The goal of this work is to develop a method to reproducibly isolate cardiomyocytes from the adult heart and measure DNA content and nucleation.
The adult mammalian heart is composed of various cell types including cardiomyocytes, endothelial cells and fibroblasts. Since it is difficult to reliably identify nuclei of cardiomyocytes on histological sections, many groups rely on isolating viable cardiomyocytes prior to fixation to perform immunostaining. However, these live cardiomyocyte isolation techniques require optimization to maximize the yield, viability and quality of the samples, with inherent fluctuations from sample to sample despite maximum optimization. Here, we report a reproducible protocol, involving fixation prior to enzymatic digestion of the heart, which leads to maximum yield while preserving the in vivo morphology of individual cardiomyocytes. We further developed an automated analysis platform to determine the number of nuclei and DNA content per nucleus for individual cardiomyocytes. After exposing the chest cavity, the heart was arrested in diastole by perfusion with 60 mM KCl in PBS. Next, the heart was fixed in 4% paraformaldehyde (PFA) solution, and then digested with 60 mg/mL collagenase solution. After digestions, cells were singularized by trituration, and the cardiomyocyte fraction was enriched via differential centrifugation. Isolated cardiomyocytes were stained for Troponin T and α-actinin to assess purity of the obtained population. Furthermore, we developed an image analysis platform to determine cardiomyocyte nucleation and ploidy status following DAPI staining. Image based ploidy assessments led to consistent and reproducible results. Thus, with this protocol, it is possible to preserve native morphology of individual cardiomyocytes to allow immunocytochemistry and DNA content analysis while achieving maximum yield.
Heart disease has been the leading cause of death in the majority of western countries for many decades1,2. Although many improvements in the treatment of cardiovascular diseases have improved survival, there are currently no treatments that can replace lost cardiomyocytes. Therefore, studies related to cardiomyocyte function, proliferation, apoptosis and hypertrophy have been and continue to be a major focus of the scientific community. Since the adult mammalian heart has a very limited regenerative capacity, with an estimated cardiomyocyte renewal rate of less than 1% per year, it is crucially important to reliably identify cardiomyocyte proliferative events3,4. Most strategies that measure proliferative events rely either on staining for incorporated DNA nucleotide analogs to assess previous or current proliferation, or stain for nuclear markers of active proliferation5. It is especially important to reliably identify cardiomyocyte proliferative events since the overall number of proliferative cardiomyocytes is so low3,6. For example, based on a 1% renewal rate of endogenous cardiomyocytes per year, one can expect to find between 25 and 50 cardiomyocytes to be proliferative at any given time in the adult mouse heart7,8. Any inaccuracies in identification of cardiomyocyte nuclei might lead to false positive results. Therefore, it is critical to reliably identify cardiomyocyte nuclei, which has proven difficult and unreliable from histological sections9. Identification of cardiomyocytes is much more accurate from single cells than from tissue sections as it might be difficult to distinguish cardiomyocytes from other cell types even when using markers such as α-actinin, although PCM1 might be a reliable marker of cardiomyocyte nuclei in histological sections10.
Current protocols rely on isolating live cardiomyocytes prior to fixation, which is known to cause death of at least 30% of cardiomyocytes, and might lead to inadvertent selection of specific populations of cardiomyocytes11. Furthermore, these protocols are notoriously difficult to optimize to provide reproducible results. Even optimized isolation techniques can typically produce no more than 65% live, rod-shaped cardiomyocytes with varying yields12.
To overcome these issues, we developed a protocol that allows researchers to isolate fixed cardiomyocytes. Since the samples are fixed prior to isolation, the yield is maximized, and in vivo morphology is well preserved. Moreover, with this protocol it is possible to isolate cardiomyocytes from clinical samples, which are typically fixed immediately after procurement. Furthermore, to identify newly generated cardiomyocytes, it is important to measure the nucleation and ploidy status of individual cardiomyocytes, since only diploid cardiomyocytes are typically assumed to be newly formed. Flow cytometry cannot distinguish multinucleation from polyploidy and is a relatively time and resource-intensive protocol. Manual outlining and measurement of nuclei within images is very low-throughput and prone to human bias. Automated quantification of images of fixed, isolated DAPI-stained cardiomyocytes solves both of these problems. Imaging-based determination of nucleation and ploidy distributions can be obtained with a minimum of time and reagents using basic equipment.
All animal experiments were performed conform the National Institutes of Health guidelines and approved by the University of Minnesota Institutional Animal Care and Use Committee (IACUC).
1. Preparation of the solutions and surgical equipment
2. Perfusion and fixation of the heart
3. Isolation of fixed cardiomyocytes
4. Staining cardiomyocytes
5. Setup imaging software
NOTE: Follow along with these steps using Supplementary File 1-SoftwareScreenshots.pdf.
6. Image quantification
7. Data analysis
NOTE: The csv files that are produced can be analyzed manually. Each analyzed image subset produces a triplet of csv files named "nuclei(metadata).csv", "nucleilink(metadata).csv", and "cardiomyocytes(metadata),csv", where (metadata) is replaced with a sequence of name-value pairs of the form "_(name)=(value)", where (name) and (value) are sequences of alphanumeric characters derived from strings matched in the regular expression given earlier. (For example, if row and column were indicated in the filenames then strings like "_row=F" and "_column=8" will be present). The unnamed leftmost column of each nuclei and nucleilink file is a nucleus ID number. The "Min" column of the nucleilink file is the id of the cardiomyocyte that contained said nucleus wholly or 0 otherwise. The "Max" column of the nuclei is the ID of the highest-numbered cardiomyocyte that contained said nucleus in part, or 0 otherwise. The "Mean" column of the cardiomyocytes file is the cardiomyocyte id number.
Cardiomyocytes were isolated according to the protocol described above. Using this method, we typically get uniformly singularized cardiomyocytes that are relatively pure without contaminating non-cardiomyocyte cells (Figure 1A). Cardiomyocytes are easily identified under bright field microscopy due to their characteristic size and birefringence. This technique is easy to implement and provides consistent results from different isolations with comparable cardiomyocyte yields and quality (Figure 1B). Isolated cardiomyocytes can be stored at 4 °C for several weeks before further use.
Cardiomyocytes that were isolated according to the above protocol can be used for various downstream applications, such as measuring cardiomyocyte size, cardiomyocyte ploidy and immunocytochemistry. As a representative result, we show that cardiomyocytes isolated according to this protocol can be stained using antibodies and fluorochrome-conjugated azides for click chemistry to detect localization of specific proteins or to detect cardiomyocyte DNA replication, respectively. For example, we stained cardiomyocytes with antibodies recognizing α-actinin to show the characteristic z-line staining pattern of sarcomeres (Figure 2A). In a separate experiment, we administered the thymidine analog 5-Ethynyl-2'-deoxyuridine (EdU) to mice before isolating fixed cardiomyocytes. After cardiomyocyte isolation, we stained for incorporated EdU using standard protocols13, and were able to detect cardiomyocytes that had undergone S phase in either mononucleated, binucleated and trinucleated cardiomyocytes (Figure 2B).
To further expand the utility of the isolation method, we developed a pipeline that allows quantification of cardiomyocyte ploidy based on integrated DNA staining. To be able to measure ploidy status of cells or nuclei, we needed to segment nuclei and cardiomyocytes. Figure 3 shows a representation of the strategy we used to identify individual nuclei. First, the original image DNA stained image (Figure 3A) is thresholded based on intensity (Figure 3B). Here, we used DAPI to stain for DNA, but any other nuclear dye that shows a linear correlation with DNA content would work. The program allows for any of Fiji's intensity thresholding methods to be chosen, but in this example Otsu's method was used. Nuclear masks that are touching the edge of the image or are smaller than the specified minimum pixel area threshold are excluded. Then, ellipses are fit to the nuclear masks, segmenting individual nuclei. Figure 3C shows these ellipses overlaid on the original image. Next, holes are filled in the masks, and the pixels of the image are then partitioned into territories based on which ellipse they are most proximal to (Figure 3D). The borders of these territories are then used to draw lines through nuclear clusters, finishing the nuclear segmentation process (Figure 3E).
The next step involves detection of cardiomyocytes. For cardiomyocyte images that are obtained based on fluorescently stained cells (Figure 4A), the process is very similar to that for nuclei. The image is thresholded based on an intensity value calculated by the selected thresholding method, in this case the triangle method. Identified cardiomyocyte masks that are touching the boundary of the image or are below a certain size are excluded and holes are filled in the masks to provide properly segmented cardiomyocytes (Figure 4B). Because cardiomyocytes have a more irregular shape than nuclei, no attempt is made to segment cardiomyocyte clusters. Instead, these clusters are excluded based on their high minimum Feret's diameter during the analysis step. Segmentation from bright field images proceeds slightly differently. First, the original bright field image (Figure 4C) is processed with a Sobel edge filter. This filter calculates the absolute value of the gradient of each pixel within the image. Pixels in regions with rapid changes receive high values and pixels in smooth regions of the image receive low values. This edge-filtered image is then thresholded by intensity, using the Triangle method, resulting in masked cardiomyocytes (Figure 4D). These highly irregular masks are then smoothed and linked together via morphological closing using a circle with a radius of 2 pixels, which fills in all white regions in the image where the circle cannot fit without overlapping a black region (Figure 4E). Finally, holes in the masks are filled, regions touching the border are excluded, and small particles are removed, finishing the cardiomyocyte segmentation process (Figure 4F).
Using the outlined segmentation strategy, we can then determine the nucleation status of individual cardiomyocytes. Using this approach, we determined the nucleation status of cardiomyocytes isolated from hearts of outbred CD-1 mice at early postnatal time-points. Hearts of newborn mice (first day of life) showed that the majority of cardiomyocytes at that point are mononucleated (Figure 5: neonatal). This high frequency of mononucleated cardiomyocytes is much lower in juvenile mice (2-week old), where mononucleated cardiomyocytes make up about 25% of the total cardiomyocyte population (Figure 5: juvenile). Finally, we can measure the ploidy status of individual nuclei within cardiomyocytes, and determine whether they are diploid or tetraploid. These results show higher frequency of tetraploid nuclei in adolescent mice (Figure 6).
Figure 1: Efficiency of cardiomyocyte isolation after fixation. (A) Representative image of isolated cardiomyocytes stained with DAPI to show nuclei. (DAPI (blue), Brightfield (gray)) (B) Yield of cardiomyocytes isolated from different mice at 3 months of age. Scale bars = 50 µm. Please click here to view a larger version of this figure.
Figure 2: Immunocytochemistry of isolated cardiomyocytes. (A) Representative image of cardiomyocytes stained for α-actinin (α-actinin (red) and DAPI (blue)). (B) Cardiomyocytes stained for incorporated EdU (red) and DAPI (blue). Representative cardiomyocytes that are mononucleated (left), binucleated (middle) and trinucleated (right) and EdU positive are shown. Scale bars = 50 µm. Please click here to view a larger version of this figure.
Figure 3: Strategy for nuclear segmentation. (A) Original DAPI channel image. (B) Thresholded image (in this example, Otsu's method was used). (C) Masks that were identified from the thresholded images overlaid on the original DAPI stained image. (D) Voronoi tessellation based on nuclear masks. (E) Final segmented nuclei, with split clusters highlighted. Scale bars = 100 µm. Please click here to view a larger version of this figure.
Figure 4: Strategy for cardiomyocyte segmentation. (A) Original fluorescent Troponin I stained cardiomyocyte image. (B) Triangle-thresholded image, after filling holes and excluding small objects and those touching the border. (C) Original bright field cardiomyocyte image (D) Edge-filtered and triangle-thresholded cardiomyocyte image (E) Edge-filtered image after morphological closing with a radius of two pixels (F) Same image after filling holes and excluding small objects and those touching the border. Scale bars = 100 µm. Please click here to view a larger version of this figure.
Figure 5: Classification of cardiomyocytes based on number of nuclei. Neonatal hearts (1 day old) contain more mononucleated cardiomyocytes than juvenile hearts (14 days old). Please click here to view a larger version of this figure.
Figure 6: Distribution of cardiomyocyte DNA content per nucleus. In neonates (left), 13.5% of mononucleated CM nuclei are tetraploid and 11.9% of binucleated CM nuclei are tetraploid. In juveniles (right), 33.9% of mononucleated CM nuclei are tetraploid and 31.2% of binucleated CM nuclei are tetraploid. Please click here to view a larger version of this figure.
Supplementary File 1: Software Screenshots. Please click here to view this file (Right click to download).
Supplementary File 2: AnalyzeNucleation.py. Please click here to view this file (Right click to download).
Supplementary File 3: AnalyzeMultinucleatedServer.R. Please click here to view this file (Right click to download).
Since cardiomyocytes cannot be maintained in culture, it is important to isolate primary cardiomyocytes to be able to study their architecture and function11. Hence, cardiomyocyte isolation techniques have been widely used in the cardiac field. If the goal is to determine functional aspects of cardiomyocytes, it is important to isolate viable cardiomyocytes. These live cardiomyocytes can also be used to perform immunostaining on isolated cardiomyocytes. However, optimizing the technique of isolating live cardiomyocytes is technically challenging, and even the best techniques typically only yield 60-65% live rod-shaped cardiomyocytes, and the remaining cardiomyocytes are all balled up and dying or dead11,12. Here, we developed a technique that will allow researchers to first fix the heart, and then isolate cardiomyocytes efficiently. This new protocol allows for much higher yields of rod-shaped cardiomyocytes compared to previously published protocols. Furthermore, we developed an imaging analysis platform to categorize cardiomyocytes automatically based on nucleation and ploidy. With these new methodologies, groups can stain cardiomyocytes for different proteins, and study cardiomyocyte ploidy and nucleation status as surrogates for the regenerative potential of the heart.
The protocol described here is relatively straightforward, and can be performed without any advanced equipment. The amount of collagenase and incubation time for digestion might vary depending on the collagenase lot, and the company providing it. We used collagenase type 2, since this is most widely used to digest the heart for obtaining live cardiomyocytes. Based on our observations, we determined that overnight incubation with 60 mg/mL collagenase type 2 is optimal for almost all mouse hearts regardless of the level of fibrosis. We have never had an issue of overdigestion as intracellular proteins are fixed and not as accessible as extracellular collagen. However, if the heart is not digested properly, more vigorous trituration might be needed, which causes cell fragmentation due to shear stress. Thus, it is crucial to make sure that the heart is digested properly before moving on to trituration. Stiffness of the heart can be tested by squeezing with forceps to assess the degree of digestion. Following incubation with collagenase, hearts should be less stiff and easy to tear apart. Other types of collagenase can also be used. A previous report used a combination of collagenases B and D14.
Furthermore, we believe that this protocol can be used to assess overall number of cardiomyocytes in the heart15. However, if the goal is to obtain and quantify all cardiomyocytes from the heart, it is important to incubate the hearts for extended periods of time in the collagenase solution (e.g., 3-7 days), where the collagenase solution should be replenished once a day. This will minimize inconsistencies in isolation efficiency by eliminating the impact of the degree of trituration on cardiomyocyte yield.
The use of DNA content to measure ploidy is not new, and has been used in flow cytometry for decades. Recently, it was shown that microscopy can similarly be used to estimate DNA content per nucleus16. Here, we implemented this strategy to measure ploidy of cardiomyocyte nuclei, as a surrogate for newly formed cardiomyocytes. The dogma in the field of cardiac regeneration is that only mononucleated, diploid cardiomyocytes can undergo cytokinesis and give rise to new cardiomyocytes. Since it is very challenging to measure new cardiomyocyte formation in vivo, isolating cardiomyocytes that have been chased after administration of a DNA nucleotide analog and determining the level of mononucleated, diploid cardiomyocytes has been used as an approximation of the ability of the heart to generate new cardiomyocytes17. Here, we provide a macro for ImageJ that allows easy quantification of cardiomyocyte ploidy. At the very minimum, 500 nuclei must be measured to attain an accurate estimate of the location of the G1 peak. If care is taken to ensure that staining and imaging conditions are consistent across every well of the imaged plate, only 500 nuclei across the entire sample need to be imaged, otherwise, there need to be 500 nuclei per image group18,19. Limitations of imaging-based measurement of nucleation and ploidy include difficulty to distinguish nuclei from adherent cells from actual cardiomyocyte nuclei, when using two-dimensional images. Such adherent cells might result in overestimation of the quantity of multinucleated cells and decrease the accuracy of measurements of the tetraploid cardiomyocyte nucleus population. One possible strategy to solve this problem would be to use the cardiomyocyte nuclear marker PCM16,20. However, we have had difficulties to obtain reliable PCM1 staining on properly fixed cells or tissues.
Another potential limitation is that some nuclear stain images might have significant background cytoplasmic staining, preventing proper thresholding using Fiji's built in methods without extensive preprocessing. In addition, the irregular contribution of this background fluorescence into ploidy estimates reduces their accuracy. Moreover, if the cells are not left in DNA-staining solution for sufficient time, the fluorescent dye will not bind to saturation within the nuclei and the assumption of a linear relationship between nuclear integrated intensity and DNA content will no longer be accurate.
It should be noted that the software cannot segment cardiomyocyte clusters and instead removes them from analysis. Therefore, it is critically important to seed cardiomyocytes at a relatively low density (e.g., 1000 cells/cm2). Further, the software cannot distinguish between two cardiomyocytes lined up end-to-end and long, singular cardiomyocytes. These sorts of clusters might erroneously inflate multinucleation estimates.
Although the described method does not allow for obtaining viable cardiomyocytes and thus cannot be used to measure dynamic cellular processes, if the goal is to perform immunostaining, we believe that the described method is superior to existing protocols with higher yields of cardiomyocytes and better quality in terms of morphology and protein localization. Finally, the described method could be used to isolate cardiomyocytes from clinical samples14,21. We believe the described methodology can help different researchers to obtain high-quality cardiomyocytes and measure nucleation and ploidy as surrogates for new cardiomyocyte formation.
The authors have nothing to disclose.
JHvB is supported by grants from the NIH, Regenerative Medicine Minnesota, and an individual Biomedical Research Award from The Hartwell Foundation.
96 wells plate for imaging | Corning | 3340 | We use these plates as they are suitable for imaging, although glass bottom plates would be better for confocal imaging |
Alpha actinin | Novus Biologicals | NBP1-32462 | This antibody is used as a marker of cardiomyocyte sarcomeres |
Blunt scissors | Fine Scissor Tools | 14072-10 | We prefer blunt scissors as the possibility of tearing heart tissue is lower when exposing the heart |
C57BL/6J | The Jackson Laboratory | 664 | Used for imaging, assessing ploidy and nucleation in cardiomyocyte population |
CD-1 mice | Charles river | 22 | Used for imaging, assessing ploidy and nucleation in cardiomyocyte population |
Collagenase 2 | Worthington | LS004177 | For the purpose of this protocol, the batch to batch differences are minimal and don't affect overall yield and quality of the isolation |
Copper (II) sulfate pentahydrate | Sigma-Aldrich | 203165-10G | For edu staining |
Cy5 Picolyl Azide | Click Chemistry Tools | 1177-25 | Azide used for edu staining |
Cytation3 | BioTek | – | Used for automated imaging for DNA analysis |
DAPI | Life Technologies | D3571 | DAPI used for DNA staining. Stocks were dissolved in distilled water. |
donkey anti-mouse IgG-Alexa568 | Life Technologies | A10037 | Secondary antibody used to detect alpha actinin staining within cardiomyocytes |
Forceps | ROBOZ | RS-5137 | We use these curved, blunt forceps, although straight forceps could also be used |
Hydrochloric acid | Fisher Scientific | A144212 | To set pH of Tris-HCl buffer to pH 8.5 |
ImageJ | imagej.net/Fiji/Downloads | – | Used for analyzing images |
L-ascorbic acid | Sigma-Aldrich | 255564-100G | For edu staining |
Needle for infusion | TERUMO | SV*23BLK | We use winged infusion sets throughout the protocol as it is easy to manipulate the position of the needle with these sets during injection |
Nikon A1R HD25 | Nikon | – | Used to take confocal images of alpha actinin staining |
Nylon mesh 200 micron | Elko filtering | 03-200/54 | Mesh used for filtering regular cardiomyocytes (not hypertrophied) |
Nylon mesh 400 micron | Elko filtering | 06-400/38 | Mesh used for filtering hypertrophied adult cardiomyocytes |
Phosphate Buffered Saline (1X) | Corning | 21-040-CV | This can also be prepared in the lab. Although sterility is important in this experiment, we think it is sufficient to prepare PBS and filtering it |
Potassium chloride, Granular | Mallinckrodt | 6858 | Granular potassium chloride was preffered by us as it forms less aggregates when stored in room temperature |
R | r-project.org | – | Used for data analysis of the measurements obtained from images |
Tris Base | Fisher Scientific | BP152-5 | Used to buffer EdU staining reaction |