Microreactor Array Device

We report a device to fill an array of small chemical reaction chambers (microreactors) with reagent and then seal them using pressurized viscous liquid acting through a flexible membrane. The device enables multiple, independent chemical reactions involving free floating intermediate molecules without interference from neighboring reactions or external environments. The device is validated by protein expressed in situ directly from DNA in a microarray of ~10,000 spots with no diffusion during three hours incubation. Using the device to probe for an autoantibody cancer biomarker in blood serum sample gave five times higher signal to background ratio compared to standard protein microarray expressed on a flat microscope slide. Physical design principles to effectively fill the array of microreactors with reagent and experimental results of alternate methods for sealing the microreactors are presented.

functionalizing chemicals using non-contact piezoelectric inkjet dispensing technology [15][16][17] . Portions of these chemicals are bound to the functionalized surfaces of the microreactors. Dried printed slides may be stored for later processing. The slides may be soaked in a blocking buffer to wash away remaining unbound chemicals and to mitigate nonspecific binding. A centrifuge or vacuum chamber is used to force entrapped air out of the microreactors and fill them with the blocking buffer. After rinsing and drying, slides are inserted into the fill & seal device, Figure 1. An O-ring is placed around the periphery of the slide for vacuum or pressure sealing. A transparent, flexible, impenetrable, smooth, sealing membrane is placed over the O-ring and slide. A transparent window is placed over the sealing membrane and the assembly is clamped together in a rigid frame using fasteners.
Approximately 300 ml of degassed reagent is injected into a reagent inlet/outlet port and forced through a thin fluid gap between the slide and sealing membrane. Vacuum is applied to the fluid gap, through a reagent outlet port, before injecting the reagent. Microreactors are filled with reagent at a rate of greater than 10,000 microreactors per second. The functionalizing chemicals stay in place during filling since they are bound to the surface. High viscosity, pressurized, incompressible, sealing liquid is injected onto the sealing membrane through an inlet port in the window. The sealing membrane wipes excess reagent from the slide surface as the sealing liquid spreads out across the membrane. Excess reagent is forced out through inlet or outlet reagent ports and ,200 ml may be recovered. After spreading out across the whole surface, the pressurized sealing liquid continues to apply even pressure onto the slide surface through the sealing membrane. This uniformly seals all of the microreactors into isolated reaction containers.
The process of filling and sealing all of the microreactors takes less than five seconds. It is actuated entirely by opening or closing valves and is therefore amenable to automation via electronic actuators. Reagent reacts with functionalizing chemicals in the microreactors possibly creating untethered intermediate molecules that are contained for extended incubation periods. After incubation, the slide is removed from the fill & seal device for further chemical processing such as washing, blocking or labeling. To preserve molecular functionality, the slide is typically never allowed to dry out during an assay. The whole slide may be flooded with sample fluid to detect probe (query) molecules by their affinity to specific chemical products (targets) in the microreactors. After chemical processing the slide may be dried and imaged using a fluorescent microarray scanner for example.
Alternate methods for sealing microreactors. Three alternate methods for sealing microreactors were evaluated. Devices were fabricated to test the alternate methods. The alternate methods use the same process of filling microreactors with reagent as the preferred device of Figure 1. All of the alternate methods also use a sealing membrane but differ in how they actuate the membrane, either by: 1) silicone rubber, 2) gas or 3) hydraulic oil. Figure 2 is a schematic representation of the device based on silicone rubber. It is similar to the device reported previously 15 with the addition of the reagent outlet port and the sealing membrane. The reagent outlet port enables applying vacuum separately from the inlet port. Previously 15 vacuum was applied through Features are not drawn to scale. (B) Section view schematic of same device illustrating the process of displacing excess reagent and sealing microreactors. After completely filling the microreactors, excess reagent is displaced from the slide by pressurized sealing liquid injected through a sealing liquid inlet port acting on a sealing membrane. As the sealing liquid spreads out across the sealing membrane it first displaces excess reagent though reagent inlet & outlet ports and then seals the microreactors into isolated chemical reaction containers. The schematics (A, B) were drawn by Peter Wiktor. the reagent at the inlet port. This introduced air into the reagent which interfered with filling of the microreactors. The sealing membrane isolates the microreactors from the porous silicone rubber and thus reduces diffusion after sealing the microreactors. The other two alternate methods are similar to Figure 1 with either gas or hydraulic oil replacing the viscous sealing liquid. Pros and cons of the various approaches are summarized in Discussion.
Experimental evaluation of fill & seal methods. The alternate fill & seal methods were experimentally evaluated for NAPPA and compared to the preferred method of Figure 1 and to flat glass. Printing mixture, containing capture antibodies, is printed into microreactors followed by plasmid DNA. To clearly detect diffusion, spots containing printing-mixture & plasmid DNA are surrounded by spots containing just the printing-mixture. Any expressed proteins that diffuse away from a DNA spot are captured by the antibodies in neighboring spots and therefore easily detected. Fluorescently labeled proteins expressed from the DNA are displayed in the subarrays of Figure 3. The top subarray (A) was expressed using the preferred fill & seal device of Figure 1. There is negligible signal in the adjacent spots indicating negligible diffusion and capture of expressed proteins. This is significant since the signal at a given spot is due exclusively to the analyte at that spot and not from its neighbors. This reduces false positives during data analysis. Brightness and contrast are adjusted at the right third of the subarray in Figure 3A to show actual printing density and locations of the adjacent spots.
Standard NAPPA on flat glass is compared to NAPPA on microreactor array in Figure 3B. The same set of plasmid DNA was printed on flat glass (B) at the same time as microreactor arrays (A). Printing conditions were optimized for microreactor arrays. Genes were expressed into their corresponding proteins by flooding the whole array with IVTT reagent and then incubating for three hours. Proteins on the substrates in (A) and (B) were fluorescently labeled, scanned and displayed using identical conditions. There is more diffusion on flat glass (B) vs. microreactor array (A). However standard NAPPA typically has lower diffusion if the printing conditions are optimized for flat glass 11 . Normally, for standard NAPPA, mixtures of higher concentration DNA & printing mixture are printed using a pinspotter at four times lower spot density. Diffusion on flat glass is much lower under these standard conditions.
Protein subarrays expressed using the three alternative methods: silicone rubber, gas and hydraulic oil are shown in Figure 3 (C, D and E) respectively. Signal to background ratio of the various sealing methods are plotted in Figure 4.   Validation of microreactor arrays as a clinical screening platform. Some cancer patients produce autoantibodies against p53, a tumor suppressor protein 18 . These autoantibodies can be detected in blood serum samples using immunoassays such as enzyme-linked immunosorbent assay (ELISA) or NAPPA. NAPPA has dynamic range and limit of detection comparable to ELISA 19 . We compared sensitivity of detecting anti-p53 autoantibodies using NAPPA on microreactor array vs. flat glass. Full length genes for p53 protein were printed on both microreactor array and flat glass and processed as described in Methods. Printing conditions were optimized for their respective substrates, i.e. piezoelectric dispensing for microreactor arrays and pinspotting for flat glass. The two types of substrates were then probed with serum sample from the same colon cancer patient at four serum dilutions from 1550 to 15900. Signal to background ratio of anti-P53 response is five times higher for NAPPA on microreactor array vs. standard flat glass, Figure 5. Signal to background ratios in Figure 5 are much lower than Figure 4 since serum has many different proteins that bind nonspecifically to the surface and increase background.
A bitmap image rendered in fluorescently labeled p53 protein expressed in microreactors from plasmid DNA for the p53 gene is shown in Figures 6. The image illustrate fidelity of the microreactor array platform with density greater than 10,000 protein spots expressed on a single microscope slide format with no diffusion or evaporation during three hours incubation. Moreover, the ability to control different protein levels by adjusting the amount of printed DNA, demonstrates the nuanced ability of these microreactors to test reactions under quantitative conditions.

Discussion
Previous work. The microreactor array is a synergy of microplate and microarray technology. It miniaturizes microplates 1 down to the scale of microarrays 20-24 allowing tens of thousands independent biochemical reactions simultaneously on a single microscope slide format. It is based on an array of microreactors that are accessed in parallel by a thin sheet of reagent and then sealed by viscous liquid acting through a flexible sealing membrane. To our knowledge, this micofluidic configuration is the first of its kind. It does not follow the conventional 'lab-on-a-chip' paradigm, i.e. a miniaturized, serial, 2D network of micro-channels emulating an electronic circuit or conventional chemical processing plant. Such platforms are also known as 'micro total analysis systems' (mTAS), 'micro electrical mechanical systems' (MEMS) or 'microreactors' [25][26][27][28][29] . They were originally developed for polymerase chain reaction (PCR) 30,31 . Various planar microreactor array type platforms have also been developed for PCR [32][33][34][35][36][37][38][39][40][41][42][43] and other life science applications 44-52 but none use the unique microfluidic configuration presented here. Cell-free in vitro protein expression has been done in various microreactor formats 53-59 but never using the present configuration for filling and sealing the microreactors.
Microreactor filling. Physical design principles to effectively fill the array of microreactors with reagent are presented based on fluid mechanics nondimensional analysis. Reagents in the life sciences are typically aqueous solutions. Water has high surface tension which dominates other forces at the small dimensions of microreactors making them difficult to fill with reagent. For the microreactor array platform, reagent is forced into microreactors by first applying vacuum to a thin fluid gap between the microreactor array slide surface and sealing membrane and then quickly injecting reagent, under pressure, into this gap. This approach fills microreactors by overcoming surface tension through a combination of vacuum, pressure, inertial, wetting and viscous forces. Kinetic and potential energy of the reagent overcomes its surface energy. Similar issues, concerning filling small features with fluids, arise in other fields such as plastic injection molding and microimprint lithography 60 .   Physical parameter values for water are used here to calculate the non-dimensional numbers. Filling microreactors with actual reagents may be assisted by thickening agents to increase viscosity (lower R e ) or surfactants to decrease surface tension (higher C a ). Small concentrations of surfactants can dramatically decrease surface tension.
Microreactor sealing. Microreactors are sealed with viscous liquid acting through a flexible sealing membrane, Figure 1. Viscous liquid has the appropriate combination of physical properties to seal the microreactors: incompressible flow to displace excess reagent from the microreactor array surface followed by isotropic pressure to seal all of the microreactors uniformly. This is the first time that viscous liquid is used in this manner and is a key advancement for this process. It provides a simple elegant solution to a challenging problem.
As shown in Figures 3 & 4, alternate methods for sealing microreactors are not effective in preventing diffusion between microreactors. Using silicone rubber is the best alternate method but it is not completely effective. Silicone rubber presses down uniformly on the microreactor surface invariably entrapping reagent in small pockets that prevent complete sealing of the microreactors in those areas. Using either gas or hydraulic oil is not effective since each one quickly spreads out across the sealing membrane entrapping a layer of reagent that again prevents complete sealing of the microreactors. Compared to the preferred method of using viscous sealing liquid, the alternate methods clearly demonstrate the need for an effective means of first wiping reagent away from the surface before sealing the microreactors.
Conclusion. For NAPPA, the microreactor array platform provides an array of miniaturized reaction chambers that can be individually programmed to produce any desired protein. NAPPA is commonly used to screen antibody affinity to those proteins in serum samples from healthy vs. diseased individuals. The microreactor array platform has higher signal to background ratio of antibody response compared to standard NAPPA on flat microscope slides, Figure 5. It can therefore detect serum antibodies with higher sensitivity and potentially improve the accuracy of clinical studies correlating antibody response to disease.

Methods
Methods to generate NAPPA protein microarrays using microreactor arrays with the preferred fill & seal device of Figure 1 are presented. Methods to screen for antibodies in patient serum samples and analyze the results are also presented.  Functionalize surfaces. Clean residual organic materials from the microreactor array slides in piranha (151 mixture of H 2 S0 4 5H 2 O 2 ) for 30 minutes. After rinsing with DI water and drying with compressed air, immerse slides in 2% solution of (3-Aminopropyl) triethoxysilane (APTES) in acetone for 30 minutes. Thoroughly rinse slides in acetone and then DI water. Dry the slides with compressed air.
Print DNA into microreactors. Obtain plasmid DNA. Obtain a set of genes to be printed from DNASU Plasmid Repository (Center for Personal Diagnostics, Biodesign Institute, Arizona State University, Tempe AZ). Clones are sequence verified and inserted in the E.Coli pDNR-dual recombinational cloning vector open reading frames (ORFs) with the natural stop codon absent and GST-tag appended to C-terminus. Store plasmid DNA. Store plasmid DNA at 4uC until printing.
Prepare NAPPA printing-mixture. Prepare 1X printing-mixture. Thaw components and keep them on ice. Mix components together in the following order and proportions: 93% nuclease-free DEPC-treated water (Ambion, Life Technologies, Grand Island NY), 0.6% BSA (Sigma-Aldrich), 1% anti-GST antibody (GE Healthcare), 5% BS3 crosslinker (Thermo Scientific, Pierce). Keep mixture on ice. Age printingmixture at 4uC for one day to allow partial crosslinking of BS3. Aliquot printingmixture into 384-well microplate for printing. Keep microplate on ice until ready to print.
Prepare microarray pattern. Predefine a microarray printing pattern for the 96 plasmid DNA and the printing-mixture. Surround each spot containing printing-mixture and plasmid DNA with 6 spots containing just the printing-mixture without the plasmids. In this way, expressed proteins that diffuse away from a plasmid spot are captured by the anti-GST antibodies in the printing-mixture of neighboring spots and then easily detected.
Align microreactor array slides. Align APTES functionalized microreactor array slides on the deck of a non-contact piezoelectric dispensing microarrayer (Rainmaker-au302, Engineering Arts LLC, Tempe AZ). Use the au302 alignment system to align the microreactors of the slides for continuous non-contact dispensing.
Print microreactor array slides. Use either 8 or 16 piezoelectric dispensers (Engineering Arts LLC, Tempe AZ) on the au302 microarrayer. Prime dispensers with DI water. Place the 384-well microplate with printing-mixture on the deck of the au302 microarrayer. Aspirate 2 ml printing-mixture into each dispenser using on-head aspiration syringes. Dispense twelve 0.1 nl drops of printing-mixture in short bursts at 12,500 drops-per-second into microreactors using the predefined microarray pattern defined above. Use non-contact piezoelectric on-the-fly dispensing at uniform print-head speed of 175 mm/sec. Dispense the same number of drops with the same predefined pattern on flat microscope slides for Figure 2B. Clean the piezoelectric tips by flushing DI water through the dispenser while the dispenser tip is submerged in flowing DI water. Within 10 minutes, repeat the same process with the 384-well microplate of plasmid DNA. Print 3 drops of plasmid DNA in the same spots as the printing-mixture. Use the same aspirating and dispensing parameters (12,500 drops-per-second) as for the printing-mixture. Do not print plasmid DNA into the 6 surrounding microreactors containing just printing-mixture alone. Repeat the process of dispensing printing-mixture followed by plasmid DNA until all of 96 genes are printed. DNASU plate 4156 containing 96 general purpose genes was printed on the slide in Figure 2A. Print on flat glass at the same time for Figure 2B. Print plasmid DNA for the p53 gene onto the slide in Figure 6 using a PiXY piezoelectric microarrayer that prints colored bitmaps.
Print flat glass slides for p53 serum screening comparison. Mix DNA and printing mixture together. Print mixture on APTES coated glass microscope slides (Fischer Scientific, Waltham MA, Cat. No. 12-544-1) using Genetix QArray2 pin-spotter at 2,500 spot density per microscope slide.
Store microarrays. Store printed plasmid DNA microarrays at room temperature in a sealed container with desiccant. Printed plasmid DNA microarrays can be stored for several months before protein expression.
Express proteins. Block microreactor array slides. This step also washes away unbound printed molecules from slides' surfaces. Program centrifuge (Beckman Coulter model Allegra X-15R, Indianapolis IN) for 3750 RPM, maximum acceleration and maximum deceleration. Submerge slides in a tray containing 8 ml of Superblock Blocking Buffer in TBS (Thermo Scientific product #37535B, Rockford IL). Accelerate up to the maximum speed and decelerate right away. Rinse slides in DI water and dry them with compressed air.
Block glass slides. Use the same procedure as outlined above for microreactor array slides. However, the centrifuge is not required since glass slide surfaces are flat so they do not entrap air. Block slides for one hour on a rocking table (VWR model 200, Radnor PA).
Prepare in vitro coupled transcription and translation (IVTT) reagent. Prepare IVTT reagent from the ''1-step human in vitro protein expression kit'' (Thermo Scientific, Rockford, IL). Thaw out the four components of the kit and keep them on ice. Mix components in the following order and proportions: 34% HeLa lysate, 10% accessory proteins, 22% reaction mix and 34% nuclease-free water. This uses 60% HeLa lysate compared to the normal recipe. Keep mixture on ice. Degas lysate in a vacuum desiccator until bubbles are gone in approximately 5 minutes.
Apply IVTT reagent to microreactor array slides. Insert a microreactor array slide into the fill & seal device of Figure 1 (Biodesign Institute, ASU, Tempe AZ). Put the O-ring around the slide. Place sealing membrane on top of the O-ring and slide. Place window on top of the sealing membrane and clamp the fill & seal device frame together using fasteners. Apply vacuum to the reagent outlet port and inject 300 ml IVTT reagent into the inlet/outlet port. Inject pressurized sealing liquid into the inlet port in the window to displace the reagent and seal the microreactors.
Apply IVTT reagent to flat glass slides. Attach gasket (Hybriwell Sealing System item 440904, Grace Biolabs, Bend OR) to the slide. Using manual pipette, inject 150 ml IVTT reagent into the hole at one end of the gasket.
Incubate glass slides. Incubate glass slides and gasket at 30uC in a chilling incubator (EchoTherm chilling incubator, Torrey Pines Scientific, Carlsbad CA) for 1.5 hours and then 15uC for 0.5 hours.
Incubate microreactor array slide. Incubate the fill & seal device assembly at 30uC in a chilling incubator (EchoTherm chilling incubator, Torrey Pines Scientific, Carlsbad CA) for 2 hours and then 15uC for 1 hour. Incubation times are longer than standard NAPPA on glass slides to accommodate longer heating and cooling times due to the higher thermal mass of the fill & seal device.
Detect fluorescent signals. Fluorescently label proteins. To preserve molecular functionality, do not allow slides to dry out between processing steps. Prepare 5% milk-PBST (0.2% Tween) blocking buffer. Combine 500 ml 1x phosphate buffered saline (1 3 PBS) with 25 grams 100% instant nonfat powdered dry milk and 1 ml Tween 20 detergent and mix with magnetic stirring bar for 10 minutes. Store at 4uC. Fill a small tray with blocking buffer. Disassemble the fill & seal device, remove the slide and submerge it into blocking buffer. Block for one hour at room temperature on a rocking table (VWR model 200, Radnor PA) replacing blocking buffer three times. For Figures 3 & 4, prepare primary label: mix 10 ml mouse anti-GST monoclonal antibody (Cell Signaling Technologies, Danvers MA) in 3 ml blocking buffer. Incubate slides in primary label overnight at 4uC on a rocking table. Rinse slides with blocking buffer 3 times for 10 minutes each time; replacing blocking buffer each time. Prepare secondary label: mix 6 ml Alexa Fluor 647 nm goat anti-mouse IgG antibody (Life Technologies, Grand Island NY) in 3 ml blocking buffer. Incubate slides in secondary label for 1 hour at room temperature on a rocking table in the dark to prevent photo bleaching of the fluorescent dye. Prepare rinsing buffer: 500 ml 1 3 PBS and 1 ml Tween. Rinse slides in rinsing buffer three times for 1 minute each time; replacing the rinsing buffer each time. Rapidly rinse slides in DI water 6 times replacing water each time. Dry slides with compressed air. Store slides in the dark, at room temperature with a desiccator.
Probe serum samples. Following protein expression, block slides in 5% milk-PBST (0.2% Tween) on a rocking table at room temperature for 1 hour. For Figure 5, incubate slides with diluted serum sample in proplate 4-well tray set (Grace Bio-Labs, OR) at 4uC overnight. Wash slides in 5% milk-PBST (0.2% Tween) three times, 5 min each time. Detect protein display with Alexa Fluor 647 labeled goat anti-human IgG secondary antibody (Jackson ImmunoResearch, West Grove, PA). Finally, wash slides in DI water and dry them with compressed air.
Image microreactor array slides. Program fluorescent microarray scanner (PowerScanner, Tecan, Mannedorf Switzerland) for microreactor array slides. Define coordinates of the empty areas at the top and bottom of slides for autofocus. Define reflectivity of those areas. Define 70 mm offset for microreactor depth after autofocus. Apply self-adhesive 325 mm thick backing to slides to accommodate 1 mm slide thickness requirement of the scanner. Scan slides at 10 mm resolution, 25% laser power (out of 30 mW) and 25% (out of 800% maximum) photo multiplier tube (PMT) detector gain. Extract data (Array-Pro, Media Cybernetics, Rockville MD). Analyze data (Excel, Microsoft, Redmond www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 8736 | DOI: 10.1038/srep08736 WA). Plot signal to background ratio of anti-P53 response, Figure 5. Signal for each dilution point is the average signal of six spots on the array. Signal of each spot is the median pixel value of that spot. Background is the median signal of all of the spots on the array.