We report a reliable method to isolate and culture primary tumor-specific endothelial cells from genetically engineered mouse models.
Freshly isolated tumor-specific endothelial cells (TEC) can be used to explore molecular mechanisms of tumor angiogenesis and serve as an in vitro model for developing new angiogenesis inhibitors for cancer. However, long-term in vitro expansion of murine endothelial cells (EC) is challenging due to phenotypic drift in culture (endothelial-to-mesenchymal transition) and contamination with non-EC. This is especially true for TEC which are readily outcompeted by co-purified fibroblasts or tumor cells in culture. Here, a high fidelity isolation method that takes advantage of immunomagnetic enrichment coupled with colony selection and in vitro expansion is described. This approach generates pure EC fractions that are entirely free of contaminating stromal or tumor cells. It is also shown that lineage-traced Cdh5cre:ZsGreenl/s/l reporter mice, used with the protocol described herein, are a valuable tool to verify cell purity as the isolated EC colonies from these mice show durable and brilliant ZsGreen fluorescence in culture.
Endothelial cells (EC) are essential during the development of solid tumors. From initiation of the angiogenic switch in dormant tumors to dissemination and seeding of metastases at distant sites, EC form the conduits that provide blood, oxygen, and nutrients to sustain tumor growth 1. As recently suggested, EC also have perfusion-independent functions and form a niche that supports the growth of cancer stem cells and other tumor stromal cells 2-5. Thus, highly purified tumor-specific EC (TEC) for in vitro culture allows for routine functional studies that will shed light on novel molecular mechanisms mediating tumor angiogenesis and cross talk with tumor cells.
EC are highly specialized depending on the tissue of origin 6. Due to the heterogeneous nature of different tumor types and the tumor microenvironment, TEC may also display unique features that reflect a tumor-specific specialization of the vasculature. For example, there is striking variability in the gene expression signatures in TEC isolated from different types or grades of tumors 7,8. However, frequent co-purification of non-EC, especially tumor-associated fibroblasts and tumor cells, with TEC can confound genome-wide expression analyses. These unwanted cell types are especially problematic in studies that rely on long-term in vitro expansion of TEC cultures.
Described here is a high-fidelity method that consistently produces pure EC cultures from tumors and other tissues. Following immunomagnetic column enrichment of EC fractions and removal of co-purified non-EC, an additional cloning-ring step to capture pure EC colonies is used 9. Each colony can be expanded in culture for multiple passages without the emergence of contaminating non-EC. This method also yields multiple EC clones from a single isolation procedure, which is ideal for the study of endothelial heterogeneity. In addition, it is shown that Cdh5cre:ZsGreenl/s/l reporter mice are a valuable tool for generating “fate-mapped” and indelibly-marked EC which maintain ZsGreen fluorescence in culture 10. With minor adjustments to the protocol, this method should be adaptable to different tumor types or normal tissues.
The following protocol is carried out according to guidelines established by the Department of Laboratory Animal Medicine at the University of North Carolina at Chapel Hill.
1. Prepare the Following Material and Reagents Before Starting
2. EC Isolation (Day 1, ~5 hr)
3. Colony Selection Using Cloning Rings
EC represent only minor a fraction of the total cell population in most adult tissues 11. It is therefore important to fully digest the harvested tissue into a single-cell suspension that ensures the maximal release of EC from extracellular matrix (ECM) and connective tissues. In our experience, CD31-mediated immunomagnetic selection only provides enriched but not pure EC fractions; therefore, another crucial step is the physical removal of co-purified non-EC and selection/expansion of EC colonies using cloning rings (Figure 1). For example, when CD31-enriched EC were plated without further colony selection, non-EC rapidly proliferated and intermingled with EC, producing impure EC cultures as revealed by Dil-Ac-LDL+ and DiI-AC-LDL– populations (Figure 2A). However, when column eluates were plated at low density, the EC colonies grew with relatively few surrounding non-EC, which were removed by scraping with a pipet tip (Figure 2B). The growth and purity of the expanded colonies can be further monitored by addition of Dil-Ac-LDL. Small EC colonies can be observed at ~day seven after CD31-mediated column enrichment (Figure 2B, top panels). After selection and expansion of EC colonies, Dil-Ac-LDL– cells are eliminated, and the cell population is highly pure for Dil-Ac-LDL+ EC as indicated by Dil-Ac-LDL uptake (Figure 2B, bottom panels).
To test if the isolation method can produce consistent results, multiple clones derived from either mammary tumors resected from C3-TAg (FVB/N C31-TAg) mice or normal mammary glands from age-matched wild-type FVB mice were isolated and expanded 12. The purity of each TEC and normal EC (NEC) population was then carefully characterized to ensure purity. Flow cytometry analyses of three independent samples showed a single, uniformed CD31+ population that was distinct from IgG isotype controls (Figure 3A). The mRNA expression of EC markers, including Cd31, Cdh5 (VE-Cadherin), Cd133, and Vegfr2 in all four EC clones tested was ~200 to 7,000 times higher than that of mouse embryonic fibroblasts (MEFs) which was used as a negative control (Figure 3B). As expected, all EC clones expressed nearly undetectable levels of the mesenchymal marker genes Col1a1 and Tagln when compared to MEFs. Immunofluorescent staining showed that all cells in a representative TEC clone were uniformly CD31+ (Figure 3C). Furthermore, these EC clones were able to form spontaneous vessel-like structures in vitro, indicating they retain endothelial functions after culturing (Figure 3D).
The isolation method was further validated using Cdh5cre:ZsGreenl/s/l mice where all EC are indelibly labeled with the green fluorescent protein, ZsGreen (Figure 4A, a). Lung tissue that contains abundant EC was used as proof-of-principle for the isolation procedure (Figure 4A, b). ZsGreen+ cells comprised ~30% of the total cell population in the lung homogenate, and were partially enriched after CD31-mediated immunomagnetic column selection (Figure 4B). As Cdh5cre:ZsGreenl/s/l mice carry a constitutive cre gene, a proportion of hematopoietic cells are also labeled with ZsGreen 10. Thus, the actual EC percentage in the lungs may be lower than the observed ~30%. Notably, approximately 20% of cells co-isolated with ZsGreen+/CD31+ EC were ZsGreen– (Figure 4B). Because contamination by these non-EC may persist in culture, the enriched EC population at this stage is not suitable for studies that require further in vitro cell culture. However, after applying cloning rings to capture pure EC colonies, a 100% pure ZsGreen+ population was obtained that could be further expanded in culture (Figures 4C and 4D).
Figure 1. Flow Chart and Overview of the EC Isolation Procedure.
Figure 2. Dil-Ac-LDL Distinguishes EC from Contaminating Non-EC in Live Cultures. (A) Phase contrast and fluorescent images showing EC and co-isolated non-EC without cloning-ring selection of EC colonies. (B) Phase contrast and fluorescent images of EC colonies at different stages of isolation. Dotted lines mark the boundaries of Dil-Ac-LDL+ EC and surrounding contaminating cells. White arrowheads indicate non-EC outside an EC colony that should be removed with a pipette tip. Scale bars represent 100 µm. Please click here to view a larger version of this figure.
Figure 3. Cloning Rings and Physical Removal of Non-EC Produce Pure and Functional Long-term EC Cultures. (A) Representative FACS dot plots of CD31 staining of different EC clones. Three representative samples are shown. The open black rectangle in each plot outlines the CD31+ population. An IgG isotype is used as a negative control. (B) Measurement of endothelial gene expression in selected NEC and TEC clones. The mRNA levels of endothelial genes Cd31, Cd133, Cdh5, and Vegfr2 are expressed relative to those of mouse embryonic fibroblasts (MEFs), and the mRNA levels of mesenchymal genes Col1a1 and Tagln are expressed relative to those of NEC-1. (C) Representative immunofluorescent staining of an expanded TEC clone. CD31 is shown in green and nuclei are stained blue with 4’6’-diamidino-2-phenylindole (DAPI). (D) Representative images of tube formation by different EC clones plated on Matrigel. Scale bars represent 100 µm. Please click here to view a larger version of this figure.
Figure 4. EC Isolated from Cdh5cre:ZsGreenl/s/l Mice Maintain Brilliant ZsGreen Fluorescence in Culture. (A) The endothelial-lineage tracing mouse Cdh5cre:ZsGreenl/s/l carries a floxed ZsGreen transgene that is induced by Cdh5cre expressed by EC. (a) Lung tissue from the Cdh5cre:ZsGreenl/s/l mice were harvested and digested for EC isolation. (b) Representative images of Cdh5cre:ZsGreenl/s/l mouse lung tissue. ZsGreen (green) labels the endothelium and nuclei are stained with DAPI (blue). (B) FACS dot blots showing the percentages of ZsGreen+ cells before (middle panel) and after (right panel) the CD31-mediated column enrichment from the homogenized lung tissue of a Cdh5cre:ZsGreenl/s/l mouse. Unstained wild-type (WT) lung homogenate (left panel) was used as a negative control for gating. The large open rectangles indicate ZsGreen– cells and the small open rectangles indicate cells double positive for ZsGreen (FL1-H channel) and CD31 (FL2-H channel). (C) FACS histogram plot showing that ZsGreen+ EC (green peak) remain 100% pure after in vitro culturing. A ZsGreen– EC population was used as a negative control (gray peak). (D) Representative phase contrast and fluorescent images of an early-stage ZsGreen+ EC colony and an expanded ZsGreen+ EC colony. Scale bar represents 100 µm. Please click here to view a larger version of this figure.
Cell number | FACS buffer (μL) | CD31-PE antibody (μl) |
1 x 107 | 100 | 5 |
2 x 107 | 200 | 6 |
3 x 107 | 300 | 7 |
4 x 107 | 400 | 8 |
5 x 107 | 500 | 9 |
6 x 107 | 600 | 10 |
Table 1. PE-conjugated Anti-CD31 Antibody Volumes Required for Different Cell Numbers.
Cell number | FACS buffer (μl) | Anti-PE magnetic microbeads (μl) |
1 x 107 | 80 | 20 |
2 x 107 | 160 | 40 |
3 x 107 | 240 | 60 |
4 x 107 | 320 | 80 |
5 x 107 | 400 | 100 |
6 x 107 | 480 | 120 |
Table 2. Anti-PE Microbead Volumes Required for Different Cell Numbers.
Due to the difficulties in obtaining pure primary TEC cultures, many in vitro studies substitute TEC with commercially available EC lines or primary EC such as human umbilical vein EC (HUVEC) 13. However, these EC populations from normal tissues may only serve as a proxy for TEC which differ markedly from their normal counterparts. For example, TEC are phenotypically and functionally abnormal in vivo and some of these abnormalities may be transmittable in vitro 14-18. TEC have aberrant growth, migratory, and differentiation abilities and may coalesce with CD31+ tumor cells to form vessel-like structures 16,19,20. Studies using either cultured TEC or laser capture micro-dissected tumor vessels have demonstrated that TEC deviate from normal EC in gene expression, cytogenetic, and epigenetic profiles 12,18,21-23. Despite deepening our understanding of tumor angiogenesis over the past few decades, very few functional and mechanistic studies using cultured primary TEC are available.
EC were isolated from human umbilical veins in the early 1970s 24. Subsequent isolation and culturing of microvascular EC from other human and mouse tissues has provided a powerful tool for investigating endothelial functions in both health and disease 25-28. Most EC isolation protocols involve three major steps: obtaining a single-cell suspension after mechanical dissociation or enzymatic digestion, labeling EC with an endothelial-specific antibody that is conjugated to either a fluorophore or magnetic microbeads, and enriching the EC population using cell sorting or magnetic columns. However, isolation of mouse EC, especially from tumors, has proven difficult due to the low yield of viable EC and frequent contamination of other cell types including tumor cells and fibroblasts. In addition, in vitro phenotypic drift of EC into mesenchymal-like cells (endothelial-to-mesenchymal transition, EndMT) poses an additional challenge for long-term culturing of EC from normal tissues, tumors, and reprogrammed precursors.
The major challenge during TEC isolation is contamination by tumor cells, which can easily outcompete the slower-growing EC colonies that typically appear after ~7 – 10 days in culture. In addition, a commonly used selection marker for isolating EC is CD31 which is abundantly expressed in the endothelium but also marks hematopoietic cells and in some cases tumor cell subpopulations 20,30. Furthermore, while enriching for EC, CD31-mediated immunomagnetic column selection cannot remove every single contaminating cell that can eventually take over EC cultures after a few passages. By adding a cloning-ring step, one is able to select and expand pure clonally-derived EC populations that are free of these contaminating cell types. A second challenge is the long-term maintenance of pure EC without promotion of EndMT. EndMT occurs during development, may be recapitulated during vascular dysfunction, and is common in cultured EC (especially in the presence of TGFβ) 31-33. To minimize EndMT, remove contaminating cells that may act as a source TGFβ, keep cultures at high densities, and maintain high concentrations of bFGF in the media at all times. As we have shown recently, bFGF is essential to preserve EC specification as it antagonizes TGFβ-driven expression of smooth muscle actin (an EndMT marker) 34.
Using this methodology, isolation of EC from a “fate-mapped” reporter mice was also shown. These mice are particularly useful in studies where intravital imaging is needed 35. Although some labeling of hematopoietic cells may occur in the Cdh5cre:ZsGreenl/s/l mice used here, this model provides a relatively rapid and easy method for generating fluorescent EC in vivo. An alternative EC lineage-labeling model is a tamoxifen-inducible mouse (Cdh5cre/ERT2) crossed with a floxed reporter mouse strain (ZsGreenl/s/l) that will have EC faithfully and irreversibly marked 36,37. Fluorescent EC from the inducible cre mice may also be purified using FACS or using the methodology we describe herein.
EC are heterogeneous across different vascular beds and “intra-vessel” heterogeneity may also exist among EC within the same vessel 38,39. In addition, a hierarchy of EC with different proliferative potential has been observed in EC populations isolated from veins, and a small fraction of clonal EC can be passaged beyond 40 population doublings 40. Therefore, limitations of the described method here include: 1) possible selection of these highly proliferative EC residing in the vessel wall; and 2) enrichment of only one specific vascular subtype derived from arteries, veins, or capillaries, which may produce clones that do not completely represent the total EC population in vivo. Nevertheless, as multiple clonal subpopulations can be obtained from one isolation procedure, this method can be useful for functional and genetic analyses of EC heterogeneity that exists within tumors and other tissues 12. Taken together, described here is an EC isolation method that produces pure EC colonies devoid of contaminating non-EC. The isolated cells should be useful for in vitro functional studies, genome-wide expression profiling, and molecular characterization of new pathways that control tumor angiogenesis.
The authors have nothing to disclose.
ACD is supported by a grant from the National Institute of Health (R01-CA177875). LX is a fellow in the HHMI-funded translational medicine program at UNC Chapel Hill. JVM is supported by a T32 pre-doctoral fellowship from the Integrative Vascular Biology Program at UNC Chapel Hill. We thank Clayton Davis for assistance with confocal microscopy.
Antibiotic-Antimycotic | Sigma-Aldrich | A5955 | |
Dulbecco's Modified Eagle's medium (1 g/L D-glucose) (LG-DMEM) | Gibco | 11885-084 | |
EGM-2 Bullet Kit | Lonza | CC4176 | Not all components used |
Fetal bovine serum (Hyclone) | Thermo Scientific | SH30071.03 | Heat inactivated at 56°C for 30 min |
Nu-Serum IV | Corning | CB-51004 | |
Hank's Balanced Salt Solution (HBBS) | Gibco | 14175-095 | |
Phosphate-buffered saline (PBS) | Gibco | 14190-144 | |
FACS buffer | – | – | 0.5 % BSA and 2 mM EDTA in PBS, filtered through a 0.22 μm filter |
75% v/v ethanol for disinfection | – | – | |
Anti-PE microbeads | Miltenyi Biotech | 130-048-801 | |
Bovine serum albumin (BSA) fraction V, 7.5% | Gibco | 15260-37 | |
Cell freezing media (Bambanker) | Wako Chemicals | 302-14681 | |
Collagenase type II | Worthington Biochemical | LS004176 | Make stock concentration 2 mg/ml in HBSS |
Deoxyribonuclease I (DNase) | Worthington Biochemical | LS002004 | Make stock concentration 1 mg/mL in PBS |
Dil-Ac-LDL | Biomedical Technologies | BT-902 | |
EDTA, 0.5M, pH 8.0 | Cellgro | 46-034-CL | |
Enzymatic cell detachment solution (Accutase) | Sigma-Aldrich | A6964-100ML | |
Gelatin, 2 % in water, tissue culture grade | Sigma-Aldrich | G1393-100ML | Dilute in PBS to make 0.5 % gelatin solution |
Mouse FcR Blocking Reagent | Miltenyi Biotech | 130-092-575 | |
Neutral protease (Dispase) | Worthington Biochemical | LS02104 | Make stock concentration 2.5 U/mL in HBSS |
PE-rat anti-mouse CD31 antibody | BD Pharmingen | 553373 | |
RBC lysis buffer (BD Pharm Lyse) | BD Pharmingen | 555899 | |
Sterile water | – | – | |
Trypan blue, 0.4 % | Life Technologies | 15250-061 | |
10 mm tissue culture dishes | Corning | – | |
15 mL conical tubes (sterile) | Corning | – | |
50 mL conical tubes (sterile) | Corning | – | |
6-well tissue culture plates | Corning | – | |
Tissue-dissociator tubes (gentleMACS) C tubes) | Miltenyi Biotech | 130-093-237 | |
Cell Separator (MidiMACS) | Miltenyi Biotech | 130-042-302 | |
Cell strainer 100 μm | Corning | 352360 | |
Cloning rings (assorted sizes) | Bel-Art Products | 378470000 | |
Cryotubes | Thermo Scientific | – | |
Dissecting board | – | – | Sterilize or disinfect with 75% v/v ethanol before use |
Dissecting forceps and scissors | – | – | Sterilize before use |
Dissecting pins 2" | – | – | Sterilize before use |
FACS tubes with 35 μm filter cap | Corning | 352235 | |
Filter cup (Stericup, 0.22 μm) | Millipore | SCGPU05RE | |
Fine-tip marker | – | – | |
Hemocytometer | – | – | |
LS Columns | Miltenyi Biotech | 130-042-401 | |
Magnetic Multistand | Miltenyi Biotech | 130-042-303 | |
Tissue adhesive (Vetbond) | 3M | 1469SB | |
Centrifuge | Eppendorf | 5810R | Or a centrifuge with similar capacity for 15 mL and 50 mL conical tube centrifugation |
Tissue culture hood | – | – | |
Tissue dissociator (gentleMACS) | Miltenyi Biotech | 130-093-235 | Preset program "m_impTumor_01" used for tissue dissociation |
Liquid nitrogen freezer | – | – | |
Microplate or rotary shaker | – | – | |
Phase contrast light microscope | – | – |