Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2024 Mar 5;58(11):4859–4871. doi: 10.1021/acs.est.3c08250

Plastic Food Packaging from Five Countries Contains Endocrine- and Metabolism-Disrupting Chemicals

Sarah Stevens †,*, Molly McPartland †, Zdenka Bartosova †, Hanna Sofie Skåland †, Johannes Völker †, Martin Wagner †,*
PMCID: PMC10956434  PMID: 38441001

Abstract

graphic file with name es3c08250_0006.jpg

Plastics are complex chemical mixtures of polymers and various intentionally and nonintentionally added substances. Despite the well-established links between certain plastic chemicals (bisphenols and phthalates) and adverse health effects, the composition and toxicity of real-world mixtures of plastic chemicals are not well understood. To assess both, we analyzed the chemicals from 36 plastic food contact articles from five countries using nontarget high-resolution mass spectrometry and reporter-gene assays for four nuclear receptors that represent key components of the endocrine and metabolic system. We found that chemicals activating the pregnane X receptor (PXR), peroxisome proliferator receptor γ (PPARγ), estrogen receptor α (ERα), and inhibiting the androgen receptor (AR) are prevalent in plastic packaging. We detected up to 9936 chemical features in a single product and found that each product had a rather unique chemical fingerprint. To tackle this chemical complexity, we used stepwise partial least-squares regressions and prioritized and tentatively identified the chemical features associated with receptor activity. Our findings demonstrate that most plastic food packaging contains endocrine- and metabolism-disrupting chemicals. Since samples with fewer chemical features induce less toxicity, chemical simplification is key to producing safer plastic packaging.

Keywords: endocrine disruptor, food contact, in vitro, mixture toxicity, nontarget chemical analysis, nuclear receptor, plastic

Short abstract

This study highlights the presence of harmful chemicals in plastic food packaging, indicating the need for a safer—chemically simpler—plastic design to address environmental and human health concerns.

1. Introduction

Plastics are complex mixtures of polymers and a multitude of chemicals used during production either to produce or to enhance the properties of the materials. These chemicals are intentionally added and include residual solvents, monomers, and catalysts, as well as a range of additives. In addition, plastic products contain nonintentionally added substances (NIAS), such as impurities and reaction or degradation byproducts generated during the production, use, and end-of-life phase of plastics.1 In fact, more than 13,000 plastic chemicals are known.2−4

Indeed, plastics are considered a main source of chemical exposure to humans and the environment.5 This is because most plastic chemicals are not chemically bound to the polymer matrix, resulting in their release from plastic via migration or volatilization. Within that broader context, plastic food contact articles (FCAs), that is, plastic used to package or process food, are particularly relevant for human exposure.6 For instance, certain plastic chemicals, such as bisphenol A (BPA) and phthalates, have been detected in more than 90% of the US population.7−9

Plastic chemicals have adverse health effects across the full life cycle of plastics.10 Here, endocrine-disrupting chemicals, compounds “that interferes with any aspect of hormone action”,11 are of particular concern. A chemical disruption of the endocrine system contributes to a wide range of adverse health effects, including reproductive, developmental and metabolic disorders, and cancer.12 There is robust evidence that links exposures to BPA and phthalates to such adverse outcomes13 resulting in substantial societal costs.14,15 Moreover, emerging evidence suggests that metabolism-disrupting chemicals (MDCs) represent another relevant class of compounds.16,17 MDCs promote obesity, type 2 diabetes, or other metabolic disorders, thereby contributing to the increase in noncommunicable diseases.16,18,19 Metabolic disruptions can be mediated via nuclear receptors, such as peroxisome proliferator-activated receptor γ (PPARγ) which is pivotal in lipid metabolism and adipogenesis.20,21 Additionally, pregnane X receptor (PXR), besides its role as xenobiotic sensor, is involved in regulating energy homeostasis, including glucose and lipid and bile acid metabolism.22−24 Its ability to bind diverse chemicals also renders PXR an interesting target to screen for baseline toxicity. Notably, BPA and phthalates also function as MDCs.18,25 Accordingly, plastic products, including FCAs, can be a source of exposure to endocrine disrupting chemicals (EDCs) and MDCs.

While much focus is placed on well-studied compounds, the endocrine- and metabolism-disrupting properties of most plastic chemicals remain unknown. There are multiple reasons for that, including the unresolved identity of many chemicals in plastics, especially of NIAS, and gaps in regulatory frameworks.26 Thus, consumers are exposed to mixtures of plastic chemicals with unknown compositions and toxicity. Given the vast number of chemicals present in real-world plastic products, bioassays represent a powerful tool to assess the joint toxicity of such complex mixtures27 and can be combined with nontarget mass spectrometry.28 In our previous work, we applied such an approach and demonstrated that thousands of mostly unknown chemicals are present in a single plastic product which induce a range of toxicological responses in vitro.29

Nontarget analysis (NTA) provides a comprehensive characterization of the chemical composition of plastics, yet it produces vast amounts of data that are challenging to interpret. This is because high-resolution mass spectrometry generates data on thousands of chemical features in a sample, rendering it difficult to pinpoint and identify the active compounds. Statistical models for data reduction, such as partial least-squares (PLS) regression, can help address this challenge. Unlike traditional multiple linear regressions, PLS regression models can manage data sets with numerous covarying variables and limited sample sizes.30,31 Here, a stepwise method based on variable influence on projection (VIP) scores can be used to select variables that are important for the PLS component.31 This approach has been successfully implemented by Hug et al. to characterize the chemicals in wastewater effluents inducing mutagenicity.32 Thus, integrating bioassays with NTA and multivariate statistics can improve our understanding of chemical mixtures in plastics.

Given our limited knowledge about EDCs and MDCs in plastics, this study aims to investigate the receptor activity induced by all chemicals present in plastic FCAs, representing a real-world scenario that covers unknown compounds and mixture effects. We characterized the receptor activity of chemical mixtures extracted from plastic FCAs of multiple polymer types in a set of reporter-gene assays relevant to human health covering PXR, PPARγ, estrogen receptor α (ERα), and androgen receptor (AR). To gain a more representative picture of the FCAs used globally, we analyzed 36 FCAs from the countries with the highest plastic waste generation per capita.33 Further, we used NTA to quantify the chemical features and tentatively identify the chemicals present in the FCAs. We employed PLS regressions as an approach to handle the large chemical complexity encountered in the FCAs and explored a potential relationship between the chemical features and the receptor activity of the samples. Our study confirms the widespread presence of EDCs and MCDs in plastic FCAs and led to the identification and prioritization of several known and unknown chemicals.

2. Materials and Methods

2.1. Samples

We purchased 36 plastic FCAs covering the seven polymer types with the highest global market share,34 including high- and low-density polyethylene (HDPE, LDPE), poly(ethylene terephthalate) (PET), polypropylene (PP), polystyrene (PS), polyurethane (PUR), and poly(vinyl chloride) (PVC) from domestic retailers in five countries (USA, U.K., South Korea, Germany, and Norway) (Table 1). We selected four of these countries because of their high plastic consumption, using the plastic waste share per capita as a proxy,33 and included Norway because of a local interest. Five to 12 items per country were purchased between winter 2020 and spring 2021. The samples consist of single-use packaging (cups, films, trays, etc.) and FCAs for repeated use (food containers, hydration bladders, etc.). Seventeen FCAs contained food, which was removed by washing with tap water at the local sites. To test the influence that food content has on chemical composition and toxicity, three products were acquired in duplicate, one item per product without and one with food. PET 6 and PS 5 contained broccoli salad or cream cheese, respectively, which was packaged by a local shop, which also provided the empty packaging. The third product (PP 5) contained yogurt, and we used the lid (same material) that was not in contact with the content. All samples were transported to the laboratory in PE bags (VWR) and their polymer type was determined using Fourier-transformed infrared spectroscopy and differential scanning calorimetry in the case of HDPE and LDPE (see Supporting Information, SI S1.1). In cases where the polymer type was provided on the packaging, we used that information to label the products.

Table 1. Plastic Food Contact Articles Analyzed in this Study and Number of Chemical Features.

polymer sample name product previous food contacta country chemical featuresb
HDPE HDPE 1 chewing gum tray yes Germany 164
  HDPE 2 food container no Germany 37
  HDPE 3 freezer bag no South Korea 260
  HDPE 4 milk bottle yes USA 297
LDPE LDPE 1 lid food container no Germany 74
  LDPE 2 zip lock freezer bag no Germany 1588
  LDPE 3 zip lock freezer bag no South Korea 1211
  LDPE 4 sausage packaging yes South Korea 2317
  LDPE 5 zip lock freezer bag no UK 1243
  LDPE 6 cling film no USA 366
  LDPE 7 frozen blueberry bag yes USA 3474
  LDPE 8 fruit netting yes Norway 1504
PET PET 1 oven bag no Germany 528
  PET 2 water bottle yes South Korea 640
  PET3 dairy cup yes UK 140
  PET 4 oven bag no UK 231
  PET 5 water bottle yes USA 379
  PET 6a food container no Norway 174
  PET 6b food container yes Norway 251
PP PP 1 dairy cup yes Germany 1220
  PP 2 instant food cup yes Germany 194
  PP 3 bowl no South Korea 90
  PP 4 coffee cup yes South Korea 751
  PP 5a yogurt cup lid no USA 1464
  PP 5b yogurt cup yes USA 1327
PS PS 1 dairy cup yes Germany 326
  PS 2 bowl no Germany 124
  PS 3 cup no South Korea 264
  PS 4 cupc no USA 1752
  PS 5a trayc no Norway 504
  PS 5b trayc yes Norway 1516
PUR PUR 1 hydration bladder no Germany 8762
  PUR 2 hydration bladder no Norway 9175
PVC PVC 1 drinking tube no Germany 2283
  PVC 2 drinking tube no Germany 769
  PVC 3 cling film no UK 9936
  PVC 4 cling film no USA 8946
PET/PE comp. 1 sausage packaging yes South Korea 3203
PUR/PE comp. 2 cheese packaging yes UK 6803
Open in a new tab
a

Note: these items had contact with food prior to analysis.

b

Number of chemical features detected in nontarget mass spectrometry.

c

Items made of extruded polystyrene. Comp. = composite material consisting of two polymers.

2.2. Sample Extraction

Methanol (99.8%, Sigma-Aldrich) was used to extract the chemicals from the FCAs as it allows for the extraction of compounds with a large polarity range and does not dissolve the polymers. To avoid sample contamination, all consumables used for the extraction (except plastic pipet tips) were made of glass or stainless steel, rinsed with ultrapure water (18.2 MΩ cm, PURLAB flex, ELGA) and acetone, and heated for at least 2 h at 200 °C. Prior to extraction, the FCAs that had food content and FCAs of repeated use were rinsed with ultrapure water and air-dried. 13.5 g of each FCA was cut into smaller pieces (0.5–0.8 × 2 cm, thickness ≤0.4 cm) and extracted with 90 mL of methanol in two 60 mL glass vials with polytetrafluoroethylene lined lids (DWK Life Sciences). Extraction was performed by sonication (Ultrasonic Cleaner USC-TH, VWR) for 1 h at room temperature. Immediately after, 1 mL of extract was removed for the chemical analysis and stored in glass vials at −20 °C. For the bioassays, 60 mL of extract was transferred to empty glass vials and evaporated under a gentle stream of nitrogen. When about 0.5 mL was reached, 600 μL of dimethyl sulfoxide (DMSO) was added, and the evaporation was continued until that volume was reached (i.e., 100-fold concentrated extracts). In parallel, four procedural blanks (PB 1–4) not containing plastics but only methanol underwent the same procedure as the samples to control for potential contamination during the extraction.

2.3. Reporter-Gene Assays

We used CALUX reporter-gene assays (BioDetection Systems B.V., Amsterdam) for human PXR, PPARγ, ERα, and AR to analyze the receptor activity. The assays were performed as described in Völker et al.35 with minor modifications (see SI S1.2). On each plate, negative controls (assay medium), vehicle controls (assay medium with 0.2% DMSO), and a concentration series of the reference compounds were included (PXR: nicardipine, PPARγ: rosiglitazone, ERα: 17β-estradiol, AR: flutamide; Table S1 and Figure S1). The AR assay was conducted in antagonistic mode with 0.5 μM dihydrotestosterone (corresponding to the EC80 in agonistic mode, see Figure S2, CAS 521-18-6, Sigma-Aldrich) as background agonist. Plastic extracts were diluted 500-fold in assay medium and analyzed in five concentrations serially diluted 1:2. Accordingly, the highest analyzed concentration was 1.5 mg plastic well–1 (PPARγ, ERα, and AR) or 0.75 mg plastic well–1 (PXR), which means that the response observed in the assays was caused by the chemicals extracted from that mass of FCA. Cytotoxic samples were further diluted 1:2 until reaching noncytotoxic concentrations.

After exposure for 23 h, high-content imaging was used to assess the cytotoxicity and normalize the reporter-gene response. The cells were stained with NucBlue (Thermo Fisher Scientific) for 30 min and imaged (Cytation 5 Cell Imaging Multimode reader, BioTek) and the nuclei were counted (CellProfiler 4.04). A 20% reduction in the nuclei count compared to the pooled negative and solvent controls was used as the cytotoxicity threshold. The receptor activity was subsequently analyzed by measuring luminescence (Cytation 5) of the lysed cells for 1 s following injection of 30 μL of illuminate mix containing d-luciferin as substrate. Afterward, the reaction was quenched in each well with 30 μL of 0.1 M NaOH. The extracts were analyzed in a minimum of three independent experiments, each with four technical replicates.

2.4. Data Analysis of Reporter-Gene Assays

GraphPad Prism (v10, Graph Pad Software, San Diego, CA) and Microsoft Excel for Windows (v2021–2306) were used for analyzing the bioassay data. Receptor activity was expressed as luminescence normalized to the number of cells well–1 and then normalized to the dose–response relationship of the reference compound analyzed on the same plate. Negative and solvent control data were pooled, as they were not statistically different (p > 0.05, Kruskal–Wallis with Dunn’s post hoc test), and the mean was set to 0%. The maximal response to the reference compound (100%) was set to the upper plateau of the dose–response relationship calculated using four-parameter logistic regressions. To derive the effect concentrations of the samples, the data were not extrapolated. The limit of detection (LOD) was determined as the average luminesce cell–1 of the pooled controls plus 3× the standard deviation. Samples that induced a receptor activity > LOD were considered active.

For data visualization (Figure 1E), the effect concentrations were normalized to the highest tested concentration (0%, 1.5 mg well–1) by calculating: (1 – EC/1.5) × 100. A Spearman rank correlation matrix was calculated between assay endpoints (normalized EC20/50) and feature count. A comparison of samples with and without previous food contact was done using Student’s t-tests. To assess if the FCA’s country of purchase or polymer type influences receptor activity, EC20/50 values of the different categories were compared using Kruskal–Wallis with Dunn’s post hoc tests. A p < 0.05 was considered statistically significant throughout these analyses.

Figure 1.

Figure 1

Open in a new tab

Activity of plastic chemicals extracted from FCAs at (A) PXR, (B) PPARγ, (C) ERα, and (D) Anti-AR. (E) Overview of the receptor activity and cytotoxicity normalized to the highest tested concentration and number of chemical features of the FCAs normalized to the highest detected feature number. Accordingly, lower EC20/50 values representing more potent receptor activity/cytotoxicity are indicated by darker red colors. EC20/50 data are derived from at least three independent experiments, each with four technical replicates per concentration (n ≥ 12). Note: LOD = limit of detection, a = samples with activity between LOD and 20% for PXR, PPARγ, and ERα or 50% for Anti-AR.

2.5. Chemical Analysis

The NTA was performed with an ultrahigh performance liquid chromatography system (Acquity I-Class UPLC, Waters) coupled to a high-definition hybrid quadrupole/time-of-flight mass spectrometer Synapt G2-S (Waters). The separation was performed on an Acquity UPLC BEH C18 column (150 × 2.1 mm ID, 1.7 μm, Waters) in a linear gradient with water and methanol as mobile phases, both containing 0.1% formic acid (Table S2). The mass spectrometer was equipped with an electron spray ionization source operated in positive mode. Data were acquired over the mass range of 50–1200 Da using a data-independent acquisition technique in high resolution (35 000, further details in Table S3). Mass spectral data of all samples can be assessed under https://doi.org/10.18710/LZNLFX. Data treatment and compound identification were performed as described previously36 with minor modifications (see SI S1.3 and S1.4). The PUR and PVC mass spectra and the PE, PET, PP, and PS spectra were processed separately because their different chemical compositions prevented a joint retention time alignment. Features (ions with a unique m/z and retention time) that had an abundance of less than 10-fold the highest across PBs and solvents were excluded from further analysis. Additionally, the abundance of the features was corrected by subtracting the maximum abundance of the respective features detected in the PBs.

2.6. Compound Identification, Toxicity, and Use Data

The features remaining after filtering were tentatively identified using the Metascope algorithm in Progenesis QI. The experimental spectra were compared with empirical spectra from MassBank (14 788 unique compounds, release version 2021.03), and spectra predicted in silico. For the latter, four databases were used as previously described.36 In addition, we constructed a fifth database containing the plastic chemicals reported by Wiesinger et al.4 as described in S1.3. For the identification, the spectra of each feature in the samples were compared to the spectra in the database, with a precursor ion tolerance of 5 ppm and a fragment ion tolerance of 10 ppm. The results of the tentative identification were filtered for hits with a score of ≥40 choosing the highest score in the case of multiple identifications. Compounds comprising these criteria are referred to as tentatively identified and correspond to the identification level 3.37

Publicly available toxicity data of all tentatively identified compounds were retrieved from the ToxCast and Tox21 databases as described previously.29 304 of the tentatively identified compounds had entries in the ToxCast database, and we extracted activity concentrations 50 (AC50) from 45 assays corresponding to the receptors analyzed here (SI S1.4, Table S4).

2.7. Partial Least-Squares Regression

To reduce the chemical complexity and prioritize chemical features that covary with the receptor activity, we conducted PLS regressions with the R package “mdatools”.38 Here, we included all features detected in ≥3 samples30 and that had an abundance higher than the 25th percentile across all features. The effect concentrations were normalized to the highest (0%) analyzed concentration of the samples. For each receptor, a separate model was applied. The PVC and PUR samples were excluded because three very cytotoxic samples had to be diluted and thus could not be analyzed at the same concentrations as the other samples. Through a stepwise variable selection method based on the variables’ influence on projection (VIP), the complexity of the model can be reduced, improving model performance, and selecting features that are important for describing both, the dependent and independent variables.31 In an iterative way, we excluded features with a VIP < 0.8 until the best model was found or the number of features included in the model was not further reduced.39 Model performance was validated by calculating 500 iterations with a random set of features containing the same number of features as those in the optimized models. To further identify and prioritize features that positively correlate with the receptor activity, we selected the 25% features that cluster closest to the receptor activity in the first and second components of the optimized model.

3. Results and Discussion

3.1. Receptor Activity of FCA Extracts

The chemicals extracted from 33 of 36 plastic FCAs interfered with one or more receptors (Figure 1). We detected PXR agonists in 33 products and PPARγ agonists in 23 products. Cytotoxicity was less prevalent across the samples but might have masked the responses induced by active chemicals in some cases (SI, S2.1, Figure S3). This indicates that plastic food packaging contains chemicals that activate the xenobiotic metabolism and interfere with energy homeostasis and metabolic functions. We detected estrogenic and antiandrogenic compounds in 18 and 14 products, respectively. The four procedural blanks were inactive across all assays and experiments, demonstrating that the sample processing did not result in contamination with chemicals interfering with the receptors investigated here. Taken together, these results imply that most of the plastic FCAs analyzed in this study contain EDCs and MDCs.

3.1.1. PXR Activity

PXR is the predominant target of plastic chemicals extracted from FCAs. All but three extracts activated this receptor, and 75% of the extracts produced an EC20 (Figure 1A). Three samples, including two PE containers (LDPE 1, HDPE 2) and a PP bowl (PP 3), did not activate PXR nor any other receptor, while eight extracts activated PXR only. This broad activation of PXR is unsurprising given its promiscuous nature. PXR plays a key role in cellular detoxification and can bind structurally diverse chemicals due to a large ligand-binding pocket with several loops in the ligand-binding domain.40,41 PXR has important cellular functions beyond its role as xenobiotic sensor, such as energy homeostasis and inflammation.42−44 A drug-induced dysregulation of PXR is associated with adverse health effects, including hypercholesterolemia and cardiovascular disease.45 Along this line, the plastic chemical dicyclohexyl phthalate was shown to induce PXR-mediated atherosclerosis in mice.46 Interestingly, PXR activation correlates positively with the number of chemical features and with all analyzed biological endpoints, except the estrogenic activity (Figure S4). Screening for PXR activity, therefore, provides a good initial representation of the general toxicity as well as the chemical complexity of mixtures of plastic chemicals.

3.1.2. PPARγ Activity

The chemicals in 23 FCAs covering every polymer type activated PPARγ (Figure 1B). Compounds extracted from LDPE and PVC products caused a strong receptor activation (>75%), whereas the chemicals in PET, PP, and PS FCAs induced effects above the LOD but below 20%. The abundant presence of PPARγ agonists in plastics in this study is interesting since fewer plastic extracts activated that receptor in our previous work.35 Since we used the same extraction method and reporter-gene assay as before, it seems plausible that we simply selected plastic products in which PPARγ agonists were more prevalent. PPARγ is considered the master regulator of adipogenesis20 and its activation by MDCs has been implied in the development of overweight, obesity, and associated metabolic disorders.18 Accordingly, we here demonstrate the widespread presence of MDCs in plastic FCAs.

3.1.3. Estrogenic Activity

In total, the chemicals present in 18 FCAs activated ERα (Figure 1C), including four out of five PS samples that induced a potent estrogenic activity (EC20 of 0.21–0.91 mg plastic well–1). In addition, the compounds in a frozen blueberry package (LDPE 3) and a yogurt cup lid (PP 5a) activated the ERα > 20%, and another 11 samples induced weak estrogenic effects (>LOD and <20%). This widespread detection of estrogenic chemicals in plastics contrasts with our previous work in which 4 out of 34 plastic products contained ERα agonists.29 Here, the 10-fold higher sensitivity of the CALUX system compared to the yeast-based reporter-gene assay results in lower detection limits and, thus, more detects. However, our findings align with previous studies that observed estrogenic activity leaching from plastic products, including toys.47−51 The prevalence of estrogenic compounds in plastics raises health concerns due to their potential to disrupt the endocrine system, which can, among others, result in developmental and reproductive issues, and an elevated risk of hormone-related cancers, such as breast and prostate cancer.12

3.1.4. Antiandrogenic Activity

We also detected significant antiandrogenicity in 14 samples (Figure 1D). Several LDPE, PVC, and PUR products contained chemicals inducing potent antagonistic effects at the AR, while the compounds extracted from PET and PP articles did not contain antiandrogens. At the highest tested concentration, exposure to four of the antiandrogenic samples (LDPE 2 and 5, PVC 1 and 4) resulted in 8–16% fewer cells than in the controls. Such reduced cell numbers can lead to false-positive responses in antagonist assays.52 However, an additional dilution confirmed that lower, noncytotoxic concentrations were also antiandrogenic (Figure S5). Our results are in accordance with Zimmermann et al.29 and Klein et al.53 who found a similar prevalence of antiandrogenicity in plastics. Further, antiandrogenicity has been detected in plastic baby teethers.48 Notably, the antiandrogenic and the PPARγ activities of the extracts correlate significantly (Figure S4) indicating that antiandrogens and metabolism-disrupting chemicals co-occur in FCAs. As for the other nuclear receptors, these results indicate that chemicals with an antiandrogenic mechanism of action are prevalent in plastic food packaging and containers.

Our findings indicate that EDCs and MDCs are frequently present in plastic FCAs from five countries and the compounds present in products of each polymer type interfere with PXR, PPARγ, and ERα, and most (HDPE, LDPE, PS, PUR, PVC) inhibited the AR. Nonetheless, we observed some interesting patterns with regard to the polymers: The chemicals in PET and HDPE FCAs activated fewer nuclear receptors (<1/3 active) as compared to the LDPE, PVC, and PUR products (>2/3 active, Figure 1E). Similar to our previous research,29 we found individual FCAs made of HDPE, LDPE, PET, and PP that did not contain chemicals interfering with the nuclear receptors. Interestingly, these products also contained very few chemical features (Table 1). This provides two important learnings: (1) it is feasible to produce plastic FCAs from an array of materials that do not contain EDCs or MDCs, and (2) simpler chemical formulations are key to achieving this.

3.2. Chemical Composition of Plastic FCAs

3.2.1. Individual Samples Contain a Wide Range of Chemical Features

Using high-resolution mass spectrometry, we detected 16,846 unique chemical features in seven PUR and PVC samples and 8665 features in the 29 PE, PET, PP, and PS samples. The number of features differed markedly between FCAs with a minimum of 37 features in a food container (HDPE 2) to a maximum of 9936 features present in a cling film (PVC 3, Figure 2A and Table 1). Across all samples, the median number of features was 696 and one-fourth of the samples had ≤238 or ≥2150 features, most of which produced robust mass spectrometry signals (Figure 2B). Also, the chemical fingerprints, that is, the presence and abundance of features in each sample, varied greatly between the FCAs (Figure 2C,D). None of the features was detected across all samples in the PE, PET, PP, and PS set, while 42% (3592 features) are present in only a single FCA (Table S5). Among the samples made of PUR and PVC larger clusters of abundant features are shared between samples (Figure 2D).

Figure 2.

Figure 2

Open in a new tab

Chemical composition of plastic FCAs. (A) Number of chemical features per sample, (B) abundance of features per polymer type (excluding composite samples), clustered heatmap of chemical features for (C) PE, PET, PP, PS samples, and (D) PVC and PUR samples.

These results are in line with prior research reporting large numbers of chemicals in consumer plastics using NTA.36,53 Other studies have reported lower numbers of features in plastic products,54,55 that align with our samples that have less features. Notably, the total number of features as well as the full data analysis parameters are rarely reported in nontarget studies of plastic chemicals, somewhat limiting our ability to put our results into context. Nonetheless, the fact that FCAs contain hundreds to thousands of features highlights one dimension of the chemical complexity of plastics, namely, the presence of a plethora of chemicals.

3.2.2. Polymers Differ in Number of Chemical Features

The number of features differs across the polymer types with a gradient ranging from HDPE (616 features), PET (1320), PS (2284), PP (2711), LDPE (5495), and PVC (12,683) to PUR (13,004, Table S6). We observed a similar pattern with regard to the features’ abundance in the mass spectrometry: the median abundances in HDPE (26) and PET (43) were significantly lower than in PUR and PVC (439 and 508). This indicates that the latter polymers do not only contain more chemicals but also higher levels of those. This is due to the fact that PVC and PUR require more additives in their production compared to other polymers.5,56−58

3.2.3. Polymers Are Chemically Diverse

Despite the general trend of more plastic chemicals being abundant in certain polymer types, we observed striking heterogeneity in the chemical composition of individual FCAs made of the same polymer (Figures 3A and S6). Typically, these products share <2% of the features, and in fact, most features (44–82%) are unique to a single product of a given polymer (Table S6). Only PUR-containing samples are an exception to this; they share 22% of features (Figure 3B). With regard to intentionally added substances, this heterogeneity may be explained by the wide range of additives with similar functionality available on the market.4 In terms of NIAS, this finding is more surprising, because one would expect similar reaction and degradation products to form in products made of the same polymer type. This indicates that very few common chemicals are indeed used or are present in a specific plastic type.

Figure 3.

Figure 3

Open in a new tab

Overlap of chemical features in (A) HDPE samples, (B) PUR samples, including the PE–PUR composite (comp. 2), (C) LDPE freezer bags, and (D) PVC drinking tubes. An overlap of <1% is not shown.

When comparing the chemical fingerprints, the samples did not cluster according to polymer type in the PE, PET, PP, and PS samples (Figure 2C). A frozen blueberry packaging (LDPE 7) has the largest number and abundance of features and is most dissimilar to all other samples. Nonetheless, it shares characteristic clusters of features with the three LDPE zip lock freezer bags (LDPE 2, 3, 5) which form a distinct family on its own. These samples share 835 features (41%) and seven of the ten most abundant features (Figure 3C) pointing toward a joint manufacturer. The seven PVC and PUR-based samples, which comprise less diverse articles, cluster better according to the polymer type (Figure 2D). Similar to the LDPE freezer bags, samples of the same polymer and product type tend to share larger fractions of chemical features (76% PUR hydration bladders and 53% PVC cling films). However, the two PVC drinking tubes share only 307 chemical features (11%, Figure 3D). Accordingly, neither the polymer type nor the product type is a good predictor of chemical composition. This demonstrates that products with the same functionality can be produced with different numbers and abundances of chemicals.

3.2.4. Tentatively Identified Compounds

In total, we tentatively identified 4137 chemicals (17% of all features, identification level 3).37 However, this corresponds to 2146 unique identifications only, indicating that multiple features were identified as the same compound (Table S7). In the PE, PET, PP, and PS samples, 1760 chemicals (20%) were identified, comprising 1182 unique chemicals. Among the FCAs made of PVC and PUR, 2377 features (14%) were tentatively identified corresponding to 1371 unique chemicals (Tables S7 and S14, more details in S2.3).

Of the ten most abundant features per sample, we tentatively identified 69 chemicals and retrieved use and toxicity information from PubChem. Our analysis shows that 43 of these chemicals are probably used in plastics as colorants, plasticizers, flame retardants, antioxidants, and processing aids (Table S8). The remaining features either had no use data (n = 12) or are unlikely to be used in plastic but as cosmetics, pharmaceuticals, or pesticides (n = 14). Among the plastic chemicals, we detected several known toxic or persistent and bioaccumulative compounds with high abundances. One such example is the plasticizer and flame retardant triphenyl phosphate (TPP, CAS 115-86-6) that was detected in both PUR hydration bladders (PUR 1 and 2) with high abundance. TPP is very persistent, very bioaccumulative, and toxic to aquatic life.4 In addition, the compound interferes with all the receptors analyzed here.59 This demonstrates that known hazardous chemicals are likely used and are present in plastic FCAs.

3.3. Predictors of Receptor Activity

We considered the factors of previous food contact, country of purchase, polymer type, and presence of known active chemicals as potential predictors of the receptor activity and, in addition, used PLS regressions to identify features potentially contributing to it.

3.3.1. Impact of Previous Food Content on Receptor Activity and Chemical Composition

We analyzed three matched samples of which we purchased one item with and another one without previous food content, each, to investigate the impact food storage has on toxicity and chemical composition. The overlap of chemical features in these paired samples ranged from 28% (PS, 1086 features) to 70% (PP, 422 features), with unique features in both conditions (Figure S7A). This indicates the migration of chemicals from food to the packaging and vice versa. If active, these chemicals will confound the bioassay results. We found that previous food contact increases PXR activity by 39% at the highest concentration for the PET and PS FCAs but decreased it by 12% for the PP cups (Figure S7B–D). However, all FCAs without previous food content had significant activity on their own. The activity at the more specific receptors PPARγ, ERα, and AR was less affected by the food content with an increase of up to 11, 12, and 18%, respectively, and a decrease of estrogenicity for the PP cups (24%, Figure S7).

We demonstrated that chemicals migrating from food into the FCAs can contribute to the receptor activity, particularly for PXR. This should be considered when testing FCAs, but it can be challenging for researchers to gain access to FCAs on the market that was not in contact with food. However, the previous food content had no significant effect on receptor activity across all extracts (Figure S8). Thus, while the food content can be a confounding factor, the general trends observed in receptor activity cannot be attributed to chemicals originating from the food.

3.3.2. Country of Purchase Does Not Influence Receptor Activity

Across the 36 FCA, the country of purchase did not significantly affect the receptor activity (Figure S9). Accordingly, differences in regulations between Germany, Norway, South Korea, the U.K., and the US concerning food contact materials or plastic production60 do not seem to impact receptor activity. These results are not surprising given the globalized nature of the manufacturing of plastic products.61 While the generalizability of our results is limited by the small sample size, these results demonstrate the global dimension of this issue.

3.3.3. Polymer Type Affects the Receptor Activity

Contrarily to the country of purchase, we observed a significant effect of the polymer type on the PPARγ and estrogenic but not on the PXR and antiandrogenic activity (Figures 4 and S10). The chemicals in LDPE and PVC induced a significantly stronger PPARγ activity than the ones in PET and PP and the estrogenic activity is significantly stronger in PS than for products made of HDPE, PET, and PVC. This indicates that these estrogenic chemicals are specific to PS products. This is consistent with previous reports on estrogenic compounds in PS migrates62 and migrating styrene mono- and oligomers may be causative.63,64 Cytotoxic chemicals are significantly more prevalent in PUR than in the other polymer types, except PVC. This might be due to the large number of chemicals present in both polymers or due to the presence of residual, toxic monomers in PUR.57,65

Figure 4.

Figure 4

Open in a new tab

Impact of the polymer type on receptor activity at (A) PPARγ, (B) ERα, and (C) cytotoxicity. EC20 calculated from at least three independent experiments with four technical replicates per concentration (n ≥ 12). Kruskal–Wallis tests with Dunn’s multiple comparison tests for statistical differences (p < 0.05) indicated by letters. Note: HTC = highest tested concentration.

Our results indicate that the polymer type can predict certain receptor activities and cytotoxicity. While we did not find specific polymers that were free of receptor activity, some polymers (LDPE, PS, PVC, PUR) contain more EDCs and MDCs than others. This means that such polymers could be prioritized for redesign or regulation.

3.3.4. Presence of Known Active Compounds in FCAs

To explore if the receptor activity can be explained by known active compounds, we compared 2146 tentatively identified chemicals with their receptor activity using ToxCast data. Here, 304 compounds detected in our samples are listed in ToxCast of which 298 were active at one or more receptors investigated in this study. However, some of these activities might be overestimated due to a cytotoxicity burst, which we did not assess here.52 These include 117 PXR agonists, 51 PPARγ agonists, 76 ERα agonists, and 69 AR antagonists. To prioritize the active compounds, we ranked the compounds based on their abundance in samples as a proxy for concentration and their potency at a receptor according to ToxCast. Within the highest-ranking chemicals, we found plastic-related chemicals, such as triphenyl phosphate (CAS 115-86-6) present in 10 samples of six different polymers, octrizole (CAS 3147-75-9) present in both PUR hydration bladders and tributyl 2-acetyloxypropane1,2,3-tricarboxylate (CAS 77-90-7) present in three samples (LDPE 2, PS 5a, comp. 2, Table S9, further information in S2.5). These results highlight that tentatively identified chemicals with known receptor activity are present in plastic FCAs from across the globe.

To further investigate whether these known active chemicals would predict the observed receptor activity, we compared the detection and bioassay results. The lack of a clear pattern between the presence of active chemicals in the samples and their respective activity (Table S10) indicates that known active chemicals tentatively identified here cannot explain the observed effects. A limitation of this approach is that it is limited to known compounds and ignores their potency and concentration, as well as mixture effects. To account for these limitations, we employed PLS regressions, including all chemical features, to explore a potential relationship between the occurrence and abundance of chemical features and the receptor activity of the samples.

3.3.5. Data Reduction to Identify Relevant Chemical Features

We used PLS regressions to handle the large chemical heterogeneity of the samples and identify features covarying with receptor activity. Through a stepwise exclusion of features, we optimized the PLS models for PXR, PPARγ, ERα, and Anti-AR (Figure 5 and Table S11). The optimization process reduced the number of features to 1533, 729, 332, and 661 features, respectively, that were identified as important contributors to the receptor activity (Table S12). Compared with the original number of all detected features (8819), this approach reduces the chemical complexity by 82–96%. The optimized models performed better than the initial models and a randomly selected set of features was used for validation. The model for PXR activity (Figure 5A,E) resulted in the lowest feature reduction (51%) and the lowest predictive power (cross-validated R2 of 0.64) suggesting that multiple compounds contribute to receptor activation in line with the promiscuous ligand-binding domain of PXR. The ERα model (Figure 5C,G) exhibited the best performance, with a pronounced improvement of the cross-validated RMSE and good predictability (cross-validated R2 of 0.9), along with a large reduction in the number of features (89% reduction). This indicates that specific chemicals are present in the active samples, significantly covarying with their estrogenicity, while these compounds are absent in the inactive samples. These results align well with the significantly stronger estrogenic activity of PS samples, pointing to the presence of specific estrogenic compounds related to the polymer PS.

Figure 5.

Figure 5

Open in a new tab

Prioritization of chemical features covarying with the receptor activity using stepwise PLS regressions. The analysis refers to food contact articles made of PE, PET, PP, and PS. Model 0 (A–D) and the optimized models (E–H) for the respective receptors. The receptor activity is represented by red diamonds, chemical features as circles, and features clustering with receptor activity in orange.

To further narrow down to covarying features that positively correlate with the receptor activity, we selected the 25% features closest to the receptor activity in the optimized model (Figure 5). This additionally reduced the number of features to 383, 95, 83, and 165 for PXR, ERα, and AR, respectively. 264 of these features were tentatively identified before, corresponding to 152 compounds, of which 26 are related to plastics4 and six were active at the respective receptor according to ToxCast data (Table S13). The latter include the four plastic-related chemicals triethylene glycol (CAS 112-27-6) clustering with PXR, tetradecanoic acid (CAS 544-63-8) clustering with PPARγ, and triphenyl phosphate (CAS 115-86-6) clustering with ERα. The active compound clustering with the antiandrogenic activity, 1-dodecyl-2-pyrrolidinone (CAS 2687-96-9), is a surface-active agent and was detected in all three LDPE freezer bags and the frozen blueberry packaging (LDPE 7). The identification of known active compounds in the variable selection process strengthens the confidence in using PLS regressions to narrow down the chemical complexity of plastics. This indicates that the known active compounds contribute to the explanation of the variance of the modeled receptor activity. In addition, other features selected in the optimized models may contribute to the activity. By identifying latent variables that explain the variability in the response, PLS can capture complex relationships within mixtures.30 However, the covarying of features with receptor activity does not confirm a cause-effect relationship and requires experimental confirmation. Further, our approach does not consider compounds that occur only in one or two samples, and the data reduction strategy based on VIP influences feature selection.31 Nevertheless, PLS regressions provide a promising approach to prioritize potentially relevant chemicals that lack identity and activity data for further research.

3.4. Implications

Our results confirm that many plastic FCAs contain EDCs and MDCs that interfere with nuclear receptors crucial to human health. The chemicals present in food packaging made of PVC, PUR, and LDPE induced most effects, whereas the extracts of HDPE, PET, and PP were less active. Nonetheless, we cannot conclude that a particular polymer type is free of toxic chemicals as methanolic extracts of samples of each polymer activated most receptors. This research highlights the importance of analyzing the toxicity of whole chemical mixtures of finished plastic products because it covers all extractable chemicals, including unknowns. Using the stepwise PLS regressions, we were able to prioritize chemical features covarying with the observed receptor activity. This represents an important step toward reducing the chemical complexity of chemicals in plastic products. At the same time, our work also highlights the limited knowledge about the compounds present in plastics. Since many relevant features remain unidentified, we recommend identifying the active compounds in these complex mixtures to enable better monitoring of human exposure and downstream effects. Moving forward, it is essential to consider chemical simplicity as a guiding principle in plastic design and production. This is supported by our findings according to which chemically less complex plastic products induced lower toxicity. By the use of fewer and better-characterized chemicals, the safety of plastic products can be significantly improved.

Acknowledgments

This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No. 860720. Special thanks to Andrea Faltynkova for her valuable help in developing the PLS regression models, to Vaibhav Budhiraja for additional differential scanning calorimetry analyses, and to Dru Jagger, Jaeho Lee, and Mara McPartland for their help with obtaining our samples. Your contributions were essential to the success of this study.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c08250.

  • Additional methodological details on polymer determination, reporter-gene assays, chemical analysis, and database search. Results on in vitro toxicity data (PXR and PPARγ activity, estrogenicity, antiandrogenicity, cytotoxicity) of reference compounds and samples, number (all and tentatively identified) of chemical features in plastic extracts, overlap of chemical features between polymers, list of tentatively identified compounds among the top ten features per samples and their functionality, tentatively identified chemicals that induce PXR and PPARγ activation, estrogenicity and antiandrogenicity based on ToxCast data, effect of food content on chemical composition and receptor activity of plastic extracts, comparison of origin of purchase and polymer type on in vitro toxicity, model parameters of partial least-squares regressions and list of tentatively identified compounds positively correlated with receptor activity (PDF)

  • Tables of all tentatively identified compounds and compounds that induce PXR and PPARγ activation, estrogenicity, and antiandrogenicity based on ToxCast data (XLSX)

Author Present Address

‡ Innovative Environmental Services (IES) Ltd., Benkenstrasse 260, 4108 Witterswill, Switzerland

Author Contributions

S.S., J.V., and M.W. conceived the study, S.S., M.M., and H.S. performed the sampling and sample preparation, S.S. performed the bioassay experiments, Z.B. and H.S. performed the chemical analyses, S.S. and H.S. analyzed the data, S.S. and M.W. wrote the manuscript, and all authors provided comments on the manuscript.

The authors declare the following competing financial interest(s): M.W. is an unremunerated member of the Scientific Advisory Board of the Food Packaging Forum Foundation and received travel support for attending annual board meetings.

Supplementary Material

es3c08250_si_001.pdf (1.4MB, pdf)
es3c08250_si_002.xlsx (690.1KB, xlsx)

References

  1. Muncke J. Exposure to Endocrine Disrupting Compounds via the Food Chain: Is Packaging a Relevant Source?. Sci. Total Environ. 2009, 407 (16), 4549–4559. 10.1016/j.scitotenv.2009.05.006. [DOI] [PubMed] [Google Scholar]
  2. Weber R.; Ashta N. M.. et al. United Nations Environment Programme and Secretariat of the Basel Rotterdam and Stockholm Conventions. In Chemicals in Plastics: A Technical Report; ETH Zurich, 2023. [Google Scholar]
  3. Aurisano N.; Weber R.; Fantke P. Enabling a Circular Economy for Chemicals in Plastics. Curr. Opin. Green Sustainable Chem. 2021, 31, 100513 10.1016/j.cogsc.2021.100513. [DOI] [Google Scholar]
  4. Wiesinger H.; Wang Z.; Hellweg S. An Extensive Overview of Plastic Monomers, Additives and Processing Aids. Environ. Sci. Technol. 2021, 55 (13), 9339–9351. 10.1021/acs.est.1c00976. [DOI] [PubMed] [Google Scholar]
  5. Hahladakis J. N.; Velis C. A.; Weber R.; Iacovidou E.; Purnell P. An Overview of Chemical Additives Present in Plastics: Migration, Release, Fate and Environmental Impact during Their Use, Disposal and Recycling. J. Hazard. Mater. 2018, 344, 179–199. 10.1016/j.jhazmat.2017.10.014. [DOI] [PubMed] [Google Scholar]
  6. Biryol D.; Nicolas C. I.; Wambaugh J.; Phillips K.; Isaacs K. High-Throughput Dietary Exposure Predictions for Chemical Migrants from Food Contact Substances for Use in Chemical Prioritization. Environ. Int. 2017, 108, 185–194. 10.1016/j.envint.2017.08.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Silva M. J.; Barr D. B.; Reidy J. A.; Malek N. A.; Hodge C. C.; Caudill S. P.; Brock J. W.; Needham L. L.; Calafat A. M. Urinary Levels of Seven Phthalate Metabolites in the U.S. Population from the National Health and Nutrition Examination Survey (NHANES) 1999–2000. Environ. Health Perspect. 2004, 112 (3), 331–338. 10.1289/ehp.6723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Calafat A. M.; Ye X.; Wong L. Y.; Reidy J. A.; Needham L. L. Exposure of the U.S. Population to Bisphenol A and 4-Tertiary-Octylphenol: 2003–2004. Environ. Health Perspect. 2008, 116 (1), 39–44. 10.1289/ehp.10753. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Zota A. R.; Calafat A. M.; Woodruff T. J. Temporal Trends in Phthalate Exposures: Findings from the National Health and Nutrition Examination Survey, 2001–2010. Environ. Health Perspect. 2014, 122 (3), 235–241. 10.1289/ehp.1306681. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Landrigan P. J.; Raps H.; Cropper M.; Bald C.; Brunner M.; Canonizado E. M.; Charles D.; Chiles T. C.; Donohue M. J.; Enck J.; Fenichel P.; Fleming L. E.; Ferrier-Pages C.; Fordham R.; Gozt A.; Griffin C.; Hahn M. E.; Haryanto B.; Hixson R.; Ianelli H.; James B. D.; Kumar P.; Laborde A.; Law K. L.; Martin K.; Mu J.; Mulders Y.; Mustapha A.; Niu J.; Pahl S.; Park Y.; Pedrotti M. L.; Pitt J. A.; Ruchirawat M.; Seewoo B. J.; Spring M.; Stegeman J. J.; Suk W.; Symeonides C.; Takada H.; Thompson R. C.; Vicini A.; Wang Z.; Whitman E.; Wirth D.; Wolff M.; Yousuf A. K.; Dunlop S. The Minderoo-Monaco Commission on Plastics and Human Health. Ann. Glob. Health 2023, 89 (1), 1–215. 10.5334/aogh.4056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Zoeller R. T.; Brown T. R.; Doan L. L.; Gore A. C.; Skakkebaek N. E.; Soto A. M.; Woodruff T. J.; Vom Saal F. S. Endocrine-Disrupting Chemicals and Public Health Protection: A Statement of Principles from The Endocrine Society. Endocrinology 2012, 153 (9), 4097–4110. 10.1210/en.2012-1422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Kahn L. G.; Philippat C.; Nakayama S. F.; Slama R.; Trasande L. Endocrine-Disrupting Chemicals: Implications for Human Health. Lancet Diabetes Endocrinol. 2020, 8 (8), 703–718. 10.1016/S2213-8587(20)30129-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Martínez-Ibarra A.; Martínez-Razo L. D.; MacDonald-Ramos K.; Morales-Pacheco M.; Vázquez-Martínez E. R.; López-López M.; Rodríguez Dorantes M.; Cerbón M. Multisystemic Alterations in Humans Induced by Bisphenol A and Phthalates: Experimental, Epidemiological and Clinical Studies Reveal the Need to Change Health Policies. Environ. Pollut. 2021, 271, 116380 10.1016/j.envpol.2020.116380. [DOI] [PubMed] [Google Scholar]
  14. Attina T. M.; Hauser R.; Sathyanarayana S.; Hunt P. A.; Bourguignon J. P.; Myers J. P.; DiGangi J.; Zoeller R. T.; Trasande L. Exposure to Endocrine-Disrupting Chemicals in the USA: A Population-Based Disease Burden and Cost Analysis. Lancet Diabetes Endocrinol. 2016, 4 (12), 996–1003. 10.1016/S2213-8587(16)30275-3. [DOI] [PubMed] [Google Scholar]
  15. Trasande L.; Liu B.; Bao W. Phthalates and Attributable Mortality: A Population-Based Longitudinal Cohort Study and Cost Analysis. Environ. Pollut. 2022, 292, 118021 10.1016/j.envpol.2021.118021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Heindel J. J.; vom Saal F. S.; Blumberg B.; Bovolin P.; Calamandrei G.; Ceresini G.; Cohn B. A.; Fabbri E.; Gioiosa L.; Kassotis C.; Legler J.; La Merrill M.; Rizzir L.; Machtinger R.; Mantovani A.; Mendez M. A.; Montanini L.; Molteni L.; Nagel S. C.; Parmigiani S.; Panzica G.; Paterlini S.; Pomatto V.; Ruzzin J.; Sartor G.; Schug T. T.; Street M. E.; Suvorov A.; Volpi R.; Zoeller R. T.; Palanza P. Parma Consensus Statement on Metabolic Disruptors. Environ. Health 2015, 14 (1), 54 10.1186/s12940-015-0042-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Heindel J. J.; Blumberg B.; Cave M.; Machtinger R.; Mantovani A.; Mendez M. A.; Nadal A.; Palanza P.; Panzica G.; Sargis R.; Vandenberg L. N.; vom Saal F. Metabolism Disrupting Chemicals and Metabolic Disorders. Reprod. Toxicol. 2017, 68, 3–33. 10.1016/j.reprotox.2016.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Heindel J. J.; Howard S.; Agay-Shay K.; Arrebola J. P.; Audouze K.; Babin P. J.; Barouki R.; Bansal A.; Blanc E.; Cave M. C.; Chatterjee S.; Chevalier N.; Choudhury M.; Collier D.; Connolly L.; Coumoul X.; Garruti G.; Gilbertson M.; Hoepner L. A.; Holloway A. C.; Howell G.; Kassotis C. D.; Kay M. K.; Ji Kim M.; Lagadic-Gossmann D.; Langouet S.; Legrand A.; Li Z.; Le Mentec H.; Lind L.; Monica Lind P.; Lustig R. H.; Martin-Chouly C.; Munic Kos V.; Podechard N.; Roepke T. A.; Sargis R. M.; Starling A.; Tomlinson C. R.; Touma C.; Vondracek J.; vom Saal F.; Blumberg B. Obesity II: Establishing Causal Links between Chemical Exposures and Obesity. Biochem. Pharmacol. 2022, 199, 115015 10.1016/j.bcp.2022.115015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kumar M.; Sarma D. K.; Shubham S.; Kumawat M.; Verma V.; Prakash A.; Tiwari R. Environmental Endocrine-Disrupting Chemical Exposure: Role in Non-Communicable Diseases. Front. Public Heal. 2020, 8, 553850 10.3389/fpubh.2020.553850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Rosen E. D.; MacDougald O. A. Adipocyte Differentiation from the inside Out. Nat. Rev. Mol. Cell Biol. 2006, 7 (12), 885–896. 10.1038/nrm2066. [DOI] [PubMed] [Google Scholar]
  21. Egusquiza R. J.; Blumberg B. Environmental Obesogens and Their Impact on Susceptibility to Obesity: New Mechanisms and Chemicals. Endocrinology 2020, 161 (3), 1–14. 10.1210/endocr/bqaa024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Wada T.; Gao J.; Xie W. PXR and CAR in Energy Metabolism. Trends Endocrinol. Metab. 2009, 20 (6), 273–279. 10.1016/j.tem.2009.03.003. [DOI] [PubMed] [Google Scholar]
  23. Lv Y.; Luo Y. Y.; Ren H. W.; Li C. J.; Xiang Z. X.; Luan Z. L. The Role of Pregnane X Receptor (PXR) in Substance Metabolism. Front. Endocrinol. 2022, 13, 1–19. 10.3389/fendo.2022.959902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lustig R. H.; Collier D.; Kassotis C.; Roepke T. A.; Ji Kim M.; Blanc E.; Barouki R.; Bansal A.; Cave M. C.; Chatterjee S.; Choudhury M.; Gilbertson M.; Lagadic-Gossmann D.; Howard S.; Lind L.; Tomlinson C. R.; Vondracek J.; Heindel J. J. Obesity I: Overview and Molecular and Biochemical Mechanisms. Biochem. Pharmacol. 2022, 199, 115012 10.1016/j.bcp.2022.115012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Biemann R.; Blüher M.; Isermann B. Exposure to Endocrine-Disrupting Compounds Such as Phthalates and Bisphenol A Is Associated with an Increased Risk for Obesity. Best Pract. Res., Clin. Endocrinol. Metab. 2021, 35 (5), 101546 10.1016/j.beem.2021.101546. [DOI] [PubMed] [Google Scholar]
  26. Muncke J.; Backhaus T.; Geueke B.; Maffini M. V.; Martin O. V.; Myers J. P.; Soto A. M.; Trasande L.; Trier X.; Scheringer M. Scientific Challenges in the Risk Assessment of Food Contact Materials. Environ. Health Perspect. 2017, 125 (9), 1–9. 10.1289/EHP644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Vinggaard A. M.; Bonefeld-Jørgensen E. C.; Jensen T. K.; Fernandez M. F.; Rosenmai A. K.; Taxvig C.; Rodriguez-Carrillo A.; Wielsøe M.; Long M.; Olea N.; Antignac J.-P.; Hamers T.; Lamoree M. Receptor-Based in Vitro Activities to Assess Human Exposure to Chemical Mixtures and Related Health Impacts. Environ. Int. 2021, 146, 106191 10.1016/j.envint.2020.106191. [DOI] [PubMed] [Google Scholar]
  28. Escher B. I.; Stapleton H. M.; Schymanski E. L. Tracking Complex Mixtures of Chemicals in Our Changing Environment. Science 2020, 392, 388–392. 10.1126/science.aay6636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Zimmermann L.; Dierkes G.; Ternes T. A.; Völker C.; Wagner M. Benchmarking the in Vitro Toxicity and Chemical Composition of Plastic Consumer Products. Environ. Sci. Technol. 2019, 53 (19), 1–40. 10.1021/acs.est.9b02293. [DOI] [PubMed] [Google Scholar]
  30. Kettaneh-Wold N. Analysis of Mixture Data with Partial Least Squares. Chemom. Intell. Lab. Syst. 1992, 14 (1–3), 57–69. 10.1016/0169-7439(92)80092-I. [DOI] [Google Scholar]
  31. Mehmood T.; Sæbø S.; Liland K. H. Comparison of Variable Selection Methods in Partial Least Squares Regression. J. Chemom. 2020, 34 (6), 1–14. 10.1002/cem.3226. [DOI] [Google Scholar]
  32. Hug C.; Sievers M.; Ottermanns R.; Hollert H.; Brack W.; Krauss M. Linking Mutagenic Activity to Micropollutant Concentrations in Wastewater Samples by Partial Least Square Regression and Subsequent Identification of Variables. Chemosphere 2015, 138, 176–182. 10.1016/j.chemosphere.2015.05.072. [DOI] [PubMed] [Google Scholar]
  33. Law K. L.; Starr N.; Siegler T. R.; Jambeck J. R.; Mallos N. J.; Leonard G. H. The United States’ Contribution of Plastic Waste to Land and Ocean. Sci. Adv. 2020, 6 (44), eabd0288 10.1126/sciadv.abd0288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Geyer R.; Jambeck J. R.; Law K. L. Production, Use, and Fate of All Plastics Ever Made. Sci. Adv. 2017, 3 (7), 25–29. 10.1126/sciadv.1700782. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Völker J.; Ashcroft F.; Vedøy Å.; Zimmermann L.; Wagner M. Adipogenic Activity of Chemicals Used in Plastic Consumer Products. Environ. Sci. Technol. 2022, 56 (4), 2487–2496. 10.1021/acs.est.1c06316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Zimmermann L.; Bartosova Z.; Braun K.; Oehlmann J.; Völker C.; Wagner M. Plastic Products Leach Chemicals That Induce In Vitro Toxicity under Realistic Use Conditions. Environ. Sci. Technol. 2021, 55 (17), 11814–11823. 10.1021/acs.est.1c01103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Schymanski E. L.; Singer H. P.; Slobodnik J.; Ipolyi I. M.; Oswald P.; Krauss M.; Schulze T.; Haglund P.; Letzel T.; Grosse S.; Thomaidis N. S.; Bletsou A.; Zwiener C.; Ibáñez M.; Portolés T.; De Boer R.; Reid M. J.; Onghena M.; Kunkel U.; Schulz W.; Guillon A.; Noyon N.; Leroy G.; Bados P.; Bogialli S.; Stipaničev D.; Rostkowski P.; Hollender J. Non-Target Screening with High-Resolution Mass Spectrometry: Critical Review Using a Collaborative Trial on Water Analysis. Anal. Bioanal. Chem. 2015, 407 (21), 6237–6255. 10.1007/s00216-015-8681-7. [DOI] [PubMed] [Google Scholar]
  38. Kucheryavskiy S. Mdatools—R Package for Chemometrics. Chemom. Intell. Lab. Syst. 2020, 198, 103937 10.1016/j.chemolab.2020.103937. [DOI] [Google Scholar]
  39. Wold S.; Sjöström M.; Eriksson L. PLS-Regression: A Basic Tool of Chemometrics. Chemom. Intell. Lab. Syst. 2001, 58 (2), 109–130. 10.1016/S0169-7439(01)00155-1. [DOI] [Google Scholar]
  40. Le Maire A.; Bourguet W.; Balaguer P. A Structural View of Nuclear Hormone Receptor: Endocrine Disruptor Interactions. Cell. Mol. Life Sci. 2010, 67 (8), 1219–1237. 10.1007/s00018-009-0249-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. di Masi A.; Marinis E. De.; Ascenzi P.; Marino M. Nuclear Receptors CAR and PXR: Molecular, Functional, and Biomedical Aspects. Mol. Aspects Med. 2009, 30 (5), 297–343. 10.1016/j.mam.2009.04.002. [DOI] [PubMed] [Google Scholar]
  42. Hakkola J.; Bernasconi C.; Coecke S.; Richert L.; Andersson T. B.; Pelkonen O. Cytochrome P450 Induction and Xeno-Sensing Receptors Pregnane X Receptor, Constitutive Androstane Receptor, Aryl Hydrocarbon Receptor and Peroxisome Proliferator-Activated Receptor α at the Crossroads of Toxicokinetics and Toxicodynamics. Basic Clin. Pharmacol. Toxicol. 2018, 123, 42–50. 10.1111/bcpt.13004. [DOI] [PubMed] [Google Scholar]
  43. Kodama S.; Negishi M. PXR Cross-Talks with Internal and External Signals in Physiological and Pathophysiological Responses. Drug Metab. Rev. 2013, 45 (3), 300–310. 10.3109/03602532.2013.795585. [DOI] [PubMed] [Google Scholar]
  44. Hakkola J.; Rysä J.; Hukkanen J. Regulation of Hepatic Energy Metabolism by the Nuclear Receptor PXR. Biochim. Biophys. Acta, Gene Regul. Mech. 2016, 1859 (9), 1072–1082. 10.1016/j.bbagrm.2016.03.012. [DOI] [PubMed] [Google Scholar]
  45. Karpale M.; Hukkanen J.; Hakkola J. Nuclear Receptor PXR in Drug-Induced Hypercholesterolemia. Cells 2022, 11 (3), 1–22. 10.3390/cells11030313. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Liu J.; Hernandez R.; Li X.; Meng Z.; Chen H.; Zhou C. Pregnane X Receptor Mediates Atherosclerosis Induced by Dicyclohexyl Phthalate in LDL Receptor-Deficient Mice. Cells 2022, 11 (7), 20–24. 10.3390/cells11071125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kirchnawy C.; Hager F.; Piniella V. O.; Jeschko M.; Washüttl M.; Mertl J.; Mathieu-Huart A.; Rousselle C. Potential Endocrine Disrupting Properties of Toys for Babies and Infants. PLoS One 2020, 15 (4), e0231171 10.1371/journal.pone.0231171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Berger E.; Potouridis T.; Haeger A.; Püttmann W.; Wagner M. Effect-Directed Identification of Endocrine Disruptors in Plastic Baby Teethers: Endocrine Disruptors in Plastic Teethers for Babies. J. Appl. Toxicol. 2015, 35 (11), 1254–1261. 10.1002/jat.3159. [DOI] [PubMed] [Google Scholar]
  49. Coffin S.; Huang G. Y.; Lee I.; Schlenk D. Fish and Seabird Gut Conditions Enhance Desorption of Estrogenic Chemicals from Commonly-Ingested Plastic Items. Environ. Sci. Technol. 2019, 53 (8), 4588–4599. 10.1021/acs.est.8b07140. [DOI] [PubMed] [Google Scholar]
  50. Bittner G. D.; Yang C. Z.; Stoner M. A. Estrogenic Chemicals Often Leach from BPA-Free Plastic Products That Are Replacements for BPA-Containing Polycarbonate Products. Environ. Health: Global Access Sci. Source 2014, 13 (1), 41 10.1186/1476-069X-13-41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yang C. Z.; Yaniger S. I.; Jordan V. C.; Klein D. J.; Bittner G. D. Most Plastic Products Release Estrogenic Chemicals: A Potential Health Problem That Can Be Solved. Environ. Health Perspect. 2011, 119 (7), 989–996. 10.1289/ehp.1003220. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Escher B. I.; Henneberger L.; König M.; Schlichting R.; Fischer F. C. Cytotoxicity Burst? Differentiating Specific from Nonspecific Effects in Tox21 in Vitro Reporter Gene Assays. Environ. Health Perspect. 2020, 128 (7), 1–10. 10.1289/EHP6664. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Klein K.; Hof D.; Dombrowski A.; Schweyen P.; Dierkes G.; Ternes T.; Schulte-Oehlmann U.; Oehlmann J. Enhanced in Vitro Toxicity of Plastic Leachates after UV Irradiation. Water Res. 2021, 199, 117203 10.1016/j.watres.2021.117203. [DOI] [PubMed] [Google Scholar]
  54. Tisler S.; Christensen J. H. Non-Target Screening for the Identification of Migrating Compounds from Reusable Plastic Bottles into Drinking Water. J. Hazard. Mater. 2022, 429, 128331 10.1016/j.jhazmat.2022.128331. [DOI] [PubMed] [Google Scholar]
  55. Sapozhnikova Y.; Nuñez A.; Johnston J. Screening of Chemicals Migrating from Plastic Food Contact Materials for Oven and Microwave Applications by Liquid and Gas Chromatography-Orbitrap Mass Spectrometry. J. Chromatogr. A 2021, 1651, 462261 10.1016/j.chroma.2021.462261. [DOI] [PubMed] [Google Scholar]
  56. Bradley E.; Coulier L.. Report FD 07/01: An Investigation into the Reaction and Breakdown Products from Starting Substances Used to Produce Food Contact Plastics; Food Standards Agency: London, 2007; pp 1–628. [Google Scholar]
  57. Lithner D.; Larsson Å.; Dave G. Environmental and Health Hazard Ranking and Assessment of Plastic Polymers Based on Chemical Composition. Sci. Total Environ. 2011, 409 (18), 3309–3324. 10.1016/j.scitotenv.2011.04.038. [DOI] [PubMed] [Google Scholar]
  58. Groh K. J.; Backhaus T.; Carney-Almroth B.; Geueke B.; Inostroza P. A.; Lennquist A.; Leslie H. A.; Maffini M.; Slunge D.; Trasande L.; Warhurst A. M.; Muncke J. Overview of Known Plastic Packaging-Associated Chemicals and Their Hazards. Sci. Total Environ. 2019, 651, 3253–3268. 10.1016/j.scitotenv.2018.10.015. [DOI] [PubMed] [Google Scholar]
  59. United States Environmental Protection Agency (US EPA). ToxCast Database (invitroDBv3), 2022. https://www.epa.gov/chemical-research/generating-toxcast-data-chemical-lists#DownloadableData.
  60. Muncke J.; Backhaus T.; Geueke B.; Maffini M. V.; Martin O. V.; Myers J. P.; Soto A. M.; Trasande L.; Trier X.; Scheringer M. Scientific Challenges in the Risk Assessment of Food Contact Materials. Environ. Health Perspect. 2017, 125 (9), 95001. 10.1289/EHP644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Wang C.; Liu Y.; Chen W. Q.; Zhu B.; Qu S.; Xu M. Critical Review of Global Plastics Stock and Flow Data. J. Ind. Ecol. 2021, 25 (5), 1300–1317. 10.1111/jiec.13125. [DOI] [Google Scholar]
  62. Kirchnawy C.; Mertl J.; Osorio V.; Hausensteiner H.; Washüttl M.; Bergmair J.; Pyerin M.; Tacker M. Detection and Identification of Oestrogen-Active Substances in Plastic Food Packaging Migrates: Detection of Oestrogen-Active Substances in Food Packaging. Packag. Technol. Sci. 2014, 27 (6), 467–478. 10.1002/pts.2047. [DOI] [Google Scholar]
  63. Ohyama K. I.; Nagai F.; Tsuchiya Y. Certain Styrene Oligomers Have Proliferative Activity on MCF-7 Human Breast Tumor Cells and Binding Affinity for Human Estrogen Receptor α. Environ. Health Perspect. 2001, 109 (7), 699–703. 10.1289/ehp.01109699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Kitamura S.; Ohmegi M.; Sanoh S.; Sugihara K.; Yoshihara S.; Fujimoto N.; Ohta S. Estrogenic Activity of Styrene Oligomers after Metabolic Activation by Rat Liver Microsomes. Environ. Health Perspect. 2003, 111 (3), 329–334. 10.1289/ehp.5723. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Úbeda S.; Aznar M.; Vera P.; Nerín C.; Henríquez L.; Taborda L.; Restrepo C. Overall and Specific Migration from Multilayer High Barrier Food Contact Materials-Kinetic Study of Cyclic Polyester Oligomers Migration. Food Addit. Contam.: Part A 2017, 34 (10), 1784–1794. 10.1080/19440049.2017.1346390. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es3c08250_si_001.pdf (1.4MB, pdf)
es3c08250_si_002.xlsx (690.1KB, xlsx)

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES