Comparison of in utero phthalate exposure on fetal rat testis transcript and endocrine alterations with apical reproductive alterations in male rats

Abstract

In utero administration of some diortho phthalate esters (PEs) produces reproductive tract malformations in male and female rat offspring via unknown molecular initiating events. Although the molecular initiating event(s) for these effects are unknown, the PEs consistently alter several key endocrine and gene expression events in the fetal male rat providing a signature of in utero PE exposure. We compared the dose-related alterations of in utero PE exposure on gene expression levels, measured with targeted RT-qPCR custom arrays, and ex vivo testosterone production (T Prod) with the reproductive alterations seen in F1 male rats from three different PE studies. The PEs studied included dicyclohexyl (DCHP) and dipentyl phthalate (DPeP) and a mixture study with five phthalates (DCHP, diethylhexyl (DEHP), dibutyl (DBP), butyl benzyl (BBP), and diisobutyl phthalate (DiBP)). These results demonstrate that targeted testis gene expression and/or T Prod data from short-term prenatal studies conducted during a critical window of fetal masculinization can be used to determine points-of-departure (PODs) for PEs and that these PODs are several fold more protective than the apical effects in postnatal animals. The clear linkage of these gene transcript and testosterone changes to the adverse effects of in utero exposure can facilitate acceptance of the use of gene expression and endocrine data to determine PODs for risk assessment. Furthermore, the use of gene expression and endocrine data from short-term in vivo fetal studies in place of multigenerational reproductive studies for POD determination for PE and mixtures of PEs could increase the rate of POD development for risk assessment and reduce use of animals and other resources.

Introduction

In utero administration of some diortho phthalate esters (PEs) produces reproductive tract malformations in male and female rat offspring via unknown molecular mechanism(s) (Hannas et al. 2013). In the F1 males, some abnormalities are unique to the PEs like gubernaculum testis agenesis (Gray et al. 2016; Wilson et al. 2004), whereas other effects also are seen after in utero exposure to drugs and toxicants that act as androgen receptor (AR) antagonists (Gray et al. 1994; Ostby et al. 1999) or 5α reductase inhibitors (Clark et al. 1990). Although the molecular initiating event(s) for these effects are unknown, the PEs consistently alter several key endocrine, genomic (Barlow et al. 2003; Bowman et al. 2005; Euling et al. 2013; Hannas et al. 2012; Hannas et al. 2011b; Johnson et al. 2007; Lahousse et al. 2006; Lehmann et al. 2004; Liu et al. 2005; Makris et al. 2013; Thompson et al. 2005) and proteomic events (Ivell et al. 2022; Lehmann et al. 2004; Liu et al. 2005) in the fetal testis that are causally related to adverse effects seen in male rat offspring. In contrast`, little is known about the mechanistic pathway for the PE-induced vaginal and uterine agenesis in F1 female rats (Hannas et al. 2013).

The genomic and endocrine signature we have characterized is based upon studies from our laboratory and other laboratories that identified consistent alterations of mRNA expression of key genes and reduction in production of testosterone by the fetal testis (Barlow et al. 2003; Bowman et al. 2005; Earl Gray et al. 2024; Gray et al. 2021; Johnson et al. 2007; Lehmann et al. 2004; Liu et al. 2005; Ovacik et al. 2010; 2013; Shultz et al. 2001; Thompson et al. 2005). Other laboratories have shown that phthalates also reduce the concentrations of several key fetal testis proteins coded by these genes. These proteins include the insulin-like 3 (Insl3) peptide hormone (Ivell et al. 2022), cholesterol transport proteins scavenger receptor class B member (Scarb1) and steroidogenic acute regulatory (StAR) protein, several key enzymes regulating androgen synthesis including cytochrome P450 side-chain cleavage (Cyp11A1), 3 beta-hydroxysteroid dehydrogenase (3βHsd), and cytochrome P450c17 (Cyp17A1) (Lehmann et al. 2004), and Testin (Tes) (Liu et al. 2005), a protein secreted by Sertoli cells whose increased mRNA levels correlate with the disruption of the integrity of inter-testicular cell junctions (Grima et al. 1998; Grima et al. 1997). These genes are highly conserved among mammals and disruption is known to cause disorders of sexual development. Studies have shown that knockout mice that lack these genes display reproductive tract abnormalities (Aherrahrou et al. 2020; Caron et al. 1997; Hasegawa et al. 2000; Hu et al. 2004; Hu et al. 2002; Zimmermann et al. 1999), and mutations of these genes in 46,XY boys also results in disorders of sexual development (Acien and Acien 2020; Bertelloni et al. 2021; Delot et al. 2017).

Initially, we designed a custom RT-qPCR array that measured levels of mRNA for 88 target genes involved in steroid transport and synthesis, sex determination, and peroxisome proliferator-activated receptor (PPAR) activation (described in detail in Hannas et al 2012) to identify key events disrupted in the fetal testis by in utero PE administration (S2). In addition, we evaluated the effects of PEs on 20 different RT-qPCR arrays and one whole genome study (Waterbury et al. 2024). Based upon the genomic data from our studies and those in the literature (Johnson et al. 2011; Liu et al. 2005) we recently designed a more targeted RT-qPCR array that measures mRNA levels for 19 fetal testis target genes per litter (4 litters per 96 well plate) to identify PEs that disrupt sexual differentiation. Similarly, we have measured fetal testis testosterone production (T Prod) following ex vivo incubation of fetal testes using radioimmunoassay and more recently LC-MS/MS. The dose responsive changes in testis target gene expression and T Prod have been compared along with the postnatal effects of exposure to phthalates to define relative potency factors (RPFs) for PEs among those that disrupt reproductive development in utero.

The first objective of the current investigation was to use the RT-qPCR arrays and testis T Prod measurements to determine genomic and endocrine points of departure (PODs) for PEs and PE mixtures. In the first fetal study, we used the original custom array (Gray et al. 2021; Hannas et al. 2012) to determine PODs for di-cyclohexyl phthalate (DCHP). Then we used both sets of custom arrays to determine PODs for a mixture of five PEs in a second fetal study including dibutyl phthalate (DBP), di-isobutyl phthalate (DiBP), benzyl butyl phthalate (BBP), diethyl hexyl phthalate (DEHP), and DCHP, which are designated as High-Priority Substances under EPA’s Toxic Substances Control Act (TSCA) and are actively undergoing risk evaluation under TSCA (hereinafter referred to as the TSCA mixture) (https://www.federalregister.gov/documents/2019/08/23/2019-18134/proposed-high-priority-substance-designations-under-the-toxic-substances-control-act-tsca-notice-of). In the second study, we also evaluated the postnatal effects of the TSCA mixture on reproductive tract alterations following maternal gestational exposure. In the third study, we evaluated the genomic, endocrine, and postnatal data from previously published studies with dipentyl phthalate (DPeP) (Gray et al. 2016; Gray et al. 2021; Gupta et al. 2025; Hannas et al. 2011a).

The second objective was to compare the genomic and endocrine PODs to the dose related adverse effects of in utero exposure to the PEs used in these three studies. Several of the affected transcripts displayed similar dose responsive changes to fetal testis T Prod which has been used to set PODs for reproductive toxicity (Canada 2015; EPA 2023; 2024c; 2025). In addition, an objective of the TSCA mixture study was to determine if the effects of the 5 PE mixture on the F1 male rat offspring could be accurately predicted with dose-addition modeling.

These results demonstrate that targeted testis gene expression and/or T Prod data from short-term prenatal PE studies conducted during the critical window of fetal masculinization can be used to determine PODs that are several-fold more protective than the effects in postnatal animals. The clear linkage of these genomic and endocrine changes to the adverse effects of in utero exposure can facilitate regulatory acceptance of the use of genomic and endocrine data to determine PODs for PE risk assessments. The use of genomic and endocrine data from short-term in vivo studies, replacing standard reproductive and developmental guideline studies for POD determination, would significantly increase the rate of development of PODs for risk assessment and also reduce use of animals and other resources.

Methods

Animals

Time-mated Sprague-Dawley rats (Crl:CD(SD)), ~11–12 weeks old, were purchased from Charles River Laboratories (Raleigh, NC, USA) and shipped to USEPA (Research Triangle Park, NC, USA) on gestation day (GD) 2 (GD0 = bred date; GD1 = confirmed mated). Dams were housed individually in clear polycarbonate cages (20 × 25 × 47 cm) with heat-treated, laboratory-grade pine shavings (Northeast Products, Warrensburg, NY) and fed NIH07 Rat Chow and treated municipal tap water (Durham, NC tap water, 50μm particle filtration, granular activated charcoal filtration, and rechlorinated) ad libitum. Animals were housed in a facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care and maintained at 20–22°C, 45–55% humidity, and a 12:12 h photoperiod (06:00 – 18:00 EST). This study was conducted in accordance with a protocol (Animal Care and Use Plan #22–03-001) approved by the USEPA Center for Public Health and Environmental Assessment Institutional Animal Care and Use Committee.

Test chemicals

Dosing solutions were prepared using laboratory-grade corn oil (CAS 8001–30-7; Sigma-Aldrich, St. Louis, MO, USA). Dicyclohexyl phthalate ((DCHP), CAS 84–61-7; Lot 17518JB; ≥98% purity) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Dipentyl phthalate ((DPeP), CAS 131–18-0; Lot 1431420; RTI Log 040109-A-14), diethylhexyl phthalate ((DEHP), CAS 117–81-7; Lot 01514TH; RTI Log 121208-B-16), dibutyl phthalate ((DBP), CAS 84–74-2; Lot 91997PJ; RTI Log 031609-A-17), butyl benzyl phthalate ((BBP), CAS 85–68-7; Lot 03405JH; RTI Log 121208-A-15), diisobutyl phthalate ((DiBP), CAS 84–69-5; Lot 07425BJ; RTI Log 070208-A-18) were gifted from the National Toxicology Program (Research Triangle Park, NC, USA). All gifted phthalates from NTP were independently verified for purity (>99%) by Research Triangle Institute (Durham, NC, USA) using gas chromatography with flame ionization detection. Dosing solutions were administered daily via oral gavage at a dosing rate of 2.5 mL/kg between 07:00 – 09:00 EST.

Fetal Studies. DCHP (Study 1) and TSCA mixture (Study 2, Experiment 1)

The fetal experiment with DCHP was comprised of one block with 15 pregnant rats. Dams were weight-ranked and randomly assigned to treatment groups using PROC OPTEX in SAS 9.4 to produce five treatment groups with similar mean and variances in initial body weight (n = 3 dams per treatment group). Dams were exposed to DCHP at 0, 100, 300, 600 and 900 mg/kg/d from GD 14–18 by daily oral gavage (2.5 mL/kg).

In the second fetal experiment, two blocks of dams with 15 dams/block were exposed to the TSCA mixture of five phthalates selected from the 2019 EPA TSCA High Priority list including DBP, BBP, DiBP, DCHP and DEHP (EPA 2019). The top dose of the mixture contained 120 mg/kg of each PE (total PEs=600 mg/kg/d). Lower dose groups included 50, 5, 0.5 and 0.05 % of the top dose equal to 60, 6, 0.6 and 0.06 mg of each PE or a total PE dose of 300, 30, 3 and 0.3 mg/kg/d, respectively.

On GD18 dams and fetuses were euthanized by decapitation 2–4 hours after the final oral dose. The euthanasia order was stratified such that the timing of necropsy was equally distributed across treatment groups. We evaluated maternal weight gain from GD14–18, reproductive output (number of fetuses, resorptions), measured testis T Prod from 3 males per litter, and collected and pooled testis mRNA from the remaining males.

Testis gene expression in GD18 fetal rats

GD18 fetal testis mRNA was pooled by litter for analysis of gene expression as previously described (Gray et al. 2021). GD18 fetal testis mRNA samples were evaluated using 96-well custom designed RT2 Profiler PCR Arrays from Qiagen (Germantown, MD).

The testis mRNA samples from the DCHP study were analyzed on our custom arrays containing 88 target genes per plate and we interrogated one litter per plate (described in detail in Hannas et al. 2012). The mRNA samples also were used for transcriptomic analyses using the Templated Oligo Detection Assay platform (TempO-Seq, Biospyder Technologies, Carlsbad, CA, USA) (Waterbury et al. 2024). This approach uses detector oligo probes for 21,119 protein coding genes in the rat genome and does not target non-coding RNA (Waterbury et al. 2024).

The testis mRNA samples from the TSCA mixture study were analyzed on two sets of custom arrays. The first array was our original custom array with 88 target genes per plate. The second array was a newly designed array which interrogated mRNA for 19 target genes per litter (plus two housekeeping genes, a reverse transcriptase control, a genomic DNA contamination control, and a positive PCR control) with one litter from four different treatments groups evaluated on each new array plate. The target genes included on these new arrays were selected because our studies (Gray et al. 2021) (Waterbury et al. 2024) and others (Barlow et al. 2003; Johnson et al. 2007; Liu et al. 2005) demonstrate statistically significant alterations of expression levels of these specific genes from PE exposure. PCR reactions were run using RT2 SYBR Green qPCR Master Mix (Qiagen; Hilden, Germany) on a CFX96 Touch Real-Time Detection System (Bio-Rad; Hercules, CA).

Testis testosterone production

Fetal testicular testosterone production (T Prod) was determined using an ex vivo culture method established at EPA (Furr et al. 2014). One freshly isolated testis from each of three males per litter was individually transferred into 500 μL of modified M-199 media without phenol red (Gibco Life Technologies, Product #A31224DK), supplemented with 10% dextran-coated charcoal stripped fetal bovine serum (Hyclone Laboratories, Logan, UT) for 3 h in a humidified incubator at 37°C on an orbital shaker, to allow testosterone to diffuse into media (n= 3 independent testis incubations per litter).

Following incubation, media was removed and stored at −80 °C until measurement of T Prod in media. T Prod in media was quantified using an RIA kit in the DCHP study using ALPCO kits (Salem NH, Cat # 72-TESTO-CT2). The intra-assay coefficient of variation for this kit was 5.6% (based on variability of the standard curve) and the inter-assay coefficient of variation was 8%. Cross-reactivity of the T antibody with dihydrotestosterone (DHT) was 2.6%. The limit of detection was about 0.08 ng testosterone/ml. In the TSCA study, T Prod was measured by LC-MS/MS. Mobile phase components were 0.1% formic acid in water and acetonitrile. Diluents were analyzed on an AB Sciex 6500+ QTRAP Linear Ion Trap mass spectrometer using electrospray ionization in positive ion, multiple reaction monitoring modes. Chromatographic separation was performed on a Restek Raptor Biphenyl column (2.1 mm × 50mm, 1.8 μm). The coefficient of correlation of the curve was ≥ 0.995. The calculated concentration for each testosterone calibration point was within 80–120% of the actual spike amount. The calibration was verified by a second source standard prepared similar to the calibration standards at a concentration of 1.00 ng/mL. The acceptance range was 70–130%. The actual recovery for the second source standard was 82%.

Study 2, Experiment 2: TSCA mixture one generation experiment

Postnatal evaluation of phthalate mixture effects was conducted similarly to previously reported methods (Gray et al. 2016). Briefly, five dams per dose were dosed by oral gavage with the TSCA mixture as in the fetal experiment (n=30 litters) from GD14 to 18. In the offspring, anogenital distance (AGD) and body weights were measured at two days of age (PND 2) (24 to 34 pups/sex/dose) and the numbers of nipples/areolae were counted at 13 days of age (PND 13) (both were done with the observer blind to treatments). On PND 23 dams and female pups were weighed and euthanized by decapitation. Dams were necropsied to count uterine implantation sites and female pups were examined for the presence of vaginal and uterine agenesis. On PND 23 male pups were weighed and weaned to 2–3 per cage per treatment. The F1 males were examined from PND 36 to 45 and the age at full preputial separation (PPS, an index of puberty in the male rat) was determined and body weight at PPS measured. At 65 days of age all male offspring were weighed, euthanized by decapitation, testes weights were measured, and animals were examined for the presence of reproductive tract malformations (histopathological lesion of the testis and epididymis, and agenesis of the vas deferens, ventral prostate, seminal vesicles, gubernaculum testis, and testis, and epididymis). These endpoints were selected because they have been among the most sensitive effects in our PE postnatal studies. Paired testes and epididymides were evaluated for histopathological lesions by a board certified pathologist at Experimental Pathology Laboratories (Research Triangle Park, NC). Testis findings were graded according to 5 grade criteria (minimal, mild, moderate, marked, and severe lesions corresponding to numerical scores of 1–5 where 1 is the most minimal change and 5 represents the most severe change observed. Grade 1 = 1– 10 %, 2 = 11–25 %, 3 = 26–50 %, 4 = 51–75 %, and 5 = 76 % of tissue examined).

Study 3: Dipentyl Phthalate (DPeP) fetal and postnatal study data

Data used in the present study to evaluate genomic, endocrine, and postnatal effect PODs for DPeP were previously published (Gray et al. 2016; Gray et al. 2021; Hannas et al. 2011a). Briefly, timed-pregnant SD rats were purchased from Harlan Laboratories (Indianapolis, IN) and DPeP was administered daily via oral gavage at doses of 0, 11, 33, 100, and 300 mg/kg/d from GD 14–18 in the fetal study and from GD 8–18 in the postnatal study. As detailed above, testis mRNA was assessed on our original custom RT-qPCR arrays with 88 target genes and T Prod was measured on GD 18 in the fetal study. The mRNA and T Prod data are from two blocks of pregnant rats with 3 to 8 litters per dose group (3 males/litter for T Prod). In the postnatal study male rat offspring were necropsied as adults (≥120 days of age) and evaluated for male reproductive tissue weights and malformations.

Data analyses

All data analyses were conducted using GraphPad Prism (version 8.4; GraphPad, Inc.; La Jolla, CA, USA) and SAS (version 9.4; SAS Institute; Cary, NC, USA). All maternal, fetal and postnatal data were analyzed by analysis of variance (ANOVA) using PROC GLM in SAS, followed by pairwise comparison of individual dose levels to vehicle controls using LSMEANS (α=0.05). Individual dams were the unit of statistical analysis whereas litter means were used for analyses of all fetal and postnatal data.

In addition, in the TSCA mixture study we averaged the fold values for 22 affected transcripts to summarize the relationship between the effects on gene expression with the postnatal alterations (Genomic_signature = (Nr5a1 +Nr0b1 +Star +Cyp11a1 +Cyp11b2 +Hsd3b +Cyp17a1 +Lhcgr +Scarb1 +Insl3 +Dhcr7 +Tspo +Cyp11b1 +Inha +Inhba +Cyp51 +Ebp +Hmgcr +Hmgcs1 +Idi1 +Mvd +Tm7sf2)/22)). For the DCHP and DPeP studies the Genomic_signature = ((Nr0b1 + Star + Cyp11a1 + Cyp11b2 + Hsd3b + Cyp17a1 + Lhcgr + Scarb1 + Insl3 + Dhcr7 + Cyp11b1 + Inha)/12) using data from the original custom arrays.

Fetal testis RT-qPCR gene expression data from samples collected on GD 18 were analyzed using the comparative cycle threshold (C_T) method. Delta C_T values were calculated using the equation 2^−ΔΔC_T and normalized to the mean C_T value of the housekeeping genes for each litter array. We selected two genes that are not affected by PE exposure (Actb, B2m) to normalize the data to delta C_T values. Delta C_T values were then converted to fold-induction by dividing the treated replicate delta C_T by the mean delta C_T of the control replicates for each gene.

Gene expression fold values were analyzed using BMDExpress Version 3, Build 0017 (https://github.com/auerbachs/BMDExpress-3/wiki/) to identify BMD (benchmark dose)/BMDL (lower confidence limit on the BMD) and model fit values. Data on the fold expression values from the custom gene arrays were initially filtered using the one-way ANOVA (p<0.05) and the selected transcripts analyzed using the EPA BMD MLE ToxicR EXP 5, Hill, linear, EXP 3 and power models to determine the BMD, BMDL, and BMDU (upper 95% confidence limit on the BMD) values from the model with the lowest AIC value that was a good visual fit to the data (benchmark response (BMR) was1 standard deviation (SD), BMRF was 1). These results were then filtered to include only transcripts with a BMD less than 250 mg/kg, a BMD:BMDL ratio of less than 20, a BMDU:BMDL ratio of less than 40, the dose response was not a step function below the lowest dose, best fit p value > 0.5 and the R²> 0.7.

All the transcript data were also analyzed using PROC GLM in SAS 9.4 and to identify significantly (ANOVA p <0.01) affected transcripts and no observed transcript effect levels (NOTELs) were determined using LSMEANS. In addition, EPA BMDS 3.3.2 software was used to compare to the results from BMDExpress for the transcripts that displayed acceptable BMDExpress model results. One factor that contributed to the uncertainty of the BMD values in most of the transcripts was that the lowest dose of DCHP used in the current study was more than 10 fold high than the ED₂₀ and BMDs from the BMDExpress “best fit models” (Table 1).

Table 1.

BMDExpress analyses of the effects of DCHP exposure on fetal testis mRNA expression dose response models, benchmark dose (BMD), BMD lower bound (BMDL), BMD upper bound (BMDU), model fits (R² and fit p value), and if the model displayed an all or none response (Step Function). Transcripts having flagging issues that create uncertainty in the BMD values are shaded.

1.a
Probe ID	Uncertainty Flag	Best BMD	BMDL	BMDU	fit P Value	BMD/BMDL	BMDU/BMDLB	MDU/BMD	Low dose/BMD	Is Step Function Below lowest dose
T Prod	None	17.7	9.3	53.0	1.00	1.9	5.7	3.0	5.6	FALSE
Cvp11a1	None	22.1	5.1	81.6	1.00	4.4	16.1	3.7	4.5	FALSE
Lhcgr	BMDU/BMDL, Step Function	6.4	0.0	40.8	0.19	434.2	2790.0	6.4	15.7	TRUE
Inha	BMDU/BMDL, Low Dose/BMD	0.0	0.0	0.4	0.98	315.4	10813.1	34.3	9269.8	FALSE
Cvp11b1	BMDU/BMDL, Low Dose/BMD	0.1	0.0	60.1	1.00	332.7	263859.3	793.0	1319.2	FALSE
genetic_signature	BMDU/BMDL, Low Dose/BMD	0.1	0.0	6.5	0.99	58.2	2585.4	44.4	685.7	FALSE
Cvp17a1	BMDU/BMDL, Low Dose/BMD	0.4	0.1	2.2	0.96	4.7	26.8	5.7	258.3	FALSE
Insl3	BMDU/BMDL, Low Dose/BMD	1.3	0.0	21.5	1.00	45.6	779.9	17.1	79.6	FALSE
Star	BMDU/BMDL	4.9	0.0	51.1	1.00	141.8	1473.3	10.4	20.3	FALSE
Scarb1	BMDU/BMDL	6.1	0.1	53.9	0.98	56.1	492.3	8.8	16.3	FALSE
Hsd3b	BMDU/BMDL	12.7	0.0	81.7	0.98	434.2	2790.0	6.4	7.9	FALSE
Dhcr7	BMDU/BMDL	73.1	0.8	291.5	1.00	90.8	362.1	4.0	1.37	FALSE
Nr0b1	BMD, fit P value	360.2	225.0	883.3	0.25	1.6	3.9	2.5	0.28	FALSE
Cvp11b2	BMD, fit P value	694.2	341.8	NaN	0.01	2.0	NaN	NaN	0.14	FALSE
Rhox10	BMD, fit P value	820.5	372.7	NaN	0.21	2.2	NaN	NaN	0.12	FALSE
1.b
Probe ID	Uncertainty Flag	MA BMD	BMDL	BMDU	BMD/BMDL	BMDU/BMDL	Low dose/BMD	Low dose/BMDL	RSquared	Is Step Function Less Than Lowest Dose
Cvp17a1	None	5.7	1.0	27.3	5.62	26.8	17.5	98.2	0.99	FALSE
genetic_signature	None	10.9	1.5	46.4	7.17	30.5	9.2	65.7	0.98	FALSE
Insl3	None	21.0	4.1	68.3	5.09	16.6	4.8	24.3	0.97	FALSE
Star	None	34.0	7.9	111.4	4.32	14.1	2.9	12.7	0.96	FALSE
Scarb1	None	36.6	8.3	129.7	4.40	15.6	2.7	12.0	0.96	FALSE
Cvp11a1	None	52.9	17.8	136.5	2.97	7.7	1.9	5.6	0.99	FALSE
Dhcr7	None	95.8	27.7	311.3	3.45	11.2	1.0	3.6	0.99	FALSE
Cvp11b1	Low dose/BMDL, Step Function	14.9	1.7	40.5	8.63	23.5	6.7	58.0	0.98	TRUE
Inha	Low dose/BMDL	15.2	0.9	44.9	16.45	48.6	6.6	108.1	0.96	FALSE
Nr0b1	BMDU/BMDL, Rsquared	88.4	1.6	607.5	53.72	369.1	1.1	60.7	0.91	FALSE
Cvp11b2	BMDU/BMDL, Step Function	10.7	0.2	78.3	44.14	322.8	9.3	412.4	0.97	TRUE
Lhcgr	BMDU/BMDL	13.6	0.1	39.9	112.74	331.9	7.4	831.8	0.95	TRUE
Hsd3b	BMDU/BMDL	30.1	0.3	84.9	94.07	265.1	3.3	312.0	0.96	TRUE

Abbreviations: BMD – benchmark dose, BMDL – benchmark dose lower bound, BMDU – benchmark dose upper bound, MA – model average. Table 1.a reports the results of the analyses to determine the “best fit” model and Table 1.b reports the results of LePlace Bayesian Model Averaging that included the Hill and EXP5 models.

Since many of the BMD models of the DCHP data were uncertain for a variety of factors (Table 1.a), we opted to reanalyze these data using the Bayesian Benchmark Model Averaging (BMA). This method “provides more adequate results in terms of less extreme values and no failure in BMD and BMDL calculations” (Ji et al. 2022).

Maternal and litter data from the TSCA mixture postnatal study were analyzed using PROC GLM in SAS 9.4 and each dose group was compared to control using the LSMEANS option. Litter means values were used to analyze the data when more than on male from a litter was evaluated. Dose response data were also analyzed using a four-parameter logistic regression (4PL) model in GraphPad Prism 8.4.3 to identify the ED₂₀ and slope parameters for each variable. Data were normalized to percent of control in Prism and logistic regression models were run with the top constrained to 100% and 0% at the bottom.

Finally, the observed effects in the TSCA study were compared to dose- and response addition mixture predictions (Altenburger et al. 2000; Gray et al. 2022; Kortenkamp 2007; Metzdorff et al. 2007; NRC 2008) based on the effects seen with each of the individual phthalates (models published previously) (Earl Gray et al. 2024)}.

Results

Study 1: Fetal effects of DCHP

Administration of DCHP from GD 14–18 did not significantly reduce maternal weight gain, fetal viability or litter size (S.1). Fetal testis T Prod and several transcripts on the custom gene arrays were significantly reduced in a dose-related manner including Star (p < 0.002), Cyp11a1 (p <.0003), Cyp11b2 (p < 0.05), Hsd3b (p < 0.007), Cyp17a1 ( p < 0.0001), Lhcgr (p < 0.0001), Scarb1 (p < 0.002), Insl3 (p < 0.0002), Dhcr7 (p < 0.05), Cyp11b1 (p < 0.0001), and Inha (p <0.0001) (Figure 1). The averaged transcript, Genomic_signature, was significantly reduced (p < 0.0001) by as much as 70% of control with a CV of 14%. mRNA transcripts for three Ppar receptor genes (p > 0.85), androgen receptor (AR (p> 0.98)), anti-Müllerian hormone receptor type 2 (Amhr2 (p > 0.75) and estrogen receptors α and β (ESR1 and ESR2 (p > 0.75) were unaffected by treatment (S.2, 3, 4).

Figure 1.

Dose-related effects of dicyclohexyl phthalate (DCHP) on fetal rat testis mRNA expression levels, expressed as percent of control. The column labels are the dose of DCHP administered to the dam in mg/kg/d. The data used to calculate these values, the SEs and sample sizes are provided in Supplemental files 3 and 4.

When the gene transcript and T Prod data were analyzed in BMDExpress to identify the “best fit” model for transcripts that were significantly reduced by ANOVA and with the filters in place, only T Prod, Cyp11a1 and Cyp17a1 passed these filters. However, the BMD and BMDL values for Cyp17a1 were about 250- and 1,200-fold below the lowest DCHP dose of 100 mg/kg, respectively.

When all filters were turned off, 10 more transcripts were identified as being significantly reduced by ANOVA (Star, Hsd3b, Cyp11b2, Lhcgr, Scarb1, Insl3, Dhcr7, Inha, Dkk3 and Cyp11b1). Among these there was uncertainty about the values for several reasons. For example, the dose response for Lhcgr was a step function below the lowest dose with a high BMDU:BMDL ratio, Cyp11b2 had a low fit value (0.008) and R² (0.27), Dkk3 had a high BMD of 576 mg/kg/d, and the rest had BMDU:BMDL ratio values greater than 350 (Table 1). In summary, T Prod and Cyp11a1 mRNA were the only measures that had robust BMD/BMDL estimates and fit the modeling criteria. The BMD/BMDL/BMDU estimates for the most sensitive genes were questionable since the estimates were well below the lowest dose of DCHP in the current study and the estimates for T Prod and Cyp11a1 mRNA were about 5 fold below the lowest dose tested in the study.

BMDExpress LePlace Bayesian Model Averaging included the Hill and EXP5 models (Table 1.b). This method provided more adequate results in terms of less extreme values and acceptable fits the transcripts. The BMD/BMDL/BMDU values for the pooled transcript genomic_signature were 10.9/1.5/46.4 with an R² = 0.978, however the BMD is nearly 10 fold below the lowest dose of DCHP.

When the data were reanalyzed with EPA BMDS 3.3.2 software the BMD/BMDL/BMDU for Cyp11a1 (17/10/42 mg/kg/d) versus (52/17/137) with BMDExpress Model Averaging (Table 1.b). All BMDS models for Cyp17a1 were questionable or unacceptable, likely due to the strong reduction in expression of this gene at the lowest dose tested in the present study.

The BMDExpress BMDLs for T Prod and Cyp11a1 were 9 and 5 mg/kg, respectively. For comparison, the US Consumer Product Safety Commission (CPSC (2010)) reported that BMD₁₀/BMDL₁₀ values for neonatal AGD were 101/68 and 130/79 mg/kg (from 2 studies, respectively), and for testis histopathology were 169/91 mg/kg, indicating that the genomic and endocrine BMDLs were about 10-fold lower than the BMDLs for Phthalate Syndrome abnormalities in male rat offspring.

When the fetal testis T Prod and gene transcript data were compared to the male reproductive malformation effects of DCHP (compiled from the literature as described in (Gray 2023)) the postnatal reproductive alterations were far less sensitive than the fetal genomic and endocrine measures (Figure 2).

Figure 2.

Dose related reductions in fetal testis mRNA and testosterone production (T Prod) on gestational day (GD) 18 following in utero exposure to DCHP compared with postnatal reproductive alterations in F1 male rats (compiled from the literature as previously reported in Gray 2023) anogenital distance (AGD), percent of 13-day old males with female-like areolae/nipples (NIPS), hypospadias, epididymal and testicular malformations, and testis histopathology (severity greater than minimal (score 1). Values are means with standard errors for mRNA expression and T Prod levels as percentage of control. The mean, SEs and sample sizes are provided in Supplemental files 6.

Examination of the ED₂₀ (± 90% CI) values for the postnatal effects reported in the literature and reductions in mRNA seen in the current study indicates that the genomic and T Prod reductions were more sensitive than were the postnatal effects (Table 2). The dose response and ED₂₀s for Cyp17a1 (Figure 3c) and Inha, Star, Insl3 and Scarb1 mRNA are slightly lower than the ED₂₀ for T Prod and are, on average, about 10-fold below the ED₂₀s for any apical Phthalate Syndrome effect on the F1 males.

Table 2.

DCHP effective dose 20% (ED₂₀) and 90% upper (CIU) and lower (CIL) confidence interval values for fetal testis mRNA expression and testosterone production (T Prod) on gestational day 18 compared to the postnatal alterations in F1 males. Values are estimates from GraphPad Prism 8.4 after fitting the data to four parameter logistic regression models (4PL).

4 PL LOGISTIC REGRESSION
DCHP INDUCED EFFECT	ED20	95% CIL	95%CIU
Cyp11b2	3.8	0.004	48.6
Lhcgr	7.7	1.6	25.7
Cyp11b1	9.7	3.0	27.3
Hsd3b	17.1	1.3	75.9
Cyp17a1	18.3	8.8	31.7
Inha	20.3	7.3	44.6
Star	20.8	2.0	69.6
Insl3	25.3	6.3	64.0
Scarb1	25.6	3.0	79.0
TPROD	42.0	14.5	74.6
Cyp11a1	75.9	29.7	134.2
Dhcr7	80.2	4.4	240.4
% of Males with areolae/nipples PND 13	285.6
% with Testis with Testis Histopatholgy	328.8
% with Gross Testis/Epididymal Malformations	369.0
AGD 2	461.9
% with Hypospadias	504.4

Figure 3.

A 10% increase in Phthalate Syndrome abnormalities in F1 male rats was associated with 53 to 82% reductions in T Prod (A), 34 to 61% reductions in CYP17a1 (B), and 58 to 82% reductions in CYP11a1 mRNA (rate limiting step in steroidogenesis; C).

Study 2, Experiment 1: Fetal effects of the TSCA mixture

Administration of the TSCA mixture from GD 14 to 18 had no effect on maternal weight gain, number of viable fetuses or resorptions (Table 3).

Table 3.

Maternal and fetal effects of gestation day (GD) 14 to 18 oral exposure to the five-phthalate mixture (TSCA mixture).

Total PE	Number of Fetuses			Number of Resorptions			GD18 Maternal Weight (g)			Maternal Weight Gain (g)
Dose	Mean	Std Error	N	Mean	Std Error	N	Mean	Std Error	N	Mean	Std Error	N
0	14.33	0.80	6	0.50	0.22	6	410.2	9.1	6	40.7	3.5	6
0.3	14.00	0.00	4	0.50	0.29	4	403.8	16.8	4	37.9	1.5	4
3	14.50	0.50	4	0.75	0.48	4	402.8	15.6	4	39.0	1.8	4
30	15.00	1.22	4	0.75	0.25	4	409.6	14.2	4	45.2	3.8	4
300	11.83	2.57	6	1.33	1.15	6	390.7	19.0	6	35.3	8.7	6
600	15.33	0.61	6	0.00	0.00	6	418.6	15.8	6	46.1	3.1	6

For analysis of the mRNA transcript data from the TSCA mixture study, the fold values for transcripts found on the new and the original custom gene arrays were averaged and the data analyzed using SAS 9.4 to determine LOELs/NOELs and with BMDExpress for BMDs/BMDLs/BMDUs.

The SAS analysis of the expression data from the combined arrays contained 99 target transcripts (S9). The fold data was analyzed for dose-related effects using PROC GLM using untransformed values. Twenty three of these genes were significantly affected (p<0.01 for one transcript, p<0.001 for 4 transcripts and p<0.0001 for 18 transcripts) including Nr5a1 (p<0.01), Nrob1 (p<0.006), Star (p<0.0001), Cyp11a1 (p<0.0001), Cyp11b2 (p<0.0001), Hsd3b (p<.0002), Cyp17a1 (p<0.0001), Lhcgr (p<0.0001), Scarb1 (p<0.0001), Insl3 (p<0.0001), Dhcr7, Tspo (p<0.0008), Cyp11b1 (p<0.0001), Inha (p<0.0001), Inhba (p<0.0006), Cyp51 (p<0.0001), Ebp (p<0.0001), Hmgcr (p<0.0001), Hmgcs1 (p<0.0001), Idil1 (p<0.0001), Mvd (p<0.0001), Testin (p<0.0001) and Tm7sf2 (p<0.0001). The Cyp11b2 expression was significantly reduced in all dose groups, Rhox10 was only reduced in the highest dose group whereas all other transcripts were significantly reduced at 300 and 600 mg/kg/d (Figure 4). Genomic_signature, an average of the transcripts that were reduced by the mixture, was significantly reduced at 300 and 600 mg/kg/d (F(5, 21df) = 41.4, (p<0.0001), R² = 0.91, CV = 11%). T Prod also was significantly reduced at 300 and 600 mg/kg/d (F(5,23 df) = 17.3 p<0.0001), R² = 0.79, CV = 22.7%).

Figure 4.

Effects of the TSCA mixture following gestational exposure (GD 14–18) on fetal rat testis gene mRNA expression levels (GD 18), testosterone production (TPROD) levels (GD 18), anogenital distance (PND 2), nipple retention (PND 13), and reproductive tract malformations (PND 65). The data used to calculate these values, the SEs and sample sizes are provided in Supplemental files 9 and 10.

In addition, the dose of the mixture that induced a 20% change in expression (ED₂₀) was estimated using logistic regression modeling. ED₂₀ values ranged from 5.7 to 437 mg total phthalates/kg (Table 4).

Table 4.

Fetal rat testis ED₂₀ and 95% lower and upper confidence intervals (CI) for gene mRNA and testosterone production (T Prod) on GD 18 and postnatal Phthalate Syndrome alterations after gestational exposure to the TSCA mixture (GD 14–18). Values are estimates from GraphPad Prism 8.4 using the “Find ECanything” model.

Transcript	ED20	ED20 95%CI
Cyp11b2	5.737	1.3 to 26
Cyp11b1	20.01	5.3 to 76
Star	68.99	30 to 157
Scarb1	87.68	38 to 200
Cyp17a1	98.32	43 to 226
Hsdb3	98.43	27 to 353
Genetic Signature	104.7	56 to 197
Tm7sf2	109.8	51 to 238
Ebp	130.2	54 to 314
Hmgcs1	133.6	63 to 283
Cyp11a1	136.1	77 to 242
T Prod	138.3	54 to 356
Insl3	143.6	44 to 474
Mvd	152	71 to 324
Ild1	160.1	80 to 321
Inha	164.3	74 to 367
Testin	168	111 to 259
Lhcgr	169.9	74 to 390
Cyp51	205.5	74 to 572
Hmgcr	275.1	149 to 509
Rhox10	436.9	226 to 843
Postnatal Phthalate Syndrome	ED20	ED20 95%CI
% of males with any areolae	289	wide
Number of areolae/nipples/male	300	291 to 371
% of males with reproductive malformations	422	312 to 572
AGD	456	327 to 636

The filtered transcript data analysis in BMDExpress identified 15 affected transcripts. The expression of all transcripts and T Prod were reduced with the exception of the Sertoli cell gene Testin which was increased by 5.4-fold (Table 5).

Table 5.

BMDExpress analyses of the effects of the TSCA mixture on fetal testis mRNA expression dose response models, benchmark dose (BMD), BMD lower bound (BMDL), BMD upper bound (BMDU), model fits (Rsquared and fit p value), no observed transcript effect level (NOTEL) and if the model displayed and all or none response below the lowest dose (Step Function Below the Lowest Dose).

										Model Is Step Function
	Best Model	BMD	BMDL	BMDU	BMD/BMDL	BMDU/BMDL	BMDU/BMD	RSquared	NOTEL	Below The Lowest Dose
Star	Hill	30	14.8	88.8	2.03	6.01	2.96	0.99	30	FALSE
Inhba	Hill	30.4	25.6	108.7	1.19	4.25	3.57	0.91	30	FALSE
Cyp17a1	Exp 3	57.1	42.8	102.7	1.33	2.4	1.8	0.99	30	FALSE
Dhcr7	Exp 5	66.3	8.5	282.8	7.8	33.29	4.27	0.96	30	FALSE
T PROD	Exp 3	81.5	59	177.2	1.38	3.00	2.17	0.98	30	FALSE
Testin	Linear	87.1	69.2	115.1	1.26	1.66	1.32	0.99	0	FALSE
Insl3	Exp 5	100.8	29.1	285.7	3.46	9.81	2.84	0.93	30	FALSE
Cyp11a1	Hill	111.3	31.9	263.7	3.49	8.28	2.37	1.00	30	FALSE
Inha	Linear	156.3	119.6	222.6	1.31	1.86	1.42	0.97	300	FALSE
Tm7sf2	Hill	165.9	31.9	257	5.21	8.07	1.55	0.99	30	FALSE
Ebp	Hill	188.6	42.8	264.2	4.4	6.17	1.4	0.99	30	FALSE
Idi1	Hill	213.2	47	268.1	4.54	5.71	1.26	0.98	30	FALSE
Hsd3b	Hill	222.4	32.4	283.3	6.87	8.75	1.27	0.98	30	FALSE
Lhcgr	Hill	228.7	41.4	276.7	5.53	6.69	1.21	0.97	30	FALSE
MVD	Hill	240.7	31	258.8	7.77	8.35	1.08	0.98	30	FALSE
Scarb1	Hill	241.3	31.2	257.8	7.74	8.26	1.07	1.00	30	FALSE

Study 2, Experiment 2: TSCA mixture one generation study

Administration of the TSCA mixture on GD 14 to 18 did not affect maternal weight or weight gain, or PND 2 pup bodyweight (Table 6). Anogenital distance (AGD) (adjusted for cube root of bodyweight with ANCOVA) at PND 2 was significantly reduced (p<0.003) at 600 mg/kg/d in male pups but not in female pups (Absolute AGD, Figure 5) (Table 6).

Table 6.

Effects of the TSCA mixture on maternal body weight at gestational day (GD 18), body weight gain (GD 14 to 18), post implantation loss (PIL = uterine implants-litter size) at postnatal day (PND) 23, litter sizes and pup body weights (neonatal (PND2), infant (PND13), weanling (PND23) and peri-pubertal (PND45)), and the percentage of males with full preputial separation (PPS) or genital malformations at PND 45. Note: Genital malformations precluded assessment of preputial separation (PPS).

EFFECTS OF IN UTERO ADMINISTRATION OF THE TSCA MIXTURE ON THE DAM AND F1 MALE RATS DURING PREGNANCY, NEONATAL, INFANTILE AND PUBERTAL LIFE STAGES
Dose in Total Phthalates mg/kg/d	0		0.3		3		30		300		600
n = 5 litters per Dose Group	Mean	SE	Mean	SE	Mean	SE	Mean	SE	Mean	SE	Mean	SE
Maternal weight gain during dose GD 14 to 18 (g)	39.1	3.2	43.5	4.2	45.9	6.7	40.6	4.6	44.2	1.1	30.0	5.5
Implantation Sites (at weaning)	14.0	0.9	14.0	0.5	12.2	1.6	10.6	1.4	12.6	0.5	12.2	1.0
Male AGD (mm)	3.83	0.17	4	0.11	3.94	0.15	4.1	0.11	3.79	0.17	3.29	0.16
Female AGD (mm)	1.73	0.07	1.88	0.07	2.02	0.14	1.83	0.05	1.87	0.10	1.85	0.10
Pup body weight at 2 days of age	7.92	0.37	8.34	0.40	8.48	0.43	8.64	0.27	8.76	0.20	8.45	0.35
Pup body weight at 13 days of age	27.2	1.21	31.3	1.09	31.3	2.17	31.9	2.52	33.2	1.5	32.5	1.76
Pup body weight at 23 days of age	62.4	2.1	70.8	1.7	69.6	4.3	70.4	5	74.0	3.9	70.5	3.7
Pup body weight at 45 days of age	241	7.7	257	6.3	256	10.8	262	13.9	285	18	256	12.3
Litter size at 2 days of age	13.6	1.08	12	0.84	11.2	1.46	11.2	1.02	11.4	0.4	10.4	1.78
Litter size at 13 days of age	12.6	1.29	11.4	0.98	11.0	1.41	10.6	1.44	9.6	1.5	8.8	1.24
Litter size at 23 days of age	12.0	1.3	11.0	0.6	10.6	1.4	10.6	1.4	9.6	1.5	8.4	1.2
Post-implantation loss (Implants - day 23 litter size)	2.0	0.8	3.0	0.3	1.6	0.7	3.0	1.3	3.0	1.8	3.8	1.1
Areolae/Nipples per male at 13 days of age	0.1	0.1	0.25	0.158	0.08	0.08	0.125	0.13	1.68	0.5	6.87	0.64
% of males with full Preputial Separations at 45 days	5	5	10	6.1	5	5	5	5	45	12	100	0
% of males with Genital malformationss at 45 days	0	0	0	0	0	0	0	0	0	0	33.3	9.8

bolded values differ significantly from the control group

yellow = p<0.01; blue = p<0.05

Figure 5.

In utero exposure to the TSCA mixture (GD 14–18)significantly reduced absolute anogenital distance (AGD) in F1 male (p<0.05 at 600 mg/kg/d) but not female rats at two days of age. Bars above the histograms are standard errors of the mean and the values are the AGD litter means in mm.

At PND 13 males displayed a significant increase in retained female-like areolae/nipples at 300 and 600 mg/kg/d (Figure 6A,B). In addition, litter sizes were reduced (Figure 6C) and body weights (Figure 6D) were significantly increased in a dose related manner (by linear regression). The increase in infant rat body weights on PND 13 was highly correlated with the reduction in litter size (p < 0.0001) which likely resulted from reduced competition during nursing in the smaller litters. For reference, T Prod was not reduced in the 0.3 to 30 mg/kg/d groups whereas at 300 and 600 mg/kg/d T Prod was reduced by 45 and 70%, respectively (Figure 4). Most transcripts showed similar dose-related reductions whereas Testin mRNA was increased by 3.2- and 5.4-fold at 300 and 600 mg/kg/d, respectively (Figure 4).

Figure 6.

Dose-related reproductive and developmental effects of the TSCA mixture following gestational exposure (GD 14–18) on (A, B) female-like areolae/nipple retention, (C) F1 litter size, and (D) male rat body weight at 13 days of age.

At weaning on PND 23, litter sizes were reduced in a dose-dependent manner (p<0.03 linear regression) and male rat bodyweights were slightly increased by 10% to 20% as compared to control weanlings (Table 6). The trend in increased body weight persisted throughout the study up to 65 days of age when they were necropsied (F value not significant, but slightly non-monotonic being significant by LSMEANS only at 300 mg/kg/d (p <0.05), but not 600 mg/kg/d). In addition, the age and weight at preputial separation (PPS) was not delayed by the mixture in males without genital malformations (data not shown).

At necropsy, 51% of the males in the highest dose group displayed reproductive malformations (Figure 7). Gross malformations were not seen in any other dose group. In addition, bilateral histopathological lesions (rating of moderate to severe) were seen in one of 25 males exposed to 30 mg total phthalates/kg/d and nine of 24 males in the highest dose group. Testis histopathological lesions were generally bilateral seminiferous tubule atrophy and degeneration noted in animals in the 600 mg/kg/d dose group. Degeneration was characterized by several features including disruption of the orderly arrangement of the germinal epithelium, cellular necrosis, loss of spermatocytes and spermatids, and the presence of multinucleated giant cells. Severe atrophy represented a testis without any remaining tubule germinal epithelium (S15).

Figure 7.

Effects of in utero exposure to the TSCA mixture on (A) the induction of reproductive tract malformations in the high dose (600 mg/kg/d), (B) dose-related effects on litter mean, and (C) individual testes weights in 65-day old F1 males.

The Phthalate Syndrome phenotypic effects that were modeled using EPA BMDS 3.3.2 software (BMR =1 SD) were the number of retained female-like areolae/nipples per male pup, percent of males/litter with retained female-like areolae/nipples, and the incidence of reproductive tract malformations. The dichotomous model, extra risk = 0.1, model average of six models that provided excellent fits and the BMD/BMDL/BMDU values for these biomarkers of demasculinization were 84/31/260 mg total phthalates/kg for the number of areolae/male (Hill model), 240/108/293 mg total phthalates/kg (four viable/alternative models) for the percent of males with areolae, 365/217/545 mg total phthalates for AGD (seven viable/alternative models) and 552/337/544 mg total phthalates/kg for the percent with reproductive tract malformations (seven viable/alternative with good visual fits). In summary, the BMDLs for several of the sensitive genomic effects ranged from 8 to 43 mg/kg/d and were about 3 to 10-fold below the BMDLs for most sensitive Phthalate Syndrome alterations.

The observed (OBS) effects of the mixture on AGD, nipple retention, hypospadias and epididymal and testis malformations were accurately predicted by dose-addition (DA) but not response addition (RA) mixture models. RA underpredicted the effects of the phthalates (Figure 8). We also calculated the quantitative relationship between the changes in fetal testis mRNA with the abnormalities seen in F1 male rats with the administered dose of the TSCA mixture (in total phthalates) (Figure 9).

Figure 8.

Observed effects of gestational exposure to the TSCA mixture on (A) AGD ED₂₀, (B) ED₅₀ values for nipple retention, (C) hypospadias, and (D) epididymal and testis malformations. Dose and response addition model predictions are shown below each figure along with the observed values. Observed = OBS (yellow circles), DA = dose addition (dashed red line), RA = response addition (blue solid line).

Figure 9.

Observed effects of gestational exposure to the TSCA mixture on fetal testis mRNA, as indicated by “genomic_signature” mRNA (the average of significantly down-regulated transcripts) with the postnatal abnormalities seen in male rat offspring (displayed as percent affected). Adverse effects were noted when the mRNA levels were reduced by about 55%.

Study 3: DPeP fetal and postnatal study data

SAS 9.4 Proc GLM analysis of the dose response fold values for the mRNA for several genes including the averaged transcript, the genomic_signature, were statistically significant (genomic_signature (p < 0.0001, CV = 10.8%), NrOb1 (p < 0.0001), Star (p< 0.0001), Cyp11a1 (p< 0.0001), Cyp11b2 (p< 0.0001), Hsd3b (p< 0.0001), Cyp17a1 (p< 0.0001), Lhcgr (p< 0.002), Scarb1 (p< 0.0001), Insl3 (p< 0.0001), Dhcr7 (p< 0.0001), Cyp11b1 (p< 0.0001), Rhox10 (p< 0.0001), Wnt7a (p< 0.0007), and Inha (p< 0.0001). In contrast, AR (p > 0.35), Esr1 (p > 0.13), Amhr2 and Esr2 (increased by 1.14 and 1.3 fold versus control, p < 0.02 and 0.03) respectively, Pparα (p > 0.25), Pparδ (p > 0.89), and Pparγ (p > 0.55) were not significantly reduced.

A summary of the BMDS, BMDExpress, and logistic regression analyses are shown in Table 7. The ED₂₀ was 76 mg/kg/d for the most sensitive postnatal effect (% of males with reproductive tract malformations), 16.3 mg/kg/d for reduced fetal T Prod, and 9.6 mg/kg/d for Scarb1 mRNA expression (Table 8). The BMD value for Scarb1 was 2.01 mg/kg/d with a BMDL of 0.75 mg/kg. BMDS BMD/BMDL/BMDU values for AGD at two days of age in neonatal male rats (continuous, BMR=1 SD, BMRF=1) were 62/47/128 mg/kg/d and 83/57/106 for reproductive tract malformations in adult males (dichotomous, BMR=0.1 extra risk), similar to the ED₂₀ values for these endpoints (Table 7).

Table 7.

BMDExpress analyses of the effects of the DPeP on fetal testis mRNA expression dose response models, benchmark dose (BMD), BMD lower bound (BMDL), BMD upper bound (BMDU), model fits (Rsquared and fit p value), no observed transcript effect level (NOTEL) and if the model displayed and all or none response (Step Function). ED₂₀ values for a 20% reduction are also shown (from GraphPad Prism model “FindECanything”.

		Best	Best	Best	Best	Best	Best	Best	Best	Best
Probe ID	ED20	Model	BMD	BMDL	BMDU	BMD/BMDL	BMDU/BMDL	BMDU/BMD	RSquared	Is Step Function
CVp11a1	17.8	Exp 5	6.49	1.99	16.69	3.26	8.39	2.57	1.00	FALSE
Cyp11b1	10.2	Exp 3	11.62	6.45	33.79	1.80	5.24	2.91	0.97	FALSE
Cyp11b2	24.1	Exp 3	37.31	22.71	75.26	1.64	3.31	2.02	0.92	FALSE
Cyp17a1	17.4	Exp 3	14.30	9.08	33.18	1.58	3.65	2.32	0.98	FALSE
genetic_signature	14.7	Exp 5	2.84	0.79	7.31	3.58	9.20	2.57	1.00	FALSE
Harlan SD T Prop	16.3	Exp 5	10.38	4.29	21.54	2.42	5.02	2.08	1.00	FALSE
Hsd3b	10.2	Exp 5	2.44	0.23	8.21	10.53	35.33	3.36	1.00	FALSE
Inha	19.7	Hill	8.27	3.69	18.93	2.24	5.13	2.29	0.97	FALSE
Insl3	18.6	Exp 5	4.35	1.70	12.59	2.57	7.43	2.89	0.99	FALSE
Lhcgr	34.6	Linear	96.43	69.58	154.63	1.39	2.22	1.60	0.92	FALSE
Nr5a1	21.0	Exp 5	4.06	0.37	5.14	11.10	14.04	1.26	0.93	FALSE
Rhox10	105.7	Linear	60.07	45.63	86.06	1.32	1.89	1.43	0.90	FALSE
Scarb1	9.6	Exp 5	2.01	0.75	4.59	2.67	6.11	2.29	1.00	FALSE
Star	13.3	Exp 5	7.15	2.34	17.25	3.05	7.36	2.41	0.99	FALSE

Table 8.

In utero exposure to dipentyl phthalate (DPeP) induces reproductive tract malformations in F1 adult male rats (Gray et al. 2016). ED₂₀ (with lower and upper 95% confidence intervals (CIL, CIU), from GraphPad Prism) and benchmark dose (BMDS) [with 95% lower (BMDL) and upper bounds (BMDU)].

GRAPHPAD PRISM ECF VALUES FOR DPEP				BMDS 3.3.2
4 PL LOGISTIC REGRESSION	ED20 %	95 % CIL	95 % CIU	BMD	BMDL	BMDU
Reproductive Tract Malformations	76	50.2	114.0	83	57	106
Gubernacular agenesis	90	34.9	230.9	64	44	88
AGD2	95	65.4	138.5	62	47	128
Undescended testes	106	9.3	1209.0
Gross Testis Abnormalities	107	48.8	235.9	90	54	130
Number of areolae/nipples/male PND1	109	68.5	173.9	31	22	40
Epididymal Agenesis	113	92.8	138.6	101	70	38
Epididymal weight reduction	134	101.4	177.6	128	87	248
Vas Deferens Agenesis	166	94.0	291.5
Testis Weight reduction	189	82.2	436.2
Hypospadias	195	42.3	896.0	158	89	215
Seminal Vesicle Agenesis	206	58.1	732.5	105	95	272
Ventral Prostate Abnormalities	381	145.3	999.3

In summary, the genomic and endocrine BMDs for the composite transcript genomic_signature was about 11 fold below the most sensitive measure of the Phthalate Syndrome, the number of retained female-like areolae/nipples/males at 13 days of age. BMDs for AGD, epididymal and testicular malformations and hypospadias (the least sensitive effect measured) were about 20, 30 and 55 fold below the BMD for the genomic_signature mRNA, respectively.

We also calculated the quantitative relationship between the changes in fetal testis mRNA with the abnormalities seen in F1 male rats with the maternal administered dose of (Figure 10). Adverse effects were noted when the mRNA levels were reduced by about 40% whereas a reduction of about 10% was not associated with any postnatal effects.

Figure 10.

Observed effects of gestational exposure to DPeP on fetal testis mRNA, as indicated by the “genomic signature” mRNA (the average of significantly down-regulated transcripts) with the postnatal abnormalities seen in male rat offspring.

DISCUSSION

In the current investigation, one objective was to derive BMD/BMDL and ED₂₀ values from three studies on the effects of PEs using targeted genomic mRNA and endocrine measures and compare these to the adverse reproductive alterations known as the Phthalate Syndrome in male rat offspring. The first study was an in utero DCHP dose response study, the second study was a mixture of 5 different phthalates (DBP, BBP, DiBP, DEHP and DCHP, referred to as the TSCA mixture) and included in utero and one-generation experiments with a the TSCA mixture, and the third study included previously published dose response data on DPeP. The phthalates in the TSCA mixture were selected from US EPA’s 2019 list of 20 “High-Priority substances to undergo risk evaluation under the Toxic Substances Control Act (TSCA)” and were included in EPAs Draft Technical Support Document for Cumulative Risk Analysis for phthalates (EPA 2023)(https://www.epa.gov/system/files/documents/2025-01/draft-phthalate-cumulative-risk-analysis-.-public-release-.-hero-.-dec-20-2024.pdf). In this regard, a second objective of the TSCA mixture study was to determine if the effects of the mixture on the F1 male rat offspring could be accurately predicted with dose-addition modeling.

Currently, alterations in gene expression have not gained wide regulatory acceptance as a method to establish PODs for hazard assessments of toxic substances and pesticides. PODs traditionally are based on adverse apical effects on growth, viability, abnormalities, organ weights and/or histopathology, for example. Recently, however, several regulatory agencies have used or proposed to use reductions in fetal testis testosterone levels to establish PODs and/or relative potency factors (RPFs) for diortho phthalate esters (EPA 2023; NICNAS 2012; 2013), (EFSA 2019). In a draft document on a proposed approach for cumulative risk assessment (CRA) of phthalates the USEPA considered several methods for establishing relative potency factors (RPFs) including fetal testis testosterone and gene expression levels (EPA 2023). The document states “EPA is considering several options to derive RPFs based on gestational (i.e., reduced fetal testicular testosterone content and reduced testicular steroidogenic gene expression, Options 1–4) and postnatal effects (i.e., AGD, NR, seminiferous tubule atrophy, hypospadias, Options 5 and 6)”. The document summarized the ED₅₀ values for T Prod and gene expression for six phthalates (Tables 3.4 and 3.6) (EPA 2023). However, EPA noted that “One challenge with developing RPFs based on gene expression data is that this type of data is typically not used to derive PODs for use in regulatory risk assessment. Generally, gene expression data are used by EPA as part of the weight of evidence analysis to support the human relevance of an effect or support a hypothesized MOA. However, transcriptomic dose-response modeling approaches that enable transcriptomic PODs to be calculated for use in risk assessment have been proposed, and this is an active area of research at EPA and NTP.” However, EPA subsequently derived draft RPFs for CRA of several phthalates under TSCA using fetal testicular testosterone data (EPA 2024b; 2024c).

In 2024, EPA (EPA 2024a) based the draft POD determination on a NOEL (10 mg/kg/d) for DCHP based upon “Phthalate Syndrome-related effects (e.g., ↓ fetal testicular testosterone; ↓ AGD; Leydig cell effects; ↓ mRNA and/or protein expression of steroidogenic genes; ↓ INSL3)” (Li et al. 2016). The BMDs/BMDLs in the current study were 17.7/9.3 mg/kg/d for T Prod in BMDExpress and 32.3/12 mg/kg with EPA’s BMDS software (BMR SD, BMRF 1), which are consistent with EPA’s draft proposed POD of 10 mg/kg/d for DCHP. Previously, EPA’s (2024) CRA draft document reported a BMD/BMDL (lower bound CI) for DCHP of 8.4/6 mg/kg for 5% extra risk and 17/12 for a 10% extra risk based upon meta-analysis of several studies (EPA Document #EPA-740-D-24–018, 2024, Table 4–20) (EPA 2024b).

As compared to the fetal testis genomic and endocrine effects of in utero DCHP, the ED₂₀ values for the postnatal Phthalate Syndrome alterations in the F1 male offspring were about 10- and 6-fold higher than the ED₂₀s for the genomic transcripts and T Prod, respectively. Similarly, the US CPSC reported that BMD₁₀/BMDL₁₀ values for DCHP for neonatal AGD were 101/68 and 130/79 mg/kg (2 studies), testis histopathology 169/91, and hypospadias 489/233 (from (Hoshino et al. 2005; Saillenfait et al. 2009; Yamasaki et al. 2009) (CPSC 2011) (https://www.cpsc.gov/content/Toxicity-Review-of-DCHP). In summary, the BMD/BMDL values for testis T Prod and gene expression are about 10 to 20-fold lower than those reported for these three reproductive alterations in F1 males. However the accuracy of the transcript BMD/BMDL/BMDU values is questionable because they are well below the lowest dose of DCHP of 100 mg/kg/d used in this experiment.

In contrast to the DCHP study, in the TSCA mixture study, the experimental design included doses that covered more than three orders of magnitude from 0.3 to 600 mg total phthalates/kg/d and the lowest dose was well below the transcript BMD values and the lowest dose (0.06 mg/kg of each PE) is below the PE values reported in several rodent diets, bedding and enrichment devices (Blystone et al. 2010; Jarosova et al. 2009).

Analysis of the TSCA mixture data using PROC GLM in SAS 9.4 identified 23 significantly affected transcripts from the two RT-qPCR arrays. The BMD/BMDL values for 18 of these transcripts that were adequately modeled using BMDExpress and T Prod ranged from 30 to 241 mg total phthalates/kg (BMD) and from 8 to 120 mg total phthalates/kg (BMDL). None of the BMDExpress models of the TSCA mixture transcripts were flagged as “step function below the lowest dose” and the BMD/BMDLs were all acceptable being within 3-fold below the lowest dose. Both of these “flags” were limitations with the results of the DCHP experiment (Table 5). An average fold change of the significantly affected transcripts (Genomic signature) was reduced by 45 % at 300 mg/kg and 3% at 3 mg/kg with an ED₂₀ of 112 mg/kg and BMD/BMDL values of 40/18 mg/kg. In the TSCA mixture study AGD, testis weight, testis histopathology, and the incidence of reproductive tract malformations were only significantly affected at the top dose group of 600 mg total phthalates/kg/d (Figure 5), whereas the LOEL/NOEL for the number of and percent of males with retained female-like areolae/nipples was 300/30 mg/kg (Figure 6).

In the DPeP study, the genomic and endocrine ED₂₀ and BMD values were roughly 3-fold lower than the postnatal ED₂₀ and BMD values (Table 6). Taken together, these results from all three studies indicate that PODs based on NOELs, ED₂₀s or BMDLs for the genomic and endocrine alterations of the fetal rat testis would be lower than PODs based upon Phthalate Syndrome abnormalities in male rat offspring.

Recently, the laboratory of Dr. D. Spade of Brown University published a similar study that evaluated the transcriptomic and histological effects of DPeP on the fetal testis, administered at 1, 11, 33, 100 and 300 mg/kg/d from GD 17 to 21 (Gupta et al. 2025). Ex vivo T Prod was measured, multinucleated germ cells (MNGs) were quantified, and fetal testes were used for RNA-seq, immunofluorescence, and in situ hybridization. The BMD estimates from BMDExpress2 based on RNA-seq indicated that DPeP altered gene sets related to steroidogenesis, gonadal development, epithelial cell differentiation, and vasculature development involving genes expressed in Sertoli, Leydig, endothelial, germ and progenitor cell types. BMD analysis was used to compare apical endpoints and gene expression. DPeP dose-dependently reduced T Prod and increased MNG density. The estimated BMD10 for MNGs was 2.68 mg/kg/d, (BMDL/BMDU = 0.79/ 6.7 mg/kg/d) which overlapped with 14.01 mg/kg/d (4.756, 33.303) for T Prod. Other datasets had BMD10 estimates between 8.96–14.6 mg/kg/d except for the seminiferous cord diameter, which had a BMD10 of 68.7 mg/kg/d. They reported BMD Media values for gene sets including GO:0120178, steroid hormone biosynthetic process; and GO:0006694, steroid biosynthetic process of 5.9 and 7.3 mg/kg/d. Individual gene BMD estimates for Scarb1, Cyp11a1, and Inha were 5.9, 7.3, and 3.4 mg/kg/d, respectively, which compares well with the BMD values for those transcripts in the current study from GD 18 fetal testes which ranged from 2 to 11 mg/kg/d (Table 7).

In conclusion, in utero exposure to PEs disrupts fetal testis genomic, proteomic (Ivell et al. 2022; Lehmann et al. 2004; Liu et al. 2005) and endocrine (T Prod and Insl3 hormones) key events in a unique AOP/MOA (EPA 2024c; Gray 2023; Gray et al. 2021). These key events are causally linked to the reproductive abnormalities in male rat offspring (Amato et al. 2022) and ED₂₀ and BMD/BMDLs values can be accurately quantified and used to develop PODs for phthalate risk assessment (as shown in the draft EPA document on cumulative risk for high priority phthalates (EPA 2023)). Developing PODs for risk assessment based upon the described genomic and endocrine signature would reduce and/or replace the need to conduct long-term, animal intensive, rat reproduction studies, conserving animals and other resources.

Supplementary Material

Appendix A.Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.taap.2025.117447.

Data availability

Data will be made available on request.