Microfluidic Polyacrylamide Gel Electrophoresiswith in Situ Immunoblotting for Native ProteinAnalysis

Mei He and Amy E. Herr*

Department of Bioengineering, University of California, Berkeley, California 94720

We introduce an automated immunoblotting method thatreports protein electrophoretic mobility and identity in asingle streamlined microfluidic assay. Native polyacryla-mide gel electrophoresis (PAGE) was integrated withsubsequent in situ immunoblotting. Integration of threePA gel elements into a glass microfluidic chip achievedmultiple functions, including (1) rapid protein separationvia on-chip PAGE, (2) directed electrophoretic transferof resolved protein peaks to an in-line blotting membrane,and (3) high-efficiency identification of the transferredproteins using antibody-functionalized blotting mem-branes. In-chip blotting membranes were photopatternedwith biotinylated antibody using streptavidin polyacryla-mide (PA) thus yielding postseparation sample analysis.No pressure driven flow or fluid valving was required, asthe assay was operated by electrokinetically programmedcontrol. A model sample of fluorescently labeled BSA(negative control), r-actinin, and prostate specific antigen(PSA) was selected to develop and characterize the assay.A 5 min assay time was required without operatorintervention. Optimization of the blotting membrane (ge-ometry, operation, and composition) yielded a detectionlimit of ∼0.05 pg (r-actinin peak). An important ad-ditional blotting fabrication strategy was developed andcharacterized to allow vanishingly small antibody con-sumption (∼1 µg), as well as end-user customization ofthe blotting membrane after device fabrication and stor-age. This first report of rapid on-chip protein PAGEintegrated with in situ immunoblotting forms the basis fora sensitive, automated approach applicable to numerousforms of immunoblotting.

Immunoaffinity methods are essential to clinical diagnostics,as well as fundamental life science investigations. Among immu-noaffinity methods, both immunoassays and immunoblottingapproaches are workhorse techniques used for decades to detectspecific analytes in complex biological fluids.1-4 Whereas immu-

noassays report the presence of protein using immunorecognition,immunoblotting reports both apparent mobility (i.e., protein size,charge-to-mass ratio) and antigen-antibody interaction. Immu-noblotting comprises a suite of assays including: Western and FarWestern blotting for protein presence and protein-protein interac-tion, Northern blotting for RNA, and Southern blotting for DNA.To obtain both mobility and immunorecognition characteristicsfor proteins, slab polyacrylamide gel electrophoresis (PAGE) istypically coupled with “blotting” of separated proteins on amembrane. Subsequently, blotted proteins are probed withreporter antibodies and detected via various stains includingchemiluminescence.5 Information about the mobility of the proteinfrom PAGE and the specificity of the antibody-antigen interactionenables a target protein to be identified in the midst of a complexprotein mixture. Immunoblotting of proteins is routinely used inbasic biological research studies, as well as for confirmatoryclinical diagnostic assays (i.e., HIV and Lyme disease). While apowerful assay platform, one major disadvantage of conventionalimmunoblotting (i.e., slab-gel PAGE and subsequent manual“blotting”) is the requirement for tedious labor-intensive and time-intensive manual intervention at various steps of the assay.6

Surprisingly, immunoblotting formats have changed little sincefirst introduction by Towbin in 1979.7

Recently, automation of immunoblotting has been reportedusing a capillary, not slab-gel, format.8,9 Capillary isoelectricfocusing (IEF) was used to separate proteins, including proteinisoforms.10 The capillary IEF separation was followed by UVsurface-immobilization of resolved proteins on a photoactivatablecapillary surface. Subsequent detection was accomplished byflushing out all nonsurface cross-linked materials and flushing in

* To whom correspondence should be addressed. Phone: (510) 666-3396. Fax:(510) 642-5835. E-mail: aeh@berkeley.edu.

(1) Gershoni, J. M.; Palade, G. E. Anal. Biochem. 1983, 131, 1–15.(2) Anderson, D. J.; Lente, F. V.; Apple, F. S.; Kazmierczak, S. C.; Lott, J. A.;

Gupta, M. K.; McBride, N.; Katzin, W. E.; Scott, R. E.; Toffaletti, J.;Menendez-Botet, C. J.; Schwartz, M. K.; Castellani, W. J.; Hage, D. S.; Allen,R. C.; Griffiths, J. C.; Hepler, B. R.; Touchstone, J. C.; Skogerboe, K. J.;Wang, J.; Kuesel, A. C.; Kroft, T.; Smith, I. C. P.; Haas, R. G.; Chou, D.Anal. Chem. 1991, 63, 165R–270R.

(3) Tissot, J. D.; Vu, D. H.; Aubert, V.; Schneider, P.; Vuadens, F.; Crettaz, D.;Duchosal, M. A. Proteomics 2002, 2, 813–824.

(4) Righetti, P. G.; Castagna, A.; Antonucci, F.; Piubelli, C.; Cecconi, D.;Campostrini, N.; Rustichelli, C.; Antonioli, P.; Zanusso, G.; Monaco, S.;Lomas, L.; Boschetti, E. Clin. Chim. Acta 2005, 357, 123–139.

(5) Sun, L.; Ghosh, I.; Barchevsky, T.; Kochinyan, S.; Xu, M. Q. Methods 2007,42, 220–226.

(6) Wu, Y.; Li, Q.; Chen, X. Z. Nat. Protoc. 2007, 2, 3278–3284.(7) Towbin, H.; Staehelin, T.; Gordon, J. Proc. Natl. Acad. Sci. U.S.A. 1979,

76, 4350–4354.(8) Guzman, N. A.; Park, S. S.; Schaufelberger, D.; Hernandez, L.; Paez, X.;

Rada, P.; Tomlinson, A. J.; Naylor, S. J. Chromatogr., B 1997, 697, 37–66.(9) Peoples, M. C.; Phillips, T. M.; Karnes, H. T. J. Pharm. Biomed. 2008, 48,

376–382.(10) ONeill, R. A.; Bhamidipati, A.; Bi, X.; Deb-Basu, A.; Cahill, L.; Ferrante, J.;

Gentalen, E.; Glazer, M.; Gossett, J.; Hacker, K.; Kirby, C.; Knittle, J.; Loder,R.; Mastroieni, C.; MacLaren, M.; Mills, T.; Nguyen, U.; Parker, N.; Rice,A.; Roach, D.; Suich, D.; Voehringer, D.; Voss, K.; Yang, J.; Yang, T.; Horn,P. B. V. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 16153–16158.

Anal. Chem. 2009, 81, 8177–8184

10.1021/ac901392u CCC: $40.75 2009 American Chemical Society 8177Analytical Chemistry, Vol. 81, No. 19, October 1, 2009Published on Web 09/04/2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

reagents for antibody-based identification with chemilumi-nescence.11,12 A robotic interface was required to integrate fluidexchange steps with the capillary tube geometry (i.e., single inlet,single outlet) used. One difficulty in integration of multiassay stepsin capillary systems arises from a need for external fixturing tomultiple reservoirs and interfaces. In a like-minded effort toadvance the ease and speed of immunoblotting, we note thatunified, integrated design is a hallmark of microsystems. Integra-tion of multiple functions in a single microfluidic system substan-tially benefits automation of multistage assays through facile on-chip interfacing.13 Dexterity in sample manipulation and the abilityto fabricate interconnecting channels with zero dead volumecontribute to high performance and low sample loss. Additionalcompelling advantages of microfluidic systems include reducedmanual intervention, nominal sample consumption, rapid results,improved assay precision, and digital data collection essential forarchiving and comparison.14-16 Coupled with automated fluidmanipulation, the aforementioned factors make microfluidicdevices exceptionally well-suited formats for multistep analyses.17-19

An area of notable recent growth in reports of multistep assaysincludes on-chip integration of sample handling and pretreatmentwith electrophoretic protein analysis, an area relevant to the studyreported here. Examples of the “upstream” integration of functionwith assays include sample concentrators,20 mixers,21 and mi-croreactors.22 These versatile preparatory functions can greatlyenhance the final assay performance and yield automated, mul-tifunctional protein measurement tools.23-27 In this article, weintroduce a new approach to immunoblotting that takesadvantage of microfluidic technology to simplify integration and,hence, multistep assay operation. Further, and most impor-tantly, while tremendous enhanced functionality has beengained from “upstream” sample preparation using integratedpolymer features, we focus here on enhanced postseparation

or “downstream” functionality enabled through functionalpolymer features. Here we integrate sample analysis (PAGE)with postseparation affinity-based protein blotting in an effortto develop a “hands-free” microanalytical immunoblot.

The reported PAGE in situ immunoblotting technology relieson multiple, functionalized polyacrylamide (PA) gels photopat-terned in specific regions of a planar, glass microfluidic device.This assay integrates rapid, high-resolution native PAGE withantibody-functionalized PA blotting membranes to yield in situreporting of protein mobility and antibody-affinity. After PAGE,electrophoretic transfer of resolved species to a blotting membraneis demonstrated as a directed, efficient method for proteinidentification, without a need for pressure-driven flow and valving.Using the PAGE in situ immunoblot, we demonstrate limitedconsumption of expensive detection reagents (i.e., antibodies) andprecious starting samples, as expected from microfluidic formats.14

Postseparation blotting takes advantage of the high surface area-to-volume ratio and fast mass transport available in said geom-etries, which leads to a significant decrease in analysis time (fromhours to seconds).15 The platform technology introduced formsa promising basis for our development of quantitative, automatedmicrofluidic immunoblotting.

MATERIALS AND METHODSReagents. The water-soluble photoinitiator 2,2-azobis[2-methyl-

N-(2-hydroxyethyl) propionamide] (VA-086) was purchased fromWako Chemicals (Richmond, VA). 3-(Trimethoxysilyl)-propylmethacrylate (98%), perchloric acid (70%, ACS grade), hydrogenperoxide (30%, ACS grade), glacial acetic acid (ACS grade),methanol (ACS grade), and 30% (29:1) acrylamide/bis-acrylamidewere purchased from Sigma. Streptavidin-acrylamide (SA) waspurchased from Invitrogen (Carlsbad, CA). Premixed 10× Tris-glycine native electrophoresis buffer (25 mM Tris, pH 8.3, 192mM glycine) was purchased from Bio-Rad (Hercules, CA). AlexaFluor 488 conjugated bovine serum albumin (BSA) and FITC-biotin were used as negative and positive controls, respectively(Sigma Aldrich, St. Louis, MO). R-Actinin and biotin conjugatedanti-actinin were purchased from Cytoskeleton, Inc. (Denver, CO).Free prostate specific antigen (PSA) and biotinylated monoclonalanti-PSA were purchased from EXBIO (Praha, Czech Republic).The proteins R-actinin and PSA were fluorescently labeled in-houseusing Alexa Fluor 488 protein labeling kits per the supplier’sinstructions (Life Technologies, Carlsbad, CA) and purified byP-6 Bio-Gel columns (Bio-Rad, Hercules, CA). Labeled proteinswere stored at 4 °C in the dark until use.

Chip Fabrication. Glass microfluidic chips were designed in-house and fabricated using standard wet etch processes by CaliperLife Sciences (Hopkinton, MA). The sample (S), sample waste(SW), buffer (B, B1, B2), and buffer waste (BW) reservoirs areindicated in Figure 1a. Chip layouts consisted of a double-Tjunction having a 2.5 mm separation channel connected to asecond junction needed for patterning of the downstream blottingmembranes. Channels were ∼15 µm deep and ∼80 µm wide. Aninitial sample volume of 5 µL was employed. Channel surfaceswere first functionalized for covalent linkage to PA gel using a2:3:2:3 ratio mixture of 3-(trimethoxysilyl) propyl methacrylate,glacial acetic acid, deionized water, and methanol. After a 20 minstatic incubation, methanol was flushed through all channels for30 min followed by a drying nitrogen purge.

(11) Knittle, J. E.; Roach, D.; Horn, P. B. V.; Voss, K. O. Anal. Chem. 2007, 79,9478–9483.

(12) Fan, A. C.; Deb-Basu, D.; Orban, M. W.; Gotlib, J. R.; Natkunam, Y.; ONeill,R.; Padua, R. A.; Xu, L.; Taketa, D.; Shirer, A. E.; Beer, S.; Yee, A. X.;Voehringer, D. W.; Felsher, D. W. Nat. Med. 2009, 15, 566–571.

(13) Hou, C.; Herr, A. E. Electrophoresis 2008, 29, 3306–3319.(14) Mauk, M. G.; Ziober, B. L.; Chen, Z.; Thompson, J. A.; Bau, H. H. Ann.

N.Y. Acad. Sci. 2007, 1098, 467–475.(15) West, J.; Becker, M.; Tombrink, S.; Manz, A. Anal. Chem. 2008, 80, 4403–

4419.(16) Peoples, M. C.; Karnes, H. T. J. Chromatogr., B 2008, 866, 14–25.(17) Yeung, S. H.; Liu, P.; Bueno, N. D.; Greenspoon, S. A.; Mathies, R. A. Anal.

Chem. 2009, 81, 210–217.(18) Meagher, R. J.; Hatch, A. V.; Renzi, R. F.; Singh, A. K. Lab Chip 2008, 8,

2046–2053.(19) Fan, R.; Vermesh, O.; Srivastava, A.; Yen, B. K. H.; Qin, L.; Ahmad, H.;

Kwong, G. A.; Liu, C. C.; Gould, J.; Hood, L.; Heath, J. R. Nat. Biotechnol.2008, 26, 1373–1378.

(20) Hatch, A. V.; Herr, A. E.; Throckmorton, D. J.; Brennan, J. S.; Singh, A. K.Anal. Chem. 2006, 78, 4976–4984.

(21) Johnson, T. J.; Ross, D.; Locascio, L. E. Anal. Chem. 2002, 74, 45–51.(22) Kawabata, T.; Wada, H. G.; Watanabe, M.; Satomura, S. Electrophoresis

2008, 29, 1399–1406.(23) Phillips, T. M.; Wellner, E. F. Electrophoresis 2007, 28, 3041–3048.(24) Srivastava, N.; Brennan, J. S.; Renzi, R. F.; Wu, M.; Branda, S. S.; Singh,

A. K.; Herr, A. E. Anal. Chem. 2009, 81, 3261–3269.(25) Herr, A. E.; Hatch, A. V.; Throckmorton, D. J.; Tran, H. M.; Brennan, J. S.;

Giannobile, W. V.; Singh, A. K. Proc. Natl. Acad. Sci. U.S.A. 2007, 104,5268–5273.

(26) Hu, G.; Gao, Y.; Sherman, P. M.; Li, D. Microfluid. Nanofluid. 2005, 1,346–355.

(27) Clark, M. A.; Sousa, K. M.; Jennings, C.; MacDougald, O. A.; Kennedy,R. T. Anal. Chem. 2009, 81, 2350–2356.

8178 Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

Fabrication of Blotting Membranes. Chrome mask-basedlithography via a UV objective (UPLANS-APO 4×, Olympus)microscope system (IX-70, Olympus, Melville, NY) was used tofabricate the blotting membranes. A mercury bulb was used asthe excitation source (330-375 nm) and mask alignment to thechip was performed using a manual adjust x-y translation stage.28

For photopatterning, a 5 min UV excitation was used to form theblotting membranes (8% T, including SA).

To yield immunoaffinity-based protein detection on the blottingmembrane, SA covalently linked to the PA gel matrix was usedto immobilize biotinylated antibodies. As illustrated in Figure 1b,antibody immobilization in the PA blotting membranes wasconducted by one of two methods, a prepatterning strategy or acustom patterning strategy. The single-step prepatterning strategyconsisted of a concurrent photopatterning and antibody im-mobilization step that employs PA gel precursor solutions thatcontain both SA and biotinylated antibody. The binding capacityof the streptavidin blotting membrane was demonstrated byintroducing FITC-biotin (Figure 1a, inset). The alternate, two-stepcustom patterning strategy consisted of (i) photopatterning of agel precursor solution containing SA, followed by (ii) antibody

immobilization that utilized a slow electrophoretic introductionof biotinylated antibody through the streptavidin-decorated PA gelmembrane (i.e., 10 V/cm electrophoresis from well B2, Figure1a). The custom patterning strategy resulted in the immobilizationof biotinylated antibodies after fabrication and storage, makingcustomization possible by the end-user. Characterization of thecustom patterning strategy for a range of patterning timessuggests that exposed streptavidin binding sites in the blottingmembrane are sufficiently occupied after a 10 min antibodypatterning step (10 V/cm, 1 µM anti-actinin).

Fabrication of Loading and Separation Gels. After pattern-ing of the blotting membrane, all unpolymerized open channelswere flushed with buffer for 1 min using vacuum in preparationfor fabrication of the loading and separation gels. The chip wasthen soaked in a buffer solution to allow diffusive dilution ofunpolymerized precursor between the blotting membrane andreservoir BW, although complete removal of the unpolymerizedprecursor was not necessary. PA gels of various acrylamideconcentrations were photopatterned in the microdevices in amanner similar to our previous reports.20,29,30 Briefly, degassedPA gel precursor solution for the separation gel (8% T) was wickedor gently pressure-filled (via syringe) into the channels. After allchannels were loaded, high viscosity 5% 2-hydroxyethyl cellulose(HEC, Sigma, average MW ∼720 000) drops were gently appliedonto each fluid-containing reservoir. The 5% HEC solution quicklyeliminated hydrostatic flow resulting in quiescent fluid conditionsinside the channels, as is needed for high-resolution photopolym-erization (1 min photopatterning via the UV objective setup). Afterpolymerization of the separation channel, the HEC solution oneach reservoir was flushed off by buffer solution. Then, a largerpore-size sample loading gel was formed using 3% T acrylamidesolution and an 8 min flood exposure of the chip to a filteredmercury lamp (300-380 nm) located 15 cm away (100 W, UVPB100-AP, Upland, CA) with cooling fan. The photopolymerizationtimes reported were determined empirically based on the intensityof each UV light source, composition of acrylamide precursorsolution, and desired pore-size to achieve optimal gel performancefor the desired function. PA gels were visually inspected andtypically showed a well-defined, uniform blotting structure withsharp interfaces and channels free of bubbles and voids (Figure1a, inset).

Chip Reuse. After use, the glass chips were routinelyregenerated through removal of the cross-linked PA gels that formthe loading and separation gels, as well as the blotting membranes.An empirically determined regeneration protocol consisted ofsoaking used chips (containing PA gels) in a 2:1 ratio solution ofperchloric acid and hydrogen peroxide at 75 °C overnight. Aftersoaking, channels were flushed with 0.1 M sodium hydroxide for30 min. Visual inspection showed negligible residual PA gel insidethe channels after treatment. Subsequent fabrication of cross-linked PA gels in the recycled glass chips yielded >90% success,which is on par with multistep photopatterning device yields innew glass chips.

Apparatus and Imaging. Electrophoretic transport was usedto mobilize species through the gel elements for all steps of the

(28) Das, C.; Zhang, J.; Denslow, N. D.; Fan, Z. H. Lab Chip 2007, 7, 1806–1812.

(29) Herr, A. E.; Throckmorton, D. J.; Davenport, A. A.; Singh, A. K. Anal. Chem.2005, 77, 585–590.

(30) Lo, C. T.; Throckmorton, D. J.; Singh, A. K.; Herr, A. E. Lab Chip 2008,8, 1273–1279.

Figure 1. Fabrication of microfluidic native PAGE in situ immuno-blotting device: (a) schematic of chip layout (not to scale). Fluidreservoirs are labeled according to contents: S, sample; B, buffer;SW, sample waste; BW, buffer waste. Polyacrylamide gel compositionis indicated by grayscale (% T and % C are percentage of totalacrylamide and cross-linker, respectively). The inset images show a10× view of a streptavidin functionalized blotting membrane photo-patterned within the channel. (b) Schematic depicting fabrication stepsfor blotting membrane: one-step prepatterning strategy and two-stepcustom patterning strategy. In the custom patterning strategy, loadingof the biotinylated antibody is via applied electric current (indicatedby i).

8179Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

PAGE in situ immunoblot. After sample introduction, assayoperation was programmable and controlled via a high voltagepower supply equipped with platinum electrodes (LabsmithHVS448, Livermore, CA). Samples were loaded by applying a +800V potential at the SW reservoir and grounding the S reservoir for∼2 min. During loading, both the B and BW reservoirs weregrounded to form a well-defined “pinched” injection plug.31 Forseparation, various potentials were applied at the BW reservoirwhile grounding the B reservoir with a push-back voltage atreservoir S and SW.29 To ensure minimal cross-contaminationbetween samples and reproducible runs, all channels wereelectrophoretically flushed with buffer every other run. Imageswere collected using an inverted epi-fluorescence microscope (IX-70, Olympus, Melville, NY) equipped with a 10× objective (NA0.3), filter cube optimized for GFP detection, and an x-y transla-tion stage. A 1392 × 1040 Peltier-cooled interline CCD camera(CoolSNAP HQ2, Roper Scientific, Trenton NJ) was employed tomonitor protein migration and blotting. Unless otherwise stated,the CCD exposure time was 400 ms. Electropherograms weregenerated by measuring fluorescence intensity (FL signal) in adetection region of interest (ROI) in the captured CCD image timesequences. The ROI occupied a small region of the channel (∼10µm wide and 80 µm long), and intensity values were extractedfrom the ROI using ImageJ. Nonlinear least-squares fitting of thesignal was performed using data analysis and graphing software(OriginPro 8.0, Northampton, MA).

RESULTS AND DISCUSSIONMicrofluidic PAGE in Situ Immunoblot Assay. The native

PAGE in situ immunoblot consists of three photopatterned PAgel regions in one glass device (Figure 1a). Each region isoptimized for a specific function, namely, sample loading,protein separation, transfer, and in situ immunoblotting. Theassay reports both apparent electrophoretic protein mobilityand protein identity through immunodetection. To yield thesetwo pieces of information, the assay proceeds in a electroki-netically controlled three-step sequence: (1) sample is loadedinto the chip through a large pore-size “loading” gel, (2) nativeproteins are electrophoretically injected and separated througha smaller pore-size separation gel matrix, and (3) proteins aresubsequently electrophoretically transported through an antibody-functionalized blotting membrane. During the separation step,protein species are resolved based on differential migrationthrough the gel matrix. As the PAGE demonstrated here isunder native conditions, species resolve owing to differencesin both charge-to-mass ratio and the overall size of the protein.After the PAGE separation, electrophoresis is used to transportresolved proteins to and through the contiguous blottingmembrane which has a composition similar to the separationmatrix albeit with immobilized antibodies present.

As illustrated schematically in Figure 2a, proteins lackingspecific affinity for the immobilized antibody migrate throughthe blotting membrane, while proteins having specific affinityfor the immobilized antibody can be retained on the blottingmembrane. For characterization of the blotting capture ef-ficiency, electropherograms were collected both upstream and

downstream of the blotting membrane. The specificity and theselectivity of the blotting membrane provided the capability tocapture target protein, which was demonstrated by in situimmunoblotting of a mixture of proteins integrated with nativePAGE analysis. Characterization results of biotinylated anti-actinin decorated blotting membranes (Figure 2b,c) revealappreciable resolution (R ) 7.1) between the two model species

(31) He, M.; Zeng, Y.; Sun, X.; Harrison, D. J. Electrophoresis 2008, 29, 2980–2986.

Figure 2. Operation of microfluidic native PAGE in situ immuno-blotting assay: (a) schematic depicting assay operation. A nativeprotein sample is electrophoretically separated and electrophoreti-cally transferred to a contiguous antibody-functionalized blottingmembrane, and specific target proteins are identified by interactionwith antibodies immobilized in the blotting membranes. (b) Time-sequence of micrographs show native PAGE analysis of BSA (leftpanel) and R-actinin (right panel) with subsequent in situ immu-noblotting. The blotting membrane is marked with an asterisk (*).R-Actinin was blotted to membrane (+) with negligible capture ofnegative control BSA (-). (c) Electropherograms collected fromupstream (left panel) and downstream (right panel) of the blottingmembrane in (b) show selective extraction of the target protein,R-actinin (slowest peak). Unmarked peak is attributed to animpurity. Applied electric field is 100 V/cm.

8180 Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

within a 30 s native PAGE analysis time. Electropherogramscollected upstream and downstream of the blotting membranesuggest selective blotting of R-actinin with negligible captureof BSA (negative control), as observed in Figure 2b,c. Themeasured apparent mobilities for R-actinin and BSA were 1.07× 10-4 and 1.57 × 10-4 cm2 V-1 s-1, respectively.

In the sample analysis presented here, the time requiredfor native PAGE in situ immunoblotting was less than 5 min,including sample loading ∼2 min, separation <60 s, transferand blotting ∼30 s. In addition to rapid protein electrophoresis,the microfluidic transfer of specific proteins to the blottingmembrane contributes significant time savings, as comparedto conventional membrane sheet blotting using either diffusion(4 h) or electrophoresis (0.5 h).32 Importantly, the use ofdirected electrophoretic transfer in combination with appropri-ate pore-size PA blotting membrane (8% T) results in an abilityto blot large molecular weight proteins, including R-actinin(∼100 kDa). Such large species are commonly difficult totransfer and blot using conventional immunoblotting (specifi-cally, Western blotting).32

Because of the use of electrokinetic transport and digitaldata collection, the PAGE in situ immunoblotting assay is fullyintegrated and programmable, yielding an automated assay. Theseamless integration of the two gel features (separation gel andblotting membrane) provides lossless manipulation of minutesample mass with zero dead-volume. The blotting membranesutilized here take advantage of the large surface area-to-volumeratio inherent in three-dimensional PA gel “plugs”, in contrastto methods that detect analyte by cross-linking to a channel orcapillary wall.

Blotting Membrane Selectivity. A systematic evaluation ofblotting selectivity and cross-reactivity was conducted. Severalmembrane configurations were assessed, including SA contain-ing membranes photopatterned both with and without biotiny-lated antibodies present; here biotinylated anti-actinin or anti-PSA (Figure 3). Four samples including BSA, biotin, R-actinin,and PSA were electrophoretically transported through eachblotting membrane, yielding the 4 × 3 cross-reactivity matrix.Typical capture efficiency for on-axis target detection (i.e.,

matched antibody-antigen pairs) was ∼90% (Figure 3), whichindicates significant binding compared to the off-axis capture(i.e., nonmatched antibody-antigen pairs). The results indicatelittle to no discernible cross-reactivity or nonspecific adsorptionfor the proteins and antibodies considered. For native PAGEin situ immunoblotting of more complex mixtures or differentspecies than those considered here, further assessment ofcross-reactivity of specific antibodies will be necessary, as iscommonly done in protein microarrays, ELISAs, and slab gelimmunoblotting. Our initial characterization results support theuse of biotinylated antibodies immobilized in PA gels (contain-ing SA) as a means to specifically identify antigens.

Blotting Membrane Capture Efficiency. As a metric ofblotting membrane performance, capture efficiency (capture%) was defined as

capture % )Aint - Aafter

Aint× 100% (1)

where Aint is the normalized peak area for target protein relativeto the negative control (BSA) just prior to transfer throughthe blotting membrane and Aafter is the normalized peak areafor target protein after migration through the blotting membrane.

In this study, three major factors were investigated tomaximize the capture % of the PA blotting membranes: (1) thedensity (i.e., concentration) of antibody immobilized in theblotting membrane, (2) the strength of the electric field appliedto transfer proteins through the blotting membrane, and (3)the (axial) length of the blotting membrane. Each of thesefactors contributes to determining the degree of bindingbetween migrating species and the PA blotting membrane-immobilized antibodies.

To gauge the capture % of the two antibody immobilizationstrategies (prepatterning strategy and custom patterning strat-egy), an evaluation of binding density was performed (Table1). The amount of SA in the polymer precursor solution wasvaried. Fluorescence signal from a 120 s load of ∼0.15 µMR-actinin with BSA was measured according to eq 1. Theresulting capture % increased with the amount of SA presentin the blotting membrane precursor (Table 1). These resultssuggest that the concentration of SA is a dominant factor indetermining the binding site density, which is critical to achievehigh antigen capture performance and, thus, appreciable de-tection sensitivity. Both prepatterning and custom patterningprovide acceptable capture efficiency (>80%) when the blottingmembrane precursor contains at least a 4 µM streptavidin-acrylamide (SA). For blotting membranes fabricated with 2 µMbiotinylated anti-actinin present without SA in the precursor(Table 1, strategy A), an 8.6% capture efficiency (RSD ) 4.1%,n ) 5) was observed for R-actinin. The detection of R-actinineven when no SA was present may be attributable to physicalcopolymerization of large molecular weight biotinylated anti-actinin in the cross-linked PA gel. Nevertheless, the resultsreported here suggest incorporation of SA in the PA gelprecursor solution yields specific attachment of biotinylatedantibodies as immuno-recognition sites. When no SA orbiotinylated antibody was present in the blotting membrane,no significant binding of R-actinin was observed.(32) Kurien, B. T.; Scofield, R. H. Methods 2006, 38, 283–293.

Figure 3. Microfluidic native PAGE in situ immunoblotting is ahigh specificity assay: (a) protein signal on blotting membranesallows assessment of blotting specificity. Inverted grayscale CCDfluorescence images show fluorescence of proteins bound to in-channel antibody-functionalized blotting membranes. Each analysisconsidered a single protein challenge to each blotting membraneconfiguration through a 60 s loading duration at 300 V/cm. Theprotein concentration was ∼0.2 µM in each case.

8181Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

Two important factors that impact the residence time of theanalyte in the blotting membrane include the axial length ofthe blotting membrane and the migration speed of speciesthrough the blotting membrane. The slit size on the photomask,the distance between the mask and channel surface, and theexposure duration determine the final length of the photopo-lymerized region.33 Membranes as short as 20 µm and as longas 4 mm have been fabricated using this approach (data notshown). Figure 4 shows results from characterization of twoblotting membrane geometries having different axial lengths(200 and 500 µm), both being compatible with the channellayouts used in this study. Low electric field strength operation(<100 V/cm) allowed sufficient capture of target protein on the200 µm long blotting membrane, as compared to operation atelevated electric field strengths. In addition to the increasedresidence time of migrating analytes in the blotting membraneduring the low field strength operation, enhanced retention ofthe target protein on the blotting membrane may also arisefrom reduced “stripping” of bound analyte from the antibodybinding sites; as “stripping” has been observed during electro-transfer of low molecular weight proteins to blotting mem-branes in conventional Western blotting.34 When higher electricfield strengths were desired (>100 V/cm), a longer membrane(>200 µm) was employed so as to yield sufficient capture. Ahigher % T PA gel (>8% T) can also be used to yield a blottingmembrane with smaller pore-size. In the case of the longer500 µm blotting membranes, a ∼100% capture efficiency wasobserved over the range of electric field studied (25-275V/cm). While the optimal conditions required for maximizingcapture % are dependent on several factors (i.e., gel pore size,

the affinity between the migrating species and the immobilizedantibodies, and the blotting antibody immobilization density),a blotting membrane length of 200 µm was sufficient forefficient capture under operating conditions utilized in thiswork.

Toward Protein Quantitation on the Blotting Membrane.To relate the concentration of target protein in the sample tothe fluorescence signal generated at the blotting membrane, acalibration curve was obtained through analysis of a proteindilution series (Figure 5). Dose-response behavior was ob-tained by fluorescence imaging of blotting membranes duringanalysis of BSA (against streptavidin-functionalized blottingmembranes), R-actinin (against anti-actinin blotting mem-branes), and PSA (against anti-PSA blotting membranes). Freedye was spiked into the dilution series at a set concentration

(33) Throckmorton, D. J.; Shepodd, T. J.; Singh, A. K. Anal. Chem. 2002, 74,784–789.

(34) Duchesne, L.; Fernig, D. G. Anal. Biochem. 2007, 362, 287–289.

Table 1. Blotting Membrane Capture Efficiency: Comparison between Prepatterned and Custom-Patterned AntibodyImmobilization Strategiesa

strategy A (prepatterning) strategy B (custom patterning)

[streptavidin-acrylamide, SA]b 0 µM 0.5 µM 4 µM 0.5 µM 2 µM 4 µMratio of SA to biotin anti-actinin 0:2 1:4 2:1 saturated saturated saturatedaverage capture % of R-actinin ± RSD % (n ) 5) 8.6 ± 4.1 68.4 ± 3.1 82.4 ± 2.3 76.4 ± 4.1 87.5 ± 3.5 85.2 ± 3.6

a Blotting voltage 100 V/cm and 200 µm long membrane. b Concentration in polyacrylamide precursor solution.

Figure 4. Capture efficiency of the blotting membrane is influencedby the applied electric field strength and membrane axial length.Two blotting membrane lengths (O, 200 µm; b, 500 µm) wereconsidered for a range of applied blotting electric field strengths.Inset images show inverted grayscale CCD images of blottingmembrane after capture of R-actinin at a concentration of 0.1 µM.

Figure 5. Blotted proteins show a dose response behaviorallowing development of quantitative immunoblotting. (a) Invertedgrayscale CCD images show fluorescence of proteins bound toin-channel antibody-functionalized blotting membranes. E ) 50V/cm. (b) Calibration curve of R-actinin shows enhanced detectionsensitivity on the blotting membrane (b), as compared to signalwithout the blotting membrane (O). Curves were fit by the nonlinearleast-squares method.

8182 Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

and used to normalize the fluorescence signal detected on andoff the blotting membrane. To avoid sample carry-over, newimmunoblot chips were used for each new sample. Figure 5ashows CCD images of membranes blotted with selectedconcentrations of BSA, R-actinin, and PSA. As is visuallyreported in the images, fluorescence signal from blottingmembranes for both R-actinin and PSA showed a dose-dependent response, as higher concentrations of protein wereintroduced. Note that the increase in intensity is more subtlefor PSA than the response for R-actinin, an observation thatmay be attributable to the variation in response inherent tospecies-specific binding affinities. A 1.8 pg mass of PSA wasreadily detectable in the CCD images. For the concentrationof R-actinin considered, blotting membranes successfully bounda substantial mass of analyte (19.8 pg mass in the peak). Nosuch signal dependence on concentration was observed for thenegative control, BSA. Figure 5b reports the full dose-responsebehavior of anti-actinin blotting membranes. A linear responsein fluorescence signal was observed over a 101-103 nMconcentration range. Interestingly, target protein enrichmentoccurred on the blotting membrane and thus yielded ∼5-fold enhanced detection sensitivity (Figure 5b) above corre-sponding protein solutions detected without the blottingmembrane. A similar effect has been observed in proteinmicroarrays.35 The lowest R-actinin sample mass captured anddetected on the blotting membrane was ∼0.05 pg. Our resultsindicate the potential to improve detection limits over slab-gelimmunoblotting36 through lossless manipulation of minutesample masses (picograms), as is especially relevant to singlecell studies.37

Alternate Strategy for Antibody Patterning of BlottingMembrane. The development of the custom patterning strategy(described in Materials and Methods) allows end-user flexibilityin the identity of biotinylated blotting reagents (e.g., antibodies,aptamers, Fab fragments), even after device fabrication, storage,and transport. The flexibility of the custom patterning strategywas demonstrated in Figure 6. Both CCD images and electro-pherograms showed the selective recognition of the targetprotein according to the antibodies patterned by an end-user.Additionally, the custom patterning strategy typically required10 µL of ∼1 µM biotinylated antibody reagent in the patterningwell. On the basis of loading estimates, ∼1 µg of biotinylatedantibody reagent was consumed during the immobilizationprocess. In comparison, conventional slab-gel Western blottingcommonly requires ∼8 µg of antibody reagent for blotting.7

CONCLUSIONSWe introduce a new approach to immunoblotting that takes

advantage of microfluidic technology to simplify integration ofmultistep assays. Here we introduce postseparation (i.e.,“downstream”) functionality that allows reporting of analyteapparent mobility as well as identification through a down-

stream immunoblotting assay. The seamless microfluidic inte-gration of multiple steps (separation, transfer, blotting) yieldsa streamlined workflow with potential for programmable,“hands free” operation. Initial characterization of the PAGE insitu immunoblotting assay indicates that the scheme is a highlyspecific and sensitive approach that yields no discerniblesample loss, suggesting a well-suited format for automated,quantitative microfluidic immunoblotting. Successful assess-ment of R-actinin in a multiprotein sample further suggests thatthe native PAGE in situ immunoblot provides the capability toassay large molecular weight species, a challenge in conven-tional membrane-based blotting. A flexible custom patterningstrategy was demonstrated for postfunctionalization of in-chipblotting membranes. The custom patterning strategy yieldedsubstantially greater versatility than the prepatterning strategyfor customization of the immunoblot by an end-user.

Currently, optimization of the blotting membrane function,enhanced applicability through integration of the immunoblotwith protein sizing (sodium dodecyl sulfate PAGE), and analysisof complex protein samples relevant to both basic science andclinical questions are underway. We see great potential toextend the basic assay developed and presented here to panelsof unique blotted proteins, as well as further improvements indetection sensitivity through on-chip protein enrichment. Ad-ditional design goals include robust quantitation (immunoblot-to-immunoblot reproducibility in quantitation) not readilypossible without fine control of protein transfer. While furtheroptimization is underway, the approach demonstrated hereholds substantial promise as the basis for a suite of automated

(35) Rubina, A. Y.; Pan’kov, A. V.; Dementieva, E. I.; Pen’kov, D. N.; Butygin,A. V.; Vasiliskov, V. A.; Chudinov, A. V.; Mikheikin, A. L.; Mikhailovich,V. M.; Mirzabekov, A. D. Anal. Biochem. 2004, 325, 92–106.

(36) Delaive, E.; Arnould, T.; Raes, M.; Renard, P. J. Immunol. Methods 2008,334, 51–58.

(37) Dishinger, J. F.; Reid, K. R.; Kennedy, R. T. Anal. Chem. 2009, 81, 3119–3127.

Figure 6. Custom patterned blotting membranes show captureof specific target proteins after native PAGE. Inverted grayscaleCCD images and companion electropherograms were collectedboth upstream and downstream of membranes. R-actinin (a) andPSA (b) were selectively captured by corresponding antibodies. E) 80 V/cm, [PSA] ) ∼20 nΜ, [R-actinin] ) ∼10 nΜ.

8183Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u

immunoblotting technologies, including microfluidic Westernblotting.

ACKNOWLEDGMENTThe authors greatly appreciate equipment and glass device

fabrication support from Caliper Life Sciences. The authors thankChenlu Hou and Akwasi Apori at UC Berkeley for their assistance.The authors also thank the California Institute for QuantitativeBiosciences (QB3) at UC Berkeley and UC San Francisco, TheRogers Family Foundation, and the National Science Foundation

Center for Integrated Nanomechanical Systems (COINS) forgenerous financial support. A.E.H. also thanks the UC BerkeleyRegents’ Junior Faculty Fellowship and the Hellman FamilyFaculty Fund Award.

Received for review June 25, 2009. Accepted August 18,2009.

AC901392U

8184 Analytical Chemistry, Vol. 81, No. 19, October 1, 2009

Dow

nloa

ded

by U

OF

CA

LIF

OR

NIA

BE

RK

EL

EY

on

Sept

embe

r 30

, 200

9 | h

ttp://

pubs

.acs

.org

P

ublic

atio

n D

ate

(Web

): S

epte

mbe

r 4,

200

9 | d

oi: 1

0.10

21/a

c901

392u