|Home | About | Journals | Submit | Contact Us | Français|
Trophic factor signaling is important for the migration, differentiation and survival of enteric neurons during development. The mechanisms that regulate the maturation of enteric neurons in postnatal life, however, are poorly understood. Here, we show that transcriptional cofactor HIPK2 (homeodomain interacting protein kinase 2) is required for the maturation of enteric neurons and for regulating gliogenesis during postnatal development. Mice lacking HIPK2 display a spectrum of gastrointestinal (GI) phenotypes, including distention of colon and slowed GI transit time. Although loss of HIPK2 does not affect enteric neuron in prenatal development, a progressive loss of enteric neurons occurs during postnatal life in Hipk2−/− -mutant mice that preferentially affects the dopaminergic population of neurons in the caudal region of the intestine. The mechanism by which HIPK2 regulates postnatal enteric neuron development appears to involve the response of enteric neurons to bone morphogenetic proteins (BMPs). Specifically, compared to wild type mice, a larger proportion of enteric neurons in Hipk2−/−mutants have abnormally high level of phosphorylated Smad1/5/8. Consistent with the ability of BMP signaling to promote gliogenesis, Hipk2−/− mutants show a significant increase in glia in the ENS. In addition, numbers of autophagosomes are increased in enteric neurons in Hipk2−/−mutants and synaptic maturation is arrested. These results reveal a new role for HIPK2 as an important transcriptional cofactor that regulates the BMP signaling pathway in the maintenance of enteric neurons and glia, and further suggest that HIPK2 and its associated signaling mechanisms may be therapeutically altered to promote postnatal neuronal maturation.
Enteric neurons are derived from precursors that originate in the neural crest, migrate to the gut and undergo a long process of development and maturation (Burns and Pachnis, 2009; Gershon, 2010). Although a great deal is known about the early events in enteric neuron development, much less is known about later-acting mechanisms for the acquisition of enteric neuronal phenotype diversity, maturation, and survival (Enomoto et al., 1998; Flynn et al., 2007; Uesaka et al., 2007; Uesaka et al., 2008; Burns and Pachnis, 2009; Uesaka and Enomoto, 2010). Several studies indicate that bone morphogenetic proteins (BMPs), members of the transforming growth factor β (TGFβ) superfamily, regulate the development of enteric neurons (Chalazonitis et al., 2004b; Fu et al., 2006; Chalazonitis et al., 2008). Interestingly, the effects of BMPs on enteric neurons are context-dependent; outcome is dictated by the timing of BMP signals and the interaction between BMPs and their antagonist Noggin. BMPs, for example, promote the development of the neurotrophin-3 (NT-3)-dependent, TrkC+ enteric dopaminergic neurons. In contrast, transgenic mice expressing Noggin under the control of the neuron specific enolase (NSE) promoter have reduced numbers of enteric dopaminergic neurons (Chalazonitis et al., 2008). These results support the idea that BMP signaling is essential for subtype specification in the enteric nervous system (ENS) and provide important insights into the control of maturation, survival and long-term maintenance of enteric dopaminergic neurons (Li et al., 2004; Li et al., 2006). Moreover, increased BMP signaling promotes glial differentiation in cultured enteric crest-derived precursors and in NSE-BMP4 transgenic mice (Chalazonitis et al., 2011).
The signal transduction pathways of TGFβ and BMP involve the Smad proteins, which are highly evolutionarily conserved transcription factors that are activated upon the engagement of TGFβ or BMPs with their receptors (Derynck and Zhang, 2003; Massague et al., 2005). Following their phosphorylation, Smads are translocated to the nucleus where they regulate the expression of TGFβ (Smad2/3) or BMP (Smad1/5/8) target genes. In addition, Smads can interact with a number of transcriptional co-activators or co-repressors to activate or silence the transcription of target genes, thus adding layers of complexity to the final outcomes of TGFβ or BMP signals. HIPK2 can negatively regulate BMP-dependent reporter gene activity in cell lines and interact with Smad1, Smad2, Smad3 and c-Ski (Harada et al., 2003). HIPK2 can also regulate survival and apoptosis in sensory, sympathetic and midbrain DA neurons (Doxakis et al., 2004; Wiggins et al., 2004; Zhang et al., 2007). The present study focuses on the role of HIPK2 in the maturation of enteric neurons and, in particular, the dopaminergic subset. These neurons were chosen because they play an important role in regulating intestinal peristalsis (Li et al., 2004; Li et al., 2006) and may be affected in Parkinson’s disease (Kuo et al., 2010). Our data indicate that enteric neurons require HIPK2 in postnatal life for the formation of autophagosomes and for synaptogenesis. These results underscore the important function of HIPK2 in regulating BMP signaling in the maturation of enteric neurons, especially those of the dopaminergic subtype.
The Hipk2+/− mutants were maintained on a mixed 129;C57BL6 background. The procedures used to generate and genotype Hipk2−/− mutants have been described previously (Wiggins et al., 2004; Zhang et al., 2007). Animal care was approved by the Institutional Animal Care and Use Committee at UCSF and Columbia University and followed NIH guidelines.
Both wild type and Hipk2−/− mutant mice of either sex ranging in age from 10–24 weeks were perfused with 4% paraformaldehyde (PFA). The submucosal and myenteric plexuses were dissected separately. Laminar preparations of the gut wall were prepared as whole mounts to include either the submucosa with the submucosal plexus or the myenteric plexus attached to the longitudinal muscle. Neurons were identified with the neuronal ribosomal marker cuprolinic blue as previously described (Heinicke et al., 1987; Karaosmanoglu et al., 1996; Chalazonitis et al., 2004; Chalazonitis et al., 2008). Regions of the gut examined included stomach, duodenum, jejunum, proximal and distal ileum and colon.
Acetylcholinesterase activity was demonstrated histochemically according to procedures described previously (Enomoto et al., 1998). Briefly, the gut was dissected as a single piece from the proximal esophagus to the distal colon; mesenteric attachments and pancreas were removed, and the tissue was post-fixed with 4% PFA for 1–2 hr at 4°C. Tissues were then transferred to saturated sodium sulfate and stored overnight at 4°C. Preparations were then incubated in buffer (0.2 mM ethopropazine HCl, 4 mM acetylthiocholine iodide, 10 mM glycine, 2 mM cupric sulfate, and 65 mM sodium acetate [pH 5.5] for 2–4 hr. Staining for acetylcholinesterase was developed by incubating in sodium sulfide for 1.5 min (1.25%, pH 6). Tissue was rinsed extensively with water before photographing under a dissecting microscope. Photomicrographs were captured using an Olympus BX51 microscope connected to a DP70 CCD camera and DP Manager 3.1.1 software. For pictures, the gut was opened along the mesenteric border, flattening the tissue with the serosal side up, and mounted in 50% glycerol. Images obtained in the Gershon laboratory were captured with a Leica DML6000B microscope connected to a Retiga Exi CCD camera using Improvision™ Volocity 5.4.1 software.
HIPK2 protein and HIPK2LacZ were detected using procedures described previously (Wiggins et al., 2004; Zhang et al., 2007). To identify DA neurons in the ENS, segments of duodenum, jejunum, ileum and colon were dissected to expose the myenteric or submucosal plexuses in laminar preparations. Prior to labeling with antibodies to dopamine, the tissue was exposed for 24 hours to a fixative modified to preserve tissue DA that included 4% PFA in 0.1 M phosphate buffer saline, 9% sucrose, and 0.025% L-ascorbic acid (pH 7.0). Following a 5 hr rinse in 0.1M phosphate buffer, the tissue was incubated with rat antibodies to DA (1:1,000, NT-104, Protos Biotech Corporation, New York, NY) for an additional 24 hours at 4°C. Following rinses in PBS, laminar preparations were incubated for 4 hrs with donkey secondary antibodies to rat IgG conjugated to Alexa 594 (1:750, Invitrogen, Eugene, OR), rinsed and mounted with Vectashield (Vector Laboratories, Burlingame, CA). Because DA neurons are relatively rare in the myenteric plexus but more abundant in the submucosal plexus (Li et al., 2004; Chalazonitis et al., 2008), quantitative analyses were limited to the laminar preparations of submucosal plexus. Double immunolabeling of dopamine (DA) and tyrosine hydroxylase (TH) was carried out to verify the coincident location of these two antigens in the same neurons.
The primary antibodies in this study included the following: a polyclonal sheep anti-tyrosine hydroxylase antibody (1:200, P60101, Pel-Freeze), anti-BrdU antibody (1:200, MAB3228, Millipore), anti-NF150 antibody (1:400, AB1981, Millipore), anti-cleaved caspase 3 antibody (1:1000, AF835, R&D Systems), anti-HuC/D antibody (1:20, A-21271, Molecular Probes), anti-LC3 antibody (1:50, M152-3, Medical and Biological Laboratories), anti-PGP9.5 antibody (1:20, ab8189, Abcam), anti-SNAP25 antibody (1:20, ab41455, Abcam), anti-HIPK2 antibody (1:500, Cat #ARP32586-P050, Aviva, San Diego, CA), anti-GFRα1 (1:1000, AB5963, Chemicon), anti-PSD95 antibody (1:100, Cat #2507, Cell Signaling Technology), anti-p75NTR antibody (1:2500, G3231, Promega), anti-synapsin antibody (1:20, Cat #2312, Cell Signaling Technology), TuJ1 anti-class III β-tubulin antibody (1:1000, MMS-435P, Covance), anti-pSmad1/5/8 antibody (1:625, Cat #9511, Cell Signaling Technology), anti-pSmad2 antibody (1:200, Cat #3010, Cell Signaling Technology), anti-TrkC antibody (gift from Dr. Moses Chao, Skirball Institute, New York University) and anti-S100β antibody (1:3000, Cat# Z0311, DAKO, Glostrup, Denmark). The specificity of the anti-HIPK2 antibody was confirmed by the lack of staining dorsal root ganglion neurons in Hipk2−/− mutant mice (data not shown). Secondary antibodies used included donkey anti sheep IgG coupled to Alexa 488 (1:500; Invitrogen), anti-rabbit IgG-conjugated with HRP (Vector Laboratories), and goat anti rabbit antibody coupled to Alexa 594 (1:625; Invitrogen).
Following detection of enteric neurons with TuJ1 antibodies, tissue sections were incubated in labeling buffer for 3′ at room temperature, followed by the addition of TdT reaction mix (4μl dNTP mix, 4μl 50X Mn++ stock solution, 4μl of TdT enzyme, and 50 μl of TdT Labeling Buffer)(R&D Systems) in a humidified chamber at 37°C for 1.5 hours. The reaction was terminated by incubating the slides with 1x Stop Buffer for 5 min and the slides were rinsed twice in PBS for 2′. To visualize the biotinylated nucleotides, which have been transferred by the TdT enzyme onto the nicked ends of single strand DNA, samples were incubated in Strep-Fluor solution at room temperature for 1 hour. Each slide was then drained and mounted with 50–75 μl of anti-Fade mounting media.
Mice were fixed by perfusion at P0, P7, P16 and P36 with a freshly prepared mixture of 4% formaldehyde (from paraformaldehyde) and 0.2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4). Segments of stomach, duodenum, ileum and colon were dissected and post-fixed in 4% formaldehyde (from paraformaldehyde) and 2.5% glutaraldehyde in the same buffer over night at 4°C. Gut segments were treated with 2% osmium tetroxide for 1 hr at room temperature, rinsed 5 times in PBS and maleate buffer, and stained en bloc with 5% aqueous uranyl acetate on ice for 1 hr. The blocs of tissue were dehydrated with a graded series of ethanol solutions, cleared with propylene oxide and embedded in polymerized EMbed-812 (Ted Pella). Ultra thin sections (1 micron) were cut with a microtome (Reichert-Jung) and collected on copper grids. Sections were examined with a JEOL 1200EX electron microscope.
Gut motility was measured using a protocol modified from Carai and colleagues (Carai et al., 2006). Six wild type and five Hipk2−/− mice of either sex, 8–12 weeks of age, were used for this study. Mice were housed individually during the course of the study and were not deprived of food before measurements of total gastrointestinal transit time. Xylenecyanol FF and bromophenolblue (0.25% w/v, each) were suspended in water containing 0.5% methylcellulose, and administered intragastrically with a feeding needle (25 gauge, 1″ long, 1.25mm ball diameter) at a dose of 0.3 ml per mouse. Beginning immediately after xylenecyanol FF and bromophenolblue administration, fecal pellets were collected in pre-weighed tubes to determine wet weights every 30 min for up to six hours. To determine total gastrointestinal transit time (GITT), mice were monitored 8 hours after xylenecyanol FF and bromophenoblue administration at 30-minute intervals. The time required for the blue bolus to disappear and the first brown bolus to appear was recorded as GITT. Fecal pellets were vacuum dried overnight to determine dry weights. Water content in fecal pellets was determined as the difference between the wet and dry weight of stool.
Neuronal and glial cell bodies were counted using a 40X objective in 10 contiguous non-overlapping rectangular fields covering 1.254 mm2 of plexus and counted as one measurement. The investigator counting neurons was blinded with regard to the genotype of the animals. The numbers of measurements obtained from each region of the gut and in each plexus using cuprolinic blue as a neuronal marker are listed in Table 1. The incidence of autophagy was estimated as a percentage of identified autophagosome profiles, normalized to 10−3μm2 unit area of plexus. Sections from Hipk2−/− and wild type mice were examined similarly and compared. The number of synapses in the myenteric plexus was determined by enumerating structures that contain well-delineated presynaptic buttons, synaptic vesicles, synaptic clefts and postsynaptic densities. The synaptic density was calculated based on the number of synapses normalized to 10−3 mm2 unit area. Means ± S.E.M. were calculated and significance was analyzed by Student’s t-test (between 2 means of values) or analysis of variance (ANOVA; Tukey or Bonferroni-Dunn post hoc comparison) with more than 2 means of values.
Homozygous Hipk2 (Hipk2−/−) mutants were born in a normal Mendelian ratio and displayed no gross abnormalities; nevertheless, ~40% of Hipk2−/− mutants died before weaning, most within 2–3 days after birth (Fig. 1A). In contrast, heterozygous Hipk2 (Hipk2+/−) mice survived into adulthood just as their wild type littermates did. Of those Hipk2−/− mutants that died within 3 days after birth, almost all lacked milk in the stomach (Fig. 1B, upper panel). At necropsy, the small intestines were distended and contained large amounts of air (Fig. 1B, lower panel). The remaining 60% of Hipk2−/− mutants survived into adulthood but gained weight poorly and remained abnormally small (Fig. 1C). By 12 weeks of age, the body weight of Hipk2−/− mutants was only 70% of that of wild type (24.2 gm in wild type vs. 17.2 gm in Hipk2−/− mutants)(Fig. 1C). In contrast, the body length of Hipk2−/− mutants, measured by the distance from nose to anus, did not differ significantly from that of wild type mice (data not shown). Importantly, there was no evidence that the gastrointestinal phenotypes were caused by congenital malformations, such as tracheo-esophageal fistula.
The distended bowel in Hipk2−/− mutants at perinatal stages and poor body weight gain in postnatal life suggested that GI motility might be abnormal. To test this hypothesis, the gastrointestinal transit time (GITT) was determined in wild type and Hipk2−/− mutants. Results indicated that GITT was significantly longer in Hipk2−/− mutants than in wild type mice (15.3±1.3 hours vs. 11.3±0.9 hours, n=6, p < 0.05)(Fig. 1D). In addition, under normal feeding conditions, the 24 hr stool quantity and the stool pellet size of Hipk2−/− mutants were much smaller than those of wild type mice, regardless of stool water (Fig. 1E-F). The water content of the stools of Hipk2−/− mutants, furthermore, was only 50% of that of wild type mice (Fig. 1G). Taken together, these results suggest that the gastrointestinal motility is abnormal in Hipk2−/− mutants, which accounts for the distended bowel of neonates and the constipation of adult mice.
To investigate the mechanism of the abnormal GI transit of Hipk2−/− mutants, we characterized the expression of HIPK2 in the ENS. Using Hipk2LacZ as a reporter, we detected scattered HIPK2 expression in neurons along the wall of small and large intestines as early as E15.5 (data not shown). The HIPK2LacZ expression persisted in the enteric neurons through postnatal day 0 (P0)(Fig. 2A–C and 2G–I). Consistent with these results, the endogenous HIPK2 protein was detected using a HIPK2-specific antibody in speckles within the nuclei of enteric neurons (Fig. 2D–F). In adult mice, HIPK2LacZ showed extensive co-localization with clusters of neurons within the myenteric plexus in both small and large intestines (Fig. 2J-N).
Several approaches were used to determine whether the loss of HIPK2 affected the number of enteric neurons. First, we characterized the development of enteric neurons during prenatal life. We used p75NTR as a marker for crest-derived cells, BrdU, and various neuronal markers, including β3-tubulin, neurofilament (NF150), PGP9.5 and HuC/D to quantify precursors and neurons in the developing gut. The numbers of crest-derived cells and neurons in the developing ENS of wild type mice were not significantly different from those of Hipk2−/− mutants at E15.5 or E18.5 (data not shown). Second, we used AChE activity in whole mounts to visualize the ENS in the postnatal intestine. Since more than 80% of enteric neurons are cholinergic (Sang and Young, 1998; Furness, 2006), and the overwhelming majority of enteric neurons and neurites express AChE, the AChE whole mount staining provided a convenient tool to evaluate the overall organization of neuronal cell body and nerve fiber within the ENS (Blaugrund et al., 1996; Wang et al., 2010). In wild type intestine at P0, ganglia were well-developed (Fig. 3A-C, arrowheads), and neural connectives were intensely stained with AChE reaction product (Fig. 3A-C, arrows). In contrast, the size of ganglia of the Hipk2−/− mutant ENS was smaller than that in wild type mice (Fig. 3D-F, arrowheads), and the density of neural connectives as well as the intensity of AChE staining was reduced (Fig. 3D-F, arrows). A similar pattern of AChE activity was also detected in adult intestines; again, the size of enteric ganglia and density of neural connectives were noticeably reduced in Hipk2−/− mutants (Fig. 3J–L), compared with wild type controls (Fig. 3G–I).
The diminished AChE activity in the gut of Hipk2−/− mutants suggested that the number of enteric neurons might be reduced. To test this, we analyzed the packing density of enteric neurons using cuprolinic blue as a neuronal marker. The numbers of neurons in both the myenteric and submucosal plexuses in Hipk2−/− mutants at 12 weeks of age were significantly lower than those in wild type mice (Fig. 4A-D). Interestingly, quantification of neuronal packing density (number of neurons per unit area) in the myenteric plexus of small and large intestines of wild type and Hipk2−/− mutants showed no difference at P1. There was, however, a progressive reduction of neurons in the myenteric plexus of Hipk2−/− mutant at P14 and P85 (Fig. 4E-F). By P85, the myenteric plexus of Hipk2−/− mutants showed a 20–33% reduction of neuronal density compared to that in wild type mice, depending upon the region of bowel analyzed (Fig. 4F). In addition, the packing density of the submucosal neurons of P85 Hipk2−/− mutants also was similarly lower than that of wild type mice (Fig. 4G).
The ENS contains dopaminergic neurons that critically regulate intestinal motility (Li et al., 2004; Li et al., 2006); moreover, the development of the dopaminergic neurons of the ENS is regulated by members of the TGFβ superfamily, such as BMP-2, BMP-4, and their antagonists such as Noggin (Chalazonitis et al., 2008). Since our previous results had demonstrated that Hipk2−/− mutants lost a significant number of dopaminergic neurons in the ventral midbrain (Zhang et al., 2007), we postulated that the loss of HIPK2 might also negatively affect the development of dopaminergic neurons in the ENS. To test this hypothesis, we quantified dopaminergic neurons in the ENS using antibodies to dopamine and TH as markers. Duodenum, jejunum, ileum and colon were dissected from wild type and Hipk2−/− littermates of either sex that were 12–24 weeks of age (n=5 for wild type and n=6 for Hipk2−/− mutants). The packing density of dopaminergic neurons, determined from observations of dopamine immunoreactivity alone or coincident dopamine and TH immunoreactivities, was quantified per mm2 of submucosal plexus. Consistent with previous studies (Li et al., 2004), we found that dopaminergic neurons were most prevalent in the submucosal plexus of wild type mice, where dopamine and TH immunoreactivities were detected in the cytoplasm of the same neurons (Fig. 4H-J). In Hipk2−/−mutants, however, the absolute number of DA neurons was consistently reduced in the submucosal plexus of duodenum, jejunum, ileum and colon (Fig. 4K-M). To determine whether the decrease in packing density of dopaminergic neurons reflected the decrease in the overall neuronal packing density observed in the 4 regions of the gut examined, we normalized the numbers of dopaminergic neuron to the numbers of total neurons in wild type and Hipk2−/−mutants. Whereas the normalized percentage of dopaminergic neurons was not significantly reduced in the duodenum in Hipk2−/− mutants, it was decreased to 72% of the wild type level in jejunum, to 53% of the wild type level in ileum, and to 45% of the wild type level in colon (Fig. 4O-P). These data suggest that, except for the duodenum, decreases in enteric dopaminergic neurons were much more severe in Hipk2−/− mutants relative to the total neuron numbers and thus did not simply reflect the decrease in the overall number of neurons observed in all regions of the bowel. The decrease in dopaminergic neurons also appeared to be more severe toward the caudal regions of the gut (Fig. 4O-P).
We have previously reported that dopaminergic neurons constitute a subset of the TrkC- expressing (TrkC+) neurons of the gut (Chalazonitis et al., 2008) and that BMP-2 and -4 promote the development of TrkC+ neurons in cultures of enteric crest-derived precursors (Chalazonitis et al., 2004). Consistent with this notion, NT-3 and BMP-4 promote differentiation of TH-expressing neurons; moreover, the number of dopaminergic neurons is increased in transgenic mice that over-express BMP-4 (NSE-BMP-4 mice)(Chalazonitis et al., 2004; Chalazonitis et al., 2008). Conversely, the number of TrkC+ neurons and DA neurons are reduced in the ENS of transgenic mice that overexpress the BMP antagonist Noggin (NSE-Nog mice)(Chalazonitis et al., 2004; Chalazonitis et al., 2008). Because there is a partial loss of ventral midbrain and enteric DA neurons in Hipk2−/− mutants (Fig. 4)(Zhang et al., 2007), we reasoned that the differentiation and maintenance of the TrkC+ neurons might also be affected in the Hipk2−/− mutant ENS. To test this hypothesis, we quantified TrkC+ neurons in the myenteric and submucosal plexuses using whole mount preparations of duodenum, jejunum and ileum of 12 weeks old wild type and Hipk2−/− mutant mice (Fig. 5). Our results indicated that the packing density of TrkC+ neurons per mm2 of plexus in the Hipk2−/− mutants was significantly less than that in wild type mice in the myenteric plexus of jejunum (59±4 [n=21], compared to 74±3 in wild type [n=22], p < 0.005) and ileum (44±3 [n=12], compared to 78±5 in wild type [n=18], p < 0.0001). The decrease in the TrkC+ neuron packing density was also detected in the submucosal plexus of Hipk2−/− mice in the ileum (21±1 [n=21], compared to 37±2 in wild type [n=26], p < 0.0001), but not in the duodenum-jejunum (44±2 [n=36], compared to wild type 46±2 [n=29])(Fig. 5A-B). Consistent with the reduction in TrkC+ neuron packing density, TrkC+ neurons (expressed as % of wild type) in the myenteric plexus was reduced in duodenum-jejunum and even more in the ileum, while in the submucosal plexus, the decrease was significant only in the ileum, (Fig. 5C). These results were consistent with the proximo-distal gradient in the severity of the reduction of enteric dopaminergic neurons (Fig. 4), and support the notion that, similar to its essential role for dopamineergic neurons, HIPK2 is required for the maintenance of TrkC+ neurons in the ENS.
We have previously demonstrated that BMP-2 and BMP-4 regulate the development of enteric neurons, including that of the dopaminergic subset, enteric gangliogenesis and enteric glia (Chalazonitis et al., 2004; Faure et al., 2007; Chalazonitis et al., 2008). Since BMP-2 and BMP-4 are members of the TGFβ superfamily and HIPK2 has been shown to interact physically with receptor-regulated Smads, including Smad1, Smad2 and Smad3, and modulate Smad-dependent gene expression (Harada et al., 2003; Zhang et al., 2007), we postulated that loss of HIPK2 might influence BMP signaling. To test this hypothesis, we characterized BMP signaling in enteric neurons by examining the expression of phosphorylated Smad1/5/8 (pSmad1/5/8) in enteric neurons in the myenteric and submucosal plexuses of wild type and Hipk2−/− mutants. Whereas no difference in pSmad1/5/8 staining was found in the enteric neurons of wild type and Hipk2−/− mutants at P0, there were more neurons with positive nuclear pSmad1/5/8 staining in the myenteric plexus of the ileum and colon in 12-week old Hipk2−/− mutants than in wild type littermates (Fig. 6A vsB and E vs F , and data not shown). When normalized to the total number of neurons in the corresponding region of the gut, a much higher percentage of neurons in the myenteric plexus of the ileum in Hipk2−/− mice (20.8±1.3%) showed positive pSmad1/5/8 staining than in wild type mice (5.97±0.5%)(Fig. 6G, left panel). pSmad1/5/8 staining was also greater in submucosal plexus neurons in Hipk2−/− mice (32.8±2.7%) than in wild type animals (16.7±1.3%)(Fig. 6C vsD and G, right panel, p < 0.005). A similar increase in pSmad1/5/8-positive neurons was also detected in the myenteric plexus of the colon in Hipk2−/− mutants (47.2±4.7%), compared to the wild type (16.2±1.4%)(Fig. 6G). In contrast to the higher percentage of neurons with pSmad1/5/8 immunoreactivity, we found no detectable increase in pSmad2 staining in neurons of Hipk2−/− mutants (data not shown). The inverse relationship of more severe neuronal loss and a higher percentage of neurons with pSmad1/5/8 immunoreactivity in the ileum and colon of Hipk2−/− mutants suggest that an aberrant increase in BMP signaling in enteric neurons may have contributed to the observed phenotype.
In addition to their roles in enteric neuron development, BMP-2 and BMP-4 can also regulate gliogenesis in the ENS. Mice over-expressing BMP-4 (NSE-BMP-4) showed increased glial density and glia-to-neuron ratio (Chalazonitis et al., 2011). Because loss of HIPK2 leads to a significant increase in BMP signaling, we reasoned that the glial development in the ENS of Hipk2−/− mutants might be perturbed. To test this, we first determined if HIPK2 is expressed in the enteric glia. Using antibodies that detected β-galactosidase (for HIPK2LacZ) and S100β (for glia), we found that HIPK2 was present in very few glia at P0, but showed a progressive increase from P16 to P32 (Fig. 7A-C). Remarkably, the density of glia in the myenteric plexus of adult (P32) Hipk2−/− mice (expressed per mm2 of plexus) was significantly increased in the duodenum (198±10.4 [n=15] in Hipk2−/− mutants compared to 167±7.8 [n=12] in wild type, p < 0.05) and in ileum (284.3±16.5 [n=12] in Hipk2−/− mutants compared to 181.3±11.6 [n=12] in wild type)(Fig. 7D, E, H). Similar increase in glial density was also detected in the submucosal plexus in the ileum of Hipk2−/− mice (103±7 [n=12] compared to 78.2±3.4 [n=12] in wild type), but not in the duodenum (Fig. 7F, G, I). When normalized as % of wild type littermates, the glial density in Hipk2−/− mutants continued to show significant increases in the myenteric plexus of duodenum (118.4±6%) and ileum (156.8±9%), and in submucosal plexus of the ileum (131.4±8%)(Fig. 7J). Thus, similar to the decrease in neuronal density, the increase in glial density in the ENS of Hipk2−/− mutants also showed a proximal-to-distal gradient.
Our previous results have shown that loss of ventral midbrain DA neurons in Hipk2−/−mutants is due to increased apoptosis during the period of programmed cell death (Zhang et al., 2007). As is true of ventral midbrain DA neurons, the development of enteric neurons is also a long process that extends from the fetal period to postnatal life (Pham et al., 1991; Liu et al., 2009). It is unclear, however, whether enteric neurons, including those with a dopaminergic phenotype, exhibit programmed cell death during development (Gianino et al., 2003). To determine whether loss of enteric neurons in Hipk2−/− mutants was due to an increase in neuronal death, we used TUNEL assays in combination with the demonstration of β3-tubulin immunoreactivity to quantify the number of neurons undergoing apoptosis. Because loss of enteric neurons in Hipk2−/− mutants does not become significant until the 2nd postnatal week, we focused our analyses on enteric neurons at P7 and P14 mice. In contrast to the ventral midbrain dopaminegic neurons, apoptosis in the myenteric or submucosal plexus of Hipk2−/− mutants was not significantly different from that of WT mice at either age (data not shown). These observations suggest that the loss of neurons in Hipk2−/− mutants is not due to an increase in apoptosis.
To determine whether the loss of HIPK2 influenced maturation of enteric neurons in postnatal life, we used electron microscopy (EM) to examine the ENS of Hipk2−/− mutants. At P16, there was a significant increase in lysosomes and autophagosomes in the axons and cell bodies of neurons in Hipk2−/− mutants (Fig. 8B and D). There were also many swollen axons that contained abnormal mitochondria and cytoplasmic vacuoles in the distal ileum. In contrast, axons in the wild type myenteric plexus were organized in tight bundles and only very few small autophagosomes were noticed in axons or in cytoplasm (Fig. 8A, C). Interestingly, as in the more severe neuronal phenotype in the distal parts of gut, there was a progressive increase in neurons with autophagosomes in the ileum and colon (Fig. 8E). Consistent with the increase in autophagosomes, there was a significant increase in the intensity of microtubule-associated protein-1 light chain 3 (LC3) immunofluorescence, an indicator of autophagy activity, at P7 and P17 in myenteric neurons of the colon in Hipk2−/− mutants (Fig. 8F, G).
Myenteric neurons are known to form intraganglionic synapses and elaborate microcircuits, which enable the ENS to manifest integrated neuronal activity and independently control the behavior of the bowel. The phenotype in Hipk2−/− mutants, including abnormal GITT, reduced density of neural connectives, and decreased numbers of neurons, suggested that the loss of HIPK2 might also affect synapse formation in the myenteric plexus. To test this hypothesis, we characterized the development of intraganglionic synapses in the intestines of Hipk2−/− mice. Whereas the expression of synaptic markers, such as SNAP-25, synaptotagmin and PSD-95, showed similar robust staining in neurons in wild type submucosal and myenteric plexuses, these markers were substantially reduced in the intestines of adult Hipk2−/− mutants (Fig. 9A-C vs D-F). To characterize further the cause of the reduced expression of synaptic markers in Hipk2−/−mutants, we employed EM to examine synaptic growth at various postnatal stages. As early as P0, discrete synaptic structures were found on enteric neurons in wild type and Hipk2−/− mice; these structures, which contained synaptic vesicles, synaptic clefts and post-synaptic density, were observed in duodenum, ileum and proximal colon (Fig. 9G-J). Interestingly, the density of synapses in the myenteric plexus increased progressively in postnatal life from P0 to P36, indicating that enteric neurons continue to mature postnatally for a considerable period of time (Fig. 9K and data not shown). Compared to P0, the number of synapses in P36 wild type duodenum, ileum and colon increased by 2-, 2.3- and 2.5-fold, respectively (Fig. 9K). In contrast, synapses in the myenteric plexus of Hipk2−/− mice virtually failed to increase from P0 to P36 (Fig. 9K). Importantly, whereas the density of synapses in Hipk2−/− mutants was not significantly reduced in duodenum and ileum, the synaptic density was reduced in the proximal colon at P0 (p = 0.0068, n = 4, Fig. 9K). The few synapses that were identified in the myenteric plexus of Hipk2−/− mice at P36 appeared poorly organized and lacked discrete synaptic structures, such as synaptic clefts and post-synaptic densities (Fig. 9H, J). Taken together, the EM analyses of postnatal enteric neurons supports the idea that loss of HIPK2 leads to a prominent increase in autophagosomes and an arrest in synaptogenesis and maturation.
Our results underscore the important role of HIPK2 in the postnatal development of enteric neurons and glia. Loss of HIPK2 has no effect on the prenatal development of enteric neurons or its progenitors, but leads to a progressive loss of enteric neurons and an arrest in synaptic maturation in postnatal life. Although there is no detectable reduction in enteric neurons at P0, Hipk2−/− mutants show distended stomachs and small intestines at perinatal stages (Fig. 1). These phenotypes are not caused by congenital malformations, such as tracheo-esophageal fistula, which may have interfered with feeding. Instead, two factors may potentially contribute to the gut motility problems in Hipk2−/− mutants at P0. First, there is a significant reduction in the synaptic density in the proximal colon of Hipk2−/− mutants at P0 (Fig. 9). In addition, modest reductions of synaptic density are also detected in the duodenum and ileum in Hipk2−/− mutants at P0 (Fig. 9). These synaptic defects may lead to functional impairments in the gut motility at perinatal stages and may persist into adulthood. Second, it is possible that defects in ventral midbrain DA neurons in Hipk2−/− mutants at perinatal stages may interfere with their ability to feed (Zhang et al., 2007).
The significant loss of dopaminergic neurons in the ENS of Hipk2−/− mutants in postnatal stages is reminiscent of the dopaminergic neuron phenotype in the substantia nigra pars compacta (SNpc) and the ventral tegmental area (VTA)(Zhang et al., 2007). There are, however, several distinct differences in the Hipk2−/− mutant dopaminergic phenotype in the ENS and ventral midbrain (vMB). For example, the onset of dopaminergic neuron deficits in SNpc and VTA in Hipk2−/− mutants coincides with a significant increase in apoptotic neurons during the period of programmed cell death (PCD). Ventral midbrain dopaminergic neurons from Hipk2−/− mutants also fail to survive in the presence of TGFβ, suggesting that HIPK2 is required to transmit the pro-survival signals from TGFβ in these neurons. Consistent with this notion, both in vitro and in vivo studies support the notion that TGFβ isoforms are pro-survival trophic factors for the midbrain dopaminergic neurons (Poulsen et al., 1994; Krieglstein et al., 2000; Zhang et al., 2007). In contrast, there is no detectable increase in apoptosis in the ENS of Hipk2−/− mutants and the time course for the loss of enteric neurons in Hipk2−/− mutants follows a rather long postnatal interval. While the underlying mechanism of the neuronal deficit in Hipk2−/− mutants remains unclear, it is interesting to note that, unlike sympathetic and sensory neurons, deprivation of trophic factor support for enteric neurons does not seem to trigger conventional apoptotic cell death. In a previous study, no increase in caspase-3 immunoreactivity was observed in the enteric neurons at fetal, early postnatal, or adult stages in mice (Gianino et al., 2003). Interestingly, loss of the GDNF receptor, GFRα1, in the ENS leads to unconventional cell death that is not associated with the activation of common cell death executors, such as caspase-3 or caspase-7. Neuronal death induced by GDNF deprivation, furthermore, cannot be blocked by caspase inhibitors or Bax deficiency (Uesaka et al., 2007). The only documented example of apoptotic cell death that can be related to the ENS occurs in early vagal neural crest-derived cells, which are migrating progenitors en route to the bowel. Blocking cell death in migrating vagal crest-derived cells, using dominant negative caspase-9 in developing chick embryos increases the number of enteric neurons (hyperganglionosis) in the proximal foregut (Wallace et al., 2009). Together, these results suggest that the manifestation of apoptosis in the ENS may be stage-dependent, and that loss of trophic factor support in postmitotic neurons may lead to removal of these neurons via unconventional degenerative processes (Enomoto, 2009).
Although TGFβ isoforms support the survival of vMB dopaminergic neurons, their roles in the survival and maturation of enteric neurons are not clear. In cultures of enteric neural crest-derived cells, we have shown that low doses of BMP-2 or BMP-4 promotes, while high doses and prolonged exposure inhibit the development of enteric neurons. These results suggest that the effects of BMPs on enteric neuron development may be context-dependent (Chalazonitis et al., 2004). Consistent with this idea, transgenic mice that over-express the BMP antagonist Noggin show an increased number of neurons that exit the cell cycle early, but have a reduced number of neurons, such as the dopaminergic neurons, that exit the cell cycle late (Chalazonitis et al., 2004; Chalazonitis et al., 2008). It is important to note that higher than normal levels of pSmad1/5/8 are found in the remaining enteric neurons of Hipk2−/− mutants (Fig. 6); however, there is no detectable change in the pSmad2 level (A.A. Tang & Y. Shang, unpublished observations). While it is unclear how loss of HIPK2 leads to upregulation in pSmad1/5/8, we do know that HIPK2 is a transcriptional corepressor that can directly interact with Smad1 and suppress BMP-dependent reporter gene expression (Harada et al., 2003). Consistent with the results in Figure 6, pSmad1/5/8 is upregulated in mouse embryonic fibroblasts (MEFs) from Hipk2−/− mutants upon treatment with BMP. In contrast, when treated with TGFβ, Hipk2−/− MEFs show significantly less pSmad2 than wild type MEFs (Y. Shang, unpublished observations). It is thus possible that HIPK2 may be required to prevent excessive BMP signaling by suppressing the activation of Smad1/5/8. As a consequence, loss of HIPK2 results in a constitutive upregulation of the BMP signaling mechanism, which may influence neuronal survival and maturation, and promotes gliogenesis in postnatal life.
Our observations indicate that the enteric neurons of Hipk2−/− mutants contain increased number of autophagosomes and exhibit synaptic growth arrests from P0 to P36 (Figs. 8–9). One potential contributing factor to the synaptic deficit in Hipk2−/− mutants could be the marked upregulation of pSmad1/5/8 in Hipk2−/− enteric neurons (Fig. 6). In support of this hypothesis, several studies have indicated that the BMP ortholog, Glass bottom boat (Gbb), and the BMP type II receptor, Wishful thinking (Wit), in Drosophila promote the growth and synaptic homeostasis of the neuromuscular junction (NMJ)(Aberle et al., 2002; McCabe et al., 2003; Goold and Davis, 2007). TGFβ1, furthermore, can also regulate long-term synaptic potentiation, neuronal excitability and synapsin distribution in Aplysia neurons (Zhang et al., 1997; Chin et al., 2002; Chin et al., 2006). It is possible that constitutive activation of the BMP signaling pathway in the absence of HIPK2 adversely influences synaptic growth and homeostasis in postnatal enteric neurons. Several recent studies indicate that, in addition to their roles in regulating synaptic growth, TGFβ and BMP signaling can influence the induction of autophagy in both normal and cancer cells (Kiyono et al., 2009; Suzuki et al., 2010; Xavier et al., 2010). Perturbations in BMP and/or TGFβ signaling pathways in the enteric neurons of Hipk2−/− mutants may thus have contributed to the increase in autophagy in these neurons. It is intriguing to note that the cellular and synaptic phenotype of Hipk2−/− mutants is more prominent in the ileum and colon than in the proximal intestine. While the exact cause of this phenotype is still unclear, it is possible that there is a rostro-caudal gradient in enteric expression of BMPs (Goldstein et al., 2005), similar to the dorsal-ventral gradient of BMPs that contributes to patterning of the forebrain and spinal cord (Liu and Niswander, 2005).
Finally, our analyses of the DA neuronal phenotype in Hipk2−/− mutants provide a potential insight into the pathogenesis of Parkinson’s disease. It is well-documented that patients with Parkinson’s disease show manifestations of symptoms and Lewy body pathology in the ENS at the very early stage of disease progression, including reduced dopaminergic neuron numbers and dopamine level in the muscularis externa (Wakabayashi et al., 1990; Singaram et al., 1995). Expression in transgenic mice of mutated α-synuclein, which is found in genetic forms of Parkinson’s disease, leads to degenerative changes in the ENS, abnormal motility, and aggregates of α-synuclein in enteric neurons (Kuo et al., 2010). The intestinal dysfunction of these mice is not dissimilar to that seen in Hipk2−/− mutants (Fig. 1). These observations lead to the hypothesis that the ENS plays an important role in the pathogenesis and pathophysiology of Parkinson’s disease (Lebouvier et al., 2009). Despite these exciting observations, there is a persistent lack of evidence that the dopaminergic phenotype in Parkinson’s patients or animal models can be causally related to a given signaling mechanism. The results from our current and previous studies indicate that the Hipk2−/− mutant mice provide an intriguing example where loss of HIPK2, a key signaling component in the TGFβ and BMP signaling pathways, leads to structural and functional impairments in dopaminergic neurons in vMB and the ENS (Zhang et al., 2007). Many prior studies, which support our results, have implicated TGFβ in the development, survival and neuroprotection of dopaminergic neurons in a number of experimental paradigms (Roussa et al., 2009). Future studies are needed to determine whether components of the TGFβ or BMP signaling pathways can be used as therapeutic targets to prevent or treat congenital dysmotility syndromes or acquired conditions, such as Parkinson’s disease that compromise the function of the ENS.
We thank Dr. Moses Chao for the TrkC antibody, members of the Huang Lab and Gershon Lab for many helpful discussions, and Dr. David Sulzer, Dept Neurology, Columbia University for helpful comments on the manuscript. This work was supported by grants from NIH (NS48393 and RR24858 to E.J.H., and NS12969 and NS15547 to M.D.G.) and Dept of Veterans Affairs Merit Review Award (to E.J.H.).