A Novel Omniphobic Platform for Multicellular Spheroid Generation, Drug Screening, and On-Plate Analysis

Abstract

Multicellular spheroids are superior to other culture geometries in reproducing critical physiological conditions of tumors, such as the diffusion of oxygen, nutrients, waste, and drugs, leading to a more precise model of in vivo drug sensitivity and resistance. Previously reported spheroid culture platforms are often difficult to use, expensive, single use or mechanically unstable. Here we report a facile, mechanically stable, high-throughput spheroid culture platform based on hierarchically textured omniphobic surfaces. The developed omniphobic surfaces display very high contact angles with a range of different liquids, including the cell-laden culture media, and thereby minimize the cell-surface contact area. Additionally, these surfaces maintain these high contact angles for extended periods of time to ensure cell aggregation. Using this novel platform, we demonstrate the generation and maintenance of robust multicellular spheroids, as well as heterogeneous, multi-cell-type spheroids. The platform is extremely robust, resistant to mechanical shock, allows for on-plate imaging, and also the first-ever spheroid generation platform that can be reused repeatedly. Finally, the platform is suitable for on-plate drug screening and enables the first-ever, on-plate immunofluorescence staining and imaging of spheroids.

Introduction

Fast, inexpensive and accurate replication of in vivo cancer cell biology using in vitro models allows for the rapid screening of drugs and drug delivery systems. This in turn can enable greater individualization of cancer treatments.^1–6 Given the failure of various 2D models to successfully predict in patient drug efficacy, numerous 3D in vitro methods have been developed. These methods more accurately replicate the in vivo cell microenvironment in terms of geometry, mechanical properties, and chemistry.^{1, 2, 7, 8} One such 3D model, free-floating multicellular spheroids, demonstrates high-fidelity to in vivo tumors, allows for long-term cultures, and is widely applicable for cancer drug screening, metabolism, discovery, and preclinical studies.^{2, 4, 9} Spheroids reproduce critical physiological conditions of tumors, such as the diffusion of oxygen, nutrients, waste, and drugs.^{2, 4, 9, 10} They are also a more precise model for in vivo drug sensitivity, as well as resistance, and display increased chemoresistance when compared to 2D platforms.^{7, 10} As a result, these model systems have been utilized in many biomedical applications, including fundamental cancer biology, mechanistic studies, translating discoveries to delivery systems, and precision medicine. Cells in 3D suspension self-organize into free-floating microscale spherical aggregates in the absence of any exogenous extracellular matrix – a process which closely resembles the growth and metastasis of many malignant cancers, including ovarian, breast, brain, prostate, colon, lung, melanoma, and others.^{6, 11} Additionally, spheroids can be combined with microfabrication techniques to rapidly generate complex tissues.¹²

There are numerous ways of generating multicellular spheroids, including well plates with non-adhesive surface treatments, agitated culture vessels, hanging drops, microgravity, and microfluidic devices.^{2, 13–16} One of these techniques, hanging drop plates, are the most physiologically relevant, relatively facile to fabricate, and can produce uniformly shaped and sized individual spheroids in each drop, enabling precise assays.⁴ In these plates, the bottom surface of the droplet, where cells tend to aggregate due to gravity, only contacts air and prevents cell attachment to the surface (see Figure S1).

Recently, specialized, commercial, high-throughput hanging drop plates have greatly improved spheroid culture efficiency within this platform.^{13, 15, 17, 18} We have previously established the efficacy of the 384-hanging drop spheroid platform in generating spheroids from a small number (1–100) of patient-derived cells for anti-cancer drug screening.^3–6 This platform, while excellent for generating a physiologically relevant cancer cell environment, is unstable when droplet volumes increase to > 25 µL, which can happen readily with the repetitive addition of medium, drugs, or reagents for molecular biology assays.^{3–6, 19} Even at smaller droplet volumes, there is a risk of cell-laden droplets falling or spreading due to any mechanical shock from routine plate handling. This can be catastrophic for the many different multicellular spheroid experiments that can take several days to weeks to complete. 384-hanging drop plates are also expensive and can only be used once reliably. To address these challenges, and to allow for on-plate assays, we have created a new spheroid generation platform based on hierarchically textured omniphobic surfaces.

Omniphobic surfaces utilize re-entrant texture to entrap pockets of air underneath a contacting liquid, forming the so-called Cassie-Baxter state or a composite solid-liquid-air interface. This composite interface allows the omniphobic surfaces to simultaneously achieve high contact angles (θ ) and low contact angle hysteresis (∆θ – the difference between contact angles measured when a liquid advances or recedes on a surface) with a broad range of contacting liquids.^20–23

The ideal omniphobic surface for spheroid generation must enable the cell-laden culture media droplet to remain in a stable, non-wetted Cassie-Baxter state for extended periods of time. This would ensure the non-attachment of cells to the underlying surface, and cell aggregation to form spheroids. However, retaining omniphobicity in a cell-culture environment can be challenging for multiple reasons, including: 1) Cell culture growth medium contains salts and other additives that reduce its surface tension (see Table S1). 2) Cells secrete proteins that can readily adhere to even low-surface-energy materials.^{24, 25} 3) The incubation environment is held at high relative humidity to minimize the loss of droplet volume due to evaporation. Under such humid conditions, omniphobic surfaces are vulnerable to wetting due to condensation of droplets within their porous texture.²⁶ 4) The small, spherical droplet of cell growth medium (typically < 20 µL) applies a significant Laplace pressure on the entrapped air pockets, which can lead to liquid imbibition or surface wetting. We overcame these challenges by designing a surface with hierarchical scales of texture with the smallest texture features having nanoscale dimensions.²⁷ Hierarchically structured surfaces show significantly decreased contact angle hysteresis with a wide range of liquids due to reduced solid-liquid contact area, while maintaining a high breakthrough pressure (the applied pressure that causes a transition from the non-wetting Cassie-Baxter state to the wetted Wenzel state). ^28–33 Further, nanotextured omniphobic surfaces can exhibit extremely high resistance to wetting even when immersed for extended periods at elevated pressures.^34–36

Experimental Section

Omniphobic aluminum mesh spheroid platform fabrication:

A highly pressure-resistant, hierarchically nanotextured omniphobic surface was produced on aluminum by a 3-step process. The first step was acid etching, followed by immersion in boiling water and treatment of the surface with a fluorosilane to lower the surface energy. This process was adapted from a method previously reported in the literature for superhydrophobic and superoleophobic surfaces.³⁷

The substrates were 6061 T6 aluminum alloy plates or 5052 alloy aluminum meshes with varying dimensions (McMaster-Carr). Results are reported with a size 14×14 mesh, which has 51% open area with a wire diameter of 0.02”, and 196 pores per square inch. First, the surfaces of the plates and meshes were abraded with a fine wire brush, washed with soap and deionized water, then dried with compressed air. This removes any surface oxides and contaminants and exposes pure aluminum. The substrates were then immersed in 2.5 M hydrochloric acid (Fisher Scientific) held at 60°C on a hot plate until they were uniformly etched, and all of the original planar surface was removed. This created a micro-structured surface (Figure 1C).³⁸ Depending on the effectiveness of pre-cleaning, the exact aluminum alloy, and the geometry of the sample, complete etching was achieved between 5 and 20 minutes, and was monitored by periodic visual inspection of the sample. While there is some evident variation in the morphology of the microtexture due to variation of the effectiveness of the acid etch, this does not appear to have a significant effect on the contact angles or pressure stability of the resulting substrates. Following etching, the majority of loosely adhered aluminum particles were removed with a combination of agitation in a water bath sonicator, rinsing in water and drying with compressed air. The etched aluminum was then boiled in deionized water for one hour, which caused the conformal growth of boehmite (γ-(AlO(OH))) nano-platelets (Figure 1D).

Figure 1. Novel Omniphobic Platform for Multicellular Spheroid Generation.

A) A schematic illustrating the developed platform. The inset shows an image of the complete device with the lid removed. B, C, D) SEM images for the omniphobic aluminum mesh. The three images highlight the three hierarchical length scales of texture: B) the mesh geometry, C) microscale texture on the wire surface, D) nano-platelets of boehmite. E) 10 µL droplets of water, hexadecane, and dodecane within the wells of the omniphobic spheroid platform, as well as on top of the treated aluminum mesh plate. F) A schematic diagram for the impact test set-up. G, H) Images of 4 wells on the omniphobic mesh plate with 20 µL droplets of cell media G) before and H) after 6 successive impact tests. Droplets are fully intact, and display high contact angles both before, and after impact.

To make the etched substrates omniphobic, they were exposed to 1H,1H,2H,2H-Heptadecafluorodecyl triethoxysilane (Gelest) vapor for 48 h at 60°C at 5 mTorr in a vacuum oven. All samples exhibited water contact angles >160° with ∆θ approaching 0°. The resistance to wetting of the samples under pressure and time was evaluated by placing the samples into a cylinder of DI water at a depth of 30 cm (P = 3 kPa). The samples appeared highly reflective, indicating the presence of trapped air, or a plastron.³⁹ While this diminished over time due to dissolution of the excess plastron into the water,³⁶ the surfaces remained completely non-wetted when removed from the water column over 30 hours later.

Modified design of omniphobic mesh for increased throughput:

To improve ease of handling, particularly during imaging, and to increase overall throughput, we adapted the mesh substrates to a higher density multi-well cell culture plate form factor. Sidewalls to contain the droplets were formed by drilling an array of sixty 6.5 mm diameter holes in a 2 mm thick sheet of 6061 aluminum alloy (McMaster), using standard 96-well plate spacing. The sidewalls were treated as described above for the aluminum mesh to make them omniphobic. The meshes were clamped to the base of the sidewalls, using screws and laser-cut poly(methyl methacrylate) mounting plates, and secured within a modified well plate incorporating a water reservoir along its perimeter (Figure 1A). This reservoir reduces droplet evaporation during cell culture by maintaining a saturated environment. This assembly was placed between a standard well plate lid and a 6 well plate filled with water, and the sides were sealed with Parafilm during culturing. For imaging, the center assembly was removed from this bottom well plate, to accommodate the working distance of the inverted microscope. All parts were deemed to be re-usable with sterilization, and the meshes could be easily exchanged in the assembly if necessary.

Droplet stability analysis on the omniphobic spheroid platform:

The stability of droplets plated on both the omniphobic mesh plate and the 384-hanging drop plate was evaluated via an impact test. A 127.3 g, 11 cm metal rod, was set up to rotate around a fixed, horizontal axis oriented at the end of the rod (Figure S1A). The rod was raised to a position horizontal to the bench top and released. The drop plate being tested was placed such that it was impacted by the swinging rod when the rod was at a position perpendicular to the benchtop (Figure S1A). The kinetic energy (E_k) of the impacting rod was calculated by the formula $E_k=\frac{1}{2}\left(\frac{m{l}^2}{3}\right){\omega }^2$, where m, l, and ω are the mass, length and angular velocity of the rod, respectively. The angular velocity of the rod was determined by recording the time required for the rod to rotate from its released position to the point of impact with the plate. Each impact test was recorded by slow-motion camera to improve the accuracy of this time measurement. Each tested plate was held in place by a heavy, metal backstop on the side opposite the impacting rod. Each platform was plated with 20 µL droplets of cell media. In one test cycle, each plate was subjected to two successive impacts by the swinging rod. Between cycles, each plate was rinsed with deionized water, then dried with compressed air. The number of lost or damaged droplets was counted between impacts. Three testing cycles were executed for each plate.

Omniphobic aluminum mesh plate and 384-hanging drop plate sterilization:

Prior to any cell culture experiments or droplet stability tests, the omniphobic substrates were cleaned by thoroughly washing in sterile phosphate-buffered saline solution (PBS) and deionized water, then sterilized either with a germicidal ultraviolet lamp on both sides or with ethylene oxide gas. For reusability analyses, the mesh plate was also thoroughly washed and sterilized by this method, before being re-used. The 384-hanging drop plate was sonicated for 20 minutes before submerging in 0.1% Pluronic solution overnight. The plate was then washed with deionized water and sterilized by the above method before use.

Cell culture:

Cell culture reagents were purchased from Life Technologies (Carlsbad, CA), unless otherwise specified. High grade serous ovarian adenocarcinoma cell line, OVCAR8 and Human Embryonic Kidney 293, labelled with Green Fluorescence Protein (HEK-GFP 293) were obtained from ATCC (Manassas, VA). The ovarian cancer patient sample Pt224, was obtained under IRB approved protocol with written consent, from the ovarian tumor bank at the University of Michigan. OVCAR8 were cultured in RPMI 1640 supplemented with 10% fetal bovine serum and 1X antibiotics/antimycotics. HEK-239 were cultured in DMEM supplemented with 10% FBS along with 2 µg/mL puromycin and 200 µg/mL hygromycin B. Patient primary sample derived cancer stem cells (FACS sorted for CD133+ and ALDH+) were cultured in B-27 serum-free medium, composed of a 1:1 mix of DMEM and Ham’s F-12, supplemented with 1X B-27 supplement without vitamin A, 1X MEM Non-Essential Amino Acids, 1X insulin-transferrin-selenium, 5 ng/mL EGF, and 5 ng/mL bFGF.^{3, 5} These cells were maintained in a humidified 37 °C incubator with 5% carbon dioxide.

Spheroid formation:

Spheroids were formed on the 384-hanging drop plates, purchased from XCentric Mold and Engineering (Clinton Twp, MI) as previously described.^3–6 For initial spheroid formation, cells were trypsinized from plates of about 70% confluency, and counted using a hemocytometer. Cells were diluted to contain 100 cells per 20 µL and plated in 20 µL drops, following the procedure described previously.^3–6 The spheroids on the hanging drop plates were imaged every 2–3 days and 3 µL of medium was added to them. Spheroids were maintained using previously established methods.^4,7 For the omniphobic mesh plate, OVCAR8 cells were cultured in growth medium until 70% confluency, then diluted to the final seeding density of 500 or 1000 cells per 20 µL droplet of media. Cell growth was monitored with periodic imaging, and cell medium was replenished on alternate days as necessary. For in situ imaging in the fabricated well plate, the bottom reservoir plate was removed to accommodate the working distance of the inverted microscope (Olympus IX81 equipped with an ORCA R2 cooled CCD camera and CellSens software).

Cell viability and quality of spheroids formed on omniphobic high throughput platform:

Cell viability and quality of spheroids formed on aluminum mesh was quantified using the Live/Dead viability kit (Life Technologies, Carlsbad CA). A staining solution of 2 μM Calcein AM and 4 μM ethidium homodimer-1 was introduced to the droplet, with a subsequent incubation of 1h. Viable cells were selectively stained fluorescent green with calcein-AM due to ongoing intracellular esterase activity, while the plasma membrane breakdown in dead cells enabled their DNA to be stained fluorescent red with ethidium homodimer-1. The stained cells were carefully transferred to glass coverslips and imaged using a laser scanning confocal microscope (Olympus IX81, equipped with a Yokogawa CSU-X1 confocal scanning laser unit, Andor iXon x3 CCD camera, and Metamorph 7.8 software), with constant gain and exposure settings between samples. Fluorescence images were obtained at every z-axis encompassing the spheroids, at 488 nm for calcein-AM (live cells; green) and 561 nm for ethidium homodimer-1 (dead cells; red). A composite z-stack image was obtained using Metamorph for live and dead cells within the spheroid. ImageJ (National Institutes of Health) was used to quantify the percentage of live and dead cells within the spheroid from the z-projection images. The numbers of live and dead cells were expressed as a percentage of the total number of cells within each spheroid.

Transduction of patient-derived cancer stem cells with mTomato and heterogeneous spheroid generation:

Pt224 cancer stem cells were plated at a density of twenty thousand per 6 well and left to attach to the bottom of the plate, to reach confluency of 40%–50% to ensure transduction efficacy. Medium was aspirated and 1.45 mL of fresh media was added to each well along with 4 μg/mL polybrene (Sigma TR-1003 –G) stock and virus stock at 1X concentration. After incubation for 24 hours, virus was aspirated to a bleached container and fresh medium is added to the wells. Plates were imaged 48 hours after transduction to assay for transduction efficiency. For co-culture spheroids, Pt224-mTomato and HEK-239-GFP cells were cultured in their respective growth medium until ∼70% confluency, trypsinized per regular passage and counted on a hemocytometer. Cells were re-suspended in complete medium and 20 μL of 1:1 cell suspension with 500 cells each of MTOM and 500 GFP cell suspension was added to each well of the hanging drop array plate and the aluminum mesh.

Drug treatment of spheroids:

On day 3 after seeding, spheroids were imaged to ensure formation and then treated with 10 µM paclitaxel or 20 µM doxorubucin, an established chemotherapeutic drug used in the clinic, or with media only as a control. Forty-eight hours after this treatment, phase-contrast images of the drug-treated spheroids were obtained, and the cells harvested for counting. Ten spheroids were harvested from each drug treatment group, disassembled via mechanical disruption and counted with trypan blue using a hemocytometer to assess viability.

On-plate immunofluorescence on omniphobic high-throughput spheroid platform:

All steps were performed on spheroids sitting on the aluminum mesh, while being monitored under a microscope. Spheroids on the mesh were fixed in 100% ice-cold methanol for 10 minutes. After fixing, drops containing spheroids were washed with wash buffer (0.05% Tween 20 in 1X PBS) for 5 minutes at room temperature. Blocking of non-specific staining was done by incubating in 10% Horse serum, 5% Goat serum and 0.3% Bovine serum albumin for 2 hours at room temperature. Samples were washed again with wash buffer before incubating with antibodies diluted in 5% goat serum in 1XPBS of RAD51 conjugated to Alexa Fluor 546 (Santa Cruz, CA), and to visualize F-actin Phalloidin conjugated to Alexa Fluor 488 (Life Technologies, CA) and nucleus was visualized using 4′6′-diamidino-2-phenyle indole (DAPI) (Molecular probes, USA) for 3 hours in the incubator. The spheroids were then washed with wash buffer and carefully transferred to a cover glass bottom plate in 2% agarose to be imaged using a laser scanning confocal microscope (Olympus IX81, equipped with a Yokogawa CSU-X1 confocal scanning laser unit, Andor iXon x3 CCD camera and Metamorph 7.8 software), with constant gain and exposure settings between samples.

Statistical analysis:

All experiments were repeated with n > 3 biological replicates in order to carry out statistical analysis. Statistical data was analyzed using GraphPad Prism 7.0 (www.graphpad.com, San Diego, CA). All data is represented as mean ± standard error of the mean. Where appropriate, one-way or two-way ANOVA were performed to assess statistical significance between means, with post-hoc Tukey tests for comparison between means. A p ≤ 0.05 was considered significant. Levels of statistical significance are indicated in graphs, where appropriate, with asterisks.

Results and Discussion

As part of our design, we first incorporated an open mesh substrate geometry within our omniphobic platform to allow for the unobstructed imaging of spheroids from the underside of the droplet using an inverted microscope. Figure 1A provides a schematic illustration of the overall spheroid generation platform that incorporates the omniphobic mesh surface. The selected aluminum mesh inherently possesses the re-entrant texture (Figure 1B) that is essential for repelling low surface tension liquids. Next, we etched the aluminum mesh surface³⁷ at different length scales to incorporate air pockets at multiple scales of texture (see Figure 1B, C, D). Finally, the etched mesh surface was reacted with a fluorosilane to lower its surface energy (additional experimental details are provided in Supporting Information). The fabricated omniphobic mesh surface could repel a wide range of contacting liquids including water, cell growth medium, and multiple low surface tension oils (Figure 1E, Table S1). To improve ease of handling, particularly during imaging, and to increase overall throughput, we adapted the omniphobic mesh substrates to a high-density, 60-well cell culture plate form factor, with reservoirs for DI water to prevent droplet evaporation (Figure 1A). This design maintained sufficient humidity to successfully prevent even 15 µL droplets from evaporating during 7 days of spheroid culture.

For spheroid culture, the mesh porosity was selected to maximize the open area beneath plated droplets to facilitate imaging, while still maintaining a high breakthrough pressure (see Table S2). The stability of the platform was further evaluated by subjecting the 60-well mesh plate loaded with 20 µL droplets of cell growth medium to mechanical shock. The plate was evaluated for its ability to retain droplet integrity in comparison to the previously reported 384-hanging drop spheroid platform. When impacted by a swinging rod (Figure 1F) with a kinetic energy of 18.7 mJ (equivalent to a 1 kg mass travelling at 20 cm/s), a 384-hanging drop spheroid platform lost 58 ± 12% of droplets after a single impact, and 72 ± 5% after two consecutive impacts (Figure S2). In contrast, the omniphobic plate kept all droplets intact after six successive impacts (Figure 1G, H). Droplets in the 384-hanging drop plate were observed to either fall off or spread across the hydrophilic plate surface during impact (Figure S2). Conversely, droplets resting in the wells of the omniphobic platform were untethered, yet supported from underneath by the omniphobic mesh surface, allowing them to oscillate or bounce due to mechanical shock and then return intact to their original location under gravity. These results demonstrate the superior stability of droplets plated in the 60-well mesh plate, in providing improved ease of handling and decreasing the risk of accidental sample loss over the course of long-term spheroid culture experiments.

For a direct comparison between our omniphobic mesh plate and a 384-hanging drop plate, we seeded 1000 high grade serous ovarian cancer OVCAR8 cells within a 20 µL cell culture medium (containing RPMI 1640 and 10% fetal bovine serum) droplet in both sets of plates and observed them over 7 days. This cell line represents chemoresistant and recurrent high grade serous ovarian cancers.⁴⁰ Spheroids with similar size and morphology formed by day 3 in both the 384-hanging drop plate and the omniphobic mesh plate (Figure S3A, B). On the omniphobic mesh plate, some cell aggregates were observed on the periphery of the main spheroid. However, by day 7, all cells had compacted into a single spheroid. This can be attributed to the increased cell-cell communication, cell aggregation and compaction in 3D culture.^{3, 4, 6}

The omniphobic mesh plate surface remains non-reactive, non-toxic and inert during spheroid culture, making this an ideal platform for generating multicellular tumor spheroids. This non-reactivity and lack of toxicity is best illustrated by analyzing the cellular viability within the generated spheroids (Figure 2A, B). At day 7, robust calcein-AM (green) live cell stain was observed in all spheroids. Few scattered dead cells were detected with ethidium homodimer (red fluorescence), primarily within the oxygen-deficient spheroid core. Additionally, the spheroids growing on the omniphobic mesh plate displayed a similar rate of proliferation as those growing on the 384-hanging drop plate. This was confirmed via phase contrast images taken on day 7 (Figure S3A, B) and alamarBlue proliferation.

Figure 2: Omniphobic aluminum mesh platform can successfully generate mono- or heterospheroids that are comparable to those created on a 384-hanging drop plate.

A) Representative confocal image of live (Calcein AM, green) and dead (Ethidium Homodimer, red) cells within 1000 OVCAR8 cells/20 μL spheroids generated on an omniphobic mesh plate, imaged on day 7. B) Quantification of live and dead cells within spheroids generated on the omniphobic mesh plate, from confocal images (two-tailed t-test, n>3, *p<0.0002). C) Representative phase images of spheroids on the omniphobic mesh plate (OMP) and 384-hanging drop plate (384-HDP) after 48 hours treatment with 10 μM of the anti-cancer drug paclitaxel. D) Quantification of percent viable cells within spheroids after paclitaxel treatment, as analyzed by resorufin fluorescence. The viability was normalized to control treatment with cell culture medium (one-way ANOVA, n>3, *p<0.0001). The omniphobic mesh plate demonstrated equivalent spheroid viability on the first (OMP-1) and fourth (OMP-4) use of a plate. E) Representative confocal microscope images of heterogeneous spheroids generated with 1:1 ratio of mTomato labeled Pt224 ovarian cancer stem cells (500 cells, CD133+ and ALDH+) and GFP labeled human embryonic kidney (HEK) cells (500 cells) on day 5. F) Graph representing viability of 1000 cells spheroids generated using OVCAR8 and OVCAR3-GFP cell lines, in response to 24 hr of chemotherapeutic drug (20 µM doxorubicin, 10 µM paclitaxel) across three different spheroid generating platforms: omniphobic mesh (OMP), 384 hanging drop plate (384-HDP) and commercially available, 96 well U-bottom plate (U-bottom). A significant difference in viability was observed (two way ANOVA, n=4, ****p< 0.0001) when comparing no treatment control to Doxorubicin and Paclitaxel across all platforms. Additionally, a significant difference was observed in (two way ANOVA, n=4, **** p<0.0001) OVCAR3-GFP OMP & HDP, compared to OVCAR3 U-bottom, (two way ANOVA, n=4, **** p<0.0001), and in OVCAR8 OMP compared to OVCAR8 U-bottom. There was also significant change (two way ANOVA, n=4, *** p<0.0008) OVCAR8 HDP compared to OVCAR8 U-bottom.

Tumor microenvironments are highly complex, with a high degree of cellular heterogeneity. In order to demonstrate the ability of the omniphobic mesh plate to generate heterogeneous multicellular spheroids, which closely mimic clinical tumor microenvironments, we generated fluorescently labeled co-culture spheroids using patient-derived cells. The heterogeneous spheroids composed of two cell types were fabricated with lentivirally transduced primary patient-derived ovarian cancer stem cells (Pt 224-mTomato; red fluorescence, sorted via flow cytometry for CD133+ ALDH+ ^{3–5, 19, 41, 42}) and human embryonic kidney (HEK 239-GFP; green fluorescence) cells, on the 384-hanging drop and omniphobic mesh plates. As seen in Figure S3C, there is no morphological difference in the spheroid phase contrast images obtained on the two different platforms. The confocal projections demonstrated a core of the HEK293-GFP within the spheroids on both platforms, which was surrounded by the mTomato labeled patient-derived ovarian cancer stem cells (Figure 2E).

One of the major disadvantages of the conventional 384-hanging drop plate is its lack of re-usability. To evaluate if the developed omniphobic mesh platform has similar constraints, we used the same mesh plate multiple times to form spheroids after sterilization subsequent to each use. Representative phase contrast images showing spheroids obtained from a 384-hanging drop plate, as well as those from the 1^st and 4^th use of an omniphobic aluminum mesh plate, are provided in Supporting Information (Figure S4A). Multiple uses of the omniphobic mesh plate did not seem to deter spheroid formation or cause any marked difference in spheroid morphology. After 4 uses, the substrate retained its omniphobic nature and showcased comparably viable and stable cultured spheroids.

3D culture mimics in vivo drug sensitivity and chemoresistance far more accurately than any 2D platform, making it a superior approach for in vitro testing.^1–6 To investigate whether spheroids grown on our omniphobic mesh plate display similar chemoresistance to those grown in a hanging drop plate, 1000 OVCAR8 cells in a 20 µL cell culture medium droplet were allowed to form into spheroids (day 3) on both types of plates. Next, the spheroids were exposed to 10 µM paclitaxel, a standard chemotherapeutic used for the treatment of many cancers, including ovarian cancers. Following drug treatment, live cell confocal microscopy was used to monitor the viability of spheroids on both platforms. Representative images for the untreated spheroids and paclitaxel-treated spheroids following 48 hours of drug treatment are shown in Figure 2C. We found no significant difference between the two platforms in terms of decreased cell viability of spheroids after administration of the high dose of paclitaxel, with only 31% ± 2% and 29 ± 0.6% of cells viable on the omniphobic mesh plate and the 384-hanging drop plate respectively. The untreated control maintained 100% cell viability (Figure 2D). An identical drug treatment executed on the 4^th use of an omniphobic aluminum mesh plate also demonstrated equivalent spheroid viability, again demonstrating the reusability of the platform (Figure 2D, S4B). In comparison, an order of magnitude smaller dose of 1 µM paclitaxel left only 13.6 ± 0.3% of cells grown on a 2D plate viable. Similar to spheroids grown on the 384-hanging drop plate, spheroids on the omniphobic mesh plate demonstrated loss of clear boundary, circularity and break-up due to drug treatment (Figure 2C). We also conducted experiments in order to compare three platforms – U-bottom plates, hanging drop plate and omniphobic plate, in the context of drug screening and efficacy assays (Figure 2F). Spheroids generated in any of the three platforms were treated with paclitaxel and doxorubicin, which are clinically relevant drugs for ovarian cancers. We found significant difference in the ovarian cancer cells viability in response to doxorubicin and paclitaxel, when comparing spheroids generated on the omniphobic and hanging drop platforms, to the commercially available U-bottom plate (n=4 biological replicates, with at least 30 technical replicates for each condition per experiment, two way ANOVA). These results concur with our previous characterization of the differences in the response to active pharmaceutical compounds, due to the methods utilized for generating the spheroids⁴. These differences in the three-dimensional microstructure, morphology and cell-cell interactions cause significant changes in the drug dose-response experiments⁴.

One of the most time-consuming aspects of using a 384-hanging drop plate is harvesting the spheroids for downstream processing. On-plate imaging and analysis within a spheroid generating platform would be extremely valuable, as it would enable the direct observation of spheroids in the environment in which they were formed and help maintain the physiological integrity of the cell-cell interactions. However, droplet instability and limited drop volume in each well on the 384-hanging drop plate have thus far prevented on-plate fluorescence imaging. The omniphobic platform developed here supports droplets from below, maximizing stability even at increased droplet volume. As a result, we, for the first time, successfully performed immunofluorescence staining steps of fixation, washing, blocking and antibody incubation on the mesh platform itself, without having to harvest the spheroids onto a different surface (Figure 3).

Figure 3: The omniphobic aluminum mesh spheroid platform, for the first time, can enable the performance of all steps required for the on-plate immunofluorescence staining.

A) Images of OVCAR8 1000 cells/20 μL spheroids while undergoing the different steps involved in immunofluorescence staining on the omniphobic mesh plate. B) Confocal microscopy images of a representative OVCAR8 spheroid, showing nuclear staining by DAPI, F-actin staining by Alexa Fluor 488 phalloidin and RAD51 staining by anti-RAD51 conjugated to Alexa Fluor 546.

With on-plate staining, we observed clear cortical F-actin (phalloidin, labeled with green fluorescence) and Rad51 (localized in the nucleus and cytoplasm, labeled with red fluorescence) in the spheroids along with distinct nuclei (labeled with DAPI). It is clear that the on-plate immunofluorescence staining of the spheroid on the omniphobic platform maintained the spheroid integrity, and also the cell-cell interactions. Low fluorescence of Rad51 indicated lack of DNA damage to the cells, since no DNA damage was induced in the imaged spheroids. Additionally, phalloidin staining on spheroids indicated cortical actin staining, which highlights the cell-cell interactions in 3D. Moreover, dense packing of nuclei and interwoven cortical actin staining indicated three-dimensional morphology, as well as, enhanced cell-cell interactions in the spheroids.

Conclusion:

In summary, we have developed a simple, reusable, and exceptionally robust platform for the high-throughput culture of robust multicellular spheroids, and heterogeneous multi-cell-type spheroids. This novel platform is based on a hierarchically textured omniphobic aluminum mesh, which provides a non-reactive, inert, non-wetting surface that enables the long-term culture of spheroids. The developed platform also dramatically improves the resistance to mechanical shock when compared with the some of the current state of the art technologies. Moreover, it enables on-plate spheroid imaging, drug screening, and treatment, as well as, for the first time, on-plate immunofluorescence staining. These achievements are expected to significantly improve the applicability and adoption of the spheroid-generating platform for anti-cancer drug discovery, screening and pre-clinical studies.

Supplementary Material

Movie S1

Movie S1: Impact test executed on a 384-hanging drop plate.

Movie S2

Movie S2: Impact test executed on an omniphobic spheroid platform.

3

Description of omniphobicity of the aluminum mesh spheroid platform. Table S1: Omniphobicity analysis of the mesh spheroid platform. Table S2: Variation in the spacing ratio (D*) and the robustness factor (A*) for different wire meshes. Figure S1: Omniphobic spheroid platform is mechanically stable. Figure S2: Omniphobic aluminum mesh platform can successfully generate mono- or heterospheroids that are comparable to those created on 384-hanging drop plate. Figure S2: Omniphobic aluminum mesh spheroid platform successfully facilitates multiple re-use of the plate and is an ideal platform for drug studies.