Multiscale light-sheet organoid imaging framework

Gustavo de Medeiros, Raphael Ortiz, Petr Strnad, Andrea Boni, Francisca Maurer, Prisca Liberali

Posted on: 29 August 2021 , updated on: 1 September 2021

Preprint posted on 12 May 2021

Article now published in Nature Communications at http://dx.doi.org/10.1038/s41467-022-32465-z

Light-sheet illumination of the organoid family tree with LSTree

Selected by Osvaldo Contreras

Categories: biochemistry, bioengineering, cell biology, developmental biology

Background

Organoids, multicellular in vitro models of their corresponding organs, are an excellent resource for understanding the cellular and developmental mechanisms that regulate tissue and organ growth. In recent years, organoids have relieved some of the difficulties and complexities of working with organs in vivo. Intestinal organoids are a great example of a reproducible system capable of recapitulating not only major morphological features (e.g., crypt formation) [1] but also one of the major intestine capabilities — its regenerative response [2].

One advantage of working with organoids is that their development can be precisely followed under a microscope, potentially allowing the description of their cells’ genealogical tree or lineage. However, imaging and analyzing organoid growth remains challenging, as it requires long-term imaging microscopy techniques coupled with cell-tracking and dedicated analysis pipelines to recover morphological or cell lineage information.

In this exciting new preprint, Liberali and colleagues leverage their extensive organoid-imaging knowledge [3] to present LSTree (an acronym for Light-Sheet Tree), a light-sheet imaging framework to analyze and digitally visualize organoids. LST goes from image pre-processing, cell lineage tracking, and segmentation to analytical tools to extract morphological and microscopy features. The authors used this light-sheet microscopy-based imaging framework to analyze and reconstruct, at an unprecedented spatiotemporal resolution, the cell growing dynamics and cell lineage relationships during intestinal organoid development.

About the preprint

Imaging framework

Medeiros et al. took advantage of the abundance of knowledge on intestinal organoids and used it to provide quantitative information on intestinal organoid growth dynamics using light-sheet and spinning disk microscopy.

The authors based their initial work on previously published protocols for imaging organoids, using cultured mouse intestinal crypt organoids and single-cell dissociation, followed by fluorescent-activated cell sorting (FACS) and embedding into a Matrigel drop. Hence, organoid-derived single cells were cultured into a sample holder specially designed for light-sheet imaging using a dual-illumination inverted light-sheet microscope. In addition, the authors developed a position-dependent illumination alignment step to obtain the best image quality possible. Aiming to reduce the storage needs, the authors compressed and cropped the images. The low signal-to-noise ratio, due to the very low laser intensity to avoid phototoxicity, was improved by denoising and deconvoluting the images.

To track and record the cell and cell lineages, Medeiros et al. worked with intestinal organoid-derived cells expressing the nuclear Histone 2B and mem9 membrane peptide, each tagged with fluorescent proteins (Fig. 1). Then, the authors recorded the time-lapse growth of the intestinal organoids every 10 min for about four days to study the individual cell dynamics by semi-automated tracking and single-cell segmentation. Owing to the 3D spatiotemporal dissection of the intestinal cellular dynamics, the authors were able to extract 3D information and plot the data over this time period, shedding light on cell division, cell and nucleus volume, and density, as well as cell changes induced by the daily media changes. Remarkably, they found nuclei volume decreased with each cell locating more distantly to the organoid lumen with each LSTree generation.

Fig. 1. Light-sheet high-resolution imaging framework. Modified from Medeiros et al. 2021.

Deep learning 4D organoid segmentation and multiscale digital viewer

With their protocol established, the authors added their image processing and data analyses into LSTree (for details visit: https://github.com/fmi-basel/LSTree), which was implemented as a Luigi (i.e., a workflow engine) (https://github.com/spotify/luigi). The next challenge was to perform organoid segmentation with their framework. The authors introduced a number of different segmentation strategies for organoid segmentation (cells and nuclei) that use convolutional neural networks and a trained deep learning model. Using this strategy, they extracted several distinct features from the organoid cells. Continuing with the framework development, Medeiros et al. trained and refined a deep learning model together with developing a segmentation-tracking approach to map the association between nuclei in two consecutive frames in 4D. With this achieved, the authors created a digital organoid viewer to facilitate data exploration. Remarkably, the viewer combines lineage cell trees and cell meshes, representing a multiscale organoid digital web-viewer (Fig. 2).

Fig. 2. Highlights of the organoid digital viewer developed by the authors.

Functional imaging through fixation and backtracking

Aiming to address the end state of the organoid cells and to overcome several problems involved with imaging multiple fluorescent organoid lines (i.e., reporter lines), the authors included fixation and immunolabelling as the last step after the recording. Following live recording, they integrated fixation and immunofluorescence to detect the cells expressing their proteins of interest. Hence, post-fixation staining of DLL1 (also known as Delta-like ligand 1 precursor, a marker for secretory lineage cells) and Lysozyme (a marker for differentiated Paneth cells) allowed the authors to ask about symmetry breaking of the intestinal organoid. Of note, the authors also calculated that on average an organoid cell has 5 closest neighbors. Thus, the authors were able to image and backtrack the stained cells with the light-sheet live recordings, enhancing the discovery capabilities of LSTree.

Polyploidy events during early gut organoid development

A challenging observation was that some cells went through a few rounds of failed cell division, meaning that two sister nuclei divided into two instead of four nuclei at the end of their cell cycle. The authors showed that the cause of these failed division events was an issue during cytokinesis. Interestingly, the organoids seem to exclude these cytokinesis failure events from the crypt, but a high proportion of the “failed” cell progeny localized to the villus. Thus, the authors proposed that the organoid system possesses an intrinsic self-preserving mechanism by avoiding cell division failure or damaged cells within the crypt – something they speculate could also occur in the gut. To finish, Medeiros et al. showed that LATS1 and YAP participate in regulating successful mitosis and cytokinesis, whereas perturbations of these resulted in inefficient mitosis and cell division.

What I liked about this preprint

This impressive multivariate and multiscale analytical framework was applied from a single cell to hundreds of cells and encompassed image pre-processing, lineage tracking, and segmentation coupled with analytical tools. Therefore, LSTree allowed the researchers to extract single-cell morphological features in 4D. In addition to developing a light-sheet microscopy pipeline combined with a web-based “digital organoid viewer”, the authors proposed a model of tissue development and integrity maintenance that challenges the concept of polyploidy in tissue homeostasis and regeneration. This remarkable approach will open new avenues in our understanding of organoid development and might allow further multi-parametric functional characterization of different organoid growth.

Future directions and questions to the authors

Is there any potential for combining your LSTree approach with spatial transcriptomics?
Have you applied this light-sheet imaging strategy and LSTree to study other kinds of organoids? I wonder whether other tissue-specific organoids might represent hurdles and challenges to further test LSTree applicability and feasibility.
What was the most challenging aspect of creating LSTree?
Have you explored the roles of YAP-mediated mechanotransduction or ECM/nuclear mechanotransduction of YAP as regulators of successful cell division/cytokinesis? Any thoughts if not explored yet?

Acknowledgments

Osvaldo Contreras acknowledges Irepan Salvador for initial discussions of the Medeiros et al. preprint and Helen Robertson for valuable comments on this PreLight.

References

[1] Serra, D., Mayr, U., Boni, A. et al. Self-organization and symmetry breaking in intestinal organoid development. Nature 569, 66–72 (2019). https://doi.org/10.1038/s41586-019-1146-y

[2] Lukonin, I., Serra, D., Challet Meylan, L. et al. Phenotypic landscape of intestinal organoid regeneration. Nature 586, 275–280 (2020). https://doi.org/10.1038/s41586-020-2776-9

[3] Yang, Q., Xue, SL., Chan, C.J. et al. Cell fate coordinates mechano-osmotic forces in intestinal crypt formation. Nat Cell Biol 23, 733–744 (2021). https://doi.org/10.1038/s41556-021-00700-2

Tags: gut, imaging, microscopy, organoids, regeneration, segmentation, single cell

doi: https://doi.org/10.1242/prelights.30401

Read preprint

(2 votes)

Author's response

Gustavo de Medeiros shared

Dear Osvaldo,

Thank you for selecting our work for this issue of preLights!

Regarding your questions:

Is there any potential for combining your LSTree approach with spatial transcriptomic? One of the things we really like about LSTree is the power of combining different data representations (lineage trees and segmented cells) within the same visualization tool, and this is one aspect that could be translated to spatial transcriptomics: to have a visualization of the RNA-seq data – e.g. via TSNE or UMAP – side by side with the corresponding imaging data. This way one could e.g. select particular cells / regions of interest from the imaging data and the corresponding points on the RNA-seq data would be highlighted, with all other properties of the cells embedded as a list of features on the RNA-seq plot, thus giving a more complete understanding of the information at hand.
Have you applied this light-sheet imaging strategy and LSTree to study other kinds of organoids? I wonder whether other tissue-specific organoids might represent hurdles and challenges to further test LSTree applicability and feasibility. Right now we are applying our LSTree method to other already acquired 3D+T imaging data in order to properly benchmark it and test its limits. As the recording strategy and the LSTree framework are not directly coupled, one can apply 4D datasets coming from different microscope types into the framework. The main care here is then to have good training sets for the neural networks.
What was the most challenging aspect of creating LSTree? Although tracking is one of the most basic features needed in live recorded data, most tracking approaches rely on utilizing many partial tracks in order to create lineage trees that in turn resemble an average of all of the observed systems. In our case, we needed to have trees that are 100% correct, and for that we tried to get lineage tree predictions that can avoid us having too much time on the correction / extension of these initial tracks. Turns out this fine tuning can be quite delicate depending on the data at hand, and one still needs some training sets to have a good enough network.
To facilitate this, we are now looking at ways to combine other lineage tree prediction methods to the following steps from the LSTree framewrok, making the framework more modular and open.
Have you explored the roles of YAP-mediated mechanotransduction or ECM/nuclear mechanotransduction of YAP as regulators of successful cell division/cytokinesis? Any thoughts if not explored yet? This is an interesting point, as YAP-mediated mechanotransduction or ECM/nuclear mechanotransduction of YAP may have interesting roles on the systematic appearance of mitotic defects during early organoid growth from single cells.
From looking at the recordings, during the moments where we see mitotic errors the intestinal organoids usually seem quite “malleable”, i.e. cells do not seem very stiff overall. One then could speculate that the very variability in YAP, necessary from a molecular perspective for proper symmetry breaking, might lead some cells to change their biomechanical properties to a point where cytokinesis cannot be completed. Definitely worth checking further!

Kind regards,
Gustavo

1 comment

3 years

Gustavo de Medeiros

Dear Osvaldo,

thank you for selecting our work for this issue of preLights!

Regarding your questions:

1) One of the things we really like about LSTree is the power of combining different data representations (lineage trees and segmented cells) within the same visualization tool, and this is one aspect that could be translated to spatial transcriptomics: to have a visualization of the RNA-seq data – e.g. via TSNE or UMAP – side by side with the corresponding imaging data. This way one could e.g. select particular cells / regions of interest from the imaging data and the corresponding points on the RNA-seq data would be highlighted, with all other properties of the cells embedded as a list of features on the RNA-seq plot, thus giving a more complete understanding of the information at hand.

2) Right now we are applying our LSTree method to other already acquired 3D+T imaging data in order to properly benchmark it and test its limits. As the recording strategy and the LSTree framework are not directly coupled, one can apply 4D datasets coming from different microscope types into the framework. The main care here is then to have good training sets for the neural networks.

3) Although tracking is one of the most basic features needed in live recorded data, most tracking approaches rely on utilizing many partial tracks in order to create lineage trees that in turn resemble an average of all of the observed systems. In our case, we needed to have trees that are 100% correct, and for that we tried to get lineage tree predictions that can avoid us having too much time on the correction / extension of these initial tracks. Turns out this fine tuning can be quite delicate depending on the data at hand, and one still needs some training sets to have a good enough network.
To facilitate this, we are now looking at ways to combine other lineage tree prediction methods to the following steps from the LSTree framewrok, making the framework more modular and open.

4) This is an interesting point, as YAP-mediated mechanotransduction or ECM/nuclear mechanotransduction of YAP may have interesting roles on the systematic appearance of mitotic defects during early organoid growth from single cells.
From looking at the recordings, during the moments where we see mitotic errors the intestinal organoids usually seem quite “malleable”, i.e. cells do not seem very stiff overall. One then could speculate that the very variability in YAP, necessary from a molecular perspective for proper symmetry breaking, might lead some cells to change their biomechanical properties to a point where cytokinesis cannot be completed. Definitely worth checking further!

Kind regards,
Gustavo

Reply Report

Permalink

Have your say Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Also in the biochemistry category:

Triglyceride metabolism controls inflammation and APOE4-associated disease states in microglia

Roxan A. Stephenson, Kory R. Johnson, Linling Cheng, et al.

Selected by 22 August 2024

Gustavo Stelzer, Marcus Oliveira

Impaired 26S proteasome causes learning and memory deficiency and induces neuroinflammation mediated by NF-κB in mice

Christa C. Huber, Eduardo Callegari, Maria Paez, et al.

Selected by 22 August 2024

Gustavo Stelzer, Marcus Oliveira

Discussion

Notch3 is a genetic modifier of NODAL signalling for patterning asymmetry during mouse heart looping

Tobias Holm Bønnelykke, Marie-Amandine Chabry, Emeline Perthame, et al.

Selected by 06 June 2024

Bhaval Parmar

Discussion

Also in the bioengineering category:

Engineered Nanotopographies Induce Transient Openings in the Nuclear Membrane

Einollah Sarikhani, Vrund Patel, Zhi Li, et al.

Selected by 23 September 2024

Sristilekha Nath

Scalable and efficient generation of mouse primordial germ cell-like cells

Xinbao Ding, Liangdao Li, Jingyi Gao, et al.

Selected by 05 March 2024

Carly Guiltinan

Generalized Biomolecular Modeling and Design with RoseTTAFold All-Atom

Rohith Krishna, Jue Wang, Woody Ahern, et al.

Selected by 24 January 2024

Saanjbati Adhikari

Discussion

Also in the cell biology category:

Long range mutual activation establishes Rho and Rac polarity during cell migration

Henry De Belly, Andreu Fernandez Gallen, Evelyn Strickland, et al.

Selected by 22 October 2024

Vibha SINGH

Intracellular diffusion in the cytoplasm increases with cell size in fission yeast

Catherine Tan, Michael C. Lanz, Matthew Swaffer, et al.

Selected by 18 October 2024

Leeba Ann Chacko, Sameer Thukral

Discussion

Deciphering the nanoscale architecture of presynaptic actin using a micropatterned presynapse-on-glass model

Sofia Tumminia, Louisa Mezache, Theresa Wiesner, et al.

Selected by 18 October 2024

Felipe Del Valle Batalla

Also in the developmental biology category:

Actin-based deformations of the nucleus control multiciliated ependymal cell differentiation

Marianne Basso, Alexia Mahuzier, Syed Kaabir Ali, et al.

Selected by 30 October 2024

Ryan Harrison

HIF1A contributes to the survival of aneuploid and mosaic pre-implantation embryos

Estefania Sanchez-Vasquez, Marianne E. Bronner, Magdalena Zernicka-Goetz

Selected by 11 October 2024

Anchel De Jaime Soguero

Axis reset is rate limiting for onset of whole-body regenerative abilities during planarian development

Clare L.T. Booth, Brian C. Stevens, Clover A. Stubbert, et al.

Selected by 10 October 2024

Isabella Cisneros

Discussion

preLists in the biochemistry category:

BSCB-Biochemical Society 2024 Cell Migration meeting

This preList features preprints that were discussed and presented during the BSCB-Biochemical Society 2024 Cell Migration meeting in Birmingham, UK in April 2024. Kindly put together by Sara Morais da Silva, Reviews Editor at Journal of Cell Science.

Multiscale light-sheet organoid imaging framework

Background

About the preprint

Imaging framework

Deep learning 4D organoid segmentation and multiscale digital viewer

Functional imaging through fixation and backtracking

Polyploidy events during early gut organoid development

What I liked about this preprint

Future directions and questions to the authors

Acknowledgments

References

Share this:

1 comment

Have your say Cancel reply

Sign up to customise the site to your preferences and to receive alerts

Also in the biochemistry category:

Triglyceride metabolism controls inflammation and APOE4-associated disease states in microglia

Impaired 26S proteasome causes learning and memory deficiency and induces neuroinflammation mediated by NF-κB in mice

Notch3 is a genetic modifier of NODAL signalling for patterning asymmetry during mouse heart looping

Also in the bioengineering category:

Engineered Nanotopographies Induce Transient Openings in the Nuclear Membrane

Scalable and efficient generation of mouse primordial germ cell-like cells

Generalized Biomolecular Modeling and Design with RoseTTAFold All-Atom

Also in the cell biology category:

Long range mutual activation establishes Rho and Rac polarity during cell migration

Intracellular diffusion in the cytoplasm increases with cell size in fission yeast

Deciphering the nanoscale architecture of presynaptic actin using a micropatterned presynapse-on-glass model

Also in the developmental biology category:

Actin-based deformations of the nucleus control multiciliated ependymal cell differentiation

HIF1A contributes to the survival of aneuploid and mosaic pre-implantation embryos

Axis reset is rate limiting for onset of whole-body regenerative abilities during planarian development

preLists in the biochemistry category:

BSCB-Biochemical Society 2024 Cell Migration meeting

Peer Review in Biomedical Sciences

CellBio 2022 – An ASCB/EMBO Meeting

20th “Genetics Workshops in Hungary”, Szeged (25th, September)

Fibroblasts

ASCB EMBO Annual Meeting 2019

EMBL Seeing is Believing – Imaging the Molecular Processes of Life

Cellular metabolism

MitoList

Also in the bioengineering category:

CSHL 87th Symposium: Stem Cells

EMBL Synthetic Morphogenesis: From Gene Circuits to Tissue Architecture (2021)

3D Gastruloids

ASCB EMBO Annual Meeting 2019

EMBL Seeing is Believing – Imaging the Molecular Processes of Life

Lung Disease and Regeneration

Advances in microscopy

Also in the cell biology category:

BSCB-Biochemical Society 2024 Cell Migration meeting

‘In preprints’ from Development 2022-2023

preLights peer support – preprints of interest

The Society for Developmental Biology 82nd Annual Meeting

CSHL 87th Symposium: Stem Cells

Journal of Cell Science meeting ‘Imaging Cell Dynamics’

9th International Symposium on the Biology of Vertebrate Sex Determination

Alumni picks – preLights 5th Birthday

CellBio 2022 – An ASCB/EMBO Meeting

Fibroblasts

EMBL Synthetic Morphogenesis: From Gene Circuits to Tissue Architecture (2021)

FENS 2020

Planar Cell Polarity – PCP

BioMalPar XVI: Biology and Pathology of the Malaria Parasite

Cell Polarity

TAGC 2020

3D Gastruloids

ECFG15 – Fungal biology

ASCB EMBO Annual Meeting 2019

EMBL Seeing is Believing – Imaging the Molecular Processes of Life

Autophagy

Lung Disease and Regeneration

Cellular metabolism

BSCB/BSDB Annual Meeting 2019

MitoList

Biophysical Society Annual Meeting 2019

ASCB/EMBO Annual Meeting 2018

Also in the developmental biology category:

BSDB/GenSoc Spring Meeting 2024

GfE/ DSDB meeting 2024