We adapted a permeable microporous membrane insert to mimic the tumor microenvironment (TME). The model consists of a mixed cell culture, allows simplified generation of highly enriched individual cell populations without using fluorescent tagging or cell sorting, and permits studying intercellular communication within the TME under normal or stress conditions.
Understanding the early heterotypic interactions between cancer cells and the surrounding non-cancerous stroma is important in elucidating the events leading to stromal activation and establishment of the tumor microenvironment (TME). Several in vitro and in vivo models of the TME have been developed; however, in general these models do not readily permit isolation of individual cell populations, under non-perturbing conditions, for further study. To circumvent this difficulty, we have employed an in vitro TME model using a cell growth substrate consisting of a permeable microporous membrane insert that permits simple generation of highly enriched cell populations grown intimately, yet separately, on either side of the insert's membrane for extended co-culture times. Through use of this model, we are capable of generating greatly enriched cancer-associated fibroblast (CAF) populations from normal diploid human fibroblasts following co-culture (120 hr) with highly metastatic human breast carcinoma cells, without the use of fluorescent tagging and/or cell sorting. Additionally, by modulating the pore-size of the insert, we can control for the mode of intercellular communication (e.g., gap-junction communication, secreted factors) between the two heterotypic cell populations, which permits investigation of the mechanisms underlying the development of the TME, including the role of gap-junction permeability. This model serves as a valuable tool in enhancing our understanding of the initial events leading to cancer-stroma initiation, the early evolution of the TME, and the modulating effect of the stroma on the responses of cancer cells to therapeutic agents.
The tumor microenvironment (TME) is a highly complex system comprised of carcinoma cells that co-exist and evolve alongside host stroma. This stromal component typically consists of fibroblasts, myofibroblasts, endothelial cells, various immune components, as well as an extracellular matrix1. A significant constituent, often the majority of this stroma, are activated fibroblasts, frequently referred to as cancer-associated fibroblasts or carcinoma-associated fibroblasts (CAF)2,3. Unlike normal, non-activated fibroblasts, CAFs contribute to tumor initiation, progression, angiogenesis, invasion, metastasis, and recurrence4-11 in a wide variety of carcinomas, including breast, prostate, lung, pancreas, skin, colon, esophagus, and ovary5,6,12-17. Yet, the exact nature of the contribution of CAFs throughout cancer pathogenesis remains poorly defined. Furthermore, clinical evidence has demonstrated a prognostic value of CAFs, correlating their presence to high-grade malignancies, therapy failure, and overall poor prognosis10,18,19.
Clearly, enhancing our understanding of the initiating events in CAF development, as well as the intercellular communications mediating their role within the TME, may provide exciting new therapeutic targets and enhanced strategies that could improve patient outcomes. Towards this goal, several in vivo and in vitro models have been developed. While in vivo approaches are more reflective of patients' TME, they possess limitations, including the immense complexity and heterogeneity both within and between tumors. Furthermore, tumor samples from human subjects often represent highly developed TME and do not permit an understanding of the TME initiating events. Experimental animal studies offer some advantages, however generalization of animal data to humans should be done with caution due to differences in physiology between humans and animals such as rodents (e.g., thiol chemistry20, metabolic rate21, tolerance to stress22, etc.). Further, unlike the human population, which is genetically heterogeneous in nature, laboratory animals are typically bred to homogeneity. Also, it is often difficult to examine transient physiological variations and cell phenotype changes, as well as to control for specific experimental parameters using animals such as rodents. Thus, in vitro 2- and 3-dimensional (2D and 3D) tissue culture models are frequently utilized to advance the basic understanding of TME development. In spite of their lack of an accurate portrayal of the complexity of in vivo systems, these models offer advantages that greatly facilitate mechanistic investigations. In vitro models allow for a more simplified, focused, and cost-effective analysis of the TME, whereby statistically significant data can be generated in cells free of systemic variations that arise in animals.
There are several varieties of in vitro systems. The two most commonly used TME in vitro models consist of mixed monolayer or spheroid cell cultures. Both culture methods are advantageous for basic studies of intercellular interactions (e.g., normal cells with tumor cells) and for the analysis of various TME specific cell phenotype changes (e.g., emergence of cancer-associated fibroblasts from normal fibroblasts). Additionally, the spheroids are able to create a more reflective tissue-like structure of the TME, and can be representative of tumor heterogeneity23. However spheroids often produce widely varying oxygen tension gradients across layers, which may complicate experimental conclusions24. Unfortunately, both models are extremely limited in their ability to isolate pure cell populations for further characterization and study following co-culture. To do so would require at least one cell type to be fluorescently-tagged or labeled with an identifying maker, and then subjecting the mixed co-culture to extensive processing and cell sorting to separate the cell populations. While a cell sorter is capable of isolating a rather pure cell population, one must be cognizant of cellular stress and potential microbial contamination risks25.
To facilitate the understanding of intercellular communication, great efforts have been devoted towards developing and optimizing in vitro systems that closely mimic the in vivo environment, while permitting a simplified approach. One such tool is the permeable microporous insert, a membrane substrate that was first developed in 195326 and subsequently adapted for diverse applications and studies (e.g., cell polarity27, endocytosis28, drug transport29, tissue modeling30, fertilization31, bystander effect32,33, etc.). This system permits the growth of cells with in vivo-like anatomical and functional differentiation, as well as expression of many in vivo markers34,35 that are not observed when cultured on impermeable plasticware. Furthermore, the extremely thin porous membrane (10 µm thick) permits rapid diffusion of molecules and equilibration times, which simulates the in vivo environment and permits independent cellular functioning at both the apical and basolateral cell domains. An additional advantage of the insert's utility as a TME system is its physical separation of two heterotypic cell populations grown on either side of the membrane in the same environmental conditions, while maintaining various modes of intercellular communication through the membrane pores. Though physically separated, the two cell populations are metabolically coupled via secreted elements and, as described here, also through gap-junctional channels. Additionally, by maintaining the inserts at in vivo partial oxygen tension (PO2), the model reduces the complications of oxygen and chemical gradients observed in other systems. Rather, it increases the understanding of natural mechanisms controlling the TME. Notably, the two cell populations can be easily isolated with high purity, without fluorescent tagging and/or cell sorting following extended periods of co-culture.
Here we describe an in vitro TME protocol consisting of human breast carcinoma cells and human fibroblasts grown, respectively, on either side of a permeable microporous membrane insert, but yet in continuous bi-directional communication through the membrane pores. We show that by using membranes with different pore sizes, the contribution of a specific type of intercellular communication (e.g., secreted factors versus gap junctions) to the development of the TME can be investigated.
1. Preparation of Culture Media and Cells
2. Preparation of Inserts
NOTE: To ensure sterile conditions, work in a laminar flow biological safety cabinet dedicated to cell culture.
3. Collecting Cells from Insert by Trypsinization
NOTE: In addition to the ease of obtaining enriched cell populations, the insert TME model also allows for similar experimental treatments (e.g., incubation with chemicals, exposure to oxygen conditions that are above or below ambient atmosphere, ionizing radiation treatment, etc.) as other in vitro TME co-culture models. Furthermore, the permeable insert co-culture substrate can be analyzed by procedures already developed for standard 2D tissue culture models. For example, cells can be harvested from either side of the insert membrane to obtain highly enriched populations, which can then be utilized for analysis of endpoints (e.g., in situ immuno-detection, Western blotting) or propagated for subsequent experiments36.
4. Characterization of Cell Purity by Flow Cytometry
NOTE: To characterize the ability of the permeable microporous membrane inserts to maintain the purity of the two cell populations (i.e., top and bottom) for up to 120 hr, sections 1-3 were performed, with green fluorescent protein (GFP)-tagged MDA-MB-231 cells being plated on the top side of the insert, and non-GFP-tagged AG1522 cells being plated on the bottom side of 0.4 µm-, 1 µm-, or 3 µm pore inserts.
5. Characterization of Cells by In Situ Immunofluorescence
6. Characterization of Cells by Western Blot
NOTE: The Western blot procedure is described elsewhere38. A brief outline is described here:
7. Characterization of Intercellular Communication by Flow Cytometry
NOTE: To characterize the adaptability of the permeable microporous membrane inserts to examine different modes of intercellular communication (e.g., secreted factors, gap junctions). Sections 1-2.12 were performed with non-fluorescent AG1522 cells being plated on the bottom side of 0.4 µm-, 1 µm-, or 3 µm-pore inserts.
Here we adapted a permeable microporous membrane insert to develop an in vitro heterotypic cell co-culture system that mimics the in vivo tumor microenvironment (Figure 1). This system allows for two different cell populations to be grown on either side of the insert's porous-membrane for extended periods of time (up to 120 hr, in our use). Importantly the system is capable of maintaining the purity of the cell populations, as determined by plating GFP-tagged MDA-MB-231 cells on the top side of 0.4-, 1-, or 3 m-pore inserts and allowing them to grow in co-culture with AG1522 cells growing on the bottom side of the insert, for 120 hr. After this time, the cells were collected from the bottom side of the insert and analyzed by flow cytometry to identify the presence of any GFP-positive MDA-MB-231 cells, which would indicate a loss of cell population purity. Analysis revealed 99.9% and 99.8% pure populations from 0.4 and 1 µm-pore inserts, respectively; however, the insert with 3 µm pores was unable to maintain the separation of the cell populations, as the pores were large enough to allow a significant number of the GFP-positive MDA-MB-231 cells to migrate across the membrane (Figure 2).
Importantly, through in situ immunofluorescence and Western blot analysis we were able to demonstrate the capability of this system to effectively generate cancer-associated fibroblasts from normal human diploid AG1522 fibroblasts following co-culture with MDA-MB-231 breast cancer cell, as determined by reduced expression of Caveolin-1, a well-established CAF biomarker39-42 (Figure 3). CAV1 is a scaffolding protein that forms the caveolae in the lipid raft domain of the plasma membrane, which is involved in multiple cellular processes43. In the Western blot, two distinct bands were observed, corresponding to the α and β isoforms of CAV144. Unfortunately, little is known about the difference(s) between the isoforms, especially their role as CAF biomarkers.
We have also shown the proficiency of the insert co-culture system to modulate intercellular communication between the two heterotypic cell populations (Figure 4). Non-labeled, gap-junction competent human diploid AG1522 fibroblasts were cultured to confluency on the bottom side of the insert. A second set of AG1522 cells were loaded with Calcein AM, a green fluorescent gap-junction permeable dye, and plated on the top side of the insert. The two cell populations were allowed to communicate for 6 hr, and then the cells on the bottom side of the insert were isolated and analyzed by flow cytometry. Calcein-positive cells from the bottom side of the insert would be reflective of cells that established gap-junctional coupling through the insert pores. The size of the 0.4 µm pore insert limited functional gap junction formation between cells growing on either side of the insert, thus constraining communication to secreted factors (e.g., growth factors, microvesicles, etc.). Alternatively, the size of the 1 µm and 3 µm pores are apparently large enough to allow functional coupling of the cells through gap junctions. Therefore, by modulating the pore size, one can begin to understand the diverse role(s) that different modes of intercellular communication may have on TME development and evolution.
Figure 1: Permeable Microporous Membrane Insert Model Scheme. (a) Apparently normal human diploid AG1522 fibroblasts (low passage cells) destined to become CAFs are seeded onto inverted permeable microporous inserts consisting of a 10 µm thick polyester membrane with 0.4 µm, 1 µm, or 3 µm pores. Following attachment, the inserts are inverted and placed into the wells of plates and cultured to confluency. (b) Cancer cells and control fibroblasts are grown separately in flasks prior to being seeded on the top side of the inserts with fibroblasts growing intimately on the bottom side. (c) The co-cultured cells are fed every other day and maintained for the desired time. (d) At the respective times following co-culture and/or experimental treatments, the CAFs (bottom side of the inserts) and/or cancer cells (top side of the insert) are carefully isolated by trypsinization, yielding highly pure cell populations, which can be used for analysis of endpoints or propagated for subsequent studies. Please click here to view a larger version of this figure.
Figure 2: Generation of Highly Enriched Cell Populations. Representative flow cytometry results demonstrate the ability of the insert to maintain highly enriched, yet independent heterotypic cell populations following 120 hr co-culture. (a) AG1522 cells harvested from the bottom side of the 0.4 µm pore insert show 99.9% of the population to be free of GFP-labeled MDA-MB-231 cells growing on the top side of the insert (lower left quadrant). (b) AG1522 cells harvested from the bottom side of the 1 µm pore insert also show 99.8% enriched population (lower left quadrant). (c) AG1522 cells harvested from the bottom side of the 3 µm pore insert reveal that only 68.5% of the population is free of GFP-labeled cells (lower left quadrant), indicating that a large number of GFP-positive MDA-MB-231 cells were able to migrate through the 3 µm pores. Please click here to view a larger version of this figure.
Figure 3: Generation of Cancer-Associated Fibroblasts. Reduced expression of CAV1 is a consistent and reliable CAF biomarker. (a) Microscopic images of in situ immunofluorescence (scale bar = 20 µm), and (b) Western blot analysis of CAV1 in AG1522 cells that had been in co-culture with AG1522 cells (control) (left), or with MDA-MB-231 breast cancer cells (right) for 120 hr. Staining with Ponceau S Red was used for loading control and to determine fold change in Western blots. The results are representative of three separate experiments. Please click here to view a larger version of this figure.
Figure 4: Modulation of Intercellular Communication by Insert Pore-Size. Flow cytometry analyses demonstrating the adaptability of the permeable microporous membrane insert TME model to select for different modes of intercellular communication between Calcein AM loaded AG1522 cells growing on top of the insert and AG1522 control cells growing on its bottom side. (a) Cells harvested from the bottom side of 0.4 µm pore inserts show very low levels of Calcein-positive cells (0.57%), indicating limited gap-junction communication through the insert pores. (b) Cells harvested from the bottom side of 1 µm pore inserts show higher levels of Calcein-positive cells (9.98%), indicating formation of functional gap junctions. (c) Cells harvested from the bottom side of 3 µm pore inserts show high levels of Calcein-positive cells (87.8%), indicating extensive gap-junctional communication. Please click here to view a larger version of this figure.
The protocol described here is a simple, adaptable in vitro procedure (Figure 1) that utilizes a permeable microporous membrane insert to generate highly enriched individual cell populations from a co-culture of heterotypic cells. Significantly, the model is suitable for investigating various modes of intercellular communication. The critical steps include selecting the appropriate pore-size insert for specific experimental interest(s), seeding the first cell population on the bottom side of the insert in medium supplemented with 50% FBS, seeding the second cell population on the top side of the insert, and co-culturing the two distinct cell populations for an appropriate duration of time. We show that this system is capable of mimicking various in vivo characteristics, thus providing a greater opportunity to understand the molecular, phenotypic, and latent changes occurring during early TME evolution, which addresses a major disadvantage of many established TME in vitro models (Figure 2).
The model can be easily modified to permit investigation of various aspects of cell-cell interactions in cancer-stroma development and TME evolution (Figure 4). While the 0.4 µm pore insert permits intimate coupling between the two cell populations45, it significantly restricts formation of functional gap junctions between cells grown on either side of the insert membrane (Figure 4). Therefore, thus facilitating investigation of the role of diffusible factors (e.g., growth factors) and/or extracellular vesicles (e.g., exosomes), which mediate CAF development46-48. Alternatively, the 1 µm and 3 µm pores are large enough to allow formation of functional gap junctions, thus permitting the study of both direct metabolic coupling via junctional channels and indirect communication through diffusible secreted factors. However, as the 3 µm pore insert permits cell migration through the pores (see Figure 2), it does not maintain the purity of two separate heterotypic cell populations during extended co-culture times (e.g., 120 hr). Whereas the present model enables the study of several aspects of the TME, it is limited in its ability to account for many of the complex physiologic interactions that occur within an in vivo TME (e.g., vascular fluid exchanges, etc.).
Understanding the early events of cancer-stroma activation and development of the TME is important in elucidating the steps implicated in primary tumor initiation and progression, as well as metastasis. It is now widely accepted that the cancer stroma, especially CAFs, possess a critical role in supporting carcinogenesis (reviewed in 49). This has led to the development of many in vitro 2D and 3D models, aimed at illuminating strategies to target the early mechanisms contributing to stromal-activation and tumor progression. Unfortunately, many of these models are limited by their inability to permit isolation of individual cell populations, without time-consuming and stressful processing, for further study. Thus this protocol provides a significant addition to the field of TME research, as it combines in vivo like properties with the ability to easily and rapidly obtain highly enriched or pure cell populations for immediate analysis or propagation for subsequent studies.
Once the technique is mastered, the insert TME model can be utilized for investigating diverse TME bi-directional interactions (e.g., cancer to endothelial, cancer to immune cells, etc.). Additionally, the complexity of the co-culture can be enhanced by placing a mixed cell population on one side of the insert and a single population on the alternate side, thus obtaining a more in vivo like TME cell population complexity. However, caution should be exercised in selecting an insert with the proper pore-size if using cells of especially small diameter (e.g., human lymphocytes with a diameter of ~7 µm50), as they may be capable of migrating through the smaller pores (e.g., 0.4 µm or 1 µm). Prior to co-culture, migration experiments (Figure 2) may be warranted to ensure maintenance of pure cell populations throughout extended co-culture experiments.
This model is also amenable for investigating the effects of various therapeutic challenges (e.g., ionizing radiation, chemotherapeutics, hyperthermia) and physiological alterations (e.g., hypoxia, hyperoxia, biologic inhibitors) in mixed cell populations that are coupled across the permeable microporous membrane insert. The model has demonstrated its ability to successfully generate CAFs, identified by a widely accepted CAF biomarker (i.e., down-regulation of CAV1)39-42 (Figure 3), which was further reduced by 55% when the insert system was maintained at in vivo like PO2 (7 mm Hg; 0.5% O2)51, highlighting the model's adaptability for various experimental designs and conditions (data not shown). In sum, this model provides a simplified approach and an opportunity to modulate temporal and external factors in order to obtain deeper insights into cancer-stroma activation, the early evolution of the TME, and cellular responses to therapeutic or environmental agents.
The authors have nothing to disclose.
This research was supported by grants from the New Jersey Commission on Cancer Research (Pre-Doctoral Fellowship DFHS13PPCO17), the National Institutes of Health (CA049062), and the National Aeronautics and Space Administration (NNX15AD62G).
For Cell Culture | |||
AG01522 (i.e., AG1522) human diploid fibroblast | Coriell | 107661 | Passage 8-13 |
MDA-MB-231-luc-D3H1 breast adenocarcinoma cell line | PerkinElmer | 119261 | Parental line: ATCC (#HTB-26) |
MDA-MB-231/GFP breast adenocarcinoma cell line | Cell Biolabs | AKR-201 | |
Eagle's minimal essential medium (MEM) | Corning Cellgro | 15-010-CV | |
Fetal Bovine Serum (FBS), Qualified | Sigma | F6178-500mL | |
Corning Glutagro Supplement (200mM L-alanyl-L-glutamine) | Corning Cellgro | 25-015-Cl | |
Penicillin Streptomycin Solution, 100X | Corning Cellgro | 30-002-Cl | |
Transwell Insert (i.e., permeable microporous membrane insert) (0.4 μm pore) | Costar | 3450 | |
Transwell Insert (i.e., permeable microporous membrane insert) (1 μm pore) | Greiner bio-one | 657610 | |
Transwell Insert (i.e., permeable microporous membrane insert) (3 μm pore) | Costar | 3452 | |
6-well Culture Plate | Greiner Bio-One Cellstar | 657160-01 | |
75 cm2 cell culture flask | CellStar | 658 170 | |
Phosphate-Buffered Saline (PBS), 1X | Corning Cellgro | 21-040-CV | without calcium & magnesium |
0.25% (vol/vol) Trypsin, 2.21 mM EDTA, 1X | Corning Cellgro | 25-053-Cl | |
15 mL Centrifuge Tube | CellTreat | 229411 | |
35 x 10 mm Cell Culture Dish | Greiner bio-one | 627 160 | |
Name | Company | Catalog Number | Comments |
For Immunofluorescent Microscopy | |||
Mouse anti-Caveolin 1 | BD Transduction Laboratories | 610406 | In situ Immunofluorescence – 1:5000 |
Goat anti-Mouse IgG (H+L) Secondary Antibody, Alexa Fluor 488 conjugate | ThermoFisher Scientific | A-11029 | In situ Immunofluorescence – 1:2000 |
Bovine Serum Albumin – Fraction V | Rockland | BSA-50 | Immunoglobulin and protease free |
16% (wt/vol) Formaldehyde Solution | ThermoFisher Scientific | 28908 | Dilute to 4% with 1X PBS |
Premium Cover Glass (22×22 mm No.1) | Fisher | 12548B | |
Triton X-100 | Sigma | T8787-50ML | |
SlowFade Gold antifade reagent with DAPI | Invitrogen | S36938 | |
Name | Company | Catalog Number | Comments |
For Flow Cytometric Analysis | |||
Calcein, AM | Molecular Probes | C3100MP | |
Hanks' Balanced Salt Solution (HBSS) | Gibco | 14025-076 | |
Name | Company | Catalog Number | Comments |
For Western Blot Analysis | |||
Mouse anti-Caveolin 1 | BD Transduction Laboratories | 610406 | Western Blot – 1:10000 |
Tween-20 | BioRad | 170-6531 | |
Nitrocellulose Membrane (0.2 μm) | BioRad | 162-0112 | |
Western Lightning Plus-ECL | PerkinElmer | NEL104001EA | |
BioRad DC Protein Assay | BioRad | 500-0116 | |
Sodium dodecyl sulfate (SDS) | BioRad | 161-0302 | |
Sodium deoxycholate monohydrate (DOC) | Sigma | D5670 | |
IGEPAL CA-630 (NP40) | Sigma | I8896 | |
30% Acrylamide/Bis Solution, 37.5:1 | BioRad | 161-0158 |