1

Synthetic Biology in a Chip: A Digital Microfluidic Platform for Cell

Transformation, Culture and Expression

Philip C. Gach,1, 2 Steve C.C. Shih,1, 2 Jess Sustarich,1, 2 Jay D. Keasling,1,3,4,5,6,7
Nathan J. Hillson,1,3,8 Paul D. Adams,1,3,4 and Anup K. Singh1, 2 *

1 Technology Division, Fuels Synthesis Division, Joint Bioenergy Institute (JBEI), Emeryville, California 94608, United States

2 Applied Biosciences and Engineering, Sandia National Laboratories, Livermore,

California 94550, United States

3 Physical Biosciences Division, Lawrence Berkeley National Lab, Berkeley, California

94720, United States

4 Department of Bioengineering, University of California, Berkeley, California 94720, United States

5 Department of Chemical & Biomolecular Engineering, University of California,

Berkeley, California 94720, United States

6 QB3 Institute, University of California, Berkeley, California 94720, United States

7 Synthetic Biology Engineering Research Center, Emeryville California 94608, United

States

8 DOE Joint Genome Institute, Walnut Creek, California 94598, United States

* Corresponding Author email: aksingh@sandia.gov tel: (925) 294-1260
fax: 925-294-3020

  2

ABSTRACT

Synthetic biology experiments require optimization of pathways consisting of many genes and other genetic elements. Given the large number of alternatives available for each element, optimization of a pathway can require a large number of experiments. Currently, these experiments are done manually using fairly large amounts of costly reagents per experiments making the process very expensive, extremely slow and irreproducible. We have developed a digital microfluidic platform that uses aqueous droplets suspended in an oil phase as discrete reaction chambers (or incubators) to completely automate the molecular biology steps. It integrates all critical steps of transformation and culture in one chip including plasmid addition, transformation by
heat-shock, addition of selection medium, culture and phenotypic readout. The flexibility of digital microfluidics and peltier modules allows quick optimization of heat-shock parameters permitting transformation of a variety of plasmids and cell types. The chip was validated by introducing a variety of plasmids into E. coli including; plasmids containing genes for fluorescence proteins GFP, BFP and RFP, plasmids with selection markers for ampicillin or kanamycin, and plasmids built using golden gate assembly. Compatibility of this platform with multiple organisms was demonstrated by transforming DNA plasmids followed by culture of eukaryotic cells including; S. cerevisiae and A. niger. Employing the microfluidic device for heat-shock of GFP plasmid into E. coli cells gave excellent transformation efficiencies up to 4.4 � 106 ± 0.2
� 106 CFU/µg DNA, similar to benchtop results, though requiring ~100-fold less reagent.
Reducing the duration of heat-shock steps afforded a throughput of 0.1 droplets/s, though with slightly compromised transformation efficiencies (9.0 � 105 ± 0.9 � 105 CFU/µg

  3

DNA). This afforded the autonomous generation and processing of up to 100 discrete heat-shocked droplets in just seventeen minutes. The microfluidic device, because of its ability to integrate and automate molecular biology steps and use ~100-fold less reagents, will be of great use to researchers for optimization of synthetic biology pathways to produce biofuels, pharmaceuticals, and other chemicals or biochemicals.

KEY WORDS Digital Microfluidics, Synthetic Biology, Transformation, Cell Culture

  4

INTRODUCTION

Synthetic biology is a relatively new discipline that has demonstrated substantial advances in the development of pharmaceutical drugs,1 biofuels,2 chemicals3 and biomaterials.4 The key steps in synthetic biology involve design and characterization of standard biological parts (such as promoters, 5'UTR and terminator parts) and assembly of genes onto a plasmid to express novel pathways in host organisms for producing materials (including fuels, pharmaceuticals, chemicals and novel materials). The core set of technologies utilized to clone DNA has remained largely unchanged for the last 20 years. Successful synthetic biology experiments require a large number of genes or parts (and the variants/mutants for each gene) to be tested. Currently, the genes/parts are assembled manually using an appropriate combinatorial assembly strategy (BioBricks, SLIC/Gibson/CPEC, Golden-gate etc.). The assembled plasmids are then introduced into host organisms by transformation and the successful transformants are grown in liquid cultures in large enough amounts to quantify desired products. In summary, the overall process in designing, creating and optimizing a metabolic pathway involves many steps, each optimized by trial and error and consequently, a successful effort requires a large amount of time (months to years) and cost ($million in reagents and labor).
Microfluidic devices have attracted significant attention towards developing high throughput screening platforms using greatly reduced amounts of reagents.6 These devices are capable of performing multiple reactions on a single platform through
compartmentalizing reagents into discrete chambers or droplets. Digital microfluidics is a field of microfluidics in which individual aqueous droplets, separated by an immiscible phase such as oil or air, can be generated and manipulated on-chip. Advances in

  5

microfluidics have allowed very high throughput droplet generation, mixing multiple different reagents and using minimal quantities of reagents (down to femtoliter volumes). These technologies have been employed for automating several steps of the synthetic biology pipeline including synthesis of DNA,7, 8 cell transformation and cell sorting.
Transformation of exogenous DNA into cells is a key aspect of this process,
generally achieved through chemical heat-shock or electroporation. Electroporation and chemical heat-shock create porous cellular membranes to introduce DNA into cells and have been routinely employed in microfluidic devices. These devices have demonstrated equal to enhanced transformation efficiencies with much lower reagent volumes, compared to bulk methods.9, 10 Electroporation within microfluidic devices has been performed both in open channels11, 12 and in droplets.13-16 Chemical heat-shock has also been demonstrated in open microfluidic channels10, 17 and in droplets.18, 19 Unfortunately, only the gene introduction step is consistently executed on the device leaving other key procedures performed off-chip, including: DNA addition to cells, incubation, post- transformation addition of selection antibiotics, culture and analysis. A system capable of transformation and culture on a single device is necessary for in-line integration with existing gene assembly and cell sorting technologies thereby providing all steps of standard synthetic biology on a single device.
To meet these requirements, we developed a microfluidic platform which affords droplet generation, mixing, heating and long-term storage. Droplet movement was achieved by digital microfluidics (DMF). This strategy for droplet manipulation is beneficial because it provides simple fabrication, biocompatibility, operational customization and consistent droplet volume generation. DMF has previously been

  6

demonstrated to be effective for DNA synthesis.20 In a proof-of-principle experiment, Au and coworkers demonstrated a DMF platform could transform bacteria with a fluorescent gene.19 However, this device did not possess integrated thermal elements; necessitating physical transfer of the device between a hot plate, ice bath and incubator and then moving the droplets to agar plates for culture. In this report, we describe a DMF device offering cell transformation and culture, which could seamlessly integrate with existing microfluidic tools to provide all steps of standard synthetic biology on a single device. RESULTS AND DISCUSSION

Device construction and operation

Device requirements. Development of new genes in synthetic biology involves three main operations: 1) DNA synthesis 2) cell transformation and culture and 3) isolation of desired traits. Numerous methods have been developed for automating the synthesis of DNA,7, 8 cell transformation and sorting positive cells. Many studies have reported the use of microfluidic platforms for high-throughput addition of exogenous DNA to cells. However, these technologies do not allow for completely autonomous gene delivery, transformation and culture. A system capable of transformation and culture on a single device is necessary for in-line integration with gene assembly and cell sorting. To meet these requirements a microfluidic platform was developed which affords droplet generation, mixing, heating and long-term storage (Figure 1A).

Thermal regulation. Heat-shock is a common method for gene delivery into cells. It has the advantage of not requiring additional reagents or electrical shock to introduce DNA. Effective heat-shock generally requires three temperatures: 1) cold temperature (4°C) for DNA/cell mixing and recovery 2) hot temperature (37°C or 42°C) for heat-

  7

shock and 3) warm temperature (30°C or 37°C) for cell culture. Unlike previous systems, which employ temperature regulation over the entire device or reaction chamber,19 the current device has three discrete thermal zones to allow sequential operation of the three steps namely, DNA and cell addition at 4C, heat shock at 37 or 42c, and culture at 30 or
37. Three peltier modules attached to the bottom of the DMF chip provided the three thermal zones with exquisite temperature control. Peltier supply voltages were controlled by an Arduino board, which also afforded real-time temperature adjustments of device regions. The peltiers uniformly heat the glass slide and channels directly above the module after one-minute actuation. Tightly segmented temperature regions can be maintained on the device for a long period of time with this method (Figure 1C).

Droplet movement. DMF was employed for all aspects of droplet manipulation including: droplet generation, merging and movement. The DMF device was fabricated with chrome electrodes and SU-8 channels to afford a sturdy substrate that was also biocompatible (Figure 1B). This material is also non-reactive with the biocompatible fluorinated oil (HFE 7500) used for droplet segmentation. Channels were coated with Aquapel, which provided a hydrophobic surface ensuring aqueous droplets would not stick to the substrate. Aquapel is also resistant to degradation by the HFE 7500 and stable over several hours in contact with the fluorinated oil. A custom C++ program automated the control of an Arduino board, which provided the actuation voltages to the chrome electrodes on the device. Heat-shock on the DMF device consists of 6 primary steps: 1) generation of DNA and cell droplets 2) mixing DNA and cell droplets 3) DNA/cell droplet movement to heat-shock region 4) generation of media droplet 5) mixing of DNA/cell droplet with a media droplet and 6) DNA/cell/media droplet transfer to culture

  8

region (Figure 1D-K). Media storage on device consisted of a (10�7�0.175 mm) chamber, which supported up to 3 µL reagent at a time. From these reservoirs, 235 ± 11 nL droplets were generated at a rate up to 1 droplet/s. At this point, 54 individual 1150
�1150 µm2 electrodes drove the droplets across the device using 250 ms electrode pulses.
Reagent Delivery. Two strategies were employed for reagent delivery to the reservoirs; pipetting prior to assembly or constant fluid addition during operation, as described previously.21 Experiments requiring few reactions could be prepared through manually pipetting 3 µL liquid directly onto the reservoirs followed by channel filling with HFE 7500 and enclosing with an ITO-coated glass slide. This arrangement was
beneficial for quick device reuse and was employed for generation of up to 7 individual
235 nL droplets per reservoir. Conversely, automated replenishment of reservoir reagents was utilized when large quantities of droplets needed to be generated. This configuration utilized a neMESYS syringe pump to continuously inject reagents directly into the reservoirs through a capillary imbedded in a channel on the device.

Droplet incubation. At the end of the DMF reaction region is a (50Ã?20Ã?0.175 mm) serpentine chamber for long-term culture of cells on the device. Droplets would naturally drift from the last electrode into this chamber due to convective flow. However, this movement was very slow on the order of minutes resulting in electrically induced droplet coalescence during high-speed operation. To overcome this limitation, a channel was constructed at the interface of the DMF electrodes and culture chamber. A

neMESYS syringe pump connected to the channel via a capillary provided a constant flow of oil (0.1 µL/s) at this interface that propelled the droplet from the last electrode into the culture chamber. This addition eliminated droplet electrocoalescence at the

  9

culture chamber entrance. The constant addition of oil to the culture chamber had the additional benefit of replenishing oil lost to evaporation during prolonged experiments. During on-chip culture the oil flow rate was reduced to 0.01 µL/s, which used only 0.86 mL oil for 24 hr cell culture.

E. coli Transformation

DMF heat-shock optimization. We tested the transformation efficiency in our DMF device by transforming E. coli cells (DH5α) with a GFP-expression plasmid. Unless otherwise stated, 5 individual 705 nL droplets were generated for each experiment. We started by performing chemical heat-shock following recommended protocols for the benchtop method, with minimal modifications to accommodate the droplet-based system. In this trial, 235 nL droplets containing cells and DNA were
merged and mixed at 4oC for 15 min, moved to 42oC for 45 s, allowed a 2 min recovery
at 4oC, merged with a 235 nL droplet containing media and antibiotics (ampicillin) then relocated to a 37oC culture chamber for 1 day. Our device can potentially generate libraries of discrete purified DNA plasmids in individual droplets thereby eliminating the conventional requirement of cell plating on agar plates to achieve monoclonal colonies. All droplets possessed GFP positive cells following on-chip culture and afforded a
fluorescence increase of 2072 ± 290 relative fluorescence units (RFUs) (Figure 2C-D). However, this transformation protocol only afforded a transformation efficiency of 1.4 �
106 ± 0.1 � 106 CFU/µg DNA (Figure 2F, 2H1), significantly lower than the standard
benchtop procedure under the same conditions (4.0 � 106 ± 0.4 � 106 CFU/µg DNA) (Figure 2G, 2H1). Through simply changing the heat-shock temperature to 37oC and heat-shock duration to 10 s on the DMF device, the transformation efficiencies were

  10

increased to 2.3 � 106 ± 0.3 � 106 CFU/µg DNA and 4.4 � 106 ± 0.2 � 106 CFU/µg DNA, respectively (Figure 2H2, 2H3). These results are likely due to the much faster thermal change of the droplets within the DMF chip with respect to the a 50 µL volume in a
centrifuge tube. These results are consistent with recommended heat-shock procedures for microtiter plates, where shorter heat-shock times are generally employed. These results demonstrate transformation efficiencies in our chip to be comparable to those achieved in benchtop heat-shock with the benefit of using ~100 times less reagent.

Improving heat-shock throughput. While recapitulating standard heat-shock procedures with the DMF chip produced high transformation efficiencies, this approach afforded extremely low throughputs (~1 droplet / hr). Unfortunately, this strategy would make experiments requiring large quantities of droplets infeasible. Time intensive steps

in traditional heat-shock include a 15 min DNA/cell mixing period, 2 min post heat-shock recovery and at least a 30 min outgrowth step. Several heat-shock protocols have demonstrated the length of various steps can be limited, though resulting in compromised transformation efficiencies. We adjusted the droplet duration at each chip region to efficiency investigate the effects of shorting each step on the transformation efficiency (Figure 2H, Supplementary Table 1). Employing ampicillin as a selection marker allows direct droplet transfer to the culture region, bypassing the outgrowth period, because ampicillin is a bacteriostatic thereby affording cells time to take up the DNA and
replicate in itâ??s presence with no effect on the transformation efficiency. Reducing the DNA/cell mixing time and recovery period on the chip to 10 s each resulted in a mere 5 fold reduced transformation efficiency (9.0 Ã? 105 ± 0.9 Ã? 105 CFU/µg DNA) (Figure
2H.5) and reduced fluorescence of droplet cultured on chip to 1570 ± 180 RFUs. This is

  11

significantly better than reducing benchtop incubation times by the same factor which resulted in a 54-fold lower transformation efficiency (7.3 � 104 ± 1.3 � 104 CFU/µg DNA). This is likely a result of the enhanced mixing and temperature change in the droplets that quickens DNA delivery and cell recovery. Although a 5-fold reduction in the transformation efficiency is significant, this approach allows heat-shock to be completed in just 30 s. Furthermore, by serially processing of the droplets we obtained a throughput of 360 droplets / hr. As a demonstration of large sample processing this protocol was used for preparing 100 individual droplets in just 17 min. Filled reagent
reservoirs were maintained throughout operation by continuously supplying fluids with a syringe pump. Following 24 hrs culture all 100 of the heat-shocked droplets possessed GFP expressing E. coli cells. Efforts are underway to develop a larger DMF chip with larger number of electrodes operating in a linear and/or parallel manner to improve the throughput by 2 orders of magnitude.
Varying DNA concentrations. It has previously been shown that changing the quantity of DNA plasmids added to cells has a direct correlation to the number of transformants generated and resulting transformation efficiency.  22 We varied the plasmid
DNA concentration in each droplet to examine its effect on E. coli transformation in the
DMF chip. These experiments were performed on the DMF chip operating at the highest throughput (10 s DNA/cell mixing, 10 s heat-shock, 10 s recovery and with ampicillin selection marker). As the DNA concentration in each DNA/cell droplet was changed from 0.001ng/µL to 1 ng/µL the DMF device afforded transformation efficiencies ranging from 2.4 � 106 ± 0.7 � 106 CFU/µg DNA to 1.9 � 105 ± 0.7 � 105 CFU/µg DNA, respectively (Figure 2I). These colony counts equate to each heat-shocked droplet with

  12

0.001ng/µL to 1 ng/µL DNA generating 1.7 ± 0.5 to 114 ± 50 transformed cells, respectively. Conversely, standard benchtop heat-shock afforded transformation efficiencies from 1.0 � 107 ± 0.2 � 107 CFU/µg DNA to 9.4 � 105 ± 1.2 � 105 CFU/µg DNA at DNA concentration of 0.001ng/µL to 1 ng/µL, respectively (Figure 2I). The benchtop and DMF systems each generate similar linear regression curves with increasing DNA concentrations, -0.76 and -0.79, respectively, with the benchtop approach affording 3.8 to 4.8 times greater transformation efficiencies though at significantly prolonged heat-shock durations and reagent consumption, as discussed above. Importantly, all droplets cultured on the DMF chip for 24 hrs following heat- shock with 0.001ng/µL to 1 ng/µL DNA possessed GFP expressing E. coli. Transformation of a single DNA plasmid in each isolated droplet at these low plasmid concentrations offers the possibility of library screening without requiring multiple rounds of cell dilution and sorting; each droplet would contain a unique plasmid colony.
Using different plasmids. To demonstrate the versatility of platform, cells were transformed with plasmids containing genes for blue fluorescent protein (BFP) or red fluorescent protein (RFP). Cells transformed with BFP or RFP under identical conditions as the GFP plasmid above afforded slightly lower transformation efficiencies, 4.8 � 105 ±
0.8 � 105 CFU/µg DNA and 5.5 � 105 ± 1.3 � 105 CFU/µg DNA, respectively (Figure 3).
Previous reports have shown a direct correlation between transformation efficiency and plasmid size; it is more difficult for larger plasmids to transverse the created pores and enter cells.22 The small GFP plasmid (3.6 kb) likely could more easily enter the E. coli cells than the larger BFP plasmids (4.8 kb) and RFP plasmids (5.3 kb). Successful transformation of BFP and RFP plasmids could also be confirmed after 24 hr on-chip

  13

culture. Following imaging with a blue filter set or red filter set the BFP and RFP
transformed cells afforded droplets with increased fluorescence, 620 ± 130 and 2830 ±
500 RFUs, respectively.
Kanamycin selection. To this point, ampicillin was used as the selection antibiotic because of its role as a bacteriostatic allowing immediate culture following heat-shock. However, during prolonged culture E. coli can secrete beta-lactamase that inactivates the ampicillin resulting in potential growth of non-selective cells. When this is the case researchers turn to other antibiotics, such as kanamycin, which is bacteriocidal and immediately primes cells for death.23 To ensure cell survival, an outgrowth step in antibiotic free media is required following heat-shock and prior to the addition of kanamycin. The outgrowth period on the DMF device was recapitulated by holding the droplet on the last DMF electrode for 30 min following heat-shock of E. coli with a GFP plasmid with a kanamycin resistance gene. Following outgrowth, a 235 nL droplet
containing 150 µg/mL kanamycin was added and the 940 nL droplet moved to the culture region. After 24 hr culture, droplets were imaged through a green filter set and a fluorescence increase of 1170 ± 230 RFUs was obtained, demonstrating successful transformation and culture of GFP cells. Furthermore, droplets transferred to a
kanamycin selective agar plate following outgrowth on-chip and cultured for 24 hrs afforded a transformation efficiency of 4.1 � 105 ± 0.6 � 105 CFU/µg DNA (Figure 2H.6), slightly lower than GFP transformation with ampicillin selection.

Golden gate assembled DNA. To demonstrate the compatibility of the DMF heat- shock device with assembled DNA plasmids we examined the on-chip transformation of pProm1_BCD1-GFP expressing an ampicillin resistance gene into E. coli. DNA

  14

assembly was initially performed off-chip utilizing the golden gate assembly method24 and then added to the DMF device and 30 s serial droplet heat-shock performed, as described above. Following 24 hrs on-chip culture, all droplets possessed GFP positive cells and the droplets afforded a 890 ± 210 RFU increase when fluorescently imaged with a green filter set. Successful on-chip transformation and culture show the DMF heat- shock device could seamlessly integrate with current microdevices working with other aspects of the synthetic biology pipeline.

Fungal Transformation

Yeast Transformation. Eukaryotic systems are frequently used for expressing proteins and pathways that are difficult to produce in bacterial systems with desired yields. Introduction of exogenous DNA into S. cerevisiae is routinely performed by heat- shock or electroporation. A common procedure for delivering DNA into S. cerevisiae is to heat-shock competent cells with DNA in a PEG solution at 30oC followed by culture on an agar plate.25 Flexibility of the DMF heat-shock device was demonstrated by adapting its operation to transform CEN.PK2 cells with a mCherry generating plasmid. The same heat-shock steps as the 30 s serial droplet transformation protocol was used except for changes in reagent use and operating temperatures. For on-chip culture, a 4
day incubation period was used to account for the longer replication time of yeast cells. Following 4 days incubation, all 5 of the cultured droplets possessed mCherry expressing yeast cells (Figure 4). Transformation efficiencies were assessed by transferring 3 droplets generated on the device to separate â??URA agar plates. 20 ± 5 colonies were observed after 4 days culture on the agar plate corresponding to a transformation efficiency of 1.0 Ã? 104 ± 0.2 Ã? 104 CFU/µg DNA. Bulk heat-shock following Zymo

  15

protocol afforded higher transformation efficiencies (1.0 � 105 ± 0.2 � 105 CFU/µg DNA), likely a result of the increased cell/DNA incubation duration in concentrated PEG solution. Transformation efficiencies lower than those obtained with bacteria are
common with yeast due to the difficulty of DNA transport through the cell membrane. Effective transformation in droplets is possible because the PEG helps confine the DNA to the cells following addition of culture media allowing continued gene delivery during the culture period.
Filamentous Fungi Transformation. Filamentous fungi, such as A. niger, is another important system for molecular biology and chemical synthesis.26  Although laborious sample preparation was required to generate protoplasts,27 once prepared these
cells can be transformed using similar protocols as with yeast heat-shock above. Transformation of A. niger protoplasts on the DMF device was examined by testing the incorporation of plasmid DNA expressing GFP and Hygromycin B resistance. A 30 min outgrowth step was included prior to the addition of Hygromycin B to afford sufficiently high gene expression required to retain cell viability in its presence. Five droplets were processed on the device and each droplet contained GFP positive cells following 5 days culture in the DMF chip (Figure 4). Various cell morphologies indicative of fungal growth were observed in the droplet cultures including individual mycelium (hyphal mats), mycelium clumps and individual A. niger spores all expressing GFP. Similar morphologies were observed along with larger conidiophores when A. niger cells were
transformed using the traditional benchtop method with bilayer agar culture.27 Agarose
hydrogels have previously been generated by DMF, which could potentially be adapted for recapitulating the bilayer agar culture of transformed A. niger cells.28

  16

Conclusions

A thermally controlled DMF platform was developed to afford cellular heat-shock and culture in discrete droplets on a single device. Micro peltier modules were attached to the bottom of a DMF chip to provide the different temperatures required during heat- shock. Automated DMF afforded the droplet generation, mixing and movement across
the various thermal regions of the device. This high level of droplet control allowed employment of all pivotal heat-shock steps on a single device: cell/DNA mixing, heat- shock, media addition, culture and sometimes outgrowth. The flexibility of DMF and the micro peltier modules allowed easy alteration of the operating protocols based on the experimental requirements. A single device was utilized to optimize the heat-shock parameters of E. coli to afford high transformation efficiencies, high throughput sample processing and introduction of various DNAs. Likewise, the same device was used to transform S. cerevisiae and A. niger cells with exogenous DNA. Integrated gene delivery and culture on a single platform will be of great utility for screening gene variants and systematic investigation of DNA delivery strategies.

METHODS

Device assembly and operation. A digital microfluidic device was initially fabricated as described in the Supporting Information. Prior to operation the device was connected to tubing, peltiers and droplet drive electronics. Segmented control of device temperature was achieved by three peltier heaters/coolers attached to the bottom of the device by thermal adhesive tape and controlled by an Arduino microcontroller (Modern Device, Providence, RI). A total of three peltiers were attached to the device: 1) a cooler

  17

module for DNA/ cell suspension mixing and cell recovery covered by an aluminum block 2) 2 mm peltier for heat-shock and 3) 24.5 mm peltier for cell culture (TE Technology, Inc., Traverse City, MI). Two techniques were employed for liquid addition of DNA, cell suspension, media and oil to the device; pipetting before assembly and continuous flow during operation. For the majority of experiments, 3.0 mL liquid was pipetted directly onto the device reservoirs and the channels were flooded with HFE 7500 then topped with an ITO glass slide and secured with Nd magnets (K&J Magnetics). For experiments requiring more reagents, the ITO glass slide was first bound to the device then ends of capillary tubes connected to a neMESYS syringe pump (Centoni GmbH, Korbussen, Germany) were glued into channels associated with the device reservoirs using 5 minute® Epoxy Gel (Devcon, Danvers, MA). Following assembly of either technique, a capillary was inserted into a fifth channel at the entrance of the culture chamber from which HFE 7500 was flowed continuously to remove droplets from the reaction chamber and replenish evaporated oil. Finally, the device was connected to a custom-built control board activated by a RBBB Arduino microcontroller (Modern Device, Providence, RI). A customized C++ program controlled the driving potentials of
the Arduino board to allow automation of droplet movement in the device.21
Microfluidic Transformation. E. coli. The transformation chip reservoirs were first loaded with all reagents using either a pipette or the syringe pump: 1) 3 µL thawed chemically competent DH5α cells, 2) 3 µL DNA solution (0.001 â?? 1.0 ng DNA /µL) and
3) 3 µL Luria-Bertani (LB) broth with selection antibiotics. A 235 nL droplet containing cells was first mixed with a 235 nL DNA droplet at 4oC. This droplet was then moved to a 37oC or 42oC chip region for 10 s or 45 s. The cell/DNA droplet was then merged with

  18

a 235 nL media droplet. For ampicillin selection, 300 µg/mL ampicillin was included with the media and droplet was immediately moved to the devices culture region. Conversely, kanamycin selection required 30 min outgrowth following media delivery prior to addition of a 235 nL droplet containing 200 µg/mL kanamycin. Once delivered to the culture region the droplets were incubated for at least 30 min at 37oC.

S. cerevisiae. A modified procedure from the Frozen-EZ Yeast Transformation II

Kit was employed for transforming CEN-PK cells with mCherry-expressing DNA (Zymo Research Corp., Irvine, CA). The transformation chip reservoirs were initially filled with reagents: 1) 3 µL thawed chemically competent CEN-PK cells in EZ 2 solution, 2) 3 µL mCherry DNA (20 ng/µL) in EZ 3 solution (40-50% PEG) and 3) 3 µL medium minus uracil. As per Zymo protocols for yeast transformation the entire chip was heated to
30oC. A 235 nL droplet containing competent cells in EZ 2 solution was first mixed with
a 235 nL DNA droplet mixed with EZ 3 solution 30s. The heat-shocked droplet was then merged with a 235 nL droplet containing media minus uracil and moved to the devices culture region where it was incubated for 4 days at 30oC.
A. niger. Necessary reagents for A. niger transformation were first loaded: 1) 3 µL thawed protoplast, 2) 3 µL GFP DNA (20 ng/µL) in 40% PEG 3) 3 µL minimal medium with 1 M sorbitol and 4) 3 µL minimal medium with 1 M sorbitol and 400 µg/mL Hygromycin B. A 235 nL droplet containing protoplast was first mixed with a 235 nL
DNA droplet at 4oC for 1 min. This droplet was then moved to a 30oC chip region for 30
min. The cell/DNA droplet was then merged with a 235 nL media droplet and afforded
30 min outgrowth at 30oC. At this point the droplet was merged with the 235 nL

  19

Hygromycin B droplet and moved to the devices culture region. Here the droplet was incubated for 5 days at 30oC.
Transformation efficiency Off-chip culture. To assess the transformation efficiency of E. coli. by the DMF chip, a 705 nL droplet was pipetted from the culture region following 30 min outgrowth. This droplet was immediately diluted with 50 µL LB broth and spread on an LB agar plate containing antibiotics and incubated for 24 h at
37oC. Fluorescence images of the transformed colonies were taken using a FluorChem®Q
MultiImage III system equipped with a CCD camera (Alpha Innotech, Santa Clara, CA). Transformation efficiencies were calculated from the average of 5 droplets and represent the number of colony forming units (CFUs) per µg plasmid DNA. These transformation efficiencies were directly compared to the conventional benchtop method performed
following previous procedures.23

On-chip culture. The efficiency of DMF transformation and on-chip culture was initially evaluated by imaging droplets with a 4x objective following 1 to 5 days culture. A total droplet fluorescence value obtained by image analysis using ImageJ software was utilized to assess the quality of transformation with fluorescent protein generating DNA. High magnification images of cells were obtained by transferring droplets from the

device to a coverslip by a pipette. The droplet was covered by a second coverslip and then transferred to a microscope for imaging by a 100x objective. All images were taken with an inverted IX73 microscope (Olympus, Center Valley, PA) and iXON+ EMCCD camera (Andor Technology LTD, Belfast, UK).

  20

ASSOCIATED CONTENT

Supporting Information

Details are supplied in the Supporting Information for the following: reagents, device fabrication, cell preparation, plasmid preparation and transformation protocols. This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author

*E-mail: aksingh@sandia.gov

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS

This work conducted by the Joint BioEnergy Institute was supported by the Office of Science, Office of Biological and Environmental Research, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energyâ??s National Nuclear Security Administration under contract DE-AC04-94AL85000. The authors thank Dr. Anna Lechner and Dr. Leopold dâ??Espaux for assistance with yeast transformation. A. Niger protoplast and GFP plasmid were kindly provided by Dr. Saori Campen and Dr. Jinxiang Zhang.

  21

REFERENCES

[1] Paddon, C. J., and Keasling, J. D. (2014) Semi-synthetic artemisinin: a model for the use of synthetic biology in pharmaceutical development, Nat Rev Microbiol 12, 355-367.
[2] Rabinovitch-Deere, C. A., Oliver, J. W. K., Rodriguez, G. M., and Atsumi, S. (2013) Synthetic Biology and Metabolic Engineering Approaches To Produce Biofuels, Chem Rev 113, 4611-4632.
[3] Keasling, J. D. (2008) Synthetic biology for synthetic chemistry, Acs Chem Biol 3,
64-76.
[4] Bryksin, A. V., Brown, A. C., Baksh, M. M., Finn, M. G., and Barker, T. H. (2014) Learning from nature - Novel synthetic biology approaches for biomaterial design, Acta Biomater 10, 1761-1769.
[5] Ben Yehezkel, T., Nagar, S., Mackranrs, D., Marx, Z., Linshiz, G., Shabi, U., and Shapiro, E. (2011) Computer-aided high-throughput cloning of bacteria in liquid medium, Biotechniques 50, 124-127.
[6] Huang, H. Y., and Densmore, D. (2014) Integration of microfluidics into the synthetic biology design flow, Lab on a Chip 14, 3459-3474.
[7] Lee, C. C., Snyder, T. M., and Quake, S. R. (2010) A microfluidic oligonucleotide synthesizer, Nucleic Acids Res 38, 2514-2521.
[8] Kersaudy-Kerhoas, M., Amalou, F., Che, A., Kelly, J., Liu, Y., Desmulliez, M. P. Y., and Shu, W. (2014) Validation of a Fully Integrated Platform and Disposable Microfluidic Chips Enabling Parallel Purification of Genome Segments for Assembly, Biotechnology and Bioengineering 111, 1627-1637.
[9] Nagamine, K., Onodera, S., Torisawa, Y. S., Yasukawa, T., Shiku, H., and Matsue, T. (2005) On-chip transformation of bacteria, Anal Chem 77, 4278-4281.
[10] Sha, J., Wang, Y., Wang, J., Ren, L., Tu, Q., Liu, W., Wang, X., Liu, A., and Wang, L. (2011) Capillary-composited microfluidic device for heat shock transformation of Escherichia coli, J Biosci Bioeng 112, 373-378.
[11] Wei, Z. W., Zhao, D. Y., Li, X. M., Wu, M. X., Wang, W., Huang, H., Wang, X. X., Du, Q., Liang, Z. C., and Li, Z. H. (2011) A Laminar Flow Electroporation System for Efficient DNA and siRNA Delivery, Analytical Chemistry 83, 5881-5887.
[12] Lin, Y. C., Jen, C. M., Huang, M. Y., Wu, C. Y., and Lin, X. Z. (2001) Electroporation microchips for continuous gene transfection, Sensor Actuat B-Chem 79,
137-143.
[13] Zhan, Y. H., Wang, J., Bao, N., and Lu, C. (2009) Electroporation of Cells in
Microfluidic Droplets, Analytical Chemistry 81, 2027-2031.

  22

[14] Xiao, K., Zhang, M. Y., Chen, S. Y., Wang, L. M., Chang, D. C., and Wen, W. J. (2010) Electroporation of micro-droplet encapsulated HeLa cells in oil phase, Electrophoresis 31, 3175-3180.
[15] Qu, B., Eu, Y. J., Jeong, W. J., and Kim, D. P. (2012) Droplet electroporation in microfluidics for efficient cell transformation with or without cell wall removal, Lab on a Chip 12, 4483-4488.
[16] Luo, C., Yang, X., Fu, Q., Sun, M., Ouyang, Q., Chen, Y., and Ji, H. (2006) Picoliter-volume aqueous droplets in oil: electrochemical detection and yeast cell electroporation, Electrophoresis 27, 1977-1983.
[17] Hong, J. W., Chen, Y., Anderson, W. F., and Quake, S. R. (2006) Molecular biology on a microfluidic chip, J Phys-Condens Mat 18, S691-S701.
[18] Sha, J., Wang, Y. L., Wang, J. C., Liu, W. M., Tu, Q., Liu, A. J., Wang, L., and Wang, J. Y. (2011) Heat-shock transformation of Escherichia coli in nanolitre droplets formed in a capillary-composited microfluidic device, Anal Methods-Uk 3, 1988-1994.
[19] Au, S. H., Shih, S. C., and Wheeler, A. R. (2011) Integrated microbioreactor for culture and analysis of bacteria, algae and yeast, Biomed Microdevices 13, 41-50.
[20] Kim, H., Jebrail, M. J., Sinha, A., Bent, Z. W., Solberg, O. D., Williams, K. P., Langevin, S. A., Renzi, R. F., Van De Vreugde, J. L., Meagher, R. J., Schoeniger, J. S., Lane, T. W., Branda, S. S., Bartsch, M. S., and Patel, K. D. (2013) A Microfluidic DNA Library Preparation Platform for Next-Generation Sequencing, Plos One 8.
[21] Shih, S. C., Gach, P. C., Sustarich, J., Simmons, B. A., Adams, P. D., Singh, S., and Singh, A. K. (2014) A droplet-to-digital (D2D) microfluidic device for single cell assays, Lab Chip 15, 225-236.
[22] Hanahan, D. (1983) Studies on Transformation of Escherichia-Coli with Plasmids, J Mol Biol 166, 557-580.
[23] Ausubel, F. M. (2001) Current protocols in molecular biology, J. Wiley,, New York. [24] Linshiz, G., Stawski, N., Goyal, G., Bi, C. H., Poust, S., Sharma, M., Mutalik, V.,
Keasling, J. D., and Hillson, N. J. (2014) PR-PR: Cross-Platform Laboratory Automation
System, Acs Synth Biol 3, 515-524.
[25] Ito, H., Fukuda, Y., Murata, K., and Kimura, A. (1983) Transformation of Intact
Yeast-Cells Treated with Alkali Cations, J Bacteriol 153, 163-168.
[26] Lubertozzi, D., and Keasling, J. D. (2009) Developing Aspergillus as a host for heterologous expression, Biotechnol Adv 27, 53-75.

  23

[27] Szewczyk, E., Nayak, T., Oakley, C. E., Edgerton, H., Xiong, Y., Taheri-Talesh, N., Osmani, S. A., and Oakley, B. R. (2006) Fusion PCR and gene targeting in Aspergillus nidulans, Nat Protoc 1, 3111-3120.
[28] Fiddes, L. K., Luk, V. N., Au, S. H., Ng, A. H., Luk, V., Kumacheva, E., and
Wheeler, A. R. (2012) Hydrogel discs for digital microfluidics, Biomicrofluidics 6,
14112-1411211.

  24

Figure 1. Schematic of the basic operations of the DMF transformation including: droplet generation, merging and relocation to thermally controlled regions (A). Brightfield (B) and infrared (C) images of the DMF device for cell transformation. Image sequence of
on-chip DMF transformation. Images of filled reservoirs (D), generation of droplets containing DH5α cells or GFP plasmid (E) followed by mixing of droplets (F) at 4oC. Droplet transferred to 42oC region of device (G) then moved to 4oC device region during generation of droplet consisting of LB and AMP (H). Droplets are then merged (I) and allowed to recover at 4oC (J) then transferred to a 37oC culture chamber (K).

  25

Figure 2. Brightfield and fluorescence images of a droplet containing DH5α cells and pRSET-EmGFP plasmid DNA without heat shock (A,B) and following transformation with in the DMF device (C,D). Fluorescence images of LB agar plates containing DH5α cells without heat shock (E) and following transformation with GFP plasmid DNA by the DMF device (F) and the conventional tube-based method (G). Plot of transformation efficiencies obtained under various heat shock procedures (H) and with different concentrations of GFP plasmid DNA (I).

  26

BF' Blue' Green' Red'

Control'

BFP' GFP'

RFP

25'u

Figure 3. Brightfield and fluorescence images of DH5α cells transformed with blue fluorescence protein (BFP), green fluorescence protein (GFP) or red fluorescence protein (RFP) by the DMF device.

  27

Yeast#+#mCherry# A.#Niger#+#GFP##

m# m#

100#um#

Figure 4. Brightfield and fluorescence images of droplets containing CEN.PK cells transformed with mCherry (left) or Asperagillus niger cells transformed with GFP (right) following 5 days culture at 30oC on chip.

  28

Supporting Information For

Synthetic Biology in a Chip: A Digital Microfluidic Platform for Cell

Transformation, Culture and Expression

Philip C. Gach,1, 2 Steve C.C. Shih,1, 2 Jess Sustarich,1, 2 Jay D. Keasling,1,3,4,5,6,7
Nathan J. Hillson,1,3,8 Paul D. Adams,1,3,4 and Anup K. Singh1, 2 *

1 Technology Division, Fuels Synthesis Division, Joint Bioenergy Institute (JBEI), Emeryville, California 94608, United States

2 Applied Biosciences and Engineering, Sandia National Laboratories, Livermore,

California 94550, United States

3 Physical Biosciences Division, Lawrence Berkeley National Lab, Berkeley, California

94720, United States

4 Department of Bioengineering, University of California, Berkeley, California 94720, United States

5 Department of Chemical & Biomolecular Engineering, University of California,

Berkeley, California 94720, United States

6 QB3 Institute, University of California, Berkeley, California 94720, United States

7 Synthetic Biology Engineering Research Center, Emeryville California 94608, United

States

8 DOE Joint Genome Institute, Walnut Creek, California 94598, United States

* Corresponding Author email: aksingh@sandia.gov tel: (925) 294-1260
fax: 925-294-3020

  29

METHODS

Reagents and materials. Unless otherwise specified, general-use reagents were purchased from Sigma Aldrich. Fabrication reagents and supplies included SU-8-5, SU-

8-2025, SU-8-2075, S-1811 and SU-8 Developer from Microchem (Newton, MA), MF-
321 positive photoresist developer from Rohm and Haas (Marlborough, MA), CR-4 chromium etchant was from OM Group (Cleveland, OH), and AZ-300T photoresist stripper from AZ Electronic Materials (Somerville, NJ).
Device fabrication. The digital microfluidic heat shock devices were fabricated using standard photolithography as described previously with moderate adjustments.1 The device consisted of a glass bottom, gold electrodes, dielectric layer, channel layer and an ITO cover. All photolithography was performed by exposing a photopolymer through a photomask (Fineline Imaging Inc., Colorado Springs, CO) by an Oriel® 500 W Hg exposure system (Newport Corporation, Irvine, CA). Gold electrodes were initially micropatterned on a glass slide by exposing, developing and etching a gold and S1811 coated glass slide (Telic Company, Valencia, CA). Following cleaning with acetone, IPA and O2 plasma (Harrick Plasma, Ithaca, NY): devices were coated with a 5 mm layer of SU-8 5. After selective removal of the SU-8 5 over the side contacts and another cleaning step, 150 mm channels composed of SU-8 2075 were fabricated over the dielectric layer. Following development, devices were rinsed with IPA and DI water, dried with N2 gas
and hard-baked for 15 min. at 200°C. Prior to assembly, the DMF device and ITO coated
glass slide (Delta Technologies, Stillwater, MN) were coated with 0.2 mm filtered
Aquapel (TCP Global, San Diego, CA) for 15 min. A kimwipe was then used to remove

  30

the Aquapel and the device was rinsed with DI water, dried with N2 gas and kept at 21°C
for 1h.
Cell Preparation. Chemically competent E. coli. A common strain of Escherichia coli (E. coli) for laboratory cloning (DH5α) was acquired from the JBEI registry
(https://registry.jbei.org) and made chemically competent using a standard CaCl2 procedure.2 Reagents and materials were kept chilled at 4oC during all aspects of cell preparation. Briefly, a bacterial cell colony in Luria-Bertani (LB) broth was cultured in a
37°C incubator at 200 rpm until early log phase was achieved (OD 600 = 0.3) measured with a spectrophotometer. At this point, the cell suspension was placed on ice for 15 min then centrifuged for 10 min at 3300 g. The medium was then replaced with 0.1 M CaCl2 in which the cells suspended on ice for 30 min. The suspension was again centrifuged for
10 min at 3300 g and the supernatant exchanged for a 0.1 M CaCl2 solution containing
15% glycerol. The cell suspension was then aliquoted into micro-centrifuge tubes and stored at -80oC for future use.
Chemically competent yeast. A commercially available kit (Frozen-EZ Yeast Transformation IITM) from Zymo Research Corp. (Irvine, CA) was used to prepare competent Saccharomyces cerevisiae (CEN.PK) cells obtained from the JBEI registry. Following the manufactures protocol, cells were grown at 30oC at 200 rpm until mid-log phase was reached (OD600 = 0.9). The cell suspension was then centrifuged at 500 g for 4 min and media replaced with EZ 1 solution (â?¤1% Tris and â?¤1% EDTA in water). Suspended cells were again centrifuged at 500 g for 4 min and the supernatant replaced with EZ 2 solution (1M sorbitol and 7% DMSO in water). Competent cells are then aliquoted into micro-centrifuge tubes and kept frozen at -70oC until needed.

  31

Protoplast preparation. Aspergillus niger protoplasts were prepared using a previous protocol with minor modifications.3 Briefly, 106 A. niger (ATCC® 11414TM) conidia in 1 mL H2O were added to 100 mL YEPD broth and cultured in a 30oC
incubator while shaking at 150 rpm for 12 hrs. A Miracloth was used to filter the mycelia,
which were then washed with stile H2O. Mycelia were digested in 40 mL Vinoflow solution (30 mg/mL Vino Taste Pro, 0.6 M Ammonium Sulphate, 50 mM Maleic acid, pH 5.5) at 30oC, shaking at 70 rpm for 4 hrs. At this point, the culture was filtered through a Miracloth and the filtrate centrifuged at 800 g for 10 min. The supernatant was then removed and the pellet washed twice with 25 mL of ST solution (1 M sorbitol, 50 mM Tris, pH 8.0) then resuspended in 10 mL STC solution (1 M sorbitol, 50 mM Tris, pH 8.0, 50 mM CaCl2). This suspension was centrifuged at 800 g for 10 min and then
supernatant replaced with fresh STC solution until a protoplast concentration of 1.2Ã?107
spores/mL was achieved. 40% PEG 4000 in STC solution is then gently mixed with this suspension (20% v/v), which is then mixed with DMSO (7% v/v). Protoplasts are then aliquoted into micro-centrifuge tubes and frozen at -80oC.

Plasmid Preparation. E. coli. Plasmid DNA was extracted from E. coli cells

expressing blue fluorescence protein (pBAD-mTag BFP2) 4 acquired from Addgene, green fluorescence protein (pRSET-EmGFP) from Life Technologies (Grand Island, NY) or red fluorescence protein (DsRed.T3)5 and kanamycin resistant green fluorescence protein obtained from the JBEI registry (https://registry.jbei.org) using QIA miniprep from Qiagen (Valencia, CA).
pProm1_BCD1-GFP DNA fragments used for Golden Gate assembly were prepared by PCR using primers and plasmids described elsewhere.6 Briefly, 50 µL PCR

  32

reactions consisted of 2.5 µL (2.5 µM) of each forward and reverse primer, 1 µL
template, 1 µL dNTPs (10 mM), 0.5 µL high fidelity iProof phusion polymerase (BioRad; Hercules, CA), 10 µL 5�~ high fidelity phusion buffer, and 32.5 µL deionized water.
Four 50 µL PCR reactions (200 µL total) were performed for each fragment amplified. The following PCR thermocycling conditions were used: denaturation at 98°C for 30 s,
38 cycles of denaturation at 98°C for 20 s, annealing at 68 ° C for 30 s, and elongation at
72°C for 15 s each kb, and a final extension at 72°C for 10 min. Following PCR amplification the samples were digested and purified. At this point a modified Golden Gate DNA assembly method was performed. 1 µL BsaI-digested BCD1-GFP fragment was mixed with 1 µL BsaI-digested pProm1 promoter fragment, 1 µL BsaI-digested vector backbone, 1 µL T4 ligase enzyme (Thermo Scientific), 1 µL of T4 ligase buffer and 5 µL deionized water for 30 min at room temperature.

S. cerevisiae. pL211 containing a TEF promotor and ADH1, URA- and mCherry was provided by the JBEI registry (https://registry.jbei.org).

A. niger. GFP genes were cloned into an A. Niger/ E. coli Shuttle vector at JBEI

as described previously.

Supplementary Table 1. Heat-shock protocols

Protocol name Description Antibiotic DNA - cell mixing Heat-shock Recovery Outgrowth Culture duration / duration / duration / duration / duration /

temperature temperature temperature temperature temperature

1 Long heat-shock at 42 oC Ampicillin 15 min / 4 oC 45 s / 42 oC 2 min / 37 oC 0 min 1 day / 37 oC

2 Long heat-shock at 37 oC Ampicillin 15 min / 4 oC 45 s / 37 oC 2 min / 37 oC 0 min 1 day / 37 oC

3 Quick heat-shock Ampicillin 15 min / 4 oC 10 s / 42 oC 2 min / 37 oC 0 min 1 day / 37 oC

5 Quick mixing and heat-shock Ampicillin 10 s / 4 oC 10 s / 42 oC 2 min / 37 oC 0 min 1 day / 37 oC

6 Quick mixing, heat-shock and recovery Ampicillin 10 s / 4 oC 10 s / 42 oC 10 s / 37 oC 0 min 1 day / 37 oC

7 Kanamycin resistance Kanamycin 10 s / 4 oC 10 s / 42 oC 10 s / 37 oC 30 min / 37 oC 1 day / 37 oC

8 Yeast heat-shock Ampicillin 10 s / 30 oC 10 s / 30 oC 10 s / 37 oC 0 min 4 days / 30 oC

9 A. niger heat-shock Hygromycin B 10 s / 4 oC 10 s / 30 oC 10 s / 37 oC 30 min / 30 oC 5 days / 30 oC

  33

References

[1] Shih, S. C., Gach, P. C., Sustarich, J., Simmons, B. A., Adams, P. D., Singh, S., and Singh, A. K. (2014) A droplet-to-digital (D2D) microfluidic device for single cell assays, Lab Chip 15, 225-236.
[2] Gannon, F., Neilan, J., and Powell, R. (1988) Current Protocols in Molecular-Biology
- Ausubel,Fm, Nature 333, 309-310.
[3] Szewczyk, E., Nayak, T., Oakley, C. E., Edgerton, H., Xiong, Y., Taheri-Talesh, N., Osmani, S. A., and Oakley, B. R. (2006) Fusion PCR and gene targeting in Aspergillus nidulans, Nat Protoc 1, 3111-3120.
[4] Subach, O. M., Cranfill, P. J., Davidson, M. W., and Verkhusha, V. V. (2011) An Enhanced Monomeric Blue Fluorescent Protein with the High Chemical Stability of the Chromophore, Plos One 6.
[5] Sorensen, M., Lippuner, C., Kaiser, T., Misslitz, A., Aebischer, T., and Bumann, D. (2003) Rapidly maturing red fluorescent protein variants with strongly enhanced brightness in bacteria, Febs Lett 552, 110-114.
[6] Linshiz, G., Stawski, N., Goyal, G., Bi, C. H., Poust, S., Sharma, M., Mutalik, V., Keasling, J. D., and Hillson, N. J. (2014) PR-PR: Cross-Platform Laboratory Automation System, Acs Synth Biol 3, 515-524.