|Home | About | Journals | Submit | Contact Us | Français|
Disruption of E3 ubiquitin ligase activity in immature zebrafish mind bomb mutants, leads to a failure in Notch signaling, excessive numbers of neurons, and depletion of neural progenitor cells. This neurogenic phenotype is associated with defects in neural patterning and brain development. Because developmental brain abnormalities are recognized as an important feature of childhood neurological disorders such as epilepsy and autism, we determined whether zebrafish mutants with grossly abnormal brain structure exhibit spontaneous electrical activity that resembles the long-duration, high-amplitude multi-spike discharges reported in immature zebrafish exposed to convulsant drugs. Electrophysiological recordings from agar immobilized mind bomb mutants at three days postfertilization (dpf) confirmed the occurrence of electrographic seizure activity; seizure-like behaviors were also noted during locomotion video tracking of freely behaving mutants. To identify genes differentially expressed in the mind bomb mutant and provide insight into molecular pathways that may mediate these epileptic phenotypes, a transcriptome analysis was performed using microarray. Interesting candidate genes were further analyzed using conventional reverse transcriptasepolymerase chain reaction (RT-PCR) and real-time quantitative PCR (qPCR), as well as whole-mount in situ hybridization. Approximately 150 genes, some implicated in development, transcription, cell metabolism and signal transduction, are differentially regulated including down-regulation of several genes necessary for GABA-mediated signaling. These findings identify a collection of gene transcripts that may be responsible for the abnormal electrical discharge and epileptic activities observed in a mind bomb zebrafish mutant. This work may have important implications for neurological and neurodevelopmental disorders associated with mutations in ubiquitin ligase activity.
Notch is an essential component of an evolutionarily conserved signal transduction cascade mediating development. In neuroectoderm, where cells have the potential to become neurons, activated Notch inhibits proneural gene expression in a process referred to as lateral inhibition. In the absence of Notch-mediated lateral inhibition, too many early-born cells differentiate into neurons (Chitnis et al., 1995; de la Pompa et al., 1997). Recent studies identified several E3 ligases that modulate Notch signaling through ubiquitin-dependent protein degradation and endocytosis (Lai, 2002). Ubiquitination, which occurs when an E3 ligase enzyme binds to both substrate and an E2 thioesterified protein (Deshaies and Joazeiro, 2009), is a key mechanism regulating many cellular processes. Mutation or small deletions within the ubiquitin E3A ligase gene in humans has been linked to autism spectrum disorders (Glessner et al., 2009) and Angelman syndrome, a neurogenetic disorder characterized by developmental delay, severe intellectual disability, absent speech, exuberant behavior, motor impairment, and epilepsy (Clayton-Smith and Laan, 2003).
Mutations in several zebrafish mind bomb gene disrupt E3 ubiquitin ligase activity (Schier et al., 1996; Itoh et al., 2003). Here we used a zebrafish insertional mutant (mibhi904) that is allelic to the chemically induced mib mutants (mibta52b, mibm132, mibtfi91). Defects in mibhi904 are the result of a retroviral insertion in an exon of the zebrafish homolog of the human KIAA1323 gene (Golling et al., 2002).(Chen and Casey Corliss, 2004) showed that Hi904 mutation disrupts a conserved putative E3 ubiquitin ligase that regulates Notch signaling. In mib mutants, there is an excess production of early-born neurons and a concomitant reduction in the number of late-born neurons. At 2 dpf, the developing spinal cord and hindbrain is nearly filled with huC, an early neuronal differentiation marker (Itoh et al. 2003; (Park and Appel, 2003) and an over-abundance of early differentiating Rohon-Beard neurons (Bingham et al., 2003). Consistent with a lateral inhibition neurogenic phenotype, late-differentiating secondary islet GFP-expressing motor neurons and gata3-expressing interneurons are decreased in the mib hindbrain (Bingham et al. 2003). Genome-wide analysis of mind bomb mutants was recently performed to identify Notch responsive genes required for pancreatic (Hegde et al., 2008) or mesodermic (Hwang et al., 2009) development. However, identification of differentially regulated central nervous system (CNS) genes was not discussed and little is presently known about the functional or behavioral consequences of a defect in E3 ubiquitin ligase activity in mibhi904 mutants.
Here we present a phenotypic characterization of mind bomb mutant zebrafish. Electrophysiological analysis shows that mibhi904 mutants exhibit spontaneous epileptiform-like burst discharge similar to that obtained following exposure to a convulsant drug. Video observation and locomotion tracking indicates the occurrence of seizure-like behaviors (Baraban et al., 2007; Berghmans et al., 2007). Additionally, microarray analysis, quantitative real-time PCR and whole-mount in situ hybridization identified a down-regulation of several genes associated with GABA-mediated synaptic transmission.
Adult male and female zebrafish (mind bomb, line #hi904; mind bomb, line #ta52b; t-complex polypeptide 1, line #hi3564; histone deacetylase 1, line #hi1618; denticleless homolog, line #hi447) were obtained from the Zebrafish International Resource Center (Eugene, OR; http://zebrafish.org/zirc/fish/lineAll.php). Adult zebrafish were maintained according to standard procedures (Westerfield, 1993), and following guidelines approved by the University of California, San Francisco Institutional Animal Care and Use Committee. Zebrafish embryos and larvae were maintained in egg water (0.03% Instant Ocean) unless otherwise stated.
To obtain stable physiological recordings, zebrafish larvae at 3 days post-fertilization were immobilized in 1.2% low-melting temperature agarose in zebrafish egg water. Larvae were embedded so that the dorsal aspect of the brain was accessible for electrode placement. Embedded larvae were bathed in egg water and visualized using an Olympus BX50 microscope (Olympus America Inc., Center Valley, PA). Under direct visual guidance, a glass microelectrode (<1.2 mm tip diameter, 2-7 MΩ) was placed in the forebrain or optic tectum. Electrodes were filled with 2 M NaCl and electrical activity was recorded using an Axopatch 1D amplifier (Molecular Devices, Sunnyvale, CA). Voltage records were low-pass filtered at 1-2 kHz (-3 dB, 8-pole Bessel), high-pass filtered at 0.1-0.2 Hz, digitized at 5-10 kHz using a Digidata 1300 A/D interface, and stored on a PC computer running pClamp software (Molecular Devices).
Electrophysiological recordings were analyzed post hoc using Clampfit software (Molecular Devices). Spontaneous gap-free recordings, 15-30 minutes in duration, were analyzed for all fish (n = 75). A threshold for detection of spontaneous events was set at 3x noise (peak-to-peak amplitude) and 100 msec (duration); all events exceeding these thresholds were analyzed.
For locomotion tracking, single zebrafish larvae were placed individually in 96-well Falcon culture dishes (BD Biosciences, Franklin Lakes, NJ, U.S.A.). Each well contained 70 μl embryo medium. Swimming behavior was monitored at 3 dpf for 10 minutes in WT larvae (n = 22) and mib mutant larvae (n = 28) using a CCD camera (Hamamatsu C-2400, Hamamatsu City, Japan) and EthoVision 3.1 locomotion tracking software (Noldus Information, Inc., Leesburg, VA). Locomotion plots were categorized by two observers blind to larvae phenotype as displaying class 0 (no movement, baseline activity), class 1 (small increase in movement), or class 3 (considerable movement, convulsive behavior) activity (Baraban et al., 2005). Using Ethovision software (Noldus) the percentage of time spent swimming/moving was analyzed for individual locomotion plots.
A total of six microarrays were hybridized to compare gene expression patterns between mib mutants (n = 3) and age-matched WT sibling controls (n = 3), including a dye swap for each tissue; each microarray used RNA pooled from 10 zebrafish larvae. At 3 dpf, larvae were sorted by morphology and total RNA was extracted using Trizol® Reagent (Invitrogen, Carlsbad, CA), treated with DNAse (Ambion/Applie Biosystems, Austin, TX) and quantified with GeneQuant® spectrophotometer. An Agilent Bioanalyser 2100 (Agilent Technologies Inc., Santa Clara, CA) was used to assess the integrity/quality of mRNA.
Hybridization, scanning and analysis were completed by the NIH Neuroscience Microarray Consortium (http://arrayconsortium.tgen.org/np2/home.do) using an Affymetrix (Santa Clara, CA) zebrafish genome array with ~14,900 Danio rerio transcripts. Different internal quality controls were used, including hybridization controls (BioB, BioC, BioD and cre), Poly A controls (dap, lys, phe and thr) and “housekeeping” control genes (GAPDH, alpha 1 Actin). Sequence information for this array was selected from the following public data sources: RefSeq (July 2003), GenBank (release 136.0, June 2003), dbEST (July 2003), and UniGene (Build 54, June 2003). Probe sets on the array were designed with 16 oligonucleotide pairs to detect each transcript. Affymetrix Gene Chip Operation Software (GCOS) Version 1.4 was used to analyze the resulting image files. The global scaling technique was used to scale the fluorescence intensity of each chip to a target signal of 150. The data files (CEL files), resulting from analysis with the Affymetrix GCOS software, were imported into GeneSpring GX 7.3.1 software (Agilent Technologies Inc.) for further data analysis.
Transcripts were considered as differentially expressed using multiple non-parametric two-tailed unpaired Student's t-test with a Benjamin Hochberg multi-test correction (GeneSpring GX 7.3.1). A P value ≤ 0.05 was considered to be significant. Hierarchical clustering (GeneTree, Salt Lake City, UT) was applied to the data files. Categorization of genes identified by microarray analysis was carried out using GO (Gene Ontology) Slim terms.
Several genes of interest (GOI) were chosen for further analysis. 1 μg of DNAse-treated total RNA from whole zebrafish (10 fish/pool) was reversed transcribed (SuperScript™III First-Strand Synthesis System, Invitrogen) using a mix of oligo dT20 and random hexamers. Two sets of primers pairs, forward and reverse, were specifically designed using Primer 3 web software (http://frodo.wi.mit.edu/primer3/) for each investigated gene to obtain a longer sequence (primers sequences are available in Supplemental Table 1). The most conserved regions were identified by sequence alignment (ClustalW) (Thompson et al., 1994) of all available gene sequences from GeneBank including other fish species (Cyprinus carpio, Carassius auratus, Dicentrarchus labrax, Oncorhynchus mykiss, Salmo salar, etc.). Investigation of primer cross-specificity was done using BLAST software. The predicted secondary structure of the entire DNA sequence was checked using Mfold software (http://frontend.bioinfo.rpi.edu/applications/mfold; (Zuker, 2003). Each reaction cycle (32 loops) consisted of incubations at 94°C (30 sec), 60°C (30 sec), and 72°C (60 sec) with Taq DNA Polymerase (Taq PCR Core kit, Qiagen). PCR products were separated by agarose (2%) gel electrophoresis stained with ethidium bromide and cloned in pCR®II-TOPO® plasmid vector (TOPO TA Cloning System, Invitrogen). DNA sequencing was performed by Elim Biopharmaceuticals, Inc. (Hayward, CA).
For antisense and sense RNA probes, the plasmids corresponding to each gene were linearized with appropriate restriction enzymes (HindIII, SpeI or BamHI and Not I, XbaI or ApaI, New England Biolabs, UK). The linearized DNA template (1 μg) was purified (QIAquick®, Qiagen) and incubated for 3 hour at 37°C in a solution containing 10X transcription buffer, dithiothreitol (DTT; 100mM), 10X Dig NTP Mix (Roche), RNAse inhibitor (20U/μl), and RNA polymerase (20U/μl) T7 or SP6. The DNA template was digested with DNase (10U/μl) for 15 minutes at 37°C. After incubation, 30 μl of RNAse-free water and 30 μl of LiCl were added for purification and left overnight at -20°C. After centrifugation at 4°C, the pellet was rinsed with 70% ethanol (RNAse free), dried and stored in hybridization mix at -20°C until hybridization.
Embryos (3 dpf) were fixed in 4% paraformaldehyde (PFA) and then stored in 100% methanol at -20°C. In situ RNA hybridization was performed, as described (Hauptmann and Gerster, 1994). Once developed, the embryos were mounted in 70% glycerol for whole-mount imaging. All images were captured using a Zeiss Axioskop microscope equipped with a Nikon E995 or Optronics MicroFire digital camera. Raw images were imported into Adobe Photoshop and adjusted regarding the level of brightness, contrast and cropping.
Gene expression levels of seven chosen genes were examined using RNA pooled from 10 WT sibling larvae (n = 6) or 10 mib mutant larvae (n = 6). RNA was extracted in the same manner as for microarray experimentation (above) and reverse-transcription reactions (RT-PCR) were performed in the same manner as for cloning and sequencing using SuperScript™III First-Strand Synthesis System (Invitrogen). The cDNA templates were diluted 1:2 with DEPC (Diethyl pyrocarbonate) sterile water before qPCR applications to minimize the presence of potential inhibitors.
The qPCR reactions were performed using SybrGreen® fluorescent master mix on an ABI Prism® 7700 Sequence Detection System driven by ABI prism SDS v9.1 (Applied Biosystems). Primers were designed to produce amplicons ranging in size between 81bp and 130bp (Table 1) using Primer Express v3.0 (Applied Biosystems). All primers were synthesized by Invitrogen. Samples were run in triplicate and contained 1× SYBR green master mix, 10 μM of each primer and RNAse free water for a final volume of 10 μl. Samples without reverse transcriptase and samples without RNAs were run for each reaction as negative controls. Cycling parameters were as follows: 50°C × 2min, 95°C × 10min, then 40 cycles of the following 95°C × 15s, 60°C × 1min. For each sample a dissociation step was performed at 95°C × 15s, 60°C × 20s, and 95°C × 15 s at the end of the amplification phase to check for the presence of primer dimmers or non specific products (Suppl. Fig. 1).
Triplicate quantification values (CT; cycle threshold), provided from real-time qPCR instrumentation, were imported into a Microsoft Excel spreadsheet for further analysis. Raw data was analyzed using qCalculator software (programmed by Ralf Gilsbach) which estimates qPCR efficiency E= 10(– 1/slope) and the relative gene expression between samples after normalization with the most reliable EndG basing on both the Comparative ΔΔCT (Livak and Schmittgen, 2001) and the Efficiency Based (Pfaffl, 2001) methods. Similar results were obtained with both types of analyses.
For all genes, qPCR efficiencies, detection limits and dynamic ranges were assessed by mean of 4-fold serial dilutions of pooled cDNA (5 standards assayed in triplicate: 1/1; 1/4; 1/16; 1/64; 1/256). Serially diluted cDNAs were used to construct standard curves and estimates of efficiencies, slope of the curves and the correlation coefficient (Suppl. Table 2).
Given that there is no reason to expect a single gene to be the most stably expressed endogenous gene (EndG) in all samples, the reliability of one or more reference genes was determined before proceeding with quantitative mRNA expression studies. 12 samples (6 samples of WT and 6 samples of mibhi904) were used to verify expression stability of four different commonly used EndGs (Tang et al., 2007; Chen et al., 2008; Lin et al., 2009): small subunit ribosomal RNA (18S), protein elongation factor 1 alpha subunit (EF1α), beta-actin (β-act) and beta-2 microglobulin (β2M). Bestkeeper Excel-based tool (Pfaffl et al., 2004) and NormFinder (Andersen et al., 2004) software were used to rank all the EndGs. Bestkeeper assesses candidate genes by pair-wise correlations based on cycle threshold values (CT) which are then combined into an index and calculates the standard deviation (SD) of the CT values between the whole data set. NormFinder uses a model-based approach to rank all reference gene candidates based on inter- and intra-group expression variations. Analyzed with both programs (BestKeeper and NormFinder), the β-act gene was the most stable gene followed by EF1α, β2M and 18S (data not shown); β-act was used in our studies for data normalization.
Student's t test was used to determine statistical significance between the normalized relative expression values in qPCR assay. A P value ≤ 0.05 was considered to be significant.
Brain malformations resulting from errors in neurodevelopment are commonly associated with significant cognitive delay, autism and intractable epilepsy (Schwartzkroin and Walsh, 2000; Guerrini and Marini, 2006). To identify zebrafish with a brain malformation and epilepsy, a brief “shelf screen” of previously identified mutants with grossly visible brain abnormalities was performed. Adult zebrafish mutant lines (mibhi904; mibta52b; tcphi3564; hdachi1618; dtlhi447) were obtained from the Zebrafish International Resource Center (ZIRC). Offspring from these mutants were sorted at 2 or 3 dpf based on morphology e.g., small or misshapen head, smaller eyes and in many cases, a dorsally curled body. Representative WT sibling and mibhi904 zebrafish larvae at 3 dpf are shown in Figure 1A1 & 1B1. Field recordings were obtained from forebrain or optic tectum for recording periods up to 1 hr. Recurrent spontaneous multi-spike bursts > 1000 msec in duration were observed in 93% of mibhi904 mutants (n = 28) at frequencies between 0.2 and 1.5 bursts/min (Figs. 1B2 & 1C). Analysis of burst duration (control: 0.17 ± 0.07 sec; mib: 3.3 ± 0.91 sec; Student's t test; p = 0.003) and frequency (control: 0.05 ± 0.01 bursts/min; mib: 0.8 ± 0.14 bursts/min; Student's t test; p = 0.0001) confirmed that these differences were significant. Prolonged burst discharges were similar in waveform to those classified as “ictal-like” following exposure to 15 mM pentylenetetrazole (PTZ) a GABA receptor antagonist exposure (Fig. 1C; Baraban et al. 2005). Only brief and infrequent field bursts were observed in age-matched WT siblings (Figs 1A2 & 1C; n = 18). Tectal field response to paired pulse stimulation of the contralateral eye was comparable between mibhi904 and control fish (compare traces in Fig. 1D). Analysis of paired pulse data at 20 sec interpulse intervals did not reveal significant differences in facilitation (control: 107 ± 14%; mib: 106 ± 9%; Student's t test; p = 0.94). Abnormal electrical discharge was not observed in gap-free recordings from tcphi3564 mutants between 3 and 6 dpf (n = 12) or hdachi1618 mutants at 3 dpf (n = 3). Identifiable offspring from mibta52b or dtlhi447 mutants were not viable beyond 2 dpf.
Behavioral manifestations of electrographic seizures include episodes of excessive locomotor activity and myoclonus of all four limbs, and are well characterized in rodent models of epilepsy (Pitkänen et al., 2006). To further characterize epilepsy in mibhi904 mutant zebrafish larvae, offspring were placed in one well of a 96-well Falcon plate and the behaviour of freely swimming fish was observed under a stereo-microscope. Using a scoring system developed for acute seizures (Baraban et al. 2005), a CCD camera and locomotion tracking software, 10 min recording epochs were obtained for mibhi904 (n = 28) and age-matched WT siblings (n = 22) bathed in normal embryo media at 3 dpf. Behaviors consistent with Stage I (increased swim activity; Fig. 2A) and Stage 3 (brief and violent clonus-like convulsions; Fig. 2A) were consistently noted in mibhi904 mutants (Fig. 2B). WT zebrafish were primarily observed to exhibit behaviors that could be characterized as Stage 0 (little or no swim activity) (Fig. 2B). Analysis of the percentage of time moved during 10 min epochs scored as Stage 3 (mib: 0.12 ± 0.05%; n = 8) or Stage 0 (control: 0.0 ± 0.0%; n = 11) confirmed that these differences were significant (Student's t test; p = 0.015). Video of a representative Stage 3 seizure event in a mibhi904 mutant is shown in Supplemental Video 1. Electrographic seizures were confirmed in a sub-set of mibhi904 mutants that were first monitored for behavioural seizure activity and then embedded in agar for electrophysiological recording (n = 4).
To identify potential signaling pathways and transcriptional activation patterns that could give insight into the epileptic phenotype observed in mibhi904 mutants, we performed transcriptome analysis using an Affymetrix zebrafish microarray. Hierarchical clustering analysis showed that ~1.9 % (247 genes) of the genes (±13,151 array probes selected after elimination of controls and blank; Suppl. Fig. 2) were differentially expressed between wild-type sibling and mibhi904 mutants (P ≤ 0.05, T-test; GeneSpring GX 7.3.1); 97 genes were assigned to an “unknown function” category and not analyzed further. Among the 150 genes differentially expressed between these two groups, a similar amount of genes exhibited higher (n = 65; 43% up-regulated; Suppl. Table 3) and lower (n = 85; 57% down-regulated; Suppl. Table 4) expression in mibhi904. Based on literature searches, these genes were grouped into four broad categories: cellular process (n = 103), development (n = 25), transport (n = 13) and immune response/response to a stimulus (n = 9) (Table 2 and Suppl. Fig. 3).
As shown in Table 2, many of the genes in the “cellular process” category pertain to signal transduction, transcription and general (cellular) metabolic process; additional genes were involved in other cellular process such as apoptosis, cell adhesion, etc. Several gene products in our microarray screen correspond to Danio rerio homologs of previously identified activity-regulated transcripts found in mammals following seizure (Flood et al., 2004; Gorter et al., 2006), including the immediate early gene Fos, the CCAAT element binding protein encoded by c/EBP, early growth response 1 (egr1) and brain derived neurotrophic factor (BDNF). As a general category with putative roles in the generation of abnormal electrical activity a number of gene products related to inhibitory synaptic transmission were noted as down-regulated e.g., glutamate decarboxylase 1 (GAD1), similar to GABA transporter 1 (GAT1), glycine receptor α4b (GLRA4b), calbindin 2 (Calb2), and parvalbumin isoform 2a (Pvalb5).
To validate the microarray results, we analyzed a subset of genes using three independent methods for evaluating RNA expression: (i) conventional RT-PCR, (ii) qPCR and (iii) WISH. For all seven transcripts assayed, RT-PCR detection of gene expression levels were confirmed in a qualitative manner and the average-fold change in gene expression reported in the microarray data was statistically confirmed by qPCR. Spatial expression patterns for the selected genes assessed using WISH were consistent with PCR findings. In the following sections several of these genes are discussed in more detail.
Seizure induced expression of activity-dependent CNS genes has been described in animal models and tissue obtained at surgery from patients with intractable forms of epilepsy (Hendriksen et al., 2001; Elliott et al., 2003; Hunsberger et al., 2005; Rakhade et al., 2005; Xi et al., 2009). One of these previously described genes, BDNF a member of the “neurotrophin” family of growth factors, is believed to play a role in synaptic plasticity and excitability (Croll et al., 1999; Cowansage et al., 2009). In agreement with previous reports, BDNF expression is increased nearly 3-fold in mibhi904 mutants (Fig. 3B). In contrast to the low level of endogenous CNS expression of BDNF in WT sibling zebrafish a diffuse, but prominent up-regulation was observed in the telencephalon, optic tectum, midbrain and hindbrain of age-matched mibhi904 mutants (Fig. 3C). Another previously described gene, peptide YYa (PYYa), a member of the neuropeptide Y family thought to act as an endogenous anticonvulsant (Baraban et al., 1997; Vezzani et al., 1999), exhibits a nearly 3-fold increase in mibhi904 mutants (Fig. 4B). In WT sibling zebrafish, PYYa is expressed at a relatively low level near the olfactory bulb (OB), locus coeruleus (LC), ventral thalamic cluster and hindbrain (Fig. 4C1-2). In mibhi904 mutants we observed a slight reduction of PYYa expression near the OB and LC, but prominent up-regulation of gene expression in hindbrain (Fig. 4C3) extending into spinal cord (Fig. 4C4).
Reduction in expression of genes related to GABAergic signaling has been commonly described in the epilepsy, autism and schizophrenia literature (Harrison and Weinberger, 2005; Akbarian and Huang, 2006; Brooks-Kayal et al., 2009). As a general marker of GABA-mediated synaptic function, GAD1, an enzyme responsible for catalyzing the production of GABA from L-glutamic acid, was reduced by nearly 75% in mibhi904 mutants (Fig. 5B). This dramatic reduction was most prominent in a ventral view of the zebrafish head showing a near absence of GAD1 expression in the boundary of the olfactory pit and diencephalon (Fig. 5C), a region where GABAergic interneurons are thought to originate (Mione et al., 2008). Interestingly, GAD1 expression was also found to be decreased by microarray and real-time qPCR analysis of mibta52b mutants (Hegde et al., 2008). We also analyzed the Danio rerio homolog for the α1 subunit of the GABA-A receptor (Fig. 6), as a complimentary gene associated with postsynaptic GABA-mediated signaling and a subunit associated with an autosomal dominant form of juvenile myoclonic epilepsy (Krampfl et al., 2005), but not included on the Affymetrix gene array. This subunit was shown to be down-regulated nearly 2-fold in qPCR analysis of mibhi904 mutants (Fig. 6B). Interestingly, the spatial expression patterns in WISH showed a dramatic reduction in Gabra1 expression in the telencephalon, optic tectum and hindbrain (Fig. 6C). In addition, the α4 glycine receptor subunit, a signaling pathway mediating inhibitory neurotransmission in the spinal cord and brainstem (Yoshimura and Nishi, 1982; Legendre, 2001), was also found to be down-regulated in mibhi904 mutants (Suppl. Fig. 4).
Calcium-binding proteins are widely used as expression markers for specific interneuron sub-types (Xu et al., 2004; Batista-Brito et al., 2008). Our data show a significant reduction of two of these markers in mibhi904 mutants: calbindin (Fig. 7) and parvalbumin (Fig. 8). Quantitative PCR analysis reveals a greater than 50% reduction in expression for both gene products (Figs. 7B & 8B). Calbindin shows a diffuse CNS expression pattern in the telencephalon, optic tectum, cerebellum and hindbrain that is dramatically reduced in mibhi904 mutants (Fig. 7C). Parvalbumin, a marker of fast-spiking basket-type interneurons (Cauli et al., 1997), expression in a restricted region of the telencephalon and hypothalamus showed a clear reduction with a more diffuse decrease in expression across the entire brain (Fig. 8C).
E3 ubiquitin ligases have been linked to the control of many cellular processes and to multiple human diseases. Here, we present data suggesting that a mind bomb mutant deficient in E3 ubiquitin ligase activity exhibits epilepsy that may be associated with reduced GABA-mediated signaling. Decreased GABAergic signaling is among the more robust pathologies observed postmortem in schizophrenia, autism and epilepsy (Lawrence et al., 2010; van Kammen et al., 1998; Thom et al., 2004; Akbarian and Huang, 2006; Fatemi et al., 2009; Mellios et al., 2009). The phenotype of immature mibhi904 zebrafish suggests that this mutant may facilitate a mechanistic analysis of the pathogenesis of E3 ubiquitin ligase related diseases and provide a model system to search for more effective treatments.
The “neurogenic” phenotype in zebrafish mind bomb mutants is associated with defects in neural patterning and brain development (Jiang et al., 1996; Bingham et al., 2003). Because developmental brain abnormalities are recognized as an important feature of intractable childhood seizure disorders (Schwartzkroin and Walsh, 2000), we were initially interested in determining whether immature zebrafish mutants with grossly abnormal brain structure exhibit spontaneous seizure activity resembling that seen in 3 and 7 dpf zebrafish exposed to convulsant drugs (Baraban et al., 2005; 2007). Using mutants identified in chemical mutagenesis screens, epileptiform-like discharges were only observed in mibhi904 zebrafish. Although neuronal correlates of these recurrent burst discharges are not yet known, the waveforms can be considered an “EEG-like” representation of seizure activity because they appear in CNS structures and are consistent with clinical criteria used to define epilepsy (Engel, 2006). At 3 dpf wild-type zebrafish rarely move. However, intermittent spasms of convulsive behavior were consistently observed in age-matched mibhi904 mutants. By comparison to previous locomotion monitoring in PTZ-exposed zebrafish, and with appropriate caution, we can attempt to quantify these behaviors as Stage III “seizures”. Obviously, human and rodent seizure behavior differs qualitatively and quantitatively from that observed in zebrafish. Nonetheless, these behaviors have never been reported in undisturbed WT zebrafish and, taken together with the electrophysiological data, suggest a novel epileptic phenotype.
Acute zebrafish seizures, induced by exposure to PTZ, show many similarities to chemoconvulsant-induced events in rodents: progressive stages of behavior; complex electrographic discharges; induction of immediate early gene expression; prolonged waves of neuroluminescence; and similar profiles of antiepileptic drug responsiveness (Baraban et al., 2005; Berghmans et al., 2007; Naumann et al. 2010). In Drosophila, physiological correlates of behavioral seizure-like events can be observed following electroconvulsive stimulation (Kuebler and Tanouye, 2000; Lee and Wu, 2002) and in C. elegans (Williams et al., 2004) or Xenopis laevis (Hewapathirane et al., 2008) with exposure to chemoconvulsant drugs. These model organisms, though ideally suited for analysis of seizure propagation and molecular mechanisms involved in seizure generation, do not represent models of epilepsy defined as the occurrence of unprovoked recurrent seizures. In contrast, in telencephalic and tectal recordings from immobilized mibhi904 mutant zebrafish spontaneously occurring epileptiform discharges were observed in the absence of external stimuli.
How is it possible to generate epileptic activity in immature zebrafish? A simple answer is that the zebrafish brain possesses critical functional capacities thought to be necessary for generation of epileptic events in any species. Notably, excitatory neurotransmission mediated by three ionotropic glutamate receptor subtypes (Edwards and Michel, 2003) and inhibitory interneurons exerting actions via activation of postsynaptic GABAA receptors (Sajovic and Levinthal, 1983; Kim et al., 1996; Kim et al., 2004). These basic excitatory and synaptic components serve complex CNS computations in zebrafish such as “second-order motion perception” (Orger et al., 2000), long-term potentiation (Nam et al., 2004), or odor-evoked neural coding (Edwards and Michel, 2002; Mack-Bucher et al., 2007) and could be responsible for generation of abnormal electrical discharge.
Because one of the primary uses of an experimental model is to simplify a complex situation, we proceeded with a transcriptome analysis of mibhi904 zebrafish. In considering potential interpretations of these data it is important to consider a few caveats. First, zebrafish, while possessing distinct advantages as a “simple” vertebrate species are considered to have a less complex CNS with no clear homology for some mammalian structures thought to be important for seizure generation i.e., hippocampus and septum. The CNS of most vertebrates develops from a folding mechanism in which the lateral edges of the neural plate come together and fuse at the dorsal midline. However, the zebrafish neural plate is initially a solid structure which then develops a central canal by detachment of cells in the center (Blader and Strahle, 2000). Although everted in its final form, the general organization of telencephalic hemispheres and optic tectum in zebrafish suggest that these regions receive multimodal input from a variety of sources and share many organizational similarities with higher species (Vanegas and Ebbesson, 1976; Murakami et al., 1983; Ekstrom et al., 1986; Wullimann et al., 1996; Guo et al., 1999). Second, our analysis, using a single microarray, is limited by detection biases. Although we could identify transcripts in the GABA signaling pathway, the microarray is not sensitive to low abundance transcripts or those with restricted expression patterns. Moreover, this approach is not comprehensive in that many gene products involved in regulating excitability – ion channel variants and receptor subunits – are not represented on the microarray.
Microarrays have been used extensively to identify molecular processes that may underlie acute or chronic seizure activity (Hendriksen et al., 2001; Elliott et al., 2003; Hunsberger et al., 2005; Rakhade et al., 2005; Xi et al., 2009). In the epileptic brain, up-regulation of numerous genes mediating cell injury or survival is expected given the well known damage that can occur with excessive electrical activity (Sloviter, 1996; Houser, 1999; Elliott et al., 2001). Although the large percentage of “cellular process” genes identified as differentially expressed in mibhi904 mutants fit with known cellular demands associated with seizure activity, it is also possible that these findings reflect the higher representation of gene transcripts related to metabolism, immune response and injury on the microarray. Interestingly, up-regulation of activity-dependent genes with well established roles in the epileptogenic process validate our zebrafish model and further suggest a set of “commonality genes” that are up-regulated in a variety of species, including human, following seizure. Moreover, this list is not confined to the two genes confirmed here in a series of RNA expression studies (e.g., BDNF and PYY), but could be expanded to include transcripts identified in three or more independent epilepsy-related microarray studies (e.g., chemokine (C-X-F motif) ligand, protein kinase C, ornithine decarboxylase, early growth response 1, fibronectin and cathepsins). Further analysis of these transcripts in mibhi904 mutants and, perhaps, manipulation of these same genes in wild-type zebrafish using morpholino antisense oligonucleotides (Ekker, 2000) or Tol2-mediated transgenesis (Halpern et al., 2008) could help define roles in epileptogenesis.
Although our analysis did not focus on differential expression of gene transcripts related to Notch signaling and neurodevelopment, this is an important area for future investigation. In particular, CNS expression patterns for hairy-related genes (her4, her5, her6, her8a, her9 and hes5) and proneural factors (asc1b, atoh2a, ngn1, neurod) identified as down-regulated in three different microarray analyses of mib mutants (Table 2; Hegde et al., 2008; Hwang et al., 2009) warrant further analysis. Grossly abnormal development and early fatality suggest important roles for these genes in establishing the mind bomb phenotype.
Missense mutant alleles of human ubiquitin E3A ligase that result in loss-of-function have been identified in Angelman syndrome (AS) (Kishino et al., 1997; Matsuura et al., 1997; Camprubi et al., 2009) and loss-of-function point mutations have been reported in patients with autism spectrum disorder (Muhle et al., 2004). One potential model for these neurological symptoms posits that in the absence of ubiquitin ligase activity, one or more ubiquitin ligase substrates accumulate in the brain and interfere with CNS development and/or function. Drosophila E3 ubiquitin ligase mutants (dUBE3A) were recently used to test this hypothesis: (i) (Wu et al., 2008)) demonstrated that dUBE3A deletion mutants display a loss of motor coordination and cognitive impairment on olfaction and shock test paradigms that could be rescued by transgene expression and (ii) (Lu et al., 2009), also using dUBE3A-null mutant flies, reported deficits in the dendritic branching of sensory neurons. Although electroconvulsive stimulation can be used to assess seizure thresholds in fruit flies (Kuebler and Tanouye, 2000) these studies were not performed. Null mutation of Ube3a in mice (Jiang et al., 1998; Miura et al., 2002) results in impairment of hippocampal long-term potentiation (an in vitro correlate of learning and memory), increased susceptibility to induced seizures and intermittent bursts of spike-wave discharges in some animals. Although the precise nature of this disorder remains to be determined, there is a growing body of literature, supported here in our analysis of a zebrafish E3 ubiquitin ligase mutant, suggesting that dysfunction of GABA-mediated inhibitory neurotransmission is a key factor in the pathophysiology.
This work was supported by an R01 grant (NS053479-03) from the National Institutes of Health to S.C.B. The authors would like to thank Dan Lowenstein and Joy Sebe for comments on earlier versions of the manuscript, and Sally Chege for assistance with in situ hybridization studies.