Effect of drug metabolism in the treatment of SARS-CoV-2 from an entirely computational perspective

Understanding the effects of metabolism on the rational design of novel and more effective drugs is still a considerable challenge. To the best of our knowledge, there are no entirely computational strategies that make it possible to predict these effects. From this perspective, the development of such methodologies could contribute to significantly reduce the side effects of medicines, leading to the emergence of more effective and safer drugs. Thereby, in this study, our strategy is based on simulating the electron ionization mass spectrometry (EI-MS) fragmentation of the drug molecules and combined with molecular docking and ADMET models in two different situations. In the first model, the drug is docked without considering the possible metabolic effects. In the second model, each of the intermediates from the EI-MS results is docked, and metabolism occurs before the drug accesses the biological target. As a proof of concept, in this work, we investigate the main antiviral drugs used in clinical research to treat COVID-19. As a result, our strategy made it possible to assess the biological activity and toxicity of all potential by-products. We believed that our findings provide new chemical insights that can benefit the rational development of novel drugs in the future.


S.2 -OPTIMIZED STRUCTURES, BOND LENGTHS OF FUNCTIONAL GROUPS AND ESP MAPS (WITH SURFACE ISOVALUE OF 0.0004) OF a) FAVIPIRAVIR, b) GALIDESIVIR; c) NITAZOXANIDE, d) REMDESIVIR, AND e) RIBAVIRIN, GENERATED IN THE GAUSSVIEW 6.0 HTTPS://GAUSSIAN.COM/GAUSSVIEW6/
In the figure above, it is illustrated the bond lengths of the functional groups that compose the drugs' structures and their ESP maps. In general, the electronic density is distributed in a polar manner for all the molecules, in which the red areas show higher negative potential (mostly due to electron acceptor oxygenderivate groups such as nitro or hydroxide) and the blue areas positive potential (mainly due to the electrondonor nitrogen atoms). In this framework, the Favipiravir polar structure shows two regions suitable for interactions with external molecules between the carbonyl groups and the neighbor amine. For the Galidesivir drug, the charge is distributed moderately on the hydroxide groups and heavily on the outermost amines, also it is worth notice that amine groups in the aromatic rings show electron-acceptor character. The Nitazoxanide molecule is very polarized with many regions with positive potential, showing the higher positive charge next to the ester group, and heavily negatively charged in the nitro region. Due to the large molecular structure, the charge distribution of Remdesivir is slightly more homogenous than the other drugs, having polar character in the region containing the most functional groups, thus presenting a portion with high positive potential that is close to an electron-acceptor area due to the hydroxide group. It is also noticed that for the Remdesivir molecule, the sulfur center shows slightly intense negative potential because of the oxygen-derivate groups. Lastly for the Ribavirin drug, the charge is distributed mostly homogenously around the structure body and shows two poles with high negative and positive potentials, due to a hydroxide and amine groups, respectively.

S.3 -ESTIMATED a) UV-VIS AND b) ECD SPECTRA OF THE DRUGS
In the Figure S  The Table above organizes the electronic properties of the drugs, in which are in agreement with the expected values for molecules. It is observed that the Favipiravir shows the smallest LUMO-HOMO energy, followed by Nitazoxanide, Remdesivir, Galidesivir and then Ribavirin, which these energies may in principle indicate the order of reactivity of these molecules, being Favipiravir the most reactive and Ribavirin the lowest. As expected, the hardness follows the same tendency and is inverse to the softness, indicating that the calculations are in order. As for the electronegativity, it increases from Galidesivir to Remdesivir, Ribavirin, Favipiravir and Nitazoxanide, which is the same tendency as the electrophilicity.

S.5 -THEORETICAL (BLACK) AND EXPERIMENTAL (PINK) EI-MS SPECTRA OF a) CHLOROQUINE AND b) HYDROXYCHLOROQUINE, AND THEIR RESPECTIVE TRAJECTORIES, BY MEANS OF GFN2-XTB. THEORETICAL (BLACK) AND EXPERIMENTAL (PINK) EI-MS SPECTRA OF c) CHLOROQUINE AND d) HYDROXYCHLOROQUINE RESULTED FROM THE GFN1-XTB METHOD. STRUCTURES. PLOTS GENERATED IN THE GRACE SOFTWARE HTTPS://PLASMA-GATE.WEIZMANN.AC.IL/GRACE/
As seen in the S.5.A above, the EI-MS diagram from the GFN2-xTB method, in principle, presented most of the intermediaries of the Chloroquine molecule, in agreement with the experimental data profile (NIST MS 42361), with slight deviation of intensity. The most intense signal 86 m/z and its respective intermediary are also identified in the theoretical plot and the resultant trajectory described. As for the Hydroxychloroquine spectra in S.5.B, the most intense peaks from the experimental data (NIST MS 246973) are not identified in the theoretical data, still, some of the signals are found with distinct intensity from the experimental spectra.
In order to evaluate the results between the theoretical methods, S.5.C shows the EI-MS spectra of the Chloroquine obtained using the GFN1-xTB method in comparison with the same experimental profile. In this manner, the spectra from the GFN1-xTB did not match the intermediaries mass/charge rate and intensity as well as its analog method, reveling that the GFN2-xTB is the best option for this calculation. Transitioning to the Hydroxychloroquine in S.5.D, the spectra acquired from the GFN1-xTB approach did not show significant improvements over the GFN2-xTB, and the most intense signals of the NIST profile are not identified as well.
In general, an increasing on the molecular dynamics parameters could lead to a better prediction of the EI-MS spectra and its intermediaries in exchange of meaningful computational cost, however, as the current methodology with GFN2-xTB provided satisfactory results for the Chloroquine drug, it has been chosen as the default semiempirical method to the study of the other drugs. For more information regarding the characterization and details of Chloroquine and Hydroxychloroquine, please check our other works with these drugs. 1,2 Henceforth, the discussion of the EI-MS spectra and trajectories will be done in the context of xenobiotics metabolism, evaluating the obtained intermediaries as drug by-products, their metabolism and toxicity when possible. Thus, returning to the Chloroquine drug, the spectra and trajectories are shown in S.5.A. The Chloroquine molecule contains polar amine and chloride groups in its structure, showing an aromatic region with more polar character than the other extremity. The first trajectory showed the fragmentation of Chloroquine around the amine that bond the aromatic and the alkane regions, leading to the following intermediaries: the I-177 m/z 7-chloro-4-aminoquinoline, containing the aromatic region, deprotonated amine and chloride polar groups, is a toxic and major metabolite from the oxidation of Chloroquine by the cytochrome P-450 enzyme; 3,4 the deprotonated I-57 m/z butane and I-29 m/z ethane, both nonpolar hydrocarbons which can be oxidized into polar species in Phase I of metabolism; and the I-56 m/z deprotonated amine, a polar and likely water-soluble molecule that may metabolize directly in Phase II. The trajectory II leads to the high molecular mass fragment II-233 m/z similar to the I-177 m/z, with an alkane extremity that may be target of oxidation in Phase I; and the II-86 m/z, the specie also identified in the experimental spectra, show very low polar character and may be almost insoluble in water, possible target of oxidative reactions in Phase I metabolism before conjugation in Phase II. The last trajectory for Chloroquine gives the following intermediaries: the III-205 m/z, a specie like the I-177 m/z and II-233 m/z, with a shorter alkane segment which may be oxidized in Phase I, and share the behavior of its analog molecules; the deprotonated organic molecules III-29 m/z ethane and III-28 m/z ethene, both nonpolar and likely targets to oxidizing reactions in Phase I, leading to polar conjugates to metabolize in Phase II; and the protonated form of I-56 m/z. The S.5.B shows the EI-MS spectra and unique trajectory of the Hydroxychloroquine drug. The molecular structure of this drug is a more polar analog of the Chloroquine due to the addition of a hydroxide group. The calculations for the Hydroxychloroquine resulted in a single trajectory: the I-142 m/z, a deprotonated aminoquinoline similar to the 7-chloro-4-aminoquinoline from the metabolization of Quinoline, which is a metabolite from the Hydroxychloroquine; 5 the I-35 m/z chloride ion; and the I-144 m/z, with polar amine and alcohol groups, and the nonpolar extremities likely submitted to oxidative reactions in Phase I that may lead to smaller and polar fragments. This last fragment is further cleaved into two more species: the I-1-31 m/z molecule, which is deprotonated into a highly water-soluble and toxic formaldehyde form, being rapidly metabolized into formate by the alcohol dehydrogenase enzyme; 6,7