Abstract
Structural studies on living cells by conventional methods are limited to low resolution because radiation damage kills cells long before the necessary dose for high resolution can be delivered. X-ray free-electron lasers circumvent this problem by outrunning key damage processes with an ultra-short and extremely bright coherent X-ray pulse. Diffraction-before-destruction experiments provide high-resolution data from cells that are alive when the femtosecond X-ray pulse traverses the sample. This paper presents two data sets from micron-sized cyanobacteria obtained at the Linac Coherent Light Source, containing a total of 199,000 diffraction patterns. Utilizing this type of diffraction data will require the development of new analysis methods and algorithms for studying structure and structural variability in large populations of cells and to create abstract models. Such studies will allow us to understand living cells and populations of cells in new ways. New X-ray lasers, like the European XFEL, will produce billions of pulses per day, and could open new areas in structural sciences.
Introduction
In structural biology, imaging living cells at high resolution presents a fundamental "catch-22." Achieving sub-nanometer detail requires radiation doses—often exceeding 100 million Grays—that far exceed the lethal threshold for biological life (typically a few hundred Grays). Consequently, most high-resolution cellular knowledge is derived from dead, fixed, or cryogenically frozen samples.
This study addresses the critical need for high-resolution data from cells in their native, living state by leveraging X-ray Free-Electron Lasers (XFELs). By capturing a "molecular-level snapshot" before the sample is vaporized, researchers can "outrun" radiation damage. However, the massive volume of data generated requires standardized, large-scale datasets to facilitate the development of new reconstruction algorithms.
Experimental Methodology
The researchers utilized the Linac Coherent Light Source (LCLS) and the CAMP instrument to perform "diffraction-before-destruction" experiments. The core of this technique relies on femtosecond pulses (~70 fs), which are shorter than the time required for radiation-induced damage (such as atomic motion or plasma formation) to manifest structurally.
To minimize background noise, live cells were aerosolized in helium and injected directly into a vacuum chamber using an aerosol sample injector. This isolation ensures high-contrast diffraction patterns. The automated preprocessing software, Cheetah, was employed to filter hundreds of thousands of "shots" to identify "hits"—frames where the laser successfully intersected a cell.
Data Description
The study introduces two primary data records, totaling 199,000 diffraction patterns, now available in the Coherent X-ray Imaging Data Bank (CXIDB):
- Data Record 1 (Cyanobium gracile): Comprises 192,370 hits collected over 77 minutes. These cells are small (0.25–0.4 μm diameter) and robust.
- Data Record 2 (Synechococcus elongatus): Comprises 6,394 hits collected over 9 minutes. These cells share a similar diameter with C. gracile but exhibit greater length.
The data is provided in raw (XTC) and minimally preprocessed (CXI/HDF5) formats, including essential calibration and geometry files.
Discussion and Results
The fundamental insight of XFEL imaging is the ability to gather data from a sample being destroyed by utilizing ultra-fast timing to "freeze" atomic motion. This shift in methodology allows for the analysis of structural variability across large populations rather than focusing on a single specimen, enabling the identification of rare events in the cell cycle.
Key results include:
- Scale: This is the largest freely available X-ray diffraction dataset for living cells.
- Resolution: Recorded signals extend beyond 4 nm, significantly surpassing conventional optical microscopy.
- Efficiency: A hit ratio of 41% was achieved for C. gracile, validating the efficiency of the aerosol injection system.
- Validation: Previous successful reconstructions using subsets of this data confirm that the "live" state is maintained during the pulse.
Conclusion and Future Directions
A primary limitation remains the "depth information gap," as single 2D diffraction patterns provide limited 3D context. Future work will focus on developing algorithms to combine thousands of patterns into 3D models and handling structural variability within populations. As next-generation facilities like the European XFEL begin producing billions of pulses daily, the automated analysis tools supported by this dataset will become increasingly vital for moving toward abstract models of cellular interaction.
Summary updated