Introduction

Neurodevelopmental disorders (NDDs) are clinically and genetically extremely heterogeneous, and currently more than 1,000 genes have been implicated (SysID database1). Within the last years, phenotypically and biologically coherent modules within this large and heterogeneous group have been increasingly delineated, indicating that overlapping phenotypes are often caused by mutations in genes involved in the same molecular networks1,2,3,4,5. This has been demonstrated mainly for well-defined pathways or complexes such as the RAS-MAPK-pathway4 or the SWI/SNF chromatin remodelling complex6, but is less characterized for disease genes involved in a broader biological context or process such as transcriptional regulation. Identifying and characterizing such connections and correlations contributes to a better understanding of the complex mechanisms and pathomechanisms in neurodevelopment and its associated disorders.

Pitt-Hopkins syndrome (PTHS, MIM# 610954), MEF2C-related intellectual disability (MRD20; MIM# 613443), Mowat-Wilson syndrome (MOWS, MIM# 235730), ATRX-related intellectual disability (ATRX, MIM#301040; MRXHF1, MIM#309580) and Angelman syndrome (AS, MIM#105830) represent a particular group of syndromic NDDs. They are important differential diagnoses towards each other and share phenotypic characteristics such as severe intellectual disability, epilepsy, postnatal microcephaly and some facial features7 (Fig. 1a). Angelman syndrome is caused by loss of the maternal allele of UBE3A, encoding an ubiquitin-protein ligase8,9. PTHS, MRD20 and MOWS are caused by haploinsufficiency of TCF4, MEF2C or ZEB2, respectively, all encoding transcription factors10,11,12. Also X-chromosomal ATRX, implicated in ATRX-related intellectual disability, encodes a transcriptional regulator13.

Figure 1
figure 1

Genes of interest and study outline. (a) Five syndromic NDDs with considerable phenotypic overlap were selected. AS, Angelman syndrome; MOWS, Mowat-Wilson syndrome; MRD20, mental retardation, autosomal dominant 20 (MEF2C haploinsufficiency syndrome); MRHXF1, mental retardation-hypotonic facies syndrome/ATRX, alpha-thalassemia-mental retardation syndrome; PTHS, Pitt-Hopkins syndrome (b) Schematic drawing of the study outline and the work flow. (c) Overview of results from genetic interaction studies in Drosophila. Black lines indicate gene-gene connections investigated in this study by one (thin line), two (middle line) or four (thick line) different combinations. Magenta lines indicate genetic interactions between two genes, respectively. Genetic interaction was defined here as the observation of multiple phenotypic modifications upon pairwise dosage manipulation in several tissues. Solid magenta line indicates strongest evidence for genetic interaction with several consistent phenotypic modifications observed in different tissues. Dashed line in magenta indicates evidence for a possible genetic interaction with some phenotypic modifications observed. Single or inconsistent phenotypic modifications are not indicated.

Drosophila melanogaster has been demonstrated to be a powerful model to investigate functional relationships between NDD-associated genes/proteins, given a high conservation of genes, pathways and regulatory networks between flies and humans14,15,16,17. Drosophila screens for eye, wing and neuronal phenotypes upon RNAi-based knockdown of NDD-associated gene orthologues revealed robust correlations between fly and human phenotype groups1,18 in terms of phenologs19, thus indicating conservation of functional modules. More specific functional relationships between individual genes can be investigated by genetic interaction studies. Genetic interaction is defined as the observation that a double mutant’s phenotype deviates from what is expected from each individual mutant20. Such approaches, based on quantifiable phenotypes in Drosophila, were utilized to investigate multiple-hit and copy number variant (CNV) models as done e.g. for autism spectrum disorders21,22 or to validate newly identified NDD-associated candidate genes by establishing biological links to phenotypically overlapping, known NDD-associated genes in terms of a chromatin-modification module5.

By utilizing Drosophila melanogaster as a model to screen for genetic interactions, we identified specific functional links between several genes in the fly, most stringent between Ube3a and Mef2. This interaction was furthermore confirmed in a human cell line using co-immunoprecipitation experiments. These molecular commonalities might contribute to the clinically overlapping features of the investigated disorders.

Results

Ubiquitous and tissue specific dosage manipulation in Drosophila melanogaster

To systematically investigate functional links between these genes of interest, we used Drosophila melanogaster as an in vivo model system and tested genetic interaction between the fly orthologues Mef2 (MEF2C), Zfh1 (ZEB2), Daughterless (Da) (TCF4), XNP (ATRX) and Ube3a (UBE3A).

Quantitative reverse transcriptase PCR (RT-PCR) confirmed knockdown (KD) to 35–70% residual levels and 3 to 8.5fold overexpression (OE) for all used lines (Supplementary Table S1) except for Zfh1 (Supplementary Fig. S1). Overexpression of Zfh1 resulted in early lethality, preventing quantitative RT-PCR, and knockdown could not be shown (Supplementary Fig. S1), leading to exclusion of the KD_Zfh1-line from subsequent experiments. To evaluate the possibility of overlapping phenotypes resulting from dosage manipulation of Ube3a, Mef2, Da, XNP or Zfh1 and to identify quantifiable phenotypes for subsequent genetic interaction experiments, we induced knockdown (four genes) or overexpression (five genes) either ubiquitously or in several different tissues and evaluated parameters such as viability, morphological alterations, synapse development and gross neurological behaviour (Fig. 1b).

Ubiquitous knockdown of two of the four tested genes (Mef2 and Da) and overexpression of all five tested genes resulted in lethality. Wing specific knockdown resulted in morphological phenotypes for two genes (Mef2, Da), and wing specific overexpression resulted in (male) lethality or morphological wing phenotypes for all five genes (Table 1, Figs. 2, S2).

Table 1 Viability and morphology upon tissue-specific dosage manipulation.
Figure 2
figure 2

Genetic interaction of Ube3a and Mef2 or XNP in the Drosophila wing (ms1096-GAL4). (a) Knockdown (KD) of Ube3a does not cause a wing phenotype, KD of Mef2 causes abnormally curled wings in male flies with additional cross vein defects, such as missing anterior cross veins and/or ectopic cross veins (marked with an arrow). Simultaneous KD of Ube3a and Mef2 results in a milder phenotype with significantly more flies with both cross veins present and fewer flies with ectopic cross veins, quantified in (b). (c) Overexpression (OE) of Ube3a causes abnormally curled wings in females, while KD of Mef2 does not cause a phenotype in female flies (male phenotype see above). Simultaneous OE of Ube3a and KD of Mef2 results in male lethality and in a more severe disorganization of wing architecture in about 75% of females, as quantified in (d). (e) OE of XNP causes abnormally curled wings in female flies with additional cross vein defects, such as missing anterior cross veins and/or ectopic cross veins (marked with an arrow). Simultaneous KD of Ube3a (normal) and OE of XNP results in a more severe phenotype with more flies with ectopic cross veins and fewer flies with both cross veins intact, as quantified in (f). Statistical analysis was performed using Fisher’s Exact test, **p ≤ 0.001; ***p ≤ 0.0001). Flies are counted towards the more severe phenotype if at least one wing was affected. These results are from an independent experiment than in Supplementary Table S2, thus numbers are different.

Eye-specific knockdown resulted in morphological phenotypes for two genes (Mef2, Da), and eye-specific overexpression resulted in lethality or morphological phenotypes for four of the genes (Ube3a, Mef2, Da, Zfh1) (Table 1, Figs. 3, S3).

Figure 3
figure 3

Genetic interaction of Mef2 and Ube3a in the Drosophila eye (GMR-GAL4). (a) Knockdown (KD) of Ube3a does not cause a phenotype, KD of Mef2 causes rough eyes with more severely affected flies also displaying a bubble-like appearance. Simultaneous KD of Ube3a and Mef2 results in a milder phenotype with significantly fewer eyes with bubble-like appearance, quantified in (b) (***: p ≤ 0.001, Fisher’s Exact test). (c) Overexpression (OE) of Ube3a causes rough eyes, OE of Mef2 results in a mildly reduced number of bristles but grossly intact ommatidial structure. Simultaneous OE of both results in a severe phenotype with reduced eye size and dissolved ommatidia structure in all eyes as quantified in (d). (e) Simultaneous OE of Ube3a (rough eye) and KD of Mef2 (rough eyes and occasionally bubble-like appearance) results in a different and more severe phenotype with disorganized ommatidia structure and progressive necrosis as quantified for male and female flies in (f). Pictures and quantifications are from male flies if not indicated otherwise.

Nervous system specific knockdown either pan-neuronally or in glia cells did not result in reduced viability. However, pan-neuronal overexpression of either Zfh1, Da or Ube3a or glial overexpression of either Zfh1, Da, XNP or Mef2 resulted in variable lethality phenotypes (Tables 1 and S2). Evaluation of several parameters of larval neuromuscular junctions (NMJs) such as NMJ area, NMJ length, number of islands, branches, boutons or active zones did not reveal any consistent alterations in synaptic morphology upon either pan-neuronal knockdown or overexpression of any of the genes (Supplementary Fig. S4). We did not observe a satellite bouton phenotype (data not shown) as previously described in an Ube3a mutant23. This might be related to a weaker knockdown of the RNAi line. Furthermore, gross neurological functioning was tested using the climbing assay upon pan-neuronal knockdown or overexpression. We found that overexpression of Da or XNP resulted in a mild climbing defect, respectively, and that overexpression of Ube3a led to a severe climbing deficit (Supplementary Fig. S5). Additional testing of seizure susceptibility with the bang sensitivity assay (Supplementary Table S3) did not reveal consistent phenotypes for any of the tested KD and OE lines.

Genetic interaction screen

Based on the location of RNAi- or UAS-elements on chromosomes 2 or 3, we could create three lines combining two KD elements each, eight lines combining two OE elements each, and twelve lines combining KD with OE elements each (Table 1, Fig. 1c). By quantitative RT-PCR we found no indication that distribution of GAL4 between two UAS-elements would dilute its effect (Supplementary Fig. S1). We tested all combinations for lethality and wing and eye morphology phenotypes. NMJ morphology, bang sensitivity and climbing behaviour were assessed for the three KD combination constructs and for the eight OE combinations.

Phenotypic modifications were considered as a) additive, when the combined phenotype represented the sum of the two single phenotypes, b) as antagonistic, when the combined phenotype was milder than the more severe of the two single phenotypes, and c) as synergistic, when the combined phenotype was different or more severe than what was expected from the sum of the two single phenotypes.

In the evaluation of viability and gross morphologic eye and wing phenotypes upon ubiquitous and tissues specific dosage manipulation we observed several modified phenotypes in the pairwise combinations (Table 1). We considered those most indicative of a true genetic interaction, where modified phenotypes occurred at least in three different KD/OE combinations and tissues and were markedly milder (antagonistic) or more severe or different than expected from adding the single phenotypes (synergistic) (Fig. 1c).

The most robust interaction was observed between Ube3a and Mef2. Combined knockdown in either wing (Fig. 2a,b) or eye (Fig. 3a,b) resulted in an amelioration of the phenotype observed in the single conditions. Wing venation defects as well as the rough eye with loss of ommatidia integrity, resulting in bubble-like appearance of the eye surface, improved. Pairwise overexpression of Mef2 and Ube3a in either wing or eye led to a worsening of phenotypes. In the wing, the only mildly abnormal, curled wings in females from each of the single conditions were severely disorganized and not fully unfolded in the combined overexpression (Supplementary Fig. S1). While each of the single overexpression conditions was associated with a mild eye phenotype only, their combination resulted in a severe eye malformation with markedly reduced size and dissolved ommatidia structure (Fig. 3c,d). Overexpression of Ube3a and simultaneous knockdown of Mef2 in either wing or eye also resulted in increased phenotypic severity, compared to the single conditions. Wings showed a severe disorganization in 75% of the females, which was not present in the single manipulations (Fig. 2c,d, Table 1). In the eye, the same combination resulted in partial male lethality and progressive necrotic patches in eyes of females (Fig. 3e,f, Tables 1 and S2). While pan-neuronal knockdown of Mef2 did not affect viability and while pan-neuronal overexpression of Ube3a only resulted in partial lethality, a combination of these conditions resulted in complete lethality (Tables 1, S2). A similar observation was made for glial deregulation. There, simultaneous overexpression of both and simultaneous overexpression of Mef2 with knockdown of Ube3a and vice versa resulted in complete lethality, while the single conditions were viable or only lethal in males (Tables 1, S2).

In summary, we have identified multiple phenotypic modifications upon pairwise dosage manipulation of Ube3a and Mef2. The synergistic or antagonistic direction of an interaction was dependent on the combination of KD and OE conditions and consistent across multiple tissues for each specific combination.

Also for combined dosage manipulation of Ube3a and XNP, phenotypic modifications were observed for several combinations and tissues. Simultaneous overexpression of Ube3a and knockdown of XNP in the wing resulted in increased viability of male flies (6% in OE_Ube3a alone and 39% in OE_Ube3a; KD_XNP, p < 0.001, Fisher’s exact test) and therefore in an improvement of the phenotype. In contrast, simultaneous knockdown of Ube3a and overexpression of XNP in the wing led to a more severe phenotype in the females, with stronger disorganization of cross veins compared to the phenotype in the single lines (Fig. 2e,f, Tables 1, S2). Pan-neuronal overexpression of Ube3a in combination with knockdown of XNP resulted in worsening of incompletely penetrant lethality in females (20% of balancer flies in OE_Ube3a_1 alone and 10% in OE_Ube3a_1;KD_XNP, p < 0.05, Fisher’s exact test) and complete lethality in males (76% vs. 0%, compared to females, p < 0.05, and 23% vs. 0%, compared to balancer flies, p < 0.0001, Fisher’s exact test).

Pairwise manipulations of other tested genes did not result in multiple or consistent phenotypic modifications. When we evaluated the climbing assay upon pairwise dosage manipulations, we observed a significantly increased climbing impairment of the OE_Da;OE_XNP construct compared to the respective single conditions, which, however, might represent an additive effect of both single gene conditions. No other consistent modifications of the observed single phenotypes were observed (Supplementary Fig. S5). Analysis of NMJ morphology and bang sensitivity upon pairwise dosage manipulations did not reveal any phenotypic modifications (Supplementary Fig. S4, Supplementary Table S3).

Protein co-localization and interaction

To investigate whether UBE3A and MEF2C also physically interact, we performed co-localization and co-immunoprecipitation studies in human cell lines. Firstly, we could confirm that UBE3A and MEF2C both localize to the nucleus as described previously24 when simultaneously overexpressed in HeLa cells, indicating that a physical interaction is possible (Fig. 4a). Subsequently, we could show that UBE3A and MEF2C can be co-immunoprecipitated from HEK293 cells co-transfected with Myc-tagged UBE3A and HA-tagged MEF2C. This was true for immunoprecipitation of either Myc-tagged UBE3A or HA-tagged MEF2C, suggesting that they indeed form a direct or indirect interaction (Figs. 4b, S6). Using quantitative RT-PCR, we tested if UBE3A/Ube3a or MEF2C/Mef2 might regulate transcriptional levels of each other. We did not find evidence for transcriptional effects as in fly larvae Mef2 levels were unaltered upon Ube3a knockdown and vice versa (Supplementary Fig. S1). Furthermore, also in human cells (HEK293) MEF2C expression levels were unaltered upon UBE3A knockdown using siRNAs (Supplementary Fig. S7).

Figure 4
figure 4

Physical interaction of human MEF2C and UBE3A in cells. (a) Co-localization studies of overexpressed human tagged MEF2C (HA, red) and UBE3A (Myc, green) in HeLa cells show diffuse nuclear localization of both proteins. (b) Co-immunoprecipitation experiments show physical interaction of overexpressed human MEF2C and UBE3A in HEK293 cells. Immunoprecipitation of Myc-tagged UBE3A also precipitates MEF2C-HA and vice versa. Please note, that the figure panels are cropped from two different blots with different exposure times. For full blots see Supplementary Fig. S6.

Discussion

Though it has increasingly been acknowledged that similar clinical neurodevelopmental phenotypes are caused by mutations in genes/proteins connected in common networks and processes, this has mostly been characterized for specific complexes or pathways (e.g2,4.) or on a more systematic, large scale but consequently less detailed1,18 level. We now investigated possible functional links between TCF4, MEF2C, ZEB2, UBE3A and ATRX which are all implicated in clinically overlapping, severe human neurodevelopmental disorders. To do this, we utilized Drosophila melanogaster as a model system to systematically investigate functional links between the orthologues of these genes/proteins (Da, Mef2, Zfh1, Ube3a, XNP) in vivo.

Drosophila lethality and morphological phenotypes point to developmental and glial roles

Our initial screen for lethality and gross morphological phenotypes upon ubiquitous and tissue specific gene dosage manipulation in the fruit fly identified a large number of quantifiable lethality and morphological phenotypes utilizable for the subsequent genetic interaction experiments.

In general, a high frequency of lethality upon ubiquitous but also upon tissue specific dosage manipulation underlines both the dosage sensitivity of these genes and their role for developmental processes. This has been shown in mouse models before. Ubiquitous knockout of either Zeb225, Tcf426, Mef2c27 or Atrx28 resulted in early lethality, while Ube3a brain-specific maternal-deficient mice displayed neurological deficits29. Interestingly, not only the loss of the maternal UBE3A copy in Angelman syndrome but also duplication of UBE3A in terms of duplication 15q syndrome is associated with a neurodevelopmental and epilepsy phenotype30, thus demonstrating bi-directional dosage sensitivity. We observed more frequent and more severe phenotypes upon overexpression than upon knockdown. Though knockdown might be a more suitable model for loss-of-function mechanisms, it might not always be phenotypically penetrant, particularly with RNA interference approaches as these usually leave a residual expression level of 30–50%. Therefore, overexpression screens provide a valuable additional tool to investigate gene function31,32,33 and to generate quantifiable phenotypes.

Contrary to our expectations for NDD-relevant genes, we did not observe consistent synaptic or behavioural phenotypes upon pan-neuronal manipulation in the utilized assays. Although all of the investigated genes are associated with epilepsy in humans, neither pan-neuronal knockdown nor overexpression of any single condition or combination resulted in bang sensitivity, a Drosophila model for seizure susceptibility, where mechanical shock can induce hyperactivity, spasms and paralysis34. This might suggest that seizures associated with haploinsufficiency of these transcriptional regulators might be related to different pathomechanisms from seizure-associated ion channel dysfunction which is typically reflected in bang sensitivity in flies35. Interestingly, in accordance with our findings, a previous report did not observe bang sensitivity upon neuronal overexpression of Ube3a as a model for epilepsy-associated duplications 15q11.2, either, but instead upon glial overexpression concomitant with down regulation of an ion pump36. Correlating a glial role of Ube3a between Drosophila and vertebrates is, however, difficult due to a discordant imprinting status of Ube3a in Drosophila neurons37, and as in a mouse model, Ube3a has shown to be expressed in glia but not to be imprinted there38. However, a possible, so far underestimated glial role of Ube3a and other genes investigated in this study would be in accordance with our observations that glial overexpression of either Zfh1, Da, XNP or Mef2 resulted in lethality while this was only the case for pan-neuronal overexpression of Zfh1 and with reduced penetrance for pan-neuronal overexpression of Ube3a. Although a relevance of Zfh1/ZEB2, XNP/ATRX or Da for (mainly peripheral) glia development and maintenance has been reported or discussed39,40,41,42,43, it has not yet been characterized in detail. Transcriptome analysis on cell populations from mouse brain and on human brain tissues summarized in the Brain RNA-Seq database indicates expression of all five genes not only in neurons but also in astrocytes, oligodendrocytes and microglia, with a higher expression of TCF4, UBE3A and ATRX in fetal compared to mature astrocytes44,45. Our observations therefore might suggest not only a role of neuronal but also of glial defects that might contribute to the neurodevelopmental phenotypes in humans with mutations in either of these genes.

Genetic interaction between Ube3a and Mef2

Though the phenotypic overlap between the neurodevelopmental disorders investigated in this study is widely appreciated and discussed in the literature7,46, corresponding experimental follow-up of possibly underlying molecular commonalities is largely lacking. According to the mouse brain atlas47, there is overlapping expression of orthologues of UBE3A, MEF2C, ATRX, ZEB2 and TCF4 in several neuronal subtypes including excitatory neurons of the cerebral cortex and various cell types of the hippocampus. This would be in line with molecular commonalities in the pathomechanisms of the associated NDDs. However, according to the STRING database48 (status July 2019), there is no experimental evidence on interaction between the human proteins so far, and available data is restricted to co-expression or interaction in other species. In our genetic interaction screen, we observed modification of phenotypes upon combination of several manipulated genes. Severe phenotypes such as severe and fully penetrant lethality upon Zfh1 overexpression precluded further evaluation for genetic interaction. For all other combinations, we observed modified phenotypes which might point to a functional link between these proteins in terms of genetic interaction (Table 1). Often, only two tissues and/or combinations were involved. More evidence was there for genetic interaction between Ube3a and XNP with modified phenotypes in two tissues and three different combinations. Further experimental follow-up would be necessary to confirm these potential interactions.

The most stringent evidence for a functional interaction was detected between Ube3a and Mef2. Several lines of evidence point to a true genetic interaction: a) we observed modified phenotypes in several tissues and in various knockdown/overexpression combinations, b) additive effects only can be excluded as some phenotypes occurred only upon pairwise manipulation (e.g. neuronal lethality) or were different or much more severe than expected from both single conditions (e.g. eye phenotypes), and c) we observed both antagonistic (milder phenotypes in eyes and wings upon simultaneous knockdown of Ube3a and Mef2) and synergistic effects (more severe phenotypes upon all other combinations). Additionally, we confirmed a physical interaction between human UBE3A and MEF2C in an independent cell-based system. To our knowledge, this is the first evidence for a functional link between UBE3A/Ube3a and MEF2C/Mef2 that might contribute to the phenotypic overlap between Angelman syndrome and MEF2C-associated intellectual disability. Apart from that, variants in either UBE3A or MEF2C might also represent modifiers for the phenotypic expression/severity of the respective other condition as discussed for CNV models21,22.

Taking into account the diverse functional roles of UBE3A and MEF2C, there are different possibilities conceivable how their interaction or mutual regulation might work. Observations from the genetic interaction and expression studies might already provide some insights into the possible nature of these interactions. The UBE3A gene encodes a member of the large family of E3 ubiquitin ligase proteins, initially termed E6-associated protein (E6-AP), and contributing to protein homeostasis by being involved in tagging substrate proteins with ubiquitin which are then degraded in the proteasome9. MEF2C (myocyte enhancer factor 2) belongs to the subfamily of MADS (MCM1-agamous-deficiens-serum response factor) transcription factors, whose transcriptional activity relies on the recruitment of and cooperation with other transcription factors as well as on translational and posttranslational modifications49. MEF2C might therefore be a transcriptional regulator of UBE3A expression. This might be supported by the identification of MEF2 binding sites in the Ube3a promoter50. Vice versa, also for UBE3A a transcriptional co-activation function has been reported51. However, a transcriptional regulation mechanism appears unlikely as in flies expression levels of Mef2 were unaltered upon Ube3a knockdown and vice versa, and as in a human cell line UBE3A knockdown did not change MEF2C expression levels. Additionally, stem cell-derived neurons modelling Angelman syndrome, did not show significant transcriptional changes of other genes (compared to 15q duplication neurons)52. As the most likely hypothesis, we suggest that UBE3A might regulate MEF2C activity and levels by ubiquitination, leading to subsequent degradation in the proteasome. UBE3A has been found to be located in the neuronal nuclei in discrete hotspots over euchromatin, thus well-positioned to regulate active genes24, and such a ubiquitin-ligase-dependent regulation has been reported for other transcriptions factors before in mouse models, e.g. for SOD153. Interestingly, only very recently, the critical importance of the nuclear isoform of UBE3A for the Angelman syndrome phenotype was characterized. Mice lacking the nuclear isoform but not mice lacking the cytosolic isoform displayed all major behavioural phenotypes and synaptic deficits also seen upon complete UBE3A knockout in the previous Angelman syndrome mouse model54. The hypothesis of ubiquitin-ligase dependent regulation of a putative nuclear substrate such as MEF2C by UBE3A is also supported by our genetic interaction findings. Pairwise knockdown of Ube3a and Mef2 in the fly resulted in antagonistic genetic interaction with ameliorated eye and wing phenotypes. This might be explained by knockdown of Ube3a leading to decreased ubiquitination and degradation of Mef2. Subsequently increased Mef2 levels might result in a partial compensation of the Mef2 knockdown phenotype. All other combinations resulted in synergistic genetic interaction, i.e. more severe phenotypes. This would be in line with a) overexpression of Ube3a resulting in increased ubiquitination and degradation of Mef2 and thus in a further decrease of its already low knockdown levels and b) knockdown of Ube3a resulting in decreased ubiquitination and degradation of Mef2 and thus in increased Mef2 abundance even above its overexpression levels. The observed interaction of MEF2C and UBE3A in co-immunoprecipitation experiments supports such a link. However, specific experimental proof of UBE3A regulating MEF2C in an ubiquitin-dependent fashion is still required.

By using Drosophila melanogaster as a model to screen for genetic interactions and by subsequent co-immunoprecipitation in a human cell line, we identified a robust interaction between UBE3A/Ube3a and MEF2C/Mef2. These molecular commonalities might contribute to the clinically overlapping features of the associated disorders.

Material and Methods

Drosophila lines and conditions

Drosophila orthologues of TCF4 (daughterless [CG5102]), ZEB2 (Zfh1 [CG1322]), MEF2C (Mef2 [CG1429]), ATRX (XNP [CG4548]), and UBE3A (Ube3a [CG6190]) were identified with DIOPT55. Manipulation of Da/TCF4 in Drosophila has been established as a specific model for PTHS previously56.

All RNAi lines and the respective control (VDRC no. 60000) were obtained from the Vienna Drosophila Resource Center (VDRC)57. GAL4-driver stocks and transgenic lines for overexpression of Zfh1 and XNP, respectively, were obtained from the Bloomington Drosophila Stock Center. UAS-Mef2 was obtained from the Zurich ORFeome Project (FlyORF)31, and the UAS-Da line was a gift from Pascal Heitzler (IBMP Strasbourg). For generation of the UAS-Ube3a lines, the gene was amplified from fly cDNA (forward: 5’-GTAAAGTGCGCAGATTTCAGC-3’, reverse: 5’-GGTATCAGTTCCAGATGACAGAC-3’) and cloned into a pUAST fly expression vector58. After verification of the sequence, the vector was sent to BestGene Inc. for the creation of a stable transgenic line. For a complete list of used Drosophila lines see Supplementary Table S1. All overexpression lines were isogenized to the VDRC 60000 control by backcrossing for at least seven generations. Double-transgenic fly lines were generated using a double balancer line (Kr/CyO;D/TM6C) (Supplementary Table S2). Overexpression or RNAi-mediated knockdown was induced with the UAS-Gal4 system58 and confirmed by quantitative RT-PCR (Supplementary Fig. S1). In the text and figures, RNAi-lines are referred to as KD_gene, and UAS-lines as OE_gene. Flies were maintained on standard food, containing cornmeal, sugar, agar and yeast at 25 °C and bred at 28 °C because of temperature sensitivity of the UAS-GAL4 system59.

Lethality and morphology analysis

RNAi- and transgenic overexpression lines were crossed to five different driver lines: GMR-GAL4 (eye), ms1096-GAL4 (wing), repo-GAL4 (glia cells), elav-GAL4 (pan-neuronal) and Actin-GAL4 (II, ubiquitous). Resulting offspring were counted and analysed with a Carl Zeiss™ 2000C stereo microscope for gross morphological abnormalities. If the driver line contained a balancer chromosome, offspring with balancers were counted, too. In crosses with double OE/KD constructs retaining balancers, the expected ratio of balancer to non-balancer offspring may deviate from 50%. Wings and eyes were analysed per fly, and a phenotype was counted when it occurred in at least one wing or eye per fly, respectively. Wings were visually evaluated for parameters such as shape, degree of unfolding and wing vein structure. Eyes were visually evaluated for parameters such as ommatidial structure, bristles and gross size and shape. If applicable, p-values were determined using Fisher’s Exact test.

Climbing and bang sensitivity assays

Climbing behaviour and bang sensitivity was performed as described elsewhere60 upon pan-neuronal manipulation (elav-GAL4). In brief, offspring were collected within 72 h of eclosion under CO2 anaesthesia in groups of ten (5 males, 5 females, at least 40 flies tested per genotype). After 24 h, flies were transfered to the testing vial, tapped to the bottom and filmed for 30 s while climbing up. Time until 70% of the flies had crossed line at 8.8 cm height was measured from the videos. If less than 70% of flies from a vial managed to cross the line within the videotaped interval, time was considered to be 30 s. P-values were determined using the Wilcoxon-Mann-Whitney test, and Bonferroni correction was applied for multiple testing. For testing bang sensitivity, flies were vortexed for 10 s and filmed while recovering. The fraction of flies within a vial displaying spasms 5 s after vortexing was determined from the videotapes.

Analysis of neuromuscular junctions (NMJs) from L3 larvae

Analysis of type 1b neuromuscular junctions (NMJs) of muscle 4 was performed as previously described60. Male L3 non-GFP larvae upon pan-neuronal manipulation (elav-GAL4/CyO-GFP;elav-GAL4) were dissected in PBST, fixated in 4% paraformaldehyde and stained with nc82 and anti-discs large antibodies (both from the Developmental Studies Hybridoma Bank, Iowa City, IA). Secondary antibodies used were Alexa 488 labeled anti-mouse antibody and the Zenon™ Alexa Fluor™ 546 Mouse IgG1 Labeling Kit (ThermoFisher Scientific). Images were taken with a Zeiss Axio Imager Z2 microscope in z-stacks and analysed in ImageJ61. NMJ area and length, as well as the number of active zones, synaptic islands, branches, and synaptic boutons were determined from the image stacks. Per genotype, at least 11 NMJs from at least four independent larvae were analysed. P-Values were determined using the Wilcoxon-Mann-Whitney test, and Bonferroni correction was applied for multiple-testing. Upon Mef2 overexpression, we observed an additional signal in cross-segmental neurons with Dlg staining (red channel). This signal was also present in the parental Mef2 line, in another line from the same background, and was also present without Dlg staining (Supplementary Fig. S8). It therefore most likely represents a background signal from the RFP gene under control of the artificial 3xP3 promoter present in the Fly ORF lines.

RNA samples

For RNA sampling from flies, whole larvae (~5), adult flies (~4), heads (~10) or larval brains (~40) were collected and frozen at −80 °C for at least one hour. Total RNA was isolated with the RNeasy Lipid Tissue Mini Kit (Qiagen) using TRIzol™ (ThermoFisher Scientific) instead of QIAzol and QIAshredder columns (Qiagen) for homogenization. DNAse digestion was performed on-column with the RNAse-free DNAse kit (Qiagen). Reverse transcription of RNA into cDNA was performed using the SuperScript™ II reverse transcriptase (ThermoFisher Scientific). RNA from HEK293 cells was isolated using the RNeasy Minikit. DNAse digestion and reverse transcription was performed as described above.

Quantitative reverse transcriptase PCR (quantitative RT-PCR)

Expression analysis was performed using the ABsolute QPCR SYBR Green ROX Mix (ThermoFisher Scientific) and (whenever possible) exon spanning primers (sequences available on request) on a QuantStudio™ 12 K Flex System (Life Technologies). Reactions were performed in quadruplicates, and Ct values were normalized to those of the endogenous controls Actin or Tubulin for Drosophila experiments or B2M for experiments on human cells. Relative expression levels were obtained using the ∆∆Ct method with isogenic background lines (Drosophila) or cells transfected with scrambled siRNA as references. Results were confirmed in at least a second biological replicate.

Immunofluorescence

Expression plasmids expressing human MEF2C and UBE3A and respective negative control plasmids were obtained from Sino Biologicals (MEF2C-HA: HG12320-CY, UBE3A-Myc:HG11130-CM, pCMV-Myc:CV014 and pCMV-HA:CV013) and used for transient transfection. Transfected HeLa cells were grown on poly-lysine coated coverslips, fixated with 4% paraformaldehyde in PBS for 10 minutes and stained with anti-Myc (M4439, Sigma-Aldrich) and anti-HA (H6908, Sigma-Aldrich) and with Alexa Fluor™ 488 goat anti–mouse and Alexa Fluor™ 488 donkey anti–rabbit antibodies (A11001 and A10040, Thermo Fisher). Nuclei were counterstained with DAPI (Serva). Images were taken with a Zeiss Axio Imager Z2 Apotome microscope with a 63x objective and analyzed in ImageJ.

Co-Immunoprecipitation

HEK293 cells were transiently transfected with a combination of UBE3A-Myc and MEF2C-HA or the respective negative controls and harvested 48 h post transfection. Cells were scraped from the culture dish in lysis buffer (100 mM TRIS-HCl pH8, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100). Immunoprecipitation was performed with Protein A Mag Sepharose bead suspension (GE Healthcare), incubated with the sample and anti-Myc or anti-HA antibodies (M4439 or H6908, Sigma-Aldrich) at 4 °C overnight. Subsequently, beads were washed in lysis buffer, and samples were eluted with 1x Lämmli buffer.

Proteins were separated in stain-free 4–20% Mini-PROTEAN® TGX™ Precast Protein Gels (Bio-Rad), blots were stained with anti-Myc and anti-HA antibodies and imaged on a ChemiDoc™ Touch Imaging System (Bio-Rad).

siRNA knockdown

Two different siRNAs against UBE3A (Qiagen) were transiently transfected into HEK293 cells using jetPrime (Polyplus) according to the manufacturer’s instructions. 48 h post transfection, RNA for expression analysis (see description above), was harvested.