Preclinical models aim to advance the knowledge of cancer biology and predict treatment efficacy. This paper describes the generation of zebrafish-based patient-derived xenografts (zPDXs) with tumor tissue fragments. The zPDXs were treated with chemotherapy, the therapeutic effect of which was assessed in terms of cell apoptosis of the transplanted tissue.
Cancer is one of the main causes of death worldwide, and the incidence of many types of cancer continues to increase. Much progress has been made in terms of screening, prevention, and treatment; however, preclinical models that predict the chemosensitivity profile of cancer patients are still lacking. To fill this gap, an in vivo patient-derived xenograft model was developed and validated. The model was based on zebrafish (Danio rerio) embryos at 2 days post fertilization, which were used as recipients of xenograft fragments of tumor tissue taken from a patient’s surgical specimen.
It is also worth noting that bioptic samples were not digested or disaggregated, in order to maintain the tumor microenvironment, which is crucial in terms of analyzing tumor behavior and the response to therapy. The protocol details a method for establishing zebrafish-based patient-derived xenografts (zPDXs) from primary solid tumor surgical resection. After screening by an anatomopathologist, the specimen is dissected using a scalpel blade. Necrotic tissue, vessels, or fatty tissue are removed and then chopped into 0.3 mm x 0.3 mm x 0.3 mm pieces.
The pieces are then fluorescently labeled and xenotransplanted into the perivitelline space of zebrafish embryos. A large number of embryos can be processed at a low cost, enabling high-throughput in vivo analyses of the chemosensitivity of zPDXs to multiple anticancer drugs. Confocal images are routinely acquired to detect and quantify the apoptotic levels induced by chemotherapy treatment compared to the control group. The xenograft procedure has a significant time advantage, since it can be completed in a single day, providing a reasonable time window to carry out a therapeutic screening for co-clinical trials.
One of the problems of clinical cancer research is that cancer is not a single disease, but a variety of different diseases that can evolve over time, requiring specific treatments depending on the characteristics of the tumor itself and the patient1. Consequently, the challenge is to move toward patient-oriented cancer research, in order to identify new personalized strategies for the early prediction of cancer treatment outcomes2. This is particularly relevant for pancreatic ductal adenocarcinoma (PDAC), since it is considered a hard-to-treat cancer, with a 5-year survival rate of 11%3.
The late diagnosis, rapid progression, and lack of effective therapies remain the most pressing clinical problems of PDAC. The main challenge is, therefore, to model the patient and identify biomarkers that can be applied in the clinic to select the most effective therapy in line with personalized medicine4,5,6. Over time, novel approaches have been proposed to model cancer diseases: patient-derived organoids (PDOs) and mouse patient-derived xenografts (mPDXs) originated from a source of human tumor tissue. They have been used to reproduce the disease to study the response and the resistance to therapy, as well as disease recurrence7,8,9.
Similarly, interest in zebrafish-based patient-derived xenograft (zPDX) models has increased, thanks to their unique and promising characteristics10, representing a quick and low-cost tool for cancer research11,12. zPDX models require only a small tumor sample size, which makes high-throughput screening of chemotherapy feasible13. The most common technique used for zPDX models is based on complete sample digestion and implantation of the primary cell populations, which partially reproduces the tumor, but has the disadvantages of a lack of tumor microenvironment and crosstalk between malignant and healthy cells14.
This work shows how zPDXs can be used as a preclinical model to identify the chemosensitivity profile of pancreatic cancer patients. The valuable strategy facilitates the xenograft process, since there is no need for cell expansion, allowing for the acceleration of the chemotherapy screening. The strength of the model is that all the microenvironment components are maintained as they are in the patient cancer tissue, because, as it is well known, the behavior of the tumor depends on their interplay15,16. This is highly favorable over alternative methods in the literature, as it is possible to preserve the tumor heterogeneity and contribute to improving the predictability of the treatment outcome and relapse in a patient-specific manner, thus enabling the zPDX model to be used in co-clinical trials. This manuscript describes the steps involved in making the zPDX model, starting with a piece of patient tumor resection and treating it to analyze the response to chemotherapy.
The Italian Ministry of Public Health approved all the animal experiments described, in conformity with the Directive 2010/63/EU on the use and care of animals. The local Ethical Committee approved the study, under registration number 70213. Informed consent was obtained from all subjects involved. Before starting, all the solutions and the equipment should be prepared (section 1) and the fish should be crossed (section 2).
1. Preparation of solutions and equipment
NOTE: See Table 1 for the solutions and media to be prepared.
2. Fish crossing and egg collection
3. Specimen collection
NOTE: Autoclave forceps and a scalpel handle.
4. Sample processing
NOTE: Perform the steps under a sterile tissue culture laminar flow hood.
5. Establishment of zPDX
NOTE: Perform the steps under a sterile tissue culture laminar flow hood.
6. Treatment
7. Whole-mount immunofluorescent staining
NOTE: Before starting, place acetone at -20 °C and prepare the solutions listed in Table 1.
8. Imaging
9. Analysis of apoptosis by ImageJ
This protocol describes the experimental approach for establishing zPDXs from primary human pancreatic adenocarcinoma. A tumor sample was collected, minced, and stained using fluorescent dye, as described in protocol section 4. zPDXs were then successfully established by implantation of a piece of tumor into the perivitelline space of 2 dpf zebrafish embryos, as described in protocol section 5. As described in protocol section 6, the zPDXs were further screened to identify the chemotherapy sensitivity profiles of patient-derived cancer cells. For example, the chemotherapy combination FOLFOXIRI (5-fluorouracil, folinic acid, oxaliplatin, and irinotecan) was tested, since it is used as the first-line chemotherapy for advanced pancreatic ductal adenocarcinoma and in metastatic colorectal cancer. Whole-mount images of the zPDXs were acquired as Z-stacks on the confocal microscope, and the apoptosis induction was analyzed as described in protocol sections 8 and 9. As shown in Figure 1A,B, the combined therapy led to an increase in cell apoptosis compared to the control group. In the case study reported here, a statistically significant increase in the fraction of apoptotic cells in implanted xenografts was identified for the FOLFOXIRI-treated group compared to the control group (Figure 1C).
Figure 1: zPDXs from human pancreatic ductal adenocarcinoma 3 days after FOLFOXIRI treatment. The cell membranes were stained with CM-DiI (red) and the tissue was xenografted into the perivitelline space of 2 dpf AB wild-type zebrafish. Larvae were exposed to a FOLFOXIRI combination (0.216 mg/mL 5-fluorouracil, 0.013 mg/mL folinic acid, 0.006 mg/mL oxaliplatin, 0.011 mg/mL irinotecan) or not exposed (CTRL) for 3 days, fixed, and immunostained with cleaved caspase-3 antibody (cyan) and counterstained by Hoechst (blue nuclei). Larvae were mounted with Aqua Poly-Mount in glass microscope slides. (A1–3) Representative example of a control larva (not exposed to chemotherapy). No cleaved caspase-3-positive cells were observed. (B1–3) Representative example of a xenografted larva 3 days after FOLFOXIRI treatment, showing consistent activation of cleaved caspase-3. Whole-mount images were captured on a Nikon A1 confocal microscope with a digital camera. The dashed lines show the xenograft areas. (C) Quantification of cleaved caspase-3 and CM-DiI double positive cells in xenografted larvae treated with FOLFOXIRI compared with CTRL, plotted as mean ± SEM (n ≥ 12); **** p < 0.001, Mann-Whitney U test. Scale bars = 100 µm. Abbreviations: zPDX = zebrafish patient-derived xenograft; dpf = days post fertilization; FOLFOXIRI = 5-fluorouracil, folinic acid, oxaliplatin, and irinotecan; CTRL = control. Please click here to view a larger version of this figure. |
Solution/medium | Composition, comments | Purpose | ||
Tumor medium (1x) | To RPMI-1640 medium, add penicillin-streptomycin (final concentration 100 U/mL) and amphotericin (final concentration 2.50 µg/mL). | |||
E3 zebrafish medium (1x) | Dilute the 60x E3 embryo medium (NaCl 3 M, KCl 0.1 M, CaCl2 0.2 M, MgSO4 0.2 M) in deionized water to a final working concentration of 1x. | |||
E3 1% Pen-Strep | Add 1 mL of penicillin-streptomycin to 99 mL of E3 zebrafish medium. Mix by inversion. Store the solution at 4 °C | |||
PTw | 0.1% Tween in PBS | Whole-mount immunostaining | ||
Blocking buffer | 10% goat serum, 1% DMSO, 1% BSA, 0.8% Triton X-100 in PTw | Whole-mount immunostaining | ||
Incubation buffer | 1% goat serum, 1% DMSO, 1% BSA, 0.8% Triton X-100 in PTw | Whole-mount immunostaining | ||
PBS-TS | 10% goat serum, 1% Triton X-100 in PBS | Whole-mount immunostaining | ||
PBS-T | 1% Triton X-100 in PBS | Whole-mount immunostaining |
Table 1: Solutions and media used in the protocol.
Supplementary Figure S1: PDAC tumor sample was DiO stained (green) and xenografted into the perivitelline space of 2 dpf zebrafish embryos. After a 3-day incubation time, zPDXs were injected with 2 µg/mL propidium iodide (PI; red) and then subjected to 3D confocal imaging. The cell viability was checked by measuring PI-positive cells out of the total number of human cells (DiO positive). Scale bar = 100 µm. Abbreviations: PDAC = pancreatic ductal adenocarcinoma; dpf = days post fertilization; zPDX = zebrafish patient-derived xenograft. Please click here to download this File.
Supplementary Video S1: Example of apoptotic human cell counting, using the multi-point tool of ImageJ. The video is a Z-stack of a zPDX 3 days post implantation, acquired after cleaved caspase-3 and Hoechst 33258 immunostaining. Abbreviations: zPDX = zebrafish patient-derived xenograft. Please click here to download this Video.
In vivo models in cancer research provide invaluable tools to understand cancer biology and predict the cancer treatment response. Currently, different in vivo models are available, for example, genetically modified animals (transgenic and knockout mice) or patient-derived xenografts from human primary cells. Despite many optimal features, each one has various limitations. In particular, the aforementioned models lack a reliable way to mimic the patient tumor tissue microenvironment.
It has been suggested that the tumor microenvironment may play many important roles in drug response18. Therefore, to find a suitable animal model that maintains the tumor tissue microenvironment, an alternative PDX model was developed, using pieces of tumor tissue xenografted in 2 dpf zebrafish embryos. The protocol describes how to perform xenografts in a large number of embryos, allowing the high-throughput screening of drugs, both as single agents and in combinations.
The key point of this innovative approach is that it preserves both cell-cell interactions and the tumor microenvironment, compared to the commonly performed zPDXs with single-cell suspensions. The PDAC from a surgical specimen is immediately put in fresh and cold tumor media, and the tumor is chopped into small fragments. The overall procedure used to generate the zPDX model is based on the direct implantation of a fragment from a patient's tumor into the perivitelline space of a 2 dpf zebrafish embryo.
The proposed model could serve as an alternative to the PDX model based on the digestion of the patient-derived tumor tissue sample, followed by direct injection into the yolk sac or in the perivitelline space. As evidenced by the work by Fior et al., it is possible to establish a zAvatar model from the enzymatic dissection of surgically resected pancreatobilary cancers. The implantation rate ranged between 44% and 64%, due to the typical low cellularity of the tumor pancreatic sample and the unfavorable dense desmoplastic stroma19.
However, the cell suspension only partially reproduced the original tumor14. Conversely, the mPDX, constructed by directly transplanting tumor tissue into immunodeficient mice, best resembles the original tumor, which can better help clinicians to understand tumor behavior and evaluate individual responses prior to treatment20. In fact, the small, engrafted fragment maintains the tumor microenvironment and preserves the crosstalk between malignant and healthy cells, similar to the original tumor21. The chemosensitivity or chemoresistance is thus better reproduced, with a direct impact of zPDX technology in the field of translational oncology14. The zPDX approach enables each patient to have their own tumor developed in a living system. Thus, this represents a good alternative to the use of mPDX, where the tumor tissue is orthografted intact or disaggregated in cells, which are then xenografted into mouse recipient models13.
This protocol was developed in collaboration with pathologists who are responsible for selecting the most suitable tumor pieces for transplantation. Biological tissue samples have always been taken from the surgical specimen and never from a biopsy, as the latter is usually characterized by a scarcity of tissue and is therefore destined for diagnostic purposes only. The criterion for tissue sampling is usually not marginal versus core, but rather a tumor area that is devoid of hemorrhage and necrosis. For this reason, a cryostat histological section was always performed on one half of the sample for immediate quality control of the material. The present study could be limited by the capacity of the primary tumor cells to engraft into the perivitelline space of zebrafish embryos. However, an 80% engraftment rate of viable tissue was observed 3 days post transplantation, both after storage at 4 °C (12/15 cases) and defrosted from -80 °C (10/12 cases), suggesting that the tumor can be effectively cryopreserved, with a cell viability of 74% ± 2% (Supplementary Figure S1). Despite this, the success rate of tissue engraftment varies between different PDAC patient tumors; consequently, the fast processing of the tissue soon after the surgical resection, as well as preventing freezing, can enhance the efficiency.
Some technical expedients are also suggested in the protocol for successful xenografts. For instance, the use of the chopper standardized the dimensions of the xenografted pieces. To check that the pieces of tumor tissues are equal in dimension after the chopping, a potential solution is to use a calibration glass to measure the diameter of the fragments before xenotransplantation.
The use of the 1% agarose cylinder support is essential for both laying the embryos on one side and simply implanting a piece of tumor, as well as to prevent breaking the microneedle. Because tumors exhibit a high variability, another limitation of the method is that the intratumoral heterogeneity is less represented in a fragment compared to a cellular suspension. In fact, small tissue pieces could diverge in terms of benign cell populations and tumor subclones. To overcome this criticism, a high number of injected embryos should be used to recapitulate the tumor heterogeneity and reduce the variability in the analysis. According to the statistical analysis, each group requires a sample size greater than 15, considering an effect size of d = 2, and a power of 80% and 95% statistical significance. This number is reasonable, as an expert operator can process about 40 embryos per hour.
In addition to the methodology used to generate the zPDX model, it was also evidenced here that a FOLFOXIRI combination induces apoptosis of patient-derived cells in the zPDX model. As with FOLFOXIRI, it is possible to test the zPDX chemosensitivity to other chemotherapy schemes, such as those successfully tested by Usai et al.11.
In addition, the technique of whole-mount immunostaining and imaging is presented to provide the quantification of the feature of interest. To overcome some problems and to lead to successful whole-mount immunostaining, a 5% DMSO solution is recommended to permeabilize the tissues. Regarding the confocal imaging, the resolution limitations of the confocal microscope could impair the acquisition of the deeper stacks of the xenografted larvae. However, in the study, the apoptotic cells are counted out of the total number of patient cells. A three-dimensional reconstruction from the base to the top of the implanted tumor mass, with a step size of 5 µm, enables all the human nuclei to be identified, considering the probability that the same nucleus may spread within the previous or next Z-plane. Consequently, the ImageJ multi-point tool counter can be used to control and prevent counting the same cell twice and to run double counting in the same Z-stacks (apoptotic out of total human cells) and can easily trace apoptosis at the single-cell level.
Despite its limitations, the zPDX model has several advantages compared to in vivo models: the extremely rapid life cycle of zebrafish, being able to obtain large numbers of animals in a short time; the optical clarity in the development stages; and, above all, the extremely low ethical impact in the 0-120 h post fertilization time window22. Today, a co-clinical trial using zPDXs is cheaper than that conducted with mouse-based PDXs, since immunodeficient host strains require high cost of maintenance as well as complex imaging methods to monitor tumor growth23. In conclusion, all these factors highlight the potential of the zPDX model for use in co-clinical trials for predicting the chemotherapy sensitivity profiles of cancer patients and for use by researchers interested in novel models for drug screening.
The authors have nothing to disclose.
This work was funded by Fondazione Pisa (project 114/16). The authors would like to thank Raffaele Gaeta from the Histopathology Unit of Azienda Ospedaliera Pisana for the patient sample selection and pathology support. We also thank Alessia Galante for the technical support in the experiments. This article is based upon work from COST Action TRANSPAN, CA21116, supported by COST (European Cooperation in Science and Technology).
5-fluorouracil | Teva Pharma AG | SMP 1532755 | |
48 multiwell plate | Sarstedt | 83 3923 | |
96 multiwell plate | Sarstedt | 82.1581.001 | |
Acetone | Merck | 179124 | |
Agarose powder | Merck | A9539 | |
Amphotericin | Thermo Fisher Scientific | 15290018 | |
Anti-Nuclei Antibody, clone 235-1 | Merck | MAB1281 | 1:200 dilution |
Aquarium net QN6 | Penn-plax | 0-30172-23006-6 | |
BSA | Merck | A9418 | |
CellTrace | Thermo Fisher Scientific | C34567 | |
CellTracker CM-DiI | Thermo Fisher Scientific | C7001 | |
CellTracker Deep Red | Thermo Fisher Scientific | C34565 | |
Cleaved Caspase-3 (Asp175) (5A1E) Rabbit mAb | Cell Signaling Technology | 9661S | 1:250 dilution |
Dimethyl sulfoxide (DMSO) | PanReac AppliChem ITW Reagents | A3672,0250 | |
Dumont #5 forceps | World Precision Instruments | 501985 | |
Folinic acid - Lederfolin | Pfizer | ||
Glass capillaries, 3.5" | Drummond Scientific Company | 3-000-203-G/X | Outer diameter = 1.14 mm. Inner diameter = 0.53 mm. |
Glass vials | VWR International | WHEAW224581 | |
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 647 | Thermo Fisher Scientific | A-21244 | 1:500 dilution |
Goat serum | Thermo Fisher Scientific | 31872 | |
Hoechst 33342 | Thermo Fisher Scientific | H3570 | |
Irinotecan | Hospira | ||
Low Temperature Freezer Vials | VWR International | 479-1220 | |
McIlwain Tissue Chopper | World Precision Instruments | ||
Microplate Mixer | SCILOGEX | 822000049999 | |
Oxaliplatin | Teva | ||
Paraformaldehyde | Merck | P6148-500G | |
PBS | Thermo Fisher Scientific | 14190094 | |
Penicillin-streptomycin | Thermo Fisher Scientific | 15140122 | |
Petri dish 100 mm | Sarstedt | 83 3902500 | |
Petri dish 60 mm | Sarstedt | 83 3901 | |
Plastic Pasteur pipette | Sarstedt | 86.1171.010 | |
Poly-Mount | Tebu-bio | 18606-5 | |
Propidium iodide | Merck | P4170 | |
RPMI-1640 medium | Thermo Fisher Scientific | 11875093 | |
Scalpel blade No 10 Sterile Stainless Steel | VWR International | SWAN3001 | |
Scalpel handle #3 | World Precision Instruments | 500236 | |
Tricaine | Merck | E10521 | |
Triton X-100 | Merck | T8787 | |
Tween 20 | Merck | P9416 | |
Vertical Micropipette Puller | Shutter instrument | P-30 |