characterize the performance of thin metal films for the experimental parameters. The axial error of gold, for example, for a thickness of 20 nm is much lower as compared to the curve shown in figure 5.2 for the same wavelength and quantum yield of the dye.
5.2 smMIET for Structural Biology
With the help of simple experiments, we achieved a nanometer axial localization preci-sion from a thin gold film at single molecule level. Therefore, if the dye molecules are rotating freely in space and are separated spatially or show stochastic blinking, similar to what is typically required for a localization based super-resolution technique such as STORM, PALM or PAINT, then by using the fluorescence lifetime information, one can now localize each emitter with a nanometer axial precision, within the near-field range from a thin metal film (∼ 100 nm). This would allow one to study for example, the processes involved in a focal adhesions such as a cell’s adhesion to its extra cellular matrix, force transmission, cytoskeletal regulation and signaling [162–164], close to a surface, at single molecule level.
If, on the other hand, the molecules are restricted in their rotation and are some-how fixed with a random orientation, then as we saw in section 4.3.3 one introduces significant lateral localization inaccuracy, as high as ∼ 15 nm (see figure 4.33), even when the molecules are present in the focal plane of a high N.A. objective. Although one can achieve a nanometer precision in localization, the point localization based super-resolution methods suffer from these huge inaccuracies which act as major limi-tations [34, 35]. In this case, it is desirable to utilize the method of defocused imaging where the advantage when combined with smMIET is two-fold. First, it allows one to determine the orientation of the dye molecule with respect to the metal surface, which is necessary for estimating its accurate height from the surface 2.4.3. Second, as was emphasized and shown in discussion of the same chapter, fitting the defocused intensity patterns with a log-likelihood algorithm gives a high lateral localization precision. Since one takes into account the asymmetry in the angular distribution of radiation from an oriented dipole, the lateral position estimated in such a way must be close to its true location, within error limits. One has to explore the applicability of defocused imaging in combination with smMIET.
Even though the thin metal film absorbs and reflects a part of the emission from a fluorophore, preliminary experiments show the possibility of acquiring defocused pat-terns with such a metal thin film substrate. Figure 5.4 shows a few captured defocused patterns of Atto 655 molecules on top of an SiO2 spacer of 30 nm. Thus, all that is required is to combine a FLIM microscope together with defocused imaging. Extending a laser-scanning confocal FLIM system by adding a detection channel with a defocused EMCCD camera is one of the many options. The idea is much similar to the
experimen-5.2. SMMIET FOR STRUCTURAL BIOLOGY CHAPTER 5. DISCUSSION AND OUTLOOK
Figure 5.4: Defocused images of Atto 655 molecules, spin-coated on top of a 10 nm gold film with a 30 nm SiO2spacer in between, at various defocusing values. The left figure was taken with a defocusing of 0.5µm whereas the two figures on the right were taken with a defocusing value of 1.2µm. The setup details can be found in section 3.1.2.
tal method shown in the work for the determination of excitation and emission dipole orientations (section 4.4), using a linearly polarized laser for scanning. A pre-scan is performed first to locate the position of the molecules on top of the substrate using a custom written search-and-seek LabVIEW program, and thereafter, a series of defocused images is acquired for each emitter by parking the scanner at its position. Further, in order to obtain a reliable estimate of the fluorescence lifetimes, a part of the photons are focused onto a Single-Photon Avalanche Diode (τ-SPAD, PicoQuant) during the point measurements. A good compromise between the number of photons required for a good signal-to-noise ratio in the defocused image for orientation and lateral position estimation, and the number of photons for fluorescence lifetime estimation, can be ob-tained by dividing the total emission in a ratio of 7:3 using a 70 R : 30 T beamsplitter (Thorlabs). The whole setup description can be summarized into a figure, as shown below (figure 5.5).
Now consider an experiment where one would like to determine the distance between two labeled sites on a biomolecular complex or protein. This is a classical problem where one uses FRET to determine such intramolecular distances. However, as we pointed out in chapter 3, one needs a priori information regarding the mutual orientation of the acceptor and donor molecule with respect to each other in order to quantify the distance between them, which acts as a major limitation of the method. A possible solution to this problem using smMIET would be to determine the heights of the two
CHAPTER 5. DISCUSSION AND OUTLOOK 5.2. SMMIET FOR STRUCTURAL BIOLOGY
Figure 5.5: Optical setup design for performing smMIET measurements together with defocused imaging. The collected photons are split into two pathways using a beamsplitter (BS). A part of the fluorescence photons are led to a camera that is displaced from the focal plane in image space. The remaining photons are focused onto a SPAD after having passed through a pinhole (PH).
probes in two separate wavelength channels, thereby obtaining a distribution of height differences between both, allowing one to estimate the exact distance between both the sites using rudimentary statistics. This approach was already introduced in figure 3.10.
Although this approach suits the nature of the problem, the situation gets complicated if one has multiple labeled sites on a globular protein or biomolecular complex. Of course, one could proceed in a customary way by labeling two sites at a time, resolving the distances between each pair, and subsequently obtaining all distances to determine the complete geometry. Here we propose an alternative solution.
Let us consider a simpler situation where, again, we would like to determine the distance between two labeled sites, as shown in figure 5.6. Further, instead of labeling two chromatically separated dyes, let us assume that the two dyes are identical. This reduces the required knowledge of precise free space parameters such as free space
5.2. SMMIET FOR STRUCTURAL BIOLOGY CHAPTER 5. DISCUSSION AND OUTLOOK
lifetime τ0 and quantum yield Φ for the two dye species, which are vital for determining the axial distances. Several scenarios and possibilities now exist. If the dye molecules are STORM-able (blinking stochastically) then one can acquire a video of blinking defocused images, together with TTTR scheme based photon recording, using the setup described above. This opens the possibility to measure average photon arrival time together with capturing defocused images of each individual emitter during its on time.
One can later sort the frames based on the average lifetime measured and add them individually in order to separate the defocused intensity patterns, as summarized in figure 5.6. Fitting the defocused images yields the x, y-position and vertical orientation that is useful for determining the z-distance together with fluorescence lifetime. In this way, all three coordinates of the two dye molecules can be determined. By repeating such measurements on a number of such labeled biomolecules of complexes yields improved statistics for determining the correct distance between the two sites. It is worthy to
Figure 5.6: The schematic on top shows the geometry of the experiment. The globular protein with two labels is deposited on the spacer. If a buffer is required to stabilize the structure, then one binds these proteins to the surface. The orientation of the structure can be random. A transparent SiO2
spacer is required to avoid total quenching of emission from the dye molecules. In the right bottom the four subfigures are: Top left, Summation of all the simulated defocused intensity images; top right, the summation of frames where at least one dye molecule is on; bottom two, summation of frames that are sorted out based on the observed fluorescence lifetime for each dye molecule.