Captopril (C₉H₁₅NO₃S, nominal mass 217.2867), chemical structure of which is shown in Figure 1, was purchased from Sigma-Aldrich (St. Louis, MO, USA). Solvents for analytical determinations were acetonitrile (HPLC grade, JT Baker) and ultrapure water. Ultrapure water, from a Millipore Milli-Q system (Milford, MA, USA), was employed to prepare all the solutions.

Standard solution of sodium hypochlorite at 10% w/v was provided by SEM- AE (Municipal Water and Sewage Service) in Ouro Preto, Minas Gerais, Brazil. Ozone gas was generated using an ozone generator of electrical discharge with production capacity of 3 g O₃/h (Ozone Generator, model O & L3.0RM, Ozone & Life Industry, São José dos Campos-SP, Brazil) from oxygen feed gas (oxygen purity 99.99%).

Aqueous solutions of ozone were prepared by continuously bubbling ozone gas into Milli-Q ultrapure water through bottled gas scrubber for no less than 10 minutes at room temperature (20˚C - 23˚C). The quantification of ozone occurred by direct spectrophotometric measurement at 260 nm (CO₃ (mg・L⁻¹) = 14.59 × Absorbance [17] . The ozonation experiments were performed in bench scale with amber bottles of 20 mL capacity. Initially a volume of the CP solutions were added into the flask to a final concentration of 10 mg・L⁻¹ [18] . After preparation of the ozone solution, it was added to the bottles a volume of solution which contained the final concentration of ozone (8 mg・L⁻¹). Each bottle used had a contact time of 0, 5, 10, 15 and 30 minutes.

The chlorination experiments were performed in bench scale with amber bottles of 20 mL capacity. The stock CP solution (100 mg・L⁻¹) was prepared separately using ultrapure water (Milli-Q) and 2 mL of this solution were added into the amber bottles (final concentrations of 10 mg・L⁻¹ of CP). The solutions were stirred for 10 minutes before the degradation tests. All tests were performed at room temperature, which varied between 22˚C and 23˚C. In the chemical oxidation reaction system it was collected the first sample (time 0) and subsequently an amount of sodium hypochlorite was added to achieve a final concentration of 10 mg・L⁻¹. After specific contact time, sodium thiosulfate (Na₂S₂O₃) 10 mg・L⁻¹ were added to quench the residual hypochlorite and stop the reactions. The chlorination was accomplished over a period of 30 minutes. Samples were collected in amber glass bottles at the times of 0, 5, 10, 15 and 30 minutes.

The aliquots collected from the chlorination and ozonation tests were also maintained under identical conditions until the Total Organic Carbon (TOC) and mass spectrometry analyses. Although the CP concentration of 10 mg・L⁻¹ used in this study was much higher than those typically found in the environment, it was chosen to facilitate the subsequent mass spectrometry analysis, eliminating the steps of extraction and pre-concentration of the samples, thereby minimizing errors related to sample preparation.

The analysis of total organic carbon (TOC) was based on the determination of CO₂ produced through the oxidation of organic matter in the samples, and was carried out on a TOC analyzer (Shimadzu, model TOC-L, Kyoto, Japan). The TOC content of each collected aliquot was obtained by the indirect method which corresponds to the difference between the total carbon and inorganic carbon values. The TOC-L equipment employs a combustion (680˚C) catalytic oxidation method with Non Dispersive Infra Red (NDIR) detection which allows the TOC analysis in a wide range (4 µg/L to 30,000 µg/L) with a low limit of detection (4 µg/L) [19] .

The direct infusion analyses were carried out in a high resolution mass spectrometer (IT-TOF; Shimadzu Corporation, Kyoto, Japan), a hybrid ion trap and time of flight, equipped with an electrospray (ESI) ionization source operating in negative mode (−3.5 kV). The samples were directly introduced into the ESI source via the HPLC autosampler (SIL 30AC; Shimadzu Corporation, Kyoto, Japan). The mass spectrometer parameters are presented in Table 1 [19] [20] .

The evolution, formation, and degradation of each by-product during the photodegradation and photolysis processes were monitored by high performance liquid chromatography coupled to the hybrid mass spectrometry system. The liquid phase chromatograph was equipped with a binary pump (Nexera LC- 30AD; Shimadzu Corporation, Kyoto, Japan) and an autosampler (SIL 30AC; Shimadzu Corporation, Kyoto, Japan). The mass spectrometer parameters were the same as described above. The samples were introduced into the ESI source by injecting 8 µL of sample via the HPLC autosampler. For separation, it used a Nucleosil® 100-5 CN column (250 mm × 4.6 mm × 5 µm particle diameter); and as mobile phases, it used water (A) and acetonitrile (B), both contained 0.1% formic acid, at a flow rate of 1 mL・min⁻¹. The mobile phase was split so that the flow became 0.2 mL・min⁻¹ prior entering the electrospray capillary at the mass spectrometer. The gradient elution program used in separation of CP by-prod- ucts are presented in Table 2 [20] [21] [22] .

The assay is based on the capacity of mitochondrial dehydrogenase enzymes in living cells to convert the yellow water-soluble substrate 3-(4, 5-dimethylthiazol- 2-y1)-2, 5-diphenyl tetrazolium bromide (MTT) into a dark blue formazan product that is insoluble in water [23] [24] . Viable cells are able to reduce the yellow MTT under tetrazolium ring cleavage to a water-insoluble purple-blue formation which precipitates in the cellular cytosol and can be dissolved after cell lysis, whereas cells being dead following a toxic damage, cannot transform MTT. The amount of these crystals can be determined spectrophotometrically and hence the number of living cells in the sample. These features can be taken as advantage of cytotoxicity or cell proliferation assays, which are widely used in toxicology [23] [25] [26] .

The cytotoxicity of CP and its transformation products were determined with 3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay according to Mosmann (1983) and Zegura et al. (2009) [23] [25] . HepG2 cell monolayers were trypsinized, washed with culture medium, plated in 96-well flatbottomed plates (4 × 10⁴ cells per well) and incubated in a humidified atmosphere with 5% CO₂ at 37˚C. After a 24h incubation, serial dilutions of the CP samples (0.01, 0.25, 0.50, 0.75 and 1.00 mg・L⁻¹) made in phosphate buffered saline (PBS) were added to appropriate wells, and the plates were incubated for an additional 20 h. After this time, the supernatants were removed from the wells, and MTT (28 μl of a 5 mg/mL solution in PBS) was added to each well; the plates were incubated for 90 min at 37˚C; then, DMSO (130 μL) was added to each well to dissolve the formazan crystals. After shaking the plates to ensure complete dissolution of formazan, the optical density was determined at 490 nm in a multiwell spectrophotometer (Spectra max340PC-Molecular Devices) [21] [22] . The cell survival (viability) was determined by comparing the absorbance of the wells containing the cells treated with MTF solutions (treated and untreated) to the cells exposed to a negative control (culture media with cells). Positive control was done by pure DMSO. An initial solution of sodium hypochlorite 1 mg L^-1 was neutralized by sodium thiosulfate and used as blank. A 30% reduction in the viability of a given sample was considered as a positive cytotoxic response. These experiments were also performed in triplicate.

Changes in CP (negative ion m/z 216.0700) areas during ozonation and chlorination were monitored by high performance liquid chromatography coupled to the hybrid mass spectrometry system. The ozonation and chlorination promoted high degree of CP degradation, with 100% removal efficiencies after 5 min of exposure. Total organic carbon content (TOC/TOC₀) as a function of their reaction time are presented in Figure 2.

Ozone is unstable in water, however, the unique feature of ozone is its decomposition into hydroxyl radicals (•OH), which are the strongest oxidants in water. In water, •OH reacts fast with CP and the by-products formed from the reaction of •OH also reacts fast with CP. Ozonation may lead to a complete removal of organic compounds, but it does not lead to the complete mineralization of organics, which results in the formation of carboxylic acids, carbonyl compounds, and many others by-products [8] [10] [12] .

One hypothesis for CP removal by chlorination can also be attributed to the fact that reduced sulfur moieties can easily be oxidized in presence of chlorine. In the case of thiol-containing compounds, as captopril, thiols oxidation leads, mainly, to disulfide and sulfonic acid [6] [27] .

Although impressive CP degradation rates of 100% were achieved when using both systems, the TOC data revealed that CP was not mineralized to a similar extent, even after a treatment time as long as 30 min. For instance, the highest mineralization rate was only 7.63% and it was achieved upon the application of the ozonation. Moreover, the mineralization rates achieved through chlorination were lower (6.4%). These findings indicate that whereas most of the original CP was not mineralized, recalcitrant by-products were generated under these oxidative conditions.

The aliquots collected during the ozonation and chlorination experiments were analyzed by [ESI(-)-HRMS] and HPLC/HRMS. Examples of the mass spectra and extracted-ion chromatogram (EIC) containing peaks from the CP and its by-products are shown on Figure 3 and Figure 4, respectively. The mass spectra recorder in times of 0 and 30 min are depicted in Figure 3.

Figure 3 clearly shows that whereas the CP (m/z 216.0700) is fully consumed after 30 min of ozonation and chlorination experiments, the total carbon content in solution remained practically unchanged (Figure 2). The identification of the main by-products was carried out using the software formula predictor from Shimadzu, in order to propose a plausible degradation route. These ESI(-)- HRMS and HPLC/HRMS results allowed the detection of thirteen by-products, which elemental compositions were assigned based on the accurate mass measurements (<5 ppm) provided by the TOF analyzer, Figure 3(a) and Figure 3(b) [28] .

After a meticulous analysis of these data, it was possible to propose the molecular formula for the thirteen degradation by-products of CP detected in their deprotonated form, as shown in Table 3. A maximum error of 3.00 ppm between the experimental and theoretical accurate masses of CP and its proposed by-products was observed.

Based on these results as well as on the well-known reactivity of hydroxyl radical towards organic molecules in aqueous medium and isotopic ratio (Figure 5), a route for CP ozonation (Figure 6) and CP chlorination (Figure 7) could be proposed.

The by-product 1 (C₉H₁₅NO₆S, m/z 264.0547) can be formed from the oxidation of CP thiol group, and such oxidation probably happened because of sulfur’s ability of expanding its octet. The degradation progress follows different routes. The by-product 2 (C₉H₁₅NO₇S, m/z 280.0496) was probably formed after a hydroxylation of the intermediate 1, followed by its successive oxidation

forming a ketone group in by-product 3 (C₉H₁₃NO₇S, m/z 278.0340) [19] . The by-products 4 (C₉H₁₃NO₈S, m/z 294.0289) and 5 (C₉H₁₃NO₉S, m/z 310.0238),

which have an additional hydroxyl group, are probably formed from oxidation

of by-products 3 and 4, respectively. In sequence, the by-product 6 (C₈H₁₃NO₅S, m/z 234.0442) was probably formed after a decarboxylation of the intermediate 1 followed by its successive oxidation (forming a hydroxyl and then a ketone group) given that the medium is very oxidative. The by-product 7 (C₈H₁₃NO₆S, m/z 250.0391) has another hydroxyl group in the ring when compared to by- product 6. The by-product 8 (C₈H₁₃NO₈S, m/z 282.0289), and it has two hydroxyl groups in the ring when compared to by-product 7. The acyclic intermediate 9 (C₄H₈O₅S, m/z 167.0020) could be generated from intermediates (1-8). The by-product 10 (C₁₈H₂₈N₂O₆S_2, m/z 431.1316) can also be oxidized to form the CP (C₉H₁₅NO₃S, m/z 216.0700).

In the case of captopril thiol group, the oxidation mainly leads to sulfonic acid (by-product 1) and disulfide (by-product 10; detection was not possible) [6] [27] . It is noteworthy that the intermediate 1 has been previously reported by Mahmoud and Kummerer (2012), who studied the photodegradation and the aerobic biodegradation of CP and its dimer (captopril disulfide) in aqueous solution. The intermediate 1, 3, 6 and 7 has been previously reported by Freitas and co-authors (2017) who studied the photodegradation of CP in aqueous solution after oxidation induced by photolysis and photocatalysis [21] .

The evolution, formation, and degradation of each intermediate (1 - 10) during the ozonation and chlorination processes, were monitored by HPLC/HRMS. The results from EIC were analyzed to build plots of the temporal accumulation of intermediates for all processes evaluated (Figure 8). Hence, for the ozonation system practically all intermediates remained in solution after 30 min of exposure (Figure 8(a)). For the chlorination system (Figure 8(b)) only the intermediates 1 remained in solution after 30 min of reaction. In both systems it was observed an increase in the relative concentrations of by-products. Therefore the data presented in Figure 8 corroborates the results of TOC analysis and are in agreement with the proposed degradation route.

Cell survival (viability) was determined by comparing the absorbance of the wells containing the cells treated with CP solutions (treated and untreated) to

the cells exposed to a negative control (culture media with cells). The results were expressed in percentage of cell viability. It has to be noted that in order to avoid overestimation or underestimation of the toxicity of a substance, incubations with various concentrations were required to be able to accurately determine the cytotoxicity. Since none of the diluted samples tested decreased cell viability, Figure 9 shows cell viability data for undiluted (at the highest tested concentration, i.e., 1 mg・L⁻¹) CP samples before and after degradation experiments. In addition, the cells were exposed to DMSO, considered as positive controls. Figure 10 shows cell morphologies before and after the treatment of by-products. It can be seen that the morphology of the cells treated with the by-products solutions were not different when compared to the cells treated with CP. Cells treated with the positive control showed the spherical morphology thus showing that they are dead.

Cells viability exposed to all samples (raw and treated CP solutions) were the same as that of control cells. Values are expressed as mean and 95% confidence interval for the mean. This shows that there was no decrease of cells viability after their exposition to CP or its degradation by-products. Considering a 30% reduction in cell viability as cytotoxic effect [25] , it can be seen that none of the samples tested at the highest concentration showed toxicity to HepG2 cells. Unfortunately no reference on toxicity effects of CP or its by-products were found in the literature for comparison.