Interactive comment on “ Light-induced protein nitration and degradation with HONO emission ”

This MS reports on HONO formation resulting (mostly) from the interaction of NO2 with a particular protein under visible illumination in a flow tube reactor. The HONO released to the gas phase is formed both by photolysis of nitrated tyrosine and a Langmuir-Hinshelwood surface reaction involving NO2 uptake; this latter process forms HONO even in the dark. For bot dark and illuminated channels, there is a positive dependence on RH which suggests that water is involved somehow, although this may be by changing the protein surface morphology rather than as a chemical promoter. The experiments are well constructed and the results are of some interest. I do have a few comments for the authors’ consideration, however:

Abstract. Proteins can be nitrated by air pollutants (NO 2 ), enhancing their allergenic potential. This work provides insight into protein nitration and subsequent decomposition in the presence of solar radiation. We also investigated lightinduced formation of nitrous acid (HONO) from protein surfaces that were nitrated either online with instantaneous gas-phase exposure to NO 2 or offline by an efficient nitration agent (tetranitromethane, TNM). Bovine serum albumin (BSA) and ovalbumin (OVA) were used as model substances for proteins. Nitration degrees of about 1 % were derived applying NO 2 concentrations of 100 ppb under VIS/UV illuminated conditions, while simultaneous decomposition of (nitrated) proteins was also found during long-term (20 h) irradiation exposure. Measurements of gas exchange on TNMnitrated proteins revealed that HONO can be formed and released even without contribution of instantaneous heterogeneous NO 2 conversion. NO 2 exposure was found to increase HONO emissions substantially. In particular, a strong dependence of HONO emissions on light intensity, relative humidity, NO 2 concentrations and the applied coating thickness was found. The 20 h long-term studies revealed sustained HONO formation, even when concentrations of the intact (nitrated) proteins were too low to be detected after the gas exchange measurements. A reaction mechanism for the NO 2 conversion based on the Langmuir-Hinshelwood kinetics is proposed.

Introduction
Primary biological aerosols, or bioaerosols, including proteins, from different sources and with distinct properties are known to influence atmospheric cloud microphysics and public health (Lang-Yona et al., 2016;D'Amato et al., 2007;Pummer et al., 2015). Bioaerosols represent a diverse subset of atmospheric particulate matter that is directly emitted in form of active or dead organisms, or fragments, like bacteria, fungal spores, pollens, viruses and plant debris. Proteins are found ubiquitously in the atmosphere as part of these airborne, typically coarse-sized biological particles (diameter > 2.5 µm), as well as in fine particulate matter (diameter < 2.5 µm) associated with a host of different constituents such as polymers derived from biomaterials and proteins dissolved in hydrometeors, mixed with fine dust and other particles (Miguel et al., 1999;Riediker et al., 2000;Zhang and Anastasio, 2003). Proteins contribute up to 5 % of par-  Houée-Levin et al., 2015, andShiraiwa et al., 2012). Subsequent intramolecular H transfer initiated by irradiation decompose the protein and HONO is emitted (adapted from Bejan et al., 2006). ticle mass in airborne particles (Franze et al., 2003a;Staton et al., 2015;Menetrez et al., 2007) and are also found at surfaces of soils and plants. Proteins can be nitrated and are then likely to enhance allergic responses (Gruijthuijsen et al., 2006). Nitrogen dioxide ( q NO 2 ) has emerged as an important biological reactant and has been shown to be capable of electron (or H atom) abstraction from the amino acid tyrosine (Tyr) to form TyrO q in aqueous solutions (tyrosine phenoxyl radical, also called tyrosyl radical; Prütz et al., 1984Prütz et al., , 1985Alfassi, 1987;Houée-Lévin et al., 2015), which subsequently can be nitrated by a second NO 2 molecule. Shiraiwa et al. (2012) observed nitration of protein aerosol, but not solely with NO 2 in the gas phase, and demonstrated that simultaneous O 3 exposure of airborne proteins in dark conditions can significantly enhance NO 2 uptake and consequent protein nitration (3-nitrotyrosine formation) by way of direct O 3 mediated formation of the TyrO q intermediate. A connection between increased allergic diseases and elevated environmental pollution, especially traffic-related air pollution has been proposed (Ring et al., 2001). Tyrosine is one of the photosensitive amino acids and it is subject of direct and indirect photo-degradation under solar-simulated conditions (Boreen et al., 2008), especially mediated by both UV-B (λ 280-320 nm) and UV-A (λ 320-400 nm) radiation (Houee-Levin et al., 2015;Bensasson et al., 1993). Direct light absorption or absorption by adjacent endogenous or exogenous chromophores and subsequent energy transfer results in an electronically excited state of tyrosine (for details see Houée-Lévin et al., 2015, and references therein). If the triplet state of tyrosine is generated, it can undergo electron transfer reactions and deprotonation to yield TyrO q ( Fig. 1; Bensasson , 1993;Davies, 1991;Berto et al., 2016). Regardless of how the tyrosyl radical is generated, it can be nitrated by reaction with NO 2 , as well as hydroxylated or dimerized (Shiraiwa et al., 2012;Reinmuth-Selzle et al., 2014;Kampf et al., 2015).
With respect to atmospheric chemistry, Bejan et al. (2006) have shown that photolysis of ortho-nitrophenols (as is the case for 3-nitrotyrosine) can generate nitrous acid (HONO). HONO is of great interest for atmospheric composition, as its photolysis forms OH radicals, which are the key oxidant for degradation of most air pollutants in the troposphere (Levy, 1971). In the lower atmosphere, up to 30 % of the primary OH radical production can be attributed to photolysis of HONO, especially during the early morning when other photochemical OH sources are still small (Reaction R1, Kleffmann et al., 2005;Alicke et al., 2002;Ren et al., 2006;Su et al., 2008;Meusel et al., 2016).
HONO can be directly emitted by combustion of fossil fuels  or formed by gas-phase reactions of NO and OH (the backwards reaction of Reaction R1) and heterogeneous reactions of NO 2 on wet surfaces according to Reaction (R2). On carbonaceous surfaces (soot, phenolic compounds) HONO is formed via electron or H transfer reactions (Reactions R3 and R4-R6; Kalberer et al., 1999;Kleffmann et al., 1999;Gutzwiller et al., 2002;Aubin and Abbatt, 2007;Han et al., 2013;Arens et al., 2001Arens et al., , 2002Ammann et al., 1998Ammann et al., , 2005. Previous atmospheric measurements and modeling studies have shown unexpected high HONO concentrations during daytime, which can also contribute to aerosol formation through enhanced oxidation of precursor gases (Elshorbany et al., 2014). Measured mixing ratios are typically about 1 order of magnitude higher than simulated ones, and an additional source of 200-800 ppt h −1 would be required to explain observed mixing ratios Acker et al., 2006;Li et al., 2012;Su et al., 2008;Elshorbany et al., 2012;Meusel et al., 2016), indicating that estimates of daytime HONO sources are still under debate. It was suggested that HONO arises from the photolysis of nitric acid and nitrate or by heterogeneous photochemistry of NO 2 on organic substrates and soot (Zhou et al., 2001(Zhou et al., , 2002(Zhou et al., and 2003Villena et al., 2011;Ramazan et al., 2004;George et al., 2005;Sosedova et al., 2011;Monge et al., 2010;Han et al., 2016). Stemmler et al. (2006Stemmler et al. ( , 2007 found HONO formation on light-activated humic acid, and field studies showed that HONO formation correlates with aerosol surface area, NO 2 and solar radiation (Su et al., 2008;Reisinger, 2000;Costabile et al., 2010;Wong et al., 2012;Sörgel et al., 2015) and is increased during foggy periods (Notholt et al., 1992). Another proposed source of HONO is the soil, where it has been found to be co-emitted with NO by soil biological activities (Oswald et al., 2013;Su et al., 2011;Weber et al., 2015).
In view of light-induced nitration of proteins and HONO formation by photolysis of nitrophenols, light-enhanced production of HONO on protein surfaces can be anticipated, which, to the best of our knowledge, has not been studied before.
This work aims to provide insight into protein nitration, the atmospheric stability of the nitrated protein and respective formation of HONO from protein surfaces that were nitrated either offline in liquid phase prior to the gas exchange measurements or online with instantaneous gas-phase exposure to NO 2 , with particular emphasis on environmental parameters like light intensity, relative humidity (RH) and NO 2 concentrations. Bovine serum albumin (BSA), a globular protein with a molecular mass of 66.5 kDa and 21 tyrosine residues per molecule, was chosen as a well-defined model substance for proteins. Nitrated ovalbumin (OVA) was used to study the light-induced degradation of proteins that were nitrated prior to gas exchange measurements. This wellstudied protein has a molecular mass of 45 kDa and 10 tyrosine residues per molecule.

Protein preparation and analysis
BSA (Cohn V fraction, lyophilized powder, ≥ 96 %; Sigma Aldrich, St. Louis, Missouri, USA) or nitrated OVA was solved in pure water (18.2 M cm) and coated onto the glass tube.
The nitration of OVA was described previously (Yang et al., 2010;Zhang et al., 2011). Briefly, OVA (grade V, A5503-5G, Sigma Aldrich, Germany) was dissolved in phosphate-buffered saline PBS (P4417-50TAB, Sigma Aldrich, Germany) to a concentration of 10 mg ml −1 . 50 µL tetranitromethane (TNM; T25003-5G, Sigma Aldrich, Germany) dissolved in methanol 4 % (v/v) were added to a 2.5 mL aliquot of the OVA solution and stirred for 180 min at room temperature. Please note that TNM is toxic if swallowed, can cause skin, eye and respiration irritation, is suspected to cause cancer and causes fires or explosions. Size exclusion chromatography columns (PD-10 Sephadex G-25 M, 17-0851-01, GE Healthcare, Germany) were used for cleanup. The eluate was dried in a freeze dryer and stored in a refrigerator at 4 • C.
After the flow-tube experiments (see below) the proteins were extracted with water from the tube and analyzed with liquid chromatography (HPLC-DAD; Agilent Technologies 1200 series) according to Selzle et al. (2013). This method provides a straightforward and efficient way to determine the nitration of proteins. Briefly, a monomerically bound C18 column (Vydac 238TP, 250 mm × 2.1 mm inner diameter, 5 µm particle size; Grace Vydac, Alltech) was used for chromatographic separation. Eluents were 0.1 % (v/v) trifluoroacetic acid in water (LiChrosolv) (eluent A) and acetonitrile (ROTISOLV HPLC gradient grade, Carl Roth GmbH + Co. KG, Germany) (eluent B). Gradient elution was performed at a flow rate of 200 µL min −1 . ChemStation software (Rev. B.03.01, Agilent) was used for system control and data analysis. For each chromatographic run, the solvent gradient started at 3 % B followed by a linear gradient to 90 % B within 15 min, flushing back to 3 % B within 0.2 min and maintaining 3 % B for additional 2.8 min. Column reequilibration time was 5 min before the next run. Absorbance was monitored at wavelengths of 280 (tyrosine) and 357 nm (nitrotyrosine). The sample injection volume was 10-30 µL. Each chromatographic run was repeated three times. The protein nitration degree (ND), which is defined as the ratio of nitrated tyrosine to all tyrosine residues, was determined by the method of Selzle et al. (2013). Native and untreated BSA did not show any degree of nitration.
2.2 Coated-wall flow tube system Figure 2 shows a flowchart of the setup of the experiment. NO 2 was provided in a gas bottle (1 ppm in N 2 , Carbagas AG, Grümligen, Switzerland). NO 2 was further diluted (mass flow controller, MFC3) with humidified pure nitrogen to achieve NO 2 mixing ratios between 20 and 100 ppb. Impurities of HONO in the NO 2 -gas cylinder were removed by means of a HONO scrubber. The Na 2 CO 3 trap was prepared by soaking 4 mm firebrick in a saturated Na 2 CO 3 in 50 % ethanol-water solution and drying for 24 h. The impregnated firebrick granules were put into a 0.8 cm inner diameter and 15 cm long glass tube, which was closed by quartz wool plugs on both sides. A constant total flow (1400 mL min −1 ) was provided by means of another N 2 mass flow controller (MFC2) that compensated for changes in NO 2 addition. Different fractions of total surface areas (50, 70 and 100 %) of  . Nitrogen passes a heated water bath to humidify the gas and a HONO scrubber to eliminate any HONO impurities of the NO 2 supply. The overflow maintains a constant pressure through the reaction tube and the detection unit. The dotted boxes (blue, green, orange) indicate the three different parts: the gas supply, reaction unit and detection unit.
the reaction tube (50 cm×0.81 cm i.d.) were coated with 2 mg BSA or nitrated OVA, respectively. Therefore 2 mg protein was dissolved in 600 µL pure water, injected into the tube and then gently dried in a low-humidity N 2 flow (RH ∼ 30-40 %) with continuous rotation of the tube. The coated reaction tube was exposed to the generated gas mixture and irradiated with either (i) one, three or seven visible (VIS) lights (400-700 nm; L 15 W/954, Lumilux de Luxe daylight, Osram, Augsburg, Germany), which is 0, 23, 69 or 161 W m −2 , respectively; or (ii) four VIS and three UV lights (340-400 nm; UV-A, TL-D 15 W/10, Philips, Hamburg, Germany). An overview of the experiments performed during this study is shown in Table 1. Light-induced decomposition of nitrated proteins was studied on OVA. Instantaneous NO 2 transformation and its light and RH dependence on heterogeneous HONO formation were studied on BSA in short-term experiments. Extended studies on BSA were performed to explore the persistence of the surface reactivity and respective catalytic effects.
A commercial long-path absorption photometry instrument (LOPAP, QUMA) was used for HONO analysis. The measurement technique was introduced by Heland et al. (2001). This wet chemical analytical method has an unmatched low detection limit of 3-5 ppt with high HONO collection efficiency (≥ 99 %). HONO is continuously trapped in a stripping coil flushed with an acidic solution of sulfanilamide. In a second reaction with n-(1-naphthyl)ethylenediamine-dihydrochloride an azo dye is formed, whose concentration is determined by absorption photometry in a long Teflon tubing. LOPAP has two stripping coils in series to reduce known interferences. In the first stripping coil HONO is quantitatively collected. Due to the acidic stripping solution, interfering species are collected less efficiently but in both channels. The true concentration of HONO is obtained by subtracting the interferences quantified in the second channel from the total signal obtained in the first channel. The accuracy of the HONO measurements was 10 %, based on the uncertainties of liquid and gas flow, concentration of calibration standard and regression of calibration.

BSA nitration and degradation
Nitrated proteins can trigger allergic response. The nitration of proteins can be enhanced by O 3 activation (in the dark). In the atmospheric environment, about half the time sunlight is present. What happens with irradiated proteins when exposed to NO 2 ? Can they be nitrated efficiently? To investigate the degree of protein nitration under illuminated conditions, BSA coated on the reaction tube (17.5 µg cm −2 ) was exposed to seven VIS lamps (40 % of a clear-sky irradiance for a solar zenith of 48 • ; Stemmler et al., 2006) and 100 ppb NO 2 at 70 % RH. After 20 h the BSA ND (concentration of nitrated tyrosine residues divided by the total concentration of tyrosine residues) investigated by means of the HPLC-DAD method was (1.0 ± 0.1) %, significantly higher than the ND of untreated BSA (0 %). Introducing UV radiation (four VIS plus three UV lamps) resulted in a slightly higher ND of (1.1 ± 0.1) %. Note that no intact protein (nitrated and nonnitrated) could be detected by HPLC-DAD after another 20 h of irradiation without NO 2 , indicating light-induced decomposition of proteins. However, the applied HPLC-DAD technique only detects (nitro-)tyrosine residues in proteins and does not provide information about protein fragments or single nitrated or non-nitrated tyrosine residues. Hence, proteins might have been decomposed while tyrosine remains in its nitrated form, not detectable by our analysis method. Similarly, proteins (here OVA) that were nitrated with TNM in aqueous phase prior to coating (21.5 µg cm −2 ) to an extent of 12.5 % also decomposed when illuminated about 6 h (one to seven VIS lights; with and without 20 ppb NO 2 ). Thus the nitration of proteins by light and NO 2 was confirmed, but with simultaneous gradual decomposition of the proteins. Effects of UV irradiation (240-340 nm) on proteins containing aromatic amino acids were reviewed previously (Neves-Peterson et al., 2012). It was shown that triplet state tryptophan and tyrosine can transfer electron to a nearby disulfide bridge to form the tryptophan and tyrosine radical. The disulfide bridge could break leading to conformational changes in the protein but not necessarily resulting in inactivation of the protein. In strong UV light (≈ 200 nm) the peptide bond could also break (Nikogosyan and Görner, 1999). Franze et al. (2005) analyzed a variety of natural samples (road dust, window dust and particulate matter PM 2.5 ) collected in the metropolitan area of Munich, containing 0.08-21 g kg −1 proteins, and revealed equivalent degrees of nitration (EDN, concentration of nitrated protein divided by concentration of all proteins) between 0.01 and 0.1 % only. Such low nitration degree is in line with light-induced decomposition of (nitrated) proteins. In contrast, an EDN up to 10 % (average 5 %) was found for BSA and birch pollen extract exposed to Munich ambient air for 2 weeks under dark conditions, with daily mean NO 2 (O 3 ) concentration of 17-50 ppb (7-43 ppb) in the same study, possibly suggesting the deficiency of decomposition without being irradiated. BSA and OVA loaded on syringe filters and exposed to 200 ppb NO 2 / O 3 for 6 days under dark conditions were nitrated to 6 and 8 %, respectively (Yang et al., 2010). Reinmuth-Selzle et al. (2014) found similar ND for major birch pollen allergen Bet v 1 loaded on syringe filters exposed to 80-470 ppb NO 2 and O 3 . When exposed for 3-72 h to NO 2 / O 3 at RH < 92 % the ND was 2-4 %, while at condensing conditions (RH > 98 %) the ND increased to 6 % after less than 1 day (19 h). The ND of Bet v 1 was considerably increased to 22 % for proteins solved in the aqueous phase (0.16 mg mL −1 ) when bubbling with a 120 ppb NO 2 / O 3 gas mixture for a similar period of time (17 h). Shiraiwa et al. (2012) performed kinetic modeling and found that maximum 30 % (conservative upper limit) of N uptake on BSA could be explained by NO 3 or N 2 O 5 , which are generated by the reaction of NO 2 and O 3 , while overall nitration was governed by an indirect mechanism in which a radical intermediate was formed by the reaction of BSA with ozone, which then reacted with NO 2 . On NaCl surface N uptake was dominated by NO 3 and N 2 O 5 . Furthermore, NO 3 radicals, which in this study could be formed by photolysis of NO 2 (> 410 nm, disproportionation of excited NO 2 ), are not stable under the light conditions applied (400-700 nm) (Johnston et al., 1996). Therefore, in the present study reactions with NO 3 were neglected. Photolysis of NO 2 forming NO (< 400 nm) can also be neglected (Gardner et al., 1987;Roehl et al., 1994). A photolysis frequency for NO 2 of up to 5×10 −4 s −1 under similar experimental light conditions was determined by Stemmler et al., 2007. Other nitration methods investigated by Reinmuth- Selzle et al. (2014), e.g., nitration of Bet v 1 with peroxynitrite (ONOO − , formed by reaction of NO with O − 2 ) or TNM, lead to ND between 10 and 72 % depending on reaction time, reagent concentration and temperature. Similarly, high NDs of 45-50 % were obtained by aqueous-phase TNM nitration of BSA and OVA by Yang et al. (2010).

HONO formation from nitrated proteins
To study HONO emission from nitrated proteins, OVA was nitrated with TNM (see Sect. 2.1) in liquid phase. The nitrated OVA (2 mg; ND = 12.5 %) was coated onto the reaction tube and exposed to VIS lights under either pure nitrogen flow or 20 ppb NO 2 gas. Strong HONO emissions were found. A high correlation between HONO emission and light intensity was observed (50 % RH; Fig. 3). Initially, we did not apply NO 2 . Thus the observed HONO formation (up to 950 ppt) originated from decomposing nitrated proteins rather than from heterogeneous conversion of NO 2 . However, when exposed to 20 ppb of NO 2 in dark conditions, HONO formation increased 4-fold (50-200 ppt) and about 2fold with seven VIS lamps turned on (950-1800 ppt). After 7 h of flow tube experiments (4.5 h irradiation with varying light intensities (0, 1, 3, 7 lights) + 2.5 h irradiation/20 ppb NO 2 (7, 3, 0 lights)), no intact protein was found according to the analysis of HPLC-DAD.
As proteins can efficiently be nitrated by O 3 and NO 2 in polluted air (Franze et al., 2005;Shiraiwa et al., 2012;Reinmuth-Selzle et al., 2014), the emission of HONO from light-induced decomposing nitrated proteins could play an important role in the HONO budget. As proteins are nitrated at their tyrosine residues (at the ortho position to the OH group on the aromatic ring) the underlying mechanism of this HONO formation should be very similar to the HONO formation by photolysis of ortho-nitrophenols described by Bejan et al. (2006). This starts with a photo-induced hydrogen transfer from the OH group to the vicinal NO 2 group (Fig. 1),  which leads to an excited intermediate from which HONO is eliminated subsequently.

Light dependency
To investigate HONO formation on unmodified BSA coating (31.4 µg cm −2 ) dependent on light conditions, the radiation intensity (number of VIS lamps) was changed under otherwise constant conditions of exposure at 20 ppb NO 2 and 50 % RH. Decreasing light intensity revealed a linearly decreasing trend in HONO formation from about 1000 to 140 ppt (red symbols in Fig. 4). After re-illumination to the initial high light intensity the HONO formation was reduced by 32 % (blue symbol in Fig. 4). Stemmler et al. (2006) and Sosedova et al. (2011) also observed a similar saturation of HONO formation on humic, tannic and gentisic acid at higher light intensities. Stemmler et al. (2006) argued that surface sites activated for NO 2 heterogeneous conversion by light (Reaction R3) would become de-activated by competition with photo-induced oxidants (X * , Reactions R7-R8), e.g., primary chromophores or electron donors are oxidized by surface*, which is in line with the observed decomposition of the native protein presented above.
In other studies the NO 2 uptake coefficient on soot, mineral dust, humic acid and other solid organic compounds similarly increased at increasing light intensities ( (Bejan et al., 2006).

NO 2 dependency
At about 50 % relative humidity and high illumination intensities (seven VIS lamps, ∼ 161 W m −2 ), heterogeneous formation of HONO strongly correlated with the applied NO 2 concentration (Fig. 5). On a BSA surface of about 16.1 µg cm −2 (Table 1) the produced HONO concentration increased from 56 ppt at 20 ppb NO 2 to 160 ppt at 100 ppb NO 2 . Only at a threshold NO 2 level well above those typically observed in natural environments ( 150 ppb) did this increasing trend slow down to some extent, indicative of saturation of active surface sites. A similar pattern of NO 2 dependence was also observed for light-induced HONO formation from humic acid (Stemmler et al., 2006) (Stemmler et al., 2006), while the blue triangles pointing down are the humic acid aerosol with 100 nm diameter and a surface of 0.151 m 2 m −3 at 26 % RH and 1 × 10 17 photons cm −2 s −1 (Stemmler et al., 2007). The black circles are gentisic acid coating (160-200 µg cm −2 ) at 40-45 % RH and light intensity similar to that in the humic acid aerosol study (Sosedova et al., 2011). Green diamonds are ortho-nitrophenol in gas phase (ppm level) illuminated with UV/VIS light. Dotted lines are exponential fittings of the measured data points and are meant to guide the eyes.
(up to 40 ppb NO 2 ) was observed when NO 2 was applied additionally during the gas-phase photolysis of nitrophenols (Fig. 5;Bejan et al., 2006). Even though the matrix (nitro-phenols) and conditions (illuminated) of the latter is comparable to the experiment presented here, for BSA no clear indication of saturation was found up to 160 ppb of NO 2 , pointing to a highly reactive surface of BSA for NO 2 under illuminated conditions. As shown with Reactions (R7) and (R8), the concentration dependence depends on the competing channel (Reaction R8); therefore, this is strongly matrix dependent, both in terms of chemical and physical properties.

Impact of coating thickness
Strong differences in HONO concentrations were found for experiments with different coating thicknesses applying otherwise similar conditions (20 ppb of NO 2 , seven VIS lamps and 50 % RH). While only 55 ppt of HONO concentration was observed for a shallow homogeneous coating of 16.1 µg cm −2 (217.6 nm thickness, see below) applied on the whole length of the tube, up to 2 ppb was found for a thick (more uneven) coating of 31.44 µg cm −2 (435.2 nm thickness) covering only 50 % of the tube (Fig. 6). Potential explanations are that thicker coating leads to (1) more bulk reactions producing HONO or (2) different morphologies, e.g., higher effective reaction surfaces. Exposing (20 %) different coated surface areas in the flow tube, potentially introduced bias comparing different data sets. Emitted HONO might be re-adsorbed differently by proteins and glass surface. However, as the protein is slightly acidic, a low uptake efficiency of HONO by BSA can be anticipated, which should not differ too much from the uncovered glass tube surface . Accordingly, NO 2 uptake on glass is assumed to be significantly lower than on proteins. A strong increase in NO 2 uptake coefficients with increasing coating thickness was also observed for humic acid coatings (Han et al., 2016). However, they found an upper threshold value of 2 µg cm −2 of cover load (20 nm absolute thickness, assuming a humic acid density of 1 g cm −3 ), above which uptake coefficients were found to be constant. The authors also proposed that NO 2 can diffuse deeper into the coating and below 2 µg cm −2 the full cover depth would react with NO 2 , respectively.
For proteins the number of molecules per monolayer depends on their orientation and respective layer thickness can vary accordingly. One (dry, crystalline) BSA molecule has a volume of about 154 nm 3 (Bujacz, 2012). In a flat orientation (4.4 nm layer height and a projecting area of 35 nm 2 molecule −1 ) 3.64 × 10 14 molecules (40.5 µg; 0.32 µg cm −2 ) of BSA are needed to form one complete monolayer in the flow tube (i.d. of 0.81, 50 cm length, 100 % surface coating). Hence, the thinnest BSA coating applied in the experiment (16.1 µg cm −2 ) would consist of 50 monolayers, revealing a total coating thickness of 217.6 nm, and the thickest BSA coating (31 µg cm −2 ) would have 99 monolayers and an absolute thickness of 435.1 nm. At the other extreme (non-flat) orientation, more BSA molecules are needed to sustain one monolayer. With 21.7 nm 2 of pro- Figure 6. HONO formation on three different BSA coating thicknesses, exposed to 20 ppb of NO 2 under illuminated conditions (seven VIS lamps). The HONO concentrations were scaled to reaction tube coverage (black: 100 % of reaction tube was covered with BSA; light blue: 70 % of tube was covered; red: 50 % of tube was covered with BSA). The middle thick coating (22.46 µg cm −2 ) was replicated and studied with different reaction times (cyan and blue triangle). Solid lines (with circles or triangles) present continuous measurements; when those are interrupted, other conditions (e.g., light intensity, NO 2 concentration) prevailed. Dotted lines show interpolations and are meant to guide the eyes. Arrows indicate the intervals in which the shown decay rates were determined. Error bars indicates SDs from 10 to 20 measuring points (5-10 min). jected area of one molecule and 7.1 nm monolayer height, 5.86 × 10 14 molecules of BSA are needed to form one complete monolayer in the flow tube. The coatings would consist of between 31 (thinnest) and 61 (thickest) monolayers of BSA. With a flat orientation 1-2 % (number or weight) of BSA molecules would build the uppermost surface monolayer, whereas in an upright molecule orientation 1.6-3.3 % would be in direct contact with surface ambient air.
In the crystalline form several molecules of water stick tightly to BSA. As BSA is highly hygroscopic, more water molecules are adsorbed at higher relative humidity. At 35 % RH BSA is deliquesced (Mikhailov et al., 2004). Therefore the above described number of monolayers and the absolute layer thickness are a lower bound estimate.
In conclusion, the thickness dependence on HONO formation is extremely complex. Activation and photolysis of nitrated Tyr occurs throughout the BSA layer. The heterogeneous reaction of NO 2 may or may be not limited to the surface depending on solubility and diffusivity of NO 2 . Also the release of HONO may be limited by diffusion. The observed dependence on the coating thickness suggests the involvement of the bulk reactions, but the reactions can happen in both surface and bulk phase.

RH dependency
The dependence of HONO emission on relative humidity is shown in Fig. 7. Here about 25 ppb of NO 2 was applied to a (not nitrated) BSA-coated flow tube (17.5 µg cm −2 ) both in dark and illuminated conditions (seven VIS lights). HONO formation scaled with relative humidity. Kleffmann et al. (1999) proposed that higher humidity inhibits the self-reaction of HONO (2 HONO (s,g) → NO 2 + NO + H 2 O), which leads to higher HONO yield from heterogeneous NO 2 conversion.
The RH dependence of HONO formation on proteins is different to other surfaces. For example, no influence of RH has been observed for dark heterogeneous HONO formation on soot particles sampled on filters (Arens et al., 2001). No impact of humidity on NO 2 uptake coefficients on pyrene was detected (Brigante et al., 2008). For HONO formation on tannic acid coatings (both at dark and irradiated conditions) a linear but relatively weak dependence has been reported between 10 and 60 % RH, while below 10 % and above 60 % RH the correlation between HONO formation and RH was much stronger (Sosedova et al., 2011). Similar results were obtained for anthrarobin coatings by Arens et al. (2002). This type of dependence of HONO formation on phenolic surfaces on RH equals the HONO formation on glass, following the BET water uptake isotherm of water on polar surfaces (Finnlayson-Pitts et al., 2003;Summer et al., 2004). For humic acid surfaces the NO 2 uptake coefficients also weakly increased below 20 % RH and were found to be constant between 20 and 60 % (Stemmler et al., 2007).
While on solid matter chemical reactions are essentially confined to the surface rather than in the bulk, proteins can adopt an amorphous solid or semisolid state, influencing the rate of heterogeneous reactions and multiphase processes. Molecular diffusion in the non-solid phase affects the gas uptake and respective chemical transformation. Shiraiwa et al. (2011) could show that the ozonolysis of amorphous protein is kinetically limited by bulk diffusion. The reactive gas uptake exhibits a pronounced increase with relative humidity, which can be explained by a decrease of viscosity and increase of diffusivity, as the uptake of water transforms the amorphous organic matrix from a glassy to a semisolid state (moisture-induced phase transition). The viscosity and diffusivity of proteins depend strongly on the ambient relative humidity because water can act as a plasticizer and increase the mobility of the protein matrix (for details see Shiraiwa et al., 2011, and references therein). Shiraiwa et al. (2011) further showed that the BSA phase changes from solid through semisolid to viscous liquid as RH increases, while trace gas diffusion coefficients increased about 10 orders of magnitude. This way, characteristic times for heterogeneous reaction rates can decrease from seconds to days as the rate of diffusion in semisolid phases can decrease by multiple orders of magnitude in response to both low temperature (not investigated in here) and/or low relative humidity. Accordingly, we propose that HONO formation rate depends on the condensed-phase diffusion coefficients of NO 2 diffusing into the protein bulk, HONO released from the bulk and mobility of excited intermediates.

Long-term exposure with NO 2 under irradiated conditions
To study long-term effects of irradiation on HONO formation from proteins, flow tubes were coated with 2 mg BSA (17.5 ± 0.4 µg cm −2 ; 90 % of total length) and exposed to 100 ppb NO 2 , at 80 % RH at illuminated conditions for a time period of up to 20 h (Fig. 8). Samples illuminated with VIS light only (red and orange colored lines in Fig. 8) showed persistent HONO emissions over the whole measurement period. For unknown reasons, and even though the observed HONO concentrations were within the expected range with regard to the applied NO 2 concentrations, RH and cover characteristics, one sample (orange in Fig. 8) showed a sharp short-term increase in the initial phase followed by respective decrease, not in line with all other samples (compare Fig. 6). However, after 4 h both VIS irradiated samples showed virtually constant HONO emissions (−3.8 and +1.6 ppt h −1 , respectively). The sample illuminated with UV and VIS light (three UV and four VIS lamps) showed a sustained sharp increase in the first 4 h, followed by persistent and very stable (decay rate as low as −0.5 ppt h −1 ) HONO emissions at an about 3-fold higher level compared to samples irradiated with VIS only. HONO formation by photolysis of (adsorbed) HNO 3 is assumed to be insignificant in this study. With N 2 as carrier gas, gas-phase reactions of NO 2 do not produce HNO 3 . Even when small amounts of HNO 3 would be formed by unknown heterogeneous reactions, photolysis of HNO 3 is only significant at wavelengths < 350 nm, which is close to the lowest limit of the UV wavelength applied in this study. Likewise, the respective photolysis frequency recently proposed by Laufs and Kleffmann (2016) of about 2.4 × 10 −7 s −1 is very low.
If BSA acts like a catalytic surface as in a Langmuir-Hinshelwood reaction each BSA molecule can react several times with NO 2 to heterogeneously form HONO. As described in 3.1, BSA nitration is in competition with NO 2 surface reactions and only a limited number of NO 2 molecules could react with BSA forming HONO via nitration of proteins and subsequent decomposition of nitrated proteins. A BSA molecule contains 21 tyrosine residues, which could react with NO 2 . However, even a strong nitration agent such as TNM is not capable of nitrating all tyrosine residues and a mean ND of 19 % was found (Peterson et al., 2001;Yang et al., 2010); i.e., four tyrosine residues of one BSA molecule can be nitrated to form HONO. As 2 mg of BSA was applied for each flow tube coating, a total of 1.8 × 10 16 protein molecules can be inferred. In 20 h of irradiating with VIS light 13-22 % of the accessible Tyr residues (four Tyr per BSA molecule) would have been reacted. Irradiating with additional UV lights at least 56 % of the tyrosine residues would have been nitrated and decomposed. However, as NO 2 is a much weaker nitrating agent and nitration of only one tyrosine residue is probable (ND of BSA with O 3 / NO 2 6 %; Yang et al., 2010) up to 85 % BSA molecules would have been reacted when irradiated with VIS lights and even more HONO molecules as coated BSA molecules would have been generated under UV/VIS light conditions. Other amino acids of the protein like tryptophan or phenylalanine might also be nitrated but without formation of HONO (Goeschen et al., 2011). Hence, a contribution of heterogeneous conversion of NO 2 can be anticipated.

Kinetic studies
The experimental results (especially the stability over a long time) indicate that the formation of HONO from NO 2 on protein surfaces likely underlies the Langmuir-Hinshelwood mechanism in which the protein would act as a catalytic surface (Fig. 9). The first step is the fast, reversible physical adsorption of NO 2 (k 1 ) and water followed by the slow conversion into HONO.
There are two possible processes for the HONO formation. HONO is formed by heterogeneous NO 2 conversion (k 2 ) but also via nitration and decomposition of nitrated proteins (k 4 , k 5 ). The final step of the mechanism is the release of the generated HONO into the air. Since proteins are in general slightly acidic, the desorption of HONO (k 3 ) should be fairly fast. Pseudo-first-order kinetics are assumed for the reaction of NO 2 to HONO (Stemmler et al., 2007) and the reaction can be described as follows (Eq. 1).
with k eff the effective pseudo-first-order rate constant (for more detailed information check the Supplement). Figure 9. Schematic illustration of the underlying Langmuir-Hinshelwood mechanism of light-induced HONO formation on protein surface. Reaction constants for NO 2 uptake, direct NO 2 conversion, protein nitration, HONO formation from decomposing nitrated proteins and HONO release are indicated by k 1 , k 2 , k 4 , k 5 , and k 3 .
In this study, neither HONO nor NO 2 photolysis is considered, as the overlap of the applied UV/VIS or VIS range (340-700 or 400-700 nm) and the HONO and NO 2 photolysis spectrum (< 400 nm) is low. Furthermore, the applied light intensity is lower compared to clear-sky irradiance and the respective UV light is partly absorbed by the reaction tube although quartz glass was used (transmission ∼ 90 %) and the photolysis frequency would decrease down to 10 −4 s −1 . Hence, the photolysis is assumed to be not significant.
In the first 5-10 min of the long-term experiments, HONO increased ( Fig. 8 -zoomed in range). This slope was taken as d[HONO] g /dt in Eq. (6). Effective rate constants between 1.48×10 −6 s −1 (VISa) and 7.40×10 −6 s −1 (VISb) were calculated. When irradiating with VIS light only, the concentration of HONO was either constant or decreased for 2 h after this first 10 min. When irradiating with additional UV light, the HONO signal showed an enhancement in two steps. In the first 10 min it was strongly increasing (1327 ppt h −1 ) and then in the next hour it increased less with 170 ppt h −1 prior to stabilization. Therefore two rate constants of 4.10 × 10 −6 and 5.2 × 10 −7 s −1 were obtained, respectively.
Reactive uptake coefficients for NO 2 were calculated according to Li et al. (2016). For both irradiation types the uptake coefficient γ was in the range of 7 × 10 −6 at the very beginning of each experiment. After a few minutes they decreased to a mean of 1 × 10 −7 . The calculated k eff values and uptake coefficient are in the same range and match the NO 2 uptake coefficients on irradiated humic acid surfaces (coatings) and aerosols obtained by Stemmler et al. (2006/07) which were in between 2×10 −6 and 2×10 −5 (coatings) and 1×10 −6 and 6×10 −6 (aerosols), depending on NO 2 concen-trations and light intensities. Similar NO 2 uptake coefficients on humic acid were observed by Han et al. (2016). George et al. (2005) reported about a 2-fold increased NO 2 uptake coefficients for irradiated organic substrates (benzophenone, catechol, anthracene) compared to dark conditions, in the order of (0.6-5) × 10 −6 . NO 2 uptake coefficients on gentisic acid and tannic acid were in between (3.3-4.8) × 10 −7 (Sosedova et al., 2011), still higher than on fresh soot or dust (about 1 × 10 −7 ; Monge et al., 2010;Ndour et al., 2008). The NO 2 uptake coefficients on BSA in the presence of O 3 (1 × 10 −5 , for 26 ppb NO 2 and 20 ppb O 3 ) published by Shiraiwa et al. (2012) were somewhat higher than the values calculated here without O 3 but with light.
It was not possible to extract a set of parameters for a Langmuir-Hinshelwood mechanism (like Langmuir equilibrium constant, surface accommodation coefficient or second-order rate constant) from the presented data. The saturating behavior of photochemical HONO production may be due to either the adsorbed precursor on the surface or due to a photochemical competition process, which also leads to a Lindemann-Hinshelwood type kinetic expression (Minero, 1999).

Summary and conclusion
Photochemical nitration of proteins accompanied by formation of HONO by (i) heterogeneous conversion of NO 2 and (ii) decomposition of nitrated proteins was studied under relevant atmospheric conditions. NO 2 concentrations ranged from 20 ppb (typical for urban regions in Europe and USA) up to 100 ppb (representative for highly polluted industrial regions). The applied relative humidity of up to 80 % and light intensities of up to 161 W m −2 are common on cloudy days. Under illuminated conditions very low nitration of proteins or even no native protein was observed, indicating a light-induced decomposition of nitrated proteins to shorter peptides. These might still include nitrated residues of which potential health effects are not yet known. An average effective rate constant of the total NO 2 -HONO conversion of 3.3 × 10 −6 s −1 (for about 120 cm 2 of protein surface, layer thickness 240 nm and a layer volume of 0.003 cm 3 ; surface/volume ratio ∼ 40 000 cm −1 ) or 8.25 × 10 −8 s −1 cm −2 BSA layer was obtained. At 20 ppb NO 2 HONO formation of 19.8 ppb h −1 m −2 on a pure BSA surface could be estimated. While heterogeneous HONO formation of BSA exposed to NO 2 revealed light saturation at intensities higher than 161 W m −2 , the HONO formation from previously nitrated OVA was linearly increasing over the whole light intensity range investigated. The latter let assume even higher HONO formation under sunny (clear-sky) ambient atmospheric conditions. No data about representative protein surface areas on atmospheric aerosol particles are available. However, the number and mass concentration of primary biological aerosol particles such as pollen, fungal spores and bacteria, containing proteins, are in the range of 10-10 4 m −3 and 10 −3 -1 µg m −3 , respectively (Despres et al., 2012;Shiraiwa et al., 2012). Typical aerosol surface concentrations in rural regions are about 100 µm 2 cm −3 . Stemmler et al. (2007) estimated a HONO formation of 1.2 ppt h −1 on pure humic acid aerosols in environmental conditions. As NO 2 uptake coefficients and HONO formation rates on proteins are similar to humic acid, but only about 5 % of the aerosol mass can be assumed to consist of proteins, it can be anticipated that HONO formation on aerosol is not a significant HONO source in ambient environmental settings. However, proteins on ground surfaces (soil, plants, etc.) might play a more important role. Accordingly, Stemmler et al. (2006 and suggested that NO 2 conversion on soil covered with humic acid would be sufficient to explain missing HONO sources up to 700 ppt h −1 . Therefore it is difficult to estimate the importance of HONO formation on protein surface and its contribution to the HONO budget. In many studies the calculated unknown source strength of daytime HONO formation is within a range of about 200-800 ppt h −1 Acker et al., 2006;Li et al., 2012).
Data availability. Please contact the corresponding authors Hang Su (h.su@mpic.de) or Yafang Cheng (yafang.cheng@mpic.de) for more information on data.