All chemicals used in the experiments were reаgent grаde аnd were utilized without further purification. The pharmaceutical NAD ((2R, 3S)-5-[3-(tert-butylamino)-2-hydroxypropoxy]-1,2,3,4-tetrahydronaphthalene-2,3-diol, C₁₇H₂₇NO₄, M = 309.4 g/mol, ≥99% purity, Sigma-Aldrich, Germany), and chemicals H₂O₂ (30%, Sigma–Aldrich, Germany) isopropanol (IPA, ≥98%, Sigma–Aldrich, Germany) and benzoquinone (BQ, ≥98%, Sigma–Aldrich, Germany) were used in this work. More information about applied chemicals is given in the Supplementary material.

Experiments were performed using 20 mL of 0.05 mmol/L NAD. In the indirect photolysis experiments, the solution contained 3.0 mmol/L of H₂O₂. The solution was irradiated for 180 min. The SS irradiation source was a 50 W halogen lamp (Philips, with the intensity of 1.9 × 10⁻² W/cm² in the visible region and 4.5 × 10⁻⁵ W/cm² in the UV region), and the UV-LED radiation source was a 5 W UV-LED Lamp (Enjoydeal, China, type: MR16 AC 85-265V/12, with intensity of 2.3 × 10⁻³ W/cm² in the visible region, and 2.2 × 10⁻⁴ W/cm² in the UV region). As the UV radiation source, a 125 W high-pressure Mercury lamp was used (Philips, HPL-N, emission bands at 290, 293, 296, 304, 314, 335, and 366 nm, with maximum emission at 366 nm and intensity of 5.0 × 10⁻³ W/cm² in the visible region and 1.6 × 10⁻³ W/cm² in the UV region). More information about energy fluxes measurement is given in Supplementary material.

All experiments were performed at the natural pH without any adjustment (c.a. 8.0), which declines during the photodegradation (Supplementary Table S1). To investigate the degradation mechanism, reactive radicals scavengers (2.0 mmol/L IPA or BQ) have been added to a water solution of NAD/H₂O₂.

For monitoring of kinetics of NAD photodegradation, aliquots of 0.40 mL were taken from the reactiоn mixture at the beginning of the experiment and at regular time intervals. A 20 μL sample was injected and analyzed using Ultra Fast Liquid Chromatography with Diode Array Detection (UFLC−PDA, Shimadzu), equipped with an Eclipse XDB−C18 column (150 mm × 4.6 mm i.d., pаrticle size 5 μm, 30°C). The UV/Vis PDA detectоr was set at 210 nm (wavelength of maximum absorption of NAD). Experimental conditions for the UFLC—DA can be found in the Supplementary material.

Absorption spectra of NAD was recorded on a double-beam T80+ UV/Vis Spectrometer (UK) at a fixed slit width (2 nm), using a quartz cell (1 cm optical length) and computer-loaded UV Win 5 data software. The absorption maximum of NAD was 210 nm (Supplementary Figure S1).

During the degradation process, a combination glass electrode (pH-Electrode SenTix 20, WTW) connected to a pH meter (pH/Cond 340i, WTW) was used to track the initial pH values and how they changed over time (Supplementary Table S1).

The degree of hydroxyl radicals neutralization obtained by Fenton’s reaction was measured as the ability of hydroxyl radicals to induce lipid peroxidation of liposomal unsaturated fatty acids, followed by the formation of a colored adduct between malondialdehyde and thiobarbituric acid at 532 nm (Kladar et al., 2015). Catalase activity was measured after incubation (1 h at 37°C) of 100 µL of heparinized blood and 200 µL of analyzed samples (Jelić et al., 2018). The reduction rate of H₂O₂ was followed at 240 nm for 3 min at 25°C, and the percentage of catalase activity was calculated. The lipid peroxidation process in heparinized blood incubated with investigated samples was determined according to the previously described liquid chromatography based analytical method (Khoschsorur et al., 2000).

The COD concentration was measured using spectrometry by measuring the absorbance of the formed Cr⁺³ at 600 nm. Absorbance was measured on a double-beam T80+UV-Vis. The samples were digested at 150°C for 2 h.

The in vitro toxicity effects were assessed using the human cell lines colon adenocarcinoma HT-29 (ECACC 91072201) and fetal lung MRC-5 (ECACC 05090501). Cell growth was evaluated by the colorimetric sulforhodamine B (SRB) assay, according to Skehan et al. (1990), modified by Četojević-Simin et al. (2013). The effect on cell growth was calculated as (A _t/A _c) × 100 [%], where A _t is the absorbance of the test sample and A _c is the absorbance of the control. The experimental details about the cell growth activity are described in Supplementary material.

DFT calculations have been performed using the B3LYP density functional in combination with 6-31G (d,p) basis set (Ditchfield et al., 1971; Vosko et al., 1980; Lee et al., 1988; Becke, 1993; Stephens et al., 1994; Rassolov et al., 1998; Rassolov et al., 2001). All molecules were subjected to frequency calculations to check that geometrical optimizations identified the true ground states (Supplementary Figure S2). During geometrical optimizations, frequency and property calculations, the effects of water were taken into account in the framework of the Poisson-Boltzmann solver. DFT calculations were performed with the Jaguar program (Bochevarov et al., 2013; Cao et al., 2016; Jacobson et al., 2017; Cao et al., 2021), as implemented in the Schrödinger Materials Science Suite 2022-2. NMR chemical shifts were also obtained via DFT calculations, according to the approach described in the paper by Cao et al. (2005). Before geometrical optimizations via the DFT approach, the initial structure of NAD was pre-optimized using the GFN2-xTB method (Grimme et al., 2017; Bannwarth et al., 2019; Pracht et al., 2019; Bannwarth et al., 2021), as implemented in the xtb program, developed by the group of Prof. Grimme. GFN2-xTB calculations were performed by implicitly including the influence of water as the solvent in the framework of the Analytical Linearized Poisson-Boltzmann (ALPB) model (Ehlert et al., 2021). Calculations with xtb program were performed via the online molecular modeling platform atomistica.online, available at https://atomistica.online Armaković and Armaković (2022).

If the degradation results are compared under the influence of different radiation sources (Figure 1), significantly higher efficiency is observed during degradation under UV radiation than degradation along SS and UV-LED. After 180 min, 3.0% and 11.0% of NAD were degraded under SS and UV-LED radiation, respectively (Figure 1A). The degradation efficiency was slightly higher using UV radiation, wherein 14.7% of the NAD was degraded for the same time. The obtained results are expected considering that absorptive peaks of NAD are at 210 and 274 nm, which contributes to a higher degradation efficiency by applying a high-pressure Mercury lamp than a halogen lamp. For effective direct photolysis, there must be an appropriate overlap between the investigated compound absorption spectrum (Supplementary Figure S1) and the corresponding radiation source spectrum (Bonvin et al., 2013). Fujioka et al. (2020) have determined that UV-LED irradiation sources are less effective than UV irradiation sources for degrading organic compounds with absorption maximums in the UV area due to the longer emission wavelengths of the UV-LED lamps.

To increase the degradation efficiency of NAD, indirect photolysis in the presence of H₂O₂ has been investigated under SS, UV-LED, and UV radiation. H₂O_2, as an oxidant with a high oxidation potential, can oxidize many different organic and inorganic substrates under very mild conditions in a wide range of pH values (Goyal et al., 2020). The obtained results show that in comparison to the direct photolysis of NAD, adding H₂O₂ increased degradation efficiency under all three types of irradiation (Figure 1B). Further, indirect photolysis of NAD was the most efficient under UV radiation, whereby 80.2% of NAD was degraded. Lower degradation efficiency was achieved under UV-LED and SS irradiation, where 23.1% and 4.5% of NAD were degraded after 180 min, respectively. It is well known that UV irradiation of H₂O₂ in aqueous solutions leads to the production of ^•OH, and other reactive species, such as O 2 • − since H₂O₂ absorbs light in the 185–300 nm range. These radicals react with the substrate, forming intermediates that further react with ^•OH radicals until complete mineralization. ^•OH radicals are short-lived and highly reactive chemical species that indiscriminately react with NAD and its intermediates in water. ^•OH radical’s action mechanism on NAD consists of several processes: elimination of hydrogen, electrophilic bonding, and electron transfer. Therefore, the higher degradation efficiency of NAD in the presence of UV-LED/H₂O₂ and UV/H₂O₂ was anticipated.

The initial pH value of the NAD solution under UV is 8.2. During direct degradation by photolysis, the value increases to 9.2 (Supplementary Table S1), probably due to the formation of slightly basic intermediates concerning the starting compound. The initial pH value of the solution was somewhat lower in the presence of H₂O₂ than in pure NAD solution. During indirect photolysis, the pH value of the solution constantly decreases, most likely due to the formation of slightly acid intermediates in the starting compound. A similar trend is after 180 min indirect photolysis, the value decreases from 7.7 to 6.8.

Many organic compounds are mineralized during UV/H₂O₂ treatment since ^•OH radicals are non-selective, highly reactive species (Lopez-Alvarez et al., 2016; Ghanbari et al., 2021; Santolin et al., 2022). The final products include CO₂, H₂O, and inorganic ions. However, it is crucial to know that high mineralization levels are rarely reached since organic compounds, such as pharmaceuticals, are very stable in water. The mineralization of investigated solutions for different degradation processes, shown as COD measurements, determined that 4.1% and 22.2% of NAD and its degradation intermediates were removed after 180 min under UV-LED and UV irradiation (Figure 2). Based on the presented results, we can conclude that the UV/H₂O₂ treatment proved to be effective in the case of NAD degradation but not in the case of its intermediates. (Figures 1, 2).

After 180 min of UV irradiation, the content of organic intermediates in the suspension was approximately 78%. The difference in the efficiency obtained based on COD and UFLC-DAD determination results from forming various aromatic and aliphatic intermediates, considering that UFLC monitors only the NAD. The COD method monitors intermediates that arise from indirect photolysis. Also, it may be inferred that a mixture of NAD and its organic intermediates formed under UV-LED radiation are stable and not subject to decomposition under the mentioned radiation sources.

Based on the obtained results, we can conclude that the degradation of NAD is accompanied by the formation of intermediates, especially under UV irradiation, that can potentially harm the environment. Having that in mind, it is essential to identify them and evaluate reactivity and toxicity.

To understand and compare the reactivity of NAD and its photodegradation intermediates, the interaction of NAD and its degradation mixtures with H₂O₂ and free radicals was investigated. The percentage of catalase activity was used to assess the interaction with H₂O₂, the degree of inhibition of lipid peroxidation was used to evaluate the interaction with ^•OH radicals, and the degree of hydroxyl radicals neutralization was used to assess the prooxidant or antioxidant activity of obtained degradation mixtures (Figure 3).

In this part of the experiment, catalase activity reflects residual H₂O₂ in the sample mixtures, while the lipid peroxidation process reflects the presence of the intermediates with the potential ability to induce, along with hydroxyl radicals, this process or to reduce it, acting as scavengers of ^•OH radicals. Further, the degree of hydroxyl radical neutralization indicates the presence of NAD since it acts as a direct hydroxyl radical scavenger. In comparison to UPW (Figure 3), results suggest that NAD, itself, scavenges ^•OH radicals (8% vs. 22%), while the catalase activity and lipid peroxidation process have insignificant influence. Among three indirect photolysis groups, solar radiation produces the smallest amount of hydroxyl radical based on the results of CAT activity. In this group, during the incubation period, an enzyme in the blood was involved in the decomposition of H₂O₂, and as a consequence, its activity was lowest at the end of it. Our assumption is that indirect photolysis with simulated solar radiation induces degradation of NAD in the manner that degradation intermediates exert antioxidant effects since the lipid peroxidation process is drastically decreased (−21%), while the scavenging ability of ^•OH radical increased in comparison to UPW and NAD group (37% in comparison to 8% for UPW, and 22% for NAD group). Contrarily, UV-LED and UV indirect photolysis induce the conversion of H₂O₂ to hydroxyl radicals since CAT activity is near the level of UPW (9% and 0%, respectively). Additionally, during the process of NAD degradation, prooxidant intermediates are formed, in contrast to SS indirect photolysis, since the lipid peroxidation process is more than pronounced (32% and 56%). Comparing the results of UV-LED and UV indirect photolysis, our assumption is that there is a slight difference in the degradation path of NAD. Namely, our assumption is that UV induces more prooxidant intermediates able to scavenge ^•OH radicals than UV-LED, while UV-LED induced photolysis induces additional antioxidant intermediates since the degree of hydroxyl radical neutralization is higher (11% versus 29%).

In this work impact of NAD (Figure 4) and mixtures formed during indirect photolysis on the growth of human cell lines HT-29 and MRC-5 was assessed in vitro. The obtained results depended on the cell line and the type of applied radiation. In vitro bioassays using mammalian cell lines can reveal the combined effect of interacting chemical compounds and complex mixtures found in environmental samples and their potential additive, synergistic, or antagonistic effects. Mammalian cell lines are potent tools for predicting the toxicity of a single compound or a mixture of compounds towards higher trophic levels organisms and their organ-specific effects (Četojević-Simin et al., 2013).

A comparison of the in vitro toxicity of NAD and evolution of its degradation mixtures obtained under different irradiation types revealed mild cell growth stimulation (<4%) of fetal lung cells MRC-5 and mild to moderate (2%–19%) cell growth stimulation of colon adenocarcinoma cells HT-29. Both parent compound NAD and mixtures obtained after UV/H₂O₂ treatment revealed similar cell growth stimulation patterns in both cell lines, indicating no change in the cell growth effects after irradiation treatment compared to the parent compound. On the other hand, after treatment with UV-LED/H₂O₂ obtained mixture had minimal effect on the growth of both selected cell lines, indicating the evolution of non-toxic intermediates.

A reaction mechanism is a microscopic path by which reactants are transformed into products. Therefore, it is crucial to identify intermediates formed by NAD degradation. The experiments have shown that using UV/H₂O₂ leads to higher degradation efficiency due to the generation of reactive species (reactions 1-3) (Oancea and Meltzer, 2014). H 2 O 2 + h v   →   2 • O H (1) H 2 O 2 + O • H   →   H O 2 • + H 2 O (2) H 2 O 2 + O • H   →   O 2 • − + H + + H 2 O (3)

According to literature data, reactive oxygen radicals (^•OH and O 2 • − ) significantly influence the decomposition of organic compounds (Liu et al., 2018). As mentioned in the previous papers (Cai et al., 2022a; Cai et al., 2022b; Li et al., 2022b), different radical scavengers (methanol, benzoic acid, tert-butanol, furfuryl alcohol) could be used for identification of generated reactive radicals. IPA and BQ were chosen as ^•OH and O 2 • − scavengers because of the elevated reactivity of IPA with ^•OH (2×10⁹ L/(mol×s)) (Vione et al., 2010) and BQ with O 2 • − (1.1×10⁹ L/(mol×s)) (Zhu et al., 2020), respectively. When these reactive radical scavengers are present, the efficiency of NAD degradation decreases significantly, which indicates that the mechanism of degradation of the investigated compound takes place efficiently precisely in the presence of the mentioned reactive species.

To investigate the influence of ^•OH and superoxide radicals O 2 • − on NAD suspensions, we have added IPA or BQ. As shown in Figure 5, the degradation efficiency of IPA quenching decreased more than that of BQ. After 180 min of degradation, 70% and 20% NAD was less degraded in the presence of IPA and BQ scavengers, respectively. O 2 • − is an essential component in many chemical processes (Hayyan et al., 2016), but indirect photolysis of NAD has the lowest impact on its degradation.

To better understand the interaction between NAD and ^•OH radicals, we conducted a DFT computational analysis of the NAD/^•OH system. Ten configurations were obtained (Supplementary Figure S2) by optimizing the investigated system’s molecular geometry, which, together with the distance between certain atoms, are given in Table 1. The different location of the ^•OH radical concerning the NAD molecule indicates other interactions that can take place between them. Namely, it is known that the ^•OH radical reacts with organic molecules by either adding to unsaturated bonds or hydrogen abstraction (Panchenko, 2006). Since in configurations 3 and 5, the shortest distance of NAD and ^•OH radicals is between the oxygen atom of the ^•OH radical and the unsaturated carbons of the NAD ring (C13 and C16, respectively), the mechanism of addition of ^•OH radical to unsaturated bonds can be assumed.

On the other hand, in configuration 2 the ^•OH radical is closest to the hydrogen from the aliphatic chain, and in configurations 1, 4, 6, and 7, to the ring, which would mean that abstraction of those hydrogens atoms could occur. From the above, it can be assumed that the primary reaction mechanism in the observed system is hydrogen abstraction. For configurations 8, 9, and 10, it is interesting to note that ^•OH radical abstracted a hydrogen atom connected to the nitrogen atom bonded, whereby a water molecule is formed.

In terms of the suggestion of tentative degradation intermediates and predicting sensitivity towards autoxidation, bond dissociation energy for single acyclic bonds (BDE) and for hydrogen abstraction (H-BDE) are shown to be very useful. Namely, the H-BDE values in the range of 70–85 kcal/mol indicate high sensitivity towards autoxidation, and H-BDE values in the range of 85–90 kcal/mol also indicate sensitivity towards autoxidation but must be taken with caution (Lienard et al., 2015; Wright et al., 2009; Gryn’ova et al., 2011). BDE and H-BDE values are reported in Figure 6. H-BDE values indicate that NAD is not sensitive toward autoxidation. It can be presumed that NAD degradation could begin with the detachment of the -R group from the oxygen atom on the tetrahydronaphthalene ring. BDE values of C-C bonds in the NAD aliphatic chain (Figure 6) show that degradation could be continued by breaking these bonds. The bonds connecting the -OR group to the tetrahydronaphthalene ring and the C-OH bonds have the highest calculated BDE values. It was assumed that these bonds hardly break.

According to the obtained results, the formation of intermediates 1, 2, and three was proposed due to the addition of ^•OH radicals to the NAD ring. On the other hand, it was shown that there is the possibility of abstraction of hydrogen atoms by ^•OH radicals (Table 1). This mechanism could lead to the formation of NAD radical, which could further react with ^•OH and O 2 • − . In the presence of identified radicals, the ring opening mechanism is possible, so the appearance of the intermediates from 4 to 6 and from 8 to 15 is suggested. Liu and Williams (2007) proposed three major propranolol photodegradation products due to ring oxidation in the presence of oxygen. Since NAD and propranolol are the most structurally similar β-blockers (both have two rings), it can be assumed that they will have similar degradation pathways during photodegradation (intermediates 14 and 15). Uwai et al. (2005) suggested the oxidation of the propranolol aromatic ring and the aliphatic separation part of molecules. A similar could be recommended for NAD. As stated above, intermediates 8, 9, and 10 are presented in Figure 7. Yang et al. (2010) gave a general scheme for the photodegradation of β-blockers in water in which cleavage of the side chain can also occur in the presence of ^•OH radicals. Accordingly, the reaction of ^•OH radicals with NAD leads to the formation of intermediates 7 and 8. BDE values also go in favor of formation intermediates 7 and 8. Also, intermediates 11, 12, 13, and 16 were proposed.