Developmental Programming: Impact of Prenatal Testosterone Excess on Insulin Sensitivity, Adiposity, and Free Fatty Acid Profile in Postpubertal Female Sheep

Abstract

Prenatal T excess causes reproductive and metabolic disruptions including insulin resistance, attributes of women with polycystic ovary syndrome. This study tested whether increases in visceral adiposity, adipocyte size, and total free fatty acids underlie the insulin resistance seen in prenatal T-treated female sheep. At approximately 16 months of age, insulin resistance and adipose tissue partitioning were determined via hyperinsulinemic euglycemic clamp and computed tomography, respectively, in control and prenatal T-treated females. Three months later, adipocyte size and free fatty acid composition were determined. Results revealed that at the postpubertal time points tested, insulin sensitivity was increased, visceral adiposity and adipocyte size in both the sc and the visceral compartments were reduced, and circulating palmitic acid was increased in prenatal T-treated females relative to controls. In parallel studies, 20-month-old prenatal T-treated females tended to have increased basal insulin to glucose ratio. Relative to earlier findings of reduced insulin sensitivity of prenatal T-treated females during early life and adulthood, these findings of increased insulin sensitivity and reduced adiposity postpubertally are suggestive of a period of developmental adaptation. The disruption observed in free fatty acid metabolism a few months later correspond to a time point when the insulin sensitivity indices of prenatal T-treated animals appear to shift toward insulin resistance. In summary, current findings of improved insulin sensitivity and reduced visceral adiposity in postpubertal prenatal T-treated sheep relative to our earlier findings of reduced insulin sensitivity during early postnatal life and adulthood are indicative of a period of developmental adaptation.

Polycystic ovary syndrome (PCOS) is a major infertility disorder of unexplained chronic hyperandrogenism, oligoanovulation, and/or polycystic ovaries (1–4). Most women with PCOS also develop insulin resistance and are at a higher risk for developing type 2 diabetes, among other metabolic disturbances. Weight loss seems to improve insulin sensitivity and reproductive outcomes suggestive of a role for adiposity in determining the severity of the derangements of these women (5–7). Although visceral fat appears to be more prevalent in PCOS women, this is not apparent from body mass index to fat mass ratio analysis (8). Because most studies in PCOS include overweight and obese subjects, the actual impact of PCOS on visceral adiposity is unclear. Interestingly, recent studies looking at visceral fat deposition have found either no difference between PCOS women and control subjects (9) or the lean PCOS women to have lower fat mass compared with controls (10).

Animal models are great resources in the understanding of the developmental origins of human disease. The exposure of sheep (11, 12) and monkeys (13) to excess T during fetal life culminates in a reproductive and metabolic phenotype that resembles that of women with PCOS. At the metabolic front, prenatal T treatment induces insulin resistance early in life (14) and tissue-specific changes in insulin signaling in sheep (15). Visceral adiposity has been implicated in the development of insulin resistance (16, 17). Insulin resistance and visceral adiposity appear to go hand in hand in prenatal T-treated female monkeys because they manifest impaired insulin sensitivity (18) and increased adiposity (19), but this is not the case in rats (20). Visceral adiposity appears to accompany insulin resistance in women with PCOS (9, 21, 22). It is unclear whether visceral adiposity also accompanies insulin resistance in prenatal T-treated sheep.

Other studies have found that increased visceral adiposity correlates not only with insulin resistance but also with dyslipidemia (23–25). For instance, circulating cholesterol and triglycerides levels are increased in prenatal T-treated rodents (20). This raises the possibility that abnormalities in lipolysis of specific adipose tissue depots may mediate the interplay between adiposity and insulin resistance (26). Evidence exists to indicate that defective lipolysis can lead to an increase in free fatty acid release, which in turn can lead to the development of insulin resistance (27, 28). In support of this, total free fatty acid levels were found to be elevated in monkeys treated prenatally with T (29). Because prenatal T treatment also induces insulin resistance in female sheep (14) as in prenatal T-treated monkeys (18), we hypothesized that prenatal T-treatment would induce visceral adiposity, increase adipocyte size in the visceral depot, and increase total free fatty acids in parallel with development of insulin resistance. This is of relevance because such disruptions may underlie development of insulin resistance in this animal model of PCOS phenotype (12).

Materials and Methods

Animals, prenatal treatments, and adipose tissue harvest

Details of prenatal T treatment, husbandry, and nutrition have been described previously (30). In brief, pregnant Suffolk sheep were treated twice weekly with 100 mg of T propionate im (Sigma-Aldrich Corp, St Louis, Missouri) in cottonseed oil (2.0 mL) from 30 to 90 days of gestation (term ∼ 147 days). When twin births were involved, only 1 offspring from each mother was used in this study. Newborn weights were recorded at birth. All lambs were fed a pelleted diet (Shur-Gain, Elma, New York) comprised of 3.6 MCal/kg digestible energy and 18% crude protein. After weaning at 8 weeks of age, female lambs were maintained outdoors at the Research Facility (Ann Arbor, Michigan; 42°, 18′N). They were fed ad libitum until they attained 40 kg body weight, at which time they were switched to a diet with 15% crude protein until 6 months of age (Shur-Gain). Adult sheep were fed a ration consisting of 2.3 MCal/kg digestible energy and 11.3% crude protein.

At approximately 16 months of age, during the anestrus season, weights were recorded, and insulin sensitivity was determined in all females (control: n = 6; prenatal -treated: n = 6) by hyperinsulinemic euglycemic clamp and adipose tissue distribution by computed tomography (CT). Three months later, at approximately 19 months of age, to avoid potential influence from cycle stage, 2 20-mg prostaglandin F_2α (PGF_2α; 5 mg/mL, im; Lutalyse; Pfizer Animal Health, Kalamazoo, Michigan) injections were administered 11 days apart to synchronize estrus. A blood sample was collected during the follicular phase 24 hours after the second PGF_2α injection for determination of free fatty acid (FFA) composition. At approximately 21 months of age, all animals were resynchronized as described above. During the follicular phase, at 27 hours after the second PGF_2α injection, females were euthanized by administration of a barbiturate overdose (Fatal Plus; Vortech Pharmaceuticals, Dearborn, Michigan), and sc and visceral adipose tissues were procured.

Hyperinsulinemic euglycemic clamp and basal insulin to glucose ratio

All animals were fasted for 48 hours. Bilateral indwelling jugular catheters (BD Angiocath; Beckton Dickinson and Co, Franklin Lakes, New Jersey) were placed; one serving for infusion and one for sample collection. Insulin solution (Novolin R; Novo Nordisk Inc, Princeton, New Jersey) was infused at a constant rate of 4.0 mU/kg body weight per minute for 3 hours. Insulin solution was supplemented with 2.5% of potassium chloride to prevent hypokalemia. Glucose [25% (wt/vol) iv, 50% dextrose-sterile solution; Hospira Worldwide Inc, Lake Forest, Illinois] was infused, starting 15 minutes after insulin infusion. Blood samples were collected every 5 minutes, starting 15 minutes prior to insulin infusion for assessment of glucose with a glucometer (Accu-Check; Roche Diagnostics Corp, Indianapolis, Indiana). The glucose infusion rate was adjusted based on these 5-minute readings to restore and maintain euglycemia and mean glucose infusion rate normalized to body weight. In addition, blood samples were collected at 10-minute intervals in heparinized sodium fluoride/potassium oxalate tubes and stored frozen for subsequent measurement of insulin and glucose. Measures from the first 3 samples (before insulin infusion samples) were averaged for the determination of basal fasting levels of insulin and glucose. The insulin sensitivity index was calculated as the mean of the glucose infusion rate/ln (mean basal fasting insulin) as previously reported (31).

Basal insulin to glucose ratio levels were determined in 17 control and 20 prenatal T-treated 20-month-old females from 3 earlier animal cohorts subjected to a similar prenatal T treatment paradigm. All females were fasted for 48 hours, and fasting blood samples were collected. Plasma concentrations of glucose and insulin were determined and insulin to glucose ratio calculated. Plasma glucose levels were measured by the glucose oxidase method (Pointe Scientific, Inc, Canton, Michigan). The inter- and intraassay coefficients of variation (CVs) for glucose measured at 50 and 200 mg/dL were both less than 5.0% (n = 10 assays). Plasma insulin levels were measured using a RIA kit (MP Biomedicals, Orangeburg, New York). The sensitivity of the insulin assay was 2.7 ± 0.4 μU/L (n = 3 assays; mean ± SEM). Mean intraassay CVs based on 2 quality control pools measuring 38.4 ± 0.8 and 131.2 ± 2.8 μU/L were 5.1 ± 0.6% and 7.8 ± 0.7%, respectively. The corresponding interassay CVs averaged 9.9% and 10.7%.

Adipose tissue distribution

Adipose tissue distribution was assessed in all females by CT using a multislice CT scanner (16 slice Brightspeed; General Electric, Indianapolis, Indiana) at the Diagnostic Imaging Service at the Veterinary Teaching Hospital of Michigan State University. Animals were sedated with a combination of 5% guaifenesin in 5% dextrose (Michigan State University VTH Pharmacy, East Lansing, Michigan; 25–50 mg/kg, iv) ketamine hydrochloride (Fort Dodge, Iowa; 3–4 mg/kg, iv), and diazepam (Hospira; 0.1–0.2 mg/kg, iv) and administered to allow intubation with a cuffed endotracheal tube. The sheep were maintained under general anesthesia with sevoflurane (Abbott, North Chicago, Illinois; 1%–5%) administered in 100% oxygen using a precision vaporizer and a circle rebreathing system.

Animals were placed on the CT scanner in sternal recumbency with their forelegs bent and their hind legs extended. CT scans were performed from the level of the caudal skull until the ischium. The transverse (axial) scan allowed identification of skeletal landmarks for use in generating the multiplanar CT images. Two scans were performed, one using a 1.25-mm slice thickness and the other at 5 mm. Other scan parameters were constant among all animals including 120 kVp and 300–330 mA. Images were constructed using a standard algorithm. Adipose tissue assessment was restricted to the area from the 10th thoracic vertebrae to the first lumbar vertebrae (mean number of slices assessed per animal: 115.5 ± 0.6). Recognition of the adipose tissue was based on an attenuation range of −50 to −150 Hounsfield units. Once the adipose tissue was localized, the CT scan images were imported into an imaging software (Analyze 6.0; AnalyzeDirect, Overland Park, Kansas) for quantification. Differentiation of the sc (outside peritoneal cavity) from the visceral (inside peritoneal cavity) compartment was performed manually using the peritoneal cavity, ribs, vertebrae, and muscular fascia as landmarks. Serial 1.25-mm slices from identified areas were used for calculation of adipose and sc tissue volume.

FFA measures

FFA measurement involved the following steps: 1) lipid extraction, 2) preparation of methyl ester and purification, 3) analysis of nonesterified free fatty acids of plasma samples, and 4) gas chromatography (GC) of fatty acid methyl esters. In brief, total lipids from the fat samples were extracted following the Bligh and Dyer method of solvent partition (32). The methyl esters were extracted with hexane and then purified by thin-layer chromatography and identified with respect to the retention flow of the standard. The fatty acid compositions were then analyzed by GC. Analysis of nonesterified free fatty acids of plasma samples was performed using a direct method of transesterification as described previously (33). Hexane is used to extract the methyl esters and then purified by thin-layer chromatography and analyzed by GC. GC of fatty acid methyl ester was performed on an Agilent GC machine (6890N model; Agilent Technologies, Santa Clara, California) equipped with a flame ionization detector, an auto sampler, and a Chemstation software for data analysis. Hydrogen was used as a carrier gas as well as for flame ionization detector and nitrogen was used as a makeup gas.

A calibration curve was prepared using proportional amounts of C17:0 methyl ester standard. A mixture of standard methyl esters was also run to identify the components in unknown samples by comparing their retention times. The FFAs were quantified with respect to the amounts of C17:0 internal standard added and the calibration curve prepared. Intraassay CVs were 0.9% and 1.1% and interassay coefficients of variation were 2.6% and 3.3%. These measures were carried out through the Molecular Phenotyping Core of the Nutrition Obesity Research Center at the University of Michigan. Desaturation indices were calculated as the ratio of the sum of the unsaturated to sum of saturated FFAs with the same number of carbons (ie, C16:1 to C16:0).

Adipocyte morphometry

Adipose tissue was cut into small pieces and suspended in a 15-mL Falcon tube containing 12 mL of 1× PBS supplemented with 4% bovine serum albumin and 1 mg/mL of collagenase A (MP Biomedicals). The samples were incubated in an orbital shaker for 1 hour at 40°C. After complete dissociation, adipocytes were filtered through a nylon mesh (250 μm). The upper layer of the suspension containing the adipocytes was reconstituted in fresh medium. Adipocyte suspension was then transferred to a siliconized glass slide and covered with a siliconized coverslip. Adipocyte images were captured under bright-field illumination [Leica DMR microscope (Heidelberg, Germany) and Diagnostic Instruments Spot RT camera (Sterling Heights, Michigan)]. All images were captured using the same magnification and camera settings. Uniform microspheres (diameter 98.0 μm; Bangs Laboratories, Fishers, Indiana) were used as a reference. The mean cell diameter, area, and volume were determined by computerized image analysis (Image Pro Analyzer version 7.01; Media Cybernetics, Bethesda, Maryland). An estimate of the number of adipocytes in each fat depot was calculated by dividing total fat volume by the mean adipocyte volume (of each of the 2 fat depots) calculated using the following formula: 3/4 · π · r³, being r the radius.

Statistical analysis

The impact of prenatal T treatment on variables from insulin clamp, sc, visceral, and total adipose tissue depot, and free fatty acid profile were analyzed using a Student's t test. The effect of prenatal T treatment on fasting insulin to glucose ratio was analyzed using an univariate general linear model and adjusted for the 3 cohorts used. Appropriate transformations were applied for all variables to account for heterogeneity of variances. To assess the impact of prenatal T treatment on adipocyte size, an empirical cumulative distribution function was also calculated for each measurement and the difference between both groups tested using a 2-sample Kolmogorov-Smirnov test. The difference between the ratios of large vs small adipocytes between groups was tested using a permutation test with 10 000 iterations. Significance was defined as P < .05. All results are presented as mean ± SEM. All analyses were carried out using PASW Statistics for Windows release 18.0.1 (IBM, Chicago, Illinois).

Results

Effects of prenatal testosterone on insulin sensitivity

Figure 1 shows the mean glucose infusion rate and insulin sensitivity index between 90 and 180 minutes from the start of the hyperinsulinemic euglycemic clamp. The mean glucose infusion rate was higher in prenatal T-treated sheep compared with controls (Figure 1A; P < .05). The mean insulin sensitivity index assessed between 90 and 180 minutes from start of glucose infusion tended also to be higher in prenatal T-treated sheep compared with the controls (Figure 1B; P = .056). Assessment of basal insulin to glucose in the 20-month-old cohorts showed that the mean (±SEM) basal insulin to glucose ratio tended to be higher in prenatal T-treated females compared with the controls (control: 0.36 ± 0.02 vs prenatal T-treated: 0.45 ± 0.04; P = .062) (see Figure 4).

Figure 1.

A, Mean (±SEM) glucose infusion rate from 90 to 180 minutes of the hyperinsulinemic euglycemic clamp are shown on the left and mean glucose infusion rate calculated by averaging values from 90 to180 minutes on the right. B, Mean (±SEM) insulin sensitivity index from 90 to 180 minutes during the hyperinsulinemic euglycemic clamp is shown on the left and the overall mean (±SEM) across this time period on the right. *, Significant differences (P < .05). t, trend ranging between P < 0.06 and P < 0.07. C, control; T, treated.

Figure 4.

A, Developmental changes in the insulin to glucose ratio. Difference in basal insulin to glucose ratio (expressed as a percentage) between control and prenatal T-treated females at 1.2, 4, 5, 7.5, 14, 20, and 22 months of age in control and prenatal T-treated females. Except for the data point at 20 months of age derived from this study, all other data were computed from previously published papers (14, 35). The indent graph shows the insulin to glucose ratio (mean ± SEM) of 20-month-old control (open) and prenatal T-treated (closed) females. Prenatal T treatment was from day 30 to day 90 gestation for all time points except the 22-month time point, which involved T treatment from day 60 to day 90 of gestation. B, Comparison of directionality of changes in insulin homeostasis (gray boxes) and visceral adiposity (open boxes) after prenatal exposure to T in sheep (our studies) with outcomes from perinatal T/DHT exposure in other species pre- or perinatal T/DHT-treated mice (36), rats (20, 37–39), cows (40), rhesus monkeys (18, 19, 41), and women with PCOS (10, 42–46). Note the differences in age and reproductive status of animals undergoing the tests. I/G, insulin to glucose ratio; ↑ IS, increased insulin sensitivity; ↓ IS, reduced insulin sensitivity; NC-IS, no change in insulin sensitivity. Superscripts within boxes represent the reference source.

Effects of prenatal T excess on visceral adiposity

Representative images of CT scans from 2 control and 2 prenatal T-treated sheep are shown in Figure 2A. Although birth weight of prenatal T-treated females was lower compared with controls (control: 5.8 ± 0.3 vs prenatal T treated: 3.7 ± 0.1 kg; P < .05; Figure 2B), mean weight of animals at the time of scan was similar between groups (control: 76.9 ± 0.9 vs prenatal T treated: 76.0 ± 2.7 kg; Figure 2C). Mean total fat (control: 8.7 ± 0.6 vs T treated: 5.9 ± 0.7 dm³; P < .05; Figure 2D) and visceral fat (control: 5.3 ± 0.3 vs prenatal T treated: 3.4 ± 0.4 dm³; P < .05), but not sc fat (control: 3.3 ± 0.3 vs prenatal T treated: 2.5 ± 0.3 dm³; P = NS) volume was lower in prenatal T-treated females compared with controls. The ratio of visceral to sc fat also tended to be lower in prenatal T-treated females relative to controls (control: 1.6 ± 0.05 vs prenatal T treated: 1.4 ± 0.1; P = .07; Figure 2E).

Figure 2.

A, Representative CT scans from 2 control and 2 prenatal T-treated females. Visceral and sc adipose tissue depots are represented in purple and blue, respectively. B, Birth weight (mean ± SEM) of control (open bars) and prenatal T-treated females (closed bars). C, Body weight (mean ± SEM) of control (open bars) and prenatal T-treated females (closed bars) at the time of the CT scan. D, Subcutaneous, visceral, and total adipose tissue volume (mean ± SEM) from control (open bars) and prenatal T-treated females (closed bars). E, Visceral to sc adipose tissue ratio (mean ± SEM) of control (open bars) and prenatal T-treated females (closed bars). *, Significant differences (P < .05). BW, body weight; SC, subcutaneous.

Effects of prenatal T on FFA metabolism

Table 1 shows the profile of circulating free fatty acids in 19-month-old females. Circulating levels of the unsaturated palmitic acid were higher in prenatal T-treated females compared with the control group. No differences were found in any of the other circulating FFAs, the desaturation indexes, or the total saturated, mono-, and polyunsaturated free fatty acids. Table 2 shows the profile of total fatty acids in the visceral adipose depot. A higher number of fatty acids species, specifically the long-chain fatty acids, were detected in adipose tissue compared with plasma. However, no differences were found between the treatment groups in any of the fatty acids of total saturated, mono-, and polyunsaturated fatty acids in the visceral adipose depot.

Table 1.

Mean (±SEM) Circulating FFA (Nanomoles per Milliliter) and Desaturation Indices of Control and Prenatally T-Treated Females at Approximately 19 Months of Age

Analyte, nmol/mL	C	T	P value
Palmitic acid (C16:0)	2.28 ± 0.28	3.92 ± 0.58	.032
Palmitoleic acid (C16:1)	0.18 ± 0.06	0.23 ± 0.03	.401
Stearic acid (C18:0)	6.33 ± 1.45	9.21 ± 1.51	.215
Oleic acid (C18:1n9)	3.15 ± 1.05	4.42 ± 1.09	.438
α-Linolenic acid (C18:3n3)	0.14 ± 0.03	0.23 ± 0.04	.144
Arachidonic acid (C20:4)	0.33 ± 0.10	0.44 ± 0.10	.484
Docosapentaenoic acid (C22:5)	0.06 ± 0.00	0.09 ± 0.03	.731
Nervonic acid (C24:1)	0.04 ± 0.01	0.04 ± 0.00	.990
Total FFA	15.13 ± 3.70	20.71 ± 3.66	.323
Saturated FFA	9.39 ± 2.09	13.22 ± 2.07	.233
Unsaturated FFA	5.74 ± 1.63	7.48 ± 1.64	.484
Monounsaturated FFA	3.30 ± 1.13	4.72 ± 1.18	.421
Polyunsaturated FFA	2.45 ± 0.54	2.77 ± 0.49	.677
Desaturation index (C16)	0.05 ± 0.01	0.06 ± 0.01	.290
Desaturation index (C18)	0.72 ± 0.09	0.65 ± 0.08	.586

Abbreviation: C, control.

Table 2.

Mean (±SEM) Total FFA (Nanomoles per Milligram) in Visceral Adipose Tissue of Control and Prenatally T-Treated Females at Approximately 19 Months of Age

Analyte, nmol/mg	C	T	P value
Myristic acid (C14:0)	0.96 ± 0.47	0.12 ± 0.05	.988
Myristoleic acid (C14:1)	0.12 ± 0.5	0.17 ± 0.07	.612
Palmitic acid (C16:0)	21.85 ± 5.77	23.47 ± 1.66	.761
Palmitoleic acid (C16:1)	1.82 ± 0.62	1.65 ± 0.20	.810
Stearic acid (C18:0)	20.93 ± 5.17	28.99 ± 1.98	.131
Cis-vaccenic acid (C18:1n7)	0.54 ± 0.20	0.58 ± 0.23	.915
Oleic acid (C18:1n9)	49.89 ± 13.98	50.99 ± 3.42	.942
Linoleic acid (C18:2)	1.68 ± 0.49	2.23 ± 0.27	.318
α-Linolenic acid (C18:3n3)	0.254 ± 0.768	0.421 ± 0.067	.134
γ-Linolenic acid (C18:3n6)	0.189 ± 0.058	0.209 ± 0.034	.750
Arachidic acid (C20:0)	0.029 ± 0.013	0.013 ± 0.005	.222
Gadoleic acid (C20:1)	0.191 ± 0.633	0.192 ± 0.039	.989
Dihomolinoleic (C20:2)	0.028 ± 0.007	0.021 ± 0.004	.398
Arachidonic acid (C20:4)	0.097 ± 0.036	0.087 ± 0.016	.792
Erucic acid (C22:1)	0.003 ± 0.002	0.001 ± 0.001	.528
Adrenic acid (C22:4)	0.005 ± 0.005	0.015 ± 0.003	.142
Docosapentaenoic acid (C22:5)	0.017 ± 0.007	0.027 ± 0.007	.360
Lignoceric acid (C24:0)	0.037 ± 0.015	0.031 ± 0.005	.668
Nervonic acid (C24:1)	0.024 ± 0.011	0.009 ± 0.002	.264
Total FFA	98.69 ± 25.83	110.08 ± 5.75	.687
Saturated FFA	43.82 ± 10.80	53.45 ± 2.19	.362
Unsaturated FFA	54.87 ± 15.27	56.62 ± 3.71	.541
Monounsaturated FFA	52.59 ± 14.62	53.60 ± 3.41	.342
Polyunsaturated FFA	2.28 ± 0.65	3.01 ± 0.38	.527

Abbreviation: C, control.

Effects of prenatal T on adipose tissue morphometry

The mean (±SEM) number of cells analyzed per animal in visceral and sc adipose tissue was 928.5 ± 62.4 and 1041.5 ± 7.5, respectively. The cell size distribution in prenatal T-treated females was shifted toward lower diameter in both adipose tissue depots (Figure 3). The mean (±SEM) diameter of the visceral adipose tissue was lower (P < .0001) in the prenatal T-treated females (91.0 ± 0.2 μm) compared with the control females (102.6 ± 0.3 μm). Mean (±SEM) diameter and volume of the sc adipose tissue (72.7 ± 0.2 μm³ and 4.9·10⁵ ± 0.5·10⁵ μm³, respectively) were also lower (P < .0001) in prenatal T-treated females) compared with the controls (82.8 ± 0.2 μm³ and 6.7·10⁵ ± 0.7·10⁵ μm³, respectively).

Figure 3.

A, Representative images of visceral (2 left panels) and sc (2 right panels) adipocytes from 2 control and 2 prenatal T-treated females (top panel). Corresponding images of adipocytes identified by computerized image analyses (red lines) are shown to the right of each image. B, Cumulative size distributions of visceral (bottom left panel) and sc (bottom right panel) adipocytes from control (gray) and prenatal T-treated (black) females. Estimated number (mean ± SEM) of adipocytes in each fat depot are computed from mean adipocyte size and fat depot volume by a CT scan from control (white) and prenatal T-treated (black) females is shown as indents. SC, subcutaneous.

Frequency distributions were created for all groups, and statistical analysis was performed on ratios of small (below the 25th percentile) to large (above the 75th percentile) adipocytes. The size distribution was defined by the adipocyte area. In the visceral depot, the ratio of smallest to largest adipocytes was increased (P < .001) in prenatal T-treated females (2.4) compared with controls (3.3). This ratio did not differ between controls and prenatal T-treated females in the sc depot (control: 2.9, T treated: 2.6; P = .058). The number of adipocytes in the visceral and sc fat depot also did not differ between the control and prenatal T-treated groups (see indent in Figure 3).

Discussion

Findings from this study demonstrate that prenatal exposure to excess T, in contrast to previous findings during early life and adulthood (14), increased insulin sensitivity at 16 months of age. This unexpected finding at postpubertal age is suggestive of a period of developmental adaptation. The increased insulin sensitivity was accompanied by a reduction in visceral adiposity at the same age and a reduction in adipocyte cell size in both the visceral and sc compartments at 19 months of age. The premise of a developmental adaptation is consistent with later development of disruptions in FFA metabolism at 19 months and the previously reported reduction in insulin sensitivity reported in 22-month-old prenatally T-treated females (14).

Effects of prenatal T excess on insulin resistance

Prenatal T treatment was found to induce insulin resistance as early as approximately 5 weeks of age (14). Because control females, similar to humans (34), develop insulin resistance as they approach puberty (14), there is no difference in insulin sensitivity indices during the peripubertal period (puberty occurs at approximately 28 weeks of age in sheep) between control and prenatal T-treated females. The trend for increased insulin to glucose ratio (Figure 4) in prenatal T-treated females evidenced at 20 months of age in this study as they approach adulthood may be indicative of developing changes in insulin homeostasis. This is further substantiated by our earlier findings that prenatal T treatment, even for a shorter duration (days 60–90 of gestation), leads to insulin resistance in 22-month-old adult females (14). Although the insulin to glucose ratio is not considered a true surrogate for insulin resistance (47), the measure points toward this direction and helps track developmental progression of a programmed trait in large animal models such as sheep. The trajectory of changes, namely reduced insulin sensitivity at 5 and 11 weeks of age (14, 35), lack of difference around puberty (14, 35), enhanced sensitivity at 16 months of age (this study), and resurfacing of reduced insulin sensitivity during adulthood (14), is supportive of compensatory attempts by the developing offspring to overcome pathology.

Our preliminary studies also show corresponding directionality of age-specific changes in the insulin signaling cascade in metabolic tissues of prenatal T-treated sheep (48) (Lu, Y. and V. Padmanabhan, unpublished data). Consistent with these findings, prenatal T-treated infant monkeys manifest increased insulin sensitivity at 45 days of age (41), whereas adult monkeys are insulin resistant at 15 years of age (18). A similar adaptive trait is evident at the level of LH in prenatal T-treated monkeys, with circulating LH levels being elevated in gestational day 120 fetuses and newborns (postnatal day 1) (49), transient normalization of this trait at postnatal day 30 (49) and resurfacing LH excess in adult life (∼13–14 years old) (50). Such sequential testing of outcome variables at multiple developmental time points has not been undertaken in other animal models.

Effects of prenatal T excess on visceral adiposity

Pre- or perinatal T or DHT treatment has been found to disrupt adipose tissue mass and distribution, although the directionality of the effect varies between studies (18–20, 36–38, 40). Increased visceral fat mass was evident in prenatal T-treated 2-month-old (postpubertal) rats (20) but not in prenatal DHT-treated adult mice (36). Similar to findings in this study, Sprague Dawley rats (but not Wistar rats) treated with T within 3 hours of birth showed a reduction in parametrial, retroperitoneal, and inguinal fat depots with no change in the mesenteric depot at 2.5 months of age (38). In contrast to T treatment, DHT treatment within 3 hours of birth resulted in an increase in the mesenteric fat depot at 1.7–3.7 months of age (37). Supportive of the importance of the developmental window of exposure, rhesus monkeys treated with T for 15–80 days from days 40–44 of gestation had increased total intraabdominal fat mass at 20 years of age (19), whereas those treated for only 15–35 days of gestation (29) showed no change in adiposity. Similarly, heifers exposed to excess T during the first one third of gestation had increased sc fat mass at 27 months of age (40). The variability in adiposity outcomes may stem not only from species/strain differences, timing, dosage, or quality of steroid exposure but also from the developmental time point when the adiposity measures were undertaken.

Effects of prenatal T excess on adipocyte size

Both estrogens (51, 52) and androgens (53) play an important role in white adipose tissue regulation. Prenatal DHT treatment increased adipocyte size in adult mice without altering fat mass (36), whereas postnatal DHT-treated increased both mesenteric adipocyte size and visceral fat depot in Wistar rats (39). Most studies attribute androgen receptor activation to increased adipocyte size. Consistent with this premise, androgen receptor knockout mice do manifest smaller adipocytes (54). Findings from this study did not follow this prediction because prenatal T treatment led to a reduction in adipocyte size and visceral adiposity. Nonetheless, the parallel direction of changes in both these traits is consistent with the predicted positive association between fat mass and adipocyte size (55–57). Although paradoxical, it has been recently postulated that insulin resistance in obese subjects stems from failure of a subset of small adipocytes to mature as fully differentiated adipocytes (55) with different lipolytic capacities (21, 58). The findings from this study, namely the increase in the ratio of small to large adipocytes in the prenatal T-treated females in the visceral depot, support failure of a subset of small adipocytes to differentiate, thus providing support for this hypothesis.

Effects of prenatal T on free fatty acid metabolism

Although androgens have the potential to influence FFA metabolism (59), very little is known about the programming effects of androgens on free fatty acid metabolism. Our detailed profiling of FFAs found an increase in palmitic acid as well as a numerical nonsignificant increase in most FFAs and the total FFA, at an age close to when prenatal T-treated females manifest insulin resistance (22 months) (14). Of interest, palmitic acid is implicated in insulin resistance (60, 61). Studies with animal models found that different FFAs have different entry rates into circulation (62–64). Therefore, the finding that only 1 FFA (palmitic acid) was significantly increased, whereas all others only showed a numerical, nonsignificant increase, may be a function of higher entry rate of palmitic acid in circulation of ruminants compared with others as was shown for stearic acid (62), the next saturated free fatty acid (C18:0).

The increase in palmitic acid accompanied by the nonsignificant increase in total FFA observed in this study in concert with the increase in total FFA reported in prenatal T-treated rhesus monkeys (29) may represent increased lipolysis. Whether this is the result of tissue-specific insulin resistance in the adipose tissue remains unknown. In contrast to these findings in sheep and monkeys, rats treated with T within 3 hours of birth (38) or treated postnatally for 90 days (39) showed no effect on FFA metabolism. From a translational standpoint, although dyslipidemia is widely reported in women with PCOS (22), measures of the total circulating FFAs in these women are scarce. A recent study characterizing individual FFA species showed that women with PCOS had a normal FFA profile and that any observed changes in the FFA profile are a function of adiposity (65).

Relationship among metabolic variables

The reduction in visceral adiposity in prenatal T-treated females accompanied by increased insulin sensitivity at 16 months of age is consistent with the dogma that increased visceral adiposity correlates with impaired insulin resistance or β-cell dysfunction (16, 17). Whether these animals will manifest increased visceral adiposity at older ages, when animals manifest insulin resistance, remains to be addressed. Importantly, the disrupted FFA profile (increased palmitic acid and overall elevation in total FFA) that occurs at a time point preceding when these females become insulin resistant (20 months of age) (14) is supportive of the premise that FFA plays a role in the establishment of insulin resistance (27, 28).

Other studies testing the effects of pre- or perinatal administration of T/DHT also found clear deviations in this predicted relationship (see Figure 4). For instance, prenatal T-treated rats manifested increased visceral adiposity in the absence of disruption in insulin homeostasis (20). This may be representative of a benign obesity phenotype in which obese subjects exhibit a better metabolic profile than expected for their adiposity (66). On the other hand, Sprague Dawley rats treated with T within 3 hours of birth (38) manifested insulin resistance in the absence of increased visceral adiposity, whereas Wistar rats treated at the same point manifested both insulin resistance and visceral adiposity (37), supportive of differences in strain sensitivity (67). A similar discrepancy was also observed in rhesus monkeys treated during 2 different gestational windows (19, 29). The apparent lack of consistency in metabolic outcomes after pre- or perinatal programming with T or DHT (Figure 4) may reflect species/strain differences, the qualitative nature of steroid used for programming, and/or the age of testing. The reduced visceral adiposity and enhanced insulin sensitivity seen in the present study may reflect an attempt to compensate and overcome pathology at the specific age studied. This is in line with predictions of the thrifty phenotype, in which adaptive mechanisms are set in place to compensate for early insults (68).

Translational relevance

Considering that the prenatal T-treated female sheep mimic the phenotype of women with PCOS at the reproductive and metabolic level (12) and that these females are kept on a maintenance diet and not allowed to become obese (69), the metabolic phenotype of prenatal T-treated sheep described in this study, namely reduced fat mass lack of insulin resistance, is similar to those reported for the lean-PCOS phenotype (10). Because lean PCOS women are not insulin resistant (42, 43), insulin resistance in PCOS women (44–46) appears to be a trait strongly influenced by obesity (6, 65).

The reduced adipocyte size in the visceral depot found in the current study in this animal model of PCOS differs from the presence of enlarged visceral adipocytes in obese women with PCOS (70). However, studies comparing lean and obese PCOS women with non-PCOS women found an effect of obesity but not of PCOS on adipocyte cell size (71). More recent studies with lean PCOS women provide evidence in support of effect of PCOS on adipocyte morphology (9, 21) that is independent of body fat mass. Developmental changes in insulin sensitivity across the life span discussed above (Figure 4) emphasize the need for considering homogeneity of the study population and importantly the age of testing in addressing the controversy in this field. Furthermore, the positive correlation found between serum and follicular fluid FFAs, specifically palmitic and stearic, and the reduced in vitro fertilization outcomes with increased FFAs in women with PCOS (72), suggest that the disruption in ovarian (73, 74) and reproductive function (75) evidenced in prenatal T-treated females, may be mediated in part via disruption in FFA metabolism evidenced in this study.

In summary, current findings of improved insulin sensitivity and reduced fat mass in prenatal T-treated sheep females relative to our earlier findings of reduced insulin sensitivity of prenatal T-treated females during early life and adulthood are suggestive of a period of developmental adaptation.