*2.2. Cell Preparation*

A human monocytic cell line (THP-1, ATCC TIB-202, Manassas, VA, USA) was cultured in a complete RPMI-1640 culture medium supplemented with 10% fetal bovine serum and 1% penicillin. An immortal human nasopharyngeal epithelial cell line (NP 460) and a nasopharyngeal carcinoma cell line (NPC 43) were developed and donated by the research team of S. W. Tsao, from cell extracts of nasopharyngeal cancer patients [27]. NP 460 cells were cultured in a mixture of 50% complete Eplife medium (Thermo Fisher Scientific, Waltham, MA, USA), 50% complete defined keratinocyte−SFM (Thermo Fisher Scientific, Waltham, MA, USA) with 100 units/mL penicillin and 100 μg/mL streptomycin. NPC 43 cells were maintained in RPMI-1640 (Sigma-Aldrich, St. Louis, MO, USA) added with 10% fetal bovine serum, 4 μM Y27632 dihydrochloride (Alexis), 100 unit/mL penicillin and 100 μg/mL streptomycin. All the cell types were cultured in a 37 ◦C incubator (HERA cell 150, Thermo Fisher Scientific, Waltham, MA, USA) with a humidified and 5% CO2 environment. Additionally, the macrophages were derived from THP−1 cells before experiments. THP−1 cells with a density of 1 × <sup>10</sup><sup>6</sup> cells/mL was treated with 50 ng/mL of phorbol 12-myristate 13-acetate [28] (PMA; Sigma-Aldrich, St. Louis, MO, USA) for 20 h. The treated cells were then trypsinized (Sigma-Aldrich, St. Louis, MO, USA) for 3 min at room temperature and replaced with fresh media before transferring to the device.

#### *2.3. Automated Microscope and Cytokine Quantification*

All the microfluidic manipulation for cell culture, monitoring and imaging was operated by an automated microscope platform developed in our laboratory [29]. It mainly includes an inverted fluorescence microscope (IX71, Olympus, Tokyo, Japan) integrated with a microscope camera (Zyla 4.2, Andor Technology Ltd., Belfast, UK) with computercontrolled compressed air supply manifolds and a confining shield mounted on a computercontrolled movable stage of the microscope, offering stable temperature (37 ◦C), humidity and gas (5% CO2) conditions for cell culture (Figure S2 in the Supplementary Material). Cytokine concentrations were quantified using a commercial human inflammatory cytokine kit (Catalog No. 551811, BD Biosciences) in all samples.

#### *2.4. Statistics*

All experiments were conducted with at least four independent experiments. The *p*-values were calculated using Student's *t*-test in Excel (Microsoft), with *p* < 0.05 considered as statistically significant.

#### **3. Results and Discussion**

#### *3.1. Device Design and Operation*

We have developed an integrated microfluidic immunoassay device, which can implement in-situ and multiplex monitoring of time-lapsed cytokine secretions in an extracellular matrix (ECM) protein-coated topologic environment. This device consists of three main components: a cell culture chamber with parallel microgratings, cytokine detection arrays (four rows and four columns) and an array of active peristaltic mixers (Figure 1a,b). The parallel gratings (each with 18 μm in width and 18 μm in depth) fabricated by polydimethylsiloxane (PDMS) replica molding can mimic the microtopography of pterygoid muscles [23]. The optically transparent cell culture chamber allows the direct observation of cell behaviors and live cell staining under a microscope. Furthermore, it was surrounded by a barrier containing multiple micro-gaps (height: 4 μm; width: 50 μm) to confine cells inside the cell culture chamber, while allowing cell culture media and the secreted cytokines to flow to one of the cytokine detection chambers to achieve the microbeads-based immunoassay for multiple cytokines at different time points. The extracted media were incubated with antibody-conjugated microbeads for capturing cytokines, followed by mixing with phycoerythrin (PE)-conjugated detection antibodies to form fluorescent sandwich complexes for microscopic imaging and cytokine quantification (Figure 1c). In each cytokine detection chamber (volume: 160 nl), a bypass channel (volume: 30 nl) was integrated with a peristaltic mixer, which consisted of three peristaltic microvalves for active pumping and mixing to accelerate the cytokine detection based on the Taylor dispersion effect [30]. The whole assay can be conducted in less than 30 min.

Notably, we applied 20 μg/mL bovine collagen (Sigma Aldrich, St. Louis, MO, USA) [31] into the culture chamber for 2 h at room temperature before injecting the selected cells, because collagen is a key ECM protein of pterygoid muscles [32]. The nasopharyngeal cells (NP460 or NPC43) and macrophage cells were pre-mixed with a defined population ratio and seeded into the cell culture chamber through the culture inlet and incubated in a microscope incubator with stabilized temperature at 37 ◦C and 5% CO2 to facilitate immune cells secreting cytokines in response to target cells. Simultaneously, the detection microbeads were loaded into the bypass channel aside the cytokine detection chamber by opening the corresponding control valves for defining the access from the 'Microbeads' inlet (Figure 1b) to the target detection chamber. We then waited for 10 min for the microbeads sitting on the bottom of the microchamber. The device operation for extracting and quantifying cytokine is shown in Figure 2a and Supplementary video S1. In each measurement, 0.5 μL of the media with secreted cytokines in the culture chamber was extracted from the culture chamber and transported into a detection microchamber with preloaded microbeads for cytokine detection. Meanwhile, fresh media flowed into the culture chamber to supplement the media being extracted. The bypass channel corresponding

to the defined cytokine detection chamber was flushed with a PE-conjugated antibody solution, and then the peristaltic mixer was activated for 2 min to mix the PE solution and microbeads (Supplementary video S2). We removed the unconjugated antibody solution by flowing phosphate-buffered saline (PBS) along the detection chamber under gentle driving compressed air with pressure of 0.2 psi for 3 min. The microbeads expressed fluorescence at 647 nm from the bead bodies and the bound cytokine molecules induced fluorescence at 488 nm over the bead surfaces. After imaging the microbeads using an inverted fluorescent microscope, the measured fluorescence intensities at 488 nm were converted to the cytokine concentrations according to the calibration curves (Figure 2b).

**Figure 2.** (**a**) Cytokine detection procedures of the integrated microfluidic immunoassay device. Cytokines produced by NPC and immune cells on-chip were transferred from the cell culture region to the cytokine detection arrays, and specifically captured by the pre-loaded antibody conjugated microbeads, followed by washing with PBS buffer. To quantify the cytokine concentrations on-chip, fluorescent phycoerythrin (PE)-conjugated detection antibodies were mixed with the cytokine-binding microbeads using an integrated micromixer (middle panel, representative microscope images), causing the fluorescence change on the microbeads as a readout of binding events. Finally, the fluorescence change was quantified by microscope imaging, and the cytokine concentration was calculated according to the calibration curves. (**b**) The calibration curve of two different cytometric microbeads for TNF and IL12p70 in the microfluidic device by challenging microbeads with different concentrations of cytokines. The data points for 2.4 pg/mL cytokine concentration are not included in the fitting line. Each data point was obtained from the average value of *n* > 100 from 3 repeated measurements. All error bars represent the standard errors.

To obtain the calibration curves for the selected cytokines, different concentrations (2.4 pg/mL, 4.9 pg/mL, 9.8 pg/mL, 19.5 pg/mL, 39.1 pg/mL, 78.1 pg/mL, 156.3 pg/mL, 312.5 pg/mL, 625 pg/mL, 1250 pg/mL, 2500 pg/mL, 5000 pg/mL) of each cytokine were prepared by standard PBS dilution of the stocking cytokine solution (5 ng/mL). Linear regression with R2 > 0.9 was observed between fluorescence intensity (488 nm) and molecular concentrations for each cytokine type. The limit of detection was calculated as ~5 pg/mL for both cytokines based on the concentration, giving a signal equal to the blank signal (Y0) plus three standard deviations of the blank (3σ), or 4 pg/mL for TNF and 3 pg/mL for IL12p70 based on 3σ/S, where S is the slope of the calibration curve.

#### *3.2. Viability and Cytokine Secretion of Single Cell-Type Cultures*

To verify cell viability in the culture chamber, we applied live/dead-cell staining (L-3224, Thermo Fisher Scientific, Waltham, MA, USA) to the immune cells (THP-1 cells and differentiated macrophages) and nasopharyngeal cells (NP460 and NPC43) growing in the culture chamber at a cell density of 1 × <sup>10</sup><sup>5</sup> cells/mL. For instance, a stained NPC43 pseudopod attaching on a micro-grating can be captured as shown in Figure 3a, which suggests the good biocompatibility of the on-chip microtopographic environment; and three-dimensional fluorescence micrographs of stained NPC43 cells are shown in Figure 3b, where green cells are live cells and red cells are dead cells. Our results (Figure 3c) indicate that viabilities of NP460, NPC43, THP-1 and macrophages are maintained at >95% for the cells growing in the culture chamber for 24 h.

**Figure 3.** (**a**) Top view (*left* panel) and three-dimensional view (*right* panel) of the reconstructed confocal microscopy images of a NPC43 cell culture on a micro-grating array. Scale bar: 20 μm. (**b**) Brightfield (*left*) and live/dead-stained (*right*) images of NPC43 cultured on micro-gratings for 8 h. Scale bar: 30 μm. (**c**) Viability of NPC43, NP460, THP-1 and THP-1-derived macrophages in the culturing chamber for 8 h. Asterisk represents a *p*-value of <0.05 calculated by Student's *t*-test. (**d**) Cytokine secretions of mono cultured NPC43/NP460 cells and immune cells on micro-gratings after 8 h of culture. Error bars represent the standard errors.

Additionally, we have also measured the cytokine secretion levels (TNF and IL-12p70) of the single cell-type cultures as summarized in Figure 3d. As described above, the simulated macrophages produce more TNF-α and IL-12 [5,6]; and EBV-infection of the monocytes can cause suppressed IL-12 secretion [8] via the induced MCP-1 expression [7]. Therefore, we selected and quantified TNF and IL-12p70 (an active heterodimer of IL-12) for reflecting the simulated activity and the EBV-infection of the macrophages, respectively. Each of the cell types was seeded in the culture chamber at a density of 1 × 105 cells/mL and incubated for 8 h before the cytokine quantification. In brief, our results show that the unstimulated macrophages can secrete a measurable level of IL-12-p70 (65 pg/mL); whereas relatively larger portions of TNF can be contributed from macrophages (80 pg/mL) and NPC43 cells (62 pg/mL). Such measurements of the single cell-type cultures can determine the baselines of the cytokine levels for the immune-nasopharyngeal cell cocultures.
