2. Materials and methods

2.1. Chemicals

Perfluorononanoic acid (PFNA, free acid, 97% pure, primarily lin-ear isomer according to supplier) and other chemicals used in thisstudy were purchased from Sigma–Aldrich (St. Louis, MO) unlessindicated otherwise. All other chemicals and reagents used for thisstudy were of analytical grade purchased from VWR (West Chester,PA). Stable isotope 1,2-13C9-labeled PFNA (13C9F17O2H, 99% pure)was supplied as free acid in methanol (50 g/ml) by Cambridge Iso-tope Laboratories (Andover, MA) and used as internal standard forliquid chromatography/tandem mass spectrometric (LC–MS/MS)analysis of PFNA. For all studies, PFNA was dissolved in deionizedwater at 60◦C with sonication, and prepared fresh daily.

2.2. Animals

All animal studies were conducted in accordance with the Insti-tutional Animal Care and Use Committee guidelines established bythe U.S. Environmental Protection Agency’s Office of Research andDevelopment/National Health and Environmental Effects ResearchLaboratory. Procedures and facilities were consistent with the rec-ommendations of the 1996 National Research Council’s “Guide forthe Care and Use of Laboratory Animals,” the Animal Welfare Act,and Public Health Service Policy on the Humane Care and Use ofLaboratory Animals. Animal facilities were controlled for tempera-ture (20–24◦C), relative humidity (40–60%) and kept under a 12-hlight–dark cycle.Timed-pregnant CD-1 mice were obtained from Charles RiverLaboratories (Raleigh, NC), where females were bred overnight,and the sperm-positive females, considered to be at gestationalday (GD) 0, were shipped to our facility on the same day. Uponarrival, mice were housed individually in polypropylene cages withAlpha-dri (Shepherd Specialty Papers, Kalamazoo, MI) bedding andprovided pellet feed (LabDiet 5008, PMI Nutrition International LLC,Brentwood, MO) and tap water ad libitum.

2.3. Treatment of animals

The study was performed in three separate blocks, with approx-imately 40 mice per block. Upon arrival at the animal facility, micewere weighed and randomly assigned to five treatment groups:vehicle control received deionized water, and other groups received 1, 3, 5, or 10 mg PFNA/kg body weight per day on GD 1–17. Mice were weighed and dosed by oral gavage at a volume of 10 ml/kgbody weight; pH of PFNA dosing solution was approximately 3.2 for 1 mg/ml (highest dose) and 3.4 for 0.5 mg/ml, compared to a pHof 5.4 for vehicle (deionized water). These dose levels were cho-sen to approximate the effective doses of PFOA previously used in our laboratory [50], taking into consideration the half-life of PFNA(estimated at 40 days in female mice).Maternal weights and signs of toxicity were monitored daily.At GD 17, mice from each dose group were assigned to one of twogroups: some dams were sacrificed for maternal and fetal examina-tions (received PFNA GD 1–16); the rest (received PFNA GD 1–17)were allowed to give birth. For animals sacrificed by decapitationon GD 17, trunk blood from each dam was collected and serumstored at −20◦C for subsequent analysis of PFNA. The gravid uteruswas removed, weighed and examined, and individual live fetuseswere weighed and prepared for teratological evaluation. Half of the fetuses from each litter were randomly assigned for skeletal exami-nation and the other half for visceral evaluation. Maternal liver andsome fetal livers were removed, weighed, and immediately frozenon dry ice and stored at −80◦C for PCR studies or PFNA analysis.Mice without live or dead fetuses were considered as non-pregnant.To confirm non-pregnancy, uteri were removed and stained withammonium sulfide to visualize implantation sites [51]. For damsallowed to deliver, mice were monitored on GD 18–19, at frequentintervals until parturition when the date and time of birth, numberof live and dead pups, and birth weights were recorded. Postna-tal survival, growth, and development of the pups were monitoredthereafter.

2.4. Skeletal evaluation

An overdose of pentobarbital was used to kill fetuses. Fetuseswere then eviscerated, and immediately fixed in 95% ethanol. Sub-sequently all specimens were stained with alizarin red S and Alcianblue for visual examination of bone and cartilage, respectively [52].Finally skeletal morphology was evaluated by the method of Narot-sky and Rogers [52].

2.5. Visceral evaluation

Fetuses were fixed in Bodian’s solution (2% formaldehyde, 5% acetic acid, 72% ethanol, 21% water), and a free-hand razor sec-tioning and dissection technique was used to examine the head,thoracic and abdominal viscera [53].

2.6. Postnatal observation, necropsy and sample collection

All litters were carefully observed daily for any morphologicalchange and weighed on postnatal day (PND) 1, 2, 3, 4, 7, 10, 14, 17,and 21. Eye opening was monitored daily beginning on PND 10. OnPND 24, pups were weighed, weaned, and males and females werehoused separately. Vaginal opening and preputial separation weremonitored beginning on PND 25. For post-weaning growth deter-mination, four male and four female pups were selected randomlyat PND 25 from each litter for biweekly weighing, and body weightswere recorded up to PND 287.After weaning, dams were weighed and necropsies were per-formed on the 28th day postpartum. Blood samples were collected,and serum was stored frozen at −20◦C for PFNA analysis. Liverswere removed, weighed, and stored at −80◦C for PFNA analysis.Uteri were removed and stained with 2% ammonium sulfide, andsubsequently the implantation sites were counted according to themethod of Narotsky and Kavlock [51].One male and one female from each litter were weighed and sac-rificed at PND 1, 10, 24, 42 and 70. Blood samples were collected,and serum was stored frozen at −20◦C for PFNA analysis. Liverswere removed, weighed, and about 200 mg (based on availability)of fresh liver tissue was stored in RNA-Later®(AM7021 Life Tech-nologies, Grand Island, NY) and processed for specific gene analysis;the rest of the liver was stored at −80◦C for PFNA analysis.

2.7. Determination of PFNA in serum and liver

The procedures for chemical analysis have been described pre-viously [33]. In brief, mouse livers were thawed, homogenized(polytron) in six volumes of picopure water, and stored at −80◦C.For sample preparation, 25 l of serum or 50 l of liver homogenatewere aliquoted into polypropylene tubes, to which 100 l of 0.1 Mformic acid and 1 ml of acetonitrile (ACN) fortified with 12.5 ng/mlof internal standard (13C9-labeled PFNA) was added sequentially,with vigorous mixing (vortex) between each addition. Sampleswere then capped, mixed for 20 min, and centrifuged (×3500 rpm) for 5 min at room temperature. Two-hundred l of the supernatantwas transferred to a HPLC auto sampling vial, followed by the addi-tion of 200 l of 2 mM ammonium acetate.Analysis of samples was performed using a LC/MS/MS sys-tem consisting of an Agilent 1100 HPLC (Agilent Technologies,Palo Alto, CA) interfaced with a Sciex 3000 triple quadrupolemass spectrometer (Applied Biosystems/MDS Sciex, Foster City,CA). The HPLC method was a 3.5 min isocratic run at a flowrate of 200 l/min consisting of 70% ACN and 30% 2 mM ammo-nium acetate. Samples (10 l) were injected onto a PhenomenexLuna C18 (3 mm × 50 mm) HPLC analytical column fitted with a4 × 30 mm Phenomenex Security Guard column (Torrance, CA). Thecolumn temperature was kept constant at 25◦C in a temperature-controlled column compartment. Electrospray negative ionization(−1500 V) was used in the mass spectrometer source, withturbo-dryer gas maintained at 375◦C. Mass transitions for PFNA(analyte) and13C9-PFNA (internal standard) were monitored usingmultiple-reaction-monitoring (MRM) and analyte-specific massspectrometer parameters were optimized for both compounds.Two mass transitions were monitored for quantification of theanalyte (PFNA). A quantification ion (463.0 m/z → 418.8 m/z) anda confirmation ion (463.0 m/z → 218.9 m/z), were monitored forPFNA in order to ensure proper analyte identification and quan-tification. The mass transition monitored for13C9-PFNA (internalstandard) was 472.0 m/z → 427.0 m/z. The limit of quantification(LOQ) for analysis was set to the lowest calibration point on thestandard curve. For samples below the LOQ (primarily control ani-mals), we assigned a value equal to the LOQ divided by the squareroot of 2 [54]. Analyst 1.4.2 software was used for quantification ofall compounds (Applied Biosystems/MDS Sciex, Foster, CA).

2.8. Analysis and evaluation of gene expression by real-time PCR

Transcripts evaluated by real-time RT-PCR included genesknown to be regulated by PPAR in addition to several cytochromeP450 genes commonly used to assess the activity of various othernuclear receptors. Additional genes included those associated withglucose and cholesterol metabolism, two functional categoriespotentially regulated by PPAR. Pax6 was evaluated as a nega-tive control since changes in the expression of this gene were notexpected in the fetal liver based on historical data from our labora-tory.Two micrograms of total RNA were digested using 2 unitsDNaseI (#M6101, Promega Corporation, Madison, WI) accordingto the manufacturer’s protocol. The RNA was then quantifiedusing a Quant-iT RiboGreen RNA assay kit according to themanufacturer’s protocol (#R11490, Life Technologies, GrandIsland, NY) and approximately 1.5 g RNA reverse transcribedusing a High Capacity cDNA Archive Kit according to the pro-vided protocol (#4322171, Life Technologies). Amplificationwas performed on an Applied Biosystems model 7900HT FastReal-Time PCR System in duplicate using 25 ng cDNA and Taq-Man Universal PCR Master Mix (#4304437, Life Technologies)in a total volume of 12 l. Beta-2-microglobulin (B2m, Entrez#12010) was used as the endogenous reference gene. The fol-lowing TaqMan assays (Life Technologies) were included inthe study: B2m (Mm00437762 m1), A2m (Mm00558642 m1),Acox1 (Mm00443579 m1), Amy1 (Mm00651524 m1),Cyp17a1 (Mm00484040 m1), Gck (Mm00439129 m1),Hmgcs1 (Mm00524111 m1), Pdk4 (Mm00443325 m1),Pxmp4 (Mm00480657 m1), Pax6 (Mm00443072 m1),Ehhadh (Mm00470091 s1), Cyp3all (Mm00731567 m1),Cyp1a1 (Mm00487218 m1), Lss (Mm00461312 m1), Pexlla(Mm00478137 m1), Mbl2 (Mm 00487623 m1), Cyp7a1(Mm00484152 m1), and Cyp2b10 (Mm00456591 m1). Data are presented as means and standard errors. Litter wasused as the statistical unit. Statistical significance of treatmenteffects was determined for each outcome variable using maineffects analysis of variance (ANOVA) that included block anddose as factors. When a significant treatment effect was detected,pairwise differences among groups were tested post hoc usingTukey–Kramer adjustments for multiple comparisons. For mea-surements taken across time, repeated measures analyses wereperformed with mixed effects linear models using SAS Proc Mixed(SAS/STAT 9.3, Cary, NC). The models included an autoregressivecovariance structure for data across time points. Fixed effects wereage, treatment and the treatment by age interaction, and blockwas included as a random effect. Pairwise tests among treat-ment groups within each age/time point were calculated withTukey–Kramer adjustments. For maternal weight gain the damwas considered the subject for the repeated measures; for pupweight, the pup (nested within dam) was the subject and damwas included as a random factor. Gene expression data were ana-lyzed using one-way ANOVA by evaluating the change in cyclethreshold for each gene relative to the reference gene (CT).Where a significant main effect was observed, individual treat-ment contrasts were evaluated for significance by a Student’s t-testof the least squares means (SAS JMP 7.0, Cary, NC). Fold changewas calculated using the 2−CTmethod of Livak and Schmittgen[55].The U.S. Environmental Protection Agency now uses the bench-mark dose (BMD) approach [56,57] for non-cancer risk assessment(U.S. EPA, 1995). This approach is designed to provide a morequantitative alternative to dose-response assessment than the no-observed-adverse-effect-level (NOAEL) process, by constructingmathematical models to fit all data points in the dose-responsestudy, and to take data variance into consideration. In the currentstudy, BMD5and BMDL5values were calculated for the maternaland developmental toxicity of PFNA using Benchmark Dose Soft-ware 2.2 (U.S. EPA, 2011). BMD5refers to the central estimate ofthe administered dose predicted to cause a 5% increase in responseabove background, BMDL5is defined as the corresponding lowerlimit of the 95% confidence interval on the BMD5[58]. Selection ofa specific curve-fitting model for the BMD determination was basedon the Akaike’s Information Criterion (AIC) value. The AIC is equalto −2L + 2p, where L is the log-likelihood at the maximum likeli-hood estimates for the parameters, and p is the number of modelparameters estimated. The model that demonstrates “goodness-of-fit” with the lowest AIC value is presumed to be the mostappropriate.

F1 CD-1 Mice