Project Topics
Get Complete Project Material File(s) Now! »
Single Molecule tracking
To study proteins’ dynamics I performed Single Molecule (SM)tracking experiments in living mouse Embryonic Stem Cells (mESCs). The main reason behind the choice of SM microscopy is that in contrast to other approaches often used to study particles’ diffusive behaviour, such as Fluorescence Recovery After Photobleaching (FRAP) and Fluorescence Correlation Spectroscopy (FCS), SM imaging give access to the individual behaviour, avoiding ensemble averages. By doing so, different dynamic sub-population can be revealed; furthermore, looking at distributions, beyond mean values, helps maintaining the complexity and variability of many biological processes.
As stated in the previous section, there are particular requests to fulfil in order to localise and distinguish individual molecules.
To overcome the diffraction limited resolution, SM experiments rely on very low concentrations labelling, as shown in fig 2.3. Staining the cells with a nM or pM concentration of the fluorescent dye allows to assert that when a bright spot is detected, this spot is emanating from a single molecule. This ”bright spot” correspond to the PSF defined in the previous section. If all aberrations have been corrected, the PSF is a 2D Airy function, commonly approximated to a 2D gaussian profile in the lateral dimensions (x, y) (see fig 2.2). By fitting the PSF with a 2D gaussian, it is possible to identify the centre of the PSF and localise the probe, with a precision better than tens of nm (assuming few thousands of emitted photons per molecule) which is
less than ten nanometers in size (such as an organic dye), see fig 2.4.
Microscopy set-up
Single molecule imaging was performed on an epifluorescence inverted microscope (IX71, Olympus) in Highly Inclined and Laminated Optical sheet, or HILO illumination. Proper HILO was achieved using a slit (or field stop) placed in the plane conjugated to the specimen plane (OWIS, 14.021.0020, RT 40-20-R). Lasers beams were focused in the back focal plane of a 150X objective lens (UAPON 150XOTIRF, Olympus, France), selected by an excitation quadband dichroic supplied with the corresponding emission filters (Chroma, TRF89901-EMv2 – ET – 405/488/561/640nm Laser Quad Band Set for TIRF application). Lasers were tuned via an acousto-optical tunable filter (AOTFnC-400-650-TN, A&A Optoelectronic, France) and controlled by a home-made interface in Micromanager (Edelstein et al., 2014). Laser power was adjusted to have a density of 0.1kW/cm2. Signal was acquired with an EM-CCD camera (iXonEM DV860DCS-BV, Andor, Ireland) run in frame transfer mode. The setup is provided with a 405 nm laser (Cube 405- 100C, Coherent, Santa Clara, CA, USA), a 488 nm laser (35LAL030- 220, CVI, Melles-Griot, France) and a 561 nm laser (Genesis MX 561-2000 MTM, Coherent, Santa Clara, CA, USA).
Biological system
To perform live single molecule imaging an Halo-tag was encoded in the protein sequence. Thanks to the tag I could label individual proteins and follow them around the nucleus.The cell lines were edited by Elph`ege P. Nora in the laboratory of Benoit Bruneau at the Gladstone Institute in San Francisco, California (USA).
Cohesin has been tracked in the context of various alterations. In particular I studied how Cohesin dyamics is affected if we deplete factors like CTCF, Sororin and Nipbl. Figure 2.7 reports a sketch of the edited cell line where I tracked Cohesin in the absence of CTCF and the concept of the auxin inducible degradation system is represented. The different conditions, and the corresponding cell lines are listed in the table reported below.
Analysis of single molecule imaging
One single dataset corresponds to the pool of trajectories obtained from a single cell, which is of the order of thousands trajectories. For each biological condition 10-15 cells were imaged.
To localise the single emitters and build the trajectories we used a homemade software (SLIMFast, (Normanno et al., 2015)) implemented in Matlab and based on the MTT algorithm (Serg´e et al., 2008). The Point Spread Function (PSF) of a single emitter is fitted with a 2D-gaussian, whose center corresponds to the position of the fluorophore with a sub-pixel resolution.
To exhaustively characterise CTCF and Cohesin dynamics, I performed experiments at different acquisition rates. This approach enabled me to overcome the bleaching limit and cover different timescales. I performed acquisitions with an exposure of 5ms, 50ms and 500ms; when increasing the exposure time the laser power was reduced coherently (keeping the Signal to Noise qualitatively constant).
Data from experiments at 5ms exposure were used to quantify the fraction of bound molecules, since at this rate there is no bias towards one of the dynamic subpopulations. In fact at slower rates highly mobile proteins are blurred and often not properly localised. Data from experiments at 50ms served to characterise the dynamics via the computation of the Mean Square Displacement (MSD) and the binding kinetics, with the residence time distribution, or Survival Probability. Data from experiments at 500ms were used to quantify the binding kinetics on longer time-scales. At such rate I was indeed completely biasing the acquisitions and the analysis towards stably bound molecules.
The methods used to analyse the trajectories will be explained in detail in the following paragraphs.
Analysis of binding kinetics
To quantify the number of bound molecules I exploited trajectories from the acquisition at 5ms. By doing so I could include all the trajectories, even the shortest ones consisting of only 2 displacements.
The analysis was performed with SpotOn (Hansen et al., 2018). The method is based on a fit of the distribution of all the step lengths performed by the molecules. The idea behind is that proteins can be bound or freely diffusing. The state of the protein (bound or unbound) is reflected in the distribution of the steps that it perform: a bound molecules will give raise to small steps while a freely diffusing one will show longer steps. The kinetic modelling is inspired by (Mazza et al., 2012). The so-called jump distribution is resumed by the following expression: P(r,t) = Fbound r 2(Dboundt + 2).
Table of contents :
Introduction
1 On Chromatin Architecture
1.1 Chromatin structure and function
1.1.1 Chromatin structure: how to characterize it
1.1.2 Chromatin structures span over different length-scales .
1.2 On Topologically Associating Domains (TADs)
1.2.1 How are TAD formed: the loop extrusion hypothesis .
1.3 CTCF and Cohesin: a focus on chromatin organisers
1.3.1 CTCF
1.3.2 Cohesin
1.3.3 CTCF, Cohesin and chromatin structure
1.4 Goal of this work
2 Technique and system
2.1 Optical microscopy
2.1.1 Diffraction limit and fluorescence
2.1.2 Single Molecule tracking
2.2 Microscopy set-up
2.3 Biological system
2.3.1 Cell Culture
2.4 The experiments
2.4.1 The degron system
2.4.2 Labelling and imaging conditions
2.5 Analysis of single molecule imaging
2.5.1 Analysis of binding kinetics
2.5.2 Residence time
2.5.3 Analysis of dynamics
3 Results and discussion
3.1 Chromatin as a reference
3.2 CTCF
3.3 Cohesin
3.4 Cohesin without CTCF
3.4.1 Cohesin in non-cycling cells
3.5 Other mutants
3.5.1 Cohesin in absence of Sororin
3.5.2 Cohesin in absence of Nipbl (Scc2)
3.5.3 Control
3.6 Discussion
3.6.1 CTCF
3.6.2 Cohesin
3.7 Conclusion and perspectives
Acknowledgements
3.8 R´esum´e en fran¸cais
References
