Tatjana J. Kop*a,
Nataša Terzić-Jovanovića,
Željko Žižakb,
Bogdan A. Šolajacd and
Dragana R. Milićc
aUniversity of Belgrade, Institute of Chemistry, Technology and Metallurgy, Department of Chemistry, Njegoševa 12, 11000 Belgrade, Republic of Serbia. E-mail: tatjana.kop@ihtm.bg.ac.rs
bUniversity of Belgrade, Institute for Oncology and Radiology of Serbia, Pasterova 14, 11000 Belgrade, Serbia
cUniversity of Belgrade, Faculty of Chemistry, Studentski trg 12-16, 11158 Belgrade, Serbia
dSerbian Academy of Sciences and Arts, Knez Mihailova 35, 11000 Belgrade, Serbia
First published on 18th July 2022
A new oxidant, containing m-chloroperoxybenzoic acid (MCPBA) and an iron salt, was developed and used for oxidation of steroidal phenols to a quinol/epoxyquinol mixture. Reaction was optimized for estrone, by varying initiators (Fe-salts), reaction temperature, time and mode of MCPBA application. A series of five more substrates (17β-estradiol and its hydrophobized derivatives) was subjected to the optimized oxidation, providing corresponding p-quinols and 4β,5β-epoxyquinols in good to moderate yields. The obtained epoxyquinols were additionally transformed by oxidation, as well as the acid-catalyzed oxirane opening. In a preliminary study of the antiproliferative activity against human cancer cell lines, all newly synthesized compounds expressed moderate to high activity.
Our first attempt to incorporate a cyclohexadienone substructure into a steroidal skeleton resulted in a novel method of oxidation of polycyclic phenols by the MCPBA/(BzO)2/hν system.33,34 Both groups of obtained compounds, steroidal p-quinols and 4β,5β-epoxyquinols, exhibited the antitumor activity.35,36 In addition, they also proved to be good synthetic intermediates for many A-ring polyfunctionalized compounds, including quinones,34,37 diepoxides,36,38 bromo-derivatized quinols and epoxyquinols,38 rearomatized39 and A,B-spiro structures.36 All of them expressed moderate to significant antitumor activity, as well.
Willing to investigate other synthetic approaches to such useful compounds, we developed a new oxidation system and investigated it extensively. In this work, we present the oxidation of estrone (1a) and its five derivatives (of which three have an unprotected 17β-hydroxy group) with meta-chloroperoxybenzoic acid (MCPBA) in the presence of iron salts. The influence of different iron salts and reaction conditions on the reaction efficiency and distribution of products was investigated. Based on those results, a modified procedure for oxidation of steroidal phenols was optimized, producing steroidal p-quinols 2 and epoxyquinols 3 in good overall yields under mild reaction conditions.
Run | Fe-Salt | Fen+:1a (eq.) | MCPBA | T (°C) | t (h) | Yielda (%) | |||
---|---|---|---|---|---|---|---|---|---|
1a | 2a | 3a | 2a + 3a | ||||||
a Isolated yields; yields of unconverted estrone is also included, since conversion is incomplete; the rest up to 100% remained adsorbed on the column as a complex mixture of polar products.b Complex mixture of products-not isolated. | |||||||||
1 | FeSO4 | 1:1 | 3 × 1 eq. | 56 | 4 | 5 | 43 | 47 | 90 |
2 | FeSO4 | 0.2:1 | 3 × 1 eq. | 56 | 4 | 40 | 29 | 28 | 57 |
3 | FeSO4 | 1:1 | 1 × 3 eq. | 56 | 4 | 15 | 36 | 22 | 58 |
4 | FeSO4 | 0.2:1 | 1 × 3 eq. | 56 | 4 | 49 | 33 | 17 | 50 |
5 | FeSO4 | 1:1 | 1 × 3 eq. | 20 | 24 | 19 | 37 | 18 | 55 |
6 | (NH4)2Fe(SO4)2 | 1:1 | 3 × 1 eq. | 56 | 5 | 7 | 36 | 32 | 68 |
7 | (NH4)2Fe(SO4)2 | 0.2:1 | 1 × 3 eq. | 56 | 7 | 25 | 39 | 25 | 64 |
8 | K4[Fe(CN)6] | 1:1 | 3 × 1 eq. | 56 | 5 | 10 | 20 | 13 | 33 |
9 | K4[Fe(CN)6] | 1:1 | 1 × 3 eq. | 20 | 24 | 57 | 17 | 16 | 33 |
10 | FeCl2 | 1:1 | 3 × 1 eq. | 56 | 2.5 | — | — | — | —b |
11 | FeCl2 | 0.2:1 | 1 × 3 eq. | 56 | 5 | 54 | 25 | 11 | 36 |
12 | Fe2(SO4)3 | 1:1 | 3 × 1 eq. | 56 | 4 | 17 | 29 | 11 | 40 |
13 | Fe2(SO4)3 | 0.2:1 | 3 × 1 eq. | 56 | 4 | 39 | 31 | 29 | 60 |
14 | Fe2(SO4)3 | 0.2:1 | 1 × 3 eq. | 56 | 4 | 54 | 29 | 16 | 45 |
15 | Fe2(SO4)3 | 1:1 | 1 × 3 eq. | 20 | 24 | 56 | 28 | 6 | 34 |
16 | Fe2(SO4)3 × (NH4)2SO4 | 0.2:1 | 1 × 3 eq. | 56 | 4 | 25 | 43 | 19 | 62 |
17 | Fe2(SO4)3 × (NH4)2SO4 | 0.2:1 | 1 × 3 eq. | 56 | 7 | 12 | 33 | 31 | 64 |
18 | K3[Fe(CN)6] | 1:1 | 3 × 1 eq. | 56 | 4 | 45 | 30 | 24 | 54 |
19 | K3[Fe(CN)6] | 0.2:1 | 1 × 3 eq. | 56 | 5 | 45 | 30 | 20 | 50 |
20 | K3[Fe(CN)6] | 1:1 | 1 × 3 eq. | 20 | 24 | 48 | 35 | 11 | 46 |
21 | FeCl3 | 1:1 | 3 × 1 eq. | 56 | 2.5 | — | — | — | —b |
22 | FeCl3 | 0.2:1 | 1 × 3 eq. | 56 | 5 | 20 | 2 | 6 | 8 |
In all performed experiments, the same product mixture, consisting of quinol 2a and syn-epoxyquinol 3a, was formed. Their yield and distribution varied with initiators and reaction conditions. The mode of MCPBA application showed the major influence on reaction efficacy (total yield). As can be seen from Table 1, better total yield, together with a significant increase of the epoxide fraction in the product mixture, was achieved when the peracid was added in portions (runs 1 vs. 3, 2 vs. 4, and 13 vs. 14), except in the presence of complex Fe-salts where the mode of application showed no influence on reaction efficacy. The counter ion of the applied initiator was important only in the case of chloride salts, where a low yield of target compounds (run 22), or even the formation of a complex mixture of inseparable products (runs 10 and 21) was obtained, probably due to competitive reactions. The effect of Fe-ion oxidation state on reaction efficacy proved to be irregular and could not be simply generalized – Fe(III) complex salts were mainly more efficient than Fe(II) ones (runs 18 vs. 8, and 20 vs. 9), while in the case of simple salts, both the opposite effect (runs 1 vs. 12, 5 vs. 15, and 11 vs. 22) and quite similar results with Fe(II) and Fe(III) salts (runs 2 vs. 13, 4 vs. 14, and 10 vs. 21) were observed. The amount of Fe(III) ions did not affect the yield of products (run 12 vs. 13), as it was in the case of Fe(II) ions (run 1 vs. 2). In addition, the insolubility of iron salts in acetone causes reaction to take place under heterogenous conditions, so the solid phase surface area and activity could also affect the overall yield and products distribution.
Finally, in the optimal experiment to a refluxing acetonic suspension of estrone (1 eq.) and FeSO4 (1 eq.) MCPBA was added in portions (3 × 1 eq.), with 30 minutes interval between them, and reaction mixture was heated for additional 4 h. After simple work-up, a clean reaction mixture was easily separated by a dry-flash column chromatography on silica-gel, leading to the highest overall yield of products (90%), in a ratio of quinol to epoxyquinol of nearly 1:1 (Table 1, run 1). The rest of the reaction mixture consisted of unreacted estrone (5%) and inseparable mixture of polar products strongly adsorbed on silica during the chromatography.
A series of five more steroidal phenols derived from estrone, including compounds with unprotected hydroxyl-groups, as well as the ester subunits, were subjected to optimized oxidation (Scheme 1). As with estrone (1a), each reaction led to formation of only two products – para-quinol 2 and 4β,5β-epoxyquinol 3 (although theoretically p-quinol could be transformed to four epoxyquinols, with different, α- or β-orientation at different, 1,2- and 4,5-positions).
Scheme 1 Oxidation of steroidal phenols 1 into corresponding quinol/epoxyquinol mixtures (#incomplete conversion; yields recalculated to the amount of converted substrate are shown in parentheses). |
In the presence of unprotected hydroxy-, and ester moieties, as well as aromatic and alkyl side chains, the reaction proceeded in satisfactory yield, as shown in Scheme 1. Total yields of quinols and epoxyquinols ranged from 44% for 17α-benzyl-17β-estradiol (1e) up to 90% for estrone (1a), while their distribution varied from 1:1 to 2:1 approximately.
Fig. 1 EPR spectrum of the reaction of estrone with MCPBA in the presence of Fe(II) ions and DEPMPO as a radical trapper. |
Interestingly, Fe(III) ions also initiated a reaction with estrone without irradiation. It can be assumed that the reaction with Fe(III) starts with the direct one electron oxidation of estrone and the formation of a radical 1A (Scheme 2, red path). The reaction propagates according to the reaction path indicated for systems Fe(II)/MCPBA and MCPBA/(BzO)2/hν. An almost imperceptible difference (meaning overall yield and product ratio) between the reactions performed with stoichiometric and catalytic amounts of Fe(III) supports presumption that the initiating step, oxidation of estrone by Fe(III), is the slowest one and at some point of the reaction is suppressed by aroyloxy radical coming from fast reaction of the MCPBA with 1A. Therefore, starting amount of Fe(III) does not affect the efficiency of the reaction to the same extent as the starting amount of Fe(II)-ions as initiators, which contribute to the amount of m-Cl-C6H4COO˙ by direct reaction with MCPBA. The proximity of the β-hydroxyl group in position 10 directed a further oxygenation of p-quinol from β-side. The abstraction of hydroxy-radical from MCPBA via homolytic cleavage of double bond 4,5 produced a thermodynamically favored C-radical 2A and, followed by ring closure, 4,5-epoxyde rather than 1,2-isomer.40 Conversion of p-quinol to 4β,5β-epoxyquinol (like for previous reaction system) was confirmed experimentally, subjecting it under the same Fe(II)/Fe(III)-MCPBA conditions.
Oxidation by hydrogen peroxide under basic conditions introduced a second oxirane ring in epoxyquinols 3c and 3d to give diepoxides 4c and 4d, respectively, by already known synthetic transformation, applied and investigated for synthesis of analogous 17-keto diepoxyde.38 Modifications of the A-ring were achieved by the acidic oxirane ring opening of epoxyquinol 3a using the LiCl/Amberlyst 15 system. Applied mild reaction conditions provided corresponding chloroquinol 5 as the main product. It is important to note that unlike the previous synthesis of analogous 4-bromoquinol,38 where intermediate bromohydrin was not detected, chlorohydrin 6 expressed a significant degree of stability under the same reaction conditions. So, by shortening the reaction time, it could be isolated as the only product, while prolonging reaction time to 48 h resulted in its almost complete dehydration to chloroquinol 5.
All synthesized p-quinols expressed strong UV absorption around 230–250 nm due to conjugated carbonyl group at C-3 position. Weaker absorption at the similar wavelengths was observed for the corresponding epoxyquinols, while diepoxides were UV inactive compounds. The band around 280 nm, appeared in UV spectra of compounds 2e and 3e, indicated the presence of 17α-benzyl group.
IR spectra of p-quinols and corresponding epoxides contain strong wide maxima at 3100–3000 cm−1 (O–H), strong sharp maxima originated from conjugated CO, at 1660–1670 cm−1 (quinols) and 1680–1690 (epoxyquinols) and at 1610–1635 cm−1 (due to conjugated CC, much weaker for epoxyquinols than for quinols). IR spectra of diepoxides also contain wide maxima at 3100–3000 cm−1, but the characteristic carbonyl band is shifted at higher values (1710–1730 cm−1), due to saturation of the A-ring.
In the 1H NMR spectra of all p-quinols a typical d–dd–t pattern of the A-ring dienone subunit37 was found: a doublet at 7.0–7.3 ppm, with coupling constant around 10 Hz, originating from proton attached to C-1, coupled with cis-vinyl H–C(2); doublet of doublets at 6.1–6.2 ppm belonging to H–C(2), coupled with H–C(1) and H–C(4), with coupling constants around 10 and 2 Hz, indicating coplanar W-geometry of H–C(1)–C(3)–C(4)–H segment; doublet of doublets belonging to H–C(4), appearing as an irregular triplet at 5.9–6.0 ppm, with coupling constants of 2 and 1 Hz, long-range coupled with H–C(2) (W-coupling) and allylic Hβ–C(6) (Fig. 2). An axial orientation of Hβ–C(6) enables maximal overlapping of π-C(4)–C(5) and σ-C(6)–H orbitals and long-range allylic coupling. Several 2D NMR data contributed A-ring assignments: scalar couplings were confirmed from COSY correlations; NOESY correlations between H–C(1) and Hα–C(11) and Hα–C(4)–Hα–C(6) confirmed their spatial proximity (Fig. 2); correct HSQC correlations with carbon signals at expected shifts confirmed position of protons. Hβ–C(6) gave doublet of doublet of doublet of doublets (appearing as triplet of doublet of doublets) at ∼2.8 ppm, which indicates two strong couplings (geminal and anti-diaxial with Hα–C(7)), one moderate (gauche) with Hβ–C(7) and weak allylic with Hα–C(4).
Fig. 2 H,H-COSY and NOE correlations in quinols 2, epoxyquinols 3 and diepoxides 4. COSY couplings: vinylic (H–C(1)–H–C(2)), W (H–C(2)–H–C(4)) and allylic (H–C(4)–Hβ–C(6)). |
Strong coupling (COSY correlation) and shared carbon with Hβ–C(6) at ∼32 ppm (HSQC correlation), alongside with NOESY correlation with Hα–C(4), indicated that doublet of doublets of doublets at 2.3–2.4 ppm belongs to Hα–C(6). Signals at ∼1.0 ppm and ∼2.0 ppm were assigned to Hα–C(7) and Hβ–C(7), respectively, regarding the COSY, HMBC and NOESY correlations to adjacent protons attached to C(6) and C(8), and correlation of both signals with one carbon signal at ∼33 ppm in HSQC spectrum. Corresponding signals of protons attached to C(11) and C(12) were assigned in the similar way, having in mind coupling with Hα–C(9) and NOESY correlation of Hα–C(11) with H–C(1). Correlations with H–C(14) enabled assignment of protons in position 15, and consequently, in position 16. 1H NMR spectra of 4β,5β-epoxyquinols also contain characteristic signals of vinylic protons: doublet at ∼6.7 ppm for H–C(1) and doublet of doublets for H–C(2) at ∼5.8 ppm. Oxirane proton attached to C(4) gives doublet on lower chemical shift (∼3.3 ppm), comparing to quinol, without coupling with H–C(6). Saturation of 4–5 bond led to the absence of H–C(4)–Hβ–C(6) coupling and to lower chemical shifts of protons in position 6, as well. Although β-orientation of the epoxy moiety could not be doubtlessly determined from observed H–C(4)–Hα–C(6) NOE correlation35 (Fig. 2), it was verified from further transformation of epoxyquinols 3 to syn-diepoxyalcohols 4. Similar chemical shifts of protons in positions 4 and 6 were noticed for diepoxides, but 1H NMR spectra and 13C NMR spectra of diepoxides are also characterized by total loss of the vinylic signals. Signals of H–C(1), H–C(2) and H–C(4) are between 3.0 and 4.0 ppm, with significantly lower J1,2 (∼4 Hz), comparing to those of corresponding p-quinols and epoxyquinols. The presence of the W-coupling between H–C(2) and H–C(4) indicated their coplanar position. The NOE correlations of H–C(1) with Hα–C(11), and H–C(4) with Hα–C(6) (Fig. 2) indicated α-orientation of H–C(1) and H–C(4), and due to H-(2/4) coplanarity of H–C(2), as well. Consequently, both oxirane rings adopted β-orientation, as it was found earlier for 17-keto analog.36,38
All compounds expressed dose-dependent and phenotype-independent activity. Quinol 2b proved to be completely inactive, but modification of D-ring substituents, as well as the introduction of chlorine into the quinol subunit led to an increase in the antineoplastic activity, especially against human cervix carcinoma (HeLa) and human myelogenous leukemia (K562) cells (2b vs. 2d,e,f and 5). However, a moderate activity was achieved employing quinols. On the other hand, their further oxidation to epoxyquinols and diepoxides resulted in the significant increase of the antiproliferative activity against all three tested cell lines (2 vs. 3/4). The influence of oxirane fragment was not found to be additive, since the first epoxidation provoked much higher activity increase (2d vs. 3d) in comparison to the second one (3d vs. 4d). In addition, quite good antiproliferative activity was preserved even after the opening of the epoxy-ring by chloride (Fig. 2, compound 6). At the same time, there is a noticeable increase in activity of quinols and epoxyquinols with 17β-propioniloxy- (2d and 3d) and 17α-n-butyl-substituents (2f and 3f), relative to other members of two series. Diepoxides 4c and 4d showed IC50 values less than 1 μM in all cases, except for 4c against HeLa cells, where this value slightly exceeded 1 μM.
Footnote |
† Electronic supplementary information (ESI) available: Experimental part, tabular representation of the antineoplastic activity and NMR spectra (1H, 13C, COSY, HSQC, HMBC) of newly synthesized compounds. See https://doi.org/10.1039/d2ra03717c |
This journal is © The Royal Society of Chemistry 2022 |