Longitudinal Metabolic Impacts of Perinatal Exposure to Phthalates and Phthalate Mixtures in Mice

Abstract

Developmental exposures to phthalates are suspected to contribute to risk of metabolic syndrome. However, findings from human studies are inconsistent, and long-term metabolic impacts of early-life phthalate and phthalate mixture exposures are not fully understood. Furthermore, most animal studies investigating metabolic impacts of developmental phthalate exposures have focused on diethylhexyl phthalate (DEHP), whereas newer phthalates, such as diisononyl phthalate (DINP), are understudied. We used a longitudinal mouse model to evaluate long-term metabolic impacts of perinatal exposures to three individual phthalates, DEHP, DINP, and dibutyl phthalate (DBP), as well as two mixtures (DEHP+DINP and DEHP+DINP+DBP). Phthalates were administered to pregnant and lactating females through phytoestrogen-free chow at the following exposure levels: 25 mg of DEHP/kg of chow, 25 mg of DBP/kg of chow, and 75 mg of DINP/kg of chow. One male and female per litter (n = 9 to 13 per sex per group) were weaned onto control chow and followed until 10 months of age. They underwent metabolic phenotyping at 2 and 8 months, and adipokines were measured in plasma collected at 10 months. Longitudinally, females perinatally exposed to DEHP only had increased body fat percentage and decreased lean mass percentage, whereas females perinatally exposed to DINP only had impaired glucose tolerance. Perinatal phthalate exposures also modified the relationship between body fat percentage and plasma adipokine levels at 10 months in females. Phthalate-exposed males did not exhibit statistically significant differences in the measured longitudinal metabolic outcomes. Surprisingly, perinatal phthalate mixture exposures were statistically significantly associated with few metabolic effects and were not associated with larger effects than single exposures, revealing complexities in metabolic effects of developmental phthalate mixture exposures.

Obesity prevalence has been on the rise in recent decades, with more than one third of US adults being affected (1). The high prevalence of obesity is a threat to public health, due in part to its numerous comorbidities, including, but not limited to, metabolic syndrome, type 2 diabetes, nonalcoholic fatty liver disease, cardiovascular disease, and some cancers (2). Poor nutrition, sedentary lifestyle, and genetic polymorphisms are well-established risk factors for obesity, but a growing body of evidence implicates a role for exposure to endocrine-disrupting chemicals (EDCs) in the obesity epidemic (3–5). EDCs that have been suggested to play a role in the development of obesity have been termed “obesogens” (5, 6). Development is considered to be a particularly sensitive period of exposure to obesogens (3, 4). The developmental origins of health and disease hypothesis postulates that environmental perturbations during critical periods of development can result in reprogramming of cells and tissues to influence susceptibility to chronic disease (7). Exposures to obesogens during development have been linked to altered basal metabolic rate, glucose metabolism, energy storage, and food intake, thereby influencing susceptibility to obesity, nonalcoholic fatty liver disease, type 2 diabetes, and dyslipidemia in adulthood (8, 9).

Phthalates are classified as obesogens and are found in a wide variety of consumer products, including plastics and personal care products. High-molecular-weight (HMW) phthalates, such as diethylhexyl phthalate (DEHP) and diisononyl phthalate (DINP), are used primarily as plasticizers and are found in items such as children’s toys, medical equipment, and vinyl flooring, whereas low-molecular-weight (LMW) phthalates, such as dibutyl phthalate (DBP), are used primarily as solvents in items such as shampoo and nail polish (10). Human exposure to phthalates is nearly unavoidable. Biomonitoring samples collected from women of reproductive age as part of the National Human and Nutrition Examination Survey identified metabolites from >13 different phthalates and detected at least 1 phthalate metabolite in all available samples (11). Developmental exposures to one of these phthalates, DEHP, has been associated with increased body fat and impaired glucose tolerance in rodents (12–17). Despite indications that human exposures to DEHP are decreasing, whereas exposure to other phthalates, such as DINP, are increasing (11), there is a lack of in vivo data regarding metabolic impacts of developmental exposures to phthalates other than DEHP. Animal studies examining metabolic health outcomes resulting from exposures to phthalates other than DEHP are needed to understand whether other phthalates are also capable of interfering with metabolic processes.

Human birth cohort studies have found inconsistent relationships between in utero phthalate exposures and obesity-related outcomes. Some studies have found positive associations between in utero phthalate exposures and body mass index and body fat in childhood (18, 19), whereas other studies have found no association or negative associations (20). One challenge in human studies that may contribute to disparate findings is that humans are coexposed to a mixture of phthalates. Previous animal studies have indicated that exposure to multiple phthalates has a dose-additive or synergistic effect on the reproductive tract (21, 22). However, metabolic impacts following developmental exposures to phthalate mixtures are not well understood. Furthermore, existing human and animal studies have only measured metabolic outcomes during infancy, childhood, or early adulthood, despite strong trends of increased metabolic syndrome risk with increased age (23, 24). Thus, more studies are needed to better understand the extent to which developmental phthalate and phthalate mixture exposures impact metabolism across the life course.

In this study, we used a longitudinal mouse model of perinatal exposure to characterize long-term metabolic impacts of perinatal exposures to phthalates and phthalate mixtures to inform human birth cohort studies and to complement the current animal literature. We included three exposure groups with individual phthalates and two exposure groups with phthalate mixtures. DEHP was chosen because it is the most widely studied phthalate and still presents high risk of exposure despite recent bans in certain products. DINP, another HMW phthalate, was chosen because exposures to DINP have been increasing in recent years (11), and it is often used as a replacement for DEHP due to its structural similarity (25). We also examined DBP as a widely studied phthalate that represents one of the highest levels of human exposures to a LMW phthalate (11). For mixtures, we examined a mixture of the two HMW phthalates (DEHP+DINP) and a mixture of all three phthalates together (DEHP+DINP+DBP). Mice were exposed from preconception to weaning and followed out to 10 months of age, far longer than other similar rodent studies (12, 13, 15). Multiple metabolic phenotyping measures, including body composition, glucose tolerance, energy expenditure (EE), and food intake, were taken at two time points, early adulthood (2 months) and later adulthood (8 months), so that effects could be assessed in a longitudinal manner. Additionally, plasma adipokines were measured at 10 months of age. From this study, we aimed to test the following hypotheses: (i) perinatal exposures to DEHP, DINP, and DBP have long-lasting impacts on metabolism, and (ii) perinatal exposures to phthalate mixtures have exaggerated effects on metabolism compared with individual exposures.

Materials and Methods

Experimental design

To evaluate metabolic effects of developmental exposures to phthalates and phthalate mixtures, we used a longitudinal mouse model (Fig. 1). Exposure to phthalates and phthalate mixtures was carried out through adding phthalates to chow, and the exposure duration spanned from periconception until weaning. Virgin females aged 6 to 8 weeks were randomized onto one of six diets 2 weeks prior to mating: (i) phytoestrogen-free 7% corn oil control (Teklad diet TD95092; Envigo), (ii) 25 mg of DEHP/kg of chow, (iii) 25 mg of DBP/kg of chow, (iv) 75 mg of DINP/kg of chow, (v) 25 mg of DEHP plus 75 mg of DINP/kg of chow, and (vi) 25 mg of DEHP plus 75 mg of DINP plus 25 mg of DBP/kg of chow. Details regarding exposure level selection are provided below, as well as in Neier et al. (26). F₀ females remained on their assigned chow throughout gestation, birth, and lactation. At postnatal day (PND) 21, one male and one female F₁ offspring per litter were weaned onto control chow and followed to 10 months of age (n = 9 to 13 per sex per group). F₁ offspring underwent metabolic phenotyping at two time points across the life course: 2 months and 8 months. At 10 months, F₁ offspring were euthanized; tissues were harvested and blood and plasma were collected via cardiac puncture.

Figure 1.

Experimental design. Two weeks prior to mating, virgin a/a female mice (F₀) were randomly assigned to one of six exposure groups containing different combinations of phthalates. Phthalates were administered through chow on a background diet of 7% corn oil (AIN-93G, phytoestrogen-free). Exposure spanned preconception, gestation, and lactation, and at PND21, one male and one female F₁ offspring per litter were weaned onto control chow and followed until 10 mo of age (n = 9 to 13 per sex per exposure group). Metabolic phenotyping was carried out at 2 and 8 mo, and plasma adipokines were measured at 10 mo.

Animals

For this study, we used a/a mice from a genetically invariant colony of viable yellow agouti (A^vy) mice maintained by sibling mating and forced heterozygosity through the male line for >220 generations, resulting in an isogenic background 93% identical to the C57BL/6J strain (27, 28). A^vy/a mice display a spectrum of metabolic phenotypes based on epigenetic marks at the A^vy locus and expression of the A^vy allele and have an increased susceptibility to obesity and tumorigenesis (27, 29–31), and thus only a/a offspring underwent metabolic phenotyping.

A total of 108 mate pairs were set up from which a total of 98 litters were generated. Mean litter size was 6.57 pups per litter and was not significantly impacted by exposure (26). The largest a/a male and largest a/a female from each litter were selected for follow-up systematically in attempt to ensure high survival rates and to reduce chances of complications during metabolic phenotyping at 2 months. We selected one mouse per sex per litter for follow-up to avoid controlling for within-litter effects in statistical analyses. Litters were generated until a minimum of 10 male a/a offspring and 10 female a/a offspring were available for follow-up; however, the DBP exposure group had only nine females due to difficulty in generating a/a females, and one DINP male died after oral gavage during oral glucose tolerance tests (OGTTs). A sample size of 10 per sex per exposure group was selected for this study based on previous work carried out successfully using similar longitudinal mouse models examining perinatal exposures to EDCs (32). At weaning (PND21), the number of offspring in each group was as follows for females and males, respectively: control, n = 13, n = 12; DEHP, n = 11, n = 11; DBP, n = 9, n = 11; DINP, n = 12, n = 10; DEHP+DINP, n = 13, n = 12; DEHP+DINP+DBP, n = 12, n = 11; total, n = 70, n = 67. The experimental group was kept blinded to all laboratory personnel for the duration of the study.

All animal procedures were carried out in an Association for Assessment and Accreditation of Laboratory Animal Care International–accredited facility under approval by the University of Michigan Institutional Animal Care and Use Committee and in compliance with the Guide for the Care and Use of Laboratory Animals. Mice were housed in two different facilities throughout the course of the study: their primary housing location was at the University of Michigan School of Public Health, and their secondary location was at the University of Michigan Animal Phenotyping Core (APC). Mice were transported to the APC at 2 months and at 8 months of age and remained at the APC for ∼4 weeks each time. They were given 1 week of adjustment period immediately after transport prior to any metabolic testing and were given 1 week after metabolic testing before transport back to the School of Public Health. Mice were housed in polycarbonate-free static cages with corn cob bedding. All mice were housed with one nonstudy cage mate throughout the duration of the study, with the exception of 4 days at 2 months of age and 4 days at 8 months of age when animals were undergoing Comprehensive Laboratory Animal Monitoring System (CLAMS) measurements. Males were given extra enrichment (Envirodry) to prevent fighting. All mice were provided food and water ad libitum throughout the duration of the study and were kept on a 12-hour light/12-hour dark cycle at 21°C. Laboratory personnel and husbandry staff monitored animals on a daily basis and reported health conditions to veterinary staff to ensure high quality of life.

Exposures

Virgin a/a females aged 6 to 8 weeks were randomly assigned to one of six experimental groups: (i) phytoestrogen-free 7% corn oil control (Teklad diet TD95092; Envigo), (ii) 25 mg of DEHP/kg of chow, (iii) 25 mg of DBP/kg of chow, (iv) 75 mg of DINP/kg of chow, (v) 25 mg of DEHP plus 75 mg of DINP/kg of chow, and (vi) 25 mg of DEHP plus 75 mg of DINP plus 25 mg of DBP/kg of chow. To generate phthalate-containing chow, phthalates were mixed into the corn oil that was used to make the chow. Phthalate-containing chows were also comprised of 7% corn oil and were the same as the control chow in nutritional values. Phytoestrogen-free chow was chosen as the control and background diet to minimize interference of phytoestrogens with molecular pathways that may be impacted by phthalate exposures. Ingestion through chow was chosen as a means of exposure because ingestion is a major route of phthalate exposures (10, 33), to mimic the gradual exposures that humans experience throughout the day as opposed to one bolus daily dose, and to minimize stress to pregnant mice that can incur during other exposure methods (e.g., oral gavage) (34). DEHP and DBP were added to chow at 25 mg of phthalate/kg of chow, and DINP was added at 75 mg of phthalate/kg of chow, resulting in an estimated dose of 5 mg/kg-day and 15 mg/kg-day, respectively (34). These estimated doses were based on assumptions used in our previous exposure studies (28) that average consumption rates were 5 g of chow per day for a 25-g pregnant female mouse. These perinatal exposure levels were chosen based on levels anticipated to result in effects indicative of metabolic syndrome in offspring observed in previous rodent studies of DEHP and DBP (13, 15) and based on published literature, indicating that DINP is approximately threefold less potent than DEHP and DBP with respect to antiandrogenic effects (22). Furthermore, the exposure levels selected are human relevant based on extrapolations between rodent and human data of phthalate exposures (26). Previous rodent studies found mean levels of DEHP metabolites of 68 ng/mL in amniotic fluid of pregnant rats that were exposed to 11 mg/kg-day of DEHP (35), a dose similar to the phthalate doses used in this study. A variety of human studies that measured DEHP metabolites in amniotic fluid found that the median levels ranged from 1.6 to 22.1 ng/mL, with a maximum level of 100 ng/mL (36–40). Thus, the exposure levels used in this study likely result in human-relevant fetal exposure levels, albeit in the higher range.

Metabolic phenotyping

Body weight and composition

Body weights were measured in F₁ offspring on a weekly basis between weaning (PND21) and 10 months of age. Body composition was measured at ages 2 months, 8 months, and 10 months. At 2 months and 8 months, body fat, lean mass, and free fluid were measured at the University of Michigan APC via a nuclear magnetic resonance–based analyzer (Minispec LF9011; Bruker Optics). For one subset of mice at 2 months of age (n = 22, spread across all exposure groups), the instrument malfunctioned and body composition measurements were not recorded, but data were recorded for all remaining mice (n = 114, spread across all exposure groups). Body fat mass and lean mass were measured at 10 months via magnetic resonance imaging in the morning prior to euthanization (EchoMRI). Body fat, lean mass, and free fluid measured via nuclear magnetic resonance at 2 months and 8 months were used for longitudinal phenotypic analyses, whereas the EchoMRI measurements taken at 10 months were used primarily as covariates for analyzing plasma adipokine levels. Body fat, lean mass, and free fluid measures were reported as percentage of body weight, which was recorded immediately prior to body composition analysis.

CLAMS

Respiratory exchange rate (RER), EE, glucose oxidation, fat oxidation, spontaneous activity, and food intake were evaluated at 2 months and 8 months of age with CLAMS (Columbus Instruments) at the APC. Mice were placed into a CLAMS chamber alone for 4 days at each time point (2 months and 8 months). The first 24 hours was designated as an acclimation period, and therefore the first day was removed from subsequent analyses comparing experimental groups. While in the CLAMS chamber, physical activity was measured as number of times the mouse moves through laser fields, oxygen consumption (Vo₂) and carbon dioxide production (Vco₂) were measured using indirect calorimetry, and food intake was measured by weight. Vco₂ and Vo₂ were measured once every 20 minutes. RER was calculated as Vco₂/Vo₂, and EE, fat oxidation, and glucose oxidation were calculated from Vco₂ and Vo₂ based on previously published methods (41). Resting metabolic rate (RMR) was calculated for each mouse at 2 months and 8 months of age using a penalized spline model incorporating both EE and ambulatory physical activity as described in Van Klinken et al. (42) and Kochmanski et al. (43).

Glucose tolerance testing

OGTTs were carried out at 2 months and 8 months to evaluate glucose tolerance in F₁ offspring. Mice were fasted for 5 hours prior to testing, and then administered an oral dose of glucose (2.0 g of glucose/kg of body weight) via oral gavage. Blood glucose was measured at T₀ (prior to glucose administration) and 15, 30, 60, and 120 minutes after glucose challenge via blood collected from the tail vein. Glucose was measured with an AccuChek (Roche) glucometer.

Plasma adipokines

Adipokines were measured in plasma collected from F₁ offspring at 10 months. Blood was collected via cardiac puncture after CO₂ euthanasia and immediately placed in EDTA-containing microcentrifuge tubes. Blood was centrifuged at 2000g for 10 minutes to separate the plasma; plasma was then stored at −80°C. Leptin, MCP-1, PAI-1, resistin, IL-6 and TNF-α were measured in plasma using the Milliplex MAP mouse adipokine magnetic bead panel (Sigma-Aldrich) (44), a multiplex immunoassay that was carried out on a Luminex xMAP. Each plasma sample was run in duplicate, per the manufacturer’s instructions. Plasma adiponectin was measured separately via ELISA (Sigma-Aldrich) (45), with each sample run in duplicate. Three samples did not have enough sample volume left to analyze plasma adiponectin levels.

Statistical analysis

All statistical analyses were carried out using R version 3.5.0 (www.R-project.org). As EDCs, phthalates are expected to have sex-specific effects, so all analyses were stratified by sex. We first carried out cross-sectional analyses on data obtained at 2, 8, and 10 months via multiple linear regression (MLR) comparing each exposure group to controls. However, the primary goal of this study was to assess longitudinal metabolic impacts of perinatal phthalate exposures, and many of the metabolic measures were repeated at two time points (2 and 8 months). Thus, we used linear mixed effects (LME) models to assess longitudinal effects and to account for temporal dependence. To determine whether exposure significantly modified the effect of age, we carried out an additional LME model with an interaction term for exposure × age. To test the hypothesis that mice perinatally exposed to phthalate mixtures exhibit exaggerated effects longitudinally compared with mice exposed to individual phthalates, we used LME models with a simple order constraint on the exposure combinations, the so-called constrained linear mixed effects implemented in the R package CLME. We employed three models to make the following sets of comparisons, with each model containing two comparisons: (i) control vs DEHP and DEHP vs DEHP+DINP, (ii) control vs DINP and DINP vs DEHP+DINP, and (iii) control vs DBP and DBP vs DEHP+DINP+DBP. Additionally, intermouse variability in certain measures was compared at 2 months of age vs 8 months of age via an F test.

A portion of F₁ offspring had gross liver masses at dissection (46). Models that included data obtained at 8 or 10 months included a dichotomous variable to control for gross liver masses [0, no mass present; 1, mass(es) present], because the presence of gross liver masses significantly impacted several different outcomes. In evaluation of EE, fat and lean mass were also controlled for in each model, because fat and lean mass are both known to significantly impact EE. For glucose and fat oxidation, we examined models both with and without controlling for fat and lean mass. With respect to plasma adipokine levels, interaction terms were added to MLR models to test whether perinatal exposures to phthalates and phthalate mixtures modify the relationship between plasma adipokine levels and body fat percentage. Plasma IL-6 and TNF-α had a large portion of nondetects (17% and 44.3%, respectively). Thus, the Kolmogorov–Smirnov test for comparing empirical distribution functions of IL-6 and TNF-α for each group to controls was carried out.

In discussion of our findings, we considered differences with P ≤ 0.10 to be marginally statistically significant, and differences with P ≤ 0.05 to be statistically significant. Because this study included multiple groups, we also set more stringent P value cutoffs based on the means of familywise type I error control to account for multiple comparisons. For cross-sectional MLR and longitudinal LME analyses, which compared each of the five groups to controls, differences with P ≤ 0.02 were considered marginally significant and differences with P ≤ 0.01 were considered significant when taking multiple comparisons into account. For LME models with ordered constraints we made two comparisons per model, and thus differences with P ≤ 0.05 were considered marginally significant and differences with P ≤ 0.025 were considered significant.

Results

Litter parameters and life course morbidity and mortality

A total of 108 mate pairs were set up to generate mice for this study, of which 98 produced litters (90.7%) (46). The number of litters generated per exposure group ranged from 15 to 19. The DINP exposure group had a significantly lower birth rate, with 15 of 21 mating pairings resulting in pups (71.4%) compared with 17 of 17 mate pairings resulting in pups (100%) for controls (P = 0.02). Of the 137 total F₁ offspring that were followed until 10 months of age, 7 died due to complications with metabolic testing or fight wounds, and 1 died of unknown causes at 2 months of age. Upon necropsy at 10 months of age, 22 offspring had observable gross liver masses (46). One female and one male control offspring had liver masses, representing a background liver mass rate of 8.3% and 9.1%, respectively. Males perinatally exposed to DINP and DEHP+DINP had the highest rate of liver masses (33.3% for both), although this was not significantly higher than the background rate (P > 0.10).

Body weight and body composition

Longitudinally, there were no significant differences in body weight across phthalate exposure groups in females or males across the life course (P > 0.10; Fig. 2). Despite a lack of statistical significance, females exposed to individual phthalates exhibited trends toward increased body weight longitudinally. Furthermore, perinatal exposure to phthalates modified the relationship between age and body weight in females. Females perinatally exposed to DEHP only and DBP only gained more weight as they aged than did controls (interaction P = 0.006 and P < 0.0001, respectively; Fig. 2). Interestingly, there was increased variability in body weights of control and exposed females as they aged. F tests comparing the variance in body weights at 2 months vs variances in body weights at 8 months were statistically significant for control females (P = 0.0001), DEHP females (P = 0.0004), DINP females (P = 0.0002), DBP females (P = 0.0001), DEHP+DINP females (P < 0.00001), and DEHP+DINP+DBP females (P < 0.00001).

Figure 2.

Body weights across time. Weekly body weights were recorded in (A) female and (B) male offspring (n = 9 to 13 per sex per exposure group). No significant differences in body were detected via longitudinal LME models. However, females perinatally exposed to DEHP only, DBP only, and DINP only gained more weight with age than did controls (P = 0.006, P < 0.0001, and P = 0.09, respectively).

Females, but not males, perinatally exposed to individual phthalates had altered body composition in adulthood (Table 1). Longitudinal analyses indicated that females perinatally exposed to DEHP only had increased body fat percentage (P = 0.01) and decreased lean mass (P = 0.01) compared with controls. Females perinatally exposed to DINP only had increased body fat percentage (P = 0.05) and decreased lean mass percentage (P = 0.03) in comparison with controls, but these differences were not significant after accounting for multiple comparisons (P ≤ 0.01 for significance and P ≤ 0.02 for modest significance). (Table 1). On average, females in the DEHP exposure group displayed an increase of 3.63% body fat and a decrease of 2.98% lean mass relative to controls, and females in the DINP exposure group had an increase of 2.82% body fat and a decrease of 2.47% lean mass relative to controls. Cross-sectional analyses indicated similar trends in females perinatally exposed to DEHP only and DINP only at 2 and 8 months, but these trends were not statistically significant when accounting for multiple comparisons (P > 0.02) (46). Notably, females perinatally exposed to mixtures of phthalates did not exhibit altered body composition in adulthood (P > 0.10).

Table 1.

Longitudinal Impacts of Perinatal Phthalate Exposures on Body Composition

Exposure Group	Body Fat Percentage						Lean Mass Percentage
	2 mo		8 mo		Longitudinal Analysis		2 mo		8 mo		Longitudinal Analysis
	n	Mean ± SE	n	Mean ± SE	β	P Value	n	Mean ± SE	n	Mean ± SE	β	P Value
Females
Control	11	13.05 ± 0.62	13	32.94 ± 1.92	Ref.	Ref.	11	67.98 ± 0.48	13	52.76 ± 1.67	Ref.	Ref.
DEHP	9	15.14 ± 0.81	11	36.52 ± 1.24	3.63	δ 0.01a	9	66.53 ± 0.47	11	49.66 ± 1.06	−2.98	δ 0.01a
DBP	8	14.58 ± 0.95	9	37.09 ± 1.70	2.62	0.08b	8	67.17 ± 0.63	9	49.26 ± 1.37	−1.96	0.09b
DINP	10	15.10 ± 0.80	12	34.87 ± 1.57	2.82	0.05c	10	66.20 ± 0.78	12	51.06 ± 1.34	−2.47	0.03c
DEHP+DINP	11	13.61 ± 0.61	13	34.99 ± 1.93	1.68	0.21	11	67.77 ± 0.59	13	50.95 ± 1.58	−1.36	0.20
DEHP+DINP+DBP	10	14.43 ± 0.55	12	32.00 ± 1.90	0.76	0.58	10	67.55 ± 0.49	12	53.41 ± 1.55	−0.04	0.74
Males
Control	10	12.18 ± 0.61	12	27.39 ± 0.95	Ref.	Ref.	10	69.37 ± 0.48	12	57.10 ± 0.81	Ref.	Ref.
DEHP	8	13.51 ± 1.40	10	29.39 ± 1.07	1.62	0.12	8	69.15 ± 1.14	10	55.28 ± 0.87	−1.06	0.23
DBP	9	12.53 ± 0.71	10	27.34 ± 1.00	−0.19	0.86	9	69.70 ± 0.65	10	56.81 ± 0.88	0.26	0.77
DINP	9	13.68 ± 0.95	9	26.77 ± 0.85	0.51	0.63	9	68.19 ± 0.90	9	57.58 ± 0.78	−0.47	0.60
DEHP+DINP	10	12.82 ± 0.57	12	26.56 ± 1.26	0.006	1.00	10	69.29 ± 0.52	12	57.72 ± 1.15	0.1	0.91
DEHP+DINP+DBP	9	12.56 ± 0.54	11	27.79 ± 0.80	0.17	0.87	9	69.26 ± 0.66	11	56.59 ± 0.70	−0.11	0.91

β Coefficients and P values are from LME models of longitudinal metabolic parameters comparing each exposure group to controls. Body composition data were obtained at 2 and 8 mo of age. Models controlled for the presence of gross liver masses at necropsy. Body fat percentage and lean mass percentage are expressed as percentage of body weight. Each exposure group was compared with controls, resulting in five comparisons per model; to account for multiple comparisons, P value cutoffs of 0.05 for significant and 0.10 for borderline significant are adjusted to 0.01 and 0.02, respectively. Comparisons that remained statistically significant after accounting for multiple comparisons are denoted with δ.

^a

P ≤ 0.01, compared with controls.

^b

P ≤ 0.10, compared with controls.

^c

P ≤ 0.05, compared with controls.

Food intake and physical activity

Mice perinatally exposed to phthalates and phthalate mixtures did not exhibit notable differences in physical activity or food intake compared with controls longitudinally or cross-sectionally (P > 0.05) (46). Thus, differences in body composition observed in females exposed to phthalates were more likely due to alterations in intrinsic metabolic processes and not due to differences in energy intake or EE due to physical activity.

EE and resting metabolic rate

Longitudinally, there was no indication of statistically significant differences in EE rates in females or males perinatally exposed to phthalates and/or phthalate mixtures (P > 0.05) (46). Cross-sectionally at 2 months of age, females exposed to DEHP only had average EE rates of 0.378 kcal/h and controls had average EE rates of 0.359 kcal/hr (P = 0.03) (46), but this difference was not significant when accounting for multiple comparisons. There were no significant differences in EE rates across exposure groups in females cross-sectionally at 8 months of age. Males perinatally exposed to DEHP+DINP had a mean EE rate of 0.431 kcal/h during the light cycle at 2 months of age compared with control males who had a mean EE rate of 0.390 kcal/h at 2 months of age; however, this difference was not statistically significant when accounting for multiple comparisons (P = 0.05) (46). Cross-sectional analyses at 8 months of age indicated a similar, nonsignificant trend in light cycle EE rates of DEHP+DINP males (P = 0.06) (46).

Females perinatally exposed to phthalates did not exhibit significant differences in RMR longitudinally. However, at 2 months of age, the mean RMR in DEHP females was 0.318 kcal/h, which was higher than the mean RMR of 0.295 kcal/h observed in control females to a modest degree of statistical significance (P = 0.02; Fig. 3A and 3B). Alternatively, males perinatally exposed to DEHP+DINP trended toward having a higher mean RMR than did control males longitudinally (P = 0.03; Fig. 3D) and cross-sectionally at 2 months (P = 0.04; Fig. 3C), but these differences were not statistically significant after adjustment for multiple comparisons.

Figure 3.

RMRs in mice perinatally exposed to phthalates. Cross-sectional analyses of (A) females and (C) males comparing RMRs across exposure groups at 2 and 8 mo of age were carried out via MLR (n = 8 to 13 per sex per exposure group). Models for 8-mo data included a variable to control for the presence of gross liver abnormalities. Longitudinal analyses of (B) females and (D) males to examine the effects of perinatal phthalate exposures on RMR were carried out via LME models. RMR was calculated based on spontaneous activity and indirect calorimetry. Bars represent mean RMR for each group, and error bars represent ± SE. Lines track from mean RMR at 2 mo to the mean RMR at 8 mo for each group.

Respiratory exchange rate, fat oxidation rate, and glucose oxidation rate

Mice perinatally exposed to phthalates did not exhibit a statistically significant altered RER or fat oxidation rate when accounting for multiple comparisons. However, females, but not males, perinatally exposed to DEHP only, DINP only, and DEHP+DINP trended toward decreased mean RER compared with controls longitudinally across 2 and 8 months of age (P = 0.07, P = 0.08, and P = 0.05, respectively) (46). Cross-sectional analyses also indicated that there were trends toward lower RERs at 2 months in females perinatally exposed to DEHP+DINP (dark cycle P = 0.03) and at 8 months in females perinatally exposed to DEHP (light cycle P = 0.03, average of light and dark cycle P = 0.05) (46).

In concordance with decreased RERs, females perinatally exposed to only DEHP, only DINP, and a mixture of DEHP and DINP demonstrated trends toward increased fat oxidation rates longitudinally across 2 and 8 months of age by 5.17, 4.34, and 4.72 mg/h compared with controls, respectively (P = 0.05, P = 0.09, and P = 0.05, respectively) (46), but these differences were not statistically significant after adjusting for multiple comparisons (P ≤ 0.01). Similar trends were also evident in cross-sectional analyses, but none of the differences reached statistical significance at P ≤ 0.01 (46). Lastly, males did not exhibit significant differences in fat oxidation rates either longitudinally or cross-sectionally (46).

Perinatal phthalate exposure was not associated with longitudinal changes in glucose oxidation rates in females (46). Males perinatally exposed to DEHP+DINP exhibited a trend toward increased glucose oxidation rates longitudinally, with an average increase of 14.87 mg/h compared with controls (P = 0.03), but this difference was not significant when accounting for multiple comparisons, and cross-sectional analyses demonstrated only nonsignificant trends (46).

Glucose tolerance

Females perinatally exposed to DINP only exhibited modestly impaired glucose tolerance longitudinally. Across 2 and 8 months of age, females exposed to DINP only had a mean increase in glucose area under the curve (AUC) of 1404.3 compared with controls (P = 0.02; Fig. 4C). Cross-sectional analyses revealed that differences in blood glucose levels during OGTTs in females exposed to DINP only were more pronounced at 2 months of age than at 8 months of age. At 2 months, females exposed to DINP had modestly increased blood glucose levels at 60 minutes postgavage, and they had increased glucose AUC, compared with controls (P = 0.02, for both; Fig. 4A). At 8 months, alternatively, there were no significant differences in blood glucose levels following OGTTs across groups in females (Fig. 4B). Of note, DEHP+DINP+DBP females also had a trend toward higher blood glucose levels at 60 minutes postgavage in comparison with controls at 2 months (P = 0.03; Fig. 4A), but this difference was not statistically significant after adjustment for multiple comparisons. Surprisingly, there was a trend of decreased glucose AUC with age in females perinatally exposed to DEHP+DINP compared with controls (P = 0.04; Fig. 4C), suggesting that their glucose tolerance improved with age. Furthermore, at 8 months, females perinatally exposed to DEHP+DINP had a trend toward decreased blood glucose levels at 120 minutes postgavage compared with controls (P = 0.04; Fig. 4B). However, these improvements in glucose tolerance with age were not statistically significant when accounting for multiple comparisons.

Figure 4.

Oral glucose tolerance testing. OGTTs were carried out on (A–C) females and (D–F) males at 2 and 8 mo (n = 8 to 13 per sex per exposure group). Lines in (A), (B), (D), and (E) represent group means of blood glucose levels during the testing period (0 to 120 min after glucose challenge). Cross-sectional analyses were carried out via MLR for each time point (0, 15, 30, 60, and 120 min) after glucose challenge, and were also carried out for AUC, a measurement of glucose tolerance. Mean AUC was plotted across the 2 and 8 mo ages for (C) females and (D) males to examine longitudinal effects of exposure via LME models and also to determine whether phthalate exposures modified the relationship with glucose tolerance and age (INTRXN). *P ≤ 0.02.

Although males perinatally exposed to phthalates and phthalate mixtures did not exhibit significant differences in blood glucose measures longitudinally, males perinatally exposed to DINP only had a modest decline in glucose AUC with age in comparison with controls (exposure × age interaction P = 0.02; Fig. 4F). Similar to females, males perinatally exposed to DEHP+DINP also exhibited a trend toward decreased glucose AUC with age (exposure × age interaction P = 0.04), but this difference was not statistically significant after adjustment for multiple comparisons. Cross-sectionally, there were no differences across exposure groups in glucose levels at any time in 2-month-old males (Fig. 4D), but males perinatally exposed to DBP only exhibited a modestly decreased blood glucose levels at 120 minutes postgavage when compared with controls at 8 months of age (P = 0.02; Fig. 4E).

Plasma adipokines

Plasma adipokines did not significantly differ across exposure groups in males or females (46). TNF-α and IL-6 plasma levels were too low to detect in many of the samples analyzed for this study; 46.1% of the samples did not have detectable TNF-α levels and 22% did not have detectable IL-6 samples (46). Although there were no statistically significant differences by exposure group for TNF-α plasma concentrations, it was notable that none of the plasma samples from DINP males had detectable levels of TNF-α.

Because adipokines are produced by adipocytes, we next examined whether perinatal phthalate exposures modified the relationship between body fat mass and adipokine levels by examining interactions between exposures and body fat percentage via MLR, controlling for the presence of gross liver masses. Our findings indicated that perinatal exposures modified the relationship between body fat percentage and 10-month plasma resistin and MCP-1 levels in females (Fig. 5). MLR analyses indicated that perinatal exposure to DEHP only modified the relationship between body fat percentage and plasma resistin levels in females at 10 months in a positive manner to a modest degree of statistical significance after adjusting for multiple comparisons (P = 0.02; Fig. 5A). Females perinatally exposed to DEHP+DINP and DEHP+DINP+DBP also trended toward positively modifying the relationship between body fat percentage and plasma resistin levels (P = 0.08 and P = 0.04, respectively; Fig. 5A). Perinatal phthalate exposure more strongly modified the relationship between plasma MCP-1 levels and body fat percentage in females at 10 months; DBP only and DEHP+DINP had significant negative interactions between body fat percentage and plasma MCP-1 levels (P = 0.006 and P = 0.0009, respectively; Fig. 5B). Perinatal exposure to DEHP only also modified the relationship between body fat percentage and plasma MCP-1 levels in females to a modest degree of statistical significance in the negative direction (P = 0.02), and perinatal exposure to DEHP+DINP+DBP trended toward a negative modification of body fat percentage and plasma MCP-1 levels, although the effect modification was not statistically significant after accounting for multiple comparisons (P = 0.03; Fig. 5B). Thus, plasma MCP-1 levels in females perinatally exposed to phthalates increased less with increasing body fat percentage than in controls, potentially indicating that their adipocytes produced less MCP-1.

Figure 5.

Modification of relationship between plasma adipokine levels and body fat percentage at 10 mo. The relationship between body fat percentage and plasma levels of (A) resistin and (B) MCP-1 in females (n = 9 to 13 per exposure group) is shown. Dots represent individual female mice; lines represent a least squares regression curve of the relationship between body fat percentage and plasma adipokine levels for each exposure group. MLR models with interaction terms for exposure × body fat percentage were carried out to examine the effect modification of perinatal phthalate exposure. Analyses controlled for the presence of gross liver masses at necropsy.

Longitudinal mixture effects

One of the primary objectives of this study was to characterize long-term metabolic impacts of perinatal exposure to phthalate mixtures and to understand whether perinatal exposures to mixtures of phthalates had exaggerated effects compared with perinatal exposures to individual phthalates. Thus, we used an LME model with a simple order constraint to model longitudinal metabolic parameters and test whether perinatal exposure to phthalate mixtures had an exaggerated metabolic response vs individual phthalates. LME models demonstrated longitudinal impacts of perinatal exposure to either individual phthalates or phthalate mixtures on body fat percentage, lean mass percentage, and glucose AUC in females (Table 1; Fig. 4). There were also trends toward exposure-related effects on RER and fat oxidation in females (0.02 < P < 0.05) and RMR and glucose oxidation in males (0.02 < P < 0.05) that were further explored with simple order restraint models. These models were used to test the following comparisons to test the overall hypothesis that perinatal exposures to phthalate mixtures would have exaggerated effects compared with perinatal exposures to individual phthalates: (i) control vs DEHP and DEHP vs DEHP+DINP, (ii) control vs DINP and DINP vs DEHP+DINP, and (iii) control vs DBP and DBP vs DEHP+DINP+DBP.

Perinatal exposures to DEHP+DINP and DEHP+DINP+DBP did not exert exaggerated effects on body fat percentage, lean mass percentage, RER, fat oxidation, or glucose tolerance. Overall P values calculated via the bootstrap likelihood ratio test were statistically significant for body fat percentage (P = 0.028, P = 0.020) and lean mass percentage (P = 0.018, P = 0.017), for comparisons of control females to DEHP females to DEHP+DINP females, and for comparisons of control females to DINP females to DEHP+DINP females (Table 2). Overall P values were also significant for RER (P = 0.005) and fat oxidation (P = 0.003) when comparing control females to DEHP females to DEHP+DINP females, and they trended toward significance for control females vs DINP females vs DEHP+DINP females (P = 0.055 and P = 0.057, respectively). However, these significant overall P values were driven by the differences between control females and females that were exposed to individual phthalates, namely DEHP and DINP. Individual comparisons indicated that females perinatally exposed to a mixture of DEHP and DINP did not have exaggerated effects compared with females exposed to DEHP and DINP alone (P > 0.35 in all instances; Table 2). Notably, the longitudinal models did not reveal any significant differences in glucose tolerance between exposure groups in females (Table 2). Additionally, overall model P values and individual comparisons between control females, DBP females, and DEHP+DINP+DBP females were not statistically significant for any of the metabolic outcomes evaluated, which was consistent with previous longitudinal LME models (Table 2).

Table 2.

Linear Mixed Effects Models with Simple Order Constraints Evaluating Effects of Individual Phthalates Compared With Phthalate Mixtures on Longitudinal Metabolic Outcomes

Outcome	Model 1: Control vs DEHP vs DEHP+DINP					Model 2: Control vs DINP vs DEHP+DINP					Model 3: Control vs DBP vs DEHP+DINP+DBP
	Overall P Value	Control vs DEHP		DEHP vs DEHP+DINP		Overall P Value	Control vs DINP		DINP vs DEHP+DINP		Overall P value	Control vs DBP		DBP vs DEHP+DINP+DBP
		Estimate	P Value	Estimate	P Value		Estimate	P Value	Estimate	P Value		Estimate	P Value	Estimate	P Value
Females
Body fat, %	0.028a	2.53	δ 0.014a	0	1	0.020a	2.33	δ 0.014a	0	1	0.21	1.61	0.09b	0	1
Lean mass, %	0.018a	−2.04	δ 0.018a	0	1	0.017a	−2	δ 0.014a	0	1	0.23	−1.08	0.11	0	1
RER, Vco2/Vo2	0.005c	−0.03	δ 0.001c	0	1	0.055b	−0.026	δ 0.024a	−0.002	0.37	0.33	−0.011	0.18	0	1
Fat oxidation, mg/h	0.003c	5.24	δ 0.001c	0	1	0.057b	4.31	δ 0.023a	0.15	0.35	0.37	1.6	0.2	0	1
Glucose AUC, mg/dL	0.27	422.2	0.16	0	1	0.15	697.2	0.07b	0	1	0.2	582.7	0.09b	0	1
Males
RMR, kcal/h	0.005c	0	1	0.054	δ 0.004c	0.035a	0.01	0.22	0.035	δ 0.016a	1	0	1	0	1
Glucose oxidation, mg/h	0.023a	0	1	17	δ 0.003c	0.057b	2.95	0.24	10.71	ψ 0.035a	0.65	0.15	0.39	0	1

LME models with simple order restraints were used to assess whether mice perinatally exposed to phthalate mixtures had greater effects compared with those exposed to individual phthalates. Overall P value is the global P value of the model calculated via bootstrap likelihood ratio test based on 1000 bootstraps. Individual P values are Williams-type tests for each individual comparison. Individual tests had two comparisons for each model, and thus a P value of 0.025 (0.05/2) was used as a cutoff for statistical significance after accounting for multiple comparisons, and a P value of 0.05 (0.10/2) was used as a cutoff for borderline statistical significance. Comparisons that remained statistically significant after accounting for multiple comparisons are denoted with δ, and those that are still marginally statistically significant are denoted with ψ.

^a

P ≤ 0.05.

^b

P ≤ 0.10.

^c

P ≤ 0.01.

Males perinatally exposed to a mixture of DEHP and DINP had significantly increased RMR and glucose oxidation in comparison with those exposed to DEHP or DINP alone (Table 2). Across 2 and 8 months of age, DEHP+DINP males had an increased RMR by 0.054 kcal/h vs DEHP-only males (P = 0.004) and had an increased RMR of 0.035 kcal/h vs DINP-only males (P = 0.016). Similarly, males perinatally exposed to DEHP+DINP had increased glucose oxidation rates of 17 mg/h compared with those exposed to DEHP only (P = 0.003) and increased glucose oxidation rates of 10.71 compared with those exposed to DINP only (P = 0.035). However, males exposed to DEHP or DINP individually did not display significant differences in longitudinal metabolic outcomes compared with controls. Comparisons between male controls, males perinatally exposed to DBP only, and males perinatally exposed to DEHP+DINP+DBP did not exhibit statistically significant differences via LME models with simple constraints, which was consistent with findings from LME models without constraints and similar to our findings in females (Table 2) (46).

Discussion

Mice perinatally exposed to phthalates exhibited alterations in metabolism throughout their life course. Metabolic impacts of perinatal phthalate exposures were sex-specific, with females showing more sensitivity to phthalate exposures than did males. Females exposed to phthalates had increased body fat percentage, decreased lean mass percentage, and impaired glucose tolerance. One previously published study used indirect calorimetry to evaluate metabolic outcomes in mice following direct exposure to phthalates during adulthood (47), but, to our knowledge, this is the first study to use indirect calorimetry to characterize life course metabolic effects following indirect developmental exposure to phthalates. To our knowledge, this is also the first study to follow mice perinatally exposed to phthalates out until middle age, with longitudinal measurements taken at multiple points throughout the life course. Furthermore, to our knowledge, this is the first metabolic study on perinatal phthalate exposures to examine phthalates other than DEHP, as well as mixture effects.

In previous work, we identified a relationship between perinatal exposure to phthalates and phthalate mixtures and increased body weight in offspring at PND21 (26). However, the current study did not find significant persistent differences (P > 0.10) in body weight longitudinally across the life course in phthalate-exposed offspring compared with controls. It is possible that the differences were too subtle to detect, because the sample size of animals that we followed longitudinally was approximately half of the number of animals that we recorded body weights for at PND21 (n = ∼10 vs n = ∼20 per group per sex, respectively). That female body weights became more variable as they aged also may have contributed to the lack of statistically significant differences in body weight longitudinally. Despite the lack of a statistically significant direct exposure effect on body weight, we observed nonsignificant trends toward increased body weight and found that exposure to DEHP only and DBP only modified the effect of age on body weight in females, indicating that females exposed to phthalates gained more weight as they aged compared with controls. Other researchers have reported both increased and decreased body weights in adult rats and mice following perinatal exposure to phthalates (12, 13, 48). These studies spanned a range of differing mouse strains, including inbred (C3H/N, C57BL6/J) and outbred (CD-1) strains, and they also spanned a variety of ages at body weight measurement cross-sectionally (PND21, 8 weeks, 12 weeks), likely contributing to the inconsistency of findings. Our study, alternatively, used weekly weights to analyze body weight longitudinally from PND21 to 10 months of age.

The association between perinatal phthalate exposures and increased body fat percentage and impaired glucose tolerance observed in the current study are consistent with previously published studies examining perinatal exposure to DEHP in rodents. Multiple mouse studies demonstrated increased body fat in 2- and 3-month-old mice perinatally exposed to DEHP (12–14), whereas other rodent studies have reported associations between developmental DEHP exposure and impaired glucose tolerance in adults (15–17). One distinctive aspect of our study design was the systematic selection of the largest male and female offspring in each litter. This may have resulted in selectively measuring animals that were already more prone toward increased body fat accumulation. However, we selected the largest offspring in both control and exposure groups, and, as mentioned above, our findings were consistent with other studies that examined similar doses of DEHP. Interestingly, our findings also indicated that perinatal exposure to DINP in males resulted in improvement in glucose tolerance with age when compared with controls. Furthermore, the impaired glucose tolerance observed in females perinatally exposed to DINP was more apparent at 2 months of age than at 8 months. Although we did not find significant alterations in absolute levels of circulating adipokines following perinatal phthalate exposures, other studies have reported impacts of perinatal exposure to DEHP on TNF-α and leptin (14, 49). To date, no other studies to our knowledge have examined circulating plasma adipokine levels in adult rodents perinatally exposed to phthalates, and therefore our findings that perinatal phthalate exposures modified the relationship between body fat percentage and plasma MCP-1 and resistin levels in females are novel. Additional studies examining the relationship between developmental phthalate exposure and adipokine levels are needed to corroborate these findings.

HMW phthalates appeared to more consistently impact the metabolic phenotypes measured in this study than did the LMW phthalate we tested. The only statistically significant alterations observed in DBP-exposed females were the positive modification on the effect of age on body weight and a negative interaction between plasma MCP-1 levels and body fat percentage at 10 months. The only statistically significant alteration observed in males exposed to DBP was decreased blood glucose at 120 minutes after glucose administration during OGTTs at 8 months. Additionally, mice exposed to phthalate mixtures containing DBP were not significantly different from controls in any of the metabolic parameters measured, although there were trends toward a modification of the effect of body fat percentage on plasma adipokine levels at 10 months in females. These findings are consistent with in vitro studies that demonstrated increased potency in HMW phthalates vs LMW phthalates with respect to activation of peroxisome proliferator-activated receptors, which is considered to be an important mechanism for phthalates’ obesogenic effects (50, 51). Because we only investigated one LMW phthalate and two HMW phthalates in this study, additional studies examining a larger variety of phthalates are needed to confirm that developmental exposures to HMW phthalates have greater impacts on metabolism than do LMW phthalates.

To our knowledge, this was the first animal study to examine metabolic impacts of perinatal DINP exposure. Previous animal studies have indicated that DINP is capable of endocrine disruption (22, 52). Human and rodent studies have indicated that developmental exposures to DINP is associated with adverse reproductive outcomes (53, 54), and a cross-sectional study on adolescents and teens indicated that direct exposure to DINP was associated with increased insulin resistance (25). The findings in this study indicate an association between perinatal exposure to DINP and impaired glucose tolerance in females. Females perinatally exposed to DINP also had trends, albeit nonstatistically significant after adjusting for multiple comparisons, toward increased body fat percentage and decreased lean mass percentage. Additional studies are warranted to further explore relationships between developmental exposures to DINP and metabolic outcomes, particularly because exposure trends in humans indicate that DINP exposures have been increasing in recent years (11).

Our findings demonstrated striking sex-specific effects of perinatal phthalate exposure on adult metabolic phenotypes. Previously published rodent studies have also demonstrated sex-specific differences in metabolic outcomes following perinatal phthalate exposures. In concordance with our findings that females, but not males, perinatally exposed to phthalates had increased body fat percentage compared with controls, at least one other study also suggested that females are more sensitive to fat accumulation following developmental DEHP exposure (12). Some studies, however, reported no or minimal sex-specific effects with respect to fat accumulation (13, 14). Investigators who examined glucose tolerance following developmental phthalate exposures found conflicting sex-specific effects. One study exposed mice perinatally to DEHP with a subsequent high-fat diet challenge and found that males, but not females, exposed to DEHP and challenged with a high-fat diet had impaired glucose tolerance (17). Conversely, another study found that female rats, but not male rats, had impaired glucose tolerance following developmental DEHP exposure (15). Our findings were in concordance with the latter and demonstrated impaired glucose tolerance in females, but not in males, perinatally exposed to phthalates. Notably, the first study used a higher dose of DEHP (300 mg/kg-day) compared with the current study (estimated dose of 5 mg/kg-day), whereas doses used by the second study were 1.25 and 6.25 mg/kg-day, which are more comparable to the present study. Thus, the sex-specific effects on body fat percentage and glucose tolerance that were evident in this study were also generally in concordance with other similar rodent studies that have reported on these metabolic outcomes and may be dose-dependent.

During lactation, it is possible that phthalate-exposed mice experienced “catch-up” growth, which may have mediated the long-term effects of perinatal phthalate exposures on metabolism. Catch-up growth can occur when infants experience growth restriction in utero, and accelerated catch-up growth is linked with an increased risk of metabolic disease in adulthood (55). Perinatal phthalate exposures have been associated with premature birth (56) and low birth weight (57) in epidemiological studies. Thus, in utero phthalate exposures may result in in utero growth restriction and consequential catch-up growth to influence later-life metabolism. To avoid disturbing the nests, we did not measure birth weights and thus cannot determine whether catch-up growth during weaning mediated the relationships observed between perinatal phthalate exposures and long-term metabolic outcomes in this study. Note that other mouse studies did not find statistically significant changes in birth weights of pups exposed to phthalates in utero (58, 59), but a potential role for catch-up growth cannot be ruled out.

In females, perinatal phthalate mixture exposures did not result in exaggerated effects compared with perinatal exposures to individual phthalates. This was demonstrated by a lack of significant differences in females exposed to DEHP+DINP and DEHP+DINP+DBP in longitudinal models, and particularly in the simple order constrained models. Alternatively, the only exposure group that exhibited any differences that trended toward statistical significance in males was the DEHP+DINP exposure group. One possible explanation for the lack of effect observed in mixture groups in females is that the lack of mixture effect is really a nonmonotonic dose effect. Lower total doses of phthalates perinatally, such as those used in the individual exposure groups, may have larger effects on body composition and glucose tolerance than do higher doses, such as those used in the mixture groups, or, in other words, they exhibit a nonmonotonic dose effect. Nonmonotonic dose responses have been linked with EDCs, including phthalates, in several previous studies (60, 61). Future studies examining multiple doses of each phthalate individually and in mixture form are required to fully understand metabolic effects of perinatal exposures to phthalate mixtures.

Epigenetic reprogramming is considered to be a potential molecular mechanism linking developmental exposures with later-life health outcomes, and it has also been explored as a mediating factor driving sex-specific effects of developmental EDC exposures (62). Previously, we published a study that demonstrated sex-specific alterations in tail DNA methylation at repetitive intracisternal A-particles (IAPs) in weanling mice perinatally exposed to phthalates and phthalate mixtures (26). Together with the longitudinal metabolic phenotypes presented in the current study, our data suggest that altered DNA methylation at IAPs may be a link between perinatal phthalate exposures and metabolic outcomes later in life. Specifically, females in the DEHP only and DINP only mixture exposure groups exhibited increased tail DNA IAP methylation at PND21, and females in these same exposure groups had increased body fat percentage and impaired glucose tolerance, respectively, in adulthood. Females perinatally exposed to a mixture of DEHP and DINP also had increased tail DNA IAP methylation at PND21, but metabolic effects observed in DEHP+DINP females were not statistically significant after adjusting for multiple comparisons. In contrast to females, males perinatally exposed to a mixture of DEHP and DINP had decreased tail DNA IAP methylation at PND21 and had trends toward increased RMR and glucose oxidation rates longitudinally compared with controls. Thus, DNA methylation at repetitive elements in surrogate tissues, such as tail, is a potential biomarker linking perinatal phthalate exposure and metabolic health outcomes.

Although we followed a relatively large number of animals per sex per group (n = 9 to 12), we used six experimental groups, which required adjustment for multiple comparisons when comparing each exposure group to controls, resulting in relatively stringent P value cutoffs for statistical significance. Thus, we consider the metabolic phenotypes that we observed in mice perinatally exposed to phthalates to be robust. Additional metabolic outcomes that had P values <0.05 and did not reach statistical significance after adjustment for multiple comparisons included RER, glucose oxidation, and fat oxidation (46). Future studies evaluating metabolic impacts of developmental phthalate exposures should still consider these as potential outcomes of interest, but fewer experimental groups or a larger sample size may be required to achieve enough power to detect statistically significant effects.

The in-depth metabolic phenotyping measures and longitudinal nature of the study allowed us to characterize long-lasting metabolic changes in mice that were exposed to phthalates and phthalate mixtures perinatally. Females were particularly susceptible to long-term obesogenic effects and impaired glucose tolerance. Importantly, also note that there were additional effects in females and males perinatally exposed to phthalates that could be considered beneficial, although many of these effects were not statistically significant after accounting for multiple comparisons. For example, males perinatally exposed to DINP only had improved glucose tolerance with age compared with controls, and there were trends toward increased fat oxidation in females perinatally exposed to DEHP only and DEHP+DINP. Furthermore, perinatal exposures to phthalate mixtures did not result in larger effect sizes compared with perinatal exposures to individual phthalates, but it was not possible to delineate whether this was due to a nonmonotonic dose effect or to antagonistic mixture effects. A multidose mixture study focusing on one or a few of the outcomes identified here would help to disentangle these two possibilities.

Conclusion

Perinatal exposures to phthalates resulted in altered metabolism across the life course long after exposure had ceased. These effects were sex-specific, phthalate-specific, and had a larger magnitude in mice exposed to individual phthalates compared with those exposed to phthalate mixtures. Compared with controls, females perinatally exposed to DEHP only exhibited increased body fat percentage and decreased lean mass percentage longitudinally across 2 and 8 months, whereas females perinatally exposed to DINP only had impaired glucose tolerance longitudinally. Females perinatally exposed to DEHP only and DBP only had larger increases in body weight with age compared with controls, and males perinatally exposed to DINP only exhibited improved glucose tolerance with age compared with controls. Additionally, at 10 months of age, females perinatally exposed to phthalates had increased plasma resistin and decreased MCP-1 with increasing body fat percentage. The LMW phthalate examined, DBP, appeared to have fewer effects, many of which were distinctive from the effects observed in the mice perinatally exposed to the HMW phthalates. Moreover, mixture effects were difficult to interpret; multidose studies are needed to better characterize these effects. Current human birth cohort studies have yielded conflicting results regarding in utero phthalate exposures and obesogenic effects; the findings from this study suggest some possibilities for these conflicting results, including complex mixture effects. Additional animal and human studies are needed to fully understand mechanisms linking perinatal phthalate exposures and metabolic health outcomes, and to better understand the contribution of developmental phthalate exposures to the high incidence of metabolic syndrome.

Acknowledgments

Financial Support: This work was supported by the University of Michigan National Institute of Environmental Health Sciences/Environmental Protection Agency Children’s Environmental Health and Disease Prevention Center Grants P01 ES022844/P01 RD83543601, and by the Michigan Lifestage Environmental Exposures and Disease National Institute of Environmental Health Sciences Core Center Grant P30 ES017885. Animal phenotyping and adipokine multi-panels were supported by Core Services, supported by the National Institute of Diabetes, Digestive, and Kidney Diseases under Awards U2CDK110768 (MMPC), P30DK089503 (MNORC), and P30DK020572 (MDRC). K.N. was supported by the University of Michigan National Institute of Environmental Health Sciences Institutional Training Grant T32 ES007062 and by National Institute of Child Health and Human Development Institutional Training Grant T32 HD079342.

Disclosure Summary: The authors have nothing to disclose.

Data Availability:

The data sets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Abbreviations:

Abbreviations:
 
• APC

  Animal Phenotyping Core

 
• AUC

  area under the curve

 
• A^vy

  viable yellow agouti

 
• CLAMS

  Comprehensive Laboratory Animal Monitoring System

 
• DBP

  dibutyl phthalate

 
• DEHP

  diethylhexyl phthalate

 
• DINP

  diisononyl phthalate

 
• EDC

  endocrine-disrupting chemical

 
• EE

  energy expenditure

 
• HMW

  high-molecular-weight

 
• IAP

  intracisternal A-particle

 
• LME

  linear mixed effects

 
• LMW

  low-molecular-weight

 
• MLR

  multiple linear regression

 
• OGTT

  oral glucose tolerance test

 
• PND

  postnatal day

 
• RER

  respiratory exchange rate

 
• RMR

  resting metabolic rate

 
• Vco₂

  carbon dioxide production

 
• Vo₂

  oxygen consumption