Prader-Willi syndrome (PWS) has a biphasic clinical phenotype with failure to thrive in the neonatal period followed by hyperphagia and severe obesity commencing in childhood among other endocrinological and neurobehavioral abnormalities. The syndrome results from loss of function of several clustered, paternally expressed genes in chromosome 15q11-q13. PWS is assumed to result from a hypothalamic defect, but the pathophysiological basis of the disorder is unknown. We hypothesize that a fetal developmental abnormality in PWS leads to the neonatal phenotype, whereas the adult phenotype results from a failure in compensatory mechanisms. To address this hypothesis and better characterize the neonatal failure to thrive phenotype during postnatal life, we studied a transgenic deletion PWS (TgPWS) mouse model that shares similarities with the first stage of the human syndrome. TgPWS mice have fetal and neonatal growth retardation associated with profoundly reduced insulin and glucagon levels. Consistent with growth retardation, TgPWS mice have deregulated liver expression of IGF system components, as revealed by quantitative gene expression studies. Lethality in TgPWS mice appears to result from severe hypoglycemia after postnatal d 2 after depletion of liver glycogen stores. Consistent with hypoglycemia, TgPWS mice appear to have increased fat oxidation. Ghrelin levels increase in TgPWS reciprocally with the falling glucose levels, suggesting that the rise in ghrelin reported in PWS patients may be secondary to a perceived energy deficiency. Together, the data reveal defects in endocrine pancreatic function as well as glucose and hepatic energy metabolism that may underlie the neonatal phenotype of PWS.
PRADER-WILLI SYNDROME (PWS) is characterized by a distinct biphasic clinical phenotype. Infants show failure to thrive, including severe muscle hypotonia, respiratory and feeding difficulties, and hypogonadotropic hypogonadism (1), whereas children and adults display short stature, mild to moderate mental retardation with behavioral abnormalities, hyperphagia, and severe obesity (2–4). Indeed, much of the pathology that results in morbidity and mortality in PWS, including type 2 diabetes and cardiovascular disease, is secondary to obesity (3). A number of endocrine abnormalities occur in children and adults with PWS, including GH (3, 5–9), IGF-I, and IGF binding protein (IGFBP)-3 deficiencies (3, 8), and insulin levels are relatively decreased compared with individuals with common obesity (10, 11). In contrast, fasting levels of circulating ghrelin, a gut hormone believed to act on hypothalamic arcuate nucleus neurons to modulate feeding (12), are grossly elevated in children and adults with PWS, which may play a major role in the hyperphagia and obesity (13–15). Although the hormonal and clinical abnormalities in PWS have led to the idea of a primary hypothalamic origin for the disorder (2, 9, 16), hormonal and metabolic studies have not been reported in newborns with PWS and the precise pathophysiology for the disorder remains unknown.
PWS is caused by loss of function of a unique set of 10 known imprinted, paternally expressed genes from human chromosome 15q11-q13 (4). Seventy-five percent of cases are associated with a 4- to 5-Mb deletion of this region that encompasses the imprinted genes and a set of five or nine nonimprinted loci (4, 17). The other 25% of PWS cases typically have maternal uniparental disomy or, rarely, an imprinting defect (4). The PWS imprinted candidate genes include three intronless genes (NDN, MAGEL2, and MKRN3), a complex polycistronic locus (SNURF-SNRPN), and five subfamilies of box C/D snoRNAs (4). SNRPN encodes a core spliceosomal protein in postnatal neurons (18). Although some biochemical properties for the other PWS-region gene products are known or have been proposed based on sequence homologies, the precise functions of these genes and the links with clinical aspects of the syndrome are not known. The PWS chromosome region is conserved in mouse chromosome 7C, and three PWS mouse models have been generated with maternal uniparental disomy, an imprinting defect (ID), or a paternally derived chromosome deletion. These models share a very similar phenotype that models the first stage of the human syndrome and includes hypotonia, respiratory difficulties, failure to thrive, and early postnatal lethality (19–21).
Because there is reduced fetal movement and a severe clinical course in the newborn with PWS, we hypothesize a fetal developmental abnormality in PWS leading to the neonatal failure to thrive with the adult metabolic phenotype resulting from a failure in compensatory mechanisms. To begin to test this hypothesis, in the present study, we use a transgenic PWS (TgPWS) deletion mouse model to evaluate endocrinological and metabolic abnormalities in late fetal life and during the failure-to-thrive period. TgPWS mice display fetal and neonatal growth retardation associated with insulin/IGF axis abnormalities. Because the fetal changes predate the postnatal failure to thrive, this may identify a primary mechanism. Subsequently endocrine and metabolic deficiencies peak at postnatal day (P)2, leading to dramatic hypoglycemia and severe failure to thrive in TgPWS neonates, mimicking the clinical course in untreated infants with PWS.
Materials and Methods
Animals and tissue collection
The TgPWS mouse model has an approximately 5-Mb deletion including the entire PWS-homologous region (Fig. 1), created by an LMP2A transgene array insertion (21) (Stefan, M., K. C. Claiborn, E. Stasiek, J.-H. Chai, T. Ohta, R. Longnecker, J. M. Greally, and R. D. Nicholls, unpublished data). The transgenic line is maintained in the CD-1 strain by maternal transmission and shows imprinted inheritance with a TgPWS mouse phenotype after paternal transmission only. The other 50% of offspring in each litter are wild-type (WT) and served as controls. Approval for all experiments was obtained from the University of Pennsylvania and Monell Chemical Senses Center Institutional Animal Care and Use committees.
Genetic structure of the mouse chromosome 7C region homologous to the human PWS region. Circles, Genes; ovals, small nucleolar RNAs (snoRNAs); black, paternally expressed; gray, maternally expressed; open circles, biparentally expressed; gray bar, LMP2A transgene (Tg)-induced PWS/AS deletion (del) mouse model; AS, Angelman syndrome mouse model; IC, imprinting center; cen, centromere; tel, telomere (21 ) (Stefan, M., K. C. Claiborn, E. Stasiek, J.-H. Chai, T. Ohta, R. Longnecker, J. M. Greally, and R. D. Nicholls, unpublished data).
Genetic structure of the mouse chromosome 7C region homologous to the human PWS region. Circles, Genes; ovals, small nucleolar RNAs (snoRNAs); black, paternally expressed; gray, maternally expressed; open circles, biparentally expressed; gray bar, LMP2A transgene (Tg)-induced PWS/AS deletion (del) mouse model; AS, Angelman syndrome mouse model; IC, imprinting center; cen, centromere; tel, telomere (21 ) (Stefan, M., K. C. Claiborn, E. Stasiek, J.-H. Chai, T. Ohta, R. Longnecker, J. M. Greally, and R. D. Nicholls, unpublished data).
Tissues were collected between 1000 and 1400 h. Pups were weighed and then killed by carbon dioxide inhalation. After excision, tissues were immediately frozen in liquid nitrogen. For carcass analysis (water, protein, and triglyceride content), whole bodies were frozen in dry ice and stored at −80 C. Blood was collected into capillary tubes from the heart cavity. For determination of triglycerides, free fatty acids, ghrelin, glucose, insulin, glucagon, and corticosterone, EDTA was used as an anticoagulant. Heparin was used for all other blood collection. Blood was centrifuged at 4000 rpm for 10 min at 4 C and plasma was saved. Frozen liver and plasma were stored at −80 C until further analysis. Timed pregnant animals were used for all studies performed at embryonic day (E)18.
Liver tissue from five TgPWS and five WT pups at P1 was used for quantitative real-time RT-PCR (QRT-PCR) experiments, whereas tissues from four animals of each genotype were used for QRT-PCR at E18 and P5. All animals were genotyped by PCR (21).
Metabolic and hormone assays
Total water content was calculated from the difference in weight between wet and dehydrated carcasses. Protein content was measured using the DC Protein Assay Kit II (Bio-Rad, Hercules, CA), and triglyceride content was assessed by a quantitative enzymatic colorimetric procedure (Stanbio, Boerne, TX). Liver glycogen was determined using an enzymatic method (22). Plasma glucose was measured using glucose oxidase kit 510 (Sigma, St. Louis, MO). Plasma insulin and glucagon were measured using an ultrasensitive rat insulin ELISA kit (ALPCO Diagnostics, Windham, NH) and GL-32K ELISA kit (Linco Research, St. Charles, MO), respectively. Plasma corticosterone was measured using a corticosterone kit (rat/mouse RIA, ALPCO Diagnostics). Plasma immunoreactive ghrelin levels were measured using a modification of a commercial RIA kit (Phoenix Pharmaceuticals, Phoenix, AZ), as previously described (13).
Plasma triglycerides and free fatty acids were determined using commercial kit 337 (Sigma) and the NEFA-C kit (Waco Chemicals Inc., Dalton, GA), respectively. These two assays were scaled down to use 5-μl samples and performed in a flat-bottom 96-well microplate. Color density of the reaction mixtures was assessed using a Multiskan MCC/340 microplate reader (Thermo Electron Corp., Waltham, MA). The two-step Bligh and Dyer method (23) was used to estimate liver lipid content. Liver contents of ATP and ADP were determined in neutralized acid-extract using HPLC (24), and inorganic phosphate was measured using a commercial kit (kit 360-3; Sigma). Phosphorylation potential was calculated using the formula [ATP]/([ADP] × [Pi]). Plasma ketones (β-hydroxybutyrate + acetoacetate) were determined in deproteinized samples using an enzymatic assay with fluorometric detection (25).
Comparisons of measures over time between TgPWS and WT pups were made using two-way ANOVA with repeated measures. Significant differences (P < 0.05) at each time point were identified by an Independent Samples t test.
QRT-PCR
Total RNA was extracted with TRIzol (Invitrogen, Carlsbad, CA), treated with RQ1 RNase-free DNase (Promega, Madison, WI), and checked for integrity by gel electrophoresis. cDNA was synthesized using random hexamers with the SuperScript first-strand synthesis system (Invitrogen) and QRT-PCR was performed as described (26). Matching primers for each gene were designed using the Primer Express program (PE Applied Biosystems Inc., Foster City, CA) as follows: Igfbp1 (exons 3–4), 5′-AGAATGGATTTTATCACAGCAAAGAG-3′ [forward (F)], 5′-TTCCACTCCATGGGTAGACACA-3′ [reverse (R)]; Igf1 (exons 2–3), 5′-GCTGGTGGATGCTCTTCAGTT-3′ (F), 5′-GGTGCCCTCCGAATGCT-3′ (R); Igfbp3 (exons 2–3), 5′-GCAGGCAGCCTAAGCACCTA-3′ (F), 5′-CCTCCTCGGACTCACTGATGTT-3′ (R); and Gapdh (exons 1–2), 5′-CTCCACTCACGGCAAATTCA-3′ (F), 5′-ATGGGCTTCCCGTTGATGA −3′ (R). QRT-PCR data were calculated as the means of expression fold changes (±sd), with two-tailed P values and are presented as the ratio of mean fold changes for TgPWS to WT.
Results
Metabolic and endocrine evaluations
Growth
TgPWS mice have a transgene insertion/deletion transmitted through CD-1 TgAS (transgenic Angelman syndrome mouse model) males, with half of the offspring being TgPWS and half being WT. Despite their relatively normal appearance at birth, TgPWS mice were significantly growth retarded at E18 (P = 0.0007), and after birth they did not grow at the same rate as their WT littermates [F(3, 1) = 444, P < 0.0001]. Consequently, at P5 TgPWS pups reached only half the weight of their WT littermates (Fig. 2A). At E18, placental weights did not differ between genotypes (TgPWS, 0.11 ± 0.007 g, n = 5; WT, 0.13 ± 0.005 g, n = 9; P = 0.14).
Body weight and composition in TgPWS vs. WT mice. A, Reduced body weight in fetal life (E18) and postnatal growth retardation (P1-P5) of the TgPWS (n = 5–75 at each time point) vs. WT (n = 10–106) mice. Body water (B), protein (C), and triglyceride (D) percentage in TgPWS (n = 6–11) vs. WT (n = 8–24) mice. The water, protein, and triglyceride contents (milligrams) were related to wet body weight and expressed as a percentage. White diamonds, WT mice; black circles, TgPWS mice. Values are means from all animals for each time point ± sem, with significant differences (independent-samples t test) between WT and TgPWS mice at the indicated time points: *, P < 0.05; **, P < 0.001; ***, P < 0.0001.
Body weight and composition in TgPWS vs. WT mice. A, Reduced body weight in fetal life (E18) and postnatal growth retardation (P1-P5) of the TgPWS (n = 5–75 at each time point) vs. WT (n = 10–106) mice. Body water (B), protein (C), and triglyceride (D) percentage in TgPWS (n = 6–11) vs. WT (n = 8–24) mice. The water, protein, and triglyceride contents (milligrams) were related to wet body weight and expressed as a percentage. White diamonds, WT mice; black circles, TgPWS mice. Values are means from all animals for each time point ± sem, with significant differences (independent-samples t test) between WT and TgPWS mice at the indicated time points: *, P < 0.05; **, P < 0.001; ***, P < 0.0001.
Similar postnatal growth patterns for a small number of PWS ID mice have been reported (20), except that the majority of mutants in that study died within 48 h (20). In contrast, in our study the majority (80%) of TgPWS mice survived beyond P2 but then usually died between P5 and P7 (see supplemental data Fig. S1 published on The Endocrine Society’s Journal Online web site at http://endo.endojournals.org). The difference is likely due to the use of inbred (20) vs. outbred (our study) strains of mice, with the earlier lethality potentially associated with Ndn deficiency and a respiratory deficit that is pronounced in some inbred strains of mice (27).
Body composition
From P1 to P4, TgPWS contained a greater proportion of body water than the WT mice [F(3, 1) = 10.53, P = 0.001] (Fig. 2B). At P4, percent water content was the highest in TgPWS, compared with WT littermates (P = 0.04). Body protein percentage was unchanged from P1 to P3, but by P4 protein content was higher in TgPWS vs. WT (P = 0.01) (Fig. 2C). Triglycerides, comprising the majority of total body fat content, did not differ at birth (P1). However, overall body triglyceride percentage was decreased in TgPWS [F(3, 1) = 40.87, P = 0.0001] and was most pronounced at P2 (P = 0.0003) (Fig. 2D).
Liver glycogen
Liver glycogen content was similar at E18 and P1 (Fig. 3A) between TgPWS and WT pups. Both groups of mice had a significant reduction in hepatic glycogen content between P1 and P2 (−76 and −91% for WT and TgPWS, respectively; P < 0.0003). Overall, postnatal WT pups had more glycogen in liver than did TgPWS pups [F(3, 1) = 13.6, P = 0.004]. This difference was most pronounced at P2 and P3 (P < 0.01).
Glucose and endocrine homeostasis in TgPWS vs. WT mice. A, Liver glycogen content in fetal and neonatal TgPWS (n = 9–10) and WT (n = 10–18) mice. B, Plasma glucose levels in fetal and neonatal TgPWS (n = 4–21) vs. WT (n = 4–29) pups. C, Plasma insulin in TgPWS (n = 4–11), compared with WT (n = 5–26) littermates (dotted line, minimum of 0.15–5.5 ng/ml standard curve). D, Plasma glucagon levels in TgPWS (n = 2–5) vs. WT (n = 2–8) mice at E18-P4 (dotted line, minimum of 20–500 pg/ml standard curve). E, Circulating corticosterone in fetal and neonatal TgPWS (n = 4–11) and WT (n = 4–17) mice. F, Total plasma ghrelin levels in postnatal TgPWS (n = 4–25), compared with WT littermates (n = 6–23). Symbols, values, and statistics are as for Fig. 2.
Glucose and endocrine homeostasis in TgPWS vs. WT mice. A, Liver glycogen content in fetal and neonatal TgPWS (n = 9–10) and WT (n = 10–18) mice. B, Plasma glucose levels in fetal and neonatal TgPWS (n = 4–21) vs. WT (n = 4–29) pups. C, Plasma insulin in TgPWS (n = 4–11), compared with WT (n = 5–26) littermates (dotted line, minimum of 0.15–5.5 ng/ml standard curve). D, Plasma glucagon levels in TgPWS (n = 2–5) vs. WT (n = 2–8) mice at E18-P4 (dotted line, minimum of 20–500 pg/ml standard curve). E, Circulating corticosterone in fetal and neonatal TgPWS (n = 4–11) and WT (n = 4–17) mice. F, Total plasma ghrelin levels in postnatal TgPWS (n = 4–25), compared with WT littermates (n = 6–23). Symbols, values, and statistics are as for Fig. 2.
Plasma glucose
There was no difference in plasma glucose levels between TgPWS and WT pups at E18 or P1 (Fig. 3B). However, at P2 glucose concentration dropped in TgPWS pups and thereafter was significantly lower, compared with WT mice [F(3, 1) = 100.7, P = 0.0001] (Fig. 3B).
Plasma insulin
Plasma insulin levels were significantly lower in TgPWS vs. WT mice in fetal life at E18 (P = 0.04) and from P1 to P4 [F(3, 1) = 44.9, P = 0.0001] (Fig. 3C). Insulin levels were extremely low (<0.15 ng/ml, the detection limit of the assay) in 75 and 100% of TgPWS pups at P2 and P3, respectively.
Plasma glucagon
The concentration of plasma glucagon was also lower in TgPWS, compared with WT mice [F(3, 1) = 7.4, P = 0.008] (Fig. 3D). Similar to insulin levels, glucagon was also very low (<20 pg/mg, the detection limit of the assay) in TgPWS plasma from P2 to P4.
Corticosterone
Corticosterone was significantly higher in TgPWS vs. WT mice at all postnatal days [F(3, 1) = 9.31, P = 0.003] but not at E18 (Fig. 3E). The normal fall in corticosterone levels that occurs after birth was blunted in TgPWS mice (P < 0.0001).
Ghrelin
There was no difference in plasma ghrelin levels between TgPWS and WT mice at P1 and P2, but for P3, P4 and P5, ghrelin concentration was significantly higher in TgPWS pups (P = 0.0003) (Fig. 3F).
Triglycerides
Between P1 and P4, plasma triglyceride concentrations in TgPWS pups were 35–59% of those in WT pups [F(3, 1) = 91.2, P = 0.0001] (Fig. 4A). Between P1 and P2, there were substantial increases in both WT (+182%; P = 0.0001) and TgPWS (+387%; P = 0.0001) pups. Overall plasma free fatty acids were also significantly reduced in TgPWS mice [F(3, 1) = 19.8, P = 0.0001] and were most profoundly reduced on P1 and P4 (P < 0.05) (Fig. 4B). Concentrations of free fatty acids increased substantially between P1 and P2 in both WT (+95%; P = 0.0001) and TgPWS (+159%; P = 0.0001) pups. In contrast, plasma levels of ketones were higher in TgPWS than in WT pups from P1 to P4 [F(3, 1) = 6.9, P = 0.01] (Fig. 4C).
Lipid metabolism in TgPWS vs. WT mice. A, Plasma triglyceride levels in TgPWS (n = 9–12) and WT (n = 11–18) mice from P1 to P4. B, Plasma free fatty acid levels in neonatal TgPWS (n = 9–13) and WT (n = 10–18) mice. C, Plasma ketone bodies during the P1-P4 postnatal period in TgPWS (n = 8–35), compared with WT (n = 11–31) mice. D, Liver lipid content in TgPWS (n = 9–13) and WT (n = 10–13) mice. Liver total ATP content (E) and phosphorylation potential (F) in TgPWS (n = 8–35) vs. WT (n = 11–31) mice at P1-P4. Symbols, values, and statistics are as for Fig. 2.
Lipid metabolism in TgPWS vs. WT mice. A, Plasma triglyceride levels in TgPWS (n = 9–12) and WT (n = 11–18) mice from P1 to P4. B, Plasma free fatty acid levels in neonatal TgPWS (n = 9–13) and WT (n = 10–18) mice. C, Plasma ketone bodies during the P1-P4 postnatal period in TgPWS (n = 8–35), compared with WT (n = 11–31) mice. D, Liver lipid content in TgPWS (n = 9–13) and WT (n = 10–13) mice. Liver total ATP content (E) and phosphorylation potential (F) in TgPWS (n = 8–35) vs. WT (n = 11–31) mice at P1-P4. Symbols, values, and statistics are as for Fig. 2.
Liver lipids
Paralleling the increase in plasma triglyceride and free fatty acid levels, liver lipid content also increased from P1 to P2 in both WT and TgPWS pups (Fig. 4D). However, this increase was markedly blunted in TgPWS mice (97%, P = 0.0001, and 47%, P < 0.0001, respectively). Overall, between P1 and P4, liver lipid contents were higher in WT pups than TgPWS pups [F(3, 1) = 10.5, P < 0.002].
Liver ATP
Liver ATP content of WT pups did not change between P1 and P4 (Fig. 4E). There was also no difference between WT and TgPWS pups on P1. In contrast, in TgPWS pups, liver ATP content significantly increased from the first to the second day of life in TgPWS mice (P < 0.02) but then fell back to P1 levels by P3. Overall, liver ATP contents were higher in TgPWS pups than WT pups [F(3, 1) = 7.3, P < 0.01]. Between P1 and P4, liver phosphorylation potential decreased gradually in WT pups (P = 0.0001) (Fig. 4F), most likely due to a gradual increase in liver inorganic phosphate content (P = 0.0001) (data not shown). Overall, TgPWS pups had higher liver phosphorylation potential than did WT pups [F(3, 1) = 6.0, P < 0.02], which was highest at P2 and P3 (P < 0.01).
Hepatic IGF system gene expression
We examined hepatic expression of the genes encoding IGF-I, IGFBP-3, and IGFBP-1 because plasma levels are often reported abnormal in humans with PWS (3, 8, 28, 29). QRT-PCR analysis showed a 2.6-fold decreased level of Igf1 mRNA at P1 (P = 0.002) and a 2.2-fold decreased level at P5 (P = 0.006) but no difference at E18 in TgPWS vs. WT liver (Fig. 5A). Hepatic mRNA levels of Igfbp3, which encodes the protein that binds nearly 75% of circulating IGF-I (30), did not differ between TgPWS and WT mice at all three time points (Fig. 5B). In contrast, Igfbp1 mRNA levels were consistently up-regulated in postnatal TgPWS mice, by 2.3-fold, compared with WT at P1 (P = 0.02), and 5.3-fold at P5 (P = 0.002). However, there was no difference between TgPWS and WT at E18 (P = 0.8) (Fig. 5C).
Quantitation of mRNA expression levels for Igf1 (A), Igfbp3 (B), and Igfbp1 (C) in TgPWS vs. WT mice at E18, P1, and P5. Relative quantitation for expression of the three genes by QRT-PCR was performed in two groups of five TgPWS (black bars) and five WT (white bars) mice at P1 as well as four TgPWS and four WT mice at E18 and P5. A, Hepatic Igf1 expression was unchanged at E18 (TgPWS: 0.82 ± 0.05; WT: 1.02 ± 0.33, P = 0.3) but decreased at P1 (TgPWS: 0.46 ± 0.11; WT: 1.2 ± 0.34, P = 0.002) and P5 (TgPWS: 0.48 ± 0.24; WT: 1.05 ± 0.13, P = 0.006) in TgPWS vs. WT mice. B, There was no difference in Igfbp3 hepatic expression at any of the three time points in TgPWS vs. WT mice: E18 (TgPWS: 0.68 ± 0.29; WT: 0.95 ± 0.25, P = 0.2), P1 (TgPWS: 1.17 ± 0.41; WT: 0.91 ± 0.16, P = 0.2), or P5 (TgPWS: 0.94 ± 0.27; WT: 1.11 ± 0.17, P = 0.3). C, Hepatic Igfbp1 mRNA expression was increased at P1 (TgPWS: 2.43 ± 1.1; WT: 1.06 ± 0.25, P = 0.02) and P5 (TgPWS: 4.22 ± 1.31; WT: 0.81 ± 0.43, P = 0.002), whereas there was no difference in mRNA levels at E18 (TgPWS: 1.32 ± 0.13; WT: 1.28 ± 0.32, P = 0.8) in TgPWS vs. WT mice. All values are represented as the mean of the fold change ± sd. *, Significant differences between WT and TgPWS, P < 0.05 (independent-samples t test).
Quantitation of mRNA expression levels for Igf1 (A), Igfbp3 (B), and Igfbp1 (C) in TgPWS vs. WT mice at E18, P1, and P5. Relative quantitation for expression of the three genes by QRT-PCR was performed in two groups of five TgPWS (black bars) and five WT (white bars) mice at P1 as well as four TgPWS and four WT mice at E18 and P5. A, Hepatic Igf1 expression was unchanged at E18 (TgPWS: 0.82 ± 0.05; WT: 1.02 ± 0.33, P = 0.3) but decreased at P1 (TgPWS: 0.46 ± 0.11; WT: 1.2 ± 0.34, P = 0.002) and P5 (TgPWS: 0.48 ± 0.24; WT: 1.05 ± 0.13, P = 0.006) in TgPWS vs. WT mice. B, There was no difference in Igfbp3 hepatic expression at any of the three time points in TgPWS vs. WT mice: E18 (TgPWS: 0.68 ± 0.29; WT: 0.95 ± 0.25, P = 0.2), P1 (TgPWS: 1.17 ± 0.41; WT: 0.91 ± 0.16, P = 0.2), or P5 (TgPWS: 0.94 ± 0.27; WT: 1.11 ± 0.17, P = 0.3). C, Hepatic Igfbp1 mRNA expression was increased at P1 (TgPWS: 2.43 ± 1.1; WT: 1.06 ± 0.25, P = 0.02) and P5 (TgPWS: 4.22 ± 1.31; WT: 0.81 ± 0.43, P = 0.002), whereas there was no difference in mRNA levels at E18 (TgPWS: 1.32 ± 0.13; WT: 1.28 ± 0.32, P = 0.8) in TgPWS vs. WT mice. All values are represented as the mean of the fold change ± sd. *, Significant differences between WT and TgPWS, P < 0.05 (independent-samples t test).
Discussion
We measured metabolic and hormonal parameters in a TgPWS deletion mouse model in fetal life and early postnatal life to assess the abnormalities that underlie the neonatal phenotype and premature lethality in these mice. We found a constellation of abnormalities in TgPWS mice that are consistent with the clinical presentation in newborns or children with PWS, including growth retardation, insulin/IGF-I deficiencies, and hyperghrelinemia. Additional findings of abnormal glucose and energy homeostasis and pancreatic insufficiency, beginning in fetal life, provide new insights into the pathophysiological basis of PWS.
The most surprising finding of our study was the deficit in insulin and glucagon observed in fetal and postnatal TgPWS mice, suggesting that this model of PWS in mice induces a primary pancreatic defect. To determine whether TgPWS mice have defects in pancreatic development and function, we are currently examining islet architecture and β- and α-cell mass and performing analyses of global gene expression in the pancreas, compared with WT mice. Our findings have important implications for the pathophysiology of type 2 diabetes that is so often observed in patients with PWS. It has generally been hypothesized that diabetes in this population is secondary to profound obesity; however, a number of studies have demonstrated that insulin levels are actually much lower than expected for the degree of the increase in body mass index (3, 10, 11).
Glucose levels were normal in fetal TgPWS mice, suggesting that placental function is normal in these mice. The fetal requirement for glucose is met almost, if not entirely, by transplacental transport from the mother to the fetus (31, 32). At birth, there is an abrupt loss of the maternal supply of substrates and nutrients and the newborn has to mobilize glucose and other substrates to meet its energy needs. This is achieved primarily through breakdown of glycogen. A number of factors initiate liver glycogenolysis at birth; however, the precise mechanisms have not been completely elucidated. Whereas insulin and glucagon play a role in the breakdown of glycogen, increased plasma catecholamine levels have been suggested to be the primary mediators of the sudden increase in hepatic glucose output (33). This hypothesis is consistent with our findings of low levels of insulin and glucagon at birth, but a normal decrease in hepatic glycogen content in the first day of life in TgPWS mice. Because of the rapid depletion of glycogen, liver glycogenolysis can support glucose homeostasis for only a short period of time, and maintenance of the newborn’s blood glucose levels requires active liver gluconeogenesis. TgPWS mice develop severe hypoglycemia by d 2 of life that is likely due to impaired gluconeogenesis. Phosphoenolpyruvate carboxykinase, the rate-limiting enzyme of gluconeogenesis, is not turned on until several hours after birth and is activated by a decrease in the insulin to glucagon ratio (fall in insulin and rise in glucagon) and an increase in glucocorticoid levels. In response to hypoglycemia, corticosterone levels were appropriately higher in TgPWS mice, but the mutant mice were unable to increase glucose production. The low levels of glucagon and limited substrate supply are likely to be among the primary causes of the inability of the TgPWS mouse to mount an adequate gluconeogenic response. It will be important in future studies to examine the expression levels of genes encoding critical gluconeogenic enzymes or the levels and/or activity of phosphoenolpyruvate carboxykinase or pyruvate carboxylase, for example, that could be altered in the TgPWS mouse.
The low levels of glucagon may also explain the relatively low levels of triglycerides and free fatty acids observed in P1 TgPWS pups. The surge of glucagon that normally occurs after birth induces mobilization of lipids from peripheral tissues. After birth, milk provides a relatively high-fat diet, and lipids are the main energy source of the newborn. However, due to severe failure to thrive, TgPWS mice are not able to suckle normally and lipid levels continue to be low. Because ketone concentrations are actually higher in TgPWS mice, it is unlikely that these mice have fatty acid oxidation defects.
Fetal TgPWS mice are significantly growth retarded, and growth rates remain abnormally low after birth. PWS patients also have low birth weight and reduced weight gain in infancy (34–36). Placental insufficiency cannot be implicated because placental weights are normal for TgPWS animals. The growth restriction in TgPWS mice starts in fetal life and is not due to GH deficiency that occurs postnatally in PWS patients (3, 8, 9) because GH does not regulate fetal growth (37, 38). Additionally, levels of hepatic Igf1 and Igfbp1 mRNA were normal in TgPWS at E18, so these are unlikely to play a role in growth retardation of the TgPWS fetus. However, we cannot completely rule out a causative role for the IGF axis because we did not measure Igf1 receptor levels. In contrast, because insulin is significantly reduced in the TgPWS fetus, decreased activity of insulin signaling pathways (39) may be responsible for the growth restriction observed in fetal and newborn TgPWS mice. Nevertheless, because PWS patients have obesity-independent GH and IGF-I deficiencies (3–10, 28), a pituitary defect could contribute to or underlie postnatal growth retardation of PWS infants and children.
Analysis of hepatic gene expression revealed significantly deregulated Igf1 and Igfbp1 expression but normal Igfbp3 expression in TgPWS newborn mice. These changes correlate with known hormonal regulation of these genes. For example, whereas Igfbp3 production is primarily regulated by GH (29), insulin and/or corticosterone are implicated in the hepatic transcriptional regulation of Igfbp1 and Igf1 (40–43), and these genes are therefore deregulated in the expected manner in hypoinsulinemic TgPWS mice at P1 and P5. Indeed, by P5 there is a dramatic increase in Igfbp1 mRNA in TgPWS on progression of failure to thrive. Because hepatic overexpression of Igfbp1 in two mouse models have been correlated with postnatal growth retardation (44, 45), this may contribute to failure to thrive in the TgPWS mouse. Nevertheless, despite the known alteration in insulin, IGF-I and IGFBP-1 and -3 in individuals with PWS (3, 8, 28, 29), the hormonal and metabolic status of newborns with PWS is unknown, and thus, it is not clear what role the insulin-IGF axis might play in the failure-to-thrive component of PWS in the human, but it is expected that this will be revealed from retrospective or prospective patient studies.
Body composition measurements revealed that postnatal TgPWS mice maintain the same proportion of whole-body water and protein content as do their WT littermates, but whole-body fat (triglyceride) levels decrease after P2, a pattern similar to that seen in animals under conditions of energy restriction (46). This is not unexpected for TgPWS pups, given their hypotonia and failure to thrive. The absence of dehydration in TgPWS mice is consistent with the presence of milk in their stomachs at each age (21), although the amount is usually visibly less than for WT littermates. Nevertheless, intake and nutrient uptake have as yet not been examined in TgPWS mice. Although the milk intake is sufficient to prevent dehydration, a reduced intake may explain the decrease in liver lipids, plasma triglycerides, and plasma free fatty acids as well as whole body fat.
Another model generated by gene targeting of the imprinting center (Fig. 1) is PWS ID mice, which disrupts a genetic element controlling imprinting and active expression of all the paternally expressed genes in the PWS domain (20). Recently it was found that these PWS ID mice generally survive on an FVB strain but not other genetic backgrounds, with surviving mice being small and not displaying hyperphagia or obesity (47). However, if the PWS hyperphagia and obesity phase results from a compensatory attempt to overcome a metabolic basis for neonatal failure to thrive, as we propose, the lack of the latter phenotype in the PWS ID mice on the FVB strain (47) leads to the prediction that these mice would not develop hyperphagia and obesity. In contrast, TgPWS deletion mice did not survive on the FVB background, even with removal of most WT littermates (Ohta, T., and R. D. Nicholls, unpublished data).
An increase in plasma ghrelin levels occurs in postnatal TgPWS mice and appears to begin at the onset of severe hypoglycemia but is not directly coincident with hypoinsulinemia. These findings are consistent with known regulators of ghrelin expression and secretion because both glucose and insulin have been shown to suppress ghrelin levels (48). By P5, ghrelin levels in TgPWS mice are approximately 3-fold higher than in WT littermates, suggesting that high ghrelin levels in TgPWS might be a physiological adaptive mechanism in an attempt to increase feeding via its actions on the arcuate nucleus (49) to ameliorate the rapidly worsening failure to thrive. However, either this signal is unrecognized due to an unknown mechanism, or it may be too late to elicit a physiological response. The finding of high ghrelin levels in TgPWS mice echoes observations in PWS children and adults, in whom plasma ghrelin levels are 2.5- to 4.5-fold higher than those in normal lean and obese controls (13–15). Although these studies suggested that this orexigenic hormone could be a mediator for the hyperphagia observed in PWS, our mouse data are consistent with an alternative hypothesis that PWS patients metabolically do not sense the degree of adipose tissue, and hence their lean body mass is in a starvation state that induces ghrelin production (50, 51).
In accordance with severe hypoglycemia and hypoinsulinemia, postnatal TgPWS pups show evidence of increased fat oxidation, compared with their WT littermates, as revealed by lower plasma levels of triglycerides and free fatty acids as well as increased ketogenesis. Consistent with this, hepatic energy status (ATP content and phosphorylation potential) was higher in TgPWS vs. WT at P2 and P3, possibly as a result of compensatory β-oxidation due to impaired glucose homeostasis. In adults, it has been proposed that low liver ATP levels stimulates food intake through signals to the central nervous system via vagal sensory neurons (52, 53). Because a decrease in liver ATP in rats is associated with increased eating (24, 54), it is possible that a negative feedback signal of elevated hepatic ATP in TgPWS pups may contribute to a decrease in feeding behavior as part of the failure to thrive process. At least in the newborn rat, it appears that the vagal afferent connections are intact and sufficiently mature (55, 56). In contrast, leptin neural pathways are not mature until closer to the time of weaning (57, 58), suggesting that leptin signaling abnormalities do not underlie the TgPWS phenotype.
In conclusion, we characterized the endocrine and metabolic profile of a TgPWS deletion mouse model with fetal growth retardation and neonatal failure to thrive. Most surprising among the deficiencies found in TgPWS mice were fetal insulin and glucagon insufficiency, suggestive of a primary pancreatic defect. It will be important to further examine the mechanisms that might contribute to the phenotype in TgPWS mice and PWS newborns and children. Furthermore, many of the findings in TgPWS mice, such as hypoinsulinemia, low glucose and IGF-I, and high IGFBP-1 and ketones, are also consistently seen in small-for-gestational-age infants in the human (59). The TgPWS mouse therefore serves as a useful PWS and small-for-gestational-age animal model for further investigation, and additional studies on animal models and similar studies on PWS newborns will shed new understanding on the underlying pathophysiological basis for this syndrome.
Acknowledgments
We thank Brande Latney and Heather Collins (Penn RIA Core, Philadelphia, PA) for technical assistance, Jessica Knepper for contributions to early phases of this project, and Dr. Klaus Kaestner for comments on the manuscript.
Abbreviations
- E,
Embryonic day;
- F,
forward;
- ID,
imprinting defect;
- IGFBP,
IGF binding protein;
- P,
postnatal day;
- PWS,
Prader-Willi syndrome;
- QRT-PCR,
quantitative real-time RT-PCR;
- R,
reverse;
- TgPWS,
transgenic PWS;
- WT,
wild type.
