Microfluidic assembly of hydrogel-based immunogenic tumor spheroids for evaluation of anticancer therapies and biomarker release

Microfluidic device design and fabrication:

Preparation of the microfluidic devices using soft lithography was conducted as discussed previously.^[20] Negative photo resist SU-8 2100 (MicroChem, Newton, MA) was patterned on the silicon wafers to a thickness of 150 μm. This was then exposed to UV light through a transparent photomask (CAD/Art Services, Bandon, OR). Polydimethylsiloxane (PDMS) (Sylgard 184, Dow Corning, Midland, MI) and its crosslinker were mixed at a ratio of 10:1 and poured onto the device template generated on the silicon wafers. This was subsequently degassed and cured for 12 hours at a temperature of 65 °C. The cured PDMS with the etched microfluidic cannels was then peeled off from the wafers, cut into individual devices and bonded to glass slides after oxygen-plasma treatment of both surfaces. To make the microfluidic channels hydrophobic for efficient droplet generation, devices were treated with Aquapel (PPG Industries, Pittsburg, PA) for 20 min prior to the experiment. Subsequently the Aquapel was expelled by flushing the device with air. All the liquids for droplet generation (cell solutions, oil, etc.) were introduced into the device using Tygon Micro Bore PVC Tubing (dimensions: 0.010″ ID, 0.030″ OD, 0.010″ wall) (Small Parts Inc, FL, USA). Sterile, disposable 1 mL syringes were utilized to load the oil, cell and polymer solutions onto the devices connected by the tubing. The input of the solutions was regulated by syringe pumps (Harvard Apparatus, USA) to control flow rates. The oil phase used for droplet generation consisted of mineral oil and 2% (w/v) of surfactant Span80 (Sigma-Aldrich, St. Louis, MO).

Rheological characterization of alginate-puramatrix hydrogel:

Hydrogel solutions containing 1% w/v alginate and 0–0.25% v/v puramatrix were prepared and kept at 4°C. The viscoelastic properties of the hydrogels were analyzed using a Discovery HR1 hybrid rheometer (TA Instruments, Newcastle, DE) equipped with a with temperature control unit (Peltier plate, ± 0.05 °C) operated in the cone-plate mode with a cone angle of 1° and a 40 mm diameter. Storage (G’) and loss (G”) moduli were measured for solutions of 1% alginate, 1% alginate / 0.15% Puramatrix and 1% alginate / 0.25% Puramatrix in a frequency range of 0.1 – 10 Hz at 25°c, while viscosity was measured at 25°C and 37°C, after initial equilibration at 4°C, at a frequency of 1 Hz.

Scanning Electron Microscopy:

Alginate-puramatrix (1% alginate, 0.15% puramatrix) spheroids were seeded with SUDHL-10 cells at a concentration of 3 × 10⁶ cells mL⁻¹. After incubation for 1–6 days at 37°C and 5% CO₂ in RPMI media with 2% (w/v) calcium chloride, the spheroids were fixed in 2.5% gluataraldehyde, 2.5% (para)formaldehyde in 0.1 M sodium cacodylate (pH 7.2) for 1 hour at 25°C. The samples were washed with 0.1 M sodium cacodylate buffer (pH 7.2) and dehydrated through a graded ethanol series (30%, 50%, 70%, 85%, 95%, 100%) for 10 minutes each. The spheroids were subjected to critical point drying with liquid CO₂ using a Samdri-PVT-3D critical point dryer (Tousimis, Rockville MD) to complete the drying process. Dried samples were adhered onto an aluminum mount with carbon adhesive tabs and coated with 6 nm of platinum using a 208HR sputter coater equipped with a MTM-20 thickness controller (Cressington, Watford, England). SEM images were captured with a Hitachi S-4800 field emission scanning electron microscope (Hitachi, Japan) at an accelerating voltage of 2–3 kV.

Cell culture:

SUDHL-10 cells (DLBCL line) and HS-5 cells (fibroblast line) were originally purchased from American Type Culture Collection (ATCC, Manassas, VA). The SUDHL-10 cells were maintained in RPMI 1640 media supplemented with 10% fetal bovine serum and 1% antibiotic-antimycotic solution (Corning Cellgro, Manassas, VA) and the HS-5 cells were maintained in DMEM medium with 4.5 g L⁻¹ glucose and L-glutamine (Corning Cellgro, Manassas, VA) with serum and antibiotic-antimycotic solution.

Peripheral Blood mononuclear cells (PBMCs) were obtained from Astarte bio (Astarte Biologics, Bothell, WA) and subsequently treated to generate activated PBMCs. Briefly, anti-CD3 antibody was used to coat well plates at a concentration of 10 μg mL⁻¹ for a period of 2 hours. The cells were then seeded at a concentration of 15, 000 cells per well along with an activation cocktail consisting of IL-2 (10 ng mL⁻¹), IL-4 (20 ng mL⁻¹) and soluble anti-human CD-28 antibody (2 μg mL⁻¹) for two days. The plate was centrifuged and re-treated for another two days with IL-2 (10 ng mL⁻¹) and IL-4 (20 ng mL⁻¹). The cells were then harvested, washed and resuspended in fresh IL-2 containing media for experiments.

All cells were grown at 37°C under 5% CO₂ in a humidified atmosphere. Cells were routinely passaged every three days and harvested at a density of 1 × 10⁶ viable cells mL⁻¹.

Spheroid generation and maintenance in microfluidic device:

Sodium alginate (Sigma) was dissolved in media at a concentration of 1% w/v by constant agitation at 50 °C for 2 hours. The solution was stored at 4 °C for a maximum of 2 weeks. Puramatrix (BD Biosciences, Franklin Lakes, NJ) is provided as a 1% w/v solution by the manufacturer and was stored at 4 °C prior to use. It was diluted into various concentrations in the previously prepared alginate solution (0.05%−0.25% v/v) immediately prior to use and was maintained at 4 °C. Calcium chloride (2% w/v, Sigma) solution was prepared in media and sterile filtered using a 0.22 μm syringe filter. It was stored at 4 °C for a maximum of 2 weeks.

Spheroid generation in the microfluidic device was a multi-step process as follows: the first step was the generation the polymer droplets containing the cell suspension and the second step was the gelation of the droplets in the docking array by the calcium chloride solution (2% w/v ≈ 175 mM). 10 × 10⁶ cells mL⁻¹ were used for spheroid generation and for co-culture experiments the cells were used at a ratio of 3:1 (PBMC: tumor and stroma). The cells were suspended in the polymer solution and introduced into the microfluidic device at a flow rate of 75 μL hour⁻¹ and oil was flown at 500 μL hour⁻¹ through the oil inlet. Once the docking array was filled with cell laden droplets, calcium chloride solution was allowed to flow through the third inlet at the rate of 1 μL hour⁻¹ for 4–5 hours for the gelation to occur. Once gelled, the spheroids were stable in the docking array for extended periods of time. Complete media and/or drug solution was perfused through the spheroids in the array for the experimental period (5 days) at a flow rate of 50 μL hour⁻¹. For lenalidomide experiments, the spheroids were treated with a therapeutically relevant concentration of 2.3 μM for a period of 5 days by continuous perfusion. All images were obtained using Zeiss Axio Observer. Z1 Microscope (Zeiss, Germany) equipped with Hamamatsu digital camera C10600 Orca-R2.

Proliferation and cytotoxicity assessment:

Cell proliferation was determined by labeling the cells in the spheroids fluorescently with different cell proliferation trackers (Life Technologies, Carlsbad, CA). SUDHL-10, HS-5 and PBMCs (activated and non-activated) cells were labeled with CFSE cell proliferation dye (ex/em: 492⁄517 nm) for single cell type spheroid experiments. For co-culture spheroid experiments, activated and non-activated PBMCs were labeled with CFSE dye and the SUDHL-10 and HS-5 cells were labeled with CMAC cell proliferation dye (ex/em: 353⁄466 nm). For specific experiments, PBMCs were labeled with CMAC, SUDHL10 cells were labeled with CFSE and HS5 cells were labeled with CMTPX cell proliferation dye (ex/em: 577/602). Cell death was indicated by the uptake of ethidium homodimer-1 (EthD-1) (ex/em: 495/635 nm) at the end of day 5.

Cells were counted on day 1 and day 5; the increase in cell number was normalized to initial (day 1) cell counts and reported as percent proliferation. For cell death analysis, the cells with red fluorescence of EthD-1 were counted at day 5 and percent death was calculated by dividing the number of dead cells by total number of cells at day 5.

Cell response in 2D monolayers:

Viability in immunogenic microfluidic spheroids was compared with viability in 2D monolayer cultures. Cells were seeded at a density of 7500 cells per well and treated with lenalidomide as in the microfluidic experiments (2.3 μM for 5 days).

Secretion analysis:

The media perfused through the microfluidic spheroids for all treatment conditions was collected at 24 hour intervals for 5 days. These perfusions were analyzed for 92 protein biomarkers using Proseek Multiplex® immune-oncology panel (Olink Proteomics, Uppsala, Sweden). Processing, output data quality check, and normalization were performed by Olink Proteomics. All values were presented as Normalized Protein expression (NPX) values on a log2 scale. The log2 NPX values were scaled to mean 1 and SD 1 before data analysis, to facilitate comparisons between protein associations. Data values below the level of detection (LOD) are highlighted in red.

Statistical Analysis:

The cell numbers were measured using the colony counting extension of ImageJ software (https://imagej.nih.gov/). For all studies with microfluidic spheroids, a minimum of 30 spheroids was analyzed for every condition, and n=2 experimental repeats were conducted ^[20]. For 2D monolayer culture, all conditions were repeated in triplicates. Statistical analysis was done by comparing the data sets using the student’s t-test, and p value < 0.05 was considered statistically significant.

3.1. Rheological characterization of alginate and puramatrix combination polymer for spheroid generation

Alginate has been used previously in combination with collagen, fibrin and matrigel to promote cell survival and growth in tumor spheroids.^[28,29,30] However, alginate does not express cell-binding moieties, which makes it challenging to observe cell proliferation and subsequent interaction in alginate gels. In this study, we combined alginate with puramatrix to increase cell adhesion and aggregation. Puramatrix has low pH (pH=3 at 1% solution), which could potentially reduce cell viability.^[23] To minimize this effect, puramatrix was diluted (0.1–0.25%) and added to 1% w/v alginate in the combination hydrogels.

Extensive rheological characterization of the mixed hydrogel combinations was performed prior to microfluidic application (Figure 1). Since the hydrogel solution must be continuously flown in the microfluidic platform to generate droplets, it is necessary to assess the viscosity of the hydrogel prior to gelation so as to ensure smooth flow of cell and aqueous gel streams. Here, dynamic viscosity of the alginate-puramatrix hydrogel, defined as the measure of the fluid’s internal resistance to flow, was measured. The viscosity of the combination hydrogel, prepared by mixing a range of puramatrix (0.05–0.25%) dilutions with 1% w/v alginate, was compared to non-modified 1% alginate solution at 25°C and 37°C (Figure 1A). Dynamic viscosity did not vary significantly for puramatrix concentrations up to 0.15% (0.01 ± 0.005 Pa. S) but increased thereafter with increase in puramatrix concentration (0.20–0.25%). Next, the storage modulus (G’), indicating elastic behavior, and the loss modulus (G”), indicating viscous response was determined at frequencies 0.1–10 Hz. As shown in Figure 1B, G“ of alginate is greater than G’, particularly at lower frequencies (0.1–1Hz), depicting liquid-like responses. The addition of puramatrix reversed the response (G’ > G”), yielding the characteristic behavior of hydrogels. This result suggests that puramatrix increases the elasticity of the combination polymer solution and promotes formation of hydrogels. Lastly, the porosity of the hydrogels was also assessed. The pore size of 0.15% puramatrix-1% alginate combination, which supported cell clustering and proliferation, was found to be 0.005–0.2 μm, while that of 1% alginate was 0.005–0.1 μm (Figure 1C-D). We also determined cell proliferation in combination hydrogels. Cells aggregated and proliferated in clusters in 0.1–0.25% composite hydrogel, in agreement with previous studies with puramatrix only (Figure S1).^[23] Based on these data, we selected the combination of 1% alginate and 0.15% puramatrix to form microfluidic spheroids with improved hydrogel strength while maintaining the elasticity offered by puramatrix.

3.2. Fabrication of immunogenic tumor spheroids using droplet microfluidic platform

Here we describe a droplet microfluidics approach for the generation of hydrogel-based 3D lymphoma spheroids (Figure 2A-B). The microfluidic device can potentially house and monitor 250 cell-laden spheroids in the integrated docking array (Figure 2A). The generation of spheroids occurred in two stages. In the first stage, droplets containing cells and liquid hydrogel (1% alginate and 0.15% puramatrix) were formed at the T-junction and driven into the docking array (Figure 2C-D). The individual docking sites for the spheroids are 400 μm in diameter. The optimal droplet size was thus maintained at 350 ± 25 μm in diameter to prevent neighboring droplets in the array from merging prior to gelation. The second phase occurs after stabilizing the droplets in the docking array. Calcium chloride (2% w/v solution) in complete growth media was perfused at a slow and controlled rate of 2 μl/hour to ionically crosslink the hydrogel in droplets (Figure 2E).^[20] This external gelation mechanism allows for polymerization of the hydrogels in situ while the slow perfusion helps maintain the spherical structure of these constructs.

The solid-phase spheroids are trapped in the docking array for extended periods of time ( Figure 2F), which permitted us to monitor spatiotemporal dynamics in the same spheroid throughout the experimental duration. The generated spheroids were continuously perfused with media/or drug solution at 50 μL/h, equivalent to 575 μm/s, for 5 days. This flow rate mimics the blood velocity in the tumor, which has been reported to vary over a range of 100–800 μm/s, depending on the tumor size in vivo.^[31] Continuous perfusion has two advantages in our system: supplying nutrients to the spheroids for cell survival over long periods, and ability to assess cell secretions at specific intervals. Media was collected from the spheroid array outlet every 24 hours to quantify cell-secreted factors such as cytokines and chemokines as described below.

The cellular components of the spheroid were selected to mimic the TME of NHL, which is comprised of tumor cells, fibroblasts and immune cells.^[32,33] For the tumor cell component, a germinal center B cell (GCB) lymphoma line SUDHL-10 was utilized. The stromal component was represented by bone marrow fibroblast line HS-5.^[34] SUDHL10 and HS-5 cells were mixed at a ratio of 1:1.^[19] Primary peripheral blood mononuclear cells (PBMCs) from healthy donors were incorporated as the immune component. Since 75% of the cell population in the lymphoma TME are immune cells, the initial seeding density ratio of immune cells: non-immune cells (i.e., cancer cells and fibroblast) was maintained at 3:1 (Figure S2A). Prior to generating a complex co-culture spheroid, control spheroid models consisting of one cell type (SUDHL-10 cells (Figure S2B), HS-5 cells (Figure S2C), activated PBMCs (Figure S2D), and non-activated PBMCs (Figure S2E) were also formed to test the compatibility of each cell type and the viability in the microfluidic system.

3.3. Effect of immunomodulatory drug on cell survival in tumor spheroids

Several reports have suggested distinct differences in cell survival in 2D monolayers compared to 3D spheroids.^{[19, 35]} Here we compared cell survival in 2D monolayer and microfluidic spheroids (Figure S3, Figures 3–6). We used two methods to assess overall cell heath in all models: (a) cell proliferation, and (b) overall death at day 5. Given the complex mixture of cells in the immunogenic tumor spheroids, we first determined the viability of cells in monoculture spheroids (i.e., containing one cell type) before progressing to the co-culture models. We further determined the impact of the immunomodulatory drug lenalidomide on cell death, as this drug has demonstrated anti-cancer effects in various types of cancers including lymphoma.^[34,36,37] The concentration of lenalidomide tested was 2.3μM, a relevant concentration used for clinical studies.^[38]

3.4. Proteomic profile of factors released from 3D microfluidic spheroids

Release of secretory factors such as cytokines and chemokines by the immune cells in the TME plays an important role in tumor progression and response to therapies. Lenalidomide promotes tumor regression by downregulating proinflammatory cytokines, increasing recruitment and activating effector cells (cytotoxic T cells and NK cells).^[36] Since the microfluidic array was continuously perfused with media, we were able to collect conditioned media periodically (every 24 hrs) from the spheroid array. We investigated a total of ninety two protein biomarkers relevant to various biological processes as shown in Figure 7. Here, we discuss the ten key molecules released from 3D spheroids and changes in their levels in the presence of lenalidomide.

Angiopoietin-I (ANG-I), a key vascular growth factor, was detectable in all microfluidic spheroid cultures. Treatment with lenalidomide reduced ANG-1 secretion. CD137 (also referred to as Tumor Necrosis Family Receptor Superfamily member 9) and CD40 (also a member of the TNFR superfamily) were detected, although these molecules are surface proteins. Lenalidomide decreased the amount of both these molecules. Likewise, lenalidomide treatment resulted in the downregulation of proinflammatory cytokines and chemokines such as IL-6, IL-8, CCL2 and CCL3 (Figure 7A). IL-6 was detected in all on-chip spheroid models but reduced by lenalidomide in experimental conditions containing activated PBMCs by 2-fold.

IL8 is known to be predominantly secreted by tumor cells, macrophages and epithelial cells.^[40] IL-8 was released in larger quantities by ACT-NHL spheroids compared to mono-culture spheroids, and there was a 30-fold decrease in IL8 concentration from ACT-NHL spheroids due to lenalidomide treatment (Figure 7B). Galectin-9 (Gal-9) concentration was reduced one-fold in lenalidomide treated conditions of ACT-NHL as compared with control. In contrast, Granzyme B secretion was upregulated due to lenalidomide treatment in spheroids containing activated PBMCs, either singly (two-fold increase in protein concentration) or in co-culture NHL spheroids (20-fold increase in concentration) (Figure 7B). Also, the release of granzyme B in the activated NHL was relatively higher at 24 hours compared to 72 hours since the release of granzyme B from NK cells is a rapid process.^[41] Thus, secretory analysis of microfluidic spheroids clearly indicated differences in cell secretions due to immunomodulation by lenalidomide.