Nano-enabled cellular engineering for bioelectric studies

Abstract

Engineered cells have opened up a new avenue for scientists and engineers to achieve specialized biological functions. Nanomaterials, such as silicon nanowires and quantum dots, can establish tight interfaces with cells either extra- or intracellularly, and they have already been widely used to control cellular functions. The future exploration of nanomaterials in cellular engineering may reveal numerous opportunities in both fundamental bioelectric studies and clinic applications. In this review, we highlight several nanomaterials-enabled non-genetic approaches to fabricating engineered cells. First, we briefly review the latest progress in engineered or synthetic cells, such as protocells that create cell-like behaviors from nonliving building blocks, and cells made by genetic or chemical modifications. Next, we illustrate the need for non-genetic cellular engineering with semiconductors and present some examples where chemical synthesis yields complex morphology or functions needed for biointerfaces. We then provide discussions in detail about the semiconductor nanostructure-enabled neural, cardiac, and microbial modulations. We also suggest the need to integrate tissue engineering with semiconductor devices to carry out more complex functions. We end this review by providing our perspectives for future development in non-genetic cellular engineering.

1. Introduction

Rather than merely relying upon the natural behavior of cells, researchers have begun implementing techniques of cellular engineering to aid them in modifying and amplifying cellular functions, thereby opening new potential routes for disease treatment [1-3]. Cellular engineering has the advantage of creating a degree of control over cellular growth and functions that is not attainable otherwise. Synthetic biology, in particular, has been implemented for two decades to enable designer gene-circuits within synthetic and engineered cells and create responsive subunits that lead to novel biological functions [2-5]. As an example, immunoengineering has enabled several cutting-edge techniques in tumor treatment [1]. The lymphocyte T-cell can be harvested to be modified ex vivo with chimeric antigen receptors (CAR) in order to more effectively target tumor cells. Different nanostructured material systems, such as those involving carbon nanotubes [6, 7], paramagnetic iron nanoparticles [8], and mesoporous silica microrods [9], can be utilized to activate and expand T-cells. T-cell stimulating nanoparticles can also be implanted in vivo via degradable biomaterials to deliver these T-cell treatments directly to the tumor itself [10]. Similar materials-based approaches may extend to other engineered cell systems for future biomedical applications.

Semiconductor nanostructures have entered this narration of cellular engineering as a means of developing innovative electronic and photonic devices that can interface with cellular surfaces at the nanoscale level [3, 11]. Semiconductors, silicon as a key example, have the advantage of precise engineering with regards to parameters for their size, morphology, structure, and doping [12]. This exact material control creates the ability for predictable and improved device functions [13]. If being shrunk to the nanoscale, semiconductor devices would then be employed for subcellular probing, as they can quickly respond to optical, electronic and electrochemical signals to produce diverse signals to elicit or sense biological responses.

The ability of semiconductors for both sensing and modulation makes them appreciably useful in developing medical implants, studying biophysical cellular signaling, and adaptable cellular materials [11, 14]. For example, nanowires-based field effect transistors (FETs) can sense in highly localized areas, with high spatiotemporal resolution over multiple different cellular events in three dimensions [15, 16]. High-density electrical recordings of ex vivo heart and brain tissue slices have been achieved through the use of these FETs [17]. The seamless interfacing of nanoscale FETs-based injectable mesh electronics with brain tissues opens the door for unprecedented medical research and implantable devices that could one day monitor and interact with a patient’s body [18]. On the modulation side, freestanding semiconductor nanomaterials, when configured as photodiodes, have enabled subcellular control of activities of neurons [19-23], cardiomyocytes [24, 25], and even non-excitable cells and tissues [26-28]. In the future, the close biointerface with semiconductors (e.g., nanomembranes, nanowires, nanoparticles, or nanotubes) will eventually extend into all subcellular elements, including lipid bilayer, ion channel, and cytoskeletal system (Fig. 1). The intracellular semiconductor nanostructure may be used to guide molecular transport, elicit bioelectric response, or regulate enzymatic reactions upon optical stimuli. These advances suggest that a new wave of cellular engineering and synthetic biology is within the realm of possibility, due to the unique functionalities and structures of semiconductor nanostructures. This review will discuss the progress in traditional cellular engineering and the recent advance in semiconductor-enabled synthetic biology.

Figure 1.

Semiconductor-enabled synthetic biology. In the future, semiconductors can also be engineered with lipid bilayer, ion channel, and cytoskeleton to guide molecular transport, elicit bioelectric response, or regulate enzymatic reactions upon external stimuli.

2. Selected “cellular” engineering approaches

An artificial cell is an engineered or synthetic cell that can perform biological functions. Besides having analogous size, morphology, and surface characteristics of those of a real cell, the artificial cells would achieve specific tasks such as intra- and intercellular signaling, micro-environmental monitoring, and biochemical recording and memory effect in vitro or in vivo [29]. Engineered artificial cells have attracted intense interest from diverse disciplines over the past several decades, as they are promising in several clinical and technical applications such as adeno-associated virus gene therapy [30]. Thus far, the design and implementation of engineered cells have adopted both a bottom-up approach and a top-down approach. The bottom-up approach can build an artificial cell in the laboratory through a similar approach as molecular evolution [31, 32]. The fabrication starts from organic chemicals and the complexity of the whole system increases step-by-step. For example, bio-macromolecules, phospholipid, and protein are assembled to form a protocell with some fundamental biological characteristics such as homeostasis, metabolism, and the capability of evolution. The top-down approach builds an engineered cell by chemically or genetically modifying an existing cell [30]. For example, the genes of mammalian cells or bacteria can be modified to achieve diverse logic gates and memory functions. The bottom-up and top-down approaches are usually complementary to each other and can be chosen selectively according to different fundamental studies or applications.

2.1. Protocell

The bottom-up approach aims to build a living protocell from non-living components for similar cellular functions [33]. However, it is challenging to coordinate all the individual components together to mimic the functionality of a cell completely. Therefore, as a practical alternative in the laboratory, scientists intend to build protocell models with at least several essential features of a living cell. The critical feature of innate cells usually involves a semi-permeable membrane, the existence of biomacromolecules, and homeostasis [29]. First, the semi-permeable membrane forms a barrier between the interior and external aqueous environments, defining the cell boundary. Next, biomacromolecules such as DNA and RNA record the genetic information of the cell to then carry out the central dogma. Last, homeostasis is essential to achieve information and energy transfer by metabolic processes inside the cell.

Efforts have been made to build the appropriate experimental models for protocells [32, 34-36]. Early studies focused on constructing minimal systems confined by organic chemicals, such as phospholipid or fatty acid vesicles [33]. One example includes introducing the gene expression system (i.e., engineered plasmid DNA and essential enzymes) inside lipid vesicles [37]. It has been manifested that the expression of protein is able to start successfully with sufficient essential amino acid supply. Aside from gene expression, RNA synthesis, replication, and DNA amplification have also worked successfully inside a lipid or fatty acid based protocell [31]. Nevertheless, the lack of an inner organelle structure limits the lipid-supported system function in mass and energy transfer. Additionally, parameters such as pH, temperature, and ionic strength can also break the stability of fatty acid vesicles, which further impedes its potential application in complex natural environments. Therefore, alternative types of protocells have been developed recently to increase their chemical stability and biological ability [32].

For example, semipermeable inorganic colloidosomes can also build microscale capsules to form protocells. Lei et al. developed silica nanoparticle-based colloidosomes (Fig. 2(a)) that are crosslinked with a pH-responsive copolymer and tetramethoxysilane (TMOS) on the surface [34]. The change in the zeta potential of the surface copolymer layer makes the inorganic membrane able to release or take in small molecules. The enzymatic reaction inside inorganic capsules can also be controlled by this surface gating mechanism. Recently, Kumar et al. has designed an organoclay/DNA semipermeable microsystem (Fig. 2(b)) to achieve the directional motion (one key characteristic for bacterial) of the protocell [35]. The protocell system entirely utilized the electrostatically self-assemble between polyanionic DNA (with catalase) and cationic nanoclay. Functional catalases are stored inside the organoclay/membrane to achieve the directional motion of the system. Specifically, the protocell can produce oxygen bubbles inside to create the buoyancy when caught the hydrogen peroxide from the environment. Then this buoyant force controls the one directional motion of the protocells. Organoclay/DNA protocells can further achieve oscillatory movement by introducing another glucose oxidase (oxygen-dependent enzyme) to consume the oxygen bubbles inside protocells. Thus, the motion of the protocell can be controlled through the concentration of hydrogen peroxide and glucose. Protocells can be further associated and linked to form prototissue to carry out a specific function together. Gobbo et al. then assembled around 10–30 protein-polymer protocell into a prototissue (Fig. 2(c)) through bio-orthogonal adhesion with thermally responsive polymer [36]. The assembled prototissues are not a simple gathering of protocells, since they give collective interaction upon external stimuli. For example, they present the coordinated response to external temperature variation and perform reversible contractions or relaxation, correspondingly.

Figure 2.

Fabrication of protocells, (a) Schematic representation shows the assemble of inorganic-based protocells with TMOS and pH-sensitive copolymer (as shown in the top right) on the surface. The yellow part illustrates the closely packed silica nanoparticles. The red part indicates anionic carboxylate of the copolymer and the blue part indicates cationic dimethylamino. (b) Schematic diagram shows the approach to build electrostatically-induced protocells with the organoclay/DNA membrane. Anionic DNA are injected into the cationic nanoclay dispersion to form the microcapsule. Encapsulated catalase inside is used to control the directional motion of the protocells. (c) Fluorescence optical microscopy images of the assembled prototissue. Individual of the protocell is dye-labeled and tightly packed by bio-orthogonally adhesion. Red and green parts indicate the protein-polymer protocells with orthogonally crosslinking. Purple parts indicate the crosslinked host protein-polymer protocells membrane. Panel (a) is reproduced with permission from Ref. [34], © Springer Nature 2013. Panel (b) is reproduced with permission from Ref. [35], © Springer Nature 2018. Panel (c) is reproduced with permission from Ref. [36], © Springer Nature 2018.

Self-replication is another crucial characteristic for the living cell, and the Cronin group has developed a millimeter-scale oil-in-water droplet system to build a self-replicable protocell model [38]. In this protocell model, the oil-in-water droplet system made of amphiphilic imine can be divided into two parts by lowering the interfacial tension between the chloroform phase and the water phase. Amphiphilic imine is designed to be an effective surfactant through organic modification and thus can work as an organic replicator at the chloroform/water interface to induce the division of droplets. For the application of protocells, the Gu group has recently built a protocell model to achieve insulin secretion for clinic application [39]. The synthetic cells are assembled by the classic lipid-film hydration method and have vesicles-in-vesicle superstructures with 1–5 micrometer. These cells are capable of sensing glucose concentration and secreting insulin correspondingly.

2.2. Chemically and genetically modified cells

The top-down approach modifies an existing cell and aims to develop a more programmable pathway to precisely control cellular behaviors [30]. This approach has attracted considerable attention over the past decades due to the promising results demonstrated in clinical trials. For example, genetically modified T-cells with CAR are already used in clinical trials for curing certain types of leukemias and lymphomas [40]. This approach can be implemented by modifying the cells either chemically or genetically. The chemical approach includes modifying small molecule [41] and grafting synthetic polymer on the cell surface [42]. For example, the Hawker group has developed a cytocompatible controlled radical polymerization (CRP) technique to graft synthetic polymer into the surface of the living cells, which provides researchers a practical approach to direct the cellular phenotype [42]. The genetic approach includes modifying the biological modules inside cells, such as DNA and RNA, to influence the processes in the central dogma. Recent progress in target DNA base editing and demethylation through the CRISPR/Cas system offers many ways to design functional engineered cell [43-46]. For example, the Wagers group has demonstrated that disease like dystrophin can partially recover through gene modification in mouse model [47].

Optogenetics is a recently developed technique that allows either pulsed or continuous light to control cellular behavior such as action potential generation [48-50]. In this approach, opsin (such as light-sensitive microbial ion channel or pump) is introduced into cell by genetic approaches. The Deisseroth group has early identified a light-driven chloride pump (NpHR) which could be used in modulating neural activity [51]. It has shown that the expression of NpHR has the ability to “knockout” action potentials to control the swimming behavior of nematode worms. Aside from NpHR, the anion-conducting channelrhodopsins are another excellent opsin option in optogenetics approaches. The Deisseroth group further enhanced the opsin performance by designing their mutants called fast light-activated anion-selective rhodopsin [52]. The expression of this protein in vivo showed similar inhibition of swimming behavior of worms. Recent progress in the Deisseroth group has proved that optogenetic can control diverse behavior in animals. For example, when studied the thirst behavior of mice, they found that sated mice can be rapidly switched into initial thirsty state by optogenetic stimulated on specific neuron [53]. Organelle position and transport are also important information for revealing the intracellular molecules motion. The Cui group has recently recorded rotational and translational dynamics of intracellular endosomes by engulfed gold nanorods [54]. Optogenetics methods can control these organellar motion too. For example, the Kapitein group has optically controlled cytoskeletal motion and the organelle position through light-sensitive cytoskeletal motor proteins [55]. This approach can be applied to explore the relationship between the functions and positions of intracellular organelles.

Other than neurons, bacteria could also be engineered to achieve specific functions in order to extend their application in fields such as biomedical sensing. The Lu group has recently developed an ingestible optoelectronic device (Fig. 3) for remote reporting of local bleeding in the gastrointestinal environment [56]. The engineered bacteria are capable of transducing blood signals into bioluminescence. In particular, this kind of engineered bacteria are able to internalize extracellular heme to express luciferase operon. To prove its ability in an animal model, engineered bacteria are first assembled into a wireless platform and then put into the gastric environment of a porcine. The whole device is manifested to be resident and remain stable in the gastric environment of the porcine. It is shown that this device could sense gastric bleeding sensitively even in the harsh acidic environment. In addition to blood, devices can also be responsive to many biological analytes such as thiosulfate, an essential biomarker for intestinal inflammation.

Figure 3.

Application of the engineered microbial. (a) Schematic diagram shows the bacterial gene circuit for engineering microbial with the ability of blood sensing. (b) Schematic and photographs of the whole device. Engineered bacteria are integrated into a device for in-vivo wireless bleeding sensing. (c) X-ray image demonstrates that the device can stay in the porcine gastric environment for detecting the bleeding in vivo. (d) Kinetic diagram of the photocurrent for the blood sensor device in vivo. Compared with buffer control, the device is capable of sensing the gastric bleeding (*P < 0.05, student’s t test). Panles (a)–(c) are reproduced with permission from Ref. [56], © AAAS 2018.

Despite this explosive progress, it is still challenging to build engineered cells that always yield deterministic behaviors. The fabrication of engineered cell still needs a lot of design-build-test iterations before final applications [30]. Due to the complexity of genetic information, genetic designs and outcomes usually vary from cell to cell, and the process can be slow. Thus, the fabrication of artificial cells may benefit from other non-genetic tools or parts.

3. Synthesis of semiconductors for biointerfaces

The use of semiconductor materials for subcellular biophysics or clinical therapeutics has been widely studied [14, 57]. Several semiconductors have shown to form biocompatible and functional interfaces with cells and tissues readily [11]. Semiconductor nanostructures, in particular, have tunability in electronic, photonic and mechanical properties, can be tailored into various geometries that match those of biological components, and therefore offer more design space for biointerfaces. Multiple synthetic approaches are available to yield nanostructured semiconductors. Among them, solution-phase colloidal synthesis and vapor–liquid–solid (VLS) growth are the two most common methods [11]. Diverse synthetic parameters, such as pressure, temperature, dopant concentration, spatiotemporal profiles of the reactants, can be controlled with high precision (e.g., by computer program [58]) to yield semiconductor nanostructures of targeted performance. In this section, we will highlight several recent approaches for functional semiconductor synthesis.

Chemical vapor deposition (CVD) is a widely used technique for the synthesis of nanostructured semiconductors due to its controllability and scalability. Earlier work from Lieber group demonstrated the synthesis of two-dimensional (2D) kinked Si nanowires with zig-zag shapes through precise pressure modulation in the CVD system [59]. When growing along <112> orientation, the kinked Si nanowires have controllable arm lengths and uniform 120° joint angles between the two straight segments. The ability to control the angle is critical because it enables unusual device designs in cellular interfaces. For example, by joining two or three 120° kinks together with ultrashort segments (~ 50–300 nm long) in between, Si nanowire probes with 60° or 0° tip angles can form and readily penetrate into single cells for intracellular measurement.

Tuning angles represents one strategy to morphology control. By exploiting the effects of dopant gases and the surface chemistry during CVD, Cahoon and Filler groups have produced significantly more complex nanowire structures (e.g., diameter-modulated nanowire superlattice) by either direct synthesis or post-structural modification [13, 58, 60-62]. At the fundamental level, they found that the growth rate of nanowire is linear with the partial pressure of SiH₄ and the chemical etching rate has an exponential relationship with the doping level of phosphorus (an n-type dopant) inside the silicon nanowires. Thus, arbitrary morphologies of nanowires can be achieved through the rational design in structure and precise control of the synthesis process. Furthermore, the surface chemistry is also vital to keep the cylinder shape of nanowires. It has been manifested that the surface elements such as chloride on the sidewall can largely passivate the side surface and thus help the nanowire grow vertically. The absorption and desorption of passivating elements on the surface have a large effect on the stability of the seed particles.

In addition to exploit the effect of dopants and other additives, the Tian group has leveraged dynamic behaviors of gold (Au), which is also the catalyst used for nanowire growth, to yield sophisticated nanowire structures [63]. For example, through pressure control, Tian lab deposited atomic Au patterns along the sidewall of Si nanowires by iterated deposition-diffusion-incorporation process. These patterns display anisotropic and graded profiles, and can serve as a chemical resist for anisotropic wet chemical etching of Si in potassium hydroxide (KOH). Upon etching, skeleton-like morphology patterns (Fig. 4(a)) are revealed by removing the unprotected area. Similar to the shape of bee’s stinger, these spicules-shape nanowires could be rooted tightly inside biological tissues, thus enhancing their mechanical robustness in applications such as bioelectronic implants [63]. In a separate work, the Tian group found that periodic atomic gold lines can be deposited over the Si nanowire sidewalls through the stick-slip motion of the Au/Si alloy droplets [64]. In conjunction with Au-assisted wet chemical etching in HF/H₂O₂ solution, Si nanowires with massively parallel nanoporous grooves (Fig. 4(b)) were produced. These grooves significantly increase the surface area and roughness of the nanowires, as well as certain physical properties such as photothermal response. Nanowire structures similar to this have been used for cellular modulation by a photothermal process [65].

Figure 4.

Rational design of silicon structure for biointerface. (a) STEM tomography of one type silicon nanowires spicules feature (right) and 3D curvature maps (left) show the mesostructured silicon nanowire display anisotropic and graded profiles. (b) TEM image (left) and schematic diagram (right) of Si nanowires with atomic Au-catalyzed etching show the massively parallel nanoporous grooves structure on the sidewall of silicon nanowires. (c) TEM image (left) of mesostructured Si illustrates the hexagonal packing of silicon nanowires inside. End- (right top) and side-view (right bottom) schematic diagram illustrates the heterogeneous structure between nanowire (Si, as shown in green) and inner bridges (SiO_x, as shown in pink). Panel (a) is reproduced with permission from Ref. [63], © AAAS 2015. Panel (b) is reproduced with permission from Ref. [64], © Springer Nature 2017. Panel (c) is reproduced with permission from Ref. [21], © Springer Nature 2016.

Another strategy for morphology control of semiconductor nanostructures is the template-assisted synthesis, e.g., a nanocasting method. The Tian group has shown that the mesoporous Si particles can be synthesized by decomposition of SiH₄ inside the mesoporous silica template of either a hexagonal or double gyroidal structure [21]. For example, with the hexagonal template, mesostructured bundles of Si nanowires with 7–10 nm nanowire diameter and 8–15 nm nanowire spacing can be produced (Fig. 4(c)) upon template removal with hydrofluoric acid (HF). With atom probe tomography and other structural analysis, it was found that these Si-based mesostructures contain size-dependent chemical heterogeneity, which produces a deformable inorganic framework and strong photothermal response in saline. These Si particles have been used for neuromodulation through an optocapacitance mechanism.

Morphology control of nanostructured semiconductors represents only one material aspect for biointerfaces. To elicit functional cellular activities, the semiconductors need to produce physical outputs that can be coupled to cells and tissues. For the majority of these physical outputs, the dopant control in semiconductors is a key. In particular, diode junction has been incorporated into multiple Si nanowire structures for neural or cardiac modulations (see Section 4 for details). For example, a diode junction of p-type/intrinsic/n-type (PIN) configuration can be produced as a coaxial geometry in Si nanowires, by first growing p-doped Si nanowire cores via VLS method, followed by vapor–solid (VS) deposition of intrinsic and n-doped shells [66]. One key advantage of the coaxial (i.e., radially-modulated) PIN silicon nanowire is that photogenerated carriers can be collected more efficiently by the diode junction, compared to axially-modulated PIN Si nanowires [67]. Indeed, the coaxial PIN Si nanowires have been used to produce sufficient photoelectrochemical response to depolarize a target neuron locally.

Other physical properties of semiconductors, such as the bandgap, can also be tuned for biological studies. The bandgap of semiconductors can be modified by altering the size or the dimension of the material. For example, the Cui group reported that the bandgap of molybdenum disulfide (MoS₂) could be increased from 1.3 eV of an initial bulk band gap to 1.55 eV by reducing MoS₂ thickness to a nanometer scale [68]. During the synthesis, Mo precursors are sputtered on the substrate with a thickness of around 10 nanometers before sulfurization in the furnace tube at high temperature. The increase in bandgap allows the MoS₂-based device to disinfect water by generating reactive oxygen species (ROS) with solar illumination.

Finally, electrochemical synthesis represents another intriguing and compelling approach for semiconductor nanostructures, which is mostly underestimated. For example, nanoporous Si with well-defined pore sizes and alignment (i.e., either straight channels or zig-zag channels) can be synthesized by electrochemical etching of Si wafers [69]. When a constant current is applied on a p-type silicon wafer in the HF/ethanol environment, Si particles with fairly uniform pore sizes form. With sonication and filtration, the size distribution of the Si particles can be narrowed down to fit for specific biological applications such as drug delivery or intracellular imaging. These etched Si particles are promising biomaterials because they have tunable photoluminescent properties, are biocompatible and biodegradable.

4. Cellular-scale engineering through semiconductor-based nanostructures

Genetic approaches for modulation and sensing of the output of cellular circuits are primarily based on gene modification and the expression of fluorescent proteins [30, 70, 71]. However, genetic approaches can be slow and inefficient in some instances, and they could produce unexpected off-target effects. Another concern for the genetic approach is that altering of human genomes in any clinical trials would lead to various ethical issues [72]. An alternative to the genetic approach is non-genetic cellular engineering with stimuli-responsive materials such as nanoscale semiconductors or molecular clusters [3, 11, 14]. The non-genetic approach is typically controlled by physical or chemical means, thus offering many unnatural pathways to control biological behaviors across many different length scales. As semiconductors can be configured into various electronic and optoelectronic devices, incorporation of semiconductor processing would allow for faster and more accurate modulation and sensing of diverse biomechanical or bioelectric processes [11, 57]. Recent research shows that nanoscale semiconductors are able to either enter a cell or form a tight extracellular interface with a cell, thus forming non-genetically engineered constructs [11]. The non-genetically engineered constructs then can give specific responses upon external stimuli like light [22, 66], ultrasound [73], or magnetic field [74, 75].

4.1. Neuromodulation

Neuromodulation is the alteration of neural activity through targeted delivery of a stimulus, such as an electrical current or an optical pulse [19]. Neural activity is marked by the opening and closing of ion channels on the plasma membrane, producing single or a spike train of action potentials [76]. Action potentials then propagate across synapses between adjacent neurons and trigger the biological activities of the tissue or organ. Optically active inorganic semiconductors have been explored as non-genetic tools to yield electrochemical or thermal outputs for precise neural modulation [22,66,77,78]. In general, two processes are employed for neuromodulation, i.e., photoelectrochemical and photothermal modulations.

In the photoelectrochemical process, a light signal is converted by the semiconductor into capacitive or Faradaic outputs to modulate the behavior of a neuronal cell then [11]. Most of photoelectrochemical devices used in energy studies or earlier biological modulations require interconnects or internal wiring. However, the free-standing coaxial PIN Si nanowires have been recently demonstrated to show that they can be used for the photoelectrochemical modulation of single neuronal activities wirelessly [22, 66]. The coaxial PIN Si nanowires also contain atomic gold on the outer surfaces and at the grain boundaries of nanocrystalline shell; these gold species can promote photocathodic reactions at the biointerface to depolarize the neurons (Fig. 5(a)). Precisely, upon light stimulation, the photogenerated electrons move to the n-type Si nanowire shell and then get injected into the saline for reduction reaction (the detailed processes are still unknown at this stage given the complex physiological environment), which lowers the local electrochemical potential. In this manner, a neuron/Si nanowire hybrid can produce action potentials with controlled frequencies in the neuron by light pulses. The similar principle can be scaled up to the whole animal level. For example, a recently reported multi-layered and dopant modulated Si membranes are capable of optically controlling the somatosensory and motor cortex activities and the manipulation of the limb motion of anesthetized mice [22].

Figure 5.

Neuromodulation. (a) Schematic diagram (left) illustrates that rational designed PIN-silicon nanowires elicit action potential on the cell membrane upon light illumination. The blue and orange arrows indicate the action potential propagating inside PIN-silicon nanowire when stimulated by light. Membrane voltage of dorsal root ganglion (DRG) neurons with nanowires (right) can be largely changed with 0.4 ms light exposure. (b) Schematic diagram (top) shows that mesostructured Si yield capacitive currents across lipid bilayer when stimulated by light. (i) represents the pulsed optical signals input and local heat is generated by the photothermal effect of mesostructured Si. (ii) represents capacitive currents directly generated by local heating will flow across the lipid bilayer to determine membrane potential. (iii) represents the local heat also influenced the ion channel activities. (iv) represents currents will also be generated through the ion channel and then determine the membrane potential together with the current in (ii). (v) represents the ion channel activities can be influenced back by the membrane potential. (vi) represents that all of the processes discussed above can generate the output of membrane potential in time, frequency and 2D maps. Membrane potential of DRG neuron (bottom left and middle) can rapidly respond to different frequency of pulsed laser. f and f₀ are output and input frequency, respectively. Green bars indicate when the laser was delivered. The bottom right panel shows the area-based return map for the output result. (c) The differential interference contrast (DIC) image of nanowire co-cultured glial cells and neurons (left). Black arrow points out the engineered glial cell with nanowire. Red arrow and blue arrow point out the adjacent glial cell and neuron, respectively. Quantitative analysis of the result (right) provides the calcium dynamics information of glial cells and neurons under laser stimulation. (d) The confocal microscope image of the system illustrates that silicon nanowire can manipulate the microtubule network upon illumination. Red parts represent the microtubule and blue parts represent the silicon nanowires. The white star points out the laser illumination site on the nanowire. Panel (a) is reproduced with permission from Ref. [66], © Springer Nature 2018. Panel (b) is reproduced with permission from Ref. [21], © Springer Nature 2016. Panels (c) and (d) are reproduced with permission from Ref. [22], © Springer Nature 2018.

Doping level control represents only one strategy in enabling semiconductor-based biointerfaces, through mostly the photoelectrochemical mechanism [79]. As surface or the internal topography also plays an important role in dictating other semiconductor properties such as thermal and mechanical behaviors, several efforts in the structural design of Si have led to new neuromodulation materials. The Tian group has recently demonstrated that the rough surfaces and porous channels of the textured Si nanowire or mesoporous Si particles can display enhanced photothermal properties [65]. With textured Si nanowires, highly localized modulation junctions with neuronal processes can be established. Upon illumination, the local temperature at the Si/neuron junction rapidly elevates. This could lead to a fast increase of the electric capacitance on the surface and a depolarization on the membrane (i.e., optocapacitance mechanism), triggering the calcium influx through voltage-sensitive calcium channels. Taken the Si and the interfacing neurons together, this hybrid configuration represents an engineered cell system that can bring out calcium wave propagations upon the external light stimuli.

Similarly, mesoporous Si particles adhere tightly to the neuronal cell membrane, and elicit action potentials in neurons by optocapacitance mechanism [21]. Notably, by controlling the light pulse frequency, some unnatural neuronal responses such as an alternating pattern of action potential and subthreshold depolarization can be recorded with a patch clamp pipette from the Si-attaching neurons (Fig. 5(b)) [21]. This suggests a potentially new way for cellular engineering through bioelectric pattern formation in single cells or cellular populations. While Si-based biomaterials can enable engineered neurons through the formation of extracellular modulation interfaces, the Si particles or nanowires may detach from cells over time. An alternative method is to establish intercellular interfaces. For example, satellite glial cells can internalize Si nanowires and readily form intracellular modulation interfaces [22]. When a focused laser pulse was delivered onto the intracellular glia/Si interface, an instantaneous increase in cytosolic calcium concentration was observed in that cell, followed by intercellular calcium wave propagation to adjacent glial and neuronal cells (Fig. 5(c)). This initial calcium increase is likely due to photothermally-triggered transient polarization of calcium-storage organelles (e.g., endoplasmic reticulum). This coupled intracellular–intercellular modulation represents an indirect method for neuromodulation. Finally, the internalized Si nanowires can enable cytoskeletal control in a highly localized area (Fig. 5(d)), suggesting a remotely controlled biomechanical modulation of cells.

For the aforementioned studies, while other materials such as gold nanoparticles can also elicit neuronal responses [80], Si-based materials are biodegradable and can incorporate unique semiconductor device processes such as photoelectrochemical reactions into neuromodulations; this is potentially useful for transient and local biochemical control (i.e., not just bioelectric control) in the nervous system.

Aside from optically active inorganic semiconductors, organic semiconductors with excellent biocompatibility also be used to modulate neuronal behaviors. The unique advantage of the organic semiconductor is that it combines the deformability and flexibility of organic materials with the electronic or optoelectronic properties found in semiconductors. Additionally, it can offer new behaviors such as ion conductivity. For example, poly(styrenesulfonate) (PEDOT:PSS) is one of the most frequently used organic semiconductors for biointerfaces, as it is biocompatible and also conducts both electrons and ions very effectively. The Malliaras group has developed several flexible electrocorticography (ECoG) probes that contain PEDOT:PSS-based electrodes [81]. It was shown that ECoG probes elicit a very minute glial response, and the device formed tight interfaces with the rat neurons and record multiple types of brain activities.

While these studies were focused on electrical recording of neurons, organic semiconductors with a high signal-to-noise ratio, desirable mechanical properties and biocompatibility may offer new ways for cellular engineering in the future. For example, the Lanzani group has shown that biocompatible poly(3-hexylthiophene) (P3HT) conjugated polymer can enhance the photosensitivity of living cells [82]. Internalized P3HT nanoparticles can respond to external light in vivo and then modulate the elongation or contraction behavior of the cell. Furthermore, polymer-enabled optical modulation also enhances the expression of opsin likely through the photoelectrochemical reactions on the surface of conjugated polymer. In parallel, the Głowacki group also achieve the control of living cell optically through organic conjugated molecules [83]. In one latest work, they assemble the conductive electrode (ITO) and organic semiconductors (H2Pc (p-type) and PTCDI (n-type)) into an organic electrolytic photocapacitor (OEPC). The whole device can record the electrophysiology of neuron or retinal ganglion cells. Upon light illumination, they are capable of inducing the opening of the ion channel and perturbing the membrane potential. The same p-n semiconductors can also be integrated with thin trilayer metal to accomplish the direct optical modulation for light-insensitive neurons [84]. The Głowacki group also arranged non-toxic organic semiconductors into a hierarchical nanocrystal structure by van der Waals forces [85]. The nanocrystals have designed hedgehog-shaped and are able to form a seamless interface with the cell membrane, which strengthens the efficiency of photostimulation. With a large absorption coefficient in visible light, many other organic nanoscale semiconductors may also be explored for photostimulation of cellular behaviors such as the control of the bioelectric current flow and the temperature-gated channels inside the cells.

4.2. Cardiac modulation

Cardiac conduction disorders are problems associated with the cardiac electrical conduction system, the structure which keeps the heartbeat at a rhythmic rate. Therapies for cardiac conduction disorder include modulating the beating frequency of a cardiomyocyte by either optical or mechanical methods [86-89]. As we previously mentioned, optogenetics can offer the necessary mechanistic insights into cellular processes to then modulate cardiac cell excitability via control of the light-activatable ion channel. However, optogenetic cardiac modulation is rarely used clinically because of the difficulty in gene transfection. Over the past few decades, researchers have been looking for alternative biocompatible methods for cardiac pacing or resynchronization through non-genetic approaches. For example, the Rollins group has found that pulsed 1.875-μm infrared laser light can resynchronize the frequency of heart beating without causing any biological damage in the meanwhile [87]. This approach is non-invasive and have high spatial resolutions. One key advantage of laser-induced pacing is that this technique does not need to further introduce any exogenous photosensitive element into cardiac cell, which largely reduces the complexity of pacemaker.

Semiconductors with photoelectric effect or photothermal effect are also promising material for designing pacemaker. For example, the Molokanova group fully exploited the unique photoelectric property of graphene for cardiac cells optical modulation [90]. Specifically, graphene is a unique semiconductor with zero band gap thus the electron and energy transfer are faster and more efficient when stimulated by the light. They have shown that graphene is biocompatible with living organisms and can modulate the heart contraction rate rapidly. Akin to the photoelectric effect, photothermal effect is also widely used in cardiac cell modulation. The Heinemann group precipitated the Au nanoparticles on the surface of the cell to induce cardiac cell contraction [91]. Au nanoparticles have notable photothermal effect thus are capable of increasing the local temperature to modulate cardiac cells.

Recently, cardiac engineering with Si nanowires has become a promising method because these nanostructures can form either extra- or intracellular interfaces with cells in the cardiac system (e.g., cardiomyocytes or myofibroblasts). The Tian group recently demonstrated that a freestanding polymer-supported Si nanowire mesh (Figs. 6(a)-6(c)) could optically induce the cultured neonatal rat cardiomyocytes to reach a specified beating frequency in a minimally invasive manner [24]. The mesh device contains randomly oriented Si nanowires with the SU-8 polymer grid network. Cardiomyocytes readily form interfaces with the nanowires. To actuate this engineered system, cardiomyocyte/Si interfaces were periodically exposed to optical pulse stimulation until the cells beat synchronically. Although immediate synchronized pacing (i.e., one light pulse immediately results in one cell beat) was not observed, the bioelectric activities in cardiomyocytes can be altered by the applied frequency of the light pulses and the cells eventually reached the targeted frequencies. Optical stimulation of the mesh would not cause any significant cytotoxicity, which thus extends its potential application in biomedicine. However, exact biological adaptation mechanisms of the observed increases in beating frequencies still remain unknown, although they maybe similar to other electrical pacing methods.

Figure 6.

Cardiac modulation. (a) Wide-field reflected light microscopy image of the polymer-supported Si nanowire mesh. (b) Scanning electron microscopy (SEM) image of polymer-supported Si nanowire mesh with false color. Blue parts represent the cardiomyocytes. Orange parts represent the polymer grid. White parts represent the PIN-silicon nanowires. (c) Diagram of beating pattern for cardiac cells before and after stimulation. Beating frequency of cardiomyocytes can reach the target frequency upon laser illumination. (d) The schematic of the cellular engineering methodology between cardiac cells and silicon nanowires. Silicon nanowires are internalized inside the myofibroblast (MF). The whole engineered cell then cocultured with cardiomyocyte (CM) or injected into heart tissue for further characterization. (e) Diagram of calcium wave propagation velocity shows that the propagation between MF and CM is faster than MF-MF or MF-intracellular (*P < 0.0001). Panels (a)–(c) are reproduced with permission from Ref. [24], © United States National Academy of Sciences 2019. Panels (d) and (e) are reproduced with permission from Ref. [25], © United States National Academy of Sciences 2019.

Similarly to the indirect neuromodulation, engineered intracellular interfaces can be established for cardiac modulation. In a recent report, myofibroblasts are engineered by internalizing nanocrystalline Si nanowires (Figs. 6(d) and 6(e)) [25]. Those engineered cells with photoresponsivity are able to remodel the electrical activities of intercellularly connected cardiomyocytes with light pulses. Electrical coupling between the engineered myofibroblasts and cardiomyocytes largely depends on the cell density and the location of the engineered cells. The cardiac contraction frequency increased and gradually synchronized with the optical stimulation, despite the absence of immediate feedback. Additionally, the beating activities of cardiomyocytes can be adapted for different optical intensities and frequencies. These intracellularly engineered myofibroblasts were also used to probe the bioelectric coupling between myofibroblasts and cardiomyocytes in vivo.

4.3. Microbial modulation

Bacteria and yeast cells represent other cell types that have also been engineered with semiconductor nanostructures [92, 93]. This strategy usually combines biological processes in their native cellular environments and the photoelectrochemical outputs from semiconductors, for applications such as the unassisted solar reduction of carbon dioxide to useful natural products. During this process, semiconductors carry out both the light-harvesting and delivery of reducing equivalents.

For example, the Yang group has reported that Sporomusa ovata can hybridize with Si nanowire arrays for the light-driven production of acetate [94]. The biocompatible light-capturing nanowire array can form conformable interfaces with the microbial system and provide high surface areas for catalysis. One unique advantage of this engineered system is that the semiconductor array not only increases the carbon dioxide fixation activity of the bacteria, but also creates a local anaerobic environment that allows strictly anaerobic bacteria to continue carbon dioxide reduction aerobically. Enhancement in oxygen tolerance of these engineered bacterial allows exhaust gas to be directed back to the system to increase efficiency. This strategy can be programmed to produce diverse natural products without any setup change in the components for light capture. In particular, a variety of complex organic compounds or molecular targets can be produced by this biocatalytic method with engineered E. coli.

In addition to one-dimensional semiconductors like nanowires, zero-dimensional semiconductors like cadmium sulfide (CdS) nanocrystals can also be used to mimic photosynthesis. The Yang group precipitated semiconductor CdS nanoparticles on the cell wall of a non-photosynthetic bacteria to mimic the natural photosynthesis process [95]. This engineered bacterial system contains non-photosynthetic CO₂-reducing bacterium Moorella thermoacetica and CdS nanoparticles with an appropriate band gap at the surface. The precipitation of CdS is induced by Cd²⁺ and a source of sulfur, such as cysteine. Upon illumination, CdS nanoparticles can produce photoelectrons, generating a reducing equivalent to help synthesize acetic acid from carbon dioxide (Fig. 7(a)). One interesting phenomenon of this engineered system is that bacteria and semiconductors can form close interfaces without any participation of organic ligands, thus improving efficiencies in charge transfer. Significantly, the photosynthesis ability can be adjusted easily by changing the concentration of the semiconductor. This system does not have catabolic energy loss during the dark cycles, indicating that the engineered bacterial system is more efficient than natural photosynthesis.

Figure 7.

Microbial modulation. (a) Schematic diagram shows the electron transfer and total pathway of semiconductor-engineered microbial system for artificial photosynthesis. Dashed line indicates the reducing equivalents are generated outside the cell. Solid line indicates reducing equivalents generated by direct electron transport to the cell. (b) Schematic diagram of the engineered microbial system illustrates how InP nanoparticles assist the generation of NADPH and synthesis of shikimic acid. Panel (a) is reproduced with permission from Ref. [95], © AAAS 2016. Panel (b) is reproduced with permission from Ref. [96], © AAAS 2018.

Indium phosphide (InP) nanoparticles represent another example that can be integrated into microbes to achieve artificial photosynthesis. The Joshi group reported a polyphenol-based assembly method to integrate InP nanoparticles with Saccharomyces cerevisiae [96]. InP nanoparticles firmly attached to the bacterial cell wall and were able to provide reducing equivalents (e.g., NADPH) to biological reductive processes upon optical stimulation (Fig. 7(b)). The decoupling of biosynthesis and cofactor regeneration makes the system efficiently generate alkaloid natural products or drug precursors such as shikimic acid [97]. Compared to the CdS nanocrystal, InP nanoparticles are more stable with oxygen and less cytotoxic in the biological environment, which strengthens its potential application in photosynthesis. Compared to Si, the direct bandgap of InP makes it suitable for absorbing a significant fraction of the solar spectrum, mainly increasing its adaptation in different microbial culture environments.

5. Tissue-scale engineering through semiconductor-based devices

Biological structures are hierarchical and their functions span across a range of length scales. As an extension of the single cell cellular engineering, semiconductors can also be engineered with tissue or even living animals to perform new functions [26]. Tissue engineering with semiconductors is more complicated than single cellular engineering as seamless integration at different length scales between tissues and semiconductor devices is hard to achieve. Nevertheless, researchers over the past few years have developed several practical methods to engineer tissues with Si-based nanoelectronics devices [98, 99].

For example, the Lieber group has built electronically-active synthetic tissues with Si nanowire FET, which performed the electrical recording of physiological activities within the engineered tissues [16]. As verified by the three-dimensional (3D) confocal microscope, Si nanowire FETs formed seamless interfaces with high densities of cells inside the engineered tissues. To achieve these interfaces, macroporous 3D nanoelectronics mimicking the tissue-scaffolds were first fabricated using a sacrificial layer and folding/rolling techniques. The semiconductor device elements, the Si nanowire FETs were distributed inside this macroporous structure to enable electronic sensing functions. Cells were seeded into the macroporous nanoelectronics to grow into cardiac patches, 3D neural cultures, and vascular constructs, from which various tests such as cardiac drug screening and pH sensing were performed (Fig. 8). These nanoelectronics scaffolds have low cytotoxicity and can be used as 3D electronic probes for long-term in vitro monitoring. Later on, the Dvir group further improved the synthetic tissue to achieve not only recording but also remote manipulation such as electrical perturbation and drug delivery [100]. They introduced large gold electrodes with nanoscale titanium nitride (TiN) deposition in order to provide enough current in biological environment to deliver signal and manipulate cellular behavior. Although they used conducting materials as the sensing elements, the similar methodology can be applied to semiconductor-based devices as well.

Figure 8.

Semiconductor-enabled tissue engineering. (a) Photographs of the vascular engineered tissue with pH sensing function. (b) (I) Micro-CT image of a tubular engineered tissue and (II) zoom-in view of the micro-CT image. Yellow arrow indicates the position of nanowire FET inside the engineered tissue. (c) The diagram shows that the engineered tissue is capable of sensing the pH by the signal in conductance. The schematic of the whole setup for pH sensing is shown in the inset figure. Panels (a)–(c) are reproduced with permission from Ref. [16], © Springer Nature 2012.

In addition to tissue engineering, Si-based nanoelectronics have also been integrated into living rodent brains as electrophysiology probes. For example, Yang et al. firstly designed a neuron-like electronic by mimicking the subcellular structure and mechanical properties of neuron [101]. The similarity in structure and mechanics alleviate the detrimental impact of synthetic component on nearby tissues. For device delivery into native tissues, Liu et al. pioneered a class of injectable mesh nanoelectronics and used the Si nanowire FETs to monitor rodent brain activities [18]. The open framework of the mesh structures allows interpenetration between neuronal cells and sensory devices to form minimally invasive and glial-free biointerfaces. In parallel, Xie et al. also invented a frozen probe-based device delivery method, which allows for the facile coupling of the flexible mesh electronics to an input/output (I/O) connector [102]. After being frozen in liquid nitrogen to increase rigidity, the mesh electronics penetrated rodent brain stereotaxically with minimal invasion to the tissues. Besides worked as brain probes, mesh electronics are also able to be injected into mouse eyes for long-term in vivo monitoring. Hong et al. has shown that mesh electronics form a stable interface with retinal ganglion cell and record the cellular activity without intercepting the normal function of eyes [103].

Recently, nanoelectronics have also been exploited to create the 3D “cyborg” organoids. For example, the Cohen-Karni group has designed a self-rolled biosensor array with the single-layer graphene to investigate the electrophysiological activity of the cardiac tissues in 3D [104]. The devices are fabricated by a self-rolling platform and can achieve multiple 3D geometries to target multiple types of tissues. In particular, they showed that the devices were able to map the 3D electrical signal propagation inside the cardiac organoids due to seamless biointerfaces. In parallel, the Liu group has built a 3D “cyborg” organoid from 2D cell layers [105]. For the device fabrication, the metal nanoelectronics are first transferred into a cell layer and then proliferation and migration of cells fold the whole structure into a spherical morphology. It has been shown that nanoelectronics are evenly distributed and form tight interfaces with cells. Thus, the devices can achieve long-term recording from large organoids to study biological behaviors or disease mechanisms. Although these works are focused on the graphene or metal electrode, similar methods can be used for semiconductor nanostructures as well.

6. Outlook

The exquisite structural and function designs and biocompatibility of semiconductor nanostructures broaden the horizon for numerous applications within cellular biophysical research and nanomedicine. Recent studies detailing the mechanical and electrical signals, whether that be through bioelectric potentials or cytoskeletal force transductions, create the opportunity for the implementation of semiconductor nanostructures to resolve biophysical behaviors at the organelle level [14]. These nanoscale semiconductors would allow for access to previously unreachable data regarding, e.g., the mechanisms of modulating endoplasmic reticulum redox activities as they can relate to insulin resistance, or the role of polar cytoskeletal filaments in creating intracellular bioelectric heterogeneity. The data gathered by semiconductor nanostructures is also pertinent for investigative medical studies where conventional methods of data gathering, such as optogenetic photostimulation and electrode-based probes, fail in integration with large-brained animals or activate an immune response due to mechanical invasiveness. The nanoscale devices derived from these semiconductor materials have demonstrated tight biointerfaces with cells and facile surface modifications, creating the potential for targeting specific cell types for minimally invasive medical diagnoses. Finally, traditional synthetic biology could also be coupled with semiconductor nanostructures and existing nanoelectronic or optoelectronic devices for more quick and accurate feedback loops in biointerfacing with cells. This promising avenue could lead to substantial leaps in semiconductor-enabled precision medicine that could resolve many previously unanswerable fundamental research questions.