Benedikt Haas (Berlin / DE), Katia March (Paris / FR), Ondrej Krivanek (Berlin / DE; Tempe, AZ / US; Kirkland, WA / US), Christoph T. Koch (Berlin / DE), Zdravko Kochovski (Berlin / DE), Peter Rez (Tempe, AZ / US)
Abstract text (incl. figure legends and references)
INTRODUCTION
Radiation damage fundamentally limits the attainable spatial resolution in the electron microscopy of biological specimens. To determine protein structures with the resolution needed to understand biological function it is necessary to average over multiple copies of a molecule of interest. The need for averaging limits studies of how macromolecules interact e.g. in an organelle. Traditionally, the effects of radiation damage have been assessed by measuring spot fading in crystals [1], or by monitoring mass loss from EELS spectra [2].
OBJECTIVES
It is highly desirable to understand the mechanisms of radiation damage, which bonds are breaking, and when they can reform. Recent developments make it possible to record vibrational spectra in the electron microscope - equivalent to IR but at the nanometer scale. We acquired data on guanine and vitreous ice to quantitatively measure the structural changes through the vibrational signal from different chemical bonds as a function of electron exposure.
MATERIALS & METHODS
We used a Nion HERMES microscope at 60kV with Dectris ELA direct detector attached to the IRIS spectrometer. We used guanine fish scales [3] and vitreous ice from plunge freezing. Spectrum images were recorded by repeatedly scanning a defocused probe over an area (typically 100x100 nm²). The spectra from each map were aligned and summed, yielding average spectra vs. fluence at below 8 meV energy resolution. The intensities of peaks corresponding to different bonds were extracted by fitting power laws over multiple windows.
RESULTS
Spectra of guanine and vitreous ice are given in Fig. 1 and show vibrational modes linked to different bonds. Fig. 2 shows different peak heights (normalized to zero loss) versus fluence. In the early stages, the C=O intensity from guanine at room temperature does not change while the XH intensity is drastically, cf. Fig. 2 (a). This indicates that the CH and NH bonds break first and is consistent with the observation of hydrogen bubbles [4]. At higher electron exposure up to 1000 e-/ Å2 the C=O bonds are broken as seen in Fig. 2 (b). Eventually, a steady state of broken bonds is reached but mass loss from sputtering continues (increased zero-loss intensity, not shown). Fig. 2 (c) shows that cooling to liquid N2 temperatures reaches the steady state earlier but does not change the general behavior. For vitreous ice, it appears from Fig. 2 (d) that single OH bonds break first releasing free H, as seen from the decrease in the OH stretch peak.
CONCLUSION
At a low electron fluence of up to 100 e-/ Å2 CH, NH and OH bonds are the first to break. They presumably release hydrogen that can accumulate as bubbles. As the electron exposure increases other bonds such as the C=O in guanine are broken. However, many of these bonds are also being reformed as damage proceeds. We are now studying hydrogen exchanges via deuteration.
[1] H. Stark et al, Ultramicr. 63, 75 (1996).
[2] P. Li and R.F. Egerton, Ultamicr. 101, 161 (2004).
[3] P. Rez et al, Nat. Commun. 7, 10945 (2016).
[4] R.D. Leapman and S. Sun, Ultramicr. 59, 71 (1995).
Fig. 1 (a) Background subtracted spectrum of guanine and (b) spectrum of vitreous ice. Variation of peak intensities (normalized to zero-loss) versus fluence for (c) guanine at room temperature and low fluence, (d) increased fluence, (e) cooled and (f) for vitreous ice.