Synthesis, In Vitro Antiproliferative Activity, and In Silico Evaluation of Novel Oxiranyl-Quinoxaline Derivatives

The quinoxaline core is a promising scaffold in medicinal chemistry. Multiple quinoxaline derivatives, such as the topoisomerase IIβ inhibitor XK-469 and the tissue transglutaminase 2 inhibitor GK-13, have been evaluated for their antiproliferative activity. Previous work reported that quinoxaline derivatives bearing an oxirane ring present antiproliferative properties against neuroblastoma cell lines SK-N-SH and IMR-32. Likewise, quinoxalines with an arylethynyl group displayed promising antineoplastic properties against glioblastoma and lung cancer cell lines, U87-MG and A549 respectively. Here, 40 new quinoxaline derivatives bearing an oxirane ring were synthesized using a tetrakis(dimethylamino)ethylene (TDAE) strategy and a Sonogashira cross-coupling reaction. Each reaction with TDAE furnished a pair of diastereoisomers cis and trans. These new compounds formed two series according to the substitution of position 2 on the quinoxaline core, with chlorine or phenylacetylene respectively. Each of these isomers was evaluated for antiproliferative activity against neuroblastoma cell lines SK-N-SH and IMR-32 by MTT assay. All cell viability assay results were analyzed using R programming, as well as a statistical comparison between groups of compounds. Our evaluation showed no difference in drug sensitivity between the two neuroblastoma cell lines. Moreover, trans derivatives were observed to display better activities than cis derivatives, leading us to conclude that stereochemistry plays an important role in the antiproliferative activity of these compounds. Further support for this hypothesis is provided by the lack of improvement in antineoplastic activity following the addition of the phenylacetylene moiety, probably due to steric hindrance. As a result, compounds with nitrofuran substituents from the TDAE series demonstrated the highest antiproliferative activity with IC50 = 2.49 ± 1.33 μM and IC50 = 3.96 ± 2.03 μM for compound 11a and IC50 = 5.3 ± 2.12 μM and IC50 = 7.12 ± 1.59 μM for compound 11b against SK-N-SH and IMR-32, respectively. Furthermore, an in silico study was carried out to evaluate the mechanism of action of our lead compounds and predict their pharmacokinetic properties.


Introduction
Neuroblastoma is a neuroendocrine tumor of the sympathetic nervous system that develops from immature nerve tissue cells called neuroblasts. With 90% of cases diagnosed under 5 years old, it is the most common extra-cranial solid tumor, and the 4th

Chemistry
A first series (TDAE series) of epoxide diastereoisomers, with chlorine in position 2 of the quinoxaline core, was synthesized from 2-chloro-3-(dibromomethyl)quinoxaline  Previous work reported that quinoxaline derivatives bearing an oxirane ring showed interesting antiproliferative activity, with IC 50 = 3.9 ± 0.2 µM and IC 50 = 5.0 ± 0.9 µM for the most active compound, similar to the reference XK-469 (IC 50 = 4.6 ± 1.0 µM and IC 50 = 13.0 ± 2.9 µM) against neuroblastoma cell lines SK-N-SH and IMR-32, respectively [22]. Epoxides are an appropriate choice in the design of anticancer agents as they resemble aziridine ring, a well-known class of cytotoxic agents [23]. Another work reported that quinoxaline derivatives bearing an arylethynyl moiety display IC 50 of 3 µM against glioblastoma and lung cancer cell lines, U87-MG and A549 respectively [24]. Since our research activity focuses on the preparation of new potentially bioactive compounds [25][26][27], this work aimed to synthesize and evaluate from both in vitro and in silico perspectives, novel quinoxaline derivatives. These derivatives bear an oxirane ring substituted by a variety of aromatic and non-aromatic groups to evaluate the influence of this substitution. To determine whether the presence of an arylethynyl moiety can improve the antiproliferative activity of oxiranyl-quinoxaline derivatives against neuroblastoma cell lines SK-N-SH and IMR-32, two series were evaluated in this work with different substitutions on the position 2 of the quinoxaline core.

Chemistry
A first series (TDAE series) of epoxide diastereoisomers, with chlorine in position 2 of the quinoxaline core, was synthesized from 2-chloro-3-(dibromomethyl)quinoxaline and various carbonyl compounds using the organic electron donor TDAE to form the oxirane ring. The mechanism of this reaction is divided into two steps: firstly, TDAE in presence of dibromomethyl quinoxaline forms an anion that attacks the carbonyl to form an intermediate species; secondly, an intramolecular nucleophilic substitution (SNi) forms the epoxide ring ( Figure 2).

Chemistry
A first series (TDAE series) of epoxide diastereoisomers, with chlorine in position 2 of the quinoxaline core, was synthesized from 2-chloro-3-(dibromomethyl)quinoxaline and various carbonyl compounds using the organic electron donor TDAE to form the oxirane ring. The mechanism of this reaction is divided into two steps: firstly, TDAE in presence of dibromomethyl quinoxaline forms an anion that attacks the carbonyl to form an intermediate species; secondly, an intramolecular nucleophilic substitution (S N i) forms the epoxide ring ( Figure 2). Carbonyl compounds were chosen to cover a broad spectrum of chemical properties such as aromatic and non-aromatic groups, halogenated substituents in ortho, meta, para positions, electron-donating, and electron-withdrawing groups. This allowed us to obtain Carbonyl compounds were chosen to cover a broad spectrum of chemical properties such as aromatic and non-aromatic groups, halogenated substituents in ortho, meta, para positions, electron-donating, and electron-withdrawing groups. This allowed us to obtain a mixture of cis and trans diastereoisomers that led to 20 new compounds (2a-13) after purifying each isomer by flash chromatography (Figure 3). Proportions of each diastereoisomer formed, as determined by 1H-NMR, were distributed nearly 50/50 between cis and trans Pharmaceuticals 2022, 15, 781 4 of 23 isomers. For compounds 3, 6, and 12, the cis isomer could not be retrieved after purification of the reaction mixture. a mixture of cis and trans diastereoisomers that led to 20 new compounds (2a-13) after purifying each isomer by flash chromatography (Figure 3). Proportions of each diastereoisomer formed, as determined by 1 H-NMR, were distributed nearly 50/50 between cis and trans isomers. For compounds 3, 6, and 12, the cis isomer could not be retrieved after purification of the reaction mixture.
A second series (Sonogashira series) was obtained using a Sonogashira cross-coupling reaction with phenylacetylene on each purified isomer from the previous series. This enabled us to change chlorine in position 2 to the arylethynyl moiety, thereby obtaining 20 supplementary new compounds (14a-25) (Figure 3). In the end, 40 novel quinoxaline derivatives were synthesized, enabling us to evaluate multiple parameters: firstly, the influence of the position 2 substitution of the quinoxaline core to determine whether the addition of an arylethynyl group to the oxirane ring improves antiproliferative activity; secondly, the influence of stereochemistry on this biological activity since we characterized a pair of diastereoisomers cis and trans; lastly, the influence of the variation of R-groups substituting the epoxide on the antiproliferative activity.

In Vitro Antiproliferative Activity Evaluation
The antiproliferative activity of all derivatives was evaluated by cell survival experiments against two neuroblastoma cell lines, SK-N-SH and IMR-32 respectively, using conventional tetrazolium reduction assay as per previous work [22,24]. Both cell lines are of human origin, but they differ by their primary origin site, their resistance profile, and A second series (Sonogashira series) was obtained using a Sonogashira cross-coupling reaction with phenylacetylene on each purified isomer from the previous series. This enabled us to change chlorine in position 2 to the arylethynyl moiety, thereby obtaining 20 supplementary new compounds (14a-25) ( Figure 3).
In the end, 40 novel quinoxaline derivatives were synthesized, enabling us to evaluate multiple parameters: firstly, the influence of the position 2 substitution of the quinoxaline core to determine whether the addition of an arylethynyl group to the oxirane ring improves antiproliferative activity; secondly, the influence of stereochemistry on this biological activity since we characterized a pair of diastereoisomers cis and trans; lastly, the influence of the variation of R-groups substituting the epoxide on the antiproliferative activity.

In Vitro Antiproliferative Activity Evaluation
The antiproliferative activity of all derivatives was evaluated by cell survival experiments against two neuroblastoma cell lines, SK-N-SH and IMR-32 respectively, using conventional tetrazolium reduction assay as per previous work [22,24]. Both cell lines are of human origin, but they differ by their primary origin site, their resistance profile, and oncogene amplification [28]. SK-N-SH are female cells expressing the efflux pump glycoprotein-p (P-gp) responsible for multi-drug resistance but do not display the MYCN oncogene amplification associated with poor prognosis. On the contrary, IMR-32 are male cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds.
The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 µM), low activity (between 30 and 100 µM), and no activity (>100 µM).
The highest tested concentration being 100 µM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 µM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing. oncogene amplification [28]. SK-N-SH are female cells expressing the efflux pump glycoprotein-p (P-gp) responsible for multi-drug resistance but do not display the MYCN oncogene amplification associated with poor prognosis. On the contrary, IMR-32 are male cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The highest tested concentration being 100 μM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 μM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing. oncogene amplification [28]. SK-N-SH are female cells expressing the efflux pump glycoprotein-p (P-gp) responsible for multi-drug resistance but do not display the MYCN oncogene amplification associated with poor prognosis. On the contrary, IMR-32 are male cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The highest tested concentration being 100 μM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 μM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing. oncogene amplification [28]. SK-N-SH are female cells expressing the efflux pump glycoprotein-p (P-gp) responsible for multi-drug resistance but do not display the MYCN oncogene amplification associated with poor prognosis. On the contrary, IMR-32 are male cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The highest tested concentration being 100 μM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 μM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing. oncogene amplification [28]. SK-N-SH are female cells expressing the efflux pump glycoprotein-p (P-gp) responsible for multi-drug resistance but do not display the MYCN oncogene amplification associated with poor prognosis. On the contrary, IMR-32 are male cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The highest tested concentration being 100 μM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 μM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing. protein-p (P-gp) responsible for multi-drug resistance but do not display the MYCN oncogene amplification associated with poor prognosis. On the contrary, IMR-32 are male cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The highest tested concentration being 100 μM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 μM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing. H cells that do not express the P-gp but display a native amplification of the MYCN oncogene [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The highest tested concentration being 100 μM, IC 50 was not reached for some compounds (4a, 16a, 16b, 17a, 17b, 18a, 19b, 21a, 21b, 22b). They are presented in Table 1 as IC 50 > 100 μM and were considered not to have antiproliferative activity. Therefore, these molecules were excluded from further statistical testing.  [29] involved in about 20% of neuroblastoma cases and are associated with advanced disease and unfavorable biology [2]. Thus, the comparison of the results between both cell lines will allow us to anticipate the drug resistance profile of our compounds. The obtained data were analyzed with R programming [30], to determine IC 50 (concentration that inhibits 50% of cell proliferation) of all compounds. Compounds were classified into three groups according to their IC 50 on both cell lines: good activity (<30 μM), low activity (between 30 and 100 μM), and no activity (>100 μM).
The trans isomers bearing an epoxide substituted with trifluoromethyl benzene did not display any antiproliferative activity against both cell lines in the TDAE (4a) or Sonogashira (16a) series. Out of the 19 other derivatives from the TDAE series, 12 resulted in total or partial loss of activity in the Sonogashira series by combining the arylethynyl moiety (16b,  17a, 17b, 18a, 19a, 19b, 21a, 21b, 22b, 23a, 23b, 24a). This combination resulted in the loss of activity for all halogen substituents (17a, 17b, 18a, 19a, 19b), except for fluorine in the meta position (20a, 20b), on which it had the opposite effect by improving its antiproliferative activity. These variations are likely due to the steric hindrance of the phenylacetylene group. In contrast, it improved the antiproliferative activity of the remaining 6 derivatives (14a,  14b, 15a, 20a, 20b, 25).
Since each molecule was tested against two different cell lines, we explored whether the antiproliferative activity of our derivatives was significantly different between them. Data visualization of all MTT assay results that were carried out ( Figure 4) did not reveal obvious differences in IC 50 distribution between cell lines. Statistical analysis led us to the conclusion that antiproliferative activities did not significantly differ between SK-N-SH and IMR-32, with p-value = 0.09. This result suggests that our compounds are probably not substrates of the efflux pump P-gp responsible for drug resistance. From this conclusion, we did not further consider IC 50 against SK-N-SH and IMR-32 apart for statistical testing.  Visualization of the distribution between diastereoisomers of both series as above seems to indicate a difference between them ( Figure 5). Statistical comparison of TDAE series distribution and Sonogashira's returned p-value < 0.01. Regarding the difference in activity between cis and trans, the same test was performed and returned a p-value < 0.01. This significant difference in IC 50 suggested that trans isomers are significantly more active than cis isomers. All these results support our hypothesis that stereochemistry is key to our compounds' antiproliferative activity. Therefore, we divided into four groups each isomer from both series for further statistical comparisons: trans isomers of the TDAE series, cis isomers of the TDAE series, trans isomers of the Sonogashira series, and cis isomers of the Sonogashira series. Since compounds 13 and 25 are neither trans isomers nor cis ones, we compared them in each group according to their belonging series. Visualization of the distribution between diastereoisomers of both series as above seems to indicate a difference between them ( Figure 5). Statistical comparison of TDAE series distribution and Sonogashira's returned p-value < 0.01. Regarding the difference in activity between cis and trans, the same test was performed and returned a p-value < 0.01. This significant difference in IC 50 suggested that trans isomers are significantly more active than cis isomers. All these results support our hypothesis that stereochemistry is key to our compounds' antiproliferative activity. Therefore, we divided into four groups each isomer from both series for further statistical comparisons: trans isomers of the TDAE series, cis isomers of the TDAE series, trans isomers of the Sonogashira series, and cis isomers of the Sonogashira series. Since compounds 13 and 25 are neither trans isomers nor cis ones, we compared them in each group according to their belonging series.
Accordingly, when assessing the influence of epoxide substitution, we compared cytotoxic activity within the four previously formed groups. Unsurprisingly, these four tests returned significant p-values < 0.03 which led us to further investigate these differences by pairwise comparisons between all substituents using Dunn's test ( Figure 6). This significant difference in IC 50 suggested that trans isomers are significantly more active than cis isomers. All these results support our hypothesis that stereochemistry is key to our compounds' antiproliferative activity. Therefore, we divided into four groups each isomer from both series for further statistical comparisons: trans isomers of the TDAE series, cis isomers of the TDAE series, trans isomers of the Sonogashira series, and cis isomers of the Sonogashira series. Since compounds 13 and 25 are neither trans isomers nor cis ones, we compared them in each group according to their belonging series.  Accordingly, when assessing the influence of epoxide substitution, we compared cytotoxic activity within the four previously formed groups. Unsurprisingly, these four tests returned significant p-values < 0.03 which led us to further investigate these differences by pairwise comparisons between all substituents using Dunn's test ( Figure 6).
Among the trans isomers of the TDAE series, compounds 2a, 5a, 6a, 7a, and 11a displayed the highest activity. From these, halogenated benzene substituents (5a, 6a, 7a) displayed similar activities. Interestingly, fluorine in meta position (8a) differed significantly from them, with poorer cytotoxic activity. Unsubstituted benzene (2a) comparisons with the majority of the other chemical groups were inconclusive probably because of its important IC 50 standard deviation. It only showed significant differences with p-nitrobenzene (10a) and ester substituents (12a, 13), making them bad drug candidates; but also, with 5-nitrofuran (11a) which is a significantly better substituent. In the end, IC 50 of 5nitrofuran (11a), fluorine in para (7a) and chlorine in ortho (6a) substituents were not significantly different from each other, making them the best substituents in this group.
Among the cis isomers of the TDAE series, only the 5-nitrofuran substituent (11b) stood out from the others as the best one. Interestingly, in both cis and trans isomers, the introduction of the p-nitrobenzene ring (10a, 10b), as an analog of 5-nitrofuran (11a, 11b) on the epoxide ring, led to an almost complete loss of activity. This suggests that the antiproliferative activity of these compounds is not the result of the nitro group which is often associated with cytotoxic implications. Among the trans isomers of the TDAE series, compounds 2a, 5a, 6a, 7a, and 11a displayed the highest activity. From these, halogenated benzene substituents (5a, 6a, 7a) displayed similar activities. Interestingly, fluorine in meta position (8a) differed significantly from them, with poorer cytotoxic activity. Unsubstituted benzene (2a) comparisons with the majority of the other chemical groups were inconclusive probably because of its important IC 50 standard deviation. It only showed significant differences with p-nitrobenzene (10a) and ester substituents (12a, 13), making them bad drug candidates; but also, with 5nitrofuran (11a) which is a significantly better substituent. In the end, IC 50 of 5-nitrofuran (11a), fluorine in para (7a) and chlorine in ortho (6a) substituents were not significantly different from each other, making them the best substituents in this group.
Among the cis isomers of the TDAE series, only the 5-nitrofuran substituent (11b) stood out from the others as the best one. Interestingly, in both cis and trans isomers, the introduction of the p-nitrobenzene ring (10a, 10b), as an analog of 5-nitrofuran (11a, 11b) on the epoxide ring, led to an almost complete loss of activity. This suggests that the antiproliferative activity of these compounds is not the result of the nitro group which is often associated with cytotoxic implications.
Among the trans isomers of the Sonogashira series, 3 substituents stood out as the most active ones with no difference between them: unsubstituted benzene (14a), m-fluorobenzene (20a), and 5-nitrofuran substituent (23a). There is also no significant difference between these isomers and the ester derivative 25 which also has good activity. Comparisons to p-methylbenzene (15a) are not very conclusive when categorizing it in this good activity compounds group since it did not show a significant difference with p-nitrobenzene (22a) and p-fluorobenzene (19a), compounds that did not display any activity on SK-N-SH cell line. Interestingly, the trans isomer with the ester substituent 24a is significantly less active than its analog 25. This difference could be explained by higher lipophilicity of 25 than 24a.
Lastly, m-fluorobenzene (20b) and ester (25) appear the best compounds from the cis isomers of the Sonogashira series comparison group. Unsubstituted benzene (14b) and 5-nitrofuran (23b) substituents are significantly less active than the carboxylate 25, but still might be grouped as active compounds with IC 50 < 30 µM.

Molecular Docking
To predict the molecular mechanism involved in the antiproliferative activity of these molecules, molecular docking of best compounds 11a and 11b from the TDAE series, and best compounds 14a and 25 from the Sonogashira series, was performed on both crystallographic structures of human Topoisomerase IIβ (3QX3) and human Tissue Transglutaminase (4PYG) obtained from the Protein Data Bank. We selected these drug representatives because they have the lowest IC 50 of the TDAE and Sonogashira series, respectively. Molecular docking was also performed on 3QX3 for compound XK-469, and on 4PYG for compound GK13 as references on each protein, respectively. Since no crystallized structure of topoisomerase complexed with XK-469 is available, the 3QX3 entry was chosen for Topoisomerase IIβ because of its complexation with a well-known topoisomerase II inhibitor, etoposide [31]. Similarly, since no crystallized structure of human tissue transglutaminase combined with GK13 was found, the 4PYG entry was chosen for this protein because of its complexation with guanosine triphosphate (GTP) since this enzyme notably possesses a GTPase enzymatic [32]. To perform molecular docking, affinities between compounds and protein targets were calculated by a "blind-docking approach" without "a priori" binding site information, for all our docking simulations [33,34]. We evaluated the binding modes of these compounds to each protein and calculated the binding affinities using the open-source program AutoDock Vina [35,36] which uses a scoring function relying on the Broyden-Fletcher-Goldfarb-Shanno algorithm for the local optimization. The quality of protein-ligand interactions was visually examined from the resulting conformation binding mode of the lowest level of Gibbs free energy of binding (∆G).
On topoisomerase, this docking experiment revealed a very similar site of binding, close to the catalytic site of DNA cleavage, for all our compounds and XK-469 (Figure 7). This binding to the 3QX3 protein appears to be with high energy (Table 2), mostly by hydrophobic interactions. In particular, XK-469 interacts with 3QX3 primarily through Van Der Waals interaction with Tyr821 which has been described as an essential residue for the physiological activity of this enzyme [37]. All compounds interact with Lys759 which belongs to the winged helix domain (WHD) of the protein containing the catalytic tyrosine. Compound 11a also displays Glu769 and Ala768 in common with 14a and XK-469. Likewise, compound 11b displays Ala768 in common with our drug reference. Representatives from the Sonogashira series 14a and 25 both interact in the same way with His774 and Phe823, which are also amino acids interacting with 11b. In particular, compound 25 has its interaction with Gln762 in common with XK-469. All these amino acids belong to the WHD domain of the Topoisomerase IIβ which could be an explanation for the mechanism of action of these compounds. As all our compounds interact, mainly through the same residues from the same domain as the proven antitopoisomerase XK-469, we strongly suggest that they share the same mechanism of action. Regarding the molecular docking of these compounds on human tissue transglutaminase, all ligands but 25 interact with the protein in the same binding site as the reference inhibitor GK13. The binding energies were also in the same range as our reference ( Table 3). The amino acids involved in this interaction are Lys176, Ile178, Arg433, Asn586, Glu588, Lys677, and Phe679, all by hydrophobic linking for compound GK13. (Figure 8). These amino acids belonging to the catalytic core of 4PYG could explain the inhibitory effect of GK13. Compound 14a shares Lys176, Lys677, and Phe679 with GK13 in its interaction with the protein, also through Van der Waals linking. Surprisingly, compounds 11a and 11b from the TDAE series, which do not have the arylethynyl moiety, were also able to bind the protein domain like our reference. As for compound 25, molecular docking revealed that it does not share the same protein binding domain as all other compounds. Indeed, this derivative from the Sonogashira series interacts with the protein through Val249, Ser250, Ser253, Thr621, Thr623, and Glu669. action with the protein, also through Van der Waals linking. Surprisingly, compounds 11a and 11b from the TDAE series, which do not have the arylethynyl moiety, were also able to bind the protein domain like our reference. As for compound 25, molecular docking revealed that it does not share the same protein binding domain as all other compounds. Indeed, this derivative from the Sonogashira series interacts with the protein through Val249, Ser250, Ser253, Thr621, Thr623, and Glu669.

ADMET Predictions
Pharmacokinetics (PK) parameters are essential in the selection process of hit compounds in a hit-to-lead approach. Thus, we evaluated in silico these features with Simulations Plus software, ADMET Predictor ®, and GastroPlus ® . The results of this evaluation

ADMET Predictions
Pharmacokinetics (PK) parameters are essential in the selection process of hit compounds in a hit-to-lead approach. Thus, we evaluated in silico these features with Simulations Plus software, ADMET Predictor ®, and GastroPlus ® . The results of this evaluation did not reveal any differences between cis and trans diastereoisomers since the software does not consider conformation during its calculations. Therefore, they are presented as one molecule in this section and not as a and b compounds.
None of these compounds violate Lipinski's "Rule of Five", also known as Pfizer's rule of five, for potential drug candidates (Table 3). These rules allow us to evaluate the drug likeliness of our compounds. They state that most orally active drugs with good bioavailability have no more than one violation of its four criteria: molecular weight (MW) ≥ 500 Da, limited lipophilicity expressed as logP ≥ 5, number of hydrogen bond acceptors ≥ 10, and number of hydrogen bond donors (H-BD) ≥ 5. In addition, simulations of logD at a physiological pH of 7.4 returned identical values to logP, as presented in Table 3. These parameters were calculated with Simulations Plus software ADMET Predictor ® .
During the absorption phase, this modeling revealed that compounds from the TDAE series result in a predicted bioavailability (F) ranging from 82.71% for compounds 13 to 32.69% for compounds 4. The hit compounds 11 returned a bioavailability value of 35.48% from this simulation. For aromatic substituted epoxide derivatives from the Sonogashira series, F values were ranging from 12.56% for compounds 14 to 1.535 for compounds 21. In this series, only the non-aromatic substituted epoxide derivatives displayed high bioavailability with 90.05% for compounds 24 and 89.24% for compound 25. This loss of absorption between TDAE and Sonogashira series derivatives can be explained by the increase of drug lipophilicity when adding the arylethynyl moiety.
During the distribution phase, all our derivatives are susceptible to being bound by proteins in the plasma such as albumin because of their lipophilicity. Percentages of unbound drug to blood plasma proteins were less than 10%, except for compounds 11, 12, and 13, which returned 15%, 22%, and 20%, respectively. These percentages range from 71% for compound 14 to 99% for our hit compound 11. Furthermore, corroborating our in vitro results, none of our compounds are susceptible to being substrates of the P-gp except for compound 25. Interestingly, this compound is susceptible with 97% accuracy to being a P-gp inhibitor. Similarly, this modeling suggests that all compounds in the Sonogashira series are susceptible to be P-gp inhibitors.
According to our compounds' metabolism phase prediction, all derivatives are susceptible to being substrates of the superfamily of cytochromes P450 (CYPs). This metabolism may result in various metabolites with hydroxylation of the quinoxaline core or even the oxirane ring with its opening (Figure 9).  During the absorption phase, this modeling revealed that compounds from the TDAE series result in a predicted bioavailability (F) ranging from 82.71% for compounds 13 to 32.69% for compounds 4. The hit compounds 11 returned a bioavailability value of 35.48% from this simulation. For aromatic substituted epoxide derivatives from the Sonogashira series, F values were ranging from 12.56% for compounds 14 to 1.535 for compounds 21. In this series, only the non-aromatic substituted epoxide derivatives displayed high bioavailability with 90.05% for compounds 24 and 89.24% for compound 25. This loss of absorption between TDAE and Sonogashira series derivatives can be explained by the increase of drug lipophilicity when adding the arylethynyl moiety.
During the distribution phase, all our derivatives are susceptible to being bound by proteins in the plasma such as albumin because of their lipophilicity. Percentages of unbound drug to blood plasma proteins were less than 10%, except for compounds 11, 12, and 13, which returned 15%, 22%, and 20%, respectively. These percentages range from 71% for compound 14 to 99% for our hit compound 11. Furthermore, corroborating our in vitro results, none of our compounds are susceptible to being substrates of the P-gp except for compound 25. Interestingly, this compound is susceptible with 97% accuracy to being a P-gp inhibitor. Similarly, this modeling suggests that all compounds in the Sonogashira series are susceptible to be P-gp inhibitors.
According to our compounds' metabolism phase prediction, all derivatives are susceptible to being substrates of the superfamily of cytochromes P450 (CYPs). This metabolism may result in various metabolites with hydroxylation of the quinoxaline core or even the oxirane ring with its opening (Figure 9).  AUC values predicted were ranging from 3819.6 ng-h/mL (8) to 24000 ng-h/mL (10) for compounds from the TDAE series, and from 495.3 ng-h/mL (21) to 180000 ng-h/mL (14) for compounds from the Sonogashira series.

Generality
Melting points were determined on a Büchi melting point apparatus (BUCHI Corporation, New Castle, United States) and were uncorrected. High-resolution mass spectrometry analyses were carried out at the Spectropôle, Faculté des Sciences de Saint-Jérôme (Marseille, France) with a mass spectrometer SYNAPT G2 HDMS Waters (Milford, MA, United States) equipped with an electrospray ionization source (electrospray tension: 2.8 kV; orifice tension: 20 V; nebulization gas flow (nitrogen): 100 L/h). Samples were dissolved in 300 µL of dichloromethane, diluted at 1/102 in methanol solution at 0.1 mM sodium chloride, and introduced into the ionization source at 10 µL/min. High-resolution mass spectra were obtained with a time-of-flight (TOF) analyzer. Exact mass measurements were repeated in triplicate with external calibration. NMR spectra were recorded on a Bruker Avance NEO 400 MHz NanoBay spectrometer at the Faculté de Pharmacie of Marseille. (1H-NMR: reference CDCl 3 δ = 7.26 ppm, reference DMSO-d6 δ = 2.50 ppm and 13C-NMR: reference CDCl 3 δ = 76.9 ppm, reference DMSO-d6 δ = 39.52 ppm). They are presented in Supplementary Materials. TLC was performed on 5 cm × 10 cm aluminum plates coated with silica gel 60F-254 (Merck) in an appropriate eluent. Visualization was performed with ultraviolet light (234 nm). Reagents were purchased and used without further purifications from Sigma-Aldrich or Fluorochem. Ultra-High Performance Liquid Chromatography (UHPLC) analyses were performed using an Agilent 1290 Series apparatus (binary pump G4220A, autosampler G1330B, column oven G1316C, photodiode array detector G4212A).

General Procedure for Compounds 2 to 13
To 2-chloro-3-(dibromomethyl)quinoxaline (1) (1 g, 2.97 mmol), appropriate carbonyl derivative (5.94 mmol, 2 eq.) in THF (20 mL) was added in a two-necked flask under inert gas. The reaction mixture was stirred for 1 h at −25 • C and 2 h at room temperature. Then, the mixture was extracted in ethyl acetate (3 × 40 mL) and washed with H 2 O (3 × 40 mL) before being dried with sodium sulfate. Each diastereoisomer was then purified by flash chromatography puriFlash ® using an IR-50SI-F0080 regular silica column and a dichloromethane/cyclohexane gradient (40:60 to 60:40).  (12) We considered the molecule adopting the lowest energetic conformation as a promising compound. Visual analysis of the lowest energy solutions for each compound allowed us to identify the protein binding site. All the figures were drawn using the program ChimeraX.

Pharmacokinetics Modeling with Simulation Plus Software Suite
The drug database for pharmacokinetic modeling was set up with MedChem Stu-dio™ 4.0 from the Simulation-Plus software suite. Drug likeliness parameters were determined with ADMET Predictor ® 10.3. Pharmacokinetic parameters were determined with GastroPlus ® 9.8.2. From GastroPlus ® , a compartmental model was repeated for each drug as an administration to a 70 kg fasted human with normal gut physiology in a 100 mg immediate-release tablet dosage form.

Conclusions
In this work, three chemical aspects of our synthesized compounds were evaluated. Firstly, we demonstrated the influence of stereochemistry on the antiproliferative activity of our compounds. The trans derivatives were significantly more active than cis ones from both TDAE and Sonogashira series. Secondly, we evaluated the influence of the substitution of position 2 on the quinoxaline core. Combining the epoxide and the arylethynyl group within the same structure in the Sonogashira series improved the antiproliferative activity of 6 out of the 20 compounds synthesized in the TDAE series. Since this induced a loss of activity for the other quinoxaline derivatives, it seems to demonstrate that the activity of most compounds is negatively influenced by the steric hindrance from the 2arylethynyl substituent. Thirdly, we evaluated the influence of a variety of substituents on the oxirane ring in both the TDAE and Sonogashira series. As in the previously described TDAE series [22], the lowest IC 50 was observed for the derivatives on which the epoxide is substituted by 5-nitrofuran (11a, 11b). Our analysis also revealed that the nature of the epoxide's substituent and the substitution pattern of the benzene ring can have a considerable impact on the antiproliferative activity of the synthesized compounds. Indeed, halogenated phenyl (5, 6, 7) and 5-nitrofuran 11 seem to be the most appropriate options from the TDAE series. Likewise, unsubstituted benzene 14, fluorinated phenyl 20, 5nitrofuran 23, and carboxylate 25 are the most active compounds in the Sonogashira series.
Moreover, we evaluated each compound against two neuroblastoma cell lines that were different in many aspects, more specifically by their expression of the efflux pump P-gp and the MYCN gene amplification. Since no significant difference could be demonstrated between IC 50 against SK-N-SH and IMR-32, we could think that our compounds are not substrates of the P-gp which is an encouraging feature. Furthermore, most compounds are active against aggressive MYCN amplified cell line IMR-32, which is also an encouraging feature for further evaluations. In conclusion, we presented in this work multiple quinoxaline derivatives that display antiproliferative activity against resistant cell lines and aggressive ones.
Further work will allow us to dig into the mechanism of action of these molecules. From our work, several hypotheses are made. From the similarity of structure with compounds XK-469 and CQS, which are both topoisomerase IIβ inhibitors, we could think that our products have the same target. Based on the results of our molecular docking study and the structural similarity with compounds XK-469 and CQS, which are both topoisomerase IIβ inhibitors, we were able to suggest that our products have the same target. Similarly, it allowed us to identify another potential target: the tissue transglutaminase responsible for tumor resistance. According to other oxirane ring carrier molecules described in the literature [23], other mechanisms could be at stake such as intracellular epoxide opening generating reactive oxygen species inducing apoptosis or DNA alkylation.