*2.1. Chemistry*

Avarone (**1**) and avarol (**3**) were isolated from the sponge *D. avara* and purified according to the previously described procedures [21–23]. They were easily identified by comparison of their spectroscopic properties (1H and 13C NMR, HRESI-MS) with those reported in literature [21–23].

Thiazoavarone (**2**) was prepared as reported in Scheme 1. A portion of avarone was dissolved in a solution of CH3CN/EtOH (1:1) and then hypotaurine and a catalytic amount of salcomine in portion were added. The mixture was stirred for 48 h at room temperature and then was extracted with diethyl ether. The crude material was purified by HPLC on a reverse phase column (Luna 3 μm, 150 × 3.00 mm) (MeOH/H2O 75:25 v/v%) to afford the pure compound **2**. The nucleophilic addition reaction is regioselective in unsymmetrical quinones but generally leads to the formation of both regioisomers one of which is obtained in large excess with respect to the other. In the case of avarone, a single isomeric product, thiazoavarone **2**, has formed, as determined by MS and NMR spectroscopy; this could be reasonably due to the high steric hindrance of the heavy sesquiterpene moiety.

**Scheme 1.** Coupling of avarone (**1**) with hypotaurine via nucleophilic addition reaction.

Compound **2**, obtained as yellow powder, [α]D<sup>25</sup> = +19.2 (c 0.004, MeOH), had a molecular formula of C23H31NO4S as determined by the HRESI MS ion at m/z C23H31NO4SNa [M + Na]+ 440.1865 (calculated value: 440.1866). Molecular formula obtained from MS and a first survey of the 1D NMR spectra of **2** (CDCl3) and the comparison with those of the known avarone **1** quickly allowed us to hypothesize that the condensation reaction with hypotaurine has occurred. 1H and 13C NMR data of **2** indicated the same decalin ring system of **1**, and the only difference between the two compounds is in the quinonic portion. Indeed, 1H NMR spectrum of **2** lacked the signals at δH 6.51 and 6.71, whereas contained a quite deshielded methylene signals (δH3.30 and 4.05), resonating as two multiplets and each integrating for two protons. Likewise, the 13C spectrum of **2** contained two additional methilene carbon resonances at δC 39.8 and at δC 48.8 attributable to a nitrogen and sulfoxide-bearing carbons, respectively. The key HMBC cross-peaks (Figure 3) from H-3- to C-2- (δC 48.8), C-4a- (δC 143.2), and from H-2- to C-3- (δC 39.8), C-8a- (δC 111.6), from H-6- to C-4a- (δC 143.2), and from H-15a and H-15b- to C-8- (δC 177.1) indicated the regiochemistry of **2**. It should be noted that only one of possible regioisomer has been obtained.

**Figure 3.** Key 1H-13C HMBC correlations of thiazoavarone (**2**).

Chemical shifts and coupling patterns of the all signals of **2** were assigned by aid of COSY, HSQC, and HMBC experiments (Table 1). Anyway, a purity higher than 99.8% has been determined by HPLC for compounds **1**–**3**.


**Table 1.** 1H (700 MHz) and 13C (125 MHz) NMR data of thiazoavarone (**2**) in CDCl3.

Overlapped by other signals.

a

### *2.2. In Vitro Activity on P. falciparum and Cytotoxicity*

Avarone (**1**) and the semisynthetic 1,1-dioxo-1,4-thiazine analogue (**2**), as well as hydroquinone avarol (**3**) were tested for their in vitro antiplasmodial activity against asexual and sexual (gametocytes stage V) stages of *P. falciparum* (Table 2). A chloroquine-sensitive (CQ-S) D10 and a chloroquine-resistant (CQ-R) W2 strains were used to determine the IC50 against asexual stage of parasites. The most potent compound was thiazoavarone (**2**) with an IC50 value in the nanomolar range, higher than those exhibited by the previously identified synthetic lead [13]. The high potency of the thiazinoquinone **2**, specifically on the chloroquine-resistant strain W2 and compared with avarone (**1**), lacking the heterocyclic moiety, confirmed once again the high potential of the thiazinoquinone scaffold for development of new antimalarial hits. Interestingly, both natural compounds **1** and **3** also exhibited a remarkable effect against sensitive and resistant *P. falciparum* strains, showing no cross-resistance with chloroquine (see Table 2). In particular, the reduced hydroquinone form (avarol, **3**) resulted significantly more active than the oxidized quinone form (avarone, **1**).

The evaluation of the effects of compounds **1**–**3** on *Pf* gametocytes stage V, the sexual stage circulating in the bloodstream, was performed in order to evaluate their transmission blocking potential. As evidenced in Table 2, all the three compounds resulted less active against the gametocytes with respect to the parasite asexual stage; this finding is not unexpected since most of the current antimalarial drugs have no effect on the late stage of gametocytes. Avarol (**3**) resulted the most potent in the series against *Pf* gametocytes stage V with an IC50 = 9.30 μM, comparable to that of OZ27, a drug in clinical development (IC50 = 6.4 μM) [42].


**Table 2.** In vitro antimalarial activity against asexual *P. falciparum* parasites from D10 (CQ-sensitive) and W2 (CQ-resistant) strains a and against stage V *P. falciparum* gametocytes from a 3D7 transgenic line.

 a Chloroquine (CQ) has been used as positive control (D10 IC50 = 0.04 ± 0.01; W2 IC50 = 0.54 ± 0.28). b The results are the mean ± SD of IC50of three independent experiments performed in duplicate.

Finally, we tested compounds **1**–**3** for their cytotoxic effects against two different human cell lines, microvascular endothelial (HMEC-1) and acute monocytic leukemia (THP-1) cells differentiated into macrophages; for each compound, we evaluated the selectivity index (SI, Table 2), namely the ratio between the IC50 on the human cells HMEC and that on the parasite strains (see Table 3). Thiazoavarone (**2**) exhibited a high toxicity against both mammalian cell lines, with IC50 in the low micromolar concentration range and, consequently, a very low SI. Avarone and avarol (**1** and **3**) were lowly and moderately cytotoxic, respectively (Table 3) but the hydroquinone **3** exhibited a better SI.

**Table 3.** IC50 against HMEC-1 (human microvascular endothelial cells) and THP-1 (human acute monocytic leukemia cells) and Selectivity Index (SI) of the compounds **1**–**3**.


a Camptothecin has been used as positive control (IC50 (μM) = 0.018 ± 0.008 on HMEC-1). b Data are expressed as mean ± SD of three different experiments performed in duplicate. c Data are the mean of two different experiments in duplicate. d SI = IC50 HMEC-1/IC50 *P. falciparum* strain.

### *2.3. In vitro Activity on Leishmania Parasites*

To assess the antileishmanial activity of compounds **1**–**3** we tested them against promastigote stage of *L. infantum* and *L. tropica* responsible for visceral and cutaneous leishmaniasis, respectively. The results of this study are reported in Table 4 as IC50 values with the relevant selectivity indexes (SIs). Considering the couple avarone/avarol, we can notice that, as observed for the antiplasmodial effects, the reduced form (**3**) is substantially more potent than the oxidized form **1** on both investigated *Leishmania* parasites. This has been reported also for several antileishmanial naphthohydroquinones which resulted more active than the corresponding naphthoquinones [43]. Instead, the hydroquinone metabolite **3** and the thiazinoquinone **2** exhibited similar values of IC50 in the range of low micromolar; however, avarol (**3**) showed a significantly higher SI (see Table 4). Definitely, the above results displayed that introduction of the 1,1-dioxo-1,4-thiazine ring in the structure of avarone (**1**) to give thiazoavarone (**2**) meaningfully improves the antileishmanial activity, once again according to the antiplasmodial effects (see Tables 2 and 4).

In addition, compounds **1**–**3** were also investigated against intracellular amastigotes, the clinically relevant form of *Leishmania*. All compounds resulted from 2 to 4-fold more active against amastigotes than promastigotes; the hydroquinone avarol **3** confirmed itself as a promising agen<sup>t</sup> to be further investigated in efficacy and selectivity, presenting the lowest IC50 value in the series and a SI greater than 10 (Table 4) [44].


**Table 4.** Activity of compounds **1**–**3** against promastigote stage of *L. infantum* and *L. tropica* and against amastigote stage of *L. infantum.*

a Data are expressed as mean ± SD of three different experiments performed in duplicate. b Data are the mean of two different experiments in triplicate. c SIp = IC50 HMEC-1/IC50 *L. infantum* (*L. tropica*) promastigotes. d SIa = IC50 HMEC-1/IC50 L. infantum amastigotes.

### *2.4. In Vitro Activity on S. mansoni*

a

Compounds **1**–**3** were tested against larval stage (schistosomula), adult worm couples and eggs of the platyhelminth *S. mansoni*. The most potent compound on schistosomula was thiazoavarone (**2**) with a LC50 value in the low micromolar range (Table 5). The natural compounds (**1** and **3**) showed comparable activity, with avarol (**3**) slightly more active than avarone (**1**), both showing a LC50 in the high micromolar range (Table 5). Therefore, the presence of a thiazine ring was proved to be very important for activity on schistosomula.

> **Table 5.** Activity of compounds **1**–**3** against *S. mansoni* schistosomula.


Data are expressed as mean ± SD of three different experiments.

All three compounds **1**–**3** were also very active on adult worm pairs at 50 μM leading to parasites death 7 days after treatment (Figure 4). However, when used at lower concentration (20 μM), only avarol (**3**) strongly impaired parasites viability (only 20% survival), while thiazoavarone (**2**) was poorly effective against the adult stage (Figure 4), despite its strong lethal effect on the larval stage (Table 5). These results sugges<sup>t</sup> the possibility that the double lipid bilayer coating the adult worms, namely the tegument [45], can interfere with compound **2** uptake by adult parasites.

**Figure 4.** Compounds **1**–**3** impair adult *S. mansoni* viability. Worm pairs were incubated with DMSO (vehicle) (black circle) or the indicated compounds at 50 μM (yellow triangle), or 20 μM (green, square) as described in material and methods. Phenotype analysis was recorded for 7 days and % viability represents the mean ± SEM of three independent experiments.

In the process of drug discovery for schistosomiasis, strategy involving any impairment in egg production and/or development must also be taken into account. In fact, upon mating with males, mature *S. mansoni* adult females, residing in the mesenteric veins of the definitive host, can lay hundreds of eggs each day. The eggs secreted in stool or trapped in the liver respectively cause disease transmission and, as a result of inflammatory granulomas reactions, intestinal and hepato-splenic diseases [46]. Therefore, compounds **1**–**3** were also assayed against the in vitro laid eggs (IVLEs). The IVLEs produced in the first 48 h by *S. mansoni* pairs and treated for 3 days with vehicle or compounds **1**–**3** were classified by microscopic observation according to the Vogel and Prata staging system of egg maturation [47]. The thiazoavarone (**2**) resulted the most effective compound, impairing eggs maturation already at 5 μM and resulting in undeveloped and severely damaged eggs at 20 μM (Figure 5). Similar results were obtained with compounds **1** and **3** at 50 μM.

**Figure 5.** Thiazoavarone (**2**) impairs egg viability and maturation. Representative pictures of IVLEs treated with vehicle (DMSO) (**a**) or compound **2** at 5 μM (**b**) and 20 μM (**c**) for 72 h. Filled red arrows indicate viable eggs at stages III–V (intermediate/developed); filled red triangle indicate viable eggs at stages I–II (immature); red-edged arrows indicate damaged eggs at stages III–V; red-edged triangle indicate damaged eggs at stages I–II. Bar, 200 μm.

### *2.5. Computational Studies and DFT Calculations*

To rationalize the observed SARs, the steric and electronic features of compounds **1**–**3** were investigated by means of computational studies, including conformational analysis and DFT calculations.

A systematic conformational search considering all rotatable bonds was applied to generate all possible conformations of the compounds, which were, then, subjected to molecular mechanic (MM) geometry optimization using the CFF force field and a distance dependent dielectric constant value of 80 (Discovery Studio 2017, BIOVIA, San Diego USA; see the experimental Section for details) [48]. The global minimum energy conformer (GM) was identified for each compound. All the generated conformers presented an energy di fference from the GM ( Δ*E*GM) ≤ 3 kcal/mol. MM conformers were, then, subjected to density functional theory (DFT) calculations. In order to mimic an aqueous environment, all DFT calculations were performed using the conductor-like polarizable continuum model (C-PCM) as solvent model [49]. Moreover, to characterize every structure as minimum, a vibrational analysis was carried out (see the experimental Section for details). Fully optimized DFT conformers were classified into families according to the values of their torsion angles (Tables 6 and 7 and Table S1).

Results evidenced that compounds **1**–**3** present common conformational features, characterized by the electronic attraction between the hydrogen atoms of the first methylene group of the alkyl substituent and the nearby quinone oxygen, which limits the conformational freedom of R- (Figure 6). Accordingly, the torsional angle τ1 showed just two sets of possible values (~±100◦; Tables 6 and 7 and Table S1), and for each of them the rigid sesquiterpene ring could assume three orientations with respect to the thiazinoquinone/quinone/quinol ring (τ2 = ~60◦, <sup>~</sup>−60◦ and ~180◦; Tables 6 and 7 and Table S1). This determined a total number of six conformers (named I–VI; Figure 6). In the case of the thiazinoquinone derivative **2**, due to the presence of the two opposite flips of the thiazinoquinone ring (<sup>τ</sup>flip ~ ±60◦), we obtained two specular sets of conformers with the same conformational energy (i.e., conformational enantiomers; Figure S10), as previously reported for other thiazinoquinone derivatives [13,14].

**Figure 6.** Density functional theory DFT conformers of compounds **1**–**3** superimposed by the carbon atoms of the quinone/hydroquinone ring. Carbon atoms are colored according to conformer classification (I = green, II = magenta, III = pink, IV = light blue, V = orange, and VI = Yellow); heteroatoms are colored by atom type (H = white, O = red, N = blue, S = orange). Hydrogens are omitted for sake of clarity, with the exception of those of the first methylene group of the R- substituent, whose intramolecular distances from the nearby oxygen atom of the quinone are reported.


**Table 6.** ΔEGM values (kcal/mol) and torsion angle values (degrees) of the DFT conformers of **2**.


a Only the conformational enantiomers with the value of <sup>τ</sup>flip ~60◦ are reported. b τ1 torsion angle is defined by e, f, g, and h atoms. c τ2 torsion angle is calculated considering f, g, h, and i atoms.


aτ1 torsion angle is defined by a, b, c, and d atoms. b τ2 torsion angle is calculated considering b, c, d, and e atoms.

The fixed position of the methylene group combined with the presence of the rigid sesquiterpene moiety, place in the putative semiquinone radical produced upon one electron reduction/oxidation several hydrogen atoms at a distance (≤3 Å) suitable for an intramolecular radical shift from the oxygen atom to a carbon atom of R- (see below).

Then, starting from the DFT minima, the redox properties of **1**–**3** were calculated. At this aim, we considered the two electrons/two protons quinone reduction pathway in a protic solvent (Scheme S1) and all the species involved in the pathway (Q•−, QH•, QH<sup>−</sup>, QH2) were generated and DFT optimized using as starting structures the energetically favored DFT minima I and II.

Two possible protonated semiquinone species may be formed, depending on which of the two quinone oxygen atoms is reduced/oxidized at first. The location of the lowest unoccupied molecular orbital (LUMO) in **1** and **2** and of the highest occupied molecular orbital (HOMO) in **3** indicated the oxygen opposite to the alkyl chain as the most probable site to be reduced and the one close to the alkyl chain as the most probable site to be oxidized, respectively (Figure 7). It is worthy to be mentioned that, regarding the most probable oxidation pathway of **3** to its quinone form, since the deprotonation

**1, 3**

> **2**

> **2**

II

> I

II −10.68

−17.76

−17.83

step is supposed to be the first event in protic solvents (Scheme S1), we calculated the pKa values of the two hydroxyl groups, too. Results indicated the hydroxyl group nearby the alkyl substituent as the first to lose the proton (Figure S11), further supporting the formation of the hydroquinone radical reported in Figure 7.

**Figure 7.** (**A**): DFT global minimum energy conformer (GM) structure of **1** (Q), DFT conformer I of **3** (QH2) and their semiquinone radical (QH•). (**B**) DFT GM structure of **2** (Q) together with its one- and two-electron reduced species QH• and QH2. Atoms possibly involved in an intramolecular radical shift are evidenced with red dashed lines. The LUMO of **1** and **2**, and the HOMO of **3** are visualized using GaussView with an isosurface value of 0.02 e<sup>−</sup>/a.u.<sup>3</sup> The NBO spin density isosurface of the QH• species is displayed using GaussView with an isosurface value of 0.01 e<sup>−</sup>/a.u.3. The blue surface (positive spin density) corresponds to an excess of α-electron density.

The standard redox potential (E◦) and the standard Gibbs free energy (ΔG0red, aq) of each electron-transfer reaction (see Scheme S1) alongside with the standard Gibbs free energy required for the protonation of the resulting reduced species (Δ<sup>G</sup>0H+) of the redox couple **1**, **3** and **2** were calculated (for details see the experimental Section). To further evaluate the propensity of **1** and **2** to undergo a one-electron reduction, the energy of the lowest unoccupied molecular orbital (ELUMO) was also taken into account. Similarly, we calculated the ionization potential (IP; i.e., −EHOMO) of the radical anion QH− as indicative of the tendency of the deprotonated species of **3** to undergo a one-electron oxidation. Finally, we considered the energy of the single occupied molecular orbital (ESOMO) of the radical species as indicative of the ability to delocalize the unpaired electron. The resulting data are reported in Tables 8 and 9.


−106.60

−158.60

−159.93 −95.13

−95.29

−95.66 −279.87

−272.91

−273.03

−304.65

−297.63

−281.68

a

−143.42

−148.69

−148.72

**Table 8.** DFT calculated parameters and standard redox potentials (E◦; Q/Q•−) of compounds **1**–**3**.


 mV.

a kcal/mol.

**Table 9.** DFT calculated parameters and standard redox potentials (E◦; QH•/QH−) of compounds **1**–**3**.

A first consideration can be derived comparing the redox properties of the quinone-based compounds **1** and **2**. With respect to **1**, **2** showed either a higher tendency to acquire one electron (lower ELUMO and ΔG0(red, aq); higher E◦; Table 8) and a higher stability of the QH• radical (lower ESOMO; Table 8). While the *E*SOMO values of the anion radical species showed little difference (ESOMO Q•−; Table 8), on the contrary, the differences became evident after the protonation step (ESOMO QH•). These results, on one hand confirm the key role played by the 1,1-dioxo-1,4-thiazine ring on the electron affinity [11,13,14]; on the other hand, support the hypothesis that the compound activity is related to the formation of the semiquinone radical species. Indeed, the thiazinoquinone derivative **2** resulted overall more potent than the quinone derivative **1**.

A second consideration is that, as evidenced in Figure 7 and Figure S12, all the calculated semiquinone radicals showed a hydrogen atom of the first methylene group together with, at least, another hydrogen atom of the rigid sesquiterpene ring, at a distance suitable for an intra-molecular hydrogen radical shift to the semi- reduced/oxidized oxygen atom (≤3 Å) (Table S2). By consequence, as above mentioned, the presence of the sesquiterpene moiety as alkyl substituent is expected to promote the putative "through space" intramolecular hydrogen radical shift leading to the formation of the toxic radical species. In line with this hypothesis, **2** resulted the most potent thiazinoquinone developed by us against *P. falciparum* D10 and W2 strains as well as against stage V gametocytes. In addition, **1**, although lacking the 1,1-dioxo-1,4-thiazine ring of **2**, resulted still active on *P. falciparum* D10 and W2 strains as well as on schistosomula, contrarily to what previously observed by us for other 1,4-benzoquinone derivatives when compared to the corresponding thiazinoquinone analogues [13,14].

Finally, the hydroquinone **3** showed a higher propensity to be oxidized to the semiquinone radical (i.e., lower E◦ and IP of the QH− anion) with respect to the corresponding reduced form of **2** (Table 9). Thus, the absence of the 1,1-dioxo-1,4-thiazine ring favors the one-electron oxidation reaction of **3**.

The hydroquinone **3** resulted more active than the corresponding quinone **1** against all considered parasites, while presenting the highest selectivity index with respect to mammalian cells. Both the oxidized and the reduced forms could produce the same putative toxic radical species upon a one-electron transfer reaction. In this view, our results sugges<sup>t</sup> the presence in the parasite cell of a bioactivation reaction partner preferentially binding the hydroquinone (reduced) form (**3**) rather than the quinone (oxidized) form (**1**). This putative bioactivation partner seems not to be present in human cells.

To investigate the role played by molecular pharmacokinetics on the observed SARs, we calculated the distribution coefficient values of **1**–**3** (clogD, Table S3) (ACD/Percepta 2017). According to Lipinski's rules for drug absorption [50] the thiazinoquinone **2** showed a better cLogD value for cell membrane passive diffusion (cLogD ~ 4) with respect to **1** and **3** (cLogD > 5). However, in specific developmental stages (i.e., *Pf* late gametocytes, promastigote of *L. infantum* and *L. tropica*, amastigote of *L. infantum*, and adult worms of *S. mansoni*) **3** resulted more active than **2** (Tables 2, 4 and 5; Figures 4 and 5). According to what previously reported by us [14], this peculiar activity profile could be due to the significant morphological changes in the parasite during the above-mentioned developmental stages [51–53], which are likely to impair compound ability to penetrate into the parasite by passive diffusion, while it could still penetrate by exploiting the large number of transport proteins expressed on the parasite membrane.

Taken together, the results of our computational investigation indicate that a toxic semiquinone radical species [11,13,14] which can be produced starting both from quinone- and hydroquinone-based compounds could mediate the anti-parasitic effects of the tested compounds **1**–**3**.

### **3. Materials and Methods**
