You ask a very good series of questions. The specificity of intersectin for dynamin is the easiest one. Intersectin has five SH3 domains. SH3B and SH3D do not interact with dynamin so we have a very nice internal control for in vitro specificity. Intersectin and dynamin co-ip., as do Dap160 (a Drosophila homologue of intersectin and dynamin, Roos and Kelly, JBC). So far, dynamin and synaptojanin are the only proline-rich proteins demonstrated to bind to intersectin.
The rationale for different proline-rich partners for dynamin is difficult to say. Dynamin has an extensive proline-rich C-terminus (as does synaptojanin) with multiple consensus SH3 domain-binding sites. Amphiphysin I and II are certainly important interacters as they are co-localized with dynamin in nerve terminals, they co-ip with dynamin, and regulate dynamin oligomerization. Endophilin may not be an important dynamin interacter as it i.ps. with dynamin very weakly if at all, and the in vitro affinity is much less than the affinity of endophilin for synaptojanin. Grb2 interactions are likely less relevant, especially with dynamin 1 in neurons. Different SH3 partners could control different aspects such as dynamin targeting, oligomerization, enzymatic activity. They may also only be important in specific tissues and stations of the vesicular trafficking pathway, or at specific developmental stages.