Learning the dynamics of embedded clauses

Recent work by Siegelmann has shown that the computational power of recurrent neural networks matches that of Turing Machines. One important implication is that complex language classes (infinite languages with embedded clauses) can be represented in neural networks. Proofs are based on a fractal encoding of states to simulate the memory and operations of stacks. In the present work, it is shown that similar stack-like dynamics can be learned in recurrent neural networks from simple sequence prediction tasks. Two main types of network solutions are found and described qualitatively as dynamical systems: damped oscillation and entangled spiraling around fixed points. The potential and limitations of each solution type are established in terms of generalization on two different context-free languages. Both solution types constitute novel stack implementations - generally in line with Siegelmann's theoretical work - which supply insights into how embedded structures of languages can be handled in analog hardware.

GAIP participates in budding of membrane carriers at the trans-Golgi network

Galpha interacting protein (GAIP) is a regulator of G protein signaling protein that associates dynamically with vesicles and has been implicated in membrane trafficking, although its specific role is not yet known. Using an in vitro budding assay, we show that GAIP is recruited to a specific population of trans-Golgi network-derived vesicles and that these are distinct from coatomer or clathrin-coated vesicles. A truncation mutant (NT-GAIP) encoding only the N-terminal half of GAIP is recruited to trans -Golgi network membranes during the formation of vesicle carriers. Overexpression of NT-GAIP induces the formation of long, coated tubules, which are stabilized by microtubules. Results from the budding assay and from imaging in live cells show that these tubules remain attached to the Golgi stack rather than being released as carrier vesicles. NT-GAIP expression blocks membrane budding and results in the accumulation of tubular carrier intermediates. NT-GAIP-decorated tubules are competent to load vesicular stomatitis virus protein G-green fluorescent protein as post-Golgi, exocytic cargo and in cells expressing NT-GAIP there is reduced surface delivery of vesicular stomatitis virus protein G-green fluorescent protein. We conclude that GAIP functions as an essential part of the membrane budding machinery for a subset of post-Golgi exocytic carriers derived from the trans-Golgi network.