On-demand production of femtoliter drops in microchannels and their

Subscriber access provided by Iowa State University | Library

Article

On-demand production of femtoliter drops in microchannels and their use as biological reaction compartments Mostafa Shojaeian, François-Xavier Lehr, H. Ulrich Göringer, and Steffen Hardt Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b05063 • Publication Date (Web): 04 Feb 2019 Downloaded from http://pubs.acs.org on February 5, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

On-demand production of femtoliter drops in microchannels and their use as biological reaction compartments Mostafa Shojaeian,a François-Xavier Lehr,b H. Ulrich Göringer c and Steffen Hardt *a a

Institute for Nano- and Microfluidics, TU Darmstadt, 64287 Darmstadt, Germany. Bioinspired Communication Systems, TU Darmstadt, 64287 Darmstadt, Germany. c Molecular Genetics, TU Darmstadt, 64287 Darmstadt, Germany b

ABSTRACT: We present a method allowing to produce monodisperse droplets with volumes in the femtoliter range in a microchannel on demand. The method utilizes pulsed electric fields deforming the interface between an aqueous and an oil phase and pinching off droplets. Water and xanthan gum solutions are considered as disperse-phase liquids, and it is shown that the method can be applied even to solutions with a zero-shear rate viscosity more than 104 times higher than that of water. The droplet formation regimes are explored by systematically varying the pulse amplitude and duration as well as the salt concentration. The dependence of the process on the pulse amplitude can be utilized to tune the droplet size. To demonstrate the applicability of the electric-fielddriven droplet generator, it is shown that the droplets can be used as versatile biological reaction compartments. It is proven that droplets containing a cell-free transcription-translation system execute gene transcription and protein biosynthesis in a timely and programmable fashion. Moreover, it is verified that biomolecules inside the aqueous droplets such as small RNAs can be diffusionally activated from the outside to induce a ligand-driven biochemical switch.

Droplet-based protocols in microfluidic devices have found numerous applications in such different areas as bioanalytics 1, chemical synthesis 2, or drug delivery 3. Droplets can either be produced in a continuous stream or on-demand. The two most popular schemes for droplet generation in the continuous mode are based on the Tjunction and the flow-focusing configuration 4. Apart from such passive droplet production schemes, often temperature 5, acoustic 6, magnetic 7 or electric fields 8–10 are used to control the droplet production. When the task is to produce a droplet at a certain time at a specific location, or when samples that are only available in smallest amounts need to be processed, the ondemand production of droplets is favored over the continuous process. In this case, the primary goal is not to perform a high-throughput screening operation, which means that the droplet production rate is only of secondary importance. Typical applications in that context require outcomes of sample preparation steps to be wrapped into droplets. One example is the processing of droplets containing the products of an electrophoretic separation step11. Another example is transferring the product of a fluorescence-activated cell sorting step into droplets 12. The most widespread scheme for the on-demand production of droplets in microfluidic devices is based on pressure pulses. In that context, a number of systems have been presented that differ in the microchannel architecture and the method by which the pressure pulse is applied 13– 22 . That way, the production of droplets with volumes ranging from femtoliters to nanoliters has been demonstrated. The on-demand production of droplets by electric fields is especially appealing, because corresponding systems often do not require any moving parts. The principle of

electrowetting on dieletric (EWOD) can be utilized for producing droplets on demand by employing the local contact angle variation of an aqueous liquid wetting an array of microelectrodes 23,24. Gu et al. 25 combined EWOD with the hydrodynamic control of a flow-focusing device to produce droplets on demand. Droplet production from Taylor cones forms the foundation of the work reported here. When an electric field is applied to the interface of an aqueous liquid and a dielectric fluid, above a threshold field strength the interfacial tension forces are no longer strong enough to balance the Maxwell stress, which results in the ejection of a jet of the aqueous phase decaying into droplets. He et al. 26,27 used electric field pulses to generate femtoliter droplets on demand based on that principle. Chen et al. 28 produced sub-femtoliter droplets on demand by combining flow focusing with droplet ejection from Taylor cones. Apart from that, a few rather exotic principles have been applied to produce droplets on demand in microfluidic devices. Park et al. 29 applied a laser pulse to an aqueous solution, injecting droplets with volumes in the range of picoliters into a channel filled with corn oil. Collins et al. 30 integrated an interdigital transducer producing surface acoustic waves in a T-junction microfluidic device and were able to eject water droplets of picoliter volumes into an olive oil medium. The miniaturization of biological assays has become a desired feature in many biological research areas 31. Especially attractive in this context is the generation of droplets, which when produced in a high-throughput format allow the screening of very large parameter spaces, which represents an inherent characteristics of biological systems. Droplet-based systems have enabled a plethora of new applications ranging from medical diagnostics tests 32 to systems and synthetic biology 33. Microfluidic-based,

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

monodisperse droplet generators allow the formation of femto- to picoliter droplets 34,35, which can be used to compartmentalize in vitro reactions. This represents a celllike biological quality and, as a consequence, has been used to study biomolecular binding reactions 36, enzyme reactions 37, and other real-time scenarios. Monoclonal droplets containing single molecules have succesfully been used in emulsion polymerase chain reactions 38 and have been tested in single cell sequencing experiments 39. The reduction of the reaction volume down to nanoliters not only reduces reagent costs, it also avoids amplification and recombination artifacts 40, in addition to simplifying the integration with other "on chip"-devices. Droplet systems have also been used in directed in vitro evolution experiments 41 and have been applied to analyse protein/protein 42 and protein/nucleic acid interactions. Due to the efficient mixing of reactants in the droplets, these reactions proceed in a precise and quantifiable fashion, which can be used to extract high quality timeresolved kinetic data 43. Finally, water-in-oil droplets have recently been used as a promising tool to investigate cell/cell-communication in three-dimensional tissue-like materials 44. In this work, we further explore the principle of dropon-demand production from Taylor cones. Through extensive parameter studies we demonstrate that this principle may not only be applied to liquids with viscosities close to that of water, but also to highly viscous solutions with a zero-shear-rate viscosity roughly 14,000 times higher than that of water. This is especially relevant for biological fluids, which invariably contain high concentrations of cosolutes including biopolymers such as proteins and nucleic acids. Further, we explore the parameter space in which monodisperse droplets can be produced on-demand in terms of the electric pulse amplitude, pulse duration and the salinity of the solution. By deliberately choosing a specific parameter combination, droplets of a specific size can be produced. Finally, we demonstrate their use as biochemical reaction compartments. MATERIALS AND METHODS Experimental setup. Fig. 1 shows a schematic of the microfluidic device. The device consists of a T-junction microchannel with a constriction at the junction and a separate U-shaped microchannel, a part of which is directed parallel to the main channel of the T-junction. The water-filled U-channel serves as an electrode and is connected to a power supply. The other port of the power supply is connected to the inlet of the disperse-phase channel. The disperse-phase liquid is introduced into the Tchannel from one inlet where it partially displaces the continuous phase that was filled into the channel from the other inlet. The channels have a height of 20 µm, the other dimensions are specified in Fig. 1. To fabricate the microfluidic device, the soft lithography protocol was used. A SU-8 photoresist structure on a silicon wafer, purchased from Microchem, was used as a mold for the device. First polydimethylsiloxane (PDMS), purchased from Dow Corning (Sylgard 184 Silicone Elastomer), was mixed with the cross-linker in two different ratios of 10:1 and 20:1. Then the 10:1 PDMS/cross-linker mixture was poured onto

Page 2 of 19

the master structure and cured at 75 °C for 40 mins, resulting in a thick layer of PDMS containing the microchannels. The 20:1 PDMS/cross-linker mixture was spin coated on a glass slide at 2000 rpm for 30 seconds and cured for 10 minutes at 75 °C to form a sticky thin layer of PDMS. After punching the holes for inlets and outlets into the thick PDMS layer, it was gently brought in contact with the glass slide covered by the thin PDMS layer. The assembly was cured overnight at 75 °C, resulting in a microfluidic chip with channel walls fully made of PDMS.

Figure 1. Schematic of the experimental setup.

Polytetrafluoroethylene (PTFE)-tubing was used for connecting the microfluidic chip to 1 ml syringes (B. Braun Medical Inc.) filled with the disperse- and the continuousphase liquids. The tubing was connected to the PDMS inlets using metallic needles, which were connected to the power supply. A High Voltage Sequencer (LabSmith HVS448) was employed as the power supply to provide the desired voltage pulse. A high-speed camera (Redlake Motion Pro Series Y) was mounted on an inverted microscope (Nikon Eclipse Ti) to image the fluid dynamics inside the channels. Both the continuous and the disperse phase were introduced into the microchannels using syringe pumps (KD Scientific Inc.). To quantify the products of the biomolecular reactions, fluorescence microscopy was employed. For this purpose, a Nikon Ti Eclipse microscope with a Lumencor SPECTRAX light source was used. Materials. The working fluids in our experiments were water and aqueous polysaccharid solutions as the disperse phase and 1000 cSt silicone oil AP (Sigma-Aldrich) as the continuous phase. If not stated otherwise, all dispersephase liquids contained NaCl (17 mM). To prepare the


Page 3 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


polymer solutions, xanthan gum powder (from Xanthomonas campestris), purchased from Sigma-Aldrich, was dissolved in water at 40 °C, resulting in xanthan gum solutions with concentrations of 0.5, 1, 2 and 3 g/l. The rheological behavior of xanthan gum solution, a liquid with shear-dependent viscosity, has been reported in the literature and can be fitted by a number of well-known models 45. The dissolution of xanthan gum powder in water not only makes the viscosity shear-rate dependent, but also significantly increases the viscosity of the solutions at zero shear rate 46. For example, at zero shear the 3 g/l solution has a viscosity of about 13.9 Pa·s. The interfacial tensions between the aqueous phases and the oil phase were measured by a drop and bubble shape tensiometer (PAT-1, Sinterface Technologies, Germany). The measurements showed that the interfacial tension does not significantly change upon addition up to 3 g/l of xanthan gum into water. The obtained value (~35 mN/m) is consistent with literature values 47. PlasmidDNA (pBEST-OR2-OR1-Pr-UTR1-deGFP-T500; Addgene #40019) for the cell-free transcription/translation (TX/TL) reactions inside the droplets, was isolated by anion-exchange chromatography on silica NucleoBond® Xtra columns (Macherey-Nagel). DNA-concentrations were determined by UV-absorbance measurements at 260 nm. Cell-free TX/TL-reactions were performed using an E. coli-based TX-TL-competent protein extract (myTXTL®, Arbor Bioscience). In vitro transcription experiments inside the droplets were conducted using PCR-assembled DNA-fragments containing the sequence of the T7 promoter and the coding sequence of the iSpinach aptamer. All DNA-primer sequences are listed in Suppl. Table S1. PCR-fragments (50 nM) were purified (Monarch NA purification, NEB) and added to a transcription mix containing 2 mM of each NTP, 25 mM MgCl2, 40 mM Tris-HCl pH 8.0, 5 mM DTT, 1 mM spermidine, 20 nM of DFHBI and 70 μg/ml T7 RNA-polymerase. Reactions were incubated and microscopically monitored at 37 °C (TX-reactions) and at 29°C (TX/TL reactions). Experimental Protocol. During the droplet-formation experiments, the syringe pump providing the dispersephase fluid was running. The corresponding backpressure enabled keeping the aqueous liquid/oil interface at a desired location where it becomes exposed to the electric field originating from the U-shaped channel. The high voltage sequencer device was used to create a voltage pulse of several hundred volts with a duration of some milliseconds. At the beginning of an experiment, the main channel was filled with silicone oil. Then, by switching on the disperse-phase pump, the oil in about half of the channel was displaced by aqueous solution, resulting in a stationary liquid/liquid interface close to the junction. Subsequently, a voltage pulse was applied, and the jet formation and droplet pinch off were recorded by the highspeed camera at frame rates of 1000 fps or 4000 fps. The visible area of the generated droplets, , was measured using the ImageJ software (National Institute of Health, USA), and the droplet diameter was calculated by the / equation 4 / . For each case, the experiments

were repeated several times to compute the average droplet diameter and its standard deviation. Cell-free TX/TL-reactions inside the droplets were started by adding the deGFP-gene containing plasmid (pBEST-OR2OR1-Pr-UTR1-deGFP-T500) to a myTXTL® master mix at a final concentration of 5 nM. The reagents were mixed and prewarmed to 29 °C before being injected into the microfluidic device. Microscopic imaging was conducted at a constant temperature of 29 °C. To quantify the reaction product inside the droplets, the auto-fluorescence of the synthesized deGFP (λex 485 nm, λem 520 nm) was measured using fluorescence microscopy. Measurements were performed in 15 min intervals over 2 h using an exposure time of 500 ms. For quantification of the same reaction in the bulk, clear-bottom 384-well plates containing cell-free myTXTL® master mix (10 µl) and plasmid DNA (114 ng) were analyzed using a PHERAstar FSX microplate reader (BMG Biotech). Measurements were performed at 29 °C for 5 h, obtaining a readout of the deGFPfluorescence every 5 min.

RESULTS AND DISCUSSION Electric Field Distribution. Before analyzing the experimental results on droplet production, it is instructive to study the electric field distribution between the liquid/liquid meniscus and the U-shaped channel. For this purpose, the electrostatic potential was computed numerically using the commercial finite-element solver COMSOL Multiphysics. For simplicity, a two-dimensional model was set up, treating the aqueous phases filling the U-shaped channel and partially filling the main channel as conductors. These are visible as white areas in Fig. 2. Consequently, the tangential component of the electric field vanishes at the surface of the white areas. The relative dielectric permittivities of the PDMS substrate and silicon oil are about 2.8 and 2.9, respectively. The inset of Fig. 2 shows a microscopy image of the liquid/liquid meniscus. The electric field distribution was obtained by solving the Laplace equation, i.e.   0 , with  being the electric potential. The U-shaped channel was modeled as an isopotential surface with vanishing potential, and the electric potential at the surface of the water finger in the partially filled main channel was set to 400 V. At the other boundaries of the computational domain, a vanishing normal component of the electric field was imposed. An unstructured mesh of triangular elements with quadratic shape functions was employed. It was ensured that the results obtained are virtually grid-independent. The resulting electric field distribution is shown in Fig. 2. The colormap indicates the magnitude of the electric field strength in V/m. The lines are the electric field lines, and the vectors indicate the electric field vectors at the surface of the conductors. It its clearly visible that the maximum electric field strength is found at the circular-arc shaped liquid meniscus close to the lower channel wall (adjacent to the U-shaped channel). This is the position where we expect the liquid/liquid interface to break down, i.e. to form a Taylor cone from which droplets emerge.



Page 4 of 19

structure of the electric field, shown in Fig. 2. The electric field strength at the liquid/liquid interface is maximal in close vicinity of the channel wall. This is the position where the interface becomes unstable and the jet emerges. However, due to the proximity of the channel wall, the jet does not grow to a significant length before the droplet pinches off. This is presumably the reason why we obtain a much more uniform droplet size distribution than reported in the literature 26 (see discussion in section 3.3).

Figure 2. Electric-field distribution around the water finger in the main channel. The colormap indicates the electric field strength in V/m, the lines are electric field lines. The inset shows a corresponding microscopy image of the water finger.

Dynamics of Droplet Formation. One of the major challenges related to droplet production in microchannels is the droplet size distribution. On the one hand, a uniform droplet size is desired. On the other hand, while the production of droplets with volumes in the pico- or nanoliter range is well established, producing femtoliter droplets still poses some challenges. The challenges are related to the fact that in most microfluidic droplet generators, the droplet size is larger than the scale of the geometric features of the microchannels, which translates to very small features in the case of femtoliter droplets. So far there are two major technologies allowing to generate droplets significantly smaller than the smallest geometric scale, flow-focusing and electrospray technology. In flow focusing the diameter of the liquid jet from which droplets are produced is reduced by the sheath flow of a second immiscible fluid 48. Electrospray technology relies on the formation of Taylor cones 49,50. A Taylor cone results from the deformation of the surface of a conducting liquid to which an external electric field is applied. Up to a critical value of the electric field strength, a static deformation of the liquid surface is observed, which results from the balance of Maxwell and capillary stresses. When the critical value of the field strength is exceeded, a jet emerges that decays into small droplets. The electrospray approach was reconsidered by the group of Daniel T. Chiu as a method to produce small droplets inside microfluidic channels 26. Employing electric-field pulses, they demonstrated the on-demand generation of droplets significantly smaller than the channel dimensions. In our study, we have used a configuration inspired by their work to produce droplets by electric-field pulses. Fig. 3 shows time lapse images of the droplet production process taken at a frame rate of 4000 fps. Compared to Fig. 2, the frames are rotated by 90°. Water is used as the liquid from which the droplets are produced, the second liquid is silicon oil. The pulse amplitude and duration are 400 V and 10 ms. The liquid/liquid interface exposes a Taylor-conelike structure from which a single droplet with a diameter of about 8 µm pinches off. The jet is much shorter than the one reported in the literature 26 at the instant when the droplet is produced. This is due to the special asymmetric

Figure 3. Time-lapse images of the on-demand production of a water droplet, using a voltage of 400 V and pulse duration of 10 ms. The numbers below the individual frames denote the time in milliseconds.

Droplet Size Distribution. Since the uniformity of the droplet size distribution is a key feature of a microfluidic droplet generator, we performed systematic studies with different liquids for the droplet phase. Biological fluids often contain proteins, DNA and other polymers, resulting in a significant increase of the viscosity. To study the influence of viscosity, experiments with aqueous solutions of xanthan gum were performed. Xanthan-gum solutions show a shear-thinning behavior, i.e. the viscosity decreases as a function of the shear rate. Experiments with liquid/liquid systems in which one of the phases is a xanthan gum solution have the advantage that the dissolved polymer leaves the interfacial tension largely unaffected, such that the viscosity and the interfacial tension can be varied independently. The viscosity values given in table 1 are the values at zero shear rate obtained from the literature 46 , where, different from our experiments, desalinated water was used as a solvent. Since we only wish to give an


Page 5 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


indication of the order of magnitude of the aqueous-phase viscosity, we consider it sufficient to refer to the literature results. The experiments with different aqueous phases were performed with a pulse amplitude of 400 V and a pulse duration of 10 ms. Table 1 displays the properties of the different aqueous-phase liquids and the corresponding average droplet diameters obtained, together with the standard deviations of the droplet diameter. The average was obtained from at least 25 independent measurements for water and at least 10 measurements for each xanthangum solution. Overall, the standard deviation of the droplet diameter is sufficiently small, i.e. our method allows generating uniformly sized droplets. Increasing the viscosity of the aqueous phase slightly reduces the average droplet diameter, where for viscosities above 6·102 mPa·s hardly any change in the droplet diameter is observed. Remarkably, even for zero-shear rate viscosities 104-fold higher than that of water, still droplets with a size distribution comparable to that of water droplets are produced. For any microfluidic droplet generator, the question arises which parameters need to be varied to tailor the droplet size. One option would be to vary the height of the channel in which the droplets are produced. We have conducted experiments with water in channels of half the height (10 µm) of our standard channels and found the average droplet size to be reduced by a factor of 2.4. Table 1: Average droplet diameter and standard deviation of droplet diameter for aqueous-phase liquids containing different concentrations of xanthan gum. The viscosity values are the zero-shear literature values 46.

Xanthan gum concentration [g/l] 0 0.5 1 2 3

Zero shear rate viscosity [mPa·s] 1.01 6.00×102 1.46×103 3.59×103 1.39×104

Average droplet diameter [µm] 8.3 6.9 6.9 6.5 6.6

Standard deviation of droplet diameter [µm] 0.7 0.8 1 0.8 0.5

Comparing our results on the droplet size distribution with the results reported in the literature 26, we notice that we do not observe a broad size distribution with a bimodal structure, but a largely monodisperse size distribution. Fig. 4 shows the size distributions for water (blue filled bars) and for the xanthan gum solution at a concentration of 3 g/l (red open bars). Previously the formation of satellite droplets was reported 26. As already indicated, we hypothesize that the proximity of the wall limits the extension of the liquid jet and has a positive effect on the droplet size distribution, whereas in previous studies 26 the jet grows to a significant length before it decays into droplets. However, obtaining a monodisperse size distribution requires working in the right domain of the parameter space spanned by pulse amplitude and pulse duration, as will be shown in the following chapter.

Figure 4. Droplet size distribution for water (blue filled bars) and 3 g/l xanthan gum solution (red open bars). The data were normalized such that the integrals of the two histograms are the same.

Droplet Formation Regimes. The results reported so far were obtained in a domain of the parameter space in which single, virtually monodisperse droplets are produced. However, there are other domains in which different phenomena are observed. For applications of the method described in this paper it will be important to know how material and process parameters influence the results and in which domains one should work to obtain single droplets of uniform size. For this purpose, parameter studies with water droplets containing different amounts of NaCl were performed. The pulse amplitude was kept fixed at 400 V, and the pulse duration was varied. The results are displayed in Fig. 5a, outlining the different dynamic regimes. While black filled circles indicate the absence of droplet formation, open circles show the formation of multiple (≥ 2) droplets. Between these two regions lies an area where the ondemand production of single droplets is possible (colored circles). The different colors denote average droplet sizes. It can be seen that for the majority of NaCl concentrations, there is a pulse duration range that enables the formation of single droplets. Lower pulse durations result in no droplets, higher values in multiple droplets. At salt concentration ≥ 1 mol/l, the production of single droplets becomes impossible. Fig. 5b captures the dynamic regimes in the case that the pulse amplitude and the pulse duration are varied. The higher the pulse amplitude, the narrower the regime in which single droplets are formed. However, there is a threshold below which droplet formation is no longer possible. This lies between 300 V and 325 V. Of course the voltage amplitudes are specific for the design and the dimensions of the system outlined in Fig. 1. Rather than the voltage, it is the electric field strength that determines the physical response of the liquid/liquid interface. Fig. 5b also indicates that varying the pulse amplitude presents an option to tailor the droplet size. Achievable droplet diameters range from 3 µm (violet) to 9 µm (red). The



significance of the yellow symbol at 450 V is limited, since at that voltage the range of pulse durations in which single droplets are produced is already very narrow. Discarding the yellow symbol, the trend is that with increasing pulse amplitude, the droplet diameter increases.

Page 6 of 19

Figure 5. Droplet formation regimes for water. a) For varying pulse duration and NaCl concentration at a pulse amplitude of 400 V. b) For varying pulse duration and amplitude with 17 mM NaCl. Black filled circles indicate cases where no droplets are produced, the open circles denote the formation of multiple droplets. Grey filled circles indicate situations in which only in a part of the experiments a droplet was formed. The different colors encode the droplet diameters.

Droplets As Biological Reaction Compartments. Having demonstrated that the method described above allows the generation of single femtoliter-size droplets in a reproducible manner, it suggests itself to use the droplets as reaction compartments. We see an important potential application in the field of synthetic biology. Synthetic biology utilizes artificial genetic circuits, i.e. transcriptiontranslation systems are key components in that context. An important challenge is to bridge the gap between in-vivo implementations of artificial generic circuits (for example based on yeast cells) and their in-vitro counterparts. Therefore it is important to formulate in-vitro implementations that mimic the implementations in biological cells. In that context our drop-on demand generator brings along at least two key advantages. First, it allows creating biological reaction compartments with a size comparable to that of yeast cells, one of the most frequently used cell types in synthetic biology. The comparable size translates to a comparable number of biologically relevant molecules. That way, it becomes possible to mimic the stochastic behavior of biological cells. Second, it allows to mimic the crowding characteristic for biological cells. The interior of biological cells is densely packed with different molecular species, effectively providing a highly viscous environment. As apparent from Table 1, we have demonstrated the creation of droplets with a viscosity more than 104-fold higher than that of water.

To begin with, we performed two of the most fundamental biochemical reactions inside the droplets: gene transcription and mRNA-translation. For that we used an Escherichia coli-based cell-free transcriptiontranslation (TX-TL) system 51, in which a plasmid encoding a gene variant of the enhanced green fluorescent protein (deGFP) is first transcribed into mRNA and subsequently translated into protein. deGFP represents a start codon-optimized version of the enhanced (e)GFP, and its synthesis and bioactive folding can be monitored by the auto-fluorescence of the protein 52. The synthesis of deGFP was analyzed in a time-dependent manner, and the product of the reaction when performed at standard ("bulk")conditions in a microliter volume (10 µl) is quantified in Fig. 6a: deGFP-fluorescence starts at roughly 30 min. The protein concentration is a linear function of time up to 1 h and reaches a plateau roughly at 3 h. The TX-TL reaction mix was enclosed in aqueous droplets after 60 minutes of “bulk” reaction (arrow in Fig 6a), using the method described in the previous sections. Droplets with diameters between 3 µm and 8 µm were tested, corresponding to droplet volumes between 20 fl and 300 fl. As an example, Fig. 6b shows micrographs of the fluorescence signal of a 5.1 µm diameter droplet at three different points in time. A representative fluorescence signal of deGFP averaged over 23 droplets is shown in Fig. 6c. While the negative control containing no DNA template shows no traceable signal over the entire time interval, the fluorescence signal in the droplets increases over 2 h. A fast increase within the first 15 min is followed by a slower increase thereafter. As expected and previously


Page 7 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


demonstrated 53,54, the data show a high statistical variance, likely due to the stochastic repartition of plasmid DNA and TX-TL-extract components 55 in the small reaction volumes. In that sense, the droplets perfectly mimic the stochastic behavior of single cells 56.

Figure 6. On-demand droplets as biological reaction compartments. (a) Signal from the bulk TX-TL reaction generating autofluorescent deGFP-protein. (b) Microscopic images of a single droplet containing the cell-free TX-TL reaction mix at 0, 60 and 120 min of incubation. Scale bar: 2.5 μm. (c) Plot of the time-dependent accumulation of deGFP. Data points in black represent the average obtained from n = 23 droplets. Grey curve: Control experiment in the absence of plasmid DNA (n = 3). Error bars are standard deviations from the mean.

As an additional biologically relevant experiment, we tested whether low-molecular mass ligands can diffuse across the oil/water interface to bind to a cognate receptor molecule and execute a biochemical switch. For this we set up an in-vitro transcription reaction within the droplets to synthesize the "light-up" RNA aptamer iSpinach 57. iSpinach is a synthetic, intricately folded small RNAmolecule that can bind to low molecular mass compounds such as 3,5-difluoro-4-hydroxybenzylidene imidazolinone (DFHBI). RNA-binding triggers DFHBI to fluoresce. The fluorophore was solubilized in the oil phase inside the microfluidic device. About 2-3 min after droplet formation, the droplets started to fluoresce, reaching a stable plateau after 30 min. Representative fluorescence micrographs of a 5.9 µm diameter droplet at different time points are shown in Fig. 7a. Fig. 7b displays the averaged fluorescence intensity as a function of time from a sample

of 9 droplets. As before, the strong electric field required for droplet formation did not adversely affect the bioactive structure of the biomolecules, in that case the polyanionic RNA molecules. Similar to the case of the TX-TL system, this reinforces the compatibility of the droplets produced using the method described above with biomolecular reactions.

Figure 7. Droplet uptake experiment. In-vitro transcription of iSpinach RNA-aptamers inside aqueous droplets. Diffusion of the fluorogenic DFHBI-ligand from the surrounding oil-phase into the droplet results in the formation of iSpinach RNAaptamer/DFHBI complexes, which can be detected by their green fluorescence (λem 473 nm). (a) Time-lapse microscopic images (0, 10, 30 min) of a 108 fl droplet expressing the iSpinach RNA-aptamer. Scale bar: 3 μm. (b) Plot of the time-dependent formation of green fluorescent iSpinach RNAaptamer/DFHBI complexes. Data points in black represent the average from n = 9 droplets. Grey curve: Control experiment in the absence of DFHBI (n = 3). Error bars are standard deviations from the mean.

SUMMARY AND CONCLUSIONS We have demonstrated a principle that allows the ondemand formation of aqueous femtoliter-volume droplets in an oil phase. The method is based on electric field pulses deforming the aqueous-oil interface. The resulting droplets are close-to monodisperse. Using xanthan gum solutions as the aqueous phase, we have shown that even with solutions having a viscosity at zero shear rate more than 10,000 times higher than that of water, the method yields similar results as in the case of water. To achieve the production of single droplets on demand, the parameters characterizing our system have to be chosen with due diligence. In a space spanned by the salt concentration in the aqueous phase and the amplitude of the voltage pulse, the single droplet regime is delimited by regimes in which no droplets or multiple droplets are produced. The same is true in the space spanned by the pulse duration and the pulse amplitude. Varying the pulse amplitude allows tuning the droplet size. To demonstrate the applicability of the electric-fielddriven droplet generator, we showed that the droplets can be used as versatile biological reaction compartments. First, we demonstrated that droplets containing a cell-free transcription-translation system execute gene transcription and protein biosynthesis 51 in a timely and programmable fashion. Second, we verified that biomolecules inside the



aqueous droplets such as small RNAs can be diffusionally activated from the outside to induce a ligand-driven biochemical switch. As such the droplets mimick defined cell-like characteristics, making the on-demand droplet generator a promising tool for a variety of (bio)molecular studies including high-throughput sceening experiments.

ASSOCIATED CONTENT

(13)

(14)

(15)

Supporting Information Supporting information on genetic sequences and on fluorescent traces of individual droplets is provided with the submission.

(16) (17)

AUTHOR INFORMATION

(18)

Corresponding Author * Email: [email protected]

Notes

(19)

There are no conflicts to declare.

ACKNOWLEDGMENT

(20)

We thank V. Noireaux for plasmid pBEST-OR2-OR1-PrUTR1-deGFP-T500 and H. Köppl for discussions and access to the microscopy equipment in his laboratory. M. Shojaeian and F. X. Lehr kindly acknowledge the support from the LOEWE CompuGene Project, funded by the Hessian Ministry of Science and Art.

(21)

(22) (23)

REFERENCES (1) (2)

(3) (4) (5) (6) (7) (8)

(9)

(10) (11) (12)

Guo, M. T.; Rotem, A.; Heyman, J. A.; Weitz, D. A. Droplet Microfluidics for High-Throughput Biological Assays. Lab Chip 2012, 12 (12), 2146. Elvira, K. S.; i Solvas, X. C.; Wootton, R. C. R.; DeMello, A. J. The Past, Present and Potential for Microfluidic Reactor Technology in Chemical Synthesis. Nat. Chem. 2013, 5 (11), 905–915. Zhao, C.-X. Multiphase Flow Microfluidics for the Production of Single or Multiple Emulsions for Drug Delivery. Adv. Drug Deliv. Rev. 2013, 65 (11–12), 1420–1446. Seemann, R.; Brinkmann, M.; Pfohl, T.; Herminghaus, S. Droplet Based Microfluidics. Reports Prog. Phys. 2012, 75 (1), 016601. Porter, D.; Savage, J. R.; Cohen, I.; Spicer, P.; Caggioni, M. Temperature Dependence of Droplet Breakup in 8CB and 5CB Liquid Crystals. Phys. Rev. E 2012, 85 (4), 041701. Cheung, Y. N.; Qiu, H. Acoustic Microstreaming for Droplet Breakup in a Microflow-Focusing Device. Appl. Phys. Lett. 2010, 97 (13), 133111. Tan, S.-H.; Nguyen, N.-T.; Yobas, L.; Kang, T. G. Formation and Manipulation of Ferrofluid Droplets at a Microfluidic T -Junction. J. Micromechanics Microengineering 2010, 20 (4), 045004. Link, D. R.; Grasland-Mongrain, E.; Duri, A.; Sarrazin, F.; Cheng, Z.; Cristobal, G.; Marquez, M.; Weitz, D. A. Electric Control of Droplets in Microfluidic Devices. Angew. Chemie Int. Ed. 2006, 45 (16), 2556–2560. Kim, H.; Luo, D.; Link, D.; Weitz, D. A.; Marquez, M.; Cheng, Z. Controlled Production of Emulsion Drops Using an Electric Field in a Flow-Focusing Microfluidic Device. Appl. Phys. Lett. 2007, 91 (13), 133106. Shojaeian, M.; Hardt, S. Fast Electric Control of the Droplet Size in a Microfluidic T-Junction Droplet Generator. Appl. Phys. Lett. 2018, 112, 194102. Keyon, A. S. A.; Guijt, R. M.; Bolch, C. J.; Breadmore, M. C. Droplet Microfluidics for Postcolumn Reactions in Capillary Electrophoresis. Anal. Chem. 2014, 86 (23), 11811–11818. Birchler, A.; Berger, M.; Jäggin, V.; Lopes, T.; Etzrodt, M.; Misun, P. M.; Pena-Francesch, M.; Schroeder, T.; Hierlemann, A.; Frey, O. Seamless Combination of Fluorescence-Activated Cell Sorting and Hanging-Drop Networks for Individual Handling and Culturing of Stem Cells and Microtissue Spheroids.

(24) (25)

(26) (27) (28)

(29) (30) (31) (32) (33) (34)

(35)

Page 8 of 19

Anal. Chem. 2016, 88 (2), 1222–1229. Lorenz, R. M.; Edgar, J. S.; Jeffries, G. D. M.; Chiu, D. T. Microfluidic and Optical Systems for the On-Demand Generation and Manipulation of Single Femtoliter-Volume Aqueous Droplets. Anal. Chem. 2006, 78 (18), 6433–6439. He, M.; Edgar, J. S.; Jeffries, G. D. M.; Lorenz, R. M.; Shelby, J. P.; Chiu, D. T. Selective Encapsulation of Single Cells and Subcellular Organelles into Picoliter- and Femtoliter-Volume Droplets. Anal. Chem. 2005, 77 (6), 1539–1544. Jung, S.-Y.; Retterer, S. T.; Collier, C. P. On-Demand Generation of Monodisperse Femtolitre Droplets by Shape-Induced Shear. Lab Chip 2010, 10 (20), 2688. Zhou, H.; Yao, S. A Facile On-Demand Droplet Microfluidic System for Lab-on-a-Chip Applications. Microfluid. Nanofluidics 2014, 16 (4), 667–675. Yu, M.; Hou, Y.; Zhou, H.; Yao, S. An On-Demand Nanofluidic Concentrator. Lab Chip 2015, 15 (6), 1524–1532. Lin, B.-C.; Su, Y.-C. On-Demand Liquid-in-Liquid Droplet Metering and Fusion Utilizing Pneumatically Actuated Membrane Valves. J. Micromechanics Microengineering 2008, 18 (11), 115005. Zeng, S.; Li, B.; Su, X.; Qin, J.; Lin, B. Microvalve-Actuated Precise Control of Individual Droplets in Microfluidic Devices. Lab Chip 2009, 9 (10), 1340. Galas, J.-C.; Bartolo, D.; Studer, V. Active Connectors for Microfluidic Drops on Demand. New J. Phys. 2009, 11 (7), 075027. Guo, F.; Liu, K.; Ji, X.-H.; Ding, H.-J.; Zhang, M.; Zeng, Q.; Liu, W.; Guo, S.-S.; Zhao, X.-Z. Valve-Based Microfluidic Device for Droplet on-Demand Operation and Static Assay. Appl. Phys. Lett. 2010, 97 (23), 233701. Xu, J.; Attinger, D. Drop on Demand in a Microfluidic Chip. J. Micromechanics Microengineering 2008, 18 (6), 065020. Sung Kwon Cho; Hyejin Moon; Chang-Jin Kim. Creating, Transporting, Cutting, and Merging Liquid Droplets by Electrowetting-Based Actuation for Digital Microfluidic Circuits. J. Microelectromechanical Syst. 2003, 12 (1), 70–80. Srinivasan, V.; Pamula, V. K.; Fair, R. B. An Integrated Digital Microfluidic Lab-on-a-Chip for Clinical Diagnostics on Human Physiological Fluids. Lab Chip 2004, 4 (4), 310–315. Gu, H.; Murade, C. U.; Duits, M. H. G.; Mugele, F. A Microfluidic Platform for On-Demand Formation and Merging of Microdroplets Using Electric Control. Biomicrofluidics 2011, 5 (1), 011101. He, M.; Kuo, J. S.; Chiu, D. T. Electro-Generation of Single Femtoliter- and Picoliter-Volume Aqueous Droplets in Microfluidic Systems. Appl. Phys. Lett. 2005, 87 (3), 031916. He, M.; Kuo, J. S.; Chiu, D. T. Effects of Ultrasmall Orifices on the Electrogeneration of Femtoliter-Volume Aqueous Droplets. Langmuir 2006, 22, 6408–6413. Chen, I.; Tsai, H.; Chang, C.; Zheng, G.; Su, Y. Sensors and Actuators B : Chemical Electric-Field Triggered , on-Demand Formation of Sub-Femtoliter Droplets. Sensors Actuators B. Chem. 2018, 260, 541–553. Park, S.-Y.; Wu, T.-H.; Chen, Y.; Teitell, M. A.; Chiou, P.-Y. High-Speed Droplet Generation on Demand Driven by Pulse Laser-Induced Cavitation. Lab Chip 2011, 11 (6), 1010. Collins, D. J.; Alan, T.; Helmerson, K.; Neild, A. Surface Acoustic Waves for On-Demand Production of Picoliter Droplets and Particle Encapsulation. Lab Chip 2013, 13 (16), 3225. Taly, V.; Kelly, B. T.; Griffiths, A. D. Droplets as Microreactors for High-Throughput Biology. ChemBioChem 2007, 8 (3), 263– 272. Kaminski, T. S.; Scheler, O.; Garstecki, P. Droplet Microfluidics for Microbiology: Techniques, Applications and Challenges. Lab Chip 2016, 16 (12), 2168–2187. Schaerli, Y.; Hollfelder, F. The Potential of Microfluidic Waterin-Oil Droplets in Experimental Biology. Mol. Biosyst. 2009, 5 (12), 1392. Shim, J.; Ranasinghe, R. T.; Smith, C. A.; Ibrahim, S. M.; Hollfelder, F.; Huck, W. T. S.; Klenerman, D.; Abell, C. Ultrarapid Generation of Femtoliter Microfluidic Droplets for Single-Molecule-Counting Immunoassays. ACS Nano 2013, 7 (7), 5955–5964. Wang, F.; Burns, M. A. Performance of Nanoliter-Sized DropletBased Microfluidic PCR. Biomed. Microdevices 2009, 11 (5),


Page 9 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(36) (37) (38) (39)

(40)

(41)

(42) (43)

(44) (45) (46)


1071–1080. Shembekar, N.; Hu, H.; Eustace, D.; Merten, C. A. Single-Cell Droplet Microfluidic Screening for Antibodies Specifically Binding to Target Cells. Cell Rep. 2018, 22 (8), 2206–2215. Fradet, E.; Bayer, C.; Hollfelder, F.; Baroud, C. N. Measuring Fast and Slow Enzyme Kinetics in Stationary Droplets. Anal. Chem. 2015, 87 (23), 11915–11922. Williams, R.; Peisajovich, S. G.; Miller, O. J.; Magdassi, S.; Tawfik, D. S.; Griffiths, A. D. Amplification of Complex Gene Libraries by Emulsion PCR. Nat. Methods 2006, 3 (7), 545–550. Margulies, M.; Egholm, M.; Altman, W. E.; Attiya, S.; Bader, J. S.; Bemben, L. A.; Berka, J.; Braverman, M. S.; Chen, Y.-J.; Chen, Z.; et al. Genome Sequencing in Microfabricated HighDensity Picolitre Reactors. Nature 2005, 437 (7057), 376–380. Marcy, Y.; Ishoey, T.; Lasken, R. S.; Stockwell, T. B.; Walenz, B. P.; Halpern, A. L.; Beeson, K. Y.; Goldberg, S. M. D.; Quake, S. R. Nanoliter Reactors Improve Multiple Displacement Amplification of Genomes from Single Cells. PLoS Genet. 2007, 3 (9), e155. Leemhuis, H.; Stein, V.; Griffiths, A. D.; Hollfelder, F. New Genotype–phenotype Linkages for Directed Evolution of Functional Proteins. Curr. Opin. Struct. Biol. 2005, 15 (4), 472– 478. Srisa-Art, M.; DeMello, A. J.; Edel, J. B. High-Throughput DNA Droplet Assays Using Picoliter Reactor Volumes. Anal. Chem. 2007, 79, 6682–6689. Huebner, A.; Olguin, L. F.; Bratton, D.; Whyte, G.; Huck, W. T. S.; de Mello, A. J.; Edel, J. B.; Abell, C.; Hollfelder, F. Development of Quantitative Cell-Based Enzyme Assays in Microdroplets. Anal. Chem. 2008, 80 (10), 3890–3896. Booth, M. J.; Schild, V. R.; Graham, A. D.; Olof, S. N.; Bayley, H. Light-Activated Communication in Synthetic Tissues. Sci. Adv. 2016, 2 (4), e1600056–e1600056. Song, K. W.; Kim, Y. S.; Chang, G. S. Rheology of Concentrated Xanthan Gum Solutions: Steady Shear Flow Behavior. Fibers Polym. 2006, 7 (2), 129–138. Wyatt, N. B.; Liberatore, M. W. Rheology and Viscosity Scaling

(47) (48) (49) (50) (51) (52) (53) (54) (55) (56) (57)

of the Polyelectrolyte Xanthan Gum. J. ofAppliedPolymer Sci. 2009, 114, 4076–4084. Muthamizhi, K.; Kalaichelvi, P.; Powar, S. T.; Jaishree, R. Investigation and Modelling of Surface Tension of Power-Law Fluids. Rsc Adv. 2014, 4, 9771–9776. Anna, S. L.; Bontoux, N.; Stone, H. A. Formation of Dispersions Using “Flow Focusing” in Microchannels. Appl. Phys. Lett. 2003, 82 (3), 364–366. Zeleny, J. The Electrical Discharge from Liquid Points , and a Hydrostatic Method of Measuring the Electric Intensity at Their Surfaces. Phys. Rev. 1914, 3 (2), 69–91. Taylor, G. Disintegration of Water Drops in an Electric Field. Proc. R. Soc. London A 1964, 280, 383–397. Shin, J.; Noireaux, V. Efficient Cell-Free Expression with the Endogenous E. Coli RNA Polymerase and Sigma Factor 70. J. Biol. Eng. 2010, 4 (1), 8. Garamella, J.; Marshall, R.; Rustad, M.; Noireaux, V. The All E. Coli TX-TL Toolbox 2.0: A Platform for Cell-Free Synthetic Biology. ACS Synth. Biol. 2016, 5 (4), 344–355. Kato, A.; Yanagisawa, M.; Sato, Y. T.; Fujiwara, K.; Yoshikawa, K. Cell-Sized Confinement in Microspheres Accelerates the Reaction of Gene Expression. Sci. Rep. 2012, 2 (1), 283. Sharma, B.; Takamura, Y.; Shimoda, T.; Biyani, M. A Bulk SubFemtoliter in Vitro Compartmentalization System Using SuperFine Electrosprays. Sci. Rep. 2016, 6 (1), 26257. Karig, D. K.; Jung, S.-Y.; Srijanto, B.; Collier, C. P.; Simpson, M. L. Probing Cell-Free Gene Expression Noise in Femtoliter Volumes. ACS Synth. Biol. 2013, 2 (9), 497–505. Pietrini, A. V.; Luisi, P. L. Cell-Free Protein Synthesis through Solubilisate Exchange in Water/Oil Emulsion Compartments. ChemBioChem 2004, 5 (8), 1055–1062. Autour, A.; Westhof, E.; Ryckelynck, M. ISpinach: A Fluorogenic RNA Aptamer Optimized for in Vitro Applications. Nucleic Acids Res. 2016, 44 (6), 2491–2500.



Page 10 of 19

On-demand production of femtoliter drops in microchannels and their use as biological reaction compartments Mostafa Shojaeian,a François-Xavier Lehr,b H. Ulrich Göringer c and Steffen Hardt *a a

Institute for Nano- and Microfluidics, TU Darmstadt, 64287 Darmstadt, Germany. Bioinspired Communication Systems, TU Darmstadt, 64287 Darmstadt, Germany. c Molecular Genetics, TU Darmstadt, 64287 Darmstadt, Germany b

ABSTRACT: We present a method allowing to produce monodisperse droplets with volumes in the femtoliter range in a microchannel on demand. The method utilizes pulsed electric fields deforming the interface between an aqueous and an oil phase and pinching off droplets. Water and xanthan gum solutions are considered as disperse-phase liquids, and it is shown that the method can be applied even to solutions with a zero-shear rate viscosity more than 104 times higher than that of water. The droplet formation regimes are explored by systematically varying the pulse amplitude and duration as well as the salt concentration. The dependence of the process on the pulse amplitude can be utilized to tune the droplet size. To demonstrate the applicability of the electric-fielddriven droplet generator, it is shown that the droplets can be used as versatile biological reaction compartments. It is proven that droplets containing a cell-free transcription-translation system execute gene transcription and protein biosynthesis in a timely and programmable fashion. Moreover, it is verified that biomolecules inside the aqueous droplets such as small RNAs can be diffusionally activated from the outside to induce a ligand-driven biochemical switch.

Droplet-based protocols in microfluidic devices have found numerous applications in such different areas as bioanalytics 1, chemical synthesis 2, or drug delivery 3. Droplets can either be produced in a continuous stream or on-demand. The two most popular schemes for droplet generation in the continuous mode are based on the Tjunction and the flow-focusing configuration 4. Apart from such passive droplet production schemes, often temperature 5, acoustic 6, magnetic 7 or electric fields 8–10 are used to control the droplet production. When the task is to produce a droplet at a certain time at a specific location, or when samples that are only available in smallest amounts need to be processed, the ondemand production of droplets is favored over the continuous process. In this case, the primary goal is not to perform a high-throughput screening operation, which means that the droplet production rate is only of secondary importance. Typical applications in that context require outcomes of sample preparation steps to be wrapped into droplets. One example is the processing of droplets containing the products of an electrophoretic separation step11. Another example is transferring the product of a fluorescence-activated cell sorting step into droplets 12. The most widespread scheme for the on-demand production of droplets in microfluidic devices is based on pressure pulses. In that context, a number of systems have been presented that differ in the microchannel architecture and the method by which the pressure pulse is applied 13– 22 . That way, the production of droplets with volumes ranging from femtoliters to nanoliters has been demonstrated. The on-demand production of droplets by electric fields is especially appealing, because corresponding systems often do not require any moving parts. The principle of

electrowetting on dieletric (EWOD) can be utilized for producing droplets on demand by employing the local contact angle variation of an aqueous liquid wetting an array of microelectrodes 23,24. Gu et al. 25 combined EWOD with the hydrodynamic control of a flow-focusing device to produce droplets on demand. Droplet production from Taylor cones forms the foundation of the work reported here. When an electric field is applied to the interface of an aqueous liquid and a dielectric fluid, above a threshold field strength the interfacial tension forces are no longer strong enough to balance the Maxwell stress, which results in the ejection of a jet of the aqueous phase decaying into droplets. He et al. 26,27 used electric field pulses to generate femtoliter droplets on demand based on that principle. Chen et al. 28 produced sub-femtoliter droplets on demand by combining flow focusing with droplet ejection from Taylor cones. Apart from that, a few rather exotic principles have been applied to produce droplets on demand in microfluidic devices. Park et al. 29 applied a laser pulse to an aqueous solution, injecting droplets with volumes in the range of picoliters into a channel filled with corn oil. Collins et al. 30 integrated an interdigital transducer producing surface acoustic waves in a T-junction microfluidic device and were able to eject water droplets of picoliter volumes into an olive oil medium. The miniaturization of biological assays has become a desired feature in many biological research areas 31. Especially attractive in this context is the generation of droplets, which when produced in a high-throughput format allow the screening of very large parameter spaces, which represents an inherent characteristics of biological systems. Droplet-based systems have enabled a plethora of new applications ranging from medical diagnostics tests 32 to systems and synthetic biology 33. Microfluidic-based,


Page 11 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


monodisperse droplet generators allow the formation of femto- to picoliter droplets 34,35, which can be used to compartmentalize in vitro reactions. This represents a celllike biological quality and, as a consequence, has been used to study biomolecular binding reactions 36, enzyme reactions 37, and other real-time scenarios. Monoclonal droplets containing single molecules have succesfully been used in emulsion polymerase chain reactions 38 and have been tested in single cell sequencing experiments 39. The reduction of the reaction volume down to nanoliters not only reduces reagent costs, it also avoids amplification and recombination artifacts 40, in addition to simplifying the integration with other "on chip"-devices. Droplet systems have also been used in directed in vitro evolution experiments 41 and have been applied to analyse protein/protein 42 and protein/nucleic acid interactions. Due to the efficient mixing of reactants in the droplets, these reactions proceed in a precise and quantifiable fashion, which can be used to extract high quality timeresolved kinetic data 43. Finally, water-in-oil droplets have recently been used as a promising tool to investigate cell/cell-communication in three-dimensional tissue-like materials 44. In this work, we further explore the principle of dropon-demand production from Taylor cones. Through extensive parameter studies we demonstrate that this principle may not only be applied to liquids with viscosities close to that of water, but also to highly viscous solutions with a zero-shear-rate viscosity roughly 14,000 times higher than that of water. This is especially relevant for biological fluids, which invariably contain high concentrations of cosolutes including biopolymers such as proteins and nucleic acids. Further, we explore the parameter space in which monodisperse droplets can be produced on-demand in terms of the electric pulse amplitude, pulse duration and the salinity of the solution. By deliberately choosing a specific parameter combination, droplets of a specific size can be produced. Finally, we demonstrate their use as biochemical reaction compartments. MATERIALS AND METHODS Experimental setup. Fig. 1 shows a schematic of the microfluidic device. The device consists of a T-junction microchannel with a constriction at the junction and a separate U-shaped microchannel, a part of which is directed parallel to the main channel of the T-junction. The water-filled U-channel serves as an electrode and is connected to a power supply. The other port of the power supply is connected to the inlet of the disperse-phase channel. The disperse-phase liquid is introduced into the Tchannel from one inlet where it partially displaces the continuous phase that was filled into the channel from the other inlet. The channels have a height of 20 µm, the other dimensions are specified in Fig. 1. To fabricate the microfluidic device, the soft lithography protocol was used. A SU-8 photoresist structure on a silicon wafer, purchased from Microchem, was used as a mold for the device. First polydimethylsiloxane (PDMS), purchased from Dow Corning (Sylgard 184 Silicone Elastomer), was mixed with the cross-linker in two different ratios of 10:1 and 20:1. Then the 10:1 PDMS/cross-linker mixture was poured onto

the master structure and cured at 75 °C for 40 mins, resulting in a thick layer of PDMS containing the microchannels. The 20:1 PDMS/cross-linker mixture was spin coated on a glass slide at 2000 rpm for 30 seconds and cured for 10 minutes at 75 °C to form a sticky thin layer of PDMS. After punching the holes for inlets and outlets into the thick PDMS layer, it was gently brought in contact with the glass slide covered by the thin PDMS layer. The assembly was cured overnight at 75 °C, resulting in a microfluidic chip with channel walls fully made of PDMS.

Figure 1. Schematic of the experimental setup.

Polytetrafluoroethylene (PTFE)-tubing was used for connecting the microfluidic chip to 1 ml syringes (B. Braun Medical Inc.) filled with the disperse- and the continuousphase liquids. The tubing was connected to the PDMS inlets using metallic needles, which were connected to the power supply. A High Voltage Sequencer (LabSmith HVS448) was employed as the power supply to provide the desired voltage pulse. A high-speed camera (Redlake Motion Pro Series Y) was mounted on an inverted microscope (Nikon Eclipse Ti) to image the fluid dynamics inside the channels. Both the continuous and the disperse phase were introduced into the microchannels using syringe pumps (KD Scientific Inc.). To quantify the products of the biomolecular reactions, fluorescence microscopy was employed. For this purpose, a Nikon Ti Eclipse microscope with a Lumencor SPECTRAX light source was used.

Materials. The working fluids in our experiments were water and aqueous polysaccharid solutions as the disperse phase and 1000 cSt silicone oil AP (Sigma-Aldrich) as the continuous phase. If not stated otherwise, all dispersephase liquids contained NaCl (17 mM). To prepare the



polymer solutions, xanthan gum powder (from Xanthomonas campestris), purchased from Sigma-Aldrich, was dissolved in water at 40 °C, resulting in xanthan gum solutions with concentrations of 0.5, 1, 2 and 3 g/l. The rheological behavior of xanthan gum solution, a liquid with shear-dependent viscosity, has been reported in the literature and can be fitted by a number of well-known models 45. The dissolution of xanthan gum powder in water not only makes the viscosity shear-rate dependent, but also significantly increases the viscosity of the solutions at zero shear rate 46. For example, at zero shear the 3 g/l solution has a viscosity of about 13.9 Pa∙s. The interfacial tensions between the aqueous phases and the oil phase were measured by a drop and bubble shape tensiometer (PAT-1, Sinterface Technologies, Germany). The measurements showed that the interfacial tension does not significantly change upon addition up to 3 g/l of xanthan gum into water. The obtained value (~35 mN/m) is consistent with literature values 47. PlasmidDNA (pBEST-OR2-OR1-Pr-UTR1-deGFP-T500; Addgene #40019) for the cell-free transcription/translation (TX/TL) reactions inside the droplets, was isolated by anion-exchange chromatography on silica NucleoBond® Xtra columns (Macherey-Nagel). DNA-concentrations were determined by UV-absorbance measurements at 260 nm. Cell-free TX/TL-reactions were performed using an E. coli-based TX-TL-competent protein extract (myTXTL®, Arbor Bioscience). In vitro transcription experiments inside the droplets were conducted using PCR-assembled DNA-fragments containing the sequence of the T7 promoter and the coding sequence of the iSpinach aptamer. All DNA-primer sequences are listed in Suppl. Table S1. PCR-fragments (50 nM) were purified (Monarch NA purification, NEB) and added to a transcription mix containing 2 mM of each NTP, 25 mM MgCl2, 40 mM Tris-HCl pH 8.0, 5 mM DTT, 1 mM spermidine, 20 nM of DFHBI and 70 μg/ml T7 RNA-polymerase. Reactions were incubated and microscopically monitored at 37 °C (TX-reactions) and at 29°C (TX/TL reactions).

Experimental Protocol. During the droplet-formation experiments, the syringe pump providing the dispersephase fluid was running. The corresponding backpressure enabled keeping the aqueous liquid/oil interface at a desired location where it becomes exposed to the electric field originating from the U-shaped channel. The high voltage sequencer device was used to create a voltage pulse of several hundred volts with a duration of some milliseconds. At the beginning of an experiment, the main channel was filled with silicone oil. Then, by switching on the disperse-phase pump, the oil in about half of the channel was displaced by aqueous solution, resulting in a stationary liquid/liquid interface close to the junction. Subsequently, a voltage pulse was applied, and the jet formation and droplet pinch off were recorded by the highspeed camera at frame rates of 1000 fps or 4000 fps. The visible area of the generated droplets, 𝐴𝐴, was measured using the ImageJ software (National Institute of Health, USA), and the droplet diameter was calculated by the equation 𝐷𝐷 = (4𝐴𝐴/𝜋𝜋)1/2 . For each case, the experiments

Page 12 of 19

were repeated several times to compute the average droplet diameter and its standard deviation. Cell-free TX/TL-reactions inside the droplets were started by adding the deGFP-gene containing plasmid (pBEST-OR2OR1-Pr-UTR1-deGFP-T500) to a myTXTL® master mix at a final concentration of 5 nM. The reagents were mixed and prewarmed to 29 °C before being injected into the microfluidic device. Microscopic imaging was conducted at a constant temperature of 29 °C. To quantify the reaction product inside the droplets, the auto-fluorescence of the synthesized deGFP (λex 485 nm, λem 520 nm) was measured using fluorescence microscopy. Measurements were performed in 15 min intervals over 2 h using an exposure time of 500 ms. For quantification of the same reaction in the bulk, clear-bottom 384-well plates containing cell-free myTXTL® master mix (10 µl) and plasmid DNA (114 ng) were analyzed using a PHERAstar FSX microplate reader (BMG Biotech). Measurements were performed at 29 °C for 5 h, obtaining a readout of the deGFPfluorescence every 5 min.

RESULTS AND DISCUSSION Electric Field Distribution. Before analyzing the experimental results on droplet production, it is instructive to study the electric field distribution between the liquid/liquid meniscus and the U-shaped channel. For this purpose, the electrostatic potential was computed numerically using the commercial finite-element solver COMSOL Multiphysics. For simplicity, a two-dimensional model was set up, treating the aqueous phases filling the U-shaped channel and partially filling the main channel as conductors. These are visible as white areas in Fig. 2. Consequently, the tangential component of the electric field vanishes at the surface of the white areas. The relative dielectric permittivities of the PDMS substrate and silicon oil are about 2.8 and 2.9, respectively. The inset of Fig. 2 shows a microscopy image of the liquid/liquid meniscus. The electric field distribution was obtained by solving the Laplace equation, i.e. ∆φ = 0 , with φ being the electric potential. The U-shaped channel was modeled as an isopotential surface with vanishing potential, and the electric potential at the surface of the water finger in the partially filled main channel was set to 400 V. At the other boundaries of the computational domain, a vanishing normal component of the electric field was imposed. An unstructured mesh of triangular elements with quadratic shape functions was employed. It was ensured that the results obtained are virtually grid-independent. The resulting electric field distribution is shown in Fig. 2. The colormap indicates the magnitude of the electric field strength in V/m. The lines are the electric field lines, and the vectors indicate the electric field vectors at the surface of the conductors. It its clearly visible that the maximum electric field strength is found at the circular-arc shaped liquid meniscus close to the lower channel wall (adjacent to the U-shaped channel). This is the position where we expect the liquid/liquid interface to break down, i.e. to form a Taylor cone from which droplets emerge.


Page 13 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry structure of the electric field, shown in Fig. 2. The electric field strength at the liquid/liquid interface is maximal in close vicinity of the channel wall. This is the position where the interface becomes unstable and the jet emerges. However, due to the proximity of the channel wall, the jet does not grow to a significant length before the droplet pinches off. This is presumably the reason why we obtain a much more uniform droplet size distribution than reported in the literature 26 (see discussion in section 3.3).

Figure 2. Electric-field distribution around the water finger in the main channel. The colormap indicates the electric field strength in V/m, the lines are electric field lines. The inset shows a corresponding microscopy image of the water finger.

Dynamics of Droplet Formation. One of the major challenges related to droplet production in microchannels is the droplet size distribution. On the one hand, a uniform droplet size is desired. On the other hand, while the production of droplets with volumes in the pico- or nanoliter range is well established, producing femtoliter droplets still poses some challenges. The challenges are related to the fact that in most microfluidic droplet generators, the droplet size is larger than the scale of the geometric features of the microchannels, which translates to very small features in the case of femtoliter droplets. So far there are two major technologies allowing to generate droplets significantly smaller than the smallest geometric scale, flow-focusing and electrospray technology. In flow focusing the diameter of the liquid jet from which droplets are produced is reduced by the sheath flow of a second immiscible fluid 48. Electrospray technology relies on the formation of Taylor cones 49,50. A Taylor cone results from the deformation of the surface of a conducting liquid to which an external electric field is applied. Up to a critical value of the electric field strength, a static deformation of the liquid surface is observed, which results from the balance of Maxwell and capillary stresses. When the critical value of the field strength is exceeded, a jet emerges that decays into small droplets. The electrospray approach was reconsidered by the group of Daniel T. Chiu as a method to produce small droplets inside microfluidic channels 26. Employing electric-field pulses, they demonstrated the on-demand generation of droplets significantly smaller than the channel dimensions. In our study, we have used a configuration inspired by their work to produce droplets by electric-field pulses. Fig. 3 shows time lapse images of the droplet production process taken at a frame rate of 4000 fps. Compared to Fig. 2, the frames are rotated by 90°. Water is used as the liquid from which the droplets are produced, the second liquid is silicon oil. The pulse amplitude and duration are 400 V and 10 ms. The liquid/liquid interface exposes a Taylor-conelike structure from which a single droplet with a diameter of about 8 µm pinches off. The jet is much shorter than the one reported in the literature 26 at the instant when the droplet is produced. This is due to the special asymmetric

Figure 3. Time-lapse images of the on-demand production of a water droplet, using a voltage of 400 V and pulse duration of 10 ms. The numbers below the individual frames denote the time in milliseconds.

Droplet Size Distribution. Since the uniformity of the droplet size distribution is a key feature of a microfluidic droplet generator, we performed systematic studies with different liquids for the droplet phase. Biological fluids often contain proteins, DNA and other polymers, resulting in a significant increase of the viscosity. To study the influence of viscosity, experiments with aqueous solutions of xanthan gum were performed. Xanthan-gum solutions show a shear-thinning behavior, i.e. the viscosity decreases as a function of the shear rate. Experiments with liquid/liquid systems in which one of the phases is a xanthan gum solution have the advantage that the dissolved polymer leaves the interfacial tension largely unaffected, such that the viscosity and the interfacial tension can be varied independently. The viscosity values given in table 1 are the values at zero shear rate obtained from the literature 46 , where, different from our experiments, desalinated water was used as a solvent. Since we only wish to give an



indication of the order of magnitude of the aqueous-phase viscosity, we consider it sufficient to refer to the literature results. The experiments with different aqueous phases were performed with a pulse amplitude of 400 V and a pulse duration of 10 ms. Table 1 displays the properties of the different aqueous-phase liquids and the corresponding average droplet diameters obtained, together with the standard deviations of the droplet diameter. The average was obtained from at least 25 independent measurements for water and at least 10 measurements for each xanthangum solution. Overall, the standard deviation of the droplet diameter is sufficiently small, i.e. our method allows generating uniformly sized droplets. Increasing the viscosity of the aqueous phase slightly reduces the average droplet diameter, where for viscosities above 6∙102 mPa∙s hardly any change in the droplet diameter is observed. Remarkably, even for zero-shear rate viscosities 104-fold higher than that of water, still droplets with a size distribution comparable to that of water droplets are produced. For any microfluidic droplet generator, the question arises which parameters need to be varied to tailor the droplet size. One option would be to vary the height of the channel in which the droplets are produced. We have conducted experiments with water in channels of half the height (10 µm) of our standard channels and found the average droplet size to be reduced by a factor of 2.4. Table 1: Average droplet diameter and standard deviation of droplet diameter for aqueous-phase liquids containing different concentrations of xanthan gum. The viscosity values are the zero-shear literature values 46.

Xanthan gum concentration [g/l] 0 0.5 1 2 3

Zero shear rate viscosity [mPa∙s] 1.01 6.00×102 1.46×103 3.59×103 1.39×104

Average droplet diameter [µm] 8.3 6.9 6.9 6.5 6.6

Standard deviation of droplet diameter [µm] 0.7 0.8 1 0.8 0.5

Comparing our results on the droplet size distribution with the results reported in the literature 26, we notice that we do not observe a broad size distribution with a bimodal structure, but a largely monodisperse size distribution. Fig. 4 shows the size distributions for water (blue filled bars) and for the xanthan gum solution at a concentration of 3 g/l (red open bars). Previously the formation of satellite droplets was reported 26. As already indicated, we hypothesize that the proximity of the wall limits the extension of the liquid jet and has a positive effect on the droplet size distribution, whereas in previous studies 26 the jet grows to a significant length before it decays into droplets. However, obtaining a monodisperse size distribution requires working in the right domain of the parameter space spanned by pulse amplitude and pulse duration, as will be shown in the following chapter.

Page 14 of 19

Figure 4. Droplet size distribution for water (blue filled bars) and 3 g/l xanthan gum solution (red open bars). The data were normalized such that the integrals of the two histograms are the same.

Droplet Formation Regimes. The results reported so far were obtained in a domain of the parameter space in which single, virtually monodisperse droplets are produced. However, there are other domains in which different phenomena are observed. For applications of the method described in this paper it will be important to know how material and process parameters influence the results and in which domains one should work to obtain single droplets of uniform size. For this purpose, parameter studies with water droplets containing different amounts of NaCl were performed. The pulse amplitude was kept fixed at 400 V, and the pulse duration was varied. The results are displayed in Fig. 5a, outlining the different dynamic regimes. While black filled circles indicate the absence of droplet formation, open circles show the formation of multiple (≥ 2) droplets. Between these two regions lies an area where the ondemand production of single droplets is possible (colored circles). The different colors denote average droplet sizes. It can be seen that for the majority of NaCl concentrations, there is a pulse duration range that enables the formation of single droplets. Lower pulse durations result in no droplets, higher values in multiple droplets. At salt concentration ≥ 1 mol/l, the production of single droplets becomes impossible. Fig. 5b captures the dynamic regimes in the case that the pulse amplitude and the pulse duration are varied. The higher the pulse amplitude, the narrower the regime in which single droplets are formed. However, there is a threshold below which droplet formation is no longer possible. This lies between 300 V and 325 V. Of course the voltage amplitudes are specific for the design and the dimensions of the system outlined in Fig. 1. Rather than the voltage, it is the electric field strength that determines the physical response of the liquid/liquid interface. Fig. 5b also indicates that varying the pulse amplitude presents an option to tailor the droplet size. Achievable droplet diameters range from 3 µm (violet) to 9 µm (red). The


Page 15 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


significance of the yellow symbol at 450 V is limited, since at that voltage the range of pulse durations in which single droplets are produced is already very narrow. Discarding the yellow symbol, the trend is that with increasing pulse amplitude, the droplet diameter increases.

Figure 5. Droplet formation regimes for water. a) For varying pulse duration and NaCl concentration at a pulse amplitude of 400 V. b) For varying pulse duration and amplitude with 17 mM NaCl. Black filled circles indicate cases where no droplets are produced, the open circles denote the formation of multiple droplets. Grey filled circles indicate situations in which only in a part of the experiments a droplet was formed. The different colors encode the droplet diameters.

Droplets As Biological Reaction Compartments. Having demonstrated that the method described above allows the generation of single femtoliter-size droplets in a reproducible manner, it suggests itself to use the droplets as reaction compartments. We see an important potential application in the field of synthetic biology. Synthetic biology utilizes artificial genetic circuits, i.e. transcriptiontranslation systems are key components in that context. An important challenge is to bridge the gap between in-vivo implementations of artificial generic circuits (for example based on yeast cells) and their in-vitro counterparts. Therefore it is important to formulate in-vitro implementations that mimic the implementations in biological cells. In that context our drop-on demand generator brings along at least two key advantages. First, it allows creating biological reaction compartments with a size comparable to that of yeast cells, one of the most frequently used cell types in synthetic biology. The comparable size translates to a comparable number of biologically relevant molecules. That way, it becomes possible to mimic the stochastic behavior of biological cells. Second, it allows to mimic the crowding characteristic for biological cells. The interior of biological cells is densely packed with different molecular species, effectively providing a highly viscous environment. As apparent from Table 1, we have demonstrated the creation of droplets with a viscosity more than 104-fold higher than that of water.

To begin with, we performed two of the most fundamental biochemical reactions inside the droplets: gene transcription and mRNA-translation. For that we used an Escherichia coli-based cell-free transcriptiontranslation (TX-TL) system 51, in which a plasmid encoding a gene variant of the enhanced green fluorescent protein (deGFP) is first transcribed into mRNA and subsequently translated into protein. deGFP represents a start codon-optimized version of the enhanced (e)GFP, and its synthesis and bioactive folding can be monitored by the auto-fluorescence of the protein 52. The synthesis of deGFP was analyzed in a time-dependent manner, and the product of the reaction when performed at standard ("bulk")conditions in a microliter volume (10 µl) is quantified in Fig. 6a: deGFP-fluorescence starts at roughly 30 min. The protein concentration is a linear function of time up to 1 h and reaches a plateau roughly at 3 h. The TX-TL reaction mix was enclosed in aqueous droplets after 60 minutes of “bulk” reaction (arrow in Fig 6a), using the method described in the previous sections. Droplets with diameters between 3 µm and 8 µm were tested, corresponding to droplet volumes between 20 fl and 300 fl. As an example, Fig. 6b shows micrographs of the fluorescence signal of a 5.1 µm diameter droplet at three different points in time. A representative fluorescence signal of deGFP averaged over 23 droplets is shown in Fig. 6c. While the negative control containing no DNA template shows no traceable signal over the entire time interval, the fluorescence signal in the droplets increases over 2 h. A fast increase within the first 15 min is followed by a slower increase thereafter. As expected and previously



demonstrated 53,54, the data show a high statistical variance, likely due to the stochastic repartition of plasmid DNA and TX-TL-extract components 55 in the small reaction volumes. In that sense, the droplets perfectly mimic the stochastic behavior of single cells 56.

Figure 6. On-demand droplets as biological reaction compartments. (a) Signal from the bulk TX-TL reaction generating autofluorescent deGFP-protein. (b) Microscopic images of a single droplet containing the cell-free TX-TL reaction mix at 0, 60 and 120 min of incubation. Scale bar: 2.5 μm. (c) Plot of the time-dependent accumulation of deGFP. Data points in black represent the average obtained from n = 23 droplets. Grey curve: Control experiment in the absence of plasmid DNA (n = 3). Error bars are standard deviations from the mean.

As an additional biologically relevant experiment, we tested whether low-molecular mass ligands can diffuse across the oil/water interface to bind to a cognate receptor molecule and execute a biochemical switch. For this we set up an in-vitro transcription reaction within the droplets to synthesize the "light-up" RNA aptamer iSpinach 57. iSpinach is a synthetic, intricately folded small RNAmolecule that can bind to low molecular mass compounds such as 3,5-difluoro-4-hydroxybenzylidene imidazolinone (DFHBI). RNA-binding triggers DFHBI to fluoresce. The fluorophore was solubilized in the oil phase inside the microfluidic device. About 2-3 min after droplet formation, the droplets started to fluoresce, reaching a stable plateau after 30 min. Representative fluorescence micrographs of a 5.9 µm diameter droplet at different time points are shown in Fig. 7a. Fig. 7b displays the averaged fluorescence intensity as a function of time from a sample

Page 16 of 19

of 9 droplets. As before, the strong electric field required for droplet formation did not adversely affect the bioactive structure of the biomolecules, in that case the polyanionic RNA molecules. Similar to the case of the TX-TL system, this reinforces the compatibility of the droplets produced using the method described above with biomolecular reactions.

Figure 7. Droplet uptake experiment. In-vitro transcription of iSpinach RNA-aptamers inside aqueous droplets. Diffusion of the fluorogenic DFHBI-ligand from the surrounding oil-phase into the droplet results in the formation of iSpinach RNAaptamer/DFHBI complexes, which can be detected by their green fluorescence (λem 473 nm). (a) Time-lapse microscopic images (0, 10, 30 min) of a 108 fl droplet expressing the iSpinach RNA-aptamer. Scale bar: 3 μm. (b) Plot of the time-dependent formation of green fluorescent iSpinach RNAaptamer/DFHBI complexes. Data points in black represent the average from n = 9 droplets. Grey curve: Control experiment in the absence of DFHBI (n = 3). Error bars are standard deviations from the mean.

SUMMARY AND CONCLUSIONS We have demonstrated a principle that allows the ondemand formation of aqueous femtoliter-volume droplets in an oil phase. The method is based on electric field pulses deforming the aqueous-oil interface. The resulting droplets are close-to monodisperse. Using xanthan gum solutions as the aqueous phase, we have shown that even with solutions having a viscosity at zero shear rate more than 10,000 times higher than that of water, the method yields similar results as in the case of water. To achieve the production of single droplets on demand, the parameters characterizing our system have to be chosen with due diligence. In a space spanned by the salt concentration in the aqueous phase and the amplitude of the voltage pulse, the single droplet regime is delimited by regimes in which no droplets or multiple droplets are produced. The same is true in the space spanned by the pulse duration and the pulse amplitude. Varying the pulse amplitude allows tuning the droplet size. To demonstrate the applicability of the electric-fielddriven droplet generator, we showed that the droplets can be used as versatile biological reaction compartments. First, we demonstrated that droplets containing a cell-free transcription-translation system execute gene transcription and protein biosynthesis 51 in a timely and programmable fashion. Second, we verified that biomolecules inside the


Page 17 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


aqueous droplets such as small RNAs can be diffusionally activated from the outside to induce a ligand-driven biochemical switch. As such the droplets mimick defined cell-like characteristics, making the on-demand droplet generator a promising tool for a variety of (bio)molecular studies including high-throughput sceening experiments.

ASSOCIATED CONTENT

(13)

(14)

(15)

Supporting Information Supporting information on genetic sequences and on fluorescent traces of individual droplets is provided with the submission.

(16)

(17)

AUTHOR INFORMATION

(18)

Corresponding Author * Email: [email protected]

Notes

(19)

There are no conflicts to declare.

ACKNOWLEDGMENT

(20)

We thank V. Noireaux for plasmid pBEST-OR2-OR1-PrUTR1-deGFP-T500 and H. Köppl for discussions and access to the microscopy equipment in his laboratory. M. Shojaeian and F. X. Lehr kindly acknowledge the support from the LOEWE CompuGene Project, funded by the Hessian Ministry of Science and Art.

(21)

(22) (23)

REFERENCES (1)

(2)

(3)

(4) (5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

Guo, M. T.; Rotem, A.; Heyman, J. A.; Weitz, D. A. Droplet Microfluidics for High-Throughput Biological Assays. Lab Chip 2012, 12 (12), 2146. Elvira, K. S.; i Solvas, X. C.; Wootton, R. C. R.; DeMello, A. J. The Past, Present and Potential for Microfluidic Reactor Technology in Chemical Synthesis. Nat. Chem. 2013, 5 (11), 905–915. Zhao, C.-X. Multiphase Flow Microfluidics for the Production of Single or Multiple Emulsions for Drug Delivery. Adv. Drug Deliv. Rev. 2013, 65 (11–12), 1420–1446. Seemann, R.; Brinkmann, M.; Pfohl, T.; Herminghaus, S. Droplet Based Microfluidics. Reports Prog. Phys. 2012, 75 (1), 016601. Porter, D.; Savage, J. R.; Cohen, I.; Spicer, P.; Caggioni, M. Temperature Dependence of Droplet Breakup in 8CB and 5CB Liquid Crystals. Phys. Rev. E 2012, 85 (4), 041701. Cheung, Y. N.; Qiu, H. Acoustic Microstreaming for Droplet Breakup in a Microflow-Focusing Device. Appl. Phys. Lett. 2010, 97 (13), 133111. Tan, S.-H.; Nguyen, N.-T.; Yobas, L.; Kang, T. G. Formation and Manipulation of Ferrofluid Droplets at a Microfluidic T -Junction. J. Micromechanics Microengineering 2010, 20 (4), 045004. Link, D. R.; Grasland-Mongrain, E.; Duri, A.; Sarrazin, F.; Cheng, Z.; Cristobal, G.; Marquez, M.; Weitz, D. A. Electric Control of Droplets in Microfluidic Devices. Angew. Chemie Int. Ed. 2006, 45 (16), 2556–2560. Kim, H.; Luo, D.; Link, D.; Weitz, D. A.; Marquez, M.; Cheng, Z. Controlled Production of Emulsion Drops Using an Electric Field in a Flow-Focusing Microfluidic Device. Appl. Phys. Lett. 2007, 91 (13), 133106. Shojaeian, M.; Hardt, S. Fast Electric Control of the Droplet Size in a Microfluidic T-Junction Droplet Generator. Appl. Phys. Lett. 2018, 112, 194102. Keyon, A. S. A.; Guijt, R. M.; Bolch, C. J.; Breadmore, M. C. Droplet Microfluidics for Postcolumn Reactions in Capillary Electrophoresis. Anal. Chem. 2014, 86 (23), 11811–11818. Birchler, A.; Berger, M.; Jäggin, V.; Lopes, T.; Etzrodt, M.; Misun, P. M.; Pena-Francesch, M.; Schroeder, T.; Hierlemann, A.; Frey, O. Seamless Combination of Fluorescence-Activated Cell Sorting and Hanging-Drop Networks for Individual Handling and Culturing of Stem Cells and Microtissue Spheroids.

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

(32)

(33)

(34)

(35)

Anal. Chem. 2016, 88 (2), 1222–1229. Lorenz, R. M.; Edgar, J. S.; Jeffries, G. D. M.; Chiu, D. T. Microfluidic and Optical Systems for the On-Demand Generation and Manipulation of Single Femtoliter-Volume Aqueous Droplets. Anal. Chem. 2006, 78 (18), 6433–6439. He, M.; Edgar, J. S.; Jeffries, G. D. M.; Lorenz, R. M.; Shelby, J. P.; Chiu, D. T. Selective Encapsulation of Single Cells and Subcellular Organelles into Picoliter- and Femtoliter-Volume Droplets. Anal. Chem. 2005, 77 (6), 1539–1544. Jung, S.-Y.; Retterer, S. T.; Collier, C. P. On-Demand Generation of Monodisperse Femtolitre Droplets by Shape-Induced Shear. Lab Chip 2010, 10 (20), 2688. Zhou, H.; Yao, S. A Facile On-Demand Droplet Microfluidic System for Lab-on-a-Chip Applications. Microfluid. Nanofluidics 2014, 16 (4), 667–675. Yu, M.; Hou, Y.; Zhou, H.; Yao, S. An On-Demand Nanofluidic Concentrator. Lab Chip 2015, 15 (6), 1524–1532. Lin, B.-C.; Su, Y.-C. On-Demand Liquid-in-Liquid Droplet Metering and Fusion Utilizing Pneumatically Actuated Membrane Valves. J. Micromechanics Microengineering 2008, 18 (11), 115005. Zeng, S.; Li, B.; Su, X.; Qin, J.; Lin, B. Microvalve-Actuated Precise Control of Individual Droplets in Microfluidic Devices. Lab Chip 2009, 9 (10), 1340. Galas, J.-C.; Bartolo, D.; Studer, V. Active Connectors for Microfluidic Drops on Demand. New J. Phys. 2009, 11 (7), 075027. Guo, F.; Liu, K.; Ji, X.-H.; Ding, H.-J.; Zhang, M.; Zeng, Q.; Liu, W.; Guo, S.-S.; Zhao, X.-Z. Valve-Based Microfluidic Device for Droplet on-Demand Operation and Static Assay. Appl. Phys. Lett. 2010, 97 (23), 233701. Xu, J.; Attinger, D. Drop on Demand in a Microfluidic Chip. J. Micromechanics Microengineering 2008, 18 (6), 065020. Sung Kwon Cho; Hyejin Moon; Chang-Jin Kim. Creating, Transporting, Cutting, and Merging Liquid Droplets by Electrowetting-Based Actuation for Digital Microfluidic Circuits. J. Microelectromechanical Syst. 2003, 12 (1), 70–80. Srinivasan, V.; Pamula, V. K.; Fair, R. B. An Integrated Digital Microfluidic Lab-on-a-Chip for Clinical Diagnostics on Human Physiological Fluids. Lab Chip 2004, 4 (4), 310–315. Gu, H.; Murade, C. U.; Duits, M. H. G.; Mugele, F. A Microfluidic Platform for On-Demand Formation and Merging of Microdroplets Using Electric Control. Biomicrofluidics 2011, 5 (1), 011101. He, M.; Kuo, J. S.; Chiu, D. T. Electro-Generation of Single Femtoliter- and Picoliter-Volume Aqueous Droplets in Microfluidic Systems. Appl. Phys. Lett. 2005, 87 (3), 031916. He, M.; Kuo, J. S.; Chiu, D. T. Effects of Ultrasmall Orifices on the Electrogeneration of Femtoliter-Volume Aqueous Droplets. Langmuir 2006, 22, 6408–6413. Chen, I.; Tsai, H.; Chang, C.; Zheng, G.; Su, Y. Sensors and Actuators B : Chemical Electric-Field Triggered , on-Demand Formation of Sub-Femtoliter Droplets. Sensors Actuators B. Chem. 2018, 260, 541–553. Park, S.-Y.; Wu, T.-H.; Chen, Y.; Teitell, M. A.; Chiou, P.-Y. High-Speed Droplet Generation on Demand Driven by Pulse Laser-Induced Cavitation. Lab Chip 2011, 11 (6), 1010. Collins, D. J.; Alan, T.; Helmerson, K.; Neild, A. Surface Acoustic Waves for On-Demand Production of Picoliter Droplets and Particle Encapsulation. Lab Chip 2013, 13 (16), 3225. Taly, V.; Kelly, B. T.; Griffiths, A. D. Droplets as Microreactors for High-Throughput Biology. ChemBioChem 2007, 8 (3), 263– 272. Kaminski, T. S.; Scheler, O.; Garstecki, P. Droplet Microfluidics for Microbiology: Techniques, Applications and Challenges. Lab Chip 2016, 16 (12), 2168–2187. Schaerli, Y.; Hollfelder, F. The Potential of Microfluidic Waterin-Oil Droplets in Experimental Biology. Mol. Biosyst. 2009, 5 (12), 1392. Shim, J.; Ranasinghe, R. T.; Smith, C. A.; Ibrahim, S. M.; Hollfelder, F.; Huck, W. T. S.; Klenerman, D.; Abell, C. Ultrarapid Generation of Femtoliter Microfluidic Droplets for Single-Molecule-Counting Immunoassays. ACS Nano 2013, 7 (7), 5955–5964. Wang, F.; Burns, M. A. Performance of Nanoliter-Sized DropletBased Microfluidic PCR. Biomed. Microdevices 2009, 11 (5),



(36)

(37)

(38)

(39)

(40)

(41)

(42)

(43)

(44)

(45)

(46)

1071–1080. Shembekar, N.; Hu, H.; Eustace, D.; Merten, C. A. Single-Cell Droplet Microfluidic Screening for Antibodies Specifically Binding to Target Cells. Cell Rep. 2018, 22 (8), 2206–2215. Fradet, E.; Bayer, C.; Hollfelder, F.; Baroud, C. N. Measuring Fast and Slow Enzyme Kinetics in Stationary Droplets. Anal. Chem. 2015, 87 (23), 11915–11922. Williams, R.; Peisajovich, S. G.; Miller, O. J.; Magdassi, S.; Tawfik, D. S.; Griffiths, A. D. Amplification of Complex Gene Libraries by Emulsion PCR. Nat. Methods 2006, 3 (7), 545–550. Margulies, M.; Egholm, M.; Altman, W. E.; Attiya, S.; Bader, J. S.; Bemben, L. A.; Berka, J.; Braverman, M. S.; Chen, Y.-J.; Chen, Z.; et al. Genome Sequencing in Microfabricated HighDensity Picolitre Reactors. Nature 2005, 437 (7057), 376–380. Marcy, Y.; Ishoey, T.; Lasken, R. S.; Stockwell, T. B.; Walenz, B. P.; Halpern, A. L.; Beeson, K. Y.; Goldberg, S. M. D.; Quake, S. R. Nanoliter Reactors Improve Multiple Displacement Amplification of Genomes from Single Cells. PLoS Genet. 2007, 3 (9), e155. Leemhuis, H.; Stein, V.; Griffiths, A. D.; Hollfelder, F. New Genotype–phenotype Linkages for Directed Evolution of Functional Proteins. Curr. Opin. Struct. Biol. 2005, 15 (4), 472– 478. Srisa-Art, M.; DeMello, A. J.; Edel, J. B. High-Throughput DNA Droplet Assays Using Picoliter Reactor Volumes. Anal. Chem. 2007, 79, 6682–6689. Huebner, A.; Olguin, L. F.; Bratton, D.; Whyte, G.; Huck, W. T. S.; de Mello, A. J.; Edel, J. B.; Abell, C.; Hollfelder, F. Development of Quantitative Cell-Based Enzyme Assays in Microdroplets. Anal. Chem. 2008, 80 (10), 3890–3896. Booth, M. J.; Schild, V. R.; Graham, A. D.; Olof, S. N.; Bayley, H. Light-Activated Communication in Synthetic Tissues. Sci. Adv. 2016, 2 (4), e1600056–e1600056. Song, K. W.; Kim, Y. S.; Chang, G. S. Rheology of Concentrated Xanthan Gum Solutions: Steady Shear Flow Behavior. Fibers Polym. 2006, 7 (2), 129–138. Wyatt, N. B.; Liberatore, M. W. Rheology and Viscosity Scaling

(47)

(48)

(49)

(50) (51)

(52)

(53)

(54)

(55)

(56)

(57)

Page 18 of 19 of the Polyelectrolyte Xanthan Gum. J. ofAppliedPolymer Sci. 2009, 114, 4076–4084. Muthamizhi, K.; Kalaichelvi, P.; Powar, S. T.; Jaishree, R. Investigation and Modelling of Surface Tension of Power-Law Fluids. Rsc Adv. 2014, 4, 9771–9776. Anna, S. L.; Bontoux, N.; Stone, H. A. Formation of Dispersions Using “Flow Focusing” in Microchannels. Appl. Phys. Lett. 2003, 82 (3), 364–366. Zeleny, J. The Electrical Discharge from Liquid Points , and a Hydrostatic Method of Measuring the Electric Intensity at Their Surfaces. Phys. Rev. 1914, 3 (2), 69–91. Taylor, G. Disintegration of Water Drops in an Electric Field. Proc. R. Soc. London A 1964, 280, 383–397. Shin, J.; Noireaux, V. Efficient Cell-Free Expression with the Endogenous E. Coli RNA Polymerase and Sigma Factor 70. J. Biol. Eng. 2010, 4 (1), 8. Garamella, J.; Marshall, R.; Rustad, M.; Noireaux, V. The All E. Coli TX-TL Toolbox 2.0: A Platform for Cell-Free Synthetic Biology. ACS Synth. Biol. 2016, 5 (4), 344–355. Kato, A.; Yanagisawa, M.; Sato, Y. T.; Fujiwara, K.; Yoshikawa, K. Cell-Sized Confinement in Microspheres Accelerates the Reaction of Gene Expression. Sci. Rep. 2012, 2 (1), 283. Sharma, B.; Takamura, Y.; Shimoda, T.; Biyani, M. A Bulk SubFemtoliter in Vitro Compartmentalization System Using SuperFine Electrosprays. Sci. Rep. 2016, 6 (1), 26257. Karig, D. K.; Jung, S.-Y.; Srijanto, B.; Collier, C. P.; Simpson, M. L. Probing Cell-Free Gene Expression Noise in Femtoliter Volumes. ACS Synth. Biol. 2013, 2 (9), 497–505. Pietrini, A. V.; Luisi, P. L. Cell-Free Protein Synthesis through Solubilisate Exchange in Water/Oil Emulsion Compartments. ChemBioChem 2004, 5 (8), 1055–1062. Autour, A.; Westhof, E.; Ryckelynck, M. ISpinach: A Fluorogenic RNA Aptamer Optimized for in Vitro Applications. Nucleic Acids Res. 2016, 44 (6), 2491–2500.


Page 19 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For TOC only


10

On-demand production of femtoliter drops in microchannels and their

Recommend Documents