J.
Csábi
a,
A.
Martins§
b,
I.
Sinka
c,
A.
Csorba
a,
J.
Molnár
b,
I.
Zupkó
cd,
G.
Tóth
ae,
L. M. V.
Tillekeratne
f and
A.
Hunyadi
*ad
aInstitute of Pharmacognosy, University of Szeged, Eötvös str. 6, 6720 Szeged, Hungary. E-mail: hunyadi.a@pharm.u-szeged.hu; Fax: +3662545704; Tel: +3662546456
bDepartment of Medical Microbiology and Immunobiology, University of Szeged, Dóm sq. 9, 6720 Szeged, Hungary
cDepartment of Pharmacodynamics and Biopharmacy, University of Szeged, Eötvös str. 6, 6720 Szeged, Hungary
dInterdisciplinary Centre for Natural Products, University of Szeged, Eötvös str. 6, 6720 Szeged, Hungary
eNMR group, Department of Inorganic and Analytical Chemistry, Budapest University of Technology and Economics, Szt. Gellért Sq. 4, H-1111, Budapest, Hungary
fDepartment of Medicinal and Biological Chemistry, College of Pharmacy and Pharmaceutical Sciences, MS 606, University of Toledo, Toledo, OH 43606, USA
First published on 26th September 2016
Efflux pumps, like the ABCB1 transporter, play an important role in the chemo-resistance of various tumors and particularly of cancer stem cells. We have previously reported the chemo-sensitizing activity of apolar ecdysteroid derivatives on cancer cell lines of various origin and sensitivity to chemotherapeutics. Herein we report the preparation of three fluorinated derivatives and a dehydrated byproduct from 20-hydroxyecdysone 2,3;20,22-diacetonide (1) through diethylaminosulfur trifluoride (DAST)-mediated fluorination. Complete NMR assignment of the products is provided. In vitro bioactivity testing was performed on four human breast cancer cell lines, a neuroblastoma, and a mouse lymphoma cell line and its counterpart expressing the human ABCB1 efflux transporter. Fluorination increased the ABCB1 inhibitory effect of the compounds but had little effect on their limited antiproliferative action, which was, however, markedly increased by a Δ14,15 double bond. Compound 5, a new 14,25-difluoro analog of 1, exerted higher chemo-sensitizing activity to doxorubicin as compared to its parental compound.
An increasing number of fluorinated antitumor agents are becoming available for cancer treatment.12 Special properties of the fluorine atom, such as strong electronegativity, small size and the low polarizability of the C–F bond, can have great impact on the biological activity and frequently also on the metabolic stability of a molecule, and fluorination of natural products appears to be an increasingly attractive strategy to obtain new leads for drug discovery.13 Accordingly, our aim was to prepare fluorinated derivatives of 20E 2,3;20,22-diacetonide (1), an ecdysteroid with particularly strong chemo-sensitizing properties, and to investigate related changes in the antitumor potential as compared to that of compound 1. Fluorination of this compound was previously performed in an earlier study by Pascual et al. aiming to prepare 25-fluoroponasterone A, where the acetonide moieties served as protecting groups and were eventually removed by acidic hydrolysis. The 25-fluorinated target compound was tested in an insect bioassay, and it showed an activity on the oocyte development of the German cockroach (Blattella germanica) similar to that of the parental 20E.14 To the best of our knowledge, however, fluorinated ecdysteroids have never been reported for their activity on any bioassays related to mammals.
The mass spectrum of compound 2 suggested that the elimination of a water molecule from 1 took place. This was confirmed by the NMR spectra indicating the presence of a Δ14,15 double bond, represented by the appearance of the chemical shifts of HC-15 (6.08/130.1 ppm) and the quaternary C-14 (150.4 ppm) observed in the 1H and/or the 13C NMR spectra. Accordingly, the HMQC spectrum revealed seven methylene groups, one less than in compound 1, while the disappearance of a hydrogen atom from position 15 was also detected in the 1H,1H COSY spectrum leaving only four members in this structural fragment of correlated protons: H-15 (6.08 ppm), H2-16 (2.38; 2.63 ppm) and H-17 (2.08 ppm). These assignments were supported by the HMBC spectrum. Moreover, C-28 and C-29 showed HMBC cross peaks with two methyl groups each analogously to the parent compound 1, providing further evidence that the acetonide groups at position 2,3 and 20,22 remained intact after the reaction. Hence, compound 2 was identified as Stachysterone B 2,3;20,22-diacetonide.15
In case of compounds 3–5, evidence for fluorine substitution was found. It is well-known that, as a result of altered substituent increments, the exchange of an sp3C connected –OH group to a –F manifests in characteristic changes in the NMR spectrum. In the α position, namely directly on the substituted carbon, a ca. 20–25 ppm paramagnetic shift and in the β position a ca. 2–3 ppm diamagnetic shift can be observed, which effect decreases below 1 ppm in the γ position. In addition to these, both the 13C and the 1H NMR spectra show signal splitting caused by the characteristic direct (∼165 Hz), geminal (22–26 Hz) and vicinal (∼5 Hz) nJ(F,C) and 3J(F,H) (4.5 Hz) couplings. Based on these, it could be evidenced that compounds 3–5 contain a fluorine atom connected to the C-25, and that compound 5 contains another fluorine substituent also at position C-14. Fluorination of alcohols catalyzed by DAST can take place through either SN1 or SN2 reaction mechanism, which also determines the stereo-specificity of the reaction:16a priori, the 14-F substituent in compound 5 can be present in either α or β position. The latter case would also involve a change of the initially trans C/D ring junction to cis. The effect of such a configurational change on the 13C chemical shifts can well be estimated by comparing the corresponding chemical shifts of 20-hydroxyecdysone and its diastereomer 14-epi-20-hydroxyecdysone, where significant, i.e. more than 2 ppm differences could be detected on the δC-9 (+2.4), δC-12 (+9.2), δC-13 (+4.0), δC-15 (+9.0), δC-16 (+3.2) and δC-17 (+6.2) (our own unpublished NMR data for 14-epi-20-hydroxyecdysone isolated from the plant Serratula wolffii17). Considering that no such changes were detected in case of compound 5, a retained C-14 configuration can be concluded. Moreover, a detailed analysis of the NOESY spectrum revealed the steric proximity of the H3C-18 and the Hβ-15 hydrogen atoms, which is only possible in case of a trans C/D ring junction, providing further evidence for a 14α-F group in compound 5. Accordingly, the SN1 reaction mechanism is suggested for this substitution, which mechanism is also more likely whenever steric effects (e.g. due to a rigid steroid skeleton, as in our case) prevent an SN2 attack.16 Altogether, compound 5 was identified as 14-deoxy-14,25-difluoroponasterone A 2,3;20,22-diacetonide, a new ecdysteroid.
Thanks to the comprehensive one- and two-dimensional NMR techniques utilized in the structure elucidation process, a complete signal assignment could be achieved for all compounds including compounds 3 and 4, whose previously published NMR data lacked the assignment of several hydrogens, and certain carbon signals were assigned as interchangeable.14 The 1H and 13C NMR data of parent compound 1 and its derivatives 2–5 are compiled in Table 1, structures of the compounds are shown in Fig. 1.
Atom no. | 1 | 2 | 3 | 4 | 5 | |||||
---|---|---|---|---|---|---|---|---|---|---|
H | C | H | C | H | C | H | C | H | C | |
a In case of compounds 3–5 the symbol “d” after the 13C chemical shifts indicates doublet splitting (Hz) resulted from 1J(F,C), 2J(F,C) or 3J(F,C) coupling. On signal C-13 in compound 5, due to overlapping with the strong solvent signal, the determination of 3J(F,C) failed. | ||||||||||
1α | 1.99 | 39.0 | 1.95 | 38.8 | 1.97 | 39.0 | 1.96 | 38.8 | 1.99 | 38.5 |
β | 1.22 | 1.24 | 1.21 | 1.23 | 1.27 | |||||
2 | 4.27 | 73.7 | 4.24 | 73.5 | 4.24 | 73.7 | 4.23 | 73.5 | 4.28 | 73.6 |
3 | 4.30 | 73.3 | 4.30 | 73.4 | 4.29 | 73.3 | 4.28 | 73.4 | 4.30 | 73.1 |
4α | 1.98 | 27.9 | 1.83 | 28.1 | 1.96 | 27.8 | 1.83 | 28.0 | 1.93 | 27.7 |
β | 1.96 | 1.98 | 1.96 | 1.98 | 1.93 | |||||
5 | 2.24 | 52.7 | 2.22 | 52.5 | 2.22 | 52.6 | 2.22 | 52.5 | 2.31 | 52.3 |
6 | — | 205.8 | — | 205.3 | — | 205.8 | — | 205.3 | — | 204.7 |
7 | 5.80 | 122.0 | 6.06 | 121.0 | 5.78 | 122.0 | 6.06 | 121.0 | 5.92 | 124.4d 6.5 |
8 | — | 167.1 | — | 158.0 | — | 167.0 | — | 158.0 | — | 159.4d 20.5 |
9 | 2.93 | 35.9 | 2.54 | 40.0 | 2.91 | 35.9 | 2.54 | 40.0 | 2.76 | 37.3 |
10 | — | 38.9 | — | 39.7 | — | 38.9 | — | 39.7 | — | 38.8 |
11α | 1.76 | 21.8 | 1.81 | 21.7 | 1.76 | 21.7 | 1.81 | 21.7 | 1.84 | 21.8 |
β | 1.66 | 1.73 | 1.64 | 1.75 | 1.69 | |||||
12α | 2.10 | 32.5 | 1.59 | 40.7 | 2.08 | 32.4 | 1.58 | 40.7 | 1.98 | 33.0d 4.3 |
β | 1.87 | 2.25 | 1.82 | 2.25 | 1.98 | |||||
13 | — | 48.7 | — | 48.7 | — | 48.7 | — | 48.9 | — | 49.7da |
14 | — | 85.4 | — | 150.4 | — | 85.4 | — | 150.4 | — | 108.9d |
165.6 | ||||||||||
15α | 1.61 | 31.8 | 6.08 | 130.1 | 1.59 | 31.7 | 6.08 | 130.1 | 1.88 | 29.1d 24.0 |
β | 1.95 | 1.94 | 2.13 | |||||||
16α | 1.84 | 22.6 | 2.38 | 32.8 | 1.84 | 22.6 | 2.38 | 32.8 | 1.86 | 22.2 |
β | 2.04 | 2.63 | 2.02 | 2.63 | 2.12 | |||||
17 | 2.31 | 50.6 | 2.08 | 59.0 | 2.27 | 50.6 | 2.07 | 58.9 | 2.17 | 51.3 |
18 | 0.82 | 17.8 | 1.07 | 20.0 | 0.80 | 17.8 | 1.07 | 20.0 | 0.86 | 17.1d 3.5 |
19 | 0.96 | 24.2 | 0.94 | 23.8 | 0.95 | 24.2 | 0.94 | 23.8 | 1.00 | 22.6 |
20 | — | 86.0 | — | 84.8 | — | 85.9 | — | 84.8 | — | 85.3 |
21 | 1.18 | 22.8 | 1.21 | 22.0 | 1.15 | 22.8 | 1.21 | 22.0 | 1.19 | 24.2 |
22 | 3.69 | 83.5 | 3.74 | 83.2 | 3.68 | 83.0 | 3.76 | 82.7 | 3.71 | 82.8 |
23 | 1.52 | 24.9 | 1.56 | 24.9 | 1.52 | 24.5d | 1.491.56 | 24.6d 4.5 | 1.53 | 24.5d 4.5 |
1.51 | 1.49 | 1.51 | 5.1 | 1.53 | ||||||
24 | 1.49 | 42.4 | 1.46 | 42.1 | 1.61 | 40.2d | 1.63 | 39.9d | 1.63 | 40.0d 23.1 |
1.73 | 1.70 | 1.86 | 22.9 | 1.85 | 22.5 | 1.87 | ||||
25 | — | 71.3 | — | 71.2 | — | 96.5d | — | 96.6d | — | 96.0d 166.0 |
164.9 | 164.9 | |||||||||
26 | 1.20 | 29.1 | 1.17 | 29.0 | 1.34 | 26.8d | 1.35 | 26.7d | 1.33 | 26.8d 24.8 |
24.5 | 24.9 | |||||||||
27 | 1.21 | 29.0 | 1.19 | 29.7 | 1.31 | 27.4d | 1.31 | 27.5d | 1.36 | 27.3d 24.7 |
24.4 | 24.4 | |||||||||
28 | — | 109.6 | — | 109.6 | — | 109.6 | — | 109.6 | — | 109.6 |
29 | — | 108.2 | — | 108.2 | — | 108.3 | — | 108.3 | — | 108.4 |
iPr-(2,3) β | 1.47 | 29.0 | 1.45 | 28.9 | 1.45 | 29.0 | 1.45 | 28.9 | 1.47 | 29.0 |
iPr-(2,3) α | 1.32 | 26.8 | 1.30 | 26.7 | 1.30 | 26.8 | 1.30 | 26.8 | 1.31 | 26.7 |
iPr-(20,22) β | 1.32 | 27.3 | 1.28 | 27.3 | 1.30 | 27.3 | 1.28 | 27.3 | 1.32 | 27.2 |
iPr-(20,22) α | 1.39 | 29.5 | 1.38 | 29.4 | 1.37 | 29.5 | 1.39 | 29.4 | 1.40 | 29.4 |
Based on previous reports on fluorination reactions using DAST, it is not surprising, that two kinds of structural modifications took place in this setting: the OH group in position 14 was either eliminated as in compounds 2 and 4, or substituted with fluorine as in compound 5. Furthermore, fluorine substitution of the 25-OH group led to the production of compounds 3–5.
Compounds 1–5 were tested for their antiproliferative activities on a diverse set of tumor cell lines including a panel of human breast cancer cell lines (MCF-7, T47D, MDA-MB-231 and MDA-MB-361), a human neuroblastoma cell line (SH-SY5Y), and a murine lymphoma cell line (L5178) and its transfected multi-drug resistant (MDR) counterpart expressing the human ABCB1 efflux transporter (L5178MDR). The compounds were also tested for their ability to interfere with the efflux function of ABCB1 on the L5178MDR cell line, as determined by means of the intracellular accumulation of rhodamine 123, a well-known ABCB1 substrate fluorescent dye; results of these studies are compiled in Table 2. In general, fluorine substitution at C-25 or C-14 and C-25 appears to have little effect on the very mild antiproliferative activity of compound 1, as seen from a comparison of the results obtained for compounds 1, 3 and 5. Moreover, a comparison of the activities of compounds 2 and 4 leads to the same conclusion. On the other hand, a Δ14,15 double bond, formed by the elimination of the 14-OH group, markedly increased the antiproliferative activity of compounds 2 and 4, as compared to compounds 1 and 3, respectively. It is worth noting that, even though all of the above structural changes led to more lipophilic compounds, their activity did not show any correlation to the logp values, which were calculated as 4.01, 4.61, 4.91, 5.50 and 5.80 for compounds 1–5 by using ChemAxon's web based resource available at http://chemicalize.org.18 The most relevant cell line specific differences in the antiproliferative potential of the compounds were observed between the MCF-7 and T47D cells. Although receptorial status (i.e. expression of estrogenic and progestin receptors) of these two cell lines is similar, the expression levels of steroid metabolizing enzymes are substantially different.19 Interestingly, an approximately 3-times selectivity towards T47D over MCF-7 cells was observed in case of compounds with a Δ14,15 double bond (compounds 2 and 4) or a 14-F moiety (compound 5), which selectivity was not present for neither of the two 14-hydroxyecdysteroids (compounds 1 and 3). Moreover, both the fluorination (at either position) and the 14-OH elimination manifested in a significant increase in the ABCB1 inhibitory activity, as compared to the parental compound 1, particularly when both a Δ14,15 double bond and a 25-fluoride group was present as in compound 4. Concerning the chemical changes at C-14, it might be worth noting that dehydroxylation at this position is among the known primary metabolic conversions of ecdysteroids in mammals.3,20 Accordingly, one could also expect a higher metabolic stability for 14-fluorinated (5) or dehydrated compounds (2 and 4) as compared to that of the 14-hydroxyecdysteroids 1 and 3; this should certainly be clarified by future studies.
IC50 (μM) | FAR | ||||||||
---|---|---|---|---|---|---|---|---|---|
MCF-7 (n) | T47D (n) | MDA-MB-231 (n) | MDA-MB-361 (n) | SH-SY5Y (n) | L5178 (n) | L5178MDR (n) | 2 μM | 20 μM | |
1 | 75.1 ± 3.4 (5) | 84.7 ± 3.9 (5) | 106.1 ± 7.2 (4) | 69.2 ± 6.0 (4) | 126.8 ± 9.8 (3) | 82.9 ± 1.0 (2) | 106.1 ± 3.3 (2) | 3.33 | 11.28 |
2 | 30.1 ± 0.8 (5) | 10.9 ± 0.3 (5) | 38.53 ± 3.8 (5) | 13.8 ± 0.4 (5) | 20.8 ± 0.3 (3) | 14.6 ± 0.3 (3) | 19.3 ± 0.5 (3) | 1.28 | 57.36 |
3 | 63.1 ± 2.3 (5) | 70.3 ± 1.3 (5) | 48.85 ± 2.0 (5) | 30.9 ± 2.6 (4) | 70.7 ± 4.1 (3) | 48.3 ± 0.2 (3) | 50.1 ± 2.1 (3) | 1.59 | 38.62 |
4 | 43.8 ± 1.7 (4) | 17.4 ± 1.0 (5) | 50.18 ± 2.6 (5) | 11.8 ± 1.2 (5) | 18.8 ± 1.2 (3) | 14.1 ± 0.8 (3) | 17.4 ± 1.2 (3) | 31.15 | 109.37 |
5 | 127.5 ± 4.3 (5) | 49.2 ± 5.0 (5) | 98.43 ± 17.2 (4) | 53.8 ± 5.7 (5) | 125.6 ± 13.2 (2) | 88.6 ± 1.1 (2) | 82.3 ± 1.4 (3) | 11.02 | 89.78 |
C | 5.8 ± 0.3 (5) | 9.8 ± 1.0 (5) | 19.1 ± 1.8 (5) | 3.7 ± 0.5 (5) | — | — | — | — | — |
D | — | — | — | — | 0.14 ± 0.02 (3) | 0.23 ± 0.01 (3) | 3.54 ± 0.44 (3) | — | — |
Following the above observations, our next aim was to test the effect of these structural and bioactivity changes on the chemo-sensitizing properties of compound 1. Accordingly, the antiproliferative activity of all compounds was tested in combination with doxorubicin on the mouse lymphoma cell line pair, by using the checkerboard microplate method and calculating combination index (CI) values to characterize and quantify the interaction between the two drugs.21 Results of this study are compiled in Table 3.
Cell line | Drug ratioa | CI values at | D m | m | r | CIavg | |||
---|---|---|---|---|---|---|---|---|---|
ED50 | ED75 | ED90 | |||||||
a Molar drug ratios are given; serial dilutions of doxorubicin were initiated from a commercially available injection of 2 mg mL−1 (doxorubicin hydrochloride, Teva). | |||||||||
1 | L5178MDR | 10.2:1 | 0.31 | 0.18 | 0.11 | 8.926 | 1.902 | 0.937 | 0.17 (ref. 6) |
20.4:1 | 0.27 | 0.14 | 0.07 | 11.678 | 3.246 | 0.964 | 0.13 (ref. 6) | ||
81.5:1 | 0.39 | 0.22 | 0.12 | 26.598 | 3.859 | 0.970 | 0.20 (ref. 6) | ||
L5178 | 81.5:1 | 0.74 | 0.71 | 0.68 | 6.742 | 1.614 | 0.933 | 0.70 | |
163:1 | 0.67 | 0.55 | 0.46 | 11.236 | 2.103 | 0.942 | 0.53 | ||
326:1 | 0.61 | 0.55 | 0.51 | 17.795 | 1.872 | 0.945 | 0.54 | ||
2 | L5178MDR | 10.2:1 | 0.55 | 0.45 | 0.39 | 7.300 | 2.460 | 0.992 | 0.44 |
20.4:1 | 0.54 | 0.44 | 0.37 | 8.338 | 2.905 | 0.989 | 0.42 | ||
81.5:1 | 1.09 | 0.87 | 0.69 | 19.144 | 3.664 | 0.985 | 0.82 | ||
L5178 | 81.5:1 | 0.99 | 0.89 | 0.82 | 9.506 | 2.300 | 0.982 | 0.87 | |
163:1 | 0.93 | 0.89 | 0.86 | 11.103 | 2.244 | 0.970 | 0.88 | ||
326:1 | 0.90 | 0.81 | 0.73 | 12.112 | 2.720 | 0.991 | 0.78 | ||
3 | L5178MDR | 10.2:1 | 0.19 | 0.17 | 0.16 | 4.617 | 2.646 | 0.999 | 0.17 |
20.4:1 | 0.19 | 0.19 | 0.19 | 5.760 | 2.468 | 0.908 | 0.19 | ||
81.5:1 | 0.35 | 0.33 | 0.33 | 13.551 | 3.436 | 0.938 | 0.33 | ||
L5178 | 81.5:1 | 0.80 | 0.66 | 0.56 | 9.945 | 2.675 | 0.982 | 0.63 | |
163:1 | 0.85 | 0.76 | 0.71 | 15.042 | 2.488 | 0.978 | 0.75 | ||
326:1 | 1.04 | 0.96 | 0.91 | 23.554 | 2.696 | 0.981 | 0.95 | ||
4 | L5178MDR | 10.2:1 | 0.56 | 0.34 | 0.22 | 9.059 | 5.479 | 1.000 | 0.32 |
20.4:1 | 0.48 | 0.36 | 0.28 | 9.620 | 3.338 | 0.983 | 0.34 | ||
81.5:1 | 0.78 | 0.73 | 0.69 | 19.112 | 2.281 | 0.935 | 0.72 | ||
L5178 | 81.5:1 | 0.73 | 0.80 | 0.89 | 6.893 | 2.560 | 0.965 | 0.83 | |
163:1 | 0.80 | 0.87 | 0.95 | 8.861 | 2.850 | 0.987 | 0.90 | ||
326:1 | 0.79 | 0.84 | 0.90 | 9.569 | 3.217 | 0.993 | 0.86 | ||
5 | L5178MDR | 10.2:1 | 0.23 | 0.11 | 0.06 | 5.904 | 4.546 | 0.948 | 0.10 |
20.4:1 | 0.21 | 0.12 | 0.07 | 7.733 | 3.375 | 0.957 | 0.11 | ||
81.5:1 | 0.24 | 0.16 | 0.11 | 13.888 | 3.040 | 0.966 | 0.15 | ||
L5178 | 81.5:1 | 0.70 | 0.42 | 0.27 | 6.822 | 2.765 | 0.990 | 0.39 | |
163:1 | 0.60 | 0.44 | 0.34 | 9.757 | 2.248 | 0.944 | 0.42 | ||
326:1 | 0.72 | 0.59 | 0.52 | 17.179 | 2.134 | 0.990 | 0.58 |
Considering that in case of chemotherapy the desirable outcome is a complete eradication of the tumor, a weighted average CI value (where CIs at higher inhibition rates count more) has been suggested as an important measure for the relevance of synergism or antagonism on cancer cell lines.21 As compared to compound 1 that acted in strong synergism with doxorubicin, the more cytotoxic dienone compounds 2 and 4 exerted a much less profound chemo-sensitizing activity on both cell lines. This is particularly interesting in view of their high ABCB1 inhibitory activity, providing further evidence to our previous assumption that a functional efflux pump inhibition is unlikely to be the key mechanism for chemo-sensitization by ecdysteroids.6,9 A 25-fluoro substitution had a small, but rather weakening effect on the chemo-sensitizing activity as compared to the corresponding 25-hydroxyecdysteroids. On the other hand, the 14,25-difluorinated compound 5 showed a stronger synergism with doxorubicin as compared to the parental compound 1 in particularly on the non-MDR cell line L5178. Structure–activity relationships observed in our studies are summarized in Fig. 2.
Considering the major importance of the ABCB1 efflux pump in the chemo-resistance and stem-likeness of cancer stem cells,10,11 compounds that can sensitize ABCB1 over-expressing cancer cells to chemotherapeutics are of potential interest to the CSC paradigm. Based on its highly potent chemo-sensitizing activity on the mouse lymphoma cell line pair, compound 5, a new 14,25-difluorinated derivative of 1, is particularly promising in this regard; further studies are planned to investigate its potential use in the eradication of cancer stem cells.
Footnotes |
† The authors declare no competing interests. |
‡ Electronic supplementary information (ESI) available: NMR spectra of compound 5, HPLC chromatograms of compounds 2–5. See DOI: 10.1039/c6md00431h |
§ Permanent address: Synthetic Systems Biology Unit, Institute of Biochemistry, Biological Research Centre, Temesvári krt. 62, 6726 Szeged, Hungary. |
This journal is © The Royal Society of Chemistry 2016 |