Orthotopic engraftment of ovarian cancer cells mixed with human stromal cells provides a mouse model that exhibits rapid, diffuse metastatic behavior that is characteristic of human ovarian cancer. This model also allows for the study of tumor cell and stromal cell interactions, as well as their role in tumor progression and metastasis.

Ovarian cancer is characterized by early, diffuse metastasis with 70% of women having metastatic disease at the time of diagnosis. While elegant transgenic mouse models of ovarian cancer exist, these mice are expensive and take a long time to develop tumors. Intraperitoneal injection xenograft models lack human stroma and do not accurately model ovarian cancer metastasis. Even patient derived xenografts (PDX) do not fully recapitulate the human stromal microenvironment as serial PDX passages demonstrate significant loss of human stroma. The ability to easily model human ovarian cancer within a physiologically relevant stromal microenvironment is an unmet need. Here, the protocol presents an orthotopic ovarian cancer mouse model using human ovarian cancer cells combined with patient-derived carcinoma-associated mesenchymal stem cells (CA-MSCs). CA-MSCs are stromal progenitor cells, which drive the formation of the stromal microenvironment and support ovarian cancer growth and metastasis. This model develops early and diffuses metastasis mimicking clinical presentation. In this model, luciferase expressing ovarian cancer cells are mixed in a 1:1 ratio with CA-MSCs and injected into the ovarian bursa of NSG mice. Tumor growth and metastasis are followed serially over time using bioluminescence imaging. The resulting tumors grow aggressively and form abdominal metastases by 14 days post injection. Mice experienced significant decreases in body weight as a marker of systemic illness and increased disease burden. By day 30 post injection, mice met endpoint criteria of >10% body weight loss and necropsy confirmed intra-abdominal metastasis in 100% of mice and 60%-80% lung and parenchymal liver metastasis. Collectively, orthotopic engraftment of ovarian cancer cells and stroma cells generates tumors that closely mimic the early and diffuse metastatic behavior of human ovarian cancer. Furthermore, this model provides a tool to study the role of ovarian cancer cell: stroma cell interactions in metastatic progression.

Ovarian cancer is a deadly disease with the 5^th highest mortality rate of all cancers in women¹. Most women with ovarian cancer are diagnosed at an advanced stage, with metastatic spread present in 70% of patients at the time of diagnosis. Factors such as early metastases and advanced stage at diagnosis contribute to the high mortality rates seen with this disease. Moreover, these unique disease characteristics have posed a challenge for establishing ovarian cancer mouse models, including reproducing rapid disease migration into the peritoneal cavity^2,3,4.

Ovarian cancer pathogenesis, including peritoneal spread, is facilitated by the formation of a supportive tumor microenvironment (TME) that comprises many elements⁵. One critical component of the ovarian cancer TME is the carcinoma-associated mesenchymal stem cell (CA-MSC). CA-MSCs are stromal progenitor cells that enhance ovarian cancer initiation, growth, chemotherapy resistance, and metastasis^6,7. CA-MSCs also drive the formation of the ovarian cancer TME through stimulating tumor-associated fibrosis, inducing angiogenesis, and altering the immune microenvironment^6,8,9. Given the powerful functions of CA-MSCs within the ovarian cancer TME, modeling human ovarian cancer within a physiologically relevant stromal microenvironment is critical in studying ovarian cancer progression and metastasis.

Recently, transgenic mouse models have gained attraction in the study of spontaneous ovarian cancer metastases. However, transgenic mice are costly and exhibit a prolonged time course for the development of metastatic disease. While other available mouse models such as the intraperitoneal model and patient derived xenografts (PDX) have relatively short metastatic time intervals, they do not fully recapitulate ovarian cancer metastases due to the lack of a relevant stromal microenvironment^10,11,12. In an attempt to overcome this challenge, this study presents an orthotopic ovarian cancer mouse model using human ovarian cancer cells combined with patient-derived CA-MSCs. In the model described here, combining CA-MSCs with ovarian cancer cells generates tumors with early and diffuse metastases, as demonstrated by the presence of intra-abdominal metastases in 100% of mice within 30 days post-injection.

Patient samples were obtained in accordance with the protocols approved by the University of Pittsburgh's IRB (PRO17080326). Animal experimental methods were conducted under the protocol approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh.

1. Isolation and validation of patient-derived carcinoma-associated mesenchymal stem cells (CA-MSCs)

NOTE: CA-MSCs are derived from surgically resected human ovarian cancer tissue (this study uses high grade serous carcinoma), including the fallopian tube, ovary and/or omental metastatic deposits. CA-MSC medium is prepared from MEBM (mammary epithelial cell basal medium) supplemented with 10% heat-inactivated FBS, 1x B27, 20 ng/mL EGF, 1 ng/mL hydrocortisone, 5 µg/mL insulin, 100 µM β-mercaptoethanol, 10 ng/mL β-FGF, 1% penicillin/streptomycin, and 20 µg/mL gentamicin^7,13.

1. CA-MSC isolation
   1. Using a sterile scalpel and forceps, divide the tissue sample into pieces approximately 1 mm³ in size. Place three pieces of the tissue in each well of a 6-well cell culture plate.
      NOTE: 0.2 g of the tumor tissue can be cut into 12 pieces and can be distributed into four wells of a 6-well cell culture plate.
   2. Add a thin layer of CA-MSC media, approximately 1 mL for each well. Place the cell culture plate in a humidified incubator at 37 °C in 5% CO₂.
   3. After 24 h, remove the media carefully without disturbing the tissue pieces. Add approximately 1 mL of fresh media to each well. Pipette the media slowly without disturbing the tissue attachments. Place the cell culture plate back in a humidified incubator at 37 °C in 5% CO₂.
      NOTE: Examine the tissue with an inverted microscope every day and change the media as described in step 1.1.3 whenever dead cells are visualized in the culture media.
   4. After 7 days, remove the tissue pieces by picking up the tissue using sterile forceps and wash the adherent cells once with PBS (PBS without Ca²⁺/Mg²⁺, 1 mL per well). Then, add fresh CA-MSC medium (2 mL per well) and place the cell culture plate in a humidified incubator at 37 °C in 5% CO₂.
2. Expansion of CA-MSCs
   1. When CA-MSCs start to grow from the tissue within the 6-well plate, examine the cells with an inverted microscope every other day to ensure cellular growth. Once the cells reach 90% confluency, proceed with the following steps.
   2. Aspirate the media from the cell culture plate, then wash the cells once with 1-2 mL of PBS per well (PBS without Ca²⁺/Mg²⁺).
   3. Add 300 µL per well of 0.05% trypsin/0.02% EDTA solution (trypsin/EDTA). Rotate the cell culture plate to cover the cells with trypsin/EDTA and return the cell culture plate to the incubator for 2-3 min or until the cells are detached.
   4. Add fresh CA-MSC medium to inactivate the trypsin (1 mL per well) and re-suspend the cells. Centrifuge at 300 x g for 5 min. Remove the supernatant and re-suspend the cells with growth medium. Transfer the re-suspended cells to a 25 cm² flask with a total volume of 5 mL of culture media.
      NOTE: Cells collected from three wells of 6-well cell culture plate can be transferred into a 25 cm² flask. Cells should be tested for the presence of mycoplasma (following an established mycoplasma detection protocol such as PCR) since they are derived from surgically resected human tissue^14,15.
   5. Once the cells reach 70%-80 % confluency, transfer the cells from a 25 cm² flask to a 75 cm² flask following the previously described steps 1.2.2-1.2.4. Expand the cells every 5-7 days to obtain the cell number needed for CA-MSC validation and for the orthotopic inoculation procedure.
      NOTE: Since CA-MSCs are patient-derived cells, the cellular doubling time differs between cell lines. Most CA-MSCs require 3-4 days to double in number. However, some CA-MSCs require 5-7 days to double in number. CA-MSCs can be passaged 8-10 times. After passage 10, CA-MSCs may senesce or differentiate.
3. CA-MSC validation
   1. Validation of CA-MSC surface expression: Use 0.5 x 10⁶ CA-MSCs (approximately 70% confluency in a 75 cm² flask) for flow cytometric analysis to ensure that CA-MSCs express appropriate surface markers as defined by the International Society for Cellular Therapy guidelines, as previously described (CD73/CD90/CD105 positive; CD45/CD34/CD14 or CD11b/CD79a or CD19/HLA-DR negative)^7,16. Confirm the cell number after trypsinization and count using a hemocytometer with trypan-blue exclusion of dead cells.
      NOTE: If the flow cytometry analysis demonstrates that the cells are <90% pure, then sort the cells for the CA-MSC population.
   2. Validation of CA-MSC differentiation: To validate lineage differentiation, seed CA-MSCs into one well of a 6-well plate at a cell density of 5 x 10⁴ cells/well in 2 mL of MSC media (one well per differentiation condition). After 24 h, change the media to the appropriate differentiation media to induce adipocyte vs chondrocyte vs osteocyte differentiation as previously described^7,16.
      NOTE: As per ISCT guidelines, MSCs must demonstrate in vitro differentiation into at least two lines (adipocyte, chondrocyte, or osteocyte).
4. On the day of the planned mouse injection procedure, collect the cells by following steps 1.2.2 and 1.2.4. Then, count the cells using a hemocytometer with trypan-blue exclusion for viability and re-suspend the cells in CA-MSC medium at a density of 1 x 10⁸ cell/mL (0.5 x 10⁶ CA-MSCs per injection). Place the suspended cells over ice.
   NOTE: Generally, at least 1 x 10⁷ CA-MSCs can be derived from one tumor sample (0.2 g of tissue). This allows a single patient-derived CA-MSC line to be used for most mouse experiments. It is possible to combine CA-MSCs from different patients, but we recommend they are derived from the same ovarian cancer histologic subtype. Most CA-MSCs lines take 2 months and/or 6-7 passages to reach 1 x 10⁷ cells. CA-MSCs can be frozen using standard cell freezing media and stored in liquid nitrogen indefinitely for later use.

2. Preparation of the ovarian cancer cells

1. Maintain OVCAR3-Luc cells in DMEM medium with 10% fetal bovine serum (FBS) and penicillin/streptomycin (penicillin: 100 units/mL, streptomycin: 0.1 mg/mL). Culture cells at 37 °C in a 5% CO₂ humidified incubator.
2. On the injection day, collect OVCAR3-Luc cells using the following steps.
   1. First, aspirate the media from the cell culture flask and wash the cells once with 5-7 mL of PBS per 75 cm² flask (PBS without Ca²⁺/Mg²⁺). Aspirate the PBS and add trypsin/EDTA solution (2 mL per 75 cm² flask). Then, return the cell culture flask to the incubator for 3-5 min or until the cells are detached.
   2. Using microscopy, ensure complete trypsinization of the cells (ensure that all the cells are detached from the culture flask). To halt trypsinization, add fresh serum-containing DMEM to the flask (2-3 mL per 75 cm² flask) and re-suspend the cells. Using a pipette, transfer the cells to a 15 mL conical tube.
   3. Centrifuge at 300 x g for 5 min. Remove the supernatant and re-suspend the cells with growth medium. Then, count the cells and re-suspend in growth medium at a density of 1 x 10⁸ cell/mL (0.5 x 10⁶ OVCAR3-Luc cells per injection). Place the suspended cells over ice.

3. Orthotopic inoculation

1. Preparation of cell suspension
   1. Mix CA-MSCs and OVCAR-Luc cells in a 1:1 ratio.
   2. Thaw the frozen basement membrane matrix in an ice bath. Then, add the cell suspension prepared in step 3.1.1 to the basement membrane matrix at a ratio of 1:1 (final cell concentration: 1 x 10⁶/ 20 µL of medium-basement membrane matrix) and blend thoroughly. The prepared cell suspensions should be kept in an ice bath until use.
      NOTE: Described here is a 1:1 CA-MSC to OVCAR-Luc ratio based on previous experience with the mentioned model. Previous work demonstrates that CA-MSCs are generally in a 1:10 ratio to tumor cells when quantified directly from patient tissue. As discussed below, starting with a 1:1 CA-MSC to tumor cell ratio yields an approximate 1:10 ratio of CA-MSC to tumor cells at both the primary and metastatic sites at experimental end points using this model.
2. Surgery
   NOTE: Surgery should be performed in sterile conditions and sterile surgical instruments must be used. It is recommended that female NSG mice at 8 weeks of age are used for this orthotopic model.
   1. Anesthetize the mouse using inhalation with isoflurane (4% for induction, 2% for maintenance) and oxygen (100%, flow meter 0.5-1 L/min). Place the mouse on the pre-surgical nose cone system. To deliver anesthesia throughout the surgery (see detailed instructions below), insert the mouse's head into the isoflurane vaporizer nose cone system.
   2. Subcutaneously inject the mouse with 5 mg/kg of carprofen, a pre-surgical analgesic, with an insulin syringe.
   3. Using clippers, shave the left caudal portion of the mouse between the costal margin and the femur. Sterilize the shaved area with povidone-iodine solution and alcohol swabs. Then, transfer the mouse to the surgical nose cone system and adjust the isoflurane to 1.5% for maintenance anesthesia.
   4. Make a 1-2 cm horizontal skin incision approximately 2-3 cm below the costal margin. Grasp the parietal peritoneum with forceps and elevate (the peritoneum will appear glossy and be characterized by deposits of adipose tissue). Make a 1 cm incision in the parietal peritoneum to open the abdominal cavity.
   5. Observe the collection of adipose tissue surrounding the mouse ovary (the ovarian bursa) upon peritoneal entry. Use curved forceps to grasp and expose the ovarian bursa. While firmly grasping the ovary with the forceps, inject 20 µL of the cell suspension prepared in step 3.1.2 into the bursa.
   6. Carefully return the ovary to the abdominal cavity. Using an absorbable polyglycolic acid 6-0 suture attached to a needle, reapproximate the edges of the parietal peritoneal incision. Close the skin incision with wound clips.
   7. Remove the mouse from the nose cone and place the mouse in a separate clean cage under warming light until it gains consciousness and maintains sternal recumbence. Then, return the mouse to its cage.
   8. Monitor the mice daily for 3-7 days following surgery. Place a cup of carprofen gel in the mouse cages for pain control. Remove the wound clips 7-10 days post-operatively.

4. Monitoring tumor growth and mouse body weight weekly

1. In Vivo Imaging System (IVIS): Starting from week one after the orthotopic inoculation, perform in vivo imaging every week for the next 3-4 weeks.
   1. Prepare the solution of D-luciferin (substrate of firefly luciferase) by dissolving 15 mg in 1 mL PBS and filtering through a 0.22 µm membrane for sterilization.
   2. Operate the IVIS following the manufacturer's instructions. Click on the Living Image Software desktop icon. Then, on the IVIS Acquisition Control Panel, select Initialize. This step can take up to 10-15 min. During this time, prepare the mice for imaging.
   3. Anesthetize each mouse using the isoflurane chamber connected to IVIS (2%-4% isoflurane and oxygen flow meter to 0.5-1 L/min). Inject the D-luciferin (100 µL/10 g body weight) via intraperitoneal injection using an insulin syringe.
   4. Place the luciferin-injected animal back into the isoflurane chamber.
      NOTE: It is important to record the injection time in order to keep the time consistent between luciferin injection and imaging performance throughout the study.
   5. Transfer the anesthetized mouse to the imaging chamber and set the imaging system to the appropriate settings. Then, acquire the image at the appropriate time.
   6. After saving the images, remove the mice from the imaging chamber and place them back into their respective cages.
      ​NOTE: CA-MSCs can also be labeled with a fluorescent tag and migration/co-metastasis of CA-MSCs and OVCAR3-Luc cells monitored in vivo using a fluorescent channel on the IVIS.

5. Assessing CA-MSC: tumor cell metastatic deposits

1. On reaching their pre-defined endpoint (weight gain >20% with presence of ascites, weight loss <10%, tumor ulceration, body conditioning score of 2 or less¹⁷), euthanize the mice in accordance with institutional animal welfare standards.
2. Perform necropsy on each mouse individually. Place an 18 G sterile needle intra-peritoneally to aspirate any ascites. Next, expose the thoracic and the abdominal cavity. Quantify the gross metastatic disease via visual inspection of tumor involvement of murine organs¹⁸.
3. Remove the tumor tissue using a sterile scalpel and either i) fix in paraformaldehyde for eventual paraffin embedding, ii) flash freeze and embed into OTC cryostat embedding media, iii) flash freeze for downstream DNA, RNA or protein analysis, or iv) dissociate into single cells for flow cytometry or FACS isolation.
   NOTE: CA-MSCs remain viable in this model and can be re-isolated for downstream applications. Depending on the goal of the study, the CA-MSC to tumor cell interactions can be investigated via immunohistochemical approaches, flow cytometric approaches, or maintaining re-isolated cells in culture.

The described approach closely mimics the supportive microenvironment of ovarian cancer (in particular, high grade serous carcinoma) by co-injection of patient-derived mesenchymal stem cells (CA-MSCs) and ovarian cancer cells into the ovarian bursa. First, CA-MSCs were isolated from primary, surgically resected human high grade serous ovarian cancer involving the omentum (all CA-MSCs used in this experiment were derived from the same patient). After 2 weeks of plating the tissue samples, CA-MSCs were tested to meet the criteria defined by the International Society for Cellular Therapy (ISCT) for mesenchymal stem cells/multipotent mesenchymal stromal cells. As CA-MSCs are patient-derived primary cells, slight differences in morphology can be observed between cell lines (Figure 1).

Following the orthotopic engraftment of OVCAR3-Luc tumor cells combined in a 1:1 ratio with CA-MSCs into the ovarian bursa of NSG mice, tumors formed with a 100% engraftment rate. Tumor engraftment was confirmed using an in vivo bioluminescent imaging system (IVIS) by day 7 post-injection. IVIS imaging was performed weekly to document tumor growth and the presence of metastases. Bioluminescent imaging revealed progressive tumor growth over time (Figure 2A). Serial imaging also demonstrated the formation of intra-abdominal metastases by day 14 post-injection, with increased metastatic disease burden over the experimental time course (Figure 2B). Further, mice experienced significant decreases in body weight during this experiment (Figure 3A). This weight loss is a surrogate marker of systemic illness and increased disease burden. By day 30 post-injection, the mouse cohort reached the endpoint criteria of greater than 10% loss of body weight. Ascites formation can compound the use of weight loss alone as an endpoint; therefore, additional endpoints including weight gain >20% with presence of ascites, weight loss <10%, tumor ulceration, and body condition scoring of 2 or less are also considered. In this experiment, the weight loss endpoint was reached first in all mice.

Upon reaching the experimental endpoint, necropsies were performed that confirmed aggressive, diffuse intra-abdominal metastasis in 100% of mice (based on visual inspection and immunohistochemical confirmation). Distant lung and parenchymal liver metastasis were confirmed in 80% and 60% of mice, respectively. Additionally, ascites was noted in 80% of mice (Figure 3B). In comparison, based on the previous work, OVCAR3-Luc cells injected without CA-MSCs demonstrate slower engraftment (though they do demonstrate 100% engraftment over time) and decreased metastasis formation (40% reduction in intra-peritoneal metastasis and no evidence of lung or parenchymal liver metastasis). Altogether, these results recapitulate the rapid, diffuse abdominal metastases and disease burden characterized by human ovarian cancer.

Figure 1: Representative examples of morphologic variation between cell lines of patient-derived CA-MSCs (scale bar = 100 µm, magnification at 10x). Panels (A-D) are representative pictures of CA-MSCs derived from four different high grade serous ovarian cancer samples. This demonstrates that while general morphologic similarities exist, each cell line has slight variations. Please click here to view a larger version of this figure.

Figure 2: Orthotopic ovarian tumor growth and progression. (A) Bioluminescent signal (photon flux per second) showed steadily increasing ovarian tumor growth over time. (n = 10, average and SEM). (B) Representative bioluminescent imaging demonstrating tumor growth and the development of intra-abdominal metastases. Please click here to view a larger version of this figure.

Figure 3: Orthotopic ovarian tumor metastases. (A) A significant reduction in average total body weight is demonstrated during the experimental time course (n = 10, average and SEM displayed). (B) Distribution of metastatic disease and ascites formation based on necropsy findings from 10 mice (percent of mice with metastasis is displayed). Please click here to view a larger version of this figure.

Despite significant clinical and research efforts, minimal headway has been made in the treatment and prevention of ovarian cancer¹. While a variety of mouse models have been used to study ovarian cancer progression and metastases, these models have faced significant limitations. In particular, previous mouse models have been unable to fully recapitulate the natural history of ovarian cancer progression, including the hallmark feature of early, diffuse intra-abdominal metastases^11,12. Recent studies have demonstrated that the ovarian cancer stromal microenvironment plays a critical role in disease progression and metastases⁶, and thus modeling ovarian cancer within a physiologically relevant stromal microenvironment would be a highly advantageous feature of an ovarian cancer mouse model.

Here, the study presents an orthotopic ovarian cancer mouse model that incorporates the ovarian stromal microenvironment via the use of carcinoma-associated mesenchymal stem cells (CA-MSCs). The results demonstrated that when patient-derived CA-MSCs are co-injected with ovarian cancer cells, the subsequent tumors acquire an aggressive phenotype, as reflected by early and diffuse intra-abdominal metastases in 100% of mice within 14 days post-injection. Moreover, the mouse cohort experienced a significant total body weight loss over the course of the experiment, which reached more than 10% at day 30 post-injection. In addition to intra-abdominal disease spread, this experimental design also led to late-stage disease findings that are analogous to those seen in humans, including ascites in 80%, lung metastases in 80%, and parenchymal liver metastases in 60% of mice, respectively.

Given CA-MSCs are obtained from patient tissue, some variability between CA-MSC lines is expected. However, CA-MSCs derived from high grade serous ovarian cancer have very similar promotion of ovarian cancer metastasis in this model regardless of the tissue source used for isolation (ovary, fallopian tube, or omentum). However, CA-MSCs derived from other histopathologic subtypes (particularly, clear cell ovarian cancer) demonstrate differential effects on ovarian cancer cells. For this model, we recommend using high grade serous derived CA-MSCs paired with high grade serous derived ovarian cancer cells.

Limitations of this model include obtaining access to fresh patient tumor tissue and difficulties in isolating and maintaining patient-derived CA-MSCs. Also, there is need for an invasive surgical procedure for tumor injection, as well as rapid progression of disease which requires frequent and close observation of the mouse cohort. However, since this experimental method accurately mimics the aggressive behavior of human ovarian cancer, it may be more advantageous than the previously described models. Moreover, this model provides a tool for the future study of the complex interactions between ovarian cancer cells and stromal cells in the tumor microenvironment, and their role in tumor progression and metastases.

The authors declare no competing interests.

We would like to thank the Gynecologic Oncology Biospecimen Program-Promark for help with tissue collection. LGC is supported by Tina's Wish Rising Star Grant and The Mary Kay Foundation.