A Microfluidic Platform for Studying Roles of Mechanical Compression in Tumor-Immune Cell Interactions in a 3D Extracellular Matrix

Summary

We present a protocol for using a microfluidic platform to model tumor-immune interactions within a three-dimensional (3D) environment. Our results revealed that Natural Killer (NK) cells retained their anti-tumor activity under mechanical compression.

Abstract

Mechanical forces significantly influence the ability of immune cells to kill tumor cells in the context of cell-based immunotherapy. To kill tumor cells, immune cells must exert forces on the target cell and form an immune synapse, through which cytotoxic molecules —including granzyme B—are delivered. Despite their importance, how mechanical cues can be leveraged to enhance immune-mediated killing remains poorly understood. This knowledge gap is partly due to the lack of tools capable of providing well-controlled mechanical stress to cell cultures in a physiologically realistic environment. Here, we describe a microfluidic compression device that can apply static or dynamic compression to tumor spheroids embedded in extracellular matrix (ECM) while enabling real-time imaging of tumor-immune interactions via optical microscopy.

The microfluidic platform consists of 12 compartments (6 control and 6 functional). Each compartment contains a cell chamber positioned directly beneath the pressure control unit. Spheroids embedded in ECM are placed within the cell chamber. Using this platform, we investigated the killing efficiency of Natural Killer (NK) cells against breast tumor spheroids (MCF-7) under defined mechanical compression. The results showed that NK cells remained the primary drivers of tumor spheriods death regrardless of mechanical compression in 1.5 mg/mL collagen. Therefore, suggesting that NK cells can maintain their anti-tumor activity under compressive stress. These findings demonstrate the utility of this platform for investigating the role of mechanical forces in tumor-immune interactions. Ongoing studies are identifying the molecular mechanisms that allow immune cells to adapt to compressive stress. Insights gained from these studies may reveal a promising therapeutic avenue.

Introduction

Cell-based cancer immunotherapy is a therapeutic approach that utilizes immune cells to target cancer cells¹. This strategy has demonstrated remarkable success in the treatment of blood cancers^2,3,4. However, the level of success observed in blood cancers has not yet been fully translated to the treatment of solid tumors^5,6,7. One major reason for this limitation is the immunosuppressive tumor microenvironment (TME) that is specific to solid tumors.

The TME is characterized by distinct mechanical cues that can regulate cellular behavior. Effective killing of tumor cells by immune cells requires them to exert forces on target cells and establish an immune synapse through which cytotoxic molecules are delivered⁸. Therefore, understanding how mechanical forces influence immune cell function may uncover new therapeutic strategies to improve the efficacy of cell-based immunotherapies against solid tumors.

A defining mechanical feature of the solid TME is the compressive stress that the tumor microenvironment imposes on the growing tumor^9,10. To date, most compression studies have focused on the effects on tumor cell behavior^10,11. Only recently has attention shifted toward understanding how compression influences immune cells within the TME¹². As a result, the mechanisms by which compressive stress affects Natural Killer (NK) cell migration, immune synapse formation, and cytotoxicity activity remain poorly understood and insufficiently quantified. This gap represents a critical limitation in our understanding of how mechanical cues within TME can be leveraged to enhance immune targeting of solid tumors.

A major barrier to addressing this question is the lack of experimental platforms that can apply controlled and quantifiable compression within physiologically relevant 3D microenvironments while simultaneously enabling real-time imaging of tumor-immune interactions. Early experimental approaches to studying compressive stress involved embedding spheroids in agarose gels¹³. While this method effectively captured physical confinement, it eliminated native extracellular matrix (ECM)-cell interactions. Dolega et al. applied osmotic pressure to induce compressive stress; however, spheroids were suspended in culture medium and therefore remained devoid of ECM interactions⁹. Pandey et al. used a transwell insert and weights to compress spheroids embedded in collagen, which preserved ECM-cell interactions but limited the high-resolution imaging required to quantify immune cell motility and cytotoxic activity¹⁴. Together, these limitations underscore the need for experimental platforms that enable real-time imaging of the morphological responses of cells to compression within the TME.

Here, we present a protocol for a microfluidic compression device¹⁵ that enables controlled application of compressive stress while preserving physiologically relevant ECM-cell interactions and supporting high-resolution live-cell imaging of immune–tumor dynamics. The protocol consists of the preparation of the device components (Section 1), assembly of the device (Section 2), and the imaging of tumor–immune interactions (Section 3). We demonstrate the workflow using an NK killing assay against green fluorescent protein (GFP)-labeled tumor spheroids. Beyond this application, the platform can be readily adapted to investigate the effects of mechanical forces on a broad range of cell types and cellular interactions in 3D microenvironments.

Protocol

1. Preparation of the device components

1. Surface activation of the sample loading chamber
   1. Place a mask over the sample-loading chamber (Layer 1, L1) such that only the wells are exposed.
   2. Expose the masked Layer 1 to oxygen plasma for 15 s using the high-power setting (45 W).
   3. Place Layer 1 in a Petri dish, then continue the following steps in a biosafety cabinet.
   4. Treat each well with 10 µL of 1% polyethylenimine (PEI) for 10 min.
   5. Remove the PEI from the wells by vacuum aspiration from the four corners of each well to prevent scratching the glass surface.
   6. Wash the wells with 10 µL of phosphate-buffered saline (PBS), then remove the wash by vacuum aspiration from the side.
   7. Treat the wells with 10 µL of 0.1% glutaraldehyde for 30 min, then remove the glutaraldehyde by vacuum aspiration from the corners.
   8. Wash the wells with PBS three times, removing the wash from the corners after each wash.
   9. Wrap the Petri dish containing Layer 1 in foil and leave it in the biosafety cabinet overnight.
2. Sterilization of the metal frame
   1. Submerge the 38 mm × 88 mm metal frame and the twelve M2.5 pan-head screws in 70% ethanol overnight.
3. Priming of the pressure control unit
   1. Expose the pressure control unit to oxygen plasma for 30 s using the high-power setting (45 W).
   2. Submerge the pressure control unit, with the features facing downward, in de-ionized (DI) water overnight to fill the channels and the chambers of the pressure control unit.
   3. The next day, remove the pressure control unit from the DI water in which it was submerged.
   4. Fill a microfluidic reservoir with 50 mL of DI water.
      NOTE: The microfluidic reservoir consists of a 50 mL conical tube fitted with a custom lid containing two tubing connections: a pneumatic tubing line and a fluid delivery tubing line.
   5. Connect a syringe filled with air to the pneumatic tubing line.
   6. Slowly depress the syringe plunger to pressurize the reservoir and drive DI water through the fluid delivery tubing until the tubing is filled and water begins to drip from the tube.
   7. Connect the pre-filled, dripping fluid delivery tubing through the polycarbonate manifold and then to the 1.5 mm inlet located on the back of the pressure control unit.
   8. Allow the pressure control unit and manifold assembly to air dry for 2 h. Once completely dry, the pressure control unit and manifold assembly are ready for the next step.

2. Assembling the device

1. Preparation of cell-embedded collagen
   1. Remove and dry the metal frame and screws from the 70% ethanol.
   2. In the biosafety cabinet, place three 600 µL microtubes on ice. Label the tubes as follows: "N" (neutralization), "sp" (MCF-7 spheroids), and "sp+NK" (MCF-7 spheroids with NK cells).
   3. Prepare the neutralization mixture by adding 10× the required volume of Medium 199 (M199) and sodium hydroxide (NaOH) to the "N" tube.
   4. Transfer the appropriate volume of the M199:NaOH mixture to the "sp" and "sp+NK" tubes to achieve a final 1× concentration.
   5. Harvest tumor spheroids and centrifuge at 141 × g for 3 min to allow the spheroids to settle at the bottom of the tube.
      NOTE: Tumor spheroids approximately 200 µm in diameter were generated using a previously described agarose microwell platform¹⁶. Briefly, 3 × 10⁶ MCF-7 cells were seeded into a 36 × 36 agarose microwell array consisting of wells 200 µm in diameter and 200 µm in depth and then cultured for 7 days to produce the tumor spheroids used in the experiments.
   6. Carefully remove the supernatant and resuspend the spheroids in 200 µL of 50:50 MCF-7/NK complete medium (Table 1 and Table 2).
   7. Collect NK cells in a 50 mL conical tube, centrifuge at 141 × g for 4 min, and count the cells.
      NOTE: NK cell viability should be >85% the day of the experiment.
   8. Resuspend the resulting pellet in 200 µL of complete NK medium.
   9. Add the appropriate volume of collagen stock solution to each microcentrifuge tube.
   10. Add the appropriate volume of cell suspension to each tube to obtain a final volume of 200 µL containing 1.5 mg/mL collagen.
       NOTE: The "sp" suspension contains tumor spheroids only, whereas the "sp+NK" suspension contains tumor spheroids and 1.5 × 10⁶ NK cells.
   11. Mix gently by pipetting to ensure an even distribution of cells and collagen while minimizing bubble formation.
2. Device assembly and cell loading
   1. Place Layer 1 inside the dried metal frame.
   2. Load 9 µL of the collagen mixture into each well.
   3. Align the pressure control unit and manifold assembly with the cell chambers on Layer 1 using a microscope.
   4. Insert the dried screws through the manifold and into the metal frame, securing the pressure control unit and Layer 1 between the manifold and the metal frame.
   5. Place the assembled device in a 37 °C, 5% CO₂ incubator for 45 min to allow collagen polymerization.
   6. Add 50:50 MCF-7/NK complete cell culture media through the four media ports.

3. Imaging of tumor-immune interactions

1. Place the assembled device inside the stage incubator mounted on an inverted epifluorescence microscope for live-cell imaging.
   NOTE: The stage incubator maintained a temperature of 37 °C and a humidity of approximately 50% throughout the duration of imaging.
2. Place a secondary housing box over the assembled device to maintain the internal environment at 37 °C, 100% humidity, and 5% CO₂ throughout the imaging period.
3. Connect the device to the pressure controller using the pneumatic tubing.
4. Acquire brightfield and GFP images of the tumor spheroids before applying compression to enable assessment of spheroid deformation following compression.
5. Apply a constant pressure of 120 mbar throughout the imaging period.
   NOTE: The pressure controller software enables the user to define the pressure profile and magnitude applied during the experiment. Supplementary Figure 1 contains a schematic of the pressure control system.
6. Begin time-lapse imaging by acquiring brightfield and GFP images every 10 min for 18 h.
   NOTE: Images were acquired at the center Z-plane of each tumor spheroid using a 20× objective lens on an inverted epi-fluorescence microscope. Brightfield and GFP fluorescence images were captured with exposure times of 200.1 ms and 298.9 ms, respectively.

Results

For the following experiments, a microfluidic compression device¹⁵ was used to model tumor-immune interactions within a physiologically relevant 3D microenvironment, and at the same time, can apply static and dynamic compressions to tumor spheroids (Figure 1).  The device components (Figure 1A1) and their assembly order (Figure 1A2) are shown in Figure 1A. The metal frame serves as the base of the device, with the cell chamber layer (L1) positioned within the frame. The pressure control unit, consisting of L2 and L3 plasma bonded together, is placed above L1, followed by the manifold (Figure 1A2). Together, L1 and the pressure control unit form the functional microfluidic device and are sandwiched between the metal frame and the manifold with screws. The manifold contains inlet ports for fluidic tubing to supply pressure to the device, as well as media reservoirs that facilitate long-term live-cell imaging experiments.

Figure 1: Microfluidic compression device for studies of interactions between immune cells and tumor spheroids under mechanical compression. (A1) Photograph of the components required for device assembly, shown from left to right: manifold, pressure control unit (L2 + L3), cell chamber layer (L1), and metal frame. (A2) 3D schematic of the device components, shown from top to bottom: manifold, pressure control unit (L2 + L3), cell chamber layer (L1), and metal frame. PDMS layers L2 and L3 are plasma-bonded to form the pressure control unit, which is positioned above the cell chamber. The manifold and metal frame secure the assembled device. (B1) Cross-sectional schematic of a compression unit before and after compression. ECM-embedded tumor spheroids and NK cells are placed in the cell chamber. The initial distance between the bottom of L1 and the bottom of L2 before compression is denoted as h. Upon application of pressure to the pressure control unit, L2 deflects downward, causing the PDMS piston to compress the tumor spheroid. The downward displacement of L2 following compression is denoted as Δh. (B2) Merged brightfield and GFP images of MCF-7 tumor spheroid with NK cells before (left) and after (right) compression. MCF-7 cells are GFP-labeled, while NK cells are not labeled. Representative NK cells are circled in blue, while yellow outlines mark the peripheries of the tumor spheroids. Scale bar = 50 µm. Please click here to view a larger version of this figure.

Once assembled, the microfluidic compression device can be used to compress tumor spheroids as illustrated in Figure 1B. Here, the collagen-embedded tumor spheroids with NK cells are introduced into the cell chamber (L1). Upon application of pressure to the pressure chamber (L3), the deformable membrane layer (L2) deflects downward, causing the Polydimethylsiloxane (PDMS) piston to transmit compressive forces to the spheroid. The pressure chamber is controlled by an external pressure source. Additional details regarding the design, fabrication, and characterization of the microfluidic compression device have previously been published in another study¹⁵.

We define the distance between the bottom of the cell chamber (L1) and the bottom of the deformable membrane (L2) as h (Figure 1B1, left panel). Before compression, this distance is 200 µm. Upon application of fluid pressure to the pressure control unit, the deformable membrane deflects downward, reducing the distance h by Δh (Δh; Figure 1B1, right panel) and thereby compressing the collagen-embedded tumor spheroid. In this system, compressive strain is defined as Δh/h, where h is the initial distance between L1 and L2 before compression, and Δh is the downward displacement of L2 after compression. Membrane displacement was previously calibrated by visualizing fluorescent beads adhered to the bottom of L2. A detailed description of the calibration procedure and compressive strain calculations can be found in a previous study¹⁷. An example calibration curve can be seen in Supplementary Figure 2.

Figure 1B2 shows a representative merged brightfield and GFP images of an MCF-7 spheroid with NK cells embedded in 1.5 mg/mL collagen before (left panel) and after compression (right panel). Tumor spheroid boundaries were outlined in yellow. Following compression, the spheroid exhibited an increase in projected area, indicating deformation in response to the successful application of mechanical load. For all experiments presented in this manuscript, we used a compression strain of Δh/h = 50%. We note that the spheroid diameter ranges from 140 µm to 200 µm. This means that the actual compression strain on the spheroid is slightly smaller than 50%. Supplementary Video 1 and Supplementary Video 2 show representative X-Y z-stacks acquired in the GFP fluorescence channel of an MCF-7 spheroid embedded in 1.5 mg/mL collagen before and after application of the 50% compressive strain, respectively.

The microfluidic compression device was used to investigate the effects of NK cells, compression, and their combination on tumor spheroid death (Figure 2). This is made possible by the design of the device, which contains 12 culture wells. Of these wells, six can be subjected to compression, while the remaining six serve as uncompressed controls. This configuration enables four experimental conditions (Figure 2A): spheroids alone with no compression (sp), spheroids with NK cells but no compression (+NK), spheroids subjected to compression (+comp), and spheroids exposed to both compression and NK cells (+comp+NK). 

Figure 2: Mechanical compression modulates NK cell–mediated killing of tumor spheroids. (A) Time-lapse fluorescence images of MCF-7 spheroids embedded in 1.5 mg/mL collagen under four conditions: spheroid-only control (sp), spheroids with NK cells (+NK), spheroids with compression (+comp), and spheroids with both compression and NK cells (+comp+NK). GFP fluorescence labels viable MCF-7 tumor cells. The yellow outline indicates the region of interest (ROI), which was defined using the 18-h image and applied to all time points for fluorescence quantification. Total GFP fluorescence intensity within the ROI was measured and used for subsequent analysis. Scale bar = 50 µm. (B) Total GFP fluorescence intensity of one spheroid over time under 4 conditions. Each colored line represents the total fluorescence intensity of a single spheroid, normalized to its initial intensity at t = 0. The black line is the second-order polynomial fitting of the fluorescence intensity data pooled from 7 spheroids cultured under the control condition and was used as the reference for calculating tumor cell loss in (C). (C) Normalized GFP loss for the +comp, +NK, and +comp+NK conditions was calculated by subtracting the GFP intensity at t = 18 h of each spheroid from the spheroid-only reference line (sp). Therefore, each data point represents the normalized GFP loss of a single spheroid. Error bars represent SEM. Please click here to view a larger version of this figure.

The effects of NK cells, compression, and their combination on tumor spheroids can be visualized in the representative images of MCF-7 spheroids in 1.5 mg/mL collagen (Figure 2A). Because the MCF-7 cells used to generate the tumor spheroids were GFP-labeled, tumor viability could be monitored through changes in GFP fluorescence. GFP and brightfield images were acquired every 10 min for 18 h using an inverted epifluorescence microscope. Notably, spheroids subjected to compression exhibited a larger projected area than uncompressed spheroids, confirming successful transmission of mechanical load from the deformable membrane to the tumor spheroid. Over time, a progressive loss of GFP fluorescence was observed in the +NK and +comp+NK conditions (Figure 2A), indicating NK cell-mediated tumor cell killing.

We use the loss of GFP fluorescence as a measure of tumor cell death. The GFP-labeled MCF-7 cells used in this study were generated through lentiviral transduction. Briefly, a lentiviral vector was used to introduce an enhanced green fluorescent protein (eGFP) expression cassette driven by the elongation factor-1α (EF1α) promoter, along with a puromycin resistance gene driven by the Rous sarcoma virus (RSV) promoter. Following infection, the lentiviral expression cassettes were integrated into the genomic DNA of the MCF-7 cells, resulting in stable GFP expression. Puromycin selection was subsequently used to eliminate non-transduced cells. The GFP labeling of MCF-7 cells enabled tumor cell viability to be quantified by monitoring changes in GFP fluorescence intensity over time (Figure 2B). A strong correlation between GFP fluorescence loss and established markers of cell death, including the live/dead marker propidium iodide (PI) and the apoptotic marker annexin-V, has previously been established¹⁸, validating GFP fluorescence loss as a surrogate measure of tumor cell death.

To quantify the effects of the four experimental conditions on tumor viability, the total GFP fluorescence intensity of each tumor spheroid was measured at every time point, background-subtracted, and normalized to its initial value (Figure 2B). Total spheroid fluorescence was calculated within a predefined region of interest (ROI). This ROI was obtained by manually outlining the boundary of the tumor spheroid in the 18-h image (yellow outline on 18 h panel, Fig. 2A) using an in-house MATLAB script. The total fluorescence intensity within this ROI was then calculated for each time point and corrected by subtracting the average background fluorescence measured from a separate ROI selected elsewhere in the image. The resulting background-corrected fluorescence values were subsequently normalized to the initial fluorescence intensity of the spheroid. 

Each colored curve represents a normalized GFP intensity of an individual spheroid under a specific experimental condition in Figure 2B. The curve is obtained using a second-order polynomial fitting to the experimental data. The purple curves represent spheroids subjected to compression alone (+comp, Figure 2B), the red curves represent spheroids with NK cells (+NK, Figure 2B), and the blue curves represent spheroids exposed to both compression and NK cells (+comp+NK, Figure 2B). In contrast, the black curve is the second-order polynomial fit to the data of seven spheroids cultured under the spheroid-only condition serving as a control (sp, Figure 2B). In Figure 2B, only the polynomial fitted data, but not the experimental data points, are presented for clarity. Notably, the GFP fluorescence intensity of the spheroid-only (sp) and spheroid + compression (+comp) conditions exhibited a general upward trend over time (Figure 2B), consistent with continued tumor growth or maintenance of viability. In contrast, the +NK and +comp+NK conditions demonstrated a progressive decrease in GFP fluorescence intensity (Figure 2B), indicating tumor cell death.

Cumulative tumor cell loss was quantified by subtracting the fitted normalized GFP fluorescence intensity of each spheroid under a given condition (+comp, +NK, or +comp+NK) from the control condition at 18 h (Figure 2C). As a result, each data point represents the normalized GFP loss of an individual spheroid relative to the spheroid-only control (Figure 2C). Accordingly, the +comp condition reflects GFP loss resulting from compression alone (purple, Figure 2C), the +NK condition reflects GFP loss resulting from NK cells alone (red, Figure 2C), and the +comp+NK condition reflects GFP loss resulting from the combined effects of compression and NK cells (blue, Figure 2C). The +comp condition exhibited the lowest degree of GFP loss (Figure 2C), suggesting that compression alone had a minimal effect on spheroid viability. In contrast, both the +NK and +comp+NK conditions showed significantly greater GFP loss than the +comp condition (Figure 2C), indicating that NK cells are the primary drivers of tumor cell death. Although the +comp+NK condition exhibited a greater average GFP loss than the +NK condition, this difference did not reach statistical significance (Figure 2C). These findings suggest that compression does not impair NK cell-mediated killing of tumor cells. Taken together, these findings demonstrate that the microfluidic compression device can be successfully used to quantify the independent and combined effects of compression and NK cells on tumor spheroid viability within a physiologically relevant 3D environment.

In contrast, suboptimal outcomes may arise when (1) spheroids are located outside the piston boundary, (2) spheroids are positioned too close to the piston boundary, and (3) multiple spheroids are located within one spheroid diameter of each other (Figure 3). In these cases, the corresponding fields should be excluded from analysis.

Figure 3: Representative example of a sequence of images of spheroids excluded from analysis. Time-lapse brightfield (BF), GFP fluorescence, and merged images of MCF-7 spheroids with NK cells in 1.5 mg/mL collagen before (left column) and after compression (right two columns) are shown. Pre-compression images were acquired 20 min before pressure application and 10 min before the start of time-lapse imaging. Images were subsequently collected every 10 min for 18 h, with 0 h corresponding to the first image acquired immediately after compression and 18 h representing the end of the experiment. Blue arrow indicates the edge of the PDMS piston (L2). This imaging field was excluded from analysis because two spheroids were present within the compression region and were located too close to the piston boundary. As a result, spheroid deformation and GFP fluorescence intensity could not be reliably quantified during compression. Scale bar = 50 µm. Please click here to view a larger version of this figure.

(1) Spheroids located outside the piston boundary do not experience compression because pressure is applied only within the piston region. These spheroids should therefore be excluded from both compressed and control conditions. (2) Spheroids positioned on or near the piston boundary (Figure 3, top spheroid) should be excluded, as the spheroid is experiencing nonuniform mechanical loading. (3) Spheroids located within one spheroid diameter of neighboring spheroids should be excluded (Figure 3). Close proximity between spheroids complicates the quantification of GFP fluorescence intensity because the region of interest (ROI) is defined using the final time point. Over time, spheroid boundaries may become indistinguishable, making it challenging to delineate an ROI that accurately encompasses only a single spheroid. These issues can be minimized by ensuring that spheroids are centered within the compression region during device loading. Careful attention to spheroid distribution, positioning, and device setup is essential for achieving uniform compression and generating reliable, reproducible experimental results.

Components	Concentration
α-MEM media
Fetal bovine serum (FBS)	10%
Penicillin/streptomycin	1%

Table 1: Complete MCF-7 medium.

Components	Concentration
α-MEM media
Myo-inositol	0.2 mM
2-mercaptoethanol	0.1 mM
Fetal bovine serum (FBS)	12.50%
Folic acid	0.02 mM
Horse serum (HS)	12.50%
Penicillin/streptomycin	1%

Table 2: NK complete medium.

Supplementary Figure 1: Schematic of the pressure control system. An air compressor supplies compressed air (up to 20 psi) to the pressure controller. The pressure controller regulates the air pressure delivered to the pressure reservoir based on user-defined settings. The controller has a maximum operating pressure of 200 mbar, a pressure stability of 0.005% full scale (FS), a pressure resolution of 0.003% FS, and a response time of 9 ms, enabling accurate and reproducible pressure control. The controller is connected to a computer via universal serial bus (USB), allowing the user to program the magnitude and profile of the applied pressure using the accompanying software. The software supports multiple compression waveforms, including constant, sinusoidal, triangular, and periodic compression. In this schematic, the pressure is set to 120 mbar, and compressed air at this pressure is delivered to the pressurized reservoir through the pneumatic tube. The pressurized reservoir contains 50 mL of deionized (DI) water and serves to convert pneumatic pressure into fluidic pressure. Once pressurized, the reservoir drives fluid through fluid delivery tubing to the pressure control unit of the microfluidic compression device (see Figure 1 for details of the device), where the fluidic pressure is used to apply compression to the tumor spheroid. Note that fluid pressure is used to avoid air bubbles in the PDMS device. Please click here to download this file.

Supplementary Figure 2: Calibration curve for the microfluidic compression device showing the relationship between piston displacement (, µm) and the pressure applied to the pressure control chamber (KPa). Experimental measurements are represented by dots, while the solid line corresponds to the COMSOL simulation. Error bars indicate the standard error of the mean (SEM) from eight independent experimental measurements but are too small to be visible. This figure is reproduced from Suh et al.¹⁵ with permission from the Royal Society of Chemistry. Please click here to download this file.

Supplementary Video 1: X-Y z-stack of a GFP-labeled MCF-7 spheroid before compression. The spheroid was embedded in 1.5 mg/mL collagen within the microfluidic compression device and imaged using the GFP fluorescence channel with a 1 µm z-step size. Then, ImageJ was used to enhance contrast. Scale bar = 50 µm. Please click here to download this file.

Supplementary Video 2: A compressive strain of 50% (Δh/h = 50%) was then applied to the GFP-labeled MCF-7 spheroid embedded in 1.5 mg/mL collagen. Following compression, an X–Y z-stack was acquired in the GFP fluorescence channel using a 1 µm step size and processed in ImageJ to enhance contrast. Scale bar = 50 µm. Please click here to download this file.

Discussion

This method describes the use of a microfluidic compression device to investigate the effects of compressive stress on tumor–immune interactions. This system is able to apply controlled mechanical compression to tumor spheroids embedded within a 3D collagen matrix. The glass substrate that forms the base of the cell chamber (L1) enables high-resolution live-cell imaging, making the platform well-suited for studies aiming to quantify immune cell motility, infiltration, and cytotoxic activity. Importantly, the system preserves cell-cell and cell-matrix interactions while allowing precise control of the mechanical environment. In contrast, many existing experimental approaches for studying compressive stress require a tradeoff between ECM-mediated cellular interactions^9,13 and high-resolution imaging¹⁴.

In the present study, NK cells and MCF-7 breast tumor spheroids were used as a model system. However, the microfluidic compression device is not limited to these cell types. To date, the platform has been successfully applied to multiple breast cell models, including MDA-MB-231 breast tumor spheroids and MCF-10A breast spheroids^15,17. The platform can be readily adapted to investigate tumor–immune interactions in a variety of other tumor models, including brain and lung tumor spheroids. Similarly, the platform is compatible with a broad range of immune cell populations beyond NK cells, enabling investigations of diverse tumor-immune interactions under mechanical compression. Furthermore, the device is not restricted to spheroid-based studies and can be adapted for single-cell experiments. Together, these features make the platform a versatile tool for examining how mechanical forces regulate cellular behavior across a wide range of biological and disease contexts.

Another advantage of this system is the precise control over the magnitude and pattern of compression applied to the samples. In the present study, a static compressive strain of 50% was applied for the duration of the experiment. However, the level of strain can be readily modified depending on the experimental objective. In addition, dynamic compression of various wave forms, including square-wave and sinusoidal waveforms, can be applied to the spheroids using the external pressure control system. Square-wave compression involves alternating intervals of compression and release¹⁷, while sinusoidal compression involves gradual increases and decreases in compression over time¹⁵. These capabilities, when combined with quantitative imaging and predictive modeling, can be used to study the influence of dynamic compression on the immune killing efficiency.

We note that this compression device was originally developed to characterize tumor mechanical properties¹⁷. By incorporating an in situ force sensor, a polyacrylamide (PAA) gel layer underneath the tumor spheroid, the mechanical properties of the spheroid, such as stiffness and viscoelasticity, can be quantitatively characterized^15,17. This modification expands the utility of the platform beyond modeling compressive stress to include direct measurement of tumor mechanical behavior.

Overall, the microfluidic compression device has a wide range of applications from the measurement of spheroid mechanical properties to investigating tumor-immune interactions within a mechanically modulating environment. However, there are several critical steps required for successful device operation. Tumor spheroids should be positioned near the center of each well to ensure uniform compression. The pressure control unit must be properly aligned with the cell chamber layer (L1) to ensure accurate transmission of pressure. Air bubbles may be introduced when the pressure control unit is placed on top of L1, which can negatively affect imaging quality and experimental outcomes. To prevent this issue, care should be taken to avoid introducing bubbles when loading the collagen mixture, and 8 µL of collagen mixture should be added to each well to slightly overfill the chamber before placement of the pressure control unit. This helps ensure complete contact between the layers and reduces the likelihood of trapped air.

In conclusion, this device enables controlled modeling of compressive stress within a 3D physiologically relevant collagen matrix that supports tumor-immune interactions. This platform provides a tool for investigating how static and dynamic mechanical forces regulate immune-mediated killing and tumor behavior within the tumor microenvironment. Currently, we are identifying the molecular mechanism that underlies the mechanical modulation of immune killing of tumor cells. These studies may lead to potential novel therapeutic drug targets.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
1 N NaoH	Sigma-Aldrich	79724	7 µL
10× M199 medium	Sigma-Aldrich	M0650	200 µL
2-mercaptoethanol	ThermoFisher	313500100
Air compressor	Elveflow	Jun-Air Model 3 Compressor
Automated X-Y microscope stage	Applied Scientific Instrumentation	MS-2000
CCD camera	Hamamtsu Photonics	ORCA-R2
Fetal bovine serum (FBS)	GeminiBio	S11150
Fluid delivery tubing	Eleveflow	PTFE tubing spare roll	1/16 in outer diameter
Fluorescence light source	Excelitas Technologies	X-Cite series 120 PC unit
Folic Acid	Sigma	F7876
Glutaraldehyde	EM Sciences	16019	0.10%
Harrick Plasma Cleaner	Harrick Plasma	PDC-001-HP	High power setting (45W)
Horse Serum (HS)	ThermoFisher	16050-122
Inverted epi-fluorescent microscope	Olympus	IX81	20x objective lens (NA = 0.25), operated using the Olympus CellSens software
Mask	Artus	10 Sheet Aluminum Shim Stock	Thinkness is 0.0125" (0.75 mm)
MCF-7/GFP (Puromycin) stable cell line	GenTarget Inc	SC050-G
Metal Frame	Custom made	Custom made
Microfluidic reservoir	Eleveflow	50 mL microfluidic reservoir
Microscope	Olympus	CKX41	4x objective lens              (NA = 0.1)
Myo-inositol	Sigma	I7508
NK-92 MI	ATCC	CRL-2407
Penicillin/streptomycin	Gibco	15140122
Phosphate-buffered saline (PBS)	ThermoFisher	10010072
Pneumatic tubing	Eleveflow	PUR polyether clear pipe input	4" mm outer diameter
Polyethyleneimine (PEI)	Sigma-Aldrich	P3143	1%
Pressure controller	Elveflow	OB1 NK4	Operated using the Elveflow Smart Interface software
Rat tail type I collagen	Corning	354249	31.3 µL of 9.6 mg/mL stock collagen
Screws	Phillips	M2.5 Pan-head	12 screws
Stage incubator	Precision Plastic	Custom made
Syringe	Becton Dickinson	303134	Paired with a Male luer Integral lock to 3/32" OD Barb from Eleveflow
α-MEM media	ThermoFisher	12561056