Background
Model organisms are very important for understanding human diseases [
1]. Of the current available vertebrate animal models, genetic and experimental zebrafish and mouse models have contributed significantly to advancing our insights into cancer biology and therapy [
2], largely due to the high genomic similarities they share with humans [
3].
Zebrafish, described for the first time by George Streisinger, emerged as a model organism for developmental genetics in the 1960s [
4]. Since then, the zebrafish model has been extensively used in biomedical studies for several reasons, which include the large numbers of descendants, small physical size, reduced cost of maintenance, availability of genetically modified lines, embryo transparency and, most importantly, the small amount of drug required for testing new compounds in drug screening assays [
5].
The capacity of tumors to efficiently engraft in animals has been known since the first experiments were conducted in mice in the 19XXs [
6]. Tumor transplantation in animal models is very informative; not only can it provide data on tumor growth and the metastatic potential of tumor cells, but it also offers the possibility to test drugs in an in vivo animal setting, which could be putatively applied to the clinical setting to determine the best treatment for patients [
7]. While these types of studies have been carried out for nearly half a century with mouse models, it was not until 2006 that Haldi et al. [
8] performed the first xenograft of leukemia cells in zebrafish, highlighting the potential use of this alternate vertebrate animal for tumor-based studies. Since then, several studies have demonstrated that zebrafish are an excellent model for the transplantation of tumor cell lines or primary patient derived cells [
9‐
12]. Specifically, with zebrafish there are three opportunities available for tumor cell transplantation: 1) the embryo stage, when the innate immune system is present, but the acquired immune system has yet to be fully developed, 2) transplantation in adult fish lines, such as Casper, with rejection inhibiting treatments [
13] or 3) the recently developed immunocompromised adult rag2
E450fs mutant zebrafish [
12,
14].
The most commonly used methodology for cancer xenograft assays in zebrafish consists in real time tracking of tumor cells (labeled with different fluorescent dyes [
15] or that constitutively express a fluorescent protein in the cytoplasm) introduced via microinjection into particular zones of 48 h-old zebrafish embryos. Tumor cell proliferation and their invasive capacity can then be analyzed over the course of several days [
16].
When compared to the murine xenotransplantation model, the zebrafish model offers several advantages, including the high number of embryos that can be injected, allowing for statistical analysis of the aforementioned parameters in a few days; the possibility to test several concentrations/combinations of drugs in 96 well plate formats [
9,
17]; and the lack of an acquired immune system in embryos [
18], which facilitates tumor cell engraftment.
While a promising model, several drawbacks need to be considered when using zebrafish for xenotransplantation assays. One of the most important [
19] limitations is the temperature (28 °C) at which these fish are routinely maintained, which differs by 9 degrees from that of the human body (37 °C), the latter being the ideal temperature for tumor cell proliferation. To tackle this problem, several groups have described incubation temperatures for xenografts in zebrafish ranging from 31 °C to 34 °C [see Additional file
1: Figure S6], as a compromise solution between the optimal temperature for human cell proliferation and zebrafish survival.
The analysis of cellular proliferation inside the embryo is another challenge considering the high number of fish that need to be imaged in high resolution, and the short period of time available to test different compounds and examine the effect on the injected cells [
20,
21]. Different image analyses can be performed using commercial and free software to estimate the number of cells at the beginning and at the end of the experiment [
22,
23], but these techniques are not accurate enough to reliably measure the proliferation of the cells as they are dependent on user intervention in terms of manually adjusting parameters for each image. In this work, we introduce the software ZFtool, which addresses the current problems faced in zebrafish imaging as the features used to extract the proliferation index (area and mean intensity of GFP points) with ZFtool are automatically computed and adapted to the autofluorescent characteristics of each fish. In this way, the measurements are repeatable, reproducible and reliable without user intervention. Performing the necessary computations on a fish-by-fish and stage-by-stage level, and manually adjusting all the parameters results in data that are difficult to compare leading to unreliable results. To provide a solution to this inherent problem, we developed, implemented and validated the automatic ZFtool methodology as described below. At this moment, the software is a Matlab toolbox and the software interface is currently under development.
To significantly improve the technique of assaying different chemotherapeutic agents in an in vivo system, at a temperature almost equal to that of the human body, and in a fast and efficient way, in this study we present a zebrafish yolk xenotransplantation assay together with an image analysis software that provides an answer to the main problems currently faced in the zebrafish xenotransplantation community. Tumor cell injection and rearing conditions were established so that experiments were performed at 36 °C, a temperature that to our knowledge has not been reported before for this type of assay. The conditions utilized showed a good overall survival rate of the embryos, facilitated tumor growth, and together with the automated measurements obtained with the new ad-hoc imaging analysis software ZFtool, we were able to accurately monitor tumor growth with high reproducibility in order to generate reliable results.
Methods
Zebrafish handling
The care, use and treatment of zebrafish were performed in agreement with the Animal Care and Use Committee of the University of Santiago de Compostela and the standard protocols of Spain (Directive 2012-63-UE). The protocol was approved by the Animal Care and Use Committee of the University of Santiago de Compostela. One-year-old adult zebrafish (
Danio rerio, wild-type) were maintained at 28.5 °C in 30 L aquaria at a rate of 1 fish per liter of water, with a light-dark cycle of 14:10. Zebrafish embryos were obtained from mating adults according to previously described procedures [
24]. When needed, embryos were euthanized by tricaine overdose.
Reagents and cell culture
The human colorectal cancer cell line HCT116 was obtained from American Type Culture Collection (ATCC, Catalog No. CCL-247) and cultured using McCoy’s 5A Medium containing 10% FBS (GIBCO, Invitrogen) and 1% Pen/Strep (GIBCO, Invitrogen) at 37 °C with 5% CO2 in a humidified atmosphere. The HCT116 cell line was transfected to express GFP constitutively. The HCT116 line was tested monthly for contamination.
Fluorescent GFP cell labeling
HCT116 cells were transduced using a lentiviral-driven GFP construct (Sigma, Mission TurboGFP, SHC003 V). Cells were placed 72 h post infection under selective pressure using 10 μg/ml puromycin. The rate of GFP positive cells was tested using flow cytometry (BD FACS Aria I, software FACSDiva 6.0.3).
Cell proliferation assays
Cell proliferation was determined using xCELLigence Real-Time Cell Analyzer; Acea Biosciences (Roche) following the manufacturer instructions. In brief, cells were seeded on E-plates containing electric nodes in their surface that allow the measurement of changes in impedance attributed to cell proliferation. Measurements were performed in quadruplicate, normalizing the initial cell index once the cells were completely adhered.
Cell injection
Two days post fertilization (dpf), zebrafish embryos were dechorionated (if needed) and anesthetized with 0.003% tricaine (Sigma). Cells were suspended at 10,000-20,000 cells/μl in complete McCoy and maintained at room temperature for no longer than 2 h before they were injected. The cell suspension was loaded into borosilicate glass capillary needles (1 mm O.D. × 0.78 mm I.D.; Harvard Apparatus), and injections were performed using IM-31 Electric Microinjector (Narishige) with an output pressure of 34 kPa and 30 ms injection time. The injections were performed manually right into the yolk of the embryo. Incorrectly injected embryos without cells inside of the yolk, or showing them in the circulation after xenotransplantation were discarded.
Incubation, imaging and cell quantification
After injection, 2dpf embryos were incubated at two different conditions (34 °C or 36 °C) in 24-well plates with salt dechlorinate tap water (SDTW, chlorine free water obtained with a reverse osmosis filter system) for 72 h to check the proliferation of the cell line by ZFtool. Each plate contained at least 2 negative controls (injected with complete McCoy medium) and 2 blanks (not injected). Apart from those plates, another plate with 12 negative controls and 12 blanks were included in some experiments to test the viability of the embryos. No development abnormalities were observed during incubation at this temperature.
In order to reach a 36 °C incubation temperature without a large amount of embryo mortality, plates were covered with a transparent sealing tape (PCR Plastics) to prevent evaporation and reduction of dissolved oxygen. After that, plates were placed inside an incubator with minimal contact between the plate and the incubator structure to prevent water overheating.
Each embryo was photographed with AZ-100 Nikon fluorescence stereomicroscope at 0 hpi and 72 hpi to be analyzed by ZFtool software. The objective of this software is to automatize and improve the task of measuring the number and mean value of GFP pixels in order to compare them for these two conditions and compute the proliferation index. Finally, this analysis yields the number of GFP pixels in the image (nGFP), which represents the area of the cells inside the yolk sac at two different times and the GFP intensity Medium Value (GMV), which represents the medium intensity of the fluorescence inside the yolk. By multiplying the nGFP number by the GMV of each image, we determined the proliferation ratio between 0 hpi and 72 hpi to estimate the cell growth. The result obtained at 72 hpi was divided by that obtained at 0 hpi, yielding a proliferation index value (PI):
$$ \frac{nGFP_{72 hpi}\bullet {GMV}_{72 hpi}}{nGFP_{0 hpi}\bullet {GMV}_{0 hpi}} $$
A PI value =1 means that cells remain stable during incubation, a PI higher than 1 indicates tumor cell proliferation and a PI lower than 1 indicates tumor cell death.
Zebrafish embryos have variable autofluorescence, especially in the yolk area. To accurately quantify the injected cells fluorescence a pre-processing is needed to only count the GFP pixels belonging to injected cells filtering autofluorescence. To achieve this, the software counts the number of GFP pixels with different intensity thresholds, from 0 (no threshold) to 50 (Fig.
2) and the ZFtool algorithm provides a homogeneous measurement of the GFP area for all fish analyzed comparing nGFP for each threshold analyzed with nGFP for threshold = 0, where fish auto fluorescence is mostly present. When the relation between measured nGFP compared to nGFP at threshold = 0 surpass a fixed value, we consider the GFP area to be stable and the threshold is fixed at this point. In case there is no autofluorescence in the embryo, the threshold is established based on a tolerance parameter and a correction is included to assure the accuracy of the measurement in this cases. The ZFtool algorithm automatic thresholding for each analyzed embryo is one of the main automation components of the software, making it efficient in producing reliable fish to fish measurements.
Cell counting software
The ZFTool extension for cell counting was developed. A drop of cells was placed on a microscope slide and photographed to obtain a fluorescence image. The algorithm detects circular objects of the fluorescence input image with a fixed diameter. The output yields a fluorescence image with nearly every cell or group of cells delimited by a contour and an estimation of the number of cells inside the input image. This algorithm is based on the circular Hough transform and has several parameters fixing the strength of the edge, and a minimum and maximum radius of the circles to detect. As we know the approximate size of the cells, we can fix these parameters in order to obtain an estimation of the number of cells. The method will be more accurate as the cells are more isolated, but as the number of cells injected increases over 400, we do not need the exact number of cells, but only an estimation, so this method still fits our purposes.
Anticancer drugs toxicity and treatment
In order to test the toxicity of an anticancer drug (5-Fluorouracil), experiments were performed according to the OECD (Organisation for Economic Co-operation and Development) guideline for the testing of chemicals [
25]. This procedure consists of exposing 0 h post fecundation (hpf) eggs to dissolved chemicals in 24-well plates, for a period of 96 h. Various indicators (such as coagulation of embryos, lack of somite formation, non-detachment of the tail or lack of heartbeat) were checked every 24 h during the experiment, to test the mortality of the embryos and calculate the LC50 (lethal concentration 50%) at the end of the test. The drug was tested to determine a concentration range that included 0–100% mortality. Experiments were considered valid when egg fertilization was ≥ 70%. At the beginning, the oxygen concentration should have ≥ 80% saturation, and the water temperature should be 26 ± 1 °C. During the test, the negative control embryos mortality could not be ≥ 10% at any time of the experiment. Exposure to the positive control resulted in a minimum mortality of 30% at the end, and the hatching rate of the negative control embryos was higher than 80 % at 96 h. The concentrations tested were 250 μM, 500 μM, 1000 μM, 1500 μM, 2000 μM, with 1% DMSO. Another analog experiment was conducted changing the treatment starting point from 0 hpf to 48hpf in order to evaluate how the toxicity changed with a dechorionated embryo at 36 °C.
Statistical analysis
Homoscedasticity and statistical analyses were performed using the SPSS software (IBM). An excel outlier analysis was performed using interquartile range (IQR), while the outliers were discarded. One factor ANOVA for non-parametrical data was applied to non-homoscedastic data with confidence intervals of 95% or 99%, and a Student’s t-test was applied to homoscedastic data with confidence intervals of 95% or 99%. Number of embryos analyzed is represented by nrep and ntotal, being nrep the number of embryos in each replica, and ntotal the total number of embryos statistically analyzed for the experiment.
Discussion
Model organisms as zebrafish have become a very important tool in the study of human diseases in recent years. Zebrafish, due to its characteristics and advantages, has emerged as an ideal model to study the behavior of different types of cancers and to test new chemotherapeutic compounds [
28]. While the yolk does not provide the ideal microenvironment for tumor cells it is the suitable place to inject the cells to rapidly test chemotherapeutic compounds. Even this, enhancements of the xenotransplantation technique are required, together with accurate imaging analysis software to verify the fate of the cells inside the zebrafish embryo assuring a rapid analysis of xenotransplanted human cells when exposed to different treatments. This study describes an improvement in the xenotransplantation conditions in relation to temperature and the establishment of the injected cells combined with ZFtool image analysis.
Different authors reported normal development of zebrafish embryos up to 35.5 °C [
29,
30], but a range of temperatures was tested in order to reduce the mortality of the embryos [see Additional file
1: Figure S6]. Some authors noted that more assays would be needed to check the proliferation, migration, and response of the cells to drugs at higher temperatures despite the potential increase in mortality [
19].
We have set the temperature of the cells xenotransplanted into the zebrafish embryos closer to human temperature by raising the temperature from 28 °C (normal temperature at which zebrafish embryos develop) to 36 °C, with no significant change in mortality and sporadic developmental defects (curvature), due to the higher temperature, on the surviving embryos at 3 days post injection. Embryo incubation temperature is important to test the effects of anticancer drugs [
9,
31], otherwise the temperature could affect the proliferation rate of the injected cells, and the effect of the drug is underestimated. The results in this study clearly show that the proliferation of injected cells and their response to anticancer drugs is better at 36 °C than at 34 °C; 36 °C being the temperature closer to their optimal growth temperature of 37 °C [
8].
The number of injected cells is very relevant in terms of the proliferation and migration of these cells and should be considered for improved xenotransplantation and anticancer drug proliferation assays. The proliferation rate of the cells injected inside the embryos decays when the number of initial cells is insufficient at 34 °C. This may be due to cell-cell interactions: the cells injected appear to be isolated and cannot interact among themselves in order to proliferate properly. Nevertheless, even if the number of injected cells at 0 hpi is reduced, at 36 °C a higher proliferation rate of these cells exists compared to 34 °C, where the proliferation rate is absent. This previous point was assayed in vivo, demonstrating that despite the number of injected cells and mortality, 36 °C is an optimal temperature for cell growth. On the other hand, cell migration can also be modified, depending on the number of cells injected. Cells will not be able to migrate when the number of injected cells is insufficient. It is reported that 400 cells is the optimal number of injected cells to study these behaviors. Our cell line HCT116 remained in the yolk of the embryo from 0 hpi to 72 hpi, consistent with other authors that used the HCT116 cell line. This cell line has a low dissemination ratio [
9,
22].
In in vitro studies, other authors have performed in vitro proliferation assays with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide colorimetric assay (MTT). The initial cell density seeded on the plates was the same for each experiment, so there was no assessment of how the proliferation could change with different concentrations of the initial cells seeded [
9,
20]. In this study, we show that at least for the cell line HCT116, the temperature and number of initialy seeded cells are critical factors for the proliferation of injected cells.
Together with the work done for the improvement of zebrafish xenotransplantation, a method inside ZFTool was developed to measure the cell proliferation inside the yolk of the embryo. This method was designed to fill the gap present in the current methodology that does not adequately quantify cell proliferation at different times in vivo. For example, flow cytometry is not sensitive enough to quantify the number of cells in dispersed embryos [
32], and software used by other authors, such as ImageJ or Photoshop, does not automatically quantify proliferation in order to compare high number of fishes per experiment, since they require a higher amount of user intervention per fish.
In summary, we demonstrated that at 36 °C, a better proliferation rate exists for the injected cells inside the embryos, with no significant mortality changes compared with 34 °C. Our results also reveal a correlation between the number of initially injected cells and the proliferation ratio when comparing the two different temperatures. In addition, we used a new image analysis software, the ZFtool, which improves tumor cell quantification in vivo with accuracy and speed. One of the future challenges will be the quantification of these cells with a 3D method with much greater accuracy, reaching the count of each cell individually.