2dx—User-friendly image processing for 2D crystals

https://doi.org/10.1016/j.jsb.2006.07.020Get rights and content

Abstract

Electron crystallography determines the structure of two-dimensional (2D) membrane protein crystals and other 2D crystal systems. Cryo-transmission electron microscopy records high-resolution electron micrographs, which require computer processing for three-dimensional structure reconstruction. We present a new software system 2dx, which is designed as a user-friendly, platform-independent software package for electron crystallography. 2dx assists in the management of an image-processing project, guides the user through the processing of 2D crystal images, and provides transparence for processing tasks and results. Algorithms are implemented in the form of script templates reminiscent of c-shell scripts. These templates can be easily modified or replaced by the user and can also execute modular stand-alone programs from the MRC software or from other image processing software packages. 2dx is available under the GNU General Public License at 2dx.org.

Introduction

Structural biology of membrane proteins is of central importance for health, disease, and the development of new drugs. Membrane proteins represent the majority of today’s drug targets in pharmaceutical research. Nevertheless, the PDB database contains only a few hundred membrane protein structures, only a third of which can be considered unique conformations. Compared with the wealth of knowledge on the structure and function of soluble proteins, the low number of determined membrane protein structures stands in stark contrast to their biological importance.

Membrane protein structure determination faces several technical hurdles. Difficulties in over-expression, non-destructive detergent solubilization and gentle purification limit the amount of membrane protein sample available for structural studies. Structure determination by X-ray diffraction (XRD)1 of three-dimensional (3D) crystals, nuclear magnetic resonance (NMR), and cryo-electron microscopy (cryo-EM) of two-dimensional (2D) crystals has revealed an amazing array of structural concepts and mechanisms that nature employs to solve the challenging tasks that membrane proteins perform. Recent highlights include the 1.35 Å structure by XRD of the ammonium channel AmtB (Khademi et al., 2004), the structure of the waterchannel Aqp0 from cryo-EM at 1.9 Å and XRD at 2.2 Å resolution (Gonen et al., 2005, Harries et al., 2004), and the structure of Mistic (Roosild et al., 2005) by NMR (Wüthrich, 1998), to name a few.

Electron crystallography uses cryo-electron microscopy to study the structure of membrane proteins that are reconstituted into phospholipid bilayers and laterally crystallized into 2D membrane protein crystals. Atomic models for seven membrane proteins and tubulin have been determined by electron crystallography: BR (Henderson et al., 1990) LHCII (Kühlbrandt et al., 1994), AQP1 (Murata et al., 2000, Ren et al., 2001), nAChR (Miyazawa et al., 2003), AQP0 (Gonen et al., 2004, Gonen et al., 2005), AQP4 (Hiroaki et al., 2006), and MGST1 (Holm et al., 2006), and Tubulin (Nogales et al., 1998). In addition, several low-resolution structures of transporters, ion pumps, receptors and membrane bound enzymes, that reveal secondary structural motifs such as transmembrane helices are likely to produce atomic models in the near future (e.g., Hirai et al., 2002, Schenk et al., 2005, Kukulski et al., 2005, Tate et al., 2003, Vinothkumar et al., 2005, Aller and Unger, 2006).

The crystallization of membrane proteins in a 2D array within the lipid bilayer represents a valuable alternative route for structure determination. Electron Crystallography has matured into a methodology that allows the determination of membrane protein structures at a resolution of 3 Å or better (e.g., Grigorieff et al., 1996, Mitsuoka et al., 1999, Gonen et al., 2005). 2D membrane crystals offer the possibility of assessing membrane-inserted protein conformations. Existing 2D crystals can be incubated with ligands or other protein binding partners, or they can be exposed to different buffer conditions, and the structure of the complex or altered conformation can then be studied by electron diffraction. However, electron crystallography remains a labor-intensive method: beam-induced charging and/or drumhead-type movement of tilted samples in the electron microscope still affect the success rate for recording high-resolution images—despite recent advances though the use of the SpotScanning method (Downing, 1991) and/or the sandwich sample preparation method (Gyobu et al., 2004). During the screening of crystallization conditions, high-resolution data collection or computer image processing, the lack of automation also requires time-intensive operator interaction.

Computer image processing of electron crystallography data in almost all the aforementioned cases has, to date, been performed by the “MRC programs” for image processing (Crowther et al., 1996). These “MRC programs” are a compilation of individual programs, most written in Fortran-77, that were designed to process images of two-dimensional crystals as well as electron diffraction patterns (Unwin and Henderson, 1975, Henderson et al., 1990, Kühlbrandt et al., 1994, Murata et al., 2000). While this software collection offers a vast repertoire of tools for the processing of 2D crystal images, learning how to employ these programs is time-intensive, and the their usage involves a high amount of direct user interaction.

The MRC programs and bsoft programs (Heymann, 2001) are a collection of stand-alone programs written in Fortran-77 or C/C++. These programs need to be executed either manually, one-by-one in a terminal window, or from a shell script. The later has the advantage of facilitated usage, along with high flexibility and adaptability, but maintaining such scripts can be labor intensive. The execution speeds of computational tasks in scripts are slow, and readability of the scripts and interpretation of results in the form of log-files can be difficult.

A number of other software packages exist for the processing of 2D crystal images. Spectra from the Ice package facilitates the usage of the MRC software (Schmid et al., 1993, Hardt et al., 1996). Wilko Keegstra at the University of Groningen, The Netherlands, is currently developing the Groningen Image Processing Package (Grip) that can also interface with the MRC software (unpublished). The Image Processing Library and Toolkit (IPLT) is a new ground-up image processing development for electron crystallography (Philippsen et al., 2003).

We present a new software system, 2dx that is designed for the electron crystallography community. The purpose of this software system is to facilitate and streamline the processing of electron crystallography data, by providing a user-friendly interface, user-guidance throughout data processing, and a high degree of automation. In the current implementation, 2dx utilizes programs from the MRC software, as well as additional stand-alone programs written specifically for interaction with the 2dx environment as well as providing additional functions and features. 2dx is highly dynamic and can easily be used in conjunction with other image processing packages, including IPLT (Philippsen et al., 2003), bsoft (Heymann, 2001), and/or Spider (Frank et al., 1996). 2dx is developed under the Gnu Public License (GPL), and is freely available as open source. 2dx is available at http://2dx.org and runs natively on Mac OSX and Linux/X11 (Linux, IRIX and other Unix variants).

Section snippets

Software design

2dx is a collection of five programs, 2dx_manager, 2dx_image, 2dx_diffraction, 2dx_merger and 2dx_logbrowser (Fig. 1). 2dx_manager assists in the management of an image-processing project, which typically amounts to 3D structure determination of one membrane protein. 2dx_manager maintains control over the existing data (images or diffraction pattern), their parameters (e.g., resolution, sample tilt geometry) and results. 2dx_manager also launches other programs such as 2dx_image and 2dx_

Graphical user interface and work flow

In its current state, the 2dx_manager assists in the generation of a directory structure for a 2D crystal project (Fig. 2). A four-letter project code and the image number of the first non-tilted image are requested, together with a selection of sample tilt ranges that the user intends to use for data collection. The 2dx_manager then initializes the directory structure as reproduced in Fig. 2, to be used in the following conventions: Each 2D crystal image should be processed in its own

Scripting conventions

The entire 2dx_image construction is kept as user-adjustable and flexible as possible. The 2dx_image database named “2dx_image.cfg”, for example, is kept in a self-explanatory, editable text format. A user can easily add or delete variables, define their format (e.g., float, integer, pull-down menu, Boolean switch, etc.), and define the corresponding help information and web-page link. The standard and custom scripts can be modified, extended or replaced by other scripts that might launch other

Implemented algorithms

In the current state of 2dx_image we have provided a collection of standard scripts for the processing of 2D crystal images, as we use them in our laboratory—most of which are based on the MRC programs. We also added functions for automatic lattice determination (Zeng et al., 2006), spot-list determination, and crystal masking, as well as for the determination of the tilt geometry (using ctffind2; Grigorieff, 1998). The need for the determination of the optimal reference patch location is

The quality value QVal

The scripts calculate a single, one-dimensional, value QVal that attempts to describe the quality of the entire image processing. While the IQ-values that were introduced by R. Henderson (Henderson and Unwin, 1975) are defined as a function of the intensity ratio between a specific reciprocal spot and its local background, the QVal addresses the entire image processing phase. QVal is calculated by an empirical formula that combines different performance measures and indicators into a single

Conclusions

2dx is a user-friendly software system for electron crystallography. In its current state the components 2dx_image and 2dx_logbrowser allow the processing of 2D crystal images. Future development for electron diffraction pattern evaluation and 3D merging is under way. 2dx is currently employed to run the “MRC programs” (Crowther et al., 1970), but can be used in conjunction with other systems. While the focus of 2dx lies on user-friendliness, user-guidance, transparency, processing efficiency,

Acknowledgments

This work was supported by the NSF, grant number MCB-0447860 and by the NIH, Grant No. U54-GM074929. We wish to thank Per Bullough, Anchi Cheng, Andreas Engel, Bob Glaeser, Niko Grigorieff, Richard Henderson, Werner Kühlbrandt, Kaoru Mitsuoka, Ansgar Philippsen, Sriram Subramaniam, Vinzenz Unger, Jannet Vonck, and Tom Walz, who generously shared their knowledge and experience in image processing, and we thank Bob Glaeser and Rena Hill for comments on the manuscript. Development of some of the

References (40)

  • W. Kukulski et al.

    The 5 Å structure of heterologously expressed plant aquaporin SoPIP2;1

    J. Mol. Biol.

    (2005)
  • K. Mitsuoka et al.

    The structure of bacteriorhodopsin at 3.0 Å resolution based on electron crystallography: implication of the charge distribution

    J. Mol. Biol.

    (1999)
  • A. Philippsen et al.

    Iplt—image processing library and toolkit for the electron microscopy community

    J. Struct. Biol.

    (2003)
  • A.D. Schenk et al.

    The 4.5 Å structure of human AQP2

    J. Mol. Biol.

    (2005)
  • M.F. Schmid et al.

    SPECTRA: a system for processing electron images of crystals

    Ultramicroscopy

    (1993)
  • C.G. Tate et al.

    Conformational changes in the multidrug transporter EmrE associated with substrate binding

    J. Mol. Biol.

    (2003)
  • P.N. Unwin et al.

    Molecular structure determination by electron microscopy of unstained crystalline specimens

    J. Mol. Biol.

    (1975)
  • J.M. Valpuesta et al.

    Analysis of electron microscope images and electron diffraction patterns of thin crystals of phi 29 connectors in ice

    J. Mol. Biol.

    (1994)
  • S.G. Aller et al.

    Projection structure of the human copper transporter CTR1 at 6-Å resolution reveals a compact trimer with a novel channel-like architecture

    Proc. Natl. Acad. Sci. USA

    (2006)
  • R.A. Crowther et al.

    Three dimensional reconstructions of spherical viruses by fourier synthesis from electron micrographs

    Nature

    (1970)
  • Cited by (160)

    • Structural analysis of the architecture and in situ localization of the main S-layer complex in Deinococcus radiodurans

      2021, Structure
      Citation Excerpt :

      Briefly, for each micrograph the Fast Fourier Transform (FFT), defocus, and astigmatism were calculated. The reciprocal lattice relative to the crystalline regions for each micrograph was determined from the calculated FFT by the algorithms FindLattice and GetLattice (Gipson et al., 2007). For each image, after applying the Constant Transfer Function (CTF) correction, a lattice's spot list was obtained followed by two cycles of unbending and a second CTF correction.

    • CryoEM: a crystals to single particles round-trip

      2019, Current Opinion in Structural Biology
    View all citing articles on Scopus
    View full text