Pancreatic cancer remains one of the toughest cancers to treat. Therefore, it is critical that pre-clinical models evaluating treatment efficacy are reproducible and clinically relevant. This protocol describes a simple co-culture procedure to generate reproducible, clinically relevant desmoplastic spheroids.
Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest cancers with a 5-year survival rate of <12%. The biggest barrier to therapy is the dense desmoplastic extracellular matrix (ECM) that surrounds the tumor and reduces vascularization, generally termed desmoplasia. A variety of drug combinations and formulations have been tested to treat the cancer, and although many of them show success pre-clinically, they fail clinically. It, therefore, becomes important to have a clinically relevant model available that can predict the response of the tumor to therapy. This model has been previously validated against resected clinical tumors. Here a simple protocol to grow desmoplastic three-dimensional (3D)-coculture spheroids is described that can naturally generating a robust ECM and do not require any external matrix sources or scaffold to support their growth.
Briefly human pancreatic stellate cells (HPaSteC) and PANC-1 cells are used to prepare a suspension containing the cells in a 1:2 ratio, respectively. The cells are plated in a poly-HEMA coated, 96-well low attachment U-well plate. The plate is centrifuged to allow the cells to form an initial pellet. The plate is stored in the incubator at 37 °C with 5% CO2, and media is replaced every 3 days. Plates can be imaged at designated intervals to measure spheroid volume. Following 14 days of culture, mature desmoplastic spheroids are formed (i.e. average volume of 0.048 + 0.012 mm3 (451 µm x 462.84 µm)) and can be utilized for experimental therapy assessment. Mature ECM components include collagen-I, hyaluronic acid, fibronectin, and laminin.
Pancreatic cancer's poor prognosis is associated with a variety of reasons, among which is its lack of easily detectable biomarkers leading to a late detection. Another major reason is the thick stroma surrounding the tissue, which leads to reduced blood supply. The deposition of large amounts of extracellular matrix (ECM), cell-cell interaction, endothelial cells, various immune cells, pericytes, proliferating myofibroblast, and fibroblast population, and the presence of non-neoplastic cells (together constituting the desmoplastic reaction)1, constitute the thick stroma that is responsible for PDAC's chemo and radiotherapeutic resistance2. Cancer and stromal cells have a complex, dynamic, and bidirectional interaction. Although some elements either attenuate or accelerate disease progression, most processes are adaptive during the tumor's development1. This provides an environment rich with growth factors, proangiogenic factors, proteases, and adhesion molecules. These factors promote angiogenesis, cell proliferation, metastasis, and invasion3,4. Together, they are immune and drug-privileged sanctuary for the tumor, resulting in drug resistance.
The desmoplasia is a complex mixture which consists of various ECM proteins, along with immune cells and pancreatic stellate cells (PSC). Together, these tend to form a scaffold for the cells to grow. PSCs are one of the largest components of the stromal compartment5. Their ability to produce enzymes like matrix metalloproteases (MMP), tissue inhibitors of matrix metalloproteases (TIMP) and cancer associated fibroblasts (CAF)6 imply they are likely to play a critical role in development of the desmoplastic reaction. The ECM, cancer-associated fibroblasts (CAF), and vasculature are the cardinal aspects of PDAC. Among CAFs, myofibroblast and inflammatory CAFs are speculated to be involved in active crosstalk responsible for pro-tumor properties7. The more extensive the fibroblastic formations on the tumor, the poorer the prognosis8,9,10.
Monolayer cell culture through established cell lines continues to remain a useful tool for analyzing drug toxicity and is a good starting point for proof of concept and discovery studies. Established cell culture lines, however, lack germline DNA and clinical relatability11. Since they are grown on flat surfaces, they undergo different in vitro selection criteria compared to when they are a part of the tumor, divide abnormally and lose their differentiated phenotype12. Overall, single cell cultures limit tumor heterogeneity and therefore lose clinical relevance. They are unable to accurately represent the complexity of the tumor's microenvironment (for e.g., the ECM). 3D culture can more closely replicate the complex tumor microenvironment.
3D culture was introduced in the 1970's for healthy cells and their neoplastic counterparts13. Several techniques have been used to study the morphology and architecture of malignant tissues through spheroids14. Co-cultures with stromal cells can model TME signals. An upregulation of EMT markers was seen when cells were co-cultured with stellate cells15. PDAC spheroids and their interaction with the stroma can be modelled by co-culturing with ECM components. Co-culturing specifically with PSCs have been reported to produce clinically relevant drug cytotoxicity data16,17,18. PSCs also aid drug resistance by evading apoptosis and stimulating proliferation of cancer cells through various paracrine factors19 and by inducing EMT transition. It, therefore, becomes critical to include the PSCs from an early stage in the criteria used to evaluate the success of a drug or drug delivery system. The PSC's ability to enhance proliferation and support faster growth in combination, compared to pancreatic cancer cells alone, has also been seen in vivo when subcutaneous flank injections of the two cell lines were evaluated in immunocompromised mice20.
The ability of a cell type to interact with ECM components is also critical to consider when growing co-cultured spheroids. BxPC-3 and PANC-1 have been reported to have equal affinities in binding to collagen. The two cell lines also bind equivalently to laminin, although there have been reports that BxPC-3 binds better21,22,23,24,25. In terms of migration, Stahle et al.26 demonstrated a 5x faster motility for PANC-1 cells as compared to BxPC-3. PANC-1 cells were also reported to migrate primarily as single cells, whereas BxPC-3 cells migrate as a tightly packed sheet. The choice of the cells also affects the size of the tumor25. BxPC-3 tumors were shown to be larger27,28 than those obtained from PANC-1, whereas one study demonstrated the opposite29 case. Despite their differences in size and motility, both cells have been reported to need long periods of latency to form tumors in mice. This duration can be especially long for BxPC-3 ranging from 4 weeks to 4 months25. However, there is also literature where BxPC-328 or BxPC-3 cancer stem cells30 have formed visible tumors quicker, implying there could be variation seen in tumor growth durations. The durations stated here should, therefore, only serve as an initial guideline for tumor growth rates.
BxPC-3 cells form spheroids with loose cells on the surface and dense cores, whereas PANC-1 cells have been reported to form both porous but robust spheroids31 as well as compact spheroids. PANC-1 cells have also been reported to be less differentiated and more aggressive32. Keeping the aggressive nature32 at the forefront, combined with the PANC-1 cells' higher motility, ability to form compact spheroids, and ability to interact with ECM components, PANC-1 cells were chosen for spheroid studies.
In the last few years, spheroid culture has seen a lot of success in demonstrating an advantage in its clinical relevancy compared to two-dimensional (2D) cultures. Its relevance has been leveraged in using this technique as a substitute to animal studies and to better understand the tumors' biology. The clinical relevance of spheroids, especially when co-cultured with PSC's has enabled their use to study various functions of the spheroid such as stiffness33, expression of TGF-β34,35,36,37,38, E-cadherin, F-actin18,34,36,37, α-SMA34,35,37,38, lactate dehydrogenase (LDHA)32, HIF-1α35,39, drug resistance16,37,40, cell migration41, cell invasion37, fibrosis35, radiation resistance42, phenotypical changes18, heterogeneity36, cellular levels of interactions39 and demonstrate ECM components37,38,39. Many of the protocols that were used to obtain the data described rely on Matrigel, the hanging drop method, printed molds, or other scaffolds to help support spheroid and ECM growth. The studies also usually involve the use of either non-human fibroblastic cells or freshly isolated stellate cells from patients. While using stellate cells is critical for the tumors to resemble in-vivo conditions, the inter-patient variability associated with fresh extractions makes these studies difficult to replicate.
This protocol aims to demonstrate a model that is easy to develop, reproducible, clinically relevant, and free of scaffolding, thereby relying exclusively on the co-cultures' abilities to naturally generate the ECM. To do this, a simple co-culture method involving a mix of PANC-1 cells (due to their natural tendency to migrate as single cells) along with human pancreatic stellate cells (HPaSteC) was chosen, due to their ability to behave like stem cells and be highly drug-resistant. Using the studies by Durymanov et al.38as a baseline, the protocol detailed below was established after further optimizing parameters such as cell ratios and durations between media changes. The spheroids resulting from this protocol can be used as a model system for new drug candidate evaluation40.
Additionally, for users not familiar with the spheroid culture, the work Peirsman et al.43discussing the development of the MISpheroID knowledgebase may be helpful. It establishes some minimum information guidelines that could help cope with heterogeneity between lab protocols. Although with some limitations, the work demonstrated that the choice of culture media, cell lines, spheroid formation method, and the final spheroid size are critical in determining the phenotypic properties of spheroids.
1. 2D cell culture
2. Poly(2 -hydroxyethyl methacrylate (poly-HEMA) solution coating for 96 well plate
3. 2D cell culture planning
4. 3D culture growth
Figure 1: Overview of the process to grow 3D desmoplastic pancreatic cancer spheroids (Generated using BioRender). The figure gives an overview of the basic processes involved; namely, trypsinzing cells, using the initial cell count to make dilute cell suspensions, preparing a co-culture using the diluted cell suspensions, adding cell suspensions to each well, incubating the cultures, performing media maintenance and final spheroid formation as expected on day 14. Please click here to view a larger version of this figure.
Figure 2: Structure of the U bottom well. The right image is an exaggerated shape to demonstrate the "halo" portion of the well. The figure aims to define where the "halo" portion of the well is as working above the halo is critical to growing the spheroids and avoiding accidental loss. Please click here to view a larger version of this figure.
5. Evaluation of ECM components and confocal microscopy
Three of the most critical steps involved in growing the spheroids are the initial cell count, the mixing steps while seeding the spheroids, and performing timely media changes to allow the spheroids to grow (Figure 1). Additionally, being familiar with Figure 2 on media changes after day 3 is critical to allow for effective media changes due to the increased media volume per well. When all these steps are performed according to the directions given, the spheroids (Figure 3) are found to grow up to an average volume of 0.048 ± 0.012 mm3 (451 µm x 462.84 µm) (Figure 4). Instances have been encountered where cells may grow better, leading to larger initial average volumes of 0.096 ± 0.014 mm3(611 µm x 560 µm) on Day 14. A variety of reasons may lead to the spheroids having larger initial volumes, such as maintenance of the incubator's temperature and humidity or the use of earlier cell passages, which have a slightly improved vitality compared to later passages.
Figure 3: A representative spheroid as seen on day 14 (size: 451.59 μm x 462.84 μm). Scale bar = 200 µm. Each spheroid is measured along its longest and shortest axis, respectively, and used the equation described in the methods section to calculate the volume. Please click here to view a larger version of this figure.
Figure 4: Spheroid growth progress over 17 days. Plates 1 and 2 are replicates. The error bars represent the standard deviation. The figure can be used as an approximate estimate of spheroid volumes that may be seen on days 6, 10, 14, and 17. Please click here to view a larger version of this figure.
However, in cases where the cells are not counted correctly or, cell suspension is not mixed thoroughly, or the final culture is not regularly maintained, or a combination of these errors occur, it may lead to the development of smaller spheroids or incompletely grown spheroids. Spheroids with similar appearances were also seen when initial optimization studies only supplemented media in wells around the periphery of the plate on Day 3 instead of the entire plate (Figure 5). Alternatively, in cases where the cells are not mixed well, or initial counting/ dilution errors lead to seeding more cells per well than the recommendation, multiple spheroids may be seen per well. A similar finding was seen when PANC-1:HPaSteC cells were used at a 240:120 ratio (Figure 6). This may also be seen occasionally when spheroids are grown per direction, but such spheroids are fewer in number (e.g., 4/96 wells may contain two spheroids per well instead of one). In instances where multiple spheroids are encountered, either avoid using the wells with multiple spheroids or make notes of changes in volumes when these spheroids are being treated. Improper media changes where the pipette tip dips below the halo (Figure 1) may also lead to accidental loss of the spheroids.
Figure 5: A representative image of a poorly grown spheroid on day 14. Scale bar = 200 µm. Here, the cells have not aggregated well, so it becomes difficult to measure the length and width of the spheroid. Please click here to view a larger version of this figure.
Figure 6: A representative image of a well containing two poorly grown spheroids as seen on day 14. Scale bar = 200 µm. Here, one of the spheroids has a size of 246.61 µm x 178.31 µm and the other 289.13 µm x 252.35 µm. This is an example where there is likely an issue with either initial dilutions or the quality of the cells that were used to start the experiment. Please click here to view a larger version of this figure.
The spheroids also develop an ECM with four of the primary ECM components: collagen-I, laminin, hyaluronic acid, and fibronectin. In studies the lab has published previously, the advantage of the co-culture over PANC-1 alone has been demonstrated in terms of homo PANC-1 spheroids displaying diffuse collagen, no fibronectin, and lesser hyaluronic acid deposition.
Figure 7: Representative image of the four primary ECM components as seen on day 17 in untreated spheroids: Collagen I, laminin, hyaluronic acid, and fibronectin. These images are obtained after immunostaining cryosections of the spheroid. The brightness of these images has been increased uniformly to help highlight differences in appearance. Areas where tissue is damaged or missing must be adjusted while intensity measurements are being performed. Scale bars = 100 µm. Please click here to view a larger version of this figure.
The duration and cell ratios chosen to grow the spheroids were based on studies as reported previously38. When attempting to optimize these studies by substituting NIH3T3 cells for HPaSteC cells, spheroid volumes and apoptosis patterns were found to closely resemble the reported optimized parameters (reported for PANC-1: NIH3T3:: 120:12) when PANC-1: HPaSteC ratios were at 120: 60. Although these studies only measure apoptosis until day 14, the protocol described in this process continues to use spheroids until day 17, since no visible increase in debris formation is observed between day 14 and day 17. As a control, all spheroids evaluated on day 17 always include an untreated arm to serve as a baseline for natural increases in the spheroids' volume and apoptosis.
The process of seeding and growing spheroids has 3 critical steps: Proper counting, mixing prior to the addition of the cells, and maintenance for the cultures. Since the process relies on using a small number of PANC-1 and HPaSteC cells, having the right count and the right dilutions becomes critical. This is particularly important for HPaSteC cells since these cells are smaller than PANC-1 cells, and it is easy to underestimate the cell number. Therefore, it is critical to count the cells carefully. Underestimating the count results in multiple spheroids being formed, whereas overestimating can lead to poorly formed spheroids.
It also becomes critical to mix the suspension well at all steps. Working with the expectation that cells may settle helps prevent undermixing and ensures uniform cell distribution. While preparing the spheroids for more than one plate (e.g., two plates), it is recommended that the number of back-and-forth mixes (step 4.4.1) be at least 20 between each well. This can be reduced to 15 as the first plate is halfway done, followed by 10 when beginning the next plate, and 5 as progress is made towards the second half of the second plate. Performing a stirring motion with the pipette in parallel to back-and-forth mixing helps improve the extent of mixing. An idea to consider is involving a second person while preparing two plates, as this increases speed and mixing efficiency. In this scenario, two people can use the hood at the same time with their individual suspensions and plates by working on the opposite corners of the hood.
Maintaining the cell cultures is necessary to ensure the spheroids always have an adequate nutrient supply. Media supplements on days 3 and 6 are most critical as this is where the cells are growing and need to be well fed. Media changes on days 6 and 10/11 can also be tricky since they involve collecting some supernatant to discard older media. This step must be done carefully by gently pulling media out until the "halo" and never beyond. This helps avoid media suction from too close to the spheroid and the unintentional loss of spheroids during the media change.
The method has a large advantage in that it is easily reproducible and does not require any mold or scaffolding material, unlike the methods we have cited earlier32,34,35,36,37,39. Additionally, the media used is standard cell culture media, which is relatively inexpensive as opposed to EGM-2 media39, mixtures of stellate cell media, and DMEM-FBS35or either media alone, depending on cell types34. This advantage becomes critical in translatability between labs and ease of setting up. The process can also be scaled up for multiple plates but would likely require one person for every two plates. The robust nature of the spheroids also allows for them to be evaluated under stirring conditions. This step is particularly helpful in performing cellular pharmacokinetic or dug release and absorption studies where spheroids are involved.
Despite its advantages, a limitation of the method is that it needs very thorough mixing during the seeding step. This can become time-consuming and physically strenuous. A possible way to bypass this could involve using a well-sterilized magnetic bead to maintain constant stirring conditions during the seeding process. Collecting the spheroids is also tricky and requires some practice. The spheroids also need time to grow and, therefore, require careful planning to ensure personnel and supply availability for media changes.
To conclude, a simple and reproducible method to produce clinically relevant desmoplastic 3D spheroids was developed. The spheroids are robust and capable of generating ECM components: collagen I, laminin, hyaluronic acid, and fibronectin. The simplicity involved in generating these spheroids carries a unique advantage in being easily reproducible and easily adaptable to changes such as including a third cell line. Another unique advantage of the model is that it is resistant to therapy40. This helps add to its clinical relevancy and ensures that only the most effective drugs are chosen to move forward and be utilized for treating pancreatic cancer. Further, these spheroids can be collected and implanted in mice to give reproducible desmoplastic tumors for in vivo analysis38and also used to examine the penetration of varying sizes of nanoparticles44as previously described.
The authors have nothing to disclose.
The work described was supported by the South Dakota Governors' Office of Economic Development, the South Dakota Board of Regents Competitive Research Grant Program (SD-BOR-CRGP), and the Department of Pharmaceutical Sciences at South Dakota State University for their support.
Axio Observer inverted microscope | Carl Zeiss | 0450-354 | |
Cellometer Auto T4 | Nexcelom Bioscience LLC | Auto-T4 | |
DMEM, powder, high glucose | Gibco | 12100046 | |
Donkey anti-sheep conjugated with Alexa Fluor 568 | Abcam | ab175712 | |
Fetal Bovine Serum | Cytiva | SH3091003HI | |
Goat antirabbit IgG labeled with Alexa Fluor 488 | Abcam | ab150077 | |
Hanks Balanced Salt Solution (HBSS) | Gibco | 14175145 | |
Human Pancreatic Stellate Cells (HPaSteC) | ScienCell | 3830 | |
Microscope Nikon | Nikon | Eclipse Ts 100 | |
Nunc 96-Well Polystyrene Round Bottom Microwell Plates | Thermo Scientific | 12-565-331 | |
Olympus Fluoview FV1200 confocal laser | Olympus | N/A | Discontinued product |
PANC-1 | ATCC | CRL-1469 | |
Poly-HEMA | Sigma | P3932 | |
Rabbit polyclonal anti-laminin antibodies | Abcam | ab11575 | |
Rabbit polyclonal anti-type I collagen antibodies | Abcam | ab34710 | |
Sheep polyclonal anti-hyaluronic acid antibodies | Abcam | ab53842 | |
Stellate cell media complete kit | ScienCell | 5301 | |
Trypsin | MP Biomedicals, LLC | 153571 | Trypsin solution prepared according to manufacturers protocol and used at 0.25%w/v |
Trypsin Neutralization Solution (TNS) | ScienCell | 103 |