Stoletov et al. 10.1073/pnas.0703446104. |
Fig. 5. Human tumor cells with different metastatic potential display different invasive properties when injected in zebrafish. Approximately 200-300 of MDA-435 (CFP, blue) and HT1080 (DsRed, red) cells were mixed at 1:1 ratio and injected into the peritoneal cavity of Tg(fli1:egfp) zebrafish and then imaged at 0 and 48 h after injection. (A-D), Representative examples showing tumors in two independent animals A and B and C and D). (E) Plot showing number of single HT1080 or MDA-435 cells that migrated >20 mm from main tumor cell mass, see Data Analysis and Quantification in SI Text for more details. Mean values are displayed above the plot. Colors: red, HT1080 cells ; blue, MDA-435 cells; green, fish vasculature. N = 10 animals. (Scale bar, 200 mm.)
Fig. 6. Quantification of the VEGF secretion effect on zebrafish vasculature. (A-C) Quantification of VEGF-induced changes in the fish vasculature at the site of injection. Number of branch points (A), vessel diameter (B), and total vessel length (C) were measured for animals that were injected with parental MDA-435 cells (control) and MDA-435 cells that secrete VEGF before (VEGF) or after SU5416 treatment (5 mm, 24 h, VEGFSU). At least 10 animals were analyzed for each condition.
Fig. 7. VEGF-induced blood vessels display irregular vessel wall thickness and increased permeability. Tg(fli1:EGFP) fish were injected with parental or VEGF-secreting MDA-435 tumor cells, and tumor cells were allowed to grow for 4 days. At day 4 after tumor cell implantation, animals were injected with 0.5 ml of dextran-Texas red (5 mg/ml) into the caudal vein and imaged 1 h later. (A) Three-dimensional reconstruction of VEGF-secreting tumor cells and surrounding vasculature 1 h after injection. Tumor cell surfaces were rendered semitransparent to allow visualization of VEGF-induced vasculature. (B) Single optical section of A, only red (dextran) channel is shown. Note leakage of dextran within and in close vicinity of the microtumor area (solid blue line). Also note the absence of the dextran leakage in vessels that are located away from the microtumor area (dashed white squares in A and B). (C) Higher magnification optical section of vessels within the white dashed squares (not affected by VEGF secretion) in A and B. (D) Higher-magnification optical section of vessels within the yellow squares (affected by VEGF secretion) in A and B. Note dextran leakage from VEGF-affected (D) but not -unaffected vessels (C). (E) Plot shows the distribution of 20 random vessel wall thickness measurements for control (nontumor cell-injected) animals, injected with parental MDA-435 cells or with MDA-435 VEGF animals that were treated or untreated with SU5416. Horizontal lines represent mean values displayed above the plot. (F) Plot showing the pixel intensity of red channel (dextran) along the lines (20 per each condition) that were perpendicular to the randomly chosen VEGF-induced (MDA-VEGF), preexisting (control) or blood vessels that are located within the MDA-parental microtumors (MDA-parental). Mean values for each measurement point are displayed above the plot. Representative examples of the line scans can be seen in C and D, dashed lines. See Data Analysis and Quantification in SI Text for more details. Colors: red, dextran Texas-red; blue, MDA VEGF cells; green, fish vasculature. [Scale bars, 200 mm (A and B) and 20 mm (C and D).] N = 10 animals.
Fig. 8. MDA-RhoC cells show increased intravasation in the chick CAM. 2 ´ 106 MDA-435 (CFP, blue) and MDA-RhoC (DsRed, red) cells were mixed at 1:1 ratio, placed onto the chorioallantoic membrane of a 9-day-old chicken embryo, and visualized 10 days later with a confocal microscope. (A and B) Two independent fields showing cells intravasating into blood vessels within the tumor cell mass (arrows). Vessels can be seen as darker channels visualized simultaneously by using transparent light. (C and D) Two independent fields showing cells intravasating into blood vessels within the tumor cell mass after SU5416 treatment (arrows). (E and F) Plots showing number of branch points for untreated (vehicle) or SU5416-treated (SU5416) animals (E) or number of intravasating cells (F) per field for MDA-RhoC or parental cells for untreated (solid columns) or SU5416-treated (dashed columns) animals. Mean values and SEM are displayed above the plots. Colors: red, MDARhoC cells; blue, MDA parental cells. (Scale bars, 100 mm.) N = 10 animals.
To view the movies transfer to desktop and open using Quick Time Movie Player. To play movies S1, S2, S7 click on the "Play" button. To play other movies left click on the movie window, hold and rotate for the best viewing angle.
Movie 1. A z-stack of images representing MDA-435 control cells (38 optical sections, 1-mm thick, 40´).
Movie 2. Higher magnification of SI Movie 1. Invasive mesenchymal cells clustering around a remodeling blood vessel with typical intusseseptive morphology (27 optical sections, 1-mm thick, 60´).
Movie 3. Interactive 3D reconstruction of VEGF-mediated blood vessel remodeling with a representative vessel contacting tumor cell (4 dpi). Image can be rotated 270° horizontally and 45° vertically.
Movie 4. Interactive 3D reconstruction of VEGF-mediated blood vessel remodeling with a vessel contacting tumor cell. Note that this image was obtained at the same area as in SI Movie 3 but 24 h later (5 dpi). Image can be rotated 270° horizontally and 45° vertically.
Movie 5. Interactive 3D reconstruction of MDA-435 cells secreting VEGF (4 dpi). Image can be rotated 90° horizontally and 15° vertically.
Movie 6. Interactive 3D reconstruction of the same MDA-435 cells secreting VEGF same area as in SI Movie 5 but after treatment with 5 mM SU5416 for 24 h (5 dpi). Image can be rotated 90° horizontally and 15° vertically.
Movie 7. A z-stack of images representing MDA-435 RhoC cells (35 optical sections, 1-mm thick, 40´).
Movie 8. Interactive 3D reconstruction of single MDA-435 RhoC cell in contact with the blood vessel wall. Note that even though these cells are in contact with the vessel wall, they fail to protrude membranes inside the blood vessel lumen at the point of contact. Image can be rotated 45° horizontally.
Movie 9. Interactive 3D reconstruction of a single, MDA-435 cell secreting VEGF in contact with the blood vessel wall. Note that the tumor cell membrane integrates into the vessel wall but does not protrude into the vessel lumen. Image can be rotated 45° horizontally.
Movie 10. Interactive 3D reconstruction of a single MDA-435 RhoC cell secreting VEGF in contact with the blood vessel wall. Note that in this case the tumor cell membrane protrudes deep into the vessel lumen. Image can be rotated 45° horizontally.
Table 1. Characteristics of human tumor cell lines used in zebrafish model
Cell line | No. of animals | Median animal survival, days* | Animals with tumor cells,† % | Cell morphology | Invasiveness‡ | Type of angiogenesis |
MDA-435 (human breast adenocarcinoma) | 42 | 13.6 ± 0.45 | 90 | Mixed amoeboid and mesenchymal | + | Intussusceptive |
MDA-435 VEGF | 13 | 13.5 ± 0.73 | 92 | Amoeboid | + | Sprouting |
MDA-RhoC | 11 | 4.0± 0.26 | 100 | Amoeboid with extensive blebbing | +++ | No |
MDA-RhoC VEGF | 10 | 4.3 ± 0.4 | 100 | Amoeboid with extensive blebbing | + | Sprouting |
MDA Src | 3 | At least 5 | 100 | Amoeboid | + | ND |
HT1080 (human fibrosarcoma | 24 | 9.7 ± 0.52 | 25 | Amoeboid | +++ | No |
B16 (human melanoma) | 10 | 13.5 ± 0.74 | 100 | Mesenchymal | + | ND |
CT26 (mouse colon carcinoma) | 10 | At least 5 | 100 | ND | ND | ND |
MEF (mouse embryonic fibroblasts) | 10 | >2 weeks§ | 0 | ND | None | No |
ND, not determined.
*Median time until when tumor cell-injected animals were euthanized because of signs of distress. Animals injected with MDA Src and CT26 cells were sacrificed at 5 dpi.
†
Percentage of the animals that displayed microtumors after 2 days.‡
Invasiveness was measured as low (+) when tumor cells formed tight aggregates or high (+++) when tumor cells scattered throughout the tissue.§
MEF did not survive for >1-2 days when injected into zebrafish.SI Text
Experimental Procedures
Intravasation of Breast Cancer Cells in Chicken Chorioallantoic Membranes (CAMs).
A 3-cm2 square window was created in the eggshell to expose the underlying CAM at 9 dpf. A square, cotton gauze (5 mm) with a central cavity 1 mm in diameter was placed on the CAM in a visibly vascularized area. MDA-435 control (CFP) cells and MDA-RhoC (DsRed) cells (106 of each) were suspended at a 1:1 ratio in 25 ml of Matrigel at 4°C and inoculated into the central cavity of the gauze. The window was sealed, and the eggs remained in the incubator for 10 days, after which time, the tumor-bearing segment of the CAM was harvested and analyzed for the presence of intravasating tumor cells (see also Data Analysis and Quantification below). A tumor cell was considered intravasating if it was protruding >50% into the vessel lumen. For SU5416 treatment 5 ml of SU5416 (5 mM solution in water, diluted from stock DMSO solution) was applied twice a day directly onto the cotton gauze. Control eggs were treated with vehicle only (DMSO).Data Analysis and Quantification.
Images were 3D rendered and analyzed by using Imaris software (Bitplane, www.bitplane.com). All 3D reconstructions were done with the same threshold settings.Quantification of blood vessel contacting, integrating, or intravasating tumor cells.
Our criteria for a vessel contacting cell, integrating cell, and intravasating cell is defined as follows: A cell (red) contacting the vessel wall (green) was defined as no visible space between the red and green signal as seen in single optical section (0.5 mm optical section, 60´ objective). A cell is designated as integrating if it integrates its membrane at least 2 mm into the blood vessel wall (see Fig. 4B for example) but does not protrude membrane beyond the inner side of the vessel wall (0.5-mm optical section, 60´ objective). A cell is designated as intravasating if it protrudes membrane at least 5 mm into the vessel lumen beyond the inner vessel wall (0.5 mm optical section, 60´ objective).(Fig. 3C). Time-lapse analysis.
For time-lapse analyses, »60-mm-thick z-stack series (60´, 1-mm thick) containing 30-50 cells (MDA-control and MDA-RhoC) were acquired for each time point. Representative single cells together with nearby blood vessels were then digitally isolated by using Imaris Contoursurface function, and 3D rendering was performed by using Imaris Isosurface function. Dorsoventral blood vessels were used as a stationary reference points for cell invasion through the body wall.(Fig. 3 D and E). Tumor cell shape quantification.
Images were acquired by using a 60´ objective. Single cells were digitally isolated by using Imaris Contoursurface function. A separate Isosurface was then built for each cell based on Contoursurface-isolated pixels. The cell volume and sphericity were then measured. Twenty randomly chosen MDA-RhoC or MDA-control cells from three animals were analyzed. MDA-RhoC cells had significantly higher sphericity and lower volume (P < 0.05, t test) than MDA-control cells when injected separately or together. A particular cell was considered to be mesenchymal if its sphericity was £0.7. A cell was considered to be rounded if its sphericity was ³0.7. Sphericity ranged from 0 to 1. Cellular blebbing was quantified as the mean number of particles, <2 mm in size (diameter) for each specific tumor cell type (MDA-435 or MDA-RhoC). Cell size (mm, diameter) was quantified at the same microscope settings using single optical sections.Quantification of cells contacting host blood vessels.
Tg(fli1:egfp) fish were injected with MDA-435 parental, and 10 animals (2,238 cells) were analyzed for number of mesenchymal or amoeboid cells that form direct contacts (as judged from single optical sections, 60´, 0.5-mm step or in some cases 40´, 1-mm step, total z-stack size is ~60-80 mm) with the fish vasculature.Quantification of cells integrating into host vessels.
Tg(fli1:egfp) fish were injected with MDA-435 parental or MDA-VEGF cells. Ten animals were analyzed for each case (2,238 and 1,670 cells). Approximately 60- to 80-mm z-stacks were acquired (60´, 0.5-mm step), and single optical sections were analyzed for cells that integrate at least 2 mm into the vessel wall. No cell integration was found in the case of MDA-435 parental cells.(Fig. 4E) Quantification of tumor cell intravasation.
Tg(fli1:egfp) fish were injected with MDA-435 parental, MDA-VEGF, MDA-RhoC, or cells that express both VEGF and RhoC. Numbers of animals used for each condition were 10, 12, 12, and 13, respectively. Approximately 60- to 80-mm z-stacks were acquired (60´, 0.5-mm step), single optical sections were analyzed, and cells that extended 5 mm or greater protrusions are displayed as a percent of intravasating cells of the total cell number within the z-stack. Total cell numbers analyzed were 1,406, 1,766, 2,315, and 1,695, respectively. The inside of the vessel wall and associated tumor cell membranes and protrusions were visualized by using Imaris Isosurface function and digitally magnified/angled to achieve best point of view. Independent sets of animals were used for quantification of tumor cell vessel contacts/integration and intravasation.(SI Fig. 5) Quantification of relative invasive potential of cells coinjected in zebrafish.
Differentially labeled MDA-435 (CFP) and HT1080 (DsRed) human tumor cells were coinjected at 1:1 ratio (~200-300 total) and imaged at 0 and 48 h after injection (10´, 2-mm step). Tumor cells that invaded >20 mm from the outer cell layer of the main tumor cell mass were counted based on 3D reconstructions. Data represents 10 animals analyzed. HT1080 cells were significantly more invasive than MDA-435 cells (P < 0.05, t test).(SI Fig.6) Quantification of the VEGF induced changes in the fish vasculature
. Imaris Filament Tracer function was used for 3D measurements of vessel length, diameter, and determining the number of branch points contained within the 3D area imaged by using a 40´ objective (such as Fig. 2 D or E). Tumor cell size (diameter) was measured for the cells present in the z-stacks by using Imaris Spots function. Numbers of animals injected with MDA-435 cells, MDA-VEGF cells (treated with DMSO) and MDA-VEGF/treated with SU5416 were 14, 12, and 10, respectively.(SI Fig. 7) Quantification of blood vessel wall thickness and dextran leakage.
Twenty random vessel wall measurements were done on 10 animals (two vessels per each animal) for each condition. For quantification of the blood vessel leakage, red channel intensity measurements (561 nm laser line, range 0-4095) were done every 5 mm along the random 25-mm-long lines that were perpendicular to the blood vessels. Original images were acquired by using a 60´ objective in all of the three channels (408/488/561 nm). Three types of vessels were used for measurements: "VEGF" vessels that were contained within the VEGF-secreting tumor cell mass, "Control" vessels from same animals but at least 400 mm away from the tumor cell mass, or "MDA parental" vessels that were contained within non-VEGF-secreting tumor cell mass. All of the measurements were done 4 days after tumor cell injection, 1 h after dextran injection. Ten randomly chosen blood vessels from 10 independent animals for each case were used for analysis (one vessel per animal). Although there was some tissue autofluorescence in the fish gut, skin, eyes, and sensory cells, no significant signal (>20 units) was detected in the fish tissue before dextran injection.(Fig. S4) Quantification of the tumor cell intravasation in the chicken CAMs.
Chick embryos were euthanized at 19 dpf, and CAM areas containing tumor cell mass were surgically removed. Random fields (three per animal) were acquired (10´, 2-mm step) and number (blue, MDA-parental or red, MDA-RhoC) of cells that are protruding >50% of tumor cell body length into the vessel lumen was quantified manually. Chick blood vessels were imaged simultaneously by using a transparent light detector. Vessel branch point number was quantified by using Imaris Filament Traker function. Ten animals were analyzed for each condition. MDA-RhoC cells accounted for a significantly higher number of intravasation events than MDA-parental cells (P < 0.05).Data plots and statistical analysis (t test) were done by using the Prizm software (www.graphpad.com). Data plots show mean values ± SEM except Fig. 3 D and E and SI Fig. 7E, which show distributions of single measurements.