A Novel PDMS Microfluidic Spotter for Fabrication of Protein Chips and Microarrays

A novel polydimethyl siloxane (PDMS) microfluidic spotter system has been developed for the patterning of surface microarrays that require individually addressing each spot area and high probe density. Microfluidic channels are used to address each
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006 1145 A Novel PDMS Microfluidic Spotter for Fabricationof Protein Chips and Microarrays David A. Chang-Yen  , Member, IEEE  , David G. Myszka, and Bruce K. Gale  Abstract— A novel polydimethyl siloxane (PDMS) microfluidicspotter system has been developed for the patterning of surfacemicroarrays that require individually addressing each spot areaand high probe density. Microfluidic channels are used to addresseachspotregion,andlargespotarrayscanbeaddressedinparallel.Fluorescence intensity measurement of dye-spotted samples com-pared to control and pipetted drops demonstrated a minimum of three-fold increase in dye surface density compared to pin-spotteddyes. Surface plasmon resonance (SPR) measurement of protein-spotted samples as comparedto pin-spottedsamples demonstratedan 86-fold increase in protein surface concentration. The spottingsystem has been applied successfully to protein microarrays forSPR applications, in both a 12-spot linear and 48-spot two-dimen-sional (2-D) array format. This novelspottersystem can beappliedto the production of high-throughput arrays in the fields of ge-nomics, proteomics, immunoassays, and fluorescence or lumines-cence assays. [1565]  Index Terms— Microarray, microfluidic, polydimethyl siloxane(PDMS), spotter. I. I NTRODUCTION T HE market for biomolecule-based microassays has ex-panded rapidly within the last few years, with predictedexponential increases in applications in the near future [1].These microarrays can be applied over a huge field includinggenomics, proteomics, lipidomics, metabolomics, and cel-lomics. Genomics and proteomics in particular are of greatinterest to gene sequencing, immunoassays, and drug dis-covery. The challenge is to reproducibly fabricate high-qualitymicroarray spots using a massively parallel system, whileremaining relatively cost-effective. Currently, commercial highspot density arrays are produced using complex robotic spottersystems, such as the Genetix QArray and the Perkin-ElmerBioChip Arrayer. Recent studies have focused on the fabrica-tion of increasingly larger numbers of spots on small substrateareas using either microdispensing or microcontact printing, Manuscript received April 4, 2005; revised April 11, 2006. This work wassupported in part by the National Science Foundation IGERT under Grant DGE9987616 , the Utah State Center of Excellence, University of Utah, and the Col-lege of Engineering, University of Utah. The work of D. G. Myszka was sup-ported in part by the National Science Foundation under Grant EF-0427665.Subject Editor A. Lee.D. A. Chang-Yen was with the Utah State Center of Excellence for Biomed-ical Microfluidics, University of Utah, Salt Lake City, UT 84112 USA. He isnow with Wasatch Microfluidics Inc., Salt Lake City, UT 84123 USA ( Myszka is with the Protein Interaction Facility, HSC Core Laboratories,University of Utah School of Medicine, Salt LakeCity, UT 84132 USA ( K. Gale is with the Mechanical Engineering Department, University of Utah, Salt Lake City, UT 84112 USA (e-mail: Object Identifier 10.1109/JMEMS.2006.880289 but little is done to improve the individual spot quality [2]–[5].While conventional spotting systems are capable of producingmultiple spots of a controlled size, if the desired molecule fordeposition is in very low concentration then the total numberof desired molecules that can be deposited on the surface isseverely limited for a single spot. Lipid and protein solutionsin particular are difficult to purify, making the cost of solutionsrequired for conventional spotting systems prohibitively high.Protein spots produced by either pin- or drop-spotters aresusceptible to drying, causing the proteins to denature andrendering them useless. Additionally, lipid solutions are notcompatible with conventional pin and ink jet spotters, as thehydrophobicity of the solutions prevents liquid adhesion to thespotter tip or ejection from a spray head. Finally, the majorityof waste from conventional spotting systems will occur duringthe washing process following spotting, where any unboundmolecules are rinsed from the array substrate. These moleculescannot be recovered without repurification, which makesconventional spotting highly inefficient in terms of materialcosts, especially when only a small amount of spotting solutionis available. Recycling unbound molecules is possible usinga flow deposition system, which can produce high surfacedeposition density if the substrate is tailored to bond only tothe desired molecules, allowing the unwanted bulk materialto be washed off and added back into the bulk depositionsolution [6]–[8]. However, flow deposition systems generallyare not capable of producing large-scale, low-background spotarrays, let alone individually addressed arrays. Additionally,if sequential chemical processing is required for individualspots, or if layer-by-layer self-assembly (LBL) is necessary tobuild up the spot concentration, only a continuous flow spottersystem is capable of fabricating these types of arrays.The spotter system described in this study uses microfluidicchannel networks embedded in polydimethyl siloxane (PDMS),combinedwithauniquesubstrateinterfacetoflowdepositionso-lutionsoveralithographicallydefinedsubstratearea.Toachieveindividual spot addressing, each spot interface is linked to flu-idic ports by a pair of microchannels, allowing fluid to circulatepast the spotted area (Fig. 1). Since the fluid channel system ispatternedlithographically,anunlimitednumberofchannelsandspots can be produced. The channels all lie in a single plane,forming a single line of spots. To produce a two-dimensional(2-D)array,multiplelayersofmicropatternedPDMSarestackedand bonded together, forming a rectangular array.The lithographic fabrication process for the PDMS channeland interface system also allows unique techniques to be ap-plied to the flow deposition (Fig. 2). Flow conditioners suchas laminar mixing structures can be used to customize the so-lution flow characteristics, enhancing deposition. Additionally, 1057-7157/$20.00 © 2006 IEEE  1146 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006 Fig. 1. PDMS continuous  fl ow spotter operation.Fig. 2. Spotter tip variations. multiple port injection to the spotted area opens the possibilityof surface con fi ned chemical reactions, and even surface poly-merization reactions at speci fi c points. The out-of-plane natureof the spotter would allow very high aspect ratio structures tobe constructed and attached to the surface. Fabrication of suchmicroscale polymeric structures such as prisms is dif  fi cult withconventional micromachining, but would be relatively simplewith the spotter system, creating an innovative approach to mi-croscale optical components.II. M ETHODOLOGY  A. Fabrication Procedure1) Overview:  The fabrication process for the PDMS spottercan be subdivided into  fi ve main steps: SU-8 mold fabricationand preparation, PDMS casting and curing,  fl uidic port coring,channel sealing with PDMS slab, and spotter face cuttingwith a razor edge. The entire fabrication process including theSU-8 mold formation takes approximately 10 h to complete.Providing the user is careful, the mold can be reused to makean unlimited number of castings. 2) SU-8 Mold Fabrication and Preparation:  Since the SU-8mold was photolithographically constructed, an emulsion mask was fabricated prior to the mold using a high resolution printer(lithopointe) and was used as-is for the mold manufacturingprocess. The  fl uid channels were laid out on the mask in pairs,with one end of each channel leading to an exit port, and theother end joining its pairing channel at the spotting port. All of the channels were 100 m wide and the spotting port width atthe intersection was also 100 m wide. Variations could easilybe made at or close to the spotting port, such as constrictionsand laminar mixers, simply by altering the mask design. For the Fig. 3. SU-8 mold surface modi fi cation. primary test apparatus, four spots were arranged in a line, withthe spotting ports separated by 500- m gaps. The other ends of thechannelsleadingtotheexitportswerespacedapartby5mmfor easy packaging. Molds for a 48-spot system were fabricatedin the same way as the four-spot system, but molds for four sep-arate linear arrays of 12 spots were created.A 3-in single-side polished silicon wafer was used as the sub-strate for the SU-8 molds. A wafer was preheated for 10 minat 95 C to drive off the water from the surface and improveadhesion. Once the wafer had cooled, SU-8 50 (Microchem)was spun on at 1300 rpm for 60 s to produce a 100- m-thick layer. The wafer was soft-baked at 65 C for 3 min and 95 Cfor 2 h to drive off as much of the photoresist solvent as pos-sible. This long soft bake time was absolutely essential to pre-vent the formation of high stress levels in the 1:1 aspect ratiochannelstructurethatwouldbuckleunderhighstress.Followingthe soft baking process the wafer was cooled in preparation forexposure.Exposure of the SU-8 on the wafer was carried out using a365-nm light source aligner (EVG), but the exposure processhad to be altered to allow the use of the emulsion mask. Themask was laid directly on the wafer in the approximate centerwith the emulsion side facing the SU-8 and covered with a 4-inglass plate. The wafer was then inserted into the aligner and ex-posedwitha430mJ/cm dosage.Post-exposurebakingwascar-ried out for 3 min at 65 C for 3 min and 95 C for 15 min tocomplete the crosslinking of the exposed resist. The wafer wasimmersiondevelopedinpropyleneglycolmonomethyletherac-etate (PGMEA) (Microchem) for 20 min, washed in isopropylalcohol and dried with a nitrogen spray. 3) SU-8 Mold Surface Preparation:  Silicon wafers nor-mally have a thin native oxide layer on their surface to whichPDMS will bond strongly, preventing the casting from re-leasing from the mold. To prevent this unwanted bonding, the fl uorosilanizing agent (trideca fl uro-1,1,2,2-tetrahydrooctyl)triethoxysilane (Gelest) was used to coat the native oxide witha  fl uorocarbon layer (Fig. 3). The  fl uorosilane was evaporatedin a vacuum chamber containing the wafer for 2 h, allowinga surface reaction to occur at a controlled rate and form amonomolecular surface layer. The silane group binds prefer-entially to the oxide layer, leaving the  fl uorocarbon residuesticking up from the wafer surface, preventing the PDMS from  CHANG-YEN  et al. : NOVEL PDMS MICROFLUIDIC SPOTTER FOR FABRICATION OF PROTEIN CHIPS AND MICROARRAYS 1147 Fig. 4. PDMS  fl uidic port coring with modi fi ed 20-gauge needle. bonding. A blank 3-in wafer was also coated along with theSU-8 mold to provide a mold for the channel cover slab. 4) PDMS Casting and Curing:  PDMS (Sylgard 184) wasused as-is from the supplier (DowCorning), and was used asdirected. 40 mL of the base resin was mixed with the curingagent in a 10:1 ratio by volume and mixed thoroughly. Theprepolymer mixture was placed in a vacuum for 1 h to removeall air bubbles and then split into two equal parts for each of themolds.Theprepolymerwaspouredovereachwaferandallowedto settle evenly. The wafers were then placed in a vacuum for1 h to remove any air bubbles trapped between the mold andprepolymer. Once all air had been evacuated from the molds,theywereplacedinanovenat65 Cfor2htocure.Immediatelyafter the cure was complete, the castings were peeled from themold, washed in isopropyl alcohol and dried with a nitrogenspray. 5) Fluidic Port Coring:  The PDMS cover slab was placedin a sealed container to prevent dust contamination during theport coring process. The ports were cut in the PDMS channelslab from the channel side of the casting, making alignmentof the holes with the channels simple. The coring process wasperformed using a  fl at-tipped 20-gauge syringe needle that hadbeen modi fi ed at the tip to form a sharp, beveled cutting edge[9]. This edge allowed the coring tool to make a clean cut intothe PDMS, forming a cylindrical hole from the channel face of the slab to the outer face of approximately the same diameter asthe internal bore of the needle (0.023 in). To connect existing fl uidic systems to the PDMS, an unmodi fi ed 20-gauge needlewas inserted into the hole, and the appropriate luer connectionsmade to the needle (Fig. 4). The seal between the needle andthe PDMS is purely mechanical, caused by compression of thesmaller hole diameter (0.023 in) around the larger outer diam-eter of the needle (0.036 in). This  fl uidic connection has provedto be extremely robust, and is capable of withstanding severemechanical shock and handling, as well as multiple insertionsand removals of the needle. Prior to channel sealing, the needleconnections were removed from the PDMS slabs to allow themto sit level in the oxygen plasma chamber for surface treatment. 6) Channel Sealing and Spotter Face Cutting:  To bond thePDMS channels to the blank cover slab, a 73 W oxygen plasmaat 273 mTorr for 20 s was used to activate the sealing surfaces.Immediately following the plasma treatment, the treated PDMSsurfaces were placed together, forming an instant bond between Fig. 5. Cutting of spotter face with razor edge and packaging with 20-gaugesyringe needles. the channel PDMS and the cover slab. Bonding was completedby baking the PDMS in the oven at 65 C for 2 h, creating ahermetic seal. The excess PDMS was trimmed from around thedevice, and the spotting face cut with a razor edge (Fig. 5). Un-modi fi ed  fl at-tipped 20-gauge needles were then inserted intothe cored ports to package and complete the device. For the48-spot system, a similar bonding procedure was used betweeneach layer of the arrayer with four total bonding steps required.  B. Spotter Operation1) General Operation:  Operation of the PDMS spotter re-quires that the spotted surface be relatively clean and smoothto allow the spotter face to form a  fl uid seal. The spotter facemust then be pressed onto the required area and held for the du-ration of the  fl uid  fl ow. Each channel pair is connected at the fl uid connection port to a  fl uid input and output line. Fresh orrecirculated  fl uid is pumped into the  fl uid inlet and waste/ex-cess  fl uid is simultaneously pumped out, as shown in Fig. 1. Inhigher fl ow-rate depositions, infusing and withdrawing the fl uidfrom the spotter will prevent leakage at the spotter face that canoccurif onlyinfusion isused.Forspot depositionssuchasDNAwhich require a surface incubation period, the spotter can beused in a noncontinuous fashion, whereby probe solution is in- jected to the substrate, allowed to incubate, refreshed, and thecycle repeated to build up the probe surface concentration. Byexternallyswitching the depositionsolutions, multiple layering,and washings on each of the spotted areas can be performed.Additionally, by initially loading a small volume of solution tothe spotting tip, then cycling the  fl ow direction, a much smallervolume of solution can be used. This technique would be par-ticularly useful for small-volume, precious solutions. Regard-less of the solution volume or  fl ow cycling used, by applying avacuum to the outlet port of the channel, all of the unused solu-tion can be withdrawn, collected, and reused. Since the spottedarea is completely isolated from the surrounding environment,spotcrosstalk,backgroundnoise,anduncontrolleddryingofthespots is virtually eliminated.The spotter may be used for fl uid loading into other micro fl u-idic systems, simply by pressing the spotter face against a sur-face port array. Surface modi fi cation of the internal walls of the fl ow channels can be performed easily using solutions such asbovine – serum – albumin (BSA), which coat the walls with pro-tein and reduce build up of materials on the channel walls andthe associated loss of material, as well as reduce contaminationto successive solutions during reuse. However, the design and  1148 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006 Fig. 6. Four spot system, showing magni fi ed view of tip face and completedsystem with connecting  fl uidics. Each of the four spots measure 725    m 2   100   m. manufacture of the spotter system makes the system tenable todisposability, eliminating contamination to successive deposi-tion solutions.The  fi rst version of the spotter system was fabricatedas a linear array with four 100 m 725 m spots[Fig. 6(a) and (b)]. Each spot was individually addressableusing separate  fl uid channels and ports. Initial leakage testingwas performed using this system, with no detectable leakageacross the spotter face or between the separate  fl uidic channels.Ultimately, these spots proved to be too small for initial testingwith a standard SPR system, so a larger variant of the systemwas designed and fabricated. 2) DepositedProbeSurfaceConcentrationComparison:  Forprobe surface concentration testing, a larger scale single-portspotter was manufactured by casting PDMS channels arounda copper wire mold. The larger sized spot produced was com-patible with available fl uorescence test apparatus. The spot areawas de fi ned by an inserting the end of the mold wires into a2-mm by 2-mm cube of PDMS. Initially, a  fl uorescent dye wasused as a test molecule for deposition, as the level of depositionis directly proportional to the measured  fl uorescence intensity.A bulk solution of 40 mL of a solution of the  fl uorescent dyetris (2,2 ’ -bipyridyl dichlororuthenium) hexahydrate (Ru(bpy))at 2 g/L concentration was recirculated for 60 min over a glass Fig. 7. Macroscale spotter performing dye deposition on a glass substrate. slide at 2 mL/h to allow deposition to occur (Fig. 7). Althougha large volume of solution was used for continuous- fl ow spotterdeposition (approximately 2 mL total from the bulk 40 mL), thespotter was run continuously in one direction, with the used dyesolution being deposited back into the bulk solution container.Recirculating the deposition solution back through the 40-mLbulk container avoided depletion of dye molecules, and allowedrecapture of any undeposited material. This process also effec-tively allowed any of the molecules in the 40-mL bulk to accessthe surface of the spot. For low-volume, precious solutions, re-circulation can be achieved by cycling the direction of the so-lution fl ow, combined with integrated fl uidic holding chambers,reducing the required volume to the order of the channel systemvolume. For the 100 m by 725 m system, this volume is lessthan 4 L. Additionally, whereas drop or pin-spotter systemsdeplete the spotting solutions, the continuous- fl ow spotter re-captures all the unused solutions, making the system extremelycost-effective for large-scale array manufacture.To simulate current spotting techniques, 3 L of the samesolution was dropped onto a glass slide with a micropipette(VWR Scienti fi c) and dried to form a spot of the same area asthe spotter-deposited region. Calculations based on the spottedarea and average dye molecule diameter predicted that thespotter-deposited area would theoretically produce maximumdye surface coverage in hexagonal close-pack con fi guration,while the dropped-deposited area would cover three orders of magnitude less. Comparison of the dye surface density wasbased on observed  fl uorescence intensity and both spottedareas were compared to a blank control slide. The spectrometer(Oceanoptics) was set to a 5-s integration period, averaging100 scans using a 20-point boxcar average (Fig. 9). The peak  fl uorescence was observed at 614.26 nm, and two separatesets of spots were compared for consistency. To quantitativelydetermine the increase in deposition density by the spottersystem, the peak   fl uorescence values were extracted from eachscan and averaged. Standard deviations for these averages werealso calculated to quantify the signal-to-noise ratio. Once thespotter had demonstrated success in the initial test phase, depo-sition of proteins was tested. Protein A (ImmunoPure Protein  CHANG-YEN  et al. : NOVEL PDMS MICROFLUIDIC SPOTTER FOR FABRICATION OF PROTEIN CHIPS AND MICROARRAYS 1149 Fig. 8. Two-dimensional (2-D) array continuous  fl ow spotter face. Each of the48 400    m 2   400    m spots are individually addressed by a pair of microchan-nels. A, Catalog No. 21181, Pierce Inc.) was biotinylated Biotin(EZ-Link Sulfo-NHS-Biotin, Catalog No. 21217, Pierce Inc.)to provide speci fi c adhesion to a surface plasmon resonance(SPR) streptavidin gold chip (8500 streptavidin af  fi nity chip,Part No. 4346388, AB). The protein solution was diluted to aconcentration of 0.15 g/mL in 0.1X PBS buffer (0.19 mMNaH PO , 0.81 mM Na HPO , pH 7.4 and 15 mM NaCl) andsupplemented with 100 g/mL BSA to prevent nonspeci fi cadhesion. To recirculate the solution over the chip surface,200 L of protein A solution was loaded into a PhynexusMicroExtractor 100 syringe pump and  fl owed continuouslyback and forth through the spotter at 75 L/min for 1 h, in acon fi guration similar to Fig. 7. A wash step was then performedusing 800 L of 0.1X PBS with 100 g/mL BSA. At the end,the sample was removed from the surface by withdrawing airthrough the assembly, and the chip was washed with water.To compare the results of the continuous- fl ow immobi-lization, protein A was also immobilized on the same chipusing solid-pin spotting. Samples at the same concentration(0.15 g/mL) as the ones used for the continuous- fl ow deliverytest were spotted across the chip. Binding to the two sets of spots allowed a comparison of the sensitivities of the immo-bilization methods. Solid-pin spotting was carried out using aGenetix QArray Mini spotter with the same solution and washcycle as the continuous  fl ow spotter. A series of increasingprotein concentrations were deposited using the pin spotterto create a calibration curve of SPR response to depositedprotein A concentrations. This curve was used to calculate anequivalent concentration of the PDMS spotter, to determine thefactor increase in deposition density.Preliminary testing of the 2-D array continuous- fl ow mi-crospotter was also performed, using both a single active layerconsisting of 12 linear spots, as well with a 48-spot systemconsisting of four 12-spot layers (Fig. 8). Alignment of theseparate layers was achieved visually using marks within eachchannel layer. The single layer 12-spot system was fabricatedseparately using adjusted channel widths to produce equal  fl owrates (and consequently uniform deposition) over each spot.The tests were carried out using protein A at concentrations of  Fig. 9. Fluorescence test apparatus. 0.22 and 0.4 mg/mL for the 12-spot linear array and 48-spot2-D array, respectively, with deposition being performed ongold af  fi nity-binding SPR chips. To image the arrays, IgGwas  fl owed over the hydrated spots. Although imaging of thehydrated, IgG-bound linear 12-spot array was performed, the2-D array was only imaged prior to hydration and IgG binding.All protein deposition and testing was carried out by the Pro-tein Interaction Facility, Health Sciences Center Core Labora-tories, University of Utah, Salt Lake City.III. R ESULTS  A. Spotter Fabrication Completed continuous  fl ow spotting systems are shownin Figs. 6 – 8. Fig. 6 shows a four-spot system with complete packaging and a close up of the spotting face. Note the smoothface allowing for good sealing between spots on the substrate,which prevents any crosstalk between spots. Fig. 7 showsthe large spotter in operation and depositing dye on a glasssubstrate. Fig. 8 shows the face of the 48-spot system. Note thatthe spacing between layers is controlled by the thickness of thePDMS, which in this case was not perfectly controlled. Notealso that the alignment between layers is good, but not perfect.Three of the layers are aligned within a few microns, but thefourth layer is about 200 m off in alignment. A better align-ment jig is expected to help with this issue. The spotter systemsdemonstrated no leaking between channels and withstood allpressures used in the characterization experiments.  B. Dye Immobilization The averaged  fl uorescence intensity data shows a signi fi cantincreasein fl uorescenceintensityfromthespotterdropsascom-pared to the pipetted and controlled drops. The pipetted dropsdid not display signi fi cant fl uorescence intensity, demonstratingthat this form of surface deposition cannot produce signi fi cantor useful spots at very low dye concentrations (Fig. 10). Thepipetted drop and control drops show almost the same  fl uores-cence level, most likely caused by background light, and al-though the pipetted drops demonstrated a slightly lower signalthan the control drops, the difference can easily be accountedfor by the high level of noise (indicated by the error bars). De-spite the relatively large standard deviation for all of the av-eraged data, the PDMS spotter conclusively produced an in-creased deposition density. Directly comparing peak intensityvalues yields a minimum 300% increase by the PDMS spottercompared to the pipetted drop. This simple approach underes-timates the power of the spotter though. When the control spot
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks