Phosphodiester Silybin Dimers Powerful Radical Scavengers: A Antiproliferative Activity on Different Cancer Cell Lines

Silibinin is the main biologically active component of silymarin extract and consists of a mixture 1:1 of two diastereoisomeric flavonolignans, namely silybin A (1a) and silybin B (1b), which we call here silybins. Despite the high interest in the activity of this flavonolignan, there are still few studies that give due attention to the role of its stereochemistry and, there is still today a strong need to investigate in this area. In this regard, here we report a study concerning the radical scavenger ability and the antiproliferative activity on different cell lines, both of silybins and phosphodiester-linked silybin dimers. An efficient synthetic strategy to obtain silybin dimers in an optical pure form (6aa, 6ab and 6bb) starting from a suitable building block of silybin A and silybin B, obtained by us from natural extract silibinin, was proposed. New dimers show strong antioxidant properties, determined through hydroxyl radical (HO●) scavenging ability, comparable to the value reported for known potent antioxidants such as quercetin. A preliminary screening was performed by treating cells with 10 and 50 μM concentrations for 48 h to identify the most sensitive cell lines. The results show that silibinin compounds were active on Jurkat, A375, WM266, and HeLa, but at the tested concentrations, they did not interfere with the growth of PANC, MCF-7, HDF or U87. In particular, both monomers (1a and 1b) and dimers (6aa, 6ab and 6bb) present selective anti-proliferative activity towards leukemia cells in the mid-micromolar range and are poorly active on normal cells. They exhibit different mechanisms of action in fact all the cells treated with the 1a and 1b go completely into apoptosis, whereas only part of the cells treated with 6aa and 6ab were found to be in apoptosis.


Introduction
Oxidative stress can cause cell injury and death, which may be related to numerous diseases and conditions, such as liver damage, aging, cancer, stroke, Alzheimer's disease, and Parkinson's disease [1,2]. The well-known capability of flavonoids to scavenge reactive oxygen species (ROS) is frequently cited as the key property underlying the prevention of and/or reduction in oxidative stress-related chronic diseases and age-related disorders, such as cardiovascular diseases, carcinogenesis, and neurodegeneration. However, many studies have suggested that the therapeutic activity of these compounds involves other properties their ability to directly bind to target peptides [3], inducing the inhibition of key enzymes, the modulation of cell receptors or transcription factors, as well as the perturbation of protein (or peptide) aggregates, who are known to regulate many cell functions. Flavonoids have a broad spectrum of biological activities and administering high dosages could trigger side effects. One strategy to improve the potency and selectivity of flavonoids is to take advantage of the dimeric nature of biflavonoids, thereby facilitating simultaneous interactions through the binding of multiple sites of a biological target [4][5][6]. Similar to flavonoid dimers, flavonolignan dimers or simply bi-flavonolignans are also an emerging class of dimeric compounds that unlike bi-flavonoids, which are very widespread in nature, consist of synthetic dimers of few flavonolignans isolated from the milk thistle Silybum marianum [L. Gaertn. (Asteraceae)] [7]. In this frame, recently we reported the synthesis of new silibinin dimers in which the two monomer units are linked through a phosphodiester bridge, between two aliphatic OH functions (Phosphate-Linked Silybin dimers, Figure 1) [8].
ever, many studies have suggested that the therapeutic activity of these compounds involves other properties their ability to directly bind to target peptides [3], inducing the inhibition of key enzymes, the modulation of cell receptors or transcription factors, as well as the perturbation of protein (or peptide) aggregates, who are known to regulate many cell functions.
Flavonoids have a broad spectrum of biological activities and administering high dosages could trigger side effects. One strategy to improve the potency and selectivity of flavonoids is to take advantage of the dimeric nature of biflavonoids, thereby facilitating simultaneous interactions through the binding of multiple sites of a biological target [4][5][6]. Similar to flavonoid dimers, flavonolignan dimers or simply bi-flavonolignans are also an emerging class of dimeric compounds that unlike bi-flavonoids, which are very widespread in nature, consist of synthetic dimers of few flavonolignans isolated from the milk thistle Silybum marianum [L. Gaertn. (Asteraceae)] [7]. In this frame, recently we reported the synthesis of new silibinin dimers in which the two monomer units are linked through a phosphodiester bridge, between two aliphatic OH functions (Phosphate-Linked Silybin dimers, Figure 1) [8]. Silibinin is a diastereoisomeric mixture of two flavonolignans, namely, silybin A (SilA) and silybin B (SilB) (1a and 1b, Figure 1), in a ratio of approximately 1:1, extracted from milk thistle seeds [9,10]. Silibinin has been used as a traditional drug to treat a range of liver disorders, including hepatitis and cirrhosis. The manifold inhibitory effects of silibinin against various cancer cells include growth inhibition, anti-inflammation, cell cycle regulation, apoptosis induction, chemo-sensitization, inhibition of angiogenesis, reversal of multi-drug resistance, and inhibition of invasion and metastasis [11,12]. Many in vitro and in vivo reports on the activity of silibinin, clearly neglect the structure-activity relationship of the pair of diastereoisomers, using silibinin, the natural mixture of the two flavonolignans SilA and SilB, for all experiments. These studies have mostly disregarded this aspect because of the difficulty separating, on a preparative scale, two diastereoisomers (1a and 1b). In the 2021 a very in-depth study by Kren et al. [13] on the central role of stereochemistry in the pharmacological properties of silybin, Silibinin is a diastereoisomeric mixture of two flavonolignans, namely, silybin A (SilA) and silybin B (SilB) (1a and 1b, Figure 1), in a ratio of approximately 1:1, extracted from milk thistle seeds [9,10]. Silibinin has been used as a traditional drug to treat a range of liver disorders, including hepatitis and cirrhosis. The manifold inhibitory effects of silibinin against various cancer cells include growth inhibition, anti-inflammation, cell cycle regulation, apoptosis induction, chemo-sensitization, inhibition of angiogenesis, reversal of multi-drug resistance, and inhibition of invasion and metastasis [11,12]. Many in vitro and in vivo reports on the activity of silibinin, clearly neglect the structure-activity relationship of the pair of diastereoisomers, using silibinin, the natural mixture of the two flavonolignans SilA and SilB, for all experiments. These studies have mostly disregarded this aspect because of the difficulty separating, on a preparative scale, two diastereoisomers (1a and 1b). In the 2021 a very in-depth study by Kren et al. [13] on the central role of stereochemistry in the pharmacological properties of silybin, highlights how it is necessary to continue studying these flavonolignans, together with the other silymarin flavonolignans, never neglecting their optical purity.
As a part of our continuing research effort on the synthesis of newly modified silibinin [14][15][16], in 2017, starting from silibinin (1ab), dimers 3-3, 3-9 and 9 -9 phosphodiester were obtained ( Figure 1) [8]. Dimers, obtained as mixture of diastereoisomers, were very soluble in water and stable in both human serum and alkaline phosphatase. Despite silibinin (1ab) and silybins (1a and 1b) [17,18] not having strong antioxidant activity, dimers 9 -9 showed a strong radical-scavenging ability. In particular, the ability to scavenge 1 O 2 in H 2 O was tested, and a higher reactivity towards HO • (about two times) was estimated for the 3-9 and 9 -9 dimers with respect to silibinin. Starting from these results, it seemed interesting to investigate the structure-activity relationships of dimers 9 -9 , obtained from diastereoisomerically pure silybin monomers (SilA 1a and SilB 1b, Figure 1).
Herein, we report an improvement of 9 -9 PLSd dimers synthesis and a systematic study on the ability to scavenge HO • radicals as well as their antiproliferative effect on many human tumor cell lines of different histological origins or metastatic potential. Human dermal fibroblasts (HDFs) were used as healthy cells to evaluate the selectivity of action of the examined metabolites towards tumor cells. Furthermore, apoptosis induction was investigated in leukemia cells treated with the examined compounds.

Synthesis of 9 -9 Phosphodiester Silybin Dimers 6
To deepen our research efforts on 9 -9 phosphodiester silybin dimers, we chose to improve the efficiency of the initially proposed synthetic strategy [18][19][20]. For the synthesis of these highly symmetrical dimers, it was advantageous to start from the intermediate 9 -OH of the silybins; therefore, it was necessary to review the previously reported synthetic strategy [8].
In the previous method, the synthesis involves the regioselective protection of different OH functions of silybin with isobutyryl chloride, but it is laborious and not very efficient (yields ≤ 20%) [7]. To develop an efficient synthetic strategy and highlight the effects of stereochemistry on biological activity, we started from two pure diastereoisomers, silybin A and silybin B (1a and 1b), obtained by our own silibinin purification protocol [21].
Initially, we converted silybins (1a and 1b) into their 9 -ODMT ether (Scheme 1) and then applied exhaustive acylation with an excess of isobutyryl chloride in DCM and pyridine. Fully protected silybins were obtained in good yields (65-68% range, see Materials and Methods section). The next treatment with 1% I 2 in MeOH allowed the removal of the DMT protecting group to give 2 (2a or 2b, Scheme 1) in 90% and 88% yields, respectively [22]. Unlike our previously reported observations during formic acid deprotection [19,20,23], side deprotection products were not observed with this procedure. In addition, deprotection was highly reproducible and very efficient. Phosphitylation of building blocks 2a and 2b with 2-cyanoethyl-N,N-diisopropylaminochlorophosphoramidite (3) in anhydrous DCM led to derivatives 4a and 4b in 86% and 80% yields, respectively.
These were then coupled with building blocks 2a and 2b in the presence of 4,5dicyanoimidazole (DCI) in MeCN. For the synthesis of the heterodimer SilA-p-SilB, the best yield was observed by coupling silybin A phosphoramidite 4a and the protected silybin B (2b). After one-pot oxidation of the triester phosphite to phosphate with tButOOH in decane, the phosphotriester dimers were purified and obtained in good yields (5aa 83%, 5bb 80% and 5ab 77%). Treatment with ac. ammonia finally led to complete deprotection, and after RP-HPLC purification, dimers 6aa, 6bb and 6ab were obtained in 77%, 82% and 80% yield, respectively. The structures of dimers 6 were confirmed by 1D and 2D NMR ( 1 H, 13 C, and 31 P) and MS analyses. The spectra of the silybin derivatives 2a and 2b appeared as those of a silybin monomer (see Experimental Section and Supporting Material), whereas the splitting of some signals is observed in the spectra of phosphoramidites 4a and 4b, since they are a pair of diastereoisomers where the phosphorous is a stereocenter and it can be R or S. The same goes for dimers 5aa and 5bb and 5ab, which are also a mixture of diastereoisomers. As expected, for the 6ab dimer, chemical shift splitting of some nuclei was observed in both the 1 H and 13 C spectra.
derivatives 2a and 2b appeared as those of a silybin monomer (see Experimental Section and Supporting Material), whereas the splitting of some signals is observed in the spectra of phosphoramidites 4a and 4b, since they are a pair of diastereoisomers where the phosphorous is a stereocenter and it can be R or S. The same goes for dimers 5aa and 5bb and 5ab, which are also a mixture of diastereoisomers. As expected, for the 6ab dimer, chemical shift splitting of some nuclei was observed in both the 1 H and 13 C spectra. Scheme 1. Synthesis of 9′′-9′′ linked-phosphate silybin dimers 6aa, 6bb and 6ab. Scheme 1. Synthesis of 9 -9 linked-phosphate silybin dimers 6aa, 6bb and 6ab.

Radical Scavenger Activities (HO • )
Biologically, the hydroxyl radical (HO • ) is widely believed to be generated when hydrogen peroxide reacts with Fe(II) (Fenton reaction). The putative HO • ·is an extremely reactive and short-lived species that can damage DNA, proteins, and lipids. However, the Fe(II)/H 2 O 2 mixture has disadvantages in a scavenging assay because many flavonoids as well as flavonolignans are also metal chelators. When the sample is mixed with Fe(II), it may alter the activity of Fe(II) by chelation. As a result, it is impossible to distinguish if the antioxidants are simply good metal chelators or HO • scavengers. In our study the second order rate constants for HO • reactions with silybin dimers (6aa, 6bb and 6ab) have been determined by pulse photolysis method [24] using hydrogen peroxide (H 2 O 2 ) as ROS sources. The reactivity towards HO • was determined to be in the same order of magnitude for dimers 6aa, 6bb and 6ab (Table 1), always remaining significantly greater than dimers 3-3 and 3-9 [8]. This could be explained considering that in dimers 9 -9 the 3, 5 and 4 OH functions, responsible for the radical scavenger activity, are not involved in any bond. In fact, the influence of the individual hydroxy groups of silibinin (1ab) on its antioxidant and radical scavenging properties were studied in detail and the findings led to the conclusion that the 3, 5 and 4 phenolic moieties as well as the 3-OH group, are essential for the compounds radical-scavenging properties [25,26].

In Vitro Antiproliferative Activity of Dimers 6aa, 6ab, and 6bb
The anticancer efficacy of silibinin, which is mainly realized through targeting proliferation, apoptosis, inflammation, angiogenesis, and other cancer-modulating mechanisms, is clearly evident from recently published reports. Despite the high interest in the properties of this flavonolignan, which is also attributable to its non-toxicity, pharmacological studies on the two diastereoisomers, silybins A and B, are still required to develop a structure-activity profile.
In this context, the antiproliferative effect of silibinin compounds was evaluated on many human tumor cell lines of disparate histological origins or different metastatic potential, and human dermal fibroblasts (HDFs) were used as healthy cells to evaluate the selectivity of action of the examined compounds towards tumor cells. A preliminary screening was performed by treating cells with 10 and 50 µM concentrations for 48 h to identify the most sensitive cell lines. The results show that silibinin compounds were active on Jurkat, A375, WM266, and HeLa, but at the tested concentrations, they did not interfere with the growth of PANC, MCF-7, HDF or U87 ( Figure 2). This reveals an interesting tumor cell selectivity, an important feature for the optimization of therapeutic compounds. All molecules showed good activity on Jurkat cells derived from leukemia, even at the lowest used concentration (10 µM), displaying a reduction in proliferation of about 20%, which reached more than 40% when the cells were incubated with the molecules at 50 µM. Interestingly, the dimers appeared to be more active than monomers on the melanoma cell lines used (WM266 and A375) ( Figure 2).
For further studies, dose-response curves were obtained (Figure 3), and the corresponding IC 50 values were calculated on the cells resulted more sensitive to the treatment with the compounds, the leukemia cell line Jurkat. As shown in Table 2, IC 50 values were quite similar for all tested molecules. Nevertheless, the most active compound was found to be SilB, with an IC 50 of 36 µM. Importantly, all molecules were poorly active on HDFs, showing their selectivity of action towards tumor cells. In particular, monomers showed an IC 50 of about 200 µM, and dimers were even more selective with higher IC 50 ( Figure 3). These data are interesting considering that, in similar experiments, silibinin compounds are usually used in concentrations of up to 300 µM [27]. A relevant outcome is the identification of a molecule with low toxicity in healthy cells. As reported in the literature, silibinin exerts its cytotoxic effect by activating the apoptotic pathway [28,29]; therefore, we investigated whether dimers induce apoptosis to the same degree as the related monomers.
an IC50 of about 200 µ M, and dimers were even more selective with higher IC50 (Figure 3). These data are interesting considering that, in similar experiments, silibinin compounds are usually used in concentrations of up to 300 µ M [27]. A relevant outcome is the identification of a molecule with low toxicity in healthy cells. As reported in the literature, silibinin exerts its cytotoxic effect by activating the apoptotic pathway [28,29]; therefore, we investigated whether dimers induce apoptosis to the same degree as the related monomers.   For this purpose, Jurkat was incubated with the molecules at a concentration of 200 µ M, and flow cytometric analysis with annexin V/propidium iodine (PI) double staining was carried out.
The results indicate that 6aa and 6ab could induce apoptosis as effectively as the monomers, although they showed a lower percentage of apoptotic cells; in particular, 20% of cells treated with 6aa were apoptotic (early and advanced) with respect to the control, and 6ab exhibited about 25% (early and advanced) apoptotic cells, whereas 70% of cells treated with 1a or 1b were apoptotic (Figure 4). In contrast, 6bb was not able to induce apoptosis: only 4% of treated cells were apoptotic compared with the control (Figure 4). These results indicate not only that the dimers probably have a different mechanism of action from the monomers but also that the behaviour differs among the dimers, underlining the importance of the stereochemistry of the molecules and suggesting that it could affect the activity of the silybin compounds by means of specific and selective interactions with protein partners.

General Methods and Materials
All chemicals were purchased from Sigma-Aldrich (Milano, Italy). HPLC-grade MeCN and MeOH were purchased from Carlo Erba Reagents and Sigma-Aldrich, respectively. Reactions were monitored by TLC (F254 precoated silica gel plates, Merck) and column chromatography (Merck Kieselgel 60, 70-230 mesh, Milano, Italy). HPLC analysis of dimers 6aa, 6bb and 6ab, was performed with a Shimadzu LC-8A PLC system (Shimadzu Analytical and Measuring Instruments, Milano, Italy) equipped with a Shimadzu SCL-10A VP System control and a Shimadzu SPD-10A VP UV-Vis detector.  For this purpose, Jurkat was incubated with the molecules at a concentration of 200 µM, and flow cytometric analysis with annexin V/propidium iodine (PI) double staining was carried out.
The results indicate that 6aa and 6ab could induce apoptosis as effectively as the monomers, although they showed a lower percentage of apoptotic cells; in particular, 20% of cells treated with 6aa were apoptotic (early and advanced) with respect to the control, and 6ab exhibited about 25% (early and advanced) apoptotic cells, whereas 70% of cells treated with 1a or 1b were apoptotic ( Figure 4). In contrast, 6bb was not able to induce apoptosis: only 4% of treated cells were apoptotic compared with the control (Figure 4). These results indicate not only that the dimers probably have a different mechanism of action from the monomers but also that the behaviour differs among the dimers, underlining the importance of the stereochemistry of the molecules and suggesting that it could affect the activity of the silybin compounds by means of specific and selective interactions with protein partners. signment. The 1 H signals were assigned by using 1 H/ 1 H COSY, 1 H/ 13 C HSQC, and 1 H/ 13 C HMBC. NMR data were processed using Bruker Topspin 3.6.1 software. The proton-detected heteronuclear correlations were measured using a gradient heteronuclear single-quantum coherence (HSQC) experiment, optimized for 1 JHC = 155 Hz, and a gradient heteronuclear multiple bond coherence (HMBC) experiment, optimized for n JHC = 8 Hz. Silybin A and silybin B were obtained by HPLC purification of silibinin purchased from Sigma-Aldrich (S0417) as reported by us [21]. The experimental procedures to the synthesis of building blocks 2 and 4, are described in detail only for the stereoisomers of silybin A: the same reaction conditions (temperature, stoichiometric ratios, time of reaction) were used for silybin B.

Synthesis of 3,5,7,4′′-O-tetra-isobutyryl-silybin 2a (or 2b)
Silybin A (1a, 730 mg, 1.51 mmol), previously co-evaporated several times with anhydrous THF and dissolved in anhydrous pyridine (4 mL), was reacted with DMTCl (666 mg, 1.96 mmol). The reaction mixture, left at room temperature for 2 h under stirring, was then diluted with MeOH and concentrated under reduced pressure. The crude was then diluted with DCM, transferred into a separatory funnel, washed once with a saturated NaHCO3 aqueous solution, and then once with H2O. The organic phase, dried over

General Methods and Materials
All chemicals were purchased from Sigma-Aldrich (Milano, Italy). HPLC-grade MeCN and MeOH were purchased from Carlo Erba Reagents and Sigma-Aldrich, respectively. Reactions were monitored by TLC (F254 precoated silica gel plates, Merck) and column chromatography (Merck Kieselgel 60, 70-230 mesh, Milano, Italy). HPLC analysis of dimers 6aa, 6bb and 6ab, was performed with a Shimadzu LC-8A PLC system (Shimadzu Analytical and Measuring Instruments, Milano, Italy) equipped with a Shimadzu SCL-10A VP System control and a Shimadzu SPD-10A VP UV-Vis detector. Mass spectrometric analyses were performed on AB SCIEX TOF/TOF 5800 in positive or negative mode and Waters Micromass ZQ Instrument (Waters, Milano, Italy) equipped with an electrospray source in positive mode. The NMR spectra were recorded at 25 • C on an NMR spectrometer Bruker DRX, Bruker Advance (Bruker Italia Srl, Milano, Italy) and INOVA-500 NMR instrument (Varian, Milan, Italy), referenced in ppm to residual solvent signals (CDCl 3 , at δ H 7.27, δ C 77.0; CD 3 OD, at δ H 3.31, δ C 49.0 and DMSO-d 6 , δ H 2.50, δ C 39.5. 31 P NMR spectra were recorded using D 3 PO 4 (85 wt. % in D 2 O, 98 atoms %D) as an external standard, referenced to residual solvent signals (δ P 0.0 ppm). Data for 1 H NMR are reported as follows: chemical shift (ppm), multiplicity (s = singlet, br = broad, d = doublet, t = triplet, and m = multiplet), coupling constant (Hz), integration, and assignment. The 1 H signals were assigned by using 1 H/ 1 H COSY, 1 H/ 13 C HSQC, and 1 H/ 13 C HMBC. NMR data were processed using Bruker Topspin 3.6.1 software. The proton-detected heteronuclear correlations were measured using a gradient heteronuclear single-quantum coherence (HSQC) experiment, optimized for 1 J HC = 155 Hz, and a gradient heteronuclear multiple bond coherence (HMBC) experiment, optimized for n J HC = 8 Hz.
Silybin A and silybin B were obtained by HPLC purification of silibinin purchased from Sigma-Aldrich (S0417) as reported by us [21]. The experimental procedures to the synthesis of building blocks 2 and 4, are described in detail only for the stereoisomers of silybin A: the same reaction conditions (temperature, stoichiometric ratios, time of reaction) were used for silybin B.

Synthesis of 3,5,7,4 -O-tetra-isobutyryl-silybin 2a (or 2b)
Silybin A (1a, 730 mg, 1.51 mmol), previously co-evaporated several times with anhydrous THF and dissolved in anhydrous pyridine (4 mL), was reacted with DMTCl (666 mg, 1.96 mmol). The reaction mixture, left at room temperature for 2 h under stirring, was then diluted with MeOH and concentrated under reduced pressure. The crude was then diluted with DCM, transferred into a separatory funnel, washed once with a saturated NaHCO 3 aqueous solution, and then once with H 2 O. The organic phase, dried over anhydrous Na 2 SO 4 , was filtered, and then concentrated under reduced pressure. The crude was next purified on a silica gel column, eluting with DCM/MeOH (98:2, v/v) in the presence of 1% of pyridine, affording pure 9 -O-(4,4 -dimethoxytriphenylmethyl)-silybin A as a pale amorphous solid (1.14 g, 1.44 mmol) in a 96% yield.
In total, 1.14 g (1.44 mmol) of product dissolved in anhydrous DCM (20 mL), adding Et 3 N (841 µL, 6.05 mmol) and pyridine (1.2 mL, 14.4 mmol), was reacted with isobutyryl chloride (638 µL, 6.05 mmol). The mixture was left under stirring at 0 • C for 15 min and then diluted with MeOH and concentrated under reduced pressure. The crude was then diluted with DCM, transferred into a separatory funnel, washed one time with a saturated NaHCO 3 aqueous solution, and then once with H 2 O. The organic phase, dried over anhydrous Na 2 SO 4 , was filtered, and then concentrated under reduced pressure. The crude was next purified on a silica gel column, eluting with n-hexane/EtOAc (7:3, v/v) in the presence of 1% of pyridine furnishing pure 3,5,7,4 -O-tetra-isobutyryl-9 -O-(4,4dimethoxytriphenylmethyl)-silybin A as pale amorphous solid (1.06 g, 1.0 mmol) in a 70% yield.
In a solution, 0.1 M of product (1.06 g, 1.0 mmol) in MeOH/DCM (6:1 v/v) was added 1% (p/v) of I 2 (100 mg). The solution was left under stirring at room temperature for 1 h and then added Na 2 S 2 O 3 and concentrated under reduced pressure. The crude was then diluted with DCM, transferred into a separatory funnel, washed one time with a saturated NaHCO 3 aqueous solution, and then once with H 2 O. The organic phase, dried over anhydrous Na 2 SO 4 , was filtered, and then concentrated under reduced pressure. The crude was next purified on a silica gel column, eluting with n-hexane/EtOAc
The results are presented as the percentage of proliferating cells respect to the control (vehicle treated cells) and are expressed as means ± SE of, at least, three independent experiments performed in triplicate. The statistical analysis was performed using Student's t-test, unpaired, two-sided, p < 0.05 was considered significant. The IC 50 values were calculated by GraphPad Prism software.

Apoptosis Assay
The apoptosis analysis was performed on Jurkat cells seeded at 2.5 × 10 5 cells/mL in a 6-well plate. The cells were incubated in the absence or presence of 200 µM concentration of examined compounds at 37 • C and apoptosis induction was analysed after 48 h by double staining with annexin V/FITC and propidium iodide (PI) (eBioscience, Affimetrix Santa Clara, CA, USA) [33]. The cells undergoing apoptosis were quantified using a flow cytometer equipped with a 488 nm argon laser (Becton Dickinson, Franklin Lakes, NJ, USA) by Cell Quest software. All FACS analyses were performed at least 2 times.

Conclusions
In this work, we reported the synthesis of the optically pure phosphodiester dimers of silybins following a very efficient synthetic strategy using an orthogonal protection of the different OH groups. Starting from the two pure diastereoisomers and exploiting the well-known phosphoramidite chemistry, the new 9 -9 dimers of silybin A and silybin B (6aa, 6ab and 6bb) were obtained in pure form and in good yields.
Their ability to scavenge reactive oxygen species (ROS) such as hydroxyl radical (HO • ) highlights the high activity of all three dimers, comparable to that reported for a known potent antioxidant as quercetin. Although they are diastereomers, 6aa, 6ab and 6bb show very similar radical scavenger activity.
To disclosing a structure-activity relationship, dimers (6aa, 6ab and 6bb), as well as the silybin A (1a) and silybin B (1b), a preliminary screening was performed by treating cells with 10 and 50 µM concentrations for 48 h. The results indicate that both monomers and dimers present selective anti-proliferative activity towards leukemia cells at the concentrations used, in particular, all silybin compounds showing a similar IC 50 in the midmicromolar range and are poorly active on normal cells. However, the mechanism by which the different silybin compounds induce their cytotoxic activity appears to be different: all the cells treated with the monomers go completely into apoptosis, whereas only part of the cells treated with 6aa and 6ab were found to be in apoptosis. Contrarily, when the cells were treated with 6bb dimer, no significant number of cells in the apoptotic stage was observed. These results demonstrate the crucial role of the stereochemistry for these flavonolignans, in the activation of the apoptotic mechanism and opens up a new window to deeply investigate the interaction of such compounds with proteins involved in cancer metabolic pathways.
Supplementary Materials: The following supporting information can be downloaded: HPLC profile of dimers 6aa, 6bb and 6ab in Figures S1-S3; 1 H, 13