Dihydroxyquingdainone Induces Apoptosis in Leukaemia and Lymphoma Cells via the Mitochondrial Pathway in a Bcl-2- and Caspase-3-Dependent Manner and Overcomes Resistance to Cytostatic Drugs In Vitro

Isatis tinctoria and its indigo dyes have already provided highly active anti-leukaemic lead compounds, with the focus mainly being on indirubin, whereas indigo itself is inactive. There are many more indigoids to find in this plant extract, for example, quingdainone, an indigoid derived from tryptanthrin. We present here a new synthesis of hitherto neglected substituted quingdainones, which is very necessary due to their poor solubility behaviour, and a structure-dependent anti-leukaemic activity study of a number of compounds. Substituted α-phenylaminoacrylic acid was synthesised by hydrogen sulfide extrusion from an analogue mercaptoacetic acid, available from the condensation of rhodanin and a substituted tryptanthrin. It is shown that just improving water solubility does not increase anti-leukaemic activity, since a quingdainone carboxylic acid is inactive compared to dihydroxyquingdainone. The most effective compound, dihydroxyquingdainone with an AC50 of 7.5 µmole, is further characterised, revealing its ability to overcome multidrug resistance in leukaemia cells (Nalm-6/BeKa) with p-glycoprotein expression.


Introduction
Extracts from the natural dye and medicinal plants Isatis tinctoria, Strobilanthes cusia and other plants yielding the indigo dye have long been known as anti-inflammatory agents in traditional Chinese medicine (TCM). They can be applied as anti-leukaemic agents, as proven by Chinese small-scale clinical trials in the nineteen-eighties, whereas in Europe, any phytopharmaceutical use of woad has been forgotten, despite its centuries-long use here. The anti-leukaemic activity was traced to indirubin 1, an isomer of bisindoleheterocyclic structure. Some of its synthetic derivatives were recognised as potent cell cycle inhibitors [1] with efficacy far beyond that of indirubin itself. It has been found that the indirubin-5-sulfonate 1a fits perfectly in the ATP niche of CDK-1 and thereby blocks its activity [2]. All these compounds are formed from "indoxyl" precursors such as isatane and various ketones present in the plants during post-harvest treatment in the moist state or by extraction with aqueous fluids. In this manner, the presence of indigo

Chemistry
The accessibility of indoxyl derivates has always been a problem in the preparation of indigoids. The method reported within this paper provides a way to introduce different substituents based on an α-phenylaminoacrylic acid methodology. The indoxyl part is built up by using ylidenoxindoles in our case, from tryptanthrins 4, by means of the sequential introduction of anilines followed by ring-closing. This enables one to have a wide variety of different substituents. The presented syntheses have moderate to good yields and need no or only one column purification at the end. It was planned to convert obtained tryptanthrins 4 in a first step to a condensation product 7 with 2-thioxo-4-thiazolidinone (rhodanin) (see Scheme 1), followed by hydrolytic ring opening to a mercaptoacetic acid 8. The next sequence was the substitution of the mercapto group with various anilines to yield cyclisable molecules 9 to the final quingdainones 2a-f. The first step already proved to be comparably difficult. This condensation could possibly be completed in a few minutes with isatins 5, as even boiling for a long time is not sufficient for a reaction with the 6-keto group in tryptanthrins. Only a solvent with a large amount of Lewis acid (BF3) in acetic acid promoted the condensation sufficiently to yield 7. Substituents at position 4 of isatins prolong the condensation, and a substituent at position 7 in our case of tryptanthrins renders the reaction impossible. The bioactivity of 2 has been explored insufficiently and has its origin in the very poor solubility of this compound. Quingdainone has been found to possess anti-cancer activity [7]. On the other hand, one wonders whether the tumor-inhibitory activity with unsubstituted indigoids in general could be observed at all, since the solubility of this compound is nearly zero in water. We therefore synthesised quingdainones with improved solubilities and screened these compounds for anti-leukaemic properties.

Chemistry
The accessibility of indoxyl derivates has always been a problem in the preparation of indigoids. The method reported within this paper provides a way to introduce different substituents based on an α-phenylaminoacrylic acid methodology. The indoxyl part is built up by using ylidenoxindoles in our case, from tryptanthrins 4, by means of the sequential introduction of anilines followed by ring-closing. This enables one to have a wide variety of different substituents. The presented syntheses have moderate to good yields and need no or only one column purification at the end. It was planned to convert obtained tryptanthrins 4 in a first step to a condensation product 7 with 2-thioxo-4-thiazolidinone (rhodanin) (see Scheme 1), followed by hydrolytic ring opening to a mercaptoacetic acid 8. The next sequence was the substitution of the mercapto group with various anilines to yield cyclisable molecules 9 to the final quingdainones 2a-f. The first step already proved to be comparably difficult. This condensation could possibly be completed in a few minutes with isatins 5, as even boiling for a long time is not sufficient for a reaction with the 6-keto group in tryptanthrins. Only a solvent with a large amount of Lewis acid (BF 3 ) in acetic acid promoted the condensation sufficiently to yield 7. Substituents at position 4 of isatins prolong the condensation, and a substituent at position 7 in our case of tryptanthrins renders the reaction impossible. Furthermore, the cleavage of the rhodanin-adduct 7 with KOH/H2O took a different route, since it yielded more products. Intense violet flakes could be obtained from the hydrolysis solution in amounts not to be neglected. This material can be shown by DC to consist of two compounds. Tryptanthrin 4 was reported to be reduced by agents such as hypophosphorous acid to yield dihydroditryptanthrin 10 [8], which is oxidised to "ditryptanthrin" (11, (6E)-6-(12-oxoindolo[2,1-b]quinazolin-6-ylidene)indolo[2,1-b] quinazolin-12-one) ( Figure 2). Since there are enough reactive reducing SH molecules present during the hydrolysis reaction of the thioxothiazolidinones, we speculated on what the side products might be. DC supports the identity of the minor compound in the "violet flakes" being 11, while it has very low solubility like 11. Of interest is the nature of the second intense violet product obtained from the "flakes". Since it was insoluble in basic aqueous media, it is suspected not to be an ionic molecule, while there is a difference in its solubility in methanol or acetone. Bergman also reports some stable thioketones of tryptanthrin and ditryptanthrin, prepared by oxygen-sulphur exchange [9]. Regarding the idea that such an exchange might have been feasible with thioxothiazolidinone on tryptanthrin, its absolute insolubility in everything disproves such an idea. On the other hand, a simple ring-closing reaction of the thioacetic acids 8 could possibly yield a thioketone 15 of a dark colour.

Scheme 1.
Overview of all reactions of the quingdainone synthesis. The double bond (3-2 ) is portrayed with the correct Z-configuration. Precursors are also depicted with Z-configuration; however, E-configuration can be present as well.
Furthermore, the cleavage of the rhodanin-adduct 7 with KOH/H 2 O took a different route, since it yielded more products. Intense violet flakes could be obtained from the hydrolysis solution in amounts not to be neglected. This material can be shown by DC to consist of two compounds. Tryptanthrin 4 was reported to be reduced by agents such as hypophosphorous acid to yield dihydroditryptanthrin 10 [8], which is oxidised to "ditryptanthrin" (11, (6E)-6-(12-oxoindolo[2,1-b]quinazolin-6-ylidene)indolo[2,1-b] quinazolin-12-one) ( Figure 2). Since there are enough reactive reducing SH molecules present during the hydrolysis reaction of the thioxothiazolidinones, we speculated on what the side products might be. DC supports the identity of the minor compound in the "violet flakes" being 11, while it has very low solubility like 11. Of interest is the nature of the second intense violet product obtained from the "flakes". Since it was insoluble in basic aqueous media, it is suspected not to be an ionic molecule, while there is a difference in its solubility in methanol or acetone. Bergman also reports some stable thioketones of tryptanthrin and ditryptanthrin, prepared by oxygen-sulphur exchange [9]. Regarding the idea that such an exchange might have been feasible with thioxothiazolidinone on tryptanthrin, its absolute insolubility in everything disproves such an idea. On the other hand, a simple ring-closing reaction of the thioacetic acids 8 could possibly yield a thioketone 15 of a dark colour.
Despite these difficulties, substantial amounts of mercaptoacetic acids 8 after acidic work-up were obtained. Some of them could better be characterised as allylthio derivatives, which crystallise far better. We projected our tryptanthrin-derived mercaptoacetic acids to react analogously with anilines in a DME solution catalysed with trifluoroacetic acid, as reported earlier [10], but this method did not work here. Gränacher et al. already described that mercaptoacetic acids from isatins react with liquid anilines under the evolution of H 2 S [11]. Accordingly, melting 8 at 100 • C for 1 h with different para-substituted anilines for the sake of the simplification of the NMR spectra was attempted to yield 9. The evolution of H 2 S and formation of a yellow colour indicated a successful reaction. Subsequently, excess aniline was removed by sublimation. After an acid/base aqueous work-up procedure, the remaining contamination was a small amount of the para substituted aniline, which was finally removed under high vacuum (10 −3 mbar) at 130 • C. When a higher temperature than this was used, a second set of small signals next to the ones from the original compound were observed in 1 H-NMR. This can most likely be interpreted as isomerisation of the double bond of 9. The last step, a cyclisation under Nazarov conditions in polyphosphoric acid (PPA), yielded quingdainones 2a-g without further complications (see Table 1). Despite these difficulties, substantial amounts of mercaptoacetic acids 8 after acidic work-up were obtained. Some of them could better be characterised as allylthio derivatives, which crystallise far better. We projected our tryptanthrin-derived mercaptoacetic acids to react analogously with anilines in a DME solution catalysed with trifluoroacetic acid, as reported earlier [10], but this method did not work here. Gränacher et al. already described that mercaptoacetic acids from isatins react with liquid anilines under the evolution of H2S [11]. Accordingly, melting 8 at 100 °C for 1 h with different para-substituted anilines for the sake of the simplification of the NMR spectra was attempted to yield 9. The evolution of H2S and formation of a yellow colour indicated a successful reaction. Subsequently, excess aniline was removed by sublimation. After an acid/base aqueous work-up procedure, the remaining contamination was a small amount of the para substituted aniline, which was finally removed under high vacuum (10 −3 mbar) at 130 °C. When a higher temperature than this was used, a second set of small signals next to the ones from the original compound were observed in 1 H-NMR. This can most likely be interpreted as isomerisation of the double bond of 9. The last step, a cyclisation under Nazarov conditions in polyphosphoric acid (PPA), yielded quingdainones 2a-g without further complications (see Table 1).
The question as to whether the quingdainone is an E or Z isomer has been addressed [12]. Unambiguously, it is of Z-geometry, as shown by 1 H-NMR through space coupling of the 1'' NH proton to aromatic proton 2' and not to 4. No other isomer has been found. It has the same configuration lock that occurs in indigo itself by the hydrogen bridge of the indoxyl NH to O of the oxindole moiety. Since substitution sites of the synthesised compounds 2a-g are far from this hydrogen bridge area, it is assumed that all the compounds are Z isomers.
Water solubility was installed with 2f,g. As acidic molecules, 2f,g are soluble in a  The question as to whether the quingdainone is an E or Z isomer has been addressed [12]. Unambiguously, it is of Z-geometry, as shown by 1 H-NMR through space coupling of the 1 NH proton to aromatic proton 2 and not to 4. No other isomer has been found. It has the same configuration lock that occurs in indigo itself by the hydrogen bridge of the indoxyl NH to O of the oxindole moiety. Since substitution sites of the synthesised compounds 2a-g are far from this hydrogen bridge area, it is assumed that all the compounds are Z isomers.
Water solubility was installed with 2f,g. As acidic molecules, 2f,g are soluble in a dilute base from pH > 7. When precursor 9g (R,R = OMe,) was cyclised in PPA, a product 2g with a surprisingly good water solubility was obtained. Usually, quingdainones separate out of the water solution of PPA during work-up. In this case, an adsorbent had to be used to extract 2g. 1 H-NMR of this compound showed none of the expected methoxy signals, and a mass from HRMS interpreted as R,R = OH proved the notion that the methoxy groups have been demethylated. Different examples from the literature also report cleavage of OMe groups and esters in PPA at higher temperatures [13,14].

Quingdainones Cause Apoptosis in Nalm-6 Cells
In order to examine the biological activity of new quingdainone derivates, we first measured DNA fragmentation as evidence for the late apoptotic phase [15] in leukaemia cells (Nalm-6) after 72 h of treatment. Only a few compounds were able to induce apoptosis in the measured concentrations up to 50 µM (see Table 2). Dihydroxyquingdainone (2g) turned out to be the most promising compound, so we further characterised its effects in malignant cells; it has an AC 50 of 7.5 µM, which is the concentration of compounds able to induce apoptosis in 50 percent of the Nalm-6 cells after 72 h of incubation [16][17][18][19][20]. The non-substituted original quingdainone does not induce apoptosis in concentrations up to 100 µM. Table 2. AC 50 values of the synthesised quingdainone variants. They describe the concentration of compounds able to induce apoptosis in 50 percent of the Nalm-6 cells after 72 h of incubation. Compounds with no significant apoptosis induction up to 50 µM are marked with a slash. Quingdainone is not active at all in our cell lines.
With a view to clinical application, necrosis could be excluded as a non-specific form of cell death. For this purpose, the enzyme lactate dehydrogenase released by cell lysis was quantified colorimetrically after treating the cells with 2g for 1 h. It could be shown that the viability of the cells was not significantly reduced by relevant concentrations of 2g (see Figure 3).
In order to examine the biological activity of new quingdainone derivates, we first measured DNA fragmentation as evidence for the late apoptotic phase [15] in leukaemia cells (Nalm-6) after 72 h of treatment. Only a few compounds were able to induce apoptosis in the measured concentrations up to 50 µM (see Table 2). Dihydroxyquingdainone (2g) turned out to be the most promising compound, so we further characterised its effects in malignant cells; it has an AC50 of 7.5 µM, which is the concentration of compounds able to induce apoptosis in 50 percent of the Nalm-6 cells after 72 h of incubation [16][17][18][19][20]. The non-substituted original quingdainone does not induce apoptosis in concentrations up to 100 µM. Table 2. AC50 values of the synthesised quingdainone variants. They describe the concentration of compounds able to induce apoptosis in 50 percent of the Nalm-6 cells after 72 h of incubation. Compounds with no significant apoptosis induction up to 50 µM are marked with a slash. Quingdainone is not active at all in our cell lines.

<7.5 µM /
With a view to clinical application, necrosis could be excluded as a non-specific form of cell death. For this purpose, the enzyme lactate dehydrogenase released by cell lysis was quantified colorimetrically after treating the cells with 2g for 1 h. It could be shown that the viability of the cells was not significantly reduced by relevant concentrations of 2g (see Figure 3).

Dihydroxyquingdainone 2g Shows Anti-Proliferative Effects in Nalm-6 Cells
For the tumour therapy, the inhibition of proliferation of cells is important to stop the spread of malignant cells. To prove this, Nalm-6 cells were treated with different concentrations of 2g for 24 h before the cell count was determined via the CASY Cell Counter and Analyzer System. Figure 4 shows the reduction in cell count and thus an inhibition of proliferation in a dose-dependent manner.

Dihydroxyquingdainone 2g Shows Anti-Proliferative Effects in Nalm-6 Cells
For the tumour therapy, the inhibition of proliferation of cells is important to stop the spread of malignant cells. To prove this, Nalm-6 cells were treated with different concentrations of 2g for 24 h before the cell count was determined via the CASY Cell Counter and Analyzer System. Figure 4 shows the reduction in cell count and thus an inhibition of proliferation in a dose-dependent manner.

Dihydroxyquingdainone 2g Involves the Intrinsic Apoptosis Pathway
Two major pathways have been found to induce apoptosis: the extrinsic and intrinsic pathway [21,22]. The latter is mitochondrially mediated and can be triggered via, i.a., ROS or DNA damage. The permeabilization of the mitochondrial outer membrane (MOMP) results in the release of pro-apoptotic proteins, which in further steps activate procaspase 9. As a result, effector caspase-3 is processed, which then initiates cellular changes of apoptosis [23,24].
To investigate the apoptosis induction via the intrinsic pathway, mitochondrial membrane potential, which is changed as a result of MOMP, could be examined. Therefore, Nalm-6 cells were incubated with 2g for 48 h and subsequently stained with JC-1. The results show a significant increase in the proportion of cells with reduced mitochondrial membrane potential (see Figure 5), which is a direct indication of the involvement of the intrinsic pathway.

Dihydroxyquingdainone 2g Involves the Intrinsic Apoptosis Pathway
Two major pathways have been found to induce apoptosis: the extrinsic and intrinsic pathway [21,22]. The latter is mitochondrially mediated and can be triggered via, i.a., ROS or DNA damage. The permeabilization of the mitochondrial outer membrane (MOMP) results in the release of pro-apoptotic proteins, which in further steps activate procaspase 9. As a result, effector caspase-3 is processed, which then initiates cellular changes of apoptosis [23,24].
To investigate the apoptosis induction via the intrinsic pathway, mitochondrial membrane potential, which is changed as a result of MOMP, could be examined. Therefore, Nalm-6 cells were incubated with 2g for 48 h and subsequently stained with JC-1. The results show a significant increase in the proportion of cells with reduced mitochondrial membrane potential (see Figure 5), which is a direct indication of the involvement of the intrinsic pathway.
Furthermore, the induction of apoptosis by 2g depends on the level of expression of Bcl-2. BiBo cells (vincristine-resistant BJAB cells) show an overexpression of this protein.
We compared the apoptosis induction via 2g in primary BJAB and BiBo cells. It was shown that 2g induces significantly less apoptosis in BiBo cells than in BJAB cells, which is why a dependence on Bcl-2 can be assumed ( Figure 6A). This is another hint for the involvement of the intrinsic pathway. In the same way, a dependence on capase-3 was proven. The pancaspase inhibitor Z-VAD is able to inhibit the 2g-induced apoptosis significantly. Z-VAD alone do not induce apoptosis (see Figure 7). These data demonstrate that dihydroxyquingdainone 2g-induced apoptosis is dependent on the activity of caspases in tumour cells. The number of induced apoptotic cells by 2g in BJAB cells was significantly higher than in 7CCA cells (doxorubicin-resistant BJAB cells). Moreover, 7CCA cells show an underexpression of caspase-3, so 2g induces apoptosis depending on the level of caspase-3 ( Figure 6B). Furthermore, the induction of apoptosis by 2g depends on the level of expression of Bcl-2. BiBo cells (vincristine-resistant BJAB cells) show an overexpression of this protein.
We compared the apoptosis induction via 2g in primary BJAB and BiBo cells. It was shown that 2g induces significantly less apoptosis in BiBo cells than in BJAB cells, which is why a dependence on Bcl-2 can be assumed ( Figure 6A). This is another hint for the involvement of the intrinsic pathway. In the same way, a dependence on capase-3 was proven. The pan-caspase inhibitor Z-VAD is able to inhibit the 2g-induced apoptosis significantly. Z-VAD alone do not induce apoptosis (see Figure 7). These data demonstrate that dihydroxyquingdainone 2g-induced apoptosis is dependent on the activity of caspases in tumour cells.The number of induced apoptotic cells by 2g in BJAB cells was significantly higher than in 7CCA cells (doxorubicin-resistant BJAB cells). Moreover, 7CCA cells show an underexpression of caspase-3, so 2g induces apoptosis depending on the level of caspase-3 ( Figure 6B).   Furthermore, the induction of apoptosis by 2g depends on the level of expression of Bcl-2. BiBo cells (vincristine-resistant BJAB cells) show an overexpression of this protein.
We compared the apoptosis induction via 2g in primary BJAB and BiBo cells. It was shown that 2g induces significantly less apoptosis in BiBo cells than in BJAB cells, which is why a dependence on Bcl-2 can be assumed ( Figure 6A). This is another hint for the involvement of the intrinsic pathway. In the same way, a dependence on capase-3 was proven. The pan-caspase inhibitor Z-VAD is able to inhibit the 2g-induced apoptosis significantly. Z-VAD alone do not induce apoptosis (see Figure 7). These data demonstrate that dihydroxyquingdainone 2g-induced apoptosis is dependent on the activity of caspases in tumour cells.The number of induced apoptotic cells by 2g in BJAB cells was significantly higher than in 7CCA cells (doxorubicin-resistant BJAB cells). Moreover, 7CCA cells show an underexpression of caspase-3, so 2g induces apoptosis depending on the level of caspase-3 ( Figure 6B).

Dihydroxyquingdainone 2g Overcomes Resistances to Cytostatic Drugs
The resistance of malignant cells to conventional cytostatic drugs is a major problem in the therapy of tumours and lowers the chances of achieving a cure [25,26]. To illustrate this, Nalm-6 cells were made resistant to various cytostatic drugs. To test whether 2g is able to overcome resistance, these different cell lines were treated with the quingdainone derivate 2g. were carried out per concentration. The mean values ± SD are shown (* p < 0.05 vs. corresponding concentration in BJAB, t test).

Figure 7.
The apoptosis induction of 2g is caspase-dependent. Nalm-6 cells were treated with two concentrations of the pan-caspase inhibitor Z-VAD, with 2g and with 2g together with two concentrations of the pan-caspase inhibitor Z-VAD for 72 h. As the solvent, control cells were treated with the same amount of DMSO as used for 2g. After treatment, the induction of apoptosis was measured via flow cytometric analysis of DNA fragmentation. Three replicates were carried out per concentration. The mean values ± SD are shown (p < 0.05 vs. corresponding concentration in Nalm-6, ttest). The pan-caspase inhibitor Z-VAD is able to inhibit the 2g-induced apoptosis significantly. Z-VAD alone do not induce apoptosis. These data demonstrate that dihydroxyquingdainone 2g-induced apoptosis is dependent on the activity of caspases in tumour cells.

Dihydroxyquingdainone 2g Overcomes Resistances to Cytostatic Drugs
The resistance of malignant cells to conventional cytostatic drugs is a major problem in the therapy of tumours and lowers the chances of achieving a cure [25,26]. To illustrate this, Nalm-6 cells were made resistant to various cytostatic drugs. To test whether 2g is able to overcome resistance, these different cell lines were treated with the quingdainone derivate 2g.
BeKa cells, which are vincristine (VCR)-resistant Nalm-6 cells, show an expression of the p glycoprotein compared to the initial cells. Substances can be actively secreted from the cell via this membrane protein, whereby they lose their effect [27,28]. This mechanism often results in multidrug resistance (MDR) [29], which has also been demonstrated for BeKa cells. BeKa cells show a co-resistance to anthracyclines (idarubicin, daunorubicin, doxorubicin, epirubicin), mitoxantrone, fludarabine, vinca alkaloids (vincristine, vindesine, vinorelbine, vinblastine) and etoposide in vitro [30].
Nalm-6 and BeKa cells were treated with 2g for 72 h before DNA fragmentation was performed. The comparison of the percentage of apoptotic cells in these 2g-treated cell lines showed no significant differences (see Figure 8A). Thus, 2g is not a substrate of the p-glycoprotein and overcomes multidrug resistance in leukaemia cells. Similar results have been found for JeBa and JeFri cells, which are Nalm-6 cells resistant to methotrexate (MTX) and etoposide (Eto.), respectively. The mechanisms of resistance have not been elucidated yet. Dihydroxyquingdainone 2g overcomes both resistances and shows even higher effects in the resistant cells compared to Nalm-6 cells (see Figure 8B,C). Figure 7. The apoptosis induction of 2g is caspase-dependent. Nalm-6 cells were treated with two concentrations of the pan-caspase inhibitor Z-VAD, with 2g and with 2g together with two concentrations of the pan-caspase inhibitor Z-VAD for 72 h. As the solvent, control cells were treated with the same amount of DMSO as used for 2g. After treatment, the induction of apoptosis was measured via flow cytometric analysis of DNA fragmentation. Three replicates were carried out per concentration.
The mean values ± SD are shown (p < 0.05 vs. corresponding concentration in Nalm-6, t-test). The pan-caspase inhibitor Z-VAD is able to inhibit the 2g-induced apoptosis significantly. Z-VAD alone do not induce apoptosis. These data demonstrate that dihydroxyquingdainone 2g-induced apoptosis is dependent on the activity of caspases in tumour cells.
BeKa cells, which are vincristine (VCR)-resistant Nalm-6 cells, show an expression of the p glycoprotein compared to the initial cells. Substances can be actively secreted from the cell via this membrane protein, whereby they lose their effect [27,28]. This mechanism often results in multidrug resistance (MDR) [29], which has also been demonstrated for BeKa cells. BeKa cells show a co-resistance to anthracyclines (idarubicin, daunorubicin, doxorubicin, epirubicin), mitoxantrone, fludarabine, vinca alkaloids (vincristine, vindesine, vinorelbine, vinblastine) and etoposide in vitro [30].
Nalm-6 and BeKa cells were treated with 2g for 72 h before DNA fragmentation was performed. The comparison of the percentage of apoptotic cells in these 2g-treated cell lines showed no significant differences (see Figure 8A). Thus, 2g is not a substrate of the p-glycoprotein and overcomes multidrug resistance in leukaemia cells. Similar results have been found for JeBa and JeFri cells, which are Nalm-6 cells resistant to methotrexate (MTX) and etoposide (Eto.), respectively. The mechanisms of resistance have not been elucidated yet. Dihydroxyquingdainone 2g overcomes both resistances and shows even higher effects in the resistant cells compared to Nalm-6 cells (see Figure 8B,C).

Dihydroxyquingdainone 2g Causes Significantly Less Apoptosis in Healthy Human
Leukocytes Than in Nalm-6 Selectivity of cytostatic drugs is very important to ensure that only malignant cells and no healthy cells are affected. Most of the conventional agents cause side effects because of their limited selectivity against cancer cells [31]. A first test to determine the selectivity of a new potential agent is to test the effect on healthy leukocytes (ex vivo). For 2g, a significantly reduced apoptosis in leukocytes compared to malignant Nalm-6 cells was shown (Figure 9).

Dihydroxyquingdainone 2g Causes Significantly Less Apoptosis in Healthy Human Leukocytes than in Nalm-6
Selectivity of cytostatic drugs is very important to ensure that only malignant cells and no healthy cells are affected. Most of the conventional agents cause side effects because of their limited selectivity against cancer cells [31]. A first test to determine the selectivity of a new potential agent is to test the effect on healthy leukocytes (ex vivo). For 2g, a significantly reduced apoptosis in leukocytes compared to malignant Nalm-6 cells was shown (Figure 9).

Discussion
The interest in natural compounds for use as anti-cancer therapeutic agents is very high due to the wide range of biological activities found in different kingdoms of life [32]. Several of today's important cytostatic drugs are based on natural substances. Vinca alka- Figure 9. Selectivity of 2g. Nalm-6 cells and healthy human leucocytes (ex vivo) were treated with different concentrations of 2g for 72 h. As the solvent, control cells were treated with the same amount of DMSO as used for 2g. After treatment, the induction of apoptosis was measured via flow cytometric analysis of DNA fragmentation. Three replicates were carried out per concentration. The mean values ± SD are shown (* p < 0.05 vs. corresponding concentration in Nalm-6, t-test).

Discussion
The interest in natural compounds for use as anti-cancer therapeutic agents is very high due to the wide range of biological activities found in different kingdoms of life [32]. Several of today's important cytostatic drugs are based on natural substances. Vinca alkaloids such as vincristine or vinblastine as well as taxanes (paclitaxel, docetaxel) were initially isolated from plants [33,34]; nevertheless, the occurrence of resistance demands ongoing research. For centuries, indigo plants have been used in traditional medicine. Some ingredients of Qing Dai, an indigo-containing preparation still used today which consists of extracts from different herbs in traditional Chinese medicine, show anti-tumour effects [35]. The responsible molecules also include quingdainone 2 (R=R =R = H) [7]. Derivatives were investigated here in comparison with the original, since it was speculated that a more water-soluble derivative might be more suitable for the elucidation of its activity. Adding neutral substituents (2a-e) to the aromatic cores does not improve solubility in aqueous media and does not improve it so much in organic solvents, so the lack of cytotoxicity seems to reflect this. However, just improving water solubility does not improve cytotoxicity. Rather, it was abolished, since in the carboxylic acid 2f, the respective AC 50 is above 50 mmol or more like in 2a-e. The most effective compound dihydroxyquingdainone 2g showed in vitro cytotoxic effects in leukaemia cells and was able to overcome various cytostatic resistances, making it a potential new agent for the therapy of malignant diseases. Apoptosis was induced with a half maximal concentration (AC 50 ) of <7.5 µM (Table 2) after 72 h of incubation.
Another important property for treating cancer is to inhibit the proliferation of malignant cells. This could be proven for 2g in Nalm-6 cells ( Figure 4). Fortunately, necrosis was excluded as a non-specific form of cell death for 2g, which would otherwise have been an exclusion criterion for further investigation (Figure 3).
Another highly important property of 2g is the selective induction of apoptosis, which means that non-proliferating healthy human cells such as leukocytes are hardly affected (Figure 9). It would also be interesting to find out how healthy human proliferating cells react, which could be achieved in further experiments. In order to assess the efficacy of 2g, it is also very important to investigate the mechanism of action. Although further studies would be useful, it has already been shown that 2g induces apoptosis via the intrinsic signalling pathway ( Figure 5). The proapoptotic protein Bcl-2 and the effector caspase-3 are important for the mechanism of apoptosis via 2g ( Figure 6A,B).
Dihydroxyquingdainone 2g was also convincing because it overcame various cytostatic resistances. In BeKa cells (daunorubicin-resistant Nalm-6 cells), 2g induced apoptosis as high as in original Nalm-6 cells ( Figure 8A). Knowing the mechanism of resistance in these cells helps us to understand how 2g works. Overcoming this resistance based on the overexpression of p-glycoprotein as a multidrug resistance mechanism shows the importance of further investigation. Furthermore, 2g is able to overcome etoposide and methotrexate resistance in leukaemia cells ( Figure 8B,C). Amounts of 22.7 nM vincristine, 50 nM methotrexate and 0.37 µM etoposide are equivalent to the effect of about 7.5 µM 2g, which sets a base for further investigation. Closely related to quingdainone is indirubin. Of its derivatives, indirubin-5-sulfonate and indirubin-3 -oxime were recognised as potent anti-leukaemic agents. Due to the large aromatic residue joint to the indole nucleus down side from the double bond, it is hard to see a connection between the activity of quingdainones and indirubin. Indirubins are cell cycle inhibitors [1] by inhibition of CDKs: indirubin-5-sulfonate 3 -oxime has an IC 50 of 7 nM at CDK-1. Moreover, halogen substituents improve the activity of indirubin in pos. 3 [36]. In our case, the extrinsic and intrinsic apoptosis induction pathways [21,22] play a major role, together with change of the mitochondrial membrane potential. Caspase-3 is processed actively, which then initiates cellular changes of apoptosis [23,24]. We expect a quite different dependence from substituents.

Synthetic Procedures General Considerations
All the commercially available chemicals were used without further purification. NMR spectra were recorded on a JEOL ECS-400 (400 MHz, Japan). Chemical shifts are reported in ppm and refer to residual protons in the deuterated solvent as the internal standard. IR spectra were recorded on a Nicolet 380 FT-IR (Thermo Scientific, England). UV/VIS-spectra were recorded on a Specord 210 (Analytik Jena, Jena, Germany). Mass spectrometry was performed on an IT-TOF from Shimadzu, Japan with direct injection of the samples. For the original NMR, IR and mass spectrometry of compounds, please see the Supplementary Materials. Flash purification was performed with a Puriflash 4250-250 from Interchim, France.
Tryptanthrins 4, which are initially needed, were prepared with isatin 5 and isatoic anhydride according to Bergman et al. [37], which is extendable to all commercially available isatins. For tryptanthrin derivatives with substituents at positions 1-4 (R 2 -5 in quingdainones), another method was used due to the unavailability of various isatoic acids. 2-Chloroisatins 6 were prepared according to Grimshaw et al. [38]. These 2-chloroisatins reacted with 2-aminobenzoic acids at 130 • C in DMF to form tryptanthrins without any further additives [39]. Unsubstituted quingdainone was obtained according to [12].
General Procedure for the Synthesis of 2-(12-Oxoindolo[2,1-b]quinazolin-6(12H)-ylidene)-2-(phenylamino)acetic Acids In total, 400 mg of substances 8b-f was added to 2 g of a para-substituted-aniline in a 100 mL Erlenmeyer flask with a ground glass joint. This mixture was heated to 100 • C for 1 h under stirring. The aniline, which deposits on the edge of the vessel, was brought back to the medium with a spatula every 15 min. After cooling to 80 • C, the excess aniline was removed by sublimation at 80 • C and 7 mbar for 1 h. The remaining solid was stirred with 10 mL of 10% NaOH solution. After 1 h, the suspension was filtrated and the residue was allowed to dry overnight. The resulting solid was again sublimated at 140 • C at 10 −3 mbar to remove the rest of the aniline, while the product remained at the bottom of the flask.

Measurement of the Mitochondrial Membrane Potential
The intrinsic pathway leads to the permeabilisation of the mitochondrial outer membrane (MOMP), which results in a breakdown of the transmembrane potential [23,47]. For quantifying cells with a reduced mitochondrial membrane potential, 5,5 ,6,6 -tetrachloro-1,1 ,3,3 -tetraethylbenzimidazoylcarbocyanine iodide (JC-1) dye was used and cells were measured by flow cytometry [48]. After treating the cells with different concentrations of 2g for 48 h, cells were collected by centrifugation (3000 rpm, 5 min, 4 • C) and resuspended in phenol-free RPMI. Then, 6.25 µL of JC-1 (ThermoFisher Scientific Inc., Waltham, MA, USA) was added to each sample, and all were incubated for 30 min (37 • C, 300 rpm). Subsequently, cells were centrifuged (4000 rpm, 4 • C, 5 min) and resuspended in 1× PBS. The percentage of cells with reduced mitochondrial membrane potential were quantified using flow cytometry and Cell Quest Software.

Isolation of Healthy Human Leukocytes
A total of 40 mL of blood was collected and diluted with 14 mL RPMI. An amount of 4 mL of Ficoll (saccharose-epichlorohydrin-Copolymer) was placed in 15 mL of Falcons and 5 mL of diluted blood was carefully added to each. The blood was separated by density gradient centrifugation to obtain the buffy coat containing leucocytes [49,50]. After centrifugation (18 min, 18 • C, 2000 rpm), leukocytes were collected by transferring the buffy coat with a Pasteur pipette into a 50 mL tube (VWR International GmbH, Darmstadt, Germany). Buffy coat was diluted 1:1 with RPMI and centrifuged for another 5 min (18 • C, 2000 rpm). The pellet was resuspended in 10 mL of RPMI. Subsequently, the cell count and viability were determined by using the CASY ® Cell Counter and Analyzer System. Cells were seeded with a density of 3 ×10 5 cells/mL. The further treatment of the cells was carried out as described above.

Statistics
The results shown correspond to a triplicate determination. The standard deviations (SD) are shown as error bars. If the SD is very small, error bars may not be seen. Significance was calculated after normalization to DMSO using a two-tailed t-test with a significance level of 0.05. Graphs and statistics were created using Microsoft Office Excel.

Conclusions
All in all, dihydroxyquingdainone (2g) is a very interesting compound among all of the tested quingdainones, especially in terms of overcoming a wide variety of cytostatic resistances. Inhibition of proliferation and specific apoptosis induction in selected malignant cell lines (human B cell precursor leukaemia cells, human Burkitt-like lymphoma cells and human chronic myeloid leukaemia cells) could be shown. Furthermore, as a first indication of the mechanism of action, involvement of the intrinsic signalling pathway has already been proven. Dihydroxyquingdainone 2g induced significantly less apoptosis in healthy human leukocytes than in malignant cell lines. In particular, the overcoming of resistances in acute lymphoblastic leukaemia cells is of great importance, as this represents a major advantage for the treatment of cytostatic-resistant tumour cells. The data are promising for further investigation of dihydroxyquingdainone 2g for the treatment of malignancies, but also for the exploration of other quingdainones, as a strong structure-dependent anti-tumour effect was shown.