Project

Discipline

Membrane Proteins; 2D crystals; Projective Constraint Optimization; Electron Crystallography; Image Processing; Membrane Proteins

Lead
Membrane proteins are central to health and disease, and represent the major focus of intensive research efforts.The structure of two-dimensionally crystallized membrane proteins can be determined by electron microscopy and image processing. We have developed a software package for this task, called 2dx, which we will here extend to improve resolution and speed of the structure determination.
Lay summary
The structure determination of human membrane proteins from electron microscopy images requires the proteins to be available in well-ordered two-dimensional crystals (side-by-side densely packed in a membrane). These crystals are imaged in the electron microscope from different viewing angles, but the full side-view cannot be obtained due to physical restrictions. Processing of such images allows to determine the 3D structure of the proteins. But the reconstruction so far was only possible if the crystals were extremely well ordered. In addition, the reconstruction usually suffered from a worse resolution in the vertical direction, due to the absence of side-views of the protein crystals.We have created a software package, called 2dx, which allows the user-friendly computer processing of such images.Aims:Here we will extend the 2dx software package by two additional algorithms, with the aim to also allow the 3D structure reconstruction of the membrane proteins from less-well ordered crystals (1), and to strongly improve the resolution in the vertical direction (2). We will approach these by using a Maximum Likelihood (ML)-based algorithm for the first aim, and a so-called projective constraint optimization (PCO) algorithm for the latter. Both are mathematically complex software algorithms, which are in early development phases in several other scientific disciplines. Here we will develop them for membrane protein structure determination.Significance:The software will be available as open-source software in our 2dx package, which now enjoys over 400 external users from life sciences and materials sciences electron microscopy fields (http://2dx.org). As such, the algorithms will immediately be available in a user-friendly and heavily used environment to other research labs.The algorithm development will in addition be of significant value for the field of general protein structure determination by electron microscopy (ML), and for Free Electron Laser data analysis (PCO), since these research fields face similar limitations that can be overcome with these algorithms.

Direct link to Lay Summary

Last update: 21.12.2012

Active participation

Self-organised

Number	Title	Start	Funding scheme

Membrane proteins are central to health and disease. Electron crystallography can deter-mine the high-resolution structure of membrane proteins from membrane-reconstituted and two-dimensionally crystallized protein. The Transmission Electron Microscope (TEM) is used to study membrane proteins in the lipid membrane environment, which is a powerful tool to study the structure and dynamic of membrane proteins, and of membrane protein complexes that still resist high-resolution 3D crystallization or large-quantity expression efforts.Two-dimensional membrane crystals are imaged by cryo-electron microscopy (cryo-EM). The high-resolution 3D membrane protein structure can then be reconstructed by computer im-age processing. Until recently, the only available software for that task was the MRC software (Crowther et al., 1996). Starting in 2003, my group has created a software system, called 2dx, which embeds the MRC software into a user-friendly graphical user interface, and in addition offers a multitude of additional functions that enable full automation of the processing task, user guidance, interactive documentation and help, and allows automated iterative refinement of the processing results. 2dx received good acceptance in and excellent feedback from the electron crystallography community, and we now count more than 240 external users (as of 1/15/2009).We here propose to extend the 2dx software by new algorithms, to increase resolution and speed of the membrane protein structure determination by electron crystallography, and to en-able high-resolution structure determination also from badly ordered 2D crystals. We will imple-ment a single-particle approach for 3D reconstructions from tilted 2D crystal images, and extend this by a maximum likelihood algorithm. We will develop a hybrid input output algorithm imple-mentation to improve the Z resolution of electron crystallography, and to reduce the need to record images of highly tilted 2D crystals. We will also modernize 2dx, by enabling cluster CPU and cluster GPU computing, and interface it with other systems like Eman2 and IPLT.