In this new series "From Image to Results", explore various case studies explaining how to reach results from your demanding samples and acquired images in an efficient way. For each case study, we highlight different samples, imaging systems, and research questions. Find all previous episodes here.
In this fourth episode, we study a nematode (C. elegans) embryo during the very early stage of embryonic cell divisions.
|Case Study Overview|
Cell division tracking
Precise description of the distinct early waves of cell division
ZEISS Lattice Lightsheet 7
Embryogenesis describes the gradual process of developing differentiated cells, tissues, organs and eventually, specific body structures out of an initially uniform cell mass. The precise kinetics, dynamics and signaling events taking place during this period are of major interest for developmental biologists. Studies are typically performed with abundantly available and fast-growing model systems like flies, frogs or worms.
In this use case, we study a nematode (C. elegans) embryo during the very early stage of embryonic cell divisions. C. elegans is a particularly important model organism for developmental biology, because it develops with timed cell divisions and cell differentiation dynamics, reaching an adult stage with precisely 959 cells. This allows study of the function of genes or drugs by observing perturbations in the normal development scheme1.
Microscopic imaging has been established as a key method for studying early embryos, because only microscopy can decipher the spatial context of cell divisions and cell migration patterns. However, commonly used microscopy techniques often are too invasive for observing these delicate processes over extended periods of time. In this use case, we show how ZEISS Lattice Lightsheet 7 can be used to make gentle long-term imaging of sensitive specimens like these embryos feasible.
As noted above, the coordination of cell divisions is of particular interest for analysis. The very first cell cycles in C. elegans are completely synchronized and fast (15 minutes) but become increasingly longer and asynchronous1. Thus, we focused our image analysis, using arivis Vision4D, on the precise description of the distinct early waves of cell division in our data set.
1 Koreth, J., van den Heuvel, S. Cell-cycle control in Caenorhabditis elegans: how the worm moves from G1 to S. Oncogene 24, 2756–2764 (2005). https://doi.org/10.1038/sj.onc.1208607
The C. elegans embryos were created by William Okafornta, TU Dresden, Germany. They are equipped with fluorescent marker proteins for cell membranes (PH::mKate2), nuclei (H2B::mCherry) and centrosomes (gammaTub::GFP). This allows continuous imaging of the living embryo together with its cellular structures to analyze cell division cycles and cell compartmentalization.
The raw data for this use case was captured with ZEISS Lattice Lightsheet 7. This system is specifically tailored to gentle, volumetric live cell imaging, making it the perfect instrument for observing delicate structures over long periods of time. The C. elegans embryos in this use case were imaged in the green and red channels. Thus, cell membranes and nuclei were merged into one channel. Live acquisition was performed for almost 2 hours, one volume every 30 seconds, 476 planes per volume — adding up to a total of 120,000 images.
Figure 1A: Deskewed and deconvolved dataset of C.elegans embryo as a maximum projection movie. Membranes and nuclei in green, spindle poles in purple. Cell divisions can be observed by splitting the centrosomes and by membrane compartmentalization.
Figure 1B: Deskewed and deconvolved dataset of C.elegans embryo as a maximum projection movie with added graph showing number of centrosomes, mean volume, and sphericity over time.
When using lattice light sheet microscopy, the sample is imaged at an angle, due to the geometry inherent to this technology. Consequently, the raw data requires processing (deconvolved and deskewed) in ZEN (blue edition) (Figure 1A and Figure 1B) before further analyses can be performed.
The processed .czi file from ZEN (blue edition) was directly imported into Vision4D and converted into arivis' .sis file format, a format specifically designed for fast handling of large data sets. In order to reduce the original data set of ~86 GB for faster processing and presentation, we cropped the images to include only the embryo volume. We then used only every second time point from each dataset , resampled the dataset to 50% in the Z dimension and converted to dataset to 8-bit. The final dataset contained 2 channels, 79 slices, 115 time points, with a size of ~1.8 GB.
arivis Vision4D® is a modular software for working with multi-channel 2D, 3D and 4D images of almost unlimited size, highly scalable and independent of local system resources. Many modern microscope systems such as high-speed confocal, light sheet / SPIM, super-resolution, electron microscopy or X-ray instruments can produce huge amounts of imaging data. arivis Vision4D handles such datasets without constraints and in relatively short time.
For image analysis we aimed at segmenting centrosomes and the cellular compartments. Centrosomes are very straightforward to segment, because the centrosome fluorescence signal is very clear. After Background Subtraction and Denoising pre-processing, centrosomes were segmented by watershed segmentation.
Segmenting cell compartments is trickier. Vision4D has segmentation functions to detect the cell membranes and define the compartments, however, the embryo has no membrane fluorescence at the outside border. To circumvent this, an outer embryo edge was processed from the red channel background fluorescence and merged with the membrane channel. With this derived image set, cell compartments were reliably segmented by region growing from the segmented centrosomes.
Finally, segmented centrosomes and compartments were grouped per time point to extract temporal average measurements of numbers, volumes and intensities.
The sketch summarizes the image analysis procedure. Additionally, the Vision4D pipeline for performing these operations will be available at the end of the story.
To validate sufficient segmentation of centrosomes, segmented centrosomes were overlayed with the raw image (Figure 2A). Objects are color-coded based on their time of appearance. Centrosomes vary in size and intensity throughout the cell cycle. All centrosomes, including the small interphase centrosomes, were successfully segmented.
Cell compartments were more difficult to segment, however, the membrane watershed segmentation appropriately found most membrane cell borders (Figure 2B). Of note, the outline of the embryo was segmented sufficiently to perform measurements of compartment volumes.
Figure 2A: Centrosome segment overlay with raw image. Membranes and nuclei in green, spindle poles in purple. Centrosome segments are color-coded by time point ranging from blue to red.
Figure 2B: Compartment segments validation. Compartment segments are shown as transparent outlines in random colors. Centrosome segments in cyan, raw membranes in green.
An outstanding property of early embryonic development is the orchestrated mitoses that take place in a very fast succession. Thus, timing and spatial characteristics of ongoing cell divisions are a primary read-out from such data sets. Typically, evaluation of nuclei is the most straightforward way to obtain such results. In the present data set, however, nuclear fluorescence intensity is very weak and therefore not directly accessible for analysis.
To arrive at a meaningful interpretation of results, we make use of the centrosomes as a proxy for cell division. Careful inspection of Figure 2A reveals that centrosomes have a characteristic pattern of “growth” throughout the cell cycle. They are small during interphase but increase in size during mitosis. In Figure 3A and 3B, we plotted the time-dependent mean volume and mean sum intensities of segmented centrosomes, with both read-outs having almost the same shape. Within these plots, four mitotic waves are discernable as distinct peaks. Measuring the distances between peaks shows that average cycling times become increasingly prolonged, from 17.3 minutes between first and second wave to ~32 minutes between third and fourth waves. Of note, the peaks also get wider, indicating increasingly asynchronous mitotic waves. To quantify this, the width at half peak intensity (FWHM) was measured and plotted in Figure 3C. For the first mitotic wave this width is ~5 minutes, while it is ~18 minutes for the last wave.
Another read-out of interest is the quantification of cells within the embryo, typically done from counting the nuclei. But, in the absence of a good nucleus staining, quantification of centrosomes also can approximate the cell counts. This is because every cell has one centrosome until entry into mitosis, when the centrosome splits to form the two mitotic spindle poles.
Figure 4A shows the number of centrosomes over time. In each mitotic wave, we observe a doubling the number of centrosomes, which in turn indicates that in each wave nearly all cells undergo division. The horizontal lines cross the vertical lines to highlight the number of centrosomes after each mitotic wave, indicating 12 centrosomes following wave 1, 24 centrosomes following wave 2, 46 centrosomes following wave 3 and 88 centrosomes following wave 4.
Surprisingly, the sphericity of the centrosome follows the distinct pattern of mitotic waves. The data demonstrates that centrosomes in between mitotic events have a near-spherical shape but become deformed during mitotic events, resulting in a loss of sphericity (Figure 4B).
Similarly, certain features from cellular compartments can be plotted to extract more information about the developmental process. A peculiar observation about early embryogenesis is that the generation of new cells is prioritized over the absolute volume growth of the embryo. Figure 5A shows that the overall volume of the embryo does not change significantly over the course of the 4 mitotic waves (outliers in this plot show that compartment segmentation sometimes fails to detect the complete embryo). Hence, as the number of cells within the embryo increases, the average volume of each cell compartment decreases.
As for centrosomes, the sphericity of cell compartments shows an inverse pattern with mitotic waves (Figure 5B). Cells elongate and form oval shapes while splitting into two daughter cells and assume a spherical shape in interphase.
In this use case, we highlighted imaging of a C.elegans embryo during early development. ZEISS Lattice Lightsheet 7 allowed us to gently acquire the live organism with great spatial and temporal resolution.
Using arivis Vision4D, we were able to segment centrosomes and cell compartments by applying highly complex image processing techniques to derive meaningful parameters describing the dynamics of embryonic development.
Over the course of 100 minutes, the embryo underwent 4 discernible waves of cell division. By extracting and evaluating morphological features from the centrosomes (volumes, fluorescence intensity, shape) and cell compartments (volumes, shapes), we could determine the characteristics of the cell divisions: Cell divisions are synchronous and fast at the beginning of development, but then become increasingly slower and more asynchronous. Mitoses are accompanied with progressive compartmentalization of the embryo, leading to increasingly smaller cell volumes. Further, centrosomes and cell compartments undergo periodic changes in shape, from a spherical shape in interphases to a prolonged phase during mitosis. This nicely summarizes what has been reported in the literature.
Object segmentation as performed here does have its limitations. For example, we couldn’t segment cell nuclei with conventional methods because of the faint signals. This method also has set limits on for object tracking, which would allow lineage tracing of individual developing cells. In a future application story, we will show how these limitations can be overcome by using modern Deep Learning segmentation methods with ZEISS APEER machine learning embedded in arivis Vision4D.
As always, we provide here our data set and analysis pipeline for you to test and learn arivis Vision4D.