Hydroxyl Radical Production by Air Pollutants in Epithelial Lining Fluid Governed by Interconversion and Scavenging of Reactive Oxygen Species

Air pollution is a major risk factor for human health. Chemical reactions in the epithelial lining fluid (ELF) of the human respiratory tract result in the formation of reactive oxygen species (ROS), which can lead to oxidative stress and adverse health effects. We use kinetic modeling to quantify the effects of fine particulate matter (PM2.5), ozone (O3), and nitrogen dioxide (NO2) on ROS formation, interconversion, and reactivity, and discuss different chemical metrics for oxidative stress, such as cumulative production of ROS and hydrogen peroxide (H2O2) to hydroxyl radical (OH) conversion. All three air pollutants produce ROS that accumulate in the ELF as H2O2, which serves as reservoir for radical species. At low PM2.5 concentrations (<10 μg m–3), we find that less than 4% of all produced H2O2 is converted into highly reactive OH, while the rest is intercepted by antioxidants and enzymes that serve as ROS buffering agents. At elevated PM2.5 concentrations (>10 μg m–3), however, Fenton chemistry overwhelms the ROS buffering effect and leads to a tipping point in H2O2 fate, causing a strong nonlinear increase in OH production. This shift in ROS chemistry and the enhanced OH production provide a tentative mechanistic explanation for how the inhalation of PM2.5 induces oxidative stress and adverse health effects.


ELF 118
OH reacts with nearly all matter present in the ELF with a rate coefficient approaching diffusion limitation. 19 119 Because of this unspecific reactivity, effective scavenging of OH radicals, e.g. through lung antioxidants, is 120 not possible. 19,20 As an estimate, we assume that the amount of protein in the ELF corresponds to the total 121 amount of dissolved organic matter. The protein mass in the ELF amounts to approximately 10 mg per mL 122 lung fluid. 3 Using an average molecular weight of ~125 g mole -1 of a single amino acid, the total amino acid 123 concentration in the ELF can be estimated to ~80 mmol L -1 . A second-order reaction of amino acids with 124 OH is included, using a reaction rate coefficient on the order of 1.66 x 10 -12 cm -3 s -1 (R122, Tab. S1). 21

125
Because proteins are folded, not all amino acids will be surface exposed, and thus accessible reaction 126 partners for OH. In general, spherical proteins have fewer surface exposed amino acids due to a smaller 127 surface-to-volume ratio, while elongated, cuboid or conical proteins have more surface exposed amino 128 acids. Furthermore, the surface exposure of amino acids depends on the physical properties, e.g. the 129 polarity of the respective amino acid. 22 Therefore, as an order of magnitude estimation, we assume that 130 50% of all amino acids are surface exposed in the ELF, yielding an effective amino acid concentration of 131 ~40 mmol L -1 and, in turn, a lifetime of OH with respect to reaction with dissolved organic matter of 2.5 × 132 10 -8 s.

134
Pryor estimated the lifetime of OH in a cell to 10 -9 s, assuming a rate coefficient of 1 x 10 9 M -1 s -1 (equivalent 135 to 1.66 x 10 -12 cm -3 s -1 ) and an effective organic matter concentration of 1 mol L -1 . 21 From comparing the 136 ELF protein mass of ~10 mg mL -1 to the cellular protein mass of ~250 mg mL -1 , we infer that the ELF must 137 be about ~25 times more dilute compared to a cell with respect to dissolved organic matter. 2,23,24 Multiplying 138 Pryor's OH lifetime in cells with this dilution factor yields an estimate for the OH lifetime in ELF of ~2.5 × 139 10 -8 s, which is identical to the estimate above and consolidates our description of OH reactivity.

141
We find that inclusion of this second-order loss reaction of OH results in a decrease of momentary OH 142 concentrations by one order of magnitude compared to our earlier calculations (Fig. 2a). 1 This finding 143 suggests that OH will react unspecifically with organic matter and only secondarily with antioxidants (7%) 144 in the ELF. Due to the fast reaction of OH, spatial gradients of reactants could play a role in OH fate, e.g.

145
through local depletion of antioxidants around a dissolving particle or inhomogeneous distribution of organic 146 matter and PM2.5 constituents in the ELF. However, for the calculations in this study, starting 147 concentrations of antioxidants and organic matter were homogeneous across the bulk ELF.

154
Inhaled particles can reside in the ELF for several hours. 25 For this study, we assume a residence time of 155 PM2.5 of 2 hours and use this as accumulation time of inhaled particles (tacc) and simulation time (tsim) to 156 mimic a pseudo steady-state of ROS concentrations that would be achieved through continuous inhalation, 157 in line with our previous studies. 1 It should be noted that there is some uncertainty regarding the residence 158 time of PM2.5 in the ELF, with estimates on PM2.5 clearance also exceeding 2 hours. 26 As lung ventilation 159 rate, VR, 1.5 m 3 h -1 is used, and the PM2.5 deposition fraction in the ELF, dPM2.5, is assumed to be 0.45. 1,27

160
The total ELF volume, VELF, is set to 20 mL. 1,28 This study only includes copper and iron as transition metals, 161 because these are the only two that have been shown to significantly produce ROS in surrogate ELF 162 (sELF). 29,30 The fractional solubilities, SFY, of copper and iron ions were discussed and tested extensively  Tables S4-S6. Not all references in Tables S4 and S6 include PM2.5 concentrations. In such 169 cases, PM2.5 concentrations are estimated based on similar geographical locations and indicated with an 170 asterisk. Additionally, secondary organic aerosol (SOA) forms ROS and is included in the model. [32][33][34] 171 However, because the exact mechanism of ROS formation by SOA in the ELF, first order formation rates 172 of H2O2 and OH by SOA were parameterized based on experimental observations. 1,32,33 Quinones in PM2.5 173 are included in this study as previously described. 1 Three quinones are included that were shown to form 174 ROS in sELF, phenanthrenequinone (PQN), 1,2-naphthoquinone (1,2-NQN), and 1,4-naphthoquinone (1,4-175 NQN) in a molar ratio of 2:1:1. 29 176 177

S5. Gas-phase pollutant concentrations in the ELF 178
Exposure to gas-phase oxidants, O3 and NO2 is quantified in the model using a simplified breathing 179 mechanism. In our previous study, it was assumed that the concentrations of these gas-phase oxidants in 180 the lung were equal to their respective ambient concentrations. However, because of the reactivity of these 181 oxidants, we find that lung gas-phase concentrations of these oxidants are depressed and limited by supply 182 from inhalation of ambient air. In order to get accurate estimates for the amount of gas-phase oxidants transferred to the surfactant layer and ELF, an average lung volume of four L, an average breath volume 184 of 1.5 L and an average duration of a breath of 3.6 s are used to compute mass fluxes into and out of the 185 lung (Table S2). Application of this simplified breathing mechanism results in a significant decrease in the 186 amount of gaseous oxidants in the surfactant layer and ELF. Therefore, neither O3, nor NO2 are saturated 187 in ELF with respect to their ambient concentrations in this study (Fig. S5).

S6. Acid dissociation 190
In this study, corresponding acid/base-pairs are treated as a single species in the numerical computation 191 of ordinary differential equations (ODE). This effectively reduces the stiffness of the ODE system and 192 applies to glutathione (GSH/GS -), superoxide radicals (HO2/O2 -) and peroxynitrous acid (ONOOH/ONOO -).

193
Instead of treating each species explicitly with separate differential equations and explicit protonation and 194 deprotonation reactions, the pKa of these species was used to calculate the acid/base-ratio at the pH of 195 the ELF (Table S2). Then, if a reaction requires only one of the two species to react, the rate of that reaction 196 was multiplied with the inferred fraction of the reacting species.

S7. pH of the ELF 199
Following estimations by Holma (1985Holma ( , 1989, the pH of the ELF was assumed to stay constant upon air 200 pollutant exposure. 35,36 In diseased individuals such as asthmatics 37 , chronic obstructive pulmonary 201 disease, 38 or cystic fibrosis patients 39 the pH of the ELF may be decreased. Figure S6 shows the ROS 202 concentration, production, interconversion and transition metal valence state at pH 7 and pH 4 as a function 203 of PM2.5 concentration. Panel a shows that CΣROS displays a very similar behavior at pH 4 and pH 7. CO2-204 and CHO2 are slightly increased at low PM2.5 concentrations, but depressed at elevated PM2.5 205 concentrations (panel b). This reduction is due to a higher rate of Fe 2+ -mediated interconversion of HO2 206 and O2to H2O2 (Table S1, R38 and R39), which in turn is due to a higher Fe 2+ /Fe 3+ ratio (panel c). The

207
Cu + /Cu 2+ ratio shows the opposite trend at reduced pH (panel j).

208
Panel e shows that at pH 4, NΣROS is marginally reduced compared to pH 7. In panel f, NΣROS is broken 209 down to its components. PO2-is slightly reduced due to decreased Cu + -dependent O2formation (Table S1, 210 R54). POH is slightly increased due to the higher Fe 2+ /Fe 3+ ratio and the according increase in the Fenton 211 reaction (Table S1, R40).

212
Panel g shows that CFO2-→H2O2 is slightly decreased at low PM2.5 concentration and slightly increased at

216
Reduced pH in the ELF may additionally lead to a reduction in antioxidant enzyme activity, 17 increased 217 transition metal solubility, 40 and increased OH yield from the Fenton reaction. 41,42 These effects are not 218 included in the presented study, and are expected to all reduce ROS buffering and promote the PM2.5-219 controlled OH radical production regime, which may exacerbate oxidative stress. Furthermore, in the 220 presented study SOA produces H2O2 and OH, 32,33 which at lower pH may increasingly shift towards only 221 H2O2 production, without OH getting formed. 43,44 However, a thorough investigation of pH effects is beyond 222 the scope of the presented study and warrants future investigations.