Nitration of the Birch Pollen Allergen Bet v 1.0101: Efficiency and Site-Selectivity of Liquid and Gaseous Nitrating Agents

Nitration of the major birch pollen allergen Bet v 1 alters the immune responses toward this protein, but the underlying chemical mechanisms are not yet understood. Here we address the efficiency and site-selectivity of the nitration reaction of recombinant protein samples of Bet v 1.0101 with different nitrating agents relevant for laboratory investigations (tetranitromethane, TNM), for physiological processes (peroxynitrite, ONOO–), and for the health effects of environmental pollutants (nitrogen dioxide and ozone, O3/NO2). We determined the total tyrosine nitration degrees (ND) and the NDs of individual tyrosine residues (NDY). High-performance liquid chromatography coupled to diode array detection and HPLC coupled to high-resolution mass spectrometry analysis of intact proteins, HPLC coupled to tandem mass spectrometry analysis of tryptic peptides, and amino acid analysis of hydrolyzed samples were performed. The preferred reaction sites were tyrosine residues at the following positions in the polypeptide chain: Y83 and Y81 for TNM, Y150 for ONOO–, and Y83 and Y158 for O3/NO2. The tyrosine residues Y83 and Y81 are located in a hydrophobic cavity, while Y150 and Y158 are located in solvent-accessible and flexible structures of the C-terminal region. The heterogeneous reaction with O3/NO2 was found to be strongly dependent on the phase state of the protein. Nitration rates were about one order of magnitude higher for aqueous protein solutions (∼20% per day) than for protein filter samples (∼2% per day). Overall, our findings show that the kinetics and site-selectivity of nitration strongly depend on the nitrating agent and reaction conditions, which may also affect the biological function and adverse health effects of the nitrated protein.

, b) setup 2 used for experiments [11][12][13][14][15] (Table S2)    (arithmetic mean ± standard deviation). High intensity is defined as higher than 10% of the mean intensity (~10 5 to 10 6 ) and low intensity is defined as lower than 1% of the mean intensity (~10 4 ).   Tyr position  ppb O 3 at a relative humidity of 92%. The nitration and oxidation sites of four different exposure times were analyzed with four parallel exposed filters for each time. (+++) means that the modified site is found in three of three analytical replicates with high intensities, (++) means that the modified site is found in two of three analytical replicates with high intensities or found in all three analytical replicates with low intensities, (+) indicates that the modified site is found in one of three analytical replicates with high or low intensities or found in two of three analytical replicates with low intensities, and (-) means that the modified site is not found in any of the runs.
The mean intensity for the LC-MS/MS runs was (5.04 ± 0.92)10 6 (arithmetic mean ± standard deviation). High intensity is defined as higher than 10% of the mean intensity (~10 5 to 10 6 ) and low intensity is defined as lower than 1% of the mean intensity (~10 4 ).  0  7  0  0  34  1  22  3  26  21  Table S10. Triplicate LC-MS/MS analysis of Bet v 1 in aqueous solution exposed to 100 ppb NO 2 and 100 ppb O 3 . The nitration and oxidation sites of two different exposure times were analyzed with two parallel experiments. (+++) means that the modified site is found in three of three analytical replicates with high intensities, (++) means that the modified site is found in two of three analytical replicates with high intensities or found in all three analytical replicates with low intensities, (+) indicates that the modified site is found in one of three analytical replicates with high or low intensities or found in two of three analytical replicates with low intensities, and (-) means that the modified site is not found in any of the runs. The mean intensity for the LC-MS/MS runs was (5.04 ± 0.92)10 6 (arithmetic mean ± standard deviation). High intensity is defined as higher than 10% of the mean intensity (~10 5 to 10 6 ) and low intensity is defined as lower than 1% of the mean intensity (~10 4 ).  --++  -+++  -++  +  +  +  mock  Bet_m  -----------18  Bet_3  --+  -++  --++  -+++  -Bet_4  --++  -+++  +  -+  ++  +  +  mock  Bet_m  --------

S2 HPLC-HR-MS analysis of intact Bet v 1.0101
Monolithic 150 x 0.20 mm PS-DVB capillary columns were prepared according to a previous published protocol. 2 Separations were performed with acapillary -HPLC system (Model LTQ Hybrids (Sigma Aldrich).The mass spectra were analyzed by using the data evaluation software Xcalibur (Thermo Scientific) and the implemented deconvolution toolXtract. For isotopically unresolved mass spectra the software ProMass (ThermoFisher Scientific) was utilized to calculate the average molecular mass of the intact protein.

S3 Experimental details for the nitration of Bet v 1 using O 3 /NO 2 mixtures
A total of 17 experiments were conducted (see Table S2 for experimental conditions). For experiments 1-15, syringe filters (cellulose acetate membrane, 1.2 µm pore size, 30 mm diameter, sterile; FP30, Whatman GmbH, Germany) were pre-washed with ultrapure water and freezedried. The pre-treated filters were then loaded with 100 µL of Bet v 1 dissolved in 10 mM Na 2 HPO 4 buffer at different concentrations to yield the loads listed in Table S1 and freeze-dried again. For each experiment five loaded syringe filters were prepared, of which four were used for the exposure experiment and the fifth filter was treated as a mock sample. For experiments 1-10 (setup 1, Fig. S1), the protein-loaded filters were exposed to mixtures of O 3 in synthetic air plus were obtained from AIR LIQUIDE (Germany). NO 2 concentrations were adjusted by varying the source flow rate (5-20 mL min -1 ). Ozone was produced from synthetic air passing by a partly shielded mercury lamp (Jelight Company, Inc., Irvine, US) at 1.5 L min -1 . Water vapor was generated by bubbling N 2 through a washing bottle filled with autoclaved high purity water at different temperatures and flow rates of 1.0 L min -1 or 1.5 L min -1 to adjust the relative humidity (RH). Experiments 1-6 were conducted at a lower relative humidity (RH = (47 ± 3)% (arithmetic mean ± standard deviation)), while experiments 7-10 were carried out at a high relative humidity (RH > 98%). In the latter experiments (experiments 7-10), condensation was observed in the tubing and on the filters. The RH sensor has a working range of 5% -98% RH with an accuracy of ± 2% RH at 25°C.
After the respective exposure times, the proteins were extracted from the filters with 0.6 mL ultrapure water. Further analysis was performed using the extractable and soluble fraction.
To further investigate the role of the phase-state of the protein for the nitration kinetics, experiments 16-17 were conducted using setup 3 (Fig. S1) The bulk concentrations of O 3 and NO 2 in an aqueous solution can be estimated by using the Henry's law constant of K sol,cc,O3 ≈ K sol,cc,NO2 ≈ 10 -5 mol cm -3 atm -1 . 4 Here, we assumed a maximal ND of 0.28, which is equivalent to 2 fully nitrated Tyr residues in Bet v 1. The above differential equations were numerically solved using a Matlab software with varying k BR,a and k BR,b to fit to the experimental data. The formation of a ROI, most likely a Tyr radical, was found to be the rate-limiting step, which is consistent with previous studies 5, 6 , and k BR,a could be well-constrained to be 4 × 10 -17 cm 3 s -1 (= 2.4 × 10 4 M -1 s -1 ). This value is within the range of reported value for bulk reaction between amino acid and ozone in water, which vary by three orders of magnitude in constrained as the modeled ND is not sensitive to k BR,b if k BR,b is >10 -15 cm 3 s -1 .
To analyze and interpret the obtained experimental data for the heterogeneous reaction of Bet v 1 on filters at <92% RH, the kinetic multi-layer model for aerosol surface and bulk chemistry (KM-SUB 12 ) was applied. KM-SUB explicitly resolves gas-particle interactions between O 3 , NO 2 , and protein including reversible adsorption, surface-bulk exchange, bulk diffusion, and surface and bulk reactions. The required kinetic parameters include Henry's law constants and second-order bulk reaction rates were assumed to be same as for the aqueous-phase reaction case.
The other parameters are surface accommodation coefficients and desorption lifetime, which were assumed to be same as those determined in the previous study of nitration of bovine serum albumin. 5 The total surface area of deposited Bet v 1 was assumed to be the same as the geometric surface area of the filter for simplicity.
At an RH < 92% the physical state of protein is (semi-)solid with typical bulk diffusion coefficient of oxidants (D b ) on the order of 10 -10 -10 -8 cm 2 s -1 . 7 In Figure 4a the gray shaded area shows the range of simulated ND assuming D b = 10 -9 cm 2 s -1 and gas-phase O 3 and NO 2 concentrations of 100 -470 ppb. The model captured the observed data fairly well (Fig 4a).
When an increased D b value of 10 -5 cm 2 s -1 was used that is a characteristic magnitude of D b in liquid, it explains the observed enhanced ND for protein on filter under condensing conditions at RH > 98% (Fig. 4a).  Figure   4a. The ND ranged from 2% to 4% for Bet v 1 on filters (experiments 1-6 and 11-14, Table S2).
Variation of RH from 47% to 92% did not significantly influence the ND.
Intact protein analysis by HPLC-HR-MS showed mainly oxidation (∆m = 16 Da) for the heterogeneous nitration at O 3 and NO 2 mixing ratios of 100 ppb and low RH (~45%) (experiments 1-6) (see Table S11). After 70 h of exposure the relative abundance of oxidized protein reached more than 60%.
For low RH (~45%) experiments we did not observe any nitrated tyrosine containing peptides in the HPLC-MS/MS analysis. In Figure 4b the HPLC-MS/MS analyses of tryptic peptides of Bet v 1 exposed to O 3 /NO 2 at a relative humidity of 92% (experiments 11-15, Table S2) showed that four or five of seven tyrosine residues could be identified with an amino acid sequence coverage of (71.1 ± 3.4)% (arithmetic mean ± standard deviation) (for details see Table S8). Y83 and Y158 were found in nitrated form in all samples, while Y158 was also present in oxidized form (+ OH) in some samples. Y150 was found nitrated (+ NO 2 ) or oxidized (+ OH) in some samples, Y66 was found in oxidized form (+ OH), and methionine at position 139 was also found in oxidized (+ O) form (for details see supplementary Table S9). However, Y83 and Y158 were the preferred sites for the heterogeneous nitration of Bet v 1 for different reaction times (Fig. 4b). For Y83 we observed a slight decrease at reaction times up to 48 h. This might be due to enhanced degradation and will be studied in more detail in follow-up studies. In addition, we observed high levels of oxidation at Y158 for short reaction times (for details see table S8) supporting the assumption of reactive oxygen intermediates to be involved in the first step of the nitration reaction. 5 2) Reaction of gaseous O 3 /NO 2 with Bet v 1 on filters at >98% RH The ND reached up to 6% after 19 hours for >98% RH (water condensation on filters, experiment 7-10, Table S2).
Intact protein analysis by HPLC-HR-MS of filter samples with high RH under water condensing conditions (>98% RH, experiments 7 + 9), the addition of one or two nitro groups was observed, along with a smaller amount of oxidation (Table S12). Thus, oxidation seems to be favored under low RH conditions, whereas under high RH conditions nitration appears to be more prominent.

3) Reaction of dissolved O 3 /NO 2 with dissolved Bet v 1 in aqueous solution
The ND for the homogeneous reaction in aqueous solution (both the gases and the protein are dissolved in an aqueous solution, experiments 16-17,  Table S8) differed from the other samples.
Here, Y150 was observed to be the most efficiently nitrated residue and the pattern was more in line with the pattern of ONOOnitration. This discrepancy, which may suggest different rates for competing reaction pathways, will be investigated in future studies.
Y158 is located at the C-terminus of the Bet v 1 molecule within a coil structure motif. As coil structures belong to the most flexible parts of a protein, Y158 might have an increased probability for heterogeneous nitration, because hydration dynamics occur fastest in lower-order structures. 13 Y83 and Y81 are located at the hydrophobic cavity and might be a preferred site, due to 1) the stabilization of Tyr· in hydrophobic environments, and 2) because most of the reactive nitrogen compounds are hydrophobic gases. 14