Imaging of cell dynamics and interactions across scales with self-driving, multi-scale microscopy

Introduction: : Biological systems are inherently multi-scale. They integrate processes across a wide range of spatial and temporal scales, from molecules and sub-cellular dynamis to whole organism behavior. This multi-scale integration is also fundamental to disease progression such as cancer metastasis, the leading cause of cancer-associated death [1]. In cancer metastasis, localized molecular interactions drive cancer cell spreading across an entire organism.

To image sub-cellular processes, cells, and organisms, light-sheet microscopy has become the method of choice over the past decade due to its fast and gentle acquisition scheme [2,3]. However, when selecting a microscope, we often must select a specific scale for imaging. Therefore, it has remained a fundamental challenge to dynamically image entire living organisms over long-time periods with simultaneous observation of selected sub-cellular processes.

To overcome these challenges, we introduce a novel, versatile multi-resolution light-sheet microscope platform that enables observation and quantification of subcellular behaviors alongside entire organisms [4]. Importantly, the microscope is self-driving, i.e., it automatically follows and images cells and regions of biological interest. This enables imaging over hours to days as otherwise regions of interests quickly move out of view in living organisms due to sample growth and changes.

Materials and

Methods: : For self-driving, multi-scale microscopy, we developed innovative microscope hardware, control software, and data analysis tools [4]. We combined the strengths of low-resolution light-sheet imaging [5] with sub-cellular axially swept light-sheet imaging [6]. Thereby, our microscope switches within seconds from subcellular imaging (390 nm lateral and 650 nm axial resolution), to whole organism imaging at micron-level resolution (Fig. 1a).

Guided by custom Python hardware control software, the microscope is self-driving, i.e., it adjusts the region for high-resolution imaging on-the-fly as the sample changes and grows over hours of imaging without any user interactions. Importantly, our Python hardware control software is designed from scratch also with future application of smart and event-driven microscopes in mind.

To study cancer progression and metastasis in a near-physiological environment in vivo, zebrafish cancer models and xenografts are powerful models [7,8]. They particularly excel due to their fast development, high-throughput, and optical translucency, making them ideally suited for optical imaging. Here, we leverage the zebrafish model with our microscope system. We present self-driving, multi-scale imaging studies of cellular dynamics and growth of Ewing Sarcoma, osteosarcoma, melanoma, breast cancer and glioblastoma models and xenografts in zebrafish.

Results, Conclusions, and Discussions:: We demonstrate the power of our novel imaging platform with imaging of developmental processes and models of cancer invasion, metastasis, and immune-cancer cell interactions in vitro and in vivo. Specifically, the resulting multi-scale image data reveals: i) How cells divide during gastrulation and later in development form new vessels during vascular growth. ii) How cancer cells divide at the invasive front of cancer spheroids. iii) How breast cancer cells form an intriguing network of protrusions in the vasculature. iv) How local immune-cancer cell interactions (Fig. 1 b-d) induce shape changes in macrophages and trigger behavioral changes of macrophages throughout a whole organism. Moreover, our platform uncovers that the clearance of cancer cells is cell-line dependent; more aggressive melanoma cells appear to have mechanisms to evade immune responses in vivo despite close contact with macrophages, while moderately aggressive osteosarcoma cells are phagocytosed, paving the way to study immune evasion of different cancer cells within a physiological tumor microenvironment with molecular and mechanistic insights. Notably, to achieve quantitative analysis of the observed biology, we further innovated analysis workflows of low- and high-resolution data (Fig. 1b-d), including custom segmentation, application of the Hopkins statistics to immune cell distribution, and automated curvature and cell shape analysis over long-time periods.

In the future, we aim to enhance the versatility of our platform by employing event-driven strategies to guide high-resolution data acquisition, rather than imaging predefined regions of interest at the onset of an experiment. The availability of low-resolution data will thereby enable dynamic identification and imaging of specific biological processes at high resolution across the entire organism. Moreover, we will expand the capabilities of the microscope to go beyond imaging towards event-driven, multi-scale sample manipulation.

In conclusion, it has remained a fundamental challenge to observe whole organisms and simultaneously provide mechanistic insight at a subcellular resolution over many hours. Here, we introduce a self-driving multi-scale imaging platform that provides a significant advancement in integrated bioimaging, enabling long-term observation of biological dynamics from cellular to systemic levels. The applicability of this novel platform to identify and study various processes in development and disease is imminent.

Acknowledgements and/or References (Optional):: We thank Gaudenz Danuser, Dagan Segal, Etai Sapoznik, Bingying Chen, Jill M. Westcott, Rolf A. Brekken, Subhrajit Banerjee, William A. Prinz, Maria Del Carmen Lafita-Navarro, Maralice Conacci-Sorrell, Christopher Kuo, Shanti Raminani, James Amatruda, Andrew York, Bo-Jui Chang, Justine Keth, Dana Kim Reed, Kushal Bhatt for discussions, support and samples; the Animal Resource Center for taking care of the zebrafish facility; and the whole Fiolka, Dean, Amatruda and Danuser laboratories for providing feedback and comments. Moreover, this research was supported in part by the computational resources provided by the BioHPC initiative at UT Southwestern Medical Center.

References

1. Valastyan, S. & Weinberg, R. A. Tumor metastasis: molecular insights and evolving paradigms. Cell 147, 275–292 (2011).
2. Daetwyler, S. & Fiolka, R. P. Light-sheets and smart microscopy, an exciting future is dawning. Communications Biology 6, 502 (2023).
3. Daetwyler, S. & Huisken, J. Fast Fluorescence Microscopy with Light Sheets. Biol Bull 231, 14–25 (2016).
4. Daetwyler, S. et al. Imaging of cellular dynamics from a whole organism to subcellular scale with self-driving, multiscale microscopy. Nature Methods 22, 569–578 (2025).
5. Huisken, J., Swoger, J., Del Bene, F., Wittbrodt, J. & Stelzer, E. H. K. Optical Sectioning Deep Inside Live Embryos by Selective Plane Illumination Microscopy. Science 305, 1007–1009 (2004).
6. Dean, K. M., Roudot, P., Welf, E. S., Danuser, G. & Fiolka, R. Deconvolution-free Subcellular Imaging with Axially Swept Light Sheet Microscopy. Biophysical Journal 108, 2807–2815 (2015).
7. Osmani, N. & Goetz, J. G. Multiscale Imaging of Metastasis in Zebrafish. Trends Cancer 5, 766–778 (2019).
8. White, R., Rose, K. & Zon, L. Zebrafish cancer: the state of the art and the path forward. Nature Reviews Cancer 13, 624–636 (2013).