Protocol to evaluate a lineage marking system in the Drosophila testis

Summary

Accurate interpretation of genetic lineage tracing data is often confounded by the specificity and sensitivity of promoter activity and recombination rates. Thus, reported values may misrepresent targeted event frequency. Here, we present a protocol to validate a genetic tool for marking “dedifferentiation” in Drosophila testes. We describe steps for long-term live imaging to observe reporter activation and detail procedures to estimate false-positive and false-negative rates by integrating these datasets into a mathematical model. This approach enables rigorous evaluation of system performance.

For complete details on the use and execution of this protocol, please refer to Bener et al.¹

Graphical abstract

Highlights

• •Strategies for designing genetic tools to detect dedifferentiated germline stem cells
• •Instructions for extended live imaging to monitor genetic marking
• •Estimating lineage tracing specificity and efficiency via mathematical modeling

Publisher’s note: Undertaking any experimental protocol requires adherence to local institutional guidelines for laboratory safety and ethics.

Accurate interpretation of genetic lineage tracing data is often confounded by the specificity and sensitivity of promoter activity and recombination rates. Thus, reported values may misrepresent targeted event frequency. Here, we present a protocol to validate a genetic tool for marking “dedifferentiation” in Drosophila testes. We describe steps for long-term live imaging to observe reporter activation and detail procedures to estimate false-positive and false-negative rates by integrating these datasets into a mathematical model. This approach enables rigorous evaluation of system performance.

Before you begin

In many tissues, stem cells continuously generate new cells, a process essential for maintaining tissue integrity. Stem cells both self-renew and differentiate into one or more specialized cell types. Their maintenance is tightly regulated, and one established mechanism contributing to this regulation is dedifferentiation, in which committed progenitors or differentiated cells revert to a stem-like state. Increasing evidence supports the occurrence of dedifferentiation across diverse stem cell systems.^{2,3,4,5,6,7,8} In most cases, however, dedifferentiation is a rare event that is triggered only under specific physiological conditions or stress. Because of its rareness, dedifferentiation remains difficult to study. Dedifferentiated cells often lack clear morphological or molecular features that distinguish them from the original stem cell population.⁸ Moreover, although genetic lineage-tracing strategies are often used, their accuracy is difficult to assess.^4,8,9,10,11

Here, we describe a method of direct observation of genetic marking events and subsequent application of mathematical modeling. Rather than focusing solely on the experimental procedure itself, this protocol intends to guide how to select tools and design experiments to test the performance of the system by estimating false-positive and false-negative rates.

All works were done under the BSL-1 requirement under the Institutional Biosafety Committee (IBC) Registration number:17-018AB.

FLP/FRT lineage tracing

We utilize the FLP/FRT site-specific recombination system. In this system, the FLP recombinase is expressed under the control of a lineage-specific promoter and recognizes FRT sites located in the reporter cassette. We used a promoter of a differentiation factor, bag of marbles (bam). Reporter constructs are often designed to start expression of a marker gene (e.g., GFP) upon recombination of the FRT sites. For example, FLP recombinase removes FRT-flanked stop codons located upstream of a reporter gene (flip-out). This event permanently marks the cell and activates a downstream reporter gene (e.g., GFP) (Figures 1A and 1B). The marked cells increase over time in the niche (Figure 1C). The accuracy of this system depends on the combination of the promoter driving FLP and the FRT reporter cassette.

Figure 1.

Detecting dedifferentiation using the bam-FLPD5 lineage tracing system

(A) Schematic of asymmetric division and bam expression dynamics. GSCs (blue) attached to the hub (orange) divide to produce differentiating GBs, which subsequently differentiate into SGs. The bam transcription is repressed in the stem cells but upregulated in differentiating GBs and SGs. The bam-FLPD5 lineage tracing system permanently marks differentiating cells (green) by recombination.

(B) If a marked SG dedifferentiates and re-attaches to the hub, the resulting GFP positive GSC represents a dedifferentiation event.

(C) Time-course schematic showing the accumulation of GFP-positive GSCs in the niche over days as a result of dedifferentiation events.

Key components for detecting dedifferentiated germline stem cells

FLP construct

The promoter sensitivity and specificity determine the system’s accuracy. Ideally, the promoter should be effectively active in one lineage but silent in the other lineage. Insufficient activation results in false-negatives, whereas any off-target activity results in false-positives that confound data interpretation.

We use the promoter of the bam gene, which is silent in germline stem cells (GSCs) and upregulated in differentiating gonialblasts (GBs) and spermatogonia (SGs) (Figure 1A). Previous studies utilized a combination of bamGal4, UAS-FLP, which marked extremely low frequency of dedifferentiated cells compared to the estimated rate in live imaging.¹² This is likely (at least partly) caused by a delay in FLP expression due to the extra time required for Gal4 expression. Indeed, bamGal4 driven expression of UAS-GFP is detectable only after 8-cell SGs, even though expression of bam gene starts in 2-cell SGs and fully active in 4-cell SGs.¹ To minimize the delay of the FLP mediated recombination, we generated a construct driving the FLP recombinase directly under the bam promoter. In addition, we utilized the FLPD5 variant, FLP with aspartic acid at amino acid 5 to minimize the time required for FLPD5 expression to recombination. According to Nern et al. 2011,¹³ the FLPD5 version has approximately tenfold higher activity than the FLPG5 version, another broadly used construct.

Reporter cassette

The recombination threshold of the reporter cassette to FLP recombinase is another critical factor. An optimal reporter can be recombined specifically in the FLP expressing lineage, but not in the lineage where FLP is not supposed to be expressed. Even though constructs containing identical FRT sequences often show various sensitivity to FLP potentially due to different reasons,¹⁴ it is critical to test available reporters or generate several transgenic insertions and select appropriate cassettes.

To generate reporter transgenic animals, we suggest using well-characterized genomic transgenic landing sites, such as attP sites for phiC31 integrase mediated attB/P integration in flies,¹⁵ or the ROSA26 site in mice.¹⁶ This helps not only to maximize the response to FLP, but also to avoid complex responses to FLP expression levels due to position effect. Thankfully, we were able to obtain multiple FRT-flanked reporter transgenic flies available from Bloomington Drosophila Stock Center (BDSC). We selected to use two reporters with distinct recombination thresholds: Reporter-1 (nos-FRT-stop-FRT-Gal4, UAS-GFP) is sensitive to low levels of FLP, while Reporter-2 (nosGal4, UAS-FRT-stop-FRT-mCD8-GFP) requires higher FLP levels for recombination (Figures 2A and 2B). In the related publication, we examined the specificity and sensitivity of reporter marking using a long-term live imaging.¹ These allowed us to estimate false-positive and false-negative marking rates.¹

Figure 2.

Two reporter constructs perform differently for dedifferentiation marking

(A) An illustration of the Bam-FLPD5 lineage tracing system. FLPD5 expression is driven under the Bam promoter. The Bam-FLPD5 flies are crossed with the flies carrying nos-FRT-mCherry-stop-FRT-Gal4, UAS-GFP cassette (nos>>GFP, Reporter-1). The image below shows example of the mCherry positive or GFP positive (white dotted line) GSCs at day 14 after eclosion.

(B) An illustration of the Bam-FLPD5 lineage tracing system. The Bam-FLPD5 flies are crossed with the flies carrying UAS-FRT-stop-FRT-mCD8-GFP cassette and nosGal4 (nos>>mCD8GFP, Reporter-2). The image below shows example of the mCD8-GFP positive (white dotted line) GSCs at day 14 after eclosion. Germ cells are stained with Vasa. The hub (niche) is stained with FasIII. Scale bars are 10 μm. Asterisks indicate the hub.

(C) Predicted effect of dedifferentiation rate (k_dd) on GFP-positive GSCs in the Bam-FLPD5, nos>>GFP system. Model predictions for the percentage of GFP-positive GSCs over time are shown for different dedifferentiation rates (k_dd = 0, 0.2, and 0.4/day). The overlapping lines demonstrate that the system is insensitive to changes in k_dd due to high ratio of ⍺∗/⍺^True = 11.

(D) Predicted effect of dedifferentiation rate (k_dd) on GFP-positive GSCs in the Bam-FLPD5, nos>>mCD8GFP system. Model predictions for the percentage of GFP-positive GSCs over time are shown for different dedifferentiation rates (k_dd = 0, 0.2, and 0.4/day). The lines demonstrate the system’s sensitivity to changes in k_dd due to the lower ratio of α^∗/α^True = 0.0667.

Figures are adapted from the related manuscript.¹

For the experimental setup, flies carrying the bam-FLPD5 construct are crossed with flies carrying these reporter cassettes (Reporter-1 or Reporter-2) for assessing the change in GFP positive GSCs over time. Reporter activation was also monitored by live imaging alongside Vasa-mCherry. In the related publication,¹ we showed that these two reporters exhibit different levels of spontaneous flipping rates in unintended cells (such as in GSCs). As a consequence, we concluded that Reporter-1 does not reflect any dedifferentiation event (see details in Figure 2).

Innovation

This protocol describes a quantitative pipeline to measure the accuracy of genetic lineage tracing systems. First, the method combines long-term live imaging of intact Drosophila testes with direct observation of de novo recombination events. This enables precise determination of when and in which cell types recombination occurs, allowing spontaneous background activation to be empirically distinguished from lineage-specific marking. Such direct validation is rarely performed in lineage-tracing studies due to technical barriers in live tissue imaging.

Second, the protocol leverages differential reporter sensitivity by systematically comparing low- and high-threshold FLP/FRT reporter cassettes. By pairing a lineage-specific FLP driver with reporters that respond to distinct recombination thresholds, the system exposes how reporter design critically influences apparent biological outcomes, including sensitivity to rare events. Finally, the protocol integrates experimentally measured recombination rates into a mathematical model to estimate false-positive and false-negative probabilities and assess sensitivity to biological parameters such as dedifferentiation rate.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rat monoclonal anti-Vasa (1:20)	Developmental Studies Hybridoma Bank	DSHB Cat# anti-vasa; RRID: AB_760351
Mouse monoclonal anti-hu-li tai shao (1:20)	Developmental Studies Hybridoma Bank	DSHB Cat# 1b1; RRID: AB_528070
Mouse monoclonal anti-Fasciculin III (1:40)	Developmental Studies Hybridoma Bank	DSHB Cat# 7G10 anti-Fasciclin III; RRID: AB_528238
Cyanine 3 goat anti-mouse IgG + IgM (H + L) (1:400)	Jackson ImmunoResearch Labs	RRID: AB_2338684
Cyanine 5 goat anti-rat IgG (H + L) (1:400)	Jackson ImmunoResearch Labs	RRID: AB_2338263
Chemicals, peptides, and recombinant proteins
Formaldehyde, 16%	Electron Microscopy Sciences	Cat#15710
TritonX-100	Sigma	Cat#X100-500ML
10× phosphate-buffered saline (PBS)	Fisher Scientific	Cat#BP3991
Bovine Serum Albumin (BSA)	Sigma	Cat#A2153-500G
Poly-L-lysine	Sigma	Cat#P5899-5MG
Schneider’s Drosophila medium	Sigma	Cat#21-720-024
Fetal Bovine Serum (FBS)	Sigma	Cat#NC3365964
Penicillin-Streptomycin-Glutamine (100×)	Gibco (Fisher Scientific)	Cat#10-378-016
VECTASHIELD with DAPI	Vector Lab	Cat#H-1200-10
Deposited data
VCell math model	Bener et al.1	VCell model: Username: Boris; Model name: Inaba_dedifferentiation_model_public; https://vcell.org/run-vcell-software
Experimental models: Organisms/strains
D. melanogaster: nos-FRT-mCherry-stop-FRT-Gal4, UAS-GFP (Reporter-1)	Inaba lab stock17	N/A
D. melanogaster: UAS(FRT.stop)mCD8-GFP (Reporter-2)	Bloomington Drosophila Stock Center	RRID: BDSC_ 30032
D. melanogaster: pVas-Vasa mCherry	Gift from Yukiko Yamashita	FBtp0065762
D. melanogaster: Bam-FLPD5	Bener et al.1	RRID: BDSC_604597
Software and algorithms
ZEN Microscopy Software	ZEISS	RRID:SCR_013672
ImageJ Fiji	Schneider et al.18	RRID:SCR_003070 https://imagej.net/software/fiji/
Virtual Cell	Resasco et al.19; Slepchenko and Loew.20	RRID:SCR_007421 https://vcell.org/
Other
Zeiss LSM800 confocal microscope with an Airyscan module	Zeiss	RRID:SCR_015963
35mm glass bottom dishes	Nunc (Thermo Fisher)	Cat#150680
Dissection microscope	Leica	M80
Fiber optic microscope illuminator	Schott	KL 1600
Dumont #55 forceps	Fine Science Tools	1125520
Pyrex spot plate with nine depressions	Corning	Cat#7220-85
Drosophila flypads	Genesee Scientific	Cat#59-119
Drosophila CO2 bubbler kits	Genesee Scientific	Cat#59-181

The primary antibodies listed (anti-Vasa, 1b1, 7G10) are specific antibodies obtained from the DSHB and are critical for accurately identifying germline and niche cell populations. The Drosophila genotypes are also essential for the reproducing the lineage tracing results.

Unless otherwise noted, all standard laboratory chemicals (e.g., PBS, BSA, Triton X-100, Formaldehyde) and general equipment (e.g., forceps, CO2 bubbler kits, dissection microscopes) may be substituted with equivalent materials from other suppliers.

Materials and equipment

Becker Ringer’s Solution²¹

REAGENT	Final concentration (in 1× working solution)	Amount (for 1 L of 10× stock)
NaCl (5 M)	111 mM	222 mL
KCl (1 M)	1.88 mM	18.8 mL
NaH2PO4.2H2O (0.2 M)	64 μM	3.2 mL
CaCl2.2H2O (1 M)	816 μM	8.16 mL
NaHCO3 (Powder)	2.38 mM	2 g
Distilled Water	–	Bring to 1 L total volume
Total	–	1 L

Filter with 0.22 μm pore-Bottle Top Vacuum Filter. Store at 22°C–24°C. The solution is stable for several months.

To prepare 1× working solution, dilute 1:10 with ultrapure water.

Step-by-step method details

Testing lineage marking

Timing: ∼2 months; Crosses (10–12 days), aging (1–28 days), dissection/fixation/immunostaining (2–3 days), and imaging and data processing (1–2 days).

Here, we describe the steps for fixation and staining of Drosophila testes to quantify GFP-positive GSCs and validate the lineage-marked cells before proceeding to live imaging.

• 1.Cross flies carrying the bam-FLPD5 construct with flies carrying each reporter cassette (Reporter-1 or Reporter-2).
• 2.Collect adult males carrying Bam-FLPD5 and either Reporter-1 or Reporter-2 at specific time points (e.g., 0-, 7-, 14-, 21-, and 28- days post-eclosion).
• 3.Dissect testes from adult males in 1× PBS.
• 4.Fix testes in 4% formaldehyde in PBS for 30 minutes at 22°C–24°C on a nutator.
• 5.Wash samples in PBS containing 0.2% Triton X-100 (PBST) for 60 minutes (three washes of 20 minutes each).

Pause point: This incubation can be extended to ∼16 h at 4°C.

• 6.Incubate for 12–16 h at 4°C with primary antibodies diluted in 3% Bovine Serum Albumin (BSA) in PBST: rat anti-Vasa (germ cell marker, 1:20), mouse anti-Hts (to visualize connection of SGs for stage identification, 1:20), and mouse anti-FasIII (Niche cell marker, 1:40).
• 7.Wash samples in PBST for 60 minutes (three washes of 20 minutes each) at 22°C–24°C.
• 8.Incubate with fluorophore-conjugated secondary antibodies (1:400) in 3% BSA in PBST at 22°C–24°C for 2 h.

Pause point: This incubation can be extended to ∼16 h at 4°C.

CRITICAL: Protect samples from light from this step onward.

• 9.Wash samples in PBST for 60 minutes (three washes of 20 minutes each).
• 10.Mount samples on glass slides using VECTASHIELD Antifade Mounting Medium with DAPI (4′, 6-diamidino-2-phenylindole).

Pause point: Sample can be stored at 4°C in the VECTASHIELD in the tube (before mounting on the slides) for several weeks.

• 11.We used a Zeiss LSM800 to score the percentage of GFP-positive GSCs per testis.

Note: We used a 63× oil-immersion objective (NA 1.4). For GFP detection, use a 488 nm laser (power: ∼2%). Set the Gain Master to maximum to minimize photobleaching. Process images using Airyscan processing (2D processing).

Note: Z-stack imaging is required to identify the GSCs directly contacting the hub. Using a confocal microscope helps to determine GSCs based on their localization directly contacting the hub (Figure 2A and 2B). Depending on the tissue type, a widefield system can be used.

Note: GSCs are determined as the cells directly associated with the niche (Hub cell cluster). Typically, 8–10 GSCs are found to be Vasa positive germ cells direct contact with the niche cell cluster.

CRITICAL: GFP signal becomes weak after prolonged detergent treatment steps (PBST-washing). In parallel, conduct short-term live imaging to count GFP-positive cells per testis to check consistency. Usage of a GFP antibody staining can be an alternative to resolve this issue.

• 12.Compare the percentage of marked GSCs over time between Reporter-1 and Reporter-2.

Live imaging (capturing recombination events)

Timing: 1–3 months. Poly-L-lysine coating of imaging dish (2–4 h), dissection/mounting (∼1 h), and imaging and data processing (16–20 h).

This section details the dissection, mounting, and imaging of testes for long-term live imaging, adapted from.²¹ It allows for real-time quantification of de novo GFP activation in GSCs, GBs, and SGs.

• 13.
  Apply 0.5ml of 1 mg/mL poly-L-lysine solution (Sigma) to 35mm glass-bottom dishes (Nunc).

  • a.
    Incubate at 22°C–24°C for ∼4 h.
  • b.
    After incubation, replace poly-L-lysine with Becker’s Ringer’s solution.
  • c.
    Use the coated dish immediately for testis mounting.

CRITICAL: Coating beyond 7 h decreases the affinity of the surface to the testis.

• 14.
  Dissect testes from adult males (we used 0- to 7-day-old) directly in 1× Becker Ringer’s solution.

  • a.
    Perform dissection carefully to keep the testis intact.
  • b.
    Place the testis immediately onto the coated glass surface by orienting the apical tip toward the glass bottom (Figure 3).
  • c.
    Ensure proper orientation under a dissection microscope, and re-orient if necessary (Figure 3).

Note: Damaged tissues will not survive the 16-h imaging period.

CRITICAL: The orientation of the testis tip determines the imaging quality. The niche must be in contact with the glass for optimal resolution.

• 15.Remove the Becker Ringer’s solution carefully.
• 16.Add pre-warmed (22°C–24°C) Schneider’s Drosophila medium (supplemented with 10% fetal bovine serum and glutamine–penicillin–streptomycin) immediately.

CRITICAL: Add the medium gently to the side of the dish, away from the testes, to avoid dislodging them.

CRITICAL: Add sufficient medium (∼3.5 mL) to fill the dish and place the lid on top to minimize evaporation.

• 17.Place the dish on the stage of an inverted confocal microscope.

Note: We use a Zeiss LSM800 confocal microscope with an Airyscan module and a 63× oil immersion objective (NA 1.4). Calibrate the stage before each run. Use 561 nm and 488 nm lasers to detect mCherry and GFP, respectively. We use Airyscan Super-Resolution (SR) mode for all acquisitions.

• 18.
  Acquire z-stack images (10–15 stacks, 1.5∼2 μm interval) every 10∼15 min for a 12- to 16-h period.

  • a.
    Set the z-step interval as large as possible (1.5∼2 μm) to minimize scanning while still allowing accurate identification of cells in 3D geometry.
  • b.
    Use a bidirectional scan and perform sequential acquisition for the two fluorophores by frame.
  • c.
    Use the multi-position function (if available) to image multiple testes simultaneously (we imaged ∼10 testes per imaging session).

CRITICAL: Add sufficient immersion oil, especially if performing multi-position scanning.

CRITICAL: Ensure the Z-stack covers the entire niche, including all GSCs and surrounding GBs/2-cell SGs.

CRITICAL: Minimize laser power as much as possible (1–2%), otherwise testis does not survive for more than 5–6 h due to the phototoxicity.

• 19.
  Optimize image settings in advance to maximize viability of the tissue.

  • a.
    Use high-sensitivity detectors, such as Gallium arsenide phosphide (GaAsP) detectors, with a maximum detector gain (1,000 V).
  • b.
    Maximize the scan speed. Averaging is typically not necessary for Airyscan SR mode (typical speed: 2.53 s/frame at 2.06 μs/pixel).
  • c.
    Maximize the time interval between scans (10–15 min) to reduce imaging frequency while maintaining continuous tracking of individual cells.

CRITICAL: In successful imaging experiments, we do not observe any overt signs of cell death or abnormal cellular morphology during the 16-h imaging sessions. We consistently detect cells are undergoing mitosis throughout the entire imaging period. However, when the medium volume is insufficient (see point #16), all testes on the plate die simultaneously during imaging. In addition, excessive laser power causes the tissue to cease dividing and eventually undergo cell death.

CRITICAL: When using multi-position settings, limit the number of testes to ensure total acquisition cycle completes within the set interval (typically ∼10 testes, for 10 min intervals).

Note: The following section details the manual tracking of cells from the time-lapse movies to quantify de novo GFP activation events in GSCs, GBs, and SGs. These counts are necessary to calculate the “flipping ratio” (α^∗/α^True) used in the mathematical model.

• 20.Process images using Airyscan processing in the Zen software (we typically use 2D processing or 3D processing modes).
• 21.Import the processed files into ImageJ/Fiji for manual tracking.
• 22.Locate the hub (the niche cell cluster), which appears as a region surrounded by 8∼12 germ cells at the apical tip of the testis (Figure 4, asterisks).
• 23.Identify GSCs as the Vasa-mCherry positive cells directly contacting the hub interface (Figure 4, time frame 0min, inside of the white dotted line).
• 24.Identify gonialblasts (GBs) and 2-cell spermatogonia (SGs) as Vasa-mCherry positive cells positioned one or more cell layers away from the hub (Figure 4, time frame 0min, outside of the white dotted line).
• 25.Manually analyze images frame-by-frame over the entire 12- to 16-h recording period.
• 26.Detect “flipping” events, defined visually as the de novo appearance of GFP signal that clearly exceeds background levels and exhibits a sustained increase in intensity over subsequent frames (Figure 4, white arrowheads).
• 27.Classify each observed GFP activation event.

Figure 3.

Optimal positioning of the testis for live imaging

Schematic diagrams (top, Side view) and representative images (bottom, Top view) illustrate the correct and incorrect orientations of the testis on the glass-bottom dish. In the correct orientation (left panels), the apical tip (blue arrow), where the stem cell niche is located, is oriented downward to make direct contact with the poly-L-lysine coated coverslip, placing the stem cell niche within the optimal focal range of the objective. Conversely, the incorrect orientation (right panels) shows the testis lying flat on its side, which positions the niche deeper within the imaging volume. The “Top view” panels demonstrate the visual difference under a dissection microscope, where the correctly oriented testis appears anchored at the tip rather than lying flat.

Figure 4.

Visual identification of FLP recombination events during live imaging

Selected frames from a 12-h time-lapse recording of a niche illustrate the criteria for cell identification and scoring FLP recombination events. The stem cell niche (hub) is located at the apical tip (asterisk). Germ cells are identified as Vasa-mCherry positive cells (magenta). Among them, GSCs are in direct physical contact with the hub. The white dotted line (shown at 0 min time point) encircles the hub and the GSCs. GBs and SGs are defined as Vasa-mCherry positive cells positioned one or more cell layers away from the hub interface (outside the white dotted line). White arrowheads track two single GBs that initiate as GFP-negative and exhibit a gradual, sustained increase in GFP intensity (315–645 min). A recombination event is scored if a cell transitions from background-level signal to a clearly distinguishable GFP signal during the imaging session, regardless of the precise frame of onset. All scale bars are 10 μm.

Figure adapted from the related manuscript.¹

 α^∗: A de novo GFP activation occurring in a GSC (no Bam-FLP expression) without prior dedifferentiation event. This could represent the non-specific recombination rate without FLP.

 α: A de novo GFP activation occurring in a GB or SG (cells with Bam-FLP expression).

Mathematical modeling and data fitting

Timing: 3–6 months

Here, we describe steps for installing the VCell software, accessing the mathematical model, and inputting experimental data.

• 28.
  Install VCell.

  • a.
    Navigate to the VCell software download page: https://vcell.org/run-vcell-software.
  • b.
    Select the installer for your operating system and follow the installation prompts.
  • c.
    Launch the VCell application. Click “Register” on the login screen to open the registration page in your web browser.
  • d.
    Complete the free registration form.

    Note: A VCell account is required to access the public database.

  • e.
    Log in to the VCell application using your credentials.
• 29.
  Locate the specific model:

  • a.
    Click the “VCell DB” tab in the bottom-left panel.
  • b.
    Select the “MathModels” tab by clicking on the arrow to view mathematical models stored in the database.
  • c.
    Expand the list by clicking the small arrow next to “Public MathModels,” then select “Uncurated”.
  • d.
    Locate the username “boris”.
  • e.
    Find the model named “Inaba_dedifferentiation_model_public” under that username.
  • f.
    Double-click the model to open it.

Note: You will now have full access to view the model's equations and simulations as described in this protocol.

• 30.
  Create a new simulation.

  Note: The simulations panel is pre-populated with the simulation results used to generate the data for our original publication.¹ To input your own experimental data, we recommend creating a new simulation rather than overwriting the published examples.

  • a.
    Click the “New Simulation” icon in the top-left panel of the model window.
  • b.
    Locate “Simulation0” in the list.
  • c.
    Click on “Simulation0” and select “Edit Simulation” from the top panel. This will open the “Edit: Simulation0” window.
  • d.
    Click the “Parameters” tab at the top of this window.
• 31.
  Input fixed constraints and experimental data (Click the parameters tab).

  • a.
    Ensure the default values from steady-state data¹² are present (See details in related publication Bener et al.¹).
• 32.
  Configure simulation time and output settings (Click the solver tab):

  • a.
    Set time bounds:

    • i.
      Starting: 0.0.
    • ii.
      Ending: 70.0.

      Note: Use a value of 70.0 to correspond to 35 days of biological time (one time unit equals to 12 h, or 2 cycles per day). This covers the standard experimental observation window from Day 0 to Day 28.

      Note: If you wish to observe a longer experimental period, increase the “Ending” value accordingly (e.g., set to 140.0 for 70 days).

  • b.
    Set output options:

    • i.
      Select output interval.
    • ii.
      Value: 1.0.

      Note: This ensures data points are recorded every 12 h (1 cycle), providing sufficient resolution for analysis.

      Note: Following section details the use of the VCell platform to integrate the experimental data obtained in the “testing lineage marking” and “live imaging (capturing recombination events)” sections. By incorporating the “flipping ratio” derived from live imaging, the false-positive and false-negative probability can be estimated.

• 33.
  Enter the experimental variables (from “live imaging (capturing recombination events)” section):

  • a.
    α: Enter your observed rate of GFP turns on in GB/SGs (intended cells).
  • b.
    α^∗: Enter your observed rate of GFP turns on in GSCs (unintended cells).
• 34.Sum the total number of events for α^∗ and α across all analyzed movies.
• 35.Calculate the rate of GFP turns on in GB or SGs as α^True, defined as

$${\alpha }^{True}=\alpha -{\alpha }^{\ast }$$

(Rationale: Unintended background recombination is assumed to occur at the same rate in GB/SGs as it does in GSCs).

• 36.Calculate the flipping ratio (α^∗/α^True).

Note: The following rates were experimentally constrained in our previous live imaging study investigating the mechanisms of GSC replenishment in the Drosophila testis niche¹²: asymmetric division rate (p), symmetric renewal rate (q), symmetric differentiation rate (r), dedifferentiation rate (k_dd), differentiation without cell division (u), and differentiation rate of GB/SGs (β).

• 37.Input the flipping ratio (α^∗/α^True) calculated in the “live imaging (capturing recombination events)” section (Step 27) as a fixed constraint.
• 38.Run simulations to generate time-dependent curves for the fraction of GFP-positive GSCs.
• 39.Compare the simulation output against your experimental fixed-tissue data (from the “testing lineage marking” section). Calculate the Root-Mean-Square Deviation (RMSD) to assess the quality of the fit.
• 40.Calculate the sensitivities of the marking system to k_dd,α^True, and compute the probability of false-positives.

Expected outcomes

The primary outcome of the long-term live imaging is the quantification of two distinct “flipping” rates required for the mathematical model: stage-specific activation (α) and spontaneous background activation (α^∗). You should observe frequent GFP activation in GBs and SGs. For example, in our validation of the Bam-FLPD5 system with Reporter-2, we observed 32 such events across 44 testes. Conversely, there should be minimal to no GFP activation in GSCs, as evidenced by our observation of only 2 spontaneous activation events in GSCs within the same dataset. A successful lineage tracing system will yield a high ratio of specific events to background events. For the Reporter-2, the calculated background ratio (α^∗/α^True) was approximately 0.066, whereas the Reporter-1 yielded a ratio of ∼1.0, indicating high rates of spontaneous recombination.

Fitting these data to the model assesses how reliably each system captures the true dedifferentiation rate. As demonstrated in Bener et al.,¹ the frequency of GFP-positive GSCs in Reporter-1 system fails to reflect the true dedifferentiation rate (Figure 1C). In contrast, the Reporter-2 system shows a positive correlation, where the number of marked cells increases proportionally with the dedifferentiation rate (Figure 1D). Consequently, the model allows for the estimation of false-positive and false-negative rates inherent to the chosen reporter system (see¹).

Limitations

Frequency of marking events

This protocol heavily relies on live observation to determine when and where marking events happen. However, our whole tissue live imaging only allows up to 20∼24 h, thus it will be challenging to catch rare events. Previously used combination of bamGal4, UAS-FLP only marked ∼1 GSC/week in average.^10,11 Therefore, it was almost impossible to evaluate the system using live imaging. Bam-FLPD5 improved this frequency and enabled us to use this method.

Technical limitations of live imaging

All parameters used for mathematical modeling were estimated from ex vivo live imaging. However, ex vivo conditions may alter the division modes, or dedifferentiation rates compared to physiological conditions within the intact fly. Furthermore, live imaging of the complex 3D tissue presents challenges. Cellular events occurring deep within the tissue or obscured by other cells can lead missed events during tracking. Altogether, these technical factors may affect the interpretation of the results.

Troubleshooting

Problem 1

Extended live imaging of intact testes poses a significant risk of phototoxicity, which can compromise tissue viability before the 16–20 h image acquisition window is complete (related to “live imaging (capturing recombination events)” section).

Potential solution

• •Acquire images at substantially reduced laser power to minimize laser-induced damage while maintaining sufficient signal quality.
• •Use a detector with high sensitivity, such as an Airyscan gallium arsenide phosphide (GaAsP) area detector. This allows for the acquisition of high-resolution images using low laser power.

Problem 2

The probability of capturing a recombination event in one sample during a single imaging session is low (related to “live imaging (capturing recombination events)” section).

Potential solution

Mount multiple testes on a single glass-bottom dish and perform automated multi-position imaging to monitor multiple samples sequentially within every time-lapse interval. Scan full z-stacks covering entire niche volume for up to 10 to 15 testes within a 10–15 min interval.

Problem 3

Certain reporter cassettes may exhibit high levels of spontaneous (background) recombination, leading to false-positive lineage marking (related to “testing lineage marking” section).

Potential solution

• •Test multiple combinations of transgenic constructs. We found several reporter transgenic insertions have different recombination thresholds against the same FLP recombinase even if they possess same recombination targets sequence (FRT sites in this case). This is likely due to the different accessibility of FLP to each insertion site.
• •Identify the construct where the activation threshold is sufficiently high to prevent spontaneous marking in wrong cell types, yet sufficiently low to allow marking in the cells of your interest.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mayu Inaba (inaba@uchc.edu).

Technical contact

Technical questions on executing this protocol should be directed to and will be answered by the technical contacts, Muhammed Burak Bener (mburakbener@gmail.com) and Boris M. Slepchenko (boris@uchc.edu).

Materials availability

All unique reagents generated in this study are available from the lead contact without restriction.

Data and code availability

The VCell model “Inaba_dedifferentiation_model_public” is publicly available in the VCell database (https://vcell.org/) under the username “boris.” Because the model’s quantitative equations and simulations are integrated within the VCell software environment, direct access requires the VCell platform. Detailed instructions for software installation and model access are provided in the mathematical modeling and data fitting section. The data that support all experimental findings of this study are available in the BioStudies database under accession number S-SSST1903.

Acknowledgments

We thank Yukiko M. Yamashita, the Bloomington Drosophila Stock Center, and the Developmental Studies Hybridoma Bank for reagents. This research is supported by R35GM128678 from the National Institute for General Medical Sciences and start-up funds from UConn Health (to M.I.). B.M.S. was supported in part by R24GM137787 from the National Institute for General Medical Sciences. The graphical abstract was created using Biorender.com.

Author contributions

M.I. and M.B.B. conceived the project, designed and executed experiments and analyzed data. S.M.D designed and executed experiments and analyzed data. B.M.S. and M.B.B. conducted mathematical modeling and data fitting using Virtual Cell. All authors wrote and edited the manuscript.

Declaration of interests

The authors declare no competing interests.