BRIEF COMMUNICATIONS 182 | VOL.9 NO.2 | FEBRUARY 2012 | NATURE METHODS followed by standard electron microscopy sample preparation. After blotting, the residual protein in the gel matrix is sufficient for peptide mass fingerprinting by mass spectrometry. If a clear separation of bands is not feasible because of the complexity of the sample, which is usually the case for crude cell lysates, typi- cally containing hundreds or even thousands of different proteins, other means of prefractionation have to be applied before elec- trophoresis (Supplementary Fig. 1). We loaded protein samples, containing 1–5 µg protein per ana- lyzed band, in duplicate in two lanes on native gels and electro- phoretically separated them (Fig. 1). After electrophoresis, we cut the gels to separate duplicate lanes. We stained the ‘reference lane’ with Coomassie blue, which allowed us to extrapolate the position of the unstained protein bands in the ‘blotting lane’. After destaining, we aligned the two gel parts and located unstained protein bands in the ‘blotting lanes’ by manually extrapolating the position of the stained bands across horizontal lines. As the gel swells during staining, the ‘reference lanes’ were typically 5–10% elongated compared to the ‘blotting lanes’; we compensated for this by taking the overall proportions of the gels into account when locating the protein bands in the ‘blotting lanes’. Then we roughened the surface of the gel to increase the accessible blot- ting surface and placed previously glow-discharged copper grids, coated with a continuous carbon film, on the positions of the identified bands, with carbon film facing the gel. To initiate the blotting transfer, we wetted the interface between the gel and grid with a droplet of electrophoresis buffer. To evaluate the grid-blotting procedure, we selected three of 39 known complexes 6 from the archaeon Thermoplasma acidophilum, the thermosome 7 , 20S proteasome 8 and VCP-like ATPase (VAT) 9 , with respective molecular weights of 940 kDa, 680 kDa and 500 kDa. These complexes were the dominant species in two fractions obtained from crude T. acidophilum cytosolic extracts after protein enrichment. After blotting, we removed the grids and subjected them to standard electron microscopy sample preparation, followed by electron microscopy data acquisition and image analysis (Fig. 2). The obtained average images showed good agreement with available reference structures (Fig. 2a–e and Supplementary Fig. 2). We analyzed the residual protein band via standard mass spectrometric analysis (Supplementary Table 1). With this approach, we obtained particle densities of ~30% coverage for negatively stained samples (Fig. 2a–c). This corres- ponded to a transfer efficiency of ~0.05%. For vitrified samples, transfer efficiency was reduced by a factor of ~3, when we used continuous carbon film (Fig. 2d,e). When using holey carbon films (Quantifoil or Lacey carbon), we observed a severe drop in Blotting protein complexes from native gels to electron microscopy grids Roland Wilhelm Knispel 1,2 , Christine Kofler 1,2 , Marius Boicu 1 , Wolfgang Baumeister 1 & Stephan Nickell 1,2 We report a simple and generic method for the direct transfer of protein complexes separated by native gel electrophoresis to electron microscopy grids. After transfer, sufficient material remains in the gel for identification and characterization by mass spectrometry. The method should facilitate higher- throughput single-particle analysis by substantially reducing the time needed for protein purification, as demonstrated for three complexes from Thermoplasma acidophilum. Single-particle analysis is increasingly important for structural studies of molecular machines and macromolecular assemblies. It is based on the acquisition of large datasets of transmission electron micrographs 1 , which after exhaustive computational processing yield a high-resolution electron density map 2,3 . As compared to other techniques for structure determination (X-ray crystallography or nuclear magnetic resonance), single-particle analysis has two major advantages. First, required sample quanti- ties are minute: as little as 5–10 ng of a protein complex covers an entire electron microscopy grid, which is usually sufficient for obtaining a high-resolution density map, when high-throughput data-acquisition schemes are applied 4 . Second, single-particle analysis can tolerate a certain degree of sample heterogeneity because classification algorithms are used as an additional in silico purification step to ensure the rejection of outliers. A major bottleneck in high-throughput single-particle analysis, however, is the isolation and purification of protein complexes. Biochemical methods traditionally used for purification require a substantial time investment, and purification strategies often need to be adapted for every protein complex individually. Here we present a method we call ‘grid blotting’, for generic protein purification and sample preparation for electron micros- copy. The procedure is based on the separation of proteins in a polyacrylamide gel, where they are focused in bands during native electrophoresis 5 . Spatially separated proteins are blotted directly from the gel matrix onto an electron microscopy grid, which is 1 Max Planck Institute of Biochemistry, Department of Molecular Structural Biology, Martinsried, Germany. 2 Present addresses: ChemAxon Kft., Budapest, Hungary (R.W.K.), Tietz Video and Image Processing Systems GmbH, Gauting, Germany (C.K.) and Carl Zeiss NTS GmbH, Oberkochen, Germany (S.N.). Correspondence should be addressed to W.B. (baumeist@biochem.mpg.de). RECEIVED 18 MAY 2011; ACCEPTED 5 DECEMBER 2011; PUBLISHED ONLINE 8 JANUARY 2012; DOI:10.1038/NMETH.1840 npg © 2012 Nature America, Inc. All rights reserved.