Mechanism and Rate-determining Factors of Amide Bond Formation

Feb 12, 2018 - Acyl transfer of in-situ generated mixed anhydrides is an important method for the amide bond formation from short linkages with the ea...
1 downloads 11 Views 3MB Size
Article pubs.acs.org/joc

Cite This: J. Org. Chem. 2018, 83, 2676−2685

Mechanism and Rate-Determining Factors of Amide Bond Formation through Acyl Transfer of Mixed Carboxylic−Carbamic Anhydrides: A Computational Study Yuan-Ye Jiang,* Tian-Tian Liu, Rui-Xue Zhang, Zhong-Yan Xu, Xue Sun, and Siwei Bi* School of Chemistry and Chemical Engineering, Qufu Normal University, Qufu 273165, People’s Republic of China S Supporting Information *

ABSTRACT: Acyl transfer of in situ-generated mixed anhydrides is an important method for amide bond formation from short linkages with the easily removed byproduct CO2. To improve our understanding of the inherently difficult acyl transfer hindered by the large ring strain, a density functional theory study was performed. The calculations indicate that the amidation of activated α-aminoesters and N-protected amino acids is more likely to proceed via the self-catalytic nucleophilic substitution of the two substrates and the subsequent 1,3-acyl transfer. By comparison, the mechanism involving 1,5-acyl transfer is less kinetically favored because of the slow homocoupling of activated α-aminoesters. Furthermore, we found that the detailed mechanism of 1,3-acyl transfer on the mixed carboxylic−carbamic anhydrides depends on the catalysts. Strong acidic catalysts and bifunctional catalysts both lead to stepwise pathways, but their elementary steps are different. Basic catalysts cause a concerted C−N bond formation/decarboxylation pathway. The calculations successfully explain the reported performances of different Brønsted-type catalysts and substrates, which validates the proposed mechanism and reveals the dependence of the reaction rates on the acid−base property of catalysts and the acidity of substrates.

1. INTRODUCTION Robust and efficient amide bond formation is the core issue of peptide synthesis1 and also finds wide applications in synthetic organic chemistry and medical chemistry for the preparation of various amide-containing compounds.2,3 Although amides are traditionally synthesized through direct condensation of amines and activated carboxylic acids in the presence of various coupling reagents,4 the related methods suffer from toxic or sensitive substrates, e.g., acyl halides and anhydrides. To solve this problem, considerable attention has recently been paid to the amide bond formation processes involving more friendly and stable carbon and/or oxygen materials such as alcohols,5 ketones,6 carboxylic acids,7 carbon monoxide,8 and water.9 However, these methods usually involve the use of transition metal catalysts or external oxidants, which are generally not very compatible with bioactive peptides. In this context, more effort has been spent on alternative strategies of stoichiometric or organocatalytic amide bond formation reactions under redox-neutral conditions.10 Some earlier methods include native chemical ligation (NCL)11 and Staudinger ligation12 in which the amide bond is generated by using thioesters or phosphine-containing esters as substrates. More recently, novel substrates such as α-ketocarboxylic acids,13 acyltrifluoroborates,14 selenoesters,15 and in situ-generated amides16 have been developed to enrich the set of tools used for amide bond formation. Intramolecular acyl transfer is one common core step in the new amidation methods described above, which is usually limited to 1,4- or 1,5-acyl transfer due to the consideration of ring stain (Scheme 1).10 Examining whether other types of acyl © 2018 American Chemical Society

Scheme 1. Amidation via Intramolecular Acyl Transfer

transfer can also be used in the amidation process, which may improve the synthetic flexibility for peptide and amide synthesis, remains interesting. In this respect, the 1,3-acyl transfer shown in Scheme 1 represents a unique process in the amidation reactions. For example, the coupling of (thio)carboxylic acids or thioesters with iso(thio)cyanates generates mixed anhydrides, which further rearrange to amides with the easily removed CO2, COS, or CS2 as the byproduct (Scheme 2a).17 Similarly, the couplings of carboxylic acids with isonitriles followed by 1,3 O-to-N acyl transfer (Scheme 2b)18 and the couplings of thioacids and dithiocarbamate-terminal amines Received: December 8, 2017 Published: February 12, 2018 2676

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry

indicates that the relevant mechanism is variable and depends on the properties of catalysts. Strong acidic catalysts and bifunctional catalysts lead to stepwise mechanisms but via different intermediates. Basic catalysts accomplish this reaction via a concerted C−N bond formation/decarboxylation mechanism. These results are consistent with the reported efficiencies of different catalysts and substrates, from which the relationship between the acid−base properties of catalysts and substrates and the kinetics of 1,3-acyl transfer were validated.

Scheme 2. Representative Amidation Reactions via 1,3-Acyl Transfer

2. COMPUTATIONAL DETAILS All calculations were performed by employing Gaussian0924 in the solution phase with solvation model SMD (dichloromethane solvent),25 density functional theory method M06-2x,26 and basis set def2-SVP.27 An ultrafine grid was assigned to the DFT calculations to avoid possible integration grid errors.28 At the same level of theory, frequency analysis was conducted to verify that the optimized structures are intermediates or transition states, and also to obtain thermodynamic corrections. Intrinsic reactant coordinate (IRC) analysis was performed to ensure that the transition states connect suitable intermediates;29 1.9 kcal/mol was added to all species to account for the standard state change from 1 atm to 1 mol/L at 298.15 K.30 Meanwhile, a moderate correction of addition of 2.6 kcal/mol to the calculated solution-phase Gibbs free energies was used because the entropy effect was generally considered to be overestimated for the reactions from m to n components (m ≠ n).31 Natural bond orbital (NBO) analysis was performed by employing NBO version 3.1 implemented in Gaussian 09.32 The three-dimensional (3D) diagrams of the optimized structures were generated using CYLView.33 Note that recalculating solution-phase single-point energies based on larger basis set def2-TZVP27 and 6-311+G(2d,p) was tested; however, the corresponding results overestimate the overall energy barrier, whereas the relative feasibilities of 1,3- and 1,5-acyl transfer mechanisms were not changed (Table S1). Therefore, unless mentioned otherwise, the solution-phase Gibbs free energies calculated with M06-2x/def2-SVP/ SMD after the corrections described above, referring to 1 mol/L and 298.15 K, were used for the following discussion.

followed by the removal of CS2 were developed (Scheme 2c).19 Recently, amidation with N,N′-carbonyldiimidazole (CDI)activated α-aminoesters and N-protected amino acids with CO2 extrusion was also disclosed (Scheme 2d).20 The amidation is found to be promoted by both Brønsted acids and Lewis acids and works at room temperature in the absence of a base. The common amine protecting groups, i.e., Fmoc, Boc, and Cbz, are compatible with this method, and no epimerization was detected for the sensitive cysteine residue. The synthesis of a tetrapeptide in the reverse N → C direction (instead of the conventional C → N direction) was performed with good overall yield and purity. Previous mechanistic studies of acyl transfer were focused on the processes experiencing smaller ring strain,21 whereas less effort has been spent on the more challenging 1,3-acyl transfer.22,23 Therefore, several fundamental questions such as the detailed mechanism of 1,3-acyl transfer, the relative feasibility of the mechanism involving 1,3-acyl transfer compared with that of other possible mechanisms, and the factors benefiting the 1,3-acyl transfer have not been fully answered for some reported systems. To improve our understanding of the acyl transfer involving strained transition states, we performed a mechanistic study with the aid of density functional theory (DFT) calculations. Considering the satisfactory performance of the amidation of CDI-activated αaminoesters and N-protected amino acids as well as its potential usage for peptide synthesis in an unconventional direction, we chose the 1-hydroxybenzotriazole (BtOH)catalyzed relevant reaction reported by Campagne, de Figueiredo, and co-workers20a for the mechanistic study first. Our study supports the idea that the pathway involving 1,3-acyl transfer is indeed kinetically favored over the one involving 1,5acyl transfer overall because of the slow homolytic nucleophilic substitution of the activated α-aminoesters. More interestingly, further investigation of the rate-determining 1,3-acyl transfer of mixed anhydrides under different experimental conditions

3. RESULTS AND DISCUSSION In the following discussions, CDI-activated α-aminoester 1 and acetic acid 2 were chosen as the model reactants and BtOH was chosen as the catalyst for mechanistic study first (Scheme 3). 1 was proposed to react with 2 to generate mixed anhydride Int1. Thereafter, intramolecular 1,3-acyl transfer from Int1 can lead to the amide product. On the other hand, we hypothesize that the coupling of Int1 and another 1 can generate Int3, from which a more common 1,5-acyl transfer can occur to lead to the amide product. Meanwhile, the homocoupling of 1 also possibly generates Int2 that further evolves into Int3. In the following, the generation of Int1 and Int3 was considered first. Then, the two competitive acyl transfer processes with their associated subsequent steps were compared to find the most favorable pathway among them. Thereafter, the favorable acyl transfer mechanisms under different conditions were investigated to explore the relationship of the catalysts, the substrates, and the reaction rates. 3.1. Generation of the Precedent Intermediates of Acyl Transfer. 3.1.1. Generation of Int1. The nucleophilic substitution of 1 with 2 can generate Int1 and an imidazole. Because no extra bases were used in the model reaction, the concerted C−O bond formation/proton transfer via the fivemembered-ring transition state TS1 was considered first (see Figure 1 for the energy profile and Figure 2 for the 3D diagram). In this process, no prior deprotonation of 2 occurs while the nucleophilicity of 2 is expected to be much lower than 2677

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry

deprotonation of 2 by 1 to generate 3 and an acetate is considered. However, completely separating cationic 3 and acetate in CH2Cl2 causes a free energy increase of 36.3 kcal/ mol, suggesting that concerted deprotonation/substitution is more plausible. In this case, proton shuttles are expected to be involved in this concerted process because of the long distance between the C1 atom and the N2 atom. Taking these factors into account, we identified two transition states (TS2 and TS3) in which one and two acetic acids act as proton shuttles, respectively,34 for the generation of Int1. TS2 and TS3 are both late transition states with short N2−Hx (x = 2 or 3) bonds and relatively long O2−H1 bonds, indicating that the deprotonation of the carboxylic acid and protonation of the heteroaromatic ring are important for the substitution. Because TS3 has two acetic acids as the proton shuttles, all the H··· Ocarbonyl hydrogen bonds in TS3 are allowed to be shorter than 1.8 Å. By comparison, only one acetic acid acts as the proton shuttle in TS2, and the H2···O3 bond distance has reached 2.22 Å, suggesting a weaker intramolecular hydrogen bond. Possibly for this reason, TS3 is favored over TS2 by 3.8 kcal/mol. After TS3, two acetic acids and one imidazole are released and the mixed anhydride Int1 is formed. It is noted that the long-range proton transfers facilitated by various proton shuttles were also reported in other types of reactions.35 The generation of Int1 from 1 and 2 causes a free energy increase of 4.4 kcal/mol, and this result is consistent with the lower stability of anhydride with respect to amide. 3.1.2. Generation of Int3. The further coupling of Int1 and another 1 can generate Int3. The acidity of the N−H bond of an amide is much lower than that of the O−H bond of a carboxylic acid,36 and thus, the predeprotonation of Int1 should be more difficult than that of carboxylic acids under the reported experimental conditions. Hence, a concerted proton transfer/C−N bond formation assisted by two acetic acids (via TS4) was also considered. In TS4, the N3−H1 bond length is 1.51 Å, which is already close to the O2−H1 bond distance (1.55 Å) in TS3. Because of the lower acidity of the N−H bond of an amide, heterolytic cleavage of the N−H bond to the same degree should be more difficult than that of the O−H bond. Meanwhile, the emerging C1−N3 bond in TS4 is quite weak with a long distance of 1.91 Å. As a result, the energy barrier for the generation of Int3 from Int1 reaches 32.6 kcal/mol, much higher than that for the generation of Int1 from 1. Associated with the weak nucleophilicity of the amide, TS4 affords a tetrahedral intermediate 4, from which the elimination of an imidazole smoothly occurs via TS5 to generate Int3. The transition states of C1−N3 bond formation (TS4) and C1−N1 bond cleavage (TS5) have close relative Gibbs free energies (37.0 and 35.6 kcal/mol, respectively). By contrast, the previous theoretical study of the S-to-N acyl transfer indicates that C−N bond formation is more facile than C−S bond cleavage.21b These different results probably result from the different strengths of C−N and C−S bonds. Similar to the transformations from Int1 to Int3 via 4, the homocoupling of 1 to yield Int2 can be completed with the assistance of two acetic acids via TS6, TS5, and TS7 successively (Scheme 4b). The free energy barrier of these transformations is ∼35 kcal/mol. Thereafter, Int2 can evolve into Int3 via TS8 with a free energy barrier of 14.1 kcal/mol. This barrier is lower than that of the transformation from 1 to Int1. We attribute the difference to the substitution of the N− H bond of 1 with CONHMe, by which the carbonyl group of 1 is more electronic-deficient (see NBO charges in Scheme S1).

Scheme 3. (a) Model Reaction for Mechanistic Study and (b) Possible Mechanisms for Amide Bond Formation of Carboxylic Acids and CDI-Activated α-Aminoesters

Figure 1. Calculated energy profile for the generation of Int1 from 1 and 2. Solution-phase Gibbs free energies (ΔGsol) are given in kilocalories per mole.

that of its deprotonated form. Probably for the reason mentioned above, the C1−N1 bond is almost broken in TS1 (the C1−N1 bond length is 2.47 Å) to improve the electrophilicity of the C1 atom, but no significant C1−O1 bond formation is observed in TS1 (the C1−O1 bond length is 2.59 Å). Therefore, the corresponding energy demand for the concerted process becomes quite high (50.9 kcal/mol), and we excluded this pathway accordingly. We noticed that the heteroaromatic ring of 1 can act as a base; thus, the 2678

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry

Figure 2. Optimized structures of key intermediates and transition states for the generation of Int1 (bond lengths are given in angstroms).

3.1.3. Comparison of the Generation of Int1 and Int3. According to the calculations described above, the generation of Int3 either from Int1 directly or via Int2 is less kinetically favorable than the generation of Int1 from 1 and 2 by more than 12.8 kcal/mol. On the other hand, considering the high free energy barrier for the generation of Int3, further investigation of the hypothesized 1,5-acyl transfer on Int3 appears to be not necessary. However, it should be pointed out that the calculated absolute energy barriers are less reliable than calculated relative kinetic feasibility.37 Thus, the comparison of the calculated relative energy barriers of different mechanisms is more appealing for judging the feasibility of a mechanism. In light of this, the remaining steps shown in Scheme 3 are still examined in the next section. 3.2. Amide Bond Formation from Int1 and Int3. 3.2.1. 1,3-Acyl Transfer from Int1. First, the direct 1,3-acyl transfers on Int1 via the four-membered-ring transition state TS9 and TS10 in the absence of any other species were considered first (Scheme 5 and Figure 3). The two methyl groups of Int1 are trans to each other in TS9 and are cis to each other in TS10. Because of the steric repulsion of the two methyl groups, TS10 is slightly less stable than TS9 by 1.3 kcal/mol. After TS9 and TS10, a CO2 is completely released and the amide product forms immediately. However, TS9 and TS10 lead to high overall energy barriers of 37.1 and 38.4 kcal/ mol, respectively, suggesting that these pathways are less possible. Next, one BtOH was added to TS9 to form a hydrogen bond with the carbonyl group to increase its electrophilicity (Scheme 5b and Figure 3). As a result, C−N bond formation via TS11 is remarkably more facile than that via TS9 by 7.6 kcal/mol. In addition, TS11 does not directly generate the amide product and CO2 but affords a transient intermediate 6 that is very energetically close to TS11. In 6, the C2−N1 bond is only partially formed with a Wiberg bond order of 0.49. From 6, another transition state TS12 is reached before the formation of the final product with a puny barrier of 2.9 kcal/mol. In this

step, the C1−N1 bond and C2−O1 bond stretch while the C2− N1 bond is further shortened. A similar stepwise acyl transfer process has been found in our recent mechanistic study of chlorosilane-catalyzed amidation reactions.23c Akin to the case described above, the stepwise acyl transfer via TS13 and TS14 in which the two methyl groups are cis to each other was considered. Different from the case of TS9 and TS10, TS13 and TS14 are more stable than TS11 and TS12, respectively. We noticed that BtOH forms two hydrogen bonds with Int1 in TS13 and TS14, and thus, the disfavored steric repulsion in the cis configuration is overcome. In addition to the TS13 → 7 → TS14 pathway, we further considered a pathway that starts from the proton transfer on 6 to generate a cyclic hemiaminal, which further converts into acetyl(methyl)carbamic acid, then to N-methylacetimidic acid, and finally to the amide product. The overall energy barrier of this pathway is 28.7 kcal/mol, and these results appear in Scheme S2. Inspired by the results presented above, we examined the stepwise acyl transfer via TS15 and TS16 (Scheme 5c and Figure 3). In the two transition states, the two methyl groups are trans to each other to prevent their steric repulsion. Meanwhile, two BtOH groups are present to form a hydrogen bond with the carbonyl group and the N−H bond of Int1, and thus, the electrophilicity of the carbonyl group and the nucleophilicity of the amide are both expected to be improved. For these two reasons, the overall free energy barrier of acyl transfer is further lowered to 27.1 kcal/mol. 3.2.2. Amide Bond Formation from Int3. The 1,5-acyl transfer on Int3 is the first step for the generation of the amide product from Int3, and two pathways assisted by BtOH or HOAc were considered for the transformation of Int3 into Int5 (Figure 4). In the presence of BtOH, the 1,5-acyl transfer proceeds via concerted proton transfer/C−N bond formation (TS17) and the slower concerted proton transfer/decarboxylation (TS18). In a similar pattern, HOAc promotes the 1,5acyl transfer via concerted proton transfer/C−N bond formation (TS19), proton transfer (TS20), and the rate2679

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry Scheme 4. Calculated Energy Profiles for the Generation of Int3 (a) from Int1 Directly and (b) via Int2a

Scheme 5. Calculated Energy Profile of Amide Formation from Int1

a

Solution-phase Gibbs free energies (ΔGsol) are given in kilocalories per mole.

groups and the p orbital of the nitrogen atom. By contrast, the conjugation of the p orbital of the nitrogen atom and only one carbonyl group is wrecked in TS23, and thus, TS23 is more stable than TS22. The thermodynamic stability of 12 and 13 is close to that of Int5 and imidazole. If no other reactions occur to transform 12 and/or 13 into more stable species, a fast equilibrium between the two states is expected under the experimental condition while a sufficiently high temperature can produce the final amide product from Int5 via TS22. 3.3. Overall Mechanistic Insights. On the basis of the all calculations, a simplified overall energy profile of amide formation from CDI-activated aminoester 1 and carboxylic acid 2 via 1,3- or 1,5-acyl transfer is shown in Figure 5. The rate-determining steps of the amidation via 1,3-acyl transfer are the C−N bond formation from the mixed anhydride Int1 as well as the following decarboxylation, with a calculated overall energy barrier of ∼27 kcal/mol. The rate-determining steps of the amidation via 1,5-acyl transfer lie in the homocoupling of 1 to generate Int2 with a calculated overall energy barrier of ∼36 kcal/mol. In addition, even if Int3 is directly used as a reactant for amidation via 1,5-acyl transfer, the overall efficiency is still lower than that of the catalytic coupling of 1 and 2 due to slow decarboxylation (via TS21) and the possible competitive reactions, i.e., the faster generation of 12 and 13, and their

determining concerted proton transfer/decarboxylation (TS21). The large energy gaps between 9 and TS18 and between 11 and TS21 possibly result from the twist of the sixmembered ring. 9 and 11 are in half-chair form because of the conjugation of the two carbonyl groups and the amine. TS18 and TS21 are in boat form because the amine is going to be protonated and the hybridization of the amine nitrogen atom changes from sp2 to sp3. Meanwhile, BtOH is a shorter shuttle for the proton transfer than HOAc, and the twist of the sixmembered ring is expected to be more serious in TS18 than in TS21, making TS21 more stable than TS18 accordingly. We calculated the electronic energies of the twisted six-memebered ring in TS18 and TS21 and found that the former is 6.4 kcal/ mol higher, which supports our proposal (Scheme S3). Int5 can react with an imidazole and two HOAc groups to yield the amide pro and regenerate a 1 via TS22. This process is akin to the reverse process of the transformation of two 1 compounds into Int2 (Scheme 4b). On the other hand, imidazole possibly attacks the other carbonyl group of Int5 via TS23 to generate 12 and 1,3-dimethylurea 13. TS23 is energetically lower than TS22 by 11.2 kcal/mol.38 We notice that the hybridization of the carbon and nitrogen atoms of the cleaving C−N bonds changes from sp2 to sp3 in both TS22 and TS23. This factor wrecks the conjugation of the two carbonyl 2680

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry

Figure 3. Optimized structures of selected transition states in BtOH-catalyzed 1,3-acyl transfer from Int1. Bond lengths are given in angstroms. Some parts of the structures have been omitted for the sake of clarity.

Figure 4. Calculated energy profile of amide formation from Int3. Solution-phase Gibbs free energies are given in kilocalories per mole.

unknown downstream transformations. Although C−N bond formation via 1,5-acyl transfer (via TS19) is more facile than that via 1,3-acyl transfer (via TS15) with respect to the corresponding preceding intermediates, our calculations indicate that the previously proposed amide bond formation mechanism via 1,3-acyl transfer is still kinetically favored overall. It should be pointed out that the calculated energy barrier for the amidation via 1,3-acyl transfer is overestimated as this reaction works at room temperature. The overestimated energy barrier possibly results from the overestimated entropy effect

mentioned above. If a stronger correction, i.e., adding 4.3 kcal/ mol (instead of 2.6 kcal/mol) to the energy of each species, which has also been widely used in recent computational studies,39 is employed herein, a formally more reasonable overall energy barrier of 23.7 kcal/mol can be present while the 1,3-acyl transfer is still favored over the 1,5-acyl transfer overall. On the contrary, if no entropy correction is employed, 1,3-acyl transfer is also favored over 1,5-acyl transfer by 12.5 kcal/mol (Figure S1) but the overall energy barrier of 1,3-acyl transfer increases to 30.9 kcal/mol. Therefore, in this case, the entropy correction is important only for obtaining a formally reasonable 2681

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry

Figure 5. Simplified overall energy profile for amide bond formation from 1 and 2 via 1,3- or 1,5-acyl transfer. Solution-phase Gibbs free energies are given in kilocalories per mole.

overall energy barrier but does not impact the conclusion that 1,3-acyl transfer catalyzed by BtOH proceeds via a stepwise mechanism and is favored over 1,5-acyl transfer. 3.4. 1,3-Acyl Transfer of Mixed Anhydrides under Other Conditions. The overall energy barrier of the 1,3-acyl transfer of Int1 without the aid of other species is higher than that of the BtOH-assisted one by ∼10 kcal/mol (TS9 vs TS15 and TS16). In the latter case, BtOH acts as a Brønsted acid catalyst and Brønsted base catalyst to form two hydrogen bonds in TS15 and TS16. As a result, both the electrophilicity of the carbonyl group and the nucleophilicity of the amide of Int1 increase, by which the inherently difficult 1,3-acyl transfer can be realized. It is noted that the amide product can also form in the absence of BtOH with a lower yield.20a We speculate that the carboxylic acid itself acts as a catalyst in this case. Indeed, replacing the BtOH with HOAc slows the stepwise 1,3-acyl transfer via TS24 by 3.0 kcal/mol, though HOAc is expected to be more acidic than BtOH (see Scheme 6 for the key transition states and Scheme S4 for the remaining steps). Meanwhile, TS25, the transition state of the 1,3-acyl transfer assisted by two HOAc groups, is even slightly less stable than TS24, and this trend is different from those for TS13 and TS15. We speculate that the activation effect of the less basic HOAc toward the amide N−H bond is weaker than that of BtOH, and the extra entropy decrease induced by the second HOAc cannot be overcome at the current level of entropy effect correction. Interestingly, if the acidity of the catalyst is further increased, e.g., using p-toluenesulfonic acid (pTsOH) as the catalyst, the relevant 1,3-acyl transfer becomes more kinetically favored than the BtOH-catalyzed one by 1.4 kcal/mol. In this case, C−N bond formation associated with a concerted proton transfer via TS26 occurs, after which a hemiaminal intermediate forms. Thereafter, concerted proton transfer/C−O bond cleavage to generate an acetyl(methyl)carbamic acid, decarboxylation, and tautomerization of N-methylacetimidic acid occur in turn to yield the amide product (see Scheme S5 for full details). The calculated energy barrier also explains the good performance of pTsOH in the reported amidation reactions.20a The aforementioned results indicate that the activation by an acidic catalyst toward the carbonyl group is important to the

Scheme 6. Calculated Results of 1,3-Acyl Transfer on a Mixed Anhydride under (a) Acidic and (b) Basic Conditionsa

a

Solution-phase Gibbs free energies are given in kilocalories per mole.

1,3-acyl transfer. However, it has been known that similar 1,3acyl transfer reactions proceed well under basic conditions.17 To check the validity of our proposed mechanistic model, the 1,3-acyl transfer with Me3N as the catalyst was investigated. Different from the acid-catalyzed cases, a concerted C−N bond formation/decarboxylation via TS27 was found for the amide bond formation. The related energy barrier is 26.3 kcal/mol with respect to Int1 and even higher than that of the HOAccatalyzed case by 0.6 kcal/mol. If we change the methyl group on the amide nitrogen atom to a phenyl group, the energy barrier of 1,3-acyl transfer via TS28 decreases to 20.2 kcal/mol with respect to the mixed anhydride Int6. These results qualitatively explain the different phenomena reported by Campagne, Crich, Gaertner, and their co-workers; i.e., using amines as the catalysts, no reaction was observed for the Nalkylated CDI-activated α-aminoesters,20a while the amide bond formation of isocyanatobenzene derivatives and carboxylic acids occurs facilely.17a,b 2682

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry

BtOH promote the 1,3-acyl transfer by simultaneously acting as a hydrogen bond donor and a hydrogen bond acceptor to activate the carbonyl group and the amide N−H bond, respectively. In contrast to the three types of catalysts, the weaker acids like carboxylic acids or weaker bases are expected to be less efficient. This work provides the first systematic theoretical study of decarboxylative 1,3-acyl transfer. These calculations clarify the detailed mechanism and further reveal the self-catalytic role of carboxylic acids, the relationship of the mechanism, the reaction rates, and the acid−base properties of catalysts and substrates. These results are qualitatively consistent with the experimental observations, which supports our proposed reaction models. We believe this work also improves our understanding of the 1,3-acyl transfer of mixed anhydride analogues and hope that it can provide some inspiration for the further development of amide bond formation via acyl transfer with respect to catalyst design and the choice of substrates.

Taking all the factors into account, we propose that three classes of catalysts can effectively promote the 1,3-acyl transfer of the mixed anhydrides. One class consists of strong acids that can effectively improve the electrophilicity of the carbonyl group by protonating it. The second class consists of strong bases that can effectively improve the nucleophilicity of the amide nitrogen by a hydrogen bond or even deprotonation of it. The third class consists of bifunctional catalysts that can activate both the carbonyl group and the amide nitrogen, and the activation effect from each of two aspects is not necessarily as strong as that of the first and second classes of catalysts. In contrast, the acids with acidities between those of the first and third classes or the bases with basicities between those of the second and third classes are expected to be inferior. With respect to the substrates, it is easy to imagine that the mixed anhydrides bearing more electronic-deficient carbonyl groups and more acidic amide N−H bonds are superior. Finally, it should be pointed out that the success of the 1,3-acyl transfer of mixed anhydrides or the analogues tightly connects with the facile release of gaseous byproducts (Scheme 2). Without such a process, the formation of secondary amides is blocked or tertiary amides are formed instead as shown in Scheme 2b.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.joc.7b03107. NBO charges, calculated energy profile of less favored pathways, electronic energy difference between the parts of TS18 and TS21, HOAc- and pTsOH-catalyzed 1,3acyl transfer on Int1, simplified overall energy profile without entropy corrections, relative energies of key transition states calculated with larger basis sets, calculated solution-phase Gibbs free energies after corrections, and Cartesian coordinates of all species (PDF)

4. CONCLUSIONS To gain deeper mechanistic insights into the amide bond formation through acyl transfer of in situ-generated mixed carboxylic−carbamic anhydrides, we performed a DFT study and reached the following major conclusions. (1) Amide bond formation from CDI-activated aminoesters and carboxylic acids was found to occur more easily by 1,3-acyl transfer on the mixed carboxylic−carbamic anhydride Int1 than by the 1,5-acyl transfer on the mixed anhydride Int3 overall. This is because, in the latter pathway, the homolytic nucleophilic substitution of the CDI-activated aminoesters is more kinetically difficult. (2) The more favored amidation mechanism involves two major steps, i.e., the nucleophilic substitution of CDI-activated aminoesters by carboxylic acids to generate Int1 and the subsequent 1,3-acyl transfer. Furthermore, we found that carboxylic acids themselves can act as a proton shuttle to assist the nucleophilic substitution and the 1,3-acyl transfer proceeds in different pathways based on the catalysts. The bifunctional catalyst BtOH accomplishes the 1,3-acyl transfer via a stepwise mechanism in which C−N bond formation and decarboxylation are both rate-determining steps. With a strong acid (like pTsOH) as the catalyst, the 1,3-acyl transfer also proceeds via a stepwise mechanism that involves the rate-determining concerted proton transfer/C−N bond formation to generate a hemiaminal intermediate, concerted proton transfer/C−O bond cleavage to generate an N-acylated carbamic acid, decarboxylation, and tautomerization of imidic acid. With a base like amine as the catalyst, the 1,3-acyl transfer is achieved via a concerted mechanism in which C−N bond formation and decarboxylation occur spontaneously. (3) The intrinsically difficult 1,3-acyl transfer of the mixed anhydride can be facilitated by three types of catalysts. Strong acids promote this process by directly protonating the carbonyl group to improve its nucleophilicity. Bases can realize the 1,3acyl transfer of the mixed anhydrides by improving the electrophilicity of the amide nitrogen via a hydrogen bond or directly deprotonating it. If the basicities of the catalysts, e.g., amines, are not sufficiently high, relatively acidic N−H bonds of the mixed anhydrides are required. Bifunctional catalysts like



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Yuan-Ye Jiang: 0000-0002-4763-9173 Siwei Bi: 0000-0003-3969-7012 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China (21702119, 21473100, 21603116, and 21703118), the Natural Science Foundation of Shandong Province, China (ZR2017QB001 and ZR2017MB038), the Special Fund Project for Postdoctoral Innovation of Shandong Province (201602021).



REFERENCES

(1) (a) Burkhart, B. J.; Schwalen, C. J.; Mann, G.; Naismith, J. H.; Mitchell, D. A. Chem. Rev. 2017, 117, 5389−5456. (b) Bode, J. W. Acc. Chem. Res. 2017, 50, 2104−2115. (c) Burke, H. M.; McSweeney, L.; Scanlan, E. M. Nat. Commun. 2017, 8, 15655. (d) Huang, Y. C.; Fang, G. M.; Liu, L. Natl. Sci. Rev. 2016, 3, 107−116. (e) Li, H.; Dong, S. Sci. China: Chem. 2017, 60, 201. (2) (a) de Figueiredo, R. M.; Suppo, J.-S.; Campagne, J.-M. Chem. Rev. 2016, 116, 12029−12122. (b) De Risi, C.; Pollini, G. P.; Zanirato, V. Chem. Rev. 2016, 116, 3241−3305. (c) Ulbrich, K.; Holá, K.; Šubr,

2683

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry V.; Bakandritsos, A.; Tuček, J.; Zbořil, R. Chem. Rev. 2016, 116, 5338− 5431. (3) (a) Takise, R.; Muto, K.; Yamaguchi, J. Chem. Soc. Rev. 2017, 46, 5864−5888. (b) Aghazadeh Tabrizi, M.; Baraldi, P. G.; Borea, P. A.; Varani, K. Chem. Rev. 2016, 116, 519−560. (c) Kays, D. L. Chem. Soc. Rev. 2016, 45, 1004−1018. (d) Wang, Q.; Su, Y.; Li, L.; Huang, H. Chem. Soc. Rev. 2016, 45, 1257−1272. (e) Meng, G.; Szostak, M. Org. Biomol. Chem. 2016, 14, 5690−5707. (4) (a) Dobereiner, G. E.; Crabtree, R. H. Chem. Rev. 2010, 110, 681−703. (b) Chen, C.; Hong, S. H. Org. Biomol. Chem. 2011, 9, 20− 26. (5) (a) Liang, Y.-F.; Jiao, N. Acc. Chem. Res. 2017, 50, 1640−1653. (b) Iqbal, N.; Cho, E. J. J. Org. Chem. 2016, 81, 1905−1911. (c) Papadopoulos, G. N.; Kokotos, C. G. J. Org. Chem. 2016, 81, 7023−7028. (d) Whittaker, A. M.; Dong, V. M. Angew. Chem., Int. Ed. 2015, 54, 1312−1315. (e) Vora, H. U.; Rovis, T. J. Am. Chem. Soc. 2007, 129, 13796−13797. (f) Bode, J. W.; Sohn, S. S. J. Am. Chem. Soc. 2007, 129, 13798−13799. (6) (a) Reddy, C. N.; Krishna, N. H.; Reddy, V. G.; Alarifi, A.; Kamal, A. Asian J. Org. Chem. 2017, 6, 1498−1504. (b) Zhou, W.; Fan, W.; Jiang, Q.; Liang, Y.-F.; Jiao, N. Org. Lett. 2015, 17, 2542−2545. (c) Zhu, C.; Wei, W.; Du, P.; Wan, X. Tetrahedron 2014, 70, 9615− 9620. (d) Tang, C.; Jiao, N. Angew. Chem., Int. Ed. 2014, 53, 6528− 6532. (7) (a) Noda, H.; Furutachi, M.; Asada, Y.; Shibasaki, M.; Kumagai, N. Nat. Chem. 2017, 9, 571−577. (b) Huang, B.; Zeng, L.; Shen, Y.; Cui, S. Angew. Chem., Int. Ed. 2017, 56, 4565−4568. (c) Wood, A. J. L.; Weise, N. J.; Frampton, J. D.; Dunstan, M. S.; Hollas, M. A.; Derrington, S. R.; Lloyd, R. C.; Quaglia, D.; Parmeggiani, F.; Leys, D.; Turner, N. J.; Flitsch, S. L. Angew. Chem., Int. Ed. 2017, 56, 14498− 14501. (d) Hu, L.; Xu, S.; Zhao, Z.; Yang, Y.; Peng, Z.; Yang, M.; Wang, C.; Zhao, J. J. Am. Chem. Soc. 2016, 138, 13135−13138. (8) (a) Willcox, D.; Chappell, B. G. N.; Hogg, K. F.; Calleja, J.; Smalley, A. P.; Gaunt, M. J. Science 2016, 354, 851−857. (b) Chow, S. Y.; Odell, L. R. J. Org. Chem. 2017, 82, 2515−2522. (c) Kannaboina, P.; Raina, G.; Anil Kumar, K.; Das, P. Chem. Commun. 2017, 53, 9446−9449. (d) Zhang, L.; Wang, C.; Han, J.; Huang, Z.-B.; Zhao, Y. J. Org. Chem. 2016, 81, 5256−5262. (9) Wu, Z.; Laffoon, S. D.; Nguyen, T. T.; McAlpin, J. D.; Hull, K. L. Angew. Chem., Int. Ed. 2017, 56, 1371−1375. (10) (a) Koniev, O.; Wagner, A. Chem. Soc. Rev. 2015, 44, 5495− 5551. (b) Tailhades, J.; Patil, N. A.; Hossain, M. A.; Wade, J. D. J. Pept. Sci. 2015, 21, 139−147. (c) Otaka, A.; Sato, K.; Shigenaga, A. Top. Curr. Chem. 2014, 363, 33−56. (d) Malins, L. R.; Payne, R. J. Top. Curr. Chem. 2014, 362, 27−87. (e) Panda, S. S.; Jones, R. A.; Hall, C. D.; Katritzky, A. R. Top. Curr. Chem. 2014, 362, 229−265. (f) Zheng, J.-S.; Tang, S.; Huang, Y.-C.; Liu, L. Acc. Chem. Res. 2013, 46, 2475− 2484. (11) (a) Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. Science 1994, 266, 776−779. (b) Warren, J. D.; Miller, J. S.; Keding, S. J.; Danishefsky, S. J. J. Am. Chem. Soc. 2004, 126, 6576−6578. (c) Blanco-Canosa, J. B.; Dawson, P. E. Angew. Chem., Int. Ed. 2008, 47, 6851−6855. (d) Fang, G.-M.; Li, Y.-M.; Shen, F.; Huang, Y.-C.; Li, J.-B.; Lin, Y.; Cui, H.-K.; Liu, L. Angew. Chem., Int. Ed. 2011, 50, 7645−7649. (e) Fang, G.-M.; Wang, J.-X.; Liu, L. Angew. Chem., Int. Ed. 2012, 51, 10347−10350. (f) Blanco-Canosa, J. B.; Nardone, B.; Albericio, F.; Dawson, P. E. J. Am. Chem. Soc. 2015, 137, 7197−7209. (g) Wang, J.-X.; Fang, G.-M.; He, Y.; Qu, D.-L.; Yu, M.; Hong, Z.-Y.; Liu, L. Angew. Chem., Int. Ed. 2015, 54, 2194−2198. (h) Pan, M.; Gao, S.; Zheng, Y.; Tan, X.; Lan, H.; Tan, X.; Sun, D.; Lu, L.; Wang, T.; Zheng, Q.; Huang, Y.; Wang, J.; Liu, L. J. Am. Chem. Soc. 2016, 138, 7429−7435. (i) Shelton, P. M. M.; Weller, C. E.; Chatterjee, C. J. Am. Chem. Soc. 2017, 139, 3946−3949. (12) Saxon, E.; Bertozzi, C. R. Science 2000, 287, 2007−2010. (13) Rohrbacher, F.; Wucherpfennig, T. G.; Bode, J. W. Top. Curr. Chem. 2014, 363, 1−31. (14) (a) Noda, H.; Erő s, G.; Bode, J. W. J. Am. Chem. Soc. 2014, 136, 5611−5614. (b) Dumas, A. M.; Molander, G. A.; Bode, J. W. Angew. Chem., Int. Ed. 2012, 51, 5683−5686.

(15) (a) Temperini, A.; Piazzolla, F.; Minuti, L.; Curini, M.; Siciliano, C. J. Org. Chem. 2017, 82, 4588−4603. (b) Mitchell, N. J.; Malins, L. R.; Liu, X.; Thompson, R. E.; Chan, B.; Radom, L.; Payne, R. J. J. Am. Chem. Soc. 2015, 137, 14011−14014. (c) Raj, M.; Wu, H.; Blosser, S. L.; Vittoria, M. A.; Arora, P. S. J. Am. Chem. Soc. 2015, 137, 6932− 6940. (d) Durek, T.; Alewood, P. F. Angew. Chem., Int. Ed. 2011, 50, 12042−12045. (16) (a) Mhidia, R.; Boll, E.; Fécourt, F.; Ermolenko, M.; Ollivier, N.; Sasaki, K.; Crich, D.; Delpech, B.; Melnyk, O. Bioorg. Med. Chem. 2013, 21, 3479−3485. (b) Beshore, D. C.; Dinsmore, C. J. Org. Lett. 2002, 4, 1201−1204. (17) (a) Schragl, K. M.; Forsdahl, G.; Gmeiner, G.; Enev, V. S.; Gaertner, P. Tetrahedron Lett. 2013, 54, 2239−2242. (b) Sasaki, K.; Crich, D. Org. Lett. 2011, 13, 2256−2259. (c) Venkataramanarao, R.; Sureshbabu, V. V. Tetrahedron Lett. 2006, 47, 9139−9141. (18) (a) Li, X.; Danishefsky, S. J. J. Am. Chem. Soc. 2008, 130, 5446− 5448. (b) Li, X.; Yuan, Y.; Berkowitz, W. F.; Todaro, L. J.; Danishefsky, S. J. J. Am. Chem. Soc. 2008, 130, 13222−13224. (19) Chen, W.; Shao, J.; Hu, M.; Yu, W.; Giulianotti, M. A.; Houghten, R. A.; Yu, Y. Chem. Sci. 2013, 4, 970−976. (20) (a) Suppo, J.-S.; Subra, G.; Bergès, M.; Marcia de Figueiredo, R.; Campagne, J.-M. Angew. Chem., Int. Ed. 2014, 53, 5389−5393. (b) de Figueiredo, R. M.; Suppo, J.-S.; Midrier, C.; Campagne, J.-M. Adv. Synth. Catal. 2017, 359, 1963−1968. (21) (a) Kolakowski, R. V.; Shangguan, N.; Sauers, R. R.; Williams, L. J. J. Am. Chem. Soc. 2006, 128, 5695−5702. (b) Wang, C.; Guo, Q.-X.; Fu, Y. Chem. - Asian J. 2011, 6, 1241−1251. (c) Sun, X.-H.; Yu, H.-Z.; Pei, S.-Q.; Dang, Z.-M. Chin. Chem. Lett. 2015, 26, 1259−1264. (d) Ali Shah, M. I.; Xu, Z.-Y.; Liu, L.; Jiang, Y.-Y.; Shi, J. RSC Adv. 2016, 6, 68312−68321. (e) Patil, M. Org. Biomol. Chem. 2017, 15, 416−425. (f) Jiang, Y.-Y.; Zhu, L.; Man, X.; Liang, Y.; Bi, S. Tetrahedron 2017, 73, 4380−4386. (g) Jiang, Y.-Y.; Wang, C.; Liang, Y.; Man, X.; Bi, S.; Fu, Y. J. Org. Chem. 2017, 82, 1064−1072. (22) For experimental mechanistic studies, see: (a) Hou, J.-L.; Ajami, D.; Rebek, J., Jr J. Am. Chem. Soc. 2008, 130, 7810−7811. (b) Restorp, P.; Rebek, J., Jr J. Am. Chem. Soc. 2008, 130, 11850−11851. (23) For computational mechanistic studies, see: (a) Jones, G. O.; Li, X.; Hayden, A. E.; Houk, K. N.; Danishefsky, S. Org. Lett. 2008, 10, 4093−4096. (b) Di Santo, E.; Alberto, M. E.; Russo, N.; Toscano, M. ChemCatChem 2015, 7, 2309−2312. (c) Jiang, Y.-Y.; Zhu, L.; Liang, Y.; Man, X.; Bi, S. J. Org. Chem. 2017, 82, 9087−9096. (24) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Keith, T.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, revision D.01; Gaussian, Inc.: Wallingford, CT, 2013. (25) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2009, 113, 6378−6396. (26) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215−241. (27) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297−3305. (28) For comments on utilizing the ultrafine grid in DFT calculations, see: Jiang, Y.-Y.; Man, X.; Bi, S. Sci. China: Chem. 2016, 59, 1448−1466 and references cited therein. (29) (a) Fukui, K. J. Phys. Chem. 1970, 74, 4161−4163. (b) Fukui, K. Acc. Chem. Res. 1981, 14, 363−368. (30) (a) Li, Z.; Zhang, S.-L.; Fu, Y.; Guo, Q.-X.; Liu, L. J. Am. Chem. Soc. 2009, 131, 8815−8823. (b) Li, H.; Hall, M. B. J. Am. Chem. Soc. 2684

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685

Article

The Journal of Organic Chemistry 2015, 137, 12330−12342. (c) Jiang, Y.-Y.; Yan, L.; Yu, H.-Z.; Zhang, Q.; Fu, Y. ACS Catal. 2016, 6, 4399−4410. (31) See: Wang, Z.; Zhou, Y.; Lam, W. H.; Lin, Z. Organometallics 2017, 36, 2354−2363 and references cited therein. (32) Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. NBO, version 3.1; University of Wisconsin: Madison, WI, 1996. (33) Legault, C. Y. CYLView, version 1.0b; Universitéde Sherbrooke: Montreal, 2009 (http://www.cylview.org/Home.html). (34) We also tried using the hydroxyl group of BtOH as a proton shuttle to realize the concerted proton transfer/nucleophilic substitution, but the related transition states could not be located. It should be noted that the amidation reaction also works in the absence of BtOH with a lower yield, which was further found to result from the slower HOAc-catalyzed 1,3-acyl transfer from the mixed anhydride Int1 (see section 3.4). Therefore, we believe that BtOH is not essential for the generation of Int1 from 1 and 2, though we cannot exclude the related BtOH-catalyzed pathway. (35) (a) Li, J.; Li, J.; Zhang, D.; Liu, C. ACS Catal. 2016, 6, 4746− 4754. (b) Jiang, Y.-Y.; Yan, L.; Yu, H.-Z.; Zhang, Q.; Fu, Y. ACS Catal. 2016, 6, 4399−4410. (c) Petrone, A.; Cimino, P.; Donati, G.; Hratchian, H. P.; Frisch, M. J.; Rega, N. J. Chem. Theory Comput. 2016, 12, 4925−4933. (d) Liu, J.; Zheng, Y.; Liu, Y.; Yuan, H.; Zhang, J. J. Comput. Chem. 2016, 37, 2386−2394. (e) Ma, J.; Chen, K.; Fu, H.; Zhang, L.; Wu, W.; Jiang, H.; Zhu, S. Org. Lett. 2016, 18, 1322−1325. (36) Reich, H. J. Bordwell pKa Table (Acidity in DMSO). https:// www.chem.wisc.edu/areas/reich/pkatable/index.htm (accessed November 10, 2017). (37) Zhang, X.; Chung, L. W.; Wu, Y.-D. Acc. Chem. Res. 2016, 49, 1302−1310. (38) Inconsistent with the relative stability of TS23 and TS24, the MeC(O)−N bond is shorter (1.39 Å) than the MeNHC(O)−N bond (1.43 Å) in Int5, suggesting that the former C−N bond is stronger than the latter. (39) See: Yu, J.-L.; Zhang, S.-Q.; Hong, X. J. Am. Chem. Soc. 2017, 139, 7224−7243 and references cited therein.

2685

DOI: 10.1021/acs.joc.7b03107 J. Org. Chem. 2018, 83, 2676−2685