Long-term live imaging of epithelial organoids and corresponding multiscale analysis reveal high heterogeneity and identify core regulatory principles

Organoids are morphologically heterogeneous three-dimensional cell culture systems. To understand the cell organisation principles of their morphogenesis, we imaged hundreds of pancreas and liver organoids in parallel using light sheet and bright field microscopy for up to seven days. We quantified organoid behaviour at single-cell (microscale), individual-organoid (mesoscale), and entire-culture (macroscale) levels. At single-cell resolution, we monitored formation, monolayer polarisation and degeneration, and identified diverse behaviours, including lumen expansion and decline (size oscillation), migration, rotation and multi-organoid fusion. Detailed individual organoid quantifications lead to a mechanical 3D agent-based model. A derived scaling law and simulations support the hypotheses that size oscillations depend on organoid properties and cell division dynamics, which is confirmed by bright field macroscale analyses of entire cultures. Our multiscale analysis provides a systematic picture of the diversity of cell organisation in organoids by identifying and quantifying core regulatory principles of organoid morphogenesis. Graphical Abstract Created with BioRender.com.


25
Organoids are morphologically heterogeneous three-dimensional cell culture systems. To 26 understand the cell organisation principles of their morphogenesis, we imaged hundreds of 27 pancreas and liver organoids in parallel using light sheet and bright field microscopy for up to 28 seven days. We quantified organoid behaviour at single-cell (microscale), individual-organoid 29 (mesoscale), and entire-culture (macroscale) levels. At single-cell resolution, we monitored 30 formation, monolayer polarisation and degeneration, and identified diverse behaviours, 31 including lumen expansion and decline (size oscillation), migration, rotation and multi-organoid 32 fusion. Detailed individual organoid quantifications lead to a mechanical 3D agent-based 33 model. A derived scaling law and simulations support the hypotheses that size oscillations 34 depend on organoid properties and cell division dynamics, which is confirmed by bright field 35 macroscale analyses of entire cultures. Our multiscale analysis provides a systematic picture 36 of the diversity of cell organisation in organoids by identifying and quantifying core regulatory 37 principles of organoid morphogenesis. 38 Introduction 42 Understanding the principles of collective cell behaviour in mammalian organs during 43 development, homeostasis, regeneration and pathogenesis requires simplified models 44 mimicking the in vivo cell-cell and cell-matrix interactions. To this aim, organoids provide an 45 ideal in vitro model. Organoids are three-dimensional (3D) cultures obtained from pluripotent 46 stem cells (embryonic or induced pluripotent stem cells) or organ-derived adult stem cells (Clevers, 2016). Organoid systems recapitulating the brain and the majority of epithelial organs 48 have been established. These systems reproduce aspects of organ-specific development and 49 disease (Kretzschmar et al., 2016;Lancaster et al., 2019) and are valuable for personalised 50 (Broutier et al., 2017;de Winter-De Groot et al., 2018) and regenerative medicine (Takeda et 51 al., 2013). Multicellular self-organisation determines organoid behaviour and morphology. For 52 instance, epithelial organoids can acquire a spherical ("monocystic" (Box 1)), budding 53 ("branched"), but also a dense ("polycystic" (Box 1)) phenotype (Loomans et al., 2018;Serra 54 et al., 2019). Organoids are therefore a valid model to understand the principles of tissue self-55 organisation at the mesoscale, which are largely unknown (Trepat et al., 2018). 56 In order to fill this knowledge gap, a quantitative analysis at single-cell resolution is essential. 57 A multiscale approach is required, capturing the cell-to-cell variability while monitoring the 58 entire organoid system (Xavier da Silveira dos Santos et al., 2019) (Box 1). 59 The advancement of molecular biology allows quantifications of large-scale omics data at 60 single-cell resolution. For example, high-throughput single-cell transcriptomics detect rare cell 61 populations and trajectories between distinct cell lineages (Grün et al., 2015). Unlike most 62 single-cell molecular characterisations, time-resolved advanced microscopy enables both 63 spatio-temporal analysis of the organoids' global morphology as well as "zooming-in" on the 64 fates of a single cell. In previous studies, Bolhaquiero et al. (Bolhaqueiro et al., 2019) were 65 able to combine single-cell molecular and image-based analyses and proved chromosome 66 segregation errors with up to 18 hours long image acquisitions in a confocal microscope. In an 67 approach using an inverted light sheet microscope, Serra et al. (Serra et al., 2019) were able 68 to perform 5-day long live acquisitions of individual organoids. 69 Ultimately, the experimental quantitative data on organoid dynamics should serve as a 70 foundation for mathematical models, which predict the experimental outcome and test 71 hypotheses about underlying mechanisms of observed behaviours by altering controllable 72 parameters in silico (Sasai, 2013). 73 In our study, we focus on two types of organoids initiated from adult progenitor cells of the 74 pancreas and liver tumour as representatives for a spherical as well as a polycystic phenotype. 75 Murine pancreas-derived organoids (mPOs), are used as a model to study pancreas 76 development and the regeneration of pancreatic β-cells (Huch et al., 2013). Human 77 cholangiocarcinoma-derived organoids (hCCOs), are promising models to study personalised 78 connected via cell-cell junctions (Harris et al., 2010) and the cell layer ruptures if the internal 115 pressure reaches a critical point. 116 The single-cell resolution achieved by the light sheet pipeline is necessary for studying single-117 cell dynamics and collective cell dynamics in individual organoids in depth. However, the large 118 amounts of data acquired by this pipeline require considerable computational resources, which 119 hinder the extraction and quantification of macroscale (entire organoid culture) features. We 120 therefore developed the bright field pipeline that measures luminal size changes at individual-121 organoid resolution based on projected luminal areas (Box 1). This pipeline enables the 122 observation of entire organoid cultures (approx. 100-200 organoids within 25 µl ECM droplets, 123 depending on seeding density) over several days while retaining optimal physiological 124 conditions. In addition, the bright field setup allows label-free image acquisition, which ensures 125 minimal exposure to phototoxic effects. Quantification of the projected luminal areas over time 126 yields features on a mesoscale level, such as minimal and maximal area of individual 127 organoids, which are used to determine the median area increase of the entire culture at the 128 macroscale level. 129 Our light sheet data indicate that epithelial organoids show size oscillations (expansion and 130 decline phases) (Box 1), which are frequently observed in small organoids (diameter < 400 131 µm), but much less in large organoids (diameter > 400 µm). This is reflected in our 3D agent-132 based model, which indicates the size oscillations arise in response to an interplay of an 133 increase of the internal pressure, the cell division dynamics and the mechanical properties of 134 the single cells. The critical internal pressure due to release of osmotically active substances 135 into the lumen is reached earlier in organoids with increased surface-to-volume ratios (small 136 organoids) compared to organoids with reduced surface-to-volume ratios (large organoids). 137 We further verified these findings by quantifying the size oscillations in entire organoid cultures 138 using the bright field pipeline. 139 In summary, our approach reveals the dynamics of organoid cultures from single-cell and 140 single-organoid scale to the complete culture scale, ascertaining the core regulatory principles 141 (Box 1) of their multicellular behaviour. 142

144
Long-term live imaging with LSFM allows detailed visualisation of dynamic processes in 145 organoid morphogenesis and reveals high heterogeneity in single-cell and individual-146 organoid behaviour.

147
To gain deeper insights into the dynamic cellular processes occurring within organoid systems, 148 we developed Z1-FEP-cuvette holders for live imaging with the Zeiss Lightsheet Z.1 149 microscope system (Supplementary Figure 2). As previously described (Hötte et al., 2019), 150 ultra-thin FEP-foil cuvettes are sample holders for LSFM which preserve physiological culture 151 conditions for organoid cultures and allow the acquisition of high resolution images at single-152 cell level. Using the Z1-FEP-cuvette, we recorded the formation and development of hCCOs 153 expressing H2B-eGFP (nuclei marker) and LifeAct-mCherry (F-actin cytoskeleton marker) and 154 mPOs expressing Rosa26-nTnG (nuclei marker) for up to seven days. The medium was 155 exchanged every 48 hours to ensure sufficient nutrient supply. Temperature and CO2 levels 156 were controlled to ensure optimal growth conditions (Supplementary Figures 3, 4). The setup 157 enabled us to monitor dynamic processes at high temporal and spatial resolution in up to 120 158 organoids simultaneously contained in one Z1-FEP-cuvette (in this example in a total volume 159 Visual inspection of the acquired data revealed that the initially seeded organoid cell clusters 164 contract before the cells within the clusters start to rearrange and form spherical structures 165 (Figure 1, Formation). The cells within these spherical structures begin to polarise and form 166 a lumen (in this example around 13.5 hours), indicated by a stronger F-actin signal at the apical 167 (luminal) side of the cell membranes. Potentially dead cells accumulate within the lumen, 168 indicated by loss of the LifeAct-mCherry signal and by smaller nuclei with stronger H2B-eGFP 169 signals, hinting towards apoptotic nuclear condensation (Mandelkow et al., 2017). The 170 polarisation of cells in the epithelial monolayer is maintained during luminal expansion and is 171 still clearly visible at later stages of organoid development (in this example around 41.0 hours) 172 (Figure 1, Polarisation). The recording interval of 30 minutes, also allows us to visually track 173 single cell division events (here: over a time course of 2.5 hours) (Figure 1, Cell division). We 174 were also able to observe polarisation and cell division events in isolated single cells 175 (Supplementary Figure 5b), which remained dormant for relatively long periods during 176 observation. We identified an overall shrinking of the organoid, nuclear condensation and a 177 fading nuclei signal to be hallmarks of organoid degeneration (Box 1) (Figure 1, 178 Degeneration; Supplementary Movie 1). This process is initiated upon extended culturing 179 without further medium exchange (here: after about 100 hours). 180 Image-based segmentation and three-dimensional (3D) volume rendering of the acquired data 181 allow for even more detailed inspections of features observed in the highly dynamic organoid 182 system. While most processes can already be followed in maximum intensity z-projections of 183 the acquired image data, 3D volume rendering facilitates a more detailed understanding of the 184 underlying cellular dynamics from different perspectives. We identified organoid fusion to be a 185 frequent phenomenon in the investigated cultures (Figure 2 Single-cell tracking also revealed that the cells within an organoid move with different speeds 203 during organoid expansion. Cells at the organoid's "poles" tend to show less or slower 204 movement (2 µm per hour) compared to cells located at the organoid's equatorial plane (7 µm 205 per hour) (Supplementary Movie 7). Furthermore, we observed that prior to organoid 206 formation, some of the initially seeded cell clusters migrate through the ECM before they start inspections of dynamic morphological processes in organoid development. hCCOs and 215 mPOs were seeded into Z1-FEP-cuvettes for long-term live observations. They expressed the 216 nuclei marker H2B-eGFP (magenta) or Rosa26-nTnG (grey) and the F-actin cytoskeletal 217 marker LifeAct-mCherry (green). About 120 organoids were recorded in image stacks up to 218 900 z-planes deep for at most seven days. Long-term single-cell analysis of pancreas-derived organoids reveals cell-to-cell 239 heterogeneity in cell proliferation points. From the segmented data, we chose to analyse three representative organoids. One 248 small organoid (diameter < 400 µm), one large organoid (diameter > 400 µm) and one which 249 was size-comparable to the large organoid but showed a higher cell number. The three mPOs 250 expressed Rosa26-nTnG as a nuclei marker (Figure 3). lines) showed seven oscillations, whereas the two larger organoids showed three (Figure 4a, 276 b, red lines) and two (Figure 4a, b, green lines) events, respectively. We did not observe 277 any correlation between the size oscillation events and the changes in cell number or number 278 and distance of neighbouring cells. Further, we did not observe any size oscillation events in 279 the first 80 hours (within the 5% threshold) nor did we observe any synchronised oscillation 280 behaviour between the organoids within one culture. 281 282 Figure 3: Long-term single-cell analysis of mPOs reveals heterogeneity of proliferation 283 potentials. mPOs (seeded and maintained in one Z1-FEP-cuvettes) expressed the nuclei 284 marker Rosa26-nTnG (grey). Organoids were imaged for six days and analysed with our 285 previously published nuclei segmentation pipeline (Schmitz et al., 2017 To solve the mechanical principles underlying size oscillation events, we developed a 315 mathematical model based on the following assumptions. Since organoids are spherical 316 single-layer multicellular clusters, they are described by their volume ( ) and the number of 317 superficial cells ( ) at time point . We propose a functional relationship for an organoid's 318 increase in volume ̇( ), which is derived from two processes: a) The internal pressure of an 319 organoid increases with time, due to an influx following the segregation of an osmotic active 320 substance by the cells. b) Due to mitosis, the cell number ( ) grows and the surface area 321 ( ) increases (Figure 1, Cell division). 322 We hypothesise that the increase of the cell number ̇( ) can balance the increase in inner 323 pressure of an organoid and prevent size oscillation events. In the following we show that this 324 requires the cell count ( ) to grow faster than or equal to ( )~ 2 . In return, we expect the 325 occurrence of size oscillations in the case where the cell number increases slower than 326 ( )~ 2 . Our estimation is based on the following relations and simplifying assumptions: 327 328 i.
Organoids form spheres with a volume of 329 Every cell produces substances, which are secreted into the lumen of the organoid. 338 We assume that the production rate is constant in time and the same for all cells, 339 and therefore proportional to the number of superficial cells. In order to avoid a rupture, the growth of the surface ( ) has to balance the resulting osmotic 360 pressure Π arising from constant production of by . We can compute the functional form of 361 ( ) which leads to a constant osmotic pressure Π. We require 362 This relationship is fulfilled when ( )~2. This scaling law provides the following direct 366 implications: A constant cell division rate causes the cell count to increase exponentially. 367 Exponential growth is faster than quadratic, due to the theoretical considerations here we 368 expect no rupture and subsequent size oscillation events. Some of the organoids, however, 369 show a quasi-linear increase in cell numbers, which corresponds to a mitosis rate that is 370 decreasing with 1/ . A linear increase is slower than 2 , hence, in these organoids we expect 371 rupture and size oscillations. 372 In addition, we point out that the surface to volume ratio of a sphere changes with the radius, 373 since the volume grows much faster than the surface. Since the osmotically active substance 374 in the lumen is produced by the surface, this implies that smaller organoids reach a critical 375 internal pressure earlier than large organoids. forces (i.e. repulsion and adhesion), internal pressure of the organoids (due to an osmotic 383 imbalance), a bending potential of the cells in order to maintain the spherical shape of the 384 organoid, and cell division. Further, we assume the monolayer of the organoid to break if the 385 mean distance of neighbours exceeds a certain limit. 386 We hypothesise that the organoids can be represented as elastic spheres with a growing 387 surface due to cell division (Figure 5b). Further, we assume that the cells constantly secrete 388 a substance into the lumen which leads to osmotic influx. This leads to an increase in the 389 internal pressure which, however, can be balanced by an increase in volume. Based on these 390 assumptions, we derived that the organoid can balance the inner pressure when the cell count 391 increases at least quadratically (Supplementary theoretical considerations). Furthermore, 392 for small organoids the ratio between surface and volume is smaller than for large organoids. 393 Therefore, small organoids should reach a critical pressure for leakage faster than large 394 organoids. Thus, we expect the size oscillations to critically depend on a) the cell division 395 dynamics and b) the organoid size. The latter (b) is confirmed by the data obtained through 396 bright field analysis (Figure 6e). 397 The model is used to support the theoretical considerations and to qualitatively reproduce the 398 size oscillations of the three analysed mPOs (Figure 4a-b). Hereby, the cell division rate is 399 adjusted to match the experimental data. Simulations of two large organoids do not show a 400 size oscillation during phases of exponential cell number increase but start to oscillate after 401 transitioning to a linear growth (Figure 5c-d, green and red lines). Simulations with a small 402 organoid exhibit size oscillation even during the initial exponential growth, which confirms our 403 hypothesis that small organoids are more prone to rupture and deflation (Figure 5c-d The extracted features can further be used in downstream analyses to categorise organoid 438 behaviour. We found that the size of an organoid is crucial for the number of oscillation events 439 it displays. Consistent with the LSFM data and the mathematical model, the bright field analysis 440 shows that initially smaller organoids (area < 0.01 mm²) feature more size oscillation events, 441 while initially larger organoids display less oscillation events (area > 0.01 mm²) (Figure 6e). 442 Besides that, the average expansion factor (a measure for expansion speed consistency) with 443 a median average value of 0.11 is similar between organoids with various initial areas -50% 444 of all values range between 0.09 and 0.14, while only 12% of the evaluated organoids are 445 outliers with values above 0.22 (Figure 6f). Further, the linear correlation between the initial 446 area and the final area becomes apparent (R² = 0.7445), which shows that the growth is 447 independent of the initial area (Supplementary Figure 7a). This indicates strong similarities 448 in expansion speed consistency between individual organoids within one culture despite their 449 (high) size heterogeneity. 450 451 Besides the already mentioned features, other extracted features facilitate the definition of 452 quantitative reference parameters of organoid systems. By comparing the final area to the 453 maximum area, for example, continuous growth of mPOs during the analysed time window is 454 proven. A comparison of the initial area to the minimum area identifies size oscillation events 455 or overall descending size progression within organoid cultures. In mPOs, the minimum area 456 falls only slightly below the initial area, which can be associated with oscillation events 457 (Supplementary Figure 7c). 458 Besides the average expansion factor, analysis of the maximum expansion factor indicates 459 expansion speed variations within organoid cultures. As a variable factor, the maximum 460 expansion can be used to compare different culture conditions (Supplementary Figure 7b). 461 An additional feature, which is likely to change upon differentiation or other perturbations (e.g. 462 drug treatment), is the organoid circularity. In healthy mPOs, the circularity is 0.9 on average 463 and the deviation around the average narrows over time (Supplementary Figure 7e). In previous studies, chromosomal segregation errors in organoids were monitored using 505 confocal or spinning disc microscopy, capturing single organoids in a z-range of 60 µm (3-4 506 min intervals) (Bolhaqueiro et al., 2019). The observation of single mitotic events with LSFM 507 over multiple days, from the initial seeding to the plateau-phase of growth, can therefore 508 increase the throughput and allow the analysis of larger organoids to monitor single cell 509 behaviour. Serra et al. used an inverted LSFM to analyse the development of single organoids 510 originating from single cells. By parallelisation, they were able to image multiple organoids 511 (Serra et al., 2019). Our light sheet pipeline combines a parallelised acquisition of more than 512 100 organoids, within a volume of up to 8 mm³ given by the Z1-FEP-cuvette, with a high spatial 513 (1000 z-planes, 2 µm spacing), as well as a high temporal resolution and still allows long-term 514 observations. 515 Our analysis of mPOs and hCCOs reveals their highly heterogeneous and multi-faceted growth 516 patterns and common morphological dynamics independent of their carcinogenic or healthy 517 origin. This matches the observation of intrinsic abilities of single intestinal organoid cells to 518 form asymmetric structures (Serra et al., 2019), as well as former studies that have not 519 addressed heterogeneity directly, but already showed variable organoid sizes and irregularly 520 occurring rupture events (Mahe et al., 2013;Schlaermann et al., 2016;Schwank et al., 2013;521 Sebrell et al., 2018). 522 Adult tissue-derived organoids can develop from single cells or cell clusters, although starting 523 from single cells results in lower organoids formation efficiency, which hampers the systematic 524 analysis of cellular behaviours (Serra et al., 2019). Starting from cell clusters, the cell survival 525 and therefore the multiplication-rate of cell material is higher. The production of large amounts 526 of material for a potential clinical application is therefore ensured (Dossena et al., 2020). 527 However, starting from clusters, the heterogeneity of the cultures dynamics increases. Since 528 our pipelines capture both aspects, they support the understanding of clonal formation as well 529 as the determination of quality control parameters for clinical applications of organoids. We The multi-faceted dynamic behaviour of organoids is reflected in their motion. We were able 553 to show that organoids differ in their overall rotation speed and rotation direction, in their cell 554 motion depending on their position within the organoid and that not all organoids show 555 rotational behaviour. In general, rotation in organoids is poorly described directly, however, 556 dynamic processes are already investigated in other 3D cell culture systems (Ferrari et al., 557 2008;Hirata et al., 2018;Marmaras et al., 2010;Tanner et al., 2012). Sebrell et al. related the 558 rotation of gastric epithelial organoids to the passage number and patient/donor history 559 (Sebrell et al., 2018). Wang et al. investigated the rotation of 3D human mammary epithelial 560 acini and identified that cell polarity and microtubules are essential for rotation (Wang et al., 561 2013). It remains an exciting question, why not all investigated organoids show rotation. 562 Computational and mathematical in silico models are a valuable tool to understand the 563 underlying mechanics of 3D cell culture behaviour (Eils et al., 2013). They can be used to 564 predict organoid behaviour in conditions that are challenging to implement in experiments or 565 when perturbations of normal conditions occur (Dahl-Jensen et al., 2017;Eils et al., 2013). 566 However, only relatively few models for organoid systems have been developed (Montes-567 Olivas et al., 2019). 568 Here, we implemented a mechanical 3D agent-based model that relies on a limited set of 569 assumptions (namely intercellular forces, internal pressure of the organoid, bending energy of 570 the surface and cell division). We showed that the model is a valuable instrument for the 571 description of spatiotemporal dynamics of organoids. We were able to recreate the qualitative 572 growth curve of the three segmented organoids and showed that the frequent size oscillation 573 of organoids is not directly associated with mitosis (for further experimental analysis of the 574 mechanisms underlying organoid size changes see also Yang et al. (Yang et al., 2020). 575 Instead, the model indicates that the decline process relies on cells losing cell contacts due to 576 mechanical stress exerted by the internal luminal pressure (Ruiz-Herrero et al., 2017). Further, 577 the model confirms that the size oscillation dynamics are dependent on the organoid volume-578 to-surface ratio and its dynamics with exponential and linear growth phases. The disagreement 579 between the simulation and data concerning the volume of the medium and large organoid 580 results from the fact that the cell size differs between the two organoids. The large organoid 581 exhibits a higher cell density, which implies a smaller cell size compared to the medium size 582 organoid. Different cell sizes are not considered in the current version of the model but can be 583 included to reflect these different phenotypes. 584 Our light sheet pipeline shows that the small organoid has a higher size oscillation frequency 585 (Box 1) than the larger organoids. The theoretical considerations and the mathematical model 586 support this observation: size oscillations are affected by an increased surface-to-volume ratio. 587 The bright field pipeline further confirm this observation. In summary, the simulation of 588 epithelial organoid growth predicts organoid behaviour and helps to understand the intrinsic 589 mechanisms responsible for the organoid phenotype. Further, it is straightforward to generate 590 a cost-and time-effective tool to predict possible outcomes of external stimuli like drug 591 treatments for instance (Montes-Olivas et al., 2019). 592 The bright field pipeline enables the quantification of culture dynamics on meso-and 593 macroscale level, generating robust data on organoid growth behaviour and allowing the 594 quantification of heterogeneity in whole organoid cultures. The pipeline has been extensively 595 applied in the LSFM4LIFE (www.lsfm4life.eu) and the Onconoid Hub projects to measure and 596 optimise the growth of human pancreas-derived organoids in synthetic hydrogels and identify 597 novel drug candidates for the treatment of intrahepatic cholangiocarcinoma (manuscripts in 598 preparation). In perspective, the same analysis is suitable to determine parameters of organoid 599 growth for stem cell therapy (Aberle et al., 2018;Huch et al., 2017;Lancaster et al., 2019;600 Nagle et al., 2018) and to characterise patient-specific responses for optimising personalised 601 drug treatments or assaying the onset of resistance in cancer therapy (Broutier et al., 2017;602 Fan et al., 2019;Nagle et al., 2018;Ooft et al., 2019). 603 In conclusion, our multiscale analyses of diverse organoid cultures have a great potential for 604 further investigations of epithelial organoids and many other complex culture systems. 605

606
Organoid culture 607 hCCOs were initiated from primary liver tumour biopsies of cholangiocarcinoma patients (~0.5 608 cm 3 ) collected during surgery performed at the Erasmus MC -University Medical Center 609 Rotterdam (NL) and cultured as previously described (Broutier et al., 2017). mPOs were 610 obtained from Meritxell Huch (Gurdon Institute, Cambridge, UK) and cultured as described 611 (Huch et al., 2013). 612 Transgenic murine pancreas-derived organoids: Fabrication of positive moulds for vacuum forming. 646 We designed positive moulds of the cuvettes for the use in the Zeiss Lightsheet Z.1 system by 647 using the free CAD software "123D Design" (version 2.2.14, Autodesk). We 3D-printed the 648 positive moulds by using the service of the company Shapeways. Before use, the positive 649 moulds were inspected by stereomicroscopy and cleaned by immersion in an ultrasonic bath. 650 Cuvette fabrication with vacuum forming.    pancreas-derived organoids. The mathematical model was given as a set of stochastic 712 differential equations that were solved using the Euler-Maruyama method. 713 To describe the pancreas-derived organoid, we assumed it has a roughly spherical shape, with 714 cells forming a monolayer filled with fluid at a different pressure relative to the environment. 715 The volume of the organoid is affected by two mechanisms: a) the influx of liquid caused by 716 an osmotic imbalance or active pumping of the cells, and b) cell division. While a) is increasing 717 the internal pressure, b) leads to a relaxation of the surface. 718 Each cell was described by a small set of features, i.e. a position in 3D space and a cell size 719 denoted by its radius. Displacement of the cells was described as a response to three forces: 720 1) external forces exerted by surrounding cells, given as a spring potential, 2) internal pressure 721 of the organoid pushing the cells outwards, given by the ideal gas law, and 3) a surface bending 722 energy, keeping the organoid in its spherical shape. 723 Cell division was adjusted to match the experimental data, obtained by long-term single-cell 724 analysis of pancreas-derived organoids, but can easily be adapted to other growth dynamics. 725 If the average distance of neighbouring cells exceeds a certain limit, we assumed the 726 mechanical stress to be too high and a leakage in the shell of the organoid emerges. Through 727 the rupture, the internal liquid is released and the internal pressure decreases. Thus, the 728 mechanical forces, exerted on the cells might relax and the organoid deflates. When the 729 average distances between all neighbouring cells falls below a given threshold, the shell closes 730 and the liquid stops to be released. 731 York) droplets in suspension culture plates (48-well, Greiner Bio-One, Kremsmünster, Austria), 738 overlaid with 250 µl expansion medium and cultured for 12 h before imaging. They were then 739 imaged every 30 min in a 3x3 tile imaging (15% overlap) mode using the Zeiss Cell Observer 740 Z.1, fully equipped with an incubation chamber and motorised stage using a Plan-Apochromat 741 5x/0.16 objective, with a pixel size of 1.29 µm x 1.29 µm. In total, ten planes throughout the 742 droplet were imaged, with a z-distance of 50 µm (mPOs) and 65 µm (hCCOs), respectively, 743 capturing a z-range of 450 to 585 µm. 744 Image processing and organoid segmentation 745 Organoid growth rates were determined using a python custom-made pipeline for bright field-746 based image segmentation. The whole pipeline was equipped with a general user interface. 747 The recorded time-lapse image stacks were pre-processed with Fiji (ImageJ version 1.51n, 748 Java version 1.8.0_6 (64-bit)) by reducing the dimensionality of the raw data set from 9 (3x3) 749 tiles with 10 z-planes each to one stitched image with one z-plane per time frame using the The results were plotted and statistically evaluated (Kruskal-Wallis ANOVA, p < 0.05) using 761 OriginPro 2019 or Excel. For a normalisation, the projected areas were normalised to the 762 median of the fifth time point. 763 Mesoscopic feature extraction 764 Quantitative features were extracted using a Python script and were defined as follows: A size 765 oscillation event consists of a decline phase followed by an expansion phase. The start of a 766 decline phase was defined as the time point after which the area declines by 5%, and the end 767 is marked if the area increases again. Expansion phases were defined between the end of a 768 decline phase and the start of the following decline phase. As additional criterion, the duration 769 of expansion phases is greater than or equal to five time points, and the correlation coefficient 770 of the fitted polynomial is above 0.9. The number of decline and expansion phases per 771 organoid was determined including their duration and slope. Subsequently, maximum and 772 average expansion slopes were computed. The average expansion factor is specified as the 773 average slope of all detected expansion phases per organoid. The maximum expansion factor 774 is specified as the maximum slope of all detected expansion phases per organoid. Outliers in 775 average expansion were defined as smaller than the first quartile minus 1.5 x IQR or above 776 the third quartile plus 1.5 x IQR. The circularity was monitored continuously and is defined as 777 4π(area/perimeter²). Its standard deviation is displayed as the average standard deviation in 778 all analysed wells. Organoids displaying a circularity below 0.6 were considered as deficiently 779 segmented and were excluded from further analyses. Due to deficient segmentation during 780 organoid formation the projected area was normalised to the fifth time point of acquisition. 781 Box 1.