A Review on Microfluidic Platforms Applied to Nerve Regeneration

Abstract

In recent decades, microfluidics have significantly advanced nerve regeneration research. Microfluidic devices can provide an accurate simulation of in vivo microenvironment for different research purposes such as analyzing myelin growth inhibitory factors, screening drugs, assessing nerve growth factors, and exploring mechanisms of neural injury and regeneration. The microfluidic platform offers technical supports for nerve regeneration that enable precise spatio-temporal control of cells, such as neuron isolation, single-cell manipulation, neural patterning, and axon guidance. In this paper, we review the development and recent advances of microfluidic platforms for nerve regeneration research.

1. Introduction

According to the World Health Organization, up to 500,000 people suffer spinal cord injuries (SCI) per year worldwide. SCI can cause associated diseases that decrease patients’ life quality and even endanger their lives in ways such as chronic pain, deep vein thrombosis and respiratory complications. Depending on the degree of injures, SCI can be classified as complete SCI and incomplete SCI. With complete SCI, the spinal cord is completely severed, which leads to complete loss of function and sensation below the injury parts. Meanwhile, incomplete SCI patients are able to retain some unilateral sensory and motor function although spinal cord is partially damaged. Among all SCI patients, the incomplete SCI patients account for 65% and the complete SCI patients account for 35% [1]. The recovery rate of incomplete SCI patients is higher than complete SCI patients. About 20% to 75% of incomplete SCI patients regain some degree of walking ability and 20% patients restore motor function completely in 1 year post-injury [2]. Only 5% complete SCI patients are able to recover to their pre-injury status, while the remaining 95% patients have to suffer from various physical disabilities including urinary dysfunction and loss of locomotor function. However, incomplete SCI patients are able to restore physical functions through long-term post-healing treatment and rehabilitation because of neuroplasticity. Neuroplasticity lays an important theoretical foundation for incomplete SCI treatment; thus, the treatment principle is to reactivate the residual neural circuits to modulate the impaired functions of body. Treatments such as the spinal epidural electrical stimulation is one of the applications based on the neuroplasticity theory, which can effectively improve the lower limbs locomotor functions of incomplete SCI patients. However, these treatments are not available for complete SCI patients due to the lack of residual nerves. Nowadays, the treatment for complete SCI remains in the laboratory stage instead of clinical applications, and a priority research area of treatment for complete SCI is nerve regeneration. However, traditional methods for in vitro neuron culture cannot sufficiently simulate or provide precisely control of the different neuronal microenvironments. Therefore, it is difficult to study the molecular and cellular mechanisms that differentially affect different parts of neuronal cells using traditional culture methods [3].

Recent developments in microfluidics and related applications to neurobiology have led to a significant advance in neuron culture. Microfluidic platform can realize precise spatio-temporal control of cellular microenvironment [4], allowing neurobiologists to explore a wide variety of neuronal events such as axon elongation [5,6], local signaling events [7] and interactions with other cells [8]. Microfluidic platform can manipulate nanoliter-scale liquids by using micro-channels ranging in sizes from ten to hundreds of micrometers [9] that enables single cell manipulation. In addition, microfluidic platform can utilize polydimethylsiloxane polymers materials (PDMS) for microfabrication, which has advantages in biocompatibility, gas permeability and surface-bound gradients to achieve microfabrication of complex tissue/organ cells [10,11]. Furthermore, the environment in microfluidic devices is enclosed, which is similar to the in vivo environment. These airtight, closed, non-convective devices allow for local accumulation of cellular secreted material in contrast to classical open micropore structure. Compared with traditional neuron culture platform, microfluidic-based platform has many advantages such as single neurons manipulation [12], isolation of axons [13,14], cultured neuron patterning [15] and neurite outgrowth control [16]. The microfluidic platform can be utilized to detect and analyze chemical factors that inhibit nerve axon regeneration, create different in vitro microenvironments and evaluate candidate drugs that promote nerve growth. Microfluidic platforms enable safe, high-throughput and accurate experiments at the microscopic scale. Some experiments that require many volunteers with risk of injury, such as electrical stimulation experiments [17,18], can be performed on microfluidic platforms [19].

In recent years, commercial microfluidic platforms have been widely used in many cytological pathology studies. Teotia, P. [20] used microfluidic device SND 450 (Xona Microfluidics, Inc., Durham, NC, USA) to test the hypothesis that the mTOR pathway regulates retinal ganglion cells (RGC) and promotes axonal regeneration after injury. Wang, J et al. [21] investigated the regenerative factors of chemically damaged neurons through a commercial microfluidic platform. Nagendran, T. [22] adopted a modified rabies virus encoding a fluorescent protein to retrogradely label neurons through isolated axons. Then he created an isolated microenvironment within a compartment, and performed axotomy and immunocytochemical analysis on a pre-assembled XC450 XonaChip (Xona Microfluidics, Inc.). Yu Yong [23] proposed a method for assessing axonal degeneration using microfluidic device. De Vincentiis [24] conducted an experiment on the effects of external magnetic field gradients to generate drag forces on axon growth on R150 chip (Xona Microfluidics, Inc.) to investigate strategies for axon regrowth. Gladkov, A. [25] used microfluidic device to investigate the mechanisms of axon–axon interaction and developed axons to guide the growth of young axons during embryonic development.

The current commercial microfluidic platforms can perform several basic operations for in vitro regenerative culture of neuronal cells, including neuron loading, axon isolation, axon injury and drug/nerve growth factor (NGF) screening. Although commercial microfluidic platforms can be used to perform high-quality experiments in neural regeneration study, they still cannot fulfil the specific requirements in some experiments. For example, experiments that involve precise damage to the axon require a special coating on the top of the micro-channels. Experiments with solution gradients for axon guidance require a modification of micro-channels, by adding extra inlets and micro-wells. Therefore, many papers have proposed self-developed microfluidic platforms to satisfy the specified research requirements such as experimental protocols, study models, and specific methods. Those self-developed microfluidic platforms are introduced in following sections.

In Section 2, we discuss the properties of commonly used microfluidic fabrication materials and introduce different structures for microfluidic devices, including 2D and 3D culture structures. Section 3 introduces microfluidic devices for related research according to two subcategories of nerve regeneration research: nerve repair and nerve regrowth. The studies of nerve repair often involve precise damage to neural axon. Therefore, a subsection is listed to introduce different microfluidic platforms to generate axonal injury. Axonal guidance is a significant technical point in the research of nerve regeneration; we therefore describe in detail the different platforms that guide the growth of axon in Section 4. Finally, in Section 5, we summarize this manuscript, whereafter we discuss the existing challenges and difficulties, and future perspective.

2. Design of Microfluidic Devices

2.1. Materials

The traditional method for the manufacturing process of cell culture devices uses glass, plastic and silicone as common raw materials. Silicone has advantages, such as resistance to organic solvents, ease of metal deposition, high thermal conductivity and stable electroosmotic mobility. However, its properties such as of impermeability, opacity and high cost make it an undesirable candidate for manufacturing microfluidic devices. Besides this, its high stiffness makes it difficult to manufacture microfluidic components [26]. Glass is an optically transparent, electrically insulating and impermeable material. Glass has relatively low non-specific adsorption and is compatible with biological samples. However, glass has similar disadvantages to silicone, wherein its high hardness and high manufacturing costs lead to many limitations for its application in microfluidics [27]. Low temperature co-fired ceramic (LTCC) technology is a commonly used fabrication method for ceramic microfluidic device. Ceramic material has high themostability, high surface stability and are lower in price than glass and silicone. However, ceramic does not have gas permeability or optical transparency either.

Some manufacturers have attempted to use thermoplastic materials such as polystyrene (PS), polycarbonate (PC), poly-methyl methacrylate (PMMA) or poly-ethylene glycol (PEG) diacrylate to manufacture microfluidic device. Thermoplastics are fabricated by thermomolding that allows the manufacture of a large amount of duplicates at high speed and low cost which is profitable for commercial production, but not economical for production of prototypes. Thermoplastoic material has many merits such as water stability, optical clarity, rigid mechanical property and compatibility to electrophoresis, which make it one of the common materials for microfluidic devices [28]. However, different thermoplastic materials have different characteristics. PS has the advantages of being optically transparent, biocompatible, chemically inert and having easily functionalized surfaces. Its hydrophobic surface can be hydrophilized by physicochemical means. However, PC is expensive to manufacture and process and is suitable for large-scale manufacturing [29]. PC material is made from bisphenol and phosgene polymerization. Its advantages of high transparency, high glass transition temperature, high impact resistance and low moisture absorption make it suitable for DNA thermal cycling applications. However, PC has poor resistance to some organic solvents and poor absorption of UV light [30]. The advantages of PMMA include strong mechanical properties, high optical transparency and electrolysis compatibility. Therefore, it is often used in the fabrication of disposable microfluidic chips. However, the hydrophobicity of PMMA is the worst among all thermoplastic materials [31,32]. As PDMS, PEG diacrylate shares similar advantages, such as water stability, optical clarity and low background fluorescence. However, PEG diacrylate shows less nonspecific adsorption and has greater resistance to permeation of small hydrophobic molecules than PDMS material [33].

Over past two decades, PDMS has gradually been used for fabrication and prototyping of microfluidic chips, because of its convenient fabrication, low production cost, good optical clarity, high gas permeability and relative biocompatibility. PDMS provides a chemically inert surface with low interfacial free energy, and its surface properties tend to modify with surface coating such as plasma, covalent and dynamic modification [26]. In addition, due to the superior biocompatibility and nontoxicity, PDMS materials can be made into chips that will not be immunologically rejected by human body and have the potential to be applied in the in vivo environment for long-term coexistence with the body. There are some disadvantages of PDMS, such as permeability to water vapor, which makes it difficult to control evaporation in PDMS devices. In addition, metals such as electrode resistors are difficult to deposit on PDMS, which limits the integration of resistive electrodes. However, PDMS can be bonded to glass slides by plasma treatment and dielectric deposition can be performed on glass slides. In summary, PDMS is an ideal material for microfabrication of the microfluidic device for cell culture and nerve regeneration research.

In recent years, hydrogels have been becoming popular materials in microfluidic devices. There are various types of hydrogels, and most of hydrogels are hydrophilic, such as agaros, polyacrylamide and alginate. Some hydrogels are hydrophobic association hydrogels, such as hydrophobic collagen and synthetic hydrogels using molecules with hydrophobic properties as segments [34,35,36]. In addition, there are hydrogel materials that have both hydrophilic and hydrophobic features. For example, Thomas, B.H. [37] presented a new type of hydrogels combining both hydrophilic and hydrophobic structures as a replacement for cartilage Abdurrahmanoglu, S. [38] proposed a hydrosol using acrylamide as a hydrophilic monomer and lauryl acrylate as a hydrophobic monomer. The advantage of adopting hydrogels for cell culture is the 3D network structure of hydrophilic polymer chains that allows the diffusion and transmission of small molecules and biological particles. Hydrogels can be used to construct microchannels for the transport of solutions, cells and other materials [26]. Hydrogel is an ideal material to create chemical and physical gradients. For example, collagen, agarose and poly ethylene glycol are preferable for generating chemical gradient in microenvironments due to their permeability allowing fluid diffusion. Besides this, by placing different concentrations of hydrosols, such as collagen, in adjacent chambers, physical gradients can be created, which can determine the direction of cell migration and growth [39]. However, there are some disadvantages of hydrogels for neural cell culture, such as unfavorable diffusion of nutrients and poor permeability, which leads to necrosis of cells in deep layers due to starvation and lack of oxygen [40]. Due to the low density and low strength at the macromolecular scale, hydrogels support only micron scales lower than the nanoscale of other polymers in microfabrication.

The properties of microfludic materials discussed above are listed in

Table 1. Properties of different materials for microfluidic devices applied in neurobiology.

Fabrication Material	Categories	Themo-Stability	Surface Stability	Optical Transparency	Gas Permeability	Biocompatibility	Hydrophilicity	Cost
Silicon	Inorganic	High	High	No	No	✘	Hydrophilic	High
Glass	Inorganic	High	High	High	No	✘	Hydrophilic	High
Ceramic	Inorganic	High	High	No	No	✘	Hydrophilic	Low
Thermoplastic	polymeric	High	Medium	High	Low	✓	Hydrophobic	Low
PDMS	polymeric	Medium	Low	High	High	✓	Hydrophobic	Low
Hydrogel	Hydrogel	Low	N/A	Low	No	✓	Hydrophobic/ Hydrophilic	Low

.

2.2. Structures

The geometry of the microenvironment varies a lot in nerve regeneration experiments with different conditions. As shown in Figure 1, the basic structure of a microfluidic device for neural regeneration contains at least two compartments for cell soma and distal axon storage and several micro-channels connecting the two compartments, in which the neural axons can grow. Cell somas are attached to the flat surface in the somal compartments, where they can adhere and allow the axon to grow in micro-channels.

2.2.1. 2D Culture Structure

There are many kinds of 2D culture structures. Figure 2A shows a modified structure of microfluidic device, which is added a pair of reservoirs on both sides of each compartment to store sufficient volume of medium. This structure is the first generation of microfluidic-based compartmentalized neuron culture platform designed by Taylor, A.M. [13] and Park, J.W. [41]. The structure can make a small volume difference between the two compartments, achieving fluid isolation of the axon in the microenvironment. The role of microchannels is to directly isolate the axons. Their elongated structures are designed based on the property that axons grow faster and longer than dendrites so that the axon, rather than the dendrite, can extend to the opposite isolation compartment. By varying the length of the microchannel, the axon tip can access proximal or distal sites. There are many microfluidic platform-based neurobiological researches that have adopted this structure [42,43,44]. Subsequent improvements to microfluidic devices include increasing the number of compartments and optimizing the alignment of compartments, micro-channels and reservoirs on the chip. As shown in Figure 2B, Deleglise, B. [45] designed a microfluidic device with three compartments (soma, axonal, axonal) to conduct comparison experiments. Wang, J. et al. [21] presented a microfluidic device with four compartments (glial, glial, axonal, somal) to control the injection of solutions in micro-channels and somal/axonal compartments, which is shown as Figure 2C. Samson, A.J. [46] presented a five-compartment microfluidic device to screen drugs. As shown in Figure 2D, fluidic isolation between compartments allows for investigation of effects of neural networks after application of chemical injury to one of the compartment. Kim, H. S. [47] designed a high-throughput microfluidic device with 24 axon compartments and one somal compartment. As shown in Figure 2E, every six axon compartments share a common reservoir for culture media. This design facilitates the use of batch culture and the conduct of controlled trials. Park, J. [48] introduced a circular structure microfluidic device with axon guiding microgrooves for quantified axon growth and regeneration, which is shown as Figure 2F. These kinds of improvements are made to meet the requirements of experiments such as drug/NGF screening, spatio-temporal control of cell and axonal axotomy. Takesuke, S. [49]. presented a device constructed of eight chambers in the shape of an eight-spoked asterisk. The device is characterized by the ability to limit the interface between the cell-filled hydrogel and the chemical solution of interest to improve the time difference between the chemical solution reaching the cells. The advantages and disadvantages of different cell culture structures are listed in

Table 2. Advantages and disadvantages of different cell culture structures.

Methods	Advantages		Disadvantages
Structure of Classical methods			It cannot mimic in vivo environment. A large number of samples and reagents are required. Large-scale experiments are difficult to be conducted.
Structure in Figure 1	East fabrication of the single layer in PDMS microfluidic device can be realized easily. Applicable to non-adherent cells. It can be combined with basic flow control.		It requires precise flow control to reduce the medium switch time without flushing cells. Injection of cells within microchannels can exert stress on cells. Cells density in each microchannel will affect the drug functions to cells. Flow rates during flow changes can arouse mechanical stress on cells.
Structure in Figure 2A,B,D	Faster diffusion speed than basic structures. Applicable to non-adherent cells. Fast medium switch time. Flow rates during flow changes will not arouse mechanical stress on cells.		Injection of cells within channels can exert stress on cells. It requires quite precise flow control to limit medium convection in cells channels that leads to cells movements.
Structure in Figure 2C	Applicable to non-adherent cells. Flow rates during flow changes will not arouse mechanical stress on cells.		Injection of cells in channels can exert stress on cells. Medium change speed may be difficult to adjust.

.

2.2.2. 3D Culture Structure

Most existing microfluidic devices for nerve regeneration studies are constructed in 2D culture structure. However, with increasing requirements to mimic in vivo environment, 2D culture devices have revealed some drawbacks. The 2D culture devices neither accurately recapitulate the structures, functions or physiology of living tissues, nor the highly complex and dynamic 3D environments in vivo [50]. Recent years, to better represent the intricate human physiology and relevant conditions, 3D cell culture platforms have gradually gained attentions. In 3D culture, cells are grown on a scaffold-based matrix that mimics the extracellular matrix (ECM) [51]. Compared with 2D culture, 3D culture can realize better simulations of actual complex environment in vivo, including cell matrix components, intercellular and cell-substrate interactions. The 3D culture is able to reproduce more physiological equilibration and transport of soluble factors [52]. In some experiments, such as drug screening, the results of 3D culture are closer to clinical trials than experiments with 2D culture [53]. The 3D culture structure can improve the simulation of the in vivo setting of cells inside organs and tissues. This helps to simulate the complex environment within the vertebral cavity of post-SCI, which has significant implications for SCI research.

Generally, 3D cell culture is formed on pore plates or the biocompatible extracellular scaffolds (hydrogels) of trans-well membranes. Figure 3A,B show two typical structures for 3D culture microfluidic devices chemical gradients-based 3D culture and 3D scaffolds culture using hydrogels. By using hydrogels containing natural molecules of extra-cellular matrix (ECM) or synthetic polymers, cells are induced to polarize and interact with adjacent cells, randomly interspersed in ECM.

Tang, Y. [54] developed a microfluidic device to investigate regeneration of CNS neurons in response to neurotoxic natural small molecules. The device consists of three parallel gradient units, and each unit has a central main channel interconnected by two asymmetric peripheral channels through 3D microgrooves. As shown in Figure 3C, this structure drives the flow direction of the fluid, which constructs a stable 3D concentration gradient in the central channel. Romano, N.H. [55] presented a platform that combines a gradient-generated microfluidic device with 3D protein-engineered hydrogels to study the effects of RGD ligand density on neurite outgrowth and neuronal 3D pathways, shown as Figure 3D. Choi, J. [56] proposed a gel-free 3D microfluidic cell culture system in order to improve neurogenic potency. The structure is consisted of a fluidic channel with a size of 1000 mm × 0.6 mm. An array of ellipticals micro-pillars with size of 30 mm × 50 mm and interval of 20 mm, is fixed on the center of the microfluidic channel. The central part of cell culture compartment is connected to a reservoir inlet for loading cells. Kunze, A. [57] presented a four layers microfluidic device to investigate neuronal cell culture in 3D hydrogel scaffolds. The hydrogel is structured in parallel layers to reconstruct cell layers that close to the natural environment. Renaud, P. [58] designed a 3D layered hydrogel scaffold microfluidic device using an agarose–alginate mixture that enables the primary cortical cell culture in micro-patterned multi-layered scaffold. As shown in Figure 3E, the hydrogel or cell loaded hydrogel flow in through four inlet channels, through the main channel and exit through the outlet channel. The 3D multilayer scaffolds are micro-patterned to achieve two cellular hydrogels separated by a cell-free hydrogel layer. Kunze, A. and Renaud, P. [59] presented a novel microfluidic based cell culture method that combines 3D micro-patterning of hydrogel layers with linear chemical gradient formation.

3. Microfluidic Platforms Applied to Nerve Repair and Outgrowth

Once the nerve is damaged, it will try to repair itself. The degree of nerve recovery depends on a number factors such as mechanism of injury, the time since the injury and the mechanism of repair [60]. According to the mechanism of nerve damage, microfluidic devices for nerve regeneration can be categorized into two types, nerve repair and neurite outgrowth. Different from studies of in vitro reconstitution and functional replication of human tissues [61,62], nerve repairing studies require precise artificial physical or chemical damage to the neuron axon (without removing the severed part). Then drugs or NGFs are used to repair the axon and restore its function. In contrast, neurite outgrowth studies require excision of the axon and then promoting its regrowth by drugs or reduction of nerve growth inhibitory factors. In the following paragraphs, we describe microfluidic platforms applied to the two types of studies.

3.1. Nerve Repair

We present the different microfluidics experimental platforms and their studies. These studies used different injury models to cause precise damage to nerve cells or axons and different means to promote neuronal repair. Microscopy combined with time-lapse photography was adopted to capture nerve cells before and after injuries and axonal regeneration processes.

3.1.1. Microfluidic Platforms for Different Repairing Methods

There are basically three methods for regeneration of nerve cells after injury: drug-directed regeneration, NGF-directed regeneration and elimination of factors that inhibit axonal regeneration to induce regeneration. Some experimental protocols use a combination of those methods; thus, the microfluidic device has to be changed and adapt to these protocols.

Sala-Jarque, J. [63] designed a vacuum-assisted axotomy microfluidic device for studying the effect of neuromuscular activities on axonal repair after axotomy. NFGs such as brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF) and ciliary neurotrophic factor (CNTF) were added to trigger the lesioned area repair. This device enables NGF testing and environmental testing for assessing conditions for regrowth after neuronal axonal injury. As shown in Figure 4A, the microfluidic device consists of two compartments and an axotomy channel (100 μm × 100 μm × 12,000 μm) to perform vacuum-assisted axotomy. The medium inside the axotomy channel was aspirated using a P20 micropipette set at 15 μL, forming a tight seal between the tip and the outlet well (Ø = 1.25 mm). Air bubbles were introduced inside the axotomy channel and fresh medium was then reintroduced through the inlet well. Hosmane, S. [64] designed a novel valve-based microfluidic device for axon injury and regrowth. this device can regulate air pressure through several valves to cause controlled micro-compression damage on the axon. In the work of Tang, Y. [54], the microfluidic device can provide a 3D microenvironment with temporal and spatial adjustable gradients of NGF concentration to repair CNS neural axon injured by 6-hydroxydopamine (6-OHDA). 6-OHDA was injected from three inlets to cause neurodegeneration to the neurons. After rewashing the residual 6-OHDA in the central channel, different concentrations of collagen hydrogel and natural products were used to promote the regeneration of the axon. In the research of Wang, J. et al. [21], the PDMS microfluidic chip was designed as a four-layered structure with size of 3.0 cm × 3.0 cm. Figure 4B shows the schematic diagram of the chip and experiment process of neuronal degeneration and regeneration after chemical ACR injury. Tong, Z. [65] presented a reproducible microfluidic method to simulate in vitro mechanical lesion of hundreds of axons simultaneously in a controlled manner. As shown in Figure 4C, the structure of the device is based on the basic model with the addition of compression damage device. The dimensions of the induced axonal injury and its distance from the neuronal cell body are precisely controlled while preserving both the proximal and distal portions of axons. Mika, J.K. [66] presented a microfluidic–microelectrode device to study the regeneration of neurons and its process by different concentrations of neurotrophic factors. The model is based on the structure shown in Figure 2A; the microelectrode array is equipped above the somal compartments and records the electrical activity of growing neurites.

3.1.2. Microfluidic Platforms for Different Injury Models

It is very important to choose suitable injury models for different studies purposes. The regeneration of neurite axon in vitro and depends not only on the experimental method but also on the injury model of nerve cells. Since the level and type of neural injury affect the effectiveness of different repair and assessment of the strategies. The common injury models for study are biochemical injuries (hemolytic injury, excitotoxic injury, immunotoxic injury), physical injuries (stretch injury, compression injury, incision injury), laser-based injuries and vacuum-assisted injuries.

Table 3. Injury mechanisms and repair methods of different researches.

Research	Repairing Method	Injury Mechanism
Dollé, J. P. [69]	NGF	Physical injury (compression)
Yap, Y. C. [70]	NGF	Physical injury (compression)
Bang, S. [71]	Drug + NGF	Physical injury (puncture)
Taylor, A. M. [72]	Nan	Vacuum-assisted injury
Zhang, J. N. [73]	NGF	Vacuum-assisted injury
Lezana, J. P. [74]	Drug + NGF	Vacuum-assisted injury
Kim, Y. T. [75]	Drug	Laser injury
Hellman, A. N. [76]	Drug	Laser injury
Gokce, S. K. [77]	Drug + neurodevelopmental genes	Laser injury
Wang. J. [21]	NGF	Biochemical injury
Samson, A. J. [46]	Drug + NGF	Biochemical injury

lists some studies according to the damage mechanism and repair method

Physical injury is the most commonly used injury model for in vitro nerve regeneration and the leading cause of SCI and CNS disorder. Physical injury to nerve axons includes stretch injury, compression injury, and incision injury. Although there are many methods to reproduce tensile/strain injury in organ-type slices such as Morrison, B. [67] and Pfister, B.J. [68]. However, limited by reproducibility, these methods cannot be applied to tensile injuries to single isolated axon. Microfluidic platforms can allow precise delivery of stress/strain-based physical injury to isolated axons, diffuse portions of axons, or axons of different lengths. Dollé, J.P. [69] designed a microfluidic device to apply aerodynamic strain-based damage to axons grown within micro-channels. This design facilitates subsequent microscopic observation of injured axons to examine the dynamics of axonal degeneration and regeneration. Yap, Y.C. [70] designed a microfluidic device to induce strain injury to isolated rat cortical axons. Unlike Dollé’s strain injury device, this device can cause smooth localized tensile damage to axons. It caused more than 2 mm of damage to axons and more than 90 μm of local tensile damage to axons, including 0.5% of strain or 5% of strain. Bang, S. [71] designed a microfluidic device that reproduces the 3D alignment of axons in CNS and performs precise 3D transection of nerve axons. The device can reproduce mechanical damage to axons by puncturing the neuron-containing Matrigel with a pin. In the research of Hosmane, S. [64], a pneumatic compression based microfluidic device was developed to generate accurate injury on a single axon. Finite element modelling is utilized to parameterize and normalize the degree of axonal damage by estimating the forces between the injury pad and glass substrate and quantitively characterizing the performance of device. This device enables different levels of deformation of single axon from minor injury to major injury. For example, minor injury (<55 kPa) allows the vast majority of axons to remain health and able to continue to grow. Moderate injury (55–90 kPa) causes the degeneration of majority of axons, with axonal swelling. Major injury (>95 kPa) causes rapid and complete transection of all axons.

The vacuum-assisted injury model is one of the common models used to study neuronal injury and regeneration, allowing for precise damage to single neurons. Taylor, A. M. [13] developed the first generation of vacuum-assisted neural injury microfluidic device. The device uses an aspiration tube to perform vacuum aspiration at the entrance of the axonal compartment, generating an air bubble that shears the axon. The device allows damage to the axon without affecting the soma, which is not feasible in conventional non-microfluidic methods. The vacuum injury model has been applied to many studies to screen for potential drugs for axonal regeneration. Taylor, A. M. [72] performed microarray analysis of cortical axons without somatic cell contamination. It showed a rise in cytoskeletal and intracellularly transported transcripts in vacuum. Zhang, J. N. et al. [73] used a vacuum-based microfluidic injury device to simulate and study acute axonal degeneration verifying that calreticulin and disintegrin response mediator protein-2 can act as regulators of acute axonal degeneration in vitro. Lezana, J. P. et al. [74] used a vacuum-assisted injury microfluidic device to study the effect of peroxisome proliferator-activated receptor γ in axon regeneration.

Laser-based injury is a highly accurate and reproducible method that enables precise fine-tune of position, scale and degree of damage of the injury spot. Kim, Y. T. [75] described a neuro-optical microfluidic platform for studying injury and subsequent regeneration of individual mammalian axons. This platform consists of three components integrated on an inverted microscope that includes a neuronal culture microfluidic device, a femtosecond laser for precise axon excision, and a microcell culture incubator for continuous long-term observation of events following injury. The degree of damage can be adjusted by tuning the exposure time of the laser to achieve a high degree of reproducible axotomy. Hellman, A. N. [76] used pulsed laser micro-beam irradiation and microfluidic cell culture to study the dynamics of axonal injury and regeneration in vitro. Laser pulses of 400 nJ and 800 nJ were applied to produce partial and complete transection of the axon, respectively. EGTA was used to chelate extracellular calcium and potentially reduce the severity of axonal degeneration after injury. Gokce, S. K. [77] presented a multi-trap microfluidic device to perform neural regeneration study in Caenorhabditis elegans (C. elegans). The device enables immobilization of 20 C. elegans at the favorable orientation for precise laser surgery and optical pathway required for high-resolution imaging.

Biochemical injury is a commonly used injury method to investigate neurodegeneration disease and nerve regeneration. This method can holistically affect the biological activity and survival of the neuron. In the study of Wang, J. [21], neurotoxic drugs were used to cause varying degrees of biochemical damage to neuronal axon. In the research of Samson, A.J. [46], biochemical injury was utilized to damage neural axons in the central compartment. This method uses concentration and duration of the neurotoxic solution to control the degree of damage. The major disadvantage of biochemical injury is the inability to cause precise damage to the axon. In addition, it needs to use large amounts of nutrient solution to flush out isolation chambers and micro-channels.

3.2. Neurite Outgrowth

Gokce, S. K. [77] developed a microfluidic in vitro culture platform to simulate axon entrapment during nerve regrowth and to investigate the mechanism by which glial cell line-derived neurotrophic factor (GDNF) causes the axon entrapment. As shown in Figure 5A, the device consists of three culture chambers connected by two sets of micro-channels that prevent cell soma from moving between chambers, but allow neurons to grow between chambers. The structure allows the transport of proteins from the distal chamber to the somatic chamber to ensure that factors in the distal chamber elicit responses from neurons separated by micro-channels. The platform features a superior ability to enable cell signaling through exogenous growth factors and cell-cell interactions. Hyung, S. et al. [78] proposed a 3D motor neuron–schwann cell (MN-SC) co-culture on a microfluidic biochip to evaluate the influence of neural activity on axon outgrowth triggered by optogenetically mediated light stimulation. The microfluidic biochip consists of five different channels separated by four micropillar arrays. There are three central channels for hydrogel formation, one channel for SC culture, and one channel for MN culture. The biochip channels guide MN axon growth from the MN reservoir toward the SC reservoir and allow myelin formation from the SC in the hydrogel. Hesari, Z. [79] developed a hybrid microfluidic platform that generates a dynamic microenvironment by placing neatly aligned PDMS microgrooves on the surface of degradable polymers as a physical guidance cue to control the neural differentiation of hiPSCs. The device consists of a network of micro-channels with 32 cell culture compartments integrated and bonded to a poly lactic-co-glycolic acid (PLGA) coated substrate. The top layer is PDMS and the bottom layer is a glass substrate coated with PLGA nanofibers.

4. Microfluidic Platforms Applied to Axonal Guidance

Axonal guidance is an important research field on nerve regeneration and also has significant implications for the treatment of complete SCI. There are many studies focusing on guiding axon growth in the specified orientation. Microfluidic platforms provide conducive experimental conditions for these studies. There are currently three mainstream approaches to neural guidance based on microfluidic platforms including: topographical guidance, physical guidance and chemical guidance.

4.1. Topographical Guidance

Topographical guidance method is to construct micro-topography in the environment of axon to induce the growth direction of axon tips (growth cones). Li and Folch [80] proposed an axon guidance method based on micro-topographical and biological cues on 3D substrates. The 3D substrate is constructed by using Matrigel and poly-d-lysine (PDL) to coat on PDMS. Directing axon growth by utilizing the preference of neuronal cells for growth cones that tend to grow toward the most favorable substrate in the absence of chemotactic factors. Ristola et al. [81] proposed a PDMS microfluidic chip integrated with a light patterned substrate to achieve isolated and unidirectional axonal outgrowth of human pluripotent stem cells derived neurons. This microfluidic device achieves isolation of the axon from the soma and dendrites and robust growth of the axon toward the adjacent axon region by optimizing the cross-sectional area and length of the PDMS micro-channels. Axon regrowth is guided by the photoinscribed nanotopography on an azobenzene-containing molecular film. The innovative point of this microfluidic device is the integration of nanotopographic patterns with a compartmentalized microfluidic chip, which creates a model based on human neuron that supports superior axonal alignment in isolated microenvironment. In the research of Hesari, Z. et al. [79], PLGA-based nanofibers were electron-pinned and coated on the bottom layer of the chip, shown as Figure 6A. These nanofibers form porous, bead-free, uniform scaffolds that guide axonal growth.

4.2. Mechanical Guidance

Mechanical guidance method is to guide the growth direction of the axon by applying precise forces on the axon, such as pressure and suction. Gu, L. [82] put forward a microfluidic platform taking advantages of flow-controlled axonal guidance techniques to direct axon regrowth by micro-syringe nano-pump and micro-tubing. As shown in Figure 6B, the platform can direct cell culture medium towards neurons to fasciculate one advancing axon onto another. The linear microfluidic flow is induced by a nano-pump into sample chamber at a flow rate of 2.5 μL/min through a micro-tube of 50 μm inner diameter. The lower part of Figure 6B shows the distribution of radial forces on the axon at different locations of microfluidic flow. The highlight of this microfluidic device is using precisely controlled culture medium flow to promote formation of neuronal fasciculation and guide orientation of axonal and growth cone. Nguyen, T. D. [83] established a miniaturized microfluidic platform that can induce parallel axonal growth of neurons in specific directions along well-defined geometries. The device consists of a growth chamber for placing cell bodies and storing growth medium; a vacuum chamber for applying suction to impart tension; and an array of micro-channels connecting the growth chamber to the vacuum chamber. The suction force generated by vacuum chamber is applied to multiple neurons to induce neurite growth. This microfluidic platform uses vacuum stress to guide axon growth.

4.3. Chemical Gradients Guidance

Chemical guidance method uses the gradient of solution concentration to direct the orientation of axon growth. Wang, C. J. [84] designed a microfluidics-based turning-assay chip to against shear stress generated by chemical gradients affecting the migration of axon cone growth. This platform generates a smooth gradient of surface-bound laminin to fine-tune the polarity of growth cone responses to gradients of BDNFs, thereby realizing the guidance of axonal growth cone. The platform consists of a custom microfabricated PDMS chip and a glass coverslip. The PDMS chip consists of two layers. The bottom fluidic layer contains a gradient generation network, a symmetric cell seeding network and a chemotaxis observation chamber. The top layer contains pneumatic control valves which can control the flow in the specified fluidic channels on the bottom layer. The glass coverslip is wet etched with an array of micro-vias. The mechanism of gradient generator is using a serpentine mixing network to split solution of different concentrations introduced from three inlets into nine gradients and pass them into micro-wells connected to micro-channels, where the axons are cultured. The key feature of this platform is using flow network to generate different gradients of solutions to enable the axon guidance. Yin, B. S. [85] designed a pyramid-shaped microfluidic platform capable of analyzing the effectiveness of tacrolimus to neural regeneration. As shown in Figure 6C, the platform consists of an upstream concentration gradient generator and a downstream cell culture chamber. As two kinds of solutions are injected into the device through two inlets, they pass through a serpentine channel where diffusive mixing is diluted and different gradients are formed. In the study of Tang, Y. [54], the special design of asymmetric side channel makes the flow rate stable and adjustable so that molecule gradients can be generated within the 3D collagen. This allows quantitative investigation of growth promoting and regeneration of injured dopaminergic neurons by 3D chemical gradients formed by three natural molecules.

5. Challenges, Future Perspectives and Conclusions

In this review, we introduce the development and recent advances of microfluidic platforms for neural regeneration research. Various microfluidic platforms and devices are classified and discussed according to neural injury models, approaches of axon guidance and mechanisms of neural regeneration such as axon regeneration/repair and neurite outgrowth/regrowth. Here, we outline the characteristics and advantages of cell cultures in microfluidic platform over traditional in vitro culture methods. The technological development from 2D culture models to complex 3D culture models is described in terms of materials and structures.

The microfluidic platforms provide strong technological support for in vitro cell culture research. As seen from the trend of the last three decades, microfluidic platforms are gradually becoming an important tool for the study of neural regeneration. Microfluidic platforms enable accurate simulation of the in vivo micro-environment for different neurobiological studies such as analyzing myelin growth inhibitory factors, screening drugs and NGFs that promote axonal growth, and exploring mechanisms of injury and regeneration. Precise and complex manipulation of neuronal cells can be achieved through microfluidic platforms, such as isolation and manipulation of single neuron, cultured neuron patterning, control of neurite outgrowth and axon guidance. In some studies, microfluidic platforms are integrated with other systems, such as fluid concentration gradient generation devices, optical microscope and millisecond laser, to meet different experimental requirements.

However, microfluidic platforms have not yet revealed their full potential in the field of biological experiments. There are still some challenges and limitations of microfluidic platforms for neural regeneration. First, there are no suitable 3D injury models for complete SCI. Current microfluidic-based 3D culture models cannot closely simulate the complex environment of inside the damaged vertebral canal. Moreover, it is deemed incapable to accurately reconstruct the spatial structure of post-injury spinal cord tracts. It results in nerve regeneration methods that work well at the cellular level in vitro but are ineffective in the in vivo environment. Second, current regeneration studies based on microfluidics are not sufficiently standardized. Numerous nerve injury models and repair approaches have been developed; however, there is not a unified assessment criteria to quantitatively detect and evaluate the efficacy of drugs. Microfluidic nerve regeneration is a highly interdisciplinary research field. Models for neural regeneration study are proposed by biologists; however, microfluidic cell culture platforms often rely on complex external control systems, which involves many physicochemical parameters and variables. This leads to different standards and parameters involved in similar experiments. Finally, microfluidic platforms can greatly mimic the in vivo micro-environment and provide ideal experimental conditions for nerve regeneration. However, the study of nerve regeneration still remains in vitro, especially in CNS regeneration studies. There are many technical obstacles to progress from in vitro to in vivo studies.

We expect that microfluidic platforms will be more widely used, not only in the field of neural regeneration, but also in all cellular and neurobiological research. Although, 2D cell culture are currently widely used and many breakthroughs are made on 2D culture microfluidic platforms. The 3D culture microfluidics have a much greater potential for future cell regeneration research. More microfluidic platforms for 3D culture will be developed to break through the limitations of existing 3D culture models and better mimic the 3D environment in vivo. Since microfluidics itself is an interdisciplinary subject involving many fields such as micromechanics, fluids, physics, materials, chemistry and biomedical engineering, it is reasonable to predict that there will be more new technologies and applications that will be integrated with microfluidics, such as artificial intelligence or new material. The integration of new technologies could overcome the barriers of existing problems and provide new solutions for research on neural axon regeneration and neurite outgrowth/regrowth. The advancement of microfluidics will continue to facilitate research, such as probing neural mechanisms, selecting targeted signaling pathways and screening drug molecules to promote regeneration of the CNS or block inhibitory signals from injured neurons. Furthermore, the future development of microfluidic platforms is promising to explore effective SCI rehabilitation and treatment, as well as CNS clinical treatment options.