This cover image shows a reconstruction of part of the endoplasmic reticulum (ER) by using a novel process of fitting a smooth surface through single-molecule localizations. This approach is described in our paper in this issue of Biophysical Journal, and allows us to recover a realistic approximation of the ER from singlemolecule localizations and, in turn, use this to understand changes in structure at the nanometer scale. The image was composed by overlaying three screen captures from our super-resolution visualization software, PYthon Microscopy Environment (PYME) Visualize, and by using gradient transparency masks in Adobe Illustrator to create smooth transitions. From the left to right, the image changes from localizations to a wireframe mesh to a smooth membrane. The surface does not follow the exact locations of all points, but instead represents an attempt to have high fidelity to the underlying structure sampled by these localizations. Localizations are Gaussian distributed about the membrane, indicating a good fit to the mean of the positions measured along a tubule.
The surfaces reconstructed by using our new algorithm are of similar quality to surfaces reconstructed from 3D electron microscopy (EM) images (e.g., serial-block face or tomography). Unlike in EM, related proteins, such as ER structural proteins, can be imaged in a second color channel, and we can examine their relationship with membrane curvature, genus, and connectivity. This can be used to quantify the distribution of structural proteins along the ER, the area of ER-Golgi contact sites, and myriad other fascinating biological questions that require quantification at the nanoscale. A first application of our method to a concrete biological problem, in which we investigate the influence of the curvature-regulating protein reticulon4 on ER morphology, appears in the Journal of Cell Biology.
We are excited to not only provide a fluorescence-based alternative to EM imaging for studying membrane shape, but also to build on existing graphics literature and provide a method for handling a new class of noise sources in point cloud reconstruction. Applications of this and other tools in the PYME can be found at https://python-microscopy.org/citations.html.
— Zach Marin, Lukas A. Fuentes, Joerg Bewersdorf, and David Baddeley