User Manual for Tomography-Guided 3D Reconstruction of Subcellular Structures (TYGRESS)


 Recent advances in cryo-electron microscopy (cryo-EM) are paving the way to determining isolated three-dimensional (3D) macromolecular structures at near-atomic resolution using single-particle cryo-electron microscopy (SP-cryo-EM). However, determining the subcellular structures in intact cells and organelles using cryo-electron tomography (cryo-ET) and subtomogram averaging, another cryo-EM technique, with comparable resolution remains a challenge. Current methodologies can only reach a resolution of several nanometers in most samples studied. Here, we introduce a new hybrid method, called Tomography-Guided 3D Reconstruction of Subcellular Structures (TYGRESS) that is able to achieve structural determination of subcellular structures within their natural crowded environment with nanometer-resolution by combining the advantages of cryo-ET and SP-cryo-EM.


Introduction
SP-cryo-EM and cryo-ET both generate three-dimensional (3D) reconstructions of native biological macromolecules under cryogenic conditions. However, whereas SP-cryo-EM can be used to determine the structure of isolated macromolecules with up to atomic resolution (~3Å), the resolution achieved by cryo-ET of intact cells or organelles paired with subtomogram averaging lags approximately one order of magnitude behind (~30Å). Here, we introduce a hybrid-method called "TomographY-Guided 3D REconstruction of Subcellular Structures" (TYGRESS), which combines the advantages of both SPcryo-EM (images with good signal-to-noise ratio/contrast and minimal radiation damage of the sample) and cryo-ET averaging (extraction and 3D alignment of macromolecules contained in a complex cellular sample). Ultimately TYGRESS is a SP-reconstruction, but the parameters from subtomogram averaging are critical to guide particle picking and image alignment -steps that usually prohibit SP-cryo-EM of complex cellular samples.
In this method, a typical high dose (HD) single particle cryo-EM micrograph is acquired (i.e., with an electron dose that is higher than for individual tile series images) immediately prior to a conventional low dose (LD) cryo-ET tilt series. The LD cryo-ET data set are processed, including subtomogram averaging of the particle of interest. The parameters, i.e., the particle position and alignment, are 3 then used to guide the particle picking in the HD single-particle cryo-EM data set and initialize its angular alignment. With this information from cryo-ET, single particle image processing techniques can be applied to subcellular samples, despite the superposition of many structures in the SP projection image.

Hardware requirements and software installation
A computer cluster and/or workstation with GPU with shared storage is recommended for running TYGRESS. Before installation of TYGRESS, please ensure the software packages below are installed by following the installation guide of each package. You can also find the hardware and operating system requirements in the distribution pages of the following software packages.

1: Preparation to run TYGRESS
1.1 To use TYGRESS on the example data set, download and unzip the TYGRESS source code and example data set package (www.tygress.org), and then follow Procedure step 2 ("Preparation of TYGRESS input files" below).
Note TYGRESS is also available on Code Ocean as a compute capsule (https://doi.org/10.24433/CO.2034333.v1). Please read the readme file on the Code Ocean TYGRESS page for more details.
1.2 To use your own data set, please finish the data collection, tomogram reconstruction, and subtomogram averaging before following Procedure 2, below.
1.2.1 Data collection. During data collection, two data sets need to be acquired for each region of interest. (1) First, acquire a 2D image at 0 degree using an electron dose typical for conventional SPcryo-EM (HD image), which is used for final TYGRESS reconstruction; (2), immediately after collecting the HD image, collect a traditional low-dose tilt series (LD images) of the same region, which is used for retrieving the position information for each particle of interest after tomogram reconstruction and subtomogram averaging. Note: to minimize the effects of radiation damage in the final reconstruction, the HD image must be recorded prior to the LD images.

Motion correction.
If movie stacks were taken during the data collection, run the whole-frame motion correction using IMOD scripts.

Alignment of the combined tilt series. Each HD image needs to be inserted into its
corresponding tilt series at the angle where the HD image was taken using the "newstack" command in the IMOD package, resulting in a "combined tilt series". For example, an HD image recorded at 0 degrees should be inserted just before the LD image at 0 degrees. After tilt series alignment using the IMOD package, the combined HD and LD images ensure a common reference frame for the later steps in the TYGRESS procedure.

Tomogram reconstruction.
After alignment, only the LD images are used to generate the tomogram using the IMOD package.
1.2.5 Subtomogram averaging. The subtomogram averaging can be performed using the PEET package in IMOD. Run PEET averaging for each tilt series and all of the particles of interest, respectively.
2.1.1 There is one file named as "TYGRESS_input.txt" in the TYRESS source code folder (as shown below in the example data set), which contains all the parameters to run TYGRESS, such as the work directory, file path, 3D reconstruction parameters, etc. Update these parameters as needed to reflect the location and organization of your files. If you are following along with the example dataset, modify 2.1.2 The file named "DataList.txt" in the TYGRESS source code folder (as shown below in the example data set) has three columns containing the "data set main name" + "data set sub name" + "total .mod number", which you should change as needed.
2.3 Modify the path of your FREALIGN installation in the "mreconstruct_noMask.com" file. Please see the example "mreconstruct_noMask.com" file for details.

3: Run TYGRESS
Open the Matlab interface, and set the TYGRESS source code folder as the work path. Then, in Matlab, open the TYGRESS_GUI (Figure 1). Click the "Tygress parameter file" button, input the "TYGRESS_input.txt" file, click the "Data list" button, and input the "DataList.txt" file. Click "Yes" on the corresponding step to have TYGRESS run that step (Figure 1). All steps should be run in order if it is your first time processing your TYGRESS data. (Note: You can process the steps one by one, or you can run all selected steps together.) Final 3D reconstruction is performed with FREALIGN, and all output is saved under the "Reconstruction" folder. For more information about FREALIGN, please visit (http://grigoriefflab.janelia.org/frealign).

Time Taken
In the example data set, which includes five tilt series, each tilt series includes all the files generated after tomogram reconstruction and subtomogram averaging. The PEET averaging for the whole dataset is also included. Therefore, you can run TYGRESS directly after changing the work directories in the "TYGRESS_input.txt" file. The total TYGRESS processing time from CTF finding to final reconstruction will take less than 30 minutes on a common computer cluster.
Anticipated Results 5.1 *.mod(1) and *.csv(1) are the output files after running PEET for each individual tilt series.
5.2 You can copy the *.mod(1) and *.csv(1) to the Combined_mod_csv folder as *.mod(2) and *.csv (2) to run the PEET for the whole datasets, and then you will get *.csv(3) after PEET. All these .mod and .csv files are obtained before you starting to run TYGRESS.

After you run
Step 1 in the TYGRESS GUI, the "ctffind" folder will be created and the defocus information will be saved in the *.txt(1) file. Moreover, the "HD_particle" folder will be created and all HD images will be saved in its sub_folder.

After you run
Step 2 in the TYGRESS GUI, the program will use the information from *.mod(2) and *.csv(3) to pick particles from the HD images, and then save all picked 2D particles as *.mrc file and save the coordinate parameters of each picked particle as *.txt(2) file, and alignment parameters as *.par file.

After you run
Step 3 in the TYGRESS GUI, the program will align the 2D particles picked and update the *.par file. Note: for further constrained alignment, you need to update your 'sx' (for x, y shift) and 'sa' (for angular shift) values in the "TYGRESS_input.txt" file, use the *.par file created in Step 3 to replace the *.par file generated in Step 2, and re-run Step 3.

After you run
Step 4 in the TYGRESS GUI, the program will generate the 3D reconstruction and save the *.mrc(2) file.