3D printing of functional bioengineered constructs for neural regeneration: a review

Three-dimensional (3D) printing technology has opened a new paradigm to controllably and reproducibly fabricate bioengineered neural constructs for potential applications in repairing injured nervous tissues or producing in vitro nervous tissue models. However, the complexity of nervous tissues poses great challenges to 3D-printed bioengineered analogues, which should possess diverse architectural/chemical/electrical functionalities to resemble the native growth microenvironments for functional neural regeneration. In this work, we provide a state-of-the-art review of the latest development of 3D printing for bioengineered neural constructs. Various 3D printing techniques for neural tissue-engineered scaffolds or living cell-laden constructs are summarized and compared in terms of their unique advantages. We highlight the advanced strategies by integrating topographical, biochemical and electroactive cues inside 3D-printed neural constructs to replicate in vivo-like microenvironment for functional neural regeneration. The typical applications of 3D-printed bioengineered constructs for in vivo repair of injured nervous tissues, bio-electronics interfacing with native nervous system, neural-on-chips as well as brain-like tissue models are demonstrated. The challenges and future outlook associated with 3D printing for functional neural constructs in various categories are discussed.

These authors contributed equally to this work. * Authors to whom any correspondence should be addressed.

Introduction
The nervous system is one of the most intricate biological structures and plays critical roles in the communication and interaction between the body and environmental changes [1]. The CNS which consists of the brain and spinal cord is the most complex and main part of the nervous system; the PNS which originates from the CNS and extends to other parts of the body can coordinate and interact with CNS to govern human behavior [2]. Once the nervous system is damaged due to acute physical trauma, ischemic or chemical factors, the connection of motor, sensory, and autonomic neurons to and from the neural circuit would be interrupted, leading to paralysis, organ dysfunction and death. From a cellular level, neurons, oligodendrocytes, and other components essential for neuronal transmission are physically insulted, and the disrupted vascular components can induce the infiltration of inflammatory cells [3]. The low regeneration capacity of the nervous system on account of the damage of the neuronal cells and the inhibitory microenvironment requires efficient interventions to re-establish and repair functional neural connections, nervous tissue, and cells [4,5]. Transplantation of autologous nerve graft is regarded as the golden standard for the treatment of severe nerve injury [6,7]. Nevertheless, insurmountable disadvantages such as limited sources of donors and donor nerve mismatch exist all along in the application of autologous nerve grafts [8].
Previous studies have shown the great potential of tissue engineering strategies as possible therapeutic methods to accomplish neural regeneration by constructing TENGs [8]. TENG is often fabricated to be a biomaterial-based scaffold seeded with cells and NTFs, which could not only offer physical structural support for nerve growth but also further enhance the therapeutic effect of recapitulating and modulating the microenvironment for nerve regeneration [9,10]. An ideal TENG should meet the demands including good biocompatibility, appropriate biodegradability and mechanical properties, good permeability to nutrients and surface activity [11,12]. In addition, TENGs are expected to direct axonal growth through the damaged region of nerves and rebuild the nerve conduction pathway by the 3D-structured guidance 'bridge' scaffolding [8]. Accordingly, nerve grafts with different configurations such as hollow conduits [13], filament-reinforced conduits [14] and multi-channel conduits [15] have been constructed by many fabrication strategies, including gas foaming/particulate leaching [16], microengineering based on lyophilization [17], and fiber engineering [18]. However, it is challenging to accurately imitate the complex nervous system via conventional molding techniques, especially when the placement of materials and cells are required to be orderly designed considering the various anatomical structure in neural tissues.
In recent years, the 3D printing technique has been applied in neural autograft engineering and revolutionized TENG manufacturing in a new era. The 3D printing is an additive manufacturing process whereby layers of materials are built up to create a 3D solid part from a CAD model. It provides incomparable advantages to overcome the challenges in producing TENGs for neural regeneration which are not possible to be conquered by traditional fabrication techniques. The significant superiorities of 3D printing include: (1) designing personalized scaffolds with anatomical accuracy to match individual injury site combing with 3D imaging tools; (2) precise printing and positioning of cells or biomolecules in desired localization to accurately recapitulate complex cellular matrix in the nervous system; (3) controllable manufacturing process with a high printing resolution in micro/nanoscale to accurately mimic the 3D architectures of fibrous decellularized ECM of nervous tissues; (4) integrating multiple components (e.g. materials, cells, biomolecules) into a single printed construct to realize the concordance of the nervous system and its surrounding tissues such as vessels and mussels. All those outstanding features provide an unprecedented engineering methodology for reproducing the complex neural matrix over conventional approaches. The biomaterials selected for 3D printing of neural constructs include biocompatible polymers [19,20], composites [21][22][23] and hydrogels [24,25] which should satisfy the specific printability and biocompatibility, appropriate physicochemical and mechanical strength. Despite the promising investigations about the advantages of 3D printing in achieving neural regeneration, many of those fail to fully capture physiologically relevant nerve-mimic functions and this has triggered the development of functional TENGs which could be of critical significance for clinical applications. In the past five years, several studies have been published that reviewed the current technologies and materials for 3D printing of the neural-engineered scaffolds/-devices, and the mechanisms for repairing the neural damages by 3D-printed biomimetic constructs, but mainly on the repair and regeneration of peripheral nerves [26][27][28]. Besides, limitations remain to be addressed regarding the integration strategies encompassing 3D printing and modulation cues for the functional recovery of nervous tissues.
This review intends to summarize and analyze the development of current 3D-printed functional bioengineered constructs with desired functions considering the practical demands for nervous circuit reconstruction (figure 1). We first examine the emerging 3D printing techniques for fabricating the tissue-engineered neural scaffolds and cell-laden constructs, discussing the major advantages and limitations of each route. Following this, we explore the current and future steps toward mimicking salient functionalities of neural tissues through the integration of critical cues including the creation of hierarchically guided structures, biologically modulated microenvironment, and electroactive matrix capable of transmitting electric signals with 3D printing technology. Analysis of these existing artificial grafts based on functional design principles will pave the way for appropriate design specifications for constructing neural tissue devices. We then summarize the recent applications of those 3D-printed functional neural constructs and devices in vitro/vivo. Finally, we present a perspective on the potential of 3D printing to develop better neural devices, particularly focusing on function recovery.

3D printing techniques for fabricating neural tissue-engineered scaffolds or cell-laden constructs
The 3D printing, as a biofabrication technology capable of constructing complex and controllable 3D structures, has emerged as a powerful manufacturing strategy to address the structural challenges of neural tissue-engineered scaffolds. It enables precise deposition of multi-materials and even cells based on pre-designed patterns, which provides highly controllable physicochemical cues for neural cells such as porosity, mechanical properties, and bioactive molecule concentrations or gradients [48]. Currently, various 3D printing technologies have been used to construct well-defined implants for in vivo nerve injury repair and models or devices for in vitro nervous tissue engineering, including extrusionbased printing, light-based printing, EHD printing, embedded printing and 4D printing [5]. The overview of printing techniques for functional nervous tissue-engineered constructs is summarized in table 1.

Extrusion-based printing of neural tissue-engineered scaffolds or cell-laden constructs
Extrusion-based printing is the most widely used 3D printing technology in nerve tissue engineering, in which continuous material is extruded from a syringe needle and deposited layer-by-layer to form a 3D structure under computer control [59,60]. A successful and continuous printing process mainly relies on the coordination of a motion control system and controllable extrusion of inks by pneumatic, piston or screw  [29], Copyright (2020), with permission from Elsevier. Reproduced from [30]. © IOP Publishing Ltd. All rights reserved. Reprinted from [31], Copyright (2018), with permission from Elsevier. [ [42] with permission from the Royal Society of Chemistry. Reproduced from [43] with permission from the Royal Society of Chemistry. Reprinted by permission from Springer Nature Customer Service Centre GmbH: Nature, Nature Physics [44], Copyright (2016). Reprinted from [45], Copyright (2015), with permission from Elsevier. Reprinted from [46], Copyright (2022), with permission from Elsevier. Reprinted with permission from [47]. Copyright (2020) American Chemical Society. 6 driving (figure 2(a) (i)) [61]. After the material is deposited on the receiving substrate and cured, the above process is repeated until the 3D structure is assembled. A wide range of materials including molten/semi-molten polymers (e.g. PCL [62]), polymer solutions (e.g. polylactic acid [63]), hydrogels (e.g. GelMA [24], HA hydrogels [49], collagen hydrogels [25]) are compatible with extrusion-based printing of neural constructs corresponding to specific rheological requirements based on different solidification mechanisms including crystallization, chain rearrangement, recovery of non-covalent bonds, and chemical cross-linking under particular conditions such as UV irradiation, temperature changes or ion crosslinking [64]. Extrusion-based printing provides a flexible design strategy for composing controllable architecture by using multi-type materials as inks, which enables the fabrication of customized implants and devices exhibiting tunable porosities, desired mechanical properties, and diverse functionalities. For example, the porous polyester nerve conduit scaffolds were constructed by using extruded printed multilayer oriented linear 3D structures as perfusion molds, to provide topographyguided cues for the directional regeneration of axons [29]. In addition, cellular secretions or NTFs were added to the extruded inks to improve the microenvironment for neural tissue growth and the functionality of the printed structures [25,49,68]. For example, a 3D porous scaffold prepared by a collagen-chitosan hybrid material encapsulated with secretome-secreted MSCs was reported to not only promote nervous fiber regeneration and vascular remodeling but also regulate the levels of systemic inflammatory factors [68].
In order to further arrange cell locations precisely for better mimicking the nervous tissue structure [24,69], hydrogels with low modulus composed of living cells are extrudedprinted together to form artificial structures that imitate natural nervous tissues, which is known as 3D bioprinting [50]. Due to the accurate spatial distribution of cells and materials in the bioprinted scaffolds, optimal axon reconnection would be achieved. The preferred types of cells encapsulated are stem cells such as BMSCs [51] and NSCs [70] due to their differentiation potentials, while other neural cells like neurons [71], gliocytes and RSCs (figure 2(a) (ii-iv)) [30] were reported to be embedded as bioinks as well. It has been reported that the bioprinted NSC-laden scaffolds can offer a benign microenvironment to facilitate cell-material interactions and neuronal differentiation for neural network formation compared to the printed scaffolds without cells [72]. In vivo experiment has also been implemented to further demonstrate the effect of bioprinted scaffolds on promoting axon regeneration and decreasing glial scar deposition, which shows better performance on locomotor recovery of the SCI rat model. The complex brain-like structures were able to be engineered by extrusion-bioprinting methods. By carefully designing the layered structure and selecting the bioinks, a superior artificial culturing environment can be created for the growth of neural cells [45,73].

Light-based printing of neural tissue-engineered scaffolds
Nervous systems have complex geometries which require high manufacturing accuracy of the desired neural tissueengineering scaffolds, and this is difficult to realize with traditional extrusion-based printing. Light-based printing technology can break the limitations of printing resolution to about 50-200 µm [74] and further improve the structural control ability to form various complex shapes for optimal geometric control of neural tissues (table 1) [52]. In the light-based 3D printing process, liquid photosensitive polymers are solidified when exposed to laser or UV light to build 3D models [75,76]. Depending on the scanning and curing method, lightbased printing has developed from SL to DLP and µCPP.
SL equipment typically consists of a UV laser source, a reservoir filled with liquid resin, a system that controls the beam movement in a horizontal direction, and a printing platform that allows for vertical shift (figure 2(b) (i)) [65]. Before curing, the printing platform is located below the liquid level of the resin reservoir, and the layer thickness from the platform to the liquid level can be controlled by computer. The liquid resin is solidified rapidly point by point due to covalent crosslinking as soon as exposed to the UV light following the preset scanning route of the laser spot [65]. After finishing the solidification of one layer, the platform will drop the height of another single layer for scanning, so that the 3D construct can be formed by repeating the cycle.
The SL method is effective in building complicated nano-scale scaffolds and thus promoting seeded cells to differentiate into neuronal subtypes [77]. For example, SL-printed DA-functionalized GelMA scaffolds with highly porous and interconnected 3D environments were shown to improve neural differentiation (figure 2(b) (ii-iv)) [66]. Besides, the inks applied for SL were recently expanded to electroactive materials and even bioinks with living cells. As reported by Heo et al [53], neuronal differentiation could be enhanced in an SL-printed conductive PEDOT:PSS hydrogel scaffolds, by transferring ES toward encapsulated DRG cells systematically. However, the long printing time of SL largely limits its application in neural tissue engineering [59].
DLP has emerged as a more efficient light-based printing strategy with increased printing speed and output capabilities compared to SL [74,78]. DLP technology uses spatial light modulating elements, such as LCD or DMD, to project a 2D pixel pattern onto a liquid resin reservoir allowing a complete layer of resin to be cured simultaneously (figure 2(c) (i)) [65], which is different from the point-by-point solidification of SL [79]. Consequently, the printing time of a single layer by DLP can be greatly reduced and is only determined by the required exposure time and layer thickness. Moreover, the printing resolution of DLP can be improved to about 1 µm depending on the pixel size of the digital pattern. Because of its high precision and efficiency, DLP is widely used to fabricate 3D neural constructs in a controllable manner, such as the multi-channel nerve conduit supporting directional migration of nerve cells (figure 2(c) (ii-iv)) [54], and drug-loaded nerve conduits enabling continuous drug release to promote axonal extension [35,80,81]. However, DLP is still limited in printing speed due to the lift-up step of the print head in which the resin needs to flow back after each printed layer [82]. Meanwhile, photochemistry is urgently needed to enhance the versatility of DLP for neural tissue engineering [83].
µCPP technique was created recently which allows the use of various biomaterials and cells to create complex 3D architectures in the continuous printing process with high resolution and speed. It has been selected for constructing the butterflyshaped spinal cord structure by mimicking the distinguished regions of white and grey matter and the scaffold can be fabricated in 1.6 s [55], which can better improve axon regeneration and the formation of new 'neural relays' across SCI sites in vivo. Nevertheless, the materials suitable for µCPP technique are very limited especially when neural cells need to be encapsulated into the bioinks. Unwanted diffusion needs to be reduced by optimizing precursors and light scattering should be eliminated to further improve the printing resolution [83].

EHD printing of microscale neurobiomimetic scaffolds
An ideal neurobiomimetic scaffold should mimic the fibrous nerve ECM with preferred cellular orientation, which is crucial for guiding cellular behavior leading to tissue formation for neural regeneration [84]. EHD printing is developed by combining the principles of EHDs and layer-by-layer additive manufacturing and provides an ideal strategy for the fabrication of microscale neuro-biomimetic architectures of nervous tissues because of its capability to controllable deposition of micro/nanoscale filaments [85][86][87][88]. Compared to extrusionbased printing equipment, the EHD printing system commonly has an additional high-voltage generator [89][90][91] which creates a high-voltage electric field from the nozzle to the collecting substrate, enabling the formation of a Taylor cone of the inks at the liquid-air interface (figure 2(d) (i)) [67,92,93]. Once the electric field force at the tip of cone exceeds the surface tension of the solution, a continuous jet or droplet can be generated and deposited on the moving collecting substrate to form desired structures. Various biopolymers can be used in EHD printing [94] as summarized in table 1, among which PCL is most commonly selected due to its proper viscosity and relatively stable melting point [95][96][97][98]. It has been reported that the patterned PCL microstructures prepared by EHD printing allow SCs to migrate and extend along the fibers, thereby promoting directional regeneration and myelination of axons (figure 2(d) (ii-v)) [56]. Furthermore, EHD-printed mesh-like microfibers have been applied to improve the mechanical properties of a spinal cord cell culture model with an ultrasoft matrix to support the continuous culture of neural networks in vitro [57].

Embedded printing of soft neural constructs derived from biologically-relevant hydrogels
Preferred neural constructs require low young's modules and high shape fidelity at the same time due to the soft nature of nervous tissues (neonatal brain tissue 110 Pa and adult brain tissue <1 kPa, 90-230 kPa for spinal cord) [31]. However, traditional extrusion-based printing by using low-viscosity materials is prone to collapse and difficult to fabricate soft constructs [99], and light-based printing is largely limited by the selective range of biocompatible resins. Embedded 3D printing has emerged as an optimal strategy to further expand the options of low-viscous inks with elastic moduli of <100 kPa [100] or even ECM-based bioinks with the viscosity of 2.80 Pa s [101] to be suspended in a support bath, which facilitates the establishment of complex neural networks closer to the mechanical strength of brain tissue. In the embedded 3D printing process, the collecting substrate would be converted into a support bath with appropriate rheological or mechanical support properties, and the stable interaction between the support bath and ink ensures the controlled shaping of the printed structure. For example, Kajtez et al [23] constructed a neural model in vitro with a tunable microenvironment and defined spatial arrangement by directly embedding human NSCs into self-healing annealed particle-ECM composites for embedded 3D printing (figure 3(a) (i)). The support material not only served as physical support for embedded printing, but also provided a cell-interactive microenvironment for active cellar activities such as growth and maturation. This printing technology enables accurate patterning of extremely soft brain models and has broad prospects in the functional modeling of mechanosensitive neural structures (figure 3(a) (ii)).

4D printing of stimulus-responsive neural tissue constructs
A critical limitation of 3D printing is the static nature of the fabricated materials and the resulting constructs, causing difficulties in implantation and surgical suture. By integrating smart materials into the initial form of 3D-printed structures [103,104], 4D printing is developed to produce stimulus-responsive neural scaffolds which can deform and evolve in an expected manner when exposed to environmental stimuli (e.g. electric field [105], temperature [106,107], and chemical reaction [108,109]) or through human interference, causing a time-dependent change in shape or physical properties [110,111]. The precise control of this stimulation process facilitates the construction of multifunctional integrated stimulus-responsive neural scaffolds with physical guidance, dynamic self-regulation and seamless integration. For instance, a self-forming multi-channel NGC was constructed with topographical cues to guide cell growth based on a temperature-intelligent responsive shaping process of shape memory polymers (figure 3(b) (i)) [102]. Uniform cell loading can be achieved on the nerve guidance conduit with a planar shape followed by automagical transformation into a tubular shape triggered by a temperature rise to 37 • C (figure 3(b) (ii)) [47]. The rat sciatic model shows its superior performance in terms of cell orientation and nerve defect repair (figure 3(b) (iii, iv)). Miao et al [58] developed a graphene hybrid nerve conduit structure by SL-based 4D printing. The transformation of the structural shape was accomplished by light-induced graded internal stress changes within the structures, then it can relax induced by a subsequent solvent. Deep research on neural-cell maturation and functionality requires smart-neural tissue constructs, and 4D printing with shape programming ability may provide a way to respond to various neural tissue microenvironments and can evolve with environmental changes.

Integrating regulation cues into 3D-printed constructs for functional nerve regeneration
Within the past decade, there has been a growing number of researches focusing on 3D printing techniques and materials for the creation of intricate architectures and continuously improving resolutions for neural tissue engineering. However, successful nerve injury repair and neural function recovery require efficient neural substitutes that should not only satisfy the geometry characteristics of injury sites, but also faithfully reproduce the microenvironment of nervous tissue, accelerating the formation of functional nervous tissue analogues [112]. Therefore, 3D-printed constructs combined with additional regulation cues including topographical, biochemical and electroactive stimuli have been explored and tuned to effectively enable a wide range of approaches to influence neural cell response and drive functional nervous tissue maturation.

3D-printed functional bioengineered constructs with topographical cues for directing neural regeneration
Severe nerve damage would lead to intergap scarring within less efficient and more disordered axon regeneration, resulting in fewer axons reaching distal sensory or motor targets [113]. Although nerve autografting has been used for treating nerve transection as the gold standard procedure [114], the grafted nerve contains thousands of directionally linear basal lamina endoneurial tubes and would result in the loss of direction specificity of the regenerated axons and subsequent poor functional recovery [115]. For non-autografts, hollow tubes are simple in structure and easy to manufacture, and have been maturely used in the clinical neural restoration of short distances and small diameters. Whereas it is difficult to guide axons elongation in hollow tubes and axon scattering, incompatible nerve reinnervation or even multi-innervation cannot be avoided [116]. Therefore, it is imperative to construct NGCs with surface topographies inside to conduct axonal extension and decrease axonal dispersion, thus improving the accuracy and efficiency of nerve repair significantly.
As one of the most attractive manufacturing technologies, 3D printing provides tissue engineers with a way to design highly-ordered scaffolds capable of improving the spatial orientation of neural cells and fast reconnection of the damaged nervous pathway [56]. Via 3D printing strategies, individualized scaffolds containing various topographic cues can be built for different types and degrees of nerve damage to accurately match defects, enormously promoting the development of functional conduits for neural regeneration. So far, a variety of topographic patterns have been developed and applied in neural tissue-engineered implants within the effectively directive function.
Scaffolds with microchannels are produced to not only facilitate nutrients and oxygen exchanges, but also efficiently prevent fibrous scar tissue invasion to offer an unperturbed environment for neural regeneration [117], leading to the improvement of functional recovery of nerves [118]. As reported previously, porous multi-channels polyester scaffolds at high porosity (>90%) (figure 4(a) (i, ii)) were fabricated inside sacrificial constructs, within a controllable and aligned topography for guiding axon regeneration [29]. It was found in the in vivo experiment that the axon fascicules originated from cell clusters and extended in the resulting scaffolds, demonstrating highly ordered arrangement parallel to the microchannels direction, which was very different from thorough growth isotropy displayed by regenerating axons in the spinal cord transection injected with fibrin hydrogel (figure 4(a) (iii, iv)). In another study, multichannel GelMA NGCs with different inner diameters were fabricated by DLP or projection-based 3D printing [119], and it was found that larger inner diameters are more conducive to the longitudinal migration of cells along the channels [54,120]. Differ from regular and uniform microchannels, Li et al [121] designed cryogenic 3D-printed collagen/silk fibroin implants with four irregular channels, the biomimetic internal microarchitecture of which precisely simulated the spatial microstructure of the spinal tracts, presenting fewer lesions and disordered structures of the injured spinal cords than those prepared by freeze-drying technology by HE analysis. By µCPP, Koffler et al [55] created a bionic scaffold with 20 µm-diameter microchannels loaded with NPCs and implanted in the disrupted spinal cord of rats. It was found that the host axons in scaffolds recovered linearly resembling the axonal arrangement of the intact spinal cord.
Contact guidance by grooves/ridges structures is known as another beneficial cue assisting the directional alignment and migration along the anisotropic direction. It has previously been reported that the orientations of axons and SCs, which are the two main components of regenerating peripheral nerve, can be evidently affected by microgrooves [122,123]. Blake et al [32] demonstrated a one-pot 3D printing approach based on a microextrusion-system for generating custom NGCs with luminal microgrooves which are axially oriented in the radial dimension (figure 4(b) (i)). The microgroove architectures were measured as shown in figure 4(b) (ii) and qualitatively resembled naturally occurring physical cues presented in degraded nerve pathways. Figure 4(B) (iii, iv) and (v, vi) show that both the dissociated SCG neurons and SCs cultured on the 3D-printed physical cue can form alignment in the vicinity of microgrooves. In another study, Lee et al [124] combined extrusion printing technique with photocrosslinking system to print arrayed gelatin hydrogels with adjustable expansion coefficients, establishing micro-grooved surface patterns inner PLCL nanotubes. The morphology acted as an effective direction for axonal stretching which can enhance the adequate connection of the nerve through the NGC. Except for conventional extruding method, SL as a distinguished 3D-printing technique in biomedical engineering can be used to build furrow patterns by sculpturing the tubes themselves [32,125]. It was reported that NGCs with topographical grooves produced by SL considerably increased the directionality of neurite growth [126]. In addition, dryjet wet extrusion was also adopted to fabricate NGCs with microgrooves by modifying the inner tube of a spinneret into a 12-grooved type [127], which largely helped the parallelized axonal propagation of PC12 cells. Multi-grooved patterns provided physical limitation for neurite outgrowth, which cultivated a neurites morphology of longitudinal alignment along the wall.
Another way to guide axonal growth direction is to integrate microfibrous patterns inside NGCs. It has been reported that parallel microfibers inside neural scaffolds can increase the proportion of pro-healing phenotypes in activated macrophages, boost the proliferation of SCs and improve the expression of related genes, leading them to migrate in specific directions and subsequent oriented axonal growth [128][129][130]. The combination of parallel electrospun microfibrous flats and 3Dprinted hydrogel supports can be the most common strategy for the manufacture of fibers-integrated scaffolds [131]. For example, Lee et al [132] embedded electrospun aligned PCL/gelatin microfibers in stereographically porous hydrogel scaffolds and demonstrated that the neurite orientation of primary cortical neurons cultured with these fibers was highly consistent, extending along the direction of fiber arrangement. The unidirectional microfibers provided an obstacle-free path to the growing axons, performing satisfactory neuroregeneration. Furthermore, advanced micro/nano printing techniques such as NFEP and EHD printing have been developed in recent years as discussed in section 2 to realize the fabrication of single fiber with high positioning resolution [133,134]. In this way, Wang et al [36] built rGO-encapsulated PLCL microfiber templates with a certain interfiber overlay angle (90 • ) and adjustable fiber diameters (from 15 to 150 µm) (figure 4(c) (i)). It was found that as the diameter of the PLCL fibers raised, the axonal expansion was guided from preferred alignment to gradual disorder (figure 4(c) (ii)). A python-based analysis in figure 4(c) (iii) was carried out and indicated that on the fibers with larger diameters (72 and 150 µm), the neurite distribution was more diffused with a high inclination to branch and wind around fibers. In comparison, on small-diameter nanofibers, the neurite had fewer branching and more localized protraction to enhance neuron elongation. Besides, Vijayavenkataraman et al [135] observed that synapses expanded along parallel fibers in porous PCL scaffolds manufactured by EHD printing, whereas they grew randomly in electrospun PCL scaffolds. Liu et al [84] used EHD jet printing, dip-coating and electrospinning to fabricate triple-layered scaffolds. The inner layer of PCL printed by EHD jet presented a uniform porous structure with directional fibers and high porosity, and regular directional grooves were built between adjacent filaments, inducing cells to grow orderly along the aligned fibers.
Except for the traditional topographically engineered scaffolds printed by various methods, 3D constructs printed with living cells are more challenging and promising for meliorative neural injury treatment, considering the additional guiding functions [136]. By sequential point-dispensing, Joung et al [33] successfully embedded OPCs and sNPCs into a multi-channel hydrogel scaffold (figure 4(d) (i)), assembling an NGC architecture with oriented-cell design. Analysis of representative sections implied the longitudinal distribution of cells throughout the channels, exhibiting clear adherence, survival and differentiation states (figure 4(d) (ii-vii)).
The 3D printing technology provides a more convenient and beneficial strategy that can be applied to manufacture different morphological NGCs to guide neural cell proliferation and migration. The integration of microstructures such as multi-channel, grooves/ridges, and fibrous pattern with biomaterials and even living cells by 3D printing are potential to build functional NGCs for rapid recovery of injured nerves.

3D-printed functional bioengineered constructs with chemical cues for enhancing neural regeneration
In addition to topographical guidance, immobilization of the biological or chemical cues would create a permissive microenvironment to accelerate neural rehabilitation [8]. The controlled release of bioactive substances (e.g. NTFs, drugs, proteins, inhibitors, cell-derived bio-products) in neural constructs can enhance the bridging of gaps between nerve stumps and the functional restoration of rebuilt neural tissues. Numerous studies have indicated that the dose, concentration gradient, synergistic effect and release kinetics of exogenous curative molecules delivered by neural constructs affect the regenerative capacity and rate of injured nerves [32,[137][138][139]. Combining with 3D printing technology, more functional TENGs can be created by loading with therapeutic agents which may eventually be more closely recapitulated in the neural regeneration microenvironment to facilitate successful neural injury repair. Currently, the neurotrophic biomolecules are normally introduced into a 3D-printed system in four ways: physical impregnation, integrated printing with chemicals, grafting to the support materials by chemical bonds, or loading nanoparticles encapsulated with restorative molecules.
A common procedure to immobilize chemicals into TENGs is submerging TENGs prepared beforehand in a solution with dissolved NTFs [140]. Mishchenko et al [49] placed extrusionprinted HA-based hydrogel scaffolds in an aqueous solution of growth factor to adsorb the NTFs GDNF and BDNF. It was revealed that the gradual release of BDNF and GDNF can activate the spontaneous calcium activity of primary hippocampal cultures, and thus promote the modulation of synaptic transmission and activate the functional activity of neuronglial networks. In another study [141], aFGF was grafted on the microporous/microgrooved chitosan-grafted PLA nerve conduits by immersing the plasma-treated scaffolds into an aFGF solution for further increasing the loading rate, and this aFGF-immobilized conduit was shown to be more beneficial to myelinated axons regeneration compared to the aFGF-free counterpart. More recently, the secretome derived from stem cells was found to play a critical role in neural regeneration because it is composed of growth factors, cytokines, vesicle parts of exosomes and microvesicles which can benefit neural damage repair significantly [68,142,143]. Chen et al fabricated 3D-printed collagen/silk fibroin scaffolds carrying MSCs secretome by soaking the scaffolds in secretome solution [142]. Those secretome-modified scaffolds could facilitate nerve fiber regeneration, enhance remyelination and accelerate the establishment of synaptic connections at the injury site compared to the unmodified scaffolds. However, this impregnation method often has poor adhesion and uneven distribution of molecules on the scaffolds, inducing inadequate or uncontrolled release [144].
In the process of integrated 3D printing, biomolecules as a part of ink compositions can be programmed and printed to control the manufacturing route precisely within a well-designed distribution inside TENGs, creating haptotactic gradient surfaces in scaffolds [26]. Consequently, scaffolds can control-release NTFs and attenuate the burst release phenomenon in vivo. For example, Huang et al [34] fabricated a customized porous scaffold with NGFs gradient by printing the silk fibroin/collagen solution together with NGF using double-head printing systems (figure 5(a) (i-iv)). The ingenious G-NGF contributed to axonal arrangement and reduced axonal scattering, thus possibly enhancing target reinnervation and later recovery of motor function (figure 5(a) (v)). Another report utilized low-temperature extrusion 3D printing technology to actualize the conjugation of BDNF directly to collagen/chitosan scaffolds, and retained the biological activity of BDNF to the maximum extent [25]. Similar to bioactive factors, other small molecules such as inhibitors or hormones can be added to the raw material during the 3D printing process as well [145,146]. For example, OSMI-4 as a small molecule OGT inhibitor has been successfully induced into a supramolecular bioink to create an NSCladen spinal cord-like neural scaffold [146]. The sustained delivery of OSMI-4 could be achieved and can remarkably enhance the intrinsic neuronal differentiation of the encapsulated NSCs by inhibiting notch signaling pathway. Recently, a 3D-printed injury-preconditioned secretum/collagen/heparan sulfate scaffold was fabricated for traumatic brain injury repair by mixing HUCMSCs secretome and injury-preconditioned secretome with collagen/heparan sulfate as hybrid bioinks for low-temperature printing. The rat traumatic brain injury model showed that the scaffolds loaded with injury-preconditioned secretome were in favor of the regeneration of nerve fibers, synaptic structures, and myelin sheaths of the endogenous neurons, as well as the reduction of apoptotic response [147]. Apart from secretum, other cell-derived bio-products such as extracellular vesicles are emerging as potential therapeutic mediators and hold great promise for advancing neural regeneration, which can be combined with 3D printing technology for functional neural restoration in the future [148,149]. Those results hinted that 3D printing of biomaterials encapsulated with a wide range of bioactive molecules can create multibiochemical gradients during fabrication and release them through an ECM degradation-based mechanism [32,124].
Despite the successful delivery of chemical cues by integrated 3D printing, the stable immobilization of drugs/NGFs with sustained release is still challenging by the simple physical mixing method. Recent research tried to compound the chemical chains of delivered molecules with matrix polymers through chemical reactions to obtain a steady exitance of the combination. For instance, Chen et al [18] created a spatiotemporal delivery aligned fibrous hydrogel by establishing a 'middle-to-bilateral' SDF1α gradient on the hydrogel. SDF1α was immobilized onto the scaffold both by chemical bonding inside nanofibers with the cross-linkers and EHD jet printing technique outside the scaffolds ( figure 5(b) (i)). As expected, the sequential release of SDF1α from the gradient hydrogels engendered a more substantial proliferation of NSPCs to the center compared to the non-gradient counterpart (figure 5(b) (ii)), coming to a conclusion that SDF1α/paclitaxel spatiotemporal release had a positive influence on NSPC migration on the aligned fibrous hydrogels.
Apart from the direct grafting of chemicals to scaffolds, introducing biodegradable chemical-encapsulated nanoparticles into TENGs for local release is a potent strategy to accelerate peripheral nerve repair [150,151]. Tao et al [35] fabricated a hydrogel matrix pipeline comprising GelMA hydrogels and XMU-MP-1-loaded poly (ethylene glycol)poly (3-caprolactone) (MPEG-PCL) nanoparticles by DLPbased technology (figure 5(c) (i)). It was revealed that XMU-MP-1 as a small-molecule inhibitor, could facilitate SCs migration (figure 5(c) (ii)) and obvious enhancement of the YAP nuclear translocation could be found, contributing to the up-regulation of YAP target genes (figure 5(c) (iii)). Similarly, Zhang et al [81] used DLP technology to fabricate a biodegradable SAB consisting of a rectangular bandage layer N 3 -GelMA combined with an XMU-MP-1-encapsulated nanoparticle grating layer. The nanoparticles touched the damaged nerve and continuously released the drug for nerve rebirth, and the hydrogel outer layer effectively controlled the direction of drug delivery and reduced the drug concentration in the surrounding tissues. Those results indicated that by integrating DLP with nanotechnology, the nerve conduits profitably preserved functional drugs and slowed the drug delivery in local tissues. Furthermore, it must be noted that the detailed structures of 3D-printed nerve tubes should be finely designed to enhance accommodation for microspheres, which would also have a considerable impact on drug loading and release [152]. Lee et al [153] embedded core-shell nanoparticles loaded with NGF in SL-printed nerve scaffolds which had alterable porous structure, establishing a sustained bioactive agent delivery platform. The pore interconnectivity allowed for NGF diffusion and reinforced cell attachment and suitable porosity improved the growth rate of SCs. In contrast with the random orientation of PC12 neurites cultured on blank control scaffolds and NGF-sprayed scaffolds, neurites inclined to extend along the pattern on nanoparticle-immobilized scaffolds.
The 3D printing technology is capable of printing specific biomolecules in a regional arrangement to mimic the native microenvironment of nervous tissues. Exquisite control over these biochemical cues could ameliorate the highly ordered structures, which can be beneficial to the re-establishment of neural networks and connections.

3D-printed functional bioengineered constructs with electroactive cues for restoring neural function
Biochemical and physical cues as discussed above are both essential to reconstruct defective nervous tissues by providing specific scaffolds to achieve a mimic microenvironment. Alongside those well-known cues, the electrically conductive matrix is another non-ignorable element, as the nervous system has the most active electrical activity among the tissues/organs in the human body which transmits signals from neurons to the targets by synapses. The electrically active TENGs would help transmit electrical signals between the proximal and distal ends of injured neurons after implantation in the body, thus improving the repair process of injured nerves. Furthermore, an electroactive microenvironment is in favor of long-term electrical signal exchange by ES, which could modulate the function of channels that are affected by voltage in excitable neural cells and accordingly affects neurons discharge at a physiological level. ES has been regarded as an efficient and accurate way to guide the orientation, migration, and differentiation of desired cells at a molecular level and provide a neural signal connection for cell interactions.
The spatial 3D printing setup makes it possible to build 3D electroactive matrix that supports both structural sustaining microenvironments and cooperative electrical signaling guidance at the same time. The well-defined architectures with simultaneously incorporated electrical cues and even nanotopographical or biochemical (regulatory factors) cues can be the most promising therapy to direct the fate of stem cells for neural tissue engineering. Electrically conductive biomaterials used for 3D printing usually include polymers mixed with conductive inorganic nanomaterials such as graphene, carbon-nanotubes and Mexene, or conducting polymers such as PPy, PT, and PANI. Recent research has shown the positive effects of the conductive scaffold on the behavior of neural cells cultured on them. Wang et al [36] designed a conductive microfiber scaffold using PLCL encapsulated with rGO (figure 6(a) (i, ii)). It can be seen from figure 6(a) (iii-vii)) that the ES intensity of 100-150 mV was the most effective for the networking of neural cells and the growth of axons, and the growth rate of neural cells was closely related to the intensity of ES (figure 6(a) (viii)). In another study, PCL/rGO scaffold was produced by EHD jet printing and could largely improve the in vitro neurite elongation of PC12 cells [21]. Uz et al [154] fabricated a gelatin graphene catheter by both aqueous gel casting and 3D printing methods for neural engineering. By comparing the microstructures, conductivity and biological effects of catheters shaped by the above two ways, it was found that although there were no big differences in the structure and electroconductibility, the 3D-printed gelatin graphene scaffold had a stronger role in promoting mesenchymal stem cells (MSCs) differentiation due to the intenser electric field generated inside. In addition, those conductive nanoparticles are usually surface-modified to avoid their possible cytotoxicity when directly touching the cell membrane. For instance, PDAcoated rGO [155] and DA-coated CNFs [156] were respectively mixed with PCL for 3D printing of neural scaffolds to not only increase neural cell adhesion and proliferation, but also improve the electrical signal transmission for higher healing potential. With regard to the 3D printing applications of conductive copolymers, PPy-b-PCL scaffolds were EHD-jet 3Dprinted and showed the highest conductivity of 1.15 mS cm −1 and had to enhance the effect on the differentiation and maturation of hESC-NCSCs [19]. It was clarified in another report about EHD jet 3D-printed PCL/ PAA NGCs with a conductivity of 10 −6 S cm −1 that PAA, as a polyanion polymer, can mimic the function of the neurocortical gel layer and act as a cation exchanger, thus affecting nerve excitation and nerve conduction [63]. In recent years, conducting hydrogels are more likely to be selected as 3D printing materials due to their good hydrophilicity, biocompatibility, biodegradable properties as well as soft mechanical characteristics similar to neural tissues. Heo et al [53] prepared SL-printed conductive hydrogel structures based on PEDOT:PSS and PEGDA that maintains high conductivity and can be photocrosslinked. The conductive scaffolds fabricated by 3D printing provided superb structural support and systematically transfer the ES toward encapsulated DRG cells for enhanced neuronal differentiation.
In recent years, constructing electroactive scaffolds with highly precise deposition of biomaterials and cells encapsulated by 3D bioprinting technology using conductive hydrogels have become the desired therapeutic strategy for nervous tissue engineering. A conductive construct that consists of GelMA, HAMA and PEDOT:LS hydrogels encapsulated with NSCs was bioprinted with excellent conductivity of 0.69 S cm −1 (figure 6(b) (i, ii)), and it could remarkably promote neuronal differentiation of NSCs in vitro [37]. A complete transection model of a rat spinal cord was built to investigate the effect of bioprinted conductive NSCs in vivo and the results in figure 6(b) (iii-v) further showed that the bioprinted conductive biomimetic scaffold effectively promoted the neurons regeneration at the injury site, reduced the formation of the glial scar, and enhanced nerve axon regeneration and myelination. In another research [157], 3D bioprinted conductive scaffolds composed of GelMA, BNC and sodium alginate wrapped with RSCs (RSC96 cells) were fabricated with a conductivity of 5.5 × 10 −4 S cm −1 . After being implanted in nude mice for 4 weeks, the nerve growth level was higher in the bioprinted NSC-encapsulated scaffold group than that in the pure GelMA control group. Despite those pioneer works on conductive hydrogels with cells loaded to fabricate neurobiomimetic scaffolds by 3D bioprinting, the development of bioprinted conductive constructs with appropriate mechanical properties, highly electrical conductivity and cytocompatibility for nerve repair is still very essential and challenging due to the fragility and sensitivity of neural cells [5].
Alongside the widely studied pathways of structural and chemical regulation in 3D-printed functional scaffolds for nerve repair, electrical/electromagnetic signaling is another significant factor that should be integrated because the nervous systems are extremely rich in electrical activity. The prominent function of conductive 3D matrices must be paid extra attention to for neural relay restoration. with nerve conduits implanting in SCI rats upon immunofluorescent staining for glial fibrillary acidic protein (GFAP) (green), Tuj-1 (red), and DAPI (blue). Quantitative analysis of GFAP (iv) and Tuj-1 (v). * p < 0.05, * * p < 0.01, and * * * p < 0.001. Reprinted from [37], Copyright (2023), with permission from Elsevier.

Typical applications of 3D-printed bioengineered constructs for nerve repair and therapeutics
Current neural science has advanced a lot but still has some unmet needs. Common surgeries cannot mimic microenvironments, micro alignments, or some functionalities, as well as the high cost for customized flexible treatments while 3D printing or hybrid bioprinting technologies hold great promise [27,158]. The 3D printing offers an ideal option to replicate the complex nerve-like structures for the re-establishment of functional neural connections by not only implantation in vivo but also model fabrication in vitro. This part summarizes the current advanced applications in repairing, monitoring, and enhancing the neural system utilizing 3D printing technologies.

3D-printed functional biological constructs for repairing the injuries of nervous tissues
Nervous injuries often cause catastrophic results to the whole body, such as the loss of sensation or control. Therapeutic methods, such as utilizing free nerve endings to reconstruct the end-to-end connections or other autograft therapies often lead to a high cost, lack flexibility, and cannot be customized [159][160][161]. The 3D printing induces a bunch of regenerating methods in this field.
The regeneration of CNS is a hard task owing to the complex architecture of CNS and its inherently poor regeneration capabilities. CNS regeneration utilizing 3D printing is not accomplished especially in repairing the brain, but many efforts for the regeneration of the spinal cord have been done and showed big development prospects [162]. Joung et al [33] utilized an extrusion-based multi-material printing method to manufacture a bioengineered spinal cord scaffold while Shahriari et al [163] used drown fibers as the 3D printing material to fabricate porous implants with more complex geometries to enhance the spinal cord nerve regeneration. They all made some efforts to accelerate the repair of spinal cord injuries and the regeneration of central nervous tissues. Koffler et al used µCPP method to fabricate the PEGDA-GELMA scaffolds with 200 µm-diameter multi-channels for mimicking the grey and white matter in the spinal cord (figure 7(a) (i)) [55]. The µCPP 3D-printed scaffold loaded with NPCs were implanted into the T3 spinal cord in rat to investigate its repair effects. After 4 weeks' transplantation, it was found that the implants formed new relays across the whole spinal cord transection and reconstructed the synaptic transmission compared to the control group ( figure 7(a) (ii)). Furthermore, neural functions were recovered by testing the BBB scale and motor-evoked responded in animals implanted with 3D biomimetic scaffolds loaded with NPCs (figure 7(a) (iii-v)), indicating that the designed scaffold had long-term compatibility and reconstructed function of the spinal cord well. Regarding PNI repair, 3D printing strategies are more widely investigated and some inspiring outcomes have been achieved. For example, Singh et al [164] reported an aligned cryomatrix-filled PCL scaffold utilizing the SL method and incorporated it with NGFs for reconstructing the traumatic sciatic nerve ( figure 7(b) (i, ii)). The walking trance analysis in rat models showed that the aligned conduits guidance led to better functional recovery than the other group with autografts by observing paw spreading and gait patterns ( figure 7(b) (iii)). Besides, improvement of nerve repair can also be revealed by the corresponding regenerated sciatic nerve and gastrocnemius muscle ( figure 7(b) (iv)).
In the past decades, various 3D printing strategies brought new ideas to design different types of functional neural implants, which have been considered as a perspective technology and promising option for neural repair and regeneration. The combination of different materials, cells, and various biomacromolecules using additive manufacturing processes is expected to fully recapitulate the microstructural, compositional and functional complexity of a healthy nerve for better functional recovery.

3D-printed bio-electronics for monitoring and modulating neural signals
Bioelectrical signal processing is essential in neural regeneration by either monitoring the neural signals or stimulating the circuits by biocompatible electronics. There are some topics have been addressed by trying to electrically interface the neural pathway [165]. Nowadays, 3D printing shows promise in fabricating electronics with flexibility and capability, improving mechanical, electrical, and biological properties.
BCI is an emerging technology that has been studied in many aspects. Compared to other neuroimaging methods like EEG, ECoG is a conventional method with high resolutions in both spatial and temporal, as well as the signals are not so susceptible to artifactual contamination from muscle movements and eye blinks, which regularly impair the quality of EEG signal recordings, especially during language production. Traditional ECoG probes are rigid and harmful to target regions, but 3D printing can help to address the shortcomings. Many methods are developed to manufacture flexible ECoG arrays with biocompatible materials like photolithography, thin-film deposition, etching, and others to construct conductive arrays on flexible platforms [166]. However, these methods are high-cost in both time and price due to the complex processes and expensive facilities. Yuk et al reported a high-performance PEDOT:PSS ink for 3D printing conductive devices ( figure 8(a)). They fabricated an array with nine PEDOT:PSS electrode channels in less than 30 min to record brain activity in a mouse model [38]. The soft neural probes were implanted in a mouse and local field potential was recorded with free activities.
Applications using 3D printing or hybrid processes have been investigated by several studies and it is noted that the structural scale is important to make minimally invasive procedures and long-term biocapability. Afanasenkau et al [39] created a neural implant that was used to stimulate the spinal cord of animal models with spinal cord injuries. The  implant was printed by silicone and microparticles of platinum or another electrically conductive element, and then the surface was activated by cold plasma ( figure 8(b) (i, ii)). An in vivo experiment applied to a rat model demonstrated its bio-integrability and stability ( figure 8(b) (iii, iv)). The hybrid printing technique which integrates soft materials and composites into the bioelectronic devices can be adapted to various anatomical structures. Otchy et al [40] designed and fabricated a microscale nanoclip with electronic laden to interface with small nerves in PNS. The nanoclip, fabricated by direct laser writing with 200 nm resolution, had trapdoors to secure itself to the destinated nerve and could stimulate and record nerve activities ( figure 8(c) (i-iv)). This design was tested in a zebra finch for acute and long-term recording and the results demonstrated that the nanoclip can accurately modulate the function of specific nerves with a high signal-to-noise ratio and longterm biocapability as shown in figure 8(c) (v, vi).
Recently, a universal model of electrochemical safety limits in vivo for electrophysiological stimulation was also established by Vatsyayan and Dayeh, which might promote standardization in neural implants [167]. Not only online modulating implants, but there are also some studies focusing on biodegradable [168], wireless [169], and self-powered [170] devices providing much more flexible choices for neural system modulation. In summary, 3D printing technology or hybrid printing with other processes can be used to record bioelectrical signals and stimulate them, the potential of 3D printing can help to address some issues in bioengineering.

3D-printed organoids and neural-on-chips (NoCs) for neurobiomimetic modeling as neural therapeutics
The key to neural regeneration is a comprehensive understanding of the nervous mechanism from nano/micro to macro. However, traditional invasive surgeries are limited to ethical considerations for humans while classic animal models cannot reflect the physiological characteristics of human tissues. Advances in various 3D printing technologies provide some inspirations to help to overcome the impedes, improving our understanding of the nervous system. In vitro neural model is now a booming field and is considered a promising way to decipher the mysteries of humans and improve effective therapeutics for various neural diseases.
Organoid is one of in vitro 3D models to mimic specific in vivo neural tissues. It is usually constructed by selforganization which leads to several limitations such as the inability to fully mimic the in vivo environment and lack of programmable spatial ordering yet 3D printing can overcome them [171]. An engineered flat brain organoid in a honeycomb pattern was designed and 3D-printed by Rothenbucher et al (figure 9(a) (i)) [172]. Compared to the traditional spherical brain organoid, the flat structure had a larger surface-tovolume ratio that provided a better situation for oxygen and nutrient diffusion, which minimized the necrosis in the tissue (figure 9(a) (ii)). Besides, a consistent formation of neuroepithelial folding was reported in their research which is a good sign for tissue development (figure 9(a) (iii, iv)).
NoCs is another type of in vitro neural model with precise structures to establish a controllable system to stimulate specific neural systems or diseases. The development of 3D in vitro platforms from cell-to-chip scale helps recapitulate the complexity of nerve-like tissue and interaction of organ systems, which would service the translatability of neural regeneration treatments, therapeutics, and drug screening. Kajtez et al raised a novel hybrid process by utilizing a soft-lithograph and pneumatical extrusion to fabricate a compartmentalized microfluidic platform to simulate the human neural system (figure 9(b) (i)) [41]. A brain-like model was designed with six separated chambers connected by microchannels as shown in figure 9(b) (ii). The neurons differentiated and proliferated in dense chambers that constructed neural networks and connected the chambers with directional growth in microchannels (figure 9(b) (iii)). A step further, the nigrostriatal pathway model with optimized channel parameters was designed and the directional neural pathway was observed after cell culture (figure 9(b) (iv)). In another work, a 3D-printed platform was developed inspired by the inherent composition of a functional neural system including neurons, SCs, and the end of axons to peripheral tissues as shown in figure 9(c) [42]. Different types of cells were cultured into different chambers and the pathway was established by microchannels. The stained micrographs showed effective biomimicry in the spatial of the nervous system.
One of the major advantages of 3D printing is the ability to manage multiple cell/tissue types in high spatial accuracy, which makes it possible to fabricate vascularized neural tissueengineered constructs. Vascularization is essential for the long-term maintenance of cellular processes by allowing nutrition transport and waste disposal [173]. A novel 3D-printed platform was raised by Salmon et al to establish neurovascular interactions, which are compatible with any organoid systems (figure 10(a)) [43]. The designed platform allowed vascular cells to transport through the open wall via the inlet while the center is for organoid culture. The advantages of this strategy showed great improvement in the vascularization of specific neural cell aggregates and synchronized differentiation during maturation demonstrated by cerebral organoid development. Fluorescent visualizations in figures 10(b)-(e) showed that neurites descend from the organoid to interact with the vascular networks, revealing that a complex morphology was established between vascular networks and cerebral organoids. To date, most of the current strategies for vascularization like invading the organoids with host vasculatures by in vivo culture and in vitro co-culture, lack spatial organization and temporal synchronization as well [174,175]. It is still challenging to create a more accurate neurovascular model to be served for potential neural engineering applications.
As 3D printing keeps advancing in functional micro/nanostructures manufacturing, a bunch of in vitro neural or neural-integrated models have been established [176][177][178]. It is foreseeable that there might be more in vitro models achieving longer culture and more accuracy, promoting biological understanding and medical treatments with 3D printing participation.

3D-printed brain-like models for investigating neural dynamics at tissue level
The brain is the most important part of the CNS and the most complex organ in our body. Previous studies have reported brain organoids or 2D brain models established in vitro to mimic the human brain functions. However, the developed brain-like models lack the structural complexity and functionality of the real brain [73]. The 3D printing brings some hopeful attempts in this field to overcome the obstacle of understanding the function of the brain at the tissue or organ level.
It is important to construct brain-like models in vitro that mimic the highly convoluted structure of the cerebral cortex in terms of geometry and physical shaping, which is the morphological basis of our intelligence [179]. Tallinen et al successfully constructed a physical model of the brain by taking advantage of 3D printing technology, and then coated the surface of the model with an elastomer gel layer to mimic the white and grey matter (figure 11(a) (i)) [44]. The brainlike gel model was immersed in a solvent, with the outer gel expanding relative to the core to mimic the growth of the cerebral cortex. The cusped sulci and smooth gyri formed during the gel expansion are similar to the real brain in morphology and relative time ( figure 11(a) (ii)). The relationship between cortical curvature and mechanical properties was demonstrated and quantified, and it was found that there was a significant correlation between primary sulci in real and model brains (figure 11(a) (iii)). The study sets the biophysical basis for investigating functional neurological disorders related to structural cortical malformations.
Except for modeling the complex 3D wrinkled structures of the brain, another major challenge in developing in vitro brain models is to construct 3D cell-loading regions similar to natural brain architecture (figure 11(b) (i)). Lozano et al verified the feasibility of constructing hierarchical structures by multilayer 3D printing of a novel peptide-modified RGD-GG in different colors (figure 11(b) (ii-v)) [45]. Afterwards, the primary cortical neurons were encapsulated in hydrogels to construct a layered neuron arrangement to resemble the cell distribution and connection in the micro-scale (figure 11(b) (vi)). The hydrogel is capable of containing and supporting the growth and network formation of neurons in specific layered structures, while also being sufficient to allow neurons to penetrate the gel layers (figure 11(b) (vii)), offering the possibility of more accurately mimicking the layered arrangement of cells in brain-like tissue. Further, Song et al constructed a 3D layered brain-like tissue in vitro using 3D cell-printing, and the neural circuit was formed during culture [73]. The formation of the neural circuit enables the neural network to respond sensitively to drug stimuli, improving the functionality of the brain-like tissue constructed in vitro.

Outlooks and conclusions
In recent years, 3D printing has been recognized as a powerful modern technique for fabrication of the nerve-like constructs with similar functions to their in vivo counterparts [31,60,180,181]. It has gained impressive outcomes with the ability to generate a cytoarchitecture-mimic arrangement and electric microenvironment with navigational structures and biological simulates, for the effective re-establishment of neural networks and connections. Further, in vitro platforms or organoids as therapeutic options and nerve/BCIs can be 3D-printed to investigate the neuromodulation, drug screen, and cell-cell/matrix interactions under certain conditions, or to evocate the nerve innervation functions [40,178,182]. Thanks to bioprinting technology, the specific spatial architecture of organoids can be fabricated with high efficiency to better simulate the in vivo organization and microenvironment of neural tissues [183]. Although the progress of 3D printing in the field of nerve regeneration is conspicuous, there are still certain limitations that need to be overcome, which will be discussed in greater detail in this section.
The first challenge of 3D printing is mimicking the complexity of desired organ structures. Despite a great of work that has been carried out to create nerve-like constructs with well-designed architectures and multi-components, it is still difficult to reproduce fine structures similar to natural organs. In the adult nervous system, neural fasciculus consists of highly-oriented and densely-arranged axons with diameters varying from 0.1 to 10 µm, and the different diameters significantly influence the conduction velocity of electrical signals in a wide variation [184]. However, current printing systems and strategies are not capable to create a structure which can fully mimic the refined dimension of natural nerve tissues and be cell-friendly to neurocytes simultaneously. As a result, it restricts the efficient replication of the physiological environment prevalent to neural systems and limits the controllable guidance for nerve cell migration and axonal growth by manipulating the printing process for desired and delicate structures. Consequently, the potentials of 3D printing as in vitro bridges for reconstructing neuronal relays and in vivo models for accurately investigating biological development are restricted. Although light-based and EHD printing as discussed in section 2 have been developed as potential strategies for high-resolution printing, the biocompatible materials or inks which are suitable to those techniques are limited (table 1). With respect to 3D bioprinting, its resolution is often as low as the micrometer scale because it is usually based on extrusion mode. Despite the efforts and trials on electricity-/laser-assisted bioprinting for a higher construct resolution, it is until now not clear about the evolution of cell behaviors during those printing processes influenced by light or electricity, which requires more explorations in future research. Obviously, it is necessary to improve the printing systems and methodology to overcome current manufacturing limitations, and to extend the manufacturing capacity of 3D printing for the application of 3D printing in nerve regeneration.
Furthermore, 3D printing of hybrid constructs which integrates multiple materials, cellular types, spatial scales and techniques with different mechanical properties or biological functions has become progressively trending. Theoretically, the ideal candidate neural implants should closely recapitulate the internal organization of native nerve tissues considering the crosstalk of adjacent organs such as the sufficient nutrition supplied from vessels and motor innervation of its effector muscles, thereby developing more complex and highly organized structures which could better satisfy the holistically integrated body. As more diverse materials become available including polymers, hydrogels and inorganic nanomaterials with various biological functions and mechanical properties, the design and manufacture of versatile constructs become increasingly possible [112,185]. At the same time, Figure 11. 3D printed in vitro brain-like models at the tissue level. (a) A physical gel brain model utilizing 3D printing to fabricate. (i) A soft physical gel model fabricated by casting from a negative silicon mold created by a 3D printed fetal brain model. (ii) The selling process with evolving complex convolutions. (iii) Sulcal lines about numerical simulation (blue lines), 3D printed brain (red lines), and real brain model (green lines) were marked. Reprinted by permission from Springer Nature Customer Service Centre GmbH: Nature, Nature Physics [44], Copyright (2016). (b) A multilayer brain-like model. (i) The theoretical layers in a real human brain. (ii)-(v) The brain-like model was fabricated by 3D printing to test the structure's stability. (vi) A three-layer model with an acellular middle layer is constructed. After 5 d of culturing, the complex 3D neural networks were constructed and axons penetrating the middle layer were observed. Reprinted from [45], Copyright (2015), with permission from Elsevier. the development of hybrid additive manufacturing approaches and systems is in urgent need to overcome the disadvantages of single printing technique for mimicking the complex and structural hierarchy of neural architecture. Based on the physical characteristics of different anatomical structures in nerve tissues, highly biomimetic neural implants are expected to be fabricated in various scales and details combining different 3D printing technologies with different processing tools in an integrated fabrication platform. In addition, by using 3D bioprinting technique, different cell types can be distributed and arranged in any desired pattern and the whole cell spheroids can be accurately picked and assembled [69]. A variety of neurocytes (e.g. neurons, oligodendrocytes, astrocytes, and microglia) are able to be involved in the multicellular 3D bioprinted living constructs which can meet the needs of the neural tissue regeneration or recreation of neural organoids as the interactions among multiple cell types existed in natural tissues lead to their specific function in vivo [186]. More fundamental research is required in the future to optimize cell distribution and co-culture conditions in vitro. Taking advantage of the unique processing capability of hybrid 3D printing, intricately shaped and multi-functional structures with heterogeneous materials and gradients are potential to be engineered.
Another important issue that needs to be considered in the future is to realize the neurotization of artificial tissueengineered implants by 3D printing techniques. As is well known that nerve supply in the whole human body to carry stimulation and response back and forth from the brain, and neural signals direct the maturation and phenotypic expression of developing organs [187,188]. Therefore, it hypothesized that neurotization of the fabricated constructs would result in an improvement in their functionality expression, and the formed neuro-tissue junctions would allow indirect stimulation of the related tissue-like construct via the nerve. Taking advantage of the capacity to integrate multicomponent into one single complex construct, 3D printing may provide exciting and promising opportunities to replicate neutralized organs by special scaffold design, the inclusion of NTFs or nerve cells, and in vitro/vivo pre-neurotization. Nevertheless, neurotization remains one of the major obstacles that need to be overcome before large tissue-engineered construct can be applied and come into its role.
Currently, the majority of in vivo studies performed using current 3D printed neural constructs are performed in mice/rats, which do not provide a high level of physiological relevance to human neural injury/repair. Hence, large animal models must be built to study the underlying mechanisms and evaluate potential treatments using 3D-printed scaffolds for repairing human nerve injury [189]. Liu et al [190] 3Dprinted collagen/silk fibroin/hypoxia-pretreated HUCMSCsderived exosomes scaffolds and implanted it into the injured brains of beagle dogs, showing enhancement for neuroregeneration, inhibition of inflammatory factors and promotion of motor function recovery. They also tested the effect of 3D-printed collagen/chitosan/secretome scaffolds on the injured brain regeneration process in canines, demonstrating significant regeneration of nerve fibers, and promoted endogenous neuronal differentiation and synapse formation after TBI in the scaffold-treated canines [68]. Nevertheless, there is still an urgent need for the standardization or rationalization of large animal models for studying nerve repair.
The full clinical translation of 3D-printed nerve-like implants has a long time to accomplish as some outstanding problems still existed. First, the in vivo safety of materials, printing process and exogenous cells adopted are still not fully clarified. Adverse acute immune response or immunemediated tissue rejection must be carefully considered when foreign neural substitutes are implanted [191], and chronic inflammatory responses depending on the characteristics of implanted constructs (materials, structures, surface properties, degradation, etc) lasting in a long duration need to be monitored and regulated. Second, specific regulations and guidelines to govern medical devices produced by 3D printing are very limited, which also hinders the clinical translation of the 3D-printed products on neural repair applications [192]. Bioprinting makes the therapeutic potential more ethically complicated cell considering the cell source and its processing procedures [193]. Previous long-term in vivo studies indicated that the implanted cells may lead to the risk of teratoma and cancer-related issues [194]. Therefore, clinical translation of 3D printing, especially bioprinting, would have to go a long way through pre-clinical and clinical evaluations while considering the social ethical constraints, reliable regulations and standardization. Finally, despite the great promise of 3D printing in the applications of neural injury repair, the specific clinical treatment methods should be carefully selected and reformed by taking full consideration of the types of neural injuries/diseases encountered in the CNS/PNS. For example, printing in situ might boost the repair of PNS damage as a time-consuming therapy and can eliminate any risks of contamination or destruction of the mechanical integrity of the scaffolds during post-fabrication [180,195]. While dealing with CNS diseases/trauma is more easily realized by injecting functional hydrogels releasing cells and NGFs [196][197][198].
In addition, customized bioelectronics and devices have seen bright potentials thanks to 3D-printing approaches and functional inks, and intelligent integration and built-in computational processing are expected to provide new types of treatment paradigms in the future [199].