A K+-promoted Diels–Alder reaction by using a self-assembled macrocyclic boronic ester containing two crown ether moieties

A K+-promoted Diels–Alder reaction of 1,4,9,10-anthradiquinone with various dienes is achieved in the presence of a self-assembled macrocyclic boronic ester.


Introduction
One of the major goals for utilization of synthetic host molecules is their application as catalysts. 1 Among various organic reactions, the Diels-Alder reaction, which is one of the most useful reactions for the construction of 6-membered carbocycles, has been studied using synthetic host molecules, such as hydrogen-bonded capsules, 2 cyclic metalloporphyrin trimers, 3 coordination cages, 4 and so on. 5 Through these studies, reaction characteristics such as acceleration of the reaction rate, 1-5 formation of products with unique regio-or stereo-selectivity, 4b-d catalytic turnover, 2c,d,4b,h or asymmetric induction 4f were demonstrated depending on the host molecules. These reaction characteristics were mostly achieved by a strategy in which both a diene and a dienophile were pre-organized inside the host molecules. However, this strategy oen led to a problem known as product inhibition, because the host usually binds the product more strongly than the substrates. [1][2][3][4][5] To solve this problem, we devised a new type of supramolecular catalyst using a macrocyclic boronic ester containing two crown ether moieties. The macrocyclic host binds and activates the dienophile through M + -crown ether moieties and promotes the Diels-Alder reaction with various dienes. Aer the reaction, the bent-shaped product that showed weaker binding affinity could be replaced by the dienophile easily through the open framework. In this strategy, the entropic disadvantage arising from the need to bind both substrates is also reduced. Thus, the exchange process of the product with the dienophile is thought to be energetically neutral. Recently, the group of Lusby introduced this simple dienophile-binding strategy and realized the efficient catalytic turnover by using the coordination cage that activated the dienophile by hydrogen bonding. 4h, 6 Research is being carried out on dynamic self-assembly utilizing boronic ester formation of diboronic acids with an indacene-type tetrol. 7 During the examination of the possibility of utilizing our hosts as catalysts, we observed a concise selfassembly of the macrocyclic boronic ester [2+2] crown , containing two dibenzo-18-crown-6 moieties, and a hitherto unknown K + -accelerated Diels-Alder reaction, which showed not only acceleration of the reaction rate but also enhancement of the internal regioselectivity in the reaction of 1,4,9,10-anthradiquinone and various dienes (Fig. 1).

Results and discussion
The macrocyclic boronic ester [2+2] crown was quite easily prepared in high yield by simply mixing a diboronic acid of dibenzo-18-crown-6 2 and optically pure tetrol (+)-3 (ref. 8) in MeOH-CH 2 Cl 2 (1 : 1), followed by GPC purication (Fig. 1b). 9,10 Formation of the desired [2+2] crown was fully conrmed by 1 H NMR and FAB-MS, and the structure was conrmed by X-ray crystallographic analysis (Fig. 1c), which suggested that its cavity size (16.9 Â 12.7Å) was suitable for inclusion of quinone type compounds (Fig. 1c). 11 With the desired macrocyclic boronic ester [2+2] crown in hand, the binding behavior of [2+2] crown containing alkaline metal salts with several quinonoid compounds was examined. From 1 H NMR and isothermal titration calorimetry (ITC) studies, 1 : 1 complexation behavior was revealed for 1,4,9,10anthradiquinone 1 and [2+2] crown $2K + with a high association constant (K a ¼ 3.54 Â 10 3 M À1 ; CHCl 3 /CH 3 CN (1 : 1)) 12 ( Fig. 2) 9,13 (see the ESI †). It should be noted that the use of other metal cations (Li + or Na + ) instead of K + or a mixture of monomeric pinacol ester of dibenzo-18-crown-6 diboronic acid 2 pin , 14 KOTf, and 1 did not show an obvious shi of the signals of 1 by 1 H NMR. These results suggested that the macrocyclic structure of [2+2] crown with K + ions was essential for the efficient binding of 1. 9,13 As it is known that the Diels-Alder reaction of 1 with dienes occurs at both the C-2 and C-4a double bonds giving an internal and a terminal regioisomeric adduct and sometimes a diadduct, 15 the reaction of 1 with anthracene 4 in the presence or absence of [2+2] crown and K + was examined with the expectation that high regioselectivity owing to the restriction of the host framework would be achieved. 16 When a mixture of 1 and 4 was kept at room temperature in a mixed solution of CHCl 3 and CH 3 CN (1 : 1) for 1.5 hours, the Diels-Alder reaction proceeded very slowly (2% conversion) and a mixture of the internal adduct and terminal adduct was obtained in about a 4 : 1 ratio (Table 1, entry 1). Addition of [2+2] crown without K + did not affect the result of the reaction (Table 1, entry 2). On the other hand, when the same reaction was carried out in the presence of a stoichiometric amount of [2+2] crown and 2 equivalents of KOTf, the reaction was dramatically accelerated to give the product in 95% yield under the same reaction conditions, and the rate constant k was 206 times larger than that of the reaction without [2+2] crown and KOTf. Importantly, only the adduct with the internal alkene was obtained selectively (entry 3). Control experiments were carried out using monomeric 2 pin with KOTf, and almost no accelerating effect was observed (entry 4). Thus, the macrocyclic host structure of [2+2] crown was essential for the acceleration. Potassium cations were much more effective than sodium cations, which only had a small accelerating effect (entry 3 vs. 5).
It is well known that Lewis acids accelerate the Diels-Alder reaction by coordinating with the lone pair of the carbonyl group resulting in lowering of the LUMO level of the dienophile. 17 TiCl 4 , SnCl 4 , BF 3 $OEt 2 , etc. are employed as typical Lewis acids, and the use of alkaline metals for the acceleration of the Diels-Alder reaction has been mostly limited to the reaction using LiClO 4 in ether. 18 In fact, Na + or K + has not been employed for this purpose due to their very low ability to accept electron pairs in the vacant orbital of the metal to activate dienophiles. 19,20 To our knowledge, this is the rst example that the Diels-Adler reaction was effectively promoted by potassium ions 19 and this unique effect was specic to the combination of [2+2] crown and potassium ions.
The catalytic version of this reaction was also examined, and even 5 mol% of [2+2] crown $2K + was sufficient to promote the reaction (Scheme 1). Aer 3 hours, 45% of 1 was transformed into the Diels-Alder adduct 5 (5 int /5 ter ¼ 30 : 1), while only 5% conversion was observed without the catalyst. From the secondorder plot, aer 90 minutes, the reaction rate became slightly slower than that in the initial period, suggesting moderate product inhibition (Fig. 3). 21 However, reasonable catalytic activity was maintained under the conditions even where the amount of the product was considerably larger than that of the catalyst.
The acceleration of the reaction was observed with various 2-mono and 2,3-di-substituted 1,3-butadienes (Table 2). When dienes with small substituents 6-8 were used, the reactions were remarkably accelerated and the reaction rate constants k cat were approximately 20-50 times larger than those of the reaction without [2+2] crown $2K + (entry 1-6). On the other hand, the reactions with dienes with bulky substituents 9-12 were less accelerated (k cat /k no cat $ 10) as shown in entries 7-14. Steric repulsion between the [2+2] crown framework and the bulky substituent has made it difficult to promote the reaction smoothly. However, in all cases, enhancement of the regioselectivity of the product was observed in the presence of [2+2] crown $2K + . In particular, in the case of diene 12 (entry 13 and 14), the ratio of the internal adduct/terminal adduct was dramatically increased from 1 : 1 to 45 : 1 by the addition of [2+2] crown $2K + . Furthermore, chirality induction was observed when 2-mono-substituted 1,3-dienes were used (see the ESI, Table S1 †). This also suggested that the [2+2] crown framework recognized the substituent of dienes and the reaction proceeded inside the host, although the enantioselectivities were low (up to 19% ee). 1-  Mono-and 1,4-di-substituted 1,3-butadienes were not applicable to this reaction probably because these dienes could not approach the quinone inside [2+2] crown $2K + due to the steric hindrance of the [2+2] crown framework (see the ESI, Table S2 †).
The reason why anthracene 4 shows excellent activity compared to the other dienes 6-12 was also investigated. In the reaction of anthracene (Table 1, entry 3), the signal of the anthracene shied slightly up-eld compared to that of free anthracene in 1 H NMR spectra, suggesting the formation of a weak Michaelis complex 1$4@[2+2] crown $2K + (Fig. S10 †), while no obvious shi was observed without [2+2] crown $2K + (Fig. S11 †). 22 The formation of a substrate pair inside the host was thought to contribute to the acceleration of the reaction of anthracene as a result of the high local concentration of substrates. Subsequently, the possibility of whether the stabilization of the transition state (TS) was involved in the high activity of anthracene was also investigated. The adducts were used as TS analogues and their association constants with the host [2+2] crown $2K + were measured by ITC study. 4h Interestingly, the association constant of the anthracene adduct 5 int was 1.08 Â 10 3 M À1 in CHCl 3 /CH 3 CN (1 : 1) (Fig. S12 †), 21 while the association constants of other internal adducts derived from acyclic dienes such as 6, 9 and 11 that show lower activity were almost 0 M À1 (Fig. S25 †). These results suggested that the TS stabilization effect of [2+2] crown $2K + could also contribute to the high activity of anthracene.
Finally, we examined the possibility of the self-assembly protocol for this reaction. All the components, 1, 2, (+)-3, and 4, were mixed together and the process of the formation of the Diels-Alder adduct was monitored by 1 H NMR (Scheme 2). The internal adduct 5 int was obtained as the exclusive product in 83% yield aer 90 min. From the second-order plot shown in Fig. 4, the gradual increase of the reaction rate was observed and the reaction rate reached 5.95 M À1 min À1 from 60 min later. This value is comparable with that obtained by the reaction using preformed [2+2] crown $2K + (6.88 M À1 min À1 , Table 1 entry 2). Thus, the self-assembly of the active [2+2] crown $2K + occurred rapidly in the reaction mixture simply by mixing the component molecules to accelerate the reaction.  a Reaction conditions: 1 (6.5 mM) in the presence of the catalyst; 1 (15 mM) in the absence of the catalyst. b 3 equiv. of diene were used. c 1 (13 mM). d reaction rate k was estimated by using a second-order kinetic model with the assumption that 1 is completely complexed with [2+2] crown $2K + , although the estimated value of complexation based on the association constant is about 80% in the beginning. e The value of k cat or k no cat is taken from the reaction in the presence or absence of the catalyst. f Ratio of internal adduct/terminal adduct and conversion were determined by 1 H NMR.

Conclusions
The present study shows that the macrocyclic boronic ester [2+2] crown $2K + efficiently promotes the Diels-Alder reactions of 1,4,9,10-anthradiquinone and various dienes with high regioselectivity. It is noteworthy that four-point binding of the carbonyl groups with potassium cations in the [2+2] crown framework effectively accelerated the Diels-Alder reaction. Furthermore, the self-assembly protocol was successfully demonstrated by utilizing the dynamic nature of boronic ester linkages, offering the possibility of a novel catalytic system combined with the reversibility of boronic ester formation.

Conflicts of interest
There are no conicts to declare.

Acknowledgements
Thanks are due to Dr Hidehiro Uekusa and Dr Kohei Johmoto for performing X-ray analysis. We are grateful to a referee for useful comments about the high reactivity of anthracene. This work was supported by a Core Research for Evolutional Science and Technology (CREST) project from the Japan Science and Technology Agency (JST) and by JSPS KAKENHI Grant Number 15H05800.