Endogenous Nitric Oxide Can Enhance Oxidative Stress Caused by Air Pollutants and Explain Higher Susceptibility of Individuals with Inflammatory Disorders

Air pollution causes morbidity and excess mortality. In the epithelial lining fluid of the respiratory tract, air pollutants trigger a chemical reaction sequence that causes the formation of noxious hydroxyl radicals that drive oxidative stress. For hitherto unknown reasons, individuals with pre-existing inflammatory disorders are particularly susceptible to air pollution. Through detailed multiphase chemical kinetic analysis, we show that the commonly elevated concentrations of endogenous nitric oxide in diseased individuals can increase the production of hydroxyl radicals via peroxynitrite formation. Our findings offer a molecular rationale of how adverse health effects and oxidative stress caused by air pollutants may be exacerbated by inflammatory disorders.


S1. Overview KM-SUB-ELF
The kinetic multi-layer model of surface and bulk chemistry in the epithelial lining fluid (KM-SUB-ELF) 1 describes chemical reactions and mass transport in the lungs, at the intersection of atmospheric and physiological chemistry, with the goal of providing a chemical rationale for the adverse health effects or air pollution.A schematic overview of the model is shown in Figure S3.
The temporal evolution of reactants and reaction products is calculated by solving a set of differential equations at high temporal resolution with MATLAB software using the ode23tb solver that employs TR-BDF2 to consider the stiffness of chemical mechanisms.Compared to previous publications, KM-SUB-ELF was extended by the molecules • NO and CO 2 .The gas-phase respiratory tract CO 2 concentration was assumed to be 3.8%.Table S1 shows the chemical mechanism used in this study, which includes redox chemistry of transition metals, HO x and NO x radical chemistry, antioxidant redox couples and redox reactions in a total of 139 reactions.Of these, 23 are gas-phase chemical reactions adopted from the Master Chemical Mechanism (MCM), 2,3 and 116 are aqueous-phase reactions within the ELF, of which six occur in the surfactant layer.

S2. Exhaled • NO
Exhaled • NO (eNO) originates from inducible nitric oxide synthase (iNOS), 4 an enzyme that is present on macrophages, and produces • NO from L-arginine and oxygen. 5Humans typically exhale • NO in concentrations that exceed those in the atmosphere.Thus, the ELF is a source of • NO to the gas phase of the respiratory tract.In the model, ambient • NO levels have a similar effect on P OH as endogenously produced • NO, where a 10 ppb increase leads to an increase P OH of about 1 nM in a standard pollution and exposure scenario in both healthy and diseased individuals (Figure S4a).Accordingly, the relative increase of P OH due to eNO in diseased individuals (ΔP OH ) is slightly reduced at higher levels of ambient • NO (Figure S4b).However, ambient • NO levels are typically in the single-digit ppb range, much lower than the eNO levels found in diseased individuals.At such low levels, the effect of ambient • NO is close to negligible.As for all volatile species, mass transfer of • NO between the ELF and the gas phase of the respiratory tract is determined through explicit adsorption and desorption fluxes, which depend on molecular collision rates and Henry's law coefficient. 6In our model, a constant flux of • NO from the underlying epithelium to the ELF is fitted to achieve • NO concentrations reported in the exhaled breath of healthy volunteers, as well as rhinitis, COPD, bronchitis and asthma patients, as listed in Tab. 1 ('Controls' and 'Patients').It is noted that differences in measured • NO may arise from differences in sampling technique, as detailed in Kharitomov et al. (1997).For instance, nasal breathing (NB) results in readings that may be one order of magnitude higher compared to oral breathing because of • NO production in the sinuses.

S3. Particulate pollutant concentrations in the ELF
The ELF concentrations of redox-active PM 2.5 constituents (C ELF,Y ) are calculated as described previously, 8 using Eq.S1.
Mass fractions (w) of redox-active PM 2.5 constituents Y are derived using field observations of approximately 70 sampling sites which have previously been reported and which are tabulated in Tabs.S2-S4. 8The sampling sites represent a large range from pristine Amazonian rainforest air to the heavily polluted, hazy conditions in Beijing, China.For 'standard' PM 2.5 composition, we derive the median mass fractions of PM 2.5 constituents from all sampling sites.However, to show model sensitivity to PM 2.5 composition, and transition metals in particular, the reported PM 2.5 compositions are used explicitly in the calculations for Figure 4A.The data from sampling sites we used for Figure 4A are marked with an asterisk in Tab.S2.It was assumed that 45% of all the respired PM 2.5 deposits in the respiratory tract during a 2-hour exposure episode (f dep,PM2.5 ).
Solubilities (S) of the PM 2.5 constituents copper and iron were assumed to be 40% and 10%, respectively.The ventilation rate (Q) of the respiratory tract is calculated using a breath volume of 1.5 L and a breathing rate of 16 breaths per minute.

S4. Parameterization of ROS formation from secondary organic aerosol
0][11] Because the exact reaction mechanism has not been fully elucidated, ROS formation by SOA in the ELF is parameterized using formation rates of H 2 O 2 and • OH based on experimental observations. 1,9,10In this study, a first-order rate coefficient of H 2 O 2 production is inferred from experimental observations using αand β-pinene that found a 0.6% H 2 O 2 mass yield from SOA, 9 following Lakey et al. 1 Tong et al.
(2016) quantified • OH production of SOA, from which we infer first-and second-order rate coefficients that reproduce a molar yield of 0.1% in the absence of iron (R137; Tab.S1), and a 1% yield of • OH in the presence of iron (R139; Tab.S1), respectively, in the experimental data.

S5. Antioxidant concentrations
In this study, four antioxidants in the ELF are included.Ascobate, glutathione, uric acid have concentrations of 40, 108 and 200 µM in the aqueous ELF, respectively, whereas α-tocopherol has a concentration of 200 µM in the surfactant layer of the ELF. 1,8,12,13Previous studies suggest that the antioxidant concentrations in the ELF may vary within a factor of two between healthy and diseased individuals. 14,15For instance, the antioxidant levels in patients suffering from COPD have been found to be slightly increased compared to healthy individuals, 14 whereas asthmatics have slightly decreased antioxidant levels in their ELF. 15Here.we perform a model sensitivity study and show the effect of antioxidant concentrations on P OH at diseased and healthy eNO levels (Figure S5).We find that in diseases in which the antioxidant levels decrease (FC below 1), the relative effect of eNO becomes stronger (Figure S5).In diseases in which the antioxidant levels increase (FC above 1), the effect of eNO compounds with the transition-metal recycling effect of antioxidants thus increases the susceptibility of individuals to air pollution.
Studies using healthy volunteers suggest that even in extreme pollution scenarios, i.e. 1 ppm NO 2 for several hours, the antioxidants in the ELF do not fully deplete, 16 likely due to fast replenishment.All concentrations of antioxidants were thus kept constant in the model simulations in this study.

S6. Enzyme reactions and concentrations
The catalytic activity of enzymes in the ELF is implemented as previously described. 8In brief, the molar concentrations of enzymes are calculated from enzyme activity in enzyme units (U), the catalytic constant, k cat , and Eq.S2.
One U is defined as the quantity of enzyme needed to catalyze 1 micromole of substrate per minute.
In experiments quantifying U in biological samples (e.g. in the ELF), the substrate of the enzyme is kept in vast excess, thus U = .k cat describes the maximum number of chemical conversions   a single active site (in our case a single enzyme) can carry out in a second.
[Enzyme] =     (Eq.S2) In the ELF, two enzymes are mainly responsible for the interconversion and scavenging of ROS. 8,17In a dismutation reaction, two superoxide ( • O 2 ˉ) molecules are converted into H 2 O 2 and O 2 by superoxide dismutase (SOD; R127; Tab.S1).U SOD,ELF is 36.8 ± 2.0 U mL -1 , 17 and k cat,SOD is reported to range between 10 5 -10 6 s -1 . 18,19These values translate to a concentration range of SOD between 0.58 -6.5 nM (~1 nM) in the ELF.Previous experiments have shown that catalase (CAT) is the most important endogenous H 2 O 2 scavenger in the ELF. 17U cat,ELF is 3.7 ± 0.6 U mL -1 , 17 and k cat,CAT is reported to range between 10 5 -10 6 s -1 .These values translate into a molar concentration ranging from 1.3 -24 pM (~5 pM) in the ELF.The glutathione peroxidase (GPx) concentration in the model ELF is 50 nM based on a molar mass of 2.19 × 10 4 g mol -1 and a mass concentration of 1 µg mL -1 in ELF. 20We note that for the ELF, U, k cat , and protein masses are difficult to determine because the ELF is very difficult to sample.We thus acknowledge that the enzyme concentrations used in this study are subject to uncertainty.Table S1.Chemical reactions, rate constants as used in the KM-SUB-ELF, with reference.

# Reaction
Rate constant (cm -3 s -1 or s -1 ) Ref.     See SI text SOA + Fe 2+ → • OH + Fe 3+ 7.90 × 10 - 23 See SI text 188 S1 Table S2.PM 2.5 and transition metal concentrations with mass fractions as quantified in PM 2.5 collected at different sampling sites throughout the world.Sampling sites used in Figure 4A are indicated with an asterisk and were selected to represent diverse parts of the globe, a large variability in PM 2.5 mass and composition.

FiguresFigure S1 :Figure S2 :Figure S3 :Figure S4 :
FiguresFigure S1: Products of peroxynitrite and peroxynitrous acid decomposition and sink reactions Figure S2: Sensitivity of P OH on PM 2.5 mass and transition metal mass fraction Figure S3: Schematic overview of KM-SUB-ELF and principal model inputs and typical model outputs Figure S4: Sensitivity of P OH on ambient • NO Figure S5: Effect of decreased and increased antioxidant levels on • OH production

Figure S1 .
Figure S1.Relative contribution of various chemical reactions that contribute to the loss of peroxynitrite (ONOO -) and peroxynitrous acid (ONOOH)."Oxidized AOs" stands for reaction partners of ONOO -that are grouped as antioxidants.Those antioxidants include glutathione, uric acid and ascorbate as well as the antioxidant enzyme glutathione peroxidase.Nitrate (NO 3 -) and• OH result from first-order decomposition of ONOOH, and the former also results from CO 2 reacting with ONOO -.The carboxyl radical anion (CO 3 •-) is also a result of CO 2 reacting with ONOO -.

Figure S2 .
Figure S2.Gross chemical • OH production, P OH , using the PM 2.5 mass and composition measured in many locations worldwide, as presented in Tab.S2.The color-coding of data points indicates the mass fractions of (A) iron and (B) copper in PM 2.5 .

Figure S3 . 8 Figure S4 :
Figure S3.Schematic overview of (A) the kinetic model KM-SUB-ELF and (B) principal model inputs and typical model outputs of KM-SUB-ELF.Concentrations are indicated using C. PM 2.5 mass fractions of iron, copper, secondary organic aerosol and quinones are indicated using w Fe , w Cu , w SOA , and w Quinones , respectively.f dep,PM2.5 represents the fraction of PM2.5 deposited in the ELF.S indicates the PM2.5 soluble fractions in the ELF.Q is the ventilation rate of the lung, and t the exposure time to pollutants.N ΣROS indicates the cumulative ROS produced subtracted by ROS interconversion reactions.P stands for gross chemically produced.Units and additional information are supplied in the main and SI text, and in Lelieveld et al. 2021. 8

Figure S5 :
Figure S5: The effect of decreased and increased antioxidant levels on • OH production, P OH , in healthy and diseased individuals.All antioxidant concentrations are scaled with a factor (FC), including the low-molecular weight antioxidants ascorbate, glutathione and uric acid, as well as the enzymes catalase, SOD and GPx.For healthy and diseased individuals, we assume eNO concentrations of 10 and 30 ppb, respectively.Panel (B) shows the change in P OH from a typical 3-fold increase in eNO between healthy and diseased individuals as a function of antioxidant levels.Calculations are performed using the standard air pollution scenario.

Table S2 :
PM 2.5 and transition metal mass fractions

Table S3 :
PM 2.5 and Secondary Organic Aerosol (SOA) mass fractions

Table S4 :
PM 2.5 and quinone mass fractions

Table S4 . Phenanthrenequinone (PQN), 1,4-naphthoquinone (1,4-NQN) and 1,2-naphthoquinone (1,2-NQN) concentrations with mass fractions (MF) as quantified in PM 2.5 collected at different sampling sites throughout the world
. Studies that did not report the PM 2.5 concentration at the sampling site are marked with an asterisk, and a concentration was estimated based on other observational data at that sampling site.Note, however, that these PM 2.5 concentrations are not used as input for calculations in this study.