Transcription factor pancreatic and duodenal homeobox 1 (Pdx1) plays an essential role in the pancreas to regulate its development and maintain proper islet function. However, the functions of Pdx1 in mature small intestine are less known. We aimed to investigate the intestinal role of Pdx1 by profiling the expression of genes differentially regulated in response to inactivation of Pdx1 specifically in the intestinal epithelium. Pdx1 was conditionally inactivated in the intestinal epithelium of Pdx1flox/flox;VilCre mice. Total RNA was isolated from the first 5 cm of the small intestine from mature Pdx1flox/flox;VilCre and littermate control mice. Microarray analysis identified 86 probe sets representing 68 genes significantly upregulated or downregulated 1.5-fold or greater in Pdxflox/flox;VilCre mice maintained under standard conditions. Ingenuity Pathway Analysis revealed that functions of the differentially expressed genes are significantly associated with metabolism of nutrients including lipids and iron. Network analysis examining the interactions among the differentially expressed genes further supports the notion that Pdx1 may modulate metabolism of lipids and iron from mature intestinal epithelium. Following forced oil feeding, Pdx1flox/flox;VilCre mice showed diminished lipid staining in the duodenal epithelium and decreased serum triglyceride levels, indicating reduced lipid absorption compared with control duodenal epithelium. Blood samples from Pdx1flox/flox;VilCre mice have significantly lower mean values for mean corpuscular volume and mean corpuscular hemoglobin, consistent with iron deficiency. The absence of nonheme iron in the villous epithelium and lamina propria of Pdx1flox/flox;VilCre duodenum indicates that the duodenal epithelium lacking Pdx1 may have defects in importing iron through enterocytes, resulting in iron deficiency in Pdx1flox/flox;VilCre mice.
- intestinal epithelium
- lipid metabolism
- iron metabolism
homeodomain-containing transcription factor pancreatic and duodenal homeobox 1 (Pdx1) is best studied for its essential role in embryonic development of pancreas and subsequent maintenance of islet function for normal glucose homeostasis. Mice homozygous for Pdx1-null mutation (Pdx1−/−) fail to form a pancreas and die in the neonatal period within a week of birth (36, 54). Pdx1 regulates expression of genes in the pancreas necessary for maintaining pancreatic identity and function including insulin (55), glucose transporter 2 (68), glucokinase (70), islet amyloid polypeptide (11, 16, 62), and somatostatin (38, 49). Mutations in the human PDX1 gene are linked to maturity-onset diabetes of the young, type 4 (19, 65), and type 2 diabetes mellitus (32, 40).
In the intestine, Pdx1 is expressed most abundantly in the anterior duodenal region and decreases in expression distally (31). A role for Pdx1 in intestinal development is highlighted by the findings that, at the stomach/duodenum junction and in the rostral duodenum, Pdx1−/− null mutant embryos and neonates exhibit dilated cystic malformations, absence of Brunner's glands, areas of glucose transporter 2-positive cuboidal epithelium resembling bile duct epithelium, and reduced numbers of enteroendocrine cells (54). Because of the lethal pancreatic agenesis phenotype in Pdx1−/− null mutant neonates, mice with Pdx1 inactivation restricted to the intestinal epithelium (Pdx1flox/flox;VilCre) were generated to investigate the role of Pdx1 in mature proximal small intestine (17). Pdx1flox/flox;VilCre mice survive through adulthood with morphologically indistinguishable pancreas and proximal small intestine from the littermate control (17). However, intestinal epithelium-specific Pdx1 inactivation significantly decreases the expression of gastric inhibitory polypeptide (Gip), somatostatin (Sst) and alkaline phosphatase 3 (Akp3), intestine, not Mn requiring Akp3 in mature Pdx1flox/flox;VilCre duodenum (17).
Incretin hormone Gip is released primarily from enteroendocrine K cells dispersed throughout the mucosa of the duodenum and proximal jejunum (15). Jepeal et al. (35) demonstrated that Pdx1 is involved in regulating Gip expression in K cells and interacts with the Gip promoter fragment nt −156 to −151. In response to nutrient ingestion, Gip is released from the small intestine and stimulates insulin secretion from the pancreas in a glucose-dependent manner (45). Intestinal enteroendocrine D cells containing Sst are scattered throughout the small and large intestine (22, 61). Approximately 90% of intestinal Sst is present in the mucosa, whereas 10% is present in the muscular layer (59). Intestinal Sst inhibits basal and stimulated bicarbonate secretion from rat duodenal Brunner's glands (37), intestinal transport of ions (23), and secretion of almost every gut hormone tested including Gip, ghrelin, secretin, cholecystokinin, and motilin (4, 12, 57).
Intestinal alkaline phosphatase (IAP) encoded by Akp3 is associated with the brush border membrane of intestinal epithelial cells (33, 71); it is most highly expressed in the duodenal villi and crypts (33, 71). IAP is thought to transport dietary lipids as a component of the surfactant-like particles (SLPs), which surround neutral fat droplets in villous enterocytes during fat absorption (42, 43, 75). Fatty meals increase the levels of IAP in serum and lymph (28, 48). Mice homozygous for an Akp3-null mutation demonstrate faster body weight gain in response to a high-fat diet and increased lipid clearance after a single oral administration of corn oil (50, 52).
The present study represents the first to identify genome-wide changes in mature duodenum in response to the lack of Pdx1 in the intestine. The genes exhibiting significant differential expression were further analyzed with Ingenuity Pathways Analysis software. The data indicate that Pdx1 may regulate metabolism of nutrients including lipids and iron from the mature duodenal epithelium. Mature Pdx1flox/flox;VilCre mice exhibit diminished lipid staining in the duodenal epithelium and a decreased average serum triglyceride level, indicating reduced lipid absorption, after a single dose of forced oil feeding. The lack of nonheme iron staining in the duodenal epithelium of mature Pdx1flox/flox;VilCre mice, in combination with significantly lower mean values for mean corpuscular volume (MCV) and mean corpuscular hemoglobin (MCH) in the blood samples, indicates iron deficiency in response to intestine-specific Pdx1 inactivation.
MATERIALS AND METHODS
Mice with intestinal epithelium-specific Pdx1 inactivation (Pdx1flox/flox;VilCre) were generated by intercross mating between VilCre and Pdx1flox/flox mouse strains as previously described (17). The VilCre mouse strain carries a Villin-Cre transgene and expresses Cre recombinase under the control of a 12.4 kb mouse villin 1 gene promoter fragment (41). Pdx1 exon 2 encoding the DNA-binding homeodomain is flanked by loxP target sites in Pdx1flox/flox mice (27). The exon 2 of Pdx1 flanked by loxP sites is thus excised by Cre recombinase driven by the 12.4-kb mouse villin 1 gene promoter in the intestinal epithelium of Pdx1flox/flox;VilCre mice, resulting in Pdx1 inactivation. The absence of Pdx1 protein expression in the epithelium of mature Pdx1flox/flox;VilCre proximal small intestine was confirmed by immunohistochemical staining (17). Pdx1flox/flox;VilCre and littermate control mice were fed a standard diet (4.5% fat and 380 ppm iron, Prolab RMH 3000; LabDiet, Brentwood, MO) and maintained under normal, nonstressed condition. The protocol for animal use was reviewed and approved by the Stanford University Institutional Animal Care and Use Committee (IACUC).
Intestinal RNA isolation.
The first 5 cm of small intestine immediately adjacent to the pylorus was harvested from a litter of 6 mo-old, male Pdx1flox/flox;VilCre (n = 4) and control Pdx1flox/flox mice (n = 4). Genotype of each mouse was confirmed by PCR using genomic DNA isolated from tail biopsies as previously described (17). The small intestine tissue was preserved in RNAlater (Qiagen, Valencia, CA) during harvest and before processing for RNA isolation. Total RNA was then extracted and treated with DNase I using the RNeasy Mini kit (Qiagen) according to manufacturer's protocol. Total RNA samples were further purified using the RNeasy MinElute Cleanup Kit (Qiagen). RNA concentrations were determined by optical densitometry at 260 nm, and absence of RNA degradation was confirmed by 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Real-time quantitative PCR (RT-qPCR) was performed with the cDNA reverse transcribed from the intestinal RNA samples to validate that Pdx1 mRNA abundance was diminished in mature Pdx1flox/flox;VilCre duodenum (Fig. 2B of Ref. 17).
The intestinal RNA samples were submitted to the Stanford School of Medicine Protein and Nucleic Acid (PAN) Facility for 2100 Bioanalyzer analysis, biotin labeling, microarray hybridization, and scanning. The total RNA samples were prepared for microarray hybridization using GeneChip 3′ IVT Express Kit (Affymetrix, Santa Clara, CA), according to manufacturer's recommendation. Briefly, each total RNA sample isolated from mature duodenum of the Pdx1flox/flox;VilCre (n =4) and littermate control Pdx1flox/flox mice (n = 4) was reverse transcribed into double-stranded cDNA, followed by an in vitro transcription reaction that produced amplified amounts of biotin-labeled antisense mRNA (cRNA).
The labeled cRNA prepared from each mouse was fragmented and hybridized to a single GeneChip Mouse Genome 430 2.0 Array (Affymetrix, Santa Clara, CA). The Mouse Genome 430 2.0 Array detects more than 39,000 transcripts from known mouse genes and potentially expressed sequences, measuring the abundance of each transcript with multiple independent oligonucleotide probes, or probe set. The microarrays were washed and stained with fluorescent molecule SAPE that binds to biotin in combination of biotinylated antistreptavidin antibody on GeneChip Fluidics Station 450 (Affymetrix). Subsequently the arrays were scanned with GeneChip Scanner 3000 7G (Affymetrix) to generate fluorescence intensity image for data analysis. The microarray data in the present study have been deposited in the NCBI Gene Expression Omnibus (accession no. GSE29048).
The fluorescence intensity image files obtained from scanning each of the arrays were reviewed and processed initially by the PAN Facility with GeneChip Operating Software (Affymetrix). The software grids the array image to specify the region and location of each probe and calculates a single intensity for each probe based on the 75th percentile of remaining pixel intensities after excluding the border pixels. The resulting files, containing a single intensity value for each probe region delineated by a grid on each array image, were imported into Partek Genomics Suite (version 6.4, Partek, St. Louis, MO) for probe set summarization and statistical analysis. Model based Robust Multichip Analysis was performed as the method for probe set summarization to obtain a single intensity value representing transcript abundance for each probe set and thus enable comparisons between arrays, by normalizing and logarithmically transforming array data and stabilizing variance across the arrays.
ANOVA was used to calculate a P value from evaluating statistical significance of the difference observed in mean transcript abundance for each probe set between Pdx1flox/flox;VilCre (n = 4) and littermate control mice (n = 4). p < 0.05 indicates statistically significant differential expression. The probe sets exhibited significant, greater than 1.5-fold difference in mean transcript abundance between the Pdx1flox/flox;VilCre, and littermate control mice were visualized by average linkage hierarchical clustering using agglomerative method, based on dissimilarity measured by Euclidean distance. The clustering was plotted with standardized intensity determined by the algorithm in Partek software, setting the mean value to 0 and standard deviation to 1. The height of a cluster indicates the distance between the two members of the cluster. The similar members are combined with short clusters, whereas tall clusters separate dissimilar groups.
Functional and network analysis of the differentially expressed genes.
To interpret the possible roles of Pdx1 in mature duodenum functions and interactions of genes exhibiting significant differential expression 1.5-fold or greater were analyzed with a web-based software and database, Ingenuity Pathways Analysis (IPA version 8.8; Ingenuity Systems, Redwood City, CA). IPA database, the Ingenuity Knowledge Base, stores millions of manually reviewed biological interactions and functional annotations from the literature and other databases such as Entrez Gene, RefSeq, OMIM, Gene Ontology, KEGG Pathway, and BIND.
To investigate the possible changes in physiology, metabolism and intestinal functions predisposed by Pdx1 inactivation restricted to the intestinal epithelium, IPA functional analysis was performed to find significant associations of the differentially expressed genes to three primary categories of functions: molecular and cellular functions, physiological system development and function, and diseases and disorders. Under the primary categories, 85 subcategories were classified, consisting of specific, basic-level functions populated with a group of genes or chemicals, based on the findings stored in the Ingenuity Knowledge Base. Statistically significant, nonrandom associations of the differentially expressed genes with the specific functions and subcategories were indicated by a P value >0.05 following right-tailed Fisher's exact test.
Using IPA Network analysis, interactions among genes exhibiting significant changes in expression to Pdx1 inactivation in mature duodenum were examined and visualized by generating statistically significant, nonrandom networks. The differentially expressed genes served as “seeds” and connected to other genes or chemicals in the Ingenuity Knowledge Base via direct or indirect interactions. Direct interactions refer to direct physical contact such as protein-DNA binding. In contrast, indirect interactions do not require direct evidence of physical binding. Examples include activation, transcription, phosphorylation, and localization. A network is limited to 35 genes or chemicals to maximize specificity of the connections.
Additional genes or chemicals that were not included on the Affymetrix Mouse Genome 430 2.0 Array chip or did not show significant differential expression were considered to specifically merge two or more smaller networks. Inclusion of additional genes or chemicals through specific interactions with the differentially expressed genes aids in identifying potential functional roles of Pdx1 in mature duodenum. Network analysis complements functional analysis because functional analysis considers the differentially expressed genes alone. The statistical significance, or scores, of generated networks were calculated with right-tailed Fisher's Exact Test. The higher the score, the lower the probability of finding the observed number of differentially expressed genes in a given network by random chance.
Oil gavage, oil red O staining, and serum triglycerides levels.
Virgin female, 2-mo-old Pdx1flox/flox;VilCre and littermate control Pdx1flox/flox mice were fasted overnight and force fed 10 ml/kg vegetable oil by oral gavage. The mice were killed 7 h after the oil gavage, and the proximal small intestine immediately below the pylorus was collected, followed by fixation overnight in 10% buffered formalin. The intestinal tissue samples were then immersed in PBS solution containing 10, 15, and 20% sucrose, followed by embedding in optimal cutting temperature compound on dry ice and maintained at −80°C. To visualize fat vesicles, frozen intestinal sections (8 μm) were stained with Oil Red O and counterstained with hematoxylin. For determination of serum triglyceride levels, virgin female, 3-mo-old Pdx1flox/flox;VilCre and littermate control Pdx1flox/flox mice were fasted overnight and force fed 10 ml/kg vegetable oil by gavage. Blood samples were collected retroorbitally under Isoflurane anesthesia immediately and at 7 h after forced oil feeding. Serum samples were then analyzed by the Stanford University Department of Comparative Medicine Diagnostic Laboratory for triglycerides levels.
Blood cell analysis and tissue iron staining.
Blood samples were collected intracardially from female and male, 3- and 6-mo-old Pdx1flox/flox;VilCre (n = 8) and littermate control Pdx1flox/flox mice (n = 3) and subsequently analyzed by the Stanford University Department of Comparative Medicine Diagnostic Laboratory for hemoglobin, MCV, and MCH. Duodenal tissue samples with attached stomach and pylorus were collected from 6-mo-old female Pdx1flox/flox;VilCre and littermate control Pdx1flox/flox mice, followed by fixation in 10% buffered formalin and embedding in paraffin. To visualize nonheme iron in the duodenal epithelium, deparaffinized sections were stained with the Perls Prussian blue stain.
Genes were differentially expressed in the duodenum of mature mice in response to intestinal-epithelium-specific Pdx1 inactivation.
To identify potential gene targets of Pdx1 in mature intestine, RNA was isolated from the first 5 cm of the small intestine adjacent to the pylorus from male Pdx1flox/flox;VilCre and littermate control mice at 6 mo of age (n = 4 for each genotype). The RNA samples were subsequently analyzed with the Affymetrix Mouse Genome 430 2.0 Array chip. Compared with the control genotype, 86 probe sets representing 68 genes showed significant changes (P < 0.05) in expression by 1.5-fold or greater in the duodenum of mature Pdx1flox/flox;VilCre mice. Among the 68 genes, 44 were downregulated (Table 1), whereas 24 were upregulated (Table 2), by intestinal epithelium-specific Pdx1 inactivation. Hierarchical clustering clearly separated the differentially expressed genes into two clusters, representing upregulated and downregulated genes, respectively (Fig. 1). In addition, the total RNA samples isolated from mature duodenum of Pdx1flox/flox;VilCre and littermate control mice were clustered into two distinct groups corresponding to the respective genotypes (Fig. 1).
As expected, Pdx1 is included among the genes significantly decreased in mature duodenum of Pdx1flox/flox;VilCre mice (Table 1). Genes whose expression is known to be affected by Pdx1 in the intestine, such as Gip (35) and S100 calcium binding protein G (S100g) (5), were also identified as genes significantly downregulated (Table 1). In agreement with the results of the present microarray analysis, independent RT-qPCR performed previously with the same set of RNA samples (17) also showed that mRNA abundance of Akp3, Gip, and Sst was significantly decreased in Pdx1flox/flox;VilCre mature duodenum (Table 1). Conversely, selected genes representing enterocyte, goblet, enteroendocrine, and Paneth cell lineages previously showing no significant changes in expression with Pdx1 inactivation in mature duodenum (Table 1 of Ref. 17) were not identified by the present microarray analysis among the 68 differentially expressed genes (Tables 1 and 2).
Functions of the differentially expressed genes were significantly associated with nutrient metabolism.
The microarray data were analyzed by a computer software, IPA, for significant association (P < 0.05) with biological functions represented by the genes differentially expressed 1.5-fold or greater by intestinal epithelium-specific Pdx1 inactivation. The association was examined with three categories of functional analysis: molecular and cellular functions, physiological system development and function, and diseases and disorders (Table 3). The results were comparable between the analyses including or excluding Pdx1, indicating that the presence of Pdx1 did not skew the results of functional analysis (Table 3). In mature duodenum, cellular movement, cellular growth and proliferation, cell death, gene expression, and lipid metabolism were most likely affected by Pdx1 inactivation. Endocrine system function appeared to be most significantly impacted in response to intestinal epithelium-specific Pdx1 inactivation. The differentially expressed genes were involved in gastrointestinal and hepatic system dysfunction and nutritional disease.
The functional analysis revealed that Pdx1 may be involved in metabolizing nutrients and drugs in mature duodenal epithelium, by affecting the expression of genes with functions involving metabolism of lipids, carbohydrates, nucleic acids, amino acids, vitamins and minerals, and drugs (Table 4). In addition, intestinal epithelium-specific Pdx1 inactivation also significantly altered the expression of genes involving nutritional diseases including adult onset obesity and iron deficiency (Table 4). Products of the differentially expressed genes involved in nutrient metabolism include enzymes, transporters, kinases, phosphatases, hormones, growth factors, cytokines, ion-binding proteins, transmembrane proteins, GTPase activators, and transcription regulators (Table 4).
Network analysis of the differentially expressed genes indicates that Pdx1 inactivation in mature intestinal epithelium may impact iron and lipid metabolism.
The relational interactions among differentially expressed genes were analyzed by generating networks. Gene networks visualize the relationships among genes differentially expressed in response to intestine-specific Pdx1 inactivation and therefore reveal possible roles of Pdx1 in the duodenum. The relationships examined include direct and indirect interactions between the genes of interest. Direct interactions refer to physical binding relationships such as protein-DNA binding, whereas examples for indirect interactions include activation, transcription, phosphorylation, or localization. Networks were generated using the differentially expressed genes as seeds and connecting as many of them into a network. Other molecules (genes or chemicals) in the Ingenuity Knowledge Base are included to connect multiple smaller gene networks into a larger network, hence providing insights into possible mechanisms or signaling pathways in which Pdx1 may be involved in the duodenum.
The representative network (Fig. 2) has a high significance score of 32 and contained a high number (15) of genes differentially expressed 1.5-fold or greater in the duodenum of mature Pdx1flox/flox;VilCre mice. The score of 32 indicates that the chance is 1 in 1032 to form a network of 35 molecules by randomly selecting from the Ingenuity database and including at least 15 differentially expressed genes associated with intestinal epithelium-specific Pdx1 inactivation. This network reveals a possible functional role of Pdx1 in modulating iron and lipid metabolism from mature duodenal epithelium, by connecting iron and fatty acid with the rest of the network. Indeed, Akp3 (42, 43, 50, 52, 75), Gip (26), kallikrein 1 (31a, 39a), metallothionein 1 (Mt1) (7, 39), and UDP glucuronosyltransferase 2 family, polypeptide B38 (Ugt2b38) (8, 65a) have functional roles in lipid metabolism or weight gain/loss. In addition, solute carrier family 40 (iron-regulated transporter), member 1 (Slc40a1) and solute carrier family 46, member 1 (Slc46a1) are duodenal transporters involved in iron metabolism (1, 63).
Molecules located at the center of a network or “hub” molecules highly connected to other molecules within a network are proposed to play a major role in the biological processes represented by the network. The central hub of the representative network (Fig. 2) is ion-binding protein Mt1. Mt1 expression in mature Pdx1flox/flox;VilCre duodenum was significantly increased greater than twofold on average, and this upregulation was detected by two independent probe sets (Table 2). Mt is a family of cysteine-rich, low-molecular-weight proteins with the capacity to bind both physiological (such as zinc, copper, selenium) and xenobiotic (such as cadmium, mercury, silver, arsenic) heavy metals. Mt proteins are localized to the membrane of the Golgi apparatus and mitochondrial intermembrane space. The role of Mt1 in the intestine has not been extensively studied. Interestingly, Pdx1 suppression in rat islets decreases the mRNA abundance of Mt1a (27a).
The duodenal epithelium of mature Pdx1flox/flox;VilCre mice has altered lipid absorption.
Results from functional and network analyses of the microarray data indicate that Pdx1 may modulate lipid metabolism in mature duodenum (Fig. 2, Tables 3 and 4). A possible mechanism would be that Pdx1 inactivation alters the rate of lipid absorption across the mature intestinal epithelium. In support of this potential mechanism, mature Pdx1flox/flox;VilCre mice have reduced brush border Akp3 activity in the duodenal epithelium (17) and decreased Akp3 mRNA abundance in the duodenum (Table 1 and Ref. 17). Akp3 transports dietary lipids as a component of the SLPs during fat absorption (42, 43, 75). Mice homozygous for an Akp3-null mutation have increased intestinal lipid clearance and faster body weight gain following high-fat-diet feeding (50, 52). In contrast, another possible mechanism for altered lipid metabolism may be that Pdx1 inactivation restricted to the intestinal epithelium results in attenuation of the incretin effect mediated by Gip after the ingestion of nutrients (Fig. 2). Using the lymph fistula rat model, Yoder et al. (74) reported that secretion of triglyceride and Gip into lymph increases dose dependently in response to increasing dietary lipid. In Wistar rats fed a high-fat diet, oral glucose tolerance testing induces a progressive, significant increase in plasma Gip levels compared with control diet (29). Most importantly, GIP/DT mice with targeted ablation of Gip-producing cells resist development of high-fat-diet-induced obesity (3). Consistent with the findings that Pdx1 interacts with a regulatory sequence located within the Gip promoter region and regulates Gip expression in the intestine (35), Gip mRNA abundance is significantly decreased six- to eightfold in the first 5 cm of Pdx1flox/flox;VilCre small intestine (Table 1 in the present study and Ref. 17). The first 5 cm of the small intestine encompasses the duodenum, corresponding to the region where Gip-expressing enteroendocrine K cells are dispersed. Incretin Gip is released from K cells in response to nutrient absorption.
To investigate how Pdx1 inactivation restricted to the intestinal epithelium affects lipid absorption, mature Pdx1flox/flox;VilCre and littermate control Pdx1flox/flox mice were gavaged with vegetable oil. At 7 h following forced oil feeding, reduced Oil Red O staining of fat droplets (red) was observed in the epithelium throughout the first 5 cm (Fig. 3B) and the subsequent 5-cm (Fig. 3D) segments of Pdx1flox/flox;VilCre small intestine, compared with the matching locations in control epithelium (Fig. 3, A and C, respectively). Throughout the entire control small intestine, only two segments express Pdx1: the first 5 cm contains the highest abundance of Pdx1 mRNA, and the abundance decreases significantly in the second 5-cm segment (17). Pdx1 expression is absent from the Pdx1flox/flox;VilCre epithelium of both segments. The results indicate that the epithelium lacking Pdx1 has altered lipid absorption within the first 10 cm of small intestine encompassing the duodenum and corresponding to regions where expression of Gip and Akp3 is also significantly reduced.
To investigate whether altered lipid clearance by the absence of Pdx1 in mature duodenal epithelium results in decreased or enhanced lipid absorption, serum triglyceride levels were measured in mature Pdx1flox/flox;VilCre and littermate mice. Seven hours after oral gavage of vegetable oil, higher levels of serum triglycerides were observed in control Pdx1flox/flox mice than Pdx1flox/flox;VilCre mice, indicating decreased lipid absorption in the epithelium lacking Pdx1. The results suggest that, after ingestion of lipids, reduced lipid absorption resulting from intestinal epithelium-specific Pdx1 inactivation may be mediated through an impaired incretin effect by Gip rather than through decreased Akp3 expression and IAP activity.
Intestinal epithelium-specific Pdx1 inactivation alters iron metabolism in mature mice.
Analysis of the genes significantly upregulated or downregulated 1.5-fold or greater in the mature duodenum indicated that Pdx1 inactivation restricted to the intestinal epithelium may affect vitamin and mineral metabolism (Table. 4). Network analysis further showed that metabolism of iron may be altered in response to the lack of Pdx1 in mature intestinal epithelium (Fig. 2). Specifically, expression of iron and heme transporter genes including solute carrier family 11 (proton-coupled divalent metal ion transporters), member 2 (Slc11a2), Slc40a1, and Slc46a1 is significantly reduced by Pdx1 inactivation in the duodenum of mature Pdx1flox/flox;VilCre mice (Table. 1). To investigate whether iron homeostasis in mature Pdx1flox/flox;VilCre mice is affected, hemoglobin, average red blood cell size (MCV), and hemoglobin amount per red blood cell (MCH) were measured in blood samples from mature Pdx1flox/flox;VilCre and littermate control Pdx1flox/flox mice. Despite being fed a standard diet (380 ppm iron), Pdx1flox/flox;VilCre mice have significantly lower mean values for MCV and MCH than control littermates (Fig. 4, A and B). In addition, average hemoglobin was decreased in 6-mo-old Pdx1flox/flox;VilCre mice (13.8 gm/dl) compared with control Pdx1flox/flox littermates (14.4 gm/dl) (P = 0.07). The results indicate development of iron deficiency anemia in Pdx1flox/flox;VilCre mice by 6 mo of age.
To investigate whether impaired intestinal iron absorption is a cause of the iron deficiency that develops in mature Pdx1flox/flox;VilCre mice, nonheme iron uptake by the duodenal epithelium was examined by Perls Prussian blue staining (Fig. 4, C–F). In the villous epithelium (Fig. 4, C and E) and lamina propria (Fig. 4F) of Pdx1flox/flox duodenum, the presence of nonheme iron (visualized as blue staining) indicates iron absorption across the control duodenal epithelium. In contrast, no nonheme iron was visible in the villous epithelium or lamina propria of Pdx1flox/flox;VilCre duodenum (Fig. 4D). The results suggest that duodenal epithelium lacking Pdx1 has impaired iron absorption resulting in iron deficiency in mature Pdx1flox/flox;VilCre mice, in support of the functional association with nutritional disease iron deficiency (Table 4) and the development of iron deficiency anemia (Fig. 4, A and B, and data not shown).
The present study identifies novel target genes in mature duodenum not previously shown to be regulated by Pdx1. The functions of these genes are significantly associated with nutrient metabolism. Analysis of the relationships among the genes also supports a novel role for Pdx1 in modulating metabolism of lipids and iron from the intestinal epithelium of mature mice. Duodenal epithelium lacking Pdx1 in mice exhibits an altered rate of lipid clearance and impaired iron absorption. Blood samples from mature Pdx1flox/flox;VilCre mice have reduced serum triglyceride levels after forced oil feeding and iron deficiency with significantly lower mean values for MCV and MCH than control littermates.
Interestingly, solute carrier (Slc) family members represented the single largest group of genes showing significant differential expression by Pdx1 inactivation in mature duodenum. Slc family members are membrane transport proteins including facilitative transporters and secondary active transporters. Twenty-three percent of the 86 probe sets detecting significant differences (1.5-fold or greater, P < 0.05) in the abundance of gene transcripts comparing Pdx1flox/flox;VilCre to control duodenum represented Slc genes (Tables 1 and 2). Intestine-specific Pdx1 inactivation in the duodenal epithelium appeared to significantly downregulate the expression of Slc family member genes in mature mice, with the exception of solute carrier family 5 (sodium/glucose cotransporter), member 12 (Slc5a12) (Table 2). Expression of Slc genes with functions transporting nutrients iron, copper, zinc, calcium, vitamins, and amino acids was altered in response to the lack of Pdx1 in mature intestinal epithelium.
Iron absorption by the duodenal mucosa is initiated by uptake of ferrous Fe(II) iron across the brush border membrane and results in transfer of the metal across the basolateral membrane to the portal vein circulation. Pdx1 inactivation in mature Pdx1flox/flox;VilCre duodenum significantly reduced mRNA abundance of several Slc family of transporter genes involved in iron metabolism, namely Slc46a1, Slc11a2, and Slc40a1 (Table 1). Slc46a1 is an intestinal heme transporter, and its protein is localized to the brush-border membrane of duodenal enterocytes in iron deficiency (63). Dietary iron deprivation increases the mRNA expression of apical Slc11a2 in mice (18) and in rat duodenum from the suckling period through adulthood (20). A mutation in the SLC11A2 can cause hypochromic microcytic anemia (OMIM ID no. 206100).
Solute carrier family 40 (iron-regulated transporter), member 1 (Slc40a1) is localized to the basolateral membrane of the duodenal epithelial cell. Iron deficiency increases Slc40a1 expression in the duodenum (1). Slc40a1 is upregulated in the iron overload disease, hereditary hemochromatosis (47). Mutations in the SLC40A1 are associated with autosomal dominant hemochromatosis (69). Chen et al. (18) reported that Slc11a2 can be modulated by the enterocyte iron level, whereas Slc40a1 expression responds to systemic rather than local signals of iron status. The basolateral transport step appears to be the primary site at which the small intestine responds to alterations in body iron requirements (18).
Mature Pdx1flox/flox;VilCre mice fed a standard diet containing 380 ppm iron in the present study showed absence of nonheme iron in the duodenal villous epithelium or lamina propria, accompanied by decreased hemoglobin levels and significantly lower mean values for MCV (average red blood cell size) and MCH (hemoglobin amount per red blood cell). The findings suggest a possible hypochromic microcytic anemia in Pdx1flox/flox;VilCre mice attributable to iron deficiency resulting from impaired iron absorption by the intestinal epithelium. Impaired iron absorption by the enterocytes may thus be due to the downregulation of iron and heme Slc transporter genes Slc46a1, Slc11a2, and Slc40a1 in response to the lack of Pdx1 in the mature intestinal epithelium.
Lipid metabolism was the function most highly linked with Pdx1 inactivation restricted to the intestinal epithelium, as revealed by analyzing the functional association and relationships among the differentially expressed genes in the duodenum of mature Pdx1flox/flox;VilCre mice (Fig. 2, Tables 3 and 4). In agreement with the microarray findings, the duodenal epithelium of mature Pdx1flox/flox;VilCre mice have decreased lipid absorption (Fig. 3). Interestingly, mice homozygous for an Akp3-null mutation instead have increased intestinal lipid clearance (52) although in mature Pdx1flox/flox;VilCre duodenum there is reduced brush border Akp3 activity in the epithelium (17) and decreased Akp3 mRNA abundance (Table 1 and Ref. 17). The present microarray analysis detected a 1.3-fold decrease (P < 0.05) in the expression of Akp5, an alkaline phosphatase isoform, in mature duodenum of Pdx1flox/flox;VilCre mice fed a standard diet, whereas the detection of Akp6 was not included on the arrays. In the small intestine, Akp5 expression is not affected by high-fat feeding, whereas jejunal-ileal Akp6 expression is increased particularly in Akp3−/− mice after corn oil administration or long-term high-fat feeding (51). The residual IAP activity observed in mature Pdx1flox/flox;VilCre duodenal epithelium may result from Akp6 (17).
Although IAP does not seem to play a major role in mediating duodenal lipid absorption in mature Pdx1flox/flox;VilCre mice after oil force feeding, analysis of the 68 genes showing significant changes in expression (1.5-fold or greater) revealed a strong association with LPS-mediated inhibition of retinoid X receptor function pathway (data not shown), implicating an altered response of Pdx1flox/flox;VilCre mice to LPS in the intestine likely attributable to diminished IAP (such as Akp3) expression and activity. This finding is consistent with the report that IAP-deficient zebrafish are hypersensitive to LPS toxicity, as vertebrates harbor abundant LPS in their gut microbiota and alkaline phosphatases can dephosphorylate and detoxify the endotoxin component of LPS (6). Similarly, in mammalian systems, brush-border enzyme IAP is a gut mucosal defense factor, having the ability to detoxify LPS and prevent bacterial invasion across the gut mucosal barrier (30). Furthermore, Akp3−/− mice have dramatically fewer and different types of aerobic and anaerobic microbes in their stools compared with wild-type mice, indicating that IAP is involved in the maintenance of normal gut microbial homeostasis (44).
Various genes with functions associated with lipid metabolism have significant changes in expression resulting from Pdx1 inactivation in mature duodenum (Table 4 and Fig. 2). These genes include enzymes, acyl-CoA thioesterase 12 (Acot12), acyl-CoA oxidase 2, branched chain (Acox2), glutathione S-transferase α4 (Gsta4), monoacylglycerol O-acyltransferase 1 (Mogat1), pancreatic lipase-related protein 2 (Pnliprp2), stearoyl-CoA desaturase 1 (Scd1), UDP glucuronosyltransferase 2 family, polypeptide B38 (Ugt2b38), transporter transthyretin (Ttr), kinase phosphoenolpyruvate carboxykinase 1, cytosolic (Pck1), hormone Sst, growth factor and cytokine secreted phosphoprotein 1 (Spp1), ion-binding protein Mt1, and transcription regulator nuclear receptor coactivator 4 (Ncoa4) (Table 4).
The present study identified genes that have not been previously shown to have a role in intestinal lipid metabolism. For example, mice carrying a null mutation for Mt1, the central hub of the representative gene network (Fig. 2), have moderate obesity (7). Ncoa4 is an androgen receptor coactivator, binding to androgen receptor in a ligand-dependent manner and enhancing its transcriptional activity. Androgens such as testosterone and dihydrotestosterone are hubs highly connected to the other molecules in the representative gene network (Fig. 2). Ncoa4 mRNA abundance is significantly decreased by an average of approximately twofold in mature Pdx1flox/flox;VilCre duodenum, and this downregulation was detected by two independent probe sets (Table 1). Ncoa4 mRNA is expressed in rat stomach (2). The intestinal function and expression of Ncoa4, however, is not previously known.
For known factors involved in intestinal lipid metabolism, the present study detected a significant 1.8-fold decrease in the expression of Mogat1 (LOC100047046), also known as MGAT1 (Table 1). Hydrolysis of dietary lipids is completed in the intestinal lumen by mixing with bile salts and pancreatic juice containing pancreatic lipases. Monoacylglycerol (MAG), diacylglycerol , lysophosphatidic acid, and free fatty acids are then released to form mixed micelles for dietary fat absorption at the brush border of the enterocytes. To be absorbed after entering the enterocyte, MAG has to be reacylated and synthesized into triacylglycerol (TAG). Monoacylglycerol acyltransferase MGAT catalyzes the first step in TAG synthesis (64). Interestingly, expression of another MGAT isoform MGAT2 was significantly increased 1.2-fold in mature duodenum lacking Pdx1 (data not shown). Selected factors known in mediating intestinal lipid metabolism such as CD36, diacylglycerol acyltransferases DGAT1 and DGAT2, however, showed no significant changes in mRNA abundance in response to intestinal epithelium-specific Pdx1 inactivation (data not shown).
A possible mechanism underlying altered lipid absorption by mature duodenal epithelium lacking Pdx1 may be impaired enteroinsular axis signaling via Gip (Fig. 2). Intestinal epithelium-specific Pdx1 inactivation may thus exert a broader impact on lipid metabolism even beyond the intestine via enteroinsular axis signaling. Enteroinsular axis is a term used to describe the hormonal link between the intestine and endocrine pancreas, where nutrients, autonomic nerves, and incretin hormones all contribute to the signaling (45). During a meal, incretins such as Gip are released from the gastrointestinal tract and stimulate insulin secretion from the pancreas in a glucose-dependent manner (24, 67). It is estimated that up to 60% of total insulin secreted during a meal results from the incretin response (60). Moreover, there is a reduced incretin response in many patients with type 2 diabetes (53, 67).
The 42-amino-acid hormone Gip is the first identified incretin, originally for its ability to inhibit gastric acid secretion (13, 14). Subsequently Gip is shown to stimulate glucose-dependent insulin secretion from the islet β-cells (25, 58). Gip is predominantly expressed in the stomach and the K cells of the proximal small intestine (24). Fat is a potent Gip secretion stimulus in humans, whereas carbohydrates are more effective in rodents (73). Gip exerts its actions through the Gip receptor (GIPR), and GIPR is widely expressed in the pancreas, stomach, small intestine, adipose tissue, adrenal cortex, lung, pituitary gland, heart, testis, vascular endothelium, bone, and brain (66, 72).
It has been shown that GIP regulates adipocyte lipid metabolism (46, 73). There is also evidence that Gip may regulate bone formation (9). As expected, Gip mRNA abundance is significantly reduced in mature Pdx1flox/flox;VilCre duodenum (Table 1 and Ref. 17). Given the wide tissue expression of GIPR and the biological importance of endogenous Gip for the control of glucose homeostasis, intestinal epithelium-specific Pdx1 inactivation may exert a broad impact beyond the intestine on lipid and carbohydrate metabolism (Table 4) via impaired enteroinsular axis and attenuated Gip signaling, in response to changes in nutritional status such as high-fat-diet feeding. The combination of reduced lipid absorption in mature Pdx1flox/flox;VilCre intestinal epithelium and the evidence that mice with targeted ablation of Gip-producing cells resist development of high-fat-diet-induced obesity (3) suggests that mature Pdx1flox/flox;VilCre mice may also resist development of nutritional diseases including adult-onset obesity (Table 4) following a high-fat diet.
This research was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK72416 and DK60715 (to E. Sibley) and DK56339 (to the Stanford Digestive Diseases Center).
No conflicts of interest are declared by the authors.
The authors thank Dr. Corrine Davis for excellent veterinary assistance, Dr. Michael S. Kilberg for helpful suggestions, and Tripp Leavitt for animal husbandry work.
- Copyright © 2012 the American Physiological Society