The potential pathways underlying the association of propyl-paraben exposure with aeroallergen sensitization and EASI score using metabolomics analysis

Propyl-paraben exposure is associated with aeroallergen sensitization, but its association with atopic dermatitis (AD) is inconclusive. No studies have been conducted on the metabolomic pathways underlying these associations. We investigated the associations between propyl-paraben exposure and aeroallergen sensitization, AD, and Eczema Area and Severity Index (EASI) score and identified the underlying pathways using untargeted metabolomics analysis. We enrolled 455 children in a general population study. Skin prick tests were performed with the assessment of EASI score. Urinary propyl-, butyl-, ethyl-, and methyl-paraben levels were measured. Untargeted metabolomics analysis was performed on the first and fifth urine propyl-paraben quintile groups. The highest urine propyl-paraben quintile group was associated with aeroallergen sensitization, but not with AD. Glycine, threonine, serine, ornithine, isoleucine, arabinofuranose, d-lyxofuranose, citrate, and picolinic acid levels were higher, whereas palmitic acid and 2-palmitoylglycerol levels were lower in the highest quintile propyl-paraben group, than in the lowest quintile group. The propyl-paraben-induced metabolic perturbations were associated with serine and glycine metabolisms, branched-chain amino acid metabolism, and ammonia recycling. Propyl-paraben exposure was associated with aeroallergen sensitization and EASI score, partially via metabolomic changes related with oxidative stress, mTOR, peroxisome proliferator-activated receptors pathway, aryl hydrocarbon receptor signaling pathways, and tricarboxylic acid cycle.

. Demographic and clinical characteristics of the participants in the first and fifth quintile groups of urinary propyl-paraben levels. BMI body mass index, EASI Eczema Area and Severity Index, IQR Interquartile range, SD standard deviation. *BMI categories were classified using the BMI z-score. Obesity was defined as a BMI z-score of ≥ 1.62 and overweight was defined as 1.03 ≤ BMI z-score ≤ 1.61. ‡ AD was defined as the presence of AD symptoms in the preceding 12 months. † Missing = 4. Data in bold means statistically significant values (P < 0.05).

Variables
Children in the first quintile of urine propyl paraben (n = 91) Children in the fifth quintile of urine propyl paraben (n = 91) P value Age, years ± SD 10.9 ± 0.8 11 www.nature.com/scientificreports/ Associations between EASI score and urinary propyl-paraben levels. To identify the associations between the severity and extent of skin inflammation in AD and urinary propyl-paraben levels, we measured EASI score of the study participants, regardless of the presence of AD. When the lowest quintile of urinary propyl-paraben levels was considered as a reference, the EASI score was significantly associated with the highest quintile group of urinary propyl-paraben, both in children with AD (aOR, 4.568; 95% CI 1.215-17.172, P = 0.025) as well as in all study participants (aOR, 2.441; 95% CI 1.014-5.876, P = 0.046; Table 3). However, there were no significant associations between EASI score and urinary levels of other types of paraben in the total population (Table S2).

Urinary metabolomics analysis with global metabolomic profiling.
To identify the possible biomarkers and altered pathways linking propyl-paraben exposure with aeroallergen sensitization and EASI score, urinary metabolites in the highest and lowest quintiles of urinary propyl-paraben were analyzed and compared.
We validated the reliability of the analytical performance by using the QC samples tightly clustered in the principal component analysis (PCA) plot (Fig. S2A). The global metabolomic profiling, which represents the differences in the distribution of the metabolomic datasets between the two groups, identified no apparent discrimination between the two groups. We detected the 1543 metabolic features using gas chromatography time-of-flight mass spectrometry (GC-TOFMS). The features were involved in data processing; peak deconvolution, noise reduction, removal of artificial, and data normalization. Afterward, we performed statistical analysis using the processed data. As a result, six amino acids, two fatty acids, four carbohydrates, one citric acid, one glycolic acid, and one carboxylic acid Table 2. Odds ratios (95% confidence intervals) for the associations between aeroallergen sensitization and atopic dermatitis symptoms in the previous 12 months by quintile levels of urinary propyl-paraben. aOR adjusted odd ratio, CI confidence interval, OR odds ratio, Ref. reference group. Numbers in bold indicate a significant difference (P < .05). *Logistic regression analysis was performed with adjustment for confounding factors, including age, gender, body mass index z-score, the presence of visible mold at home, and exposure to environmental tobacco smoke.  Table 3. Association between urinary propyl-paraben levels and EASI score in the total population as well as children with atopic dermatitis. aOR adjusted odds ratio, CI confidence interval, EASI Eczema Ares and Severity Index, Ref. reference population group. *P values calculated by logistic regression analysis for positive EASI score. Numbers in bold indicate a significant difference (P < .05). *Logistic regression analysis was performed with an adjustment for confounding factors, including age, gender, body mass index z-score, the presence of visible mold at home, and exposure to environmental tobacco smoke. www.nature.com/scientificreports/ were found to differ significantly between the two groups and they were identified through the identification of metabolic markers method (Figs. 1 and S4; Table S3).
Possible pathways involved based on the metabolomics analysis. Among the 15 metabolites, glycine, threonine, serine, ornithine, isoleucine, arabinofuranose, d-lyxofuranose, ribose, ethyl d-galactofuranoside, citrate, and picolinic acid levels were higher in the highest quintile group of urinary propyl-paraben than in the lowest quintile group (Fig. 1, Fig. S2B; Table S3). Meanwhile, the levels and mean peak areas of asparagine, palmitic acid, and 2-palmitoylglycerol were lower in the highest quintile group of urinary propyl-paraben, than in the lowest quintile group. To determine the biological relevance of the associations between propyl-paraben exposure levels and metabolic changes, pathway enrichment analysis was performed. The 15 metabolites that changed significantly were involved in multiple perturbed metabolic pathways, including the metabolism of amino acids, carbohydrates, lipids, and tryptophan, as well as energy with the urea cycle (Fig. S2). In addition, the propyl-paraben induced metabolic perturbations were significantly related to glycine and serine metabolism and ammonia recycling (Figs. S3 and S4). These results indicate that exposure to higher propyl-paraben levels may lead to more defects in the glycine and serine metabolism and ammonia recycling than that associated with exposure to lower levels of propyl-paraben. www.nature.com/scientificreports/

Discussion
Summary of the results of the present study. The present study found a significant association between aeroallergen sensitization, EASI score, and urinary propyl-paraben levels. However, we did not observe a significant association between AD and urinary propyl-paraben levels. Moreover, we identified the possible pathways that might underlie these associations using untargeted metabolomics analysis. However, there were no significant associations between levels of other types of parabens, including methyl-paraben, ethyl-paraben, and butyl-paraben, and aeroallergen sensitization, AD, and EASI score. The metabolites that differed between the highest and lowest quintile groups of urinary propyl-paraben included palmitic acid, 2-palmitoylglycerol, picolinic acid, and various hydroxyl acids, carbohydrates, and amino acids, suggesting that propyl-paraben exposure might disturb these pathways. The results of the present study might contribute to a better understanding of the associations among propyl-paraben exposure, aeroallergen sensitization, and EASI scores, while also providing suggestions, using an untargeted metabolomics approach, regarding the pathways that might underlie these associations. To our knowledge, this study is the first to identify the possible pathways that link aeroallergen sensitization, the degree of AD symptoms, and exposure to environmental parabens by using the untargeted metabolomics approach.
The meaning of the results of the present study with comparison with previous studies. Thus far, there have been no studies demonstrating a positive association between paraben exposure and AD, although only one study showed that there were no significant associations between eczema and triclosan and parabens 5 .
In our cross-sectional study, we did not observe a significant association between the presence of AD symptoms during the preceding 12 months and urinary propyl-paraben, methyl-paraben, ethyl-paraben, and butylparaben levels. However, EASI scores, which include the degree of erythema, edema or papulation, excoriation, and lichenification, as well as the extent of the skin affected, were associated with the highest quintile group of urinary propyl-paraben, when the lowest quintile group was used as a reference. The lack of consistency regarding the associations between AD, EASI score, and urinary propyl-paraben levels might be partially owing to the questionnaire-based investigation on the presence of AD symptoms in the preceding 12 months. In children with mild AD symptoms, AD symptoms might not be recognized without visiting a clinic. Moreover, the significantly increased levels of blood periostin in children in the highest quintile of urinary propyl-paraben levels when compared to those in the lowest quintile might support a positive association between the severity of AD symptoms and exposure levels of propyl-paraben, as children with severe and persistent AD have higher levels of blood periostin than those with mild AD 14,15 .
Since the EASI score was measured by pediatric allergists in all participants after confirmation of no intraand inter-differences in scoring, the EASI score has its own significance, regardless of the self-reported AD symptoms. Skin barrier defects, indirectly reflected in the positive EASI score, might be linked with aeroallergen sensitization 16 . Therefore, the consistent association between aeroallergen sensitization, EASI score, and urinary propyl-paraben levels suggest that propyl-paraben exposure affects allergic outcomes, including aeroallergen sensitization and skin inflammation in AD. In addition, since assessment items in the EASI score include clinical signs of skin barrier dysfunction and might, therefore, predict the later development of AD as one of the steps in allergic march 17 , long-term follow-up studies are needed to determine whether the association between propyl-paraben exposure and allergic outcomes affects the prognosis of AD or leads to the development of other allergic diseases.
Possible pathways underlying the associations between propyl-paraben exposure levels and allergic outcomes. Our data suggest that at least five altered metabolic pathways might be linked to exposure to propyl-paraben and aeroallergen sensitization and EASI score (Figs. 2 and S5). First, changes in carbohydrates and amino acids levels, including asparagine, threonine, serine, glycine, and ornithine, cause oxidative stress 18 , thereby playing an essential role in the pathogenesis of allergic inflammatory skin disease and allergen sensitization [19][20][21] . In addition, paraben itself can generate reactive oxygen species and induce oxidative DNA damage 22 . Second, the mammalian target of rapamycin (mTOR) might be involved in the link between exposure to propyl-paraben and aeroallergen sensitization and EASI score. Amino acid levels, including glycine, serine, and isoleucine, were increased in the highest quintile group of urinary propyl-paraben than those in the lowest quintile group. Specifically, isoleucine levels, which is one of the branched-chain amino acids (BCAA), showed the most significant difference between the two groups (Table S3 and Fig. S4). Changes in the metabolites involved in the BCAA pathway are known to influence the regulation of mTOR 23,24 , which plays a vital role in epidermal barrier formation and the signaling axis for the control of filaggrin as the fundamental pathophysiology of AD [25][26][27][28] . Third, propyl-paraben exposure is associated with aeroallergen sensitization and increased EASI score via the peroxisome proliferator-activated receptors (PPAR) pathway, which is based on decreased levels of palmitic acid and 2-palmitoylglycerol found in our study 29,30 . The decreased levels of palmitic acid and 2-palmitoylglycerol in the highest quintile group of urinary propyl-paraben, compared with those in the lowest quintile group, can result in the reduced induction of the PPAR pathway, which might be linked with decreased anti-inflammatory responses, increased allergic skin inflammation in AD, and aeroallergen sensitization 29 .
Furthermore, the increased levels of picolinic acid, a catabolite of tryptophan metabolism and an interim metabolite of kynurenine, in the highest quintile group of urinary propyl-paraben might suggest that the aryl hydrocarbon receptor (AhR) signaling pathway might be involved in the pathophysiology of aeroallergen sensitization and increased EASI score in exposure to propyl-paraben. AhR is an active ligand transcription factor affected by metabolisms or pollutants 31 , and activation of this pathway plays a role in immune-modulation due to its action on diverse types of immune cells 31 . After the binding of AhR with propyl-paraben, the subsequent cascade of immunomodulation and epidermal differentiation might play a role in aeroallergen sensitization and www.nature.com/scientificreports/ increased EASI score [32][33][34] . Finally, the tricarboxylic acid (TCA) cycle, reflected in the altered citrate level found in this study, can shape immune cell responses via changes in the metabolic pathways of immune cells 35 , which may play a role in aeroallergen sensitization and increased EASI score.

Limitations of the present study.
Our study has several noteworthy limitations. First, the presence of AD symptoms in the preceding 12 months was assessed using the ISAAC questionnaire. However, the ISAAC questionnaire has been widely used in the epidemiologic studies of allergic diseases to identify the prevalence of allergic diseases, including AD 36 . Second, we did not investigate the direct mechanisms of action of propyl-paraben exposure on aeroallergen sensitization and increased EASI score. Instead, the pathways that possibly underlie the associations were inferred based on the untargeted metabolomics analysis results, which can measure as many urine metabolites as possible without bias 37,38 . Third, we could not conclude whether propyl-paraben exposure might induce the metabolites changes in a dose-dependent manner because only some key metabolites, including glycine, threonine, and palmitic acid, showed significant associations with propyl-paraben levels (data not shown). Fourth, the urinary metabolomics analysis was performed only in the lowest and highest quintile groups of urinary propyl-paraben. Thus, by investigating the changes in metabolites between the lowest and highest quintile, the differences in metabolites between groups with different propyl-paraben exposure levels can be maximized. Exposure to different types of paraben is affected by personal preference in terms of products 4 , and therefore, the associations can differ based on race and ethnicity. Moreover, further studies are needed to confirm our study results due to the small sample size.

Conclusions.
In conclusion, we have identified that exposure to propyl-paraben was associated with aeroallergen sensitization and increased EASI score, partially through the metabolomics changes linked with altered oxidative stress, PPAR pathway, AhR signaling pathway, and TCA cycle. These novel findings provide new insights into the relationship between propyl-paraben exposure, aeroallergen sensitization, and EASI score and might help develop therapeutic targets for inflammatory skin diseases caused by environmental pollutants.

Methods
Study population. This study was performed as a part of the general population-based study (Seongnam Atopy Project for Children's Happiness 2017), which were performed in 11 elementary schools in Seongnam City, South Korea [39][40][41][42] . Among the 620 children (10 to 12 years old), 455 (73.4%) children, which had complete information on the questionnaires, blood and urine samples and results of skin prick tests (SPTs) were enrolled in this study. The study protocol was approved by the Institutional Review Board of CHA University (2017-04-049) and written informed consent was obtained from all parents or guardians of the children who participated. All methods performed out in accordance with relevant guidelines and regulations.
Questionnaire. The International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire was used to investigate the presence of allergic diseases, including AD 43 . AD was defined as the presence of AD www.nature.com/scientificreports/ symptoms in the 12 months preceding the questionnaire and was investigated using the following question: "During the previous 12 months, have you had AD symptoms, such as itchy eczema?".
EASI score. The EASI score is one of the most effective and reliable tools for the evaluation of AD 44,45 . To identify the skin barrier status, the EASI score was measured in all participants regardless of AD diagnosis ever or the presence of AD symptoms in the previous 12 months by pediatric allergists, who have no information regarding the presence of AD symptoms in the previous 12 months. The EASI score for each participant was measured in a closed space at the time of enrollment. Before measuring the EASI score, no significant difference was identified in the inter-rater reliability of the EASI score (P > 0.05). The EASI score was defined as negative or positive when the values were 0 or ≥ 0.1, respectively 46 .

Assessment of urinary parabens concentrations.
To avoid the possible confounding factors on the urinary paraben concentrations, urine samples were collected between 9 and 11 AM for all participants along with the measurement of the EASI score. Urinary concentrations of four parabens, including butyl-paraben, ethyl-paraben, methyl-paraben, and propyl-paraben, as well as creatinine concentrations, were measured from spot urine samples in all participants. The quantification of urinary parabens was performed according to the previously reported method, which used an Ultimate 3000 UHPLC system (Dionex, Sunnyvale, CA, USA) with a Thermo Scientific TSQ Quantiva Triple Quadrupole mass spectrometer (Thermo Scientific, San Jose, CA, USA) in multiple reaction monitoring modes 47 . The quantification limit for all analyzed parabens was 0.1 μg/L. The levels of each paraben were corrected using the urine creatinine concentrations to adjust for urinary dilution 48 .  www.nature.com/scientificreports/ NIST, LECO-Fiehn Rtx5, and Wiley 9. The mass spectra of the markers were matched with the libraries and then, authentic standards were analyzed to compare the spectra with the markers. Finally, the retention time of standards was compared with the markers by calculating the relative retention index 49 .

Measurement of blood periostin and chitinase levels.
Statistical analysis. Data were analyzed using covariance or logistic regression analysis and presented as mean differences or odds ratios with 95% confidence intervals (CIs). The logistic regression analyses were performed after categorizing of the EASI score into two groups according to an EASI score of 0 or ≥ 0.1 to identify the associations between urine propyl-paraben levels and EASI score. The logistic regression analysis was performed to determine the association of quintile levels of urinary propyl paraben with aeroallergen sensitization and EASI score with adjustment for confounding factors, including age, gender, body mass index z-score, the presence of visible mold at home, and exposure to environmental tobacco smoke. Multivariate analysis and pathway analysis were performed to determine significantly different metabolites between the highest and lowest quintile urine propyl-paraben groups. Random forest was used for the analysis of the metabolome profiles. All data analyses were conducted using SPSS version 25.0 (IBM Co, Armonk, NY, USA). A P-value below 0.05 was considered statistically significant.

Data availability
Scientific Reports | (2021) 11:3772 | https://doi.org/10.1038/s41598-021-83288-9 www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.