Droplet Array-Based 3D Coculture System for High ... - ACS Publications

Feb 12, 2018 - Lastly, because angiogenesis involves multimolecule and multicell interactions, the model must serve as a high- throughput system that ...
0 downloads 10 Views 1MB Size
Subscriber access provided by MT ROYAL COLLEGE

Article

Droplet array-based 3D coculture system for high-throughput tumor angiogenesis assay Xiaohui Du, Wanming Li, Guan-Sheng Du, Hansang Cho, Min Yu, Qun Fang, Luke P. Lee, and Jin Fang Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b04772 • Publication Date (Web): 12 Feb 2018 Downloaded from http://pubs.acs.org on February 12, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Droplet array-based 3D coculture system for high-throughput tumor angiogenesis assay

Xiaohui Du a,1, Wanming Li a,1, Guansheng Du b1, Hansang Choc, Min Yu a, Qun Fang b**, Luke P. Leec***, Jin Fang a* a

Department of Cell Biology, Key Laboratory of Cell Biology, Ministry of Public Health, and

Key Laboratory of Medical Cell Biology, Ministry of Education, China Medical University, Shenyang, 110122, China b

Institute of Microanalytical System, Department of Chemistry, Zhejiang University, Hangzhou,

310058, China c

Department of Bioengineering, University of California, Berkeley, CA, USA

1

These authors contributed equally to this work.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT Angiogenesis is critical for tumor progression and metastasis, and it progresses through orchestral multicellular interactions. Thus, there is urgent demand for high-throughput tumor angiogenesis assays for concurrent examination of multiple factors. For investigating tumor angiogenesis, we developed a microfluidic droplet array-based cell-coculture system comprising a two-layer polydimethylsiloxane chip featuring 6 × 9 paired-well arrays and an automated droplet-manipulation device. In each droplet-pair unit, tumor cells were cultured in 3D in one droplet by mixing cell suspensions with Matrigel, and in the other droplet, human umbilical vein endothelial cells (HUVECs) were cultured in 2D. Droplets were fused by a newly developed fusion method, and tumor angiogenesis was assayed by coculturing tumor cells and HUVECs in the fused droplet units. The 3D-cultured tumor cells formed aggregates harboring a hypoxic center—as observed in vivo—and secreted more vascular endothelial growth factor (VEGF) and more strongly induced HUVEC tubule formation than did 2D-cultured tumor cells. Our single array supported 54 assays in parallel. The angiogenic potentials of distinct tumor cells and their differential responses to anti-angiogenesis agent, Fingolimod, could be investigated without mutual interference in a single array. Our droplet-based assay is convenient to evaluate multicellular interaction in high-throughput in the context of tumor sprouting angiogenesis and we envision that the assay can be extensively implementable for studying other cell-cell interaction.

Keywords: Tumor angiogenesis; High-throughput assay; Droplet array; Microfluidics; HUVEC; 3D coculture

ACS Paragon Plus Environment

Page 2 of 26

Page 3 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

INTRODUCTION Angiogenesis is one of the predominant biological hallmarks of tumor and is closely related to tumorigenesis, tumor progression, and metastasis.1,2 Therefore, establishing an effective angiogenesis-mimicking system is crucial for elucidating tumor angiogenesis mechanisms, discovering new therapy targets, and developing effective anticancer drugs. Currently, tumor angiogenesis is investigated in most cases by using in vitro models rather

than in vivo models, because the high complexity of the animal host environment makes it extremely challenging to resolve and separately evaluate the effects of distinct factors on the angiogenesis process; this has led to the obtained results being poorly reproducible.3-6 An optimal in vitro model of angiogenesis must meet several requirements: First, the model must be able to recapitulate the in vivo angiogenesis process, mimicking, in particular, cell behaviors in the micro-scale, which includes tissue architecture (e.g., 3D structure), cell-growth microenvironment (e.g., hypoxic status), and cellular interaction. Second, the model must enable visualization of angiogenesis, including real-time observation of the morphogenic changes occurring during angiogenesis, such as cell migration, sprouting, and tubule formation. Lastly, because angiogenesis involves multi-molecule and multi-cell interactions, the model must serve as a high-throughput system that allows concurrent multi-factor and multi-sample analysis. Since endothelial cells (ECs) and tumor cells (TCs) are considered to dominate tumor angiogenesis, the coculture of ECs and TCs has been used as a main approach for analyzing angiogenesis.7-14 In most of these investigations, TC-conditioned medium is supplied to ECs to induce angiogenesis,10-12 a method that is applied for parallel measurement of multiple samples, and for this purpose, several commercial kits are also now available, such as Cultrex® In Vitro Angiogenesis Assay Tube Formation Kit (BioVision, San Francisco, CA, USA), and Endothelial Tube Formation Assay (In Vitro Angiogenesis) (Cell Biolabs, San Diego, CA, USA). However, the genuine in vivo microenvironment during angiogenesis cannot yet be readily replicated because the conditioned medium prepared before the coculture fails to reproduce the real-time interaction between TCs and ECs, and the addition of necessary proteinase inhibitors into the conditioned medium can potentially alter the molecular environment. Another valuable coculture method involves the use of the Boyden chamber, in which TCs are seeded on the upper layer of a permeable polycarbonate membrane and ECs are cultured in the lower chamber.13 Although this

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 26

allows real-time interaction between TCs and ECs, the TCs are typically present in the chamber as a 2D layer rather than in the form of a 3D aggregate as found in vivo, and this results in their biological behavior being considerably different from that of genuine tumor tissues.14 Moreover, the Boyden chamber cannot be readily combined with a confocal microscope-based live-cell-analysis system for dynamic and high-resolution visualization of the angiogenesis process. Furthermore, for all of the aforementioned in vitro models, cells are cultured in routine dishes or multiwell plates featuring dimensions distinct from those of tissues in vivo, and this results in an inability to replicate the in vivo tumor microenvironment in terms of spatial interaction between cells, gradient diffusion of biomolecules, and hypoxic conditions. For tumor angiogenesis research, microfluidic technology has been widely used owing to its superiority in achieving system miniaturization and integration, which enable genuine tissue structures to be mimicked and complex angiogenesis environments to be created through fluid-flow control and multi-geometry design.15-20 In several previous studies, a device featuring three parallel channels was used, in which TCs and ECs were cultured in the two side channels and ECs were induced to migrate toward the middle channel for vascular organization.18,19 Dai et al. used a similar device design to examine and verify EC tubule formation induced by VEGF gradients.20 Nguyen et al. engineered a model to reconstitute the morphogenetic steps of angiogenic sprouting by creating an endothelium in microfluidic channel.21 Furthermore, a few studies have attempted to improve analysis throughput by developing microfluidic-based array systems. Zheng et al. fabricated an integrated chip featuring a 4 × 4 array for coculture by using a pneumatic pump, and this allowed 16 samples to be assayed in parallel.22 Dickinson et al. cocultured ECs and multiple units of breast cancer cells by generating a cancer-cell array.23 However, in these systems, the distinct coculture units shared a common microenvironment, and thus, the systems lacked the ability to provide an interference-free space for concurrent multi-factor analysis. Lastly, previous work has demonstrated that hypoxia might occur within a tumor tissue when the tumor diameter exceeds 1 mm; this would induce the secretion of VEGF, which, in turn, would promote angiogenesis.24 However, no system developed to date can mimic this environment at the micro-scale. Droplet-based

microfluidics

is

a

technology

that

can

create

discrete,

nanoliter-to-picoliter-scale droplets by manipulating two incompatible liquid phases. Systems

ACS Paragon Plus Environment

Page 5 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

designed using this technology can separate and encapsulate reaction solutions or cells into independent micro-droplets and generate functional units of various sizes and composition, and thereby facilitate multi-factor (or multi-sample) analysis without mutual interference.25-27 Accordingly, such systems have been used as micro-reactors for bioactive component analysis, cell culture, drug screening, etc. For angiogenesis analysis, droplet-based microfluidic technology is expected to offer unique advantages. First, a controllable multimode cell culture could be performed in a single assay unit to match the diverse requirements of distinct types of cells based on their in vivo behaviors, such as ECs in 2D culture and TCs in 3D culture. Second, a hypoxic microenvironment favoring angiogenesis could be built by simulating the minute size of a tumor mass in vivo. Third, interactions between different types of cells that are highly similar to interactions in vivo could be mimicked by spatiotemporally controlling droplet formation. However, all of these research aims cannot be readily achieved by currently available methods. Another advantage of droplet-based microfluidic technology is that it has demonstrated superior ability to other technologies in facilitating high-throughput and automated assays.28-31 McMillan et al. created an array of multicellular spheroids and investigated the response of cells to X-ray irradiation and anticancer compounds.32 Li et al. constructed a 3D cell microenvironment array for high-throughput analysis of cell intereaction.33 In our previous work, we encapsulated lung cancer A549 cells into droplets and implemented multi-anticancer drug screening in combination, using a droplet-based automated microfluidic system.34 Recently, we also developed a droplet chain array system for performing multimode cell-migration analysis.35 However, no study thus far has reported angiogenesis analysis by using droplet-based microfluidics. Here, we developed a novel 3D-2D coculture droplet array system to analyze tumor sprouting angiogenesis, in which HUVECs are cultured in a monolayer and TCs are cultured in 3D by loading in Matrigel in order to create a hypoxic tumor aggregate. Moreover, we constructed a 6 × 9 coculture droplet array by using the automated SODA (sequential injection droplet array) system, and this array allows concurrent mimicking of angiogenic sprouting and analysis of multiple samples and multiple factors.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

EXPERIMENTAL SECTION Cell lines and cell culture. Human umbilical vein endothelial cells (HUVECs, ScienCell, San Diego, CA, USA) were cultured in Endothelial Cell Medium (ScienCell) supplemented with 5% fetal bovine serum (FBS) and 100 U/mL penicillin/streptomycin. Only HUVECs at passages 6–8 were used in our experiments. Rat glioma cell line C6 and human colorectal cancer cell lines LoVo and HT29 were cultured in RPMI1640 medium (Gibco, Invitrogen, San Diego, CA, USA) containing 10% FBS and 100 U/mL penicillin/streptomycin. The growth medium for the human embryonic kidney cell line HEK293 was high-glucose DMEM (Gibco, Invitrogen) containing 10% FBS and 100 U/mL penicillin/streptomycin. All cells were cultured at 37 °C in a humidified atmosphere containing 5% CO2. Fabrication of the polydimethylsiloxane (PDMS) chip. A two-layer PDMS chip composed of a substrate layer and a holding frame was used in our experiments to harbor the droplet array; the chip was fabricated using standard soft lithography, according to previous studies.36 Briefly, the PDMS base and curing agent (Sylgard184, Dow Corning, Midland, TX, USA) were mixed at a ratio of 10:1 (w/w) and poured on the substrate mold and the holding mold. The ~0.5-mm-thick substrate was molded on a 6 × 6-cm glass plate bearing a mold of AZ P4620 photoresist (AZ Electronic Materials, London, UK). The holding layer was produced by cutting a 5-mm-thick PDMS plate. The substrate and the holding frame were bonded together irreversibly by exposing them to oxygen plasma for 30 s. The chip was sterilized in an autoclave at 121 °C for 30 min and dried in an oven at 60 °C overnight before use. The substrate layer contained an array of 6 × 9 wells (Figure 1, right). Each well was composed of two equal-sized wells (called paired-wells) designed for coculturing cells. In each unit, one well was for loading TCs (named as TC well) and the adjacent well for loading ECs (named as EC well). All the wells were 1 mm in diameter and 40 µm deep and the distance between the wells was 150 µm. The PDMS holding frame was used for the containment of FC-40 oil (Fluorinent, 3M, USA). Setup of the droplet-manipulation system. The droplet array was manipulated using the SODA system, which was established and used in our previous studies.36 As shown in Figure 1, during the experiments, the prepared PDMS chip was placed on an X-Y-Z translation stage controlled by a home-made LabVIEW program (National Instruments, Austin, USA). A tapered

ACS Paragon Plus Environment

Page 6 of 26

Page 7 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

fused-silica capillary (tip size: ~120 µm i.d., ~150 µm o.d.; Reafine Chromatography Co., Yongnian, China) connected with a syringe pump (PHD2000, Harvard Apparatus, Holliston, USA) through a Teflon tubing (530 µm i.d., 690 µm o.d.) was used to quantitatively load the cell droplets and induce droplet fusion. Before use, the inner and outer surfaces of the tapered capillary were silanized with 1% 1H,1H,2H,2H-perfluorodecyltrichlorosilane (Alfa Aesar, Ward Hill, USA) in isooctane (v/v). Droplet formation was visualized using a charge-coupled device (CCD) camera (G300UMD, Hangzhou Glory Technology, Hangzhou, China) connected to a microscope (SMZ850T, Hangzhou Oplenic Optronics, Hangzhou, China). 3D cell culture in droplets. To mimic the in vivo tumor microenvironment, a droplet-based 3D cell-culture model was established by loading TCs encapsulated in Matrigel (BD Biosciences, USA) into the TC well. The procedure is depicted in Figures 1 and 3A. First, the PDMS chip was filled with 1 mL of FC-40 oil in order to facilitate droplet generation and avoid droplet evaporation, and then fixed on the center of the X-Y-Z stage. Next, 5 µL of a mixture of C6 cell suspension (7.5×106 cells/mL) with Matrigel (2:1, v/v) was aspirated into the capillary, and then the mixture was ejected from the capillary and sequentially deposited into the TC wells on the PDMS chip to generate a tumor-cell droplet array—in which the volume of each droplet was 500 nL—under control of the SODA system. Lastly, the in-droplet cells on the PDMS chip were incubated at 37 °C in a humidified atmosphere containing 5% CO2. Cell images were captured using a phase-contrast microscope (Nikon Eclipse Ti-S, Nikon, Tokyo, Japan) every 3 h over the next two days. Cell viability assay. To evaluate the status of cell growth within droplets, cell viability was analyzed using a live/dead cell assay kit (Life Technology, San Diego, CA, USA). Cells were stained with EthD-1 (Ex/Em: 528/617 nm) and Calcein-AM (Ex/Em: 494/517 nm) as per the manufacturer’s instructions. Continuous serial-section images with a 6-µm interval were acquired using a laser-scanning confocal microscope system (BX61W1-FV1000, Olympus, Tokyo, Japan) in order to visualize the status of cells at distinct layers of the droplets. Droplet-based cell coculture for analyzing tumor angiogenesis. In this study, TCs and HUVECs were cocultured in a fused droplet unit for assaying sprout and tubule structure formation during tumor angiogenesis. To mimic the in vivo microenvironment during tumor angiogenesis, TCs were 3D-cultured by mixing with Matrigel and HUVECs were cultured as a

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

monolayer. Both TCs and HUVECs were cultured in serum-free medium. The process used for assembling the coculture droplet is shown in Figure 3A. First, C6 cells were loaded into the TC well (as described in Section 2.4) and cultured for 24 h to generate 3D spheroids. Next, we transferred 500 nL of HUVECs (2×106 cells/mL) into the EC well. After incubating for 10 min to allow cell adhesion onto the PDMS surface, a coculture unit was formed by fusing the two cell droplets through a reciprocating, horizontal movement of the capillary between the two droplets (Video 1). Subsequently, the chip was placed in an incubator for cell coculture and the cells were visualized under a microscope. For real-time imaging of tubule formation, the chip was placed in the culture chamber of a live-cell imaging system (PerkinElmer, CA, USA) and images were acquired using a microscope (Olympus IX51). Angiogenic tubule formation was quantified from at least three independent experiments by calculating the total tubule area of complete vessels by using ImageJ software, and is expressed here in square milimetre. To verify the characteristics of the 3D tumor-cell culture model, a 2D coculture droplet was formed in parallel by fusing C6 cell droplets and HUVEC droplets in their respective media without Matrigel. After incubation for 2.5 h, images of both 2D and 3D coculture droplets were acquired for comparison. Intra- and inter- droplet solute diffusion assay. To investigate the diffusion behavior of angiogenesis-related soluble factors within the coculture array, FITC-labeled dextran (MW: 40 kDa; Sigma, St. Louis, MO, USA), which is similar in size to human VEGF protein (hVEGF), was used. For the intra-droplet (within a coculture unit) assay, 500 nL each of the FITC-dextran solution (50 µg/mL) and FITC-dextran-free culture medium were loaded into the TC well and the EC well, respectively. After droplet fusion (as mentioned in the preceding section), time-lapse imaging was performed using a fluorescence microscope. For the inter-droplet assay, a 3 × 3 single-droplet array was generated. Briefly, one FITC-dextran droplet was loaded into the central well, and droplets of FITC-dextran-free medium were loaded into the surrounding wells. Images were acquired using the fluorescence microscope. Multi-sample analysis using the coculture droplet array. To investigate whether the coculture droplet array can be used for multi-sample analysis in parallel, we assayed for the uniformity of cell numbers in different droplets, using a chip containing a 6 × 6 array. The droplet array was formed by continuously seeding C6 cell suspensions at densities of 4 × 105 or 2 ×

ACS Paragon Plus Environment

Page 8 of 26

Page 9 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

105 cells/mL (300 nL/droplet) into the wells by using the SODA system. The cells in each droplet were then counted and analyzed. In another set of assays, various tumor cell lines were separately cocultured with HUVECs in a single 5 × 9 array chip for multi-sample analysis. Briefly, 500-nL droplets of Matrigel-coated cells (C6, LoVo, HT29, and HEK293; 5 × 106 cells/mL) were loaded into distinct TC wells, and after 24-h incubation, droplets containing HUVECs were loaded into the paired EC wells; subsequently, the coculture array was created through the fusion of TC and EC droplets as described above. The angiogenesis-inducing capacity of different cell lines was analyzed using the individual 1× 9 cocultured units prepared with each cell line. Furthermore, the 5 × 9 array chip was applied to evaluate the anti-angiogenesis effects of Fingolimod (FTY720, Selleckchem, Milan, Italy) on different tumor cells at various treatment concentrations. Briefly, the coculture units containing tumor cells (LoVo, MCF7 and HT29; 5 × 106 cells/mL) and HUVECs were generated as described above. Besides, increasing concentrations of Fingolimod (0, 0.05, 0.1, 0.2, 0.4 µM) were added into the EC wells with HUVECs together and subjected to 2.5 h incubation. In both cases, the tubule structures formed by HUVECs in the droplets were imaged using a phase-contrast microscope (Olympus). To quantify the ability of the cells to induce HUVEC tubule formation and the effects of Fingolimod on anti-angiogenesis, the total tubule area of complete vessels in each obtained photograph was calculated using Image J software. ELISA for hVEGF. To compare VEGF secretion by TCs in 3D versus 2D droplets, a coculture droplet array composed of HUVECs and LoVo cells was generated (as described in the preceding subsection), in which the LoVo cells were loaded in 2D or 3D, and HUVECs were in a monolayer. After coculture for 2.5 h, supernatants from 20 droplets were collected and diluted 50-fold with FBS-free medium in a 96-well plate, and then hVEGF was quantified by using an ELISA kit specific for hVEGF (R&D Systems, USA) according to the manufacturer’s protocol. To measure the hVEGF secreted by different types of cells, LoVo, HT29, and HEK293 cells were individually cultured in Petri dishes at a density of 1×105 cells/mL. After incubation for 24 h to allow cell adhesion, the medium was removed and replaced with serum-free medium. After 36 h, supernatants were harvested and diluted 10-fold for the ELISA assay. Meanwhile, an ELISA kit specific for rat VEGF (Boster Biological Technology, China) was used to quantify the VEGF secreted by C6 cells using the same procedure.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The 450-nm absorbance of the reaction mixture was measured using an ELISA reader (LT4000 microplate reader, Labtech, Cambridge, UK). The ELISA was independently replicated three times for each sample. Immunofluorescence Staining. To evaluate the molecular changes of HUVECs along with tubule formation when cocultured with different types of cells, the expression levels of endothelial cell markers, CD31 and CD105, were detected by immunofluorescence. First, a coculture array with the TC wells containing LoVo, HT29, and HEK293 cells was generated as described above. After induced for tube formation by different types of cells, the HUVECs in the EC wells were fixed with 3.7% (wt/wt) formaldehyde and then permeabilized with 0.1% Triton-X. After blocked with 3% (wt/wt) BSA, they were incubated with primary antibodies: anti-CD31 (1:100; proteintech, China) and anti-CD105 (1:50; Santa Cruz Biotechnology), respectively, at 4 °C overnight, and subsequently stained by appropriate FITC-labeled secondary antibodies (1:500, CST, USA) for 1 h at room temperature. Finally, the cells were counterstained with Rhodamine phalloidin (1:100; Sigma Aldrich, USA) for 20 min and DAPI (1:500; Sigma Aldrich, USA) for 5 min. The images were taken with a confocal microscope (Leica TCS SP8, Germany). Data analysis. Statistical analysis was performed using Origin 7.5 (OriginLab Data Analysis and Graphing Software). All data are presented as the means ± SD of at least three independent experiments. Comparisons were performed by Student’s t tests, and P < 0.05 was considered statistically significant.

ACS Paragon Plus Environment

Page 10 of 26

Page 11 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

RESULTS AND DISCUSSION Droplet-based 3D cell culture. Within a solid tumor tissue in vivo, an extracellular matrix surrounds the TCs and thereby generates a 3D microenvironment, which plays a crucial role during cancer progression and angiogenesis.37 Moreover, several studies have shown that as compared with TCs cultured in 2D, TCs that are cultured under 3D conditions display differential phenotypes and functions that are more similar to their in vivo status.38,39 Therefore, 3D tumor models have garnered increasing attention recently. Laura et al.40 developed a 3D bioengineered microenvironment model of tumor angiogenesis and found that it more closely mimicked the in vivo scenario than 2D cultures did, and the model helped provide an increased understanding of vascular signals in cancer progression. To analyze tumor angiogenesis in vitro effectively, we established a droplet-based 3D cell-culture platform, in which the size-controlled Matrigel-coated cell droplets were generated using the automated SODA system. Our results showed that the in-droplet rat glioma C6 cells grew normally within the culture period (Figure 2). Moreover, we observed that the seeded cells distributed uniformly at first, then aggregated gradually toward the center, and eventually organized into a spheroid-like aggregate structure that was similar to a genuine tumor in vivo.41 Serial-section images (Figure S1) revealed the presence of cells in all sections, which clearly indicated that the cells formed a 3D tissue-like structure. Solid tumors have been shown to harbor a hypoxic center when their size reaches ~1 mm in diameter; the low oxygen tension induces VEGF secretion and activates the “angiogenic switch” for new vessel growth.42,43 However, few studies to date have focused on creating tumor-like 3D aggregates that are similar in size to the minute tumors found in vivo. Furthermore, for mimicking the in vivo hypoxic environment, the strategies commonly employed thus far have involved using a low-oxygen incubator or adding certain drugs, such as cobalt chloride (CoCl2). Therefore, no previous study has reproduced the genuine in vivo tumor status. By exploiting the advantages of microfluidic devices, in this study, we generated controllable droplets featuring a diameter of ~1 mm; this is highly similar in size to in vivo tumors exhibiting a hypoxic character. To evaluate the hypoxic environment within the formed tumor aggregates, the dead/live assay was performed after culturing for 48 h, which revealed that the number of cells stained red (dead cells) in the aggregate center was higher than that at the edge (Figure 2). Several studies44-46 have demonstrated that conditions of inadequate O2 supply, such as hypoxia (pO2 < 7 mmHg), might induce cell

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

senescence or apoptosis due to the increased generation of reactive oxygen species by glycolysis, and this, in turn, would stimulate angiogenesis through the secretion of various growth factors. All of these findings together indicated that the 3D cell droplets generated in our system might reproduce the hypoxic state of a tumor tissue in vivo. Analysis of tumor angiogenesis in the coculture droplet. Interactions between TCs and ECs play a key role during tumor angiogenesis. Thus, tumor angiogenesis has been extensively investigated using coculture of TCs and ECs.11,18,47 Considering that TCs in vivo are present in 3D aggregates and ECs assemble into monolayer, Chrobak et al.48 spread human dermal microvascular EC monolayers on gel-loaded TCs in order to explore the TC-EC interaction; however, because of the relatively large geometric dimensions involved, this model did not closely mimic the genuine in vivo hypoxic environment. Furthermore, direct contact between ECs and TCs was also reported to potentially result in EC apoptosis.49,50 We addressed this issue related to TC-EC interaction by exploiting the SODA system for generating controllable droplets and thus constructed a droplet array-based coculture unit—by inducing fusion of individual TC droplets in 3D and EC droplets in monolayer without direct intercell contact—to mimic angiogenesis conditions (Figure 3A). As shown in Figure 3B and Video 2, after coculture with rat glioma C6 cells for 40 min, HUVECs started sprouting and then formed tubules, and the area of these vessels gradually increased during 2.5 h of observation; this indicated functional interaction between the two types of cells. Moreover, the coculture of HUVECs with increasing densities of C6 cells led to a corresponding increase in tubule area, except at the density of 1×107 cells/mL (Figure S2). This result further demonstrated that the observed tubule formation depended on the induction by the TCs. To verify that TCs cultured in 3D exhibited an enhanced ability to induce angiogenesis, we compared them with C6 cells cultured in 2D (Figure 3C and D). Quantification of the results (Figure 3D) revealed that the area of the tubules induced by C6 cells in 3D was significantly larger than that induced by the cells in 2D, which suggested that the angiogenesis-inducing potential of TCs cultured in 3D was higher than that of TCs cultured in 2D. Furthermore, when we used colon cancer LoVo cells in these cocultures, the results obtained were the same as those with C6 cells (data not shown). Lastly, because VEGF is considered a key factor that induces tumor angiogenesis under hypoxia,24 we measured the VEGF levels in the supernatants of both 3D and 2D droplet coculture units. Our results showed that the hVEGF level of the 3D model was

ACS Paragon Plus Environment

Page 12 of 26

Page 13 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

markedly higher than that of the 2D model (15.41 ± 0.04 vs 8.42 ± 0.03 ng/mL). Thus, VEGF secretion in the 3D model was increased considerably, which was probably due to the hypoxia inside the tumor aggregates; this finding indicated that the droplet array-based coculture system could potentially simulate the in vivo angiogenesis microenvironment. In addition to the aforementioned advantages of our model, the small size of the droplet chip makes it highly suitable for incorporation into live-cell imaging systems for real-time monitoring of the angiogenesis process. Video 2 shows that the key steps of HUVEC angiogenesis process, from sprouting to tubule formation, could be visualized when the cells were cocultured with TCs, and even single cells in specific regions of interest could be tracked. Furthermore, if combined with molecular fluorescent probes, the system could potentially be used for investigating the molecular changes that occur during angiogenesis, which has not been achieved thus far. High-throughput angiogenesis assay by establishing a coculture droplet array system. Previously, the Fang group34 developed the SODA system for multiplex drug screening. We made improvements based on this system in order to automatically achieve the cell-coculture operation by using the droplet assembly and generation technique and, in particular, to newly develop the droplet-fusion technique by using the relative movement of the capillary and droplet array chip. Video 1 and Figures 1 and 3A display the process of droplet array formation and assembly. The droplets could be accurately loaded in different wells, and then fused into independent units without inter-droplet or inter-unit disturbance. To investigate the suitability of the droplet array system for the analysis of multiple cell samples in parallel, 36 cell droplets were continuously generated in a 6 × 6 array chip, and the cells in each droplet were counted for uniformity evaluation. At the two cell densities tested, the average cell numbers in droplets were 115 with a 7.5% coefficient of variation, and 60 with 8.9% of coefficient of variation (Figure 5A); the result indicated that the system could create cell droplets containing a uniform number of cells. To demonstrate that each paired-droplet unit in the array was independent when used for the coculture, we performed intra- and inter-droplet diffusion assays using FITC-labeled dextran (40 kDa), which is similar in size to hVEGF. The intra-droplet assay revealed that within a fusion droplet unit, the fluorescent FITC-dextran diffused from one side to the other within 10 min and then achieved equilibrium after 1.5 h; this indicated that the paired droplets were capable of sharing common molecules in cell-interaction assays (Figure 4A). Moreover, in the inter-droplet

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

assay, performed using a 3 × 3 single-droplet array, we did not detect any interference between droplets: the FITC-dextran within the central droplet did not diffuse into any adjacent unlabeled droplets during a 5-day period (Figure 4B); this result indicated that all droplets in the array were closed units featuring a separate internal environment, which would be suitable for concurrent multi-sample analysis. We used the array system and investigated the angiogenesis-induction effect of different types of cells on HUVECs; a 5×9 coculture droplet array on a single chip was used (Figure 5B), and droplets of Matrigel-coated cells (C6, LoVo, HT29, and HEK293) were patterned so as to be cocultured with HUVECs loaded in different rows. The distinct features of the tubules induced by the different cells are shown in Figure 5C, and the results are quantified in Figure 5D. The areas of the tubules induced by TCs were significantly larger than the areas of tubules induced by normal cells (HEK293). The inducibility order was C6 > LoVo > HT29 > HEK293 cells. Highly metastatic TCs (C6 and LoVo) exhibited stronger tubule induction than did weakly metastatic TCs (HT29).51,52 The hVEGF level in the supernatants of the cells (LoVo, HT29 and HEK293) was quantified by ELISA, which revealed that cells with higher metastatic potentials also secreted higher levels of hVEGF, which agreed with the detected tubule-formation abilities (Figure 5D). In addition, a high level of VEGF (4.20 ± 0.13 ng/mL) was also detected in the supernatants of rat glioma cells C6 using a rat VEGF-specific ELISA assay. To evaluate the molecular changes of HUVECs when tubule formation occurred, the expression levels of endothelia cell markers (CD31 and CD105) in the HUVECs cocultured with distinct types of cells were analyzed by immunofluorescence staining. In general, CD31 is present in all endothelial cells,53 while CD105 is found to preferentially express in activated endothelial cells and has been suggested as a appropriate marker for tumor-related angiogenesis.54 In our system, CD31 did not display detectable difference in different coculture groups, while CD105 had a significant higher expression in the HUVECs under coculture with colon cancer cell lines LoVo and HT29 than that with normal cell line HEK293 and control (Figure 5F), indicating that these cancer cells could effectively activate interacted endothelial cells. The data also demonstrated our system’s capability to investigate the molecular changes of cocultured cells along with their morphogenesis. Based on the above results, we further applied the array system to test the effect of multi-factors on angiogenesis. An anti-angiogenesis agent, Fingolimod, was added into coculture units loaded with

ACS Paragon Plus Environment

Page 14 of 26

Page 15 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

distinct types of TCs and HUVECs in various treatment concentrations. As shown in Figure S4, Fingolimod differentially decreased tube formation ability of HUVECs in the context of different tumor cells (LoVo, MCF7 and HT29). Relevant reports have demonstrated that Fingolimod inhibited tumor angiogenesis via blocking sphingosine-1-phosphate receptor expressed on ECs which could be regulated by tumor-derived VEGF.21,55 Here, using the droplet-based coculture system, we observed the inhibition effect of Fingolimod on HUVEC tubule formation in the present of tumor cells. Further, we found that Finglimod did not exhibit a standard concentration-response curve but a bell-shaped curve, partly losing its activity at higher concentration of 0.2 µM. We envisioned that Fingolimod might form colloidal aggregates at relevant concentrations,56 enriching the ability of this system to investigate multi-factor effects. All these results suggested that the automatic array system effectively analyzed angiogenesis in vitro and enabled concurrent multi-criterion analysis.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

CONCLUSION In this study, a droplet array-based cell-coculture system was developed and successfully used to mimic the interaction between TCs and ECs during capillary-like structure formation. Using the SODA system, multistep liquid manipulations were automatically performed, including cell-suspension aspiration, droplet formation, and droplet fusion. To our knowledge, this is the first time a droplet array-based coculture system has been developed for high-throughput tumor angiogenesis assays. Moreover, the highly controllable features of this system enable various droplet parameters to be set on demand, including droplet size, droplet dimension, and distance between droplets. For example, ECs were presented here in 2D (as monolayers), whereas TCs were loaded as 3D aggregates that were as small as 1 mm in diameter, which provided a hypoxic microenvironment similar to that in genuine tumor tissues in vivo. Notably, when used in combination with live-cell imaging, our system allows real-time visualization of cell behaviors, and this could be extended to investigate dynamic changes in molecules by adding appropriate molecular probes. Furthermore, by fusing additional types of droplets containing different cells, multiple cell-cell interactions could be explored.

ASSICIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: Figure S1: Serial-section images of a 3D cell droplet; Figure S2: Analysis of HUVEC tubule formation induced by various densities of TCs; Figure S3: Analysis of the effect of Fingolimod on anti-angiogenesis in the present of different types of cancer cells. Video1: Generation of a coculture droplet array; Video2: Tubule formation of HUVECs in the coculture droplet.

ACKNOWLEDGMENTS Authors thank Jiabin Li for assistance with video production. This work was supported by National Natural Science Foundation of China (Grant number 21375149, 81672920, and 21435004).

AUTHOR INFORMATION Corresponding Author * Address: Department of Cell Biology, China Medical University, Shenyang, 110122, China. Tel. +86-24-31939077. E-mail: [email protected] (J. Fang)

ACS Paragon Plus Environment

Page 16 of 26

Page 17 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

** Address: Institute of Microanalytical System, Department of Chemistry, Zhejiang University, Hangzhou, 310058, China. Tel. +86-571-88206771. Fax. +86-571-88273572. E-mail: fangqun@ zju.edu.cn (Q. Fang)

*** Address: Department of Bioengineering, University of California, Berkeley, CA, USA. E-mail: [email protected] (LP, Lee) ORCID

J. Fang: 0000-0001-5188-5206 Notes The authors declare no competing financial interest.

REFERENCE (1) Folkmen, J. J.Natl.Cancer Inst. 1990, 82, 4-6. (2) Lambert, A.W.; Pattabiraman, D.R.; Weinberg, R.A. Cell 2017,168, 670-691. (3) Auerbach, R.; Shinners, B.; Kubai, L.; Akhtar, N. Clin Chem.2003, 49, 32-40. (4) Folkman, J.; Haudenschild, C. Nature 1980, 288, 551-556. (5) Morgan, J.P.; Delnero, P.F.; Zheng, Y.; Verbridge, S.S.; Chen, J.; Craven, M. Nat Protoc. 2013, 8, 1820-1836. (6) Luginbühl, J.; Montaño, I.; Schiestl, C.; Schneider, J.; Pontiggia, L.; Biedermann, T.; Sophie B.H.; Braziulis, E.; Meuli, M.; Reichmann, E. Tissue Eng. Part A 2010, 16, 269–282. (7) Francescone, R.; Ngernyuang, N.; Yan, W.; Bentley, B.; Shao, R. Oncogene 2014, 33, 2110-2122. (8) Zhang, S.; Chitu, V.; Stanley, E.R.; Elliott, B.E.; Greer, P.A. Cancer Res. 2011, 15, 1465-1473. (9) Chiew, G.G.; Wei, N.; Sultania, S.; Lim, S.; Luo, K.Q. Biotechnol Bioeng. 2017 [Epub ahead of print] (10) Lo, P.H.; Lung, H.L.; Cheung, A.K.; Apte, S.S.; Chan, K.W.; Kwong F.M.; Yee Ko, J.M.; Cheng, Y.; Law, S.; Srivastava, G.; Zabarovsky, E.R.; Tsao, S.W.; Tang, J.C.; Stanbridge, E.J.; Lung M.L. Cancer Res. 2010, 70, 5567-5576. (11) Jang, H.K.; Kim, B.S.; Han, J.; Yoon,J.K.; Lee,J.R.; Jeong,G.J.; Shin,J.Y. Biotechnol Prog. 2016,32, 456-464. (12) Li, Y.; Sun, W.; Liu, H.; Qi, G.Y.; Wang, Q.; Sun, X.R.; Chen, B.Q.; Liu, J.R. J Nutr Biochem. 2011, 22, 1127-1136.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(13) Li, Y.; Zhu, C. Clin. Exp. Metastasis 1999, 17, 423-429. (14) Wong, K.H.; Chan, J.M.; Kamm, R.D.; Tien, J. Annu Rev Biomed Eng. 2012, 14, 205-230. (15) Hasan, A.; Paul, A.; Vrana, N.E.; Zhao, X.; Memic, A.; Hwang, Y.S.; Dokmeci, M.R.; Khademhosseini, A. Biomaterials 2014, 35, 7308-7325. (16) Song, J.W.; Munn, L.L. Proc Natl Acad Sci USA. 2011, 108, 15342-15347. (17) Bischel, L.L.; Young, E.W.; Mader, B.R.; Beebe, D.J. Biomaterials 2013, 34, 1471-1477. (18) Chung, S.; Sudo, R.; Mack, P.J.; Wan, C.R.; Vickerman, V.; Kamm, R.D. Lab Chip 2009, 9, 269-275. (19) Sudo, R.; Chung, S.; Zervantonakis, I.K.; Vickerman, V.; Toshimitsu, Y.; Griffith, L.G.; Kamm, R.D. FASEB J. 2009, 23, 2155–2164. (20) Dai, X.; Cai, S.; Ye, Q.; Jiang, J.H.; Yan, X.Q.; Xiong, X.; Jiang, Q.F.; Wang, A.Chih.; Tan, Y. Chin Sci Bull. 2011, 56, 3301-3309. (21) Nguyena, D.T.; Stapletona, S.C.; Yangb, M.T.; Chab, S.S.; Choib, C.K.; Galie, P.A.; Chen. C.S. Proc Natl Acad Sci USA. 2013, 110, 6712-6717. (22) Zheng, C.; Zhao, L.; Chen, G.; Zhou, Y.; Pang, Y.; Huang, Y. Anal. Chem. 2012, 84, 2088-2093. (23) Dickinson, L.E.; Lutgebaucks, C.; Lewis, D.M.; Gerecht, S. Lab Chip. 2012, 12, 4244-4248. (24) Shweiki, D.; Itin, A.; Soffer, D.; Keshet, E. Nature 1992, 359, 843-845. (25) Li, L.; Wang, Q.; Feng, J.; Tong, L.; Tang B. Anal. Chem. 2014, 86, 5101-5107. (26) Liang, Y.R.; Zhu, L.N.; Gao, J.; Zhao, H.X.; Zhu, Y.; Ye, S.; Fang, Q. ACS Appl Mater Interfaces 2017, 9, 11837-11845 (27) Niu, G.; Ruditskiy, A.; Vara, M.; Xia, Y. Chem Soc Rev. 2015, 44, 5806-5820. (28) Tumarkin, E.; Tzadu L.; Csaszar, E.;. Seo, M.; Zhang, H.; Lee, A.; Peerani, R.; Purpura, K.; Zandstra, P.W.; Kumacheva, E. Integr Biol. 2011, 3, 653–662 . (29) Brouzes, E.; Medkova, M.; Savenelli, N.; Marran, D.; Twardowski, M.; Hutchison, J.B.; Jonathan, M.; Rothberg, D.R.; Link, N.P. Proc Natl Acad Sci USA. 2009, 106, 14195-14200. (30) Gach, P.C.; Shih, S.C.; Sustarich, J.; Keasling, J.D.; Hillson, N.J.; Adams, P.D.; Singh, A.K. ACS Synth Biol. 2016, 5, 426-433. (31) Leung, K.; Zahn, H.; Leaver, T.; Konwar, K.M.; Hanson, N.W.; Pagé, A.P.; Lo, C.C.; Chain, P.S.; Hallam, S.J.; Hansen, C.L. Proc Natl Acad Sci USA. 2012, 109, 7665-7670.

ACS Paragon Plus Environment

Page 18 of 26

Page 19 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

(32) McMillan, K.S.; Boyd, M.; Zagnoni, M. Lab Chip 2016, 16, 3548-3557. (33) Li Y.W.; Chen P.; Wang Y.C.; Yan S.Q.; Feng X.J.; Du W.; Koehler S.A.; Demirci U.; Liu B.F. Adv. Mater. 2016, 28, 3543-3548. (34) Zhu, Y.; Zhang, Y.X.; Cai, L.F.; Fang, Q. Anal. Chem. 2013, 85, 6723-6731. (35) Ma, Y.; Pan, J.Z.; Zhao, S.P.; Lou, Q.; Zhu, Y.; Fang, Q. Lab Chip. 2016, 16, 4658-4665. (36) Du, G.S.; Pan, J.Z.; Zhao, S.P.; Zhu, Y.; den Toonder, J.M.; Fang, Q. Anal. Chem. 2013, 85, 6740-6747. (37) Dvorak, H.F.; Nagy, J.A.; Dvorak, A.M. Cancer Cells 1991, 3, 77-85. (38) Yamada, K.M.; Cukierman, E. Cell 2007, 130, 601-610. (39) Hanahan, D.; Weinberg, R.A. Cell 2011, 144, 646-674. (40) Laura, J.B.; Binner, M.; Holzheu, A.; Friedrichs, J.; Freudenberg, U.; Hutmacher, D.W.; Werner, C. Biomaterials 2015, 53, 609-620. (41) Mehta, G.; Hsiao, A.Y.; Ingram, M.; Luker, G.D.; Takayama, S. J Control Release. 2012, 164, 192–204. (42) Vaupel, P. The Oncologist 2004, 9, 10-17. (43) Malda, J.; Klein, T.J.; Upton, Z. Tissue Eng. 2007, 13, 2153-2162. (44) Vaupel, P. The Oncologist 2008, 13, 21-26. (45) Hanahan, D. Lancet 2014, 383, 558-563. (46) Welford, S.M.; Giaccia, A.J. Mol Cancer Res. 2011, 9, 538–544. (47) Chiew, G.G.; Fu, A.; Low, K.P.; Luo, K.Q. Sci Rep. 2015, 5, 10801-10815. (48) Chrobak, K.M.; Potter, D.R.; Tien, J. Microvascular Research 2006, 71, 185-196. (49) Sanpaio, P.C.; Auslaender, D.; Krubasik, D. PLoS One. 2012, 7, e30753. (50) Kebers, F.; Lewalle, J.M.; Desreux, J.; Munaut, C.; Devy, L.; Foidart, J.M. Noël,A. Exp Cell Res. 1998, 240,197-205. (51) Li, W.; Bing, T.; Wei, J.; Chen, Z.Z.; Shangguan, D.H.; Fang, J. Biomaterials 2014, 35, 6998-7007. (52) Yao, N.; Chang, C.; Lin, H.; Yen, C.T.; Chen, J.Y. Sci Rep. 2016, 6, 26050-26061. (53) Saraswati S.; Kumar S.; Alhaider A.A. Mol Cancer. 2013, 22, 147-164. (54) Nassiri F.; Cusimano M.D.; Scheithauer B.W.; Rotondo F.; Fazio A.; Yousef G.M.; Syro L.V.; Kovacs K.; Lloyd R.V. Anticancer Res. 2011, 31, 2283-2290.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(55) Igarashi, J.; Erwin, P. A.; Dantas, A. P. V.; Chen, H. J.; Michel, T. Proc Natl Acad Sci USA. 2003, 100, 10664-10669. (56) Owen, S.C.; Doak, A.K.; Ganesh, A.N.; Nedyalkova, L.; McLaughlin, C.K.; Shoichet, B.K.; Shoichet M.S. ACS Chem. Biol. 2014, 9, 777-784.

ACS Paragon Plus Environment

Page 20 of 26

Page 21 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 1. Photograph of the SODA system for automated droplet manipulation. The system consists of an XY-Z translation stage and a tapered fused-silica capillary connected with a syringe pump (left). The twolayer PDMS chip is fixed on the stage, and an array of 6 × 9 droplet pairs is generated using the system. The magnified image (right) shows the formed droplet array for cell coculture, with color dyes used as a model sample instead of cell suspensions. 170x127mm (300 x 300 DPI)

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Cell culture in 3D droplets. A 3D cell droplet (500 nL) was generated by loading C6 cells mixed with Matrigel into the TC well, using the SODA system, and images were acquired at different time points. After incubation for 48 h, cell viability was analyzed by means of dead/live staining; the dead cells were visualized in red under a fluorescence microscope. 83x62mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 22 of 26

Page 23 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 3. Analyzing HUVEC tubule formation using the droplet array-based coculture system. (A) Formation process of a coculture droplet. (B) Real-time imaging of HUVEC tubule formation induced by TCs (C6 cells) in the coculture droplet. The red arrows indicate the locations of tubule structures. (C) Assay for HUVEC angiogenesis induced by TCs in 2D and 3D droplets. The images were captured using a phase-contrast microscope. (D) Quantification of angiogenesis by calculating the area of vessels formed by HUVECs. ***P < 0.001 compared with 2D. 106x66mm (300 x 300 DPI)

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Intra- and inter-droplet solute-diffusion assays. (A) Intra-droplet diffusion assay. A suspension of FITC-dextran was loaded into one side, and culture medium was loaded into the other side. After fusion of the two droplets, images were acquired at different time points over 3 h using a fluorescence microscope. (B) Inter-droplet diffusion assay. The FITC-dextran suspension was loaded into the central well and PBS was loaded into the surrounding wells. Solute diffusion was evaluated by examining the diffusion of the fluorescence signal from the center droplet toward the other droplets over 5 days. 145x153mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 24 of 26

Page 25 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 5. Droplet array-based coculture system for analysis of HUVEC tubule formation induced by different cancer cells. (A) Uniformity analysis of the number of cells loaded in droplets. First, two 6×6 droplets arrays were continuously generated using the SODA system, and then the cells in each droplet were counted. (B) Analysis of multiple samples with the coculture array. A 5×9 coculture-droplet array was created by fusing different types of cells in individual TC wells and HUVECs in EC wells using the SODA system. The TC wells in the 9 coculture-droplet units on each column were loaded with the same type of cells for this parallel assay. The images were captured under a phase-contrast microscope. (C) Magnified images of the frame in B. (D) Quantification of tube formation by calculating the area of vessels in each EC well by using ImageJ software. (E) Quantification of hVEGF levels in the supernatants of cells cultured in Petri dishes by using ELISA. *P < 0.05, ***P < 0.001 compared with Matrigel. (F) Analysis of endothelial marker expressions in HUVECs cocultured with different types of cells (LoVo, HT29 and HEK293) using immunofluorescence assay. After incubation of the fused droplets for 2.5 h, the HUVECs were stained by FITC-labeled antibodies against (CD31 and CD105), respectively, followed by phalloidin staining F-actin (red) and DAPI staining nuclei (blue). The images were captured under a confocal microscope. (i) CD31 expression imaging, (ii) CD105 expression imaging. 138x112mm (300 x 300 DPI)

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

47x26mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 26 of 26