Diastereodivergent asymmetric Michael-alkylation reactions using chiral N,N′-dioxide/metal complexes

Abstract

A diastereodivergent asymmetric Michael-alkylation reaction between 3-chloro-oxindoles and β,γ-unsaturated-α-ketoesters has been achieved using L-RaPr₂/Sc(OTf)₃ and L-PrPr₂/Mg(OTf)₂ metal complexes as catalysts. Both rel-(1R,2S,3R) and rel-(1S,2S,3R) chiral spiro cyclopropane oxindoles were constructed in good yields, diastereoselectivities and ee values. The diastereodivergent control may originate from different alkylation pathways after the Michael addition, with either intramolecular trapping of the aza-ortho-xylylene intermediate or direct S_N2 substitution.

---

Tuning diastereoselectivity in catalytic asymmetric synthesis is challenging due to the inherent preference for forming one type of diastereomer in most reactions.¹ However, relative configurations are as important as absolute configurations in pharmacology and drug discovery because both can influence the physiological activity of a molecule.² In general, diastereodivergence³ can be realized by changing the catalyst,⁴ additive and solvent,⁵ substrate,⁶ and other methods. Nevertheless, diastereodivergent synthesis is still in its infancy. It’s desirable to develop new strategies and discover more diastereodivergent reactions.

Oxindoles with a unique spirocyclopropane moiety exhibit diverse biological activities, such as non-nucleoside reverse transcriptase inhibitor and antitumor activity.⁷ Among the synthetic methodologies to prepare these molecules,⁸ the cascade Michael-alkylation reactions⁹ of 3-chlorooxindole with α,β-unsaturated olefins provide an operationally simple, stepwise pathway for diastereodivergence. Currently, only organic proline-based silyl ethers,¹⁰ cinchona alkaloid-derived thioureas¹¹ and squaramide¹² catalysts have been developed. These reactions proceed via the intramolecular trapping of the chiral aza-ortho-xylylene intermediate, A, after the Michael addition to afford the thermodynamically favored rel-(1S,2S,3R)¹³ products (Scheme 1, path a). On the other hand, if the Michael addition products follow the direct S_N2 substitution pathway (intermediate B), the rel-(1R,2S,3R) products will be formed (Scheme 1, path b). To the best of our knowledge, there is no precedent for synthesizing the rel-(1R,2S,3R) products as the major diastereomer, much less synthesizing both rel-(1S,2S,3R) and rel-(1R,2S,3R) products in high efficiency without drastically changing the reaction conditions. Herein, we reported a diastereodivergent asymmetric Michael-alkylation reaction between 3-chloro-oxindoles and β,γ-unsaturated-α-ketoesters using chiral N,N′-dioxide/metal complexes,¹⁴ synthesizing rel-(1S,2S,3R) and rel-(1R,2S,3R) spirocyclopropane oxindoles in high yields, diastereoselectivities and enantioselectivities.

Scheme 1 Proposed processes for the diastereodivergent Michael-alkylation.

Initially, the cascade reaction of 3-Cl oxindole, 1a, with the β,γ-unsaturated-α-ketoester 2a was chosen as the model reaction to optimize the reaction conditions. First, a series of metal salts were investigated by complexing with chiral N,N′-dioxide L-PrPr₂ in the presence of Na₂CO₃ as the base at 30 °C. It was found that both Sc(OTf)₃ and Mg(OTf)₂ complexes could catalyze the reaction with a preference for forming a different diastereomer. The enantioselectivity was moderate for each major diastereomer, which could be isolated by column chromatography. Next, other conditions were screened.¹⁵ The L-RaPr₂/Sc(OTf)₃ complex elevated the isolated yield of the product rel-(1R,2S,3R)-3aa to 92% with 78% ee (entry 3). After lowering the reaction temperature to 0 °C, increasing the stoichiometry of the base to 1.3 equivalents, and prolonging the reaction time to 72 h, the rel-(1R,2S,3R)-3aa could be obtained in 97% yield and 92% ee (entry 4, Table 1).

Table 1 Optimization of the reaction conditions^a

Entry	Metal salt	L*	Yield^b (3aa/3aa′) (%)	ee^c (3aa/3aa′)
a Unless otherwise noted, the reaction proceeded with 1a (0.1 mmol), 2a (0.1 mmol), metal salt/ligand (1 : 1.1, 10 mol%), and Na2CO3 (1.0 equiv.) in toluene (1.0 mL) at 30 °C for 24 h. b Isolated yield. c Determined by chiral HPLC on a chiral stationary phase (Chiralcel IA and IE). d The reaction proceeded at 0 °C for 72 h and with 1.3 eq. of Na2CO3. e At 40 °C for 72 h.
1	Sc(OTf)3	L-PrPr 2	92/7	76/0
2	Mg(OTf)2	L-PrPr 2	14/28	0/88
3	Sc(OTf)3	L-RaPr 2	92/—	78/—
4^d	Sc(OTf)3	L-RaPr 2	97/—	92/—
5^e	Mg(OTf)2	L-PrPr 2	—/77	—/95

---

On the other hand, using chiral L-PrPr₂/Mg(OTf)₂ as the catalyst and increasing the reaction temperature to 40 °C, the corresponding rel-(1S,2S,3R)-3aa′ could be afforded in 77% yield and 95% ee after 72 hours (entry 5). It is worth mentioning that the metal cations dictate the diastereoselectivity and variation of the chiral ligand structure and the reaction temperature did not change the major diastereoisomer once Sc(OTf)₃ or Mg(OTf)₂ were identified as suitable catalysts. The absolute configurations of the major enantiomers were determined separately by X-ray crystallographic analysis of the corresponding N-Boc protected derivatives 5 ¹⁶ (Fig. 1). Through this method, rel-(1S,2S,3R)-5aa′ obtained from the L-PrPr₂/Mg(OTf)₂ catalytic system was determined to be (1S, 2S, 3R), and rel-(1R,2S,3R)-5aa generated from L-RaPr₂/Sc(OTf)₃ was found to be (1R, 2S, 3R).

Fig. 1 A wire-stick representation of the N-Boc derivatives 5aa⁽′⁾ from the products 3aa⁽′⁾.

Next, the enantioselective and diastereodivergent synthesis of a series of spirocyclopropane oxindoles was carried out using these two chiral catalyst systems (Table 2). Under the optimized conditions, all of the substrates gave one major diastereomer (higher than 94:6 diastereoselectivity) with moderate to excellent yields and enantioselectivities. It is worth noting that most of the rel-(1R,2S,3R) products 3 were unstable under the HPLC analysis conditions. Fortunately, the enantiomeric excess could be determined after conversion of rel-(1R,2S,3R)-3 into the corresponding derivatives rel-(1R,2S,3R)-4 (Scheme 2). Generally, the rel-(1S,2S,3R)-diastereoisomers prepared from the L-PrPr₂/Mg(OTf)₂ catalyst were delivered in higher enantioselectivities than the rel-(1R,2S,3R) isomers from the L-RaPr₂/Sc(OTf)₃ catalyst. For the synthesis of rel-(1S,2S,3R) cyclopropanes, 3-chlorooxindoles 1 with halo-substituents at the C4 and C5-positions gave higher enantioselectivities than C6-substituted ones (entries 3–8). Electron-donating or -withdrawing substituents on the aromatic β,γ-unsaturated-α-ketoester 2 had a slight influence on the enantioselectivity (entries 10–22). 2-Naphthyl- and 2-thiophenyl substituted β,γ-unsaturated-α-ketoesters were also tolerated well (entries 21 and 22). Moreover, when the aliphatic substrates 2t and 2u were subjected to the reaction conditions, the corresponding rel-(1R,2S,3R)-products could be obtained in excellent yields with excellent enantioselectivities (up to 99% yield and 99% ee; entries 23 and 24).

Table 2 Substrate scope of the diastereodivergent Michael-alkylation reaction^a

Entry	3: R^1; R^2	rel-(1S,2S,3R)-3′ (Cat A)		rel-(1R,2S,3R)-3 (Cat B)
		Yield^b (%)	ee^c (%)	Yield^b (%)	ee^d (%)
a Unless otherwise noted, reactions were performed with 1 (0.1 mmol), 2 (0.1 mmol), chiral catalyst (10 mol%) and Na2CO3 in toluene (1.0 mL) for 72 h. For rel-(1S,2S,3R)-3′, L-PrPr2/Mg(OTf)2 (1.1/1) and 1.0 eq. of Na2CO3 were used at 40 °C. For rel-(1R,2S,3R)-3, L-RaPr2/Sc(OTf)3 (1.1/1, 10 mol%) and 1.3 eq. of Na2CO3 were used at 0 °C. b Isolated yield. c Determined by HPLC on a chiral stationary phase. d Determined by HPLC on a chiral stationary phase after transformation into 4. e The diastereoselectivity of 3ah was determined to be 94:6 from HPLC.
1	3aa^(′^): H; C6H5	77	95	97	92
2	3ba^(′^): 4-Me; C6H5	—	—	67	86
3	3ca^(′^): 4-F; C6H5	84	96	76	86
4	3da^(′^): 5-Me; C6H5	72	96	96	89
5	3ea^(′^): 5-F; C6H5	67	92	96	81
6	3fa^(′^): 5-Cl; C6H5	61	88	—	—
7	3ga^(′^): 6-F; C6H5	70	89	91	90
8	3ha^(′^): 6-Cl; C6H5	63	84	98	88
9	3ia^(′^): 6-Br; C6H5	—	—	83	84
10	3ah^(′^): H; 2-MeOC6H4	69^e	95^e	71	91
11	3ai^(′^): H; 3-MeOC6H4	70	96	98	95
12	3aj^(′^): H; 4-MeOC6H4	59	94	96	72
13	3ak^(′^): H; 2-MeC6H4	50	91	76	96
14	3al^(′^): H; 4-MeC6H4	77	93	93	87
15	3av^(′^): H; 4-FC6H4	—	—	96	90
16	3am^(′^): H; 3-ClC6H4	77	92	76	87
17	3an^(′^): H; 4-ClC6H4	63	91	79	93
18	3ao^(′^): H; 2-BrC6H4	62	94	—	—
19	3ap^(′^): H; 4-BrC6H4	81	94	83	94
20	3aq^(′^): H; 4-PhC6H4	74	98	96	92
21		71	93	91	93
22		52	95	—	—
23	3at^(′^): H; ^nBu	71	93	95	99
24	3au^(′^): H; cyclohexyl	—	—	99	99

---

Scheme 2 The stabilization of spiro-cyclopropane oxindole by amination.

Preliminary mechanistic studies were conducted to confirm our proposed diastereodivergent control mode. The relationship between the ee values of the ligand and product showed a linear correlation.¹⁵ Additionally, X-ray crystallographic analysis of the catalysts also showed a 1 : 1 ratio of ligand to metal.¹⁷ Both of these findings imply that the monomeric catalysts might be the main catalytically active species. What’s more, crystal structures of both catalysts displayed similar geometries and showed no significant differences in the accessibility of the substrate’s coordination site. To check whether L-PrPr₂/Mg(OTf)₂ and L-RaPr₂/Sc(OTf)₃ coordinated with different substrates, in situ HRMS analysis was performed.¹⁵ Based on the HRMS spectra, both of them coordinate with the same substrate, 3-Cl oxindole, to initiate the reaction. All of these experiments excluded the possibility that the observed diastereodivergence resulted from different coordinative styles of the two catalysts.

The intermediates of the Michael addition products were then synthesized and subjected to the optimized reaction conditions. The relative configuration of 6aa was also confirmed by X-ray crystallographic analysis.¹⁸ As summarized in Table 3, the two diastereomers of 6 only transformed to the same 3aa′ with 76% yield and 31% yield, respectively, under the L-PrPr₂/Mg(OTf)₂ system (Table 3, entries 2 and 4), which may be due to the chiral match or mismatch effect. On the contrary, 3aa or 3aa′ could be afforded from either 6aa or 6aa′ in the L-RaPr₂/Sc(OTf)₃ system with 18% yield and 74% yield, respectively (Table 3, entries 1 and 3). By comparing the different results between the L-RaPr₂/Sc(OTf)₃ and L-PrPr₂/Mg(OTf)₂ catalyzed reactions, L-PrPr₂/Mg(OTf)₂ should promote this reaction with aza-ortho-xylylene intermediates, as has been reported previously, which induced thermodynamically favored rel-(1S,2S,3R) 3aa′ from different diastereomers of 6. However, in the L-RaPr₂/Sc(OTf)₃ catalytic system, the chirality inversion at the quaternary carbon from 6aa to rel-(1R,2S,3R) 3aa and from 6aa′ to rel-(1S,2S,3R) 3aa′ illustrated that alkylation proceeded through a direct S_N2 substitution pathway. Moreover, chiral 6aa (−73% ee) could produce another enantiomer of 3aa′ in 75% yield and 76% ee under the L-PrPr₂/Mg(OTf)₂ conditions (Table 3, entry 5). Undoubtedly, the same enantioselectivity revealed that the first Michael addition step should be the chirality-determining step. For the result of entry 1, the product 3aa was formed with 85% ee and the starting material was recovered in 80% yield with 84:16 diastereoselectivity (6aa:6aa′) from racemic 6aa under the L-RaPr₂/Sc(OTf)₃ catalytic system (Table 3, entry 1). We propose that this result came from the chemical equilibrium between the retro-Michael reaction and Michael addition, which could also account for the lack of formation of rel-(1S,2S,3R) 3aa′. The low yield also implied that the rate of S_N2 substitution from the enolate ion is high and Na₂CO₃ is not a strong enough base to deprotonate the Michael intermediate. To evaluate the role of the N–H group, N-methylated 3-Cl oxindole was also subjected into the reaction, which showed a poor result.¹⁵

Table 3 The comparative experiments for diastereodivergent control

Entry	6	Condition^a	3aa [yield, ee]	3aa′ [yield, ee]
a Unless otherwise noted, reactions were performed with corresponding 6 and catalyst (M/L = 1/1.1 10 mol%) and Na2CO3 (1.0 eq.) in toluene (1.0 mL) at 0 °C for 72 h. b Isolated yield. ee was determined by HPLC on a chiral stationary phase. c The recovered 6aa had a yield of 80%, 84:16 dr, and −17% ee/19% ee, and the diastereoselectivity of corresponding 3aa was 84:16. d The ee of 6aa was −73% ee.
1^c	(±) 6aa	L-RaPr2/Sc(OTf)3	18%, 85%	—
2	(±) 6aa	L-PrPr2/Mg(OTf)2	—	76%, −7%
3	(±) 6aa′	L-RaPr2/Sc(OTf)3	—	74%, 3%
4	(±) 6aa′	L-PrPr2/Mg(OTf)2	—	31%, race
5^d	(+) 6aa	L-PrPr2/Mg(OTf)2	—	75%, −76%

---

Conclusions

A diastereodivergent asymmetric Michael-alkylation reaction between 3-Cl oxindoles and β,γ-unsaturated-α-ketoesters was accomplished by tuning metal catalysts and adjusting the ligands and temperature. Under the optimized conditions, both rel-(1R,2S,3R) and rel-(1S,2S,3R) spiro cyclopropane oxindoles were synthesized with high yields, diastereoselectivities and enantioselectivities. Mechanistic studies also revealed that the diastereodivergent control should come from either trapping the aza-ortho-xylylene intermediates or direct S_N2 substitution in the alkylation step, which may be caused by the different characteristics of the metal catalysts. Developing other diastereodivergent asymmetric methodologies with this strategy is ongoing.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We thank the National Natural Science Foundation of China (21432006, 21290182 and 21772127) for financial support.

Notes and references

1. E. M. Carreira and L. Kvaerno, Classics in Stereoselective Synthesis, Wiley, Weinheim, 2009.
2. (a) C. Wolf, Dynamic stereochemistry of chiral compounds: principles and applications, Royal Society of Chemistry, 2008; (b) C. G. Wermuth, The Practice of Medicinal Chemistry, Elsevier, Amsterdam, 3rd edn, 2008; (c) W. O. Foye, Foye’s Principles of Medicinal Chemistry, Lippincott Williams & Wilkins, 7th edn, 2013.
3. Reviews in asymmetric diastereodivergent synthesis: (a) L. Lin and X. M. Feng, Chem.–Eur. J., 2017, 23, 6464; (b) M. Bihani and J. C. G. Zhao, Adv. Synth. Catal., 2017, 359, 534; (c) G. Zhan, W. Du and Y. C. Chen, Chem. Soc. Rev., 2017, 46, 1675.
4. For selected examples of diastereodivergent asymmetric synthesis by changing the catalysis and dual catalyst strategies, see: (a) E. E. Maroto, S. Filippone, M. Suárez, R. Martínez-Álvarez, A. de Cozar, F. P. Cossío and N. Martín, J. Am. Chem. Soc., 2014, 136, 705; (b) S. Krautwald, D. Sarlah, M. A. Schafroth and E. M. Carreira, Science, 2013, 340, 1065; (c) M. T. Oliveira, M. Luparia, D. Audisio and N. Maulide, Angew. Chem., Int. Ed., 2013, 52, 13149; (d) E. Conde, D. Bello, A. de Cózar, M. Sánchez, M. A. Vázquez and F. P. Cossío, Chem. Sci., 2012, 3, 1486; (e) B. Simmons, A. M. Walji and D. W. C. MacMillan, Angew. Chem., Int. Ed., 2009, 48, 4349; (f) S. Filippone, E. E. Maroto, A. Martín-Domenech, M. Suarez and N. Martín, Nat. Chem., 2009, 1, 578; (g) X. X. Yan, Q. Peng, Y. Zhang, K. Zhang, W. Hong, X. L. Hou and Y. D. Wu, Angew. Chem., Int. Ed., 2006, 45, 1979; (h) Y. Huang, A. M. Walji, C. H. Larsen and D. W. C. MacMillan, J. Am. Chem. Soc., 2005, 127, 15051.
5. Selected examples: (a) X. Wu, Z. Chen, Y.-B. Bai and V. M. Dong, J. Am. Chem. Soc., 2016, 138, 12103; (b) X. Li, M. Lu, Y. Dong, W. Wu, Q. Qian, J. Ye and D. J. Dixon, Nat. Commun., 2014, 5, 4479; (c) X. Tian, C. Cassani, Y. Liu, A. Moran, A. Urakawa, P. Galzerano, E. Arceo and P. Melchiorre, J. Am. Chem. Soc., 2011, 133, 17934.
6. J. Zhu, Y. Liang, L. Wang, Z. B. Zheng, K. N. Houk and Y. Tang, J. Am. Chem. Soc., 2014, 136, 6900.
7. (a) Y. He, T. Jiang, K. L. Kuhen, Y. H. Ellis, B. Wu, T. Y. H. Wu and B. Bursulaya, Oxindoles with Anti-HIV Activity, U.S. Pat., WO 2004/037247A1, 2004; (b) T. Jiang, K. L. Kuhen, K. Wolff, H. Yin, K. Bieza, J. Caldwell, B. Bursulaya, T. Y. H. Wu and Y. He, Bioorg. Med. Chem. Lett., 2006, 16, 2105; (c) P. B. Sampson, Y. Liu, S. W. Li, B. T. Forrest, H. W. Pauls, L. G. Edwards, M. Feher, N. K. B. Patel, R. Laufer and G. Pan, Kinase Inhibitors and Method of Treating Cancer with Same, U.S. Pat., WO 2010/115279A1, 2010; (d) L. Chen, L. Feng, Y. He, M. Huang and H. Yun, Spiro Indole Cyclopropane Indolinones Useful as Ampk Modulators, U.S. Pat., WO2011/70039A1, 2011; (e) H. W. Pauls, S. W. Li, P. B. Sampson and B. T. Forrest, Plk-4 Inhibitors and Methods of Treating Cancer with Same, U.S. Pat., WO 2012/048411A1, 2012; (f) P. B. Sampson, Y. Liu and N. K. Patel, et al. , J. Med. Chem., 2015, 58, 130; (g) P. B. Sampson, et al. , J. Med. Chem., 2015, 58, 147.
8. Selected examples: (a) Z. Y. Cao, F. Zhou, Y. H. Yu and J. Zhou, Org. Lett., 2013, 15, 42; (b) Z. Y. Cao, X. M. Wang, C. Tan, X. L. Zhao, J. Zhou and K. Ding, J. Am. Chem. Soc., 2013, 135, 8197; (c) A. Noole, A. Malkov and T. Kanger, Synthesis, 2013, 45, 2520; (d) A. Awata and T. Arai, Synlett, 2013, 24, 29; (e) J. Guo, Y. B. Liu, X. Q. Li, X. H. Liu, L. L. Lin and X. M. Feng, Chem. Sci., 2016, 7, 2717.
9. Examples of an asymmetric Michael-alkylation reaction in three membered ring formation: (a) G. J. Sun, J. Gong and Q. Kang, J. Org. Chem., 2017, 82, 796; (b) B. L. Zhao and D. M. Du, Eur. J. Org. Chem., 2015, 5350; (c) C. Kim and S. G. Kim, Tetrahedron: Asymmetry, 2014, 25, 1376; (d) J. I. Martinez, E. Reyes, U. Uria, L. Carrillo and J. L. Vicario, ChemCatChem, 2013, 5, 2240; (e) L. Q. Lu, J. R. Chen and W. J. Xiao, Acc. Chem. Res., 2012, 45, 1278; (f) X. W. Dou and Y. X. Lu, Chem.–Eur. J., 2012, 18, 8315; (g) F. Pesciaioli, P. Righi, A. Mazzanti, G. Bartoli and G. Bencivenni, Chem.–Eur. J., 2011, 17, 2842; (h) H. Xie, L. Zu, H. Li, J. Wang and W. Wang, J. Am. Chem. Soc., 2007, 129, 10886.
10. (a) A. Noole, M. Ošeka, T. Pehk, M. Öeren, I. Järving, M. R. J. Elsegood, A. V. Malkov, M. Lopp and T. Kanger, Adv. Synth. Catal., 2013, 355, 829; (b) R. C. da Silva, I. Chatterjee, E. Escudero-Adán, M. W. Paixăo and P. Melchiorre, Asian J. Org. Chem., 2014, 3, 466.
11. (a) M. Ošeka, A. Noole, S. Žari, M. Öeren, I. Järving, M. Lopp and T. Kanger, Eur. J. Org. Chem., 2014, 3599; (b) X. W. Dou, W. J. Yao, B. Zhou and Y. X. Lu, Chem. Commun., 2013, 49, 9224.
12. (a) J. H. Li, T. F. Feng and D. M. Du, J. Org. Chem., 2015, 80, 11369; (b) A. Noole, I. Järving, F. Werner, M. Lopp, A. Malkov and T. Kanger, Org. Lett., 2012, 14, 4922.
13. J. H. Fletcher, Nomenclature of Organic Compounds, Advances in Chemistry, American Chemical Society, Washington, DC, 1974.
14. Selected examples of N,N′-dioxide in a catalytic asymmetric reaction, see: (a) X. H. Liu, L. L. Lin and X. M. Feng, Acc. Chem. Res., 2011, 44, 574; (b) X. H. Liu, L. L. Lin and X. M. Feng, Org. Chem. Front., 2014, 1, 298; (c) X. H. Liu, H. F. Zheng, Y. Xia, L. L. Lin and X. M. Feng, Acc. Chem. Res., 2017, 50, 2621; (d) K. Zheng, L. L. Lin and X. M. Feng, Acta Chim. Sin., 2012, 70, 1785; (e) X. H. Zhao, X. H. Liu, H. J. Mei, J. Guo, L. L. Lin and X. M. Feng, Angew. Chem., Int. Ed., 2015, 54, 4032; (f) X. Hao, L. Lin, F. Tan, C. Yin, X. Liu and X. Feng, ACS Catal., 2015, 5, 6052; (g) J. H. Feng, L. L. Lin, K. R. Yu, X. H. Liu and X. M. Feng, Adv. Synth. Catal., 2015, 357, 1305; (h) J. H. Feng, X. Yuan, W. W. Luo, L. L. Lin, X. H. Liu and X. M. Feng, Chem.–Eur. J., 2016, 22, 15650; (i) J. F. Zheng, L. L. Lin, L. Dai, Q. Tang, X. H. Liu and X. M. Feng, Angew. Chem., Int. Ed., 2017, 56, 13107.
15. For more details, see the ESI.†.
16. CCDC 1411858 (5aa) and CCDC 1412691 (5aa′) contain the supplementary crystallographic data for this paper.†.
17. The ORTEP diagram of the catalyst had been reported in our former research, CCDC 804337 [Mg(OTf)₂/L-RaPr₂], CCDC 704000 [Sc(OTf)₃/L-PrPr₂] and CCDC 882608 [Sc(OTf)₃/L-RaPr₂].
18. CCDC 1545249 (6aa) contains the supplementary crystallographic data for this paper.†.

---

Footnotes

† Electronic supplementary information (ESI) available. CCDC 1411858, 1412691 and 1545249. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7sc02757e

‡ These authors contributed equally to this work.

---