ABSTRACT
Zbtb20 is a member of the POK family of proteins, which function primarily as transcriptional repressors via interactions mediated by their conserved C2H2 Krüppel type zinc finger and BTB/POZ domains. To define the function of Zbtb20 in vivo, we generated knockout mice by homologous recombination. Zbtb20 null mice display a stark phenotype characterized by postnatal growth retardation, metabolic dysfunction, and lethality. Zbtb20 knockout mice displayed abnormal glucose homeostasis, hormonal responses, and depletion of energy stores, consistent with an energetic deficit. Additionally, increased serum bilirubin and alanine aminotransferase levels were suggestive of liver dysfunction. To identify potential liver-specific Zbtb20 target genes, we performed transcript profiling studies on liver tissue from Zbtb20 knockout mice and wild-type littermate controls. These studies identified sets of genes involved in growth, metabolism, and detoxification that were differentially regulated in Zbtb20 knockout liver. Transgenic mice expressing Zbtb20 in the liver were generated and crossed onto the Zbtb20 knockout background, which resulted in no significant normalization of growth or glucose metabolism but a significant increase in life span compared to controls. These data indicate that the phenotype of Zbtb20 knockout mice results from liver-dependent and -independent defects, suggesting that Zbtb20 plays nonredundant roles in multiple organ systems.
Precise regulation of gene expression is an essential point of control during a diverse range of biological processes including development, cancer, and immune responses. The orchestrated recruitment of protein factors to sequence-specific DNA regulatory elements is required for adoption of permissive chromatin structures and the assembly of the molecular machinery required for transcription of nascent RNA species. During the course of evolution a multitude of protein families have evolved to regulate these processes in higher eukaryotes. While our understanding of the basic processes of gene regulation has grown immensely, genome-wide DNA sequencing initiatives have revealed the presence of many putative transcriptional regulators in mammalian genomes whose functions are currently undefined. A more complete understanding of transcriptional mechanisms awaits the assignment of function to such molecules.
The C2H2 Krüppel-type zinc finger is a 25- to 30-amino-acid protein domain and is one of the most common types of DNA-binding domains, with over 600 human genes encoding proteins containing this motif (3, 15). A subset of Krüppel type zinc finger proteins also contain the BTB/POZ (Broad Complex, Tramtrack, Bric a brac [BTB] or poxvirus and zinc finger [POZ]) domain, which mediates protein-protein interactions and is required for the assembly of homomeric and heteromeric complexes between BTB/POZ domain family members. This subfamily of POK (POZ and Krüppel) transcriptional regulators comprises 5 to 10% of the Krüppel family, or some ∼40 to 60 members in humans and mice (4). The functions of numerous POK family members have been defined by biochemical, molecular, and genetic approaches. The prevailing theme from these studies is that POK family members function in a diverse range of biological contexts, acting predominantly as transcriptional repressors that modulate gene expression via recruitment of corepressor complexes, including proteins such as histone deacetylases, mSin3a, SMRT, and NCoR (3, 4, 15).
The human Zbtb20 (also known as DPZF, HOF, and Zfp288) gene is located on chromosome 3 and codes for a 741-amino-acid protein (34), while the mouse homolog is located on chromosome 16. To date only a few literature reports defining the function of Zbtb20 exist. Yeast two-hybrid systems demonstrated that the BTB/POZ domain of Zbtb20 was able to mediate homomeric interactions (24). These experiments also characterized Zbtb20 expression patterns in various regions of the brain, including the hippocampus, and identified a potential role for Zbtb20 in neurogenesis (24). In a subsequent study, expression in mice of Zbtb20 transgenes under the control of a forebrain-specific promoter led to the formation of hippocampus-like neuronal structures and resultant behavioral abnormalities (25). These data indicate that regulation of Zbtb20 expression level can control pathways of development/differentiation, potentially by the recruitment of protein complexes to sequence-specific DNA elements and regulation of gene expression.
The aim of this study was to define the in vivo function of the novel POK family member Zbtb20 via the generation of Zbtb20 knockout mice. The resultant phenotype of the Zbtb20-deficient mice indicates that Zbtb20 plays a nonredundant role in the control of postnatal metabolism, growth, and survival. Complementation of elements of this phenotype by a liver Zbtb20 transgene identified liver-dependent and liver-independent functions of Zbtb20 in vivo.
MATERIALS AND METHODS
Mice.All mice were housed in microisolator cages under specific-pathogen-free conditions at the Harvard School of Public Health and Second Military Medical University (Shanghai, China). All animal studies were performed according to institutional and National Institutes of Health guidelines for animal use and care. Mice were kept on Picolab Mouse Diet 20 no. 5058 (Labdiet) containing 9% fat.
Anti-Zbtb20 antibody generation.Rabbit polyclonal antiserum was generated by Sigma Genosys (Sigma Aldrich) via immunization with a purified N-terminal Zbtb20 protein fragment and affinity purified over a Zbtb20-coupled AffiGel-10-AffiGel-15 mixture (Bio-Rad).
Knockout mouse construction and maintenance.A 14.5-kb genomic DNA fragment of the Zbtb20 gene was isolated from a 129 mouse genomic library (Stratagene, La Jolla, CA) with a probe containing full-length cDNA of the human Zbtb20 gene and used to construct the targeting vector. The 1.1-kb short arm was derived from a sequence upstream of the XhoI site in exon 6, and the 6.4-kb long arm was derived from intron 7. A loxP-flanked neomycin resistance cassette was installed in the middle of the arms to replace a 1.8-kb genomic fragment containing most of Zbtb20 gene exon 6 by homologous recombination. The targeting plasmid was linearized by NotI and electroporated into PC3 embryonic stem cells (129/SvJae background) with protamine-Cre recombinase. Homologous recombination at the Zbtb20 gene locus was confirmed by Southern blotting, and Zbtb20+/− embryonic stem cells were injected into C57BL/6 blastocysts and transferred into CD1 pseudopregnant mice. The resulting male chimeras transmitted the mutant allele to the progeny, and the neomycin resistance cassette was removed by Cre-mediated recombination in the male germ line. The Zbtb20 null allele was backcrossed to the BALB/c and 129 backgrounds for 10 and 11 generations, respectively, and Zbtb20 null mice and controls from these inbred lines were used for experiments presented in Fig. 2, 4, and 5 and Tables 1 to 3. In addition, the Zbtb20 null allele was maintained on a C57BL/6 × 129 mixed background, and these mice were used for the experiments presented in Fig. 3 and 6.
Quantitation of serum analytes in Zbtb20 knockout mice and littermate controls
Transgenic-mouse construction and maintenance.The Zbtb20 transgenic construct was generated by cloning the full-length murine Zbtb20 gene cDNA into the pLIV.7 transgenic expression vector between the human ApoE gene 5′ 3 kb plus exons 1 and 2 and the human ApoE gene 3′ 254 bp plus a liver element (19). The purified plasmid was linearized using the NotI and SpeI restriction sites and injected into C57BL/6 fertilized embryos and implanted into pseudopregnant females. Founder lines were identified by Southern blotting and maintained as heterozygotes for experimentation.
Tissue collection and quantitative reverse transcription-PCR (RT-PCR).Mice were euthanized by CO2 narcosis, and harvested tissues were snap-frozen in liquid nitrogen. Total RNA was purified using Trizol reagents, and 2 μg was used for cDNA synthesis. Real-time quantitative PCR was performed using a Stratagene Mx3005P quantitative PCR system.
Protein extraction and Western blotting.Total protein lysates were prepared from 1 × 107 HEPA1-6 cells or ∼100 mg of tissue in 500 μl of 1× radioimmunoprecipitation assay buffer with protease inhibitors (Roche). Western blots were preblocked in 5% nonfat skim milk for 1 h and incubated with a 1:1,000 dilution of affinity-purified anti-Zbtb20 rabbit polyclonal antisera (Sigma Genosys). Blots were washed three times with Tris-buffered saline-Tween 20 (TBS-T), incubated with 1:10,000-diluted horseradish peroxidase-conjugated goat anti-rabbit antibodies (Jackson Laboratories), washed three times with TBS-T, and visualized with SuperSignal West Pico chemiluminescent substrate (Thermo Scientific).
Glucose measurements and glucose and glucagon tolerance tests.Blood glucose levels in tail vein blood samples were measured using the Ascensia Contour glucometer (Bayer). Glucose tolerance tests (2 g/kg of body weight, intraperitoneal) were performed in the fasting state, and insulin tolerance testing (1 U/kg body weight) was performed in the fed state. Blood glucose measurements were taken over the subsequent 1-h period. Glucagon tolerance tests were performed with a single bolus of glucagon delivered intraperitoneally, at a dose of 30 μg/kg body weight, and blood glucose measurements were taken over the subsequent 2-h period.
Serum metabolites and enzyme-linked immunosorbent assays.Serum samples were collected from ∼17-day-old Zbtb20 knockout and control mice that had been fed ad libitum, and metabolic parameters were assessed by Charles River Laboratories Clinical Pathology services using test panels CH1 and CH5 (see Table 1). Lactate, glycerol, nonesterified fatty acids, β-hydroxybutyrate (BHB), and glucagon were detected with commercial assay kits according to the manufacturer's instructions.
Serum hormones shown in Table 3 were quantitated with a mouse adipokine Luminex kit (Lincoplex).
Glycogen quantitation.Glycogen measurements were performed on tissue isolated from animals fed ad libitum. Six volumes of 1 M NaOH (∼300 μl) was added to ∼100 mg of tissue, and the mixture was mixed well and incubated at 55°C for 1 h. It was neutralized with an equal volume of 1 M HCl and spun at 13,000 rpm for 10 min at 4°C in a microcentrifuge. The lysate supernatant was removed and incubated for 2 h with 0.2 μl amyloglucosidase (1 mg/ml stock; Sigma) at 37°C. Twenty to 50 μl of the digested lysate was used in a glucose oxidase quantitation assay (Sigma), and total glycogen per milligram of tissue was determined accordingly. Periodic acid-Schiff staining was performed by placing deparaffinized sections into 0.5% periodic acid solution for 5 min. Slides were then incubated in Schiff reagent for 15 min, washed in lukewarm tap water for 5 min, and counterstained in hematoxylin for 1 min.
Transcript profiling.Total RNA was extracted from isolated liver tissue with Trizol reagent. Four micrograms of total RNA was used in each GeneChip in vitro transcription one-cycle labeling reaction, and 15 μg of labeled cRNA was added to each hybridization cocktail for subsequent hybridization to Affymetrix mouse 430 2.0 experimental arrays. Data were analyzed with the Affymetrix microarray suite, version 5.1 (MAS 5.1), to determine signal and call parameters; further data manipulation was performed in Microsoft Excel. Gene ontology analyses were performed with GeneSifter software (http://www.genesifter.net ). HeatMap images were prepared with GenePattern software (http://www.broad.mit.edu/cancer/software/genepattern/ ). The minimum (blue), median (white), and maximum (red) probe signals were calculated and graphed for each probe set independently.
Statistical analysis.Statistical significance for data presented in Tables 1 and 3 (except glucagon) was determined using one-way analysis of variance; for in vivo studies it was determined using the log rank (Mantel-Cox) test with Graph Pad Prism 5.0.1 software. Statistical significance for all other parameters was determined using Student's t test to compare Zbtb20 knockout animals with wild-type controls.
RESULTS
To determine the function of Zbtb20 in vivo, we generated Zbtb20 knockout mice via homologous recombination. The Zbtb20 gene comprises seven exons, with the majority of the coding sequence present in the final two exons. Exon 6 contains the coding sequences for the BTB/POZ domain and some zinc finger motifs, with those for the remaining DNA binding elements present in exon 7. We chose to target exon 6 for deletion, as we reasoned that this would yield a protein that lacked the protein domains most likely to be essential for function. Figure 1A displays the cloning strategy used to generate the Zbtb20 knockout construct and resultant mice. Homozygotes for the Zbtb20 null allele were generated by intercrossing heterozygous parents. Mice were genotyped for the wild-type and null alleles via Southern blotting (Fig. 1B) and PCR (Fig. 1C). Liver tissue was harvested from Zbtb20 knockout and control mice, nuclear protein lysates were isolated, and Zbtb20 protein expression was defined by Western blotting using a polyclonal anti-Zbtb20 antiserum (Fig. 1D). These data indicate that Zbtb20 knockout mice lack expression of Zbtb20 protein, indicating that the strategy was successful in targeting Zbtb20 and validating this model for study of the in vivo functions of Zbtb20.
Zbtb20 knockout mouse generation and validation. (A) Schematic demonstration for the gene targeting of the Zbtb20 gene. The probe 5′ is used for Southern blot analysis of EcoRV-digested genomic DNA, which detects a 7.3-kb band for the wild-type (wt) allele, a 5-kb band for the homologously recombined allele with a neomycin resistance cassette (Neo), or a 3.2-kb band for a Neo-deleted allele created by Cre recombinantion. The loxP sites are represented as filled triangles. V, EcoRV; X, XhoI; H, HpaI; E, EcoRI; D, DraIII; C, ClaI; S, SalI; N, NotI; HSV-TK, herpes simplex virus thymidine kinase cassette. (B) Southern blotting identified products specific to the wild-type (WT; 7.3-kb) and knockout (KO; 3.2-kb) alleles in homozygous and heterozygous mice. (C) PCR-amplified products specific to the wild-type (∼100-bp) and knockout (∼300-bp) alleles in homozygous and heterozygous mice. (D) Western blotting was performed on nuclear fractions from the livers of Zbtb20 knockout mice and littermate controls using an anti-Zbtb20 polyclonal antiserum. An antibody against the nuclear protein Sp1 was used as a protein loading control.
Zbtb20 knockout pups were born at Mendelian frequencies (data not shown) but began to display a series of abnormalities compared to wild-type and heterozygous controls at ∼10 to 15 days of age. Most prominently, Zbtb20 knockout mice exhibited severe growth retardation, as demonstrated by reduced body mass (Fig. 2A) and decreased axial growth (Fig. 2B). These growth deficiencies were accompanied by prolonged hypoglycemia (Fig. 2C) and postnatal lethality (Fig. 2D), the latter occurring in two waves, the first around 3 weeks of age and a second between 8 and 12 weeks of age. Comparable amounts of milk were consistently present in the stomachs of dead knockout pups and their normal littermate controls, indicating that the lethality was not the result of a feeding defect. No animals from the experimental cohort survived past 12 weeks of age, and no pups were reared from homozygous parents.
Zbtb20-deficient mice display growth retardation, hypoglycemia, and lethality. (A) Zbtb20 null mice and littermate controls were weighed weekly from 2 to 12 weeks of age. (B) Visual inspection reveals reduced axial growth at 2.5 weeks of age. (C) Blood glucose levels were measured every ∼2 weeks from 2.5 to 12 weeks of age. (D) Survival curves for Zbtb20 null mice and littermate controls (+/+, n = 16; +/−, n = 30; −/−, n = 11).
We next sought to identify potential causes for the observed phenotype by conducting histological and serological analyses on Zbtb20 null and control mice at ∼17 days. This time point was used as it was directly prior to the first wave of mortality, suggesting that the underlying molecular defects should be observable while still yielding sufficient animals for study. Examination of tissue from a variety of organs failed to identify any obvious defects at the level of tissue architecture and integrity, except disrupted hippocampal development (data not shown). A comprehensive serum chemistry analysis, which identified a number of analytes with altered levels between Zbtb20 null mice and controls, was performed (Table 1). Most notably, we observed significant increases in blood urea nitrogen, aspartate aminotransferase, alkaline phosphatase, total bilirubin, and potassium and a significant decrease in glucose. We further defined the levels of additional metabolites and hormones in Tables 2 and 3, respectively. These data revealed significantly increased levels of BHB, glucagon, and resistin, while insulin and leptin levels were significantly reduced.
Quantitation of serum metabolites in Zbtb20 knockout mice and littermate controls
Quantitation of serum hormones and proteins in Zbtb20 knockout mice and littermate controls
Glucose is a fundamental fuel source for mammals, and its levels in the serum are subject to stringent regulatory mechanisms. The temporal correlation between growth retardation, mortality, and hypoglycemia in Zbtb20 mice indicated that the observed phenotype may partially result from disruptions in glucose metabolism. To further define this element of the phenotype, we first monitored glucose levels in response to overnight starvation (Fig. 3A). This analysis revealed similar reductions in serum glucose levels in both Zbtb20 null and control animals in response to starvation. We next performed intraperitoneal glucose tolerance tests, which revealed a striking increase in glucose tolerance in Zbtb20 null mice (Fig. 3B). This was accompanied by normal induction of insulin in response to glucose injections (data not shown), suggesting that the increase in glucose tolerance may lie at the level of increased insulin sensitivity in peripheral tissue rather than increased insulin production in response to the glucose bolus. Indeed intraperitoneal insulin tolerance tests revealed an increase in insulin sensitivity in Zbtb20 knockout mice (Fig. 3C). As hypoglycemia was observed in both the fed and fasted states, we hypothesized that there may be impairments in the capacity of Zbtb20 null mice to buffer serum glucose levels via the actions of counterregulatory hormones such as glucagon. We next performed intraperitoneal glucagon tolerance tests after a 4-h fast, which demonstrated a significantly reduced release of glucose into the circulation in Zbtb20 null mice (Fig. 3D). Quantitation of liver glycogen in animals fed ad libitum revealed an almost complete absence of glycogen stores in Zbtb20 null mice compared to wild-type controls, as measured by both a glucose oxidase activity assay and periodic acid-Schiff staining of fixed liver tissue (Fig. 3E).
Zbtb20 has altered glucose metabolism consistent with nutrient deficiency. (A) Blood glucose levels of Zbtb20 null and littermate controls were measured prior to (fed; P < 0.01) and after (fast; P < 0.005) an overnight fast. (B to D) Glucose tolerance tests (2 mg/kg; +/+, n = 5; +/−, n = 15; −/−, n = 10; P < 0.005) (B), insulin tolerance tests (0.75 U/kg; +/+, n = 5; +/−, n = 5; −/−, n = 5; P < 0.005) (C), and glucagon tolerance tests (30 μg/kg; +/+ n = 4, +/−, n = 24;−/−, n = 5; P < 0.005) (D) were performed on Zbtb20 null and littermate controls. (E) Total glycogen was quantitated in liver from Zbtb20 null mice and littermate controls by a glucose oxidase activity assay (top) (+/+, n = 5; +/−, n = 3; −/−, n = 4, P < 0.05) and periodic acid-Schiff staining of fixed tissue (bottom).
A number of pieces of evidence suggested that the liver dysfunction may have been partially responsible for the observed phenotype in Zbtb20 null mice. While we see no obvious differences in wild-type and Zbtb20 livers upon histological examination (data not shown), we observe increased levels of aspartate aminotransferase and total bilirubin, which suggested that there was underlying damage to the liver. The reduced levels of glycogen in the liver may have been suggestive of a glycogen storage defect which may be intrinsic to the liver. To more fully define the transcriptional signature of liver from Zbtb20 null mice, we performed transcript profiling experiments using Affymetrix GeneChip microarrays. Livers were harvested from four Zbtb20 null and four wild-type control males on the BALB/c background at 17 days of age, and cRNA generated from these tissues was then hybridized to mouse 430 arrays. Analysis of these arrays revealed that there were a total 583 upregulated and 287 downregulated unique probe sets that exhibited a minimum twofold change and reached statistical significance (P < 0.05 by Student's t test). The most highly upregulated and downregulated groups of transcripts are shown in Fig. 4A and B, respectively, in order of decreasing magnitude of change. Further bioinformatic analyses using gene ontology databases to identify regulation of molecular pathways revealed significant changes in the expression of transcripts involved in a number of relevant biological processes such as growth, glucose metabolism, and detoxification, which are shown in Fig. 4C, D, and E, respectively.
Transcript profiling of liver tissue from Zbtb20 null mice reveals putative target genes and altered regulation of pathways involved in growth, glucose metabolism, and detoxification. HeatMaps were generated from the transcript profiling data comparing Zbtb20 null and wild-type control liver tissue (+/+, n = 4; −/−, n = 4). The 40 transcripts with the highest upregulation (A) and downregulation (B) in Zbtb20 null liver are shown. Gene ontology analysis revealed transcripts involved in pathways controlling growth (C), glucose metabolism (D), and detoxification (E). WT, wild type; KO, knockout.
As the transcript profiling studies revealed a transcriptional signature consistent with the disruption of multiple relevant biological processes, we tested the functional contribution of Zbtb20 deficiency in the liver to the overall phenotype of Zbtb20 null mice. To achieve this, we constructed a liver-specific Zbtb20 transgene. Murine Zbtb20 gene cDNA was cloned into the pLIV.7 expression construct, which contained 5′ and 3′ control elements from the human ApoE gene (Fig. 5A) and was expected to direct expression of the transgene specifically to the liver. PCR of tail DNA biopsy samples reveal a band of ∼250 bp in Zbtb20 transgenic mice corresponding to the Zbtb20 gene cDNA sequence (Fig. 5B). Analysis of liver tissue from Zbtb20 transgenic mice and wild-type littermate controls indicated that there was a significant overexpression of Zbtb20 in transgenic mice at both the mRNA and protein levels (Fig. 5C and D, respectively). Analysis of a panel of tissues from transgenic mice indicated that expression of the transgene was highest in the liver (Fig. 5E).
Zbtb20 transgenic mouse generation and validation. (A) Murine Zbtb20 gene cDNA was cloned under the control of regulatory elements from the human ApoE gene. (B) PCR of tail DNA of Zbtb20 transgenic and littermate control animals yields an ∼250-bp band in transgenic animals. (C) Zbtb20 mRNA levels in liver tissue isolated from Zbtb20 transgenic (Tg) and littermate control (WT) mice were measured by quantitative PCR (WT, n = 4; Tg, n = 9). (D) Western blots of total liver lysates from Zbtb20 transgenic and littermate controls. Two distinct bands corresponding to Zbtb20 are detected at ∼90 and ∼100 kDa, respectively, as well as a faster-migrating nonspecific band (NS). (E) Zbtb20 mRNA levels were measured in a panel of tissues isolated from Zbtb20 transgenic and littermate control mice by quantitative PCR (WT, n = 2; Tg, n = 2).
We then tested the ability of the Zbtb20 transgene to complement elements of the phenotype of Zbtb20 null mice by crossing onto the Zbtb20 null background. Analysis of these crosses revealed that expression of the Zbtb20 transgene on a Zbtb20 wild-type or heterozygous background had no significant effect on growth rate (Fig. 6A), blood glucose levels (Fig. 6B), or survival (Fig. 6C). In contrast, the Zbtb20 transgene exhibited effects in the Zbtb20 null background. Zbtb20−/−,+ mice demonstrated a small but nonsignificant increase in body weight compared to Zbtb20−/−,− mice between 6 and 10 weeks of age (Fig. 6A) and a transient normalization of blood glucose at 4 weeks of age (Fig. 6B). Most strikingly, Zbtb20−/−,+ mice demonstrated significantly increased survival compared to Zbtb20−/−,− controls, with Zbtb20−/−,+ mice displaying a median survival of 80 days compared to 45 days for Zbtb20−/−,− controls (Fig. 6C) (P = 0.017 in a log rank [Mantel-Cox] test). We next tested whether the Zbtb20 transgene was able to restore the expression of genes in the liver that may be responsible for the reduction in life span in Zbtb20 knockout mice. Quantitative RT-PCR revealed that Zbtb20−/−,+ mice displayed reduced expression of the alpha fetoprotein (AFP) gene, a previously identified Zbtb20 target gene in the liver (32), in comparison to Zbtb20−/−,− controls, although this did not reach statistical significance (Fig. 6D) (P = 0.067). In addition Zbtb20−/−,+ mice displayed increased expression of several cytochrome P450 family members, shown to be reduced in liver of Zbtb20−/−,− controls via transcript profiling (Fig. 5E). These included Cyp450 2f2, Cyp450 1a12, and Cyp450 2c54 (Fig. 6E) (P = 0.055, 0.065, and 0.052, respectively). These data indicate that some important liver intrinsic functions are impaired in the Zbtb20 null mice and that complementation via transgenesis significantly increased survival, potentially by restoring the expression of cytochrome P450 family members.
The Zbtb20 liver transgene partially complements elements of the Zbtb20 null phenotype. (A) Zbtb20 transgenic mice, Zbtb20 null mice, and littermate controls were weighed weekly from 2 to 12 weeks of age. (B) Blood glucose levels were measured every ∼2 weeks from 2.5 to 12 weeks of age for mice fed ad libitum. (C) Survival curves for Zbtb20 null mice and littermate controls (+/+,−, n = 19; +/+,+, n = 12; +/−,−, n = 50; +/−,+, n = 27; −/−,−, n = 16; −/−,+, n = 14; P = 0.017). Total RNA was extracted from liver tissue harvested from a cohort of mice at 4 to 6 weeks of age from the Zbtb20 knockout to Zbtb20 transgenic cross. (D and E) Levels of transcripts for AFP (D) and a number of cytochrome P450 family members (E) were measured by quantitative RT-PCR (+/+,− and +/−,−, n = 10; −/−,−, n = 5; +/+,+ and +/−,+, n = 5; −/−,+, n = 3).
DISCUSSION
In this study we have identified nonredundant in vivo functions for the POK transcription factor Zbtb20 in higher vertebrates. Zbtb20 null mice exhibit a severe postnatal phenotype that involves concordant hypoglycemia, growth retardation, and lethality. We hypothesized that these three distinguishable elements of the phenotype were likely to be functionally related, with impaired metabolic processes central to the pathology. The phenotype is not a result of a feeding defect because comparable amounts milk are consistently present in their stomachs and those of their normal littermate control. The nature of the phenotype suggested a number of possible interpretations, including but not restricted to (i) altered distribution and usage of glucose among sites of glucose disposal due to defects in organs such as the endocrine pancreas, liver, muscle, and adipose tissue and (ii) impaired digestion/uptake of nutrients or altered feeding behavior, leading to malnutrition, due to defects in the exocrine pancreas, small intestine, or neuroendocrine tissue.
Systemic histological examination failed to identify any obvious defects of tissue architecture and integrity, except the impaired hippocampal development (data not shown). The hippocampus is essential for cognitive functions, such as learning and memory. The disruption of the hippocampus may alter feeding behavior but putatively would not lead to a feeding defect. Conditional ablation of Zbtb20 in neurons could recapitulate the impaired hippocampal development, with apparently normal feeding behavior and glucose homeostasis (W. J. Zhang et al., unpublished data). In addition to trying to define the nature of the metabolic defects observed in Zbtb20 null mice, we chose the liver as a target organ for further study primarily because (i) serum chemistry revealed the upregulation of molecular markers associated with liver dysfunction and damage (e.g., aspartate aminotransferase, bilirubin), suggesting impaired liver function in Zbtb20 null mice; (ii) the liver plays a crucial role in maintaining euglycemia in the context of both fasting and feeding, and liver-specific disruptions to glucose metabolism could result in whole-body defects in energy metabolism (6, 17, 23); (iii) the liver is the major site of insulin-like growth factor 1 (IGF-1) production in response to growth hormone, and thus changes in the function of the somatotropic axis in the liver may have direct bearing on both glucose metabolism and growth (6, 21); and (iv) the liver plays an important role in the detoxification of a variety of chemical species, which may result in toxicity in the context of liver dysfunction.
Closer examination of the hypoglycemic phenotype revealed that glucose levels in Zbtb20 null mice were indistinguishable from those in wild types before the age of 10 days (data not shown), indicating that the reduced serum glucose levels developed postnatally. The correlation with significantly decreased insulin levels in the fed state (∼5% of control mice) (Table 3) and lack of normalization after overnight fasting indicated that the reduction in serum glucose levels was not at the level of overproduction of insulin. Glucose tolerance tests indicated that Zbtb20 null mice were extremely tolerant of glucose, which, accompanied by normal insulin production, suggested an increase in insulin sensitivity in peripheral tissues. In contrast, blood glucose levels in Zbtb20 null mice were almost impervious to a bolus of glucagon. Quantitation of requisite glycogen stores in liver and muscle (data not shown) indicated an almost complete absence of this energy store. Decreased levels of circulating leptin and qualitative decreases in adipose tissue deposits in Zbtb20 null mice (data not shown) indicated that increased adipose tissue was not a compensatory site of energy storage. Together with a threefold increase in the level of circulating ketone bodies, these data suggest that Zbtb20 null mice at ∼17 days of age are in a state of malnutrition (18). The observed data could be rationalized by the following schema: reduced glucose leads to reduced circulating insulin levels, which in turn lead to an upregulation of the insulin signaling machinery in peripheral tissues. This is supported by the transcriptional profile of Zbtb20-deficient liver tissue, which shows increased expression of positive signaling molecules downstream of insulin receptor signaling such as Akt1 and Irs-1 and downregulation of inhibitory components such as phosphatidylinositol 3-kinase p85 (7, 22, 28, 30, 31). Thus, in the context of a glucose bolus, normal levels of insulin production result in increased glucose tolerance as peripheral tissues are primed for glucose uptake. The lack of response to glucagon is the result of depleted glycogen stores and, together with depletion of adipose tissue, is a manifestation of an overall energetic deficit. This model suggests that the defects in glucose metabolism are unlikely to be the result of cell-autonomous defects in the liver, skeletal muscle, or adipose tissue, and indeed the liver-specific Zbtb20 transgene had little effect on serum glucose levels. Our preliminary studies tentatively confirm this hypothesis, as tissue-specific conditional deletion in liver, skeletal muscle, and adipose tissue failed to recapitulate the hypoglycemic phenotype (32; A. P. R. Sutherland and W. J. Zhang, unpublished data). Instead, this model may point to potential defects in tissues that regulate nutrient digestion and uptake. We are currently pursuing this hypothesis by conditional deletion of Zbtb20 in exocrine pancreas and enterocytes of the small intestine.
Transcript profiling of liver tissue revealed significant alterations in the expression of genes involved in both glucose metabolism and the regulation of the somatotropic hormonal axis. A number of genes involved in the counterregulatory response to reduced glucose also showed altered expression, consistent with a shift from glycolysis to glycogenolysis/gluconeogenesis in the context of low glucose and insulin levels (2, 5, 8-10, 16, 33). While we do not see an upregulation of glucose-6-phosphatase and phosphoenolpyruvate carboxykinase, the inducible rate-limiting enzyme required for gluconeogenesis (12-14), their genes are upregulated at the transcriptional level in response to overnight starvation in Zbtb20 null mice (data not shown), indicating that pathways required for their induction are functional in the absence of Zbtb20. Liver expression of the growth hormone receptor is also significantly reduced, as are levels of IGF-1 and IGF binding protein (IGFBP) acid-labile subunit, whose expression is positively correlated with growth (20, 26). In contrast the expression of molecules that antagonize growth hormone signaling (e.g., SOCS2, SOCS3) or IGF-1 activity (e.g., IGFBP-1, IGFBP-2) is increased (1, 11, 27). As growth hormone receptor and the various antagonists of IGF-1 function are regulated in response to nutritional status (29), it seems likely that these effects can be accounted for by the metabolic and hormonal environment that we have already described. This correlates with the lack of significant restoration of growth in Zbtb20 null mice in the context of a liver-specific transgene.
Notable among the pathways identified by gene ontology analysis was the downregulation of a group of genes involved in detoxification, primarily comprising members of the cytochrome P450 family. We show that the expression of a number of these genes was partially normalized in Zbtb20−/−,+ mice (Fig. 6E), suggesting that reduced expression of these genes may be a factor contributing to the lethality observed in Zbtb20 knockout mice and that restored expression in Zbtb20−/−,+ mice may contribute to the observed extension of life span.
The transcript profiling experiments presented here led to the identification of the AFP transcript as the most highly differentially expressed transcript between Zbtb20 null and control liver tissue. Subsequent studies of liver-specific Zbtb20 knockout mice revealed that the normal postnatal extinguishment of AFP expression in liver is temporally correlated with increased Zbtb20 expression and that AFP was a direct target of Zbtb20 (32). Thus, it would appear that in this instance Zbtb20 functions as a developmentally regulated transcriptional repressor, in keeping with the body of evidence to suggest that POK family proteins mediate repression of particular target genes. As such it is likely that Zbtb20 would coordinately control the expression of a range of postnatally regulated genes in the liver. Thus, a combined approach using transcript profiling of liver-specific conditional knockout mice and genome-wide chromatin immunoprecipation (ChIP on chip) to further define Zbtb20 target genes in liver may illuminate the subset of genes that are regulated by Zbtb20 analogously to the AFP gene.
ACKNOWLEDGMENTS
We thank Kirsten Sigrist, Jennifer Donovan, Diana Pascual, and the staff at the Harvard School of Public Health Animal Testing Facility for assistance with animal care.
This work was supported by gift from the G. Harold and Leila Y. Mathers Charitable Foundation (to M.J.G.) and National 973 Program of China, grant 2006CB503910 (to W.J.Z.). A. P. R. Sutherland is a CJ Martin Fellow of the NH&MRC of Australia.
We have no conflicting financial interests.
FOOTNOTES
- Received 27 October 2008.
- Returned for modification 23 November 2008.
- Accepted 21 February 2009.
- Accepted manuscript posted online 9 March 2009.
- Copyright © 2009 American Society for Microbiology
