Live-Cell Imaging and Single-Cell Tracking to Monitor Clonal Dynamics and Fate Transitions in Human and Murine Intestinal Organoids

Summary

We describe confocal imaging and tracking pipelines applied to two reporter strategies: mosaic organoids for resolving individual cell behavior and fate reporter organoids for monitoring cell type transitions in human and murine intestinal organoids.

Abstract

The intestinal epithelium undergoes rapid self-renewal through stem cell division, cell differentiation, and migration. Understanding how individual cells commit to specific fates and how clonal dynamics emerge within this tissue requires methods that capture cellular behavior in real time and at single-cell resolution. Intestinal organoids recapitulate key features of epithelial self-organization and cell-type diversity, making them a powerful system for studying these processes in a controlled setting. Here we present a protocol for long-term confocal live-cell imaging (up to 72 h) and single-cell tracking in human and murine intestinal organoids, built around two complementary reporter strategies. First, we describe the generation of mosaic organoids by combining differentially labeled cell populations, enabling single-cell resolution of membrane-localized and cytoskeletal reporters that cannot otherwise be attributed to individual cells in a dense epithelium. Second, we use cell-type-specific fate reporters (MUC2 for goblet cells and DEFA5 for Paneth cells) to monitor secretory cell type transitions in real time. The protocol covers organoid culture, mosaic organoid formation, sample preparation with strategies to minimize phototoxicity, image acquisition over several days, and semi-automated single-cell tracking using OrganoidTracker, which reconstructs cell trajectories and lineages over time. While demonstrated in intestinal organoids, this framework is readily adaptable to other epithelial organoid systems, including gastric, pancreatic, and colonic models, broadening its utility for studies of epithelial biology, homeostasis, and disease.

Introduction

The intestinal epithelium is one of the most rapidly self-renewing tissues in the body, with most cells replaced within three to five days. This continuous turnover is sustained by Leucine-rich repeat-containing G-protein coupled receptor 5 (Lgr5)-positive stem cells residing at the base of intestinal crypts, which give rise to transit-amplifying progenitors that differentiate into the specialized cell types of the intestinal lining as they migrate toward the villus tip¹. The major differentiated lineages include absorptive enterocytes, mucus-secreting goblet cells, hormone-producing enteroendocrine cells, and antimicrobial Paneth cells². How individual cells commit to specific fates, how collective cell behavior emerges from individual cell dynamics, and how tissue-level coordination is achieved remain fundamental questions in intestinal biology. Answering these questions requires experimental approaches that capture cellular behavior at single-cell resolution, in real time, and over timescales relevant to differentiation and migration³.

Intestinal organoids, three-dimensional (3D) self-organizing structures grown from single stem cells or tissue fragments, recapitulate key aspects of epithelial architecture, self-organization, and cell-type diversity^4,5,6. Since their development, organoids have become an indispensable tool for studying intestinal biology, disease modeling, and drug responses. Both murine and human intestinal organoids can be maintained long-term in culture, genetically modified using Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-based approaches, and combined into mosaic structures by mixing dissociated single cells of different genotypes or reporter backgrounds prior to re-embedding in basement membrane extract^7,8,9. Mosaic organoids combine cells of distinct genetic or fluorescent backgrounds within a single structure, so that the populations experience identical niche signals and mechanical context and can be compared with built-in internal controls. This genetic tractability makes them particularly suitable for live-imaging studies that require precise labeling of specific cell populations^6,10,11,12. Reporters that label cells sporadically, such as cell-type-specific fate markers, produce isolated fluorescent cells that can be traced and analyzed individually through single-cell tracking methods. However, a persistent challenge arises when membrane-localized and cytoskeletal proteins, such as cell junctions or cortical Myosin-II, are labeled uniformly across the epithelium. Their signals cannot be assigned to individual cells because neighboring cells share their boundaries at the resolution of confocal microscopy. Mosaic organoids overcome this limitation: when a cell expressing a fluorescent reporter is surrounded by unlabeled neighbors, its membrane and cytoskeletal signals can be unambiguously attributed to that single cell, revealing how it remodels its cytoskeleton during migration, division, or extrusion^9,13. Adjusting the labeling ratio inverts this view, allowing the neighborhood response around a unlabeled cell to be captured instead. Histone H2B nuclear reporters such as H2B-mCherry or H2B-iRFP serve as tools to identify the mosaic pattern and locate informative cell constellations for analysis. Beyond single-cell resolution, mosaic organoids also enable direct fitness comparisons between genotypes: because both populations share the same niche, differences in cell lifetime or extrusion rate provide an internally controlled readout of competitive fitness^9,14,15.

Live-cell imaging of organoids, however, presents specific technical challenges. Their 3D geometry requires confocal or light-sheet microscopy for optical sectioning^16,17,18,19. Capturing slow processes such as differentiation and migration demands imaging durations of 24 h or more, with careful optimization of laser power, exposure time, and imaging intervals to limit phototoxicity^16,18. The embedding matrix and culture medium must support organoid health while remaining compatible with the microscope optics. Furthermore, extracting quantitative single-cell information from dense 3D time-lapse datasets requires dedicated computational tools to detect and link cells across hundreds of time points. Beyond confocal microscopy and light-sheet microscopy, multiphoton microscopy and fluorescence lifetime imaging microscopy (FLIM) offer additional capabilities such as reduced phototoxicity and label-free metabolic readouts^20,21, though confocal microscopy remains the most widely accessible and versatile platform for multi-color live imaging of fluorescently labeled organoids.

Several approaches have been developed for tracking cells in organoids, ranging from manual annotation to semi-automated and fully automated deep learning-based pipelines²². OrganoidTracker is a semi-automated software tool that uses neural networks to detect nuclei in 3D confocal stacks and links detections across time points to reconstruct cell trajectories and lineage trees^3,23. It has been validated for tracking hundreds of cells over multiple days in intestinal organoids and provides a flexible framework for quantifying cell division, migration, differentiation, and extrusion events. Its support for multi-channel imaging makes it directly compatible with both mosaic organoids carrying spectrally distinct nuclear reporters and organoids expressing cell-type-specific fate markers.

In this protocol, we describe a modular workflow that integrates confocal live-cell imaging and single-cell tracking with two complementary reporter strategies in human and murine intestinal organoids. The first strategy uses mosaic murine organoids generated by mixing two cell populations carrying distinct H2B nuclear reporters (H2B-mCherry and H2B-iRFP) at defined ratios (Figure 1A). One population additionally carries a co-expressed reporter of interest, such as Myosin-II-mNeonGreen, whose membrane-localized signal can be unambiguously attributed to individual cells when they are surrounded by neighbors carrying only the second nuclear label. This enables direct observation of how a single cell remodels its cytoskeleton during migration, division, or extrusion within an otherwise intact tissue. The second strategy uses cell-type-specific fate reporters in human intestinal organoids, Mucin 2 (MUC2) to label goblet cells and Defensin Alpha 5 (DEFA5) to label Paneth cells, enabling real-time monitoring of secretory cell type transitions (Figure 1B). Together, these two systems illustrate how a shared imaging and tracking pipeline can be adapted to diverse biological questions simply by swapping the reporter strategy (Figure 1C,D). The sections that follow provide all steps needed to implement this workflow, from organoid culture through mosaic generation, image acquisition, and quantitative single-cell analysis.

Protocol

All experiments involving animals were performed in compliance with institutional guidelines approved by the Central Committee for Animal Experimentation (CCD) of the Dutch government and the KNAW/Hubrecht Institute Animal Welfare Body (IvD). Human intestinal organoids were generated from patient tissue obtained at the University Medical Center Utrecht (UMCU) with informed patient consent. The study was approved by the UMCU ethical committee (Utrecht, the Netherlands) and conducted in accordance with the Declaration of Helsinki and Dutch law.

1. Organoid culture and maintenance

1. Prepare Epidermal growth factor, Noggin, and R-spondin (ENR) growth medium for murine intestinal organoids (Table 1).
2. Prepare Expansion Medium (Table 2), Patterning Medium (Table 3), and Maturation Medium (Table 4) for human intestinal organoids.
3. Thaw a vial of growth factor-reduced basement membrane extract (BME) on ice overnight before use. 
4. Always keep BME on ice to prevent premature gelation.
5. Culture murine intestinal organoids in ENR medium embedded in BME, passaging every 7 days at a 1:4 to 1:8 split ratio by mechanical dissociation⁴.
6. Culture human intestinal organoids in Expansion Medium embedded in BME, passaging every 7–10 days⁵.
   NOTE: For expansion or recovery of murine organoids following single-cell dissociation, culture in ENR medium supplemented with Wnt-conditioned medium (supplemented with 10 mM nicotinamide and 10 µM Y-27632 dihydrochloride) for the first 2–3 days to promote organoid outgrowth and viability.
7. To induce differentiation of human intestinal organoids, sequentially culture organoids in Patterning Medium for 14 days, followed by Maturation Medium for 7–10 days⁶.
   NOTE: For detailed protocols for organoid establishment and culture, refer to Sato et al.4 for murine intestinal organoids, Sato et al.⁵ and He et al⁶ for human intestinal organoids. 

2. Generation of transgenic intestinal organoids

1. Passage organoids to ensure actively proliferating cultures at the time of electroporation: three days before electroporation for murine organoids and five to seven days for human organoids. For each electroporation reaction, use organoids from 4–8 wells of a 24-well plate (approximately 200–400 µL of BME containing organoids, yielding approximately 500–1,000 organoids). Dissociate organoids to single cells using recombinant trypsin-like dissociation reagent for 5–10 min at 37 °C. 
2. Generate nuclear reporter organoid lines by co-electroporating a transposon-based expression construct (p2T-H2B-mCherry-IRES-BlastR or p2T-H2B-iRFP-IRES-PuroR; 5 µg each) together with a transposase vector (p2T-CAG-mT2TP; 5 µg) into the dissociated cells using an electroporation system following established protocols²⁴. 
   NOTE: The amounts of plasmid DNA specified in step 2.2 (5 µg per construct) are optimized for approximately 500–1,000 organoids per electroporation reaction. During electroporation, verify that the impedance measured by the electroporator falls within 25–55 Ω. Impedance below 25 Ω indicates excessive cell density or buffer conductivity issues; impedance above 55 Ω indicates insufficient cell number or poor cell viability. Adjust cell concentration or replace the electroporation buffer if impedance falls outside this range.
3. After electroporation, culture cells in Wnt-conditioned medium supplemented with 10 µM Y-27632 dihydrochloride for 3 days to support cell recovery and organoid reformation. 
4. Begin antibiotic selection for overexpression constructs (puromycin (2 µg/mL) for H2B-iRFP lines or blasticidin (10 µg/mL) for H2B-mCherry lines) one week after electroporation to allow sufficient recovery before selection pressure. 
5. Maintain continuous selection for two weeks. 
6. Pick fluorescent organoid clones under an epifluorescence-equipped stereo microscope using appropriate filter sets. 
7. Dissociate selected clones to single cells using recombinant trypsin-like dissociation reagent and expand in Wnt-conditioned medium for two weeks before transitioning to ENR medium.
8. Validate reporter lines by confirming uniform nuclear localization of the fluorescent signal and stable expression across passages. 
9. Verify that reporter introduction does not alter organoid growth rate, morphology, or differentiation capacity by comparing to unlabeled parental lines. 
10. To enable readouts of membrane-localized or cytoskeletal proteins, co-express a fluorescent reporter of interest (e.g., Myosin-II tagged with mNeonGreen) in the same organoid line as the H2B nuclear reporter. 
    NOTE: Established H2B reporter organoid lines can be expanded and cryopreserved. These lines serve as a versatile baseline for downstream mosaic experiments: additional reporters or genetic mutations can be introduced into the same H2B-labeled background, generating a library of lines compatible with the mosaic workflow described in Section 4.

3. Generation of endogenous fate reporter human intestinal organoids

1. Generate the MUC2 fluorescent knock-in reporter in human intestinal organoids using CRISPR-mediated homology-independent targeted integration (CRISPR-HOT) as described in Artegiani et al.⁷
2. Generate the DEFA5 fluorescent knock-in reporter in human intestinal organoids by homology-directed repair (HDR) as described in He et al.⁶
   NOTE: CRISPR-HOT (step 3.1) and HDR (step 3.2) produce endogenous knock-in reporters. 
3. Validate both reporter expression by confirming co-localization of fluorescent signal with antibody staining for MUC2 and DEFA5, respectively. 
4. Verify that reporter introduction does not alter organoid growth, morphology, or differentiation capacity. 
   NOTE: The CRISPR-HOT approach is broadly applicable beyond fate markers and can be used to generate endogenous fluorescent reporters for membrane-localized and cytoskeletal proteins such as Myosin-II, E-cadherin, or ZO-1 in both human and murine organoids. 

4. Generation of mosaic organoids

1. Three days before mosaic organoid generation, passage both organoid lines into Wnt-conditioned medium to promote stem cell expansion and maximize cell viability during dissociation.
2. On the day of mosaic generation, coat each well of a 48-well plate with 20 µL of 5% BME in DMEM to provide a thin adhesive layer for cell attachment. Incubate at 37 °C for 1 h.
3. Dissolve the BME by adding ice-cold base medium and collect organoids from each line separately by centrifugation at 300 × g for 5 min at 4 °C. Dissociate organoids to single cells by incubating in enzymatic cell-detachment solution for 5–7 min in a water bath at 37 °C. Repeatedly pipette the suspension using a P1000 pipette to aid mechanical dissociation. Monitor dissociation under a brightfield microscope and stop the reaction by adding excess ice-cold base medium when predominantly single cells or small organoid fragments of 2–10 cells are visible. 
4. Pellet cells by centrifugation at 300 × g for 5 min at 4 °C. OPTIONAL: Filter each cell suspension through a 40 µm cell strainer to remove residual clumps, cellular debris, and BME gel fragments. 
   NOTE: This ensures a clean cell suspension, improves mosaic homogeneity, and reduces the risk of large clonal patches arising from pre-formed cell clusters. To create larger boundaries between cell populations, filtering is not recommended.
5. Assess cell viability and count viable cells using trypan blue exclusion. Combine the two cell populations at the desired ratio (e.g., 1:1, 1:3, or 1:9) to a total of 300,000 cells in 500 µL of Wnt-conditioned medium supplemented with 10 µM Y-27632 dihydrochloride. Mix thoroughly by gentle pipetting.
   NOTE: The mixing ratio determines the type of analysis: a low reporter-to-unlabeled ratio (e.g., 1:9) isolates individual labeled cells among unlabeled neighbors, enabling unambiguous attribution of membrane and cytoskeletal signals to single cells. A high fraction of reporter cells (e.g., 9:1) inverts this view, allowing the neighborhood response around a unlabeled cell to be captured.
6. Seed the cell suspension in the coated 48-well plate. Centrifuge the plate at 200 × g for 5 min at room temperature and incubate at 37 °C.
7. After 24 h, mechanically detach the cell layer by scraping with a P200 pipette tip and flushing thoroughly with ice-cold base medium or using a cell scraper. Collect the detached aggregates by centrifugation at 300 × g for 3 min at 4 °C without additional mechanical shearing. Resuspend aggregates in BME and plate into imaging-compatible sample carriers. Culture in ENR medium.
   NOTE: The 2D aggregation step without BME is critical for generating well-mixed mosaic organoids with a salt-and-pepper distribution of the two cell populations. 
8. For human mosaic organoids, replace Wnt-conditioned medium and ENR with the human Expansion Medium (Table 2); all other steps remain unchanged. 

5. Sample preparation for confocal live-cell imaging

1. Use glass-bottom imaging chambers (e.g., 24-well glass-bottom plate with high-performance #1.5 borosilicate coverslip) to ensure optical compatibility with high numerical aperture (NA) confocal objectives.
2. Plate the organoid-BME suspension as a thin dome (≤30 µL per well) near the center of the glass bottom. Before transferring to 37 °C, incubate the plate at 4 °C for 5–10 min to allow organoids to sink toward the glass bottom by gravity prior to BME gelation. 
   NOTE: This reduces the working distance between the organoid and the coverslip, maximizing image quality with inverted microscopes. 
3. Transfer to 37 °C and allow the BME to solidify for 20–30 min. Then add 500 µL of pre-warmed growth medium per 24-well. Allow organoids to recover and establish their morphology for at least 24 h after plating before beginning live imaging.
4. To minimize evaporation during long-term imaging, use a humidified stage-top incubation chamber. Place the sample on the microscope stage 1–2 h before imaging to allow thermal equilibration and minimize Z-drift during acquisition.
5. Ensure the stage-top incubator maintains 37 °C and 5% CO₂ throughout imaging.
6. Select organoids located close to the glass bottom (≤50 µm from the coverslip) with healthy morphology: clear crypt-budding architecture for murine organoids or cystic/budding structure for human organoids. Avoid organoids positioned deep in the BME, as signal quality and tracking accuracy degrade significantly with imaging depth.
   NOTE: The presence of phenol red in both the BME and the culture medium does not cause significant autofluorescence or imaging artifacts, as the fluorescent reporters used in this protocol (H2B-mCherry, H2B-iRFP, MUC2-mNeonGreen, and DEFA5-dsRed) produce sufficient signal-to-noise ratio for reliable nuclear detection and single-cell tracking. However, for experiments using weaker reporters or additional fluorescent channels, switching to phenol red-free BME and culture medium may be beneficial.

6. Confocal image acquisition

1. Use a laser-scanning confocal microscope equipped with a stage-top incubator (37 °C, 5% CO₂), motorized Z-drive, and laser lines appropriate for the fluorophores in use.
2. Select the objective based on the experimental requirement. Use the 40×/1.10 NA water immersion objective with a working distance of 0.62 mm to image deeper into the BME. Use the 20×/0.75 NA air objective with a working distance of 0.62 mm for multi-organoid or large field-of-view acquisition. 
   NOTE: Higher NA improves light collection, resolution, and optical sectioning, which is why the 40×/1.10 water immersion objective resolves fine detail deeper within the BME, while its higher magnification restricts the field to one or a few organoids. The 20×/0.75 air objective collects less light and sections less tightly, but its wider field captures many organoids in a single acquisition, making it the better choice for population-scale or multi-position imaging.
3. Allow the sample and immersion medium to thermally equilibrate on the microscope stage for at least 1–2 h before starting acquisition (step 5.8). 
4. Verify focus at all positions during the first few time points and manually correct any positions showing drift before it accumulates.
   NOTE: Loss of focus and image drift are common challenges during long-term confocal acquisitions and should be actively managed. For water immersion objectives, use an immersion oil with a refractive index matched to water (n = 1.333) instead of water itself, eliminating evaporation while maintaining the optical properties of water immersion.
5. Configure excitation and emission settings for each fluorophore. To minimize cross-bleed, acquire fluorescence channels sequentially rather than simultaneously, ensuring that only one laser line is active at a time. Use bi-directional sequential scanning to reduce total acquisition time per time point while maintaining channel separation. 
   NOTE: Include single-color control organoids expressing only one reporter to quantify residual cross-bleed and apply spectral unmixing corrections if necessary.
6. For two-color murine mosaic organoids, use 561 nm excitation for H2B-mCherry and 638 nm for H2B-iRFP670 with appropriate emission filters to minimize spectral bleed through. For the human intestinal organoids, use 488 nm excitation for MUC2-mNeonGreen and 561 nm for DEFA5-dsRed. 
   NOTE: Live nuclear dyes such as Hoechst 33342 or a  far-red fluorescent DNA dye can substitute for a genetic nuclear marker, as the detection pipeline operates on nuclear signal rather than a specific fluorophore. Note that DNA-binding dyes can become phototoxic and antiproliferative under repeated illumination, particularly Hoechst, which has a ~400 nm excitation, so keep concentrations low and favor low-phototoxicity far-red dyes for extended live imaging.
7. Set Z-stack as 25–35 optical sections at 2 µm intervals. Set time interval as 5–10 min for murine intestinal organoids and 15–30 min for human organoids.
   NOTE: For reliable nuclear detection using OrganoidTracker, we recommend the following settings as a starting point: 40x water objective, scan speed 600 Hz, zoom factor 0.75×, resulting in a pixel size of 0.38 µm in XY. These settings provide sufficient spatial sampling to resolve individual nuclei while keeping acquisition time per z-stack within acceptable limits for multi-position long-term imaging. 
8. Keep the laser power as low as possible, but increase till nuclei are clearly resolved. Start the multi-position time-lapse acquisition. Ensure the total acquisition time per time point across all positions does not exceed 90% of the set time interval to accommodate stage movement and autofocus adjustments. Image for 12–72 h, depending on the biological questions.
   NOTE: Before starting a long OrganoidTracker run, acquire a single z-stack and test the identification. Reliable identification with few missed or false detections confirms that the settings are adequate. Otherwise, adjust the laser power, detector gain, or z-step before proceeding. Capturing differentiation events such as the emergence of MUC2-positive goblet cells or DEFA5-positive Paneth cells typically requires 48–72 h. Short-term dynamics such as cell extrusion can be captured within 12–24 h. Phototoxicity is the primary limiting factor for long-term imaging. Signs include slowed cell division, apoptosis, nuclear condensation, and organoid collapse (Figure 2). Including unimaged control organoids as an internal reference is recommended. If observed, reduce laser power or increase the time interval.

7. Single-Cell tracking and lineage reconstruction using a semi-automated organoid cell-tracking software tool such as OrganoidTracker

1. Install OrganoidTracker following the instructions at https://github.com/jvzonlab/OrganoidTracker.
   NOTE: The software requires Python 3.8 or later with numpy, scipy, matplotlib, and tensorflow.
2. Load the time-lapse dataset into OrganoidTracker. Accepted file formats include TIFF stacks, ND2, and LIF. Run the neural network-based nucleus detection on the nuclear channel (H2B-mCherry or H2B-iRFP for murine organoids). 
   NOTE: The pre-trained model provided with the software is optimized for H2B-labeled intestinal organoid nuclei and cannot be directly applied to automatically identify MUC2-mNeonGreen- or DEFA5-dsRed-positive cells. 
3. Review and manually correct automated detections, focusing on false positives (detections on debris, autofluorescent material, or double-counted mitotic figures), false negatives (missed nuclei in dense regions or at the organoid periphery), and identity switches (incorrect cell identity assignments when two cells pass in proximity).
4. Run the linking algorithm to connect detections across time points and reconstruct cell trajectories. Review links manually, paying particular attention to cell division events (one track splitting into two daughter tracks) and cell extrusion events (track termination as a cell leaves the epithelium).
   NOTE: For mosaic organoids, run detection and tracking on the primary nuclear channel, then classify each tracked cell as belonging to population A or B based on the fluorescence intensity ratio between the two nuclear channels. 
5. Use “Manual Tracking” option in the software to track the human fate reporter organoids. Quantify reporter fluorescence by measuring the mean signal intensity within a defined ROI around the cell center position inserted during manual tracking. The time point at which reporter intensity crosses a defined threshold marks the commitment to fate. The onset of MUC2-mNeonGreen signal defines goblet cell commitment, while the onset of DEFA5-dsRed signal defines Paneth cell commitment.
6. Save the tracking project regularly using the software’s incremental saving function, which allows the correction process to be paused and resumed. Export tracking data (including cell positions, lineage trees, and cell classifications) for downstream analysis in Python or R.

8. Quantification of Cell Fate Transitions and Single-Cell Behavior

1. To quantify cell type transitions, identify the time point at which each DEFA5-dsRed negative cell first becomes DEFA5-dsRed positive (Paneth cell commitment) based on reporter channel intensity in the tracking. For each tracked cell, extract the fluorescence intensity of both the MUC2-mNeonGreen and DEFA5-dsRed channels over time to monitor the temporal relationship between the two reporters. Plot dual-channel intensity traces for individual cells to identify correlated reporter dynamics.
   NOTE: Monitoring both channels simultaneously in individual tracked cells is essential for distinguishing goblet cell commitment (sustained MUC2-mNeonGreen activation) from Paneth cell commitment (transient MUC2-mNeonGreen activation followed by DEFA5-dsRed onset). Single-channel analysis alone is insufficient to capture this dynamic and may lead to misclassification of early Paneth cell precursors as goblet cells.
2. In mosaic organoids, extract per-cell behavioral metrics for each tracked cell, including cell lifetime (time from last division to extrusion or end of imaging), division rate, and migration speed. Where a co-expressed reporter of interest is present, quantify its fluorescence intensity over time to capture dynamic changes during specific events such as division, migration, or extrusion. Relate these measurements to the cell’s local neighborhood context by comparing the behavior of the labeled cell to its immediate unlabeled neighbors.
   NOTE: When two genotypes are mixed in the same organoid, systematic differences in any of these metrics between the two populations can indicate a competitive fitness advantage of one genotype over the other. 
3. Perform statistical analysis using appropriate tests. For comparing the timing of fate commitment or per-cell metrics such as lifetime, migration speed, and division rate between two groups, use the Mann-Whitney U test. For comparing reporter intensity dynamics over time, use a linear mixed-effects model to account for repeated measurements within the same cell. 
   NOTE: All quantifications must be performed on manually corrected tracking data. Uncorrected or partially corrected datasets introduce systematic errors in the timing of fate commitment, spatial assignments, and behavioral metric calculations. Budget sufficient time for manual correction before proceeding to quantification (see Section 7).

Results

Following the protocol described in Section 4, mosaic organoids containing H2B-mCherry- and H2B-iRFP-labeled cell populations form within 48 h of seeding. Successful mosaic formation is characterized by either a salt-and-pepper distribution or defined boundaries between the two populations, depending on the degree of dissociation used. A salt-and-pepper distribution refers to a random, single-cell intermixing of the two populations, in which cells of each population are interspersed at the single-cell level, with no large clonal domains. Defined boundaries, by contrast, result in coherent patches of each cell population within the same organoid. The degree of cell dispersal is controlled by the duration of the dissociation step and the vigor of pipetting: thorough single-cell dissociation produces salt-and-pepper mosaics, while gentle fragmentation preserves small clonal clusters (Figure 3A).

The mosaic labeling enables single-cell analysis of processes that cannot be resolved in uniformly labeled tissues. When an individual cell co-expressing a reporter of interest (e.g., a fluorescent Myosin-II reporter) is surrounded by unlabeled neighbors, its cytoskeletal dynamics during migration, division, or extrusion can be unambiguously attributed to that cell (Figure 3B). Nuclear H2B reporters facilitate the identification of informative cell constellations within the mosaic, and per-cell measurements, such as lifetime, division rate, and reporter intensity dynamics, can be extracted from the tracking data (Figure 3C). When two genotypes are mixed, these per-cell metrics can be compared across populations to assess competitive fitness differences.

In human intestinal organoids carrying the MUC2-mNeonGreen and DEFA5-dsRed fate reporters, secretory cell fate commitment is evident as a gradual increase in reporter fluorescence over days. Organoids cultured in Maturation Medium were transferred to glass-bottom plates on day 7 and imaged by confocal live-cell microscopy (Figure 4A). Time-lapse imaging reveals two distinct reporter dynamics: goblet cells activate and maintain high MUC2-mNeonGreen expression throughout the imaging period, while Paneth cells display initial MUC2-mNeonGreen activation followed by progressive signal decrease and subsequent DEFA5-dsRed activation⁶ (Figure 4A). Single-cell quantification confirms this sequential transition at the level of individual cells (Figure 4B), suggesting that Paneth cell commitment proceeds through a transient MUC2-expressing intermediate state rather than directly from an unlabeled progenitor. Averaged intensity curves across multiple tracked cells confirm that this pattern is consistent at the population level, with Paneth cell precursors displaying a characteristic inverse relationship between MUC2-mNeonGreen and DEFA5-dsRed signals (Figure 4C). Together, these data demonstrate that the dual reporter system resolves distinct secretory fate trajectories at single-cell resolution.

Figure 1: The workflow that integrates confocal live-cell imaging and single-cell tracking with two complementary reporter strategies in human and murine intestinal organoids. (A) Schematic overview of the mosaic organoid generation workflow. Murine intestinal organoids expressing H2B-mCherry (magenta) or H2B-iRFP (green) are dissociated into single cells, mixed at defined ratios, and seeded onto a BME-coated plate for 24 h aggregate formation before re-embedding in BME. (B) Schematic overview of human intestinal organoids differentiation. Human organoids expressing MUC2-mNeonGreen (goblet cells, green) and DEFA5-dsRed (Paneth cells, magenta) endogenous knock-in reporters are differentiated using a two-step protocol. Organoids are first cultured in Patterning Medium for 14 days, followed by Maturation Medium for 7–10 days. During differentiation, organoids transition from a cystic, spherical morphology to an extensively budded structure that harbors major intestinal cell types, including MUC2-mNeonGreen-positive goblet cells and DEFA5-dsRed-positive Paneth cells. (C) 3D time-lapse confocal imaging of a murine mosaic intestinal organoid expressing nuclear reporters. (D) Single-cell tracking and lineage reconstruction using a semi-automated organoid cell-tracking software tool. Detected nuclei are linked across time points to reconstruct cell trajectories and lineage trees, enabling quantification of cell division, migration, and extrusion events. Please click here to view a larger version of this figure.

Figure 2: Comparison of healthy and phototoxic human intestinal organoids during live confocal imaging. (A) Representative brightfield and fluorescence images of a healthy human intestinal organoid. Healthy organoids display clear cystic or budding morphology with no cell debris in the surrounding medium. (B) Representative brightfield and fluorescence images of a human intestinal organoid showing phototoxic damage. Signs of phototoxicity include accumulation of cell debris in the surrounding medium, nuclear condensation, and arrested cell division. H2B-mCherry fluorescence images are shown as maximum intensity projections. Scale bars, 100 µm. Please click here to view a larger version of this figure.

Figure 3: Mosaic organoid generation and single-cell resolution of cytoskeletal dynamics in murine intestinal organoids. (A) Schematic of mosaic organoid generation by dissociation, mixing at defined ratios, and re-aggregation. Representative confocal images of mosaic organoids generated at cell ratios of 1:9, 1:5, and 1:2 (H2B-mCherry, green; H2B-iRFP, magenta) using single-cell dissociation (top row) or small-fragment dissociation (bottom row). Single-cell dissociation produces a salt-and-pepper mixing pattern, while small-fragment dissociation results in larger clonal patches. Scale bars, 50 µm. (B) Confocal images of mosaic organoids in which a dense (left) or sparse (right) population expresses Myosin-II-mNeonGreen (green), while the other population expresses only H2B-iRFP (magenta). Dashed boxes indicate the regions magnified in (C). Scale bars, 20 µm. (C) Time-lapse series showing cell extrusion at 8 min intervals, aligned to the moment of cell removal (t = 0 min). Top row (arrowheads): extrusion of a unlabeled cell surrounded by Myosin-II-mNeonGreen-expressing neighbors, revealing the neighborhood response. Bottom row (asterisks): extrusion of an individual Myosin-II-mNeonGreen-expressing cell surrounded by unlabeled neighbors, enabling unambiguous attribution of cortical myosin dynamics to the extruding cell. Scale bars, 10 µm. Right: normalized Myosin-II-mNeonGreen fluorescence intensity over time for extruding cells, aligned to the moment of cell removal (t = 0 min). Myosin-II accumulates progressively in the minutes preceding extrusion, peaking at the time of removal. Shaded area indicates standard deviation. Please click here to view a larger version of this figure.

Figure 4: Real-time monitoring of secretory cell fate transitions in human intestinal organoids using MUC2-mNeonGreen and DEFA5-dsRed fate reporters. Please click here to view a larger version of this figure.

(A) Representative time-lapse confocal images of a human intestinal organoid cultured in Maturation Medium and imaged from day 7 of maturation. MUC2-mNeonGreen (green) marks goblet cells and DEFA5-dsRed (magenta) marks Paneth cells. Scale bar, 20 µm. (B) Single-cell quantification of MUC2-mNeonGreen (green) and DEFA5-dsRed (magenta) fluorescence intensity over time for a representative Paneth cell, showing initial MUC2-mNeonGreen activation followed by progressive signal decrease and subsequent DEFA5-dsRed onset. Fluorescence intensity is shown in arbitrary units (a.u.). (C) Averaged fluorescence intensity curves for goblet cells (n = 7 cells) and Paneth cell precursors (n = 8 cells) showing MUC2-mNeonGreen (green) and DEFA5-dsRed (magenta) signal dynamics over time⁶. For Paneth cell precursors, time is aligned to the first appearance of DEFA5-dsRed signal (t = 0 h). Goblet cells maintain consistently high MUC2-mNeonGreen intensity over time, while Paneth cell precursors display a characteristic inverse relationship between MUC2-mNeonGreen and DEFA5-dsRed signals. Shaded areas indicate standard deviation.

Component	Stock concentration	Final concentration
Advanced Dulbecco's Modified Eagle Medium/F12 (Advanced DMEM/F12)		Base medium
HEPES	1M	10mM
Glutamax	200mM	2mM
N2 supplement	100x	1x
B27 supplement	50x	1x
N-acetylcysteine	500mM	1mM
Penicillin/Streptomycin	10000U/ml, 10000ug/ml	100U/ml, 100ug/ml
Recombinant murine EGF		50ng/ml
Recombinant murine Noggin		100ng/ml
Recombinant human R-spondin 1		500ng/ml

Table 1: Growth medium for murine intestinal organoids.

Component	Stock concentration	Final concentration
Advanced Dulbecco's Modified Eagle Medium/F12 (Advanced DMEM/F12)		Base medium
HEPES	1M	10mM
Glutamax	200mM	2mM
B27 supplement	50x	1x
Penicillin/Streptomycin	10000U/ml, 10000ug/ml	100U/ml, 100ug/ml
Primocin	50mg/ml	100ug/ml
R-spondin conditioned medium (U-Protein Express)		2%
Noggin conditioned medium (U-Protein Express)		2%
N-acetylcysteine	500mM	1mM
Nicotinamide	1M	10mM
Human recombinant EGF	500ug/ml	50ng/ml
A83-01	5mM	500nM
SB202190 (p38 inhibitor)	30mM	3uM
Prostaglandin E2	10mM	1uM
Wnt surrogate		1000x

Table 2: Expansion medium for human intestinal organoids.

Component	Stock concentration	Final concentration
Advanced Dulbecco's Modified Eagle Medium/F12 (Advanced DMEM/F12)		Base medium
HEPES	1M	10mM
Glutamax	200mM	2mM
B27 supplement	50x	1x
Penicillin/Streptomycin	10000U/ml, 10000ug/ml	100U/ml, 100ug/ml
Primocin	50mg/ml	100ug/ml
R-spondin conditioned medium (U-Protein Express)		2%
Noggin conditioned medium (U-Protein Express)		2%
N-acetylcysteine	500mM	1mM
Human recombinant EGF	500ug/ml	50ng/ml
A83-01	5mM	500nM
Wnt surrogate		1000x
CHIR99021	3mM	3uM
Recombinant Human IL-22	10 ug/ml	2ng/ml

Table 3: Patterning medium for human intestinal organoids.

Component	Stock concentration	Final concentration
Advanced Dulbecco's Modified Eagle Medium/F12 (Advanced DMEM/F12)		Base medium
HEPES	1M	10mM
Glutamax	200mM	2mM
B27 supplement	50x	1x
Penicillin/Streptomycin	10000U/ml, 10000U/ml	100U/ml, 100U/ml
Primocin	50mg/ml	100ug/ml
R-spondin conditioned medium (U-Protein Express)		2%
Noggin conditioned medium (U-Protein Express)		2%
N-acetylcysteine	500mM	1mM
Human recombinant EGF	500ug/ml	50ng/ml
A83-01	5mM	500nM
Wnt surrogate		30000x
Recombinant Human IL-22	10 ug/ml	2ng/ml

Table 4: Maturation medium for human intestinal organoids.

Discussion

This protocol provides a step-by-step workflow for live-cell confocal imaging and single-cell tracking in human and murine intestinal organoids, built around two complementary experimental systems: mosaic organoids that resolve individual cell behavior within a dense epithelium, and fate reporter organoids expressing MUC2-mNeonGreen and DEFA5-dsRed for monitoring secretory cell type transitions in real time.

Compared to fixed-timepoint methods such as immunofluorescence or single-cell RNA sequencing, live imaging captures the dynamics of transitions rather than snapshots. Compared to Cre-based lineage tracing, it provides the temporal resolution to identify individual fate decisions. The mosaic approach adds internally controlled comparisons within the same tissue context, avoiding confounding organoid-to-organoid variability.

Several steps in the protocol are critical for success. For a mosaic with salt-and-pepper distribution dissociation of organoids to single cells must be thorough but gentle: incomplete dissociation produces large clonal patches, while excessive dissociation reduces viability (Section 4.3). The 24 h aggregation step on a flat surface before re-embedding in BME is necessary for proper mixing of the two populations; omitting it results in poorly integrated mosaics (Section 4.6–4.8). Phototoxicity management during imaging requires careful calibration, starting with conservative laser settings and increasing only if the tracking software produces excessive false negatives (Section 6.8).

The main limitations relate to the inherent constraints of organoid systems and live imaging. Organoids lack stromal, vascular, and immune components that may influence cell behavior in vivo. For confocal microscopy, imaging depth is limited to approximately 100 µm from the coverslip, beyond which light scattering degrades signal quality and tracking accuracy. Multiphoton and light-sheet microscopy can substantially extend this depth range at the cost of additional instrumentation and protocol adaptation. The semi-automated tracking approach, while substantially more efficient than manual annotation, still requires considerable correction time, particularly in dense tissues (Section 7.4).

Both strategies are readily extensible. By varying the reporter co-expressed with the nuclear marker, different aspects of single-cell behavior can be examined, from cortical Myosin-II for mechanical dynamics to junctional E-cadherin for adhesion remodeling, calcium biosensors such as tq-Ca-FLITS and ColBD-Twitch, or metabolic biosensors such as SoNar for cellular NAD+/NADH redox state, enabling simultaneous readout of cell identity, lineage, and physiological state within the same organoid^9,25,26,27. Adjusting the mixing ratio tunes the analysis: sparse labeled cells enable single-cell attribution of membrane signals, while sparse unlabeled cells reveal neighborhood responses. The fate reporter strategy is extendable to any endogenous locus amenable to CRISPR-HOT or HDR-based tagging. Combining both approaches within the same organoid would allow simultaneous tracking of lineage commitment and single-cell dynamics.

More broadly, the imaging and analysis pipeline is applicable to organoids derived from other epithelial tissues, including gastric, colonic, hepatic, and pancreatic models. Induced pluripotent stem cell (iPSC)-derived organoids are a promising extension, as H2B-based nuclear labeling and mosaic generation by cell mixing are directly transferable, and the pipeline may enable patient-specific studies of cell fate and competitive dynamics. Synthetic embryo models such as gastruloids and stem cell-derived embryoids are also amenable to single-cell tracking, though their more complex geometry and higher nuclear density may require retraining of the a semi-automated organoid cell-tracking software tool neural network model and adaptation of imaging parameters. Brain organoids present additional challenges due to their large size and optical density, which limit confocal imaging depth. Furthermore, brain organoids require weeks to months of differentiation before meaningful imaging experiments can be performed, substantially increasing experimental timelines compared to intestinal organoids. Some brain organoid protocols also rely on suspension culture rather than BME embedding, which is incompatible with the sample preparation approach described here and would require adaptation to matrix-embedded or surface-attached formats before imaging. Light-sheet microscopy, which offers reduced phototoxicity and greater imaging depth, may be better suited for these systems¹³. Together, these extensions highlight the broad potential of the pipeline across diverse 3D model systems beyond the epithelial organoids described here.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
A83-01	Tocris	2939
Accutase Cell Detachment Solution	Innovative Cell Technologies	AT104-500
Advanced Dulbecco's Modified Eagle Medium/F12 (Advanced DMEM/F12)	Invitrogen	12634-010
B27 supplement	Thermo Fisher	17504044
Blasticidin	InvivoGen	ant-bl-05
Cell strainer (40um)	Greiner Bio-one	542140
CHIR99021	Stemgent	04-0004-base
Confocal microscope	Leica	SP8	with climate control for temperature and CO2
Cultrex Basement Membrane Extract (BME)	R&D Systems, Bio-Techne	3533-001-02
Fiji	open source		open source https://imagej.net/software/fiji/
Glass bottom plates	CellVis	P24-1.5H-N
Glutamax	Gibco	35050-061
HEPES	Gibco	15630-080
N2 supplement	Thermo Fisher	17502048
N-acetylcysteine	Sigma-Aldrich	A9165
NEPA electroporation system	NEPAGENE	NEPA21
Nicotinamide	Sigma-Aldrich	N0636
NucSpot Live			far-red live-cell nuclear dye
Noggin conditioned medium (U-Protein Express)	IPA Therapeutics	N002
OrganoidTracker	open source		open source https://organoidtracker.org/
p2T-CAG-mT2TP	homemade		protocol to generate the plasmid DOI: 10.1126/science.adr8753
p2T-H2B-iRFP-IRES-PuroR	homemade		protocol to generate the plasmid DOI: 10.1126/science.adr8753
p2T-H2B-mCherry-IRES-BlastR	homemade		protocol to generate the plasmid DOI: 10.1126/science.adr8753
Penicillin/Streptomycin	Gibco	11548876
Primocin	InvivoGen	ant-pm-05
Prostaglandin E2	Tocris	2296
puromycin	InvivoGen	ant-pr
Recombinant human EGF	Peprotech	AF-100-15
Recombinant Human IL-22	Peprotech	200-22
Recombinant human R-spondin 1	Thermo Fisher	120-38-5UG
Recombinant murine EGF	Thermo Fisher	315-09-500UG
Recombinant murine Noggin	Thermo Fisher	250-38-100UG
R-spondin conditioned medium (U-Protein Express)	IPA Therapeutics	R001
SiR-DNA			far-red live-cell DNA dye
SB202190 (p38 inhibitor)	Sigma-Aldrich	S7076
Suspension plate for organoid culture	Greiner	665102
Trypan blue	Gibco	15250061
TrypLE	Gibco	12605-010
Wnt surrogate	U-Protein Express	custom order
Wnt3A conditioned medium	homemade		The production of Wnt3A conditioned medium can be found here:  https://doi.org/10.1002/cpim.106
Y-27632 dihydrochloride	Abmole Bioscience	M1817