Proc. Natl. Acad. Sci. USA Vol. 85, pp. 453-457, January 1988 Biophysics Coloration of silver-stained protein bands in polyacrylamide gels is caused by light scattering from silver grains of characteristic sizes CARL R. MERRIL*, MARGARET E. BISHERt, MICHAEL HARRINGTON*, AND ALASDAIR C. STEVENt *Clinical Neurogenetics Branch, National Institute of Mental Health, and tLaboratory of Physical Biology, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, MD 20892 Communicated by James V. Neel, October S, 1987 ABSTRACT This study investigates the physical basis of of silver-stained protein bands in gels-the formation of color effects in the detection of proteins in polyacrylamide gels specific protein-silver complexes (8). In this case, colors by silver staining. Specifically, the hypothesis that different would presumably result from conjugated bond systems colors may correlate with the development of silver grains of formed through appropriate interactions of the silver atom(s) characteristic sizes was investigated by electron microscopy. with functional group(s) on the protein molecules. To inves- Protein bands that stained brown, yellow, and blue were tigate this phenomenon and to distinguish between these excised from stained gels and prepared for electron micros- alternative explanations, we have employed electron micros- copy by thin-sectioning. In each case, the size distributions of copy to determine directly the sizes of silver grains associ- globular silver grains were determined directly from the ated with protein bands of different colors. electron micrographs. We found that blue bands have larger silver grains (with diameters of 40-100 nm) than yellow (21-39 nm) or brown bands (17-35 nm). On the basis of these and MATERIALS AND METHODS other observations, a general mechanism is proposed whereby Proteins. Human serum albumin (fraction 5) was obtained chemical specificity of electrophoretically separated proteins is from Miles, and human apolipoprotein A-II was obtained expressed in color-specific silver staining. from Sigma. These proteins were dissolved in a solution containing 1% (wt/vol) sodium dodecyl sulfate (NaDodSO4), The observation that silver nitrate, the main ingredient of 2.5% (vol/vol) 2-mercaptoethanol, and 1% (vol/vol) Triton silver stains, can stain or blacken organic substances, includ- X-100 prior to electrophoretic separations. Dilutions of the ing human skin, is credited to Count Albert von Bollstadt in proteins were made with 70% (vol/vol) glycerol. the 12th century (1). Modem scientific applications of silver Electrophoresis. Slab gels (0.75 mm thick) containing staining began with Krause's histological staining of fresh 12.5% polyacrylamide and 1% NaDodSO4 were run in a tissues in 1844 (2). Silver staining was introduced as a vertical apparatus as described by Studier (11), using the general detection method for proteins separated by poly- discontinuous buffer system of Laemmli (12), with bromphe- acrylamide gel electrophoresis in 1979 (3, 4). It is more nol blue as a tracking dye. sensitive than the most commonly used organic protein Silver Staining. The gels were first fixed for 20 min in a stain, Coomassie blue, by a factor that can, with the most solution containing 50% (vol/vol) methanol and 10% responsive proteins, exceed 100. (vol/vol) acetic acid. They were then rinsed for 20 min with Enhanced color development with silver stains has been a fixer containing 10% methanol and 5% acetic acid. Staining used as an aid in the identification of certain proteins (5-8). was begun by treating the gels with a 32.6 mM dithiothreitol Production of color with silver stain depends on many mM variables. It is, for instance, possible to promote the devel- solution for 20 min, followed by gentle agitation in a 12 opment of colors by lowering the concentration of reducing silver nitrate solution for 15 min. The image was developed agent in the image-development solution, prolonging the in a 0.283 M sodium carbonate solution containing formal- development time, adding alkali, or elevating the tempera- dehyde (0.5 ml of a 37% formaldehyde solution per liter). ture during staining. Variations in protein concentration may Image development was terminated by repeated washes in also produce color shifts. Moreover, color may depend on deionized water. intrinsic properties of the individual proteins: Nielsen and Electron Microscopy. Desired bands were excised from Brown (8) have shown that charged amino acid side groups gels, dissected into blocks of approximately 3 mm x 6 mm play a major role in color formation. x 0.8 mm, and dehydrated by serial transfer into concen- The current study was designed to investigate the basis of trations of50% through 100% ethanol. As monitored by light color effects in silver staining of proteins in polyacrylamide microscopy, dehydration resulted in shrinkage of the sam- gels. One possibility is that they may be analogous to ples to 45% (± 4% SD, n = 13) of their original dimensions. light-induced silver-based color effects in photography, The samples and embedding resin were placed in BEEM which have been observed as long ago as 1840, with Her- capsules, the capsules were tightly capped to prevent con- schel's development of a photographic image on a paper tact with air, and the resin was allowed to polymerize for 24 saturated with silver halides. In this way, he was able to hr at 65°C. Thin sections, nominally =70 nm, were cut with record the colors of the prismatic spectrum of sunlight, glass knives on a Sorvall MT2 ultramicrotome. To prevent culminating in a "full and fiery red" (9). It has since been distortion of the sections upon exposure to water (see proposed that color development in silver-based photochem- Results), ethanol was used in the collection trough, and the istry depends primarily on two variables: the size of the sections were salvaged from the bottom of the trough on silver grains and the average intergrain spacing (10). How- hexagonal-mesh grids. To flatten the sections and to prevent ever, a second explanation has also been given for the colors them from slipping off the grids, they were sandwiched between two such grids. Specimens were observed in a The publication costs of this article were defrayed in part by page charge Philips EM300 electron microscope at a nominal magnifica- payment. This article must therefore be hereby marked "advertisement" tion of x 16,000, which was precisely calibrated by using the in accordance with 18 U.S.C. §1734 solely to indicate this fact. 2.49-nm striations of crystallites of the dye olive-T (13). 453 Downloaded by guest on September 29, 2021 454 Biophysics: Merril et al. Proc. Natl. Acad. Sci. USA 85 (1988) Image Processing. Electron micrographic prints at magni- viscosity resin, chosen on the premise that low viscosity fications of x 30,000 or x 60,000 were digitized by means of should facilitate infiltration, it was established that embed- a Hamamatsu C-1000 video camera (Hamamatsu, Waltham, dings with adequate cutting properties could be obtained MA) and displayed on a television monitor. The pixel sizes with LR White or Spurr resins (Polysciences, Warrington, used corresponded to 2-4 nm at the specimen. Image PA). However, upon exposure to water in the collection analysis was performed on a VAX 11/780 computer (Digital trough during ultramicrotomy, these sections underwent Equipment, Maynard, MA) with a DeAnza IP8500 image considerable expansion and distortion, presumably as a processor (DeAnza Systems, San Jose, CA). Diameters of result of rehydration of the polyacrylamide (15), suggesting grains were calculated by using an interactive program the lack of complete infiltration of the gels by the resins. By and substituting ethanol for the water in the cutting trough, we (written by B. L. Trus, Division of Computer Research found that sections could be retrieved with minimal distor- Technology, National Institutes of Health) that automati- tion from the bottom of the trough (Materials and Methods). cally detected the edges of manually designated grains and A typical micrograph showing the polyacrylamide matrix computed their areas (A). Diameters were calculated as with a dense distribution of silver grains, permeated with 2(A/ir)1/2. Approximately 500 randomly chosen grains were (grain-poor) pores, is shown in Fig. 2. Because these sec- measured for each color band, and these data were analyzed tions were not stained for electron microscopy with heavy statistically by using the MLAB program (14) on a DEC-10 metal salts or subjected to any other treatment that might computer (Digital Equipment). alter the sizes of the silver grains developed during the protein staining, the dimensions and (globular) morphology RESULTS of the silver grains should be correctly represented in these images. On the other hand, it is not possible to draw Preliminary experiments had established that when silver- inferences as to grain density (i.e., typical inter-grain spac- stained according to the protocol used in this study, human ing) on better than a semiquantitative basis. This difficulty is serum albumin stained yellow, apolipoprotein A-II stained due to two factors: first, the distortions of the sections, deep brown, and a minor band running slightly ahead of primarily shrinkage upon dehydration, estimated to be a apolipoprotein A-II stained blue. These and other bands factor of 2 or more; and second, variability in the projected were excised from one-dimensional gels (e.g., Fig. 1) for thickness of the sections, which are often crumpled and analysis by electron microscopy. After some experimenta- folded. tion with resins, including Epon/Araldite and Ladd's low- In sections from background regions of the gel, which do not contain a stained protein band, very small (5-15 nm), micrograms are detected (Fig. 3d). These micrograms are ...~~~~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~\........ presumably responsible for the brownish background hue of silver-stained gels, since they are present throughout the gel, even (albeit sparsely) in regions occupied by negatively stained protein bands, such as area 6 of Fig.