We describe a more consistent and expeditious method to quantify lung metastasis in the 4T1 breast cancer model by using Fiji-ImageJ.
Breast cancer is a devastating malignancy, accounting for 40,000 female deaths and 30% of new female cancer diagnoses in the United States in 2019 alone. The leading cause of breast cancer related deaths is the metastatic burden. Therefore, preclinical models for breast cancer need to analyze metastatic burden to be clinically relevant. The 4T1 breast cancer model provides a spontaneously-metastasizing, quantifiable mouse model for stage IV human breast cancer. However, most 4T1 protocols quantify the metastatic burden by manually counting stained colonies on tissue culture plates. While this is sufficient for tissues with lower metastatic burden, human error in manual counting causes inconsistent and variable results when plates are confluent and difficult to count. This method offers a computer-based solution to human counting error. Here, we evaluate the protocol using the lung, a highly metastatic tissue in the 4T1 model. Images of methylene blue-stained plates are acquired and uploaded for analysis in Fiji-ImageJ. Fiji-ImageJ then determines the percentage of the selected area of the image that is blue, representing the percentage of the plate with metastatic burden. This computer-based approach offers more consistent and expeditious results than manual counting or histopathological evaluation for highly metastatic tissues. The consistency of Fiji-ImageJ results depends on the quality of the image. Slight variations in results between images can occur, thus it is recommended that multiple images are taken and results averaged. Despite its minimal limitations, this method is an improvement to quantifying metastatic burden in the lung by offering consistent and rapid results.
One in eight women will be diagnosed with breast cancer in her lifetime, and yet despite multiple treatment options breast cancer is the second leading cause of cancer-related deaths in American women1. These women are not dying from the primary tumor in their breast. Instead, the metastatic burden is responsible for the mortality of this disease as it commonly spreads to the lung, bone, brain, liver, and lymph nodes2. Because of this, breast cancer models need to evaluate metastasis to contribute to curbing the mortality of this disease. The 4T1 murine breast cancer model is a superb protocol to accomplish this. The method described here offers an improvement to the 4T1 model by using Fiji-ImageJ to quantify lung metastasis, producing consistent and expeditious results.
The 4T1 model is well-established, with most labs using protocols such as those described by Pulaski and Ostrand-Rosenberg in 20013. The 4T1 cell line is 6-Thioguanine (6TG) resistant and representative of stage IV, triple negative breast cancer3,4,5. It is clinically relevant as it is an orthotopic model and spontaneously metastasizes to the same organs as in human breast cancer3,4. The 4T1 cells spontaneously metastasize at a predictable rate based on the quantity of cells injected3,4. Importantly, genetic differences between mice used here caused expected inter-individual variability in metastatic burden. To evaluate metastasis, tissues are harvested to collect and quantify cancer cells in distant sites using 6TG selection and methylene blue staining. The result is a collection of tissue culture plates with blue dots representing metastatic colonies. However, the Pulaski and Ostrand-Rosenberg protocol quantifies metastatic colonies by manually counting them, and therefore this has been the standard means of evaluating metastasis in this model. While this is easy for tissues with low metastatic burden, tissues like the lungs are often laden with metastases. As lung plates can be highly confluent, accurately and precisely quantifying metastatic colonies by manual counting is difficult and prone to human error. To better quantify metastatic burden, we describe using Fiji-ImageJ for a computer-based solution to human counting error. Histopathological analysis with hematoxylin and eosin (H&E) staining is another means to quantify lung metastases, and interestingly has also been improved with Fiji-ImageJ software6,7. However, because histopathological analysis observes a single slice of the lung, it can be inaccurate and unrepresentative. This is because the 4T1 model causes several metastatic lesions throughout the organ that are not evenly distributed. While overall trends between histopathological analysis and manual counting can be similar8, individual values can differ and therefore histopathological analysis should not be used as the sole means of quantification. We demonstrate the benefit compared to histopathological analysis and the inconsistencies in manual counting between different counters, while also demonstrating the consistency of using Fiji-ImageJ. Additionally, we show that this method can reduce the incubation time from 10-14 days to 5 days, meaning researchers can analyze data from their study much sooner than when relying on manual counting.
This method is a collection of simple adjustments to the Pulaski and Ostrand-Rosenberg protocol3. Because the 4T1 model is widely used, and because lung metastasis is a critical parameter to measure in preclinical models, we believe this method can be widely used and is highly valuable to breast cancer researchers. The only additional supplies needed are a camera and access to a computer with Fiji-ImageJ, a free software used frequently in image analysis9. This method specifically focuses on lung metastasis, but it could be used for other tissues with significant metastatic burden.
All methods described here have been approved by the Institutional Animal Care and Use Committee (IACUC) of Virginia Tech and in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Performing this protocol requires permission from the appropriate institutions and adherence to all appropriate guidelines.
1. Cell Culture
2. Injections
3. Monitoring
4. Necropsy
5. Processing Tissues
NOTE: All steps in this section should be done using sterile technique.
6. Staining plates
7. Image analysis
This method contains simple adjustments from the Pulaski and Ostrand-Rosenberg 4T1 protocol3 and can be visualized in Figure 1. When 3 separate researchers manually counted metastatic colonies for 12 lung plates (1:10 dilution), the results were very inconsistent between different counters (Figure 2A). All researchers were directed to “count the metastatic colonies that appear as blue dots”, yet the inconsistencies demonstrate the issue with manually counting highly-metastatic plates. The researchers had varying levels of experience with the 4T1 model. A board-certified veterinary pathologist analyzed H&E stained lung slides for metastasis as another method to compare to Fiji-ImageJ lung plate analysis (Figure 2B).
Using the Fiji-ImageJ analysis, 3 separate researchers analyzed 3 separate images of the collection of 12 plates (1:2 dilution). Images were taken in two separate lab spaces with slightly different lighting. The arrangement of the plates or the angle from which the picture was taken were different between each image. In contrast to the manual counting results, the Fiji-ImageJ results were consistent between counters for each of the 3 images (Figure 3A). To determine if there were inconsistencies between the 3 images, the results from the 3 images and the 3 counters were combined per lung plate (Figure 3B). There are differences between images for some plates, but the overall trends are similar and it offers more consistency than manual counting. To account for the variations between the 3 different images, results from each image were averaged for each plate (Figure 3C). These averages provided consistent results between counters that accurately and precisely analyze metastatic burden. Therefore, this protocol suggests taking at least 3 images of the plate collection in different arrangements, from different angles, or in slightly different light settings, and then analyzing and averaging the results. The contrast between manual counting and Fiji-ImageJ analysis is visualized when comparing Figure 2A 세스 Figure 3C.
Another way to demonstrate the improvements offered by this protocol is comparing the ranking of the plates from most to least metastatic burden between counters, based on the counts from Figure 2 and Figure 3. Manual counting agreed on the most confluent plate, but all following ranks were inconsistent between counters (Figure 4A). Contrastingly, the ranks from Fiji-ImageJ analysis for each image were much more consistent between counters (Figure 4B). The consistency is also seen when results from each image for each plate were averaged (Figure 4C). We acknowledge that this protocol does not offer complete consistency between counters, but it is an improvement from manual counting when comparing Figure 4A 세스 Figure 4C. Histopathological analysis differed from both manual and Fiji-ImageJ counting (Figure 4D).
To demonstrate the importance of avoiding reflections in the images, an image with a reflection of a hand and its subsequent Fiji-ImageJ analysis is shown (left) opposed to the same plate without a reflection (right) (Figure 5A). Other dark blemishes from a dirty background surface or blood sample residue on the plates can negatively impact Fiji-ImageJ analysis too. The blood plate in Figure 5B only has 2 metastatic colonies (noted by white arrows), but the dark residue (noted by black arrows) caused Fiji-ImageJ to consider it as 31.6% metastatic. Therefore, it is important to have a clean, light surface and to not use this method for blood samples as blood samples will typically leave residual dark spots on the plate that are not metastatic colonies.
Figure 1: Protocol Schematic. This protocol focuses solely on analyzing lung metastasis in the 4T1 model. The general flow of this protocol includes growing 4T1 cells in culture, injecting BALB/c female mice with 4T1 cells in the left abdominal mammary fat pad, monitoring mice according to IACUC and institutional protocols, sacrificing mice and collecting the lung, collecting cells from the lung samples, plating and incubating cells in 6TG selection media, fixing and staining cells after 5 days, taking pictures of the plates, and analyzing using Fiji-ImageJ. Please click here to view a larger version of this figure.
Figure 2: Manually counting metastatic cells and histopathological analysis have inconsistent results. A. 12 lung plates with a 1:10 dilution were manually counted by 3 separate researchers instructed to count metastatic colonies the same way, although experience with the model varied between researchers. The number of metastatic colonies counted varied greatly between researchers. B. Histopathological analysis identified and quantified individual tumor cell aggregates, classified as metastases, present in H&E stained lung slides. High, medium, and low magnification images of one representative slide are shown. Please click here to view a larger version of this figure.
Figure 3: Fiji-ImageJ analysis is accurate and precise in determining metastatic burden. A. 12 lung plates with a 1:2 dilution were analyzed by 3 separate researchers in 3 separate images of the 12 lung plates. B. Results from each of the 3 images by each of the 3 researchers were combined. C. Results from each lung plate from the 3 images were averaged. One-way ANOVA with Tukey’s multiple comparison test determined no significant differences between counters for each lung plate. Data are shown as mean + SD. Please click here to view a larger version of this figure.
Figure 4: Fiji-ImageJ analysis provides more consistent ranking of metastatic burden compared to manual counting and histopathological analysis. A. The same lung plates from Figure 2 were ranked from most to least metastatic based on the manual counts from Figure 2. B. The same 12 lung plates from Figure 3 were ranked from most to least metastatic based on the Fiji-ImageJ analysis from Figure 3A. C. The averages from Figure 3C were ranked from most to least metastatic. D. Lung slides were ranked from most to least metastatic based on histopathological evaluation. Please click here to view a larger version of this figure.
Figure 5: Reflections and non-metastatic dark spots will negatively impact results. A. An image with a reflection of a hand taking the picture disrupts the Fiji-Image J analysis, as shown in comparing the reflection Fiji-ImageJ analysis (left) to the correct Fiji-ImageJ analysis (right) B. Blood plates often leave leftover stains (black arrows) on the plates that are not metastatic colonies (white arrows). Please click here to view a larger version of this figure.
As demonstrated, manually counting the metastatic colonies on each lung plate can be an inaccurate and imprecise method to quantify lung metastasis, demonstrating the need for a better means of quantification (Figure 2). Histopathological analysis differed slightly from both manual counting and Fiji-ImageJ analysis (Figure 2B and 4D), likely because the H&E slides are not a representative sample of the entire organ. The protocol harvests an entire lung, and therefore is more representative of total lung metastasis, and is more consistent than manual counting. Several different approaches to Fiji-ImageJ analysis were attempted and are discussed below, but the protocol outlined above appears to be the superior method.
Lung, blood, and brain samples were collected for this study. However, the blood and brain samples had very few metastatic colonies, if any at all. We determined that manually counting the metastatic colonies is optimal for these less-metastatic tissues, and therefore blood and brain data were not included. When the metastatic burden is easy to manually count (e.g., ten or twenty metastatic colonies as opposed to thousands), the original issue of human error is not relevant, and therefore this protocol is not needed. Also, blood samples can leave dark spots on the plates after fixation, which interferes with the Fiji-ImageJ analysis (Figure 5). Importantly, the quantity of cells injected can influence the metastatic burden. For instance, if fewer cells are injected and the mice can survive longer, the cancer has more time to spread to the traditionally less-metastatic sites like the brain3,4. Therefore, this protocol could be modified to include the metastatic burden of other tissues if they are given time to become highly-metastatic. If trying the 4T1 model for the first time or changing the quantity of cells injected, we recommend trying at least two dilutions when plating cells. For this study, we used a 1:2 and 1:10 dilution. The 1:2 dilution would have been difficult to count manually, but was counted easily in Fiji-ImageJ. The 1:10 dilution was still difficult to count manually and therefore led to inconsistent results. Dilutions can be modified based on the specific study parameters.
Pictures were taken of individual lung plates and the 12 lung plates together. Individual plates were analyzed in two ways: either cropping the image to a central square of the plate prior to uploading to Fiji-ImageJ, or using the circle selection tool in Fiji-ImageJ to select the central circle of the plate in the uncropped image. We found that using the circle selection tool in Fiji-ImageJ offered the easiest, most consistent way to create a same-sized area for analysis for all plates. Furthermore, analyzing the entire collection of lung plates in the same image was superior to analyzing individual images of single lung plates. Having all of the lung plates in the same image allows for the same-sized circle to be used easily between the lung plates. It ensures all lung plates are the same distance from the camera and therefore the same-sized circle for analysis should be the correct size for all lung plates in the image. It also makes analysis quicker as redrawing the circle is not necessary between plates. It is simply dragged to the next plate in the image without changing its size, which guarantees the same size is used for all plates in the picture. When selecting the size of the circle, it is important to make it large enough to analyze the majority of the plate while small enough to avoid the background noise from the edges of the plate. Furthermore, in an attempt to save reagents, cells were also plated in 6 well plates and compared to the 10 cm tissue culture plates. The Fiji-ImageJ results from the 6 well plates were less consistent and did not correlate to the 10 cm dishes (data not shown). One explanation is the smaller surface area provides a smaller area to analyze, leading to less representative data. Another is that reducing the surface area allows the cells to grow more quickly as they are closer to other surviving cells. Therefore, we do not recommend using any tissue culture reagents other than what we have described in the protocol.
As mentioned before, avoiding reflections and having a clean, light background are absolutely critical to this method. Figure 5A demonstrates how a reflection is analyzed in Fiji-ImageJ and therefore shows the critical importance of avoiding reflections. As tissue culture plates are highly reflective, it is beneficial to take the picture at a slight angle to avoid reflections from either yourself taking the picture or from the light sources above. The lighting conditions of the specific work area will need to be accounted for. We suggest taking multiple pictures of the plates to be analyzed, trying slightly different arrangements and/or angles, in a well-lit area. Study the pictures intensely for any reflections. If there are inconsistencies in the analysis, it is likely due to a picture quality issue. To troubleshoot, compare the normal picture to the black and white picture. If areas that are not blue in the normal picture are appearing as white in the black and white picture, there is likely a reflection or blemish that is altering the results.
In addition to consistency, another notable benefit of this method is that it produces data much more quickly than manual counting. Manually counting multiple plates is very time-consuming, while Fiji-ImageJ analysis can be done quickly. It also allows for a shorter incubation time. Pulaski and Ostrand-Rosenberg recommend a 10-14 day incubation period for the plated cells, adding a substantial amount of time to the study3. The 10-14 day incubation period allows for larger, easier-to-count colonies to form. However, many lung plates can become confluent before then. Instead, 5 days of incubation gives enough time for the 6TG selection to kill non-cancerous cells (proven by healthy control mice not having any colonies on their lung plates, data not shown), and for the cells to grow enough to be easily quantified with Fiji-ImageJ. This significantly decreases the time between the mice being sacrificed and analyzing essential metastatic data.
To conclude, the benefits of this method far outweigh the limitations. We acknowledge this method does not offer perfect consistency. While this is not the ideal method for less-metastatic tissues, those tissues can easily be counted manually. While getting a picture without reflections can require some careful photography, the consistency gained with this method is significant. It is possible that this method could be used for other tissues that are highly-metastatic and other protocols that require counting stained objects. The study design could also allow for analyzing the rate of metastasis or effect of anti-cancer treatments on metastasis. This method will provide highly-consistent, reliable metastasis data and represents a significant refinement to the 4T1 model. The application of this model to upcoming breast cancer metastasis research is of utmost importance in arming researchers with tools to battle against breast cancer mortality.
The authors have nothing to disclose.
This work was supported by the Virginia-Maryland College of Veterinary Medicine (IA), the Virginia Tech Institute for Critical Technology and Applied Science Center for Engineered Health (IA), and National Institutes of Health R21EB028429 (IA).
Anesthesia chamber | See comments | See comments | Use approved materials in your institution's policies |
Anesthetic agent | See comments | See comments | Use approved materials in your institution's policies |
BALB/c Female Mice | The Jackson Laboratory | 651 | |
Blunt scissors | Roboz | RS-6700 | |
Calculator | Any | Any | |
Camera | Any | Any | Minimum of 8 megapixels |
Centrifuge | Any | Any | Needs to be capable of 125 x g and 300 x g |
CO2 euthanasia setup | See comments | See comments | Use approved materials in your institution's policies |
Cold room, refrigerator, cold storage | Any | Any | |
Computer with Fiji-ImageJ | Any | Any | Needs to be capable of running Fiji-ImageJ |
Counting Chamber | Fisher Scientific | 02-671-10 | |
Curved scissors | Roboz | RS-5859 | |
Distilled water | Any | Any | |
Elastase | MP Biomedicals | 100617 | |
Electronic scale | Any | Any | |
Fetal Bovine Serum (FBS) | R&D Systems | S11150 | |
Forceps | Roboz | RS-8100 | |
Ice | N/A | N/A | |
Incubator | See comments | See comments | Needs to be capable of 5% CO2 and 37 °C |
Methanol | Fisher Scientific | A412SK-4 | |
Methylene blue | Sigma-Aldrich | 03978-250ML | |
Penicillin Streptomycin | ATCC | 30-2300 | |
Pins or needles | Any | Any | For pinning down mice during necropsy |
Plastic calipers | VWR | 25729-670 | |
RMPI-1640 Medium | ATCC | 30-2001 | |
Rocker or rotating wheel | Any | Any | |
Sharp scissors | Roboz | RS-6702 | |
Sterile disposable filter with PES membrane | ThermoFisher Scientific | 568-0010 | |
T-150 Flasks | Fisher Scientific | 08-772-48 | |
T-25 Flasks | Fisher Scientific | 10-126-10 | |
T-75 Flasks | Fisher Scientific | 13-680-65 | |
Tri-cornered plastic beaker | Fisher Scientific | 14-955-111F | Used to weigh mice |
Trypan blue | VWR | 97063-702 | |
Trypsin-EDTA | ATCC | 30-2101 | |
Type IV collagenase | Sigma-Aldrich | C5138 | |
1 cm tissue culture plates | Nunclon | 153066 | |
1 mL syringe | BD | 309659 | |
1.7 mL microcentrifuge tubes | VWR | 87003-294 | |
10 cm tissue culture plates | Fisher Scientific | 08-772-22 | |
12 well plate | Corning | 3512 | |
15 mL centrifuge tube | Fisher Scientific | 14-959-70C | |
1X Dulbecco's Phostphate Buffered Saline (DPBS) | Fisher Scientific | SH30028FS | |
1X Hank’s Balanced Saline Solution (HBSS) | Thermo Scientific | SH3026802 | |
27 g 1/2 in needles | Fisher Scientific | 14-826-48 | |
4T1 (ATCC® CRL2539™) | ATCC | CRL-2539 | |
50 mL centrifuge tube | Fisher Scientific | 14-959-49A | |
6-Thioguanine | Sigma-Aldrich | A4882 | |
70 μM cell strainer | Fisher Scientific | 22-363-548 | |
70% ethanol | Sigma Aldrich | E7023 | Dilute to 70% with DI water |