Abstract
Delta catenin (CTNND2) is an adhesive junction associated protein belonging to the family of p120ct catenins. It is located on the short arm of chromosome 5, a region deleted in Cri-du-chat syndrome. Heterozygous loss of CTNND2 function has been linked to autism, schizophrenia, and mild intellectual disability with or without dyslexia-like learning difficulties. To date, most functional studies have focused on homozygous loss of the gene, contradictory to the dominant effect of loss of a single allele observed in neurodevelopmental disorders. Here we show that heterozygous loss of ctnnd2b results in a disorganisation and imbalance of neuronal subtypes in forebrain specific regions. Using the zebrafish model, we show that CRISPR/Cas9-induced loss of ctnnd2b but not ctnnd2a results in an increase in isl1-expressing cells and a local reduction of GABA expressing neurons in the optic recess region of the embryonic zebrafish forebrain. Using time-lapse analysis, we found that the disorganised distribution of is1l-expressing forebrain neurons was not due to migration defects, but rather an increase in the number of isl1-GFP neurons in the optic recess region. Upon closer analysis, these neurons appear disorganised and show an altered morphology and orientation. Overall this data suggests that ctnnd2 may affect the differentiation cascade of neuronal subtypes in specific regions of the vertebrate brain.
Introduction
The vertebrate nervous system is a highly organised structure consisting of various neuronal and non-neuronal cell types that need to interact in order to execute complex locomotor, sensory and behavioural tasks. Its function relies on the establishment of a spatial architecture during development which can be divided into gross patterning of the neural tube, proliferation, differentiation and specification of individual neurons, migration to their defined locations in various functional compartments of the brain, dendritic morphogenesis, targeted projection of axons and synapse formation. Neurons are highly diverse and their functional specialisation can be attributed to both their localisation and molecular signature. Broadly, they can be divided into interneurons, relaying signals locally or projection neurons, sending axons to form synapses in other regions of the brain. They are generally further classified into inhibitory and excitatory neurons, characterised by the expression of specific neurotransmitters. Most inhibitory interneurons use the neurotransmitters GABA while most excitatory projecting neurons use Glutamate. Other major neurotransmitters with modulatory actions in the CNS are Acetylcholine, Dopamine, Serotonin, Histamine and Norepinephrine. Understanding the molecular pathways involved in subtype specification, function, and maintenance is of importance.
CTNND2 belongs to the family of p120ct catenins and is an important neural developmental gene which is deleted in Cri-du-chat syndrome. Hemizygous loss of CTNND2 has been reported to increase the severity of Cri-du-chat (Medina et al., 2000), while isolated heterozygous loss of CTNND2 function has been linked to autism (Turner et al., 2015), mild intellectual disability with or without dyslexia-like learning difficulties (Hofmeister et al., 2015) (Belcaro et al., 2015), and schizophrenia (Vrijenhoek et al., 2008). During development of the nervous system, the encoded protein ∂-catenin appears to be involved in various neurodevelopment processes. These include neuronal differentiation (Lee et al., 2016), synaptic dysfunction (Israely et al, 2004), dendritic branching, spine morphogenesis (Kim et al., 2008), and maintenance of mature dendrites and dendritic spines (Matter et al., 2009) (Gilbert and Man, 2016). Homozygous mice expressing the GFP-tagged N-terminal domain of CTNND2 show spatial learning difficulties and abnormal fear conditioning (Israely et al., 2004), possibly due to abnormalities in dendritic arbour maintenance (Matter et al., 2009), suggesting that loss of ∂-catenin affects behaviour. However, no early developmental abnormalities were detected and no phenotype was reported in heterozygous mice, contradicting the autosomal dominant role of heterozygous loss of CTNND2 seen in human patients.
Here we show, using the zebrafish model, that heterozygous loss of ctnnd2b but not ctnnd2a results in a disorganisation and imbalance of neuronal subtypes in the zebrafish forebrain optic recess region (ORR). Moreover, we observe a local increase in isl1-expressing cells and a local reduction of GABA expressing neurons. Using time-lapse analysis, we found that the disorganised distribution of isl1-expressing forebrain neurons was not due to migration defects. These clusters of neurons presented an altered morphology and orientation upon closer analysis, suggesting that ctnnd2 may affect the differentiation cascade of neuronal subtypes in vertebrates.
Results and Discussion
ctnnd2b LOH but not ctnnd2a LOH results in misplaced neurons
The zebrafish genome contains two ctnnd2 orthologs, due to a genome wide duplication event that occurred during teleost evolution. We have previously shown that reduction of ctnnd2b expression but not ctnnd2a using antisense morpholinos leads to ectopic/misplaced isl- expressing neurons in the zebrafish forebrain in the optic recess region (Hofmeister et al., 2014). To further investigate these findings we created stable CRISPR/CAS9 knockout lines in an isl1:GFP transgenic background. These lines, named lines 1-4, contain a combination of short deletions and insertions of both ctnnd2 orthologs, resulting in premature stop codons (refer to methods, supplementary Fig. S1-S4 and Table. S1). To validate the genotype to phenotype penetrance of different allele combinations, double ctnnd2a and ctnnd2b heterozygotes (line 1 and line 2) were in-crossed and analysed. Heterozygous loss of ctnnd2b results in ectopic isl1-GFP neurons at 54hpf as previously observed (orange arrows, Fig. 1G and H), compared to wild-type isl1:GFP embryos (Fig. 1E), wildtype siblings (Fig. 1F) or embryos with a heterozygous loss of ctnnd2a (Fig. S5), confirming our previous finding (Hofmeister et al., 2014). Similar results were seen in mutants with different out of frame deletions in ctnnd2a and ctnnd2b (Line 2, Fig. S5). The abnormal isl:GFP cells are located in the optic recess region (ORR), which can be identified as a distinct morphological entity (Fig. 1A and B) located behind the optic chiasm, rostral to the supraoptic tract (SOT, Fig. 1B) and between the anterior commissure (AC) of the telencephalon and the post-optic commissure (POC) of the diencephalon which gives rise to the preoptic area (Affaticati et al., 2015). Although some isl1-positive neurons are occasionally seen in the optic recess region of wildtype isl1:GFP animals (Fig. 1I, Fig. S7, Fig. 3D), the GFP-positive population is significantly expanded in knockout fish both in number and intensity (Fig. 1E-M, Fig. 3D’). Confocal analysis at a higher magnification shows that the ectopic neurons still project axons into the post-optic commissure (Fig. 1N-P’). However, the neurons display a vertical rather than horizontal orientation and the cluster seems disorganised compared to neurons in wildtype embryos that form part of the diencephalic population (Fig. 1N-P’). Ectopic neurons are also still present at later stages (6dpf) analysed (Fig. 1L, M).
Schematic showing isl:GFP neuron population in the zebrafish forebrain at 2dpf (A, B). Anterior view in (A) and lateral view in (B) with rostral to the left. Confocal z-stacks of zebrafish embryos taken from an anterior view are shown (E-H and J-M). 54hpf embryo brains are shown in (E-H) and 6dpf brain in (L and M). Brightfield in situ images of laterally mounted 54hpf brains stained with a riboprobe for nkx2.1 are shown in (C, D) with anterior views in the inserts. No difference in nkx2.1 staining was observed. Wildtype isl:GFP transgenic zebrafish are shown in (C and E). Non-carrier sibling (wildtype larvae) from a cross of a double ctnnd2a and ctnnd2b heterozygotes are shown in (F, L, N and N’), a heterozygous ctnnd2b carrier in (G, J, K, M, P and P’) and a double ctnnd2a and ctnnd2b heterozygote in (H). Count of misplaced isl:GFP cells at the optic recess region is shown in (I) with heterozygotes (A(+/+)/B(+/-); n=17) and double heterozygotes (A(+/-)/B(+/-); n=16) significantly increased in number (Wilcoxon test; p<0.001) when compared to non-carrier siblings (A(+/+)/B(+/+);n=8) or non-sibling isl1:GFP embryos (n=11) (I). Confocal optical slices at the level of ectopic neurons is shown in N and P from an anterior view with lateral projections in N’ and P’ with rostral to the right. Scale bar =10μm
Ectopic isl:GFP neurons do not express ChAT in early development
In mammals, ISL1 has been shown to form a complex with LHX3 to promote cholinergic gene expression in basal forebrain cholinergic neurons (Hyong Ho Cho et al., 2013). This neuronal population is involved in attention, memory, reward pathways, and motor activity (for review see (Allaway and Machold, 2017). Forebrain cholinergic neurons have an embryonic origin in the ventral telencephalon (subpallium), in a region that expresses the transcription factor nkx2.1. In mammals the nkx2.1 expression domain encompasses the medial ganglionic eminences (MGE), septum, and post optic area. The equivalent nkx2.1 expression domain in zebrafish where the isl1 forebrain cells reside encompasses the ventral telencephalon, optic recess region, ventral diencephalon, and hypothalamus (Manoli and Driever, 2014). In situ analysis revealed no difference in patterning of this nkx2.1 region between wildtype and heterozygous embryos (Fig. 1C, D) suggesting normal patterning of this region. In order to assess whether zebrafish isl1 forebrain neurons also express acetyl choline we used an antibody against choline acetyl transferase (ChAT). However no staining was evident in the majority forebrain isl:GFP neurons at 54hpf (Fig. 1J). Instead CHAT staining at 54hpf is confined to a cluster of neurons in the hypothalamus (Fig. 1J and K), some of which but not all express isl1 at the specific stage imaged (Fig. 1J). In drosophila loss of isl1 results in loss of dopamine and serotonin synthesis (Thor and Thomas, 1997). Despite ectopic isl:GFP expression in neurons we see no change of dopaminergic, or serotoninergic cells in the zebrafish forebrain as assessed by TH1 and 5-HT staining respectively, and there is no overlap between isl1:GFP expression and TH1 or 5-HT in any of the misplaced cells at 2dpf (supplementary Fig. S6A-B and E-G).
Ctnnd2b LOH results in abnormal GABAnergic neurogenesis
The ventral telencephalon is also the birthplace of GABAergic neurons. The choice as to whether nkx2.1 expressing cells in the ventral telencephalon become GABAergic or cholinergic is due to a distinct expression of a combination of transcription factors. Post mitotic neurons fated to become GABAergic neurons express high levels of lhx6, a transcription factor essential for differentiation to GABAergic neurons (Liodis et al., 2007). Conversely, in post-mitotic nkx2.1 positive neurons expressing high levels of isl1 and lhx8, the expression of lhx6 is supressed and these neurons become forebrain cholinergic neurons (Cho et al., 2014). Thus, the differentiation into GABAergic neurons is inhibited by the expression of isl1 (Hutchison 2006). We therefore decided to assess the expression GABAergic neurotransmitters. Staining with an antibody against GABA shows that GABAergic neurons in diencephalon and ORR are intermingled but rarely overlapping with isl1:GFP as expected (Fig. 2A,B,D,E). Conversely, ectopic isl1 neurons in regions were GABA is normally expressed do not show GABA expression, and populations of GABA stained neurons appear to be more loosely packed together in both heterozygous ctnnd2b or double heterozygous embryos then in controls (compare Fig. 2A and D to B and E). A similar result is seen in double homozygous embryos (supplementary Fig. S6H). Resliced side projections confirm the presence of ectopic isl-GFP cells in the normally GABA+ ORR (compare Fig. S6I and J). Count of GABAergic neurons throughout the Z-sections taken at the ORR region confirmed a reduction in GABA+ neurons in heterozygous ctnnd2b embryos in addition to the disorganisation (Fig. 2C).
Confocal optical slices at the level of ectopic isl:GFP neurons (green), showing GABA stained cells (red) in (A,B,D,E). Equivalent confocal z-stack of (D) and (E) showing the positioning of normal isl:GFP cluster (D’) and abnormal ectopic cells (E’) is shown and the lateral view of the positioning of equivalent ectopic isl:GFP neuron cluster in (E’’) and (B’’’). anterior views are shown in (A, B, D, D’, E, E’) and resliced Lateral views in (D’’ and E’’). Count of anti-GABA stained cell throughout the ORR is shown in (C). Where the optic recess region was defined as the region from the POC where we start seeing ectopic isl1:GFP cells to the start of isl:GFP cluster of cells branching from the SOT. A significant reduction is observed (Wilcoxon test, p<0.01) when comparing siblings with wildtype ctnnd2b (n=5) compared to those heterozygous loss of ctnnd2b (n=16) irrespective of ctnnd2a genotype. Scale bar =10μm
Increased isl1 expressing cells at optic recess region due to specification not migration
In order to asses if the observed phenotype was due to migration or abnormal specification we imaged embryos between 32 and 48 hpf. Staging was additionally assessed by growth of visible axon tracts such as the trigeminal and post-optic commissure. Initially the number of GFP expressing cells were similar between isl:GFP in-crossed control larvae (Fig. 3B and Movie S7) and in animals missing one functional copy of ctnnd2b (Fig. 3B’ and Movie S7). However, as development proceeds we see a decrease in cells expressing GFP in controls, with only a few cells remaining at the end of the recording period 48hpf (Fig. 3D and Movie. S7). In contrast, in mutant embryos we observe an increase in the number of GFP-positive cells arising in the optic recess region, with no indication of downregulation of GFP throughout the developmental stages analysed (Fig. 3D’ and Movie. S7). In addition, GFP-positive cells appear less organised, supporting the observations from fixed embryos imaged using confocal microscopy at a slightly later stage at 54hpf (Fig. 1).
3D renders of timelapse imaged wildtype (A-D) and abnormal ctnnd2b heterozygous embryos (A’-D’) with highlighted ORR (insets). Embryos at 32hpf prior to appearance of neurons at ORR are shown in (A). Similarly, numbers of neurons start appearing at the optic recess region by 37hpf (B) and 42.5hpf (C). By the end of recording at approx. 48hpf only a few isl1:GFP cells are evident in wildtype embryos compared to heterozygous embryos (D).
In summary our results show that knockout of ctnnd2 results in in the overexpression of isl1 positive cells at the optic recess region which are highly disorganised and either directly or indirectly affect the local differentiation and organisation of GABA neurons in that region. Recently CTNND2 has shown to be a target gene of REST and CoREST (Abrajano et al., 2009), transcription factors which are thought to modulate neuronal subtype specification (Lee et al., 2016). Our study supports a role for ∂-catenin in subtype specification. Studies have implicated ∂-catenin as a modulator of canonical Wnt and RhoA signalling as well as modulating Cadherin stability. However, its exact role in these pathways is still a matter of debate. The transcription factor ISL1 is known to be downstream of canonical WNT signalling in cardiac development (Qyang et al., 2007)(Lu et al., 2014). Whether loss of CTNND2 causes abnormal ISL1 /GABA neuronal balance in specific regions of the mammalian brain and if it acts through an increase in canonical WNT signalling is yet to be determined.
CTNND2 haploinsufficiency has been linked to autism and mild neurodevelopmental disorders specifically reading difficulties. Reduced GABAergic action has been observed in the brain of autistic individuals (Fatemi et al., 2009)(Robertson et al., 2016) (Oblak et al., 2011) and genes encoding GABA receptor subunits have been associated with autism in linkage and copy number studies (Ma et al., 2005) suggesting imbalance or dysfunction of the GABAergic system a potential cause of autism. Our study supports a role for CTNND2 haploinsufficiency in the aetiology of previously associated neural developmental disorders due to abnormal GABA neurons differentiation in specific brain regions.
Materials and Methods
Zebrafish husbandry
All fish were maintained on a 14 h day/night cycle at the Karolinska Institute zebrafish core facility, located in the department of Comparative Medicine. Embryos were produced by light-induced spawning. Tg(isl1:GFP)rw0 transgenic strains (isl:GFP for short) or TL strains were used for all experiments.
CRISPR/Cas9-mediated knockout of ctnnd2 orthologues
Zebrafish mutant lines were established using a protocol adapted from Varshney et al (2015). sgRNAs and Cas9 protein mix was injected into 1-cell embryos and efficiency was tested using the PCR-STAT method. The target sequence of active sgRNAs were in exon 5 of ctnnd2a (5’-GGT TCA GCG AGC GAT ACA CC-3’) and exon 9 of ctnnd2b (5’-GGT GGA AGC GCC CGG GCC AG-3’). Primers used for screening were ctnnd2a-F (5’-TCA CTC TCC TTA CAG GTG ACA-3’) and ctnnd2a-R (5’-GGT TGC GTA GCC AGG TGT AT-3’) or ctnnd2b-F (5’-CCA GGT CCT GAA CTT TCT GC-3’) and ctnnd2b –R (5’-GGT CCT TCT GGA TGC TGT CC-3’), respectively. Embryos were raised and founders identified using the same PCR-STAT method. Founders were outcrossed to wildtype or isl:GFP fish and raised to adult stage. Heterozygous F1 adults were identified using fin clipping and subsequent Sanger sequencing.
Characterisation of mutant lines
We identified a total of 2 lesions for ctnnd2a, an 1bp deletion and an 8bp deletion. For ctnnd2b we identified a total of three, a 7bp deletion, a combined 6bp deletion and 10bp insertion, and a 1bp deletion, the latter was not used for further analysis. All of these lesions were predicted to cause a frameshift resulting in premature termination of the protein sequence (Fig. S1-4) Combinations of these resulted in five distinct F1 carrier lines., Line1; 8bp deletion in ctnnd2a and 6bp deletion/10bp insertion in ctnnd2b, Line2; 1bp deletion in ctnnd2a and 7bp deletion in ctnnd2b, Line3; with only the 7bp deletion in ctnnd2b, and Line 4 a with only the 6bp deletion/10bp insertion in ctnnd2b (Table. S1). F1 fish were incrossed or outcrossed and larvae analysed for phenotypes as indicated throughout the manuscript. Embryos were fixed and phenotyped via confocal microscopy before being genotyped by sanger sequencing. Alternatively, larvae were anesthetized and their tails were cut to allow genotyping before tissue fixation and image analysis
Immunofluorescence and in situ hybridisations
Embryos were fixed in 4% (w/v) paraformaldehyde (PFA) overnight at 4°C. Whole larvae or dissected brains were processed for immunofluorescence using standard protocol (Devine et al., 2003). Primary antibodies used were rabbit anti-Th1 (cat# P40101; Peel-freez biologicals), mouse anti-acetylated tubulin (cat# T6793; Sigma), rabbit anti-GABA (cat# A2052; Sigma), goat anti-CHAT (cat# AB144P Sigma/Merck), rabbit anti-Serotonin (cat#S5545; Sigma). Secondary antibodies used were goat anti-mouse Alexa 594 (cat # A11032 Life Technologies, Carlsbad, California, USA), donkey anti-goat Alexa 546 (cat# A11056 Life technologies). In situ staining was performed as previously described (Hjorth et al., 2001). The nkx2.1 riboprobe was synthesised using the DIG labelling kit (Roche) using the following PCR template with a T7 RNA polymerase site attached to the 5’ end of the reverse primer (forward primer: 5’-TTG CCT CGT ACA GAC AAC CC-3’, reverse primer: 5’-TAA TAC GAC TCA CTA TAG GGG TCA CCA CGT CCT GCC ATA A-3’). Embryos were mounted in gelvatol for imaging.
Confocal imaging and Data analysis
Transgenic embryos fixed in 4% PFA or immunostained embryos were mounted rostrally or laterally in gelvatol mounting medium and imaged using a Zeiss LSM 700,800 or Olympus FV1000. Images where blinded and isl:GFP or GABA stained cells counted throughout individual confocal sections. Cell counts were subsequently corelated back to the genotype of the embryos as identified by sanger sequencing. ISL1-positive cell count from multiple experiments/crosses were pooled together (supplementary table S2). For GABA-positive cell counts, data was obtained from two separate crosses of double heterozygous adults and one cross of double homozygous adults (table S3). dplyr package in R-studio was used for data handling, statistical analysis and visualisation using the ggplot2 package. The Mann-whitney-wilcoxon tests was used to compare between samples.
Lightsheet microscopy
31hpf embryos were anesthetised in tricaine (MS222) and mounted in 1% low melting point agarose containing 1:2000 F-Z 050 fluorescent microspheres (Merck) in a glass capillary. The agarose cylinder was extruded into a sample chamber containing E3 medium at 28.5°C. Two channel images were acquired using a Light Sheet Z.1(Carl Zeiss, Germany) with a water dipping 20X detection objective (W-Plan-APOCHROMAT-1.0NA) and dual side 10X illumination objectives (LSFM, 0.2NA). Samples were illuminated from two sides and Z-stacks were acquired from 3 angles (covering 90°) every 20 min for 16 hours. Data from multiple angles was registered and deconvolved in the Multiview Reconstruction Fiji plugin (Preibisch et al., 2010). Briefly, beads were detected in one channel and used to register views and correct drift. Transformations were duplicated to the GFP channel and images were fused by multi-view deconvolution, performed on the GPU (Preibisch et al., 2014). Deconvolved images were rendered and movies created in Imaris 9.0 (Bitplane).
