A cohort of 120 children born at a single Kaiser Permanente Southern California (KPSC) hospital in 2007 was identified. Children were selected based on estimated PM_2.5 exposure during the entire pregnancy. Children with low and high exposures were oversampled to increase the variability of exposure and increase the power to detect associations (54). We obtained the newborn DBS samples collected at birth from these children from the California Biobank archive in the Department of Public Health. Children's sex and gestational weeks at delivery based on ultrasonography were obtained from KPSC electronic medical records databases. The Institutional Review Board from KPSC and the California Health and Human Services Agency approved this study and waived individual subject consent. A summary of cohort characteristics, air pollution exposure estimates, and adduct concentrations is provided in Table 2.

Average monthly air pollution exposure, including PM_2.5, PM₁₀, NO₂, and O₃ was estimated based on residential addresses at birth recorded in KPSC electronic medical record. The birth addresses were geocoded using MapMarker USA Version 28.0.0.11. Exposure data was compiled from the EPA regional air quality monitoring network across Southern California. To estimate the exposure at each residential location, we used the inverse distance-weighted monthly average from four closest monitoring stations within 50 km, except for geocoded locations within 0.25 km of a monitor, for which only data from the nearest monitoring station were used (55). Although the distance-weighted approach has limited accuracy in areas with sparse monitoring networks, performance is acceptable in Southern California due to the dense geographical network of historical measurements covering the region. In a previous Southern California study evaluating this method using leave-one-out validation for monthly monitoring station data, the coefficients of determination (R²) were 0.76, 0.73, 0.53, and 0.46 for O₃, NO₂, PM_2.5 and PM₁₀, respectively, with lower R² values for PM attributed to the local (primary emission) dust component that is not regional (56). Bias was <1 ppb or 1 μg/m³. Each address was assigned the monthly average of the 24-h concentrations of PM_2.5, PM₁₀, and NO₂. For O₃, the monthly average of daily maximum 8-h concentrations was estimated. Averages of the monthly concentrations during each trimester and last 30 days of pregnancy were then aggregated from these monthly estimates, with each specific time window determined based on the last menstrual period (LMP) date estimated from fetal ultrasound used in clinical care. For months overlapping different exposure windows (e.g., first and second trimesters), the exposure was assigned proportional to the number of days in each window.

HSA was isolated from DBS samples using the methods described by Yano et al. (48) with minor modifications. Briefly, 3.2 mm DBS punches were extracted in a 55 μL solution of 45% methanol in deionized water to precipitate hemoglobin and other interfering proteins. All solvents including deionized water were HPLC grade or higher. Samples were agitated at room temperature for 30 min and centrifuged at 20k RCF for 15 min at 4°C. The supernatant was diluted with 95 μL of a buffer consisting of 50 mM triethylammonium bicarbonate, 1 mM EDTA, with a pH of 8.0. 130 μL of filtered solution was then trypsin digested using 1 μL of 10 μg/μL trypsin. Protein digestion was performed using a Pressure Biosciences™ Barozyme HT48 with 30 pressure cycles of 20 kpsi for 50 s, then atmospheric pressure for 10 s for each cycle. Following pressure digestion, 3 μL of 10% formic acid was added to each sample to denature the trypsin and samples were centrifuged for 2 min at 10,000 × g. Analyses were then performed using 100 μL transferred to 300 μL borosilicate silanized glass vials (Microsolv Technology Corporation).

A panel of adducts were chosen based on previous associations with air pollution exposure and oxidative stress-related processes, described in Table 1. Chromatographic separations were performed using an Agilent 1,260 Infinity system with an Agilent Poroshell 120, 3 × 50 mm column, with 2.7 μm diameter particles. An Agilent 1,260 Infinity Binary Pump A was used with a 10-min gradient, with solvents A: 0.1% v/v formic acid in deionized water; and B: 100% acetonitrile and the flow rate used was 0.7 ml/min.

The LC gradient was: 0–1 min 15% B, 1–3 min 30% B, 3–6 min linear gradient from 30 to 45% B, 6–8 min 100% B, 8–10 min 15% B. The column was maintained at 37? C using an Agilent Infinity 1,260 Thermostatted Column Compartment (G1316A). Samples were injected using an Agilent 1,260 Infinity MicroHP sample handling system, using 25 μl of sample per injection. To reduce sample carryover between runs, the needle was automatically rinsed for 10 s using 30% methanol prior to each injection.

Analyses were carried out on the Agilent 6,490 QqQ mass spectrometer using an iFunnel electrospray source with Jetstream technology. Mass spectrometer parameters were based on those used in Domanski et al. (57). Adduct levels were derived from the standard (not scheduled) SRM peak area of the [precursor (3+) –>y17 (2+)] transition for the principle peak coeluting with the [precursor (3+) –>b3+] qualifier peak (18). Collision energies were calculated using default parameters creation in Skyline (58) with no further optimization and the overall cycle time was ~750 ms. Other instrumental parameters were set using the Agilent Autotune functionality in positive ion mode, with Agilent ESI-L low concentration tuning mix (Catalog# G1969-85000) for mass calibration.

Data acquired from targeted runs on the Agilent 6,490 QqQ was imported into a Skyline document containing the full set of transitions for each run. To quantitate the level of adducts across samples, the summed peak areas of 3 transitions per adduct species was divided by the combined peak areas of 3 housekeeping peptide transitions. The tryptic housekeeping peptide is adjacent in sequence to the T3 tryptic peptide and is used to estimate the concentration of HSA in each injected sample. Housekeeping peptide (Sequence: ⁴¹LVNEVTEFAK⁵⁰) undergoes little enzymatic and non-enzymatic modification, is adjacent to the T3 peptide on HSA, and has similar tryptic digestion efficiency and ionization as the T3 peptide. As a result, this approach accounts for digestion and ionization efficiency across samples, as described in detail by Grigoryan et al. (18).

The limit of quantification (LOQ) for each adduct was calculated as 10 times the standard deviation of ion intensities from 46 blank samples (59). Blank samples were analyzed within each sample batch.

Descriptive statistics were used to examine cohort distribution. The distributions of adduct levels were not normally distributed and natural log transformation improved the distribution. The relationships between log of adduct levels and air pollution exposures were first examined by scatter plot followed by regression analysis. Data plots did not suggest strong non-linear associations and non-linear exposure terms in regression models were also largely not significant. Thus, linear regression models were used to explore associations between log of adduct levels (as dependent variable) and PM_2.5, PM₁₀, NO₂, and O₃ exposures adjusted for sex and gestational weeks at delivery. Additional adjustment for Hispanic ethnicity (46.7% of the cohort) did not affect the results. As this study was conducted as a proof of concept with limited sample size and limited previous knowledge about potential confounders, no adjustment for other factors was considered. All analysis was performed using SAS Enterprise Guide 7.1 (SAS Institute, Inc., Cary, NC). We assessed statistical significance based on a 2-sided p-value <0.05.

S-sulfinic acid was associated with air pollution exposure during pregnancy. Possitive associations were observed between S-sulfinic acid and PM₁₀ exposure in the second trimester, S-sulfinic acid and O₃ exposure in the first trimester, and S-sulfinic acid and NO₂ in the third trimester [mean difference (MD), representing the change in the log of adduct associated with change in one unit of air pollution = 0.0032, 0.0043, 0.0146]. In contrast, negative associations were observed between S-sulfinic acid and O₃ in the 3rd trimester and with O₃ exposure in the last 30 days of pregnacy (MD = −0.0049, −0.0037).

S-GSH adducts were positively associated with exposure to O₃ during the 3rd trimester (MD = 0.0035). In addition, significant associations were observed with S-CysGly and S-γ-GluCys adducts, which are two-amino acid precursors of GSH. S-CysGly was positively associated with PM_2.5 exposure in the 1st trimester and S-γ-GluCys was positively associated with PM_2.5 exposure duiring the last 30 days of pregnancy (MD = 0.0353, 0.0202). In contract, S-γ-GluCys was negatively associated with O₃ exposure duiring the last 30 days of pregnancy (MD = −0.0131). Finally, S-methanethiol was negatively associated with NO2 exposure during the 1st trimester (MD = −0.0439).

Significant relationships were also observed between S-crotonaldehyde and air pollution constituients thoughout all trimesters of pregnacy. While strong positive assocaitions were observed between S-crotonaldehyde and O₃ during the 3^rd trimester and last 30 days of pregnancy (MD = 0.0164, 0.0138), associations between S-crotonaldehyde and PM_2.5, PM₁₀, and O₃ were negative during earlier stages of pregnancy.

Air pollution exposure during the last 30 days of pregnancy was associated with decreased S-sulfinic acid, decreased S-γ-GluCys, and increased S-crotonaldehyde, which are all consistent with increased levels oxidative stress. Because HSA has a residence time of about a month (40–42), HSA-Cys³⁴ adducts measured in newborn DBS samples (collected a day or two after birth) reflect an integration of circulating reactive compounds in the blood during this last 30 days of pregnancy. Therefore, adducts associated with oxidative stress triggered by air pollution exposure late in pregnancy can be directly captured in the newborn adductome.

Exposure to O₃ during the last 30 days of pregnancy was associated with decreases in S-sulfinic acid. While it seems counterintuitive that HSA-Cys³⁴ oxidation products would decrease with oxidative stress (i.e., S-sulfinic acid), these finding are consistent with previous reports of decreases in S-sulfinic acid in association with cigarette smoking (18), chronic obstructive pulmonary disease (19), and ischemic heart disease (19). In addition, decreases in other direct oxidation products (i.e., S-sulfonic acid) have been associated with O₃ exposure (19), which is consistent with our findings. It has been proposed that negative relationships between HSA-Cys³⁴ oxidation products and increased oxidative stress may be due to perturbations to the redox proteome and/or are the result of different rates of adduction depending on the oxidation status of HSA-Cys³⁴ (e.g., oxidized vs. reduced HSA) (18, 60–62).

O₃ exposure in the last 30 days of pregnancy was negatively associated with S-γ-GluCys. S-γ-GluCys is a two-amino acid precursor of GSH, which contains a free thiol that binds to HSA-Cys³⁴ forming a mixed disulfide. Oxidative stress can result in the depletion of small thiol compounds in plasma, and therefore decreased levels of S-γ-GluCys is consistent with increased oxidative stress caused by exposure to O₃ in late pregnancy (63).

Finally, Exposure to O₃ during the last 30 days of pregnancy was associated with increases in levels of S-crotonaldehyde. Since S-crotonaldehyde is a reactive aldehyde produced through lipid peroxidation, increased levels of S-crotonaldehyde are indicative of increased levels of oxidative stress. These findings are consistent with previous findings, where biomarkers of lipid peroxidation were positively associated with O₃ exposure (64).

In contrast with air pollution exposures occurring during the last 30 days of pregnancy, we observed an opposite trend with air pollution exposure in earlier stages of pregnancy (i.e., not in the last 30 days of pregnancy).

S-sulfinic acid showed an opposite trend earlier in pregnancy with a negative associated with O₃. While a positive association was observed between NO₂ exposure and S-sulfinic acid in the 3rd trimester, this association was not significant in the last 30 days. Therefore, we infer that NO₂ exposure earlier in the 3rd trimester (i.e., not in the last 30 days) is most likely responsible for this trend, which is consistent with our O₃ findings.

Exposures to PM_2.5, PM₁₀, and O₃ in the 1st and 2nd trimesters were consistently negatively associated with S-crotonaldehyde, suggesting that exposures to air pollution earlier in pregnancy are associated with a decrease in lipid peroxidation around the time of birth.

The residence time of HSA in blood of ~1 month is important to the interpretation of these findings. Adducts measured in newborn DBS samples capture an integration of adducts formed through reactions with circulating small molecules in the infant's blood during the last 30 days of pregnancy. As a result, the newborn DBS adductome does not directly capture adducts related to environmental exposures or biological responses to exposures earlier in pregnancy, but rather likely reflect adaptive mechanisms that may have been primed by air pollution exposures earlier in fetal development. In a previous study, Tissot van Patot et al. (65) observed that infants born at high altitude appeared to develop protective mechanisms against oxidative stress compared with infants born at sea level. In that study, infants born at sea level displayed evidence of oxidative stress during labor, while infants born at high altitude had minimal or no oxidative stress at the time of delivery. The authors hypothesized that fetuses developing at high altitude exposed to chronic levels of oxidative stress during pregnancy developed compensatory responses to oxidative stress by the time of delivery. Along these same lines, we speculate that infants exposed to oxidative stress from air pollution during pregnancy may develop similar mechanisms for combating oxidative stress around the time of birth. If this is true, decreases in oxidative stress-related adducts measured around the time of birth may be due to a heightened antioxidant response primed by earlier exposures to air pollution.

It is well-established that exposures to air pollution during pregnancy causes oxidative stress (9, 15, 66). Because HSA-Cys³⁴ adducts measured in newborn DBS samples provide a record of small reactive molecules circulating in the infant's blood during the last month of pregnancy, it is not surprising that exposure to air pollution in the last 30 days of pregnancy was associated with increased levels of oxidative stress captured in the newborn DBS adductome. Adducts that were associated with air pollution in the last 30 days of pregnancy were products of lipid peroxidation (S-crotonaldehyde), ROS (S-sulfinic acid), and small thiol compounds that provide antioxidant defense (S-γ-GluCys). Thus, our adductomics approach captures three distinct layers of oxidative stress using a single assay. Because directly obtaining biospecimens from newborns before and around the time of delivery is rarely possible, measuring protein adducts in newborn DBS samples provides new opportunities to assess prenatal environmental exposures and biological responses to chemical stressors prior to birth.

We also observed an opposite trend with air pollution exposures occurring earlier in pregnancy (i.e., not in the last 30 days), when adducts related to oxidative stress were consistently found to be inversely related to air pollution exposure. We are not suggesting that exposures to air pollution earlier in pregnancy is beneficial to early development. In fact, mounting evidence supports a strong link between exposure to air pollution during all stages of pregnancy and adverse children's health. Rather, we speculate that chronic exposures to air pollution may trigger adaptive changes in redox biology during critical stages of fetal development that may help to shield the infant from environmental stressors around the time of birth.