Journal of Cell Science 118, 1279-1290 2005

A dynamin-3 spliced variant modulates the actin/cortactin-dependent morphogenesis of dendritic spines

Noah W. Gray, Anne E. Kruchten, Jing Chen and Mark A. McNiven

Immature dendrites extend many actin-rich filopodial structures that can be replaced by synapse-containing dendritic spines as the neuron matures. The large GTPase dynamin-3 (Dyn3) is a component of the postsynapse in hippocampal neurons but its function is undefined. Here, we demonstrate that a specific Dyn3 variant (Dyn3baa) promotes the formation of immature dendritic filopodia in cultured neurons. This effect is dependent upon Dyn3 GTPase activity and a direct interaction with the F-actinbinding protein cortactin. Consistent with these findings, Dyn3baa binds to cortactin with a 200% higher affinity than Dyn3aaa, a near identical isoform that does not induce dendritic filopodia when expressed in cultured neurons. Finally, levels of Dyn3baa-encoding mRNA are tightly regulated during neuronal maturation and are markedly upregulated during synaptogenesis. Together, these findings provide the first evidence that an enhanced interaction between a specific Dyn3 splice variant and cortactin modulate actin-membrane dynamics in developing neurons to regulate the morphogenesis of dendritic spines.

enlarge

Fig. 1. The expression of distinct Dyn3 variants induces profoundly different postsynaptic morphologies. (A,C) Neurons transfected with Dyn3aaa-GFP produce numerous mushroom-shaped dendritic spines by 18DIV (arrows in A). (B,D) Neurons expressing Dyn3baa-GFP exhibit morphologically immature dendritic spines at a mature culture age, with long filopodia remaining as the dominant structures instead of spines (arrows in B). (E,F) Although Dyn3aaa- and Dyn3baa-expressing neurons appear different at 18DIV, they exhibit remarkably similar morphologies at 11DIV, when filopodia normally persist. Images in C-F represent CFP fluorescence from Dyn3- spliceoform-co-transfected cells captured by confocal microscopy. Scale bars, 5 μm.

enlarge

Fig. 4. Dyn3baa-GFP induces filopodial outgrowth in non-neuronal cells. COS-7 cells transfected with Dyn3aaa-GFP (A) maintained a normal morphology, with a smooth lamellar perimeter (arrows in A, close-up in A′) and a rounded shape. By contrast, COS-7 cells transfected with Dyn3baa-GFP exhibited radical shape changes (B), with the smooth perimeter often replaced with numerous large membranous protrusions and long filopodia that were remarkably similar to the filopodia in neurons (arrows in B, close-up in B′). Transfection with Dyn3baa(KA)-GFP caused a phenotype resembling Dyn3aaa-GFP transfection, causing cells to maintain their normal smooth perimeter (C). Interestingly, mutant Dyn3baa remained at the membrane perimeter (arrows in C, close-up in C′) but did not induce the shape changes observed with Dyn3baa wild-type transfection. Scale bar, 10 μm.

enlarge

Fig. 5. Dyn3baa directly binds with a greater affinity than Dyn3aaa to the actin-associated protein cortactin. (A) Comparing the known cortactin binding site of Dyn2 with the correlative site in the Dyn3- PRD reveal that these sequences are almost identical. (B) Cortactin directly binds to both Dyn3aaa and Dyn3baa. GST alone failed to pull down the cortactin protein. (C) Dyn3aaa and Dyn3baa directly bind to cortactin in a blot overlay experiment using a range of recombinant Dyn3 protein amounts. (D) Binding curves representing densitometry measurements demonstrating the greater affinity for cortactin of Dyn3baa compared with Dyn3aaa. Results are normalized to the average maximum in each independent experiment and plotted on a scale of 0 to 1. Error bars represent the s.d. (E) Fold differences in binding of cortactin to Dyn3baa compared with Dyn3aaa. Results were from the average of three experiments and error bars represent the s.e.m.

enlarge

Fig. 7. Dyn3 transcripts are differentially regulated during neuronal development. (A) Isolated hippocampal neurons: primers flanking a 30-nucleotide spliced insert in Dyn3 were used in RT-PCR reactions to assess the relative levels of Dyn3aaa and Dyn3baa in maturing cultured neurons at each age as labeled (‘DIV’ means ‘days in vitro’). The Dyn3aaa product is 370 bp, whereas the Dyn3baa product is 400 bp. The PCR product at 430 bp is an RT-PCR artefact representing a heteroduplex of strands from each spliceoform. GAPDH RT-PCR demonstrates the use of equal amounts of template in each reaction. MW, molecular weight markers; Br, brain; –con, negative control. (B) Identical RT-PCR reactions were prepared using template derived from whole rat hippocampus at various developmental stages as listed above (P, postnatal day). GAPDH RTPCR provides evidence for equal transcript copy number in the loading of template for each reaction. (C) RT-PCR using the cultured neuron template set and primers designed to exhibit total levels of all Dyn3 transcripts. (D) RT-PCR using the whole hippocampi template set and primers designed to exhibit total levels of all Dyn3 transcripts.