Proc. Natl. Acad. Sci. USA Vol. 77, No. 6, pp. 3139-3143, June 1980 Biochemistry Electron microscopy of negatively stained and unstained fibrinogen (scanning transmission electron microscopy) LEONARD F. ESTIS* AND RUDY H. HASCHEMEYER Department of Biochemistry, Cornell University Medical College, 1300 York Avenue, New York, New York 10021 Communicated by Alton Meister, December 26,1979

ABSTRACT Electron microscopic images of negatively are always met: (i) The great majority of images of "apparent stained fibrinogen are predominantly asymmetric rods 450 A particles" in any field are of fibrinogen molecules. (Ui) The in length and about 60 A in width. The molecules appear to have considerable flexibility, and mass distribution along the major number of identifiable fibrinogen molecules is defined by fi- axis is not uniquely distinguished despite apparent beading in brinogen concentration and application time and not by the some particles. Scanning transmission electron microscopy of buffer, pH, or staining conditions. (Mii) Large-field views of unstained fibrinogen again demonstrates that a majority of fibrinogen contain predominantly well-separated molecules molecules are rodlike. The results differ from those obtained with clearly defined morphology and dimensions. by negative staining in that a substantial fraction of images are Conditions trinodular with striking resemblance to those obtained by C. required to achieve these goals are primarily E. Hall and H. S. Slayter IJ Biophys. Bioehem. Cytol. (1959) 5, related to attachment of fibrinogen to the substrate film. Once 11-161 using the mica replictecaique. The above results were this problem was solved, images of the unstained molecule were obtained on glow-discharged carbon substrate films by a simple also obtained by high-resolution scanning transmission electron low-concentration, lon&-attachment-time modification of microscopy (STEM). standjard deposition methods that is diffusion controlled and The data presented here may be combined with published depnds on concentration and time but is independent of pH, buffer, and other staining conditions. Evidence is presented gat work to present an overall interpretation of the general mor- standard attachment procedures result in artifactual images. phology of the fibrinogen molecule that is free of ambigui- Any model of fibrinogen in solution consequently must en- ties. compass properties that permit its visualization as an asym- metric rod by electron microscopy as first suggested by Hall and MATERIALS AND METHODS Slayter 20 years ago. The fibrinogen used in this work was a kind gift of Birger Electron microscopic investigation of the size and shape of the Blomback. For the neutral pH studies, bovine or human fraction fibrinogen molecule and fibrin clot substructure has led to a I-4 (>95% clottable), lyophilized or frozen in 0.1 M Tris.HCl multitude of images and conflicting interpretations (1-10). pH 7.4, was dialyzed against 0.02 M sodium phosphate buffer. Consistently well-defined large-field images have to date re- For the low pH studies, the fibrinogen was dialyzed against 0.01 sulted primarily from the shadowed images obtained by the M acetic acid. Covalent crosslinking of fibrinogen was ac- mica replica technique. This work was originally done by Hall complished by adding 10% glutaraldehyde to a solution of bo- and Stayter in 1959 (2) and was later reproduced by Gorman vine fibrinogen at 1.5 mg/ml (0.2 M sodium phosphate buffer, et al. (11), and essentially identical results are readily obtained pH 7.0), to give a final glutaraldehyde concentration of 2%. The in our own laboratory. From the Hall and Slayter work, the solution was allowed to react at room temperature for 90 min rodlike trinodular structure of fibrinogen was proposed. and was then extensively dialyzed at 40C against 0.02 M sodium Shadowed replicas of freeze-etched fibrinogen recently ob- phosphate buffer, pH. 7.0. Despite long dialysis and numerous tained by Bachmann et al. (10) also indicate that the molecules buffer changes, negative staining of the glutaraldehyde-fixed have a rodlike structure with length similar to that obtained by protein often resulted in "glutaraldehyde-induced background" Hall and Slayter but with a greater width indicative of more images as if the fixative had not been removed. extended protein domains. Fibrinogen was negatively stained by first diluting the pro- Electron microscopic studies involving the high-resolution tein to a concentration of about 1-2 ,g/ml in either the 0.02 technique of negative staining have resulted either in over- M sodium phosphate buffer or 0.01 M acetic acid. A copper grid crowded images wherein one cannot see isolated molecules or with attached thin carbon film (100-150 A), which had been in small viewing fields where it is difficult to be certain if a subjected to a low-vacuum (t10-1 torr; 1 torr = 1.333 X 102 "particle" is fibrinogen or the result of electron microscopic Pa) glow discharge at about 5000 V for 2 min, was floated on artifacts. These studies have been interpreted in terms of a the surface of the fibrinogen solution. The grid was then number of models for the fibrinogen molecule, ranging from transferred to the surface of the staining solution, either 2% a much more elongated multibeaded molecule (4, 7) to an iso- sodium phosphotungstate (NaPTA) or 0.5% uranyl acetate. The tropic pentagonal dodecahedron (5). We present evidence here grid was removed after 1 to several minutes and excess stain was that conventional negative staining methods, which are so removed by touching the edge with a piece of filter paper. In successful in imaging well-behaved macromolecules, fail to a typical time series experiment, four grids were floated on four provide negatively stained fields of fibrinogen that fulfill rea- separate fibrinogen solutions at identical concentration and sonable prerequisites for meaningful interpretation. were removed after various times to permit variation in the time Our major purpose here is to define precise conditions for the allowed for fibrinogen attachment to the carbon film-e.g., 5, negative staining of fibrinogen wherein the following criteria 10, 15, and 20 min. The publication costs of this article were defrayed in part by page Abbreviations: STEM, scanning transmission electron microscopy; charge payment. This article must therefore be hereby marked "ad- NaPTA, sodium phosphotungstate. vertisemenz" in accordance with 18 U. S. C. §1734 solely to indicate * Present address: Waksman Institute of Microbiology, Rutgers Uni- this fact. versity, P.O. Box 759, Piscataway, NJ 08854. 3139 Downloaded by guest on September 28, 2021 3140 Biochemistry: Estis and Haschemeyer Proc. Natl. Acad. Sci. USA 77 (1980)

C. Ni\ $

*r-J ?t mw~~~~~

2ak

FIG. 1. Fibrinogen negatively stained with NaPTA. (Upper) Human. (X225,000.) (Lower) Bovine. (X440,000.) Electron microscopy was performed on a JEOL 100-B mi- using procedures to minimize beam damage to the specimens. croscope equipped with an anticontamination device. The An area near to that to be photographed was visualized at high operating voltage was 80 kV and a 20-um objective aperture magnification (X100,000) to correct astigmatism. The same was used. The typical electron optical magnification employed nearby area was then brought into near focus at the photo- for photography was X50,000. Photography was performed by graphing magnification of X50,000. The desired area was then

4 $ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~is

i 4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1r<. ;..Ad<*4'ta

FIG. 2. Bovine fibrinogen negatively stained with uranyl acetate. (X225,000.) Downloaded by guest on September 28, 2021 W.**>4 Biochemistry: Estis and Haschemeyer Proc. Nati. Acad. Sci. USA 77 (1980) 3141

4

qrich

FIG. 3. Bovine fibrinogen negatively stained with NaPTA after varying time of protein attachment: 1 min (Left); 5 min (Center); 10 min (Right). (X235,000.) brought into position, the electron beam was deflected, and, Fig. 1 Upper shows the characteristic reproducible images after a short period of time to permit stage stabilization, the obtained when a grid is floated for 10 min on a dilute solution beam was recentered and the noncontaminated area was of human fibrinogen (-1-Ig/ml) in 0.02 M phosphate buffer quickly photographed. (pH 7.0), followed by floatation on 2% NaPTA (pH 8.0) for 2 Areas chosen for photography were typically near the edge min. A similar result is obtained with bovine fibrinogen (Fig. of a smooth wavelike depression in the carbon film. These de- 1 Lower). The predominant images in the field of view are seen pressions, or channels, presumably formed as the carbon film to be rodlike with dimensions of approximately 450 A in length was attached by sweeping the grid smoothly upward from and 60 A in width. Some of the images appear to be bent or below the water surface, show a gradation in density of negative S-shaped and, in some, apparent beading is evident. Sometimes stain. The spreading of negative stain on the glow-discharged images appear to have a different staining intensity in the two grids presents a final smooth surface with very light stain over halves defined by the center of the rod out to each end along most of the grid to thick stain in the center of the channels. the long axis. Results obtained with highly purified human and Fibrinogen is poorly contrasted in areas of very light stain bovine fibrinogen show no distinguishable differences. compared to other proteins, whereas near the center of the Other images in the field of view, which are generally channel the stain is too thick. Areas near the periphery of the spherical in appearance with a wide range of diameters (10-200 channels provide excellent contrast and were selected for most A), are often seen in negatively stained preparations of other of the negatively stained photographs. proteins and in protein-free controls. Their origin appears to Magnification calibration utilized catalase crystals (Worth- be due to minor contamination of the buffers used. ington) as an external standard. A value of 84 A for the crystal Fig. 2 shows that similar results are obtained at low pH (3.2). spacing was assumed (12). The negative staining procedure used-i.e., low concentration Unstained fibrinogen was examined at the Brookhaven (;1 ,gg/ml) and a long time of attachment (10 min)-is iden- STEM Resource with the microscope operating at 40 kV with tical to that employed for Fig. 1 except that the fibrinogen was a resolution of 2.8 A. Thin carbon films (;20 A) made in a clean in 0.01 M acetic acid and the negative stain was 0.5% uranyl room by using an ultrahigh vacuum evaporator were stretched acetate (pH 4.5). Except for more apparent beading in some over films with ;5-,gm holes and mounted on titanium grids. particles, the observations are similar. Extensive fixation with The grids were usually submitted to glow discharge for about glutaraldehyde did not alter any of the above observations, with 10 s and the molecules were applied by the float procedure the exception that crosslinked molecules were often observed described above. Freeze-dried specimens were prepared by as expected (data not shown). leaving a thin film of water on the grid, immersing the grid in Fig. 3 shows the images obtained when molecules from three liquid nitrogen, placing the grid in a liquid nitrogen-cooled separate dilute (-1 Atg/ml) solutions of fibrinogen in phosphate block, and drying it under high vacuum with a warming rate buffer (pH 7.0) were attached for 1, 5, and 10 min, respectively, of about 20C per min. The dried sample was transferred to the and then stained for 2 min on 2% NaPTA (pH 7.0). After at- microscope stage under vacuum. The specimen was cooled to tachment for 1 min (Fig. 3 Left), no rodlike images were seen. -120'C during examination and most images were obtained Only a few of the previously mentioned spurious images ap- in the dark-field mode by utilizing a 0.04-0.20 radian detector peared. After a 5-min attachment (Fig. 3 Center), a few rodlike capable of quantum efficiency. images were seen. The "background" is similar to that observed in Fig. 3 Left. In Fig. 3 Right (10-min attachment) the rods RESULTS were more numerous and again a constant background ap- Negative Staining. By utilizing a seemingly simple, yet peared. unusual variation of standard negative staining methods, it is In contrast to such control over the number of molecules in possible to obtain clear and distinct images of fibrinogen. The the field, normal staining methods using higher concentrations essential feature of this procedure is the use of very dilute (e.g., 100,gg/ml) and a 1-min attachment produced an ap- protein solutions and a long grid floatation time on such solu- parent carpet of fibrinogen (Fig. 4 Upper), often with super- tions. imposed spherical images (Fig. 4 Lower). However, after a Downloaded by guest on September 28, 2021 3142 Biochemistry: Estis and Haschemeyer Proc. Natl. Acad. Sci. USA 77 (1980)

FIG. 5. Large-field STEM view of unstained human fibrinogen obtained with low-dose (t7 electrons/A2) irradiation at 512 scan lines per 1.04 ,m. (X125,000.) FIG. 4. Bovine fibrinogen negatively stained with NaPTA after being applied to the grid at high concentration. The apparent "carpet" pendent folding domains to distort separately as an artifact of of protein (Upper) and a similar field with superimposed spherical specimen preparation. images (Lower) are different fields from the same grid. (X225,000.) DISCUSSION 1:100 dilution of the protein and a 1-min attachment, no mol- We have defined precise conditions for the electron microscopic ecules were seen (Fig. 3 Left). visualization of both negatively stained and unstained fibrin- STEM Results. Grids with fibrinogen attached by the floa- ogen. These conditions involve a combination of very low tation procedure used for negative staining were placed on protein concentrations (-1 ,gg/ml) and long attachment times distilled water for 5 min or more to remove salts. In some cases (5-20 min). This procedure has allowed us to obtain clear and intermediate floatation on 10 ,M uranyl chloride was used for distinct images of fibrinogen that unambiguously constitute the positive staining. Final images did not differ significantly in majority of images in any viewing field. Of particular impor- appearance if unstained, positively stained (except for the tance in expressing this confidence is the fact that the number greater contrast imparted by bound uranyl), air-dried, or of identifiable fibrinogen images is defined primarily by fi- freeze-dried preparations were used. Fixation with glutaral- brinogen concentration and application time and not by buffer, dehyde also had no effect. pH, or staining conditions. Fig. 5 shows a typical field of unstained fibrinogen scanned In contrast, we have demonstrated that more conventional at relatively low magnification. A substantial fraction of par- attachment procedures produce viewing fields that are difficult ticles bear a striking resemblance to the familiar trinodular to interpret. These methods result in viewing fields containing shadowed molecules of Hall and Slayter (2). These molecules crowded images of a dense protein carpet or spherical images are the same length as those observed by negative staining, but superimposed upon these crowded images. When the fibrino- are substantially wider (89 + 9 A). gen solution is diluted to a much lower concentration (i.e., 1 Particles with less well-defined beads are frequently ob- ,gg/ml) and allowed to attach for 1 min, sparse viewing fields served, and considerable protein has poorly contrasted outlines are obtained containing only a few spurious images similar to and represents less mass per unit area on the grid. This flattened those observed in protein-free controls. We feel that interpre- or "puddled" appearance is sometimes seen at the end of a tation of images similar to those in Fig. 4 is generally responsible two-beaded structure or in between two beads. A reasonable for the many and different models proposed for the structure explanation would be that fibrinogen has sufficiently inde- of fibrinogen. Downloaded by guest on September 28, 2021 Biochemistry: Estis and Haschemeyer Proc. Natl. Acad. Sci. USA 77 (1980) 3143 We do not know why the standard protein attachment pro- The major emphasis of our work has been to obtain clear, cedure does not allow satisfactory visualization of fibrinogen. electron-microscopic images of isolated fibrinogen molecules. One ad hoc explanation is that fibrinogen, at typical concen- More subtle questions concerning the apparently independently trations, can form a protein film at a solution/air or solution/ folding domains of the molecule, their extent in aqueous solu- substrate interface. Attachment to the substrate film then results tion, and the structure of the fibrin clot remain to be an- in a "carpet" of protein. At low concentration (1 gug/ml), the swered. protein film does not form and a relatively long period of time is required for individual molecules to attach through a diffu- We thank the entire staff of the Brookhaven Scasining Transmission sion-controlled process. Whether this explanation is correct or Electron Microscope Research Facility. We especially thank Dr. Joseph not, fibrinogen behaves as if it were correct. S. Wall for his assistance and stimulating advice. This work was sup- Our results demonstrate that negatively stained or unstained ported by National Institutes of Health Biotechnology Research Branch fibrinogen is rodlike with a length of about 450 A. The 90-A Grant RR 00715 and National Institutes of Health Grant HL 11822. width of the rods observed by STEM is identical to that ob- served by Bachmann et al. (10) on freeze-etched fibrinogen. 1. Hawn, C. V.-Z. & Porter, K. R. (1947) J. Exp. Med. 86, 285- The calculated volume far exceeds that required to accom- 292. modate the fibrinogen molecule if it were compact. Under the 2. Hall, C. E. & Slayter, H. S. (1959) J. Biophys. Biochem. Cytol. conditions of attachment used here, fibrinogen does not sig- 5, 11-16. nificantly aggregate in solution and association on the grid 3. Bang, N. U. (1964) Thromb. Diath. Haemorrh. Suppl. 13,73- seems incompatible for a diffusion-controlled process. We thus 80. conclude, in agreement with Bachmann et al. (10), that the 4. Kay, D. & Cuddigan, B. J. (1967) Br. J. Haematol. 13, 341- molecular domain of fibrinogen is a "loose" one that extends 347. beyond expected boundaries. The most direct explanation of 5. K6ppel, G. (1967) Z. Zellforsch. Mikrosk. Anat. 77,443-517. the lesser width observed by negative staining (60 A) is that the 6. Karges, H. E. & Kuhn, K. (1970) Eur. J. Biochem. 14,94-97. stain partially penetrates the molecule to obscure its edge. 7. Belitser, V. A., Varetska, T. V. & Manjakov, J. Ph. (1973) Thromb. Considerable flexibility in the rod is consistently observed under Res. 2, 567-578. all conditions. In almost any field of isolated molecules, the 8. Krakow, K., Endres, G. F., Siegel, B. M. & Scheraga, H. A. (1972) can see J. Mol. Biol. 71, 95-103. observer some molecules that are relatively straight, 9. Pouit, L., Marcille, G., Suscillon, M. & Hollard, D. (1973) Thromb. some that are bent to varying degrees, and, occasionally, some Diath. Haemorrh. 27,559-572. whose ends are touching. 10. Bachmann, L., Schmitt-Furnian, W. W., Hammel, R. & Lederer, Trinodular structures similar to those first described by Hall K. (1975) Makromol. Chem. 176,2603-2618. and Slayter (2) are frequently observed by STEM. Many mol- 11. Gorman, R. R., Stoner, G. E. & Catlin, A. (1971) J. Phys. Chem. ecules, however, show zero, one, or two "beads." Flattening or 75,2103-2107. uncoiling of these beaded domains appears to be a common 12. Ferrier, R. P. & Murray, R. T. (1966) J. R. Microsc. Soc. 85, feature of the molecule. 323-328. Downloaded by guest on September 28, 2021