"2.1. Rat husbandry and diabetes surveillance

We used the UCD-T2DM rat model of T2DM since this rat model more closely represents the pathophysiology and progression of T2DM in humans than other available rodent models of the disease [18]. UCD-T2DM rats develop progressive hyperglycemia along with hyperinsulinemia resulting from insulin resistance at T2DM onset accompanied by ß-cell dysfunction and decompensation [18]. Because the incidence of T2DM among male and female rats is 92% and 43%, respectively, with an average age of onset of diabetes of 6 months in males and 9.5 months in females [18], we only used male rats for the diabetes study (Fig. 1). This model has been validated and characterized for the investigation of the pathophysiology of T2DM, as well as T2DM prevention and treatment in numerous peer-reviewed publications [18], [20], [21], [22].

Animal care and maintenance was conducted in accordance with the applicable portions of the Animal Welfare Act and the U.S. Department of Health and Human Services’ Guide for the Care and Use of Laboratory Animals and approved by the University of California, Davis Institutional Animal Care and Use Committee. Study rats were kept in a climate controlled room at 25 °C and 45%–60% average relative humidity on a 12 h reversed light cycle. All rats were fed Harlan diet 2018 (3.1 kcal/g, 0.57 kcal fat, 1.8 kcal carbohydrates, 0.74 kcal protein; Teklad Diets, Madison WI).

Adult non-pregnant female UCD-T2DM rats (n = 16; 3 months old) were paired with males (n = 10; 3–4 months old) for a 24 h period at which point males were removed. This was defined as gestational day zero (G0) if a sperm plug was observed or if the female rats gained at least 30 g of body weight over the next 7 days. The day of birth was designated postnatal day zero (P0). Pregnant dams were randomly assigned to an exposure group (n = 8 per group), and received daily oral TPhP or ethanol vehicle exposure from G8 through weaning (P21) as described in Section 2.2 below. Gestational length and litter sizewere recorded on P0 and the sex of pups was determined and recorded on P4. Body weights of all pups in each litter were obtained periodically from P4–21.

On P4 the litters were culled to 8 pups ensuring up to 4 males and 2 females in each litter by random selection (Fig. 1A & B). This was done to ensure consistent exposure of pups between litters [13], [23]. The time it takes to develop T2DM is accelerated among UCD-T2DM rats with higher body weights on P21. Hence at weaning the largest pups were housed in same sex littermate groups of two females and up to four males as available (Fig. 1A & B). Urine was collected from the dams using an adapted plastic wrap method outlined by Kurien [24], 60 mins after final dose. Dams were placed in clean cages without bedding for at least 20 min then using a pipette up to 500 μL of urine was collected in ethanol rinsed glass vials and placed on ice. At weaning all dams and remaining weanlings were sacrificed (90–330 min post-exposure) by CO2 asphyxiation and rapid decapitation.

Two male rats weighing between 350–400 g on P61, from the TPhP group and the vehicle group were weight-matched across treatments for the diabetes study to eliminate confounding effects of body mass on the association between TPhP and T2DM onset (Fig. 1B). This weight range was selected because male UCD-T2DM rats that are between 350 and 400 g at 8 weeks of age develop T2DM at approximately 23 weeks of age [18]. Weight-matched rats were followed until 26 weeks or until they developed T2DM, which was defined as two consecutive weekly non-fasting glucose measurements of ≥200 mg/dL [18] in accordance with the American Diabetes Association (ADA) guideline of diagnosing diabetes with a random plasma glucose of 200 mg/dL or higher [19]. The remaining rats were not weight-matched and followed for the 3.5 month obesity study (Fig. 1A).

The terminal procedures for rats in both the obesity and diabetes studies were identical. Prior to sacrifice, rats were fasted between 8 and 12 h, their body weights recorded, and blood was collected from the tail as above prior to euthanasia. Rats were anesthetized with sodium pentobarbital and euthanized with a 1 mL/kg intracardiac injection of saturated potassium chloride. Once cardiac movement had stopped for 30 s the rat was decapitated and the hypothalamus, liver, pancreas, heart, mesenteric adipose tissue, quadriceps, kidney, gonadal adipose tissue, inguinal adipose tissue, and brown adipose tissue were collected. All tissues were removed in the order listed above, wet weighed, and snap frozen in liquid nitrogen.

2.2. Exposure

Given that FM550 contains 17% TPhP [16], our oral TPhP dose of 170 μg per day is equivalent to the amount of TPhP present in the 1 mg FM 550/rat/day dosed over the same developmental period that produced metabolic disruption in a previous study [17]. On average an adult ingests 28 mg of dust per day [25] making oral TPhP administration highly relevant to humans [12], [25]. At 28 mg of dust per day, over the course of pregnancy and breastfeeding a woman would ingest a total of 17.5 g of dust [26]. Therefore it would require only 90 μg of TPhP per gram of dust for a woman to receive 21 mg/kg of TPhP. This is the same total exposure as rats in this study and well below 1800 μg of TPhP/g of dust, which to the best of our knowledge is the maximum observed level reported from samples collected in 2007 and 2012 [12], [27], [28]. Triphenylphosphte (100% purity; Accustandard, New Haven, CT) was weighed to the nearest tenth of a mg, diluted into 100% ethanol (Fisher Scientific, Pittsburg, PA), and vortex mixed to yield a dose solution of 8.5 mg/mL.

Due to their taste aversion to the treat used as a dose vehicle by Patisaul [17], we instead dispensed 20 μL of ethanol with or without TPhP onto a ¼ mini Nilla wafer cookie vehicle, 1 quarter equivalent to 1.75 kcal total, 0.65 kcal fat, 1.05 kcal carbohydrates, 0.05 kcal protein; Nabisco, East Hanover, NJ [29] and allowed the ethanol to completely evaporate before administration to UCD-T2DM rats (Fig. 1A & 1B). Dams were habituated to eating the ¼ mini Nilla wafer prior to exposure from G4-7 (Fig. 1A & B), and all rats readily consumed the ¼ mini Nilla wafer within 5 min for all groups by the onset of the exposure period.

2.3. Urine analysis for the analysis of diphenylphosphate (DPhP)

Urine analysis for the main and non-volatile metabolite of TPhP, diphenylphosphate (DPhP), was performed with small modifications of a method as previously described [30]. Briefly, DPhP was measured using mixed – mode anion exchange solid phase extraction and a labeled internal standard (deuterated (d10)-DPhP). The modifications included that only 500 μL samples were used for each analysis rather than 1 mL and the samples had to be diluted 200 times with the elution solvent, after the clean-up preparation, due to high levels of DPhP detected (values exceded the upper limits of the quantitation curve). The analysis was performed on a liquid chromatographic system (LC) coupled with tandem mass spectrometry (MS/MS) applying negative electrospray ionization (N-ESI). Recoveries for both DPhP and the labeled analog were 105% ± 13% and 122% ± 25%, respectively. For quality assurance purposes, DPhP was evaluated in laboratory blanks and six levels of quality control samples (QCs) during each analysis. DPhP levels in laboratory blanks and materials averaged to 0.3 ng/mL. The method detection limit was calculated by serial dilutions of the lowest calibration standard (0.1 ng/mL) and spiked into 1 mL of pool urine samples that were processed with the whole clean up protocol. The method detection limit (MDL) was set as the lowest amount of analyte that provided a response with: (a) at least 5 times the response of the matrix sample; (b) both precision (relative standard deviation, RSD) and accuracy (error, %Er) ≤20%. Limit of detection (LOD) was the lowest value detected with RSD > 3 So, where So was the standard deviation of the signal received from a urine sample processed and analyzed three times with the same chromatographic method and detected with %Er ≤ 20%. Those two values for DPhP were 0.2 and 0.1 ng/mL respectively.

2.4. Metabolic and physical parameters

Metabolic parameters were evaluated across the lifespan as detailed here and in Fig. 1A and B. Body weight and food intake per cage were recorded weekly in rats from the obesity and diabetes studies, from P21. Every 2 weeks body temperature was assessed in rats from the obesity and diabetes studies by inserting a thermocouple probe into the rectum (BAT-10, Physitemp, Clifton, NJ). Cumulative energy intake (kcal) was calculated as the difference in food weight divided by the number of rats that had access to it and the number of days they had access, and transformed to kcal [31].

2.4.1. Glucose

Non-fasting blood glucose (AlphaTrak; Abbott Laboratories, Abbott Park, IL) was recorded weekly beginning in P21 aged rats from the obesity and diabetes studies. Tail blood was collected from all rats at 3 months of age fasted for between 8 and 12 h to determine fasting blood glucose (AlphaTrak).

At 13 weeks of age, 1 rat/sex/litter/treatment group in both the obesity and diabetes study were randomly assigned to receive an intraperitoneal glucose tolerance test (ipGTT) (Fig. 1A & B). The rats receiving the ipGTT were administered 1.0 g d-glucose/kg body weight after an overnight 14 h fast, and blood was collected from the tip of the tail prior to euthanasia to determine fasting blood glucose [32] at 0, 15, 30, 45, 60, 90, and 120 min after the d-glucose bolus. We determined total area under the curve (AUC) using the trapezoidal approach.

At sacrifice, blood was collected from the tail of rats in the obesity and diabetes studies, after fasting for 8–12 h (Fig. 1A & B) to determine fasting blood glucose. Fasted plasma collected at sacrifice in the diabetes study (Fig. 1B) was tested for glycated albumin (Diazyme; Powey, CA), a glycemic indicator.

2.4.2. Hormones

At 13 weeks of age 1 rat/sex/litter/treatment group in both the obesity and diabetes studies were randomly assigned to receive an intraperitoneal insulin tolerance tests (ipITT) (Fig. 1A & B). The rats receiving the ipITT were fasted for 4 h and then 0.5 U Humulin R/kg body weight was administered i.p. and blood was collected from a tail cut to determine fasting blood glucose at 0, 15, 30, 45, 60, 90, and 120 min after the Humulin R bolus.

At sacrifice, tail blood was collected from obesity and diabetes study-rats (Fig. 1A & B) fasted between 8 and 12 h and thyroid stimulating hormone (TSH) (Alpco Diagnostics; Boston, MA), T4 (CalBiotech; Spring Valley, CA) leptin (Meso Scale Discovery; Gaithersburg, MD), and adiponectin (Meso Scale Discovery) were evaluated with blinding to treatment. Tail blood was collected from all rats at 3 months of age (Fig. 1A & B) and from diabetes study (Fig. 1B) rats at the time of sacrifice after fasting for between 8 and 12 h to determine plasma insulin by ELISA (Mercodia; Uppsala, Sweden) while blinded to treatment. HOMA-IR, an indicator of insulin resistance, and HOMA-%B, an indicator of pancreatic beta cell function, were calculated using the HOMA2 Calculator, available at https://www.dtu.ox.ac.uk, using fasting blood glucose and specific fasting plasma insulin [33].

2.4.3. Lipids

All lipid measures described in this section were performed by investigators blinded to treatment. Tail blood was collected from rats in the obesity and diabetes studies at 13 weeks of age and at sacrifice, respectively, after fasting for between 8 and 12 h (Fig. 1A & B) to determine total cholesterol, total non-esterified fatty acids (NEFA), and plasma triglycerides by enzyme based colorimetry (Wako Chemicals; Richmond, VA).

2.4.3.1. Non-esterified fatty acid profiling

Non-esterified fatty acids (NEFAs) were further isolated from fasted plasma, and collected from diabetes study rats, as previously described by Smedes [34] and Gladine [35]. Specifically, plasma aliquots (50 μL) were enriched with a suite of extraction surrogates, including isotopically labeled triglycerides and phospholipids, along with rare cholesteroyl esters and fatty acids as previously reported by Grapov [36]. Lipids were then extracted twice with 10:8:11 cylcohexane/2-propanol/1 M ammonium acetate (v/v/v), and the organic extract was dried and reconstituted in 2:1 methanol/toluene (v/v), including pentadeca-(10Z)-enoic acid (Nu-Chek Prep, Elysian, MN) to track methylation efficiency. Fatty acids were derivatized with 2 M (trimethylsilyl) diazomethane (Sigma-Aldrich, St. Louis, MO) in hexanes, reduced to dryness and reconstituted in hexane containing methyl tricosanoate (Nu-Chek Prep)and analyzed by GC–MS on an Agilent 6890/5973 N MSD (Agilent Technologies, Santa Clara, CA) with electron impact ionization using a 30 m × 0.25 mm × 0.25 μm DB–225 ms column to separate analytes. Quantification was accomplished with ChemStation vE.02.14 (Agilent Technologies) using internal standard methodologies.

2.4.3.2. Lipid mediator profiling

Non-esterified oxylipins, endocannabinoids, endocannabinoid-like monoacylglycerols and N-acylethanolamines, and nitrated fatty acids were quantified by tandem quadrupole mass spectrometry using slight modifications of previously reported methods [36]. Briefly, analytes were isolated from hypothalami collected from female obesity study rats, as previously described [37] but without ester hydrolysis. Hypothalamus aliquots (∼20–25 mg) aliquots were enriched with isotopically labeled surrogates and extracted with 1:10:5 methanol/ethyl acetate/deionized water (v/v/v). Extracts were dried under vacuum, re-constituted in 1:1 MeOH/acetonitrile (v/v) containing internal standards, and filtered at 0.1 μm through a PVDF membrane. Analytes were separated on a 2.1 × 150 mm, 1.7 μm Acquity UPLC BEH column and detected by negative mode (oxylipins and nitrated fatty acids) and positive mode (endocannabinoids and endocannabinoid-like compounds) electrospray ionization using multiple reaction monitoring on an API 4000 QTrap (AB Sciex, Framingham, MA). Analytes were quantified against calibration curves and correcting for surrogate losses using Sciex MultiQuant version 3.0.2.

2.5. Semi-quantitative PCR

The RNeasy kit (Qiagen; Hilden, Germany) was used to extract mRNA from aliquots of pulverized mesenteric fat to be used in reverse transcription PCR (Applied Biosystems; Foster City, CA) and semi-quantitative PCR while blinded to treatment. RNA was quantitated by absorbance at 260 nm wavelength and retained with a purity threshold >1.8 for the 260/280 ratio (NanoQuant, Tecan; Männedorf, Switzerland). Semi-quantitative PCR was performed using SYBR Green primers: Peroxisome proliferator activated receptor gamma (Pparg, F: TGTGAAGGATGCAAGGGTTT, R:CATTCGCCCAAACCTGATGG), CCAAT/enhancer binding protein (C/EBP) alpha (Cebpa, F:TGCGCAAGAGCCGAGATAAA, R:GCGGTCATTGTCACTGGTCA), CCAAT/enhancer binding protein (C/EBP) beta (Cebpb, F:CAAGATGCGCAACCTGGAGA, R:AGCTGCTTGAACAAGTTCCG) in mesenteric fat using beta actin (Actb, F:CTGACAGGATGCAGAAGGAG, R:GATAGAGCCACCAATCCACA) as an endogenous control; the endogenous control did not vary significantly across treatment groups. All primers were designed for Rattus norvegicus by querying the Pubmed gene library with the NCBI Primer-BLAST tool [38]. Primers were designed to span an exon-exon junction with each pair separated by at least one intron in the corresponding genomic DNA. The 2−ddCT method was used to approximate relative transcript fold change between treatment groups [39].

2.6. Statistical analyses

Relative tissue weights were determined by dividing each tissue weight by body weight. Metabolic parameters were statistically analyzed as follows to determine means ± standard error of the mean (SEM) at a significance threshold of P < 0.05. Analyses of non-fasting blood glucose, HOMA-IR, HOMA-%B, body weight, temperature, tissue weights, relative tissue weights, TSH, adiponectin, insulin, triglycerides, cholesterol, total NEFA, and RNA expression were assessed by modeling the fixed effect of perinatal TPhP and the random effect of litter while stratifying by sex and study group shown in Fig. 1A and B using fixed effects for treatment and litter as a repeated event (PROC MIXED, SAS). All mixed models used LSMEANS to find the mean and SEM and PDIFF to test significance. We also tested for sex interactions at threshold p < 0.1 and if it was significant it was included in the model statement in addition to the sex-stratified models described above. The distribution of the data was checked for normality before T-tests were used to evaluate gestation length, litter size, sex ratio, dam body weight, luciferase E2, cumulative energy intake, and targeted lipids because only one rat per litter was included (PROC TTEST, SAS). ITTs and GTTs were evaluated by stratifying by sex and study group (Fig. 1) without random effects because only one rat per litter was included (PROC GLM, SAS). We evaluated whether perinatal TPhP accelerated the onset T2DM by using survival analysis with treatment as a fixed effect and litter as a repeated event (PROC LIFETEST, SAS). This is the best method of analyzing differences in the onset of T2DM because survival analysis accounts for the non-normal distribution of time to event data and can account for censoring, that is, ending the observation period before all rats have developed T2DM."

F1 Weight-Matched Male UCD-T2DM Rats