Here, we present a protocol to construct a three-dimensional in vitro model of the lining of the peritoneal cavity, composed of primary human mesothelial cells and fibroblasts layered with extracellular matrix, as a tool to investigate ovarian cancer cell adhesion, invasion, and proliferation.
The pattern of ovarian cancer metastasis is markedly different from that of most other epithelial tumors, because it rarely spreads hematogenously. Instead, ovarian cancer cells exfoliated from the primary tumor are carried by peritoneal fluid to metastatic sites within the peritoneal cavity. These sites, most notably the abdominal peritoneum and omentum, are organs covered by a mesothelium-lined surface. To investigate the processes of ovarian cancer dissemination, we assembled a complex three-dimensional culture system that reconstructs the lining of the peritoneal cavity in vitro. Primary human fibroblasts and mesothelial cells were isolated from human omentum. The fibroblasts were then mixed with extracellular matrix and covered with a layer of the primary human mesothelial cells to mimic the peritoneal and omental surfaces encountered by metastasizing ovarian cancer cells. The resulting organotypic model is, as shown, used to examine the early steps of ovarian cancer dissemination, including cancer cell adhesion, invasion, and proliferation. This model has been used in a number of studies to investigate the role of the microenvironment (cellular and acellular) in early ovarian cancer dissemination. It has also been successfully adapted to high throughput screening and used to identify and test inhibitors of ovarian cancer metastasis.
Ovarian cancer is the deadliest gynecologic malignancy1. The majority of patients are diagnosed after the cancer has disseminated throughout the peritoneal cavity. Once the cancer has spread throughout the peritoneal cavity, cytoreductive surgery and chemotherapy are often not sufficient treatment to prevent cancer recurrence and chemoresistance, resulting in a less than 30% 5-year survival rate. Ovarian cancer metastasis is predominantly limited to the peritoneal cavity, and several other cancer types, including gastric, pancreatic, and colon cancers, metastasize to the same anatomic sites in the peritoneal cavity. In general, ovarian cancer cells detach from the in situ carcinoma in the fallopian tube or the primary ovarian tumor, travel in peritoneal fluid as single cells or spheroids, and attach to mesothelium-lined surfaces of the omentum, bowel, and abdominal wall2.
The tumor microenvironment plays an important role in disease progression and chemoresistance in many cancers3-6. The peritoneal cavity is a unique microenvironment, with a mesothelial cell monolayer covering the majority of surfaces (Figure 1A)7. The mesothelial lining acts as a barrier that creates a low-friction surface, which tends to be protective against cancer cell adhesion8. Immediately underneath this mesothelial-lined surface is a layer made predominantly of fibroblasts and extracellular matrix (ECM), which promote cancer cell adhesion and invasion8. Ovarian cancer cells secrete factors that induce changes in the mesothelial cell lining that enhance ovarian cancer cell adhesion, invasion, and metastasis9,10. Ovarian cancer cells adhere to the mesothelial surface via integrin and CD44-mediated mechanisms (Figure 1B)11-16.
Historically, several 3D models have been developed to investigate ovarian cancer interactions with the microenvironment. Some of the first models studied ovarian cancer-ECM interface17-21, ovarian cancer-mesothelial cell communication13,14,21-24, or both25 (reviewed by us 26). Niedbala et al. discovered that ovarian cancer cells display a quicker and firmer adhesion to ECM than to mesothelial cells or to plastic alone25. However, these models did not histologically resemble the peritoneal microenvironment. Therefore, we established a 3D organotypic model to more thoroughly replicate the ovarian cancer microenvironment. In order to better understand the role of the microenvironment and the interaction between cancer and peritoneal cells in the peritoneal dissemination of ovarian cancer, we have developed a 3D organotypic in vitro culture model of the peritoneal cavity lining (schematic in Figure 1C). The proposed model is composed of primary human fibroblasts and ECM, covered with a layer of primary human mesothelial cells-each cell type is isolated from human omentum. Histologically, this model resembles the normal peritoneal or omental lining, and provides a surface on which we can study the tumor microenvironment, the interaction between cancer cells and normal tissue, and the processes of cancer cell adhesion, invasion, and proliferation8.
All research protocols described have been reviewed by the University of Chicago Institutional Review Board (IRB). Informed consent was obtained from each patient before surgery and the study was approved by the University of Chicago IRB. A biological safety cabinet type 2 and gloves should be used when handling human tissue for protection and to reduce risk of contaminating cells.
1. Isolation and Culture of Primary Untransformed Stromal Cells
2. Plating the Organotypic Culture8
3. Adhesion Assay8
4. Proliferation Assay10
5. Invasion Assay8
The organotypic culture was assembled by first mixing primary human fibroblasts with collagen type I and then overlaying this culture with 5 times the number of mesothelial cells. The culture was incubated for at least 18 hr before ovarian cancer cells were added to study adhesion, invasion or proliferation. Each assay was repeated with multiple (n=3-5) 3D cultures obtained from different patients and numerous wells were tested in each condition for the adhesion (n=5), proliferation (n=5) and invasion assays (n=3). Ovarian cancer cells adhered to the 3D organotypic culture after 4 hr (Figure 4), proliferated on the culture after 72 hr (6 times the HeyA8 cell doubling time) (Figure 5), and invaded after 24 hr (Figure 6). The adhesion, proliferation and invasion of HeyA8 ovarian cancer cells was integrin-dependent (Figure 4-6). Specifically, the RGD peptide and blocking antibodies to α5-integrin, β1-integrin and αvβ3-integrin inhibited the ovarian cancer cell adhesion to the organotypic culture, while the RAD peptide, control antibody and β4-integrin antibody did not (Figure 4)10. In the proliferation assay, the RGD peptide and blocking antibodies to α5– and β1-integrin inhibited ovarian cancer cell proliferation on the organotypic culture, while the RAD peptide, control antibody, αvβ3-integrin antibody and β4-integrin antibody did not (Figure 5). In the invasion assay, the RGD peptide and blocking antibodies to α5– and β1-integrin inhibited, while the αvβ3-integrin antibody enhanced ovarian cancer cell invasion through the 3D organotypic culture. The RAD peptide, control antibody, and β4-integrin antibody did not affect invasion (Figure 6). Rarely, these assays will be un-interpretable (i.e., α5-integrin blocking antibody will not inhibit adhesion, proliferation or invasion when compared to the control antibody). Previously, we have found that certain organotypic cultures will wash away upon changing of the media or PBS washing during the assays. In addition, if the ovarian cancer cells are dead or have not recovered long enough from the trypsin digest to re-express integrins, the assays will not work. Therefore, it is always important to have a positive and negative control in each assay so the quality of the organotypic culture and ovarian cancer cell assay can be assessed. As shown here, the RGD peptide, α5-integrin and β1-integrin blocking antibodies all inhibited ovarian cancer cell adhesion, invasion and proliferation to the organotypic culture and could be used as positive controls in future assays. Likewise, β4-integrin blocking antibody did not affect ovarian cancer cell adhesion, invasion and proliferation to the organotypic culture and could be used as a negative control in future assays.
Figure 1. Histology of human omental metastasis. Hematoxylin and eosin stain of (A) normal human omentum (MC, mesothelial cells, Fib, fibroblasts, Adi, adipocytes) and (B) ovarian cancer omental metastasis (Met, ovarian cancer metastasis). Bar in panels = 100 μm in length. (C) Schematic of the 3D organotypic model of ovarian cancer metastasis mimicking early ovarian cancer cell metastasis to the surface of the human omentum. Please click here to view a larger version of this figure.
Figure 2. Isolation of primary human omental cells. (A) A piece of omentum is collected at surgery in PBS and transported to the laboratory. The tissue is pelleted at 0.5 x g x 3 min and transferred to fresh PBS. (B) The piece of omentum is transferred to a 10 cm petri dish, washed with PBS, and scalpels are used to cut the omentum into small pieces (0.5 cm3). (C-D) The tissue pieces are collected using a 25 ml serological pipette and transferred into a 50 ml conical tube for isolation of HPMC. PBS is removed from the tube and HPMC are pelleted from the PBS a 0.5 x g at 37°C. (MC, mesothelial cells, RBC, red blood cells, Fib, fibroblasts) (E) NOF are isolated from the remaining omental tissue after incubation with hyaluronidase and collagenase type III. The image shows the tissue after a 6- to 12-hr digestion. NOF are pelleted by spinning at 0.5 x g at RT. Please click here to view a larger version of this figure.
Figure 3. Characterization of HPMC and NOF isolated from human omentum tissue. Phase-contrast images of (A) mesothelial cells (HPMC) and (B) fibroblasts (NOF) after isolation from omentum tissue. Human HPMC stained for (C) cytokeratin 8 and (E) vimentin. Human NOF stained for (F) vimentin, but did not stain for (D) cytokeratin 8. Bar = 100 µm, applies to all panels. Please click here to view a larger version of this figure.
Figure 4. Ovarian cancer cell adhesion to the organotypic culture is integrin-dependent. (A) The adhesion assay was performed with 50,000 fluorescently-labeled ovarian cancer cells, HeyA8 cells, as described in the Protocol section. The total fluorescence in each well is reported. The control antibody is mouse IgG. Each bar represents the mean +/- standard error mean. Representative (B) phase-contrast and (C) fluorescent images of ovarian cancer cell adhesion to the organotypic culture. Bar = 100 µm, applies to both panels. Please click here to view a larger version of this figure.
Figure 5. Ovarian cancer cell proliferation on the organotypic culture is integrin-dependent. (A) The proliferation assay was performed with 4,000 fluorescently-labeled ovarian cancer cells, HeyA8 cells, as described in the Protocol section. The total number of cells is reported. The control antibody is mouse IgG. Each bar represents the mean +/- standard error mean. Merged phase-contrast and fluorescent images of ovarian cancer cell proliferation on the organotypic culture at (B) 24 and (C) 96 hr. Bar = 100 µm, applies to both panels. Please click here to view a larger version of this figure.
Figure 6. Ovarian cancer invasion through the organotypic culture is integrin-dependent. (A) The invasion assay was performed with 40,000 fluorescently-labeled ovarian cancer cells, HeyA8 cells, as described in the Protocol section. The number of invading cells (moved through the 3D culture to the bottom of the insert) is reported. The control antibody is mouse IgG. Each bar represents the mean +/- standard error mean. (B) Phase-contrast and (C) fluorescent images of ovarian cancer cells that invaded through the organotypic culture. Bar = 100 µm, applies to all panels. Please click here to view a larger version of this figure.
An organotypic model of the peritoneal microenvironment was established to assess the individual and collective function(s) of both the cellular and innate components of the microenvironment in ovarian cancer dissemination. The specific protocols for plating and customizing the 3D organotypic culture to investigate ovarian cancer cell adhesion, proliferation, and invasion are provided. Primary human omental mesothelial cells and fibroblasts were isolated from patients and used at an early passage to preserve normal morphology and patient variation. In this model, normal omental fibroblasts and ECM (typically collagen type I) were layered with a primary human mesothelial cell monolayer. This model histologically mimics the peritoneal lining, and allows us to recreate and examine key events in ovarian cancer metastasis including ovarian cancer cell adhesion, proliferation, and invasion8,27.
Previously, limited 3D models have been established to investigate ovarian cancer interactions with the microenvironment13,14,21-25, while this 3D organotypic model is the first to recreate the construction of the mesothelial-lining of the peritoneal cavity. The human omentum is collected from patients varying in age and health. The health of the tissue directly affects the mesothelial cell and fibroblast isolation techniques. The healthier tissue appears lighter in color with very small lobules (1-3 mm) compared to tissue that is more orange in color with large lobules present (>5 mm). The majority of HPMC slough off in the initial PBS as the tissue is carried from surgery to the laboratory. The healthier the tissue, however, the more resilient the HPMC are to removal, and additional PBS washes and the 1:1 PBS:trypsin digest may result in a mixed culture of HPMC and NOF cells. In the healthier tissues, some HPMC can survive the 6-hr collagenase digest, and a digest O/N in 1x collagenase solution in full growth media is recommended to obtain a pure NOF culture. Note the omentum must be placed immediately in PBS upon removal or no HPMCs will be recovered from tissue. Furthermore, if the tissue is stored in PBS for an extended period of time (>2 hr) at RT or in a refrigerator, the number of viable HPMCs isolated is significantly reduced.
A 1:5 ratio of fibroblasts to mesothelial cells is plated to mimic the ratio of fibroblasts to mesothelial cells in vivo8. It is important to note that the size of isolated mesothelial cells varies from patient to patient and this may require adjustment to the plating ratio. The mesothelial cells of some patients are much smaller; in these cases, we plate double the number of mesothelial cells (20,000). In addition, the primary cells are always used at an early (1-2) passage to preserve morphology. Mesothelial cells can be cuboidal or spindly, and cuboidal mesothelial cells become more spindly as they are passaged, or with greater exposure to conditioned media from ovarian cancer cells or treatment with TGFβ10.
Ovarian cancer cells may adhere to and invade through the mesenchymal mesothelial cells more efficiently than the epithelial mesothelial cells from the same or different patients. Typically, we set up the organotypic culture for 18 hr before starting the functional assays with ovarian cancer cells. The longer the mesothelial cells and fibroblasts are cultured together, the more time they have to secrete different proteins including ECM proteins, which can affect the behavior of the ovarian cancer cells. Another important factor to take into account when performing these assays is the differential proliferation rate and expression level of proteins associated with increased adhesion, invasion, and/or proliferation, including integrins, proteases, and growth factors, in each cancer cell line or primary human cancer cell types. The time and number of cancer cells used in each of the functional assays with the organotypic culture needs to be optimized for each cancer cell line or type of primary cancer cell. For example, in the described adhesion assay, ovarian cancer cells could be invading the culture during this time.
Clearly, this organotypic model does not contain all components of the normal peritoneal microenvironment such as immune, endothelial, and adipocyte cells. Additionally, the model depends on the accessibility and availability of primary human tissue, as the primary cells are only used within the first three passages. However, the model has many advantages in its use of primary human cells and incorporation of multiple cell types that have been demonstrated to play important roles in metastasis. This model has been used to investigate gene and protein regulation in the individual cell types (i.e., cancer cells, normal mesothelial or fibroblast cells) after co-culture. In particular, the roles of c-Met28, MMP-29,29, E-cadherin30, β1-integrin10, β3-integrin16, α5-integrin10,31, and fibronectin10 in early metastasis have been explored. Additionally, the model has the potential to expand to include more cell types, such as the incorporation of adipocytes6. The model has been modified and miniaturized for use in high-throughput screening to identify small molecule inhibitors of ovarian cancer cell adhesion/invasion to the peritoneal microenvironment27. Future applications of this model will include the incorporation of immune cells into the assay, including macrophages for mechanistic studies and high-throughput screening. In summary, the organotypic cell culture provides a useful model to evaluate the interactions between the human peritoneal microenvironment and ovarian cancer cells, with the goal of elucidating important molecular mechanisms of dissemination and disease progression, and identifying potential therapeutic targets.
The authors have nothing to disclose.
We thank all residents and attending physicians, notably Dr. A.F. Haney (the University of Chicago, Department of Obstetrics and Gynecology) for collecting omental biopsies. Also, we thank Stacey Tobin and Gail Isenberg for carefully editing this manuscript. This work was supported by Bears Care, the charitable beneficiary of the Chicago Bears Football Club, the National Institute of Neurological Disorders and Stroke (NINDS) R21 NS075702, and the National Cancer Institute grant R01 CA111882 to E.L.
1. Isolation and culture of primary cells | |||
PBS | Fisher Scientific | SH3001304 | |
Single-edged razor blades | Fisher Scientific | 12-640 | |
15 cm culture dishes | BD Biosciences | 353025 | |
Glass flask | ? | ? | |
Fetal Bovine Serum (FBS) | Life Technologies | 16000044_3616914956 | |
DMEM with L-Glutamine | Corning | 10-013-CV | |
MEM Vitamins | Corning | 25-020-Cl | |
MEM Nonessential amino acids | Corning | 25-025-CI | |
Penicillin-Streptomycin | Corning | 30-002-CI | |
Shaker | Thermo-Fisher | MaxQ 4450 | |
Centrifuge | Eppendorf | 5702 | |
Incubator | Thermo-Fisher | Forma Series II Water Jacketed CO2 Incubator Model 3100 | |
Trypsin EDTA, 1x (0.25%) | Corning | 25-053-CI | |
Hyaluronidase | Worthington Biochemical | LS002592 | |
T-75 Flasks | BD Biosciences | 353136 | |
T-175 Flasks | BD Biosciences | 353112 | |
Pipet tips | Rainin | P2, P10, P20, P200 and P1000 | |
Pipet tips | Corning | Filtered tips P2, P10, P20, P200 and P1000 | |
Name of Reagent/ Equipment | Company | Catalog Number | Comments/Description |
2. Plating 3D culture | |||
Cell Counter | Invitrogen | Countess | |
Countess Cell Counting Chamber Slides | Invitrogen | C10313 | |
Trypan Blue Stain (0.4%) | Gibco | 15250-061 | |
Collagen Type I (Rat Tail) | BD Biosciences | 354236 | |
96 well plate, clear bottom, black | BD Biosciences | 353219 | |
Name of Reagent/ Equipment | Company | Catalog Number | Comments/Description |
3. Adhesion assay | |||
Multichannel pipet | Eppendorf | Xplorer 300 | |
Paraformaldehyde solution 4% in PBS | Santa Cruz Biotechnology | sc-281692 | |
Plate reader | Molecular Devices | Minimax | |
Name of Reagent/ Equipment | Company | Catalog Number | Comments/Description |
5. Invasion assay | |||
Cell Culture Inserts (8um, 24-well) | BD Biosciences | 353097 | |
Cotton swabs | Q-tips | cotton swabs | |
Microscope | Zeiss | Axiovert 200m | |
Cell Profiler | public domain | ||
24 well plate | BD Biosciences | 353047 | |
Name of Reagent/ Equipment | Company | Catalog Number | Comments/Description |
6. Antibodies | |||
Anti-Integrin αVβ3 Antibody, clone LM609 | EMD Millipore | MAB1976 | |
Beta 1 | Oncosynergy | OS2966 | |
Alpha 5 [CD49e] | ID Pharmingen | 555615 | |
Beta 4 [CD104] | EMD Millipore | MAB 2058 |