A novel oxaphenalenone, penicimutalidine: activated production of oxaphenalenones by the diethyl sulphate mutagenesis of marine-derived fungus Penicillium purpurogenum G59

Abstract

A novel oxaphenalenone, penicimutalidine (1), was isolated from a fungal mutant generated through the diethyl sulfate (DES) mutagenesis of marine-derived Penicillium purpurogenum G59 along with three known oxaphenalenones (2–4). The structure of 1, including the absolute configuration, was determined by spectroscopic methods, including TDDFT electronic CD calculations. HPLC-UV/HPLC-MS analyses verified that 1–4 were only produced in the mutant through the activation of silent biosynthetic pathways in the parent strain by DES mutagenesis.

---

Oxaphenalenones (OPs) are an important class of phenolic natural products, which demonstrate various bioactivities and are frequently isolated from fungi and plants.¹ Neonectrolide A,¹ SF226,² and corymbiferan lactones A–E³ are examples of OP monomers from fungi. Some of them showed cytotoxicity¹ and inhibitory effects on a series of DNA polymerases.² Bacillosporins A–D^4,5 and bacillisporins D–E⁶ are the OP dimers from fungi, which displayed antibacterial activity⁴ and a growth inhibitory effect on several human cancer cell lines.⁶ Though a series of simple OP monomers with a single benzo[de]isochromen-1(3H)-one ring system have been reported,^1–3 none of them had a C–C bond substituent at the sp³ carbon on their δ-lactone moiety.

The activation of silent biosynthetic pathways for secondary fungal metabolites has become more and more important for discovering novel compounds from fungi because major gene clusters for the biosynthesis of secondary metabolites in fungi are generally silenced under standard culture conditions.⁷ In contrast with the one-strain-many-compounds,⁸ chemical epigenetics,⁹ and cocultivation¹⁰ strategies that activate silent fungal pathways through variation of environmental factors for the growth of the producing strains, ribosome engineering¹¹ could activate silent bacterial pathways to obtain new compounds by selecting drug-resistant mutants with activated pathways that work under standard culture conditions.¹² The same ribosome engineering strategy has recently been extended to fungi by our group with the development of a new series of specific methods for fungi to introduce drug resistance.^13a–c This has resulted in the discovery of several new bioactive compounds derived from silent fungal pathways.^13c–e In the course of this particular study, we have also developed a simple and practical mutagenesis strategy using diethyl sulfate (DES) to activate silent fungal pathways.¹⁴

Previously we reported that the DES mutagenesis of marine-derived fungus Penicillium purpurogenum G59 resulted in the production of diverse secondary metabolites in mutants by activating pathways that were silent in the parent strain.^14a We also reported a series of new bioactive compounds produced by several mutants via the activation of pathways that were silent in the G59 strain.¹⁴

Continuing our work in this area, we report herein a new OP monomer penicimutalidine (1) (Fig. 1) produced by another mutant, AD-1-1, of the G59 strain, selected by treating G59 spores with 0.5% (v/v) DES in 50% (v/v) DMSO at 4 °C for 1 day.^14a Compound 1 is novel in carrying a C–C bond substituent at C-10, and was produced along with three known OPs, 2–4 (Fig. 1), by the mutant via activating silent pathways in the G59 strain by DES mutagenesis. The absolute configuration of 4 was also determined in the present study for the first time.

Fig. 1 Structures of 1–4.

In this study, the mutant AD-1-1 and parental G59 strains were concurrently fermented using rice as a solid-substrate fermentation medium at 28 °C for 45 days, prior to obtaining ethyl acetate (EtOAc) extracts from their cultures. The extract from the mutant weakly inhibited K562 cells with an inhibition rate (IR%) of 21.7% at 100 μg mL⁻¹, whereas the parent extract exhibited hardly any inhibition with an IR% of 1.7% at 100 μg mL⁻¹. Chromatographic separation of the mutant extract, tracking new metabolites in the mutant by comparison with the parent extract using both bioassays and HPLC analysis of the typical UV absorption peaks, resulted in the isolation of compounds 1–4 (Fig. 1). Among them, 2 and 3 were identified as the known OPs, SF226 (2)² and corymbiferan lactone A (3),³ respectively, according to their spectroscopic data.

Penicimutalidine (1) was obtained as pale yellow needles (from MeOH), mp 104–105 °C, [α]²⁵_D +184.5 (c 0.21, MeOH), and the molecular formula, C₁₅H₁₂O₇ (Ω = 10), was assigned by HRESI-MS (m/z 305.0668 [M + H]⁺, calcd. for C₁₅H₁₃O₇ 305.0661) combined with the ¹H and ¹³C NMR data (Table 1). The typical UV absorption peaks of 1 in MeOH at 219 (log ε 3.91), 234.5 (4.00), 277 (3.71), 344 (3.25) and 364 nm (3.23) indicated the presence of the OP skeleton in the molecule.³ The IR absorptions indicated hydroxyl (3474 and 3278 cm⁻¹), ester (1720 cm⁻¹), and conjugated carbonyl (1657 cm⁻¹) groups in 1.

Table 1 400 MHz ¹H/100 MHz ¹³C NMR data for 1 and 2^a

No.	1		2
	δH (J in Hz)	δC	δH (J in Hz)	δC
a The δH/δC values were recorded in acetone-d6 using the acetone-d6 signals (δC 29.84/δH 2.05) as references.
1	—	98.0 s	—	97.7 s
2	—	163.9 s	—	163.9 s
3	6.68 q (0.8)	117.6 d	6.65 q (0.8)	117.2 d
4	—	148.7 s	—	148.5 s
4a	—	113.0 s	—	112.8 s
5	—	159.1 s	—	157.8 s
6	6.58 s	100.6 s	6.57 s	100.6 s
7	—	155.2 s	—	153.3 s
8	—	98.3 s	—	99.7 s
8a	—	133.8 s	—	133.9 s
9	—	170.6 s	—	171.2 s
10	6.37 s	75.1 d	5.62 2H, s	67.6 t
11	2.82 3H, d (0.8)	25.3 q	2.82 3H, d (0.8)	25.1 q
12	—	170.3 s	—	—
13	3.64 3H, s	52.9 q	—	—
2-OH	11.73 s	—	11.99 s	—
5-OH	9.47 br s	—	9.23 br s	—
7-OH	9.38 br s	—	9.13 br s	—

---

The ¹H and ¹³C NMR data of 1 in acetone-d₆ were similar to those of 2 except for some additional methyl carboxylate signals in 1 (δ_H 3.64/δ_C 52.9 for the C-13 methyl group and δ_C 170.3 for the C-12 ester carbonyl carbon) that coupled with the signals ascribed to an oxygen-bearing C-10 methine group (δ_H 6.73 1H, s and δ_C 75.1 d) in 1 instead of the C-10 methylene signals (δ_H 5.62 2H, s and δ_C 67.6 t) in 2 (Table 1). The chemical shifts of the C-5, C-7, C-8 and C-9 carbons also changed accordingly (Table 1). These ¹H and ¹³C NMR data revealed that 1 was most likely a derivative of 2, with a methyl carboxylate group at C-10. Then, the ¹H and ¹³C NMR data were analysed by 2D ¹H–¹H COSY, HMBC and HMQC spectroscopy [Table S1 in the ESI†]. This enabled us to deduce the planar structure of 1 according to the key NMR data as illustrated in Fig. 2.

Fig. 2 The planar structure and key NMR data for 1.

The absolute configuration of C-10 in 1 could be determined according to the [α]_D and CD data of 1. Compound 1 has only a single chiral center C-10. Thus, the chirality of the C-10 carbon determined the sign of optical rotation of 1. The same is true of the case of the known enantiomer pairs, (3S)- and (3R)-3,4-dihydroisocoumarin-3-carboxylic acids¹⁵ (5a and 5b) and (S)- and (R)-mandelic acids¹⁶ (6a and 6b). All of them also had a single chiral carbon with the local situation similar to that of the C-10 carbon in 1. Of them, the S enantiomers 5a and 6a showed a positive sign of optical rotation in the literature, [α]²⁵_D +52.2 (c 1, MeOH)¹⁵ for 5a and [α]²⁵_D +152.0 (c 1.0, H₂O)^16a for 6a, while the R enantiomers 5b and 6b showed a negative sign of optical rotation, [α]²⁵_D −51.0 (c 1, MeOH)¹⁵ for 5b and [α]²⁵_D −147.8 (c 0.5, MeOH)^16b for 6b. Thus, the positive sign of [α]²⁵_D +184.5 (c 0.21, MeOH) of 1 indicated the 10S configuration of 1. This was further supported by TDDFT electronic CD (ECD) calculations¹⁷ in conjunction with measured CD data. The CD of 1 in MeOH showed Cotton effects at 205.5 (Δε −3.26), 238.5 (+10.16), 256 (shoulder peak, −2.65), 284.5 (−9.71), 340.5 (+2.56) and 371.5 nm (+1.86), as shown in Fig. 3. We performed TDDFT ECD calculations for 1 and its enantiomer ent-1.¹⁸ The calculated ECD spectrum of 1 reproduced well the measured data, whereas ent-1 afforded an opposite ECD spectrum, as shown in Fig. 3. Thus, the absolute configuration at C-10 in 1 was assigned as S.

Fig. 3 Measured and calculated ECD spectra for 1 in MeOH.

Compound 4 was obtained as a pale yellow solid (from MeOH) and showed [α]²⁵_D +442.5 (c 0.20, MeOH). It gave pseudo-molecular ion peaks at m/z 513 [M + Na]⁺ and 489 [M − H]⁻ in the positive and negative ESI-MS, respectively, indicating a molecular weight of 490 Dalton. The ¹H and ¹³C NMR data of 4 in acetone-d₆ (see Table S2 in the ESI†) were identical to those of bacillosporin C in acetone-d₆ in the literature.⁵ Furthermore, the specific rotation of 4, [α]²⁵_D +442.5 (c 0.20, MeOH), was also consistent with the data for bacillosporin C, [α]²⁸_D +451 (c 0.20, acetone)⁴ and [α]²⁷_D +490 (c 0.2, MeOH),⁵ in the literature. This indicated that they have the same absolute configuration, and thus compound 4 was identified as bacillosporin C.^4,5

Although bacillosporin C has been repeatedly isolated from different fungal species,^4–6 the absolute configuration had not yet been determined to date. Therefore, we performed TDDFT ECD calculations¹⁷ on 4 and its enantiomer ent-4 in order to determine the absolute configuration by comparison with the measured CD data. In the CD measured in MeOH, 4 showed Cotton effects at 216 (Δε −26.50), 246.5 (+27.12), 262 (shoulder peak, +13.20), 270 (shoulder peak, +10.45), 292.5 (−3.82), 323 (+2.78) and 351.5 nm (shoulder peak, +1.74), as shown in Fig. 4. In the ECD calculations, the conformational searches and the re-optimization at the B3LYP/6-31+G(d) level were conducted as done for 1/ent-1.¹⁸ The TDDFT ECD calculations were carried out on a set of the two lowest-energy conformations (population > 5%) for 4 and ent-4 (Fig. S2, in the ESI†) each with 20 excited states using a PCM in MeOH. The CD spectra were generated by applying a Gaussian band shape with a 0.33 eV width and then combined for each configuration, as described for 1/ent-1,¹⁸ to obtain the calculated ECD spectra of 4 and ent-4. As given in Fig. 4, the ECD calculated for 4 matched the measured data, whereas the ECD calculated for ent-4 gave just the opposite peaks. Thus, the absolute configuration of bacillosporin C (4) was assigned as 7′S and 8′S.

Fig. 4 Measured and calculated ECD spectra for 4 in MeOH.

To determine whether 1–4 were also produced in the parent strain, the EtOAc extracts of the mutant AD-1-1 and the parent G59 strains were analyzed using HPLC-ESI-MS and HPLC-photodiode array detector-UV scanning. We examined 1–4 in both the EtOAc extracts according to their chromatographic behaviour in the HPLC using compound 1 as a reference standard. In the HPLC-MS/HPLC-UV analyses, 1–4 were detected only in the mutant extract and not the parent extract as determined by the retention times and the MS/UV spectra (Fig. S3 and S4 in the ESI†). This result provided evidence that the production of compounds 1–4 in the mutant AD-1-1 strain was caused by the activation of biosynthetic pathways that were silent in the parent G59 strain and subsequently activated by the DES mutagenesis process in the mutant. Plausible biosynthetic pathways for the production of 1–4 that were likely activated in mutant AD-1-1 are proposed in Scheme 1.

Scheme 1 Plausible biosynthetic pathway for production of 1–4.

A biosynthetic precursor of 1–4 was proposed to be I-1 (Scheme 1), a heptaketide formed entirely from acetate building blocks.¹⁹ In a similar manner to the biosynthesis of sclerodione and sclerodin,¹⁹ the precursor I-1 would give I-3 via the ring-contractive rearrangement of I-2. Then, differing from the decarboxylation and oxidation in the biosynthesis of sclerodione and sclerodin,¹⁹ an intramolecular nucleophilic addition reaction of I-3 followed by C–C bond cleavage in I-4 via an intramolecular S_N2-like reaction and a subsequently occurring keto-enol tautomerism of I-5, as shown in Scheme 1, would afford I-6. Then, the methylation or decarboxylation of I-6 would produce 1 or 2, respectively. Further methylation/oxidation of 2 would produce 3, whereas coupling of two molecules of 2 as described below would give 4. Deprotonation of the 7-OH proton in 2 followed by the rearrangement of the 7,8-double bond could form an electrophilic species ES-2 shown in Scheme 1. Then, electrophilic aromatic substitution of H-6 in 2 by ES-2 would form I-7. Further cyclization of I-7 via an intramolecular nucleophilic addition would produce 4 (Scheme 1).

The inhibitory effect of 1–4 on the growth of tumour cells was tested by the MTT method using the human cancer cell lines K562, HL-60, BGC-823 and HeLa. The cells were treated with samples for 24 h at 37 °C in the present study, and the MTT assay was performed according to the procedure that we have repeatedly used in previous studies.^13,14 In the MTT assay, compounds 1–4 gave IR% values at 100 μg mL⁻¹, listed in Table 2. Among them, only the inhibitory effects of 1 and 2 on the HL-60 cells (64.5% for 1 and 81.6% for 2 at 100 μg mL⁻¹, in Table 2) were stronger than the effect of the positive control 5-FU (5-fluorouracil) on the same HL-60 cells. The positive control 5-FU inhibited the HL-60 cells with an IR% value of 44.7% at 100 μg mL⁻¹ (796.2 μM) under the same conditions (IC₅₀ > 796.2 μM). The IC₅₀ values for 1 and 2 on the HL-60 cells under the same conditions were determined to be 95.2 μg mL⁻¹ (313.2 μM) and 14.0 μg mL⁻¹ (56.9 μM), respectively. Compounds 1 and 2 also weakly inhibited the K562 cells with IR% values of 20.8% and 28.1% at 100 μg mL⁻¹ (328.9 μM for 1 and 406.5 μM for 2), respectively. The positive control 5-FU inhibited the K562 cells with an IR% of 40.3% at 100 μg mL⁻¹ (796.2 μM).

Table 2 IR% values for 1–4 at 100 μg mL⁻¹

Cell lines	K562	HL-60	HeLa	BGC-823
1	20.8	64.5	13.3	11.2
2	28.1	81.6	13.9	6.6
3	5.8	19.7	16.8	2.4
4	—	35.7	16.1	11.9

---

Overy and Blunt^3a have previously shown that the cultivation of phytopathogenic Penicillium hordei using plant tissue media could stimulate the production of four new OPs, corymbiferan lactones A–D. This study^3a was likely considered as a special case of the one-strain-many-compounds strategy.⁸ The idea and performance of the study were similar to the one-strain-many-compounds,⁸ chemical epigenetics⁹ and cocultivation¹⁰ strategies. These strategies, including the study of Overy and Blunt,^3a could activate the production of dormant secondary metabolites in fungi by varying environmental factors for the growth of the producing strains in the cultivation processes. In contrast, in the present study we have presented the activated production of OPs, 1–4, by the DES mutagenesis of a marine-derived fungus P. purpurogenum G59. DES mutagenesis¹⁴ and ribosome engineering¹³ are both mutation-based strategies that allow for the facile selection of fungal mutants with activated biosynthetic pathways. The activated pathways in the mutant strains worked normally under standard culture conditions, and thus those mutant strains could be used to search for new compounds derived from the generally silent biosynthetic pathways under standard culture conditions.^12–14 In comparison with the chemical epigenetics,⁹ one-strain-many-compounds⁸ and cocultivation¹⁰ strategies for activating silent biosynthetic pathways, the mutation-based strategies have been shown to possess the advantage that not only the secondary metabolites derived from activated fungal pathways could be repeatedly obtained by fermentation under standard culture conditions^13,14 but also that the metabolic potential of the mutant strains could be further elicited by changing the fermentation conditions.^13a,e,14d,e The discovery of the novel OP monomer penicimutalidine (1) from the mutant AD-1-1 strain in the present study further demonstrates the effectiveness of our previously reported DES mutagenesis strategy¹⁴ for obtaining new compounds from inactivated fungal pathways. The results of this study coupled with our previously reported results¹⁴ suggest that the DES mutagenesis strategy could be applied to other fungal species to discover new bioactive compounds by activating otherwise silent biosynthetic pathways.

Conclusions

In summary, in the present study we have presented a novel OP penicimutalidine (1) derived from a marine-derived fungus P. purpurogenum G59 by DES mutagenesis. The production of the three known OPs, SF226 (2), corymbiferan lactones A (3) and bacillosporin C (4), was also shown to be activated in the mutant AD-1-1 by the DES mutagenesis of the G59 strain. The absolute configuration of 4 was determined for the first time.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This work was financially supported by grants from the NSFC (30973631, 81573300), NHTRDP (2013AA092901, 2007AA09Z411), NSTMP (2009ZX09301-002/2012ZX09301-003) and AMMS (2008), China, and the NSFC-Shandong Joint Fund for Marine Science Research Centers (U1406402), China.

References

1. J. Ren, F. Zhang, X. Liu, L. Li, G. Liu, X. Liu and Y. Che, Org. Lett., 2012, 14, 6226.
2. Y. Mizushina, T. Takeuchi, S. Kamito, I. Kuriyama, F. Sugawara and H. Yoshida, Jpn. Kokai Tokkyo Koho, JP 2013194044 A, September 30 2013.
3. (a) D. P. Overy and J. W. Blunt, J. Nat. Prod., 2004, 67, 1850; (b) M. Leyte-Lugo, M. Figueroa, M. C. González, A. E. Glenn, M. González-Andrade and R. Meta, Phytochemistry, 2013, 96, 273.
4. M. Yamazaki and E. Okuyama, Chem. Pharm. Bull., 1980, 28, 3649.
5. Z. Guo, C. Shao, Z. She, X. Cai, F. Liu, L. L. P. Vrijimoed and Y. Lin, Magn. Reson. Chem., 2007, 45, 439.
6. T. Dethoup, L. Manoch, A. Kijjoa, M. S. J. Nascimento, P. Puaparoj, A. M. S. Silva, G. Eaton and W. Herz, Planta Med., 2006, 72, 957.
7. (a) C. Hertweck, Nat. Chem. Biol., 2009, 5, 450; (b) R. H. Cichewicz, Nat. Prod. Rep., 2010, 27, 11; (c) A. A. Brakhage and V. Schroeckh, Fungal Genet. Biol., 2011, 48, 15; (d) K. Ochi and T. Hosaka, Appl. Microbiol. Biotechnol., 2013, 97, 87.
8. H. B. Bode, B. Bethe, R. Hofs and A. Zeeck, ChemBioChem, 2002, 3, 619.
9. R. B. Williams, J. C. Henrikson, A. R. Hoover, A. E. Lee and R. H. Cichewicz, Org. Biomol. Chem., 2008, 6, 1895.
10. A. Marmann, A. H. Aly, W. Lin, B. Wang and P. Proksch, Mar. Drugs, 2014, 12, 1043.
11. (a) K. Ochi, S. Okamoto, Y. Tozawa, T. Inaoka, T. Hosaka, J. Xu and K. Kurosawa, Adv. Appl. Microbiol., 2004, 56, 155; (b) K. Ochi, Biosci., Biotechnol., Biochem., 2007, 71, 1373.
12. (a) T. Hosaka, M. Ohnishi-Kameyama, H. Muramatsu, K. Murakami, Y. Tsurumi, S. Kodani, M. Yoshida, A. Fujie and K. Ochi, Nat. Biotechnol., 2009, 27, 462; (b) P. Fu, M. Jamison, S. La and J. B. MacMillan, Org. Lett., 2014, 16, 5656.
13. (a) Y.-J. Chai, C.-B. Cui, C.-W. Li, C.-J. Wu, C.-K. Tian and W. Hua, Mar. Drugs, 2012, 10, 559; (b) C.-J. Wu, L. Yi, C.-B. Cui, C.-W. Li, N. Wang and X. Han, Mar. Drugs, 2015, 13, 2465; (c) Y. Dong, C.-B. Cui, C.-W. Li, W. Hua, C.-J. Wu, T.-J. Zhu and Q.-Q. Gu, Mar. Drugs, 2014, 12, 4326; (d) N. Wang, C.-B. Cui and C.-W. Li, Arch. Pharmacal Res., 2016, 39, 762; (e) L. Yi, C.-B. Cui, C.-W. Li, J.-X. Peng and Q.-Q. Gu, RSC Adv., 2016, 6, 43975.
14. (a) S.-M. Fang, C.-J. Wu, C.-W. Li and C.-B. Cui, Mar. Drugs, 2014, 12, 1788; (b) S.-M. Fang, C.-B. Cui, C.-W. Li, C.-J. Wu, Z.-J. Zhang, L. Li, X.-J. Huang and W.-C. Ye, Mar. Drugs, 2012, 10, 1266; (c) M.-W. Xia, C.-B. Cui, C.-W. Li and C.-J. Wu, Mar. Drugs, 2014, 12, 1545; (d) C.-J. Wu, C.-W. Li and C.-B. Cui, Mar. Drugs, 2014, 12, 1815; (e) M.-W. Xia, C.-B. Cui, C.-W. Li, C.-J. Wu, J.-X. Peng and D.-H. Li, Mar. Drugs, 2015, 13, 5219; (f) C.-W. Li, C.-J. Wu, C.-B. Cui, L.-L. Xu, F. Cao and H.-J. Zhu, RSC Adv., 2016, 6, 73383.
15. H. Maeda, M. Suzuki, H. Sugano, M. Yamamura and R. Ishida, Chem. Pharm. Bull., 1988, 36, 190.
16. (a) H.-R. Huang, J.-H. Xu, Y. Xu, J. Pan and X. Liu, Tetrahedron: Asymmetry, 2005, 16, 2113; (b) L. Chang, L.-M. Ouyang, Y. Xu, J. Pan and J.-H. Xu, J. Mol. Catal. B: Enzym., 2010, 66, 95; (c) D. F. Colon, S. T. Pickard and H. E. Smith, J. Org. Chem., 1991, 56, 2322.
17. (a) N. Berova, L. Di Bari and G. Pescitelli, Chem. Soc. Rev., 2007, 36, 914; (b) G. Bringmann, T. Bruhn, K. Maksimenka and Y. Hemberger, Eur. J. Org. Chem., 2009, 2009, 2717.
18. In the ECD calculations, conformational searches for 1 and its enantiomer ent-1 were performed using the ‘systematic’ procedure implemented in Spartan’14 (Wavefunction: Irvine, CA, USA, 2013), using a Merck molecular force field (MMFF). All of the MMFF minima from the conformational searches were further re-optimized according to the DFT calculations at the B3LYP/6-31+G(d) level using the Gaussian 09 software package (Gaussian 09, Revision A.01. Gaussian: Wallingford, CT, USA, 2009). The TDDFT ECD calculations were performed on a set of the three lowest-energy conformations (population > 5%) for 1 and ent-1 (Fig. S1 in the ESI†) each with 20 excited states using a polarizable continuum model (PCM) in MeOH. The CD spectra were generated using the SpecDis program (T. Bruhn, Y. Hemberger, A. Schaumlöffel and G. Bringmann, SpecDis, Version 1.53. University of Wuerzburg: Wuerzburg, Germany, 2011) by applying a Gaussian band shape with a 0.26 eV width from dipole-length rotational strengths. The spectra of the conformers were combined according to the Boltzmann weighting for each conformer to obtain the ECD spectra of 1 and ent-1.
19. W. A. Ayer, M. S. Pedras and D. Ward, Can. J. Chem., 1987, 65, 760.

---

Footnotes

† Electronic supplementary information (ESI) available: Experimental procedures, DFT-optimized structures for the low-energy conformers of 1/ent-1 and 4/ent-4, figures for the HPLC-MS/HPLC-UV analyses, ESI-MS, HR ESI-MS, UV, IR, and 1D and 2D NMR spectra for 1, and ¹H and ¹³C NMR spectra for 4. See DOI: 10.1039/c6ra17087k

‡ These authors contributed equally to this work.

---