Order out of disorder: working cycle of an intrinsically unfolded chaperone.

The redox-regulated chaperone Hsp33 protects organisms against oxidative stress that leads to protein unfolding. Activation of Hsp33 is triggered by the oxidative unfolding of its own redox-sensor domain, making Hsp33 a member of a recently discovered class of chaperones that require partial unfolding for full chaperone activity. Here we address the long-standing question of how chaperones recognize client proteins. We show that Hsp33 uses its own intrinsically disordered regions to discriminate between unfolded and partially structured folding intermediates. Binding to secondary structure elements in client proteins stabilizes Hsp33's intrinsically disordered regions, and this stabilization appears to mediate Hsp33's high affinity for structured folding intermediates. Return to nonstress conditions reduces Hsp33's disulfide bonds, which then significantly destabilizes the bound client proteins and in doing so converts them into less-structured, folding-competent client proteins of ATP-dependent foldases. We propose a model in which energy-independent chaperones use internal order-to-disorder transitions to control substrate binding and release.

A Bayesian generative model for learning semantic hierarchies.

Building fine-grained visual recognition systems that are capable of recognizing tens of thousands of categories, has received much attention in recent years. The well known semantic hierarchical structure of categories and concepts, has been shown to provide a key prior which allows for optimal predictions. The hierarchical organization of various domains and concepts has been subject to extensive research, and led to the development of the WordNet domains hierarchy (Fellbaum, 1998), which was also used to organize the images in the ImageNet (Deng et al., 2009) dataset, in which the category count approaches the human capacity. Still, for the human visual system, the form of the hierarchy must be discovered with minimal use of supervision or innate knowledge. In this work, we propose a new Bayesian generative model for learning such domain hierarchies, based on semantic input. Our model is motivated by the super-subordinate organization of domain labels and concepts that characterizes WordNet, and accounts for several important challenges: maintaining context information when progressing deeper into the hierarchy, learning a coherent semantic concept for each node, and modeling uncertainty in the perception process.