Cancer Associated Fibroblasts (CAFs) facilitate tumor initiation, growth and progression through signaling that promotes proliferation, angiogenesis, and inflammation. Here we describe a method to isolate pure populations of normal fibroblasts and CAFs from fresh mouse and human tissues by cell sorting, using PDGFRα as a surface marker.
Cancer-associated fibroblasts (CAFs) are the most prominent cell type within the tumor stroma of many cancers, in particular breast carcinoma, and their prominent presence is often associated with poor prognosis1,2. CAFs are an activated subpopulation of stromal fibroblasts, many of which express the myofibroblast marker α-SMA3. CAFs originate from local tissue fibroblasts as well as from bone marrow-derived cells recruited into the developing tumor and adopt a CAF phenotype under the influence of the tumor microenvironment4. CAFs were shown to facilitate tumor initiation, growth and progression through signaling that promotes tumor cell proliferation, angiogenesis, and invasion5-8. We demonstrated that CAFs enhance tumor growth by mediating tumor-promoting inflammation, starting at the earliest pre-neoplastic stages9. Despite increasing evidence of the key role CAFs play in facilitating tumor growth, studying CAFs has been an on-going challenge due to the lack of CAF-specific markers and the vast heterogeneity of these cells, with many subtypes co-existing in the tumor microenvironment10. Moreover, studying fibroblasts in vitro is hindered by the fact that their gene expression profile is often altered in tissue culture11,12 . To address this problem and to allow unbiased gene expression profiling of fibroblasts from fresh mouse and human tissues, we developed a method based on previous protocols for Fluorescence-Activated Cell Sorting (FACS)13,14. Our approach relies on utilizing PDGFRα as a surface marker to isolate fibroblasts from fresh mouse and human tissue. PDGFRα is abundantly expressed by both normal fibroblasts and CAFs9,15 . This method allows isolation of pure populations of normal fibroblasts and CAFs, including, but not restricted to α-SMA+ activated myofibroblasts. Isolated fibroblasts can then be used for characterization and comparison of the evolution of gene expression that occurs in CAFs during tumorigenesis. Indeed, we and others reported expression profiling of fibroblasts isolated by cell sorting16. This protocol was successfully performed to isolate and profile highly enriched populations of fibroblasts from skin, mammary, pancreas and lung tissues. Moreover, our method also allows culturing of sorted cells, in order to perform functional experiments and to avoid contamination by tumor cells, which is often a big obstacle when trying to culture CAFs.
1. Dissecting Mammary or Skin Tissue from Mice
2. Preparation of Mammary Gland/skin Single Cell Suspension
Note: optimal incubation times may vary if digesting other tissues.
3. Red Blood Cells Lysis
Note: this step is not necessary if mice were heart-perfused with PBS before they were sacrificed.
4. Blocking of Endogenous Fc
5. Staining
Note: although PDGFRα is a robust marker for fibroblasts, it is recommended to include antibodies for immune cells and for epithelial cells in order to exclude any double positive populations and thus enhance the purity of isolated fibroblasts.
Optional: Cells may be resuspended in FACS buffer II for the duration of sorting. This may improve cell viability. However, exposure to factors in serum may have an effect gene expression, which should be tested, and taken into account.
Note: To avoid wasting cells, you can add 7AAD/PI to the un-stained sample after recording in the FACS, and use again as 7AAD-only control.
6. Isolation of Cells by FACS
(Note: this part of the protocol requires previous knowledge in operating a FACS sorter, e.g. BD FACS AriaII, or the assistance of a skilled technician).
Note: If sorting a large number of cells, it is not recommended to sort directly into RNA lysis buffer, as the large volume of sheath fluid accompanying the cells will dilute the lysis buffer and reduce the yield.
Using PDGFRα as a marker for fibroblasts results in isolation of highly enriched populations of tissue fibroblasts. The level of purity after sorting was 99%, as quantified by post-sort analysis (Figure 2A). Estimating the percentage of contaminating non-fibroblast cells by the relative expression of cell-specific control genes (Figure 2B) typically shows 0.1-0.6% contamination. This level of purity allows high quality transcriptome profiling of isolated fibroblasts9.
In mammary glands, the percentage of fibroblasts in the tissue varies between 5-20% in normal mammary tissue, and 1-5% in MMTV-PyMT tumors. The relative percentage of fibroblasts in tumor tissue is lower than in pre-neoplastic tissue, despite increase in the total number of fibroblasts, due to the massive expansion of the epithelial compartment. The reduced % of fibroblasts isolated from tumor tissue is also a result of the added technical difficulty and decreased efficiency of cell sorting from a highly necrotic tissue.
Isolated fibroblasts can be cultured directly after sorting, but may require a few days to recover (Figure 2D). It is essential to perform all further experiments with low passage cells, as primary fibroblasts undergo senescence during propagation in culture.
Figure 1. Purification of fibroblasts from fresh mammary glands. A- FVB/n /PyMT mouse. B- Exposed mammary glands. C- Fresh mammary glands were dissected from FVB/n /PyMT mice and digested with collagenase to a single cell suspension. Cells were immune-labeled with antibodies for fibroblast surface marker (PDGFRα) and macrophages surface marker (F4/80) followed by separation by FACS. Sorted fibroblasts were either cultured or lysed for RNA purification. Click here to view larger figure.
Figure 2. Profiling of sorted fibroblasts. A- Single cell suspensions of mouse mammary glands were stained with PDGFRαand F4/80 antibodies. FACS sorting plot is shown (left panel). P2 is the fibroblasts gate (PDGFRα+ cells) and P4 is the macrophages gate (F4/80+ cells). A post sort analysis was performed to determine the purity of sorted fibroblasts and macrophages (middle and right panels). B- Analysis of sorting purity by qRT-PCR of cell-specific control genes for fibroblast (PDGFRα, Col-1α), immune cells (CD45, CSF-1R) and epithelial cells (E-cadherin). Results were normalized to two house keeping genes (GAPDH and mGUS). Relative expression to GAPDH is shown. C- Quantification of PDGFRα expression in total unsorted population of mammary fibroblasts. D- Tissue culture of sorted fibroblasts. Click here to view larger figure.
While experiments performed in tissue culture can be informative and suggest functional principles that can be verified in vivo, it is known that large changes occur in gene expression of cells in culture11,12 . In order to avoid a tissue culture step when profiling gene expression in fibroblasts, we developed a protocol that allows isolation of normal, as well as cancer-associated fibroblasts from fresh mouse or human tissue. This protocol was successfully applied with skin, pancreas and breast tissues from mouse and from human9. Isolation of fibroblasts by FACS sorting requires a surface marker that labels fibroblasts. We established that PDGFRα is a robust surface marker that enables the isolation of highly enriched populations of fibroblasts9. Moreover, PDGFRα is expressed on both normal tissue fibroblasts and on CAFs, thus allowing unbiased isolation and profiling of tissue fibroblasts rather than focusing on a specific subpopulation.
Fibroblasts are a heterogeneous cell population in regard to the expression of marker molecules as well as in their origin and signaling properties8,10,18. While PDGFRα is a robust surface marker for fibroblasts, it does not label 100% of fibroblasts in all tissues resulting in an unlabeled subpopulation of cells, depending on tissue type. In skin, approximately 90% of fibroblasts are PDGFRα+ (not shown) while in mammary glands approximately 85% of total tissue fibroblasts are PDGFRα+ (Figure 2C). The percentage of PDGFRα expressing fibroblasts may vary in other tissue types, and needs to be defined for each tissue. Nevertheless, in skin and in mammary glands, utilizing PDGFRα as a cell surface marker will result in highly enriched populations of the majority of tissue fibroblasts, allowing expression profiling or culturing of sorted cells.
The authors have nothing to disclose.
We thank Dr. Yitzchak Oschry and Dr. Orit Sagi-Asif for their help with FACS sorting. This research was supported by grants to NE from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement n ° [276890], from the Israel Cancer Association (#20110078), and from the Israel Cancer Research Fund (Research Career Development Award).
Name of the reagent | Company | Catalogue number | Comments (optional) |
DMEM | Gibco | 41965 | |
PBS | Biological Industries | 02-023-1A | |
Collagenase II | Worthington | LS4176 | |
Collagenase IV | Worthington | LS4188 | |
Deoxyribonuclease | Worthington | LS2007 | |
PharmLyse | BD | 555899 | |
Cell strainer 70 μm | SPL | 93070 | |
Purified anti-mouse CD16/CD32 | BD Pharmingen | 553142 | |
Via probe (7AAD) | e-Bioscience | 00-6993-50 | |
Anti-mouse CD140a-PE (PDGFRa) | e-Bioscience | 12-1401-81 | |
Anti-mouse F4/80- FITC | Cederlane | CL8940F | |
DMEM w/o Phenol Red | Gibco | 31053 | |
Collagen Type I | BD Biosciences | 354236 |