Kellermayer et al. 10.1073/pnas.0704305105. |
Fig. 4. Starting (a) and ending (b) AFM images of the scanning force kymography experiment documented in Fig. 1b of the main text. Fibrils i and ii grow into opposite directions despite their identical axial orientation. The time difference between the two images is 18 min.
Fig. 5. 3D-rendered image of the assembly of an Ab25-fibril on top of an already existing one. Height values indicate that both the underlying and the top fibril are mature Ab2535 fibrils.
Fig. 6. Pause time (a) and step size (b) statistics measured as a function of total peptide concetration. Open triangles refer to the slow-growing end, and filled circles to the fast-growing one.
Fig. 7. Trigonally oriented assemblies of Ab1-42 peptide on HOPG surface. Phase image.
Fig. 8. Amplitude contrast mode AFM image of Ab1-42 fibrils growing on HOPG surface. The fibrils indicated with white arrowheads go through the dynamic stepwise length fluctuations shown in SI Movie 4.
Fig. 9. Examples of drift in kymograms. (a) Drift in a direction perpendicular to fibril axis. (b) Nonlinear drift in a direction parallel with the fibril axis.
Fig. 10. Analysis of fibril height. (a) Scanning force kymogram. Red dotted line indicates the position of the linear region of interest for height analysis. (b) Height versus distance plot.
Fig. 11. National Institutes of Health Image macro code for converting SFK image information into data.
Fig. 12. Steps of converting image information into numerical data. (a) Grayscale scanning force kymogram. (b) Binary image. (c) Distance versus time plot.
Movie 1. Discontinuous assembly of Ab25-35 fibrils on mica surface. Sequence of 3D-rendered, 2 ´ 2 micrometer images separated by 3 min. Stage drift is negligible.
Movie 2. Off-rate measurement with a gently scanned AFM tip. Time-dependent sequence of 3D-rendered, 1 ´ 1 micrometer images of a mica surface precoated with Ab25-35 fibrils following the removal of free peptides. Images are separated by 6.5 min. Scan rate, 1.3 Hz; drive amplitude, 0.34 V; set point, 0.72 V. Despite considerable stage drift, the surface density of the fibrils can be assessed.
Movie 3. Off-rate measurement with a more forcefully scanned AFM tip. Time-dependent sequence of 3D-rendered, 1 ´ 1 micrometer images of a mica surface precoated with Ab25-35 fibrils following the removal of free peptides. Images are separated by 2.8 min. Scan rate, 3 Hz; drive amplitude, 0.32 V; set point, 0.61 V. The second half of the video shows the scanned area after zooming out. Despite stage drift, the surface density of the fibrils can be assessed.
Movie 4. Time-dependent sequence of Ab1-42 fibrils growing on HOPG surface. Images are separated by 1 min. Scan rate, 1.4 Hz; drive amplitude, 0.2 V; set point, 0.2 V. Amplitude contrast images are shown. Fibrils going through dynamic length changes are indicated with arrowheads in SI Fig. 8.
SI Text
Atomic Force Microscopy and Scanning Force Kymography.
Amyloid beta 25-35 (Ab25-35) fibrils attached to and assembled on mica surface were imaged by using non-contact-mode atomic force microscopy (AFM). The fibrils displayed a trigonal arrangement dictated by the lattice structure of the underlying mica surface (Fig. 4). Because the fibrils maintained their axial orientation during growth for considerable distances (Fig. 4 and SI Movie 1), the dynamics of their assembly could be followed by repetitively scanning with the AFM tip along the fibril axis, in a method called scanning force kymography (SFK). In typical experiments we obtained 2D images before and after the acquisition of kymograms, examples of which are shown in Fig. 4a and Fig. 4b, respectively. The assembly of Ab25-35 proceeding on the surface of a preexisting fibril can often be observed (Fig. 5). The 3D-rendered image of such a kymogram indicates that both the underlying fibril, which is bound to mica, and the fibril growing on its surface are mature species.Concentration-Dependent Measurements.
The mean pause time and mean step size as a function of peptide concentration are shown in Fig. 6. Both parameters remained unchanged upon increasing peptide concentration. Although mean step size is smaller for the slow-growing fibril end than for the fast-growing, the position of the peaks in their histograms is identical. Thus, the smaller mean step size in the case of the slow-growing end may be explained by a smaller number of long step sizes.Off-Rate Measurements.
In many cases, we observed back-steps in kymograms. Furthermore, at low peptide concentrations, sequential disassembly events that shorten the net fibril length, could be observed. However, a systematic analysis, in which free peptides were completely washed away, did not result in an unambiguous measurement of the off-rate of the polymerization reaction. Apparently, the disassembly process is influenced by the speed and pushing force of the AFM tip. At low speed and pushing force (high set point value) essentially no disassembly is observed (SI Movie 2). If the speed is increased and the set point lowered (increased pushing force), then fibril disassembly becomes appreciable (SI Movie 3).Measurements on Ab1-42.
To investigate amyloid growth on peptides and surfaces different from Ab25-35 and mica, respectively, we studied the assembly of the full-length, Ab1-42 peptide on a highly ordered pyrolytic graphite (HOPG) surface. Samples (100-150 ml) of Ab1-42, dissolved 10 mM Na-citrate (pH 5), were added to freshly cleaved HOPG surface. After 30-60 min lag period, up to several hundred nanometer long filamentous assemblies appeared within a short time period, suggesting cooperative growth mechanisms. The sample surface was scanned in non-contact mode with silicon nitride micro cantilevers (Olympus OMCLTR400PSA-1). Because the fibrils were relatively weakly attached to the surface, height contrast images were difficult to collect. Highest quality data were obtained in amplitude and phase contrast modes. Phase and amplitude contrast mode images of a HOPG surface coated with Ab1-42 fibrils are shown in Figs. 7 and 8, respectively. The temporal changes in Fig 8 are shown in SI Movie 4. Because of the long lag phase and the relatively short growth phase, kymograms were relatively difficult to obtain, because thermal drift interfered with imaging.Thermal drift.
To alleviate the problem of thermal drift, we usually carried out thermal relaxation of the piezoelectric stage. The effect of drift on kymograms is shown in Fig. 9. Drift along the axis perpendicular to the fibril axis caused the AFM tip to wander off the fibril (Fig. 9a). In such experiments valuable data could not be obtained. Drift along the axis parallel to the fibril axis resulted in distorted kymograms (Fig. 9b). Because fiducial points or lines are often present (e.g., non-growing end or coss-section), it was possible to correct for this type of drift in subsequent analysis. Typically, however, thermal drift of the piezoelectric stage relaxed after a full field of scan, and remained minimal thereafter.Identifying fibrils in kymograms.
The nature of the filamentous structures on mica surface was assessed by measuring their height. Such a height analysis is shown in Fig. 10. The 2.4 nm ± 0.2 nm SEM height observed indicates that the imaged fibrils are mature Ab25-35 fibrils.Conversion of images to datapoints.
SFK image data were converted to numerical data with a user-developed macro algorithm for the public-domain National Institutes of Health image software (http://rsb.info.nih.gov/nih-image). The macro code is displayed in Fig. 11. The simple conversion procedure consists of transforming the kymogram (Fig. 12a) into binary image (Fig. 12b), then sequentially scanning along the distance axis to find the coordinate of transition from foreground color (corresponding to the amyloid fibril) to the background color (corresponding to the mica surface). The obtained data were plotted as distance versus time (Fig. 12c).