Synthesis and Pharmacological In Vitro Investigations of Novel Shikonin Derivatives with a Special Focus on Cyclopropane Bearing Derivatives

Melanoma is the deadliest form of skin cancer and accounts for about three quarters of all skin cancer deaths. Especially at an advanced stage, its treatment is challenging, and survival rates are very low. In previous studies, we showed that the constituents of the roots of Onosma paniculata as well as a synthetic derivative of the most active constituent showed promising results in metastatic melanoma cell lines. In the current study, we address the question whether we can generate further derivatives with optimized activity by synthesis. Therefore, we prepared 31, mainly novel shikonin derivatives and screened them in different melanoma cell lines (WM9, WM164, and MUG-Mel2 cells) using the XTT viability assay. We identified (R)-1-(1,4-dihydro-5,8-dihydroxy-1,4-dioxonaphthalen-2-yl)-4-methylpent-3-enyl 2-cyclopropyl-2-oxoacetate as a novel derivative with even higher activity. Furthermore, pharmacological investigations including the ApoToxGloTM Triplex assay, LDH assay, and cell cycle measurements revealed that this compound induced apoptosis and reduced cells in the G1 phase accompanied by an increase of cells in the G2/M phase. Moreover, it showed hardly any effects on the cell membrane integrity. However, it also exhibited cytotoxicity against non-tumorigenic cells. Nevertheless, in summary, we could show that shikonin derivatives might be promising drug leads in the treatment of melanoma.


Introduction
Malignant melanoma belongs to the most dangerous type of skin cancer and arises from melanocytes [1]. Only 2-3% of all diagnosed skin cancers are melanomas, however, they account for approx. 75% of all skin cancer deaths [2]. Melanomas are divided into four different stages: stages I and II display local primary tumors, stage III represents tumors with locoregional metastases, and stage IV tumors with distant metastases. In addition, tumors are typically categorized by thickness and ulceration [1,3]. When melanomas are detected at an early clinical stage and treated appropriately, the 10-year survival rate is around 98% [4]. However, when the tumor metastasizes, the survival rate drops dramatically and therapeutic success is limited due to tumor recurrence, severe side effects, and/or the development of resistances [2]. According to the Melanoma Research Alliance, the five-year survival rate for metastatic melanoma is currently only 22.5%. Therefore, the search for novel drug leads remains an important task. The first strategy was the optimization of the cyclopropylacetate in 3 [17]. In the bicyclus 4, 1′ and 2′ positions of cyclopropylacetate are connected as part of a cyclohexane. Replacement of the methylene spacer of 3 with a CO group, a C,C double bond, and an ethylene group resulted in 5, 6, and 7, respectively. The first strategy was the optimization of the cyclopropylacetate in 3 [17]. In the bicyclus 4, 1 and 2 positions of cyclopropylacetate are connected as part of a cyclohexane. Replacement of the methylene spacer of 3 with a CO group, a C,C double bond, and an ethylene group resulted in 5, 6, and 7, respectively.
The cyclopropane precursor acids were synthesized in our laboratory. As outlined in Figure 2, the bicyclic acetic acid p1 was prepared from the corresponding (1-cyclohexenyl)- 2-acetic via Furukawa modification of the Simmons-Smith reaction analogously to a procedure of Renaud and Fox [28].  was prepared by α-alkylation of acetic acid in analogy to a procedure described by Barczak and Jarvo [29] ( Figure 3). The Knoevenagel condensation of cyclopropanecarbaldehyde with malonic acid [30] resulted in β-cyclopropylacrylic acid p3.  was obtained by oxidation of acetylcyclopropane with KMnO 4 in a combination of the procedures given by Prokopenko et al. [31] and Xu et al. [32] (Figure 4). Details of the syntheses of the precursor acids are described in material and methods. Their NMR spectra can be found in the supplementary material (chapter 3). In all cases (precursor acids and shikonin derivatives), the structure and purity of each compound were analyzed using 1D and 2D NMR, LC-MS, and IR experiments. The first strategy was the optimization of the cyclopropylacetate in 3 [17]. In the bicyclus 4, 1′ and 2′ positions of cyclopropylacetate are connected as part of a cyclohexane. Replacement of the methylene spacer of 3 with a CO group, a C,C double bond, and an ethylene group resulted in 5, 6, and 7, respectively.
The cyclopropane precursor acids were synthesized in our laboratory. As outlined in Figure 2, the bicyclic acetic acid p1 was prepared from the corresponding (1cyclohexenyl)- 2-acetic via Furukawa modification of the Simmons-Smith reaction analogously to a procedure of Renaud and Fox [28].  was prepared by α-alkylation of acetic acid in analogy to a procedure described by Barczak and Jarvo [29] (Figure 3). The Knoevenagel condensation of cyclopropanecarbaldehyde with malonic acid [30] resulted in β-cyclopropylacrylic acid p3.  was obtained by oxidation of acetylcyclopropane with KMnO4 in a combination of the procedures given by Prokopenko et al. [31] and Xu et al. [32] (Figure 4). Details of the syntheses of the precursor acids are described in material and methods. Their NMR spectra can be found in the supplementary material (chapter 3). In all cases (precursor acids and shikonin derivatives), the structure and purity of each compound were analyzed using 1D and 2D NMR, LC-MS, and IR experiments.    The first strategy was the optimization of the cyclopropylacetate in 3 [17]. In the bicyclus 4, 1′ and 2′ positions of cyclopropylacetate are connected as part of a cyclohexane. Replacement of the methylene spacer of 3 with a CO group, a C,C double bond, and an ethylene group resulted in 5, 6, and 7, respectively.
The cyclopropane precursor acids were synthesized in our laboratory. As outlined in Figure 2, the bicyclic acetic acid p1 was prepared from the corresponding (1cyclohexenyl)- 2-acetic via Furukawa modification of the Simmons-Smith reaction analogously to a procedure of Renaud and Fox [28].  was prepared by α-alkylation of acetic acid in analogy to a procedure described by Barczak and Jarvo [29] (Figure 3). The Knoevenagel condensation of cyclopropanecarbaldehyde with malonic acid [30] resulted in β-cyclopropylacrylic acid p3.  was obtained by oxidation of acetylcyclopropane with KMnO4 in a combination of the procedures given by Prokopenko et al. [31] and Xu et al. [32] (Figure 4). Details of the syntheses of the precursor acids are described in material and methods. Their NMR spectra can be found in the supplementary material (chapter 3). In all cases (precursor acids and shikonin derivatives), the structure and purity of each compound were analyzed using 1D and 2D NMR, LC-MS, and IR experiments.    The second goal was to produce a broad spectrum of structural features within the acyl residue of shikonin to find potential novel drug leads. For better comparison and a more complete picture, we also included some already known derivatives. Phenylacetate 8 and cinnamate 9 showed, for example, cytotoxicity in previous studies and were synthesized for comparison reasons [26,33]. As there have been no reports about any shikonin alkynylacyl esters yet, we synthesized tetrolate 10, 2-butynoate 11, and 3phenylpropiolate 12 to investigate the influence of a triple bond.
Except for three recently published studies about benzoylacrylates [34], succinamides, and maleinamides [35,36], all carbonyl groups in the acyl chain of shikonin or alkannin esters were part of acetoxy groups mainly derived from β-hydroxy acylates. However, the substances of Sun et al. [34] are of unknown chirality, among them the p- The second goal was to produce a broad spectrum of structural features within the acyl residue of shikonin to find potential novel drug leads. For better comparison and a more complete picture, we also included some already known derivatives. Phenylacetate 8 and cinnamate 9 showed, for example, cytotoxicity in previous studies and were synthesized for comparison reasons [26,33]. As there have been no reports about any shikonin alkynylacyl esters yet, we synthesized tetrolate 10, 2-butynoate 11, and 3-phenylpropiolate 12 to investigate the influence of a triple bond.
Except for three recently published studies about benzoylacrylates [34], succinamides, and maleinamides [35,36], all carbonyl groups in the acyl chain of shikonin or alkannin esters were part of acetoxy groups mainly derived from β-hydroxy acylates. However, the substances of Sun et al. [34] are of unknown chirality, among them the pmethylbenzoylacrylate 13. We also had a look on derivatives with the carbonyl carbon as a part of the carbon chain-either as a ketone or as part of a diester. Pyruvate 14 and 2-oxo-2-phenylacetate 15 represent α-ketoacylates. γ-Carbonyls are found in the keto esters 13 and 16 as well as in the diesters derived from monomethyl succinate 17 and monoethyl fumarate 18.
The known short chain alkyl and alkenyl esters 19 (isobutyrate) [37], 20 (isovalerate) [38], 21 (crotonate) [39], and 22 (sorbate) [38] were used to examine the influence of the saturation and branching on the activity. Compounds 19, 20, and 22 are already known to inhibit the growth of other types of cancer cells [23,37]. However, there have been no reports about their effects in melanoma cells. Derivative 21 was one of the very first non-natural 1 -O-acyl shikonines and shialkines [39], but no data about the biological activity have been reported so far.
Other features, which we investigated, were carbocyclic and heterocyclic aryl groups. The most basic structures, i.e., phenylacetyl, benzoyl, and diphenylacetyl (8,23,24), were among the early synthetic 1 -O-acyl shikonines and shialkines. They were reported to bind to tubuline [38] and were cytotoxic against several cancer cell lines [23,28,40], however, no data were published about their effects in melanoma cells. Shikonin cinnamate 9 was intensively explored for its anticancer properties, too [26,41,42]. We used these derivatives to investigate the influence of the α,β double bond with the help of 3-phenylpropionate 25 [36].

Results of the XTT Screening
All prepared shikonin derivatives (see Table 1, compounds 4 to 34) were subjected to a cytotoxicity screening using the XTT viability assay. This assay is based on the activity of mitochondrial dehydrogenases. These enzymes cleave the yellow tetrazolium salt XTT leading to an orange formazan. This conversion only occurs in viable cells and can be directly quantified by measuring the absorbance [47]. Melanoma cells were treated with 1.0 µM, 5.0 µM, and 10.0 µM of each derivative for 72 h ( Figure 5; Figures S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC 50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2). directly quantified by measuring the absorbance [47]. Melanoma cells were treated wi 1.0 µM, 5.0 µM, and 10.0 µM of each derivative for 72 h ( Figure 5; Supplementary Figur S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table  Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figur S1 and S2). 1.0 µM, 5.0 µM, and 10.0 µM of each derivative for 72 h ( Figure 5; Supplementary Figures S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures  S1 and S2).  In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures  S1 and S2).  In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures  S1 and S2). S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures  S1 and S2). S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures  S1 and S2). S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures  S1 and S2). S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2). 10 S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2). S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2). 12 S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2). 13 S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2). 14 S1 and S2 in the Supplementary Material). In brief, the most active derivatives were 5, 6, 9, 14, 24, and 29, with 5 being the most cytotoxic. Its IC50 values are listed in Table 2. Moreover, the derivatives 11 and 28 showed no cytotoxicity at 10 µM, the derivatives 4, 16, 20, and 27 exhibited only a very weak cytotoxicity at 10 µM (Supplementary Figures S1 and S2).

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG-Mel2 cells reacted in the case of some compounds (18,

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG-Mel2 cells reacted in the case of some compounds (18,

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG

34
Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG-Mel2 cells reacted in the case of some compounds (18,   Discussed in more detail, the results showed that the sensitivity of the three cell lines used was different. In general, WM164 cells reacted least to the treatment. Only a few compounds showed moderate activity, for example, the known derivatives 9, 24, and 29, as well as the novel derivative 14. WM9 cells were most affected by the derivatives. Next to 5 (novel), compounds 6 (novel), 9 (known), and 14 (novel) were the most cytotoxic derivatives in this cell line. MUG-Mel2 cells reacted in the case of some compounds (18, 19,  21, 22, 30, 33, and 34) more sensitive to the treatment than the WM9 and WM164 cells. This is of special interest, because MUG-Mel2 cells are NRAS mutated, while WM9 cells and WM164 cells are BRAF mutated. The mutational status is another category in melanoma diagnosis because some mutations lead to a poorer prognosis than others do. Around 50% of all melanomas exhibit a mutation in the BRAF gene and another 25% a mutation in the NRAS gene [48]. Tumors carrying a BRAF mutation are currently typically treated with a combination of MEK and BRAF inhibitors, however, tumor resistances often develop [49]. In the case of NRAS mutated melanoma, the therapeutic success is even lower because they are more difficult to treat [50]. This means that the most active derivatives of our study, can also display lead compounds for the development of further novel shikonin derivatives with a special focus on NRAS mutated melanoma cells.
Returning to the present results, we firstly modified the structure of our previous hit 3. Further substituents at the cyclopropane reduced the activity against all cell lines significantly. The modification of the spacer showed unequal effects: replacement of the methylene group with carbonyl (5) or ethylene (7) showed a similar activity on the cell lines with 5 being more cytotoxic towards WM9 cells. An α,β unsaturation resulted in less activity towards MUG-Mel2 cells but in an increased effect in WM9 cells.
When analyzing the other investigated structural features, it became obvious that hydrogenation of the acyl side chain of 2 reduced the cytotoxicity to a very low level in all cell lines. Shortening the side chain and removal of the branching β-methyl are among the few modifications, which had a bigger influence on the activity in WM9 cells than in MUG-Mel2 cells. The additional conjugated double bond in the sorbate 22 had no significant effect. However, the attachment of a phenyl group restored the properties to the activity level of 2. The p-dimethylamino group had no effect on the activity and, again, hydrogenation of the exocyclic double bond of the cinnamic residue lowered activity significantly. The removal or shortening of the spacer slightly improved the activity (compare 8 and 23). The replacement of the phenyl ring of 9 with a furan moiety (30) abolished the activity in WM9 cells but kept some effects in MUG-Mel2 cells. The diphenylacetate 24 and indolylpropionate 29 appeared as active as the cinnamate 9 and as 2. The replacement of the phenyl ring in benzoate 23 by a 3-pyridine or 4-pyridine abolished the cytotoxicity.
Moreover, the activity of the γ oxo esters 13, 16, 17, and 18 was generally low. While WM9 cells were less affected by saturated and unsaturated compounds, MUG-Mel2 cells showed a slight sensitivity towards unsaturated compounds. The introduction of an α carbonyl in 1 -O-acetylshikonin resulted in pyruvate 14 which showed an improved activity in WM9 cells, but a reduced one in MUG-Mel2 cells. Also, α-carbonylation of the benzoate 23, resulting in 2-oxo-2-phenylacetate 15, affected the cell lines differently. All the alkynylacyl esters (10, 11, and 12) showed no or only marginal effects on all cell lines. Even though ω-bromo compounds 31, 32, 33, and 34 appear more active, their potency is inferior to 2 and display no regularity concerning chain length and activity.
In summary, 5 appeared as the most cytotoxic compound in this series and, therefore, was investigated pharmacologically in more detail. First, 5 was also tested in nontumorigenic HEK293 cells. As shown in Table 2, the IC 50 value after 72 h was 3.4 µM, which is 2.3-fold higher than in WM9 cells, but also 1.3-fold lower than in WM164 cells. The cytotoxicity of chemotherapeutics against non-tumorigenic cells is a known problem in cancer therapy and one reason of undesirable side effects. For example, vinblastine and doxorubicin are two well-known and often clinically administered chemotherapeutics. In in vitro experiments, both also show cytotoxicity towards non-tumor cells to a similar or even greater extent [14,51,52]. However, it is difficult to extrapolate from in vitro to in vivo effects. Nevertheless, there is a certain risk that 5 will also cause side effects as they are already known for clinically used chemotherapeutics. However, it has already been shown for shikonin that there are ways to overcome this problem. Fayez et al. [53] reported recently that a combination of shikonin and silver nanoparticles synergistically inhibited the growth of lung cancer cells. Wang et al. [54] have shown that shikonin and JQ1, a bromodomain and extra-terminal motif (BET) inhibitor, encapsulated in lactoferrin nanoparticles changed the tumor immune microenvironment, activated immunogenic cell death, repolarized protumor phenotype, tumor-associated macrophages, and repressed glucose metabolism. They concluded that their system can be developed as a novel cancer immunotherapy due to several synergistic advantages. Another strategy could be a further development of our compounds to oxime derivatives. Huang et al. [55] demonstrated that such derivatives exhibited strong cytotoxicity towards cancer cells, but only a low cytotoxicity towards human skin fibroblasts.

Pharmacological Effects of Cyclopropyloxoacetate 5
To investigate the effects of 5 in more detail, several pharmacological assays were performed. Since 5 was most active in the metastatic cell line WM9 (BRAF mutated), we decided to use this cell line and, in addition, another BRAF mutated cell line (WM164) for comparison. Using the ApoToxGlo™ Triplex Assay, we investigated the effects regarding cell viability, cytotoxicity, and apoptosis induction in more detail ( Figure 6). The advantage of this assay is the combination of three assays in one single assay well. It has already been shown by our [14,16] and other groups [8,9] that several shikonin derivatives induced apoptosis in a variety of cancer types. We treated the cells with up to 20 µM of 5 for up to 48 h. Regarding cell viability, we found no statistically significant changes up 10 µM after 4 h and 24 h and up to 5 µM after 48 h. In the case of cytotoxicity, the fluorescence intensity decreased time-and dose-dependently, which can be an indicator for primary necrosis in combination with a reduced viability. Concerning apoptosis, we found a clear increase in caspase 3/7 activity after 24 h and 48 h in WM9 cells and a slight increase after 24 h in WM164. This agrees with the effects of our previous hit 3 [17] even if the activation of caspase 3/7 was weaker in case of 5. In summary, our results indicate that caspases are activated during the treatment with 5 further indicating that the cells undergo apoptotic cell death.
To investigate necrosis induction, we performed the CytoTox 96 ® Non-Radioactive Cytotoxicity Assay (LDH assay) (Figure 7). Lactate dehydrogenase (LDH) is released into the cell culture medium when the cell membrane is damaged. Therefore, this enzyme can be used as a marker for measuring necrotic cell death [56]. Cells were treated with 5 with up to 20 µM for up to 72 h. We found no LDH release up to 10 µM, which is 6.7-fold higher than the IC 50 in WM9 cells and 2.2-fold higher than the IC 50 in WM164 cells after 72 h. When the cells were treated with 20 µM 5, a slight increase of LDH release was found. However, compared to the maximal possible LDH release during complete cell lysis, the measured amount of LDH released by treated cells was generally quite low. In addition, these changes at 20 µM were not statistically significant in WM164 cells: 24 h: p = 0.182; 48 h: p = 0.115; 72 h: p = 0.269. In WM9 cells, the changes were statistically significant only at the highest tested concentration of 20 µM (p = 0.0197). In summary, necrosis seems to play only a minor role during the observed cells death at the concentrations tested. Nevertheless, it has been reported for shikonin that it induced necrosis in cancer cells such as lung and gastric cancer cells [57,58]. Cellular lysis appears also during the process of necroptosis [59] which has also been reported for shikonin [60]. Therefore, the increase of LDH release could also point to the induction of necroptosis by 5, which should be investigated in future studies.

Chemicals
Shikonin was purchased from Chengdu Biopurify Phytochemicals Ltd. (Chengdu, China). β,β-Dimethylacrylshikonin (2) was isolated from dried roots of Onosma paniculata Bureau & Franchet (Boraginaceae) and identified as reported previously [14]. Finally, we investigated the effect of 5 on the cell cycle. It has been reported that shikonin derivatives are able to bind to tubulin and, therefore, lead to cell cycle arrest [43,61]. Also, in the case of 2, we found a cell cycle arrest in different types of melanoma cell lines [14]. Therefore, we treated the cells with up to 20 µM of 5 for up to 48 h (Figure 8). Only at higher concentrations, 5 changed the cell cycle distribution statistically significant (Table 3). We also tested the effect of 5 µM 5 on the cell cycle. However, 5 had no effect on the cell cycle at this concentration (data not shown). Our results agree with other studies since Baloch et al. [43] reported a quite high IC 50 of 25.28 µM for shikonin concerning its inhibitory effect on tubulin polymerization.

Synthesis of Shikonin Derivatives
The synthesis of shikonin derivatives is described below. Their NMR spectra can be found in the supplementary material, chapter 2. The purity of all compounds was analyzed using NMR experiments and always exceeded 95%. LC-ESI-MS measurements were performed on a Dionex Ultimate 3000 UHPLC (Thermo, San José, CA, USA). It was coupled with a Thermo LTQ XL linear ion trap mass spectrometer equipped with an H-ESI II probe (negative mode). The acquisition wavelength was 500 nm, source heater temperature: 250 • C, capillary temperature: 200 • C, source voltage: 3.5 kV, sheath gas flow: 50 arbitrary units, capillary voltage: −14 V, and auxiliary gas flow: 10 arbitrary units. A Kinetex C18 column (2.6 µm, 100 × 2.10 mm, Phenomenex, Torrance, CA, USA) was used as stationary phase. Water (A) and acetonitrile (B) were used as mobile phases (gradient program: 0-45 min: 55-100% B, flow rate: 0.2 mL/min, column temperature: 30 • C).

General Procedure for the Acylation of Shikonin
A solution of shikonin in abs. CH 2 Cl 2 (0.1 mmol/5 mL) was cooled to 0 • C under argon atmosphere and DCC was added. After 15 min of stirring, DMAP was added. After an additional 15 min stirring, the corresponding acid was added and stirred for another 5.5 h to 5 days with slowly warming up to room temperature. Afterwards, 1 mL cyclohexane/0.1 mmol shikonin was added and the mixture was concentrated at room temperature and under reduced pressure to ca. 0.5 mL/0.1 mmol shikonin. The mixture was then filtered over 3 mm silica and 2 mm celite ® (eluent: petroleum ether/CH 2 Cl 2 = 1:0 to 1:2). The resulting fractions were evaporated and subjected to flash CC and/or repeated PTLC (cyclohexane/CH 2 Cl 2 mixtures). Cl 2 ); 9, yield: 17%. 9: R f = 0.31 (silica, CH 2 Cl 2 ); 1 H-NMR data fit with literature values [26] and the 13 C-NMR data fit with those of the corresponding alkannin derivative [25].

ApoToxGlo TM Triplex Assay
The ApoToxGlo TM Triplex Assay was purchased from Promega (Fitchburg, WI, USA, cat. no. G6320) and performed according to the manufacturer's instructions. In brief, 10,000 cells/well (WM9 or MUG-Mel2 cells) in 100 µL of medium were pipetted in 96well plates (white plates, flat bottom), incubated for 24 h and then treated with different concentrations of 5 for 4 h, 24 h and 48 h. Staurosporine (Abcam, Cambridge, UK) at a concentration of 25.0 µM is known to induce apoptosis and served as positive control. Subsequently, the viability/cytotoxicity reagent was prepared and 20 µL added to each well. After mixing for 30 s (orbital shaking, 300-500 rpm), the plates were incubated for another 30 min at 37 • C. Fluorescence was then measured at 400 Ex /505 Em (viability) and 485 Ex /520 Em (cytotoxicity) using a Hidex Sense Microplate Reader. Afterwards, the Caspase-Glo ® 3/7 reagent was prepared and 100 µ were added to each well, followed by mixing (orbital shaking, 300-500 rpm, 30 s), and incubation at room temperature for another 30 min. Finally, luminescence was measured using a Hidex Sense Microplate Reader. The assay was performed at least two times, with three replicates each.

LDH Assay
To measure LDH release, the CytoTox 96 ® Non-Radioactive Cytotoxicity Assay (Promega) were used in accordance with the manufacturers protocol. Cell suspension of 100,000 c/mL were prepared and seeded into 96 well plates (clear, flat bottom, 100 µL/well). To allow the cells to adhere, cells were incubated for 24 h before the treatment was started. Afterwards, cells were treated with 1 µM, 5 µM, 10 µM, or 20 µM of 5 for 24 h, 48 h, or 72 h. To measure maximum LDH release, cells were lysed using the lyse reagent included in the assay kit. To quantify LDH release, the plates were centrifuged and 50 µL of each well were transferred into another 96 well plate. After addition of 50 µL CytoTox96 Reagent for 30 min, 50 µL of the stop solution were added and absorption were measured at 490 nm using a Hidex Sense Microplate Reader. Finally, using the following formula, the amount of LDH release was calculated: 100 × OD 490 test well/OD 490 maximum LDH release.

Cell Cycle Analysis
For cell cycle analysis, 2 mL of a 150,000 c/mL cell suspension were seeded into 6 well plates and incubated for 24 h. Afterwards, the cells were treated with 10 µM or 20 µM of 5 for 24 h and 48 h. The cells were collected including supernatants and by using trypsinization, centrifugation, and washing with RPMI cell culture medium. Finally, cells were resuspended in 500 µL PBS and fixed with 5 mL ice cold ethanol for 10 min. For FACS analyses, cells were centrifuged and resuspended with 200 µL PI lysis buffer. After incubation for 20 min in the dark, cells were analyzed using a LSRII™flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA) and ModFit software.

Conclusions
In previous studies, shikonin and derivatives have been demonstrated to be potent cytotoxic substances. In this study, we prepared 31 shikonin derivatives by synthesis. Most of them are novel derivatives, which have not yet been reported. One goal was to further optimize our previous hit 3. Another goal was to synthesize a broad spectrum of structural features to gain more insight into the structure-activity relationship. All derivatives were screened for their cytotoxicity in several melanoma cell lines. The results indicate that there is no strict structure-activity relationship and the different cell lines exhibited distinct sensitivities towards the derivatives. The most potent derivative was 5, which is a cyclopropyloxoacetate derivative of shikonin and, thus, structurally related to our previous hit 3. Compared to 3, 5 was more cytotoxic. Subsequent pharmacological investigations revealed that 5 leads to caspase 3/7 activation, no significant LDH release, and to a G2/M phase cell cycle arrest at higher concentrations. Nevertheless, it was also cytotoxic towards non-tumorigenic cells, which needs to be evaluated in more detail in future studies. In summary, our results indicate that shikonin derivatives might be potential drug leads for the development of novel melanoma treatment options.