Abstract
Bilirubin-induced neurologic dysfunction (BIND) and kernicterus has been used to describe moderate to severe neurologic dysfunction observed in children exposed to excessive levels of total serum bilirubin (TSB) during the neonatal period. Here we use a new mouse model that targets deletion of the Ugt1 locus and the Ugt1a1 gene in liver to promote hyperbilirubinemia-induced seizures and central nervous system toxicity. The accumulation of TSB in these mice leads to diffuse yellow coloration of brain tissue and a marked cerebellar hypoplasia that we characterize as kernicterus. Histologic studies of brain tissue demonstrate that the onset of severe neonatal hyperbilirubinemia, characterized by seizures, leads to alterations in myelination and glia reactivity. Kernicterus presents as axonopathy with myelination deficits at different brain regions, including pons, medulla oblongata, and cerebellum. The excessive accumulation of TSB in the early neonatal period (5 days after birth) promotes activation of the myelin basic protein (Mbp) gene with an accelerated loss of MBP that correlates with a lack of myelin sheath formation. These changes were accompanied by increased astroglial and microglial reactivity, possibly as a response to myelination injury. Interestingly, cerebellum was the area most affected, with greater myelination impairment and glia burden, and showing a marked loss of Purkinje cells and reduced arborization of the remaining ones. Thus, kernicterus in this model displays not only axonal damage but also myelination deficits and glial activation in different brain regions that are usually related to the neurologic sequelae observed after severe hyperbilirubinemia.
Introduction
Hyperbilirubinemia is a common clinical condition occurring in the neonatal period. Over 60% of term and virtually all premature infants experience temporary, mild-to-moderate “physiological” jaundice, owing to excessive production of unconjugated bilirubin (UCB) and defective bilirubin clearance (Stevenson et al., 2001; Cohen et al., 2010). In the vast majority of cases neonatal jaundice represents a benign condition, but in some newborns the concentration of serum bilirubin may increase with passage of the pigment into brain causing various degrees of acute or chronic bilirubin-induced neurologic dysfunction (BIND), which can expand to UCB-induced encephalopathy (kernicterus) (Shapiro, 2003, 2005, 2010). In this situation bilirubin leads to cellular neuroinflammation with activation of astrocytes and microglia, as well as gliosis. (Shapiro, 2005; Yueh et al., 2014; Johnson and Bhutani, 2011; Brites, 2012). The benchmark of extreme bilirubin neurotoxicity or kernicterus is an irreversible posticteric sequelae, presented as icteric or yellow staining of brain tissue resulting from the accumulation of UCB in selective regions of the brain (Johnson and Bhutani, 2011). Hence, it is important to understand the molecular mechanisms by which bilirubin exerts such neurodevelopmental abnormalities.
Humanized UGT1 (hUGT1) mice express the human UGT1 locus in a Ugt1-null background (Fujiwara et al., 2010). Mutation of the Ugt1 locus by interruption of exon 4 in the common region of the Ugt1a1 gene leads to inactivation of the UGT1A proteins, including UGT1A1 (Nguyen et al., 2008). Since UGT1A1 is the only glucuronosyltransferase responsible for the glucuronidation of bilirubin (Bosma et al., 1994), Ugt1−/− mice accumulate high levels of UCB, resulting in neonatal lethality approximately one week after birth. Incorporating the human UGT1 locus through transgenic technology into the UGT1-null background leads to recovery of neonatal lethality, although the newborn hUGT1 mice all display severe hyperbilirubinemia (Fujiwara et al., 2010). Approximately 10% of neonatal hUGT1 mice progress into seizures, succumbing to central nervous system (CNS) toxicity as evidenced by accumulation of bilirubin in brain tissue. The stark accumulation of bilirubin in brain tissue coincides with neuroinflammation and reactive gliosis, a term we have defined as “kernicterus” in these mice, which we have also found linked to Toll-like receptor 2 (TRL2) control of bilirubin-induced signaling (Yueh et al., 2014). TRL2-mediated gliosis correlated with the development of a bilirubin-induced proinflammatory environment in which there is upregulation of inflammatory markers such as tumor necrosis factor α (TNFα), interleukin (IL)-1β, and IL-6 in the CNS. Expression of TRL2 is a key intermediate in regulating the onset of BIND, since deleting the Tlr2 gene in hUGT1 mice leads to a dramatic increase in the neonatal death rate in hUGT1/Trl2−/− mice. These findings represent the first mechanistic link between hyperbilirubinemia and CNS toxicity, demonstrating that Toll-like receptor 2 signaling and microglia-associated neuroinflammation are linked to a repair and protection mode against BIND.
The early onset of hyperbilirubinemia during the neonatal period in hUGT1 mice is identified by severely elevated levels of total serum bilirubin (TSB) that coincide with the onset of seizures. As approximately 1 in 10 hUGT1 mice develop these symptoms (Fujiwara et al., 2010), we sought to develop a more consistent mouse model to examine the impact of hyperbilirubinemia on gliosis and CNS damage. This was accomplished by using the Cre/loxP system (Lewandoski, 2001) to target the deletion of the Ugt1a1 gene in liver tissue. All of these mice develop elevated levels that occur during neonatal development. This condition results in the development of kernicterus with a visible reduction in cerebellum volume. Microscopically, kernicterus formation presents a reduced number of axons and reduced myelination in different brain regions. In parallel, myelination impairment was associated with increased astroglial and microglial reactivity in the same brain regions. The development of kernicterus in these mice provides important new clues toward understanding the early events leading to neonatal toxicity by bilirubin.
Material and Methods
Generation of the Kernicterus Mouse Model.
The targeting construct consisted of a phosphoglycerate kinase (PGK)–neomycin resistance gene cassette (PGK-neo) that was flanked by Flp/FRT recombinase sites in the intron region between exons 3 and 4 of the Ugt1a1 gene. Positioned in intron 2 and then again outside of the Flp/FRT recombinase sites are Cre/loxP recombinase sites. This construct was electroporated into embryonic stem cells, and neomycin-positive clones were injected into C57BL/6 blastocysts. The chimera mice were out-crossed with wild-type C57BL/6 mice for five generations and then in-bred to generate mice homozygous for the Ugt1LoxP[FRTneoFRT]LoxP allele. These mice were identified as UFP mice. To delete exons 3 and 4 in liver tissue, UFP mice were crossed with mice that express as a transgene albumin-Cre, generating UFP/albumin-Cre mice, and were designated UAC mice. Deletion of the Ugt1a1 exons 3 and 4 in liver tissue leads to severe neonatal hyperbilirubinemia in UAC mice. Since exons 3 and 4 encode the “common” region of the Ugt1 locus, the other Ugt1a genes are also inactivated. All mouse experiments and procedures were in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by the University of California San Diego Animal Care Committee.
Immunoblotting.
Liver and small intestinal microsomes from mice were prepared as previously described (Chen et al., 2013). All Western blots were performed by using NuPAGE 4–12% BisTris-polyacrylamide gels as outlined by the supplier (Invitrogen/ThermoFisher Scientific, Waltham, MA). Following transfer of the proteins, membranes were blocked with 5% nonfat dry milk in Tris-buffered saline for 1 hour and then incubated with primary antibodies in Tris-buffered saline overnight. Membranes were washed and incubated with horseradish peroxidase–conjugated secondary antibody for 1 hour. Antibodies used were a rabbit anti-human UGT1A1 (Chen et al., 2013) and a mouse anti-bovine myelin basic protein (Bio-Rad, Hercules, CA). Both antibodies crossreacted with the mouse proteins. The conjugated horseradish peroxidase was detected using the ECL plus Western blotting detection system (Amersham Biosciences, Piscataway, NJ) and visualized by the Bio-Rad gel documentation system.
Bilirubin Glucuronidation Assays.
Bilirubin glucuronidation activity assay was performed as described previously (Chen et al., 2013). In brief, mouse liver microsomes or small intestinal microsomes were incubated with different concentrations of bilirubin in reaction buffer containing 2 mM UDP-glucuronic acid. All reactions were performed in the dark. Bilirubin diglucuronide formation was determined by liquid chromatography (Alliance 2695; Waters Corporation, Milford, MA ) coupled to tandem mass spectrometry (API 4000; Applied Biosystems/MDS SCIEX, Concord, ON, Canada). The high-performance liquid chromatography system used was equipped with a 50- × 3.2-mm Columbus C18 column (Phenomenex, Torrance, CA). Data are expressed as peak areas (AUC).
Total RNA Preparation and RNA Analysis by Reverse Transcription–Polymerase Chain Reaction.
Mouse tissues were collected and snap-frozen into liquid nitrogen, and then pulverized. Aliquots of the pulverized samples were homogenized in TRIzol (Invitrogen) for RNA isolation (Chen et al., 2013). Reverse transcription (RT) was conducted by using iScript Reverse Transcriptase (Bio-Rad) as outlined by the manufacturer. Following synthesis of cDNA, RT products were used for polymerase chain reaction (PCR). Primer sequences for the Ugt1a1 gene are presented in Chen et al. (2013). Real-time PCR analysis of Mrp gene expression was conducted as previously described (Yueh et al., 2014), using forward (5-CCATCCAAGAAGACCCCACA-3′) and reverse (5-CCCCTGTCACCGCTAAAGAA-3′) primers.
Immunohistochemistry.
Mice were anesthetized with pentobarbital (40 mg/kg, i.p.) and perfused through the right ventricle with 0.1 M phosphate-buffered saline (PBS) (pH 7.4) followed by the same buffer containing 4% paraformaldehyde. Fixed tissues were postfixed in 4% paraformaldehyde in PBS for 72 hours at room temperature (RT), dehydrated through a graded ethanol series, and embedded in paraffin. For immunostaining, 3-μm sections were submitted to antigen retrieval in 20 mM citrate buffer with 1.5% H2O2 for 15 minutes at RT in the dark, incubated for 10 minutes in Tris/EDTA buffer at 84°C, and blocked for 1 hour at RT in 1% bovine serum albumin (BSA) in PBS. Primary antibodies, mouse-anti neurofilaments-medium (1:50; AbCam, Cambridge, MA), rat-anti-myelin basic protein (MBP) (1:50; AbDSerotec, Raleigh, NC) for oligodendrocyte and myelination detection, rabbit anti-glial fibrillary acidic protein (GFAP) (1:250; Sigma-Aldrich, St. Louis, MO) for astrocyte detection, and rabbit anti-ionized calcium-binding adaptor molecule (Iba)-1 (1:250; Wako, Osaka, Japan ) for microglial staining were used in 0.5% BSA in PBS overnight at 4°C. For Purkinje cell staining, sections were blocked for 2 hours at RT in 10% normal goat serum and 0.5% Triton X-100 in PBS and incubated overnight with mouse anti–adenomatous polyposis coli (CC-1) (Merck, Darmstadt, Germany) in 2% goat serum and 0.3% Triton X-100 in PBS. After washing in PBS, sections were incubated for 1 hour at RT with antibodies anti-mouse coupled to Alexa Fluor 488 (1:1000; Invitrogen) or anti-goat IgG (H+L) Cy3-conjugated (1:3000 Jackson ImmunoResearch, West Grove, PA), anti-rat coupled to Alexa Fluor 568 (1:1000; Invitrogen) or to Alexa Fluor 488 (1:1000; Invitrogen), and anti-rabbit coupled to Alexa Fluor 568 (1:1000; Invitrogen) in 0.5% BSA in PBS, incubated for 20 minutes in DAPI and mounted with Shandon Immu-Mount Aqueous Non-fluorescing Mounting Medium (ThermoFisher Scientific).
Tissue sections were visualized in an AxioImager Z1 fluorescence microscope equipped with 10×/0.30 Ph1, 20×/0.50 Ph2, and 40×/1.30 Oil Ph3 EC-Plan-Neofluar objectives (Carl Zeiss, Oberkochen, Germany). Images were acquired with a AxioCam MRm version 3.0 camera connected to a PC running the AxioVision 4 acquisition software (Carl Zeiss).
Measuring Myelination and Counting Iba1- and GFAP-Labeled Cells.
To evaluate if kernicterus mice present alterations in brain morphology, axonal viability, and myelination along several brain regions, sections were stained with neurofilament and MBP. To evaluate overall neuronal and myelination alterations, 10× magnification images were acquired and stitched using Microsoft Image Composite Editor. Then binary masks were specified in ImgeJ software using the same cut-off intensity threshold value for each region of interest, defined as the minimum intensity resulting from specific staining above background values. Finally, the percentage of area immunoreactive for neurofilament and MBP was measured. In addition, the percentage of myelinated fibers obtained by the ratio of MBP to neurofilaments staining was also calculated. Moreover, changes in brain morphology were identified using the same software by measurement of the percentage of brain area occupied by cerebellum.
To evaluate more closely the changes in myelination in the different brain areas, the percentage of area immunoreactive for neurofilament and MBP was measured in 40× magnification images, using binary masks as described before, in different brain regions (medulla oblongata, pons, cerebellum, and corpus callosum) and in specific regions of cerebellum (white matter ramifications, middle of white matter tracts, and white matter terminals).
To evaluate microglia and astrocytes, we determined the number of cells with positive staining for Iba-1 and GFAP, respectively, in the same regions in which we evaluated myelination alterations. All results were given by averaging values determined in at least seven separate microscopic fields from three different sections from two control animals and three kernicterus animals. Values are expressed as mean ± S.E.M.
Statistical Analysis.
All results are presented as mean ± S.E.M. Significant differences between two groups were determined by the two-tailed t test performed on the basis of equal and unequal variance as appropriate and the P values of *P < 0.05 and **P < 0.01 were considered statistically significant.
Results
Animal Model Development.
Targeted deletion of the Ugt1a1 gene and the Ugt1 locus in mice was achieved by inserting a construct with the PGK-neo gene flanked by Flp/FRT recombinase sites and Cre/loxP recombinase sites flanking exons 3 and 4 of the Ugt1a1 gene (Fig. 1A). Breeding to homozygosity generates Ugt1LoxP[FRTneoFRT]LoxP or UFP mice. The FRT sites are positioned flanking the PGK-neo selection marker gene allowing for deletion of this DNA by Flp-recombinase, leaving the LoxP sites flanking exons 3 and 4 and generating Ugt1LoxP/LoxP or Ugt1F/F mice (Chen et al., 2013). Thus, the Ugt1F/F (Ugt1LoxP/LoxP) mice are the same as UFP mice (Ugt1LoxP[FRTneoFRT]LoxP) except that the neomycin gene has been deleted. From previous studies, Ugt1F/F mice show normal levels of circulating TSB. However, UFP mice have been characterized as hypomorphic for the Ugt1a1 gene, with all of the neonatal mice displaying hyperbilirubinemia and TSB levels averaging around 2 mg/dl (Fig. 1F). Hypomorphic gene expression is often seen following integration of targeting constructs through homologous recombination (Lewandoski, 2001). In UFP mice, hypomorphic expression of the Ugt1a1 gene is not lethal, with neonatal and adult mice displaying elevated TSB levels. Hypomorphic Ugt1a1 gene expression in liver tissue from UFP mice confirmed a reduction in mature RNA transcripts (Fig. 1B), which corresponded to a reduced UGT1A1 protein expression (Fig. 1C) and bilirubin UGT activity (Fig. 1D). Reduced Ugt1a1 gene expression occurs in all tissues in UFP mice, with an example being displayed in small intestine (Supplemental Fig. 1). Thus, the impact of inserting the targeting construct into the Ugt1 locus leads to delayed Ugt1a1 expression in all tissues.
Targeted deletion of the Ugt1a1 gene and the Ugt1 locus in liver tissue was achieved by crossing UFP mice with albumin-Cre transgenic mice, resulting in UFP/albumin-Cre mice, which are termed UAC mice (Fig. 1A). Expression of Cre-recombinase from the albumin-Cre transgene leads to the deletion of exons 3 and 4 in the liver tissue of the Ugt1a1 gene along with the other Ugt1a genes associated with the Ugt1 locus. Liver tissue from UAC mice at 14 days after birth have no mature UGT1A1 RNA (Fig. 1B) or microsomal UGT1A1 protein as determined by Western blot analysis (Fig. 1C), and no measureable bilirubin UGT activity (Fig. 1D). Newborn UAC mice show high levels of TSB (Fig. 1F) that peak around 14 days after birth. The elevated TSB levels in UAC mice eventually leads to a 95% lethality rate (Fig. 1E). The neonatal UAC mice show visible signs of CNS toxicity prior to death, as evidenced by tremors and seizures.
Alterations in Brain Morphology of Kernicterus Mice.
When brains were isolated from 14-day old UAC mice a diffuse yellow staining (kernicterus) throughout the brain was apparent (Fig. 1G). When compared with UFP mice, which do not accumulate bilirubin in brain tissue, UAC mice had a reduced cerebellar volume, as previously shown in humanized UGT1 mice (Fujiwara et al., 2010) as well as in the Gunn rat model (Conlee and Shapiro, 1997). When saggital brain sections were immunostained (Fig. 2), marked reduction of the cerebellum was observed in UAC mice when compared with the UFP littermates. Hematoxylin and eosin and Cresyl violet (Nissl) staining were used to quantitate cerebellum size in UFP and UAC mice (Supplemental Fig. 2). Interestingly, the global brain area was not affected, possibly owing to an increased dimension of the cortex and midbrain on top of the cerebellum.
Kernicterus Mice Present Axonal Loss and Decreased Myelination.
Brain sections were immunostained for neurofilaments and myelin basic protein (MBP) to calculate the percentage of brain area occupied by these markers. As depicted in Fig. 2A, a decrease in diseased UAC brain area occupied by neurofilament staining (0.80-fold ± 0.08, P < 0.05) was observed when compared with brain tissue from UFP mice. When changes related with myelination were determined there was a more pronounced decrease in MBP positive staining along all brain regions (0.62 ± 0.04, P < 0.01), which correlated with the percentage of myelinated fibers (0.74 ± 0.04, P < 0.05) in UAC brain tissue. These findings indicated that kernicterus mice present a reduced number of viable axons with myelination impairment, which we speculate is derived from a decreased ability of oligodendrocytes to produce myelin.
To examine if the early onset of bilirubin toxicity in the cerebellum had an impact on myelin sheath formation, we examined the expression pattern of MBP at 5-days after birth when TSB levels are already dramatically elevated (Fig. 2B). Immunocytochemistry analysis showed a reduction in myelination in newborn UAC mice when compared with healthy UFP mice. Real-time PCR analysis has confirmed that expression of the Mbp gene in the cerebellum in UAC mice is significantly induced, possibly in response to the inflammatory insult initiated by elevated bilirubin levels. Interestingly, accelerated gene expression is followed by a significant reduction in MBP accumulation as shown by Western blot analysis, which correlates with reduced myelin sheath formation.
Kernicterus Mice Present Reduced Myelination and Increased Glial Reactivity in Cerebellum, Medulla Oblongata, and Pons.
As changes in myelination along different brain regions were observed, brain regions that presented myelination at postnatal day 7 were examined. Changes in myelination in medulla, pons, cerebellum, and corpus callosum were evaluated, the last as a positive control of myelination since it is one of the first myelinated areas of the brain (Sturrock, 1980). As depicted in Figs. 3A and 4A, there was a decrease in the percentage of myelinated fibers in medulla (0.81 ± 0.03, P < 0.01), and a more marked effect in pons (0.58 ± 0.03, P < 0.01) and cerebellum (0.61 ± 0.04, P < 0.01), with no effect on corpus callosum (Supplemental Fig. 3). Brain sections prepared from UFP mice show myelinated fibers that are long and thin, whereas in diseased UAC mice the MBP staining appears as fragmented fibers and myelin agglomerates, suggesting a destruction of the myelin sheath surrounding the axons. Interestingly, the corpus callosum maintains the long myelinated fibers even in the kernicterus mice, corroborating the absence of myelin changes in that area. It is known that astrocytes (Zhang et al., 2006) and microglia (Olah et al., 2012) cooperate to create a favorable environment for myelination, and upon myelination injury they are rapidly activated (Petzold et al., 2002). To examine if gliosis takes place in kernicterus brain, microglia and astrocyte reactivity were identified by Iba-1 and GFAP staining, respectively. As shown in Fig. 3, the kernicterus mice presented a marked increase in both microglia and astrocytes in cerebellum, medulla, and pons with no effect in the corpus callosum (Supplemental Fig. 3). Regarding microglia (Figs. 3B and 4B), the major increase was observed in cerebellum (3.75-fold ± 0.04, P < 0.01), followed by pons (2.22-fold ± 0.07, P < 0.01) and medulla (1.68-fold ± 0.21, P < 0.05). Concerning astrocytes, very similar increases were observed in medulla and cerebellum (3.73-fold ± 0.17, P < 0.05 and 3.95-fold ± 0.08, P < 0.01, respectively), but the most pronounced effects were noticed in pons (29.08-fold ± 3.93, P < 0.01). Regarding the corpus callosum, a region with a considerable number of glial cells in the control animals, showed no noticeable changes in the kernicterus mice.
Distinct White Matter Regions from Cerebellum Present Different Alterations in Myelination and Glial Reactivity.
Since cerebellum is one of the most affected areas and it seems to present regional changes in myelination and glial reactivity, the cerebellum was divided into three distinct zones as depicted in Fig. 5A; the white matter nodes (region 1), the middle of the white matter tracts (region 2), and white matter terminals (region 3). When myelination was evaluated (Figs. 5B and 6B) there was a marked decrease in the percentage of myelinated fibers in the kernicterus mice. The major effects were observed at white matter terminals (0.45-fold ± 0.07, P < 0.01), followed by central white matter tracts (0.58-fold ± 0.04, P < 0.01) , and finally a less pronounced but significant alteration was observed in white matter nodes (0.72-fold ± 0.05, P < 0.05).
When the number of glial cells in these regions was evaluated there was a marked increase in both astrocyte and microglia number in cerebellum white matter of kernicterus mice (Figs. 5, C and D, and 6, C and D). In this context, there was an increased number of microglia in all three regions with a pattern similar to what was observed for myelination. Indeed, the most affected area was the white matter terminals (5.37-fold ± 0.18, P < 0.01), followed by central white matter tracts (5.05-fold ± 0.27, P < 0.01) and white matter nodes (3.11 ± 0.17, P < 0.01). On the other hand, when astrocyte number was evaluated, kernicterus mice presented a very similar increase in both white matter nodes and terminals (3.36-fold ± 0.18, P < 0.01, and 3.31-fold ± 0.25, P < 0.01, respectively), with the most affected area being the central white matter (4.28-fold ± 0.15, P < 0.01).
It deserves to be noted that although white matter terminals were the area most affected with respect to changes in myelination and microgliosis, the changes in astrogliosis were mainly observed in the middle of the white matter tracts, possibly suggesting a different role for microglia and astrocytes in response to UCB injury.
Kernicterus Mice Present Cerebellum Atrophy and a Great Reduction in Purkinje Cell Number.
Besides changes observed in myelination and glial reactivity, the cerebellum volume is greatly reduced in kernicterus animals, a feature also observed in Gunn rats traumatized by sulfadimethoxine administration to increase TSB levels (Conlee and Shapiro, 1997). As depicted in Figs. 2 and 7A, UAC mice present a marked derangement of cerebellar lobules resulting in a greater reduction in the area occupied by cerebellum when compared with total brain area (0.45-fold ± 0.03, P < 0.01). This cerebellum atrophy in kernicterus mice can result from a massive reduction in the number of Purkinje cells (0.21-fold ± 0.03, P < 0.01), mainly in the anterior lobes, and shrinkage of molecular and granular layers, as shown in Fig. 7A. Purkinje cells that remained in the kernicterus brains (Fig. 7A) presented an altered morphology with reduced number of branches, suggesting that cerebellar neuronal circuitry may be markedly affected.
Discussion
These findings confirm that deletion of the Ugt1a1 gene and the Ugt1 locus in liver tissue from UAC mice presents an animal model of severe hyperbilirubinemia that develops into BIND and kernicterus, leading to marked cerebellar hypoplasia in parallel with a reduction of myelination and an increase in astrogliosis and microgliosis in the cerebellum, pons, and medulla oblongata. The inclusion of the targeting construct into the Ugt1a1 gene led to a hypomorphic allele, resulting in altered Ugt1a1 gene expression in all tissues and a condition of mild hyperbilirubinemia. Selectively targeting the deletion of the Ugt1a1 gene in liver tissue exacerbates this condition, driving TSB levels to toxic concentrations that induce gliosis. Unlike in hUGT1 mice, the percentage of seizures and lethality observed in neonates was 5–10% of the neonates, and over 95% of the neonatal mice developed acute signs of brain damage when the liver Ugt1a1 gene was targeted, as observed in UAC mice. It should be noted that Ugt1F/F mice do not develop neonatal hyperbilirubinemia (Chen et al., 2013), and deletion of the Ugt1a1 gene in liver (Ugt1ΔHep) produces minimal hyperbilirubinemia. The dramatic difference in TSB levels between Ugt1ΔHep and UAC mice is attributable to reduced UGT1A1 expression in extrahepatic tissues in the UAC mice.
It has been demonstrated that UCB impairs oligodendrocyte differentiation and consequent myelination in vitro using dorsal-root ganglia-oligodendrocyte cocultures (Barateiro et al., 2013) and ex vivo in organotypic cerebellar slice cultures (Barateiro et al., 2012). These observations are consistent with previous reports showing a decrease in the density of myelinated fibers and a loss of axons in the cerebellum of a premature infant with kernicterus (Brito et al., 2012), along with white matter volume reduction and delayed hemispheric myelination as observed in infants with severe UCB encephalopathy (Gkoltsiou et al., 2008). In UAC mice, there was a marked impairment of myelination. Indeed, fragmentation of myelinated fibers within the cerebellum, medulla oblongata, and pons was detected, a finding which indicated that toxic levels of bilirubin led to damage of the myelin sheath. Myelination deficits have also been reported in several other perinatal conditions, including moderate perinatal systemic inflammation (Favrais et al., 2011), perinatal hypoxic-ischemia (Huang et al., 2009), and white matter injury in the premature baby (Buser et al., 2012). Since it is known that neuronal-oligodendrocyte crosstalk is crucial for proper myelination (Lee and Fields, 2009), myelination impairment observed in the kernicterus mice may be attributable in part to a reduced number of viable axons limiting oligodendrocyte differentiation and myelination. Nevertheless, the percentage of myelination in the remaining fibers was also affected in these animals, corroborating the toxic role of bilirubin on oligodendrocyte maturation, as previously seen in several in vitro models. Actually, as reported for white matter injury (Buser et al., 2012), myelination failure may arise from the inability of oligodendrocyte precursors to generate myelinating oligodendrocytes to repair the injury.
It deserves to be noted that the increased astrogliosis and microgliosis in UAC mice overlie those brain areas that present myelin damage. This fact is more pronounced in the cerebellum (Fig. 4), where recruitment of both microglia and astrocytes in the surrounding white matter is induced. It is known that upon myelination damage both astrocytes and microglia are rapidly activated and migrate to the site of injury (Petzold et al., 2002) and astrocyte presence is sustained along with myelination (Petzold et al., 2002; Miron et al., 2010), even during the perinatal period (Huang et al., 2009; Buser et al., 2012). Upon demyelination, microglia are responsible for clearing myelin debris to allow for proper remyelination (Olah et al., 2012). In addition, astrocytes and microglia produce growth factors that can facilitate oligodendrocyte survival, differentiation, and the ability to myelinate (Stankoff et al., 2002; Zhang et al., 2006; Olah et al., 2012), creating a favorable environment for repair. Indeed, we observed an acute increase in the number of microglia in the areas of greatest myelin deficit, suggesting that these cells have been recruited to clear the damaged myelin. In the cerebellum, where myelin sheath formation is inhibited, Mbp gene expression was significantly induced, possibly in response to the inflammatory insult in this brain region and the production of reactive oxygen species, or even as a compensatory response to the lack of functional MBP, which had been degraded as a result of intense gliosis.
It can also be argued that if activated, microglia and astrocytes may release high amounts of proinflammatory mediators that are known toxicants for oligodendrocytes and their progenitors. This was observed in hUGT1 mice and hUGT1/Trl2−/− mice that seizure, and show distinct glial cell activation (Yueh et al., 2014). It has been shown that activated microglia induced Purkinje neuronal death through TNF-α and IL-1β signaling (Kaur et al., 2013). Thus, myelination impairment and loss of Purkinje cells in the UAC mice may result from oligodendrocyte damage by glial-derived inflammatory factors. In addition, it has been reported that following white matter injury in the premature, astrogliosis extended into the site of injury, causing inhibition of oligodendrocyte differentiation and consequent myelination as a result of hyaluronic acid production (Buser et al., 2012). Interestingly, our results show increased astrogliosis in cerebellum central white matter tracts, which are regions that do not present the highest levels of myelination defects but are the most affected nerve fibers at white matter terminals. Thus, in the kernicterus model astrogliosis may be impairing remyelination following myelin damage.
Cerebellar hypoplasia was the more marked feature observed in the kernicterus mice, as also demonstrated with a similar Ugt1−/− mouse model (Bortolussi et al., 2015). In a single case of a preterm infant who was diagnosed with kernicterus, histologic analysis revealed a loss of myelin fibers in the cerebellum (Brito et al., 2012), indicating similarity with kernicterus in humans and our mouse model. Indeed, these findings revealed that increased circulating levels of TSB derange cerebellar lobules, with a reduction in the thickness of the molecular and granular layers, and a marked loss of Purkinje cells that coincides with reduced branching of the remaining cells. Similar features were noted in Gunn rats (O’Callaghan and Miller, 1985; Conlee and Shapiro, 1997). The rodent cerebellar lobules mature postnatally, increasing over 20-fold from birth to 21 postnatal days. Interestingly, the cells within the ventral lobes, namely Purkinje cells, mature first, followed by those in the anterior lobes (Altman, 1969), and may be susceptible to toxic levels of bilirubin. In UAC mice there is an increased loss of Purkinje cells in the anterior lobe. This finding suggests that elevated bilirubin levels early in the neonatal stages drives toxicity toward those regions of the CNS that are underdeveloped. This fact may be corroborated by earlier findings showing that immature neurons and astrocytes are more susceptible to UCB toxicity and cell death than mature ones (Falcao et al., 2006).
The cerebellum is a region of the brain that plays an important role in motor control, and contributes toward coordination, precision, and timing of movements (Fine et al., 2002). Therefore, damage to the cerebellum, as observed in UAC mice, may justify the abnormal motor control, movements, and muscle tone observed in these mice, as well as in hUGT1 mice (Fujiwara et al., 2010; Vogel et al., 2011). Similar motor deficits were observed clinically in severe neonatal hyperbilirubinemia (Shapiro, 2010). Interestingly, in UAC mice we also show alterations in several brain regions that are commonly affected in human neonatal brains that develop kernicterus (Bhutani and Stevenson, 2011). Indeed, we have observed myelin deficits and gliosis in medulla oblongata and pons of kernicterus mice. Both medulla oblongata and pons are structures located on the brainstem, with the medulla being on the lower half of the brainstem, and the pons above the medulla yet below the midbrain and ventral to the cerebellum. The medulla oblongata connects the higher levels of the brain to the spinal cord and has an important role in the control of nervous system autonomic functions, including the reflex center of swallowing (Jean, 1984). Interestingly, a history of past or present suckling and swallowing dysfunction is often described in children with kernicterus (Shapiro, 2010), which may occur as a result of altered function of the medulla nerve cells as observed in our animal model.
Overall, our study reveals a new animal model of kernicterus showing anatomic and histologic changes that may justify the clinical symptoms often detected in infants subjected to severe neonatal hyperbilirubinemia. Further studies will need to be conducted to examine the similarities and differences that occur in kernicterus syndrome from published reports in humans and those observed in our mouse model. However, we have identified deficits in myelination and enhanced gliosis in the kernicterus mice that are similar to findings in human brain sections, indicating that mechanisms responsible for the neurologic dysfunction described in moderate to severe hyperbilirubinemia may be viewed as potential targets for new therapeutic strategies.
Acknowledgments
Shujuan Chen and Andreia Barateiro contributed equally to this work. The authors would like to thank Glenn Handoko for excellent technical assistance.
Author Contributions
Participated in research design: Chen, Fernandes, Barateiro, Yueh, Nguyen, Tukey, Brites. Conducted experiments: Chen, Fernandes, Barateiro, Domingues, Barbier, Nguyen.
Contributed new reagents or analytic tools: Relvas, Barbier, Brites, Tukey.
Performed data analysis: Chen, Fernandes, Barateiro, Domingues, Relvas, Barbier, Tukey, Brites.
Wrote or contributed to the writing of the manuscript: Fernandes, Chen, Yueh, Barateiro, Tukey, Brites.
Footnotes
- Received February 5, 2015.
- Accepted October 14, 2015.
This work was supported by the National Institutes of Health National Institute of Environmental Health Sciences [ES010337], National Institute of General Medicine [GM100481, GM086713], and the National Cancer Institute [R21CA171008]. This work was also supported in part by the projects [PTDC/SAU-NEU/64385/2006] and iMed.ULisboa [UID/DTP/04138/2013], from Fundação para a Ciência e a Tecnologia (FCT). A.B. was a recipient of a PhD fellowship [SFRH/BD/43885/2008] from FCT. The funding organizations had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
↵This article has supplemental material available at molpharm.aspetjournals.org.
Abbreviations
- BIND
- bilirubin-induced neurologic dysfunction
- BSA
- bovine serum albumin
- CNS
- central nervous system
- DAPI
- 2-(4-amidinophenyl)-1H-indole-6-carboxamidine
- GFAP
- glial fibrillary acidic protein
- Iba
- ionized calcium-binding adaptor molecule
- IL
- interleukin
- MBP
- myelin basic protein
- PBS
- phosphate buffer saline
- RT
- room temperature
- TLR2
- Toll-like receptor 2
- TNF
- tumor necrosis factor
- TSB
- total serum bilirubin
- UFP
- mice homozygous for the Ugt1a1LoxP[FRTneoFRT]LoxP allele
- UAC
- UFP/albumin-Cre mice
- UCB
- unconjugated bilirubin
- UGT
- UDP-glucuronosyltransferase
- Copyright © 2015 by The American Society for Pharmacology and Experimental Therapeutics