Single-Cell-Derived Tumor-Sphere Formation and Drug-Resistance

May 31, 2019 - Herein, we present a platform of integrated microfluidics for the analysis of single-cell-derived tumor-sphere formation and drug resis...
0 downloads 0 Views 5MB Size
Download PDF
Article Cite This: Anal. Chem. 2019, 91, 8318−8325

pubs.acs.org/ac

Single-Cell-Derived Tumor-Sphere Formation and Drug-Resistance Assay Using an Integrated Microfluidics Long Pang,†,‡ Jing Ding,‡ Yuxin Ge,† Jianglin Fan,*,† and Shih-Kang Fan*,§ †

School of Basic Medical Science, The Shaanxi Key Laboratory of Brain Disorders, Xi’an Medical University, Xi’an, 710021, China Key Laboratory of Thermo-Fluid Science and Engineering of MOE, School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, 710049, China § Department of Mechanical Engineering, National Taiwan University, Taipei 10617, Taiwan Downloaded via KEAN UNIV on July 17, 2019 at 15:39:26 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: Considerable evidence points to cancer stem-like cells (CSCs) as responsible for promoting progression, metastasis, and drug resistance. Without damage to the cell biological properties, single-cell-derived tumor-sphere is encouraging options for CSCs identification and studies. Although several single cellbased microfluidic methods have been developed for CSCs studies, clarifying liaison between the biomechanics of cells (such as size and deformability) and stem (such as tumor-sphere formation and drug resistance) remains challenging. Herein, we present a platform of integrated microfluidics for the analysis of single-cell-derived tumorsphere formation and drug resistance. Tumor-spheres derived from different biomechanics (size and/or deformation) single-cells could be formed efficiently using this device. To demonstrate the microfluidic-platform capability, a proof-of-concept experiment was implemented by evaluating single-cell-derived sphere formation of single glioblastoma cells with different biomechanics. Additionally, a course of chemotherapy to study these singlecell-derived spheres was determined by coculture with vincristine. The results indicate that tumor cell biomechanics is associated with single-cell-derived spheres formation; that is, smaller and/or more deformable tumor cells are more stem-like defined by the formation of single-cell-derived spheres than more prominent and/or lesser deformable tumor cells. Also, tumorspheres derived from single small and/or more deformable tumor cell have higher drug resistance than more prominent and/or less deformable tumor cells. Our device offers a new approach for single-cell-derived sphere formation according to tumor cell different biomechanical properties. Furthermore, it offers a new method for CSC identification and downstream analysis on a single-cell level.

T

Currently, single-cell culture and analysis techniques primarily include limiting the dilution method and fluorescence-activated cell sorting (FACS). Dilution method limitation is labor intensive and limited in throughput.13 FACS can achieve higher single-cell seeding rates with mechanical manipulation but decreased cell viability due to its high shear stress.11 Hence, exploring a new approach which could avoid the above shortcomings remains a challenge. Recently, microfluidics-based culture systems have emerged as powerful methods for single-cell studies. A comparison between these rising techniques and conventional single-cell analysis approaches shows that microfluidics-based methods could reduce complicated manipulations, handle smaller sample volumes, and manage the integration of downstream analyses.14−17 Microfluidics-based culture systems essentially include hydrodynamic techniques, droplet systems, the

umor cell heterogeneity is pervasive in tumor therapy and research. Tiny amounts of “cancer stem-like/initiating cells” (CSCs) can reportedly account for promoting progression, metastasis, and drug resistance.1−3 Conventionally, CSCs are identified by using membrane surface markers and intracellular enzymatic markers.4,5 However, challenges remain in identifying CSC populations, as CSCs can carry distinct expressions from different tumors.6 Single-cell analysis techniques provide a new opportunity to identify CSC populations by using their behavior; that is, a single CSC can proliferate into a tumor-sphere in a suspension environment, whereas single non-CSCs, because they have lost the ability to adhere to substrates, cause programmed cell death (anoikis).7 3D-culture tumor sphere is of importance when studying cancer therapies, drug resistance, and metastases.8,9 Therefore, in vitro assays of single-cell-derived sphere formation have emerged as exciting propositions for the identification of CSCs and studying their drug resistance. Single-cell-derived sphere assays are more puzzling methods than old-fashioned bulk tests.10−12 © 2019 American Chemical Society

Received: February 28, 2019 Accepted: May 31, 2019 Published: May 31, 2019 8318

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry

Figure 1. Integrated microfluidic system for single-cell separation and sphere formation according to cell size and deformability. (A) Schematic diagram of the design of the microdevice. The various channels are shown in different colors to help to distinguish the different components of the microfluidic device. Green indicates the cell culture chamber array and red indicates single-cell capture channels. The first pores (R) of the MOAL filter were 2 μm wider than the second pores (r), from input to output of the device the rows are 45, 45, 48, 48, and 48, respectively, the length (L) and width (W) of single-cell culture chamber is 100 and 100 μm, respectively. (B) An enlarged view of the yellow dotted-line box in (A). (C) Schematic diagram of the operation. The whole operation includes five steps: cell infusion, single-cell capture, chip inverting, sphere formation, and reagent infusion.

dielectrophoresis force, and microwell systems. Hydrodynamic techniques can typically achieve high cell capture rates (>80%).17 However, the hydrodynamic system usually requires a large area, which limits the assay under a microscope.18 Droplet systems can achieve high-throughput analyses by encapsulating single cells in aqueous droplets.19,20 Nevertheless, droplet approaches are limited by short assay times due to the difficulty in media exchange. The dielectrophoresis force is another method that exhibits size-dependent captures, with higher capture rates but requires a sophisticated active control.21 The microwell system is a simple yet effective tool to isolate single cells for clonal culture.22,23 Still, most microwell systems rely on random seeding without considering cell biomechanical properties (such as deformability and size) that have been reported to associate with CSCs behaviors (such as sphere formation, drug resistance, and metastatic potential).11,24−26 Cheng et al. presented a new platform that enables the capture of high-throughput size-based single cells, which integrates a device that is able to capture high-throughput single-cell that utilizes highly parallelized structures for cancer stemness and cell size correlation and single-cell-derived sphere formationstudies.11 Their results indicated that tumor cell size is associated with cancer stemness by monitoring sphere formation rates of different size subpopulations. However, the correlation between cells deformation and CSCs behaviors was not mentioned in this method as the tumor cell deformability has been proved to be associated with CSCs behaviors.25,26

In a previous study, we established a device for multistage microfluidics used for the construction of arrays of single cells by cell size and deformability.24 Our results indicated that tumor cell size and deformation are associated with drug resistance. Therefore, a microfluidics-based single-cell culture method that can explore the correlation between the heterogeneity of cellular biomechanics and stem-like cell behaviors is still needed for CSCs researches. To that effect, we present a new microfluidic device (Figure 1) that forms single-cell-derived tumor spheres and enables us to analyze the drug resistant aspect of the tumor spheres. The microdevice combines a multiobstacle architecture-looking (MOAL) microstructural matrices with microwell arrays. We studied the formation of spheres by single standard U251 glioblastoma cells and induced U251 glioblastoma cells of different sizes or deformations with the help of this device. We next utilized vincristine, a clinical antitumor drug, in our investigation of drug resistance by the created single-cellderived tumor-spheres. Furthermore, we analyzed two different signals of apoptosis (i.e., mitochondrial depolarization and caspase-3 activation).

■

EXPERIMENTAL SECTION Detailed information on the materials and methods described is contained in the Supporting Information (SI) attached. On-Chip Single-Cell-Derived Sphere Formation and Chemotherapy. The device was first irradiated with UV for 1 h and subsequently rinsed with autoclaved phosphate-buffered saline (PBS, 0.01 M, pH 7.4) before each on-chip experiment. 8319

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry

mammalian cell in a suspended state, the height of the filter was fixed at 25 μm.28 Compared with other microstructures such as funnel, point pillars, rectangular or single obstacle architectures,29−31 the MOAL filter has a “‘cell capture room”’ located between the two filter pores (i.e., pores 1 and 2) which mitigates jamming and ensures stable capture of the cells even with unexpected vibrations.31 The microwell (the length, width, and height were 100, 100, and 75 μm, respectively) was designed to provide room for sphere culture. As shown in SI Figures S2 and S3, the numerical simulation results showed that flow via the microunit is comfortably controllable, benefiting which enhances the capture of single cells and avoiding prevents cell clog and injury which can resulting from the contactcollision of cells with the PDMS surface.24,29,32 The prototypical device comprise of 5-group micro-units matrices assembled in a 2D array with 45−48 MOAL filters and 45−48 microwell arrays for single-cell trapping and sphere formation. Detailed dimensions are shown in SI Figure 1A and B. Structurally, the microfluidics platform consisted of two PDMS layers (SI Figure S1): the single-cell capture layer and cell culture chamber layer. Specifically, the single-cell capture layer contained 5-group MOAL filter matrices for single-cell capture. Every array presents constant pores 1 and 2 sizes, but the pores decrease 2 μm in the successive filter arrays; that is, the size of the first pore in each matrix was 16, 14, 12, 10, and 8 μm, and that of pore 2 was 14, 12, 10, 8, and 6 μm. The cell culture chamber layer contained a 5-group microwells array for sphere formation. As shown in Figure 1A, the branch channels of about 7500 micro-units for single-cell capture and cultivation are in the chip. The inlet is used for cell suspension and reagent introduction, whereas the outlets were designed for waste exclusion. The stable and continuous microflow is formed by the tree-like microchannel network located in the inlet.29 In a typical experiment, single-cell capture and cultivation needed five phases: cell infusion, single-cell capture, chip inverting, sphere formation, and reagent infusion (Figure 1C and SI Figure S4). In the first two listed phases, the cell sample was introduced initially into the micro-unit matrices with the flow rate of 20 μL/min. Once introduced, the sample was ushered through matrices of the micro-unit at various rates of drive flow for isolating single cells by cell size and deformability. In the chip inverting and sphere formation phases, the chip was inverted to download the trapped single cells into the microwells from the MOAL filter microstructural matrices. To cultivate single-cell-derived spheres, trapped single cells were cultured using fresh DMEM perfused at a speed of 5 μL/min. In the reagent infusion phase, we loaded our device with a mix of supplemented medium and an anticancer drug (10 μL/min) for coculture of tumor-spheres at different drug concentrations (from 1.25 to 80 μM) and treatment times (from 0 to 24 h). Operating Parameter Optimization. We used two different tumor cell types: standard U251 and induced U251 cells in this experiment. According to a previous study, these two types of glioblastoma cells are of different biomechanical and drug-resistant properties.24 Before the experiment, both cell types underwent individual staining with a green fluorescence dye (SI). The single-cell isolation efficiency of the two types of cells, which depends on the density of cells and the rate of drive flow,24 was first studied to evaluate this single-cell capture platform. We assessed the density of cells by

A sterilized solution of Pluronic F127 (10 mg/mL in water) was injected from the inlet into the chambers to prevent the cells from adhering to the surface of the chambers.24,27 This treatment was sustained for 2 h at 20 °C, followed by a PBS rinse. A syringe pump (Longer pump, LSP01−1A) was then used to introduce cell suspensions at different cell densities (from 2500 cells/mL to 25 000 cells/mL) into the chambers in a period of 20 s from the inlet at infusion flow fate of 20 μL/ min. The cell samples were then infused at different driving infusion flow rates (25 to 150 μL/min) to separate the singlecells. Sphere formation experiment was then performed after single-cell arrays were constructed with different size and deformability, culminating in the invention of the device. The trapped single-cells were then cultured using fresh supplemented Dulbecco’s modified Eagle’s medium (DMEM) at a very slow perfusion rate (5 μL/min).24 On-chip chemotherapy assays were performed with vincristine, a clinical anticancer drug. Single-cell-derived spheres that formed in the device were initially cultured in the microfluidics for 10 days. DMEM containing the anticancer drug was then introduced (at the rate of 5 μL/min) into the device to treat the tumor-spheres. Seven drug concentrations (1.25, 2.5, 5, 10, 20, 40, and 80 μM) and five treatment times (0, 6, 12, 18, and 24 h) were implemented in this experiment.27 Supplemented DMEM without vincristine was used as blank controls and were run simultaneously during the experiment. Characterization of Mitochondrial Membrane Potential and Caspase-3 Activation. Fluorescence characterization of the mitochondrial membrane potential and caspase-3 activation of cells was done to assess the apoptotic dynamics of cells in single-cell-derived spheres. Each staining procedure was conducted according to the instructions provided by the kit manufacturer. Mitochondrial membrane potential was estimated using the potential-sensitive fluorescent probe 5,5′,6,6′tetrachloro-1,1′,3,3′-tetraethylbenzimidazolyl- carbocyanine iodide (JC-1, Biotium; 5 μM in fresh DMEM). Solution JC-1 (5 μM in fresh DMEM) was loaded into the chambers at a flow rate of 10 μL/min and incubated for 15 min at 20 °C.27 A final rinse with PBS was then carried out for 15 min. NucView 488 caspase-3 substrate (a cell membrane-permeable fluorogenic caspase substrate, 1 μM in fresh DMEM)24 was used to analize caspase-3 activation of tumor cells. NucView 488 caspase-3 substrate was introduced into the chambers at a flow rate of 10 μL/min and incubated for 30 min at 20 °C, followed by a PBS rinse for 15 min. Image Acquisition and Analysis. The Olympus CKX41 inverted microscope was used to obtain all images. Image analysis was done using software Image-Pro Plus 6.0 (Media Cyternetics), and data statistical analysis was performed using SPSS 12.0 (SPSS Inc.). The quantitative data was presented as mean ± standard deviation (SD). Statistical significance of data was detemined using one-way analysis of variance (ANOVA).

■

RESULTS AND DISCUSSION Chip Design and Operation. We captured and cultured single-cells in a series of micro-units combined with MOAL filters and microwells. The MOAL filters were made to capture single cells by their biomechanical properties. As shown in SI Figure S1, the MOAL filters consisted of two adjacent I-shaped microstructures containing at least two cross-sectional areas (that is, pores 1 and 2; pore 1 is 2 μm broader than pore 2) for the capture of single cells. Accommodating the size of a typical 8320

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry

Figure 2. Generation of single-cell arrays in the micro-unit matrices under the optimum conditions: the density of infused cells was 10 000 cells per mL and the infusion flow rate was 100 μL min−1. (A and B) standard (A) and induced (B) U251 cells captured in the micro-unit matrices. Merged and green fluorescence channels are shown in the first and second rows to demonstrate single-cell trapped in the micro-unit matrices. The scale bars in (A) and (B) represent 400 μm. (C and D) Quantitative analysis of the cell distribution rate (C) and number of cells (D) in different micro-unit matrices of the device. The standard deviations deduced from ten parallel experiments are shown as the error bars. The standard deviations in (C) and (D) deduced from ten parallel experiments are shown as the error bars, with the significance assessed by ANOVA. **p < 0.01; *p < 0.05; N.S., not significant.

single-cell isolation experiment. Cell viability assays (SI Figure S7) also indicated that the isolation operation had no effect on the cell viability (>93%) using the selected parameters. The cell distribution of the two different tumor cell types in the micro-unit matrices was assessed using the selected parameters. The results are shown in Figure 2. Expectedly, the distribution of standard U251 cells predominantly occurred in 14, 12, and 10 μm matrices of the micro-unit (that is, pore 2 sizes = 14, 12, and 10 μm, respectively), whereas, the induced U251 cells were distributed primarily in 14, 8, and 6 μm microunit matrices (that is, pore 2 sizes = 14, 8, and 6 μm, respectively). This observation could be ascribed to the difference in biomechanical heterogeneities of the standard and induced U251 cells,24 whereby induced U251 cells are smaller and more deformable than standard U251 cell. Based on our previous study,24 the reason may be that the size of a cell is a considerable factor for cell capture in the micro-unit matrices in relation to the second pore size over 10 μm (that is, pore 2 sizes = 14 and 12 μm, respectively), and the cell deformability mainly affects the cell capture in the 8 and 6 μm

infusing two categories of cells individually into the micro-unit matrices using the varying cell densities (2500, 5000, 10 000, 15 000, 20 000, and 25 000 cells/mL) at a driving flow rate of 40 μL/min. The results are shown in SI Figure S5. Expectedly, the single-cell isolation efficiency decreased with increasing cell density. Usually, single-cell isolations at cell densities of 2500, 5000, and 10 000 cells/mL (>80%) are more efficient than those at cell densities of 15 000, 20 000, and 25 000 cells/mL (85%) at 100 μL/min, after which increasing flow rates (i.e., 125 and 150 μL/min) then triggered diminishing isolation efficiency. Consequently, 100 μL/min was selected for the 8321

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry

Figure 3. Sphere formation rate and sphere size in chambers starting from a single cell. (A and B) images of single standard (A) and induced (B) U251 cell-derived sphere in different micro-unit matrices. The scale bars in (A) and (B) represent 25 μm. (C and D) Quantitative analysis of the sphere formation rate (C) and sphere area (D) in different micro-unit matrices of the device. The sizes of the micro-unit matrices are the sizes of the second pores in each MOAL filter matrix. The standard deviations deduced from ten parallel experiments are shown as the error bars. The standard deviations in (C) and (D) deduced from ten parallel experiments are shown as the error bars, with the significance assessed by ANOVA. **p < 0.01; *p < 0.05; N.S., not significant.

micro-unit matrices (that is, pore 2 sizes = 8 and 6 μm, respectively); for the 10 μm micro-unit matrix (that is, pore 2 sizes = 10 μm), the capture process was dually influenced by the cell size and its deformability. Single-Cell-Derived Sphere Formation. Two tumor cell types (standard U251 and induced U251) were loaded and cultured in the device for 10 days to validate the single-cellderived sphere assay. Before the experiment, the microdevice was coated with F127, which prevents the cells from adhering to the substrate, enabling single cells cultured on the chip to help investigate sphere formation rates and sphere area.27 Generally, in suspension culture, stem-like cells grow into spheres from single cells, whereas nonstem-like cells die as a result of anoikis (Figure 3 and SI Figure S8). Based on the previous studies,16−18 the sphere is the clusters of cultured cancer cells which usually consist of more than 30 cells observed under light microscopy. We observed the sphere formation efficiency of standard U251 and induced U251 cells after 10 days of culture. As shown in SI Figure S9, the sphere formation by the single induced U251 cell is relatively more efficient than the sphere formation by standard U251 cells. Also, according to sphere area assays (SI Figure S10), the average sphere area of single-induced-cell-derived spheres is comparatively higher than that of single-standard-cell-derived spheres. A control experiment was conducted to compare the cell proliferation rate of the two types of cultured cells. The results indicated that the cell proliferation rate of induced U251 cells was similar to the standard U251 cells using the conventional method, and no noticeable differences were

observed (SI Figure S11). Therefore, the reason may be that the induced U251 cells included more CSCs than that of the standard U251 cells.24 To further explore the sphere formation differences between the single tumor cells, the sphere formation rates and area of the single-cell-derived spheres in the micro-unit matrix were investigated after 10 days of culture. The results (Figure 3) showed that the sphere formation rates and area of single-cellderived spheres increased with the decrease of the micro-unit matrices, reaching apex in 6-μm micro-unit matrices (that is, pore 2 size = 6 μm). It has been reported that smaller and/or more deformed cells are characterized by more G0/G1 phases than those of more prominent and/or lesser deformable cells.33,34 Generally, cells in G0/G1 phases (gap) have lower proliferative rate than those in S (DNA synthesis) and M (mitosis).35,36 Therefore, the reason may be that the smaller and/or more deformed cancer cells exhibit more CSC property than those of more prominent and/or lesser deformable tumor cells.11,25 Additionally, all sphere-forming cells are considered cancer stem-like based on a conventional definition, implicating cellular heterogeneity even within cancer stem-like cells; that is, the single smaller and/or more deformable tumor cell which included more CSC than more prominent and/or lesser deformable CSC could form more and larger spheres. Chemotherapy. One significant employment of tumor spheres from in vitro cultures is in the evaluation of the efficacy of anticancer drugs. We made use of two different single-cellderived spheres (single standard U251 and induced U251 cells derived) and vincristine, a clinical anticancer drug, in our quest 8322

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry

Figure 4. Assays of the cell viability (A and B) and mitochondrial membrane potential (C, D, and E) of single-cell-derived sphere in different micro-unit matrices after treatment for 12 h with 20 μM vincristine. (A) Fluorescence images of standard (first row) and induced (second row) U251 cells after vincristine treatment. Live/dead (green/red) cells were visualized by staining with FDA/PI. (B) Statistical cell viability of singlecell-derived sphere in different micro-unit matrices after vincristine treatment. (C) Fluorescence images using JC-1 aggregate (red) and monomer (green) of standard U251 cells after vincristine treatment. (D) and (E) Statistical ratio of JC-1 aggregate to its monomeric form in single standard and induced U251 cell-derived sphere after vincristine treatment. The sizes of the micro-unit matrices are the sizes of the second pores in each MOAL filter matrix. The scale bars in (A) and (C) represent 25 μm. The standard deviations in (B) deduced from 10 parallel experiments are shown as the error bars, with the significance assessed by ANOVA. **p < 0.01; *p < 0.05; N.S., not significant.

to assess any liaison between the biomechanical heterogeneity of cancer cells and their drug resistance. Vincristine is a microtubule-targeting agent, and it irreversibly binds to microtubules and spindle proteins in mitotic S-phase.37−39 Thus, as vincristine interferes with mitotic spindle assembly, it has the ability to inhibit tumor cell development and activates tumor cells apoptosis.37 CSCs are famous for their drugresistance as they could evade the effects of vincristine as their cell cycle is mainly in the G0/G1 phase. Furthermore, CSCs are able to overexpress antiapoptotic factors or silence key death effectors.38 In this study, we employed the ratio of the viability of the single induced to standard U251 cell-derived sphere to mirror the quantifiable dynamics of the viability of cells in the course of chemotherapy. SI Figure S12 represents the obtained results. The viability of the two different single-cell-derived spheres decreased time-dependently with increasing drug concentrations (more extended drug treatment with the same drug concentrations led to a decrease in the viability of cells). An in-depth analysis (SI Figure S13) revealed that tumor cells’ drug resistance in single induced U251 cell-derived spheres stood comparatively higher than standard U251 cells’ resistance to drugs. Specifically, after treatment with vincristine (20 μM) for a duration exceeding 6 h, the difference in the viability of cells between standard and induced tumors was more significant than the one recorded when tumor cells were treated with the same drug at other concentrations (SI Figure S12), attaining apex after 12 h of treatment (SI Figure S12B), possibly because the cells of single induced U251 cell-derived spheres were prone to chemotherapy at higher concentrations of the drug (40 and 80 μM) but not at 20 μM or lower (1.25, 2.5, 5, and 10 μM), concentrations at which the cells resisted chemotherapy.

Further exploration of the difference in resistance to drugs by cancer cells in the single-cell-derived spheres required investigating the standard U251 and induced U251 cell viabilities of single-cell-derived spheres spread in different micro-unit matrices postvincristine treatment. Accordingly, (Figure 4A,B and SI Figure S13) tumor cells of single smaller and/or more deformable tumor cell-derived spheres were more resistant than that of the single more prominent and/or lesser deformable tumor cell-derived spheres. In particular, the viability of cells in the single-cell-derived spheres in the 14and 12- μm micro-unit matrices (that is, pore 2 sizes = 14 and 12 μm) was worse than that in the 10- and 8 μm filter matrices (that is, pore 2 sizes = 10 and 8 μm), whereas tumor cell viability in the single-cell-derived spheres of the 6 μm microunit matrices was the highest (that is, pore 2 sizes = 6 μm). Such a finding may probably be because the drug resistance of cells in tumor spheres strongly depends on sizes of spheres since the size of a smaller and/or more deformable tumor cellderived sphere surpasses the size of a single more prominent and/or lesser deformable tumor cell-derived sphere. Compare with the conventional plate-based method (SI), the result indicated that the on-chip tumor spheres had higher drug resistance than the conventional 2D culture method (SI Figure S14). The reason may be that tumor sphere provides an improved model that mimics the biological properties of tumors in vivo than the conventional method.27 In addition, the single cell-based tumor sphere contained more CSCs than those of the conventional method.11,24,27 Mitochondrial Membrane Potential and Caspase-3 Activity. We monitored the depolarization of the mitochondria and the activation of caspase-3 during chemotherapy for additional assessment of the drug-resistance of the two types of 8323

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry single-cell-derived spheres.24,27 Mitochondrial depolarization was assessed first with the help of a fluorescent mitochondrial probe JC-1. Generally, JC-1 exists as aggregates with high membrane potential red fluorescence that occur postuptake in healthy mitochondria and as green fluorescent monomers contained in unhealthy or apoptotic cells that harbor collapsed mitochondria.40,41 Hence, the ratio of the red fluorescence to green fluorescence was used to sample the performance of the fluorescence intensity dynamics of the potential of the mitochondrial membrane using 20 μM vincristine for a 12 h treatment. The results showed that the red/green fluorescence ratios of the standard and induced U251 cells significantly decreased after treatment with vincristine (Figure 4C−E and SI Figures S15 and S16). Comparison with cells cultured by the conventional plate-based method (SI) shows that the red/ green fluorescence ratios of the on-chip tumor spheres were higher than the cultured-cells produced by conventional method (SI Figures S17 and S18). Moveover, the analysis (Figures 4C−E and SI Figure S15) showed that cells in the 14-, and 12 μm micro-unit matrices (that is, pore 2 sizes = 14 and 12 μm) had lower ratios of fluorescence compared cells in the 10- and 8 μm micro-unit matrices (that is, pore 2 sizes = 10 and 8 μm). The fluorescence ratios were highest in the 6 μm micro-unit matrices (that is, pore 2 sizes = 6 μm). These results suggest that vincristine treatment enables the single smaller and/or more deformable cancer cell-derived spheres to present fewer mitochondrial abnormalities than the single more prominent and/or lesser deformable cell-derived spheres. Caspases, which encompasses cysteine proteases, can cause apoptosis of cells (programming of cell death) when activated. Here, we used NucView 488 Caspase-3, which has a fluorogenic DNA-binding dye and a DEVD substrate moiety, with specificity to caspase 3 to measure caspase 3 activation. The substrate usually does not function and is nonfluorescent in its capacity as a DNA dye. Nevertheless, the substrate has the capability of labeling the nucleus green post-caspase-3 cleavage.27 We analyzed caspase-3 positive cells (caspase-3+ cells) of single-cell-derived spheres separately during the vincristine chemotherapy. According to our findings, drug treatment caused caspase-3+ cells of single-cell-derived spheres to be elevated time-dependently (SI Figures S19, S20, and S21). When compared with cells cultured conventionally by plate-based method (SI), it was observed that the drug treatment caused caspase-3+ cells of the on-chip tumor spheres were lower than the cells cultured conventionally (SI Figures S22, S23, S24, and S25). Moreover, a quantitative assessment (Figure 5) of the caspase-3+ cells of single-cell-derived spheres caught in various micro-unit matrices registered caspase-3+ cells as being of more substantial amounts in the 14- and 12 μm micro-unit matrices than the numbers in the 10- and 8 μm micro-unit matrices. The amount of the caspase-3+ cells were lowest in the 6 μm micro-unit matrices (that is, pore 2 sizes = 6 μm). These findings suggest that vincristine treatment enables single smaller and/or more deformable cell-derived spheres to produce less caspase-3+ cells than single more prominent and/ or lesser deformable cell-derived spheres.

Figure 5. Percentages of caspase-3+ cells in single standard (A) and induced (B) U251 cell-derived sphere in the micro-unit matrices on the device after different time treatments with 20 μM vincristine. The standard deviations deduced from ten parallel experiments are shown as the error bars. The sizes of the micro-unit matrices are the sizes of the second pores in each MOAL filter matrix.

from different sizes and deformed single cells could be formed easily, and drug resistance could be investigated on a single device. We observed the sphere formation of single glioblastoma cells (standard U251 and induced U251) with different sizes and deformations. The drug resistance of these tumor spheres was also assessed using vincristine. While all sphere-forming cells are considered cancer stem-like based on the conventional definition, the growth dynamics and drug resistance of single-cell-derived spheres implicates the cellular heterogeneity within CSCs. Furthermore, the smaller and/or more deformable CSCs could have more resistance to chemotherapeutic drugs than more prominent and/or lesser deformable CSCs. This observation can only be enabled by scaling single-cell assays. In the future, the CSCs heterogeneity described in this device can be investigated further by using reporters to screen antitumor drugs and retrieve them for gene expression analysis.

■

■

ASSOCIATED CONTENT

S Supporting Information *

CONCLUSIONS In summary, we have developed a new platform for microfluidic assays that can analyze the formation of singlecell-derived tumor-spheres and drug-resistance. The microdevice combines MOAL microstructural matrices and microwell arrays. With the help of this device, tumor-spheres derived

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.9b01084. Additional information as noted in text (PDF) 8324

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325

Article

Analytical Chemistry

■

(24) Pang, L.; Liu, W.; Tian, C.; Xu, J.; Li, T.; Chen, S. W.; Wang, J. Y. Lab Chip 2016, 16, 4612−4620. (25) Zhang, W.; Kai, K.; Choi, D. S.; Iwamotod, T.; Nguyen, Y. H.; Wong, H.; Landis, M. D.; Uenod, N. T.; Chang, J.; Qin, L. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 18707−18712. (26) Gill, N. K.; Ly, C.; Nyberg, K. D.; Lee, L.; Qi, D.; Tofig, B.; Reis-Sobreiro, M.; Dorigo, O.; Rao, J. Y.; Wiedemeyer, R.; Karlan, B.; Lawrenson, K.; Freeman, M. R.; Damoiseaux, R.; Karlan, B. Lab Chip 2019, 19, 343−357. (27) Liu, W.; Xu, J.; Li, T.; Zhao, L.; Ma, C.; Shen, S. F.; Wang, J. Y. Anal. Chem. 2015, 87, 9752−9760. (28) McFaul, S. M.; Lin, B. K.; Ma, H. S. Lab Chip 2012, 12, 2369− 2376. (29) Pang, L.; Shen, S. F.; Ma, C.; Ma, T. T.; Zhang, R.; Tian, C.; Zhao, L.; Liu, W. M.; Wang, J. Y. Analyst 2015, 140, 7335−7346. (30) Ji, H. M.; Samper, V.; Chen, Y.; Heng, C. K.; Lim, T. M.; Yobas, L. Biomed. Microdevices 2008, 10, 251−257. (31) Kim, M. S.; Sim, T. S.; Kim, Y. J.; Kim, S. S.; Jeong, H.; Park, J. M.; Moon, H. S.; Kim, S.; Gurel, O.; Lee, S. S.; Lee, J. G.; Park, J. C. Lab Chip 2012, 12, 2874−2880. (32) Shen, S. F.; Ma, C.; Zhao, L.; Wang, Y. L.; Wang, J. C.; Xu, J.; Li, T. B.; Pang, L.; Wang, J. Y. Lab Chip 2014, 14, 2525−2538. (33) Lee, W. C.; Bhagat, A. A. S.; Huang, S.; Vliet, K. J. V.; Han, J. Y.; Lim, C. T. Lab Chip 2011, 11, 1359−1367. (34) Lee, W. C.; Shi, H.; Poon, Z.; Nyan, L. M.; Kaushik, T.; Shivashankar, G. V.; Chan, J. K.Y.; Lim, C. T.; Han, J. Y.; Vliet, K. J. V. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, E4409−E4418. (35) Murray, A. W. Nature 1992, 359, 599−601. (36) Hartwell, L. H.; Kastan, M. B. Science 1994, 266, 1821−1828. (37) Jordan, M. A.; Hirnes, R. H.; Wilson, L. Cancer Res. 1985, 45, 2741−2747. (38) Yuan, X.; Curtin, J.; Xiong, Y.; Liu, G.; Waschsmann-Hogiu, S.; Farkas, D. L.; Black, K. L.; Yu, J. S. Oncogene 2004, 23, 9392−9400. (39) Yu, S.; Ping, Y.; Yi, L.; Zhou, Z.; Chen, J.; Yao, X.; Gao, L.; Wang, J. M.; Bian, X. Cancer Lett. 2008, 265, 124−134. (40) Elmore, S. Toxicol. Pathol. 2007, 35, 495−516. (41) Hamacher-Brady, A.; Choe, S. C.; Krijnse-Locker, J.; Brady, N. R. Cell Death Differ. 2014, 21, 1862−1876.

AUTHOR INFORMATION

Corresponding Authors

*(J.F.) Phone: + 86-29-861 775 52; fax: + 86-29-861 775 52; e-mail: [email protected]. *(S.-K.F.) Phone: +886-233664515; fax +886-223631755; email: [email protected]. ORCID

Long Pang: 0000-0002-9621-0525 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work is supported by the National Natural Science Foundation of China (81702955 and 81770457), the Project of Shaanxi Key Laboratory of Brain Disorders (18NBZD03) and the Fundamental Research Foundation of Xi’an Medical University (2016DOC26, 2018PT16, and 2018GJFY02).

■

REFERENCES

(1) Lawson, D. A.; Kessenbrock, K.; Davis, R. T.; Pervolarakis, N.; Werb, Z. Nat. Cell Biol. 2018, 20, 1349−1360. (2) Dean, M.; Fojo, T.; Bates, S. Nat. Rev. Cancer 2005, 5, 275−284. (3) Reya, T.; Morrison, S. J.; Clarke, M. F.; Weissman, I. L. Nature 2001, 414, 105−110. (4) Singh, S. K.; Clarke, I. D.; Terasaki, M.; Bonn, V. E.; Hawkins, C.; Squire, J.; Dirks, P. B. Cancer Res. 2003, 63, 5821−5828. (5) Ricardo, S.; Vieira, A. F.; Gerhard, R.; Leitão, D.; Pinto, R.; Cameselle-Teijeiro, J. F.; Milanezi, F.; Schmitt, F.; Paredes, J. J. Clin. Pathol. 2011, 64, 937−946. (6) Batlle, E.; Clevers, H. Nat. Med. 2017, 23, 1124−1134. (7) Kruyt, F. J. P.; Schuringa, J. J. Biochem. Pharmacol. 2010, 80, 423−430. (8) Vermeulen, L.; Todaro, M.; F. Mello, S.; Sprick, M. R.; Kemper, K.; Alea, M. P.; Richel, D. J.; Stassi, G.; Medema, J. P. Proc. Natl. Acad. Sci. U. S. A. 2009, 105, 13427−13432. (9) Dieter, S. M.; Ball, C. R.; Hoffmann, C. M.; Nowrouzi, A.; Herbst, F.; Zavidij, O.; Abel, U.; Arens, A.; Weichert, W.; Brand, K.; Koch, M.; Weitz, J.; Schmidt, M.; Kalle, C.; Glimm, H. Cell stem cell. 2011, 9, 357−365. (10) Gach, P. C.; Attayek, P. J.; Herrera, G.; Yeh, J. J.; Allbritton, N. L. Anal. Chem. 2013, 85, 7271−7278. (11) Cheng, Y. H.; Chen, Y. C.; Briena, R.; Yoon, E. Lab Chip 2016, 16, 3708−3717. (12) Armbrecht, L.; Dittrich, P. S. Anal. Chem. 2017, 89, 2−21. (13) Stingl, J. J. Pathol. 2009, 217, 229−241. (14) Carlo, D. D.; Aghdam, N.; Lee, L. P. Anal. Chem. 2006, 78, 4925−4930. (15) Zhao, L.; Ma, C.; Shen, S.; Tian, C.; Xu, J.; Tu, Q.; Li, T. B.; Wang, Y. L.; Wang, J. Y. Biosens. Bioelectron. 2016, 78, 423−430. (16) Chen, Y. C.; Ingram, P. N.; Fouladdel, S.; McDermot, S. P.; Azizi, E.; Wicha, M. S.; Yoon, E. Sci. Rep. 2016, 6, 27301. (17) Chen, Y. C.; Zhang, Z.; Fouladdel, S.; Deolc, Y.; Ingramb, P. N.; McDermot, S. P.; Azizi, E.; Wicha, M. S.; Yoon, E. Lab Chip 2016, 16, 2935−2945. (18) Chen, Y. C.; Cheng, Y. H.; Kim, H. S.; Yoon, E. Lab Chip 2014, 14, 2941−2947. (19) Suteria, N. S.; Nekouei, M.; Vanapalli, S. A. Lab Chip 2018, 18, 343−355. (20) Huebner, A.; Bratton, D.; Whyte, G.; Yang, M.; Demello, A. J.; Abell, C.; Hollfelder, F. Lab Chip 2009, 9, 692−698. (21) Dittrich, P.; Ibáñez, A. J. Electrophoresis 2015, 36, 2196−2206. (22) Shen, F. M.; Zhu, L.; Ye, H.; Yang, Y. J.; Pang, D. W.; Zhang, Z. L. Sci. Rep. 2015, 5, 11937. (23) Guan, Z.; Jia, S.; Zhu, Z.; Zhang, M.; Yang, C. J. Anal. Chem. 2014, 86, 2789−2797. 8325

DOI: 10.1021/acs.analchem.9b01084 Anal. Chem. 2019, 91, 8318−8325