In vitro compartmentalization (IVC) technologies are discovered by Tawfik and Griffiths. They involve the compartmentalization of genes into aqueous microdroplets dispersed in an inert oil [
32]. The maintenance of the genotype-phenotype linkage in this system is achieved by the co-compartmentalization of single genes, the substrate for the enzymatic reaction being studied, and an in vitro transcription/translation solution for protein expression. Nowadays, this technology can be applied to compartmentalize cells or single genes in W/O emulsion droplets, W/O/W double emulsion droplets, microgels, giant liposomes, etc. Water-in-oil (W/O) emulsion droplets for directed evolution were first produced in a polydisperse system using traditional bulk emulsification devices: for single-cell experiments (
in vivo compartmentalization) [
33], or for single-enzyme experiments (
in vitro compartmentalization) [
34].
While the droplet boundary can largely restrict crosstalk, droplet sizes generated by bulk emulsification vary considerably and consequently the assay quality may differ from droplet to droplet, as differently sized droplets will contain different amounts of reagents. Nevertheless, polydisperse emulsions are still used today for protein engineering, especially if there is no need for quantitative measurement of the reaction product, i.e., if the selection is based on self-amplification or affinity “panning”. Bulk emulsification can be achieved using various laboratory techniques: with a stirring bar [
32,
35], using a tissue homogenizer (high-shear mixer) [
36], by vortexing [
37], and extrusion through a filter [
38].
4.1. Compartment manipulation
Four basic on-chip unit operations carried out during microfluidic screening of biocatalysts are reagent mixing, compartmentalization, sometimes followed by droplet consolidation by gelation, incubation, and sorting.
Droplet formulation: Emulsions must be formulated to allow droplets to reliably maintain their contents during incubation and subsequent manipulation without cross contamination or sample loss [
39]. Early research with mineral oil and silicon-based surfactant Abil EM 90 revealed transfer of fluorescent products into the continuous phase under certain conditions. However, the leakage of fluorophores can be reduced more than 10 times by adding 5% bovine serum albumin in the aqueous inner phase. Fluorinated oils such as Novec HFE7500 in combination with fluorinated surfactants offer an alternative carrier phase to prevent crosstalk [
40].
On-chip mixing of inlet streams: On-chip mixing of genetic material (cells or TT mixtures) and the reagent is usually done immediately before compartmentalization, which allows the initiation of an enzymatic reaction just prior to droplet formation. On chip mixing involves mixing a suspension of cells or enzymes with a fluorogenic substrate and lysis agent using a mixing junction located upstream of the droplet generation junction. The mixing junction can have a T-, X- or Y- shape. In the case of a cross junction [
41], cells suspended in a fermentation medium are injected through the main channel, while the substrate is delivered through two orthogonal side channels and the two streams are mixed by chaotic advection in the main channel [
42]. In the case of a Y-junction [
43], the cell suspension is delivered through one arm and the fluorogenic substrate through the second symmetrical arm of the Y-junction (
Figure 5).
Droplet generation (compartmentalization): Droplet generation units for single (two-phase) emulsions can be based on T-junctions (cross-flowing streams), flow focusing junctions (elongated flow), co-flow streams, and step emulsification. The two crucial requirements in this step are that droplets are formed in dripping regime and a single-gene encapsulation is achieved, i.e. droplets must be “monoclonal” and monodisperse.
T-Junction: In a T-junction droplet generator, the continuous phase (CP) flows within the main channel and the dispersed phase (DP) is injected from a side channel intersecting at 90° (
Figure 6 a). Droplets are generated due to shear forces induced by the continuous phase [
44,
45,
46] or pressure force if the main channel is completely obstructed by the dispersed phase [
47].
Co-flow droplet maker: Coflowing microfluidic droplet generators consist of two coaxial channels. The coaxial geometry is often formed by a smaller capillary tube being placed inside a larger concentric tube (
Figure 6 b). The inner capillary supplies the DP whilst the outer tube delivers the CP fluid [
48,
49].
Flow-focusing junction: In this geometry, the DP flows in the middle channel and is enveloped by the CP coming from either side (
Figure 6 c). The two liquid phases are usually forced through a small orifice downstream of the junction, and droplets are formed by viscous stress exerted on the inner phase by the surrounding outer phase flow. The geometry of these devices can vary from 3-D axisymmetric [
50] to quasi-2D planar [
51,
52] to completely 2D planar. Axisymmetric devices provide a DP phase through a cylindrical inlet, which is hydrodynamically focused by the CP as both phases pass through a circular baffle. In all droplet generators mentioned above, the droplet size depends on fluid flow rates, channel geometry and stream compositions [
52].
Microfluidic step emulsification: Microfluidic step droplet generators combine a shallow and narrow upstream microchannel delivering the DP and an abrupt (step-like) opening to a deep and wide reservoir (“well”) filled with the CP. The upstream narrow microchannel can terminate at the well or a wide terrace can be added between the narrow channel and the well to reduce the flow velocity of the DP and minimize their inertia [
53]. Step droplet generators can easily be parallelized by connecting many shallow channels to the same well [
54]. The well can contain a stagnant or perpendicularly flowing continuous phase [
55]. In addition, narrow channels can be etched on the top surface of the substrate as micro-grooves [
53] or through the entire cross section as straight-through channels [
56]. In step emulsification, the droplet size in dripping regime is independent on the CP flow rate, since droplet instability is not achieved due to shear forces generated by the CP but due to imbalance of capillary pressure along the DP interface when the droplet is deformed at the step [
57,
58] (
Figure 6 d).
Figure 6.
Lab on chip modules for microfluidic emulsifications (DP = dispersed phase, CP = continuous phase [
59].
Figure 6.
Lab on chip modules for microfluidic emulsifications (DP = dispersed phase, CP = continuous phase [
59].
The optimum droplet volume depends on the growth rate and incubation time of the cells. Using nanoliter- rather than picoliter-scale droplets is essential for screening of filamentous fungi [
60]. Typically, the time from which Aspergillus spores germinate to the stage where they start to display detectable enzymatic activity is 24 h and optimum levels of activity are only reached after several days of incubation. The requirement for long incubation time combined with the rapid growth of the fungal hyphae makes screening filamentous fungi in picoliter droplets difficult because the hyphal tips exit 250 pl droplets in ~15 h. In contrast, encapsulating single spores in 18 nl droplets allows growth of the branched mycelial network for up to 24 h confined in the droplet with the hyphal tips exiting the droplets only after incubation for 32 h.
Droplet incubation: Droplet incubation can last from several minutes up to several weeks and can be achieved on-chip or off-chip (in a test tube) [
43,
61]. On-chip droplet incubation can be achieved using wavy channels (with an added benefit that droplet contents are mixed during incubation period), rectangular storage chambers [
62] or microfluidic traps [
63]. An array of pillars can be utilized along the side walls of the culture chamber (
Figure 7 c) to allow carrier oil to flow while all droplets remain trapped inside, resulting in a highly packed droplet array inside the culture chamber [
64].
After incubation, droplets are reinjected and sorted based on readout, which can be based on fluorescence, image analysis, light scattering, surface tension, electrochemical signal, and buoyancy.