Metabolic Response of Mice to a Postnatal Ablation of CCAAT/Enhancer-binding Protein α*

Although CCAAT/enhancer-binding protein α (C/EBPα) is essential for initiating or sustaining several metabolic processes during the perinatal period, the consequences of total ablation of C/EBPα during postnatal development have not been investigated. We have created a conditional knock-out model in which the administration of poly(I:C) caused a virtually total deletion of c/ebpα (C/EBPαΔ/- mice) in the liver, spleen, white and brown adipose tissues, pancreas, lung, and kidney of the mice. C/EBPα itself was completely ablated in the liver by day 4 after the injection of poly(I:C). There was no noticeable change in phenotype during the first 15 days after the injection. The mice maintained a normal level of fasting blood glucose and responded to the diabetogenic action of streptozotocin. From day 16 onward, the mice developed hypophagia, exhibited severe weight loss, lost triglyceride in white but not brown adipose tissue, became hypoglycemic and hypoinsulinemic, depleted their hepatic glycogen, and developed fatty liver. They also exhibited lowered plasma levels of free fatty acid, triglyceride, and cholesterol, as well as marked changes in hepatic mRNA for C/EBPδ, peroxisome proliferator-activated receptor α, sterol regulatory element-binding protein 1, hydroxymethylglutaryl-coenzyme A reductase, and apolipoproteins. Although basal levels of hepatic mRNA for the cytosolic isoform of phosphoenolpyruvate carboxykinase and glucose-6-phosphatase were reduced, transcription of the genes for these enzymes was inducible by dibutyryl cyclic AMP in C/EBPαΔ/- mice. The animals died about 1 month after the injection of poly(I:C). These findings demonstrate that C/EBPα is essential for the survival of animals during postnatal life and that its ablation leads to distinct biphasic change in metabolic processes.

Administration of Poly(I:C)-Poly(I:C) (Sigma) was dissolved in 0.9% NaCl to a concentration of 1 mg/ml and administered to animals (10 mg/kg of body weight) at 1 day or 3 months of age. For 1-day-old neonates, the poly(I:C) solution was injected subcutaneously into their back; for older animals, it was injected intraperitoneally. Unless otherwise indicated, the animals received only a single injection of poly(I:C).
Southern Blotting-Southern blotting was done as described by Croniger et al. (32). Briefly, genomic DNA from various tissues was digested with the restriction enzyme EcoRI. The NotI fragment (528 bp) (see Fig. 1A, hatched bar), that was used as probe hybridizes with the alleles of C/EBP␣ F , C/EBP␣ ϩ , and C/EBP␣ ⌬ to produce 9.2-, 9.2-, and 4.9-kb bands, respectively (see Figs. 1A and 2A). The probe does not hybridize with the C/EBP␣ Ϫ allele because the target sequence was deleted. The Southern blotting was scanned by using Typhoon 9200 (Amersham Biosciences) and quantified with ImageQuant software (Amersham Biosciences). The intensity of each band was corrected by subtracting the background. The excision efficiency was calculated by using the following equation: excision efficiency ϭ band intensity of C/EBP␣ ⌬ allele/(band intensity of C/EBP␣ F allele ϩ band intensity of C/EBP␣ ⌬ allele). The percentage of C/EBP␣ DNA remaining in the cells was calculated by using the following equation; for C/EBP␣ F/Ϫ mice, the percentage of C/EBP␣ DNA remaining ϭ 50% Ϫ (50% ϫ the excision efficiency). A 50% value represents one copy of c/ebp␣ in the C/EBP␣ F/Ϫ mice. For C/EBP␣ F/F mice, the percentage of C/EBP␣ DNA remaining ϭ 100% Ϫ (100% ϫ the excision efficiency).
Western Blotting-Livers were isolated and fractionated into cytosolic and nuclear fractions as described by Dignam et al. (33) and used for Western blotting as previously described (34). Nuclear extracts (15 g of protein) were separated by electrophoresis in a 12.5% SDS-PAGE gel. C/EBP␣ was detected by using rabbit anti-C/EBP␣ IgG as the primary antibody (Santa Cruz Biotechnology, SC-61; 1:1000 dilution) and goat anti-rabbit IgG-HRP as the secondary antibody (Santa Cruz Biotechnology, SC-2004; 1:3000 dilution).
Measurements of Body Weight and Food and Water Consumption -Animals were transferred to metabolic cages 2 weeks before the administration of poly(I:C) so as to become acclimatized to a new living environment. Adult animals were then given a single injection of poly(I:C) (10 mg/kg of body weight) at 3 months of age. Their initial body weight, measured before the injection, was designated as the starting body weight, and their body weight and food and water consumptions were recorded every other day. Changes in body weight were calculated as the average daily body weight minus the starting body weight of the animal. Unless indicated, all animals were given free access to water and standard chow (Prolab 5P75 Isopro 3000 containing 22% protein, 5% fat, and 5% crude fiber and the remainder carbohydrate). Some control mice (genotype ϭ Mx1-Cre Ϫ and C/EBP␣ F/F ) were pairfed, starting on day 18 after the injection of poly(I:C). They had free access to water, but their daily food consumption was rationed by using the equation y ϭ z Ϫ w ϫ 0.115655 ϩ w ϫ 16.36/e (0.2765 ϫ x) , where y is the daily food supply (g), w is the starting body weight, z is the food intake on day 18 after the injection of poly(I:C), and x is the number of days after the injection.
DNA Microarray Analysis-One-day-old neonates were given a single injection of poly(I:C) (10 mg/kg of body weight), and their livers were removed 28 days later. Total RNA, isolated from the liver, was analyzed by using murine MoE430A GeneChip array (Affymetrix) at the DNA Microarray Core Facility at Case Western Reserve University, School of Medicine. Four animals, evenly divided into two groups, were analyzed simultaneously. Animals in each group were siblings of the same gender, with one C/EBP␣ ⌬/Ϫ mouse and one normal-fed control animal (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ). The data were grouped and analyzed with GeneSpring software program (Silicon Genetics). The probe sequences on the MoE430A array were confirmed by comparing them to the mouse genomic sequence data base at the NCBI (www.ncbi.nlm-.nih.gov/genome/seq/MmBlast.html). Only probes with sequences located in the coding sequence and/or the untranslated region of the intended genes were included in our analysis. The reliability of the final data was further confirmed by verifying their reproducibility in both groups and by performing Northern analysis. The data are presented as -fold change in mRNA concentration in C/EBP␣ ⌬/Ϫ mice compared with normal-fed controls (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ).
Histological Analysis of Tissues-Fresh tissue was collected from mice and placed in 10% neutral-buffered formalin solution (Sigma). Hematoxylin-and-eosin staining was performed at the Histology Core Facility at Case Western Reserve University, School of Medicine, and Oil red O staining was done at the Veterinary Diagnostic Services of Marshfield Laboratories (Marshfield, WI).
Assay of Metabolites in the Blood-Mice were anesthetized by an intraperitoneal injection of Avertin (2,2,2-tribromoethanol; 0.5 ml of 20 mg/ml solution per 25 g of body weight). The concentration of glucose in the blood was determined by using Encore Glucometer. Plasma was generated from whole blood using MICROTAINER plasma separator tubes (BD Biosciences). The plasma concentrations of insulin, leptin, and glucagon-like peptide-1 (active) were measured with a mouse insulin enzyme-linked immunosorbent assay (ELISA) kit (Linco Research, Inc.), a mouse leptin TiterZyme enzyme immunometric assay (EIA) kit (Assay Designs, Inc.), and glucagons-like peptide-1 (active) ELISA kit (Linco Research, Inc.). The measurement of triglyceride, fractionated bilirubin, albumin, ␤-hydroxybutyrate, total protein, blood urea nitrogen, cholesterol, and free fatty acid was performed at the Veterinary Diagnostic Services of Marshfield Laboratories (Marshfield, WI).
Determination of Hepatic Glycogen-Glycogen was determined according to Suzuki et al. (35) with modifications. Pieces of liver (ϳ50 mg) were dissolved in 0.5 ml of 30% KOH. Glycogen was then precipitated with 0.5 ml of 100% ethanol and 10 l of 4 M LiCl by centrifugation at 3000 ϫ g for 30 min. To degrade the glycogen (in the pellet) to glucose, 0.5 ml of 4 M HCl was added to lyse the pellet, and the glucosecontaining lysate was neutralized with 0.5 ml of 2 M K 2 CO 3 solution. The concentration of glucose was determined using a YSI glucose analyzer (Giangarlo Scientific Co.).
Hormone and Other Treatments-Hormones were administrated to animals according to Lechner et al. (36). Briefly, streptozotocin (freshly prepared, 200 mg/kg of body weight) was administrated to mice at day 11 after the injection of poly(I:C). Mice were fasted 18 h before injection and fed 2 h later. The concentration of blood glucose was measured daily using blood collected from the tail vein. Animals with a blood glucose concentration higher than 250 mg/dl were considered diabetic. Bt 2 cAMP was administered to the mice at day 28 after the injection of poly(I:C). To minimize differences caused by different feeding behavior of the animals, mice were injected with glucose (2 mg/gram of body weight) 2 h before the intraperitoneal administration of a mixture of Bt 2 cAMP (35 mg/kg of body weight) and theophylline (30 mg/kg of body weight). The mice were killed 1 h after the last injection. In the control group, saline was used instead of the mixture of Bt 2 cAMP and theophylline.

RESULTS
Creation and Characterization of Mice with a Genotype of Mx1-Cre ϩ ϩ C/EBP␣ F/Ϫ -To delete c/ebp␣ in adult mice, we developed a mouse model that was heterozygous for c/ebp␣. One allele of c/ebp␣ is flanked by a loxP site (designated as C/EBP␣ F ) (19), and the other is a knock-out allele (C/EBP␣ Ϫ ) (8). The genome of the mice also included a chimeric transgene Mx1-Cre (31), composed of the gene for Cre recombinase, linked to the Mx1 gene promoter that can be transiently activated by the administration of interferon ␣, interferon ␤, or an interferon inducer, such as poly(I:C). Use of mice with a heterozygous composition of c/ebp␣ (C/EBP␣ F/Ϫ ) complicates genotyping procedures because two additional alleles (C/EBP␣ F and C/EBP␣ Ϫ ) have to be genotyped compared with homozygous (C/EBP␣ F/F ) mice. We, therefore, developed a PCR-based protocol that allowed us to genotype the three transgenes simultaneously in a single PCR reaction. This involved the construction of three sets of primers that amplified unique regions of the C/EBP␣ Ϫ , Mx1-Cre ϩ , and C/EBP␣ F transgenes (Fig. 1A). The sizes of PCR products generated with these primers were 710 bp for C/EBP␣ Ϫ , 355 bp for Mx1-Cre ϩ , and 173 bp for C/EBP␣ F (Fig. 1B). Mice with a genotype of Mx1-Cre ϩ ϩ C/EBP␣ F/Ϫ developed normally and were indistinguishable from mice with a genotype of C/EBP␣ ϩ/ϩ (wild type), C/EBP␣ F/ϩ , or C/EBP␣ F/F . An advantage of using mice with such a heterozygous genotype is that they only have 50% of c/ebp␣ (one allele, C/EBP␣ F ) to be excised by Cre recombinase as compared with mice with two alleles of the gene (C/EBP␣ F/F ).
Efficiency and Specificity of the Excision of C/EBP␣ F in Various Tissues-To assess the excision efficiency of C/EBP␣ F allele, a DNA fragment obtained after NotI digestion of the 5Ј-region of genomic DNA for C/EBP␣ (Fig. 1A) was used as a probe for Southern blotting.
This probe hybridizes to the alleles of C/EBP␣ ϩ , C/EBP␣ F , and C/EBP␣ ⌬ , but not to C/EBP␣ Ϫ , which lacks the probe sequence (Fig.  1A). Two fragments were detected by Southern blotting; they are a 9.2-kb fragment for the C/EBP␣ ϩ or C/EBP␣ F alleles and a 4.9-kb fragment for the C/EBP␣ ⌬ allele (the excised product). After the poly(I:C) treatment, c/ebp␣ was totally deleted in the liver of mice that carried one copy of C/EBP␣ F ( Fig. 2A, lanes 2-3) or two copies of C/EBP␣ F ( Fig. 2A,  lanes 1 and 4). The specificity of deletion is shown in the last four lanes of Fig. 2A. As expected, only 50% of c/ebp␣ was excised when one allele was C/EBP␣ F and the other was C/EBP␣ ϩ ( Fig. 2A, top panel, lane 8). Quantification of the Southern blots indicated 97% deletion of c/ebp␣ in the liver of C/EBP␣ F/Ϫ mice and 91% deletion in C/EBP␣ F/F mice compared with control mice (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ) (Fig. 2B, Liv.). There was also almost total deletion of c/ebp␣ in other tissues of C/EBP␣ F/Ϫ mice: spleen (ϳ90%), white adipose tissue (WAT) (ϳ89%), pancreas (ϳ83%), lung (ϳ80%), kidney (ϳ80%), and brown adipose tissue (BAT) (ϳ80%) (Fig. 2B, filled bars). This is a marked improvement over the inefficient deletion of the gene noted in C/EBP␣ F/F mice (Fig.  2B, hatched bars). We have, thus, created a model in which the gene for C/EBP␣ has been ablated in C/EBP␣-abundant tissues of adult mice.
Kinetics for the Excision of C/EBP␣ F in Liver-To evaluate the rapidity of the excision of c/ebp␣, the levels of DNA, mRNA, and protein for hepatic C/EBP␣ were measured at days 1-4 after injection of poly(I:C). Approximately 95% of c/ebp␣ was excised from the liver within 24 h after the injection (Fig. 3A); the excision was complete by the 2nd day. Virtually no C/EBP␣ mRNA was detected in the liver 1 day after injection of poly(I:C) (Fig. 3B). C/EBP␣ protein also rapidly disappeared from the nucleus. In fact, only two isoforms, i.e. A and B1 (also known as p42) (37), were visible at days 1 and 2 after injection of poly(I:C) (Fig. 3C, lanes 3 and 4). By day 4, C/EBP␣ protein was completely ablated in the liver of C/EBP␣ ⌬/Ϫ mice (Fig. 3C, compare lanes 5 and 6 to lanes 1, 2, and 7). By comparing the intensity of B1 isoform at different time points, we estimated that the half-life of C/EBP␣ protein is less than 12 h in the liver of animals.
Postnatal Development and Hypophagia of C/EBP␣ ⌬/Ϫ Mice-To investigate the role of C/EBP␣ during postnatal development, we ablated C/EBP␣ in mice at either 1 day after birth (neonate) or at 3 months of age (adult). When poly(I:C) was administrated to neonates, the resultant C/EBP␣ ⌬/Ϫ mice showed no noticeable change in phenotype during the first 15 days (Phase I), after which the animals displayed marked growth retardation and died about 4 weeks after the injection of poly(I:C); their body weights were about one-third those of control mice (Fig. 4A). In an attempt to nutritionally "rescue" the animals, their regular diet was supplemented with 20% sucrose solution or liquid diet (Peptamen complete elemental diet) during the third and fourth weeks after the injection (Phase II). This treatment mildly improved their food consumption and extended their lifespan by 1-2 weeks (Fig. 4A, dotted line). When poly(I:C) was administrated to mice at 3 months of age, the resultant C/EBP␣ ⌬/Ϫ mice also exhibited a biphasic change in phenotype: i.e. they were normal during the first 15 days (Phase I) but during the subsequent 15 days (Phase II) had severe weight loss (Fig. 4B), hypo-FIGURE 2. Specificity and efficiency of the excision of C/EBP␣ F allele. A, detection of C/EBP␣ alleles in tissues. Mice were given a single injection of poly(I:C) (10 mg/kg of body weight) at 1 day or 3 months of age, and tissues were collected 2 days (white adipose tissue (WAT) and liver) or 28 days (all other tissues) later. Genomic DNA was then isolated and analyzed by Southern blotting as described under "Experimental Procedures." The genotypes of the animals (before the injection) are indicated at the top of the figure. B, excision efficiency of C/EBP␣ F allele in the C/EBP␣ F/Ϫ and C/EBP␣ F/F mice. The percentage of C/EBP␣ DNA remaining in various tissues was calculated from the results of Southern blotting analyses using the following equation: percentage of C/EBP␣ DNA remaining ϭ 50% Ϫ (50% ϫ the excision efficiency) (for C/EBP␣ F/Ϫ mice) or 100% Ϫ (100% Ϫ excision efficiency) (for C/EBP␣ F/F mice), where excision efficiency ϭ band intensity of C/EBP␣ ⌬ allele/(band intensity of C/EBP␣ F allele ϩ band intensity of C/EBP␣ ⌬ allele). The number of animals is indicated at the top of columns. For liver and white adipose tissue the results are expressed as means Ϯ S.E.; for other tissues the average value of two or three animals is presented. Gray box, mouse genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ; hatched box, mouse genotype ϭ Mx1-Cre ϩ ϩ C/EBP␣ ⌬/⌬ ; black box, mouse genotype ϭ Mx1-Cre ϩ ϩ C/EBP␣ ⌬/Ϫ . Liv., liver; Spl., Spleen; Pan., pancreas; Lun., lung; Kid., kidney; BAT, brown adipose tissue; Bra., brain. FIGURE 3. Kinetics of the ablation of C/EBP␣ in liver. Mice at 3 months of age were injected once with poly(I:C) (10 mg/kg of body weight). Groups of animals were killed daily for up to 4 days. Their livers were analyzed by Southern blotting (A), Northern blotting (B), and Western blotting (C) to determine the ablation kinetics of C/EBP␣ DNA, mRNA, and nuclear protein, respectively. The genotype of the mice (before the injection) is presented at the bottom of the panels, and the time after the injection is indicated on the top. A, B1, and C are different isoforms of C/EBP␣. phagia (Fig. 4C), and a mild decrease in water intake (data not shown). Taken together, these results indicate that ablation of C/EBP␣ leads to a biphasic response in metabolic phenotype and eventual death of the animals regardless of whether the gene was deleted early in life (neonates) or at 3 months of age.
To examine whether hypophagia was responsible for the severe weight loss noted in C/EBP␣ ⌬/Ϫ mice, control mice (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ) were pair-fed starting at day 18 after the injection of poly(I:C). The animals had free access to water, but their daily food consumption was rationed. An algorithm generated from the daily food consumption by the C/EBP␣ ⌬/Ϫ mice (Fig. 4C, dotted line) was used to calculate the quantity of food supplied daily to the pair-fed controls. The pair-fed controls exhibited a similar rate of weight loss as the C/EBP␣ ⌬/Ϫ mice during the pair-feeding period (Fig. 4B, dotted line), suggesting that hypophagia caused the weight loss observed in the C/EBP␣ ⌬/Ϫ mice. Although the pair-fed control mice lost weight, they did not exhibit the same severe metabolic abnormalities as did the C/EBP␣ ⌬/Ϫ mice.
Ablation of C/EBP␣ Alters the Deposition of Triglyceride in Tissues-C/EBP␣ ⌬/Ϫ mice lost triglyceride in their white but not brown adipose tissue (Fig. 5, A and B) at the end of Phase II irrespective of whether C/EBP␣ was ablated at 1 day or 3 months of age. Correspondingly, the size of the white adipose tissue was substantially smaller than in normaland pair-fed controls (Fig. 5C), but the brown adipose tissue was either larger than in normal-fed controls (injected with poly(I:C) at 1 day of age) or similar to that in normal-and pair-fed controls (injected with poly(I:C) at 3 months of age) (Fig. 5C). Additionally, C/EBP␣ ⌬/Ϫ mice, but not normal-or pair-fed controls, developed a marked fatty liver such that lipid infiltration was visible as white spots on the surface of the liver (Fig. 5D). Thus, C/EBP␣ is essential for maintaining the differentiated status of white but not brown adipose tissue in adult mice, and its ablation led to an abnormal deposition of fat in liver.
Impaired Energy Homeostasis in the C/EBP␣ ⌬/Ϫ Mice-Based on the observed hypophagia, C/EBP␣ ⌬/Ϫ mice are likely in an "energy crisis" during Phase II. Various energy-yielding metabolites in the blood were measured 28 days after the injection of poly(I:C). The concentrations of glucose, free fatty acids, and triglyceride were markedly lower (Ͻ50%) in C/EBP␣ ⌬/Ϫ mice as compared with normal-fed controls irrespective of whether the ablation of C/EBP␣ was performed on neonates or 3-month-old adults (TABLE ONE). The concentration of ␤-hydroxybutyrate in the plasma of C/EBP␣ ⌬/Ϫ mice was normal when ablation of C/EBP␣ was carried out 1 day after birth but was higher than in controls if the deletion was performed at 3 months of age (TABLE ONE). Ablation of C/EBP␣ also caused a decrease in cholesterol and an increase in total protein levels in the blood as compared with normal-or pair-fed controls. The decrease in levels of serum albumin and blood urea nitrogen appears to be dependent on the time of ablation of C/EBP␣. The hyperbilirubinemia, noted by Lee et al. (28) in their mice was not observed in our study (TABLE ONE) even though the level of mRNA for UDP-glucuronosyltransferase-1, the enzyme required for the conjugation of glucuronic acid to bilirubin, was reduced in the liver of C/EBP␣ ⌬/Ϫ mice to 40% that of the control values (TABLE TWO).
Next, we assessed the levels of hormones that regulate food intake. Surprisingly, the concentrations of leptin and glucagon-like peptide-1 were not elevated in the C/EBP␣ ⌬/Ϫ mice; the level of insulin was undetectable (injection of poly(I:C) at 1 day of age) or strikingly low (injection FIGURE 4. Growth retardation and hypophagia of C/EBP␣ ⌬/؊ mice. A, growth retardation of C/EBP␣ ⌬/Ϫ mice that were injected with poly(I:C) at 1 day of age. Body weights were recorded beginning on day 14 after a single injection of poly(I:C) (10 mg/kg of body weight) to 1-day-old mice. The dotted line represents the weights of C/EBP␣ ⌬/Ϫ mice, when 20% sucrose-water was supplemented to their diet. The values are expressed as means Ϯ S.E. for three C/EBP␣ ⌬/Ϫ mice and six controls. The inset shows a female C/EBP␣ ⌬/Ϫ mouse (7.4 g) compared with a control littermate (19.7 g) on day 28 after the injection. B, weight loss of C/EBP␣ ⌬/Ϫ mice that were injected with poly(I:C) at 3 months of age. Body weights of animals were measured every other day after poly(I:C) (10 mg/kg of body weight) was injected at 3 months of age. The change in body weight was calculated as the average daily body weight minus the starting body weight (body weight at time of the injection). The dotted line describes the change of body weight for pair-fed controls, starting on day 18 after injection. The values are expressed as means Ϯ S.E. for four to five animals. C, hypophagia of C/EBP␣ ⌬/Ϫ mice. Food consumption was measured every other day for mice that received a single injection of poly(I:C) at 3 months of age. The results are expressed as means Ϯ S.E. for four C/EBP␣ ⌬/Ϫ mice (black bars) and eight normal-fed control littermates (gray bars, genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ). The food consumption of adult C/EBP␣ ⌬/Ϫ animals, starting on day 18 after injection of poly(I:C), was fitted to a curve (dotted line). An algorithm, that describes the curve as y ϭ z Ϫ w ϫ 0.115655 ϩ w ϫ 16.36/e (0.2765 ϫ x) was used to calculate the food supply for pair-fed controls (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ), where y is the daily food supply (g), w is the starting body weight, z is the food intake on day 18 after the injection of poly(I:C), and x is the number of days after the injection. of poly(I:C) at 3 months of age) (TABLE ONE). Hypoinsulinemia accompanied by hypoglycemia (TABLE ONE) is an unusual metabolic situation. Because administration of glucose to C/EBP␣ ⌬/Ϫ mice increased the level of insulin in blood (data not shown), it is possible that the lack of C/EBP␣ hinders secretion of insulin by the pancreas or accelerates its degradation. Alternatively, the observed hypophagia (Fig. 4C), in combination with the inability of the C/EBP␣ ⌬/Ϫ mice to maintain hepatic gluconeogenesis and glycogenolysis (TABLE TWO; see Fig. 7A), could lead to a diminished concentration of blood glucose, resulting in a lowered level of insulin secretion by the pancreas.
Altered Hepatic Gene Expression Contributes to the Metabolic Defects Noted in C/EBP␣ ⌬/Ϫ Mice-DNA microarray analysis was used to measure mRNA levels for potentially relevant genes in the liver (TABLE  TWO) to explore the molecular basis for the metabolic changes caused by the absence of C/EBP␣. Because C/EBP␣ ⌬/Ϫ mice have marginal levels of hepatic glycogen at the end of Phase II (Fig. 6A), the enzymes that are involved in glycogen metabolism were studied. There was a 2.8-fold decrease in UDP-glucose pyrophosphorylase mRNA level in the liver of C/EBP␣ ⌬/Ϫ mice, as determined by using five different probes (Fig. 6B). Glycogen synthase and glycogen phosphorylase mRNA levels were also decreased by 2.1-and 4.5-fold, respectively, whereas branching enzyme I increased by 3.3-fold (Fig. 6B). Pair-fed controls also had little hepatic glycogen (Fig. 6A). However, in these animals, hepatic mRNA levels for glycogen synthase, glucokinase, glucose-6phosphatase, and PEPCK-C were markedly higher than those in C/EBP␣ ⌬/Ϫ mice (Fig. 7A), prompting the conclusion that decreased expression of enzymes for glycogen metabolism is the primary cause for the lack of hepatic glycogen in C/EBP␣ ⌬/Ϫ mice. Hypophagia presumably exacerbates this process.
C/EBP␣ ⌬/Ϫ mice also had a marked decrease in the levels of hepatic mRNA for proteins that are involved in fatty acid oxidation and lipid transport (i.e. apolipoproteins) (TABLE TWO). Such changes suggest that a decreased rate of both fatty acid oxidation and lipoprotein export are involved in the lipid infiltration observed in the livers of C/EBP␣ ⌬/Ϫ mice. A decrease in the level of mRNA for hydroxymethylglutaryl-coenzyme A reductase (TABLE TWO) is also consistent with the reduced concentration of cholesterol noted in the blood of C/EBP␣ ⌬/Ϫ mice (TABLE ONE). In addition to the changes in mRNAs for various enzymes, there were marked alterations in mRNA levels of several transcription factors: C/EBP␦ (6.8-fold increase), PPAR␣ (5.5-fold decrease), c-Jun (2.2-fold increase), Jun-B (3.1-fold increase), and SREBP-1 (7-fold decrease) (TABLE TWO, Fig. 7B). The consequence of these large changes in the mRNA levels of these transcription factors remains to be further investigated.
Effect of the Ablation of C/EBP␣ on Glucose Homeostasis-To investigate how the ablation of C/EBP␣ in adult mice affects glucose homeostasis, the concentration of glucose in blood was measured. C/EBP␣ ⌬/Ϫ mice maintained normal blood glucose levels in fed or fast- FIGURE 5. Lipodystrophy in white adipose tissue and fat deposition in the livers of C/EBP␣ ⌬/؊ mice. After receiving poly(I:C) (10 mg/kg of body weight) at 1 day or 3 months of age, all mice had free access to food and water except pair-fed controls. The daily food supply of the pair-fed mice was rationed starting on day 18 after the injection.
The genotypes of control animals are Mx1-Cre Ϫ ϩ C/EBP␣ F/F . A, lipodystrophy in the white adipose tissue (WAT) of C/EBP␣ ⌬/Ϫ mice. The abdominal area of mice, which received an injection of poly(I:C) 1 day after birth, is shown on the left side; epididymal fat from mice injected with poly(I:C) at 3 months of age is aligned on the right side. Lipodystrophy is indicated by the disappearance of triglyceride in the subcutaneous, peri-ovarian, and epididymal white adipose tissues in C/EBP␣ ⌬/Ϫ mice. B, relatively normal brown adipose tissue of C/EBP␣ ⌬/Ϫ mice. Hematoxylin-and-eosin staining of brown adipose tissue is shown (ϫ400). Solid arrow, oil droplets; broken arrow, nuclei. C, relative weights of white and brown adipose tissues. Values are plotted as percentage of body weight. The results are expressed as the means Ϯ S.E. from 6 C/EBP␣ ⌬/Ϫ mice (black bars), 16 normal-fed controls (gray bars), and 7 pair-fed controls (hatched bars). D, fatty liver in C/EBP␣ ⌬/Ϫ mice. Fat accumulation in the liver of C/EBP␣ ⌬/Ϫ mice is visible on the surface as white spots (upper) and detectable inside the tissue as intense red dots from Oil red O staining (lower) (ϫ400).
ing (18 h) conditions during Phase I (Fig. 8A). However, during the third week after injection of poly(I:C), their ability to maintain a normal concentration of blood glucose was attenuated, and it was severely impaired in the fourth week. Most of the C/EBP␣ ⌬/Ϫ mice could not tolerate an 18-h fast in the third week. C/EBP␣ ⌬/Ϫ mice that had free access to food lived up to the fourth week, but their fed blood glucose concentration was 30 -40% that noted in the normal-or pair-fed controls (Fig. 8A, TABLE TWO). To test whether C/EBP␣ is required for the normal response to a diabetogenic stimulus, we injected C/EBP␣ ⌬/Ϫ mice with streptozotocin at 11 days after the injection of poly(I:C) to induce diabetes. The C/EBP␣ ⌬/Ϫ mice became diabetic in a manner similar to the control animals (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ) (Fig. 8B). Next, the levels of mRNA for key regulatory enzymes that influence glucose homeostasis were determined at selected intervals after the injection of

Profile of metabolites in the plasma of C/EBP␣ ⌬/؊ mice and littermate controls
A single injection (10 mg/kg of body weight) of poly(I:C) was administrated to mice at either 1 day or 3 months of age. C/EBP␣ ⌬/Ϫ mice and controls had free access to food and water. The daily food consumption of pair-fed controls was rationed as described in Fig. 4C. Blood was collected when the animals were killed 28 days after injection of poly(I:C). The results are expressed as the means Ϯ S.E. for the number of animals indicated in parentheses. The p value is calculated for C/EBP␣ ⌬/Ϫ mice and normal-fed control mice, and the pЈ value is calculated for C/EBP␣ ⌬/Ϫ and pair-fed controls.   1 and 2). b The average -fold changes, noted in the DNA array analysis, were confirmed with the data from Northern blotting that were performed in this and other studies. NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 poly(I:C). Fig. 8C shows that C/EBP␣ ⌬/Ϫ mice and control animals had similar levels of mRNA for PEPCK-C, glucose-6-phosphatase, glycogen synthase, glucokinase, C/EBP␤, C/EBP␦, and apolipoprotein C3 at day 2 or 8 after the injection of poly(I:C). However, marked differences were noted on day 22 (Fig. 8C), when the levels of hepatic mRNA for the above-referenced genes were dramatically altered (with the exception of C/EBP␤); these changes resembled those noted in the DNA microarray analysis for the same mRNAs at day 28 (TABLE TWO). This pattern of hepatic gene expression most likely accounts for the changes in glucose homeostasis noted during the progression of the C/EBP␣ ⌬/Ϫ mice from Phase I to II. C/EBP␣ Is Optional for Bt 2 cAMP-stimulated Transcription of PEPCK-C in the Liver of Adult Mice-The mRNA for PEPCK, one of the key enzymes in hepatic and renal gluconeogensis, is undetectable in the liver of C/EBP␣ Ϫ/Ϫ mice at birth (8,11), and in such animals its mRNA cannot be induced by the administration of Bt 2 cAMP (25). It has been suggested that C/EBP␣ is required for the transcriptional response of the gene for PEPCK-C to the cyclic nucleotide (25)(26)(27). To further investigate this possibility Bt 2 cAMP was administrated to C/EBP␣ ⌬/Ϫ mice. The result was an increase in the levels of mRNAs for PEPCK-C and glucose-6-phosphatase (Fig. 9) to the same extent as noted in the livers of normal-fed controls, although the basal levels of mRNAs for both enzymes were much lower in C/EBP␣ ⌬/Ϫ mice (Fig. 9). This finding suggests that in livers of adult mice C/EBP␣ is critical for the basal but not the cAMP-stimulated induction of gluconeogenic enzymes, such as PEPCK-C and glucose-6-phosphatase.

Advantages of Our Animal Model for Metabolic
Studies-A number of major metabolic defects noted in C/EBP␣-deficient mice, such as hypoglycemia and hyperammonemia, have established C/EBP␣ as critical for the perinatal development of metabolic processes (8 -10). In the absence of C/EBP␣, the animals die within 30 min after birth. It is not clear, however, how important this transcription factor is after the perinatal period. One line of evidence suggests that C/EBP␤ can partially replace C/EBP␣ in the liver (7) and hematopoietic cells (38) but not in white adipose tissue (7), if the expression of C/EBP␤ is controlled by an endogenous C/EBP␣ gene promoter. The fact that one isoform of C/EBP can substitute for another complicates the effort to examine the loss of function of C/EBP␣ in animals. Another problem is that most animal models in which the gene for C/EBP␣ has been ablated have mainly focused on liver (28,29) or adipose tissue (30). Although such models provide valuable insights into the role of C/EBP␣ in those tissues, they provide no information on the effect of a total loss of C/EBP␣ activity in the whole animal.
In this study we have created an animal model that has several advantages for studying the role of C/EBP␣ in the postnatal development of metabolic processes. First, the ablation of C/EBP␣ occurs in multiple tissues of adult mice, particularly in C/EBP␣-rich tissues such as liver, white and brown adipose tissue, lung (Fig. 2), and bone marrow (19). Accordingly, the loss of function of C/EBP␣ was studied in the whole animal rather than in individual tissues. Second, the ablation of c/ebp␣ is rapid and virtually complete, presumably because Cre recombinase only has to excise one allele of c/ebp␣ in C/EBP␣ F/Ϫ mice. Third, C/EBP␣ can be ablated at any time after birth, which allowed one to study C/EBP␣ with reference to the control of metabolism at various developmental stages. To maximally delete c/ebp␣ in our model, we modified the procedure of Kuhn et al. (31) by using C/EBP␣ F/Ϫ mice instead of C/EBP␣ F/F mice. Consequently, the overall percentage of deletion of the gene was greatly improved in all tissues measured, especially in nonhepatic tissues where Cre recombinase-mediated excision has been shown to be inefficient (31) (Fig. 2B). As part of our strategy, new procedures, including PCR-based genotyping and Southern blotting with the NotI fragment, were developed to solve the increasing complexity of genotyping in C/EBP␣ F/Ϫ mice, as compared with C/EBP␣ F/F mice.
The Two-phase Response of C/EBP␣ ⌬/Ϫ Mice-Ablation of C/EBP␣ is lethal; the animals usually die approximately 1 month following the ablation. This distinguishes our model from others where C/EBP␣ is often ablated only in liver (28,29) or in adipose tissue (30). Clearly, C/EBP␣ is required in multiple tissues to maintain normal metabolic and developmental processes. We consistently noted a biphasic response of mice to the loss of C/EBP␣, no matter whether deletion of the gene occurred at 1 day or 3 months after birth. C/EBP␣ is known to control the transcription of a number of genes and is thus likely to have a direct effect on global gene transcription (39,40). We noted few overt metabolic alterations or changes in mRNA levels in the C/EBP␣ ⌬/Ϫ mice during Phase I. In Phase II, however, there was severe deterioration of health/metabolism, which was apparently associated with marked changes in mRNA levels for the enzymes that are directly involved in hepatic metabolism, such as PEPCK-C, glucose-6-phosphatase, glycogen synthase, and glucokinase. We also noted alterations in the concentrations of mRNAs for transcription factors such as C/EBP␦, SREBP-1, and PPAR␣, which regulate expression of genes coding for metabolically important proteins. The promoters of these genes have multiple C/EBP␣-binding sites, which suggests that ablation of C/EBP␣ is responsible (either directly or indirectly) for the altered concentration of mRNA of these genes. It is not clear why the loss of C/EBP␣ causes an increase in the levels of C/EBP␦ and c-Jun. The simplest explanation is that C/EBP␣ inhibits transcription of the genes for these proteins, although this remains to be tested.
Because the ablation of C/EBP␣ was rapid and extensive in most of the tissues studied (its half-life is less than 12 h in the liver) (Fig. 3), our current working model to explain the two-phase response of mice is that ablation of C/EBP␣ causes some major molecular alterations, which fully manifest themselves within a 2-week period after ablation of C/EBP␣ (length of Phase I); these alteration consequently lead to metabolic derangements seen in Phase II and the ultimate death of the animals. It is unlikely that these molecular alterations occur only in the liver, since excision of hepatic c/ebp␣ either at birth (29) or in the adult (28) does not result in the death of animals. Hypophagia is also unlikely to be a major factor, since pair-feeding the animals caused weight loss but not the other abnormalities noted in our study. Because C/EBP␣ normally controls the transcription of many genes involved in both metabolism and cellular differentiation (39,40), its absence probably initiates a pleiotropic response in a variety of tissues, not just the liver. The accumulated effect of alterations in the transcription of critical genes then leads to the general decline in health of the animal, as noted in Phase II. Interestingly, Zhang et al. (19) have deleted c/ebp␣ in mice 2 days after birth and noted the obstruction of granulocyte development and a 30-fold accumulation of blasts in bone marrow. They suggested that their C/EBP␣-deficient mice (C/EBP␣ ⌬/⌬ mice) died from sepsis as a result of granulocytopenia. This raises a possibility that C/EBP␣ ⌬/Ϫ mice also die from sepsis, because the degree of ablation of C/EBP␣ in the bone marrow of C/EBP␣ ⌬/Ϫ mice should be close to, if not greater than that in C/EBP␣ ⌬/⌬ mice. However, we have noted that the onset of hypophagia in the C/EBP␣ ⌬/Ϫ mice occurs at almost exactly the third week after ablation of c/ebp␣. This would require that sepsis develops in a virtually synchronous manner in the mice. We also determined the body temperature of C/EBP␣ ⌬/Ϫ mice both at room temperature and at 4°C and found no difference from littermate controls. Also, it is possible that the large accumulation of blasts produce cytokines that selectively alter metabolic processes and, subsequently, a decrease in appetite. The above possibilities need to be investigated.
C/EBP␣ and the Postnatal Development of Metabolic Processes-C/EBP␣ has been proposed to be "a central regulator of energy metabolism" (41), presumably because of its participation in the differentiation of adipose tissue and its regulation of glucose homeostasis and other critical metabolic processes. The results of our study are consistent with this proposition. Ablation of C/EBP␣ in adult mice clearly creates an energy crisis, which results from the extremely low level of energy reserves available to the animals in Phase II. For example, the C/EBP␣ ⌬/Ϫ mice have no hepatic glycogen, lose triglyceride in white adipose tissue, and develop hypophagia and fatty liver. An extreme manifestation of such crisis is the co-existence of hypoinsulinemia and hypoglycemia (TABLE ONE). Normally, a low level of insulin is accompanied by hyperglycemia, since a lack of insulin impairs glucose uptake in peripheral tissues and leads to an increased hepatic gluconeogenesis. However, C/EBP␣ ⌬/Ϫ mice consume abnormally low amounts of dietary glucose (resulting in hypophagia) and probably carry out decreased gluconeogenesis due to reduced levels of mRNA for gluconeogenic enzymes. Aberrations in both carbohydrate and lipid metabolism in the C/EBP␣ ⌬/Ϫ mice likely underlie the energy crisis apparent in these animals. The exact trigger for these changes remains to be determined.
Although C/EBP␣ is known to be essential for adipogenesis in white and brown adipose tissues (7,8,30,42), it is not clear whether it is required to maintain the differentiated status of these tissues. Because C/EBP␣ ⌬/Ϫ mice have severe lipodystrophy in their white adipose tissue regardless of the time of ablation, C/EBP␣ must be essential not only for postnatal development but also for the maintenance of the differentiated state of the white adipose tissue. Hypophagia appears to accelerate these processes, but it is not the cause, since pair-fed controls had relatively normal white adipose tissue. In agreement with earlier studies (7,30), the postnatal development of brown adipose tissue is not dependent on C/EBP␣. Interestingly, fat deposition in the liver of C/EBP␣ ⌬/Ϫ mice appears to be caused by an impaired hepatic function, likely resulting from decreased rates of hepatic fatty acid oxidation and apolipoprotein synthesis. This conclusion is consistent with the lowered levels of triglyceride and free fatty acids in the blood of C/EBP␣ ⌬/Ϫ mice.
During the perinatal period C/EBP␣ is thought to control glucose homeostasis, presumably by regulating the transcription of genes coding for key gluconeogenic enzymes, such as PEPCK-C and glucose-6phosphatase. Hepatic PEPCK-C is regulated at both basal (8) and stimulated (25) levels by C/EBP␣ during the perinatal period. Beyond the perinatal period, the importance of C/EBP␣ in the regulation of glucose homeostasis is not clear. In one study, Lee et al. (28) deleted the gene for C/EBP␣ in the liver of adult mice using an adenoviral vector expressing Cre recombinase and noted a decrease in PEPCK-C mRNA. In another study Inoue et al. (29) reported no change in hepatic PEPCK-C mRNA when c/ebp␣ was deleted in the liver of neonatal mice by Cre recombinase produced from the transgene of albumin-Cre. In the present study we have shown that the expression of hepatic PEPCK-C mRNA is biphasic; i.e. normal in Phase I but reduced dramatically during Phase II FIGURE 8. Glucose homeostasis in C/EBP␣ ⌬/؊ mice. A, glucose concentration in the blood. Glucose levels in the blood were measured in mice at various times after they received a single injection of poly(I:C) (10 mg/kg body weight) at the age of 3 months. The values are expressed as means Ϯ S.E. for the number of animals indicated on the top of each column. Gray bars, control animals (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ); black bars, C/EBP␣ ⌬/Ϫ mice; hatched bars, pair-fed controls (genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ). B, streptozotocin-induced diabetes in C/EBP␣ ⌬/Ϫ mice. Streptozotocin (STZ) was administrated to mice at day 11 after injection of poly(I:C). The glucose level was then measured daily using blood that was collected from the tail vein. The values are expressed as means Ϯ S.E. for three animals in each group. C, biphasic expression profile of FIGURE 9. Bt 2 cAMP-stimulated induction of gene expression for PEPCK-C and glucose-6-phosphatase in the liver of C/EBP␣ ⌬/؊ mice. Bt2cAMP (35 mg/kg of body weight) or saline was administrated to normal-fed controls (gray bar, genotype ϭ Mx1-Cre Ϫ ϩ C/EBP␣ F/F ) and C/EBP␣ ⌬/Ϫ mice (black bar) 28 days after the injection of poly(I:C). The levels of mRNA for PEPCK-C and glucose-6-phosphatase were determined by Northern blotting. The values for control animals treated with saline were arbitrarily set as 1.0. The results are expressed as means Ϯ S.E. for three animals in each group.
selected genes in the liver of C/EBP␣ ⌬/Ϫ mice. Mice were killed at day 2, 8, and 22 after the injection of poly(I:C). For the 2-and 8-day time points, the mice were fasted 18 h before killing; mice analyzed at day 22 were fed. Total RNA was isolated from liver, and the expression of selected genes was determined by Northern blotting. The mouse genotype (before the injection) is indicated at the bottom of the figure. Glc-6-Pase, glucose-6-phosphatase; GS, glycogen synthase. (Fig. 8C). In addition, the expression of PEPCK-C can still be stimulated to the same extent as noted in controls by the administration of Bt 2 cAMP even though the basal level of expression for PEPCK-C is strikingly lower than noted in the livers of control mice (Fig. 9). Thus, C/EBP␣ is important for basal transcription of the gene for PEPCK-C but not for its induction by the cyclic nucleotide in adult liver. Perhaps another member of the basic region-leucine zipper family of transcription factors (e.g. C/EBP␤, C/EBP␦, or CREB) partially assumes the function of C/EBP␣ in the liver of the C/EBP␣ ⌬/Ϫ mice.
Finally, the phenotypic differences between C/EBP␣ ⌬/Ϫ mice and the mice generated by Chen et al. (7) raise another interesting point, i.e. the time of C/EBP␣ expression is important to the development of metabolic processes. When C/EBP␤ is expressed under the control of the endogenous promoter of C/EBP␣, it can functionally replace C/EBP␣ in liver (7) or hematopoietic cells (38). However, unchanged mRNA levels of C/EBP␤ in postnatal development does not fully compensate for the loss of C/EBP␣ in C/EBP␣ ⌬/Ϫ mice. Thus, when and where the gene for C/EBP␣ is expressed are critical for the normal development of the mouse. For example, C/EBP␣ Ϫ/Ϫ mice die within 30 min after birth if c/ebp␣ is missing during fetal development (8,11). In contrast, if the gene is deleted 1 day after birth, the animals live for about 30 days. Because the promoter of the C/EBP␣ gene is functional as early as fetal day 13, C/EBP␣ is likely required for the initial transcription of genes that allow the animal to survive the perinatal period, especially as patterning of metabolic processes is probably completed between day 13 of fetal life and birth (8 -11). After this time, ablating C/EBP␣, whether it is at 1 day after birth or 3 months later, has the same consequence (i.e. death within 30 days). This underlines the key role of developmentally appropriate expression of genes that code for transcription factors, since they have a broad array of effects.