A high-throughput cell culture system based on capillary and centrifugal actions for rapid antimicrobial susceptibility testing

Taegeun Lim ab, Eun-Geun Kim c, Jungil Choi *d and Sunghoon Kwon *ab
aQuantaMatrix Inc., Seoul National University Hospital CMI, Seoul, 03082, Republic of Korea
bDepartment of Electrical Engineering and Computer Science, Seoul National University, Seoul, 08826, Republic of Korea. E-mail: skwon@snu.ac.kr
cLowendtechnologies Co., Anyang, 14056, Republic of Korea
dSchool of Mechanical Engineering, Kookmin University, Seoul, 02707, Republic of Korea. E-mail: jchoi@kookmin.ac.kr

Received 27th July 2020 , Accepted 2nd November 2020

First published on 2nd November 2020

Antibiotic resistance is a global threat to modern society. Rapid determination of suitable antibiotics that inhibit bacterial growth can effectively reduce antibiotic resistance and improve clinical treatment. The conventional methods of antimicrobial susceptibility testing (AST) depend on optical density measurements, which require long-time incubation. Various kinds of rapid AST systems which utilize various technologies from the field of lab on a chip have promised a great reduction in measurement time, but cannot achieve high-throughput, user-friendly testing due to the complexity of the testing system. Here, we introduce a capillary and centrifuge-based rapid AST system that reduces the time of loading the sample and culture media while achieving a high-throughput testing capacity. The capability of the proposed system is validated in a systematic analysis that includes sample loading characteristics and AST trials with standard strains. The proposed system provides a useful tool for drug testing in cell-culture systems with user-friendly and high-throughput analysis.


Antibiotic resistance has become a global clinical problem.1 Recently, multi-antibiotic resistant bacteria such as methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci are increasing in hospitals, and are also found in residential environments.2,3 The development of new antibiotic classes has been accelerated by administerial support.4 Along with drug development, timely and accurate administration of antibiotics is necessary for combatting antibiotic resistance.5,6 The suitable antibiotic for a bacterial infection must be determined by antimicrobial susceptibility testing (AST). In conventional AST, bacterial growth is detected by the optical density changes in a testing medium, which requires overnight incubaction.7 Bacteremia, which ultimately leads to sepsis, demands a rapid analysis that cannot be provided by conventional AST methods.8

In the past decade, various kinds of rapid AST have reduced the test times to a few hours.9 The detection principle can be phenotypic or genotypic. Phenotypic methods reduce the AST time by lowering the detection limit of bacterial growth using some techniques: microscopic imaging of bacteria,10–16 measuring the protein displacement profile,17 monitoring bacterial metabolism,18 detecting the pH of culture media,19 measuring fluorescence signals,20 laser light scattering,21 electrical resistance,22 Raman spectroscopy,23 droplet digital polymerase chain reaction,24 or detecting the bacteria-induced vibrations of a cantilever.25 Although these methods promise to reduce the AST time, they lack a user-friendly protocol and a high-throughput system for integration into clinical practice. To determine whether a bacterial strain is susceptible, intermediate or resistant, conventional AST systems test up to 20 clinically available antibiotics at 2–4 concentrations simultaneously.26 To meet this requirement, a high-throughput testing platform must handle approximately 50 testing conditions. A 96-well testing system satisfies these criteria, but the clinical environment demands a quick sample-loading system to handle multiple tests from multiple patients.

The most commonly used AST systems are Vitek2 and MicroScan, which can simultaneously test 30–100 samples. These systems employ their own processes for handling the samples and loading the culture media. However, new rapid AST systems either are incapable of high-throughput testing, or require cumbersome procedures (i.e., loading the samples and testing their antimicrobial susceptibilities). Specifically, both the bacterial sample and culture media must be loaded into each testing well, which doubles the pipetting time per well. Multiple pipetting also increases the possibility of misloading and contamination. To remove these difficulties, a simple system that loads a liquid sample into separate testing chambers is required.

A lab-on-a-disc system operated on a centrifuge platform is an attractive method for loading samples into separate testing chambers.27 A lab on a disc allows sequential sample tests through accurately designed microfabricated channels. Under the control of the centrifugal force induced by rotational acceleration, the system guides the direction of the fluid in the channel and the timing of the reaction. In addition, the small-scale capillary forces passively control the microfluidic flow.28 Recently, a fidget spinner introduced a new method for concentrating pathogens for AST that shows a centrifuge platform having broad applications including AST.29

Here, we introduce a capillary and centrifugal-based AST (C2-AST) system for quick loading of the sample and culture media. The C2-AST also includes a microscopic cell-growth detection protocol. The microfluidic channels realize a robust and accurate loading and culturing system. The loading performance of the C2-AST chip was tested in a systematic analysis. The accuracy of the system was then evaluated in comparison tests with the standard method. Tests were performed on standard strains in the presence of several antibiotics. This system can potentially reduce the burden of sample-loading and drug-testing in general cell-culture systems. The loading protocol does not require a complex pipetting system and it simplifies the overall form factor of the automated system for clinical applications. The reduced pipetting procedure will also reduce the chance of contamination of samples. The small microfluidic based testing chamber is favorable to high-throughput testing.


Chip structure design

The C2-AST chip (see Fig. 1) is designed for rapid, single-pipetting testing of antimicrobial susceptibilities to several antibiotics and their dilutions. The C2-AST chip is 80 mm in diameter and 4.2 mm thick, and its circular shape with a rotational symmetry is optimal for its purpose. The C2-AST system mixes two different fluids at different times by sequential introduction through two separate paths, one placed at either side of the chip. The two chamber inlets are also separated, and placed on the top surface of the chip for easy fluid injection.
image file: d0lc00753f-f1.tif
Fig. 1 (a) Schematic of the C2-AST chip components; (b) photograph of the injection-molded plastic C2-AST chip; (c) schematics and photographs of reagent loading and positioning; (d) schematics of dried antibiotics, agarose with bacteria, culture media, and time-lapsed microscope images of the samples in the wells.

Capillary forces along the bottom channels drive the movement of the first fluid through a length of 22 mm. At this stage, the channel is 600 μm wide and 400 μm high. The fluid then enters a vertical hole of 500 μm diameter and 1.5 mm length. The vertical hole and the second vertical hole are separated but linked at the bottom of the incubation cell. As the second fluid is injected at a much higher volume than the first fluid, its path should run along the top surface. The top channels are 1.2 mm wide, 2.4 mm high, and 17.8 mm long, and each channel stores approximately 45 μl of the second fluid.

The ends of one paired top and the bottom channel are connected to an incubation cell, which functions as an independent diagnostic laboratory. In each incubation cell, the second fluid is introduced after the gelation of the first fluid, forming an interface which must be located near the imaging plane of the first (gelated) fluid. One incubation cell is composed of an imaging well, a reservoir well, and other components. The imaging plane appears in the imaging well filled with the first fluid, which flows through the bottom path. The reservoir well contains a pre-dried antimicrobial reagent, which dissolves easily in the second fluid in a few minutes. The second fluid completely fills the reservoir well and spills into the remainder of the incubation cell. To close the channels of each path, two films with inlet and outlet holes are attached at both sides of the chip. The holes are aligned with a film-attaching jig for ease and efficiency.

Operational comparison

The system is injected with two different liquids: agarose and culture media. The liquid-state agarose mixed with a bacterial solution is injected into the bottom channel through the inlet. After falling into the bottom channel, the mixture is driven by capillary forces to the end of the bottom channel, which is linked to both vertical holes through outlets. The vertical capillary force draws the liquid from the bottom to the top of the chip. As the withdrawn liquid fills the imaging well, its temperature drops and gelation begins building a three-dimensional cell trapping matrix.

The culture media are pipetted into the top inlet hole, through which they fully fill the top channels under capillary forces. The excess media are removed by pulling back from the top buffer space. While pulling the excess liquid back, the beginning section of the top channels holds the liquid tightly, ensuring an equal amount of liquid in each channel. Finally, the chip is rotated at 1200 rpm, and the second liquid is introduced to the incubating cells. The second liquid, which is intended to simultaneously supply the microorganisms in the gelated matrix with nutrients and antimicrobials, completely covers the gelated first liquid.

Table 1 compares the estimated times of each step of the AST operation in the proposed and conventional methods. The loading processes of the first and second reagents in the C2-AST system are especially critical, as they convert the repeated time-consuming process into a single simple pipetting process. Broth microdilution (BMD) was chosen as the conventional method for it was considered as one of the gold-standard method. The BMD method is usually done with a 96-well rounded bottom plate which has individual cells in which individual pipetting is performed; therefore, it requires 48 times more pipetting than the C2-AST system does. The actual duration or workload difference was 12–18 times because the injection volume of the C2-AST is much larger, and the multi-dispensing function of electric pipettes is usually used for the BMD method. The rotation process was implemented only in the C2-AST system. The preparation time of the whole sequence in the C2-AST system was 25 s, approximately 8 times faster than in the conventional method (210 s).

Table 1 Comparison of sample preparation times of C2-AST and conventional AST
Operating time C2-AST (proposed method) Broth microdilution (conventional method)
Equipment C2-AST chip with a manual pipette 96-Well rounded bottom plate with a multi-dispensing pipette
First reagent loading (s) 5 90
Second reagent loading (s) 10 120
Rotation (s) 10
Total (s) 25 210

Investigating the agarose injection speed

The flows (especially the water flows) in sub-millimeter sized channels are mainly affected by inertia and the surface tension of the fluid. Capillary force is considered as an ideal power source in many microfluidic systems. Therefore, we introduced capillary force as the primary power source of the agarose injection sequence. When a small volume of liquid is injected, capillary action yields a mild, well-controlled, and structure-dependent flow. Despite its advantages, the motive power provided by the capillary force is sensitive to the environmental conditions, such as the fluid characteristics, substrate material types, surface treatments, and temperature. To ensure an appropriate experimental environment for the C2-AST system, the agarose injection characteristics were determined under various preparation conditions. The smoothness of the injection process can be simply quantified by the filling time. As the agarose filling is primarily driven by the capillary force between the solution and the channel surfaces, the factors with probable relatedness to capillary forces were selected for testing. After investigating the filling time of agarose loading under the selected characteristics, we decided the suitable preparation conditions (especially for maximizing the loading speed) of C2-AST. As our agarose solution possesses the unique properties of a gelling reagent, its physical properties were also factored into the analysis. The temperature of the agarose solution was maintained at 45 °C before the experiment, which was decreased to approximately 42 °C when the solution was injected into the chip. The solution went a few seconds without gelling parts, as the agarose solution can remain in the liquid phase before the temperature of the agarose solution decreased below 35 °C.30

The concentration and temperature of the agarose solution obviously affected the agarose filling time. The concentration was the dominant changeable variable, as the temperature cannot be altered far beyond the range 35–45 °C. At these temperatures, the solution remains in the liquid phase without imparting a critical heat shock to the bacterial sample. However, optimizing the temperature of the chip might additionally improve the circumstances of the liquid loading by slowing the cooling speed of the liquid. Fig. 2(a) shows the liquid filling times in the systems with different agarose concentrations and C2-AST chip temperatures. The results were clearly related to the agarose concentration but showed no correlation with the chip temperature. Theoretically, the chip temperature should affect the movement of the agarose, but its effects were apparently concealed by experimental error. In contrast, the agarose concentration dominantly affected the liquid loading, although the results at 0.375% and 0.5% were quite similar. The chip material and the surface modification method also affected the filling time, confirming that the capillary force provided the main motive power. Changing the intensity of the vacuum plasma treatment applied to the C2-AST chip altered the surface energy of the chip substrate to different extents. Increasing the treatment intensity decreased the filling time, and (by implication) increased the loading speed of the liquid (see Fig. 2(b)). This result also suggests that the highest treatment conditions did not saturate the hydrophilic surface. The result after the treatment at 100 W for 30 seconds deviated from the above trend, probably because an error in the agarose preparation process affected the physical properties of agarose. At the same treatment intensity, the hydrophilicity and original surface energies depended on the materials used. General-purpose polystyrene provided the most powerful capillary force under every plasma condition, whereas polycarbonate provided the weakest force. The contact-angle differences before and after the treatment also depended on the material (see contact angle measurements in Fig. S3). As shown in Fig. 2(c), the filling time increased with agarose concentrations above 0.5%, similar to that in Fig. 2(b). Finally, we selected an agarose concentration of 0.5%, a surface treatment of 200 W for two minutes, and a chip material of general-purpose polystyrene. The 0.5% agarose condition yielded firmer bacterial colonies than the 0.375% agarose condition, and the 200 W plasma treatment for two minutes was near the equipment limit in the manufacturer's guidelines. The only thermal process was maintaining the agarose solution at 45 °C before loading.

image file: d0lc00753f-f2.tif
Fig. 2 Loading properties of the C2-AST system: required agarose filling time into the bottom channel at various (a) agarose concentrations and chip temperatures, (b) surface treatment intensities and chip materials, and (c) agarose concentrations and chip materials. (d) Number of fully transferred wells from the top channel versus rotation speed of the chip (GPPS: general purpose polystyrene; K-resin: styrene–butadiene copolymer; PC: polycarbonate).

Investigating the medium-loading conditions

The medium filling proceeds in the following sequence: medium injection, medium fixing in the top channel, chip rotation, and medium filling in the incubation cell. The capillary valve at the end of the top channel obstructs the advance of the medium, but must be collapsed when the centrifugal force of the medium exceeds a certain level. The valve may not pass the entire volume of the entering medium at a certain rotation speed. The pressure working at the capillary valve is gradually decreased as the volume in the channel reduces; consequently, the final volume remaining in the channel decreases with increasing rotation speed. After a successful injection, the volume remaining in the channel should be less than 10%. Here, we investigated the relationship between the rotation speed and the number of successful injections.

The medium loading was influenced by the capillary force as expected; moreover, the material and surface-treatment conditions determined in the agarose loading investigation were sufficient to complete the first step of medium loading, which (as mentioned above) was powered mainly by the capillary force. The medium injection ceased when the medium reached the capillary valve located between the channel and the incubation cell. The second step bursts the capillary valve using the centrifugal force generated by rotating the chip. For this purpose, the rotational frequency should exceed a critical value, which should be minimized to reduce turbulent flows in the incubating cell. To determine the minimum required rotational speed, we increased the rotational speed from 0 to 1200 rpm, and counted the number of wells in which the liquid was successfully transferred into the incubating cell from the channel. Because the densities and viscosities vary in biological applications, these tests were performed using water and glycerol solutions in addition to the culture medium.

As shown in Fig. 2(d), all fluids were successfully loaded at 1200 rpm, but the results widely differed among the fluid compositions at lower speeds. Water made the highest contact angle with the hydrophilic treated polystyrene surface, and it was the most easily loaded fluid. The 50 wt% glycerol aqueous solution was the least easily loaded fluid at any given rotation speed. These two sequential loading steps enable precise volume control of a simple and simultaneous medium loading, thereby shortening the time of the reagent loading sequence.

Rhodamine diffusion test

Ideally, all incubating cells are completely separated, but in practice, the C2-AST system is physically imperfect because the bottom channel is filled with the agarose matrix, which connects all the cells at the middle of the chip. This situation may cause cell-to-cell contamination of the antibiotic ingredients. To quantify the extent of this contamination, the diffusion characteristics of the antibiotics were mimicked by those of a fluorescent dye, rhodamine B (83689-1G, Sigma Aldrich, Merck). Rhodamine B has a similar molar mass (472.02) to most of the antibiotics used in the C2-AST system, so it is considered as a representative example. The rhodamine solution (prepared at 1 mg ml−1) was loaded at 5 μl into the antibiotic well of each incubation cell. After the medium injection, the rhodamine solution was diluted to 100 μg ml−1, and was allowed to slowly permeate the agarose-filled channel. Time-lapse fluorescence imaging was performed at hourly intervals over a total acquisition time of 10 hours. The time-lapse images (Fig. 3) showed that the rhodamine diffused further into the agarose over time, but after 10 hours of diffusion, the penetration distance from the cell remained below 10 mm, which is much shorter than the length of the agarose channel.
image file: d0lc00753f-f3.tif
Fig. 3 Time-lapse bright field and fluorescence images of the rhodamine solution slowly diffusing into agarose prepared at different concentrations: (a) 0.375% and (b) 0.5% (scale bar = 2 mm).

Validation of C2-AST with Escherichia coli and Staphylococcus aureus

To demonstrate whether the accuracy of the C2-AST system meets that of conventional AST methods, we evaluated whether the C2-AST system assesses the antibiotic susceptibilities of E. coli ATCC 25922 and S. aureus ATCC 29213. The 2018 Clinical and Laboratory Standards Institute (CLSI) lists the antibiotics and their concentrations in various strain–drug combinations. For E. coli, we selected ampicillin, amikacin, gentamicin, cefotaxime, cefepime, ceftazidime, imipenem, meropenem, and colistin. For S. aureus, we selected penicillin, oxacillin, tetracycline, clindamycin, erythromycin, vancomycin, levofloxacin, and rifampicin.

The experimental images were translated into normalized values by the same method applied in our previous work.12Fig. 4(a) shows a matrix of images showing the gradual changes in the E. coli colonies during 0 to 4 hours of exposure to gentamicin at different concentrations (control, 0.12, 0.25, and 0.50 μg ml−1). The acquired images were converted into binary images, revealing the colonies as white pixels against a black background (Fig. 4(b)). From the binary images, the areas of each bacterial colony were easily calculated. The average values of the area of the 20 biggest colonies before being normalized at different antibiotic concentrations are plotted in Fig. 4(c). Under the control (no antibiotic), 0.12, and 0.25 μg ml−1 conditions, the E. coli colonies grew over time, but no growth was detected under the 0.5 μg ml−1 condition. The minimum inhibitory concentration (MIC) of gentamicin for E. coli was thus determined as 0.5 μg ml−1. According to the CLSI guidelines, most MICs of the E. coli/gentamicin combination are statistically within the 0.25–1.00 μg ml−1 range. As the broth microdilution (BMD) protocol of CLSI is the gold standard of AST evaluation, the above result demonstrates the suitability of the C2-AST system.

image file: d0lc00753f-f4.tif
Fig. 4 Time-lapse data of the E. coli colonies in the agarose matrix subjected to gentamicin at various concentrations (0–0.5 μg ml−1): (a) raw images, (b) binary converted images, and (c) average colony area at various antibiotic concentrations before being normalized from the binary images (n = 20, the selected 20 biggest colonies at the same concentration).

The MIC results of all the antibiotics are listed in Tables 2 and 3. All the results were within the range suggested in the CLSI guidelines. The BMD results, including those of the above-described E. coli/gentamicin combination, also appeared within the suitable range, although some of them showed a two-fold difference from the C2-AST results. In this test, the testing chip contains 4–5 concentrations per each antibiotic (Fig. S5). To test broad spectrum antibiotics, the system can reduce the number of concentrations and increase the testing antibiotics. As the system is capable of testing 48 combinations of antibiotics and their concentrations, it can be applied for high-throughput analysis of AST (Fig. S6).

Table 2 MIC results of AST with E. coli ATCC 25922 (unit: μg ml−1)
Antibiotic QC range C2-AST BMD test
Amikacin 0.50–4.00 0.5 1
Cefotaxime 0.03–0.12 0.12 0.06
Meropenem 0.008–0.06 0.03 0.015
Gentamicin 0.25–1.00 0.5 0.25
Ampicillin 2.00–8.00 8 4
Cefepime 0.015–0.12 0.06 0.06
Imipenem 0.06–0.25 0.12 0.12
Colistin 0.25–2.00 1 0.5
Ceftazidime 0.06–0.50 0.25 0.25

Table 3 MIC results of AST with S. aureus ATCC 29213 (unit: μg ml−1)
Antibiotic QC range C2-AST BMD
a The result of S. aureus/tetracycline was obtained after 6 hours of incubation. QC: quality control, BMD: broth microdilution.
Penicillin 0.25–2.00 0.25 0.5
Oxacillin 0.12–0.50 0.5 0.25
Levofloxacin 0.06–0.50 0.25 0.25
Vancomycin 0.50–2.00 0.5 1
Clindamycin 0.06–0.25 0.06 0.12
Tetracycline 0.12–1.00 0.25a 0.5
Rifampicin 0.004–0.015 0.004 0.008
Erythromycin 0.25–1.00 0.25 0.5

The C2-AST tended to give higher antibiotic resistances than the BMD protocol in the E. coli test; the opposite tendency was observed in the S. aureus test. This discordance might be caused by the different culturing methods in the C2-AST system and BMD method; in the former system, the bacteria grow and form colonies within the solid substrate, whereas in the BMD method, they grow in a liquid culture without forming colonies. The growth and antibiotic-resistance characteristics apparently differ between colonies formed from fixed cells in a solid substrate (agarose in this case) and freely growing cells. In the liquid medium, the bacteria can freely access the nutrients and metabolize at full speed, as they are homogeneously distributed before saturation. In agarose, only the outer cells of the colonies can access the nutrients; the interior cells slowly die, so the total colony mass multiplies in a linear fashion. Bacterial clumps can effectively resist chemical assaults by minimizing their surface-to-volume ratio. In addition, the agarose matrix physically (and occasionally chemically) changes the action mechanism and/or diffusion of the nutrients and drugs.

Finally, and perhaps most importantly, the culturing times before determination differ in the two methods. Some bacterial/drug combinations are strongly time-dependent, exhibiting fake, late, or slow growth, and yielding false positive or false negative results at certain antibiotic concentrations. The result of the slow-growing S. aureus/tetracycline combination was settled only in the 6-hourly data.


This study introduces a microfluidic system that realizes high-throughput and quick sample loading by exploiting the capillary and centrifugal forces in the system. The liquid flow is guided into the testing chambers through narrow, hydrophilic microfluidic channels. The culture-media flow is controlled with a capillary valve. When the valve opens under the centrifugal force, the culture media is guided into the testing chamber. Forty-eight testing chambers are simultaneously filled during one pipetting and rotation of the C2-AST system. The loading characteristics of the sample into the chip were revealed in detailed experimental studies. Finally, the MICs of two standard strains of clinically important antibiotics were determined in the system, and were compared with those of a conventional AST. Our system achieved rapid and accurate AST through the microscopic imaging process, and can potentially meet the requirements of many cell cultures and drug testing systems.

Methods and materials

Chip design and manufacturing

The three-dimensional (3D) features of the plastic C2-AST were designed using the chip 3D CAD software (SolidWorks2018, Dassault Systemes, Vélizy-Villacoublay, France). As the C2-AST chip is designed to be disposable, its manufacture requires mass production equipment (Victory120, ENGEL, Schwertberg, Austria) used in injection molding. A quick delivery mold was made of aluminium (AMS4045L-T651, Constellium, Paris, France). The C2-AST chips were fabricated from three kinds of plastic materials: general-purpose polystyrene (GPPS) (25SP(1), LG Chem., Seoul, Republic of Korea), styrene–butadiene copolymer (K-resin) (KR01, INEOS Styrolution, Frankfurt, Germany), and polypropylene copolymer (PC) (J-560M, Lotte Chem., Seoul, Republic of Korea). The C2-AST chip also included two polyethylene terephthalate (PET) films with adhesives at one side. The films were produced by the extrusion molding method and were cut by a blanking and punching process.

Comparison of the proposed and conventional methods

Agarose (0.5 wt% in aqueous solution) (AGA-101 500 mg, BIOPURE, Massachusetts, USA) and the Mueller–Hinton broth (212322, 500 g, BD, New Jersey, USA) were prepared, and the agarose was maintained at 40 °C. The conventional method for performance validation was prepared as follows, assuming that the conventional and C2-AST methods would be operated on the same equipment. The gelation reagent was loaded with a 1.0 ml multi-dispensing pipette (Multipette E3x, Eppendorf, Hamburg, Germany) into each well of a 96-well AST chip; the second liquid was injected in the same manner. The time consumptions per sample of the two methods were then compared.

Injection of different concentrations of the agarose solution into the chips made from different materials

The agarose concentration in the solution was varied as 0.375, 0.5, and 0.7 wt%, and the material of the C2-AST chip was varied as GPPS, K-resin, and PC. At each concentration, the agarose solution was maintained in the liquid phase in a 45 °C heat block (TF1, BLE – Best of Lab Equipment (ForBioKorea, Seoul, Republic of Korea)). The three chips were surface-treated at 200 W for two minutes using vacuum plasma equipment (COVANCE-1MP, Femto Science, Seoul, Republic of Korea), then maintained at 20 °C prior to loading. The agarose solution was directly transported from the heat block, then injected into the bottom channel through the inlet at approximately 42–43 °C. The time between the liquid entering the bottom channel and reaching the end of the bottom channel was measured with a 60 fps recording device and analyzed.

Injection of different concentrations of agarose solutions into the chips held at various temperatures

The temperature of the agarose solution dropped as the solution passed through the channel; accordingly, the influence of the plastic-chip temperature on the agarose filling time was investigated. Agarose solutions prepared at 0.375, 0.5, and 0.7 wt% were injected into a C2-AST chip made from K-resin and surface-treated at 200 W for two minutes. The chips were maintained at 25 °C, 35 °C, or 45 °C on a hotplate (MSH-30D, Daihan Scientific, Seoul, Republic of Korea). The agarose solution was injected into the chips on the hotplate. The injection and analysis processes were those described above.

Agarose injection into the chips made from different materials and subjected to different surface treatments

The hydrophilicity of the channel surface was investigated in the chips subjected to different surface treatments. The application of 200 W power for two minutes was regarded as the saturated condition, and the other conditions were set as follows: 200 W power for one minute, 100 W power for one minute, 100 W power for 30 seconds, and 50 W power for 30 seconds. The agarose concentration was 0.7 wt% and the C2-AST chips were fabricated from three different materials: GPPS, K-resin, and PC. The injection process and analysis were as described above.

Culture medium loading and rotation

The bottom side of the GPPS C2-AST chip was treated at 200 W for two minutes, and both sides of the chip were covered with PET films. The first fluid (agarose solution) was injected and solidified before the second fluid was injected.

The spinning can be done with any rotating devices such as spin coaters or centrifuges. The C2-AST chip was preferred to be pinned by the rotating axis, which runs through the entrance of the chip. A spin coater was preferred for its flexible speed programming over a centrifuge, though we used a centrifuge for confirming the concept of rotation as shown in Fig. S3(c). In this experiment, the spinning was applied with a spin coater (SF-100ND, RHABDOS, Seoul, Republic of Korea) with setting values of max speed and duration at the max speed. The duration was set as 10 s while the max speed was stepwise-increased from 0 to 1200 rpm. The spin coater has a vacuum chuck, which can hold the C2-AST chip while spinning. After each rotation, the remaining culture media were imaged for further analysis.

Rhodamine diffusion

Two C2-AST chips were subjected to 200 W vacuum plasma treatment for two minutes, sealed with top and bottom films, and maintained at 20 °C. Agarose solution (0.375% or 0.5%) was injected into each chip and cooled until it gelated. Rhodamine B solution (100 μg ml−1) entered the incubation cell by injection (50 μl) and rotation. Time-lapse imaging was performed at 10-minute intervals over a period of 10 hours. Images were acquired with an automatic microscopy system (CELENA X, Logos Biosystems, Seoul, Republic of Korea) equipped with a monochrome CMOS (5.86 μm per pixel, 1.92 MP), a red fluorescence protein filter, and a 2× objective lens.

AST using the C2-AST chip

Typical strains of Escherichia coli (ATCC 25922) and Staphylococcus aureus (ATCC 29213) were obtained from Kwik-Stik (Cat no. 0335P, 0365P, Microbiologics, Minnesota, USA). Antibiotics for E. coli were ampicillin (a9518), amikacin (a1774), gentamicin (g1264), cefotaxime (c7039), cefepime (a3737), ceftazidime (c3809), imipenem (i0160), meropenem (m2574), and colistin (c4461). Antibiotics for S. aureus were penicillin (p3032), oxacillin (28221), tetracycline (t7660), clindamycin (c5269), erythromycin (e5389), vancomycin (v2002), levofloxacin (28266), and rifampicin (r3501). All antibiotics were obtained from Merck (Sigma Aldrich or MilliporeSigma, Massachusetts, USA). The antibiotics were prepared at the highest concentration for the test, and were serially diluted two-fold with distilled water. 5 μl of antibiotic diluent was loaded into the reservoir wells and the testing chip was covered with a PET film.

Each cultured bacterial strain was diluted to 0.5 McFarland using a nephelometer (DensiCHEK Plus, Biomerieux, Marcy-l'Étoile, France), mixed with liquid-state agarose prepared at ∼40 °C, and injected into the bottom of the chip through the inlet. The bacterial sample (100 μl, containing 1.5 × 108 CFU ml−1) mixed with liquid state agarose (3 ml) contained 5 × 106 CFU mL−1. As 5 μl of the mixture and 50 μl of the culture medium were filled to each chamber, the final inoculation concentration was 5 × 105 CFU mL−1.

The C2-AST chip was incubated at 35 °C and time-lapse images were obtained after four hours using a microscope (Eclipse Ti, Nikon, Tokyo, Japan). All images, taken at 1-hour intervals, were processed by a binary conversion algorithm. To determine whether an antibiotic at a specific concentration blocked the colony growth, the areas occupied by colonies were calculated, and assessed whether the growth in the antibiotic wells (as a ratio of growth in the control medium) exceeded the given threshold. Here, the threshold growth ratio was set to 0.15 for all the antibiotics.

Conflicts of interest

There are no conflicts to declare.


This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (NRF-2019R1A2C1084419).


  1. B. Aslam, W. Wang, M. I. Arshad, M. Khurshid, S. Muzammil, M. H. Rasool, M. A. Nisar, R. F. Alvi, M. A. Aslam, M. U. Qamar, M. K. F. Salamat and Z. Baloch, Infect. Drug Resist., 2018, 11, 1645–1658 CrossRef CAS.
  2. A. C. Uhlemann, J. Dordel, J. R. Knox, K. E. Raven, J. Parkhill, M. T. Holden, S. J. Peacock and F. D. Lowy, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 6738–6743 CrossRef CAS.
  3. N. A. Turner, B. K. Sharma-Kuinkel, S. A. Maskarinec, E. M. Eichenberger, P. P. Shah, M. Carugati, T. L. Holland and V. G. Fowler, Jr., Nat. Rev. Microbiol., 2019, 17, 203–218 CrossRef CAS.
  4. E. Tacconelli, E. Carrara, A. Savoldi, S. Harbarth, M. Mendelson, D. L. Monnet, C. Pulcini, G. Kahlmeter, J. Kluytmans, Y. Carmeli, M. Ouellette, K. Outterson, J. Patel, M. Cavaleri, E. M. Cox, C. R. Houchens, M. L. Grayson, P. Hansen, N. Singh, U. Theuretzbacher, N. Magrini and WHO Pathogens Priority List Working Group, Lancet Infect. Dis., 2018, 18, 318–327 CrossRef.
  5. G. V. Doern, R. Vautour, M. Gaudet and B. Levy, J. Clin. Microbiol., 1994, 32, 1757–1762 CrossRef CAS.
  6. J. Barenfanger, C. Drake and G. Kacich, J. Clin. Microbiol., 1999, 37, 1415–1418 CrossRef CAS.
  7. K. Syal, M. Mo, H. Yu, R. Iriya, W. Jing, S. Guodong, S. Wang, T. E. Grys, S. E. Haydel and N. Tao, Theranostics, 2017, 7, 1795–1805 CrossRef CAS.
  8. J. Garnacho-Montero, T. Aldabo-Pallas, C. Garnacho-Montero, A. Cayuela, R. Jimenez, S. Barroso and C. Ortiz-Leyba, Crit. Care, 2006, 10, R111 CrossRef.
  9. A. van Belkum, C. A. D. Burnham, J. W. A. Rossen, F. Mallard, O. Rochas and W. M. Dunne, Nat. Rev. Microbiol., 2020, 18, 299–311 CrossRef CAS.
  10. C. Malmberg, P. K. Yuen, J. Spaak, O. Cars, T. Tangden and P. Lagerback, PLoS One, 2016, 11, e0167356 CrossRef.
  11. J. Choi, J. Yoo, M. Lee, E. G. Kim, J. S. Lee, S. Lee, S. Joo, S. H. Song, E. C. Kim, J. C. Lee, H. C. Kim, Y. G. Jung and S. Kwon, Sci. Transl. Med., 2014, 6, 267ra174 CrossRef.
  12. J. Choi, Y. G. Jung, J. Kim, S. Kim, Y. Jung, H. Na and S. Kwon, Lab Chip, 2013, 13, 280–287 RSC.
  13. J. Choi, H. Y. Jeong, G. Y. Lee, S. Han, S. Han, B. Jin, T. Lim, S. Kim, D. Y. Kim, H. C. Kim, E. C. Kim, S. H. Song, T. S. Kim and S. Kwon, Sci. Rep., 2017, 7, 1148 CrossRef.
  14. O. Baltekin, A. Boucharin, E. Tano, D. I. Andersson and J. Elf, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 9170–9175 CrossRef CAS.
  15. H. Sun, C. W. Chan, Y. Wang, X. Yao, X. Mu, X. Lu, J. Zhou, Z. Cai and K. Ren, Lab Chip, 2019, 19, 2915–2924 RSC.
  16. G. Pitruzzello, S. Thorpe, S. Johnson, A. Evans, H. Gadelha and T. F. Krauss, Lab Chip, 2019, 19, 1417–1426 RSC.
  17. T. A. Cowger, Y. P. Yang, D. E. Rink, T. Todd, H. M. Chen, Y. Shen, Y. Yan and J. Xie, Bioconjugate Chem., 2017, 28, 890–896 CrossRef CAS.
  18. M. Kittel, P. Findeisen, B. Ghebremedhin, T. Miethke, A. Grundt, P. Ahmad-Nejad and M. Neumaier, Clin. Chem. Lab. Med., 2019, 57, 1271–1279 CAS.
  19. B. Ibarlucea, T. Rim, C. K. Baek, J. de Visser, L. Baraban and G. Cuniberti, Lab Chip, 2017, 17, 4283–4293 RSC.
  20. J. Q. Boedicker, L. Li, T. R. Kline and R. F. Ismagilov, Lab Chip, 2008, 8, 1265–1272 RSC.
  21. R. T. Hayden, L. K. Clinton, C. Hewitt, T. Koyamatsu, Y. L. Sun, G. Jamison, R. Perkins, L. Tang, S. Pounds and M. J. Bankowski, J. Clin. Microbiol., 2016, 54, 2701–2706 CrossRef CAS.
  22. Y. Yang, K. Gupta and K. L. Ekinci, Proc. Natl. Acad. Sci. U. S. A., 2020, 117, 10639–10644 CrossRef.
  23. K. Yang, H. Z. Li, X. Zhu, J. Q. Su, B. Ren, Y. G. Zhu and L. Cui, Anal. Chem., 2019, 91, 6296–6303 CrossRef CAS.
  24. T. J. Abram, H. Cherukury, C. Y. Ou, T. Vu, M. Toledano, Y. Y. Li, J. T. Grunwald, M. N. Toosky, D. F. Tifrea, A. Slepenkin, J. Chong, L. S. Kong, D. V. Del Pozo, K. T. La, L. Labanieh, J. Zimak, B. Shen, S. S. Huang, E. Gratton, E. M. Peterson and W. A. Zhao, Lab Chip, 2020, 20, 477–489 RSC.
  25. H. Etayash, M. F. Khan, K. Kaur and T. Thundat, Nat. Commun., 2016, 7, 12947 CrossRef CAS.
  26. M. Hernández-Durán, L. E. López-Jácome, C. A. Colín-Castro, G. Cerón-González, S. Ortega-Peña, E. S. Vanegas-Rodríguez, J. A. Mondragón-Eguiluz and R. Franco-Cendejas, Investigación en discapacidad, 2017, 6, 105–114 Search PubMed.
  27. L. X. Kong, A. Perebikovsky, J. Moebius, L. Kulinsky and M. Madou, J. Lab. Autom., 2016, 21, 323–355 CrossRef.
  28. A. Olanrewaju, M. Beaugrand, M. Yafia and D. Juncker, Lab Chip, 2018, 18, 2323–2347 RSC.
  29. I. Michael, D. Kim, O. Gulenko, S. Kumar, S. Kumar, J. Clara, D. Y. Ki, J. Park, H. Y. Jeong, T. S. Kim, S. Kwon and Y. K. Cho, Nat. Biomed. Eng., 2020, 4, 591–600 CrossRef CAS.
  30. A. S. Ogrenci, O. Pekcan, S. Kara and A. H. Bilge, J. Macromol. Sci., Part B: Phys., 2018, 57, 364–376 CrossRef CAS.


Electronic supplementary information (ESI) available. See DOI: 10.1039/d0lc00753f

This journal is © The Royal Society of Chemistry 2020