Antiviral Activities of Halogenated Emodin Derivatives against Human Coronavirus NL63

The current COVID-19 outbreak has highlighted the need for the development of new vaccines and drugs to combat Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2). Recently, various drugs have been proposed as potentially effective against COVID-19, such as remdesivir, infliximab and imatinib. Natural plants have been used as an alternative source of drugs for thousands of years, and some of them are effective for the treatment of various viral diseases. Emodin (1,3,8-trihydroxy-6-methylanthracene-9,10-dione) is a biologically active anthraquinone with antiviral activity that is found in various plants. We studied the selectivity of electrophilic aromatic substitution reactions on an emodin core (halogenation, nitration and sulfonation), which resulted in a library of emodin derivatives. The main aim of this work was to carry out an initial evaluation of the potential to improve the activity of emodin against human coronavirus NL63 (HCoV-NL63) and also to generate a set of initial SAR guidelines. We have prepared emodin derivatives which displayed significant anti-HCoV-NL63 activity. We observed that halogenation of emodin can improve its antiviral activity. The most active compound in this study was the iodinated emodin analogue E_3I, whose anti-HCoV-NL63 activity was comparable to that of remdesivir. Evaluation of the emodin analogues also revealed some unwanted toxicity to Vero cells. Since new synthetic routes are now available that allow modification of the emodin structure, it is reasonable to expect that analogues with significantly improved anti-HCoV-NL63 activity and lowered toxicity may thus be generated.


Introduction
Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2) is a family of enveloped positive-sense RNA viruses that cause life-threatening respiratory infections and severe pneumonia in humans [1,2]. Coronavirus (CoV) entry into host cells (pulmonary and parabronchial epithelial cells) is mediated by spike protein, which is responsible for binding to receptors ACE-2 and mediating virus-host membrane fusion [3][4][5]. The development of effective antiviral drugs with a broad spectrum of activity has been hampered by viral diversity and the ability of SARS-CoV to mutate rapidly, even during an epidemic. It is therefore very important to develop antiviral drugs that effectively and safely inhibit the spread of SARS-CoV, or at least significantly alleviate the symptoms of SARS-CoV infection. In particular, the development of simple, small compounds that can be produced and administered inexpensively would be of great importance. Recently, several potential repurposed drugs against COVID-19 (SARS-CoV-2 virus) have been found, such as remdesivir, infliximab and imatinib. Remdesivir has potent antiviral activity and has already containing different halogen atoms and substitution patterns ( Figure 1) and the evaluation of their antiviral activity against HCoV-NL63. Our results demonstrate that the presence of different functional groups in the emodin scaffold has a significant impact on their anti-HCoV-NL63 activity. Herein, we report the strategies followed for the selective introduction of NO2, SO3H and halogen atoms into the anthraquinone ring of emodin. Taking into account the potency of halogenated emodins, our work focused on the selective decoration of emodin, containing different halogen atoms and substitution patterns ( Figure 1) and the evaluation of their antiviral activity against HCoV-NL63. Our results demonstrate that the presence of different functional groups in the emodin scaffold has a significant impact on their anti-HCoV-NL63 activity.

Synthesis of Emodin Derivatives
The natural product emodin E_H provides an entry point for the introduction of various functional groups on aromatic rings by electrophilic aromatic substitution. Our aim was to create a library of emodin derivatives by modifying the aromatic ring through halogenation, nitration, amination and sulfonation.
First, we investigated the selective halogenation of emodin E_H by classical reagents (N-chlorosuccinimide-NCS, N-bromosuccinimide-NBS, N-iodoosuccinimide-NIS) and by an alternative method-oxidative halogenation with hydrogen peroxide as oxidant. 2-Iodoemodin E_I was prepared according to the published methods [28] using I2 and NaHCO3 in 73% yield ( Table 1, entry 1) and in the presence of NIS reagent in 84% yield (Table 1, entry 6). Oxidative iodination with I2 and 30% hydrogen peroxide in 2-MeTHF also allowed the selective synthesis of the same product with a better yield of 93% (Table 1, entry 16). For the oxidative iodination, 2 equivalents of iodine and 2.5 equivalents of hydrogen peroxide were used. While conducting the experimental work, it was observed that the reaction does not proceed in the absence of the oxidant H2O2 (entry 11) and that a higher amount of iodine is required for the quantitative conversion to E_I (entries [12][13][14][15]. We also found that the amount of hydrogen peroxide has a minimal effect on the increase in conversion. A larger number of iodine substituents could not be introduced into the aromatic core of emodin using oxidative iodination. To introduce additional iodine atoms, I2/NaHCO3 or NIS had to be used. Nevertheless, the diiodinated emodin E_2I could not be selectively prepared, regardless of the amount of iodine or NIS used. 2,4,7-Triiodoemodin E_3I was selectively formed by applying the classical method (I2, Na-HCO3) [28] in 81% yield or using NIS reagent in 79% yield as a brown-orange powder ( Table 1, entries 3 and 10). We also attempted to synthesize the fully iodinated emodin E_4I, but this compound could not be prepared regardless of the reaction conditions. We tried increasing the temperature and adding the activator H2SO4 to the reagent NIS. In all cases, the triiodinated product appeared selectively. The positions of iodine on the aromatic rings in compounds E_I and E_3I were confirmed by 2D NMR spectroscopy (Figures S6-S13, Supplementary Information).

Synthesis of Emodin Derivatives
The natural product emodin E_H provides an entry point for the introduction of various functional groups on aromatic rings by electrophilic aromatic substitution. Our aim was to create a library of emodin derivatives by modifying the aromatic ring through halogenation, nitration, amination and sulfonation.
First, we investigated the selective halogenation of emodin E_H by classical reagents (N-chlorosuccinimide-NCS, N-bromosuccinimide-NBS, N-iodoosuccinimide-NIS) and by an alternative method-oxidative halogenation with hydrogen peroxide as oxidant. 2-Iodoemodin E_I was prepared according to the published methods [28] using I 2 and NaHCO 3 in 73% yield (Table 1, entry 1) and in the presence of NIS reagent in 84% yield (Table 1, entry 6). Oxidative iodination with I 2 and 30% hydrogen peroxide in 2-MeTHF also allowed the selective synthesis of the same product with a better yield of 93% (Table 1, entry 16). For the oxidative iodination, 2 equivalents of iodine and 2.5 equivalents of hydrogen peroxide were used. While conducting the experimental work, it was observed that the reaction does not proceed in the absence of the oxidant H 2 O 2 (entry 11) and that a higher amount of iodine is required for the quantitative conversion to E_I (entries 12-15). We also found that the amount of hydrogen peroxide has a minimal effect on the increase in conversion. A larger number of iodine substituents could not be introduced into the aromatic core of emodin using oxidative iodination. To introduce additional iodine atoms, I 2 /NaHCO 3 or NIS had to be used. Nevertheless, the diiodinated emodin E_2I could not be selectively prepared, regardless of the amount of iodine or NIS used. 2,4,7-Triiodoemodin E_3I was selectively formed by applying the classical method (I 2 , NaHCO 3 ) [28] in 81% yield or using NIS reagent in 79% yield as a brown-orange powder ( Table 1, entries 3 and 10). We also attempted to synthesize the fully iodinated emodin E_4I, but this compound could not be prepared regardless of the reaction conditions. We tried increasing the temperature and adding the activator H 2 SO 4 to the reagent NIS. In all cases, the triiodinated product appeared selectively. The positions of iodine on the aromatic rings in compounds E_I and E_3I were confirmed by 2D NMR spectroscopy (Figures S6-S13, Supplementary Material).
Next, bromination was carried out using NBS and by oxidative halogenation using H 2 O 2 /HBr. Neither of these methods could be effective for selective synthesis of monobromo-derivative E_Br, as shown by the results in Table 2. Although a reaction with 1 equivalent of NBS was carried out in an ice bath, both E_Br and E_2Br were formed simultaneously ( Table 2, entries 1 and 2). The same results were observed for oxidative bromination with HBr and H 2 O 2 . Regardless of the amounts of HBr and hydrogen peroxide used, a mixture of the two products was always formed. Using 2D NMR spectroscopy, we found that in the case of E_Br, bromine binds to site 2 in emodin, either when the NBS reagent or HBr/H 2 O 2 is used.  Next, bromination was carried out using NBS and by oxidative halogenation using H2O2/HBr. Neither of these methods could be effective for selective synthesis of monobromo-derivative E_Br, as shown by the results in Table 2. Although a reaction with 1 equivalent of NBS was carried out in an ice bath, both E_Br and E_2Br were formed simultaneously ( Table 2, entries 1 and 2). The same results were observed for oxidative bromination with HBr and H2O2. Regardless of the amounts of HBr and hydrogen peroxide used, a mixture of the two products was always formed. Using 2D NMR spectroscopy, we found that in the case of E_Br, bromine binds to site 2 in emodin, either when the NBS reagent or HBr/H2O2 is used.
E_2Br was selectively prepared according to a published procedure [29] with a slight modification consisting of treating emodin E_H with NBS in THF at 0 °C (88% yield). Due to the high reactivity of emodin, the bromination required a relatively short reaction time (30 min) and a low temperature (0 °C). The same product was obtained by the oxidative halogenation method (HBr, H2O2) in 2,2,2 trifluoroethanol (TFE) in 91% yield ( Table 2,  entry 14). Unfortunately, the oxidative method, although more environmentally friendly, did not allow the introduction of more bromine atoms. Despite the higher amount of HBr and the use of the activating solvent TFE, the reaction stopped at the dibrominated product. The use of an NBS reagent allowed a greater number of bromine atoms to be introduced, but we encountered problems with the selectivity in the preparation of E_3Br. Regardless of the temperature at which the reaction was carried out and the amount of reagent used, E_2Br and/or E_4Br were also formed ( Table 2, entries 4-6). Selectively, we prepared the orange-colored product 2,4,6,8-tetrabromo-1,3,5-trihydroxy-7-methylanthracene-9,10-dione E_4Br (   Chlorination of emodin gave similar results to bromination. The monosubstituted product could not be prepared selectively with the NCS reagent or oxidatively with HCl/H2O2, since the dichloro derivative E_2Cl was also formed. It was observed that the regioselectivity of chlorination to E_Cl depends on the method used-HCl/H2O2 or E_2Br was selectively prepared according to a published procedure [29] with a slight modification consisting of treating emodin E_H with NBS in THF at 0 • C (88% yield). Due to the high reactivity of emodin, the bromination required a relatively short reaction time (30 min) and a low temperature (0 • C). The same product was obtained by the oxidative halogenation method (HBr, H 2 O 2 ) in 2,2,2 trifluoroethanol (TFE) in 91% yield ( Table 2,  entry 14). Unfortunately, the oxidative method, although more environmentally friendly, did not allow the introduction of more bromine atoms. Despite the higher amount of HBr and the use of the activating solvent TFE, the reaction stopped at the dibrominated product. The use of an NBS reagent allowed a greater number of bromine atoms to be introduced, but we encountered problems with the selectivity in the preparation of E_3Br. Regardless of the temperature at which the reaction was carried out and the amount of reagent used, E_2Br and/or E_4Br were also formed ( Table 2, entries 4-6). Selectively, we prepared the orange-colored product 2,4,6,8-tetrabromo-1,3,5-trihydroxy-7-methylanthracene-9,10-dione E_4Br (Table 2, entry 7).
Chlorination of emodin gave similar results to bromination. The monosubstituted product could not be prepared selectively with the NCS reagent or oxidatively with HCl/H 2 O 2 , since the dichloro derivative E_2Cl was also formed. It was observed that the regioselectivity of chlorination to E_Cl depends on the method used-HCl/H 2 O 2 or NCS-as determined by conducting 1D and 2D NMR spectroscopy on the crude reaction mixture consisting of E_H, E Cl or/and E_2Cl. The results showed that when NCS was used, the Cl was bound at position 2 (E_Cl-2), whereas when HCl/H 2 O 2 was used, it was bound at position 4 (E_Cl-4) (Figures S1-S3). Good yields of the disubstituted emodin 2,4-dichloroemodin E_2Cl were obtained selectively with NCS [30] using H 2 SO 4 as catalyst in a 3-h process at room temperature (Table 3, entry 3). When emodin was treated with 6 equivalents of NCS in the presence of H 2 SO 4 at a reflux temperature, no formation of three-or four-chloroemodin occurred. Only two chlorine atoms could be introduced into emodin. Oxidative chlorination was proved to be an unsuitable procedure for the selective synthesis of chlorine products. Despite the combination of different ratios of HCl and H 2 O 2 , emodin was always converted into a mixture of different products that could not be separated (Table 3, entries 5-9).  HCl (5) Next, we investigated the selective nitration of E_H. We attempted to selectively prepare nitroemodin with one, two or three NO2 functional groups, but none of the conditions tested led to successful results. The results showed that despite the use of small amounts of nitric and sulfuric acids, a mixture of various nitrated products E_X formed that could not be separated (Table 4, entries 1-7). It was concluded that the nitration reaction is not selective towards a particular product despite the absence of H2SO4. We prepared the tetranitroemodin E_4NO2 using 6.2 equiv. HNO3 and 10 equiv. H2SO4 in 91% Entry

Reagent (Equiv.) Solvent Conditions
Relative Distribution a E_H E_Cl E_2Cl E_3Cl HCl (1) Next, we investigated the selective nitration of E_H. We attempted to selectively prepare nitroemodin with one, two or three NO 2 functional groups, but none of the conditions tested led to successful results. The results showed that despite the use of small amounts of nitric and sulfuric acids, a mixture of various nitrated products E_X formed that could not be separated (Table 4, entries 1-7). It was concluded that the nitration reaction is not selective towards a particular product despite the absence of H 2 SO 4 . We prepared the tetranitroemodin E_4NO 2 using 6.2 equiv. HNO 3 and 10 equiv. H 2 SO 4 in 91% yield (Table 4, entry 8), reducing excess reagents by an order of magnitude according to a published method [24] (Table 4, entry 9). In addition, we also prepared amino-substituted emodin E_NH 2 by the method described in an earlier publication ( Figure 2) [24]. Sulfonation was carried out with oleum at room temperature. After 24 h, the reaction mixture was analyzed by NMR; three different products were observed. According to the NMR spectra obtained from the crude reaction mixture, trisulfonated, disulfonated and monosulfonated emodin were formed. However, the composition of the mixture changed with time and reverse sulfonation occurred. Hence, following the work-up procedure, the crude reaction mixture was stirred in MeCN/hexane and after 3 h the product was completely converted to a stable monosubstituted E_SO3H, which was isolated as a brown solid in 76% yield ( Figure 3). 1,3,8-Trimethoxy-6-methyl-9,10-anthraquinone E_OCH3 was obtained in good yield (81%) by applying a previously described reaction procedure (Me2SO4 and base K2CO3) [30]. Brominated trimethoxyemodin E_Br_OCH3 was synthesized from tetrabrominated emodin E_4Br by the same method. After purification, the product E_4Br_OCH3 was obtained in 87% yield. Sulfonation was carried out with oleum at room temperature. After 24 h, the reaction mixture was analyzed by NMR; three different products were observed. According to the NMR spectra obtained from the crude reaction mixture, trisulfonated, disulfonated and monosulfonated emodin were formed. However, the composition of the mixture changed with time and reverse sulfonation occurred. Hence, following the work-up procedure, the crude reaction mixture was stirred in MeCN/hexane and after 3 h the product was completely converted to a stable monosubstituted E_SO 3 H, which was isolated as a brown solid in 76% yield ( Figure 3). Sulfonation was carried out with oleum at room temperature. After 24 h, the react mixture was analyzed by NMR; three different products were observed. According to NMR spectra obtained from the crude reaction mixture, trisulfonated, disulfonated monosulfonated emodin were formed. However, the composition of the mixture chan with time and reverse sulfonation occurred. Hence, following the work-up procedure, crude reaction mixture was stirred in MeCN/hexane and after 3 h the product was co pletely converted to a stable monosubstituted E_SO3H, which was isolated as a bro solid in 76% yield (Figure 3). 1,3,8-Trimethoxy-6-methyl-9,10-anthraquinone E_OCH3 was obtained in good y (81%) by applying a previously described reaction procedure (Me2SO4 and base K2C [30]. Brominated trimethoxyemodin E_Br_OCH3 was synthesized from tetrabromina emodin E_4Br by the same method. After purification, the product E_4Br_OCH3 was tained in 87% yield.

Antiviral Activity of Emodin Analogues
The primary objective of this study was to determine the potential of a library o emodin analogues (Figure 4) to combat the cytopathic effects of HCoV-NL63. The n 1,3,8-Trimethoxy-6-methyl-9,10-anthraquinone E_OCH 3 was obtained in good yield (81%) by applying a previously described reaction procedure (Me 2 SO 4 and base K 2 CO 3 ) [30]. Brominated trimethoxyemodin E_Br_OCH 3 was synthesized from tetrabrominated emodin E_4Br by the same method. After purification, the product E_4Br_OCH 3 was obtained in 87% yield.

Antiviral Activity of Emodin Analogues
The primary objective of this study was to determine the potential of a library of 11 emodin analogues (Figure 4) to combat the cytopathic effects of HCoV-NL63. The nontransformed primate kidney Vero cell line was used for this work. An additional series of studies was performed to evaluate the effects of each compound on the viability of Vero cells.

Evaluation of the Viability of Vero Cells in the Presence of Emodin and Emodin Analogues
Published reports indicate that emodin displays a broad spectrum of activities, including cytotoxicity [14]. Therefore, in addition to evaluating the activity of emodin and the emodin analogues against HCoV-NL63, it was important to test the potential cytotoxicity of these compounds. Viability assays were performed using Vero cells; IC50 curves for each compound and other additional data are shown in Figure 5 and in Table S1 (raw data in Table S4). Interestingly, only E_I showed higher toxicity to Vero cells compared to emodin (E_H). E_4Br, E_Cl and E_OMe were less toxic than emodin, while virtually no toxicity was observed with E_SO3H, E_NH2 and E_Br_OMe. However, it is important to consider that although all emodin analogues were soluble in DMSO at 50 mM concentrations, some

Evaluation of the Viability of Vero Cells in the Presence of Emodin and Emodin Analogues
Published reports indicate that emodin displays a broad spectrum of activities, including cytotoxicity [14]. Therefore, in addition to evaluating the activity of emodin and the emodin analogues against HCoV-NL63, it was important to test the potential cytotoxicity of these compounds. Viability assays were performed using Vero cells; IC50 curves for each compound and other additional data are shown in Figure 5 and in Table S1 (raw data in Table S4).

Evaluation of the Viability of Vero Cells in the Presence of Emodin and Emodin Analogues
Published reports indicate that emodin displays a broad spectrum of activities, including cytotoxicity [14]. Therefore, in addition to evaluating the activity of emodin and the emodin analogues against HCoV-NL63, it was important to test the potential cytotoxicity of these compounds. Viability assays were performed using Vero cells; IC50 curves for each compound and other additional data are shown in Figure 5 and in Table S1 (raw data in Table S4). Interestingly, only E_I showed higher toxicity to Vero cells compared to emodin (E_H). E_4Br, E_Cl and E_OMe were less toxic than emodin, while virtually no toxicity was observed with E_SO3H, E_NH2 and E_Br_OMe. However, it is important to consider that although all emodin analogues were soluble in DMSO at 50 mM concentrations, some analogues exhibited a tendency to precipitate when added to the cell culture medium; compounds containing methoxy groups were the most difficult to dissolve in DMSO and Interestingly, only E_I showed higher toxicity to Vero cells compared to emodin (E_H). E_4Br, E_Cl and E_OMe were less toxic than emodin, while virtually no toxicity was observed with E_SO 3 H, E_NH 2 and E_Br_OMe. However, it is important to consider that although all emodin analogues were soluble in DMSO at 50 mM concentrations, some analogues exhibited a tendency to precipitate when added to the cell culture medium; compounds containing methoxy groups were the most difficult to dissolve in DMSO and were therefore probably the least soluble in aqueous solutions. It is therefore possible that these compounds gave lower cytotoxicity indexes due to their poor solubility in the cell culture medium.

Anti-Viral Activity of Emodin and Emodin Derivatives
In another series of studies, we evaluated the potential of emodin and emodin derivatives to protect Vero cells against the cytopathic effects induced by HCoV-NL63. Dead cell protease activity was used as a measure of impaired cell viability, as described in the Methods section; chloroquine and remdesivir were used as controls. The results obtained are presented in Figure 6 and Table 5. Additional data can be found in Supplementary Materials (Table S2; raw data in Table S5). Three compounds, E_OMe, E_SO 3 H and E_Br_OMe, did not exhibit significant anti-viral activity; as described above, this could be due to their lower solubility in aqueous solutions. Emodin and the emodin analogues E_4NO 2 and E_I impaired Vero cell viability and had anti-viral effects at very similar concentrations. Five other compounds, E_3I, E_4Br, E_2Br, E_2Cl and E_NH 2 , showed anti-viral activity at concentrations lower than those at which they impaired Vero cell viability; among these, E_NH 2 exhibited the least toxicity to Vero cells. On the other hand, E_3I was the compound that demonstrated the strongest anti-HCoV-NL63 activity, along with the largest difference (about 10-fold) between the IC 50 value for anti-HCoV-NL63 activity and that for cytotoxicity, thus providing the largest therapeutic window. Nevertheless, the IC 50 value for cytotoxicity of E_3I was still considerably high under the applied experimental conditions, namely identical to that of E_H (Table S1).

Anti-Viral Activity of Emodin and Emodin Derivatives
In another series of studies, we evaluated the potential of emodin and emodin derivatives to protect Vero cells against the cytopathic effects induced by HCoV-NL63. Dead cell protease activity was used as a measure of impaired cell viability, as described in the Methods section; chloroquine and remdesivir were used as controls. The results obtained are presented in Figure 6 and Table 5. Additional data can be found in Supplementary Information (Table S2; raw data in Table S5). Three compounds, E_OMe, E_SO3H and E_Br_OMe, did not exhibit significant anti-viral activity; as described above, this could be due to their lower solubility in aqueous solutions. Emodin and the emodin analogues E_4NO2 and E_I impaired Vero cell viability and had anti-viral effects at very similar concentrations. Five other compounds, E_3I, E_4Br, E_2Br, E_2Cl and E_NH2, showed antiviral activity at concentrations lower than those at which they impaired Vero cell viability; among these, E_NH2 exhibited the least toxicity to Vero cells. On the other hand, E_3I was the compound that demonstrated the strongest anti-HCoV-NL63 activity, along with the largest difference (about 10-fold) between the IC50 value for anti-HCoV-NL63 activity and that for cytotoxicity, thus providing the largest therapeutic window. Nevertheless, the IC50 value for cytotoxicity of E_3I was still considerably high under the applied experimental conditions, namely identical to that of E_H (Table S1).  As mentioned earlier, chloroquine and remdesivir were used as controls, since these drugs are being considered as potentially effective against COVID-19. Interestingly, the  As mentioned earlier, chloroquine and remdesivir were used as controls, since these drugs are being considered as potentially effective against COVID-19. Interestingly, the anti-HCoV-NL63 activity of some of the emodin analogues was much higher than that of chloroquine. On the other hand, the emodin analogues E_3I and E_2Br displayed anti-HCoV-NL63 activity comparable to that of remdesivir; all the data obtained on the activities of chloroquine and remdesivir are presented in Supplementary Materials (Table S3 and Figure S5; raw data in Tables S6 and S7).

Evaluation of Antiviral Activity
Compound preparation. For testing purposes, all compounds were dissolved in DMSO to a final concentration of 50 mM. With these stock solutions, mother plates were prepared in DMSO and generated stocks for testing 8 point-dose responses. 1:1 dilutions were prepared, starting from the original 50 mM stock. Namely, 30 µL of stock solution was added to the first column in the plate. 15 µL of DMSO was added to all the other wells; stocks to test successive doses were produced by transferring 15 µL of compound solution from the preceding column to the next. Each mother plate held the test solutions corresponding to two compounds. Chloroquine and Remdesivir were prepared in threefold dilutions, chloroquine starting at a 30 µM concentration and Remdesivir at 10 µM.
Cell culturing and viral infection. Vero cells (Cercopithecus aethiops lung epithelial cells, ATCC, CRL-81), were seeded in EMEM medium containing 10% FBS in inner wells of 96-well white plates, at a density of 20000 cells/well. The following day, the plating medium was removed and replaced with 100 µL of fresh EMEM medium containing 2% FBS; a 100 nL solution of each compound was then also added to the wells using a Mosquito pipetting device (TTPlabtech, Cambridge, England). Cells were kept in this medium for 4 days at 33 • C in 5% CO 2 before evaluating the cytotoxicity of each compound.
When conducting experiments to evaluate the anti-viral activity of the emodin compounds, the culture medium was changed at 24 h post-plating to 50 µL of fresh EMEM medium containing 2% FBS; 100 nL solution of each test compound was also added to the wells, as described above and an additional 50 µL of the same medium containing a 1:5 dilution of a HCoV-NL63 viral stock was added (Human Coronavirus, Strain NL63, FR-304, IRR (International Reagent Resource). Cultures were maintained in this medium for 4 days at 33 • C, in 5% CO 2 ; the anti-viral activity of each compound was then measured.

Evaluation of Cytotoxic and Anti-Viral Activities
In order to evaluate any potential cytotoxicity of each of the test compounds, 50 µL/well of Cell Titer-Glo Luminescent reagent was added to each culture following a 4 day-incubation in the presence of these compounds. Luminescence was measured after 5 min of incubation using a Spectra Max i3 instrument (Molecular Devices, San Jose, CA, USA); the obtained values were considered proportional to cellular ATP content. Cytotoxic activity of the tested compounds on Vero cells was expressed as the percentage of live cells compared to that in control cultures that had not been exposed to any compound and had been assigned live cell rates of 100%.
Assays to evaluate anti-viral activity were carried out at 4 days post-infection by adding 50 µL/well of Cyto Tox-Fluor reagent to cultures priorly infected with HCoV-NL63 virus. Plates were incubated for 1 h and dead-cell protease activity was assayed by measuring the fluorescent product that had formed (480/520), using a Spectra Max i3 instrument. Anti-viral activity of each compound was presented as the percentage of live cells compared to that in control Vero cultures that had been infected with the HCoV-NL63 virus but had not been exposed to any of the test compounds; these latter cultures were assigned live cell rates of 100%. All IC 50 values were calculated using GraphPad Prism 8 software (GraphPad Software, LLC, San Diego, CA, USA).
Assays were considered valid when the signal to background ratio (signal from cells infected with the virus versus signal from uninfected cells) was higher than 2 and Z' was higher than 0.35.

Conclusions
Emodin is a natural anthraquinone commonly found in plants. Numerous studies (reviewed by Dong and coworkers [14]) show that emodin has a wide spectrum of pharmacological properties. Among these, it is now clear that it also has anti-viral activity. However, emodin can also cause toxicity, such as hepatotoxicity and nephrotoxicity. In accordance with these data, our study demonstrated the toxicity of emodin to Vero cells, a kidney cell line. The main objective of this work was to apply a series of synthetic approaches to the preparation of emodin analogues and to subsequently conduct structureactivity relation (SAR) studies on them, which could shed light on how to improve the anti-HCoV-NL63 activity and to reduce the cytotoxicity of these compounds. Our work, in agreement with the published literature reporting improved activity of halogenated emodin derivatives, showed that halogenation of emodin resulted in enhanced anti-HCoV-NL63 activity; however, cytotoxicity was still relatively high. On the other hand, the introduction of charged groups, such as SO 3 H and NH 2 , resulted in a significant decrease in undesirable cytotoxicity, although this also resulted in decreased anti-HCoV-NL63 activity. In addition, the introduction of methoxy functionalities into emodin did not seem to improve its anti-HCoV-NL63 activity; this was probably related to the lower solubility of these compounds in aqueous solutions.
In summary, we applied various synthetic routes to prepare a limited number of emodin analogues that allowed us to perform some initial studies aimed at improving the anti-HCoV-NL63 activity of emodin. Of particular importance was the establishment of SAR guidelines related to anti-HCoV-NL63 activity. The availability of new synthetic routes for the diversification of an emodin structure should allow the generation of new emodin analogues with significantly improved properties, such as increased solubility and, most importantly, stronger anti-HCoV-NL63 activity and reduced toxicity, so that a much wider therapeutic window can realistically be generated.