Team:ETHZ Basel/InformationProcessing/Microscope



Details on microscope

Figure 1: The automatized microscope Nikon ECLIPSE Ti we used for imaging.
For the microscopy, we used the automatized microscope ECLIPSE Ti from Nikon. The microscope has a motorized x/y-stage, which we can directly control from our Matlab Toolbox to follow E. coli cells when they swim out of the microscope image. The toolbox can also induce the acquisition of images, effectively controlling the shutter and the camera.

For all of the real-time imaging, we used, if not noted otherwise, a 40× oil lens. For limiting noise and data size we set the microscope internal binning to two (meaning we use only half the maximal resolution), such that the raw images are obtained in a resolution of 672×512 pixel. In this setting, the width of a pixel is equivalent to 0.32μm, and the whole image covers approximately 217×165μm. Automated light-emitting diode arrays serve as light sources for the red light (660nm) and far-red light (748nm) pulses (see below).

Details on the Flow Channel

E. coli cells live in a three dimensional world. However, standard light microscopy can only produce two dimensional images, which caused the problem that, in a simple experimental setting, the E. lemming could swim out of focus and disappear from the controller. We thus had to engineer an experimental setup so that the E. lemming stays on a two-dimensional manifold and, at the same time, is still able to swim. Furthermore, we had to minimize passive flow in the setup and fulfill the requirement that our oil lens had to be close to the object.

We decided to construct a simple, but effective flow channel by combining a cover slip with a plastic cuboid, which has a linear approximately 50μm deep cavity on one of its faces. This cuboid was cast using the mould depicted in Figure 2. The cast raw channel was then cutted in small pieces utilizing a razor plate (see Figure 3), and the channel was assembled simply by pressing the cuboid with the side of the cavity on the cover slip (see Figure 4). The liquid containing the E. coli cells was then pipetted on one of the openings of the channel. After the channel filled up completely, another drop was pipetted on the other opening of the channel to stop the flow of the liquid.

However, this method caused the problem that when a drop of liquid was pipetted on either side of the channel, the liquid not only flew through the channel, but also under the whole cuboid, lifting it and thus increasing the depth of the channel. To prevent this effect, we bonded both parts of the channel with plasma (see Figure 5). Finally, in Figure 6 you can see the complete experimental setup, namely the bonded channel filled with cells on top of the 40x oil lens.

Figure 2: The mold for the channel. Has to be cleaned after casting.
Figure 3: The cast raw channel with a cover slip and the razor plate needed for preparing the channel.
Figure 4: The assembled channel, yet not bonded.
Figure 5: Bonding the channel with plasma.
Figure 6: The ready to use channel under the microscope.

Details on the applied red and far-red light

For the experimental setup it is essential to use the right wavelengths of light for the activation and deactivation of straight swimming and tumbling of the E. lemming. Furthermore, it is important to only use light of wavelengths for the bright field microscopy (necessary for cell detection) which do not induce photoconversion between the Pr and the Pfr form of PhyB. In Figure 9 we plotted the photoconversion cross-sections of PhyB in the Pr and the Pfr forms as published in [1].

The photoconversion cross-section is basically the product of the extinction coefficient and the quantum yield and can be interpreted as a measure for the probability that a transition between the two forms take place. When the photoconversion cross-section is multiplied with the photon flux, one obtains the reaction rate constants for the photoconversion process. As one can see in Figure 9, the photoconversion cross sections of the Pr and the Pfr form overlap, but are displaced from each other. This displacement can be used to change the relative amounts of the Pr and the Pfr forms by using light with a given wavelength.

We use two LEDs as light sources with wavelengths around 660nm (red light) and 740nm (far-red light) produced by CoolLED. Unluckily these LEDs are new product development so that their spectra were not available, yet. We thus had to approximate their spectra by shifting the known spectra of the LEDs with wavelength maximums at 700m and 770nm, which are similar in construction, to the left (see Figure 7). The light of the diodes was send through two band-pass filters from Semrock (product IDs F39-651 and F39-769), whose spectra are shown in Figure 8.

Besides checking the configuration, the information in Figures 7-9 was also used to calculate the rate constants for the photoconversion reactions in the molecular model. These constants are crucial since they determine the speed with which an E. lemming reacts on changing inputs set by the controller. For maximal light strength (the diodes can be down-regulated smoothly) we obtained for red light a rate constants of 6.64s-1 (0.102s-1 for far-red light) for the photoconversion from the Pr to the Pfr form, and 1.02s-1 (5.68s-1 for far-red light) for the reverse reaction. These constants were used for the modeling.

Figure 7: The approximated spectral energy irradiances of the red light (red curve) and far-red light (blue curve) diodes.
Figure 8: The amplifications of the red-light (red curve) and far-red light (blue curve) filters.
Figure 9: The photoconversion cross-sections of PhyB in the Pr (red curve) and Pfr (blue curve) states.


[1] Kendrick and Kronenberg: Photomorphogenesis in Plants, 2nd Edition (1994). Kluwer Academic Publishers, Dordrecht, The Netherlands.