The genetic and molecular analyses in this study lead us to propose that modulation of dADAR activity by dFMR1 is important for NMJ synaptic architecture. The epistatic relationship of these two genes, the requirement of RNA editing by dADAR for normal NMJ morphology, and genetic suppression of the dfmr1
lof NMJ defects all support a model in which dFMR1 affects the editing activity of dADAR. Molecular analyses of dADAR substrates support this prediction, as we found that both loss and overexpression of dFMR1 result in changes in editing efficiency in several dADAR-dependent editing sites. While the changes in editing observed in dfmr1
mutant whole larvae are not large (eg: a ~15% change in editing was observed for lap
), they are statistically significant, and we propose that these changes would likely be larger if analyses could be performed using mRNA prepared from isolated neurons or synapses rather than whole larvae. In addition, despite a few transcripts that are highly edited throughout development, dADAR function during developmental stages is relatively low compared to its high editing activity in pupal and adult stages29
. Therefore, we cannot rule out a larger effect of dFMR1 levels on dADAR substrates with lower efficiency editing sites.
In considering how dFMR1 affects editing, an important clue might come from the fact that both dFMR1 and dADAR are RNA binding proteins that associate with secondary and higher order RNA structures. FMRP can bind to two separate complex RNA structures that are believed to allow for specificity of FMRP-associated transcripts: the RGG domain in the C-terminus of FMRP protein interacts with an intramolecular G quartet stem loop RNA structure whereas the KH2 domain associates with a complex tertiary kissing complex RNA structure40,41
. Similarly, the dADAR family of proteins contains several double-stranded RNA binding domains (dsRBD) and requires duplex RNA structures to identify, bind to, and function on its target RNAs21
. The RNA structure required for dADAR activity, however, can vary from a simple hairpin to complex pseudoknot structures.
Additionally, our immunoprecipitation experiments indicate that dFMR1 and dADAR associate on common RNA targets. The dADAR:dFMR1 biochemical interaction was reduced through both a decrease in the amount of RNA in our lysates using RNase A as well as by mutating the KH domains of dFMR1, suggesting to us that the ability for dFMR1 to bind to RNA plays an important role in its association with dADAR. Molecular analyses of lap
in the dFMR1 RNA binding-mutants further support this theory, albeit, differential effects were observed with respect to the two transcripts analyzed. It is possible that dFMR1 associates with these two particular transcripts via different RNA binding motifs. Although the analogous I307N mutation in mammals reduces FMRP’s ability to associate with both poly(U)-rich sequences and large RNP complexes42,43
, FMRP can still bind to RNA, including transcripts containing G-quartet structures, through an intact RGG RNA binding motif42
. Therefore, we propose that the I244N and I307N mutations in dFMR1 reduce particular dFMR1:dADAR complexes associating with certain edited transcripts while concurrently enriching for dFMR1:dADAR complexes associating with the dFMR1 RGG box. Further studies delving into the importance of each RNA binding domain in both dFMR1 and dADAR will give more insight into the biochemical and functional interaction between these two proteins.
Based on our results, we predict that dFMR1 and dADAR can associate in a common complex and converge on similar RNA substrates. Because the effect that dFMR1 has on the editing efficiency is context dependent, we propose that the association of dFMR1 with dADAR has no net positive or negative effect on the editing activity of dADAR, but instead maintains a balance of dADAR activity. At sites that demonstrate enhanced editing in the presence of dFMR1, dFMR1 could promote editing by either recruiting dADAR to the site via its own RNA binding activity, or it could help form and/or stabilize RNA structures that create a site for editing by dADAR. At sites that are negatively affected by the presence of dFMR1, we propose that the RNA binding activity of dFMR1 interferes with the formation of a substrate for dADAR (Supplemental Fig. 6
Our analyses revealed several transcripts whose level of editing is regulated by the interaction between dFMR1 and dADAR, however at this time we do not know how many such transcripts are key to the proper formation of the NMJ. Although many identified dADAR targets encode for proteins that function in synaptic transmission at the NMJ34
and mutations in several dADAR substrates (eg: syt-1
) affect NMJ synaptic architecture and/or function44–46
, how editing is affecting the function of most of these gene products remains unknown. It is also important to note that a role for dFMR1 in translational regulation is already proposed to be important for proper NMJ development through its interaction with the microtuble-associated protein (MAP1B) homolog futsch26
. Collectively, these studies suggest that both dADAR and dFMR1 play multifaceted roles at the NMJ.
In summary, we demonstrate that dFMR1 physically and genetically interacts with dADAR-dependent RNA editing. This is the first study to report a disease-associated protein that associates with and modulates A-to-I RNA editing. In addition, our findings introduce a novel function for FMRP with respect to neuronal architecture and expands FMRP’s predicted role as a translational regulator. Understanding all the mechanisms by which FMRP functions to regulate synaptic development and function is essential to better understand the pathogenesis of the FXS symptoms, and consequently can lead to effective therapeutic treatments for people afflicted with this disease.