Quantitative nanoscale water mapping in frozen-hydrated skin by low-loss electron energy-loss spectroscopy.

Spatially resolved low-loss electron energy-loss spectroscopy (EELS) is a powerful method to quantitatively determine the water distribution in frozen-hydrated biological materials at high spatial resolution. However, hydrated tissue, particularly its hydrophilic protein-rich component, is very sensitive to electron radiation. This sensitivity has traditionally limited the achievable spatial resolution because of the relatively high noise associated with low-dose data acquisition. We show that the damage caused by high-dose data acquisition affects the accuracy of a multiple-least-squares (MLS) compositional analysis because of inaccuracies in the reference spectrum used to represent the protein. Higher spatial resolution combined with more accurate compositional analysis can be achieved if a reference spectrum is used that better represents the electron-beam-damaged protein component under frozen-hydrated conditions rather than one separately collected from dry protein under low-dose conditions. We thus introduce a method to extract the best-fitting protein reference spectrum from an experimental spectrum dataset. This method can be used when the MLS-fitting problem is sufficiently constrained so that the only unknown is the reference spectrum for the protein component. We apply this approach to map the distribution of water in cryo-sections obtained from frozen-hydrated tissue of porcine skin. The raw spectral data were collected at doses up to 10(5)e/nm(2) despite the fact that observable damage begins at doses as low as 10(3)e/nm(2). The resulting spatial resolution of 10nm is 5-10 times better than that in previous studies of frozen-hydrated tissue and is sufficient to resolve sub-cellular water fluctuations as well as the inter-cellular lipid-rich regions of skin where water-mediated processes are believed to play a significant role in the phenotype of keratinocytes in the stratum corneum. Copyright 2010 Elsevier B.V. All rights reserved.