doi: 10.1210/en.2005-0778 Polychlorinated Biphenyls Exert Selective Effects on Cellular Composition of White Matter in a Manner Inconsistent with Thyroid Hormone Insufficiency

Developmental exposure to polychlorinated biphenyls (PCBs) is associated with a variety of cognitive deficits in humans, and recent evidence implicates white matter development as a potential target of PCBs. Because PCBs are suspected of interfering with thyroid hormone (TH) signaling in the developing brain, and because TH is important in oligodendrocyte development, we tested the hypothesis that PCB exposure affects the development of white matter tracts by disrupting TH signaling. Pregnant Sprague Dawley rats were exposed to the PCB mixture Aroclor 1254 (5 mg/kg), with or without cotreatment of goitrogens from gestational d 7 until postnatal d 15. Treatment effects on white matter development were determined by separately measuring the cellular density and proportion of myelin-associated glycoprotein (MAG)-positive, O4-positive, and glial fibrillary acidic protein (GFAP)-positive cells in the genu of the corpus callosum (CC) and in the anterior commissure (AC). Hypothyroidism decreased the total cell density of the CC and AC as measured by 4',6-diamidino-2-phenylindole dihydrochloride (DAPI) staining and produced a disproportionate decrease in MAG-positive oligodendrocyte density with a simultaneous increase in GFAP-positive astrocyte density. These data indicate that hypothyroidism reduces cellular density of CC and AC and fosters astrocyte development at the expense of oligodendrocyte density. In contrast, PCB exposure significantly reduced total cell density but did not disproportionately alter MAG-positive oligodendrocyte density or change the ratio of MAG-positive oligodendrocytes to GFAP-positive astrocytes. Thus, PCB exposure mimicked some, but not all, of the effects of hypothyroidism on white matter composition.

P OLYCHLORINATED BIPHENYLS (PCBs) are ubiqui- tous environmental contaminants that accumulate in animal tissues because of their lipophilic nature and chemical stability (1).Developmental exposure to PCBs in humans is associated with a variety of neuropsychological deficits such as lower IQ scores, motor impairments, deficits in visual recognition memory, and attention deficits (reviewed in Ref. 2).Recently, Stewart et al. (3) reported that the size of the corpus callosum (CC) in children is a strong predictor of the strength of association between PCB body burden and response inhibition.Although the mechanistic relationship between these variables is unknown, PCBs are known to interfere with thyroid function in animals (4,5), and thyroid hormone (TH) is known to be important in the development of white matter tracts in animals (6 -8) and in humans (9,10).Thus, it is possible that PCB exposure in humans causes specific cognitive deficits in part by interfering with TH action on the development of major white matter tracts such as the CC.
Considering this, we sought to test the hypothesis that PCB exposure selectively affects white matter tract development by interfering with TH action, focusing on the anterior commissure (AC) and CC.This hypothesis is based on the observation that TH controls several aspects of white matter tract development, including oligodendrocyte precursor cell proliferation and survival (reviewed in Ref. 11), oligodendrocyte number (6), and oligodendrocyte differentiation (12)(13)(14).Moreover, a specific PCB congener, PCB-118, can induce a dose-dependent increase in the number of oligodendrocytes in vitro, an effect that can be blocked with the TH receptor (TR) antagonist NH-3 (15).Thus, TH is important for white matter tract development, and PCB exposure may interfere with TH action in this tissue.
To test this hypothesis, we focused our experiments on postnatal day 15, a time when both myelination and TH levels are at their peak during development (16,17), and effects of hypothyroidism on white matter development have been documented (6,7,18,19).We began by evaluating the effect of developmental exposure to the commercial PCB mixture Aroclor 1254 on myelin-associated glycoprotein (MAG) expression in the CC and AC of male pups and comparing the observed effects with those of pups in which thyroid function was experimentally reduced.We further characterized treatment effects on oligodendrocytes by evaluating density and proportion of MAG-positive and O4positive oligodendrocytes within these white matter tracts and made comparisons with effects of hypothyroidism.Finally, because oligodendrocytes and astrocytes can be derived from a common precursor, we evaluated the effects of PCB exposure and hypothyroidism on glial fibrillary acidic protein (GFAP)-positive astrocyte density and proportion in these white matter tracts.We found that developmental PCB exposure significantly reduces cellular density and density of MAG-positive oligodendrocytes in the AC and CC but does not affect the proportion of MAG-positive oligodendrocytes, GFAP-positive astrocytes, or the density of O4-positive oligodendrocytes within these white matter tracts.These effects of PCB exposure are only partially consistent with the effects of hypothyroidism.Moreover, our findings are consistent with the hypothesis that TH controls the balance of MAGpositive oligodendrocytes to GFAP-astrocytes in these white matter tracts, perhaps by acting on a common precursor.

Materials and Methods Animals
All animal procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (20) and were approved by the University of Massachusetts-Amherst Institutional Animal Care and Use Committee.Timed pregnant Sprague Dawley rats (Zivic Miller Inc., Pittsburgh, PA) arrived at our animal facility 2 d after insemination [gestational d 2 (G2)].Upon arrival, animals were individually housed in plastic cages in a 12-h light, 12-h dark cycle (lights on 0600 -1800 h) with food and water provided ad libitum.

Experimental treatment
Dams were randomly assigned to one of four treatment groups (n ϭ 7-8 per group): control, hypothyroid (HTx), PCB-treated (PCB), or HTx plus PCB (HTxϩPCB).On G3, all animals began receiving a single untreated wafer at 1600 -1800 h daily.This initial exposure to untreated wafers trains the animals to consume them completely before PCB administration begins.Beginning on G7, wafers (Keebler Golden Vanilla Wafer-Mini, Kellogg Co., Battle Creek, MI) were dosed with Aroclor 1254 (lot no.A8120256; AccuStandard Inc., New Haven, CT) dissolved in methanol at a concentration of 5 g/l.To prepare the wafers daily, a volume equal to 1 l/g body weight was pipetted onto the wafer each morning and allowed to dry in a fume hood throughout the day.Wafers for animals in the control and HTx groups were dosed with methanol alone.Maternal PCB exposure continued from G7 to postnatal day 15 (P15).Dams in the HTx and HTxϩPCB groups were treated with a combination of methimazole (MMI) and potassium perchlorate in the drinking water.This treatment was initiated incrementally as follows: from G7-G9, 0.1% MMI plus 0.5% perchlorate; from G9 -G11, 0.1% MMI plus 0.1% perchlorate; and from G11 to P15, 0.02% MMI plus 0.5% perchlorate.On P15, one male pup from each litter was weighed and killed by decapitation after CO 2 exposure.Trunk blood was collected to measure total serum T 4 levels.The brain was removed, immediately frozen on pulverized dry ice, labeled, and stored at Ϫ80 C until sectioning in a cryostat.

RIA
Total T 4 was measured in 5 l of rat serum using a barbital buffer system as previously described (21).Each assay tube contained 100 l barbital buffer [0.11 m barbital (pH 8.6), 0.1% (wt/vol) 8-anilino-1napthalene-sulfonic acid ammonium salt, 15% bovine ␥-globulin Cohn fraction II, and 0.1% gelatin], 100 l anti-T 4 (rabbit; Sigma Chemical Co., St. Louis, MO) diluted to a final concentration of 1:30,000, and 100 l 125 I-labeled T 4 (ϳ15,000 cpm; Perkin-Elmer/NEN Life Science Products, Boston, MA).Standards were prepared from T 4 (Sigma) measured using a Cahn electrobalance; standards were run in triplicate, whereas samples were run in duplicate.Standards were calibrated to measure serum T 4 levels from 0.4 -25.6 g/dl.Tubes were incubated at 37 C for 30 min and then chilled on wet ice for 30 min.Bound counts were precipitated by adding 300 l ice-cold polyethylene glycol 8000 (20% wt/vol; Sigma).Tubes were centrifuged at 1800 ϫ g for 20 min at 4 C, and the supernatant was aspirated and the pellet counted in a ␥-counter (Packard Cobra II).
Pups with T 4 levels lower than the lowest standard (0.4 g/dl) were considered to have undetectable T 4 levels and not included in statistical analysis.The assay was run at 40 -50% binding; nonspecific binding was generally less than 8%.The assay was validated for rat serum by demonstrating parallelism between the standard curve and a dilution series of rat serum.

Experimental strategy
We initially examined the effect of PCB treatment and hypothyroidism on MAG expression because previous studies have shown that MAG mRNA levels are influenced by hypothyroidism (19,22).Thus, if PCB treatment affects oligodendrocyte development by producing a relative state of hypothyroidism, we predicted that PCB exposure would decrease MAG expression as evaluated by in situ hybridization.However, MAG is also a marker of differentiated oligodendrocytes (23), and the mRNA is localized to the perinuclear region of the soma (24).Therefore, we were able to simultaneously use MAG as a marker of terminally differentiated oligodendrocytes for analysis of oligodendrocyte cell number and density.However, this creates a potential confound in that the expected reduction in cellular levels of MAG mRNA may simultaneously be interpreted as a reduction in number or density of oligodendrocytes.Thus, we also evaluated the number of cells expressing the early marker of oligodendrocyte differentiation, O4.Moreover, because oligodendrocytes and astrocytes can be derived from a common precursor (12,25), we also evaluated treatment effects on similar characteristics of terminally differentiated astrocytes using GFAP as a marker.

In situ hybridization
Tissue.Frozen brain tissue was sectioned in a coronal plane at 12 m using a cryostat (Reichert-JungFrigocut 2800N; Leica Corp., Deerfield, IL).These sections were taken approximately 0.48 -1.0 mm caudal to Bregma (corresponding to Figs. [14][15][16].Sections were thawmounted onto twice-gelatin-coated slides and stored at Ϫ80 C until hybridization. Oligonucleotide and cRNA probe preparation.The probe used for detection of oligodendrocytes was generated by 3Ј-end labeling a 48-base oligonucleotide (IDT, Coralville, IA) directed against exon 8 of the MAG mRNA (5Ј-CAG GAT GGA GAC TGT CTC CCC CTC TAC CGC CAC CAC CGT CCC ATT CAC-3Ј; bp1116 -1163).Terminal deoxynucleotidyl transferase (TdT; Roche Applied Sciences, Indianapolis, IN) was used according to the manufacturer's directions with the following exception: the TdT labeling reaction contained a 10-fold molar excess of [ 35 S]dATP to 3Ј-ends to create an average tail length of 10 radiolabeled nucleotides per probe.Oligonucleotide probe preparation and specificity have previously been reported (24).For detection of astrocytes, a [ 33 P]cRNA probe for GFAP was generated in vitro from a cDNA fragment (bp 1365-2290, accession no.NM017009).The cDNA fragment was cloned into pCRII-TOPO using TOPO TA Cloning (Invitrogen, Carlsbad, CA) after RT-PCR using P15 cortical RNA as a template for the RT reaction and amplified using the following PCR primers: forward, 5Ј-AGCTC-CCTCCTCAGATAGTCTTGT-3Ј, and reverse, 5Ј-TGTGACTCTTCCAG-GTTGAGAAGC-3Ј.The authenticity of the PCR-generated fragment was confirmed by sequence analysis.To generate an antisense cRNA probe, pCRII-TOPO-GFAP was linearized with EcoRV and transcribed in the presence of T7 RNA polymerase.To generate a sense cRNA probe, the plasmid was linearized with HindIII and transcribed in the presence of SP6 RNA polymerase.The DNA templates were then removed with deoxyribonuclease digestion, and then the probes were purified by phenol/chloroform extraction followed by two ethanol precipitations.To compensate for differences in signal strength of MAG and GFAP mRNAs, different isotopic nucleotides were used for detection of mRNAs, allowing similar exposure times against x-ray film and photographic emulsion.
Pre-and posthybridization treatments.Prehybridization treatments were performed as previously described (27).Briefly, for each probe, two slides per animal were thawed and immersed in 4% formaldehyde/PBS for 30 min, rinsed in PBS, acetylated in 0.25% acetic anhydride/0.1 m triethanolamine/0.9%NaCl (pH 8.0) for 10 min, and rinsed in 2ϫ standard saline citrate (SSC) (1ϫ SSC contains 0.15 m sodium chloride and 0.015 m sodium citrate, pH 7.4).Slides were then dehydrated through a series of graded alcohols, delipidated in chloroform, rehydrated to 95% ethanol, and dried at room temperature.Fifty microliters of hybridization buffer containing 2 ϫ 10 6 cpm were applied to each slide, coverslipped, and incubated at 37 C (oligonucleotide) or 52 C (cRNA) for 16 h.Hybridization buffer contained 50% ultrapure formamide (Fisher Scientific, Fairlawn, NJ), 4ϫ SSC (oligonucleotide) or 2ϫ SSC (cRNA), 250 g/ml tRNA, 1ϫ Denhardt's, 10% (wt/vol) dextran sulfate, and 500 mm DTT.After the hybridization, coverslips were removed in 1ϫ SSC and washed four times in 1ϫ SSC for 15 min each at room temperature.For the oligonucleotide probe, sections were then washed four times with 2ϫ SSC/50% formamide at 40 C for 15 min each, rinsed two times in 1ϫ SSC for 30 min each, equilibrated to 70% ethanol, and then air dried.Posthybridization for the cRNA-GFAP probe was as follows: sections were washed two times with 2ϫ SSC/50% formamide at 52 C for 15 min each and then rinsed two times in 1ϫ SSC for 10 min each, incubated at 37 C for 30 min in RNase wash buffer (0.5 m NaCl, 0.01 m Tris, 1 mm EDTA) containing 100 g/ml RNase A, rinsed two times in 2ϫ SSC for 10 min each, followed by two more washes with 2ϫ SSC/50% formamide at 52 C for 15 min each, two more rinses in 1ϫ SSC for 10 min each, equilibrated to 70% ethanol, and then air dried.

Film autoradiography and signal quantification
After in situ hybridization, slides were arranged in an x-ray cassette and exposed to BioMax film (Eastman Kodak, Rochester, NY) for 24 h.To control for overexposure, 14 C-labeled standards (American Radiolabeled Chemicals Inc., St. Louis, MO) were included in the cassette and simultaneously exposed to the film.Radiographic signal was developed using an automated film processor (Konica SRX-101A; Alliance Imaging Inc., Warham, MA) and analyzed as follows.A magnified image was captured using a Nikon Macro lens mounted on a Dage-MTI-72 camera (Dage-MTI, Michigan City, IN) interfaced to a Macintosh computer operating NIH Image V 1.61 (W.Rashband, National Institute of Mental Health, Bethesda, MD) to control a Scion AG-5 capture board.Density values corresponding to the signal in the AC or CC were generated by encircling the entire AC unilaterally or by encircling the CC at the midline between the lateral margins of the lateral ventricles using the drawing tool in NIH Image (Fig. 1).The resulting density values from each brain region were averaged across all four sections for each animal and statistical analysis performed.

Single-cell mRNA measurements
After exposure to film, slides were removed from the x-ray cassette and dipped in NTB-2 photographic emulsion (Eastman Kodak), air dried, boxed in a light-tight container, and stored at 4 C for 72 h.Slides were then developed in Dektol (diluted 1:1 with water) for 2 min, rinsed in water, fixed in Kodak fixer for 5 min, and then rinsed in running cold water for 5 min, counterstained with 0.5% methyl green, and cover-slipped in Permount (Sigma-Aldrich, St. Louis, MO).Single-cell levels of MAG mRNA were determined as previously described by Zoeller et al. (28).Briefly, all slides were coded, randomized, and analyzed blindly.Images of the AC and CC were magnified ϫ200 with a Nikon ES-600 microscope equipped with a CCD camera (Dage-MTI) interfaced with a Macintosh G4 computer through an AG-5 frame grabber using NIH Image version 1.61.Images were captured using dark-field optics, and black and white polarization was electronically inverted.Using NIH Image, a threshold was set such that only silver grains were detected, and the area of silver grains associated with each of 10 single cells per brain region was determined by placing a fixed circle with an area of 0.002 mm 2 over each cell.The area over each of the 40 cells that was covered by silver grains in the AC and CC of each brain was used an index of single-cell mRNA levels (i.e.proportional to the amount of radioactive probe sequestered in the cell) (29).This method of measuring silver grain area has been demonstrated to closely resemble the results obtained from directly counting silver grains (30).After all brain sections were analyzed, the individual measurements from each slide were averaged for each animal and decoded, and statistical analysis was performed.

Cell density measurements
To determine the numerical density of oligodendrocytes or astrocytes present in the AC and CC, methyl green-stained nuclei associated with MAG-positive or GFAP-positive silver grain clusters were counted using a Nikon ES-600 microscope fitted with a 5-mm ϫ 5-mm ocular reticule.The number of MAG-positive nuclei and GFAP-positive nuclei in the CC was counted by randomly placing a reticule adjacent to the midline under ϫ400 magnification.At this magnification, the reticule represents 0.0144 mm 2 .The number of MAG-positive nuclei in the AC was determined by counting all MAG-positive nuclei within the entire AC in a digitized image captured at ϫ400 magnification using a mounted CCD camera.The total number of MAG-positive nuclei per AC was then normalized to square millimeters by measuring the area of the AC using the morphometric function of NIH Image.GFAP-positive nuclei in the AC were counted using the same strategy.
To determine the total cellular density in the AC, CC, and cingulate cortex, 4Ј,6-diamidino-2-phenylindole (DAPI)-stained nuclei were enumerated in the same sections as oligodendrocytes were counted.DAPI staining was performed by rinsing slides in PBS containing 1 g/ml DAPI for 30 min after removing coverslips in xylene and rehydrating through a series of graded alcohol.DAPI fluorescence was visualized using a Nikon Eclipse microscope equipped with a Spot RT Slider CCD camera (Diagnostic Instruments, Sterling Heights, MI) connected to a Windows-based computer.Individual ϫ400 magnified AC, CC, and cortical fields were digitized, and the numbers of DAPI-stained nuclei were counted either within a grid measuring 0.1 mm ϫ 0.11 mm (AC and CC) or 0.27 mm ϫ 0.38 mm at ϫ400 magnification.Two fields were counted in each slide of the CC, AC, and cingulate cortex.

O4 immunohistochemistry and quantification of O4-positive cells
To determine the numerical density of O4-positive cells within the AC and CC, adjacent tissue sections to those used in in situ hybridization experiments were fixed in freshly prepared 4% paraformaldehyde in PBS for 15 min followed by three washes of PBS for 5 min each.Slides containing brain sections were then blocked in 10% normal donkey serum in PBS at room temperature for 45 min and then incubated overnight at 4 C with anti-O4 monoclonal antibody (1 g/ml) (Chemicon, Temecula, CA) in PBS containing 1% normal donkey serum.The next day, sections were washed three times in PBS for 5 min each and then incubated at 37 C for 30 min with donkey antimouse fluorescein isothiocyanate-conjugated secondary antibody.Slides were then washed three times in PBS for 5 min each and then coverslipped in Vectashield mounting medium containing DAPI (Vector Laboratories, Burlingame, CA).Slides immunostained in the absence of primary antibody were used as negative controls.Stained sections were observed using a Nikon Eclipse microscope equipped with a Spot RT Slider CCD camera (Diagnostic Instruments) running MetaVue acquisition and analysis software (Molecular Devices, Sunnyvale, CA).The ϫ400 magnified images were analyzed for immunopositive cells by hand using the

Statistical analysis
Results were analyzed using a two-way ANOVA with PCB exposure and HTx treatment as the main effects unless otherwise stated.Post hoc tests, where appropriate, were performed by Bonferroni's t test.Bonferroni's t test uses the mean square error from the ANOVA table as a point estimate of the pooled variance [Graphpad Prism (ANOVA), San Diego, CA; http://www.graphpad.com/quickcalcs/(Bonferroni's post hoc and Grubb's test, respectively)].The Grubb's test was used on all data to identify statistical outliers.Statistical outliers were identified in some data sets, but the overall results were not altered by omission.A few samples were lost during processes; therefore, there are some unequal cell sizes.

Animals and T 4
Two-way ANOVA with repeated measures revealed a significant effect on maternal body weight of gestational day (F 8,248 ϭ 76.14; P Ͻ 0.001) but not of treatment (F 3,248 ϭ 1.463; P Ͼ 0.05) (Fig. 2A).However, there was a significant interaction among these two effects (F 21,248 ϭ 10.79; P Ͻ 0.0001), indicating that the effect of treatments were dependent upon gestational day.During lactation, both lactation day (F 4,124 ϭ 21.07; P Ͻ 0.0001) and treatment (F 3,124 ϭ 3.751; P Ͻ 0.05) affected maternal weight, but there was no interaction among these two main effects (F 12,124 ϭ 0.0623; P Ͼ 0.05) (Fig. 2A).
In contrast, there was a significant effect of treatment on maternal body weight gain during gestation (F 1,31 ϭ 41.18; P Ͻ 0.0001) (Fig. 2B) but not during lactation.Post hoc analysis of body weight gain revealed that HTx dams gained less weight during gestation, and this was unaffected by exposure to PCBs (Fig. 2B).
Treatment effects on maternal body weight gain was reflected in pup body weight on P15 (Fig. 2C).Two-way analysis of P15 pup body weight demonstrated an effect of HTx treatment (F 1,28 ϭ 79.64; P Ͻ 0.0001), but not of PCB exposure (F 1,28 ϭ 1.525; P Ͼ 0.05); there was no significant interaction between the two main effects (F 1,28 ϭ 0.037; P Ͼ 0.05).HTx and HTxϩPCB treatment groups exhibited lower body weights compared with controls (Fig. 2C).
Circulating levels of total T 4 were lower in pups derived from dams exposed to HTx and/or PCB; in fact, HTx animals had T 4 levels below the lowest standard of our assay (0.4 g/dl).Therefore, HTx and HTxϩPCB treatment groups were not included in the unpaired two-tailed t test used to demonstrate a significant difference between control animals and PCB-exposed animals (t ϭ 7.476; P Ͻ 0.0001) (Fig. 2D).

Treatment effects on MAG expression
Figure 3 shows the effects of HTx treatment and PCB exposure on MAG mRNA expression in the AC and CC (Fig. 3, A and B).Both HTx (F 1,24 ϭ 180; P Ͻ 0.001) and PCB exposure (F 1,24 ϭ 12.654; P Ͻ 0.001) significantly affected MAG expression in the AC; there was no significant interaction between these two main effects (F 1,24 ϭ 0.615; P Ͼ 0.05).Post hoc analysis revealed that the abundance of MAG mRNA was significantly lower in the AC of HTx animals compared with that of control animals.In addition, MAG mRNA levels were significantly lower in the AC of PCBexposed animals compared with controls, but not to the extent observed in HTx or HTxϩPCB animals (Fig. 3B, left).Likewise in the CC, two-way ANOVA revealed a significant effect of HTx (F 1,24 ϭ 1083; P Ͻ 0.0001) and PCB (F 1,24 ϭ 27.185; P Ͻ 0.001) and a significant interaction between these two main effects (F 1,24 ϭ 24.06; P Ͻ 0.0001).Post hoc analysis revealed that MAG mRNA levels in HTx or PCB-treated animals were significantly lower than in control animals (Fig. 3B, right).

Treatment effects on cellular levels of MAG mRNA
To determine whether the observed treatment effects on MAG expression identified by film autoradiography were a result of effects on cellular levels of MAG mRNA, we evaluated single-cell levels of MAG mRNA using liquid emulsion autoradiography (Fig. 4A).Cellular levels of MAG mRNA were significantly affected by HTx in both the AC (F 1,26 ϭ 352; P Ͻ 0.0001) and CC (F 1,26 ϭ 464; P Ͻ 0.0001) but not by PCB exposure (AC: F 1,26 ϭ 2.578; P Ͼ 0.05; CC: F 1,26 ϭ 0.015; P Ͼ 0.05), and there was no significant interaction between the two main effects (Fig. 4B).Cellular levels of MAG mRNA in both AC and CC were significantly reduced in HTx animals but were not affected by PCB exposure either in the presence or absence of HTx (Fig. 4B).

Treatment effects on MAG-positive oligodendrocyte density
Our finding that PCB exposure reduced the expression of MAG mRNA on film, but did not alter cellular levels of MAG mRNA, indicated that PCB exposure reduced the number of oligodendrocytes or their packing density in the AC and CC; therefore, we counted MAG-positive oligodendrocytes in the AC and CC to test this hypothesis.Figure 4C illustrates the effects of HTx treatment and PCB exposure on MAG-positive oligodendrocyte density.Both PCB exposure (F 1,25 ϭ 9.926; P ϭ 0.05) and HTx treatment (F 1,25 ϭ 151.4;P Ͻ 0.0001) significantly affected MAG-positive oligodendrocyte density in the AC, and there was a significant interaction between the two main effects (F 1,25 ϭ 7.707; P Ͻ 0.01) (Fig. 4C).In the CC, there was a significant affect of PCB exposure (F 1,25 ϭ 11.26; P ϭ 0.0025) and HTx treatment (F 1,25 ϭ 101.0;P Ͻ 0.0001) but no significant interaction (F 1,25 ϭ 3.270; P Ͼ 0.05) (Fig. 4C).In both the AC and CC, treatments reduced the density of MAG-positive oligodendrocytes.

Treatment effects on O4-positive oligodendrocyte density
To determine whether the effects of HTx treatment and PCB exposure on the density of MAG-positive cells in the AC and CC is selective to mature oligodendrocytes, we investigated oligodendrocyte precursor cells by counting O4-positive cells in the AC and CC. Figure 5 shows the effects of HTx treatment and PCB exposure on O4-positive cells.Two-way ANOVA demonstrated an effect of HTx treatment on the density of O4-positive cells in the AC (F 1,11 ϭ 39.44; P Ͻ 0.0001) and CC (F 1,10 ϭ 16.60; P Ͻ 0.001).Despite these findings, post hoc analysis demonstrated a significant reduction between control animals and HTx animals in the AC only (Fig. 5, B and D).There was a trend for an effect of PCB exposure in AC (F 1,11 ϭ 3.92; P ϭ 0.07) but not in the CC (F 1,10 ϭ 0.056; P ϭ 0.8174).Lastly, there was no interaction between the two main factors in the AC (F 1,11 ϭ 1.8; P Ͼ 0.05) or CC (F 1,10 ϭ 0.0008; P Ͼ 0.05).

Treatment effects on total cellular packing density
To test whether treatments selectively affected oligodendrocytes, we evaluated total cellular density in the AC and CC by counting DAPI-stained nuclei contained within a fixed area.Measuring total cell density verifies changes in markerspecific cell densities and provides a baseline in which to compare the changes in marker-specific cell densities.To determine whether treatment selectively targets white matter, we also measured cellular density in an area of gray matter, the cingulate cortex.Figure 6 demonstrates the effects of treatment on cellular packing density in the AC and CC.Two-way ANOVA demonstrated that PCB exposure produced a significant effect on cell density in the CC (F 1,26 ϭ 16.51; P Ͻ 0.001) but not in the AC (F 1,26 ϭ2.101; P Ͼ 0.05).In contrast, HTx exerted a significant effect on cell density in both the CC (F 1,26 ϭ 40.05;P Ͻ 0.0001) and AC (F 1,26 ϭ 36.37;P Ͻ 0.0001).There was no significant interaction between the two main effects in either the CC (F 1,26 ϭ 0.7142; P Ͼ 0.05) or AC (F 1,26 ϭ 2.844; P Ͼ 0.05) (Fig. 6, A and B).Post hoc analysis revealed that cell density in the CC and AC was significantly reduced in HTx and HTxϩPCB animals, but PCB-exposed animals exhibited a significant reduction in cell density in the CC only (Fig. 6, A and B).Cell density measurements in the cortex revealed no effect of HTx treatment (F 1,25 ϭ 0.751; P Ͼ 0.05) or PCB exposure (F 1,25 ϭ 0.966; P Ͼ 0.05) or a significant interaction between the two main effects (F 1,25 ϭ 1.732; P Ͼ 0.05) (Fig. 6C).

Treatment effects on GFAP expression
Considering that TH plays a role in oligodendrocyte differentiation from precursors that also give rise to astrocytes, it is important to determine whether treatments exerted specific effects on oligodendrocyte differentiation or whether astrocyte differentiation is also affected.Therefore, we examined the effect of treatment on GFAP-positive astrocytes, and these data are illustrated in Fig. 7.Both HTx (F 1,25 ϭ 20.79; P Ͻ 0.001) and PCB exposure (F 1,25 ϭ 8.974; P Ͻ 0.01) significantly affected GFAP expression, but there was no significant interaction between these two main effects (F 1,25 ϭ 0.41; P Ͼ 0.05).However, post hoc analyses showed that individual treatment groups were not significantly different from control but are different between PCB-exposed and HTx-treated animals (Fig. 7B, left).In the CC, HTx (F 1,25 ϭ 30.01;P Ͻ 0.001) exerted a significant effect, but there was no effect of PCB exposure (F 1,25 ϭ 1.1; P Ͼ 0.05) or of the interaction (F 1,25 ϭ 0.0001; P Ͼ 0.05).Both HTx and HTxϩPCB treatments significantly increased GFAP mRNA expression in the CC (Fig. 7, A and B, right).

Treatment effects on cellular levels of GFAP mRNA
Figure 7C illustrates the effects of treatment on cellular GFAP mRNA levels in the AC and CC.HTx treatment exerted a significant effect on cellular levels of GFAP mRNA in the AC (F 1,24 ϭ 15.49; P Ͻ 0.001) but not in response to PCB exposure (F 1,25 ϭ 0.147; P Ͼ 0.05), and there was no significant interaction between these two main effects (F 1,25 ϭ 0.412; P Ͼ 0.05).Post hoc analysis revealed that cellular levels of GFAP mRNA were significantly elevated in HTx-treated animals only (Fig. 7C, left).Similarly in the CC, there was a FIG. 7. The effects of PCB exposure and HTx treatment on GFAP expression, cellular levels of GFAP, and GFAP-positive cell density in the AC and CC.A, Representative film autoradiographs obtained after in situ hybridization; B, quantification of density measurements depicted as bar graphs showing the relative abundance of GFAP mRNA in the AC (left) and CC (right).PCB exposure had no effect on the expression of GFAP in the AC or CC.C, Bar graphs quantifying cellular GFAP mRNA levels in the AC (left) and CC (right).Dark-field images were threshholded and analyzed for silver grain area.PCB exposure had an effect on cellular levels of GFAP mRNA expression in the AC or CC.HTx treatment significantly elevated cellular GFAP mRNA levels in the AC and CC.Interestingly, HTxϩPCB cotreatment significantly elevated GFAP mRNA levels in the CC only.D, Bar graphs representing the mean density of astrocytes in the AC (left) and CC (right).PCB exposure had no effect on astrocyte density in the AC or CC.HTx treatment significantly increased astrocyte density in the AC and CC.Interestingly, HTxϩPCB cotreatment significantly elevated astrocyte density in CC only.Bars represent the means Ϯ SEM of raw density values (B), silver grain area per cell (C), and number of GFAP-positive cells per square millimeter (D).**, P Ͻ 0.01; *, P Ͻ 0.05 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).

Treatment effects on GFAP-positive astrocyte packing density
Figure 7D illustrates the effects of treatment on the density of GFAP-positive cells in the AC and CC.Treatment effects on GFAP-positive astrocyte cell density paralleled the effect of HTx treatment on cellular levels of GFAP mRNA in AC and CC.GFAP-positive cell packing density in the AC and CC was significantly affected by HTx (AC: F 1,25 ϭ 20.02; P Ͻ 0.001; CC: F 1,25 ϭ 50.27;P Ͻ 0.001), but no effect of PCB exposure (AC: F 1,25 ϭ 2.832; P Ͼ 0.05; CC: F 1,25 ϭ 0.981; P Ͼ 0.05) or a significant interaction between the two main factors (AC: F 1,25 ϭ 0.5; P Ͼ 0.05; CC: F 1,25 ϭ 0.919; P Ͼ 0.05) (Fig. 7D).GFAP-positive astrocyte density in the AC and CC was significantly increased in HTx animals; however, HTxϩPCB cotreatment significantly increased the density of GFAP-positive astrocytes only in the CC (Fig. 7D).

Treatment effects on the sum of oligodendrocyte and astrocyte packing density
If HTx and/or PCB exposure act on the common precursor giving rise to oligodendrocytes and astrocytes, then the sum of MAG-and GFAP-positive cells would not be altered by treatments.Figure 8 demonstrates that the sum of MAGpositive and GFAP-positive cell densities was affected by HTx treatment in the AC (F 1,24 ϭ 21.09; P Ͻ 0.001) but the not in the CC (F 1,26 ϭ 0.00001; P Ͼ 0.05).On the contrary, PCB exposure affected the sum of the densities for both the AC (F 1,24 ϭ 6.02; P Ͻ 0.05) and CC (F 1,26 ϭ 8.22; P Ͻ 0.01).There was no interaction among these main effects (AC: F 1,24 ϭ 0.419; P Ͼ 0.05; CC: F 1,26 ϭ 0.007; P Ͼ 0.05).However, post hoc tests demonstrated that the sum of MAG-positive and GFAP-positive cell densities was significantly reduced compared with controls only in the AC of HTx and HTxϩPCB (Fig. 8, A and B).
FIG. 8. Combined cell densities and ratios in the AC and CC.In the AC (A), note that the combined cell densities of oligodendrocytes and astrocytes is not significantly different from controls with PCB exposure.In the CC (B), no treatment effects on the sum of oligodendrocyte and astrocyte densities were detected.Note that in both the AC and CC, HTx and HTxϩPCB treatment have a virtually equal and opposite effect on the densities of oligodendrocytes and astrocytes.On the contrary, PCB exposure had a significant effect on oligodendrocyte density without altering astrocyte density.Cell density measurements were used to calculate the proportion of MAG-positive (C) and GFAP-positive (D) cells in the AC and CC.HTx treatment with or with cotreatment with PCBs reduced the proportion of total cells positive for MAG mRNA while concomitantly increasing the proportion of GFAP-positive cells in the AC and CC.No effect of PCBs on the proportion of MAG-positive or GFAP-positive cells was determined.a, Significantly lower compared with control; b, significantly increased compared with control; ***, P Ͻ 0.001 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).

Treatment effects on the proportion of oligodendrocytes and astrocytes in the AC and CC
We used our oligodendrocyte and cell density data to determine treatment effects on the proportion of oligodendrocytes in the AC and CC (Fig. 8, C and D).PCB exposure did not alter the proportion of oligodendrocytes occupying the AC (F 1,25 ϭ 1.983; P Ͼ 0.05) or CC (F 1,25 ϭ 0.550; P Ͼ 0.05).In contrast, HTx exerted a significant effect on the proportion of oligodendrocytes in both the AC (F 1,25 ϭ 66.33; P Ͻ 0.0001) and CC (F 1,25 ϭ 33.09; P Ͻ 0.0001) (Fig. 8C).Post hoc analysis demonstrated that HTx animals had a significantly lower proportion of MAG-positive oligodendrocytes in the AC and CC than control animals (Fig. 8C).Similarly, the proportion of GFAP-positive astrocytes in the AC and CC was significantly affected by HTx treatment (AC: F 1,27 ϭ39.74;P Ͻ 0.001; CC: F 1,27 ϭ 94.75; P Ͻ 0.001).HTx animals exhibited significantly greater proportions of GFAP-positive astrocytes in both the AC and CC compared with controls (Fig. 8D).No effect of PCB exposure on the proportion of GFAP-positive cells in the AC or CC was measured.

Discussion
Experimental studies indicate that PCB exposure may exert adverse effects on the developing brain by reducing circulating levels of TH, causing a state of relative hypothyroidism (31,32).This hypothesis is supported by the observation in animals that PCB exposure reduces serum TH (21,28,(33)(34)(35) and by the observation in some epidemiological studies that PCB body burden is negatively associated with measures of thyroid function (36 -39).In contrast, PCBs may also exert direct actions on the TR, independent of effects on serum TH levels (40,41).This hypothesis is based in part on the observation that PCBs can directly inhibit or enhance TR activity in vitro (42)(43)(44)(45)(46)(47).Moreover, several reports indicate that PCB exposure can exert TH-like actions in the developing brain (15,21,28,48).In the present studies, we tested the hypothesis that PCB exposure affects white matter tract development by interfering with TH action.Our results clearly demonstrate that developmental PCB exposure does not recapitulate the full effect of hypothyroidism on the cellular composition of white matter.
In the present study, developmental exposure to hypothyroidism produced an overall reduction in cell density (number of cells per unit area) in the genu of the CC and in the AC.In addition, hypothyroidism produced a disproportionately greater decrease in the density of MAG-positive oligodendrocytes in these tracts with a simultaneous increase in the density of GFAP-positive astrocytes.Furthermore, hypothyroidism reduced the density of O4-positive oligodendrocytes in the AC and CC.These effects were specific to white matter inasmuch as the cellular density of a single region of gray matter (cingulate cortex) was unaffected by hypothyroidism.Similarly, developmental exposure to PCBs produced an overall reduction in the total cell density of these white matter tracts, and this effect was selective inasmuch as the total cell density of cingulate cortex was unaffected by PCB exposure.However, PCB exposure did not affect the relative proportion of MAG-positive or GFAPpositive cells.Thus, PCB exposure mimicked some, but not all, of the effects of hypothyroidism on white matter development.
The finding that hypothyroidism induced a disproportionate decrease in the density of MAG-positive oligodendrocytes and a simultaneous and proportional increase in GFAP-positive astrocytes (Fig. 8) is consistent with the hypothesis that these two cell types are derived from a common precursor in the AC and CC (49 -52) and that TH acts on these precursor cells to trigger oligodendrocyte differentiation at the expense of astrocyte differentiation (12,14,53,54).Moreover, our finding that the sum of the densities of astrocytes and oligodendrocytes in the CC was not altered by hypothyroidism (Fig. 8) further supports this interpretation.These findings are consistent with previous work showing that TH insufficiency decreases oligodendrocyte numbers in vivo (6,55) and increases astrocyte numbers in white matter of the cerebellum (56,57).Our present data indicate that a potentially important role of TH in white matter tract development is to control the ratio of MAG-positive oligodendrocytes to GFAP-positive astrocytes in two major white matter tracts.
We observed that the effects of HTx treatment on singlecell levels of both MAG and GFAP mRNAs are mirrored in our MAG-positive and GFAP-positive cell density measurements.Specifically, HTx treatment reduced cellular levels of MAG mRNA and concomitantly the density of MAG-positive cells.Likewise, cellular levels of GFAP mRNA increased in HTx-treated animals as did the density of GFAP-positive cells.Thus, it is possible that our cell density measurements reflect the number of cells that reach a threshold level of detection for each marker rather than accurate MAG-positive and GFAP-positive cell density measurements.Although this interpretation is plausible, there are two elements of the current study that suggest the effects of TH on cellular levels of MAG and GFAP mRNAs are not a confounding variable.First, MAG mRNA levels are quite abundant on P15, and even the least intensely labeled cell is far above the threshold of detection.Second, we found that the density of cells positive for a second, independent marker of oligodendrocytes, the O4 antigen, was simultaneously reduced in the AC and CC of HTx animals.These findings are more consistent with the interpretation that oligodendrocyte differentiation was reduced in these HTx animals.
In contrast, we found that PCB exposure had no effect on the density of O4-positive cells in the AC or CC.The finding that PCB exposure did not reduce O4-positive cells is contrary to the finding that PCB exposure did reduce MAGpositive oligodendrocytes in the AC and CC.Oligodendrocyte differentiation proceeds through a defined series of antigen acquisition that characterizes various stages of the oligodendrocyte lineage, with the O4 antigen being expressed in the oligodendrocyte lineage at an earlier stage than MAG (58).Thus, our observation that PCB exposure reduced the density of MAG-positive cells, but not O4-positive cells, suggests that PCB exposure may selectively affect oligodendrocyte differentiation at a stage that differs from that of hypothyroidism , specifically between acquisition of O4 and MAG expression.
We were somewhat surprised that the effects hypothyroidism on packing density were selective to the AC and CC and unaffected in the cingulate cortex, because early reports indicated that cortical packing density was increased in HTx animals (59).This discrepancy may be attributable to technical differences in animal treatment or to technical differences in the method of analysis.Specifically, earlier reports used the DNA content of wet tissue as a measure of cell packing density; in contrast, we physically counted DAPIstained nuclei in the cingulate cortex.It is likely that our method of physically counting nuclei is a more accurate measurement.However, it is premature to assume that cell density in all gray matter regions remain unaffected in HTx or PCB-exposed animals.
Previous studies have clearly demonstrated the necessity of sufficient TH for proper myelination of white matter tracts (6,19,60).In the present study, we used a dose of PCB exposure that reduced circulating levels of TH by about 80% (from 5.85 Ϯ 0.38 to 1.37 Ϯ 0.42) in P15 pups.Crofton has reported that a 60% reduction in serum total T 4 is associated with a PCB-induced hearing loss in rats (31); thus, our procedure produced a reduction in serum T 4 that was greater than that associated with other adverse effects in the developing nervous system.However, we found that PCB exposure caused a reduction in total cell density of the CC and reduced MAG-positive cells in AC and CC, similar to that of hypothyroidism, but it did not alter the ratio of MAG-positive oligodendrocytes to GFAP-positive astrocytes or disproportionately reduce the density of MAG-positive oligodendrocytes in either white matter tract investigated.Thus, the PCB-induced reduction in circulating levels of T 4 did not fully recapitulate the effects of goitrogen-induced hypothyroidism, indicating that developmental exposure to PCBs did not affect developing white matter by causing a state of relative hypothyroidism.
It should be noted that in the present study, our goitrogen treatment reduced T 4 levels below the lowest standard of our assay.Thus, because of the severely depressed T 4 levels, synergistic or additive effects of PCBs and hypothyroidism may not be detectable.Consequently, we may not have been able to differentiate effects resulting from hypothyroidism from those that may result from the addition of PCBs.
Developmental exposure to PCBs can increase the expression of TH-responsive genes in the developing brain despite significantly reducing circulating T 4 (21,28,48).In addition, Fritsche et al. (15) have shown that PCB-118, a congener known to rapidly accumulate in brain tissue (61), can induce oligodendrocyte differentiation in a human neural progenitor cell line by acting on the TR.Interestingly, sustained neonatal hyperthyroidism leads to a reduction in the expression of myelin-associated genes (62), although it is not clear whether this observation is linked to changes in oligodendrocyte number or proportion relative to astrocytes or total numbers of cells.Considering these findings, it is possible that PCB exposure can act as an inappropriate TH-like signal during early development, ultimately causing a reduction in the numbers of cells present in these white matter tracts.However, in our current experiments, we failed to observe that PCB exposure could ameliorate the effects of hypothyroidism.Therefore, these observations are inconsistent with the hypothesis that PCBs act as TR agonists on white matter.However, it should be noted that our ability to detect TH-like effects of PCBs in hypothyroid animals may have been com-promised by the severity of TH insufficiency induced in HTx-treated animals.
It is important to recognize that PCBs may be acting on white matter development by a mechanism that does not include TH signaling.In fact, the findings reported here are largely consistent with the idea that the effects of PCBs on white matter composition are not the result of altered TH action.Some PCB congeners can bind to the ryanodine receptor, affecting calcium signaling (63)(64)(65), and calcium signaling may be important in oligodendrocyte precursor cell differentiation (66).In addition, PCBs can induce reactive oxygen species in neuronal cultures that ultimately leads to cell death (67), and oligodendrocytes are sensitive to elevated levels of reactive oxygen species (68).Regardless of the mechanism by which PCB exposure selectively reduces the cellular density of these white matter tracts that we have observed in the present experiments, it is a potentially important observation.Because the proportion of oligodendrocytes is reduced relative to gray matter, it is possible that fewer axon fibers are myelinated in these white matter tracts, and this may be true in humans as well as in animals.The size of the CC is reported to be altered in magnetic resonance imaging studies on children diagnosed with disorders associated with impulse behaviors and cognitive deficits including attention deficit hyperactivity disorder, Tourette's, and autism (69 -74).Moreover, exposure to environmental chemicals has been linked to these disorders (75,76).Finally, a recent report has documented that the size of the CC is a strong predictor for the association between cord blood levels of PCBs and response inhibition (3).Thus, it is possible that effects of PCB exposure on the cellular composition of white matter tracts may underlie some neurological deficits detected in children exposed to PCBs.
Developmental exposure to polychlorinated biphenyls (PCBs) is associated with a variety of cognitive deficits in humans, and recent evidence implicates white matter development as a potential target of PCBs.Because PCBs are suspected of interfering with thyroid hormone (TH) signaling in the developing brain, and because TH is important in oligodendrocyte development, we tested the hypothesis that PCB exposure affects the development of white matter tracts by disrupting TH signaling.Pregnant Sprague Dawley rats were exposed to the PCB mixture Aroclor 1254 (5 mg/kg), with or without cotreatment of goitrogens from gestational d 7 until postnatal d 15.Treatment effects on white matter development were determined by separately measuring the cellular density and proportion of myelin-associated glycoprotein (MAG)-positive, O4-positive, and glial fibrillary acidic protein (GFAP)-positive cells in the genu of the corpus callosum (CC) and in the anterior commissure (AC).Hypothyroidism decreased the total cell density of the CC and AC as measured by 4,6-diamidino-2-phenylindole dihydrochloride (DAPI) staining and produced a disproportionate decrease in MAG-positive oligodendrocyte density with a simultaneous increase in GFAP-positive astrocyte density.These data indicate that hypothyroidism reduces cellular density of CC and AC and fosters astrocyte development at the expense of oligodendrocyte density.In contrast, PCB exposure significantly reduced total cell density but did not disproportionately alter MAG-positive oligodendrocyte density or change the ratio of MAG-positive oligodendrocytes to GFAP-positive astrocytes.Thus, PCB exposure mimicked some, but not all, of the effects of hypothyroidism on white matter composition.(Endocrinology 147: 846 -858, 2006)

FIG. 1 .
FIG. 1. Regions analyzed in the current study.This film autoradiographic image of MAG expression in the P15 rat brain shows the areas of the CC, AC, and cingulate cortex (box) where gene expression and cell densities were measured (arrows).The areas of the CC, AC, and cortex were defined using the rat brain atlas of Paxinos and Watson (26).Other brain regions are labeled to give orientation and positioning of coronal sections.CPu, Caudate putamen; CTx, cortex; EC, external capsule; LO, lateral olfactory tract; LV, lateral ventricle; ON, optic nerve.

FIG. 2 .
FIG. 2. Treatment effects on maternal and pup body weight and pup T 4 levels.A-D, Treatment effects on maternal body weight (A), maternal body weight gain during pregnancy and lactation (B), P15 pup weight (C), and total T 4 levels in pups on P15 (D).Developmental PCB exposure was not associated with a decrease in maternal or pup body weights.In contrast, HTx and HTxϩPCB-treated animals exhibited significantly lower maternal body weight gain during pregnancy and reduced pup body weights on P15.Both HTx and PCB treatment were associated with significant reductions in serum T 4 in P15 pups (D).Data points (A) and bars (B-D) represent mean Ϯ SEM.**, P Ͻ 0.01; ***, P Ͻ 0.001 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).

FIG. 3 .
FIG. 3. Effects of PCB and HTx treatment on MAG mRNA in the P15 rat brain.A, Representative film autoradiographs obtained after in situ hybridization.Lower panels are higher magnifications of the CC and AC (inset) of the upper panels.B, Quantification of density measurements depicted as bar graphs showing the relative abundance of MAG mRNA in the AC (left) and CC (right).Statistical analysis revealed that PCB exposure significantly reduced MAG expression in the AC and CC.Similar results were seen in animals treated with goitrogens (HTx).Bars represent the mean Ϯ SEM of density values.*, P Ͻ 0.05; ***, P Ͻ 0.001 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).

FIG. 4 .
FIG.4.Treatment effects on cellular levels of MAG mRNA levels and MAG-positive oligodendrocyte density in the AC and CC.A, Dark-field micrographs of AC (top) and CC (bottom) after in situ hybridization for MAG mRNA using liquid photographic emulsion reveals the effect of treatments on cellular levels of MAG expression as well as the density of oligodendrocytes occupying these regions.Note the reduction in reflective puncta (single cells) with PCB exposure and HTx treatment.The AC and CC are outlined in HTx-treated animals for clarity.B, Quantification of relative cellular levels of MAG mRNA using silver grain area from dark-field images in A are shown for the AC (left) and CC (right).Note that hypothyroidism was associated with a significant reduction in the cellular levels of MAG mRNA in the AC and CC but that PCB exposure was not associated with altered cellular levels of MAG mRNA.C, Quantification of MAG-positive cells in AC (left) and CC (right).Bars represent the mean Ϯ SEM for silver grain area (B) or MAG-positive oligodendrocytes/mm 2 (C).*, P Ͻ 0.05; ***, P Ͻ 0.001 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).

FIG. 5 .
FIG. 5. Treatment effects on O4-positive oligodendrocyte density in the AC and CC.A and C, Representative micrographs of the AC (A) and CC (C) after immunostaining for the O4 antigen.The inset in A shows an enlargement demonstrating O4 staining.The border of the subventricular zone that outlines the lateral ventricle is outlined in C to provide orientation and clarity.B and D, Quantification of the density of O4-positive cells in the AC (B) and CC (D) demonstrates that HTx treatment with or without cotreatment with PCBs significantly reduced O4-positive cells in the AC.Two-way ANOVA analysis demonstrated that HTx treatment affected O4-positive cells in the CC (F 1,10 ϭ 16.60; P Ͻ 0.001), but post hoc analysis failed to demonstrate significance between controls and treatment.No effects of PCBs on O4-positve cells were observed.Scale bar, 200 m (A) and 50 m (B).***, P Ͻ 0.001 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).

FIG. 6 .
FIG. 6. Treatment effects on cell packing density in the AC, CC, and cingulate cortex.DAPI staining was used to visualize nuclei in the same sections that MAG-positive oligodendrocytes were quantified.A-C, Bar graphs corresponding to the number of nuclei in the AC (A), CC (B), and cingulate cortex (C).These results show that HTx and HTxϩPCB treatments significantly reduced the density of cells in AC and CC.PCB exposure significantly reduced cell density in the AC but not in the CC (A and B).No treatment effects were seen in the cingulate cortex (C).Bars represent the means Ϯ SEM of the number of nuclei per square millimeter.**, P Ͻ 0.01; ***, P Ͻ 0.001 (significantly different from control group using Bonferroni's t test after a two-way ANOVA).