Association of Halogen Bonding and Hydrogen Bonding in Metal Acetate-Catalyzed Asymmetric Halolactonization.

Cooperative activation using halogen bonding and hydrogen bonding works in metal-catalyzed asymmetric halolactonization. The Zn3(OAc)4-3,3'-bis(aminoimino)binaphthoxide (tri-Zn) complex catalyzes both asymmetric iodolactonization and bromolactonization. Carboxylic acid substrates are converted to zinc carboxylates on the tri-Zn complex, and the N-halosuccinimide (N-bromosuccinimide [NBS] or N-iodosuccinimide [NIS]) is activated by hydrogen bonding with the diamine unit of chiral ligand. Halolactonization is significantly enhanced by the addition of catalytic I2. Density functional theory calculations revealed that a catalytic amount of I2 mediates the alkene portion of the substrates and NIS to realize highly enantioselective iodolactonization. The tri-Zn catalyst activates both sides of the carboxylic acid and alkene moiety, so that asymmetric five-membered iodolactonization of prochiral diallyl acetic acids proceeded to afford the chiral γ-butyrolactones. In the total description of the catalytic cycle, iodolactonization using the NIS-I2 complex proceeds with the regeneration of I2, which enables the catalytic use of I2. The actual iodination reagent is I2 and not NIS.

Introduction

Electrostatic forces are fundamental forces that work in a wide range of molecular interactions. In catalysis, various metal cations act as the center of catalytic activity to enhance the reactivity of substrates (Figure 1A) (Yamamoto, 2001). Hydrogen bonding is the other representative electrostatic interaction observed between hydrogen (H) atoms with electronegative functionality (Figure 1B) (Pihko, 2009). Although these interactions have been widely utilized in the design and development of a wide range of catalysts, abundantly observed metal coordination and hydrogen bonding networks sometimes make the specific activation of substrates difficult. As a different type of noncovalent electrostatic interaction, halogen bonding has received much attention in organic chemistry. The origin of halogen bonding comes from the Lewis acidity of sigma holes that emerge on the opposite side of halogen-R sigma bonds, so that halogen bonding can effectively facilitate the direction of molecular recognition and can thus realize functional group selectivity (Figure 1C) (Cavallo et al., 2016).

Figure 1

Classification of Electrostatic Molecular Interactions

(A) Metal coordination.

(B) Hydrogen bonding.

(C) Halogen bonding.

Although the halogen bonding has been recently examined in several solution-phase catalyses, successful application in asymmetric reactions is limited (Beale et al., 2013, Bruckmann et al., 2008, Bulfield and Huber, 2016, Chan and Yeung, 2018, Coulembier et al., 2010, Dreger et al., 2018, Farina et al., 1999, Gliese et al., 2017, Haraguchi et al., 2018, He et al., 2014, Heiden et al., 2017, Heinen et al., 2018, Jungbauer et al., 2014, Jungbauer and Huber, 2015, Kaasik et al., 2017, Kazi et al., 2017, Kniep et al., 2013, Kuwano et al., 2018, Lim et al., 2016, Lindsay and Charette, 2012, Matsuzaki et al., 2018, Saito et al., 2017, Takeda et al., 2015, Zong et al., 2014). Here, halogen bonding and hydrogen bonding are merged with the metal-catalyzed asymmetric halolactonization (Castellanos and Fletcher, 2011, Chen and Ma, 2010, Cheng et al., 2014, Denmark et al., 2012, Hennecke, 2012, Tan et al., 2011a, Tan et al., 2011b, Tan et al., 2014).

Results and Discussion

Structure-Activity Relationship of Zn3(OAc)4-3,3′-bis(aminoimino)binaphthoxide-Catalyzed Asymmetric Iodolactonization

Among the wide range of halogen chemistry, halolactonization has been well utilized in the stereoselective synthesis of versatile natural products, biologically significant pharmaceutical compounds, and agricultural chemicals. For the catalytic asymmetric version of iodolactonization, Gao reported a unique system of a chiral salen-Co complex, in which I2 was used as an I+ source and the catalyst activity was enhanced by the addition of N-chlorosuccinimide (Ning et al., 2009). Jacobsen pioneered a tertiary aminourea-catalyzed asymmetric iodolactonization using N-iodo-4-fluorophthalimide as the I+ source and I2 as an additive (Veitch and Jacobsen, 2010). In the subsequently developed metal-catalyzed and organocatalyzed asymmetric iodolactonization reactions, the combined use of an I+ source (e.g., N-iodosuccinimide [NIS]) with I2 is often effective to improve the results (Arai et al., 2015a, Arai et al., 2015b, Brindle et al., 2013, Dobish and Johnston, 2012, Fang et al., 2012, Filippova et al., 2014, Kwon et al., 2008, Lu et al., 2018, Mizar et al., 2014, Murai et al., 2014a, Murai et al., 2014b, Nakatsuji et al., 2014, Toda et al., 2014, Tripathi and Mukherjee, 2013; Tungen et al., 2012, Sakakura et al., 2007, Suresh et al., 2018, Wang et al., 2012). In 2014, we also reported a highly efficient catalytic asymmetric iodolactonization using a 3,3′-bis(aminoimino)binaphthol ligand (L1) and Zn(OAc)2. Only 1 mol% of trinuclear Zn3(OAc)4-3,3′-bis(aminoimino)binaphthoxide (tri-Zn) was required to catalyze the asymmetric iodolactonization using NIS with the catalytic assistance of I2 to afford the products at up to over 99% enantiomeric excess (ee) (Figure 2) (Arai et al., 2014, Arai et al., 2015a, Arai et al., 2015b).

Figure 2

Chiral Zn3(OAc)4-3,3′-bis(aminoimino)binaphthoxide (tri-Zn) Catalyst for Asymmetric Iodolactonization

The structure-activity relationship of the aminoiminophenol ligands for Zn-catalyzed asymmetric iodolactonization is summarized in Table 1. The best ligand (L1) is prepared from (R, R)-diphenylethylenediamine and (R)-3,3′-diformylbinaphthol. The diastereomeric ligand (L2) prepared from (S, S)-diphenylethylenediamine and (R)-3,3′-diformylbinaphthol reduced asymmetric induction with 68% ee. Interestingly, a simple bis(aminoimino)binaphthol (L3) prepared using an achiral amine also provided an efficient chiral zinc catalyst to afford chiral 2a with 98% ee. Thus the use of expensive chiral diamine is not essential for asymmetric induction. The catalyst prepared using (R, R)-diphenylethylenediamine-derived bis(aminoimino)biphenol (L4) gave (R)-enriched 2a in 75% yield with 72% ee. From density functional theory (DFT) calculations of the L4-Zn(OAc)2 complex, the conformation with the (R)-axis L4-Zn3 complex is more stable by 6.0 kcal/mol than the (S)-axis configuration (Figure S1). The (R)-enriched formation of iodolactone 2a from entry 1 to 4 shows the importance of axial chirality for the construction of an efficient asymmetric reaction sphere. 3-Aminoiminobinaphthol (L6) for the dinuclear Zn complex and aminoiminophenol (L7) for the mononuclear Zn complex resulted in low catalyst activity. Overall, high catalytic activity for asymmetric iodolactonization is obtained when the trinuclear zinc acetates are precisely arranged on the axially chiral bis(aminoimino)biphenol ligands (Shibasaki and Yamamoto, 2004).

Table 1

Structure-Activity Relationship of the Aminoiminophenol Ligands for Zn-Catalyzed Asymmetric Iodolactonization

Entry	Ligand	X (mol %)	Time (h)	Yield of 2a (%)	ee of 2a (%)
1	L1	3	24	99	99.6 (R)
2	L2	3	24	99	68 (R)
3	L3	3	18	99	98 (R)
4	L4	3	24	75	72 (R)
5	L5	3	18	79	68 (R)
6	L6	2	24	34	89 (R)
7	L7	1	24	Trace	–

Questions on Catalytic Asymmetric Halolactonization

Despite the success with a series of efficient catalysts for asymmetric iodolactonization, there are still unsolved problems in the catalyst behavior, especially with respect to the activation mode of the alkene moiety. Conventional catalysts for asymmetric iodolactonization generally result in low asymmetric induction for other halolactonization (bromo-, chloro-, and fluorolactonization) reactions. The substrate scope is limited, and it is difficult to apply to rather complex molecules. For example, the catalyst developed for five-membered lactone formation tends to be unsuitable for the six-membered lactone system. The positive role of I2 in asymmetric catalysis is also unclear. Without the addition of catalytic I2, the tri-Zn catalyst gave 2a in only 7% yield with 97% ee. The answer to the actual role of I2 would provide realization of a true transition state, which would enable the appropriate selection of reaction substrates for the catalyst and the rational design of next-generation asymmetric catalysis.

Asymmetric Bromolactonization Using Tri-Zn Catalyst

The catalytic performance of the tri-Zn complex prepared using the best ligand (L1) was examined for asymmetric bromolactonization (Aursnes et al., 2016, Jiang et al., 2012, Jiang et al., 2018, Murai et al., 2010, Tan et al., 2011a, Tan et al., 2011b, Zhou et al., 2010). When the reaction was conducted using N-bromosuccinimide (NBS) at −78 ℃, 5 mol% tri-Zn complex catalyzed the bromolactonization to give the bromolactone 3a in 38% yield with 92% ee (entry1, Table 2). Similar to the previously studied iodolactonization, the addition of I2 or Br2 enhanced the catalytic activity to give the bromolactone in higher chemical yields while maintaining the ees. However, interestingly, when 0.2 equiv. of I2 was added to the bromolactonization of 1a, 36% iodolactone 2a (98% ee) was coproduced with 58% bromolactone 3a (93% ee) shown in entry 2.

Table 2

Catalytic Asymmetric Bromolactonization Using Zn3(OAc)4-3,3′-bis(aminoimino)binaphthoxide (tri-Zn) Catalyst

Entry	Additive	Additive	Temp (°C)	Yield (%)	ee (%)
1	–	–	−78	38	92
2	I2	I2	−78	58 (36)a	93 (98)a
3	Br2	Br2	−78	89	91
4	–	–	−40	99	94

Values in parentheses are yield or ee of iodolactone.

As a result of optimization for the bromolactonization, the reaction at −40℃ without additive was selected to afford 3a in quantitative yield with 94% ee. At this stage, it seems likely that the addition of I2 (or Br2) mainly contributes to accelerate the reaction and is not essential for asymmetric induction.

The scopes of both iodolactonization and bromolactonization were similar, and a comparison of bromolactonization and iodolactonization is summarized in Table S1.

DFT Calculation of the Transition State for Iodolactonization

DFT calculations were conducted to fully clarify the stereocontrol mechanism, the ligand structure-activity relationship, and the role of I2 for Zn-catalyzed asymmetric iodolactonization. The addition of I2 had a significant impact on the acceleration of the reaction, but no effect on stereochemical control. Therefore the transition state (TS)-A model for alkene activation by NIS was studied first (Figure 3A).

Figure 3

DFT Calculation of the Transition State for Iodolactonization

(A and B) Alkene activation by (A) NIS and (B) I2/NIS in the tri-Zn-catalyzed iodolactonization. Bond lengths are in angstroms

Negligible non-linear effect suggests the iodolactonization conducted on the unimolecular catalyst. In addition, a Job plot analysis suggests a 1: 1 interaction between tri-Zn complex and 1a (Figure S3). This is consistent with the rational TS-A model for the tri-Zn-catalyzed asymmetric iodolactonization supposed by the previous experimental data and preliminary computational studies (Figures S6 and S7) (Arai et al., 2014). The TS-A model for the tri-Zn-catalyzed asymmetric iodolactonization involved the zinc carboxylate of 1a replaced by the outer acetoxy anion on the tri-Zn complex. In TS-A, to produce the preferentially obtained (R)-2a, bidentate and strong electrostatic interaction between the Zn atom and acetoxy anion of 1a is formed to locate the bulky substituent (e.g., Ph group of 1a) in the empty space of tri-Zn. The anti-addition to the alkene portion of 1a locates NIS close to the aminoimino moiety of L1 and forms a hydrogen bond with the imino-proton. The cooperative electrostatic or hydrogen bonding interactions stabilize TS-A. In contrast, different structural features regarding these two interactions were found in TS-A to give the minor enantiomer (S)-2a. The bidentate Zn-carboxylate interaction is broken by the steric repulsion between the Ph group of 1a and the naphthyl moiety of L1. The axial chirality plays a key role in the asymmetric induction rather than the terminal (R,R)-diphenylenediamine part of L1. These computational results are consistent with the experimentally observed structure-activity relationship (entries 1 and 3 in Table 1, Figure S2). In addition, NIS stays away from tri-Zn to lose the hydrogen bond. The changes in the attractive non-covalent interactions between 1a, NIS, and tri-Zn are thus major factors in directing the stereochemical outcome. In addition, TS-A is 5.1 kcal/mol less stable than TS-A.

However, there still remains a question with respect to the tri-Zn-catalyzed iodolactonization regarding the role of I2. For the full description of the transition state, extensive studies with DFT calculations were conducted for the TS-B model involving the tri-Zn complex, 1a, NIS, and I2 in Figure 3B. The relative arrangement and electrostatic interaction between the tri-Zn complex and 1a are well stored in the TS-B model (TS-B in Figure 3B, TS-B in Figure S8). Importantly, an I2 molecule inserts between NIS and the alkene moiety of 1a by forming a bent configuration. The bent interaction of I2 with NIS is well explained by the halogen bonding observed in the I3+ species (Cavallo et al., 2016, Nakatsuji et al., 2014, Lu et al., 2018). The 1:1 interaction of the tri-Zn with the NIS-I2 species was suggested by ultraviolet-visible analysis (Figure S4). One obvious difference from the TS-A model is the lack of hydrogen bonding interaction between NIS and the imine-proton. Owing to the insertion of I2, NIS is moved to the right side and forms a new hydrogen bonding interaction with two methine protons of the diphenylethylenediamine framework and one proton on the benzene ring of diphenylethylenediamine. This plural hydrogen bonding scenario is well supported by the structure-activity relationship examined in Table 1 (entries 1, 3, and 5). The tri-Zn complex prepared from the simple ethanediamine-derived bis(aminoimino)binaphthol (L3)-tri-Zn catalyst gave 2a with 98% ee, although the catalyst prepared from tetramethyl-substituted ethanediamine-derived bis(aminoimino)binaphthol (L5), which lacked the hydrogen-bonding-donating methine protons, gave 2a with only 68% ee. In TS-B, the interaction energy analysis (Figure S9) clearly indicates that the electrostatic interaction (Zn carboxylate, 102 kcal/mol) is associated with hydrogen bonding (NIS-tri-Zn, 6 kcal/mol) and halogen bonding interactions (I2-NIS, 18 kcal/mol). Overall, in the tri-Zn-catalyzed asymmetric iodolactonization, three types of attractive non-covalent interactions cooperatively stabilize the transition state to achieve high catalytic activity and almost perfect enantioselectivity.

Design of Asymmetric Five-Membered Iodolactonization Based on DFT Simulation

The five-membered lactone system was also computationally addressed (Figure 4) based on the rational TS model for formation of the six-membered lactone (Figure 3B).

Figure 4

DFT Simulation of tri-Zn-Catalyzed Asymmetric Five-Membered Iodolactonization

There is no difference in the steric environment around the prochiral hydrogen atoms bonded to the α-carbon of the carboxyl group in the six-membered TS model (dotted circles around Ha and Hb in the left-hand side figure). On the five-membered TS model, the dotted circles around Ha and Hb in the right-hand side figure are differentiated in each space due to ring flipping. The introduction of a substituent at the prochiral hydrogen Ha would induce a repulsive interaction with the isoindoline moiety to increase the energy difference of the Ha- and Hb-substituted diastereomeric transition states and lead to high stereo-discrimination. Such transition state models predicted facilitation of the nucleophilic attack by the carboxylate anion to one side of the alkene moiety of the meso-substrate in the five-membered lactone system to yield the (3S, 5R)-product preferentially. These computational results prompted us to design the catalytic asymmetric iodolactonization of prochiral diallyl acetic acids as shown in Figure 5 (Ikeuchi et al., 2012, Jiang et al., 2018, Klosowski and Martin, 2018, Knowe et al., 2018, Murai et al., 2014a, Murai et al., 2014b, Wilking et al., 2013, Wilking et al., 2016).

Figure 5

Catalytic Asymmetric Five-Membered Iodolactonization Using Zn3(OAc)4-3,3′-bis(aminoimino)binaphthoxide (tri-Zn) Catalyst

To accomplish the asymmetric reaction shown in Figure 5, the catalyst must recognize each side of the alkene moiety in an enantioselective manner. At the same time, the catalyst must allow incorporation of the bulky branched substrate into the asymmetric reaction sphere. This requirement can be satisfied with tri-Zn catalysis because the bulky substituent of the substrate remains in the empty space, as presented in Figure 4. The tri-Zn catalyst controlled the asymmetric iodolactonization efficiently to give the five-membered iodolactone 5 in a highly diastereo- and enantioselective manner. In agreement with the computational prediction, the absolute structure of chiral γ-butyrolactone 5a was determined by X-ray crystallographic analysis to be (3S,5R)-5-(iodomethyl)-5-phenyl-3-(2-phenylallyl)dihydrofuran-2(3H)-one (see Supplemental Information, Data S1 for details.).

The synthetic utilities of the five-membered iodolactone 5 are demonstrated in Scheme 1. From 94% ee of 5a, epoxyester 6a and furan 7a were obtained, which maintain the exo-olefine functionality. A radical cyclization of 5a provided 7-oxabicyclo[4.2.1]nonan-8-one 8a.

Scheme 1

Chemical Transformation of Iodolactone 5

(A) K2CO3 (2 equiv.), MeOH, rt, 30 min; (B) LiBH4 (5 equiv.), THF/MeOH, rt, 5 h; (C) PPTS (0.1 equiv.), CH2Cl2, rt, 1 h; (D) AIBN (0.2 equiv.), Bu3SnH (1.5 equiv.), toluene, 80°C, 6 h.

Discussion of the Catalytic Cycle of Asymmetric Iodolactonization

Based on the realization of the transition state in Figure 3B, the catalytic cycle of tri-Zn-catalyzed asymmetric iodolactonization can be described as shown in Figure 6A. The tri-Zn (A) would react with the alkenoic acid substrate (1) to generate the Zn carboxylate (B). With the Zn carboxylate (B), the complex of NIS with I2 (C) would activate the alkene moiety to give the model transition state (e.g., TS-B model). When the origin of the I-source is carefully considered (as marked by purple and green), the alkene moiety is activated by I2, and not by NIS. However, by following the relay of arrows, the iodolactonization to yield product 2 would proceed with the co-production of N-succinimide and regeneration of I2. From this mechanistic hypothesis, the tri-Zn-catalyzed reaction pathways with and without I2 were compared (Figure 6B; black: with I2, gray: without I2). Ligand exchange process between tri-Zn and 1 readily proceeds without significant energy loss to generate the Zn carboxylate (B). Although additional I2 entropically destabilizes the complex of NIS with I2 (C) when compared with the complex of NIS without I2 (C′), the TS-B model is −12.6 kcal/mol more stable than the TS-A model. By receiving the insertion of I2, TS is significantly stabilized to accelerate the iodolactonization.

Figure 6

Reaction Mechanism of Asymmetric Iodolactonization

(A) Plausible catalytic cycle of tri-Zn-promoted asymmetric iodolactonization.

(B) Energy profile of plausible reaction pathways with I2 (black) and without I2 (gray).

The halogen-incorporated reaction mechanism described in Figure 6A is strongly supported by the experimental results of entry 2 in Table 2, in which 36% yield of iodolactone is produced using 0.2 equiv. of I2 for the NBS-promoted bromolactonization. This clearly indicates that both iodine atoms of I2 are rapidly introduced to the product. The phenomenon is also well explained by the DFT simulation for the second catalytic cycle (Figure 7). In the first catalytic cycle based on the halogen-incorporated reaction mechanism, I-Br should be generated. When the I-Br is incorporated into the TS, the reaction should proceed from TS-I described on the left-hand side of Figure 7 to give the iodolactone.

Figure 7

Plausible Transition States for Alkene Activation by IBr with NBS in the tri-Zn-Catalyzed Halolactonization

Bond lengths are in angstroms.

Limitations of Study

This study is limited to catalytic asymmetric iodolactonization and bromolactonization. The method based on the current approach would be potentially applicable to other chlorolactonization or fluorolactonization.

Conclusions

Excellent catalytic activity of the tri-Zn catalyst to achieve highly enantioselective asymmetric halolactonization was realized. Both activation of the metal-carboxylate and hydrogen bonding activation of NIS are harmonized by the halogen bonding with I2 to enable highly asymmetric iodolactonization. In the tri-Zn-catalyzed iodolactonization, the actual iodination reagent was identified for the first time; it is I2, and not NIS.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.