Data from: In vitro competition between two transmissible cancers, and potential implications for their host, the Tasmanian devil
Cite this dataset
Gérard, Anne-Lise et al. (2024). Data from: In vitro competition between two transmissible cancers, and potential implications for their host, the Tasmanian devil [Dataset]. Dryad. https://doi.org/10.5061/dryad.3j9kd51s6
Abstract
Since the emergence of a transmissible cancer, devil facial tumour disease (DFT1), in the 1980s, wild Tasmanian devil populations have been in decline. In 2016, a second, independently evolved transmissible cancer (DFT2) was discovered raising concerns for survival of the host species. Here, we applied experimental and modelling frameworks to examine competition dynamics between the two transmissible cancers in vitro. Using representative cell lines for DFT1 and DFT2, we have found that in monoculture, DFT2 grows twice as fast as DFT1 but reaches lower maximum cell densities. Using co-cultures, we demonstrate that DFT2 outcompetes DFT1: the number of DFT1 cells decreasing over time, never reaching exponential growth. This phenomenon could not be replicated when cells were grown separated by a semi-permeable membrane, consistent with exertion of mechanical stress on DFT1 cells by DFT2. A logistic model and a Lotka-Volterra competition model were used to interrogate monoculture and co-culture growth curves respectively, suggesting DFT2 is a better competitor than DFT1, but also showing that competition outcomes might depend on the initial number of cells, at least in the laboratory. We provide theories how the in vitro results could be translated to observations in the wild and propose that these results may indicate that although DFT2 is currently in a smaller geographic area than DFT1, it could have the potential to outcompete DFT1. Further, we provide a framework for improving the parameterization of epidemiological models applied to these cancer lineages, which will inform future disease management.
README: In vitro competition between two transmissible cancers, and potential implications for their host, the Tasmanian devil
https://doi.org/10.5061/dryad.3j9kd51s6
This dataset contains the results of two experiments: (1) direct DFT1 & DFT2 co-cultures and (2) transwell DFT1 & DFT2 co-cultures.
Description of the data and file structure
Flow cytometry dataset consists of .fcs files with associated settings.csv files. Gating events in these allowed to calculate the number of cells for the two experiments, which can be found in the files described below. Coordinates for each gate can be found in the gating.csv files.
(1) Direct co-culture
File name: DFT1&DFT2_cocultures.xlsx
File structure: long format Exel table with the following columns:
- Experiment: experiment ID - categorical variable
- Replicate: replicate number, experiments were performed in triplicates - categorical variable
- Culture: ratio of DFT1 to DFT2 cells at the start of the experiment (e.g. "4060" means 40% DFT1 and 60% DFT2) - categorical variable
- DFT: whether the sample belongs to DFT1 or DFT2 - categorical variable
- Cells: number of cells measured - continuous variable
- Time: timepoint at which the number of cells was measured, in hours - categorical variable
(2) Transwell co-cultures
File name: transwells.xlsx
File structure: long format Exel table with the following columns:
culture: cell line (4906 = DFT1, rv = DFT2) and culture conditions - categorical variable
- "4906" refers to the number of DFT1 cells in monoculture
- "4906co" refers the number of DFT1 cells being co-cultured with DFT2
- "rv" refers to the number of DFT2 cells in monoculture
- "rvco" refers the number of DFT2 cells being co-cultured with DFT1
replicate: replicate number, experiments were performed in triplicates - categorical variable
time: timepoint at which the number of cells was measured, in days - categorical variable
cells_per_mL: number of cells measured - continuous variable
DFT: whether the sample belongs to DFT1 or DFT2 - categorical variable
experiment: experiment ID - categorical variable
Code/Software
Code was written in the R programming language (version 4.1.3) to perform the following analyses and associated plots:
- flowcyto.R gating of flow cytometry and counting the number of cells for each sample
- gridsearch.R for model fitting and estimating growth parameters r and K
- stats.R for statistical analysis
Methods
Cell cultures
Representative cell lines for DFT1 (4906, also known as 88 [41]) and DFT2 (RV also known as TD467 or 202T1 [42]) were cultured as previously described [41], [42]. DFT1 and DFT2 cell lines have been shown to share many characteristics of in vivo tumours [43], [44], [45], making these cell lines a useful in vitro study system. Cells were maintained in RPMI-1640 media with GlutaMAX (Gibco, 61870036) supplemented with 10% heat-inactivated FBS (Gibco, 10500-064) and 50 ug/mL penicillin/streptomycin (Gibco, 15070063) or 50 ug/mL of gentamicin (Sigma, G1397) at 35°C and 5% CO2. Upon reaching 80-90% confluency, cells were detached using TrypLE Express (Gibco, 12605010) and passaged 1:3. Cells were maintained below Passage 30. DFT1 cell lines were tested for mycoplasma in [46]. All DFT2 cells were tested for mycoplasma [insert details such as kit name] when they entered the lab. They have been passaged successfully in a mycoplasma free tissue culture facility since then.
Transduction (Green Fluorescent Protein cell lines)
Distinguishing DFT1 and DFT2 cells using flow cytometry based solely on their size and shape is difficult, thus, we labelled one of these cell lines with Green Fluorescent Protein (GFP), allowing us to distinguish co-cultured tumour cells reliably. The DFT1-GFP (4906-GFP) and DFT2-GFP (RV-GFP) cell lines were established using lentiviral transduction. HEK293T cells were used to produce lentiviral particles with PLKO-GFP (pLKO_TRC001), psPAX2 and pMD2.G (provided by N. Divecha). Lentivirus particles were transduced into 4906 and RV cells. Following transduction, approximately 10^{5} cells expressing high levels of GFP relative to untransduced control cells were sorted by Fluorescence Activated Cell Sorting (FACS) on a BD FACS Aria II using the gating strategy presented in Fig. S1. These sorted cells were then cultured for a further two weeks before being sorted for a second time to establish a heterogeneous cell line with stable and high expression of GFP and remove cells which were not expressing GFP.
Direct co-cultures
DFT1 and DFT2 cells were cultured in 12-well plates (Corning, 3513) for 14 days. Cells were plated in 1mL of culture media per well (for a concentration of 10^{5} cells/mL of media), which was replaced every 3 days. Triplicate wells were harvested daily and counted on a Guava® easyCyte™ model 6HT. The following culture conditions with varying DFT1 to DFT2 ratios at the start of the experiment were performed: giving an advantage to DFT1 (70-80% DFT1), giving no advantage to either cell line (50-60% DFT1), giving an advantage to DFT2 (30-40% DFT1) and monoculture controls (100% DFT1 or 100% DFT2). A total number of 10^{5} cells per well were plated, meaning a 50:50 co-culture will start with 0.5x10^{5} DFT1 cells and 0.5x10^{5} DFT2 cells; while a monoculture will start with 10^{5} cells of either DFT1 or DFT2. This experiment was performed in duplicate, alternating the use of one GFP-cell line and one unlabelled cell line (i.e. DFT1-GFP was co-cultured with DFT2, and DFT1 was co-cultured with DFT2-GFP) to eliminate potential effects of the GFP-transduction process and selection by FACS on cell growth. Imaging of the cells during one representative direct co-culture experiment can be found in Fig. S5.
Transwell co-cultures
DFT1 and DFT2 cells were co-cultured using 12-well transwell plates (Corning, CLS3460). Transwells allow cells to remain in two compartments separated by a semi-permeable membrane, permitting small molecules to be exchanged but keeping cell lines separated. Monocultures (cells of a same DFT grown in the inserts and wells) were compared to co-cultures (cells of one DFT in the inserts and of the other DFT in the wells). Cells were plated at a density of 0.5x10^{5 }cells/insert and 10^{5} cells/well (the inserts having a surface about two times smaller than the wells), for both mono-cultures and for co-cultures. Cells were plated in 0.5mL of culture media in the inserts and 1mL of media in the wells (for a final concentration of 10^{5} cells/mL of media) which was replaced every 3 days. Duplicate wells were harvested and analysed on a Guava® easyCyte™ model 6HT every two days. Only cells from the wells were counted to avoid any effect of the surface size and type of the transwell inserts on cell growth. The experiment was performed in duplicate.
Flow cytometry
Cells were incubated in the dark on ice for 15 minutes with 1 ug/mL Propidium Iodide as a live/dead marker. Cells were run on a Guava® easyCyte™ model 6HT and data was analysed using the CytoExploreR R package [47]. Gating performed first selected cells from debris (FSC-Height vs SSC-Height), then singlets from doublets (SSC-Area, SSC-Height), live from dead cells (FSC vs RED), and, for the direct co-cultures, GFP-positive from GFP-negative cells (FSC vs GRN) (a representative gating strategy is shown in Fig. S2). The number of cells in each well was calculated as follows: (number of gated events / volume analysed by the flow cytometer) x volume of cells per well (1 mL).
Growth rate and carrying capacity estimation
A logistic differential equation (Eq. 1) was used to represent the growth of DFT1, DFT1-GFP, DFT2 and DFT2-GFP cell lines as this model has been shown to accurately describe DFT cell growth in vitro in a preliminary analysis [48]and in vivo [49]. In equation (1), N represents the number of cells, r the per capita cellular growth rate (per day) and Kthe maximum number of cells the space and resources can accommodate (i.e., carrying capacity).
(1) dN/dt = rN(1-N/K)
A grid search method was used to simulate growth curves using 10 000 combinations of the growth rate r (ranging from 0.01 to 1, with a step of 0.01 per day) and K (ranging from 10^{4} to 10^{6}, with a step of 10^{4 }cells). N was initialised with the number of cells at day 1 of the experiment, that is once cells have had time to attach to the surface of the plate and unattached dead cells were removed, to meet the model’s assumption that cell population grows with time. The adequacy of each combination of parameter values was then assessed on each replicate growth curve of the DFT cells in monoculture (direct co-cultures) and in co-culture (transwell co-cultures) by calculating the Root Mean Square Error (RMSE) between the simulated and the observed population dynamics of each setting. Parameters from simulations with the best fit (i.e., lowest RMSE) to the experimental data were then selected. Median values of r and K were compared between DFT1 and DFT2 cell lines using a Wilcoxon rank sum test with continuity correction. The r and Kparameters were also estimated for DFT and DFT-GFP cell lines, which, showed that transduction appears to have lowered the carrying capacity of DFT1-GFP cells (Fig. S3).
Competition coefficient estimation
The two-species competition Lotka-Volterra equations (Eq. 2 and 3 [29], [30]) were used to quantify competitive interactions between DFT1 and DFT2 cells. N_{i}, r_{i}, and K_{i} represent the number of cells, growth rate and carrying capacity for DFT1 (i = 1) and DFT2 (i = 2). The ? parameter represents the competitive impact of DFT1 on the growth rate of DFT2, and vice-versa for ?. If a competition coefficient (? or ?) is close to zero, a tumour line does not influence the growth of the other; if ? or ? is bigger than 1, a tumour line negatively impacts the growth of the other; and if ? or ? is lower than zero, a tumour line facilitates the growth of the other. Hence, ? and ? inform the type of interaction between DFT1 and DFT2: competition (alpha and beta are positive), mutualism (alpha and beta are negative), commensalism (alpha is negative and beta is close to or equal to zero, or vice versa) or parasitism (alpha is positive and beta is negative, or vice versa).
(2) dN1/dt = r1N1(1-(N1+aN2)/K1)
(3) dN1/dt = r2N2(1-(N2+bN1)/K2)
Again, a grid search method was used to simulate growth curves using 10 404 combinations of ? and ? (each ranging from -100 to 100, with a step of 1). Mean values of r_{1}, r_{2}, K_{1} and K_{2} estimated on the monocultures, as described above, were fixed in the equations to only estimate the competition coefficients. The estimation of these parameters values was performed using the same approach than previously but using this time the population dynamics of both DFTs in direct co-culture. Median values of ? and ? were compared using a Wilcoxon rank sum test with continuity correction. Model fitting and statistical analyses were performed in R (R version 4.1.3 [50]).
Predicting competition outcome
Competition outcome of the Lotka-Volterra model can be predicted by examining zero-growth isoclines of the two competing species, e.g. [51]. Briefly, the number of cells at which the DFT1 or DFT2 population stops growing can be found by solving for dN_{i}/dt = 0. The trajectory of both populations can then be represented on a phase diagram in which the zero net growth isoclines are given by Eq. 4 and 5. The coordinates of the isoclines correspond to the intercepts of both axis (i.e., [N_{1,t}=0, N_{2,t}=0]). DFT1’s zero net growth isocline has the coordinates [0, K_{1}/?] and [K_{1}, 0], and DFT2’s [0, K_{2}] and [K_{2}/?, 0]. From these, we can determine the following outcomes: one of the tumour lines always outcompetes the other, competition outcome depends on initial conditions (i.e. the number of DFT1 and DFT2 cells at the start of the experiment) or both tumour lines co-exist (Table 1).
(4) N1=K1-aN2
(5) N2=K2-bN1
Funding
University of Tasmania, Dr Eric Guiler Tasmanian Devil Research grant