Cryogenic x-ray diffraction microscopy utilizing high-pressure cryopreservation

We present cryo x-ray diffraction microscopy of high-pressure-cryofixed bacteria and report high-convergence imaging with multiple image reconstructions. Hydrated D. radiodurans cells were cryofixed at 200 MPa pressure into $\sim 10\text{−}\mu \mathrm{m}$-thick water layers and their unstained, hydrated cellular environments were imaged by phasing diffraction patterns, reaching sub-30-nm resolutions with hard x-rays. Comparisons were made with conventional ambient-pressure-cryofixed samples, with respect to both coherent small-angle x-ray scattering and the image reconstruction. The results show a correlation between the level of background ice signal and phasing convergence, suggesting that phasing difficulties with frozen-hydrated specimens may be caused by high-background ice scattering.

Figure 1

(a) A schematic of high-pressure and ambient-pressure cryopreservation in the water-ice phase diagram. ${\mathrm{T}}_{\mathrm{M}}$ stands for the melting temperature of water and ${\mathrm{T}}_{\mathrm{H}}$ stands for the homogeneous nucleation temperature, which is the low-temperature limit of supercooled water (adapted from Ref. [25]). The inset shows characteristic wide-angle x-ray scattering, measured from $\sim 10\text{−}\mu \mathrm{m}$-thick ice layers, of LDA (low density amorphous ice) with peak scattering at 3.6 Å and HDA (high density amorphous ice) with peak scattering at 3.0–3.3 Å [26], of the current case at 3.1 Å. (b) Coherent x-ray diffraction patterns: (i–iii) from high-pressure-cryofixed D. radiodurans bacteria and (iv–vi) from ambient-pressure-cryofixed ones. The total exposure time is about one hour per data set. The color scales indicate the intensity in photon counts per second. The black scale bars (bottom right) in all images indicate spatial frequency = $4\mu {\text{m}}^{-1}$. The missing data region behind the beam stop consists of $7\times 9$ pixels. Oversampling ratios were estimated to be $\sim 15$ for panels i and vi and $\sim 25$ for the others (ii–v).

Figure 2

Coherent small-angle x-ray scatterings of ice layers. Panel (a) shows the power spectra, in photon counts per second, of ice signals, (i) from a high-pressure-cryofixed ice layer and (ii–iv) from ambient-pressure ones. The total exposure time for ten sampling points per spectrum was 30 s with an x-ray flux of ${10}^9$ photons per second. The coherent scattering contrast images of corresponding ice layer is shown in panel (b). The area was scanned with $5–8 \mu \mathrm{m}$ step size of the same x-ray illumination as in panel (a). The scale bars indicate $20 \mu \mathrm{m}$ in size.

Figure 3

Reconstruction images of high-pressure-cryofixed D. radiodurans bacteria. The images show two bacteria samples: Panel $\left(\mathrm{a}{}^{\prime }\right)$ is a magnified image of the sample from panel (a) in the region marked with the black square, and panels (b) and (c) are from the second bacteria, imaged at a 15-degree rotation from each other. Color scales display between 10% and 100% of maximum pixel value in each image. Here, the pixel values are normalized to photon counts per ten-minute exposure time. Sample sizes are $\sim 1.5$ to $2 \mu \mathrm{m}$ (the scale bars indicate 500 nm). Reconstruction images also show some low-density objects around samples, which are thought to be debris in the culture solutions. Each image is the average from 30 reconstruction trials with a total of 3000 [for panels (b) and (c)] and 6000 [for panel (a)] iterations.

Figure 4

PRTF for high-pressure-cryofixed bacteria reconstructions. Curve “A” corresponds to Fig. 3 reconstruction, “B” for Fig. 3, and “C” for Fig. 3, respectively. Following the conservative measure for the resolution cutoff to be where PRTF falls below 0.5 [6], the resolutions of the reconstructions are estimated to be sub-30 nm.