Rafał Kowalczyk*a,
Przemysław J. Boratyńskia,
Aleksandra J. Wierzbaa and
Julia Bąkowiczb
aDepartment of Organic Chemistry, Wrocław University of Technology, Poland. E-mail: rafal.kowalczyk@pwr.edu.pl
bAdvanced Materials Engineering and Modelling Group, Wrocław University of Technology, Poland
First published on 30th July 2015
The regiochemical course of sulfa-Michael addition of thiols to terminally divergently activated dienes partially incorporated in a cyclic system, as in substituted 2-(3-oxocyclohexenyl)-vinyls, can be to some extent altered by the choice of activation method. The ratio of products of competing 1,4- and 1,6-addition reactions is dependent on the potency of the electron withdrawing group. The iminium ion catalysis favors 1,6-addition to the cyclic ketone group, changing the product preference for dienes terminated with ketone and aryl ester groups. Application of 9-epi-aminoquinine analogues and an acid allowed for both regioselective and enantioselective (up to 90% ee) addition at the distant δ position.
In our recent work10 on the additions of thiols to acyclic (2E,4E)-dienoates 4, exclusive β-addition with respect to the ketone carbonyl group was observed. No products consistent with vinylogous δ-addition directed either by the ketone or the ester group were detected. Moreover, the observed regioselectivity was the same regardless of the method used to activate the conjugated system. No propagation of the electronic effect in the addition of thiols to linear dienoates was also reported by other researchers.3 In contrast, effective and stereoselective δ-addition of nucleophiles to dienes 2 and 3 with cyclic conjugated motifs were performed.2,3
The literature data11 as well as our experiments indicated that in the activated dienes (Fig. 1) β-position to the ketone carbonyl group is usually the most reactive, unless it is blocked by additional substitution. The similarity between dienes 3 and 5 implies efficient orbital overlapping and consequently assures propagation of the electronic effect from the carbonyl or the electron-withdrawing group at the opposite end. In a DFT study, the LUMO orbital in dienes 5 was spread along the entire π-system.12 Thus, we recognized dienes of type 5 as suitable model substrates, in which the competing activation by two flanking electron withdrawing groups would diverge the course of nucleophilic addition. We assumed that activation of dienes 5 by various means, including iminium ion strategy13 could allow for precise functionalization at the distant position, also in a stereoselective manner.
Entry | 5, R | Catalyst, 6![]() ![]() |
|
---|---|---|---|
A: DABCO | B: 10a/11a | ||
a Reactions were performed on a 0.3 mmol scale using either (A) 20–50 mol% of DABCO at 20 °C for 20 h or (B) 10 mol% of amine 10a and 20 mol% of acid 11a at 20 °C for 20 h in dichloromethane.b Determined by 1H NMR integration for crude reaction mixtures. | |||
1 | 5a, C(O)CH3 | 0![]() ![]() |
50![]() ![]() |
2 | 5b, C(O)Ph | 0![]() ![]() |
77![]() ![]() |
3 | 5c, CO2Ph | 58![]() ![]() |
100![]() ![]() |
4 | 5d, CO2CH3 | 100![]() ![]() |
100![]() ![]() |
5 | 5e, CN | 100![]() ![]() |
100![]() ![]() |
6 | 5f, C(O)-4-FC6H4 | 0![]() ![]() |
79![]() ![]() |
7 | 5g, C(O)-4-MeOC6H4 | 0![]() ![]() |
72![]() ![]() |
8 | 5h, C(O)-1-naphthyl | 0![]() ![]() |
56![]() ![]() |
9 | 5i, C(O)-2,6-Cl2C6H3 | 0![]() ![]() |
28![]() ![]() |
Also, activation of diketone 5b with other systems14 was examined: phosphine catalyst (diphenylmethylphosphine), which is a much weaker base compared with DABCO, but still a good nucleophile did not change the product preference. On the other hand, Brönsted- and Lewis-acids as well as basic catalysts provided only slight variations in the ratio of isomers, with few notable exceptions (e.g. for zinc triflate, see ESI†). A few different activation systems composed of a primary amine and a carboxylic acid designed for the transient formation of an iminium ion were tested. Two-component catalytic systems entailing 2-fluorobenzoic acid (11a) and amines, such as (R)-α-methylbenzylamine, amino- and (R,R)-diaminocyclohexane as well as aniline hydrochloride all delivered solely the γ-addition product 7b, without discrimination of enantiomers. In contrast, application of 9-epi-amino-quinine (10a) in combination with acid 11a, similar to the system for activation of non-modified dienones 3 introduced by Melchiorre,3 gave predominantly the δ-product in 77:
23 6b
:
7b ratio with rather good enantioselectivity (vide infra). Suprisingly, when isomeric 9R-aminoquinine was used the γ-addition again became predominant (2
:
98 6b
:
7b). The 10a/11a catalytic system delivered a similar qualitative change in the regioselectivity compared to DABCO for other alkyl and aryl ketone analogues of 5b (Table 1, entries 1, 6–9). Reactions of electron poor aryl ketones were slightly more diverted towards δ-addition by the 10a/11a catalytic system than that with electron rich aryls. However, notably poorer control of the regioselectivity was observed for ortho-substituted aryl ketones, which are less likely to assume coplanarity of the aryl ring with the rest of the π-system. These findings indicate that 10a/11a activates the conjugated diketone 5b through iminium ion formation, preferentially at the cyclic ketone. The alternative structure, in which the iminium ion is formed from phenone group is much higher in energy (>8 kcal mol−1), while steric interactions distort the coplanarity of the terminal phenyl group with the π-system. When such a distortion is already present in the diketone (e.g. in 2,6-dichlorophenyl derivative, Table 1, entry 9) the reaction no longer displays preference for the δ-addition (for preliminary DFT study, see ESI†).
Application of the 10a/11a system in the reaction involving phenyl ester 5c, provided complete change of regioselectivity toward the δ-product 6c. With this catalyst, also reactions of alkyl esters and nitrile analogues (5d, 5e) remained δ-selective (Table 1, entries 4–5). The same result was observed for addition of thiophenol, which occurred at the δ-position of esters 5c and 5d, indicating that regioselectivity of the catalyzed addition is not dependent on the thiol type. Also, no impact of solvents on the ratio of products was observed (for the details, see ESI†). For all the tested epi-aminoalkaloids and esters 5 the regioselectivity was complete, with the exception for the reaction of phenyl ester 5c catalyzed by epi-aminoquinidine (10d), which gave a small but detectable quantity of the γ-product 7c.
These findings outline that in a sequence of dienes 5a, 5b, 5c, 5d, and 5e preferential reactivity site is gradually shifted from the γ- to the δ-position as the potency of the terminal electron withdrawing group decreases (Table 1). Also, iminium catalysis with 9-epi-amimo-Cinchona alkaloids favors δ-addition for all the diene substrates 5.
Entry | Diene, R | Amine/acid | Product, yield, % | eeb, % |
---|---|---|---|---|
a Reactions were performed on a 0.3 mmol scale using 10 mol% of amine 10a-k, 20 mol% of acid 11a-b at 20 °C for 20 h.b Determined using HPLC on chiral stationary phases.c The yield given in parentheses was estimated by NMR.d Performed at 0 °C for 48 h.e Major enantiomer of opposite configuration (S).f After single recrystallization from DCM/cyclohexane. | ||||
1 | 5b, C(O)Ph | 10a/11a | 6b, (67)c | 53 |
2 | 5b, C(O)Ph | 10a/11b | 6b, 50 | 67 |
3 | 5b, C(O)Ph | 10f/11b | 6b, 56 | 50 |
4 | 5c, CO2Ph | 10f/11b | 6c, 74 | 80 |
5 | 5d, CO2Me | 10a/11a | 6d, 33 | 79 |
6 | 5d, CO2Me | 10a/11b | 6d, 55 | 83 (79)d |
7 | 5d, CO2Me | 10b/11a | 6d, 27 | 68 |
8 | 5d, CO2Me | 10c/11b | 6d, 47 | 65 |
8 | 5d, CO2Me | 10d/11a | ent-6d, 42 | 57e |
9 | 5d, CO2Me | 10e/11a | ent-6d, 29 | 40e |
10 | 5d, CO2Me | 10f/11b | 6d, 48 | 87 |
11 | 5d, CO2Me | 10g/11b | 6d, 58 | 69 |
12 | 5d, CO2Me | 10h/11b | 6d, 37 | 76 |
13 | 5d, CO2Me | 10i/11b | 6d, 67 | 78 |
14 | 5d, CO2Me | 10j/11b | 6d, 90 | 71 |
15 | 5d, CO2Me | 10k/11b | 6d, 74 | 76 |
16 | 5j, CO2c-C6H11 | 10f/11b | 6j, 61 | 82 (90)f |
17 | 5k, CO2Bn | 10f/11b | 6k, 81 | 87 |
18 | 5l, CO2tBu | 10f/11b | 6l, 68 | 87 |
19 | 5e, CN | 10f/11b | 6e, 70 | 60 |
The reaction of ketoesters 5 proceeded with improved enantioselectivity compared to diketones (Table 2) and provided more space for optimization. Various 9-epi-aminoalkaloid derivatives (Table 2 entries 5–15, Fig. 2) provided enantioselectivity for the transformation.16 Both antipodes of 6d were accessible through the choice of pseudoenantiomeric amine 10, although the differences in selectivity were noticeable (Δee 6–40%). Better results were obtained for the (8S,9S)-derivatives, i.e. of quinine and cinchonidine compared to (8R,9R) quinidine and cinchonine, respectively. Further fine tuning of Cinchona alkaloid framework was made by substitution at the 2′-position with various aryl and alkyl groups (Fig. 2).17 For the additions to esters 5 the highest enantiomeric excess reaching 87% was achieved using a combination of 2′-phenyl derivative of quinine 10f and 11b (for optimization, see ESI†). The reactions run at 0 and 22 °C gave almost identical ees (Table 2, entries 5 vs. 6). Similarly, increase of catalyst loading and the use of greater excess of thiol led to no noticeable changes in the reaction outcome. Different alkyl esters (5j–l) were tested, and the enantioselectivity exhibited only slight variations to the size and electronic nature of the ester group. However, alteration of the ring in the ketoester 5d, either by the introduction of additional steric hindrance or contraction to a cyclopentenone system led to a deterioration in both yield and enantioselectivity of the addition (8, 9, Fig. 3). Similar incompatibility of such ring systems was previously reported by Melchiorre for non-divergently activated dienones 3.3
The scope of the reaction in respect to the nucleophile and its impact on both the regio- and stereoselectivity was studied for different esters of type 5. The reaction performed using amine 10f and acid 11b delivered only one regioisomer of product for all the tested mercaptans (Fig. 3).18 Moreover, the structure of the mercaptan19 had only limited impact on the enantioselectivity of the addition (ee range 84–90%). Among the tested benzyl thiol congeners, neither the electronic nature of the benzene ring nor sterically demanding substituents had notable effect on the observed enantiomeric ratios.
The optimum catalyst structure was dependent upon the type of reactant 5. Out of the tested catalytic systems, 10f/11b was superior for ketoesters 5, while for diketone 5b the unmodified scaffold of 9-epi-aminoquinine (10a) in combination with (S)-O-acetylmandelic acid (11b) gave optimum ee.
The structures of the products were confirmed in NMR experiments (HMBC, allylic correlations in COSY, see ESI†). The (R)-configuration for the product 6j obtained using 10f/11b system was assigned unequivocally by X-ray crystallography (Fig. 4) with appropriate value of Flack parameter (−0.04(8)). Tentatively, the same configuration may be ascribed to all the adducts.
![]() | ||
Fig. 4 X-ray structure for 6j, for clarity disorder is not shown (for details, see ESI†). |
Few plausible stereochemical models of induction of chirality were proposed for similar 1,6-conjugate additions that are consistent with our observations.3,17 In each, formation of stable iminium salts is postulated. Narrow catalyst scope, in particular intolerance for a change in configuration at the C9 center of the alkaloid, as well as failure of multiple other amines to even provide regioselectivity can be explained by the stabilizing intramolecular hydrogen bond within the iminium ion (Fig. 5, for details, see ESI†) previously postulated by List.17
Internal hydrogen bond that involves the quinuclidine nitrogen atom is present in 9-epi-Cinchona alkaloids (8R,9R or 8S,9S configuration), but is unlikely for the alkaloids of native configuration at C9 (8R,9S or 8S,9R).
Footnote |
† Electronic supplementary information (ESI) available: Experimental, spectral and DFT theoretical data and tables listing peripheral experiments. CCDC 1401916. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5ra09631f |
This journal is © The Royal Society of Chemistry 2015 |