Macroautophagy is morphologically characterized by autophagosome formation. Autophagosomes are double-membraned vesicles that sequester cytoplasmic components for further degradation in the lysosome. Basal autophagy is paramount for intracellular quality control in post-mitotic cells but, surprisingly, the number of autophagosomes in post-mitotic neurons is very low, suggesting that alternative degradative structures could exist in neurons. To explore this possibility, we have examined neuronal subcellular architecture by performing three-dimensional (3D) electron tomography analysis of mouse brain tissue that had been preserved through high-pressure freezing. Here, we report that sequestration of neuronal cytoplasmic contents occurs at the Golgi complex in distinct and dynamic structures that coexist with autophagosomes in the brain. These structures are composed of several concentric double-membraned layers that appear to be formed simultaneously by the direct bending and sealing of discrete Golgi stacks. These structures are labelled for proteolytic enzymes, and lysosomes and late endosomes are found in contact with them, leading to the possibility that the sequestered material could be degraded inside them. Our findings highlight the key role that 3D electron tomography, together with tissue rapid-freezing techniques, will have in gaining new knowledge about subcellular architecture.

Summary We have developed a software tool to improve the image quality in focused ion beam–scanning electron microscopy (FIB–SEM) stacks: PolishEM. Based on a Gaussian blur model, it automatically estimates and compensates for the blur affecting each individual image. It also includes correction for artifacts commonly arising in FIB–SEM (e.g. curtaining). PolishEM has been optimized for an efficient processing of huge FIB–SEM stacks on standard computers. Availability and implementation PolishEM has been developed in C. GPL source code and binaries for Linux, OSX and Windows are available at http://www.cnb.csic.es/%7ejjfernandez/polishem. Supplementary information Supplementary data are available at Bioinformatics online.

Neurons rely on a precise spatial and temporal control of protein synthesis due to their highly polarized morphology and their functional singularities. Consequently, alterations in protein translation have been widely related to the development and progression of various neurological and neurodegenerative disorders, including Huntington's disease. Here we explored the architecture of polysomes in their native brain context by performing 3D electron tomography of striatal tissue derived from a knock-in mouse model of the disease. Results showed a progressive remodelling towards a polysomal compacted architecture that parallels in time the emergence and progression of symptoms in the mouse model. The aberrant architecture is compatible with ribosome stalling phenomena and, in fact, we detected an increase in the expression of the stalling release factor eIF5A2. Polysomal sedimentation gradients showed significant excess in the accumulation of free 40S ribosomal subunits in heterozygous striatal samples. Overall the results indicate that changes in the architecture of the protein synthesis machinery might be at the basis of translational alterations associated to Huntington's disease and open new avenues for understanding disease progression.