We developed an in vitro model of dormancy in the bone marrow for estrogen-sensitive breast cancer cells. The goal of this protocol is to demonstrate use of the model for the study of the molecular and cellular biology of dormancy and for generation of hypotheses for subsequent testing in vivo.
The study of breast cancer dormancy in the bone marrow is an exceptionally difficult undertaking due to the complexity of the interactions of dormant cells with their microenvironment, their rarity and the overwhelming excess of hematopoietic cells. Towards this end, we developed an in vitro 2D clonogenic model of dormancy of estrogen-sensitive breast cancer cells in the bone marrow. The model consists of a few key elements necessary for dormancy. These include 1) the use of estrogen sensitive breast cancer cells, which are the type likely to remain dormant for extended periods, 2) incubation of cells at clonogenic density, where the structural interaction of each cell is primarily with the substratum, 3) fibronectin, a key structural element of the marrow and 4) FGF-2, a growth factor abundantly synthesized by bone marrow stromal cells and heavily deposited in the extracellular matrix. Cells incubated with FGF-2 form dormant clones after 6 days, which consist of 12 or less cells that have a distinct flat appearance, are significantly larger and more spread out than growing cells and have large cytoplasm to nucleus ratios. In contrast, cells incubated without FGF-2 form primarily growing colonies consisting of >30 relatively small cells. Perturbations of the system with antibodies, inhibitors, peptides or nucleic acids on day 3 after incubation can significantly affect various phenotypic and molecular aspects of the dormant cells at 6 days and can be used to assess the roles of membrane-localized or intracellular molecules, factors or signaling pathways on the dormant state or survival of dormant cells. While recognizing the in vitro nature of the assay, it can function as a highly useful tool to glean significant information about the molecular mechanisms necessary for establishment and survival of dormant cells. This data can be used to generate hypotheses to be tested in vivo models.
Breast cancer cells metastasize to the bone marrow before the disease is detectable1, as soon as small tumors develop blood vessels2,3. The metastatic process is rapid but inefficient. Cells enter the new blood vessels rapidly, at millions per day4 but few survive the trip to distant organs5. Nevertheless, some micrometastases survive in the bone and can be found as single cells or small cell clumps in bone marrow aspirates from newly diagnosed patients1. These cells resist adjuvant chemotherapy, which is administered for the very purpose of eliminating them6. This resistance is endowed, substantially, by survival signaling initiated by interactions with the bone marrow microenvironment7,8. Micrometastases can be found in about one third of women with localized breast cancer and represent an independent indicator of survival when analyzed by univariate analysis9. Some micrometastases are growth initiated, but recurrence patterns depend on cell type. Patients with triple negative breast cancer tend to recur between 1 to 4 years, suggesting poor control over the dormant state. Other cell types, including ER/PR+ cells, can remain dormant for up to 20 years, with a steady, continuous rate of recurrence10. While differences in dormancy gene expression signatures between ER+ and ER- breast cell lines and tumors reflect different dormancy potentials11, interactions with bone marrow stroma likely represent a significant contribution to dormancy.
The study of dormancy in vivo is exceptionally difficult because micrometastases are rare and are outnumbered by hematopoietic cells by more than 106-fold. Hence, relevant models must be generated that provide in vitro data that can suggest mechanisms and generate testable hypotheses in vivo. A number of dormancy models, including mathematical models12,13, in vitro models7,8,14,15, in vivo xenograft models16, combinations of in vitro and xenograft models17,18 and spontaneous tumor and metastasis models19, have yielded some insight into cancer cell dormancy20. Each of these models have their own limitations and are of themselves primarily useful for generating hypotheses regarding molecular signaling and interactions that govern dormancy to be tested in more biologically relevant models.
With the overall goal of defining the molecular mechanisms of dormancy, the interactions with the microenvironment that results in cycle arrest, redifferentiation and therapeutic resistance and mechanisms that result in recurrence in ER+ cells, we developed an in vitro model that provides selected relevant elements of the stromal microenvironment7. This model, while relatively sparse in its components, is sufficiently robust to permit investigators to derive specific molecular mechanisms that affect significant functions of dormancy. These experiments generate hypotheses that can be directly tested in vivo. The model relies on a few key elements that we demonstrated to be relevant in dormancy. They include the use of estrogen-dependent breast cancer cells, culture of cells at a clonogenic density where their interaction is primarily with the substratum and soluble components of the medium, a fibronectin substratum and the presence of basic fibroblast growth factor (FGF-2) in the medium.
We characterized mechanisms that govern the system in vitro, including the induction of cell cycle arrest by FGF-221, mediated through TGFβ22, survival signaling through PI3 Kinase7,8 and ERK8 and morphogenic differentiation to an epithelial phenotype, which depended on RhoA inactivation, integrin α5β1 upregulation and ligation of stromal fibronectin for survival7,15 (Figure 1). The in vitro cell cycle effects of FGF-2 on MCF-7 cells begin at concentrations at least one log below 10 ng/ml21,23. The rationale was based on the temporal control of FGF-2 expression governing mammary ductal morphogenesis, cyclic expansion and recession in a number of mammalian systems24-27. We demonstrated that FGF-2 induces differentiation, including ductal morphogenesis in 3D culture28, and that FGF-2 expression are generally lost with malignant transformation of human tumors29. The expression of FGR1 remained intact in breast carcinomas surveyed29 and MCF-7 cells continue to express all 4 FGF receptors30. In the context of dormancy, FGF-2 is exported by and heavily deposited on bone marrow stroma31,32 where it functions in the preservation of hematopoietic stem cells33. We demonstrated that FGF-2 induces a dormant state in ER+ breast cancer cells cultured on fibronectin substrata, also abundant in the marrow, where it induces morphogenic differentiation7. In the model, breast cancer cells are growth inhibited, inactivate Rho A through the RhoGap GRAF, redifferentiate to an epithelial phenotype and re-express integrins α5β1 lost with malignant progression. They bind fibronectin through integrin α5β1 and activate survival signaling that render them resistant to cytotoxic therapy7,8,15 (Figure 1). Inhibition of Rho class GTPases has been demonstrated previously to induce a dormant phenotype34.
Here we will outline the specific procedures that will permit investigators to establish the model and study specific molecular and cellular mechanisms governing dormancy of ER+ breast cancer cells. In the experiments presented here to illustrate the use of the model, we targeted the PI3K pathway (Figure 1B) with an Akt inhibitor and a PI3K inhibitor and all members of the Rho family (Figure 1B) with a pan-Rho inhibitor and a Rho Kinase (ROCK) inhibitor.
1. Clonogenic Assay
2. Clonogenic Incubation for Immunofluorescence Studies
3. Clonogenic Incubation for Molecular Studies
Experiments were conducted to recapitulate the assay. The time course of the experiment is shown in Figure 2A. Cells are incubated at clonogenic density on day -1, FGF-2 in fresh medium is added on day 0 and cells are cultured until day 6 when they are stained and colonies are counted. Any perturbations to the system are administered on day 3 in 100 μl volumes at 10x final concentrations desired. Figure 2B demonstrates the typical appearance of growing and dormant colonies. Growing colonies contain >30 cells and dormant colonies contain 12 or less cells which are many times larger than growing cells with large cytoplasm/nucleus ratios. Figure 2C demonstrates a typical experimental result conducted in quadruplicate. The predominant distribution of cells without addition of FGF-2 is represented by growing clones while, in the presence of FGF-2, the vast majority of clones are dormant.
Figure 3 is an example of an experiment in which perturbing agents were added on day 3. The figure represents the dose-dependent inhibition of dormant clones by inhibition of the Rho family of GTPases by a pan-Rho inhibitor C3 transferase and a Rho kinase (ROCK) inhibitor Y27632. The assay can assess the ED50 of added inhibitors on dormant clones as well as growing clones. In this experiment, the effect on growing clones was not determined, since it was only presented to illustrate the methodology.
Figures 4 and 5 demonstrate that the assay can be used to assess the combined effects of inhibitors on survival of dormant clones. Our prior studies have demonstrated that the PI3K pathway undergoes a sustained activation in dormant cells in this model and inhibition of both PI3K and Akt partially inhibits the survival of dormant clones7. Here, preliminary observations, demonstrated that nonspecific inhibition of the Rho family of small GTPases also partially inhibits survival of dormant clones. Data presented in Figures 4 and 5 demonstrate that combining inhibition of PI3K and one of its downstream effectors, AKT, with the inhibition of the Rho family with C3 transferase and a ROCK inhibitor can almost completely eliminate the survival of dormant clones in some circumstances. We have previously demonstrated that dormant clones surviving the inhibition of PI3K but not of Akt are comprised of cells that no longer exhibit the dormant phenotype, but rather appear mesenchymal and distressed7. Cells in dormant clones in which the Rho family has been broadly inhibited by C3 transferase and the ROCK inhibitor also lose the typical dormant appearance, assume fibrolastoid appearances and ruffled membranes and also appear distressed (Figure 6). These experiments demonstrate a model that can be queried in a very broad manner to achieve an understanding of the elements that govern dormancy and resistance to therapy.
Figure 1: Mechanisms governing our in vitro dormancy model. (A) Growing and dormant MCF-7 clones after 6 day incubation at clonogenic density on fibronectin-coated plates with and without FGF2 10 ng/ml7 (100X magnification). (B) Summary schema outlining the data that FGFR and integrin α5β1 parallel steady state signaling is required to activate and maintain dormancy7. FGF-2 upregulates cyclin dependent kinase inhibitors resulting in G1 arrest21, activates ERK8 and PI3 kinase7 that initiatesurvival signaling and upregulate integrin α5β17, which reaches steady state after several days. Dual signaling through FGFR through PI3K and independently through ligation of integrin α5β1 by fibronectin are required for activation of FAK and membrane localization and activation of the RhoA GAP GRAF15. This results in inactivation of RhoA and a permissive steady state for cortical rearrangement of F-actin (phalloidin-stained photomicrograph), epithelial re-differentiation and dormancy15. Please click here to view a larger version of this figure.
Figure 2: Elements of the in vitro dormancy assay. (A) Schema of the temporal components of the dormancy assay. Cells are incubated on fibronectin coated tissue culture plates at clonogenic density, at a maximum density of 1,500 cells/well of a 24 well plate in an incubator at 37 °C and 5% CO2 on day 1. The medium is replaced on day 0 with fresh medium or fresh medium containing FGF-2 10 ng/ml and the cells are re-incubated. Cells are stained with crystal violet solution, as described, on day 6. Any inhibitors or perturbation agents are added on day 3 in 100 μl volumes at 10x desired concentrations and cells are re-incubated until day 6. (B) The appearance of stained growing and dormant MCF-7 colonies. Growing colonies contain >30 cells and dormant colonies contain 12 or less cells. Dormant cells are pancake shaped, many times larger than growing cells with large cytoplasm/nucleus ratios. (100X magnification). (C) Graph demonstrating the clonogenic potential of 1,500 MCF-7 human breast cancer cells incubated with and without FGF-2 10 ng/ml on fibronectin-coated 24 well plates. Without FGF-2, the predominant distribution of cells is represented by growing clones while in the presence of FGF-2 the vast majority of clones are dormant. Experiments were done in quadruplicate. (Error bars are + S.D.)
Figure 3: Dose-dependent inhibition of dormant clones by pan-Rho family inhibitors C3 transferase and the ROCK inhibitor Y27632. T-47D cells were incubated at the clonogenic density of 1,000 cells per well on 24 well fibronectin-coated plates with FGF-2 10 ng/ml for 6 days. Media, C3 transferase 3 or 5 μg/ml, and the ROCK inhibitor Y27632 0.1, 1 or 10 μg/ml were on day 3 and dormant clones were counted on day 6. The graph demonstrates a dose-dependent inhibition of stained dormant clones on day 6. The ED50 of C3 was approximately 3 μg/ml while that of the ROCK inhibitor was between 1 and 10 μM in this assay. Experiments were done in quadruplicate. (Error bars are + S.D.)
Figure 4: Combined effects of pan-Rho family inhibitor C3 with an Akt inhibitor or with a PI3 kinase inhibitor on survival of dormant T-47D clones. T-47D cells were incubated at the clonogenic density of 1,000 cells per well on 24 well fibronectin-coated plates with FGF-2 10 ng/ml for 6 days. Media, C3 transferase 5 μg/ml, Akt inhibitor 25 μM and LY294002 20 μM were added individually or in combination on day 3 and dormant clones were counted on day 6. The combined effects of C3 and the AKT inhibitor appear to be additive at these concentrations but the combined effects of C3 and the PI3K inhibitor appear to be synergistic in these experiments on inhibition of dormant clone survival. Experiments were done in quadruplicate. (Error bars are + S.D.)
Figure 5: Combined effects of ROCK inhibitor Y27632 with an Akt inhibitor or with a PI3 kinase inhibitor on survival of dormant T-47D clones. T-47D cells were incubated at the clonogenic density of 1,000 cells per well on 24 well fibronectin-coated plates with FGF-2 10 ng/ml for 6 days. Media, ROCK inhibitor Y27632 3 μg/ml, Akt inhibitor 25 μM and LY294002 20 μM were added individually or in combination on day 3 and dormant clones were counted on day 6. The combined effects of Y27632 and the AKT inhibitor do not appear to be additive at these concentrations but the combined effects of Y27632 and the PI3K inhibitor appear to be more than additive in these experiments on inhibition of dormant clone survival. Experiments were done in quadruplicate. (Error bars are + S.D.)
Figure 6: The appearance of surviving dormant T-47D clones after treatment with C3 transferase and ROCK inhibitor Y27632. Colony assays were established in quadruplicate in 24 well fibronectin coated tissue culture plates at 1,000 cells/well with FGF-2, as described, inhibitors were added on day 3 at the concentrations shown, cells were stained and photographed on day 6. Cells treated with pan-Rho family inhibitors lost their spread appearance, became small, dendritic and appeared distressed. (100X magnification).
Our model is comprised of several key elements of dormancy in the bone marrow. It consists of estrogen sensitive cells, which are the type likely to remain dormant in the marrow for extended periods10, it consists of fibronectin, a key structural element of the marrow, FGF-2, a growth factor abundantly synthesized by the bone marrow stroma and heavily deposited in the extracellular matrix of the bone marrow31,32 and incubation of cells at clonogenic density where their interactions are primarily with the substratum. While the bone marrow microenvironment is far more complex, this simple system can be built up with as much added complexity as needed to ask specific mechanistic questions.
The model can be used to compare dormant with growing cells in a number of ways, including cell biologic techniques, molecular techniques and in vivo techniques. Cellular phenotypes can be assayed by re-cloning, motility and invasion studies35, nonadherent tumor initiating spherule formation36 or by functional studies for interaction with the microenvironment using specific blocking antibodies or peptides7.
The primary output of the clonogenic assay is a numeric count of colonies, either growing or dormant, which translates to the growing or dormant clonogenic potential of the cells. As implied above, a large number of variables have the potential of affecting this outcome, all of which must be controlled as stringently as possible to yield reproducible results. These include the source of cells, overall passage number, cryopreservation technique and the quality of the cell cultures that were cryopreserved, the number of passages since thawing, the confluence of cultures from which the single cell preparations were obtained, the source and quality of the fetal calf serum, the durations of trypsinization and their reproducibility, the duration of cells left at room temperature, the dexterity of pipetting equivalent cell numbers into each well, the dexterity or evenly distributing cells across the entire surface area of a well, the number of times a plate is removed from the incubator for observation, among dozens of others. In addition to representing generally accepted tissue culture quality controls, these variables affect the dormant clonogenic potential from experiment to experiment. Intra-experimental variability is typically less than 8-10%, however, resulting in percentage changes from experiment to experiment that are highly reproducible. Typically, the variance of the data begins to decrease with a direct correlation of increased experience of the experimenter with the system and the time spent conducting this assay.
Gene expression in dormant and growing cells can be assessed by RT PCR, Northern blot, Western blot, immunoprecipitation/Western and functional complex precipitation/Western, immunofluorescence or immunohistochemistry, multiplexing by microfluidics, Raman spectroscopy, flow cytometry and other techniques. These techniques can be used to determine the roles of specific receptors, other membrane proteins, signaling pathways, transcription factors, histone modifiers, organelles and mitochondrial proteins, metabolic pathways and energy utilization, tension and force and interaction with the microenvironment in countless ways on the role of dormancy.
Our prior studies have demonstrated significant roles for the PI3K pathways in survival of dormant clones7 as well as in maintaining the dormant phenotype with the epithelial distribution of cortical actin15. We also demonstrated that RhoA specifically, must be inhibited in order for the dormant phenotype to be activated. This is in contrast to the data presented here, where inactivation of the Rho family with pan-inhibitors diminishes survival of dormant clones. This underscores the danger of using chemical inhibitors to dissect mechanisms and the need for using genetic approaches to inhibit or activate specific pathways. Towards this end, the methodology lends itself to genetic manipulation of the cells, either by permanent transfection or transduction or by transient transfection15, siRNA or nanoparticles. Since the assay is 6 days in duration, transient gene expression or interference with it will result in significant phenotypic effects at 6 days that can be measured.
Recognizing its potential for studying mechanisms of entering as well as exiting dormancy, we are currently investigating mechanisms for these cells to escape dormancy. While entering dormancy is poorly understood, an understanding of the mechanisms governing the exit from the dormant state is even more elusive. We have reported preliminary observations that inflammatory responses by injured stroma can contribute to the reawakening of dormant MCF-7 cells using this model37. The model can be applied to T-47 cells as well, another ER+ breast cancer cell line. While ER- cells do not become dormant in response to FGF-2, it may be possible to adapt the methodology to other cancer cell types with different differentiation agents in future investigations.
Despite being a highly useful technique for identifying key mechanism involved in dormancy, it remains an in vitro model. However, cells that have undergone dormancy or in vitro manipulation can be tested for xenograft tumor forming capacity. The main potential benefit of the system, however, is the identification of mechanisms that will permit the development of a hypothesis which can be tested in highly complex in vivo dormancy models.
The authors have nothing to disclose.
The authors have nothing to disclose.
MCF-7 cells | ATCC | HTB-22 | |
T47D cells | ATCC | HTB-123 | |
BD BioCoat Fibronectin 24 Well Clear Flat Bottom TC-Treated Multiwell Plate | Corning | 354411 | |
BD BioCoat Fibronectin 60 mm Culture Dishes | Corning | 354403 | |
BD BioCoat Fibronectin 100 mm Culture Dishes | Corning | 354451 | |
BD BioCoat 22x22mm #1 Glass Coverslip with a uniform application of human fibronectin | Corning | 354088 | |
6 Well tissue culture plate | CellTreat | 229106 | |
Dulbecco Modified Eagle Medium High Glucose 10X Powder | Corning Life Sciences | 50-013-PB | |
Heat Inactivated, Fetal Bovine Serum | Serum Source International | FB02-500HI | |
0.25% Trypsin/2.21 mM EDTA | Corning | 25-053-CI | |
Penicillin-Streptomycin Solution, 100X | Corning | 30-002-CI | |
L-Glutamine, 100x, Liquid | Corning | 25-005-CI | |
Recombinant Human FGF basic | R&D Systems | 234-FSE-025 | |
Akt Inhibitor (1L6-Hydroxymethyl-chiro-inositol-2-(R)-2-O-methyl-3-O-octadecyl-sn-glycerocarbonate) | CalBiochem | 124005 | |
LY294002 | CalBiochem | 19-142 | Chenical PI3K inhibitor |
C3 transferase | Cytoskeleton | CTO3 | Inhibits RhoA, RhoB, and RhoC, but not related GTPases such as Cdc42 or Rac1 |
Y-27632 dihydrochloride | Santa Cruz Biotechnology | 129830-28-2 | |
BODIPY FL-Phallacidin (green) | Molecular Probes | B607 | Fluorochrome for fibrillar actin staining |
BODIPY FL-Rhodamine phalloidin (red) | Molecular Probes | R415 | Fluorochrome for fibrillar actin staining |
Alexa Fluor 488 Donkey anti-Mouse IgG Antibody, ReadyProbes Reagent | Molecular Probes | R37114 | |
ProLong Gold Antifade Mountant with DAPI | Molecular Probes | P-36931 |