfonyl azides. Sulfonyl azides exhibit powerful reactivity; however, their reaction with acyl azides to synthesize the desired acyl amidine product has not yet been realized (Scheme 1b) [14]. Interestingly, we recently found that the electron-withdrawing groups of sulfonyl azides increased the reactivity of the cycloaddition reaction [15]. This prompted us to develop a [3 + 2] cycloaddition reaction that uses the electron-withdrawing group on the less reactive acyl

Results and Discussion
The reactivity of N-silyl enamine from isoquinoline 1a was found to be promising thus, we began our study with isoquinoline [15]. Using the previously optimized cond tions for the dearomative mono-hydrosilylation of 1a, we obtained the N-silyl enamin intermediate with good yield in a nuclear magnetic resonance (NMR) cell. The result from the following addition of acyl azides with different substituents are listed in Schem 2. In contrast to the reaction of N-silyl enamine from quinoline [14], the N-silyl enamin 2a from 1a reacted with benzoyl azide 3a to produce the acyl amidine product 4a with reasonable crude yield in 48 h. Acyl azides with halogen substituents (3b-3d) also worke well with N-silyl enamine 2a to achieve cyclic acyl amidines 4b-4d. However, the electron rich azide 3e was less reactive than the electron-poor azides 3b-3d, which is consisten with our previous reports on sulfonyl azides [14,15]. We therefore examined more elec tron-poor azides (3f-3i) with nitro or trifluoromethyl substituents to improve the reactiv ity of acyl azides (4f-4i). Indeed, the strong electron-withdrawing substituents improve

Results and Discussion
The reactivity of N-silyl enamine from isoquinoline 1a was found to be promising; thus, we began our study with isoquinoline [15]. Using the previously optimized conditions for the dearomative mono-hydrosilylation of 1a, we obtained the N-silyl enamine intermediate with good yield in a nuclear magnetic resonance (NMR) cell. The results from the following addition of acyl azides with different substituents are listed in Scheme 2. In contrast to the reaction of N-silyl enamine from quinoline [14], the N-silyl enamine 2a from 1a reacted with benzoyl azide 3a to produce the acyl amidine product 4a with a reasonable crude yield in 48 h. Acyl azides with halogen substituents (3b-3d) also worked well with N-silyl enamine 2a to achieve cyclic acyl amidines 4b-4d. However, the electron-rich azide 3e was less reactive than the electron-poor azides 3b-3d, which is consistent with Molecules 2022, 27, 1696 3 of 12 our previous reports on sulfonyl azides [14,15]. We therefore examined more electronpoor azides (3f-3i) with nitro or trifluoromethyl substituents to improve the reactivity of acyl azides (4f-4i). Indeed, the strong electron-withdrawing substituents improved the reactivity of acyl azides. However, the acyl amidine products 4f-4h were relatively unstable due to hydrolysis during the work-up and purification process. Therefore, the 3,5-bis(trifluoromethyl)benzoyl azide 3i was considered to be the appropriate acyl azide to achieve the stable amidine product 4i. Notably, the conversion of 3i to 4i was achieved within a short timeframe of 16 h. We also confirmed the (Z) configuration of N-acyl amidine using an X-ray diffraction analysis of a single crystal of 4i (CCDC 2142037).
Molecules 2022, 27, x FOR PEER REVIEW the reactivity of acyl azides. However, the acyl amidine products 4f-4h were r unstable due to hydrolysis during the work-up and purification process. There 3,5-bis(trifluoromethyl)benzoyl azide 3i was considered to be the appropriate a to achieve the stable amidine product 4i. Notably, the conversion of 3i to 4i was within a short timeframe of 16 h. We also confirmed the (Z) configuration of N-a dine using an X-ray diffraction analysis of a single crystal of 4i (CCDC 2142037). Using the optimized azide 3i, we explored the substrate scope of the isoqui (Scheme 3). The 5-chloroisoquinoline 1j reacted with 3i to form the cyclic amidi terestingly, the reactivity in each step of the cascade pathway was distinct from th reaction of 1a. The conversion in the first dearomatization step was completed w h because the electron-poor quinoline derivative was more reactive toward th acid-catalyzed hydrosilylation. However, the second cycloaddition step was muc with the electron-poor N-silyl enamine 2j than 2a. This suggested that the N-silyl 2 acted as a nucleophile in the second step. Notably, the acyl amidine 4j was obtai good yield due to the cascade synthetic approach. Meanwhile, the reaction of the olines with a bromo substituent at different positions (1k-1m) proceeded smoot ducing acyl amidines 4k-4m in 48 h. Isoquinoline 1n with an alkyne substituent position reacted well with 3i to afford 4n in a good yield. The reactions of isoqu 1o-1p with an electron-donating substituent also produced cyclic acyl amidine however, the yields were moderate due to the low reactivity of 1o-1p toward hydrosilylation step. The second [3 + 2] cycloaddition step was quite fast with an rich substrate. Using the optimized azide 3i, we explored the substrate scope of the isoquinolines 1 (Scheme 3). The 5-chloroisoquinoline 1j reacted with 3i to form the cyclic amidine 4j. Interestingly, the reactivity in each step of the cascade pathway was distinct from that of the reaction of 1a. The conversion in the first dearomatization step was completed within 2.5 h because the electron-poor quinoline derivative was more reactive toward the Lewis acid-catalyzed hydrosilylation. However, the second cycloaddition step was much slower with the electron-poor N-silyl enamine 2j than 2a. This suggested that the N-silyl enamine 2 acted as a nucleophile in the second step. Notably, the acyl amidine 4j was obtained with good yield due to the cascade synthetic approach. Meanwhile, the reaction of the isoquinolines with a bromo substituent at different positions (1k-1m) proceeded smoothly, producing acyl amidines 4k-4m in 48 h. Isoquinoline 1n with an alkyne substituent at the 5-position reacted well with 3i to afford 4n in a good yield. The reactions of isoquinolines 1o-1p with an electron-donating substituent also produced cyclic acyl amidines 4o-4p; however, the yields were moderate due to the low reactivity of 1o-1p toward the first hydrosilylation step. The second [3 + 2] cycloaddition step was quite fast with an electron rich substrate.
Next, we surveyed the reactivity of various acyl azides 3 for the N-silyl enamine 6a from quinoline 5a (Scheme 4). Although the reactivity of 6a was not sufficient for benzoyl azide 3a in 2 h [14], the cyclic acyl amidine 7a was obtained within 24 h; however, the yield was low, at 13%. Meanwhile, from the screening of acyl azides 3b-3i with different electronic natures, acyl azides with 4-nitro (3f) and 3,5-bis(trifluoromethyl) (3i) were effective toward 6a, leading to a moderate to good yield of the cyclic amidines 7f and 7i, respectively. Acyl azides with a strong electron-withdrawing 3,5-dinitro substituent (3g) were converted efficiently to N-silyl enamine 6a; however, the resulting acyl amidine 7g was unstable under ambient conditions. Relatively poor electron-withdrawing (3b-3d and 3h) and electron-donating 3e acyl azides were unable to convert 6a to cyclic amidines 7 with reasonable yields. Therefore, the acyl azide 3i was considered the most appropriate acyl azide for the synthesis of 7 from 5a via Scheme 4. In addition, the N-silyl enamine 6a, which was unreactive toward benzoyl azide 3a can now be utilized for the synthesis of acyl amidines 7.
Molecules 2022, 27, x FOR PEER REVIEW Scheme 3. Substrate scope of isoquinolines 1 for the synthesis of acyl amidines 4.
Next, we surveyed the reactivity of various acyl azides 3 for the N-silyl ena from quinoline 5a (Scheme 4). Although the reactivity of 6a was not sufficient for azide 3a in 2 h [14], the cyclic acyl amidine 7a was obtained within 24 h; how yield was low, at 13%. Meanwhile, from the screening of acyl azides 3b-3i with electronic natures, acyl azides with 4-nitro (3f) and 3,5-bis(trifluoromethyl) (3i) fective toward 6a, leading to a moderate to good yield of the cyclic amidines 7 respectively. Acyl azides with a strong electron-withdrawing 3,5-dinitro substitu were converted efficiently to N-silyl enamine 6a; however, the resulting acyl am was unstable under ambient conditions. Relatively poor electron-withdrawing and 3h) and electron-donating 3e acyl azides were unable to convert 6a to cyclic a 7 with reasonable yields. Therefore, the acyl azide 3i was considered the most app acyl azide for the synthesis of 7 from 5a via Scheme 4. In addition, the N-silyl ena which was unreactive toward benzoyl azide 3a can now be utilized for the syn acyl amidines 7. Next, we surveyed the reactivity of various acyl azides 3 for the N-silyl enam from quinoline 5a (Scheme 4). Although the reactivity of 6a was not sufficient for b azide 3a in 2 h [14], the cyclic acyl amidine 7a was obtained within 24 h; howe yield was low, at 13%. Meanwhile, from the screening of acyl azides 3b-3i with d electronic natures, acyl azides with 4-nitro (3f) and 3,5-bis(trifluoromethyl) (3i) w fective toward 6a, leading to a moderate to good yield of the cyclic amidines 7f respectively. Acyl azides with a strong electron-withdrawing 3,5-dinitro substitu were converted efficiently to N-silyl enamine 6a; however, the resulting acyl ami was unstable under ambient conditions. Relatively poor electron-withdrawing and 3h) and electron-donating 3e acyl azides were unable to convert 6a to cyclic am 7 with reasonable yields. Therefore, the acyl azide 3i was considered the most appr acyl azide for the synthesis of 7 from 5a via Scheme 4. In addition, the N-silyl enam which was unreactive toward benzoyl azide 3a can now be utilized for the synt acyl amidines 7.
We investigated the substrate scope of quinolines 5 using the optimized acyl azide 3i (Scheme 5). First, an N-silyl enamine 6 from 5 was produced with diphenylsilane (Ph 2 SiH 2 ) [14]. The reactions of the electron rich N-silyl enamines (6j-6p) and electron-poor azide 3i resulted in cyclic acyl amidines (7j-7p) with low to moderate yields. The N-silyl enamines with electron-donating substituents (6j-6p) were sufficiently reactive in the [3 + 2] cycloaddition step; however, the conversions in the first hydrosilylation step were relatively slow. Especially, the 6-methoxymethyloxyquinoline 5n was not converted to 6n with Ph 2 SiH 2 . Therefore, we decided to use the more reactive methylphenylsilane (MePhSiH 2 ) for conversion in the first hydrosilylation of 5n to achieve N-silyl enamine 6n' with reasonable yield [15]. An isolable amount of acyl amidine 7n was subsequently obtained. We note that the acyl azide 3i was, however, ineffective toward other N-silyl enamines with an electron-withdrawing group. For example, the cycloaddition of 3i and bromo substituted N-silyl enamine 6q was not accomplished due to the low reactivity of 6q toward the cycloaddition step.
We investigated the substrate scope of quinolines 5 using the optimized acyl (Scheme 5). First, an N-silyl enamine 6 from 5 was produced with diphenylsilane (P [14]. The reactions of the electron rich N-silyl enamines (6j-6p) and electron-poor resulted in cyclic acyl amidines (7j-7p) with low to moderate yields. The N-silyl e with electron-donating substituents (6j-6p) were sufficiently reactive in the [3 cloaddition step; however, the conversions in the first hydrosilylation step were re slow. Especially, the 6-methoxymethyloxyquinoline 5n was not converted to Ph2SiH2. Therefore, we decided to use the more reactive methylphenylsilane (Me for conversion in the first hydrosilylation of 5n to achieve N-silyl enamine 6n' w sonable yield [15]. An isolable amount of acyl amidine 7n′ was subsequently ob We note that the acyl azide 3i was, however, ineffective toward other N-silyl en with an electron-withdrawing group. For example, the cycloaddition of 3i and bro stituted N-silyl enamine 6q was not accomplished due to the low reactivity of 6q the cycloaddition step. The reaction rate in each step was found to be relatively slow; thus, the el effect of the quinolines on the reaction rate could be observed through in situ NM itoring. (Supplementary Materials). We compared the initial reaction progression 1 H NMR using quinolines containing different substituents (Scheme 6). The bora lyzed hydrosilylation step of the quinolines (5a, 5m, and 5q) was first investigat Ph2SiH2. The initial reaction rate of 5-bromoquinoline (5q) was faster than 6-metho oline (5m) as expected. This indicated that the rate-determining step in the hydros was the dearomative borohydride addition [24]. Next, we examined the [3 + 2] cy tion step with N-silyl enamines 6′. All of the proposed N-silyl enamines (6a′, 6m′, were generated with MePhSiH2 via the proper conversion of 5a, 5m, and 5q. Du NMR reaction monitoring, we observed a faster cycloaddition of the electron-rich enamine 6m′ than the electron-poor N-silyl enamine 6q′. This result verified the tion between the electron density of N-silyl enamine 6′ and the reaction rate of cloaddition step. The reaction rate in each step was found to be relatively slow; thus, the electronic effect of the quinolines on the reaction rate could be observed through in situ NMR monitoring. (Supplementary Materials). We compared the initial reaction progression through 1 H NMR using quinolines containing different substituents (Scheme 6). The borane catalyzed hydrosilylation step of the quinolines (5a, 5m, and 5q) was first investigated with Ph 2 SiH 2 . The initial reaction rate of 5-bromoquinoline (5q) was faster than 6-methoxyquinoline (5m) as expected. This indicated that the rate-determining step in the hydrosilylation was the dearomative borohydride addition [24]. Next, we examined the [3 + 2] cycloaddition step with N-silyl enamines 6 . All of the proposed N-silyl enamines (6a , 6m , and 6q ) were generated with MePhSiH 2 via the proper conversion of 5a, 5m, and 5q. During the NMR reaction monitoring, we observed a faster cycloaddition of the electron-rich N-silyl enamine 6m than the electron-poor N-silyl enamine 6q . This result verified the correlation between the electron density of N-silyl enamine 6 and the reaction rate of the cycloaddition step. The synthetic applicability of the presented cyclic acyl amidine synthesis was th explored (Scheme 7). A 10 mmol initial scale reaction of 1a smoothly produced 1.63 g (42 yield) of cyclic amidine 4a. This result demonstrated the scalability of our methodolo for the potential preparation of practical amounts of useful cyclic acyl amidines.

Conclusions
We successfully prepared cyclic acyl amidines from versatile N-silyl enamines an acyl azides. An electronically activated acyl azide (3i) was found to be the most optim acyl azide for the [3 + 2] cycloaddition step. The substrate scope from the N-silyl enami from isoquinolines to quinolines is wide. The progression of the initial reaction was mo itored using NMR, which clearly demonstrated the relationship between the electron de sity of N-silyl enamine and reactivity in each synthetic step. Finally, the synthetic utili of the proposed methodology was demonstrated via the gram-scale reaction.

General Considerations
Unless otherwise stated, all catalytic reactions were carried out under an argon a mosphere. Chloroform-d was purchased from Cambridge Isotope Laboratories, In (Tewksbury, MA, USA), degassed and used as a solvent without additional purificatio for optimization, as a substrate scope. Tris(pentafluorophenyl)borane was purchas from TCI korea (Seoul, South Korea) and Acros (ThermoFisher Korea, Seoul, South K Scheme 6. Relative rates of hydrosilylation and [3 + 2] cycloaddition: (a) Relationship between the relative rate of first de-aromative hydrosilylation and substituents on quinoline; (b) Relationship between the relative rate of [3 + 2] cycloaddition and substituents on quinoline. * NMR yields were determined by using 1,2-tetrachloroethane as an internal standard.
The synthetic applicability of the presented cyclic acyl amidine synthesis was then explored (Scheme 7). A 10 mmol initial scale reaction of 1a smoothly produced 1.63 g (42% yield) of cyclic amidine 4a. This result demonstrated the scalability of our methodology for the potential preparation of practical amounts of useful cyclic acyl amidines.
Molecules 2022, 27, x FOR PEER REVIEW Scheme 6. Relative rates of hydrosilylation and [3 + 2] cycloaddition: (a) Relationsh relative rate of first de-aromative hydrosilylation and substituents on quinoline; (b between the relative rate of [3 + 2] cycloaddition and substituents on quinoline. * N determined by using 1,2-tetrachloroethane as an internal standard. The synthetic applicability of the presented cyclic acyl amidine synt explored (Scheme 7). A 10 mmol initial scale reaction of 1a smoothly produc yield) of cyclic amidine 4a. This result demonstrated the scalability of our for the potential preparation of practical amounts of useful cyclic acyl amid Scheme 7. Gram-scale reaction of the proposed cyclic acyl amidine synthesis.

Conclusions
We successfully prepared cyclic acyl amidines from versatile N-silyl acyl azides. An electronically activated acyl azide (3i) was found to be the acyl azide for the [3 + 2] cycloaddition step. The substrate scope from the N from isoquinolines to quinolines is wide. The progression of the initial reac itored using NMR, which clearly demonstrated the relationship between th sity of N-silyl enamine and reactivity in each synthetic step. Finally, the s of the proposed methodology was demonstrated via the gram-scale reactio

General Considerations
Unless otherwise stated, all catalytic reactions were carried out unde mosphere. Chloroform-d was purchased from Cambridge Isotope Lab (Tewksbury, MA, USA), degassed and used as a solvent without addition Scheme 7. Gram-scale reaction of the proposed cyclic acyl amidine synthesis.

Conclusions
We successfully prepared cyclic acyl amidines from versatile N-silyl enamines and acyl azides. An electronically activated acyl azide (3i) was found to be the most optimal acyl azide for the [3 + 2] cycloaddition step. The substrate scope from the N-silyl enamine from isoquinolines to quinolines is wide. The progression of the initial reaction was monitored using NMR, which clearly demonstrated the relationship between the electron density of N-silyl enamine and reactivity in each synthetic step. Finally, the synthetic utility of the proposed methodology was demonstrated via the gram-scale reaction.

General Considerations
Unless otherwise stated, all catalytic reactions were carried out under an argon atmosphere. Chloroform-d was purchased from Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA), degassed and used as a solvent without additional purification for optimization, as a substrate scope. Tris(pentafluorophenyl)borane was purchased from TCI korea (Seoul, Korea) and Acros (ThermoFisher Korea, Seoul, Korea), and was stored at −15 • C. All other reagents were directly used as purchased without further purification unless otherwise stated.
Analytical thin-layer chromatography (TLC) was performed on pre-coated silica gel 60 F254 plates (Intertechnologies, Seoul, Korea). Visualization on TLC was achieved by the use of UV light (254 nm, Collégien, France), exposure to treatment with acidic p-anisaldehyde, phosphomolybdic acid and potassium permanganate stain followed by heating. Column chromatography was undertaken on silica gel (400-630 mesh) using a proper eluent. 1 H NMR (Jeol, Tokyo, Japan) was recorded using Jeol ECZ-500R (500 MHz) for the characterization of compounds. Chemical shifts were quoted in parts per million (ppm), referenced to tetramethylsilane: 0.00 ppm (singlet). Furthermore, 13 C{ 1 H} NMR (Jeol, Tokyo, Japan) was recorded on Jeol ECZ-500R (125 MHz) and was fully decoupled by broad-band proton decoupling. Chemical shifts were reported in ppm referenced to the center of a triplet at 77.0 ppm of CDCl 3 . Infrared (IR, PerkinElmer Korea, Seoul, Korea) spectra were recorded using a Perkin Elmer Frontier ATR-FT-IR spectrometer, ν max in cm −1 . High resolution mass spectra (Jeol, Tokyo, Japan) were obtained by using EI and FAB method from Korea Basic Science Institute (Daegu). X-ray diffraction (Bruker, Billerica, MA, USA) data were collected using a Bruker D8 QUEST coated with Parabar oil under a stream of N 2 (g) at 173 K.

Substrate Scope of the Isoquinoline for the Synthesis of Acyl Amidine
Step 1: To a B(C 6 F 5 ) 3 catalyst (0.025 mmol, 5 mol%) in an NMR tube CDCl 3 (0.5 mL) and silanes (0.6 mmol, 1.2 equiv.) were added at room temperature, H 2 bubbles were observed and TCE (0.3 mmol) or mesitylene was added as internal standard. Isoquinolines (1a, 1i-1p) (0.5 mmol, 1.0 equiv.) was subsequently added to the above solution and quickly shaken once before heating up to 110 • C in an oil bath for the indicated reaction time. The mixture was subjected to an NMR to verify the conversion and yields of reactions.
Step 2: In the crude reaction mixture from the first step, acyl azide 3i (0.5 mmol, 1.0 equiv.) was added at room temperature and the NMR was indicated in the reaction time. The resulting mixture was quenched by MeOH addition, silica filter, and DCM wash. The resulting crude mixture was purified by column chromatography.