Droplet-based valveless microfluidic system for phage-display screening against spheroids

Abstract

In this study, we proposed a droplet-based valveless microfluidic system that has the necessary functions to perform the binding, washing, eluting, and collecting processes of phage-display screening against spheroids, which can be expected to present a similar repertoire and number of membrane proteins as in vivo. Although spheroids have much larger sizes than single cells, spheroids are difficult to manipulate through manual operation. The proposed microfluidic system actively controls the position and velocity of droplets using a camera, three air pumps, and three liquid pumps to perform the processes for phage-display screening. The cross section of the microchannel is large in width and height for the passage of spheroids. Valves that can close such a large cross-sectional microchannel are not readily available. Thus, we proposed valveless flow control using liquid pumps. In addition, the proposed microfluidic system involves complex flow channels with airflow subchannels to perform phage-display screening. For washing, nonspecific-binding phages remaining in the flow channels must be minimized. The proposed microfluidic system can perform selective blocking and flush washing. Selective blocking can prevent the airflow channels from becoming hydrophilic with blocking liquid, and flush washing can flush phages remaining in the flow channel. We experimentally verified the functions of the developed microfluidic device based on the proposed system.

I. INTRODUCTION

Phage display can display proteins or peptides on the surface of phages by gene recombination, and phage-display screening is a robust method for identifying specific binding proteins or peptides of target molecules^1,2 from artificially produced libraries that include millions of them with different amino acid sequences. Phage-display screening involves binding, washing, eluting/collecting, amplifying, and identifying processes. The biding process incubates phage library with target molecules to allow their specific binding; the washing process washes target molecules to remove unbound and nonspecific-binding phages; the eluting/collecting process detaches the binding phages from the target molecules with an eluent and collects the phages; the amplifying process amplifies the collected phages by infecting them with bacteria, such as Escherichia coli. The amplified phages are used as libraries in the next round of processes, which are repeated 3–6 times. The amplified phages in the last round are genetically analyzed for specific binding peptide identification.

The conformation of target molecules in phage-display screening should be the same as in vivo. The characteristics of target molecules may change in different three-dimensional (3D) conformations, even when the amino acid sequences are the same. The conformations of membrane proteins are not maintained without cell surface membranes.^3–5 To avoid this problem, cell-based phage-display screening is promising.⁶ For efficient execution of the processes of cell-based phage-display screening, a precise, stable, and uniform flow control toward cells is required. Microfluidic flow control is suitable; moreover, microfluidic devices for phage-display screening against monolayer cultured and suspended cells have been proposed.^7–9

However, the repertoire and numbers of membrane proteins expressed on suspended single cells and monolayer-cultured cells differ from in vivo tissues. Therefore, 3D cultured cells, such as spheroids,^10,11 which can be expected to present a similar repertoire and number of membrane proteins as in vivo, are preferable. Although spheroids have much larger sizes than single cells, spheroids are difficult to manipulate through manual operation and many studies have employed a microfluidic device for the spheroid formation and culture.¹² In this study, we propose a microfluidic system that has the necessary functions to perform the binding, washing, eluting, and collecting processes of phage-display screening against spheroids.^13–15 The system is based on droplet flow control^16–18 because droplet-based flow control can easily perform spheroid transportation to the succeeding process in phage-display screening. In addition, the internal flow of a droplet during transportation is advantageous for binding and washing phages.

Many droplet-based microfluidic devices employ passive control of droplets¹⁹ by pushing droplets in built-in microfluidic components in order. However, passive control is not always stable and has limitations. Meanwhile, active droplet control is suitable for complex operations.²⁰ The active droplet control in a simple microchannel²¹ realizes the forming, splitting, merging, mixing, and sorting of droplets with approximately ±10% error. Although active droplet control requires sensors and actuators, their cost has been reduced recently.

The proposed microfluidic system employs a camera, three air pumps, and three liquid pumps. We show that it can perform the functions of phage-display screening processes by actively controlling the position and velocity of a droplet with visual feedback. The cross section of microchannels is large in width and height for the passage of spheroids. Valves that can close such large cross-sectional microchannels are not readily available, except for a recently developed one.²² Valve function is necessary to implement a branch flow channel for the eluting/collecting process. We propose a valveless flow control using the liquid pumps.

For washing, nonspecific-binding phages remaining in the flow channels must be minimized. The proposed microfluidic system involves complex flow channels with airflow subchannels to perform phage-display screening. The proposed microfluidic system can perform selective blocking that can prevent the airflow subchannels from being hydrophilic with the blocking liquid as well as flush washing that can flush phages remaining in the flow channel. We experimentally verify the functions of the developed microfluidic device based on the proposed system, which is a necessary step toward phage-display screening against spheroids.

II. METHOD

A. Microfluidic system overview

Figure 1 shows the proposed microfluidic system comprising two manual syringes (S₁ and S₂), three liquid pumps (L₁, L₂, and L₃), and three air pumps (A₁, A₂, and A₃). DE is the drainage/exhaust region. The design values of the height and width of the flow channel are 350 and 500 μm, respectively, assuming that a spheroid of approximately 300-μm diameter is introduced. Several flow subchannels are connected to DE from both sides to drain liquid with L₂ and exhaust air with A₃ without sucking a spheroid. The height and width of the flow subchannels are 50 and 200 μm, respectively. The length of the DE region is 1250 μm. EC is the eluting/collecting region. B is the branch flow channel, and J is the junction of the main and branch flow channels.

FIG. 1.

Proposed microfluidic system. S₁ and S₂: manual syringes, A₁, A₂, and A₃: air pumps, L₁, L₂, and L₃: liquid pumps, DE: drainage/exhaust region, EC: eluting/collecting region, J: junction of the main and branch flow channels, B: branch flow channel.

To ensure that no spheroids are sucked from B, L₃ is connected to B through several flow subchannels. The developed microfluidic device had two flow channels with different heights of 50 and 350 μm, and its fabrication process employed two-step exposure (see the supplementary material for the fabrication details).

B. System hardware

Figure 2 shows the hardware of the system. Syringe pumps are applied as liquid pumps L₁, L₂, and L₃. Custom-made syringe pumps were assembled using a commercial ball-screw device (one-axis actuator, LS, Misumi Corporation, stroke: 140 mm, lead: 2 mm) and stepping motor (ST-42BYH1004, step: 400) microblower (Murata Inc. MZB1001T02) were used as A₁, A₂, and A₃. The microblower is a compact and inexpensive air pump, preventing the increase in the device size and cost. It drives air using pulse width modulation (PWM)-controlled ultrasonic ceramic vibration. Owing to the PWM control, the output airflow was not constant. To solve this problem, we increased the length of the silicon tube connecting the microblower and microfluidic device to 900 mm.

FIG. 2.

System hardware.¹⁴ Arduinos 1 and 2 control the microblowers and syringe pumps, respectively. Some electric cables connecting the arduinos are omitted. Manual syringes S₁ and S₂ are also omitted.

An industrial camera (DFK33UP1300, Imaging Source) with background lighting was employed to detect the droplet and measure its position and velocity. The resolution (480 × 640 pixels) was selected by considering the flame rate of 30 frame/s. One pixel corresponded to approximately 62.5 × 62.5 μm. This low resolution was permitted because functions were separately arranged in the device and a large error of 200 μm was permitted for droplet position measurement. Arduinos 1 and 2 control the microblowers and syringe pumps, respectively. The two arduinos were connected to a PC, which processed the images taken by the camera with MATLAB (The Math Works, Inc.).

C. Branch control

In the binding and washing processes, the spheroid droplet must not enter B. Thus, L₃ introduces an eluate droplet in B and makes the air volume between the spheroid and eluate droplets small [Fig. 3(a)]. Actually, the spheroid droplet enters B partially. However, if the entered volume is little, it is brought out of the branch by the main volume of the spheroid droplet due to the liquid intermolecular force. The air spring constant is given by

$$K_a=\gamma A^{2}P/V$$

where γ is the ratio of specific heat, A is the area of the cross section of B, whereas P and V are the pressure and volume of air, respectively. When the adiabatic expansion is assumed, we have PV ^γ = const. and

$$K_a\propto \gamma A^2/V^{1+\gamma }$$

FIG. 3.

Branch control: (a) little air volume between two droplets and (b) large air volume; a part of droplet remains in B.

Therefore, K_a can be made large by making the air volume little, which can prevent the spheroid droplet from entering B.

D. Operation of the binding and washing processes

Figure 4 depicts the operation of the biding and washing processes comprising eight steps (EC is omitted). The eight steps in Fig. 4 are performed as follows.

FIG. 4.

The binding (steps 1 and 2) and washing processes (steps 3–8). On pumps are colored red. Air pump A_i (i = 2 and 3) with superscript * is on to prevent liquid from entering the flow channel connected to it. Step 1: introduction of a spheroid droplet from S₁, step 2: movement to DE via A₁, step 3: drainage from L₂, (A₁and A₂ are on to maintain the position of the spheroid droplet at DE.), step 4: flushing of washing liquid from L₁, step 5: introduction of a droplet of dilution solution from L₁, step 6: merging of the two droplets, step 7: reciprocation of the merged droplet for mixing (A₁ and A₂ are switched on alternatively. When A₁ is on, A₃ is also on.), and step 8: movement to DE via A₂ then return to step 3.

Binding Process:

• -Step 1 introduces a droplet containing a spheroid (spheroid droplet for short) and phages using syringe S₁. As this is done only once, a manual syringe is used.
• -Step 2 moves the droplet toward DE with the air pressure supplied by A_1.

Note: When a droplet moves in a microchannel, an internal circulating flow occurs and its velocity can be controlled by the droplet velocity.¹³ Step 2 moves a spheroid droplet with a relatively slow speed not to remove the specific binding phages on the spheroid. The internal flow is even expected to promote contact between specific binding phages and spheroid cells when its velocity is small, and step 2 can be extended arbitrarily by moving the droplet to the zigzag flow channel, reciprocating it there and moving it back to DE.

Washing process:

This process comprises subprocesses: drainage (step 3), flush (step 4), merge (steps 5 and 6), and mixture (step 7).

• -Step 3 drains the droplet liquid at DE by applying a negative pressure via L₂ until the droplet reaches a predetermined volume. In the first round, skip step 4 and go to step 5.
• -Step 4 flushes the flow channel with a washing liquid [phosphate-buffered saline (PBS)] to remove phages remaining in the flow channel. The washing liquid is introduced via L₁ and drained via L₂.
• -Step 5 also introduces the diluent solution (PBS) into the flow channel via L₁, which is then cut to form a droplet and moved to DE by applying air pressure via A₂. The droplet of PBS moves the spheroid droplet to the right due to the increase in the air pressure between them. The spheroid droplet stops on the right of DE because air exhausts from the flow subchannels. For the air exhaust, A₃ is set off.
• -Step 6 moves the spheroid droplet to the left for merging with the diluent solution droplet and moves the merged droplet to the zigzag flow channel via A₁.
• -Step 7 reciprocates the merged droplet to mix the liquid in it using the internal flow, via A₁ and A₂, alternatively. When A₁ moves the droplet with its air pressure, A₃ is switched on such that the air does not escape from the flow channel, whereas when A₂ moves the droplet, it is switched off. This step moves the droplet with a higher speed, which is expected to increase the speed of the internal flow and wash out the low-specific binding phages on the spheroid cells.
• -Step 8 moves the droplet to DE via A₂ and performs step 3. Then, steps 4–8 are repeated as many times as needed.

E. Operation of the eluting/collecting process

Figure 5 depicts the operation of the eluting/collecting process comprising five steps. This process uses L₃.

• -Step 1: The eluate is manually introduced with S₂ and is moved to the main channel through J under the air pressure applied by L₃. In this step, L₃ is used as an air pump.
• -Step 2: The eluate is cut with A_2.
• -Step 3: The eluate droplet and the spheroid droplet are merged with A₁ and A₂.
• -Step 4: The merged droplet is moved to J via A₁.
• -Step 5: The merged droplet is mixed in the zigzag flow channel in EC and collected under a negative pressure applied via L₃.

FIG. 5.

The eluting/collecting process. On pumps are colored red. Step 1: introduction of an elute solution from S₂ and movement to the main channel, step 2: cutting off the elute solution to form a droplet via A₂, step 3: merging of the two droplets, step 4: movement to J, step 5: collection of the merged droplet via L₃.

In the experiments in Sec. III, spheroids of U87 MG (the human glioblastoma U87MG cell line, ATCC, USA) were used and PBS with 5-mol/l NaCl was used for the elute droplet.

F. Visual feedback control

Diluent solutions, such as PBS, are transparent. To detect transparent droplets, we employed the background subtraction method (see the supplementary material). The measurement error of the droplet position was ±2 pixels, ±125 μm in length. Two visual control laws were implemented.

• (1)Velocity feedforward control with on-off position control:

  This control law moved a droplet through feedforward control via an air pump and turned it off when the droplet reached the predetermined position. This control law was used in steps 1–6 and 8 to move a droplet in the binding and washing processes.

• (2)Velocity feedback control: In the mixing (step 7) of the washing process, velocity control of a droplet is necessary; hence, the proportional control of the velocity was implemented. The droplet velocity was estimated from its increment during 30 frames, i.e., every 1 s, to reduce noise.

G. Blocking

Blocking treatment is required to prevent the attachment of nonspecific-binding phages in the flow channel. Introducing the blocking liquid into the flow channel, it entered the airflow subchannels and rendered them hydrophilic. This caused problems in droplet transport and the mixing of two droplets. To solve this problem, we prevented the entrance of the air channel from becoming hydrophilic by introducing a blocking liquid as a droplet and passing it in DE at a high flow velocity.

The proposed droplet blocking method is as follows.

• -Step 1: Approximately 3-μl blocking liquid (2% Perfect-Block, PB01, MobiTec, Germany) was introduced into the liquid flow channel; this formed a droplet.
• -Step 2: The droplet was reciprocated five times at a speed of approximately 6 mm/s in R1 of the flow channel in Fig. 6.
• -Step 3: The droplet was passed through DE at a high speed of 12 mm/s and moved into R2.
• -Step 4: Step 2 was repeated in R2.
• -Step 5: Step 3 was repeated, but the droplet was moved into R1.

FIG. 6.

Blocking regions R1 and R2.

Steps 2–5 were repeated for 30 min. Blocking treatment is also performed in EC separately in the same way.

The following experiment was performed to evaluate the droplet blocking method (excluding the blocking treatment in EC):

• (1)Three devices were prepared using the same mold.
• (2)Normal blocking processing was performed in the first device. This involved the introduction of the blocking liquid into the flow channel of the device entirely. The droplet blocking processing was performed in the second device. The third device was not processed.
• (3)The first and second devices were washed using PBS.
• (4)A nonspecific-binding phage (T7 phage constructed using the T7 Select 10-3b vector, Merck Millipore, USA) solution

(titer: 1.04 × 10¹⁰ pfu/ml) was introduced and reciprocated in the entire channel five times at a flow velocity of approximately 6 mm/s.

• (5)The channel was filled with the eluate (5-mol/l NaCl solution, Wako, Japan) and let stand for 30 min.
• (6)A 250-μl PBS was introduced, and the eluate was recovered.
• (7)Plaque assay was performed for the eluate, and the titer was measured (for materials for plaque assay, see the supplementary material).

III. EXPERIMENTAL RESULTS

A. Visual feedback control

Figure 7 shows the input–output relationship, where the input is the command value to the air pump, and the output is the average velocity of a droplet. As the air pressure applied to a droplet varies depending on the position of the flow channel, the average velocity was evaluated based on the average of the velocities in the four straight portions of the zigzag flow channel. From Fig. 7, the output average velocity was approximately proportional to the air pump input. Figure 8 depicts the results of the on-off position control. The target position was at the 466-th pixel, and the steady-state error was 2 pixels (125 μm in length), which is within the target error of 200 μm.

FIG. 7.

Relationship between the micropump input and output average velocity.

FIG. 8.

Result of on-off position control (1 pixel = 62.5 μm, 1 flame = 0.33 s).

Figure 9 displays the velocity feedback control results. The desired velocity was 1.06 pixels/frame (2 mm/s). The velocity was oscillatory but controlled within 2 ± 1 mm/s, which was sufficient because the capturing and washing processes required several steps of velocity differences, e.g., low (3 mm/s), lower medium (5 mm/s), higher medium (7 mm/s), and high (9 mm/s) velocities.

FIG. 9.

Results of velocity feedback control of a droplet (1 pixel = 62.5 μm, 1 flame = 0.33 s).

B. Operation confirmation

Experiments were performed to confirm that the developed device could perform the proposed operations described in Sec. II.

• (1)Binding and washing processes using colored water:

Blue-dyed water (Food dye blue, Kyoritsu-foods, Japan) was used for a spheroid droplet to visualize the dilution with a transparent diluent solution of PBS (Wako, Japan). Figure 10 shows images of the device for steps 1–7 of the binding and washing processes, except for step 4 (which is examined in other experiments). EC was not installed in the device for this experiment.

• -Step 1: A droplet of blue pure water was introduced into the flow channel manually via S₁.

FIG. 10.

Experimental results of the binding and washing processes (EC was not installed. Db: droplet of blue-dyed water, Dd: droplet of the diluent solution, and Dm: merged droplet) On pumps are colored red. Step 1: introduction of Db from S₁. step 2: movement to DE via A₁, step 3: drainage from L₂, step 5: introduction of Db from L₁, step 6: merging of the two droplets, and step 7: reciprocation of the merged droplet Dm for mixing (A₁ and A₂ are switched on alternatively; when A₁ is on, A₃ is also on).

Then, the control program could perform the following.

• -Step 2: The droplet was moved to region DE.
• -Step 3: The droplet stayed in region DE during drainage. The initial area of the droplet was 4096 pixels, and the target area was 400 pixels. The liquid in the droplet was drained, and the area of the droplet was reduced to 416 pixels [16-pixel (4%) error compared with the target area].
• -Step 5: The diluent solution (PBS) was introduced into the flow channel via L₁ and cut to form a droplet via A₂. The dilution ratio was 1:5, and the target area of the droplet was 16 × 5= 2080 pixels. The area of the droplet was 2126 pixels [46-pixel (2.2%) error compared with the target area]. The air between the diluent solution and blue droplet pushed the blue droplet, but it stopped on the right of DE.
• -Step 6: Two droplets were merged by exhausting the air between them from DE.
• -Step 7: The liquid in the merged droplet was mixed by reciprocating it in the zigzag flow channel.

• (2) Eluting/collecting process using transparent solutions:

Figure 11 shows the experimental result of the eluting/collecting process. A PBS droplet with a spheroid of U87 MG was initially set on the left of DE, which was the drained droplet after the washing process described next [see (3)].

• -Step 1: A PBS elute droplet was introduced with S₂ and was moved into the main channel via L₃.
• -Step 2: The elute droplet was cut with air pressure via A₂.
• -Step 3: The two droplets were merged with A₁ and A₂.
• -Step 4: The merged droplet was moved to J via A₁.
• -Step 5: The merged droplet was collected via L₃.

• (3) Binding and washing using transparent solutions:

FIG. 11.

Eluting/collecting process using transparent solutions(Ds: spheroid droplet, De: elute droplet, and Dm: merged droplet). On pumps are colored red. Step 1: introduction of an elute solution from S₂ and move to the main channel, step 2: cutting off the elute solution to form the droplet De via A₂, step 3: merging of the two droplets De and Ds to form Dm, step 4: movement to J, and step 5: collection of Dm via L₃.

The binding process and step 4 of the washing process were performed using transparent PBS solutions. See a video in the supplementary material. Figure 12 shows a photo of the droplet containing the U87 MG spheroid. It could transport the spheroid and could pass the junction J. That is, a little part of the droplet entered the branch flow channel and it was brought out by the main volume of the droplet.

FIG. 12.

Photo of a spheroid droplet.

C. Blocking

Table I summarizes the results of the phage recovery ratios of normal, droplet, and no-blocking devices. Those of the normal and droplet blocking devices were 10 times smaller than that of the no-blocking device. The effect of the droplet blocking method was almost the same as that of the normal blocking.

TABLE I.

Results of the blocking evaluation experiments.

Recovery ratio	Normal blocking (%)	Droplet blocking (%)	No blocking (%)
1	0.25	0.61	8.76
2	0.25	0.34	5.32
3	0.12	0.44	6.60
Average	0.20	0.46	6.89

D. Washing experiment using phage

We evaluated the washing performance including step 4 of flush washing using the nonspecific-binding phage. Table II lists the values of the parameters in the binding and washing processes described in Subsection II D. A U87MG spheroid with a diameter of approximately 300 μm was introduced in step 1. In step 3, the drained liquid was collected in each iteration of the process, and plaque assay was applied to each collected liquid. Iteration 0 means the collection of the drained liquid after the binding process in step 2. Table III shows the results. After three iterations, it was confirmed that the reduction ratio was on the order of 10⁻³.

TABLE II.

Values of the parameters in the binding and washing processes.

Step	Parameter	Value
2	Target speed of droplet	3 mm/s
3	Target area of droplet liquid after drainage	200 pixels
4	Volume of flush liquid	30 μl
5	Ratio of dilution	1:5
7	Number of reciprocation and target speed of droplet	5 and 6 mm/s

TABLE III.

Experimental results of washing.

Iteration	First (pfu)	Second (pfu)	Third (pfu)
0	7.46 × 105	8.76 × 105	2.34 × 105
1	3.51 × 104	1.58 × 104	2.54 × 104
2	4.63 × 103	5.44 × 103	2.97 × 103
3	7.56 × 102	7.44 × 102	5.28 × 102
Ratio of 3 to 0	1.01 × 10−3	0.85 × 10−3	2.26 × 10−3

IV. DISCUSSION

In this article, we proposed a microfluidic system that has functions to perform phage-display screening against spheroids, which is a necessary step toward the goal of phage-display screening against spheroids. We should address what remains to be done to achieve this goal.

• (1)Efficient, concise, and reliable phage quantifying method: Phage-display screening requires quantifying phages many times.⁷ Plaque assay, the most typical method of quantifying phages, is time-consuming and requires technical proficiency (e.g., careful E. coli culture) to be reliable, making it difficult to perform phage-display screening. An alternative method that can quantify phages more efficiently and concisely is highly desired. Phage quantifying methods based on fluorescence staining of phages that achieved a sufficient detection limit have been proposed.^23,24 We will apply such a method to quantify phage binding to spheroid cells.
• (2)Known proteins or peptides that can specifically bind to spheroid cells: For efficient screening, it is necessary to optimize the flow rate for binding, which maximizes phages that can specifically bind to spheroid cells, and the flow rate for washing, which maximizes the ratio of specific and nonspecific-binding phages. Known proteins or peptides that can specifically bind to spheroid cells are necessary for these optimizations. However, because the status of surface proteins of the monolayer cultured cells is not necessarily the same as that of 3D cultured cells, specific binding phages to monolayer cultured cells may not be specific binding ones to spheroid cells. We should carefully investigate the differences of specific binding phages with a reliable phage quantifying method.
• (3)Uniform formation of spheroids: To perform comparative experiments, it is also necessary to make the sizes of spheroids as uniform as possible.

The following limitations of the current system should also be addressed.

• (1)The spheroid is not easily introduced into the microchannel.

Currently, a spheroid from a 96-well plate is manually introduced into the microchannel using a syringe. This transfer requires skill and can transfer only one spheroid into one droplet. Kim et al.²⁵ proposed a contact-based spheroid transfer method that can transfer spheroids formed by the conventional hanging drop method to another device. This method proved useful in drug transfer because it can be performed in stationary droplets. However, the binding, washing, and eluting of phage-display screening, which are difficult to perform in stationary droplets, can be achieved in moving droplets (as demonstrated in the proposed microfluidic device). The contact-based spheroid transfer method cannot be applied to the proposed microfluidic device. Automatic transfer of spheroid into the proposed microfluidic device could be achieved by integrating the droplet-based spheroid formation method²⁶

• (2)The clarity of the spheroid image should be improved.

Although the camera used in the present system detected the droplets, the images of an introduced spheroid were unclear and not detectable using computational image processing. Currently, the operator must visually monitor the video of the moving spheroid droplet. If the spheroid is left behind the droplet, the operator can retrieve the spheroid by halting the running program; that is, by manually moving the droplet back to the spheroid using the syringe pumps. The program is then resumed. This manual recovery operation impairs the degree of automation. A higher speed and resolution camera might resolve this problem. Fluorescent imaging can more reliably detect a spheroid than camera imaging, but requires a fluorescent microscope and possibly an electric xy-stage to cover the whole microfluidic device. Computer control of the stage is also required to follow the spheroid droplet.

• (3)Endocytosis must be suppressed in the system.

When targeting a live spheroid rather than a formalin-fixed spheroid, endocytosis must be suppressed by maintaining the device at 4 °C. In future work, we will design and develop a system that can function at 4 °C.

• (4)Limitations of spheroids

The advantages of spheroids over monolayer cell culture were mentioned in Sec. I. However, spheroids are cell aggregations with no organic structure. Recently, organ-on-a-chips and organoids have been attracting interest²⁷ as ex vivo cell models that can imitate organ functions and patient-specific diseases. Phage-display screening using such excellent ex vivo cell models is an important direction of the present study.

V. CONCLUSIONS

In this study, we proposed a valveless droplet-based microfluidic system for phage-display screening against spheroids using a camera, three liquid pumps, and three air pumps. We experimentally verified that the system could detect and actively control droplets and that it could perform the binding, washing, eluting, and collecting processes of phage-display screening with the proposed operation sequences. In addition, we evaluated the performance of the proposed blocking method and flush washing with plaque assay using phages. Finally, we discussed our perspective to achieve the goal of phage-display screening against spheroids and the limitations of the proposed system.

SUPPLEMENTARY MATERIAL

See the supplementary material for the fabrication process, background subtraction method, materials for plaque assay, and video of the experiment of binding and washing processes using transparent solutions.

AUTHOR DECLARATIONS

Conflict of Interest

There are no conflicts of interest to declare.

Author Contributions

T.S. and A.H. developed the system and performed the experiments. T.K. and S.K.-K. participated in the discussion from the biology point of view. T.O. supervised the research and wrote the manuscript.

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author upon reasonable request.