Intestinal NCoR1, a regulator of epithelial cell maturation, controls neonatal hyperbilirubinemia

Significance

In many parts of the world, especially in low- and middle-income countries, severe neonatal hyperbilirubinemia (SNH) is associated with substantial mortality and long-term morbidities. Although the immediate and rapid rise in total serum bilirubin (TSB) originating from lysis of red blood cells has been linked to genetic predisposition, preterm births, and blood type incompatibilities, the inability to efficiently metabolize bilirubin results from delayed expression of UDP-glucuronosyltransferase 1A1 (UGT1A1). In this study, the mechanism associated with delayed expression of the human UGT1A1 gene in neonatal mice that are humanized for the UGT1 locus is described. Neonatal humanized UGT1 (hUGT1) mice develop SNH and control TSB levels by nuclear receptor corepressor 1 (NCoR1)-directed repression of intestinal epithelial cell maturation, an event linked to expression of the UGT1A1 gene.

Abstract

Severe neonatal hyperbilirubinemia (SNH) and the onset of bilirubin encephalopathy and kernicterus result in part from delayed expression of UDP-glucuronosyltransferase 1A1 (UGT1A1) and the inability to metabolize bilirubin. Although there is a good understanding of the early events after birth that lead to the rapid increase in serum bilirubin, the events that control delayed expression of UGT1A1 during development remain a mystery. Humanized UGT1 (hUGT1) mice develop SNH spontaneously, which is linked to repression of both liver and intestinal UGT1A1. In this study, we report that deletion of intestinal nuclear receptor corepressor 1 (NCoR1) completely diminishes hyperbilirubinemia in hUGT1 neonates because of intestinal UGT1A1 gene derepression. Transcriptomic studies and immunohistochemistry analysis demonstrate that NCoR1 plays a major role in repressing developmental maturation of the intestines. Derepression is marked by accelerated metabolic and oxidative phosphorylation, drug metabolism, fatty acid metabolism, and intestinal maturation, events that are controlled predominantly by H3K27 acetylation. The control of NCoR1 function and derepression is linked to IKKβ function, as validated in hUGT1 mice with targeted deletion of intestinal IKKβ. Physiological events during neonatal development that target activation of an IKKβ/NCoR1 loop in intestinal epithelial cells lead to derepression of genes involved in intestinal maturation and bilirubin detoxification. These findings provide a mechanism of NCoR1 in intestinal homeostasis during development and provide a key link to those events that control developmental repression of UGT1A1 and hyperbilirubinemia.

UDP-glucuronosyltransferase 1A1 (UGT1A1), the only transferase capable of conjugating bilirubin (1), is developmentally delayed in newborn children (2). Thus, ∼80% of newborns have some form of hyperbilirubinemia (3, 4). Most cases have a benign outcome except in situations when the rapid onset of severe neonatal hyperbilirubinemia (SNH) is not monitored nor prevented. Shortly after birth, an increase in red blood cell turnover occurs where heme is released from hemoglobin and further degraded by heme oxygenase to carbon monoxide and biliverdin, which is further reduced by biliverdin reductase to bilirubin (5). Once in the circulation, bilirubin is absorbed into tissues such as the liver. Hyperbilirubinemia develops when a rise in total serum bilirubin (TSB) levels exceeds the capacity of hepatic UGT1A1 to drive bilirubin conjugation, an event that leads to the excretion of the glucuronide by MRP2 (6) into the biliary channels for deposit in the gastrointestinal tract. Thus, bilirubin glucuronidation is the rate-limiting step in bilirubin excretion. Major risk factors that lead to the onset of SNH include accelerated hemolysis (7) brought on by Rhesus disease and ABO incompatibility, glucose-6-phoshpate dehydrogenase (G6PD) deficiency, infections, in addition to breast feeding and premature birth (7–11). Extreme levels of TSB can lead to early or acute bilirubin encephalopathy (ABE) (12), presented early as lethargy and poor feeding, but can progress to hypo- and hypertonia, high-pitched crying, muscle spasms, opisthotonus, seizures, and even death (13). The more chronic form, which proceeds ABE, is termed kernicterus (14). Affected individuals are characterized by choreoathetoid cerebral palsy; display dystonic/athetoid movement disorders, hearing loss, ocular motor defects, hypotonia, and ataxia; and have been linked to cerebellar involvement (5, 13–15). Severe hyperbilirubinemia, which leads to ABE and kernicterus, has been observed world wide (14), with the highest incidences being recorded in sub-Saharan Africa and South Asia (10, 16). A report from the Nigerian Society of Neonatal Medicine suggests that extreme hyperbilirubinemia accounts for at least 5% of all neonatal mortalities in Nigeria (16). Recent evidence shows that SNH, estimated to impact over 1 million children every year, is associated with substantial mortality and permanent morbidities (10). Although conventional therapies entail aggressive light therapy or blood transfusions once SNH has been diagnosed (5), the inability to effectively conjugate bilirubin resulting from development delay in expression of UGT1A1 ultimately leads to bilirubin-induced neurological toxicity. Although genetic predisposition and environmental influences play key roles in driving TSB levels to potentially toxic levels, an understanding of the developmental events that regulate UGT1A1 expression have remained a mystery.

Humanized UGT1 locus (hUGT1) mice, which express all nine UGT1A genes including UGT1A1, develop SNH (17). It is well established that the UGT1A1 gene contains a series of nuclear receptor (NR) binding sequences recognized by the pregnane X receptor (PXR) (18, 19), the constitutive androstane receptor (CAR) (18, 20), and the peroxisome proliferator–activator receptor (PPARα) (21). Thus, activation of these receptors by chemical ligands during the neonatal period leads to derepression of UGT1A1 and clearance of TSB. Although the function of these receptors in control of UGT1A1 expression is not entirely clear, we recently linked delayed expression of hepatic UGT1A1 during the developmental stage to PXR-mediated transcriptional silencing (19). When the Pxr gene was deleted in hUGT1 mice, newborn hUGT1/Pxr^−/− mice showed elevated hepatic UGT1A1 and reduced TSB levels, demonstrating that PXR participated in developmental repression of the UGT1A1 gene. Transcriptional silencing or repression is largely achieved by two prototypical NR corepressors, silencing mediator of retinoid and thyroid hormone receptors (SMRT) and nuclear receptor corepressor 1 (NCoR1), suggesting that NCoR1/SMRT played a role in control of UGT1A1 expression.

NCoR1 interacts with NRs and collectively controls gene expression patterns by recruiting chromatin-modifying enzymes to limit nucleosomal DNA accessibility and transcriptional activation (22, 23). Upon ligand binding, NRs release NCoR1 and recruit additional coactivators that cooperate with a different set of chromatin-modifying enzymes to promote transcription activation (23). Although global NCoR1 knockout is embryonic lethal (24), studies with tissue-specific knockout mice lacking NCoR1 in liver (25, 26), muscle (27), adipocyte (28), or macrophage (29) have uncovered its role in repression of a set of NRs including the thyroid receptor (TR), retinoic acid receptor (RAR), PPARs (30, 31), estrogen related receptor α (ERRα) (32), and liver X receptor (LXR) (33, 34). These receptors and NCoR1 control numerous biological pathways involving thyroid hormone signaling, lipid homeostasis, muscle endurance, and developmental regulation.

An important role for NCoR1 in regulating UGT1A1 gene expression in neonates has emerged as we initially discovered that loss of NCoR1 in the intestine, but not the liver, ameliorates neonatal hyperbilirubinemia in hUGT1 mice. Here, we document that NCoR1 mediates repression of UGT1A1 and other genes associated with multiple signaling pathways, biological processes, and intestinal maturation. By integrating transcriptome data, genome-wide maps of histone marks, and biochemical characterization of intestinal tissue, our results show that NCoR1 orchestrates a sophisticated epigenetic regulatory scheme from which complex and dynamic biological processes take place, all of which are followed by dramatic reduction in neonatal hyperbilirubinemia.

Results

Intestinal-Specific NCoR1 Deletion Ameliorates Neonatal Hyperbilirubinemia in Mice with a hUGT1 Background.

Humanized mice harboring the UGT1 locus encoding nine functional UGT1A proteins—including UGT1A1—in a Ugt1-null background, termed hUGT1 mice, develop neonatal hyperbilirubinemia (17, 35) because of delayed UGT1A1 expression in both the liver and GI tract. We demonstrated that knockdown of SMRT and NCoR1 in primary neonatal hepatocytes by siRNA leads to derepression of UGT1A1 (Fig. S1 A and B). Similar effects are seen after HDAC1 or HDAC3 ablation (Fig. S1 C and D). These findings implicate NCoR1 along with HDAC1/3 in controlling UGT1A1 gene expression. We first generated floxed NCoR1 mice under a hUGT1 background (TgUGT1/Ugt1^−/−/NCoR1^F/F), named as F/F^UN, and then further knocked out NCoR1 in liver (ΔHEP^UN mice) and intestinal tissue (ΔIEC^UN mice) of hUGT1 mice (Fig. 1 A and B). When we examined TSB levels in neonatal mice, we found that wild-type F/F^UN and ΔHEP^UN mice exhibited a similar pattern of neonatal hyperbilirubinemia as hUGT1 mice (Fig. 1C). Bilirubin accumulation in these mice was observed as early as postnatal day 1 (P1). TSB levels gradually increased, reaching ∼12 mg/dL at P14, then declined rapidly after P18 and eventually returned to a normal range (less than 2 mg/dL) at P21. Strikingly, following intestinal-specific deletion of NCoR1, the development of neonatal hyperbilirubinemia was completely diminished in ΔIEC^UN mice (Fig. 1C). The reduction of TSB levels in ΔIEC^UN mice was exclusively the result of the NCoR1 intestinal deletion, as hUGT1/NCoR1^ΔIEC/wt mice still exhibited high levels of TSB (Fig. 1D). The resulting bilirubin clearance in ΔIEC^UN mice also had a significant effect on its accumulation in fat tissue compared with that in control F/F^UN mice (Fig. 1E).

Fig. S1.

Gene siRNA knockdown in hepatocytes isolated from hUGT1 neonates at D12. (A–D) Primary hepatocytes were isolated from hUGT1 neonates at D12 and then treated with gene-specific siRNA. Forty-eight hours later, RNA was isolated for RT-QPCR analyses. Data were analyzed using Student’s t test and are expressed as mean ± SEM (n = 3); ns, no significance; *P < 0.05; **P < 0.01.

Fig. 1.

Tissue-specific NCoR1 deletion in hUGT1 mice. (A) Scheme for the generation of tissue-specific NCoR1 deletion in hUGT1 mice (hepatocytes, ΔHEP^UN or intestinal epithelial cells, ΔIEC^UN). (B) RT-QPCR of NCoR1 in livers from F/F^UN and ΔHEP^UN mice or intestines from F/F^UN and ΔIEC^UN mice (n = 5). **P < 0.01 (Student’s t test). (C) TSB levels during neonatal development. Data are expressed as mean ± SEM (n = 4–10). Two-way ANOVA analysis for ΔIEC^UN versus F/F^UN, P < 0.0001. (D) TSB levels from mice at P12. (E) Fat tissue collected from mice at day 12; yellow staining depicts bilirubin accumulation.

Developmental Stage-Dependent Derepression of UGT1A1 Expression Resulting from Intestinal NCoR1 Deletion.

Up-regulation of the intestinal UGT1A genes in ΔIEC^UN neonates in comparison with the F/F^UN littermate controls as determined by reverse transcription-quantitative polymerase chain reaction (RT-QPCR) analysis suggests the occurrence of transcriptional derepression of the UGT1A locus in ΔIEC^UN intestines (Fig. S2 A and B). In particular, the induction of intestinal UGT1A1 in ΔIEC^UN mice is most prominent throughout the developmental stage, starting at embryonic day 19 (E19) and continuing through to P19 (Fig. 2 A and B). The prominent derepression of the UGT1A1 gene in ΔIEC^UN neonates at P12 is clearly demonstrated in both the longitudinal axis and the crypt–villi axis (CVA) (Fig. 2 C and D). However, derepression of UGT1A1 expression in the absence of NCoR1 did not occur after weaning or in adulthood (Fig. 2 A and B and Fig. S2 C–E). Eventually, the UGT1A1 gene was expressed even in NCoR1-positive mice, suggesting that derepression of intestinal UGT1A1 following NCoR1 deletion is age-dependent, only taking place in the embryonic and suckling periods and having a direct impact on bilirubin glucuronidation in ΔIEC^UN neonates.

Fig. S2.

Intestinal expression of UGT1A and NCoR1 genes. (A–C) Gene expression of human UGT1 isoforms in SI from F/F^UN and ΔIEC^UN neonates at P12 and P19 and in adult mice. (D and E) Intestinal UGT1A1 gene expression along the longitudinal axis and along the CVA in adult mice (three mice of each strain, pooled).

Fig. 2.

Developmental stage-dependent derepression of the UGT1A1 gene resulting from intestinal NCoR1 deletion. (A) RT-QPCR of human UGT1A1 gene expression in SI at E19 and IECs of mice at P1, P7, P12, P19, P21, P30, and week 10 (n = 4–10, Student’s t test). ns, nonsignificant; **P < 0.01; ***P < 0.001; ****P < 0.0001; ^##P < 0.01; ^###P < 0.001. (B) UGT1A1 fold of change (one-way ANOVA, P < 0.0001). (C) IECs isolated from different SI sections, including duodenum (D), proximal and terminal jejunum (Jp and Jt), and ileum (Ip and It) (n = 3). Shown are RT-QPCR of human UGT1A1 gene (two-way ANOVA analysis, P < 0.0001) and Western blots of UGT1A1 and α-tubulin. (D) IECs were isolated sequentially along the CVA from neonatal mice at P12. Shown are RT-QPCR of human UGT1A1 gene (two-way ANOVA analysis, n = 3) and Western blots.

NCoR1 Deletion Alters Lipid and Energy Metabolism in ΔIEC^UN Neonates, Primarily Through Modification of Histone Acetylation.

To explore the molecular events following deletion of NCoR1 in intestinal tissue, we performed global transcriptional profiling in F/F^UN and ΔIEC^UN mice using RNA-sequencing (RNA-seq) technology. The scatter plot (Fig. 3A) showed considerable variability in gene expression between F/F^UN and ΔIEC^UN mice. With a false discovery rate (FDR) of <0.05, 1,023 genes were up-regulated and 830 genes were down-regulated (n = 3). Gene ontology analysis identified that the metabolic pathway was the most significantly altered pathway in ΔIEC^UN mice, followed by oxidative phosphorylation (OXPHOS) and drug metabolism (Fig. 3B). Deleting NCoR1 promotes expression of PPARα target genes but also enhances expression of Pparα, Pparβ, and their coactivator Pgc1α (Fig. 3C), which would have an extensive impact on FA and energy metabolism. Many other lipid metabolism-associated genes that are subject to LXRα regulation were also derepressed significantly in ΔIEC^UN intestines, such as cholesterol transport-related genes, and genes associated with unsaturated FA synthesis and lipogenesis (Fig. S3A); genes related to the retinol metabolism pathway are largely regulated through RAR following the deletion of NCoR1 (Fig. S3A). Messenger RNA levels of TCA cycle genes and OXPHOS genes were coordinately increased in ΔIEC^UN neonates (Fig. S3B). When we examined mitochondria by electron microscopy, mitochondria in IECs were tethered with adjacent mitochondria and displayed more compacted cristae than those in F/F^UN IECs (Fig. S3C). The mitochondrial organization in IECs of ΔIEC^UN mice may structurally enhance the interactions between adjacent mitochondria to allow more efficient energy transfer and communication. The deletion of intestinal NCoR1 leads to derepression of genes related to lipid and energy metabolism, simultaneously boosting both FA metabolism and lipogenesis.

Fig. 3.

Global transcriptomic alterations of ΔIEC^UN neonates and ChIP-seq analysis. SI were collected from mice at P12 for RNA-seq. (A) Scatter plot analysis. (B) KEGG pathway enrichment analysis; expression heat map of a subset of key genes in the biological processes including (C) fatty acid transport and oxidation, ketogenesis (KTG), and glyconeogenesis (GNG). Pink color represents the Log₂ fold of change (Log₂FC), and the green color represents the FDR. (D–F) Chip-seq analysis on F/F^UN and ΔIEC^UN neonates at P12. (D) Average profile of H3K27Ac near up-regulated genes. (E and F) Representative distribution of H3K27Ac at selected Scd2 gene against reference genome mm9 and human UGT1A1 gene against reference genome hg19.

Fig. S3.

Representative pathways of RNA-seq analysis and the correlation of RNA-seq with ChIP-seq. (A) Expression heat map from RNA-seq analysis, including a subset of key genes in the biological processes such as chylomicron and unsaturated FA (UFA) synthesis and retinol metabolism. (B) Citric acid (TCA) cycle and OXPHOS. (C) Electron microscope images of jejunum (arrows, mitochondria–mitochondria tethering). (D and E) IECs were isolated from the entire SI from F/F^UN and ΔIEC^UN neonates at P12. Chip-seq analysis was carried out by using different histone marks. (D) Correlation analysis between Chip-seq and RNA-seq data of ΔIEC^UN neonates; data described as number of genes (number of identified peaks). (E) KEGG pathway analysis of the commonly up-regulated genes identified in H3K27ac ChIP-seq.

These global transcriptomic alterations were further confirmed when we carried out ChIP-sequencing (ChIP-seq) analyses in IECs isolated from F/F^UN and ΔIEC^UN mice to assess histone modifications across the genome. We examined histone marks associated with active transcription toward acetylated H3K27ac and H3K9ac, methylated H3K4me3 and H3K4me1, as well as transcriptional repression through H3K9me3 and H3K27me3. The most significant changes resulting from NCoR1 deletion were revealed in H3K27ac levels with the identity of 527 genes and 1,304 elevated peaks generated across the genome (Fig. S3D), which are associated with transcriptional activation of over 50% of genes in ΔIEC^UN intestines identified by RNA-seq analyses and the similar KEGG analysis (Fig. S3E). The derepressions were significantly increased in transcription start site (TSS) regions of genes that were up-regulated in ΔIEC^UN mice (Fig. 3D). The increase in identified sequence is illustrated with the Scd2 gene and the human UGT1A1 gene (Fig. 3 E and F). These data collectively indicate that removing the corepressor NCoR1 changes the chromatin structure and promoter accessibility—primarily through acetylation modification—resulting in alteration in transcription activities of NCoR1-targeted genes.

Differential Gene Regulation in Neonatal Versus Adult ΔIEC^UN Mice.

Although we have confirmed that UGT1A1 is also a target gene of PPARα and LXRα/β, we considered that the activation of these NRs may target the UGT1A1 gene and lead to its developmental-dependent derepression. To examine this possibility in greater detail, RNA-seq experiments were conducted using intestinal tissue from adult F/F^UN and ΔIEC^UN mice. From the 1,022 genes up-regulated in neonatal tissue and the 1,265 genes up-regulated in adult tissue following NCoR1 deletion, there were 462 common genes (Fig. 4A), among which key biological processes involved in membrane synthesis, FA metabolism and transport, retinol synthesis, and PPAR signaling were shared between neonatal and adult tissue in NCoR1-deleted IECs (Fig. S4A), with some specific genes demonstrated in Fig. S4B. In contrast, other regulatory patterns as illustrated by the Cyp4a10 gene were enhanced in adult ΔIEC^UN mice (Fig. 4B). More interestingly, a host of genes encoding proteins linked to drug metabolism were highly derepressed in ΔIEC^UN neonates but much less affected in adult mice. When we analyzed the expression profiles of a few representative genes in this class (i.e., Cyp2c55, Cyp2c65, Cyp3a11, Ugt2a3) at various developmental stages, we found that the derepression of these genes, which peaked in the middle suckling period, was developmentally stage-dependent in ΔIEC^UN neonates, remarkably like the derepression pattern of UGT1A1 (Fig. 4C). These results suggest that NCoR1 expression impacts gene expression differentially between neonatal and adult mice.

Fig. 4.

Common and differential gene regulation of ΔIEC^UN mice following development. (A) Venn diagram to compare RNA-seq data from both neonatal mice (N) and adult mice (A). (B) RT-QPCR analysis demonstrated the progressive enhancement of gene regulation in neonatal versus adult mice (gray bar, F/F^UN; black bar, ΔIEC^UN). ns, nonsignificant; *P < 0.05; **P < 0.01 (Student's t test). (C) Small intestines were collected and pulverized for total RNA extraction of mice at different developmental stages, followed by RT and QPCR analysis.

Fig. S4.

Neonatal to adult gene expression patterns. (A) KEGG pathway analysis of the commonly regulated genes and genes specifically regulated in neonates. (B) Mice were killed on day 12 (neonates) or week 10 (adults). Small intestines were collected and pulverized for total RNA extraction, followed by RT and QPCR analysis (n = 3). **P < 0.01, ***P < 0.001 (Student’s t test). Gray square, F/F^UN; black square, ΔIEC^UN.

NCoR1 Deletion Accelerates IEC Migration and Cell Maturation.

Analysis of gene expression profiles further enlightened our speculation that NCoR1 deletion led to precocious development of the intestines. There were no differences in total body weight between F/F^UN and ΔIEC^UN littermates at P12 (n = 10, P > 0.05) (Fig. S5A); however, the average length of the small intestine was ∼8% longer (n = 10, P < 0.01) (Fig. 5A), and the average weight of the intestines ∼15% greater (n = 10, P < 0.01) in ΔIEC^UN (Fig. 5B). H&E-stained intestinal sections displayed similar morphological characteristics between F/F^UN and ΔIEC^UN mice (Fig. S5B), and well-developed brush border and microvilli were visualized by electron microscopy (Fig. S5C). The effect of NCoR1 on cellular proliferation was evaluated by K_i-67 immunostaining and BrdU incorporation. K_i-67–positive proliferating cells were slightly increased in both duodenum and jejunum but not ileum of ΔIEC^UN mice, compared with those in F/F^UN mice (Fig. 5C). After neonatal mice were exposed to BrdU for 2.5 h, the number of BrdU-positive cells had increased by 15.6% in duodenum and 18.7% in jejunum in ΔIEC^UN mice (n = 5, P < 0.05), with no significant difference being observed in ileum (Fig. 5 D and E). After 48 h BrdU exposure, BrdU-positive cells migrated a significantly greater length in duodenum, jejunum, and ileum of ΔIEC^UN mice than those of F/F^UN mice (n = 4, P < 0.05) (Fig. 5 F and G). The results demonstrated that NCoR1 deletion accelerated cell proliferation and promoted epithelial cell migration in ΔIEC^UN mice. Sucrase isomaltase (Sis), a brush border glucosidase, only exists in differentiated duodenal and jejunal enterocytes and has been used as a marker of enterocyte maturation (36, 37). Sis gene expression along the CVA was up-regulated in ΔIEC^UN mice (Fig. 6 A and B). Elevated Sis expression was also observed longitudinally in both duodenum and jejunum in ΔIEC^UN (Fig. 6C). Up-regulation of other maturation markers in ΔIEC^UN, including the duodenal-specific Akp3 gene and jejunal-specific Krt20 gene (38–41), further supports the occurrence of cell maturation in the absence of NCoR1 (Fig. 6C). In addition, expression of Glb1, Nox4, and Lrp2, which is down-regulated at the suckling to weaning transition (38, 39), was also decreased in ΔIEC^UN (Fig. 6D). Altered expression of these genes is indicative of intestinal tissue maturation.

Fig. S5.

Impact of NCoR1 on body weight and intestinal tissue of mice at P12. (A) Body weight (n = 10). ns, nonsignificant. (B) H&E staining of the paraffin sections of duodenum, jejunum, and ileum. (C) Electron microscope images of jejunum (images were taken by Serial EM, then constructed and binned).

Fig. 5.

NCoR1 deletion accelerates migration of IECs. SI length (A) and weight (B) of F/F^UN and ΔIEC^UN mice at P12 (n = 10). (C) Immunofluorescent stainings of Ki67 in frozen sections of duodenum (D), jejunum (J), and Ileum (I), and Ki67-positive cells were counted and described as averages ± SEM (n = 5). (D and E) F/F^UN and ΔIEC^UN at P12 were treated with BrdU through i.p. injection at 0.5 mg/mice. After 2.5 h, sections of SI were prepared for paraffin embedding. Paraffin sections were stained with a BrdU antibody, and BrdU-positive cells were counted (n = 5, Student’s t test). (F and G) Mice at P10 were treated with BrdU. Forty-eight hours later, samples were prepared for BrdU staining. The migration of BrdU-positive cells was measured and described as a percentage of villi length (n = 5, Student’s t test analysis). ns, nonsignificant; *P < 0.05; **P < 0.01; ***P < 0.001.

Fig. 6.

NCoR1 deletion accelerates cell maturation of IECs. (A) IECs were isolated sequentially along the CVA in mice at P12, followed by RT-QPCR of Sis gene expression with two-way ANOVA analysis (n = 3). ***P < 0.001. (B) IF staining of Sis (red) in intestine frozen sections, counterstained with DAPI (blue). (C and D) IECs were isolated from different SI sections in mice at P12 (n = 3, samples were pooled). RT-QPCR analysis of Sis, Akp3, Krt20 (J), and Glb1, Nox4, Lrp2 (K) were carried out.

NCoR1 Expression Is Regulated by Activation of Intestinal Kinase Activity.

The linkage recently disclosed between the phosphorylation of NCoR1 by IKK leading to the cytoplasmic accumulation of NCoR1 (42) urged us to evaluate the potential involvement of NCoR1 in IKKβ deletion-induced UGT1A1 repression. We had previously demonstrated the targeted deletion of intestinal IKKβ in hUGT1/Ikkβ^ΔIEC mice leads to greater neonatal accumulation of TSB (43). The increase in TSB levels in neonatal hUGT1/Ikkβ^ΔIEC mice is a direct result of the reduction in intestinal UGT1A1 expression (Fig. 7A). RT-QPCR analysis demonstrated that following deletion of intestinal IKKβ in hUGT1/Ikkβ^ΔIEC mice, intestinal NCoR1 gene expression was significantly induced (Fig. 7B). Because of this change, superrepression initiated by NCoR1 was apparent in decreased gene expression of UGT1A1, along with other target genes, including Scd1, Scd2, Hmgcs2, Cpt1a, and Cyp4a10 (Fig. S6). Several of the small intestine maturation markers, including Sis, Akp3, and Krt20, were also significantly down-regulated in hUGT1/Ikkβ^ΔIEC mice (Fig. 7C). The absence of IKKβ in hUGT1/Ikkβ^ΔIEC mice reverses gene expression patterns that are observed in ΔIEC^UN mice, demonstrating that NCoR1 in hUGT1/Ikkβ^ΔIEC mice has enhanced transcriptional repressive properties.

Fig. 7.

Impact of IKKβ on the expressions of NCoR1 and intestinal maturation genes. (A–C) Intestine samples were collected from both 12-d-old control and hUGT1/Ikkβ^ΔIEC mice (n = 5). RT-QPCR was carried out to determine gene expression patterns of UGT1A1, NCoR1, and intestinal maturation markers. (D and E) SI were collected from mice carrying the constitutive active IKKβ (n = 6) at 12 d old. RT-QPCR was performed to determine the gene expressions of both up- and down-regulated intestinal maturation markers. *P < 0.05; **P < 0.01; ***P < 0.001 (Student's t test).

Fig. S6.

Gene expression in hUGT1/Ikkβ^ΔIEC mice. RT-QPCR of intestinal tissues collected from both hUGT1/Ikkβ^F/F and hUGT1/Ikkβ^ΔIEC mice at 12 d old (n = 5). *P < 0.05 (Student’s t test).

A cDNA clone encoding a constitutively active variant of IKKβ was cloned into plasmid containing the 12.4-kb Villin promoter and used to generate IKKβ(EE)^IEC transgenic mice that expressed constitutively active IKKβ in IECs (44). When we examined constitutively active intestinal IKKβ activity in IKKβ(EE)^IEC neonates at P12, IEC maturation marker genes Sis, Akp3, and Krt20 were dramatically induced, and Glb1, Nox4, and Lrp2 genes are significantly reduced in IKKβ(EE)^IEC mice compared with their wild-type littermates (Fig. 7 D and E), a pattern like that observed in ΔIEC^UN mice. These findings strongly implicate a role for IKKβ-directed phosphorylation in control of NCoR1 function during neonatal intestinal development.

Discussion

SNH is of significant clinical concern in many parts of the world and is associated with substantial mortality and morbidity in low- and middle-income countries (LMICs) (16). Recent comprehensive reviews that examined the burden of SNH in LMICs have shown very limited progress on the epidemiological profile since earlier reports over 55 y ago (16). For the most part, an understanding of the environmental and genetic parameters that lead to the rapid rise in TSB levels has been the primary focus in attempting to describe the mechanisms leading to hyperbilirubinemia, with limited insight on those events that developmentally limit expression of UGT1A1. An important inroad toward examining the cellular and molecular mechanisms that control the metabolism of bilirubin has been accomplished by the development of hUGT1 mice (17), an animal model that can be used to replicate SNH and characterize the events controlling UGT1A1 expression. It is now understood that delayed expression of hepatic UGT1A1 in neonatal hUGT1 mice is a controlled event, with the UGT1A1 gene being actively repressed through participation with PXR (19). We have previously reported that complete interruption of the hepatic Ugt1 locus through targeted knockout in Ugt1^ΔHEP mice is not sufficient to induce SNH (45), implicating an important role for the metabolism and clearance of bilirubin by other tissues. Indeed, selective induction of only intestinal UGT1A1 in neonatal hUGT1 mice is sufficient to reverse SNH (35, 43). This finding, and others, provides evidence that intestinal UGT1A1 is repressed during development but can be successfully induced to alter bilirubin metabolism and reverse the development of hyperbilirubinemia.

The centerpiece in controlling intestinal UGT1A1 expression is regulation of NCoR1 function, which has a direct impact on UGT1A1 gene expression and IEC maturation. When NCoR1 is selectively deleted in IECs in ΔIEC^UN mice, newborns do not display the escalating TSB levels that are prominent in neonatal hUGT1 mice. In addition, they show significantly elevated levels of UGT1A1 throughout the longitudinal range of the SI in addition to the IECs ranging from the crypts to the ends of the villi. ChIP-seq experiments validated that deacetylated histone regions were prominent in NCoR1 binding, adding support that histone deacetylases, such as HDAC3, are associated with NCoR1 binding (46). RNA-seq studies demonstrated that the deletion of NCoR1 promotes activation of PPARα and LXRα/β target genes involved in lipid and energy metabolism, as evident by increases in metabolic pathways, OXPHOS, and FA metabolism. All of these steps are necessary for successful IEC maturation. The UGT1A1 gene can be actively induced in the presence of PPARα/γ and LXRα/β ligands, so the possibility exists that deletion of NCoR1 and activation of these NRs are underlying the induction of UGT1A1 in IECs. Additionally, NCoR1 controls large sets of genes that are differentially regulated during the neonatal period or once the mice become adults. An important class of developmentally regulated genes during the neonatal period are those involved in the transition of neonatal to adult intestinal epithelium, indicating that NCoR1 functions along with the transcriptional repressor B lymphocyte-induced maturation protein 1 (38) in events involved in IEC maturation and development. Because NCoR1 deletion leads to IEC proliferation in neonatal mice, an event that does not occur in adult tissue, the action of IEC proliferation may underlie UGT1A1 derepression. NCoR1 ChIP-seq analysis can markedly increase the novelty of this work in elucidating the mechanism of NCoR1 repression on UGT1A1 gene expression. However, none of the commercial antibodies available for this work were successful in ChIP-seq experiments using neonatal intestinal tissue. Similar limitations were also encountered when Yamamoto tried to address the function of NCoR1 as an important regulator of muscle mass and oxidative function (27). To minimize this antibody limitation, our work has included ChIP-seq data by using different methylation and acetylation histone marks. H3K27ac is the primary player in response to the removal of the corepressor NCoR1 and the changes in chromatin structure and promoter accessibility.

The functional ability of NCoR1 to promote its repressive properties with NRs as well as its potential to activate gene transcription have been linked to specific phosphorylation events (23, 26), with a host of different kinases implicated in NCoR1 regulation (47–49). The loss of nuclear NCoR1 in malignant melanoma was associated with IKK-dependent phosphorylation of specific NCoR1 serine resides (42). When we deleted IKKβ selectively from IECs in hUGT1/Ikkβ^ΔIEC mice, UGT1A1 gene expression was superrepressed along with other target genes involved in IEC maturation, followed by elevations in TSB levels. Interestingly, IEC maturation genes analyzed by RT-QPCR during neonatal development were expressed in a highly concordant pattern in intestinal tissue from IKKβ(EE)^IEC mice, directly implicating a key role for IKKβ expression toward control of NCoR1 release from repressed genes, such as UGT1A1.

We are proposing that there is a link between intestinal IKK activity and events that lead to NCoR1 repression of intestinal UGT1A1 gene expression shortly after birth. Human breast milk (BM), especially colostrum, contains glycans that inhibit Toll-like receptors (TLRs) and prevent downstream mediated inflammation (50). Upon ligand binding by pathogen-associated molecular patterns (PAMPs), expressed by microflora (51), TLRs are activated and initiate intracellular signaling events resulting in regulation of genes activated by the IKK/NF-κB pathway (52). Relative to term infants or adults, preterm infants express higher concentrations of TLRs, which upon activation may increase the risk for neonatal sepsis or necrotizing enterocolitis (53) through the IKK/NF-κB–mediated inflammatory response. The components of BM block this response by inhibiting the TLRs. Based upon our findings, we hypothesize that components of BM attenuate IEC TLRs, limit PAMP-induced IKKβ activity and downstream phosphorylation events, and minimize UGT1A1 gene expression through an NCoR1-dependent mechanism. The delayed expression of the UGT1A1 gene can be reversed by agents that activate TLRs, such as LPS (35), or components of BM, such as FAs, which are ligands for activation of PPARα, a NR linked to NCoR1 repression and activation of the UGT1A1 gene (21). The oral administration of high concentrations of FAs, such as linoleic acid, oleic acid, and docosahexaenoic acid, to neonatal hUGT1 mice resulted in lowering of TSB levels (54). In addition, a diet of formula, as we have previously shown (35), lacking TLR-bound glycans, allows downstream activation of the IKK/NCoR1 loop and expression of the UGT1A1 gene.

In conclusion, we have demonstrated that the transition of intestinal epithelial tissue from its immature status shortly after birth to an advanced network of crypts and villi, as observed in adult mice, can be viewed as a carefully programmed event. Regulation of neonatal intestinal UGT1A1 and its impact toward the control and onset of hyperbilirubinemia is intimately tied with IEC maturation. The findings that we have outlined in this animal model, such as the linkage of IKK with the inhibition of NCoR1, may be of value in identifying chemicals, therapeutics, or dietary agents that would be useful in lowering TSB levels in children exhibiting SNH.

Methods

Animal Studies.

hUGT1 mice were previously generated by introducing a human UGT1 transgene into mice with a Ugt1-null background (17, 55, 56). We crossed hUGT1 with NCoR1^F/F mice (26, 27) to generate hUGT1/NCoR1^F/F (F/F^UN) mice, which were further bred with Villin-Cre (57) or Albumin-Cre transgenic mice (Jackson laboratory) (58, 59) to obtain the compound mutants Vil-Cre/TgUGT1/Ugt1^−/−/NCoR1^F/F (ΔIEC^UN) and Alb-Cre/TgUGT1/Ugt1^−/−/NCoR1^F/F (ΔHEP^UN). Generation of mouse strains hUGT1/Ikkβ^ΔIEC (43) and IKKβ(EE)^IEC (44) have been previously described. All of the mouse strains were housed in a pathogen-free University of California, San Diego (UCSD) vivarium. All animal protocols were reviewed and approved by the UCSD Animal Care and Use Committee. Cre-negative mice (F/F^UN) served as controls. Siblings were preferred for all of the experiments.

Intestinal Tissue Sections.

Entire small intestines were dissected from mice, sectioned, snap-frozen in liquid nitrogen, and stored at –80 °C. Frozen tissues were pulverized for further RNA and protein extraction. SI segments were sectioned as duodenum (D), proximal jejunum (Jp), terminal jejunum (Jt), proximal ileum (Ip), and terminal ileum (It). Markers specific for duodenum (Akp3), jejunum (Sis), and ileum (Nox4, Lrp2) were measured by RT-QPCR, and the results were consistent with a previous publication (41).

IEC Isolation and IEC Sequential Isolation Along CVA.

IECs were isolated with a modified method (60). Dissociated IECs were collected by centrifugation. The sequential isolation of IECs along the CVA was carried out according to a previous description (61). Briefly, jejunums were dissected and cut longitudinally. After washing briefly in buffer A (DPBS with 27 mM sodium citrate), tissue was incubated in buffer B (DPBS with 1.5 mM EDTA and 0.5 mM DTT) at 37 °C with very gentle shaking. Following a series of incubations (2, 2, 3, 4, 5, 7, 10, 15, 15, and 15 min), a total of 11 fractions of IECs were isolated, and the gradient of cells from villus tip to the lower villus and crypt area was confirmed by the protein expression pattern of PCNA. Cells isolated from three mice of each strain were pooled for analysis.

Total RNA Preparation, RT-QPCR, and RNA-Seq Analysis.

Total RNA from cell and tissue samples were prepared for RT by using the iScript cDNA synthesis kit (BioRad). Real-time (Q) PCR experiments were carried out on a CFX96 QPCR system (BioRad) by using Ssoadvanced SYBR Green reagent (BioRad). Primers were designed through mouse primer depot (https://mouseprimerdepot.nci.nih.gov/). Isoform-specific primers for hUGT1A are listed in Table S1. Intestine RNA was prepared from F/F^UN and ΔIEC^UN mice at either 12 d old (neonates) or 10 wk old (adults) for RNA-seq. RNA from two mice were pooled, and a total three samples per strain were used for RNA-seq studies. Polyadenylated mRNA was purified with the Dynabeads mRNA purification kit (ThermoFisher). The mRNA libraries were prepared for strand-specific sequencing as previously described (62) and sequenced using an Illumina HiSeq 2500 by running 36 cycles. Image deconvolution, quality value calculation, and the mapping of exon reads and exon junctions were performed. Sequencing reads were aligned to the Mus musculus (UCSC mm9) genome. RNA-seq data have been deposited in the GEO database under accession no. GSE86927.

Table S1.

Primers used to identify the UGT1A gene products by RT-QPCR

Targets	Sequence, 5′–3′
UGT1A1-F	CCATCATGCCCAATATGGTT
UGT1A1-R	CCACAATTCCATGTTCTCCA
UGT1A3-F	TCAACTGTGCCAACAGGAAG
UGT1A3-R	CTGAGACCATTGATCCCAAAG
UGT1A4-F	CAACGGGAAGCCACTATCTC
UGT1A4-R	TGAGACCATTGATCCCAAAGA
UGT1A6-F	TGCTTGAATATCCTAGGCCG
UGT1A6-R	ACCACAATTCCATGTTCTCCA
UGT1A7-F	TGTCATCAGGGAAAGCCAGT
UGT1A7-R	TGAGACCATTGATCCCAAAGA
UGT1A9-F	GAACATTTATTATGCCACCG
UGT1A9-R	CAACAACCAAATTGATGTGTG
UGT1A10-F	TGATGCCCAACATGATCTTC
UGT1A10-R	CCACAATTCCATGTTCTCCA

ChIP-Seq Analysis.

IECs were freshly isolated from both ΔIEC^UN and F/F^UN neonates at P12 and were subjected to ChIP (MAGnify ChIP system, ThermoFisher). After reverse cross-linking, library preparation was carried out by using the MicroPlex Library Preparation Kit V2 (Diagenode, C05010012). Cluster generation and 50 cycles of single-end sequencing were carried out on an Illumina NextSeq 500. Sequence tags were aligned (mapped) to the unmasked mouse (mm9) or human (hg19) reference genome. The following antibodies were used: anti-H3K4me1 (Abcam, ab8895), anti-H3K4me3 (Abcam, ab8580), anti-H3K9me3 (Abcam, ab8898), anti-H3K27me3 (Active Motif, AM61017), anti-H3K9ac (Active Motif, AM39137), and anti-H3K27ac (Abcam, ab4729). ChIP-seq data have been deposited in the GEO database under accession no. GSE86996.

Histology, Immunohistochemistry (IHC), and Immunofluorescence (IF).

Paraffin-embedded sections were used for routine H&E staining. BrdU IHC was performed according to the manufacturer’s instructions (eBioscience, 8800–6599-45). Briefly, neonates at day 12 (BrdU exposure for 2.5 h) or at day 10 (BrdU exposure for 48 h) were treated with BrdU at 50 mg/kg by i.p. injection. Paraffin-embedded SI sections were applied for IHC staining and detected by the ABC detection system. Slides were examined under an upright Imager A2 microscope (Zeiss). Frozen sections were prepared in OCT compound for IF staining. After the overnight incubation with primary antibodies, slides were then washed and stained with Alexa-488–conjugated secondary antibodies (Life Technologies), alongside a DAPI counterstaining. Mounted slides were examined under Leica TCS SPE confocal microscopy. The following antibodies were used: anti-Ki67 (GeneTex, GTX16667), anti–β-catenin (Santa Cruz Biotechnology, sc-1496), and anti-sis (Santa Cruz Biotechnology, sc27603).

Protein Preparation and Western Blots.

Tissue or cell lysates were prepared in a RIPA buffer for gel electrophoresis [Nupage 4–12% (wt/vol) Bis-Tris gradient gel ThermoFisher]. Western blots were developed and imaged using a ChemiDoc Touch Imaging System (BioRad). The following antibodies were used: rabbit anti-human UGT1A1 (Abcam, ab170858), anti-PCNA (BD Pharmingen, 555566), and anti–α-Tubulin (Sigma, T9026).

Statistics.

All results were subjected to statistical analysis. Student’s t test analyses were performed for most of the studies unless specified otherwise. All statistics and graphs were generated by using GraphPad Prism software. Data are expressed as mean ± SEM, and P values smaller than 0.05 were considered as statistically significant: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Electron microscope imaging and primary hepatocytes isolation are describe in SI Methods.

SI Methods

Transmission Electron Microscope.

Electron microscope imaging was performed at the National Center for Microscopy and Imaging Research (NCMIR) at UCSD according to modification of a method of Antonucci (63). Briefly described, ΔIEC^UN and F/F^UN neonates were euthanized on day 12, and jejunums were dissected and fixed in a freshly made solution of 2% (wt/vol) paraformaldehyde, 2.5% (wt/vol) glutaraldehyde, and 0.15 M sodium cacodylate buffer overnight at 4 °C. Tissues were washed with ice-cold 0.15 M sodium cacodylate containing 2 mM CaCl₂ and then postfixed with 1.5% (wt/vol) KFe(CN)₃, 0.15 M cacodylate buffer, 2 mM CaCl2, and 2% (wt/vol) OsO₄(aq). Tissues were washed with ice-cold water and stained with 1% uranyl acetate overnight at 4 °C. Tissues were dehydrated with graded ethanol solutions before embedding in Durcupan ACM resin (Fluka) for ultramicrotomy by using a Leica Ultracut UCT ultramicrotome. Sections were poststained with Sato Lead Solution before TEM imaging. Electron micrographs were taken on a FEI Technai 12 (Spirit) (120 kV) with a Teitz TemCam F224 2k by 2k CCD camera. Montages of intestinal villi were obtained using Serial EM software at 4,400× magnification in a 10 × 10 or 20 × 20 region of interest. Mosaics were constructed using the Blendmont feature in 3dmod (IMOD version 4.7, University of Colorado), and the resulting mosaics were binned.

Hepatocytes Isolation and SiRNA Treatment.

Primary hepatocytes were isolated from neonatal mouse livers and treated with gene-targeted siRNA as previously described (19, 56). Briefly, hepatocytes were dissociated through liver perfusion with Hanks buffer with liberase (Roche, 05401020001) and plated in six-well collagen-coated plates (BD). Four hours later, hepatocytes were transfected with Silencer select siRNA (ThermoFisher-Ambion) by using lipofectamine RNAiMAX reagent (ThermoFisher). The siRNA included siControl, siNCoR1, siSMRT, siHDAC1, siHDAC2, and siHDAC3. Cells were collected 48 h after transfection.