Solvent-Free Addition of Indole to Aldehydes: Unexpected Synthesis of Novel 1-[1-(1H-Indol-3-yl) Alkyl]-1H-Indoles and Preliminary Evaluation of Their Cytotoxicity in Hepatocarcinoma Cells

New 1-[1-(1H-indol-3-yl) alkyl]-1H-indoles, surprisingly, have been obtained from the addition of indole to a variety of aldehydes under neat conditions. CaO, present in excess, was fundamental for carrying out the reaction with paraformaldehyde. Under the same reaction conditions, aromatic and heteroaromatic aldehydes afforded only classical bis (indolyl) aryl indoles. In this paper, the role of CaO, together with the regiochemistry and the mechanism of the reaction, are discussed in detail. The effect of some selected 3,3′- and 1,3′-diindolyl methane derivatives on cell proliferation of the hepatoma cell line FaO was also evaluated.


Introduction
Indole is one of the most versatile heterocyclic nuclei, identified as a pharmacophore in a large number of natural and synthetic biologically active molecules [1]. Due to its electron-rich character, indole promptly reacts with hard and soft electrophiles; thus, it is fair to consider it a privileged structure, quaintly referred to as the 'lord of the rings' [2].
Interestingly, several natural [24] and synthetic 3,3′-BIMs have shown important anti-cancer properties and demonstrated the capacity to sensitize cancer cells to apoptosis by signaling various proapoptotic genes and proteins [25][26][27]. In this regard, we present a preliminarily evaluation of the effect of these novel 3,3′-and 1,3′-BIM derivatives on the growth of the hepatoma cell line FaO.

Results and Discussion
In the quest to develop a simple and eco-friendly protocol to 3,3′-BIM derivatives, we unexpectedly observed, from the very beginning of our study, that, when the reaction was carried out with formaldehyde or aliphatic aldehydes, both 3,3′-and 1,3′-bisindolyl methane (3,3′-and 1,3′-BIM) derivatives were obtained. Conversely, aromatic aldehydes gave only classical 3,3′-BIMs. To understand these unusual results, the reaction between indole 1, and formaldehyde 2a, derived from paraformaldehyde, was first investigated (Scheme 2).

Scheme 2. Reaction between indole 1 and formaldehyde 2a.
Surprisingly, apart from the expected 3,3′-diindolyl methane 3a, we observed, as shown in Scheme 2, the formation of indole-1-carbinol 5 [28] and 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a, which was fully characterized via mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy experiments (spectra available in Supplementary Materials). The best results were achieved at 100 °C, and also observed an improvement in the yields when an excess of calcium oxide (CaO) was used. The reaction did not take place at all at room temperature, both with or without CaO. Similarly, when the operative temperature was fixed at 60 °C, traces of both isomers were detectable. Thus, it was speculated that CaO, which has recently been shown to catalyse some Mannich reactions for the preparation of lariat ethers [29], might affect our experiments. Hence, the role of CaO was explored, noticing that, when it was used in stoichiometric amounts or in large excess, the reaction seemed to proceed smoothly. Use of CaO in a catalytic amount (10 mol %), did not seem to produce any effect. Reasonably, the temperature and CaO might have a synergic effect in the reaction activation. We can plausibly assume that, in these conditions, the paraformaldehyde is rapidly converted to free formaldehyde and the traces of formic acid produced during prolonged heating are rapidly neutralized by CaO. Following the experiment by means of gas chromatography-mass spectrometry (GC-MS) analysis (Figure 1), we also observed, from the very beginning of the reaction and only in the presence of CaO, the formation of the indole-1-carbinol 5, which is usually obtained in a strong basic condition [30,31]. Surprisingly, apart from the expected 3,3 -diindolyl methane 3a, we observed, as shown in Scheme 2, the formation of indole-1-carbinol 5 [28] and 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a, which was fully characterized via mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy experiments (spectra available in Supplementary Materials). The best results were achieved at 100 • C, and also observed an improvement in the yields when an excess of calcium oxide (CaO) was used. The reaction did not take place at all at room temperature, both with or without CaO. Similarly, when the operative temperature was fixed at 60 • C, traces of both isomers were detectable. Thus, it was speculated that CaO, which has recently been shown to catalyse some Mannich reactions for the preparation of lariat ethers [29], might affect our experiments. Hence, the role of CaO was explored, noticing that, when it was used in stoichiometric amounts or in large excess, the reaction seemed to proceed smoothly. Use of CaO in a catalytic amount (10 mol %), did not seem to produce any effect. Reasonably, the temperature and CaO might have a synergic effect in the reaction activation. We can plausibly assume that, in these conditions, the paraformaldehyde is rapidly converted to free formaldehyde and the traces of formic acid produced during prolonged heating are rapidly neutralized by CaO. Following the experiment by means of gas chromatography-mass spectrometry (GC-MS) analysis (Figure 1), we also observed, from the very beginning of the reaction and only in the presence of CaO, the formation of the indole-1-carbinol 5, which is usually obtained in a strong basic condition [30,31]. We postulate that, in our experimental conditions, CaO induces a partial N-H proton abstraction and, despite the modest ionic character of the N-Ca bond, formaldehyde is so reactive as to undergo a nucleophilic attack, generating indole-1-carbinol 5 [28]. As a matter of fact, when DMSO was used as polar aprotic solvent, the reaction afforded 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a as a major product in a high regioselective manner.
In addition, an interesting yield improvement, especially for isomer 4a, was noticed when KOH was used instead of CaO. A possible explanation, apart from the stronger electropositivity of K + , might be that indole-1-carbinol 5, when heated in the presence of KOH, decomposes to regenerate indole and HCHO [30] that react again to afford both 3,3′-and 1,3′-BIM isomers (Table 1). To gain a better comprehension of the reaction mechanism, we examined the possible role of compound 5. It did not demonstrate being a real intermediate in the formation of 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a. In fact, when 5 could react with indole 1 and CaO, only traces of the two isomers were detected (Scheme 3a). Interestingly, the classical bis (indolyl) methane 3a, could derive from a thermal-induced isomerisation of 4a (Scheme 3b) [20].  We postulate that, in our experimental conditions, CaO induces a partial N-H proton abstraction and, despite the modest ionic character of the N-Ca bond, formaldehyde is so reactive as to undergo a nucleophilic attack, generating indole-1-carbinol 5 [28]. As a matter of fact, when DMSO was used as polar aprotic solvent, the reaction afforded 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a as a major product in a high regioselective manner.
In addition, an interesting yield improvement, especially for isomer 4a, was noticed when KOH was used instead of CaO. A possible explanation, apart from the stronger electropositivity of K + , might be that indole-1-carbinol 5, when heated in the presence of KOH, decomposes to regenerate indole and HCHO [30] that react again to afford both 3,3 -and 1,3 -BIM isomers (Table 1). To gain a better comprehension of the reaction mechanism, we examined the possible role of compound 5. It did not demonstrate being a real intermediate in the formation of 1-[1-(1H-indol-3-yl)methyl]-1H-indole 4a. In fact, when 5 could react with indole 1 and CaO, only traces of the two isomers were detected (Scheme 3a). Interestingly, the classical bis (indolyl) methane 3a, could derive from a thermal-induced isomerisation of 4a (Scheme 3b) [20]. We postulate that, in our experimental conditions, CaO induces a partial N-H proton abstraction and, despite the modest ionic character of the N-Ca bond, formaldehyde is so reactive as to undergo a nucleophilic attack, generating indole-1-carbinol 5 [28]. As a matter of fact, when DMSO was used as polar aprotic solvent, the reaction afforded 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a as a major product in a high regioselective manner.
In addition, an interesting yield improvement, especially for isomer 4a, was noticed when KOH was used instead of CaO. A possible explanation, apart from the stronger electropositivity of K + , might be that indole-1-carbinol 5, when heated in the presence of KOH, decomposes to regenerate indole and HCHO [30] that react again to afford both 3,3′-and 1,3′-BIM isomers (Table 1). To gain a better comprehension of the reaction mechanism, we examined the possible role of compound 5. It did not demonstrate being a real intermediate in the formation of 1-[1-(1H-indol-3-yl) methyl]-1H-indole 4a. In fact, when 5 could react with indole 1 and CaO, only traces of the two isomers were detected (Scheme 3a). Interestingly, the classical bis (indolyl) methane 3a, could derive from a thermal-induced isomerisation of 4a (Scheme 3b) [20]. With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3 -diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2). With the perception that the reaction, if successful with formaldehyde, would allow similar 1,3′-diindolyl isomers when different carbonyl substrates were used, some other exploratory reactions with diverse ketones and aldehydes were carried out ( Table 2).  It was immediately evident that the reaction was highly chemoselective for aldehydes; in fact, ketones such as 2-octanone, 2-hexanone, 4′-methylacetophenone and 1-(p-methoxyphenyl)-2-propanone did not react at all. Besides, only aliphatic aldehydes afforded both 3,3′-and 1,3′-isomers, while aromatic and heteroaromatic ones produced only bis (indolyl) aryl methanes. Interestingly, the reactions efficiently proceeded under neat conditions [35] and the yields seemed to be positively affected by the absence of CaO, which demonstrated its importance mainly for the paraformaldehyde activation. Remarkably, the reactions almost did not occur or took place very slowly, with or without CaO, when a solvent (toluene, CH3CN, CHCl3, THF, DMF) was employed [35].
Mechanistically, we presume (Scheme 4) that two reaction pathways can be conceived for the formation of 3,3′-and 1,3′-BIMs isomers. Both routes share the same well-known 2-azafulvene intermediate [20][21][22][23]36,37] which may add to nucleophilic species. Specifically, if the path a could be merely assumed as a Friedel-Crafts alkylation, the path b is presumably a Mannich-type N-aminoalkylation. In this respect, the different reactivity of aliphatic and aromatic aldehydes seems plausible given their intrinsically diverse steric and electronic features, and associated with the modest nucleophilicity of the indole nitrogen. It was immediately evident that the reaction was highly chemoselective for aldehydes; in fact, ketones such as 2-octanone, 2-hexanone, 4′-methylacetophenone and 1-(p-methoxyphenyl)-2-propanone did not react at all. Besides, only aliphatic aldehydes afforded both 3,3′-and 1,3′-isomers, while aromatic and heteroaromatic ones produced only bis (indolyl) aryl methanes. Interestingly, the reactions efficiently proceeded under neat conditions [35] and the yields seemed to be positively affected by the absence of CaO, which demonstrated its importance mainly for the paraformaldehyde activation. Remarkably, the reactions almost did not occur or took place very slowly, with or without CaO, when a solvent (toluene, CH3CN, CHCl3, THF, DMF) was employed [35].
Mechanistically, we presume (Scheme 4) that two reaction pathways can be conceived for the formation of 3,3′-and 1,3′-BIMs isomers. Both routes share the same well-known 2-azafulvene intermediate [20][21][22][23]36,37] which may add to nucleophilic species. Specifically, if the path a could be merely assumed as a Friedel-Crafts alkylation, the path b is presumably a Mannich-type N-aminoalkylation. In this respect, the different reactivity of aliphatic and aromatic aldehydes seems plausible given their intrinsically diverse steric and electronic features, and associated with the modest nucleophilicity of the indole nitrogen. It was immediately evident that the reaction was highly chemoselective for aldehydes; in fact, ketones such as 2-octanone, 2-hexanone, 4′-methylacetophenone and 1-(p-methoxyphenyl)-2-propanone did not react at all. Besides, only aliphatic aldehydes afforded both 3,3′-and 1,3′-isomers, while aromatic and heteroaromatic ones produced only bis (indolyl) aryl methanes. Interestingly, the reactions efficiently proceeded under neat conditions [35] and the yields seemed to be positively affected by the absence of CaO, which demonstrated its importance mainly for the paraformaldehyde activation. Remarkably, the reactions almost did not occur or took place very slowly, with or without CaO, when a solvent (toluene, CH3CN, CHCl3, THF, DMF) was employed [35].
Mechanistically, we presume (Scheme 4) that two reaction pathways can be conceived for the formation of 3,3′-and 1,3′-BIMs isomers. Both routes share the same well-known 2-azafulvene intermediate [20][21][22][23]36,37] which may add to nucleophilic species. Specifically, if the path a could be merely assumed as a Friedel-Crafts alkylation, the path b is presumably a Mannich-type N-aminoalkylation. In this respect, the different reactivity of aliphatic and aromatic aldehydes seems plausible given their intrinsically diverse steric and electronic features, and associated with the modest nucleophilicity of the indole nitrogen.  It was immediately evident that the reaction was highly chemoselective for aldehydes; in fact, ketones such as 2-octanone, 2-hexanone, 4′-methylacetophenone and 1-(p-methoxyphenyl)-2-propanone did not react at all. Besides, only aliphatic aldehydes afforded both 3,3′-and 1,3′-isomers, while aromatic and heteroaromatic ones produced only bis (indolyl) aryl methanes. Interestingly, the reactions efficiently proceeded under neat conditions [35] and the yields seemed to be positively affected by the absence of CaO, which demonstrated its importance mainly for the paraformaldehyde activation. Remarkably, the reactions almost did not occur or took place very slowly, with or without CaO, when a solvent (toluene, CH3CN, CHCl3, THF, DMF) was employed [35].
Mechanistically, we presume (Scheme 4) that two reaction pathways can be conceived for the formation of 3,3′-and 1,3′-BIMs isomers. Both routes share the same well-known 2-azafulvene intermediate [20][21][22][23]36,37] which may add to nucleophilic species. Specifically, if the path a could be merely assumed as a Friedel-Crafts alkylation, the path b is presumably a Mannich-type N-aminoalkylation. In this respect, the different reactivity of aliphatic and aromatic aldehydes seems plausible given their intrinsically diverse steric and electronic features, and associated with the modest nucleophilicity of the indole nitrogen. It was immediately evident that the reaction was highly chemoselective for aldehydes; in fact, ketones such as 2-octanone, 2-hexanone, 4 -methylacetophenone and 1-(p-methoxyphenyl)-2propanone did not react at all. Besides, only aliphatic aldehydes afforded both 3,3 -and 1,3 -isomers, while aromatic and heteroaromatic ones produced only bis (indolyl) aryl methanes. Interestingly, the reactions efficiently proceeded under neat conditions [35] and the yields seemed to be positively affected by the absence of CaO, which demonstrated its importance mainly for the paraformaldehyde activation. Remarkably, the reactions almost did not occur or took place very slowly, with or without CaO, when a solvent (toluene, CH 3 CN, CHCl 3 , THF, DMF) was employed [35].
Mechanistically, we presume (Scheme 4) that two reaction pathways can be conceived for the formation of 3,3 -and 1,3 -BIMs isomers. Both routes share the same well-known 2-azafulvene intermediate [20][21][22][23]36,37] which may add to nucleophilic species. Specifically, if the path a could be merely assumed as a Friedel-Crafts alkylation, the path b is presumably a Mannich-type N-aminoalkylation. In this respect, the different reactivity of aliphatic and aromatic aldehydes seems plausible given their intrinsically diverse steric and electronic features, and associated with the modest nucleophilicity of the indole nitrogen. It was immediately evident that the reaction was highly chemoselective for aldehydes; in fact, ketones such as 2-octanone, 2-hexanone, 4′-methylacetophenone and 1-(p-methoxyphenyl)-2-propanone did not react at all. Besides, only aliphatic aldehydes afforded both 3,3′-and 1,3′-isomers, while aromatic and heteroaromatic ones produced only bis (indolyl) aryl methanes. Interestingly, the reactions efficiently proceeded under neat conditions [35] and the yields seemed to be positively affected by the absence of CaO, which demonstrated its importance mainly for the paraformaldehyde activation. Remarkably, the reactions almost did not occur or took place very slowly, with or without CaO, when a solvent (toluene, CH3CN, CHCl3, THF, DMF) was employed [35].
Mechanistically, we presume (Scheme 4) that two reaction pathways can be conceived for the formation of 3,3′-and 1,3′-BIMs isomers. Both routes share the same well-known 2-azafulvene intermediate [20][21][22][23]36,37] which may add to nucleophilic species. Specifically, if the path a could be merely assumed as a Friedel-Crafts alkylation, the path b is presumably a Mannich-type N-aminoalkylation. In this respect, the different reactivity of aliphatic and aromatic aldehydes seems plausible given their intrinsically diverse steric and electronic features, and associated with the modest nucleophilicity of the indole nitrogen. Electron-impact (EI) mass spectra allowed us to characterize and differentiate both isomers easily ( Figure 2). diindolyl isomers when different carbonyl substrates were used, some other exploratory tions with diverse ketones and aldehydes were carried out ( Table 2).  In fact, in the mass spectra of all 1,3 -diindolyl alkane isomers, the formation of the [M-116] + ion, presumably due to N-C bond cleavage, is the most favored fragmentation process (RA % 100), while in the case of 3,3 -isomers, the formation of [M-116] + ion (RA % 5), related to the rupture of the more stable C-C bond, is suppressed in favor of m/z 245 ion (RA % 100), generated by the radical loss of R substituent. MS/MS experiments are now in progress to investigate the most characteristic fragmentation pathways. NMR analysis included 1 H, 13 C, COSY, gHSQC, gHMQC and ROESY (spectra available in Supplementary Materials), and confirmed the structure of 1,3 -diindolyl alkane isomers.
Recent studies have reported on the pleiotropic protective properties on the chronic liver injuries steatohepatitis [38] and hepatocarcinoma [39,40], and on the multiple anti-tumour activities, including the apoptotic, anti-proliferative and anti-angiogenetic effects [24], of 3,3 -bisindolylmethane, the parent compound of 3,3 -BIMs and one of the most abundant dietary compounds derived from Brassica-genus vegetables. Hence, we decided to evaluate the effect of our novel 1,3 and 3,3 -BIMs on cell proliferation of the rat hepatoma cell line FaO by comparing their effects to those induced by 3,3 -bisindolylmethane 3a and indole-3-carbinol (I3C), the natural active precursor of 3,3 -bisindolylmethane. In experiments, carried out on a selection of compounds, FaO cells were treated with increasing concentrations of 3b, 4a and 4b as well as 400 µM of I3C for 24 h. As shown in Figure 3, the hepatoma cells were highly susceptible to the anti-proliferative effect of these BIMs. In particular, compounds 4a, 3b and 4b exhibited a concentration-dependent growth inhibitory effect in FaO cells similar to that observed after treatment with the well-characterized BIM 3a [38,39].
Electron-impact (EI) mass spectra allowed us to characterize and differentiate both isomers easily ( Figure 2). In fact, in the mass spectra of all 1,3′-diindolyl alkane isomers, the formation of the [M-116]+ ion, presumably due to N-C bond cleavage, is the most favored fragmentation process (RA % 100), while in the case of 3,3′-isomers, the formation of [M-116] + ion (RA % 5), related to the rupture of the more stable C-C bond, is suppressed in favor of m/z 245 ion (RA % 100), generated by the radical loss of R substituent. MS/MS experiments are now in progress to investigate the most characteristic fragmentation pathways. NMR analysis included 1 H, 13 C, COSY, gHSQC, gHMQC and ROESY (spectra available in Supplementary Materials), and confirmed the structure of 1,3′-diindolyl alkane isomers.
Recent studies have reported on the pleiotropic protective properties on the chronic liver injuries steatohepatitis [38] and hepatocarcinoma [39,40], and on the multiple anti-tumour activities, including the apoptotic, anti-proliferative and anti-angiogenetic effects [24], of 3,3′-bisindolylmethane, the parent compound of 3,3′-BIMs and one of the most abundant dietary compounds derived from Brassica-genus vegetables. Hence, we decided to evaluate the effect of our novel 1,3′ and 3,3′-BIMs on cell proliferation of the rat hepatoma cell line FaO by comparing their effects to those induced by 3,3′-bisindolylmethane 3a and indole-3-carbinol (I3C), the natural active precursor of 3,3′-bisindolylmethane. In experiments, carried out on a selection of compounds, FaO cells were treated with increasing concentrations of 3b, 4a and 4b as well as 400 µM of I3C for 24 h. As shown in Figure 3, the hepatoma cells were highly susceptible to the anti-proliferative effect of these BIMs. In particular, compounds 4a, 3b and 4b exhibited a concentration-dependent growth inhibitory effect in FaO cells similar to that observed after treatment with the well-characterized BIM 3a [38,39]. More importantly, our novel derivatives, as well as 3a, were 20-to 4-fold more potent than I3C in suppressing the viability of FaO cells with an IC50 value ranging from 50-100 µM after treatment with 4a and from 20-100 µM and 25-100 µM after treatment with 3b and 4b, respectively. It was noteworthy that all the tested BIMs induced a significant inhibition of growth of hepatoma cells at More importantly, our novel derivatives, as well as 3a, were 20-to 4-fold more potent than I3C in suppressing the viability of FaO cells with an IC 50 value ranging from 50-100 µM after treatment with 4a and from 20-100 µM and 25-100 µM after treatment with 3b and 4b, respectively. It was noteworthy that all the tested BIMs induced a significant inhibition of growth of hepatoma cells at lower concentrations than I3C. Furthermore, 3b was more effective in inducing loss of viability at very low doses.
Glutamax I, Invitrogen S.r.l., Milano, Italy) supplemented with penicillin, streptomycin and 10% heat-inactivated fetal calf-serum (FCS) (Invitrogen) in a humidified atmosphere of 5% CO 2 /95% air, at 37 • C. Indole 3-carbinol (I3C) and compound 3a, 3b, 4a and 4b were dissolved in DMSO and were added to the culture media to the final concentrations specified in the text. Control cells were treated with an equivalent amount of the solvent alone.

Cell Viability
Cell viability was determined by the uptake of neutral red by lysosome of viable cells. Determination of viability of adherent cells by NRU assay was performed according to Borefreund and Puerner [40]. The value obtained for treated cells was expressed as percentage of the value obtained in control cells. All experiments were performed in triplicate.

Statistical Analysis
Instant software (GraphpAd Prism 5, GraphPad Software Inc., San Diego, CA, USA) was used to analyse data. One-way analysis of variance (ANOVA) with post hoc analysis using Tukey's multiple comparison test was used for parametric data. The results of multiple observations were presented as the means ± S.D. of three experiments. A p value of <0.05 was considered statistically significant.