Recently, microfluidic approaches have been utilized for fabricating well-controlled spherical or rounded gel microparticles, as well as cylindrical and hollow gel fibers. These are shapes that cannot be easily formed using soft or photolithography techniques. Furthermore, microfluidics offers the ability to continuously alter the particle shape and size, simply by changing the flow rate of the prepolymer solution. It does so without the need for masks and masters and is compatible with cells.

Cells and particles can be manipulated with excellent fidelity inside micrometer-sized channels or chambers [90,91], enabling experiments that require tightly controlled cellular microenvironments [3,92,93,94,95,96,97]. This and other biomedical engineering applications of microfluidics have been discussed in the context of drug discovery [98,99,100], stem cell biology [101,102], oncology [103], controlling cellular behavior (i.e., angiogenesis, migration, cellular interactions) [104,105,106], and tissue culture [98,100,107,108,109,110]. Specifically, issues concerning 3D cell-tissue complexes have been addressed, including adequate perfusion of tissues with nutrients and gases [111] as well as their stimulation with chemical, mechanical and electrical signals [112].

One of the most popular applications of microfluidics for generation of cell-containing materials relies on hydrodynamic (flow) focusing in glass capillaries and nozzles to encapsulate cells inside droplets or jets of polymer solutions. These structures are based on the laminar flow of the sample framed by an immiscible phase. The sample either includes cells suspended in an uncrosslinked hydrogel or two separate streams of cell solution and a hydrogel. The inner or dispersed phase of the resulting flowing structures is usually an aqueous solution. The outer, continuous phase is commonly a non-toxic oil such as corn oil (water-in-oil emulsion) or a lower-viscosity aqueous solution (water-in-water emulsion). The flow rates of all streams are adjusted to enable shearing off of individual droplets of the inner phase, due to the shear stress caused by the continuous phase. This setup allows for the generation of droplets containing the hydrogel prepolymer, with or without cells, even in water-in-water emulsions. Such emulsions take advantage of the difference in viscosity of the two miscible phases, such that for a brief period the two phases can be considered immiscible. A successful water-in-water emulsion has been demonstrated by Capron et al., who emulsified an alginate hydrogel solution in sodium caseinate [113]. Similarly, microemulsions of dextran were generated in an aqueous PEG solution, by stirring [114] and by microfluidic flow focusing [115]. In the first case, methacrylated dextran (dexMA) was used, which allowed for photocrosslinking of the gel precursor. In addition to flow focusing, hydrogel droplets can be generated in T-junction microfluidic structures. In this case, the aqueous and lipid phases do not co-flow and aqueous droplets are periodically sheared off by the faster flowing oil stream. In both flow focusing and T-junction devices, the use of a lipid phase can act to decrease the cell viability. Even when non-toxic liquids like corn oil are applied to shear off hydrogel droplets, the encapsulated cells near the droplet and particle surface are in contact with the surfactant and the oil. Then, the amount of nutrients available inside the microparticles is often only sufficient for a short culture period, especially in densely packed particles. Hence, it is important that the crosslinked particles be washed soon after fabrication and moved to a reservoir containing cell medium. Further, the washing process should be gentle, avoiding solvents and high centrifugation rates.

Several studies highlight the encapsulation of cells in hydrogel droplets using microfluidics. For example, Shintaku et al. introduced independently an alginate solution and a solution containing Ca²⁺ ions into a microfluidic channel and established a brief co-flow of the two species. Droplets of the aqueous solutions were sheared off periodically due to the shear stress caused by the high flow rate of oil [116]. Individual alginate droplets were shown to gel by adding Ca²⁺ ions, which replace Na⁺ ions as they diffuse through the droplets. Decreasing the crosslinker flow rate and increasing the flow rate of alginate resulted in a reduced droplet size. Furthermore, mechanical properties of the gel were affected by the degree of crosslinking, which depended on the crosslinking time and the crosslinker concentration. Similarly, single Na-alginate emulsions could be formed at the tip of a rotating micronozzle by gravity and cross-linked in CaCl₂ solution [117]. The droplet size was controlled using the inner nozzle diameter and its speed of rotation.

The flow focusing setup used for simple hydrogel microdroplet generation also enables the formation of so-called Janus droplets (Figure 2a,b). These droplets contain two distinct halves, which are generated by two miscible, co-flowing sample solutions. Such heterogeneous structures are of interest in cell co-culture applications. The Janus particle generation technique is highlighted in Figure 2a. Here, suspensions of carbon black and titanium oxide particles in an acrylic monomer solution were co-flowed to produce Janus microspheres with white and black halves in an aqueous solution [118]. The gel was thermally crosslinked outside the fluidic module. Non-spherical Janus microparticles could also be generated, as shown by Prasad et al. [119], who produced dyed micro-dumbbells with distinct organic and inorganic polymer regions (Figure 2b). Finally, Seiffert et al. showed that even more complex particles could be generated, containing three distinct units [120,121]. In their experiment, the standard two miscible phases (labeled with fluorescent green and red dyes) were complemented by a third hydrogel precursor solution prior to UV-crosslinking.

Multiple emulsions can be generated by forming droplets inside other droplets. This method requires the use of two or more nozzles or capillaries for flow-focusing. The emulsification procedure is independent of the presence of cells or other particles inside the liquid streams, so we first consider examples containing only the gel precursor. We subsequently review recent works utilizing cell-containing, multiple emulsion gels. Standard double emulsification methods include the single-step and double-step double emulsion techniques. The first was used to generate cell-compatible PNIPAAm microparticles with a core and shell (Figure 2c,d) [122]. The latter technique (Figure 2d) was exemplified in the encapsulation of a hydrophobic magnetic monomer solution surrounded by an acrylamide shell in fluorocarbon oil [123]. Although this particular double-emulsion example is not cell-compatible, it depicts an emulsification method that is generally compatible with cells and hydrogels. Usually, double emulsion methods lead to water-oil-water emulsions. The exact order of the inner, middle, and outside phases can be reversed, as long as the contacting phases are immiscible. Recently, however, water-water-oil emulsions have also been demonstrated: Yasukawa and colleagues [115] created monodisperse droplets of dextran in aqueous PEG, surrounded by hexadecane. Higher level emulsions were successfully demonstrated by Weitz and coworkers [124]. Their capillary systems offer excellent control over the number of encased droplets in double-, triple- and higher emulsions. This was exemplified in the generation of a triple emulsions of the temperature sensitive PNIPAAm hydrogel [125]. Here, hydrogel shells encased up to ten aqueous droplets floating in oil. A rapid increase from room temperature to 50 °C led to shrinking of the hydrogel. As a consequence, the aqueous droplets were expelled from the microcapsules. This observation depicts such multiple emulsions as potentially suitable for cell encapsulation and controlled cell release.

In all emulsions, the microdroplets are crosslinked into solid particles by chemical, thermal, or UV-initiated reactions. There are four main advantages in utilizing drop-generating microfluidic techniques: the high throughput capacity to generate millions of droplets per hour, uniform droplet sizes achieved by added surfactants, excellent regulation of droplet size and spacing via flow rate adjustments, and control over the average number of encapsulated cells. For example, a decrease in the flow rate of the dispersed phase relative to the continuous phase yields smaller droplets. Additionally, an increase in the continuous phase flow leads to a wider droplet spacing. Moreover, the mechanical properties of the resulting microgels can be adjusted by altering the rates of the hydrogel precursor and crosslinker flows.

The individual hydrogel droplets, spherical particles and fibers are sufficiently small to enable efficient permeation by gases, water and small molecules. The resulting high cell viability is a major advantage of using such gel structures for cell encapsulation. Thus, there are many examples of encapsulating cells inside hydrogel droplets. For example, a microfluidic T-junction was used to encapsulate E. coli in monodisperse droplets of PEG-DA (Figure 3a). The droplets were then photocrosslinked and the cells remained viable after overnight incubation [126]. Similarly, PEG-based microdroplets laden with NIH-3T3 fibroblasts were formed after stirring of the aqueous solution in mineral oil and photocrosslinking [127]. In another application, PEG-DA was mixed with a suspension of mammalian cells and patterned into cylindrical microstructures inside a microfluidic device [128]. The cell-laden structures were cultured for a week, and the cells retained high viability and enzyme activity throughout that period. Furthermore, an example of monodisperse alginate beads containing Jurkat cells is shown in Figure 3b [129]. Here, the effect of CaCO₃ crosslinker concentration on cell viability was studied. It was reported that lowering the crosslinker concentration led to improved cell viability. Franco et al. [127] later demonstrated successful encapsulation of neural stem cells in PEG-DA microspheres (Figure 3c). Additionally, the microfluidic flow focusing structure was employed to generate double-emulsions from two hydrogels and selectively trap cells in the inner or outer phase [130]. In this work, a highly viscous dextran suspension of Jurkat cells was co-flowed with a low-viscosity PEG solution. Due to the mismatch in viscosity, a dextran droplet was regularly enclosed inside a PEG droplet.

Chemically polymerizing hydrogels were also used for formation of spherical cell-laden microparticles. For example, Um and others (2008) first generated a hydrogel mixture of puramatrix and alginic acid [131]. A hydrogel precursor droplet containing HepG2 cells was then merged inside a microfluidic device with a drop of CaCl₂ crosslinker. Mixing of the prepolymer solution and the crosslinker in this way led to controlled hardening of the droplet. This high-throughput particle generation technique is useful for a variety of cell-based assay applications, e.g., drug screening. Another example of cell-laden particles based on peptides is shown in Figure 3d [132]. A self-assembling peptide (SAP) hydrogel precursor was mixed with a suspension of endothelial cells. Hydrogel droplets in mineral oil were formed in a flow focusing setup. The ionic cross-linker was delivered to the droplets in the form of powdered salt inside the oil phase. The encapsulated cells were observed to migrate and proliferate within the particles after a 3-day culture. Thus, since SAP hydrogels can promote cell adhesion, spreading, and differentiation, this technique is suitable for tissue engineering applications.

In a coaxial flow arrangement of glass capillaries or inside a microfluidic device, a jetting regime can often be observed in addition to the dripping or drop-making regime. In the jetting regime [124,133], the surface tension of the inner phase needs to be overcome by its Laplace pressure before the jet becomes unstable and droplets can be formed. The droplets are usually formed far from the nozzle or flow junction. Beyond the jetting regime, the cylinder of liquid simply flows along a channel or capillary without breaking up. It is in this region that polymer fibers and hollow cylinders can be formed [134,135]. For example, Kang et al. [136] used a flow-focusing setup to generate long gel microfibers with varying topography, e.g., by including air microbubbles or altering the flow rates. They could also alter the chemical structure of the fiber by alternating flows of different alginate mixtures. In this manner the hydrogel fibers could be spatially coded with either different hydrogels or different cell types. An example is shown in Figure 4. Here, the continuous flow of an alginate precursor solution was sequentially joined with a suspension of fibroblasts, followed by a suspension of primary rat hepatocytes and alternating with a merged flow of the two cell suspensions. The precursor solution gelled during the flow, such that a single long fiber with different cell-containing regions could be produced. Furthermore, Yeh et al. [137] utilize the jetting regime to generate solid microfibers from chitosan with tripolyphosphate as crosslinker. However, since these two materials individually have non-physiological pH values, cells could only be seeded onto the fiber surface after gel crosslinking and neutralization. Fibers were also generated in microfluidic devices from amino acid based polymers, namely N-carboxyanhydrides with triethylamine as crosslinker [138]. Finally, Hu et al. used PNIPAAm to form hollow fibers in co-axial microfluidic structures [135]. Because no oils are required in this technique, as opposed to the previously discussed droplet generation methods, jetting is considered to be more cell-friendly. However, the resulting microstructures are limited to fibers and hollow cylinders.

Aside from microfluidics, electrospraying and electrospinning are two methods that can also be used for generation of cell-containing gel particles and fibers. Nanoscale polymer fibers generated through electrospinning are commonly arranged in large porous network. Cells can attach to and proliferate within such a structure. However, in this section we focus on microscale fibers, which are sufficiently large for cell encapsulation.

The physics of electrospinning and -spraying in hydrogels can be summarized as follows: Applying a large electrical potential to a prepolymer solution dripping from a syringe needle leads to charging of the droplets. When surface tension is smaller than the resulting electrostatic repulsion, the droplet stretches until a stream of liquid is formed. At low surface tension droplets begin to form (electrospraying). In contrast, at a sufficiently high surface tension, the charged liquid stream remains stable (electrospinning) [139]. In the latter case the jet dries and charges concentrate on the jet surface. Electrostatic repulsion again drives the lengthening of the jet. The generated hydrogel stream is finally deposited onto a substrate, where it gels into fibers. The final shape of the electrospun fiber is controlled by several parameters: the inner diameter of the needle; applied electrical potential and flow rate; the molecular weight, viscosity and concentration of the material; and the distance between the needle and the collection substrate. Commonly, both electrospinning and -spraying setups rely on a single syringe pump to drive the flow. However, a droplet microfluidic setup can be used as an alternative. Hong et al. demonstrated that a flow-focusing setup could be applied to generate droplets prior to their exposure to an electric field [140]. By adjusting the droplet size and spacing they were able to control the number of encapsulated particles in each droplet.

The electrical potentials applied in these techniques are usually on the order of 0.1 to a few kV. The applied currents are in the mA range. These electrospinning and -spraying conditions have been reported as cell-compatible. For example, it was shown that by co-flowing a biosuspension with a polymer solution through two nested energized needles, it was possible to electrospin fibers of medical-grade PDMS containing pockets of cells and media (Figure 5a), without deleterious effects of the electrical potential on the cell viability [12].

Furthermore, Jayasinghe and colleagues successfully encapsulated living immortalized human embryonic kidney cells inside PEO- and PVA-based electrospun fibers, and inside electrosprayed spherical alginate beads [141]. An investigation of annexin expression in these cells over 5 days showed that the cell activity was comparable to a control population. It was concluded from these results that the applied electrical potential did not strongly affect the cell viability. Embryonic stem cells have also been encapsulated in electrosprayed bio-polymer particles, and retained high viability and pluripotency [14]. Finally, the inner structure of the electrospun fibers was shown to be controllable.

As an example, Klein et al. [142] suspended Pseudomonas putida bacteria in electrospun microtubes with porous walls (Figure 5b). The polyethylene oxide (PEO) based polymer solution and the bacterial suspension were flowed together through the inner of two coaxial needles. A polycaprolactone (PCL)-PEG-based shell polymer solution flowed through the outer needle. Evaporation of the solvent rendered the generated fibers porous. This strategy could potentially be exploited to form porous cell-containing materials with improved oxygen and nutrient transport. It should be noted that the electrospun fibers are deposited on a substrate along not predictable paths, making the controlled spatial placement of such fibers difficult. This is beneficial for generating larger hydrogel units containing cells, e.g., sheets with dimensions on the order of mm and even cm. However, if spatial organization of the generated fibers is important, then bioprinting is a more appropriate approach.

A major goal of tissue engineering is to develop viable 3D models of tissues and even organs. These models would then be used to study the effects of drugs and diseases inside the body [144,145]. Generating full organs in a laboratory is extremely complex. Therefore, usually only parts of individual tissues are grown [146,147]. However, there are ways to pattern these tissues in ways that represent their natural counterparts. Bioprinting is one such recent approach to creating 3D tissues in vitro, via cell-laden microgels. The bioprinting technique utilizes computer-controlled deposition of cells and structural materials such as hydrogels [21]. By imposing cell patterns using a robotic printing head, the cells naturally migrate and proliferate in the prescribed patterns [148]. The printed matter is then polymerized and transferred into an incubating chamber for culture. The incubating chamber can be fashioned out of a simple Petri dish, or a complex microfluidic device outfitted with perfusion channels, stimulation electrodes, and pulsatile flow controls [149]. In some cases, a large cell-encasing polymer structure can be replaced with small hydrogel units. This was demonstrated by Norotte et al., who showed that mm-sized agarose rods could be deposited on a substrate in a preprogrammed pattern [150]. These rods were used to direct the placement of the printed cell aggregates. The combined hydrogel and cell patterns then generated cylindrical, functionally homogeneous and also heterogeneous structures. The bioink containing the cell suspension and the biodegradable hydrogel are printed layer by layer. As a result, complex, preprogrammed 3D cell patterns inside the resulting structures can be generated, a feature that is challenging to achieve using jetting or electrospinning. A schematic of the bioprinting process and a printed branched alginate structure are shown in Figure 5c (top and middle). A light micrograph of the structure can be seen in Figure 5c, bottom [143].

Two types of bioprinters are commonly used: retro-fitted inkjet printers and extruders. Inkjet printers are affordable and simple to adapt to printing biological matter, but are often clogged with large cell aggregates [151,152]. Conversely, pressure-driven extrusion devices tend to be costly, but are better suited to applications utilizing high cell concentrations [153]. Commercial bioprinters include high precision pressure and temperature controllers to keep cells viable and maintain a low-viscosity polymer solution. These bioprinters also utilize several printing heads for simultaneous printing of several structures. Alternatively, multiple printing heads can be used for printing different cell types and biomaterials.

Bioprinting has been used widely in tissue engineering applications [147]. Examples of generated tissues include functional, synchronously beating cardiac tissues grown from embryonic cardiac and endothelial cells [154]; mammalian bladder tissue [155]; vascular structures from porcine aortic smooth muscle cells [150]; and human skin models from keratinocyte and fibroblast layers in collagen [156]. Alternatively, bioprinters could also be used to control the placement of growth factors in cellular microenvironments. This approach was applied to direct the differentiation of stem cells into muscle and bone cells [157] and morphing of osteoblasts into bone cells [158].