Mechanical depletion (DP1-1) removes cells physically by scratching or applying pressure on cells in a defined area. This approach has the advantage of representing mechanical wounding, and it also has the merit of simplicity. The disadvantages of this approach include the poor consistency of CFZs and the difficulty of cleaning the cell debris. Further, the classification of CMA devices in mechanical depletion can be categorized based on the methods of operation: (1) the manual method and (2) automation method. The automation method is a key to improving consistency. The automation method can be further classified into three categories in terms of actuation principles: pneumatic, electromagnetic, and magnetic actuation. Figure 1 shows several devices that are based on the mechanical depletion (DP1-1) approach.

Liang et al. used the mechanical depletion approach to create a CFZ manually by using a pipette (Figure 1a) [10]. The limitation of the manual operation is poor consistency of creating CFZs. To increase the number of samples, Yue et al. developed a device based on the commercially available OrisTM CMA to create eight CFZs at the same time (see Figure 1b) [28]. The consistency of the CFZs was reported at 6%, with an average wound size of 600 μm × 7 mm. This result was achieved by creating eight CFZs simultaneously using multiple pipette tips. It is noted that the consistency (see Table 1) reported in the current literature is calculated by the relative standard deviation (RSD), as shown in the following equation [30].

From the above equation, the standard deviation is denoted as SD, and the average size of CFZs is denoted as X.

To reduce the variation in size of the CFZs, Lee et al. developed a weight stamping device to use a synthetic polymer, Polydimethylsiloxane (PDMS), as a puncher to apply pressure on a layer of cells (see Figure 1c) [27]. The fabrication of such a microfluidic device has made it possible to integrate the channel with sensors for the analysis of cell migration. However, their approach may create a large variation in CFZs because of manual operation. Overall, it is difficult for CMAs with the method of manual operation to meet the requirement of high consistency (about 5% RSD).

To achieve a high consistency of CFZs, Sticker et al. developed a CMA on the microfluidic device capable of creating four samples of CFZs simultaneously with a high consistency (4% RSD) (see Figure 1d) [29]. The CFZs were created by the deflection of the membrane. However, the device had the problem of leaking bubbles at high fluid pressures. Monfared et al. in the same research group, further studied the problem of reproducibility and revealed the limitation in matching the range of shear stresses in vivo [30]. The high pressure of up to 5 bars may result in bubble formation because of the permeability of PDMS. As shown in Figure 1e, the device created four samples of CFZs (5% RSD) with an average area of 1.51 mm² at 76 mm × 26 mm. Yin et al. developed a microfluid device, capable of creating 400 CFZs during one operation at shear stress conditions, as shown in Figure 1f [31]. However, the device has a problem of uniformity (66% in the shear stress condition, indicating high variation at the CFZs). It is difficult to achieve adjustability and uniformity because adjustability inherently produces a source of variations (i.e., non-uniformity).

Yilmaz et al. developed a scratch robot enabling us to create CFZs with different shapes and geometries of CFZs (see in Figure 2a) [32]. The robot is based on magnetic actuation (see in Figure 2b), enabling us to create CFZs of different shapes of wounds (see in Figure 2c) such as line wound, ‘plus’ shape wound, rectangle wound, and triangle wound. According to Yilmaz et al., the consistency of the CFZs can reach 2% [32]. Their study also suggested that wounds with more edges of CFZs may have a faster healing time. However, the device can only create a few samples (i.e., CFZs).

The chemical approach (DP1-2) is used to create CFZs through chemical reactions. The advantage of this approach is the possibility to clean the cell debris thoroughly [27]. Lin et al. developed a microfluidic device to create CFZs in fluid shear conditions [33]. However, the disadvantage of this approach is that the use of chemicals (e.g., trypsin) could slow the cell migration process; specifically, the average speed of cell migration in the chemical depletion approach was reduced to 12 μm/h compared to 25 μm/h in the mechanical depletion method [38].

The electrical approach (DP1-3) is used to apply an electrical field to a restricted area among a group of cells [34]. The advantage of this approach is that it increases the ability of automating the wounding process by controlling the electrical signals. However, the major drawback of this approach is that the damage is done to the adjacent cells. Another disadvantage of this approach is the difficulty of cleaning dead cells remaining on the top of electrodes.

The thermal approach (DP1-4) is used to apply heat to a restricted area among a group of cells [35]. The advantage of this approach is it provides the possibility to study healing-induced thermal wounding. The drawback of this approach is that heat may transfer to the surrounding cells, and consequently may negatively affect the viability of cells.

The optical approach (DP1-5) is used to remove cells in a restricted area using the high energy laser ablation [36] and ultraviolet light [39]. This approach has the merits of creating many CFZs, and no direct contact to cells. The disadvantage of this approach is the difficulty of the acquisition of such a specialized instrument. In [36], a commercial laser system called LEAP™ was used to create 96 CFZs simultaneously with a variation at 9% RSD on the CFZs.

Based on the actuation principle for driving fluids, there are three options: electrical actuation (DP3a-1-1), electromagnetic actuation (DP3a-1-2), and pneumatic actuation (DP3a-1-3). In the following sections, the classification of devices in the existing devices pertinent to the above principles are discussed.

For the electrical actuation (DP3a-1-1), the existing studies focus on the control of fluid-induced pressure to manipulate shear stress in cells by using pumps or valves in microfluidic devices. Polacheck et al. described microfluidic devices as promising platforms to elicit cell migration in biological systems such as blood vessels [49]. Shemesh et al. reviewed microfluidic devices integrated with analysis tools for providing the shear stress environment for cell migration [50]. Kaarj and Yoon studied devices that deliver mechanical stimuli to cell migration assays [51]. Most of the existing pumps and valves are connected externally, whereas cells are in the microfluidic system [52,53]. Shear stresses generated internally are thus not uniform in microfluidic channels. For example, Saias et al. reported a device with a uniformity of 80% and with a tolerance of 10% [54]. However, such uniformity and tolerance are far from sufficient. Schimek et al. reported on a device where there are bulky external pumps, enabling local stress in cells [55]. Li et al. studied the uniformity of shear stresses in a bio-inspired structure [56], as shown in Figure 3. The limitation with their device was that the uniformity is only in the low range of shear stresses (10−3 to 10−2 dyn/cm²), as opposed to a wide range of shear stresses (1–70 dyn/cm²) in blood vessels [72,73].

For the electromagnetic actuation (DP3a-1-2), the concept is to use the magnetic field driven by electric power to create the deformation to drive the fluid. It has the potential of controlling the flow precisely. Bhushan et al. proposed a pump system, which achieved a fluid backpressure of 20 kpa with a low energy dissipation of 0.65 mJ per switching and allowed the pump to operate at 0.45 μL/s with an average power dissipation of 1.3 mW and a temperature rise of 0.04 °C [57]. However, the drawback of the electromagnetic actuation is its difficulty in maintaining the temperature at 37 °C due to the high amplitude of the electrical currents.

For the pneumatic actuation (DP3a-1-3), the concept is to use gas or liquid to generate a high mechanical pressure on polymeric diaphragms resulting in a high flow rate at 102.05 μL/min in microfluid channels [58]. The shortcoming of this actuation principle is the formation of bubbles, which may potentially damage cell dynamics.

In the literature, there are two approaches to use cell-generated stresses in cells: the mechanical approach (DP3a-2-1) and the chemical approach (DP3a-2-2). Figure 4 illustrates various signal molecules from 1 to 140 nm in a deformation pathway between a cell and its environment [74]. As shown in Figure 4, the cell environment, often referred to as the extracellular matrix (ECM), is a complex structure serving as the connective tissue to support cells [75].

The mechanical approach (DP3a-2-1) is used to apply external force at the structure of a cell, and the cell then provides the force feedback on the deformation pathway [48]. For any moving object, it can be characterized by stiffness, inertia, and damping. Previous studies gave vague definitions about stiffness compared to “rigidity” [76], “hardness” [77], and “shape stability” [78]. It is noted that, in this paper, stiffness is defined as a system property of a cell with the ability to deform at a particular point of location on the cell along a particular direction (including both translational and rotational) when the cell is subject to a force (including moment) from the environment in which the cell interacts [79,80]. Fu et al. presented micropost arrays to investigate the effect of the stiffness of the substrate on cell shape and cytoskeletal contractility [81]. Nava et al. presented the concept of “anisotropic force” [82], which is generated by the cell on the ECM with different magnitudes of forces at varying orientations [83,84]. Steward and Kelly described the stiffness and cell shape as intrinsic mechanical cues to induce biological responses [85]. Paul et al. presented various simulators to generate stiffness for studying the effects of physical confinement on cell morphology, as well as behavior [86]. Yamada and Sixt defined the stiffness of the cell as a “local” property as opposed to a “bulky” property [87].

The chemical approach (DP3a-2-2) is used to induce the change of mechanical properties (e.g., stiffness) by the chemical cues including ions, signal molecules, and soluble factors [88]. Forero et al. presented that the adhesion forces can last for several tens of seconds at 40–70 pN forces at a protein called fimbriae (7 nm in diameter and 1 μm in length) [89]. The advantage of the chemical approach is that it can accurately target specific proteins. Sun et al. studied the different types of chemical bonds including slip bond, catch bond, and clustering bond for force generation and measurement at the molecular level [90]. However, the disadvantage of the chemical approach is that the structural change may affect cell dynamics, and thus become a confounding factor of the migration of cells.

The mechanical forces on cells can be created by various stimuli in the environment. There are two types of stimuli from the environment: physical stimuli (DP3a-3-1) and chemical stimuli (DP3a-3-2). For the physical stimuli method, this method includes optical [61], electric [59,62,64], magnetic [63], and acoustic [65] stimuli. For example, Morimoto et al. used an electric field (2.5 V/mm) to stimulate the contraction of muscle cells, capable of generating mechanical force at 13 mN [59]. The limitation for the physical stimuli method is that it lacks the first principles to generate controllable mechanical forces, making it difficult to control the mechanical forces. Several studies presented synthetic environments (i.e., substrate or matrix) to enhance the deformation at a particular structure on cells, as shown in Figure 5. Huang et al. studied the structures of the synthetic ECM at different scales [91]. Liu et al. presented the development of stimuli-responsive materials (e.g., hydrogels) for studying the mechanical interaction with cells [92].

For the chemical stimuli method, the change of the mechanical properties of the cell environment can be induced by the chemical stimuli by changing the bonds between the cell and its environment. These chemical stimuli include pH [67,70], oxygen [93], and ion concentration [66]. In addition, biomolecules such as enzymes were demonstrated to be effective in changing the stiffness of the cell environment due to the chemical reaction [69,94], crosslink activation [95], and enzymatic degradation [68]. For instance, Yin et al. built a synthetic ECM integrated with hydrogels, enabling a change in the morphology of the ECM with respect to the change of glucose level [69]. The limitation for the chemical stimuli method is that the chemical reaction could take days while the changes of cell morphology occur in minutes or hours [70]. Further, the variation in the chemical properties of the cell environment may not be desirable to the biological environment with living cells.

There are two general approaches to characterizing stress and strain: probe-based measurement (DP3b-1) and interaction force measurement (DP3b-2) [96,97]. The first approach is used to measure interaction forces between a probe tip and sample surface in a probe-based instrument, such as an atomic force microscope (AFM) [71]. This approach can provide a high resolution (5 pN to 10 nN) but has a high level of intrusiveness, which may cause irreversible damage of cell structures and eventually lead to cell death [98]. The second approach is used to measure the interaction traction force between cells and the synthetic ECM [99,100]. The interaction force instrument, such as traction force microscopy (TFM), is used to measure the adhesion force between cells and the substrate. This approach can provide the resolution between 2 and 120 nN at a 2-dimensional surface [60]. However, the limitation of this approach is the low accuracy in measuring forces due to the disturbance from the dynamic interaction between cells and their environment.

Most of the existing CMAs are focused on creating multiple chemical conditions for drug testing in wound healing assays. It is noted that in this paper, the number of conditions means that the number of varieties of geometric distributions of fluids (i.e., drugs or growth factors) and the number of samples means the number of CFZs created simultaneously. Lee et al. developed a “stamping” device based on PDMS with one condition and 20 samples [27]. Yue et al. developed a device to change 16 chemical conditions based on a commercially available CMA (OrisTM) enabling us to produce eight samples simultaneously by multichannel pipettes [28]. To combine multiple conditions with multiple samples, Sticker et al. developed microfluidic devices with four conditions and four samples [29]. The fluidic channels were boned with a deformable PDMS membrane with a diameter of 1.5 mm and height of 250 μm. Lin et al. developed a microfluidic device with three shear stress conditions and three samples [33]. Monfared et al. presented a device consisting of five substrates, and there were eight conditions and four samples on each substrate [30]. Yin et al. developed a microfluidic device with one condition and 400 samples [31]. However, none of the existing CMAs can achieve more (about 100) chemical conditions, many samples (greater than 100), and the strain condition.