Breast Cancer Extracellular Matrix Invasion Depends on Local Mechanical Loading of the Collagen Network

Abstract

Active mechanical stresses in and around tumors affect cancer cell behavior and independently regulate cancer progression. To investigate the role of mechanical stress in breast cancer cell invasion, magnetic alginate beads loaded with iron oxide nanoparticles were coated with MDA-MB-231 breast cancer cells and embedded in a three-dimensional extracellular matrix (ECM) model subjected to an external magnetic field during culture. Bead displacement, cell shape and patterns of invasion of the collagen gel, and cell proliferation were assessed over 7 days of culture. The alginate beads swelled over the first 24h in culture, creating circumferential stress akin to that created by tumor growth, while bead magnetic properties enabled local mechanical loading (compression, tension, and relaxation) and motion within the in vitro tissue constructs upon exposure to an external magnetic field. Beads displaced 0.2–1.6 mm through the collagen gels, depending on magnet size and distance, compressing the collagen network microstructure without gel mechanical failure. Invading cells formed a spatulate pattern as they moved into the compressed ECM region, with individual cells aligned parallel to the bead surface. During the first 24 hours of compressive magnetic force loading, invading cancer cells became round, losing elongation and ability to invade out from the bead surface, while still actively dividing. In contrast, cell invasion in unloaded constructs and in loaded constructs away from the compression region invaded as single cells, transversely outward from the bead surface. Finally, cell proliferation was 1.3× higher only after external magnet removal, which caused relaxation of mechanical stress in the collagen network. These findings indicate effects on breast cancer invasion of mechanical loading of ECM, both from compressive loading and from load relaxation. Findings point to the influence of mechanical stress on cancer cell behavior and suggest that relaxing mechanical stress in and around a tumor may promote cancer progression through higher proliferation and invasion.

Introduction

The mechanical microenvironment in and around solid neoplasms evolves during tumor progression, depending on cell stiffness, extracellular matrix (ECM) stiffness, internal tumor solid stress, and external solid stress ¹. Displacement of the tumor border occurs when forces from tumor growth exceed restraining forces from surrounding tissue. Beyond this simplified mechanical model, solid tumors in native tissue are characterized by complex and spatially varying ECM microstructure, anisotropic mechanical properties, and mechanical stresses that act reciprocally and contribute to neoplastic heterogeneity ^2,3,4. Cancer and cancer-associated cells experience compression, tension, and shear stresses that potentially influence cancer progression by altering cellular behavior ⁵. Beyond the influence of tissue (solid) stress on functional activity and cellular behavior, forces within the tumor microenvironment also affect ECM microstructure and interstitial fluid pressure ³. Mechanical forces in the tumor microenvironment favor invasion and lower chemotherapeutic drug transport in portions of the neoplasm and surrounding microenvironment ^1,6,7. Determining the effects of mechanical stress on the tumoral and peri-tumoral microenvironments, including the invasion front, is crucial for understanding the role of mechanobiology in neoplastic progression.

Dynamic forces in and around the tumor microenvironment influence breast cancer behavior in vitro and likely influence the dynamics of tumor progression in vivo ^8,9. Furthermore, the effects of force exerted in the tumor microenvironment are interdependent with tumor growth and tumor cell mechanical properties. For example, breast cancer cell invasion was suppressed in a collagen gel with interstitial pressure gradient controlled by a microfluidic device when the pressure at the invasion tip was lower than in the center of the neoplasm, which may be due to lower transport of soluble tumor-derived factors that enhance invasion ³. In another study, a 15 mmHg compression of MDA-MB-468 breast cancer cells significantly increased adhesion by 25% ¹⁰. When the amount of compression was as high as 50% for 3 hours, MDA-MB-231 cells in agarose elongated and aligned perpendicular to the compression direction and also expressed pro-invasive genes. These data unravel some details of the complex relationship between dynamic forces, alterations in ECM microstructure, higher interstitial pressure and flow, and availability of soluble factors within tumors ¹⁰. Importantly, no single standardized system or platform for the dynamic loading of tumor cells in vitro is widely used by researchers.

In vitro models are widely used to study cancer cell invasion and metastasis by recreating key components of the tumor microenvironment, especially the ECM that cancer cells encounter and remodel during invasion ^11,12. However, conventional models like Boyden chambers and 3D cultures often do not fully control or accurately measure the solid mechanical stresses present in tumors. Several strategies to incorporate mechanical cues include using strain-stiffening matrices, microfluidic devices, and external loading platforms ^12,13,14,15. Further development is needed to mimic heterogeneous solid stress fields in the tumor stroma, and to directly visualize cell behavior during and after dynamic force loading of tumor models. An ideal invasion model capturing dynamic force loading should recapitulate key mechanical forces during initial tumor growth, including tangential compression from spherical expansion against tissue confinement; tension perpendicular to the tumor border from aligned ECM deposited by cancer-associated fibroblasts, and radial and intratumoral stresses from cell proliferation and central tumor necrosis. Such a biomimetic model with spherical geometry and dynamic force loading could provide insights into mechanoregulation of breast cancer behavior. Particularly, such a platform should be compatible with time-lapse microscopy and biochemical/wellplate assays.

A previous study by Alshehri et al. developed a novel system to provide controlled micromechanical loading to cultured breast cancer cells in 3D collagen hydrogels. The researchers embedded magnetic alginate beads loaded with iron oxide nanoparticles within the gels. When an external magnetic field was applied, the magnetic beads attracted to the magnet displaced the surrounding collagen matrix, producing local compression and strain between bead and magnet, causing breast cancer cell alignment perpendicular to the compression ¹⁶. The next step in mimicking the geometry of cancer cells invading interstitial tissue from a small tumor would be to coat breast cancer cells on the bead surface, where, subject to mechanical stress at the bead-gel interface, the cells would experience a mechanical environment similar to that of the invading front of a stroma-bound growing tumor. Further, such cells may experience a dynamic force environment ¹⁷, similar to tumors in vivo undergoing biopsy or mediated by the tumor stroma, simply by application or removal of an external magnet, as has been done in a model of mouse colon cancer ¹⁸. This reasoning and findings of studies mentioned above led to the hypothesis that breast cancer cell invasion from the surface of a magnetic alginate bead is influenced by mechanical loading patterns in the surrounding collagen hydrogel, imparted by magnetic force.

In this study we tested this hypothesis with experimental aims to: (1) evaluate conditions leading to sufficient cell adhesion on the surface of magnetic alginate beads; (2) determine swelling and displacement of magnetic alginate beads embedded in collagen gels resulting from incubation and in situ actuation using an external magnet; (3) determine alignment and morphology of MDA-MB-231 cells in magnetically actuated tissue constructs; and (4) determine the proliferation and invasion of MDA-MB-231 cancer cells in response to local tissue construct actuation by embedded magnetic alginate beads. The highly invasive human breast cancer cell line MDA-MB-231 ^19,20 was used to coat the surface of magnetic alginate beads, which were embedded in collagen hydrogels with well-known mechanical properties and ECM microstructure that modulate the invasiveness of this cell line ^21,22,23. Experiments in this study used time-lapse phase contrast microscopy, epifluorescence microscopy, and confocal microscopy to view magnetic alginate bead displacement, cell morphology and motility. To determine the effects of compression on breast cancer cell proliferation, two separate well plate metabolic/proliferation assays were used in two independent construct compression configurations. Findings described below support the hypothesis that local mechanical stresses alter breast cancer behavior, depending on the local stress magnitude and history.

Materials and Methods

Preparation of Iron Oxide Magnetic Nanoparticles

Magnetic nanoparticles (MNPs) were synthesized by the co-precipitation of ferric to ferrous chloride in a 2:1 molar ratio (Fisher Scientific, NJ, USA). Ferrous chloride tetrahydrate (FeCl₂.4H₂O) and ferric chloride (FeCl₃) were dissolved in 80 mL of deionized water at room temperature in amounts of 2.71 and 5.41 g, respectively. On a hot stirrer plate set to 70°C, the mixture was thoroughly mixed at a speed of 1000 rpm for 30 minutes. A black precipitate was rapidly formed as a result of adding 20 mL of 29% ammonium hydroxide (NH₄OH, Fisher Scientific, NJ, USA) and mixing constantly for about 30 minutes. The pH of the reaction mixture was lowered by adding 4 mL of 2 M citric acid (Fisher Scientific, NJ, USA) and mixing constantly for 1 hour. After that, the formed precipitates were separated from supernatant three times by 30 minutes, then twice for 45 minutes using cylindrical 2” (diameter) × 1” (thickness) neodymium magnets to concentrate the nanoparticles and remove residual citric acid and ammonium chloride, replacing with deionized water. After adding deionized water, the MNPs were disrupted in a sonicator bath for 30 minutes between washes. Sterilized deionized water was used to store MNPs until needed. The steps involved in synthesizing MNPs and capping them with citric acid are shown in (Fig. 1).

Figure 1.

Schematic depicting iron oxide nanoparticles synthesis and fabrication of magnetic alginate beads.

Fabrication of Magnetic Alginate Beads

Magnetic alginate beads were prepared by mixing citrate capped iron oxide nanoparticles and soluble alginate together and then followed by polymerization of alginate. A solution of 1 % w/v sodium alginate (Carolina, NC, USA) was prepared by dissolving alginate in deionized water at room temperature with vigorous stirring. MNPs from the stock solution were then added to the alginate solution at 1:1 dilution, followed by sonication at room temperature for 30 minutes to obtain a fully homogeneous distribution of MNPs in the mixture. Using a micropipette (Fig. 1), 10 μL of alginate and magnetic nanoparticle solution was dropped into a 100 mM calcium chloride (CaCl₂) solution (Fisher Scientific, NJ, USA) to fabricate the magnetic alginate beads. To prevent nanoparticle leakage from the beads, as previously demonstrated ¹⁶, the magnetic alginate beads were coated with chitosan by an initial soak of 15 minutes in 2% glutaraldehyde (Thermo Fisher Scientific, MA, USA), followed by one hour of immersion in 2% chitosan (ICN Biomedical, OH, USA) dissolved in 2% acetic acid (Sigma-Aldrich Life Science, MO, USA). Subsequently, the magnetic alginate beads were transferred to 2% glutaraldehyde solution for an additional hour. The chitosan-coated beads were rinsed with sterilized deionized water twice for 5 minutes each time, and then stored in a large volume of sterile 100 mM calcium chloride dissolved in PBS. The size and shape of these beads were previously characterized by measurements from optical microscopy ¹⁶.

Culture of Breast Cancer Cells

The human breast cancer MDA-MB-231 cell line was cultured on tissue culture-treated polystyrene dishes in conventional tissue culture conditions of 5% CO₂, 37°C, and 100% humidity (HERAcell 150i, Thermo Fisher Scientific, MA, USA). The cells were supplied and maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) with high glucose (Sigma-Aldrich Life Science, MO, USA), containing 10% fetal bovine serum (FBS, Corning, NY, USA), 5000 I.U./mL penicillin and 5000 μg/mL streptomycin (Corning, VA, USA). After reaching confluence, the cells were split with trypsin and fed every two days.

MDA-MB-231 Cells Adhesion to Magnetic Alginate Beads

Cancer cells were attached to the surface of an alginate bead containing embedded iron oxide nanoparticles and cultured as free-floating beads in DMEM with high glucose (Sigma-Aldrich Life Science, MO, USA). To optimize coating, five conditions were evaluated: A) 100,000 cells/mL only; B) 100,000 cells/mL and 0.1 mg/mL bovine serum albumin (Sigma life science, MO, USA); C) 100,000 cells/mL and 0.1 mg/mL collagen type I (Corning, NY, USA); D) 100,000 cells/mL, 0.1 mg/mL bovine serum albumin, and 0.1 mg/ml collagen type I; and E) 100,000 cells/mL, 0.1 mg/mL bovine serum albumin, 0.1 mg/ml collagen type I and fetal bovine serum (FBS, Corning, NY, USA). The alginate beads were gently stirred and cultured at 37°C and 5% CO₂ for 1 hour and then transferred to a new well. After that, alginate beads were fixed with 1 mL of 10% neutral buffered formalin (VWR, PA, USA) for 20 minutes and stained for 1 hour with 10 μM of 4′,6-diamidino-2-phenylindole (DAPI, Life technologies, OR, USA). The cells-coated beads were washed with gentle agitation three times with 120 mM calcium chloride and resuspended in PBS (Sigma-Aldrich, MO, USA). The mean number of MDA-MB-231 cells distributed on the surface of the beads was assessed by imaging cell nuclei in seven fields of view per bead with an epifluorescence microscope using a 4× magnification objective. Cells were counted manually from epifluorescence images using Fiji software (Fig. 2).

Figure 2.

Schematic illustrating coating magnetic alginate beads with MDA-MB-231 cells, embedding cell-coated beads in collagen gels, and culturing for 7 days after exposure to an external magnetic field.

Acellular Magnetic Alginate Bead Embedded in Collagen Hydrogel

The culture chamber configuration for embedded magnetic alginate bead is demonstrated in (Fig. 2). A collagen gel at 2 mg/mL was prepared on ice by sequentially mixing the following components: 0.25 mL of 10x phosphate buffered saline (PBS) containing the pH indicator phenol red, 0.89 mL of deionized water, 0.25 mL of fetal bovine serum, 1.32 mL of type I collagen from rat tail (Corning, MA, USA), and several microliters of 1 M sodium hydroxide (NaOH) to titrate the solution to a pH of 7.4, based on a color-shift from yellow to a light red. Immediately after titration, 0.5 mL of the collagen hydrogel mixture was pipetted into each chamber, and the magnetic alginate bead was embedded into the center of collagen hydrogel mixture, followed by a 1 hour incubation at 37°C. Magnetic fields were applied to the constructs for 7 days using rectangular 13 (L) × 6 (W) × 3 (H) mm neodymium magnets (N52). The swelling of alginate beads was measured using ImageJ on time-lapse phase contrast microscopy pictures at 4× magnification registering scratched marks tracking displacement on the culture chamber bottom.

Actuating Collagen Hydrogels with Magnetic Alginate Beads

Collagen hydrogel mixture at 1.5 mg/ml was pipetted into each chamber of 8-chambered well slides. After that, acellular magnetic alginate beads were embedded into the collagen hydrogel mixture ~5 mm (near) and ~13 mm (far) from the magnet face and then incubated at 37°C for 1 hour. Magnetic fields were applied to the constructs for 5 hours and 7 days culture time using rectangular 13 (L) × 6 (W) × 3 (H) mm (small size) and cubic 10 (L) × 10 (W) × 10 (H) mm (big size) neodymium magnets. The displacement of alginate beads was assessed using time-lapse phase contrast microscopy images of the edge in reference to scratch marks on the culture chamber bottom using ImageJ.

Collagen Hydrogel Sandwich Assay

A collagen hydrogel at 2 mg/mL was prepared as described above, and 0.2 mL of the mixture was pipetted into each chamber immediately to make the first layer of collagen hydrogel. Following that, an individual acellular alginate bead was embedded into the collagen hydrogel mixture in the center of each chamber with a small sterile spatula to allow half of the alginate bead to submerge in the first layer of collagen gel, and then incubated at 37°C for 1 hour until the collagen gelled. DMEM with 100,000 MDA-MB-231 cells was added to the top of the alginate beads and the surface of the first layer of collagen gel. To achieve cell adherence to the first layer of the collagen gel, chambers were placed in an incubator for 1h. Then excess media was removed, an important step to ensure good bonding of the two collagen layers, and the second layer of collagen was added by pipetting 0.2 mL of the collagen mixture onto the first layer of collagen and then incubated at 37°C for 1 hour. After that, 0.5 mL of DMEM supplemented with 10% FBS was added to the hydrogel cultures and changed daily to feed the cells. Tissue constructs were cultured for 24 hours to allow alginate beads to swell, and cells to completely anchor in the gel, and were checked at this time for gel layer adhesion by visual observation and by focusing through the layer interface at multiple points with the phase contrast microscope. Then, magnetic fields were applied to tissue constructs using rectangular 13 (L) × 6 (W) × 3 (H) mm neodymium magnets in the same North/South orientation. MDA-MB-231 cells were assessed by time-lapse phase contrast images for 48 hours. Tissue constructs were fixed in 10% neutral buffered formalin (VWR, PA, USA) and stained for 1 hour with 10 μM propidium iodide (Thermo Fisher Scientific, MA, USA) and one drop of ActinGreen^™ 488 readyprobes^™ reagent (Thermo Fisher Scientific, MA, USA). Following that, the tissue constructs were washed three times with DMEM containing 10% FBS using gentle agitation. A solution of 70% glycerol, 1xPBS, and deionized water (optical clearing agent) was used to soak all gels for 1 hour before being imaged. To intensify the fluorescent signals, gels were placed on a microscope slide and cover slipped to create slight compression. Stained constructs were imaged using confocal microscopy.

XTT Cell Proliferation Assay in Collagen Hydrogel Sandwich Culture

The collagen hydrogel sandwich culture was prepared as described above; however, an individual acellular alginate bead was embedded into the collagen hydrogel mixture ~5 mm (near) and ~13 mm (far) from the external magnet housing. Tissue constructs were cultured for 24 hours to allow alginate beads to swell, and cells to completely anchor in the gel. Then, magnetic fields were applied to three different conditions (near to/far from magnet, and control) using rectangular 13 (L) × 6 (W) × 3 (H) mm neodymium magnets in the same North/South orientation. After 48 hours of culture, a cell proliferation kit II (XTT) (VWR, PA, USA) was used according to the manufacturer’s protocol on three culture systems when A) the magnet was still in place during XTT incubation time B) the magnet was removed immediately after adding XTT labeling mixture, and C) magnet was added only during XTT incubation time. In each of the three culture systems, old DMEM supplemented with 10% FBS was replaced with 0.2 mL of fresh media, and then 0.1 mL of XTT labeling mixture was added to the media covering the collagen hydrogel and incubated at 37°C for 4 hours. Following incubation, 0.15 mL of culture media for each condition was transferred to 96-well plates in parallel samples. As controls, acellular blanks were treated precisely like the experimental samples in each condition. Absorbance of the samples under the three conditions was measured by microplate spectrophotometer (SpectraMax 340PC384) at wavelength of 450 nm and 690 nm, and net absorbances (A₄₅₀–A₆₉₀) were calculated (Figure S3A). Replicates of 16–27 specimens per condition were performed over three trials. To eliminate background signal contributions, the net absorbance value for all experimental replicates (beads near magnet and far from magnet for each condition) was corrected against the reference net absorbance value corresponding to cell-free media controls. This was accomplished by dividing the net absorbance value of each replicate by the net absorbance value obtained for the media only, reference samples. Afterward, samples of beads near magnet and far from magnets in each condition were normalized within the control samples, which were not subjected to magnetic field, by dividing each sample by the control mean.

Alamar Blue Cell Proliferation Assessment in Gelatin Sandwich Culture

A gelatin gel at 35 mg/mL was prepared by dissolving 70 mg of gelatin powder (Knox) in 2 mL of warmed 1x phosphate buffered saline (PBS). The gelatin solution was gently mixed and placed in a water bath on a hot plate set to 37°C for 30 minutes. The resulting solution was sterile filtered through small 0.2-micron syringe filters (VWR, PA, USA) and immediately pipetted at a volume of 0.25 mL into each well of 24 well-plate to make the first layer of gelatin layer, after that incubated at room temperature for 3 hours until the gelatin firmed. After incubation time, 2% of ruthenium and 2% of sodium persulfate crosslinking reagents (Advanced BioMatrix, CA, USA) were separately sterile filtered through small 0.2-micron syringe filters and added to the total volume of DMEM supplemented with 10% FBS. After that, 0.2 mL of the mixture was added to the surface of the first layer of gelatin gel and incubated for 10 minutes at room temperature to allow diffusion through the gel. As the crosslinking mixture penetrated the gelatin gel, photocrosslinking with UV light at 400–450 nm was performed on the whole first layer of gel for 1 minute. The crosslinking process was followed by two subsequent rinses with sterile PBS with 15 minutes soak each time. Afterward, DMEM with 100,000 MDA-MB-231 cells was added on the surface of the first layer of gelatin gel and incubated at 37°C and 5% CO₂ for 1 hour to achieve cell adherence to the surface of the gel layer. After incubation time, the second layer of gelatin was added by pipetting 0.25 mL of the mixture onto the first layer of gelatin and then incubated at 37°C and 5% CO₂ for 3 hours. Then, the second layer of gelatin gel was photocrosslinked in the same way as the first layer, followed by rinsing twice with sterilized 1xPBS for 15 minutes each time. DMEM supplemented with 10% FBS was added at a volume of 0.5 mL to each well to maintain cell viability and cultured for 24 hours to allow complete cell anchoring in the gel. Magnetic fields were applied simultaneously to the top and bottom of the gelatin gel using cylindrical 6 mm (diameter)× 6 mm (thickness) neodymium magnets in the same North/South orientation. The magnet on the top of gel was covered with sterile parafilm to prevent cell contamination, and the bottom magnet was separated from gelatin gel by 12 mm using a styrofoam sheet to prevent gel breakage caused by attractive forces between the top and bottom magnets. After 48 hours of culturing, alamarBlue^™ HS cell viability reagent (AB) (Thermo Fisher Scientific, MA, USA) was applied according to the manufacturer’s protocol on three culture systems when A) the magnet was still in place during AB incubation time B) the magnet was removed immediately after adding AB reagent, and C) the magnet was added only during AB incubation time. The old DMEM supplemented with 10% FBS was replaced with 0.45 mL of fresh media. Afterward, 0.05 mL of AB reagent was added to the media covering the top of the gelatin gel, and then incubated at 37°C and 5% CO₂ for 8 hours. After incubation, 0.1 mL of culture media was transferred to 96-well plates. Each condition was accompanied by acellular controls treated like the other cultures. The absorbance of the cultures treated with magnetic fields and the control cultures was measured at wavelengths of 570 nm and 600 nm by a microplate spectrophotometer (SpectraMax 340PC384) (Figure S3B). Then, net absorbances (A₅₇₀–A₆₀₀) were calculated. Multiple samples were performed for each condition. The net absorbance of all gels exposed and not exposed to a magnetic field were corrected for background by subtracting the net absorbance of each sample from the mean net absorbance of cell-free media. Afterwards, the samples in each condition were normalized by the controls, not exposed to magnetic field.

Collagen Hydrogel Sprouting Assay

Human breast cancer cells line MDA-MB-231 (wild-type) cells attached to alginate beads containing embedded iron oxide nanoparticles, were cultured as free-floating beads in DMEM with high glucose (Sigma-Aldrich Life Science, MO, USA), containing 100,000 cells/mL, 0.1 mg/mL bovine serum albumin, and 0.1 mg/ml collagen type I. The alginate beads were stirred gently and cultured at 37°C and 5% CO₂ for 1 hour, before being transferred to a collagen hydrogel. Gently placing the cell-coated beads into liquid collagen with a spatula preserved the cell coating on the bead surface. While some cell clusters were detached from the beads during transfer to the gel, these were far enough from the bead to be distinguishable from invading cells originating from the bead. A collagen gel at 2 mg/mL was prepared on ice as before ¹⁶ [17]. After mixing all components, 0.5 mL of the pH-adjusted collagen hydrogel mixture was pipetted into each chamber immediately. After that, each coated magnetic alginate bead was embedded into the center of collagen hydrogel mixture, and then incubated at 37°C for 1 hour. After the initial incubation, the tissue constructs were subjected to magnetic fields using rectangular 13 (L) × 6 (W) × 3 (H) mm neodymium magnets in the same North/South orientation, under four conditions A) at day 5 B) at day 0 C) at day 0, and then removed at day 5 of culture, and D) without applying magnetic fields. The extent and direction of MDA-MB-231 cell invasion under all four conditions in the tissue constructs was assessed by time-lapse phase contrast images with 5 minutes time intervals for 24-hours at day 5 using an inverted phase contrast microscope (CK-2, Olympus) and a 10x magnification objective. It is important to note that cells on the bead first became confluent on the bead surface before invading the collagen gel, observable as cell sheets with occasional circular rips (caused by mechanical failure of the confluent cell sheet) in epifluorescence microscopy images, taking 5 days in culture to achieve confluence and begin invasion. The cancer cells were maintained using the Ibidi on stage cell culture system that provided matched incubator conditions on the microscope. Subsequently, the tissue constructs were fixed at day 6 in 10% neutral buffered formalin (VWR, PA, USA) and stained with 10 μM propidium iodide (Thermo Fisher Scientific, MA, USA) and one drop of ActinGreen^™ 488 readyprobes^™ reagent (Thermo Fisher Scientific, MA, USA) for 1 hour. The tissue constructs were washed three times with DMEM containing 10% FBS by gentle agitation. A solution of 70% glycerol, 1xPBS, and deionized water (optical clearing agent) was used to soak all gels for 1 hour before imaging. To intensify the fluorescent signals, gels were placed on a microscope slide and cover slipped to create compression. Then, an inverted epifluorescence microscope (EXI-310, Accuscope) was used with an objective of 10x magnification to capture images of the cell invasion patterns.

Statistic and Data Analysis

The data are graphed and reported as the mean ± standard deviation and were analyzed by SPSS statistical software to evaluate the statistical difference between groups. A statistically significant result from calculations was defined as a value p < 0.001. All quantitative data were analyzed using ANOVA analysis.

Results

Cell Adhesion on Magnetic Alginate Beads

Media composition influenced breast cancer cell adhesion to magnetic alginate beads (Fig. 3, ANOVA, F=70.2, p<0.001). Cell nuclei were distributed relatively uniformly on the magnetic alginate bead surface from epifluorescence micrographs with digitally-extended depth-of-field (Fig. 3A–E). Cells adherent on the beads did not fall off due to washing with fresh nutrient media or shaking the beads in the wellplate (data not shown), and over 5 days of culture, cells grew in a sheet tightly bound to the bead that sometimes ‘broke’ causing a small circular hole in the cell sheet. Cell adhesion density varied from 121±51 cells/mm² to 738±131 cells/mm² (Fig. 3F). While cell adhesion density was low (<200 cells/mm²) and not significantly different for serum-free media, media+BSA, and media+collagen conditions, the highest cell density resulted from coating in media+BSA+collagen at 738±131 cells/mm² (higher than all other conditions, Tukey tests, p<0.001), followed by media+BSA+collagen+FBS at 460±52 cells/mm² (higher than media, media+BSA, and media+collagen, Tukey tests, p<0.001).

Figure 3.

Stitched, focus-corrected epifluorescence micrographs (4x objective magnification) of fluorescence signal from DAPI-stained DNA in cancer cell nuclei depicting cell distribution on beads under 5 conditions: MDA-MB-231 cells incubated with beads and (A) serum-free medium (control), (B) bovine serum albumin (BSA), (C) type I collagen (Coll), (D) Coll and BSA, and (E) Coll, BSA and FBS. (F) Bar graph of average number of cells per mm² in each condition (n=5 beads/condition). a, higher than all other conditions, p<0.001; b, higher than media, media+BSA, and media + collagen, p<0.001, Tukey’s test. The red dashed line indicates the number density of MDA-MB-231 cells corresponding to roughly 50% confluence. Scale bars are indicated.

Swelling of Embedded Magnetic Alginate Beads in Collagen Hydrogels

Magnetic alginate beads embedded in collagen hydrogels swelled when exposed to standard tissue culture conditions. Bead swelling in culture depended on culture time (ANOVA, F=25.4, p<0.001), but not on the presence of a magnetic field (p=0.6) or the interaction of culture time and magnetic field (p=0.2). Beads swelled 34% between culture days 0 and 1, from projected areas of 0.58±0.07 mm² to 0.78±0.04 mm², and remained the same from culture day 3 at 0.80±0.04 mm² through day 7 at 0.77±0.04 mm² (Fig. 4C, Tukey’s tests, culture days 1, 3, and 7 vs. 0, p<0.001).

Figure 4.

Phase contrast micrographs (4x objective magnification) of magnetic alginate beads embedded in 2 mg/mL collagen gels (A) at 0 hours, and (B) after 24 hours of magnetic field exposure, illustrating bead swelling (red dashed circle indicates original bead projected border). (C) The average projected area of beads with and without magnet exposure during 7 days of culture time. a, projected area different than day zero with magnetic field, p<0.001, Tukey’s test; b, bead projected area different than day zero with no magnetic field, p<0.001, Tukey’s test. Scale bars are indicated.

Displacement of Embedded Magnetic Alginate Bead in Collagen Hydrogel

Magnetic alginate beads embedded in collagen hydrogels exhibited motion when exposed to a static external magnetic field. A localized mechanical loading of the collagen gel occurred due to displacement of the magnetic alginate beads. This displacement depended on distance separating the magnetic alginate bead from the magnet (Fig. 5), as well as the magnet size (Fig. S1). In term of proximity to the magnet, a maximum displacement of 350 μm was observed after 7 days for alginate beads located approximately 5 mm (near) from the magnetic field, whereas beads located approximately 13 mm (far) from the magnetic field exhibited a lower displacement of 120 μm following the same culture period (Fig. 5E). The separation distances between magnetic alginate beads and static external magnets significant affected total bead displacement (Fig. 5A–D). After 5 hours exposure to two different sizes of magnets, a magnet of dimensions 13 mm (L) × 6 mm (W) × 3 mm (H) (small) induced 230 μm bead displacement (Fig. S1A,B,E), while a 10 mm (L) × 10 mm (W) × 10 mm (H) magnet (big) produced 620 μm of displacement (Fig. S1C,D,E). Over 7 days, small and big magnets yielded maximum beads displacements of 570 and 1680 μm, respectively (Fig. S1F). The displacement of embedded magnetic alginate beads was highest near the big magnets and lower near small magnets. Magnetic alginate bead displacement caused compression of the collagen network, maximally nearest the bead and in between the bead and magnet (Movies S10, S11).

Figure 5.

Phase contrast micrographs of acellular magnetic alginate beads embedded in 1.5 mg/mL collagen gels with nutrient media demonstrating bead displacement (A,C) before and (B,D) after exposure to a (A,C) weak and (B,D) strong static external magnetic field. (F) Line graph of bead displacement during 7 days of culture time (n=4 bead/near, far magnet), * indicates p<0.01 by ANOVA followed by post-hoc test. Near and far magnet represent strong and weak static magnetic fields, respectively. Scale bars are indicated.

Directional Alignment and Morphology of MDA-MB-231 Cells in Magnetically Actuated Tissue Construct

Human breast carcinoma cells responded to in situ dynamic force loading by altering alignment and morphology. Cells of the MDA-MB-231 line cultured within a collagen hydrogel surrounding an embedded magnetic alginate bead altered alignment and morphology within 72h of initial seeding and 48h of continuous application of a magnetic field, starting after 24h of culture. Under normal culture conditions, MDA-MB-231 cells appeared rounded and with no alignment (Fig. 6A,D,G,J). Under a strong external magnetic field that caused the bead to move and locally load the collagen gel, cells in the compressed gel region became elongated and aligned perpendicular to the direction of compression (Fig. 6B,E,H,K), whereas under a weak external magnetic field elongation and alignment were similar but less pronounced (Fig. 6C,F,I,L). Alignment and elongation became less pronounced several cell widths (~50 μm) from the bead-gel interface. Close to the alginate bead, cells were locally aligned parallel to the bead-gel interface. Elongated cell morphology and alignment were confirmed by phase contrast microscopy (Fig. S2).

Figure 6.

Confocal micrographs (20x objective magnification) of fixed, stained MDA-MB-231 cells in collagen gels near magnetic alginate beads, with (A-C) propidium iodide (red channel) staining, (D-F) F-actin (green channel) staining, (G-I) combined red and green channels, and (J-L) magnified regions of interest taken from the dashed white rectangles in G-I. Conditions were (A,D,G,J) without applied magnetic field, or under continuous application of (B,E,H,K) a strong or (C,F,I,L) weak external magnetic field for 48h. Selected cells near the bead are highlighted by blue arrows, and further from the beads by white stars. Scale bars are indicated.

MDA-MB-231 Cells Proliferation in Response to Magnetic Actuation

Magnetic actuation of beads embedded in collagen hydrogels affected human breast cancer cell proliferation. Specifically, the proliferation of cancer cells was higher in the relaxation condition (condition 3), when the magnet was removed immediately after adding the XTT labeling mixture, (Fig. 7, 1-factor ANOVA, F(2,69) = 11.64, p<0.001), but not in other loading conditions (Fig. S3). The proximity of alginate beads to the magnet in the culture system did not significantly affect proliferation in the steady compression condition (Fig. S3C, condition 2) when the magnet remains attached to the culture system during XTT incubation time (F(2,45) = 0.25, p=0.781), or in the first time compression condition (Fig. S3D, condition 1) when adding magnet only during XTT incubation time (F(2,45) = 0.67, p=0.519). In condition 3, normalized proliferation was 1.27±0.25, vs. 1.00±0.12 for control (condition 4). The highest proliferation rate resulted from beads that were near to the magnet, at 1.27±0.25 (higher than all other conditions, Tukey pairwise comparisons, p<0.001), followed by beads that were far from magnet, at 1.19±0.2 (higher than control, Tukey pairwise comparisons, p=0.005) (Fig. 7C).

Figure 7.

Proliferation of breast cancer cells in magnetically loaded 3D tissue constructs Schematic representations of (A) horizontal and (B) vertical geometries of magnetic force loading in MDA-MB-231 cell-seeded collagen and photocrosslinked gelatin gels, respectively. (C) Normalized XTT signal from control constructs (no magnetic field), and constructs with higher (near magnet) and lower (far magnet) magnetic force loading that was released by magnet removal just before the proliferation reagents were added. (D) Normalized Alamar Blue signal from control constructs (no magnetic field), and constructs with magnet force applied to the embedded bead, released by magnet removal just before the proliferation reagents were added. a indicates p<0.001, and b indicates p<0.005, Tukey pairwise comparisons vs. control (n=16 constructs/condition). ** indicates p<0.01 by ANOVA followed by post-hoc test (n=21 constructs for magnet group and n=27 constructs for the control).

The effects of mechanical stress relaxation on breast cancer cell proliferation were confirmed in photocrosslinked gelatin constructs by the Alamar Blue assay (Fig. 7B,D). In this configuration, the magnetic force was applied vertically to a horizontal layer of MDA-MB-231 cells sandwiched in between crosslinked gelatin layers. In the relaxation condition, proliferation was significantly higher at 1.27±0.19 compared to the control 1.00±0.14 (Fig. 7D, t(46)=−5.6, Tukey pairwise comparisons, p<0.001). However, the application of a magnetic field to induce either steady compression throughout the culture period or only during the Alamar Blue incubation period did not significantly impact the cancer cells’ proliferation rate.

MDA-MB-231 Invasion in Response to Magnetic Actuation

Invasion of the collagen gel by MDA-MB-231 cells from the surface of the magnetic alginate bead began with a few cells by day 1 of culture, with en masse invasion typically by day 5–7 of culture, with a pattern that depended on the history of magnetic force applied during culture (Figs. 8, S4). Nuclear staining revealed large numbers of cells on the bead surface, with only a fraction invading the collagen gel (Figs. 8M–P, 9). With no magnetic field, breast cancer cells extended from the bead into the collagen gel in a single mass of dispersed cells, without a preferential direction. In contrast, with a continuously applied magnetic field, invading cells predominantly extended on either side of the hemisphere of the bead facing the magnet, aligned outward from the bead surface at the leading edge of the invading front. More leader cells were apparent in the gels with applied magnetic field (Fig. 8, +MF columns, Fig. 9), with occasional very long streams of invading cells (Figure S5). There was no marked difference in the number of invading cells or extent of invaded gel area between conditions.

Figure 8.

Stitched, background-subtracted phase contrast (A-L) and epifluorescence micrographs (M-P) of MDA-MB-231 breast cancer cells invading the surrounding collagen gel from the surface of an embedded magnetic alginate bead, in the absence (A,B,E,F,I,J,M,N: No-MF) and presence (C,D,G,H,K,L,O,P: +MF) of a magnetic field applied continuously during culture, on days 1 (A-D), 3 (E-H), and 7 (I-P) after embedding and culture. In phase contrast micrographs, the bead is the centrally located blank circular object. A few cells dislodged during embedding form clusters near the bead. The scale bars and direction of magnet/magnetic force on the bead are indicated. Columns of phase contrast micrographs are from the same bead, at different culture timepoints. Epifluorescence micrographs of phalloidin conjugate-stained cells represent different beads after 7 days of culture.

Figure 9.

Epifluorescence micrographs from fixed and propidium iodide-stained magnetic alginate beads seeded with breast cancer cells and embedded in collagen gels. Condition 1 (A) is no magnetic field for 5 days, then 24h with applied magnetic field (indicated as X MF ←). Condition 2 (B) is continuously applied magnetic field (←MF). Condition 3 (C) is continuously applied magnetic field, then removed for 24h (← MF X). Condition 4 (D) is no magnetic field (X MF). Each panel represents several stitched/overlapping fields-of-view. Blue dashed lines represent the magnetic bead border. The scale bars are indicated.

The pattern of invasion and morphodynamics of invading cells depended on the history of applied magnetic loading. First, epifluorescence micrographs revealed that invading cells subjected to a magnetic force only after the onset of invasion (by day 5, condition 1), formed a flattened spatulate pattern, with less invasion in the direction of the magnetic force and more invasion oblique to the force direction (Fig. 9, condition 1). This contrasts with the dispersed invasion of cells in cultured not exposed to an external magnetic field (Fig. 9, condition 4). Without a magnetic field, invading cells tended to move outward from the bead surface (Fig. 9, Movie S6, condition 4). Upon magnetic field application, invading cells between the bead and magnet became rounded and spread laterally while dividing but could not invade outward from the moving bead surface (Fig. 9, Movie S7, condition 1). In contrast, cells invading into the compressed region caused by a magnetic field continuously applied during culture had a similar invasion pattern to cells invading without applied force, but with prominent leader cells and alignment (Fig. 9, Movie S8, condition 2). Finally, cells invading during collagen gel relaxation caused by magnet removal appeared elongated but also appeared not to invade outward rapidly from the bead surface (Fig. 9, Movie S9, condition 3).

Discussion

This study investigated the effects of collagen network mechanical loading on breast cancer invasion in a novel three-dimensional in vitro tumor model. Breast cancer cells adhered to the surface of an alginate bead containing glutaraldehye-crosslinked chitosan, with maximal cell adhesion resulting after suspension in a mixture with bovine serum albumin and soluble type I collagen (Fig. 3), possibly due to rapid BSA adsorption onto the chitosan ²⁴, and then more adsorption of collagen to BSA ²⁵ versus low cell adhesion to alginate-chitosan alone. Also, electrostatic interactions promote some cell adhesion onto chitosan surfaces ²⁶. Although more conditions could be tested, and protein adsorption further characterized, this experiment as performed allows a degree of control over the seeding density of cells on the beads. These adherent breast cancer cells remained on the bead surface while proliferating to form a confluent sheet that invaded surrounding type I collagen gels after 5 days in culture. Iron oxide nanoparticles within the bead exerted a force on the surrounding hydrogel when cultured in chamber slides with strong neodymium magnets fixed to the outer chamber wall, resulting in bead displacement within the gel proportional to magnet strength (Figure S1) and bead-magnet distance (Fig. 5). Cancer cells seeded around the bead in the region of highest compression altered morphology to become elongated perpendicular to the direction of compressive stress (Fig. 6). Furthermore, cancer cell proliferation was 25% higher immediately following mechanical stress relaxation in the gel due to removal of the external magnets (Fig. 7) but was unchanged immediately after or 5 days after application of an external magnet and the associated dynamic and static mechanical loading, respectively (Figure S3). Without an applied magnetic field and mechanical force from the bead’s response, MDA-MB-231 cells invaded into the collagen gel in a disseminated fashion (Fig. 9C,D; Figure S4 condition 4). In contrast, the invasion pattern was compact, spatulate; and restricted, extending perpendicularly to the compressive stress direction of the magnetic field and force applied at day 5, when the cells starting to invade experienced a dynamic compressive mechanical load (Fig. 9A,B; Figure S4 condition 1). Under 5 days of continual magnetic force loading, cells invaded preferentially along aligned collagen network microstructure on either side of the compression region, in or near regions of low and/or tensile solid stress (Figure S4, condition 2). Finally, with stress relaxation occurring during the start of invasion (magnet removal on day 5 of culture), the cells were highly aligned perpendicular to the bead surface, potentially due to local dynamic tensile stress experienced by the leading invasive edge (Figure S4, condition 3). Taken together, these findings from an innovative cancer invasion model indicate roles of dynamic solid and fluid mechanical stress and static solid stress influencing the pattern of breast cancer invasion, as well as cell morphology and rate of proliferation.

The patterns of mechanical stress in this in vitro model mimic those found in and around native solid tumors. During growth, solid tumors experience a circumferential tensile and inward-directed compressive solid /residual stress at their periphery, estimated at 10–19 kPa for breast tumors ²⁷, along with elevated interstitial pressure of approximately 2.4–3.9 kPa ^28,29. Circumferential tensile stress at the bead surface is inferred from the bead swelling of Fig. 4, and from the fact that in a previous study, cells in the adjacent collagen gel also aligned circumferentially ¹⁶. Compressive solid stress is apparent in the Movie S7, occurring when the external magnet is present. The cells’ alignment (Figure S2D) and spreading response may result from collagen fiber tension and alignment as well as higher collagen density in the compressed region, evident from darkening of collagen in phase contrast micrographs in front of the moving bead (Figure S6), rather than being a direct response primarily to the compressive stress and pressure. On the other hand, removal of the external magnet (Movie S8) reveals stress relaxation, which clinically may be caused by surgical removal of solid tumors, after core biopsy, or due to necrotic cell lysis ^30,31. The effects of stress relaxation on cancer cell behavior are important to study when considering, for example, the effects of core biopsy on increasing lung metastases in mice ³², and in conjunction with proposed stress relaxation treatment of the tumor stroma to reduce hypoxia and increase drug transport ²⁷. Future computational studies using finite element modeling, and more extended measurements of pressure in and around the alginate bead using a piezoelectric needle probe will reveal the magnitude of mechanical stress in this in vitro tumor model. The novelty of the model and the significance of the main finding of breast cancer proliferation following stress relaxation are underscored by the relative lack of mechanobiology studies on 3D in vitro cancer models employing control of mechanical stress. However, magnetically-induced pressure in a mouse colon cancer model was found to mimic rodent tumor stress levels of ~1.2 kPa, activate β-catenin signaling and expression, and induced formation of aberrant crypt foci ¹⁸. These findings confirm the usefulness of the in vitro approach to mimicking tumor stress used here.

Invasive breast carcinoma adapts to the local mechanical microenvironment, responding to compressive stress onset and relaxation. In this invasion model, conditions 1 and 3 represent rapid loading and unloading associated with local compression and tension, respectively. These correspond to clinical scenarios of compressive solid stress within the tumor (residual stress) and from the strained peritumoral microenvironment ³³. In the present model, these two solid stress components are simulated by the bead swelling occurring in the first 24h (Fig. 4) and by the applied magnetic force, respectively. Cells invading the magnetically compressed collagen gel rounded and proliferated, resembling the pushing border in some fast-growing tumors. One histological study of basal-like triple negative breast cancers found pushing borders in 14% of cases, associated with slightly worse overall survival vs. infiltrative borders (Hazard ratio 1.16, p=0.042) ^34,35. Higher proliferation upon stress relaxation (condition 3) is consistent with apoptosis caused by high stress/compression, and proliferation in low stress/relaxation microenvironments, confirmed in tumor spheroids cultured in agarose gels ³⁶ [26]. Higher invasiveness of breast cancer cells was reported to result from higher collagen density and stiffness inducing focal adhesion/Rho kinase signaling ³⁷. Also, compression of MDA-MB-231 cells in agarose constructs caused altered gene expression favoring the invasive phenotype: lower cadherin 1 and PECAM 1, and higher caveolin 1, ITGβ1, PTEN, Rac1, and VEGF ³⁸ [11]. In the present in vitro model, marked invasion was observed in conditions 1 and 2 (Figs. 9, S5), typically just adjacent to the highly compressed region directly between bead and magnet.

Magnetic actuation of the alginate bead caused fluid stress to act on invading breast cancer cells, particularly in conditions 1 and 3, in which the magnet was applied or removed at the very beginning of invasion of the surrounding collagen gel. In condition 2, the magnetically actuated bead displacement continued over 2–7 days (Figs. 5, S1), suggesting a combined solid plus fluid stress, with a larger fluid stress component in the first several hours after magnet application/removal, and a larger solid stress component thereafter. In another in vitro platform to apply static strain to cancer cells embedded in hydrogels, 1% w/v agarose disks were compressed to 50% strain with the compressive stress peaking at 250 Pa before dropping to 50 Pa after 3 hours ³⁸. This caused alignment of MDA-MB-231 cells perpendicular to the compression direction and altered gene expression promoting a pro-invasive phenotype. Similarly, cancer cells aligned perpendicular to the compressive stress direction in conditions 1 and 2 of the present study, showing less round cell shape dependent on strength of the applied magnetic force. Another study aspirated a collagen gel with embedded MDA-MB-231 cells into a capillary tube, and found alignment of the cells with the direction of aspiration, tensile stress and interstitial fluid flow ³⁹. Similarly, bead displacement toward the magnet within the gel likely causes both interstitial fluid flow and shear and tensile stresses near the bead sides not facing the magnet, which may also explain the direction of cell alignment in those areas (Fig. 8,9). Collagen network fibers likely oriented parallel to the bead surface, similar to stretched collagen fibers parallel to tumor borders ⁴⁰. Fiber-tracking studies with confocal microscopy ^41,42 are planned for the future to confirm and quantify this. Stiffening of the soft 2 mg/ml collagen gel through crosslinking is also a subject of ongoing experiments, to better match tissue stiffness surrounding solid tumors. Thus, multiple microstructural and mechanical factors combine to influence cancer cell invasion behavior. Invading cells in the absence of external magnetic field and bead displacement did not display individual alignment although they invaded a gel region together (Fig. 8,9).

The currently described novel and useful in vitro platform to study effects of mechanical loading on cancer ECM invasion has several experimental strengths and some limitations. First, while magnetic force is easily controlled by magnet size, strength, and magnet-bead distance, allowing easy measurement of bead displacement; the exact quantification of mechanical stress and independent control of solid versus fluid stress is not presently possible. Follow-up studies will employ finite element models to estimate the magnitude of these stresses over time. Second, while the pattern of invasion from the surface of the gel-embedded bead is clearly visible using simple phase contrast microscopy and resembles invasion from a spheroid; the model lacks a basement membrane and the hypoxic, necrotic core found in patient-derived solid tumors. Future studies may use Matrigel and cells cultured within the bead to supply these features, and manual cutting or micropipette aspiration-ejection ³⁹ to vary the symmetry of the in vitro tumor border. However, the use of collagen I in this model mimics the environment experienced by invasive carcinoma cells after breaking through the basement membrane. The finding of proliferation upon stress relaxation matches the findings of a study of fibroblasts in stress-relaxed gels ⁴³. The absence of enhanced proliferation with applied stress should be interpreted with caution given the nonuniform stress profile of the bead. On the other hand, the alginate bead with chitosan shell appears to recapitulate at least some of the residual stress in native tumors, as slight cutting by a scalpel or deliberate rapid swelling results in a “split pea” appearance with an opening angle resembling that of some excised tumors ²⁷. Longer culture may catch cancer cell adaptation to the imposed mechanical loads, including altered invasion patterns, and may also facilitate necrosis and hypoxia near the bead, which will be measured in future work. It should also be noted that while breast tumors grow at estimated rates of 0 – 0.68 mm/day in diameter ⁴⁴, the magnet-induced bead displacement is toward the upper end of this range. Further work will measure interstitial stress and solid stress levels in the magnetic bead culture, tuning the magnetic force to match known stress values in solid tumors. Experiments described here were designed to demonstrate the utility of the model to detect effects of mechanical stress on breast cancer invasion and proliferation, but additional molecular assays will be used in future work to evaluate alterations of gene expression in invading cells subjected to various dynamic mechanical loads.

Several conclusions derive from the experiments described above. Magnetic actuation in vitro is suitable to analyze the effects of dynamic mechanical loading in and around tumors constructs on cancer invasion, proliferation, and drug transport through the ECM (planned in future work). Experiments imposed regions of high and low solid stress, transiently high and sustained low fluid stress, and regions of both compression and tension. Dynamic and static compressive loading created a dense border of invading cancer cells that aligned perpendicular to the compression direction and invaded/spread around the compressed region. In contrast, relaxation of compressive stress aligned cells parallel to the relaxation direction and enhanced proliferation in a horizontal collagen invasion model, and a vertical crosslinked gelatin invasion model. The patterns of invasive cells did not resemble the typical disseminated pattern found when tumor spheroids are placed in low density hydrogels but instead resembled pushing borders of fast-growing tumors and infiltrative borders consistent with coordinated migration of cancer cells through challenging microenvironments. It should be noted that a tumor with mechanical and microstructural heterogeneity may possess regions of different mechanical signaling leading to asymmetric and spatially dependent tumor progression. The resemblance of these in vitro results to histopathological features of growing and invading breast tumors suggests that the approach to modeling forces in the tumor microenvironment through non-contact magnetic loading holds promise for gaining further insights into tumor mechanobiology relevant to cancer invasion and metastasis.

Data Availability Statement

The data supporting this article have been included as part of the Supplementary Information.