Stellenbosch University
Welcome to Stellenbosch University
Confocal microscopists and data engineers collaborate to develop a new image analysis tool
Author: Lize Engelbrecht
Published: 11/11/2021

​Confocal microscopy is one of the least invasive high-resolution imaging technologies and therefore very powerful in visualising dynamic processes in live cells. It also allows for imaging different levels of a sample and eliminating all out-of-focus light in the process for better resolution in all the dimensions. Acquiring what is called a 'Z-stack' in this way is called 'optical sectioning', and during postprocessing, these images can be reconstructed to enable the researcher to visualise the sample in three dimensions. The acquisition of a Z-stack is therefore in essence a type of virtual sectioning that can be performed repetitively. This is in contrast to the once-off physical sectioning of a sample that occurs for example when imaging with an electron microscope for three-dimensional image reconstruction.

​In the Central Analytical Facilities Fluorescence Microscopy Unit, the Zeiss LSM780 ELYRA PS1 confocal microscope has been used extensively over the past nine years by various students following dynamic processes, such as cell migration and chemotaxis, cell stress and cell death, nuclear transportation and many more. Mitochondrial dynamics have been a specific focus of the Neuro Research Group led by Prof Ben Loos of the Department of Physiological Sciences. When the mitochondrial processes, such as fission and fusion, are impaired, it is usually one of the molecular indications for the onset of neurodegenerative disorders such as Alzheimer's or Parkinson's disease. 

In 2015, Prof Loos started a collaboration with Prof Thomas Niesler's group at the Department of Electronic and Electrical Engineering with a view to developing more automated, high-throughput tools for three-dimensional image visualisation and analysis. As part of this collaboration and his PhD research into virtual reality-guided visualisation and quantification of microscopy data in three dimensions, Dr Rensu Theart developed the Mitochondrial Event Localiser (MEL) tool to investigate the dynamics of the three-dimensional mitochondrial network. 


How does the Mitochondrial Event Localiser work?

The workflow firstly requires the Z-stack data to be preprocessed to ensure the best possible quality and consistency between image frames of the time lapse. Through a process called 'hysteresis thresholding', background signal is removed from each frame, resulting in a filtered binarised image. After compensating for any slight movement of the mitochondria, arrays of these stacks are ready for automatic processing.

The prepared arrays are the input to the MEL automatic image analysis algorithm that will analyse the data and produce a list of mitochondrial event locations. It will look for potential sites of (i) fusion, whereby two smaller fragments of mitochondria form a larger single structure in a subsequent frame, (ii) fission, whereby a large mitochondrial fragment separates into two smaller fragments, and (iii) depolarisation, whereby a fragment disappears in consecutive frames due to a loss of fluorescence signal. Through a process called 'back-and-forth structure matching', many locations of these events are identified and localised in the three-dimensional time lapse. The output generated by the MEL tool consists of the input Z-stacks with superimposed colour-coded event localisations.

Since the processing steps can produce false positives, the group went even further to develop a validation tool that enables a human expert to investigate each event individually by displaying the cropped image of the event identified by the MEL tool next to binary images of the frames just before and just after the event. The frames are further super-imposed, and the changed fragments are colour coded to aid the investigator to confirm that the event has indeed been identified correctly.

​In a recent publication in the academic journal PLOS One, the group showed that the number of fission, fusion and depolarisation events in healthy mammalian control cells was kept in equilibrium at an average ratio of 9.3/7.2/2.3 events. However, when treated with peroxide to perturb the mitochondrial network, the balance clearly shifted towards fusion at a ratio of 15/6/3. This was observed to settle quickly to a new equilibrium (6.2/6.4/3.4) that was more comparable to the control cells.

What can these types of results tell researchers?

The mitochondrial network and morphological changes are so dynamic that the interpretation of qualitative data and the limited quantitative data currently available is cause for debate amongst researchers. Some believe that a highly networked mitochondrial structure is indicative of improved cellular health, while a fragmented network indicates a cell that is under severe stress, which is detrimental. Others contend that a highly networked structure is only the first sign of a stress response and that fragmentation indicates increased cellular control through these adaptive mechanisms, to drive removal of dysfunctional mitochondria. Since the MEL allows tracing the development and direction of these processes, either towards fission, fusion or no change, in other words being in equilibrium, it will allow a much better understanding of this context. Hence, the truth might depend on the circumstances, and a simple observation of the network without regard to the dynamic behaviour could lead to major misinterpretations. Fission does not only separate mitochondria that are to be removed from the network but also separates mitochondria that need to be transported elsewhere in the cell, for example to the synapses in neurons. This can be a long distance, more than a meter for some neurons. In contrast, when mitochondria need to be protected against degradation, fusion may be the preferred response in the cell since material is shared and diluted, and respiration can take place across a larger network.

The accurate quantification of the relationship between fission, fusion and depolarisation in a three-dimensional cellular context will enable researchers to better describe the equilibrium under various conditions and to identify a deviation from this equilibrium and the consequent effects. This will have particular application in the study of neurodegenerative diseases and has immense potential not only to aid researchers in the development of treatments that target mitochondrial function but also to serve as a diagnostic tool for the early detection of these diseases. Coupled with powerful high-throughput platforms, the MEL could become a new standard, allowing a completely new categorisation of cells according to their fission and fusion behaviour under healthy and diseased conditions. Since the mitochondria are also the nexus of cellular fate, in other words life or death (apoptosis), this application may also be of value in the development of anticancer drugs, in which failing mitochondria are desired. 

Article published in the CAF Annual Report 2020/2021