PII: S0022-1759(01)00394-5

Ž .Journal of Immunological Methods 255 2001 1–13www.elsevier.nlrlocaterjim

Detection of multiple proteins in an antibody-based proteinmicroarray system

Ruo-Pan Huang)

DiÕision of Research, Department of Gynecology and Obstetrics, Emory UniÕersity School of Medicine, 1639 Pierce DriÕe, Room 4219,Atlanta, GA 30322, USA

Received 9 January 2001; received in revised form 27 March 2001; accepted 28 March 2001

Abstract

A new antibody-based protein array assay is described. This assay combines the advantages of the specificity ofŽ . Ž .enzyme-linked immunosorbent assays ELISA , sensitivity of enhanced chemiluminescence ECL and high-throughput of

microspot. In this system, the capture proteins, either antibodies or antigens are spotted onto membranes in an array format.Biological samples are then incubated with membranes. After antigens or antibodies in the samples bind to theircorresponding targets and unbound proteins are washed away, the membranes are exposed to Horseradish PeroxidaseŽ . Ž .HRP -conjugated antibody ies . The signals are finally visualized with ECL system. Experiments demonstrate that multiplecytokines and antibodies can be simultaneously detected using this new approach. The procedure is so simple that nosophisticated equipment is required. The concept should be able to be extended to develop a high-throughput protein arraysystem. Future applications of this new approach include direct protein expression profiling, immunological diseasediagnostics and discovery of new biomarkers. q 2001 Elsevier Science B.V. All rights reserved.

Keywords: Protein arrays; ELISA; Cytokine; Antibody; Protein expression

1. Introduction

With the complete sequence of human genome,we now should be able to analyze whole genomicexpression. This will provide vital information on thecoordinate regulation among many genes since al-most all cell activities or phenotypes are the sum of aseries of molecular and biochemical events interact-ing with each other in a complex and multifacetedfashion. However, a complete understanding of nor-mal and disease states can never be obtained from

) Tel.: q1-404-712-9673; fax: q1-404-727-8615.E-mail addresses: [email protected], [email protected]

Ž .R.-P. Huang .

genomic research alone. Although DNA is an infor-mation archive, proteins do almost all the work ofthe cell. Experimental evidence clearly shows a dis-parity between the relative expression levels of

ŽmRNA and their corresponding proteins Gygi et al.,.1999 . More importantly, the protein-based analysis

is able to study post-transcriptional control, post-translational modifications and protein–protein inter-action. Therefore, the field of proteomics with anultimate goal to assemble a complete library of allthe proteins is becoming increasingly importantŽ .Anderson et al., 2000; Legrain et al., 2000 . Unlikethe cDNA microarray technology, the methodologythat allows detecting entire pool of proteins does notexist yet. Two systems, two-dimensional polyacy-

0022-1759r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.Ž .PII: S0022-1759 01 00394-5

( )R.-P. HuangrJournal of Immunological Methods 255 2001 1–132

Ž .lamide gel electrophoresis 2-D gel coupled withŽmass spectrometry Emmert-Buck et al., 2000; Page

et al., 1999; Haynes and Yates, 2000; Celis et al.,.1999 and surface-enhanced laser desorption and ion-

Ž . Žization SELDI Kuwata et al., 1998; Bruenner et.al., 1996 are currently being used in analysis of

multiple protein expression. However, the require-ment of sophisticated devices greatly limits theiraccessibility. In this study, I demonstrated that nu-merous proteins could be detected simultaneouslyand specifically using an ELISA-based protein arraysystem. This method can be used to detect multiplesecreted proteins and antibodies.

2. Materials and methods

Ž .Immunoglobulins IgGs and correspondingŽ .Horseradish Peroxidase HRP -conjugated mono-

clonal antibodies were purchased from several differ-ent companies as shown in Table 1. IgGs wereprepared as stock solutions at a concentration of 4mgrml and diluted with TBS to 100 mgrml asworking solutions prior to experiments.

Pairs of antibodies against cytokines were ob-Ž .tained from BD PharMingen San Diego, CA . Cy-

Žtokines were purchased from Peprotech Rochy Hill,.NJ . Cytokines were prepared as stock solutions at a

concentration of 100 mgrml and diluted into suit-able working concentrations prior to experiments.

All of donkey anti-Igs against specific specieswere purchased from Jackson ImmunoResearch Lab-

Table 1Array antigens and corresponding antibodiesJIRL: Jackson ImmunoResearch Laboratories; SCB: Santa CruzBiotechnology.

Antigens Sources Antibodies Sources

Bovine IgG Sigma Anti-bovine IgG JIRLChicken IgG Sigma Anti-chicken IgG JIRLGoat IgG Sigma Anti-goat IgG SCBGuinea pig IgG Sigma Anti-guinea pig IgG JIRLHuman IgG Sigma Anti-human IgG JIRLMouse IgG Sigma Anti-mouse IgG SigmaRabbit IgG Sigma Anti-rabbit IgG SigmaRat IgG Sigma Anti-rat IgG AmershamSheep IgG Sigma Anti-sheep IgG JIRLBSA Roche

Ž .oratories West Grove, PA . HRP-conjugated anti-Ždonkey IgG was purchased from Rockland Gilberts-

.ville, PA . Agarose immobilized guinea pig IgG,goat IgG, human IgG and sheep IgG were alsopurchased from Rockland.

Membranes were provided by several companiesas indicated in Table 2.

2.1. Preparation of array membranes

A template of 504 spots with 28 spots in widthand 18 spots in length in a size of 6=8 cm wasgenerated from a computer. This template was usedas a guide to spot solution onto membranes. To spotcapture proteins onto membranes, the template wasplaced on the top of white light box. Membraneswere then put on the top of the template. Throughthe light, dark spots in the template were clearlyvisible from the membrane and used to guide to spotsolution onto membranes. 0.25 ml of solution wasmanually loaded onto a single spot by a 2-ml pipet-man. Typically it took about 30 to 50 min to spot theentire membrane. HRP-conjugated or biotin-con-jugated antibodies were spotted onto membranes aspositive control and identification of orientation ofarrays.

2.2. Array assay of different species of HRP-con-jugated IgGs

Ž .Different IgGs 0.25 ml of 100 mgrml wereloaded onto membranes as described above. Mem-

Žbranes were blocked with 5% BSA Bovine Serum. ŽAlbumin rTBS 0.01 M Tris HCl pH 7.6r0.15 M

.NaCl for 1 h at room temperature and incubatedindividually or collectively with HRP-conjugated an-tibodies for 2 h at room temperature. Arrays werethen washed three times with TBSr0.1% Tween 20and then twice with TBS. The signals were finally

Ž .imaged with enhanced chemiluminescence ECL .

2.3. Array assay of multiple cytokines

A pair of antibodies that recognize different epi-topes of same antigen was used to capture and detecta certain antigen. 0.25 ml of individual capture anti-body at a concentration of 200 mgrml was spottedonto membranes as described above. After blocking

( )R.-P. HuangrJournal of Immunological Methods 255 2001 1–13 3

Table 2Membranes for protein arraysMSI: Micron separations.

Membrane Manufacturer Cat. no. Absorption Detection of IgGs Detection of cytokines

Bkg Sensitivity Bkg Sensitivity

Biotrans ICN BNRQ3R Excellent qq qqq qqqqq ?Zeta-probe Bio-Rad 162-0155 Good qq qqqColonyrplaque screen NEN NEF-978X Very good qq qqqHybond-Nq Amersham PRN303B Very good qq qqq q qqqMagnacharge MSI NBOHY00010 Poor q qqq qqqqq qMagnaGraph MSI NJOHY00010 Excellent y qq qqqqq qHybond ECL Amersham RPN2020D Excellent qq qqq y qqq

with 5% BSArTBS, membranes were incubated witha single or a combination of different cytokinesprepared in 5% BSArTBS for 2 h at room tempera-ture. Unbounded cytokines were washed out withTBSr0.1% Tween 20 and TBS. Membranes werethen incubated individually or collectively with bi-otin-conjugated anti-cytokines antibodies. Wash wasfollowed and the arrays were imaged as done in ECLsystem.

2.4. Array assay of multiple antibodies

Ž .Different species of IgGs 100 mgrml were im-mobilized onto membranes at a quantity of 0.25 mlper spot. The arrays were then incubated with vari-ous Donkey anti-IgGs either individually or collec-tively, after blocking with 5% BSArTBS. Followingwash steps, membranes were incubated with rabbitanti-Donkey IgG, which was pre-absorbed withagarose-immobilized guinea pig IgG, goat IgG, hu-man IgG and sheep IgG, to remove cross-reactioncomponents. Images were carried out with ECL.

2.5. Protein array assay for detection of cytokinesfrom conditioned media and sera

One milliliter of twofold diluted conditioned me-dia and 1 ml of fivefold diluted sera were incubatedwith cytokine array membranes. To prepare condi-tioned media, human mammary epithelial cells, 184-A1N4 and human breast cancer cells, MDA-MB-468

Žcells as described in our previous publication Huang.et al., 1997 were plated in 100-mm tissue culture

dishes at a density of 1=106 cells per dish. MDA-MB-468 cells were cultured with Dulbecco’s modi-

Ž .fied Eagle’s medium DMEM containing 10% fetalŽ .calf serum FCS . 184-A1N4 cells were grown atŽ . Ž .DMEMrF12 1:1 containing insulin 10 mgrml ,

Ž . Ž .transferrin 10 mgrml , EGF 10 ngrml , hydrocor-Ž .tisone 0.5 mgrml and 0.5% FCS. After 48 h, the

complete culture medium was then replaced withŽ .DMEM containing 0.2% calf serum CS . Cells were

Fig. 1. Selection of membranes for protein arrays. Differentmembranes were spotted with IgGs and other controls as indi-cated. The membranes were then incubated with HRP-conjugatedanti-bovine IgG after being blocked with 5% BSA. The signalswere visualized with ECL.


then cultured for another 24 h. The supernatantswere collected, centrifuged at 1000=g, aliquotedand stored at y808C until testing. Patient’s serawere obtained from Dr. Sampath Parthasarathy atDepartment of Gynecology and Obstetrics, Emory

ŽUniversity School of Medicine Santanam et al.,.1998 .

3. Results

3.1. A simplified model: detection of HRP-con-jugated antibodies

To simultaneously detect multiple proteins andantibodies, a microspot approach was developed. In

this approach, capture proteins, either antibodies orantigens, are spotted onto a membrane. The mem-brane is then exposed to a sample containing proteinof interest. The corresponding antigen binds to itscognate antibody spotted onto membrane and de-tected by a developing antibody.

As a first step, a simplified system was applied totest the feasibility of this assay. Various known

Ž .specific immunoglobulins Igs were spotted ontomembrane and detected by incubation with HRP-conjugated antibodies specific to corresponding anti-gens. The signals are then visualized by ECL.

To select a suitable membrane for the assay,bovine IgG and other controls were spotted ontoseveral types of membranes commercially available

Ž .Fig. 2. Specificity and sensitivity of detection of HRP-conjugated anti-species-specific IgGs. A MSI magnagraph membranes immobilizedwith different species-specific IgGs were incubated with individual HRP-conjugated anti-IgGs against specific species or controls as

Ž .indicated in the figure. HRP-conjugated anti-bovine IgG was spotted onto membranes as identification of orientation. B Differentconcentrations of HRP-conjugated anti-bovine IgG as indicated beside each microarray row were used to test the detection sensitivity.


Ž .Fig. 3. Raw image data from high-density protein arrays to detect HRP-conjugated antibody. A Array layout. Membranes were spottedŽ .with different IgGs or controls as indicated in the array layout. There are 504 spots in one array. B Membranes were incubated

Ž .individually or collectively with HRP-conjugated antibodies as indicated in B .


Ž .Table 2 . Membranes were then incubated withHRP-conjugated anti-bovine IgG. As shown in Fig.1, Magnagraph and Magnacharge gave the lowestbackground in the test. Magnagraph has higher ab-sorption capacity than Magnacharge. Therefore,Magnagraph was selected in the assay.

The specificity of the assay was then tested. Dif-ferent monoclonal antibodies were incubated withthe membranes spotted with IgGs. As shown in Fig.2A, very low background levels were consistentlyseen in all cases. Specific recognitions were alsoobserved in bovine IgG, chicken IgG, donkey IgG,goat IgG, guinea pig IgG, mouse IgG, rabbit IgG andrat IgG. Cross-reactions were noticed among goatIgG, human IgG and sheep IgG.

The sensitivity of the array was demonstrated byincubation of membrane with different concentra-tions of HRP-conjugated anti-bovine IgG. As shownin Fig. 2B, as low as 5 pgrml of HRP-conjugatedanti-bovine can be detected. The same sensitivitywas also seen in other HRP-conjugated antibodiesŽ .data not shown .

To demonstrate that this approach can be used inhigh density array format, membranes with a total of504 spots immobilized with different IgGs as indi-cated in Fig. 3A were incubated with single orcombination of HRP-conjugated antibodies. Asshown in Fig. 3B, a total of 504 spots can besimultaneously detected with similar specificity andsensitivity as lower density arrays.

3.2. Application one: simultaneous detection of mul-tiple cytokines

The protein array system described above wasextended to assay human cytokines. The principle ofthis assay is based upon the sandwich ELISA andECL. A pair of antibodies, which recognize twodifferent epitopes of same antigen, is used. One

antibody is spotted onto membrane, serving as acapture. Another corresponding antibody is labeledwith biotin, serving as a detector. The signals arevisualized by ECL system.

Several different types of membranes as indicatedin Table 2 were screened for this assay. Amongthem, Hybond ECL membrane showed highest sensi-

Ž .tivity and lowest background data not shown andwas used in the array format analysis of cytokines.Six cytokines were assayed using this array format.The specificity of this assay was first demonstratedwith individual cytokine. As shown in Fig. 4, thespecificity was observed in all cytokine, consistentwith ELISA data. All six cytokines specifically rec-ognized their corresponding capture antibodies andwere detected by their cognate detection antibodies.No cross-reaction was observed among six cytokinesand other several controls, including EGF, BSA andbuffer only. When all six cytokines were used, spe-cific signals were detected in the spots immobilizedwith six cytokines, but not in EGF, BSA and bufferonly. Additional controls further demonstrated thehigh specificity of this assay, including incubation ofmembranes with EGF, followed by an unrelateddetection antibody or incubation with detection anti-body alone. High sensitivity of this assay was alsoachieved. The detection level is lower than 25 pgrml

Ž .for IL-2 Fig. 4B . Several other cytokines, includingMCP-1 and TNFa , have also shown a similar sensi-

Ž .tivity data not shown .To detect proteins from unknown samples, condi-

tioned media collected from human mammary ep-ithelial cells, 184-A1N4 and human breast cancercells, MDA-MB-468 and sera from two individualdonors were assayed for their cytokine expression.As shown in Fig. 4C, several cytokines were differ-entially expressed in different samples. As control,no signals were detected in medium alone. The

Ž .Fig. 4. Detection of cytokines in array format with high specificity and sensitivity. A Hybond ECL membranes immobilized with differentcapture antibodies against different cytokines were incubated subsequently with single cytokine or all six cytokines or controls as indicatedin the figure, with corresponding biotin-conjugated anti-cytokines or controls and with HRP-conjugated streptavidin. The signals were

Ž .detected by ECL. B High sensitivity of detection is exemplified by IL-2. Membranes spotted with cytokines were incubated with differentŽ .concentrations of cytokines. The signals were then detected with biotin-conjugated anti-IL-2 and HRP-conjugated streptavidin. C

Detection of cytokine expression from conditioned media and sera. The membranes spotted with six different anti-cytokines were incubatedwith twofold diluted conditioned media and fivefold diluted sera. The membranes were then incubated with a mixture of biotin-labeledantibodies against six different cytokines. Signals were detected by ECL.


results clearly demonstrated the potential applicationof this system in pathophysiological samples.

The possibility of constructing high-density arraysystem is demonstrated in Fig. 5. 504 spots in a


Ž .Fig. 5. Detection of cytokines in high-density protein array format. A Different cytokines or controls were spotted on Hybond ECLŽ .membranes in high-density array format as indicated in the figure. B Detection of cytokines with high-density array format was

demonstrated by incubation of the array membranes with individual or collective cytokines.


single membrane can be simultaneously detected. Allthe signals were detected as expected.

3.3. Application two: simultaneous detection of mul-tiple antibodies

The protein array system can also be used todetect antibodies. In this assay, different known anti-gens are spotted onto membranes. The membranesare then incubated with samples containing antibod-ies to be detected. After extensive wash, the mem-branes are incubated with HRP-conjugated anti-

Ž .body ies against species-specific IgGs. The signalsare visualized by ECL.

To demonstrate that this is a feasible approach todetect multiple antibodies simultaneously, different

species of IgGs were spotted onto the MSI Magna-graph membrane. After blocking with BSA, mem-branes were incubated with individual donkey anti-IgGs against given species. After extensive wash toremove the unspecific binding, membranes were in-cubated with anti-donkey IgG. Extensive wash wasthen followed and signals were visualized with ECL.As shown in Fig. 6, specific signals were detectedusing donkey anti-chicken, guinea pig, human,mouse, rabbit and rat. Some cross-reactions wereobserved between donkey anti-goat and sheep. Thesensitivity of this assay is between 5 and 50 pgrmlas demonstrated using donkey anti-mouse IgG. Theexamples of high-density arrays were demonstratedin Fig. 7. Incubation of membranes with single orcombination of donkey anti-IgGs against specific

Ž .Fig. 6. Specific and sensitive detection of antibodies in array format. A MSI Magnagraph membranes loaded with different IgGs asindicated were incubated with single donkey anti-species-specific IgGs or all eight antibodies. Membranes were then incubated with

Ž .HRP-conjugated anti-donkey IgG. B Different concentrations of donkey anti-mouse IgG were used to test the detection limit in arrayformat.


Fig. 7. High-density protein arrays to simultaneously detect multiple antibodies. Different IgGs or controls were loaded onto Hybond ECLŽ . Ž .membranes in an array format as indicated in A . Simultaneous detection of multiple antibodies was demonstrated as indicated in B .


species produced the specific signals just as ex-pected.

4. Discussion

The array format for global analysis of geneexpression, mainly cDNA microarrays and DNA chiptechnology has revolutionized biological and medicalresearch. Recently, the array format has been used insystematical analysis of protein–protein interactionŽ .Emili and Cagney, 2000 . Arrays have the advan-tage of being scalable, flexible and easy to perform.The nature of arrays allows a high-throughputscreening using robotic, imaging, or analytical meth-ods. However, no protein array method in the analy-sis of protein expression has been developed. Herethe antibody-based protein array format for analysisof multiple protein expression was described. Theprinciple of this method involves immobilization ofcapture protein onto membrane, incubation of mem-brane with protein-containing sample and detectionof protein of interest by exposure to a solutioncontaining developing antibody against a second epi-tope of protein of interest. The methodology de-scribed here has several advantages over 2-D gelcombined mass spectrometry and SELDI. It is sosimple that it can be performed at any laboratorysetting without any sophisticated device. In contrast,2-D gel combined mass spectrometry requires sev-eral technically difficult methodology and very ex-pensive equipment. A special and expensive machineis also needed in SELDI. Secondly, the array formatallows us to detect any protein of interest. While 2-Dgel combined mass spectrometry and SELDI is pow-erful in the determination of differential profile ofprotein expression, the specific protein expression ishard to detect. Finally, the antibody-based arrayapproach is particularly useful in accurately measur-ing the difference in individual protein levels be-tween two samples, which is difficult to achieveusing 2-D gel approach. Finally, the background islower. Therefore, the sensitivity is high. The detec-tion limitation for IL-2 is lower than 25 pgrml; fordonkey anti-mouse IgG, it is between 5 and 50pgrml. The sensitivity is higher than ELISA. Thesensitivity of detection will significantly increase ifwe apply other more sensitive ECL detection kits. In

fact, our unpublished data showed that the sensitivitywould enhance about one order by using Amershamplus ECL kit. Therefore, the assay system is verysensitive. Furthermore, the intensities of signals canbe adjusted by exposure times, allowing detection ofhigh and low abundant proteins at the same time.

The microspot approach makes the high-densityprotein array possible. In the system described here,as many as 504 spots can be detected simultane-ously. The limitation of spots is compromised due tothe manual manipulation of the immobilization pro-cess. The application of a microarrayer should allowfor the construction of a much higher density ofarray system.

By using densitometry or CCD imaging system,one can compare the differential protein levels amongmultiple samples. To quantitate the exact amount ofproteins, multiple standard curves can be generatedsimultaneously. According to standard curves, theexact amount of individual protein can be deter-mined.

The concept of antibody-based microarray wasproposed in the mid- to late-1980s by Drs. Ekins and

Ž .Chu Ekins and Chu, 1992 . They proved mathemati-cally that antibody-based arrays are able to simulta-neously determine multiple protein levels with high

Ž .sensitivity Ekins and Chu, 1997; Ekins, 1998 . Sincethen, several groups have attempted to establish anantibody-based array assay to simultaneously deter-mine multiple protein expression. Among them, Dr.Silzel et al. demonstrated that multiple IgG sub-classes could be detected by fluorescence imaginingŽ .Silzel et al., 1998 . Researchers at Genometrix fur-ther demonstrated a possibility to detect multiple

ŽIgGs in a high-throughput format Mendoza et al.,.1999 . Recently, the technology of protein arrays is

Žexploding Walter et al., 2000; Service, 2000; Brad-.bury, 2000 . After we completed the work described

in this paper, several groups have reported to applyprotein array format to screen antibody–antigen in-

Ž .teraction de Wildt et al., 2000 , to study protein–Žprotein interaction MacBeath and Schreiber, 2000;

.Wang et al., 2000 , to analyze yeast protein kinasesŽ .Zhu et al., 2000 and to examine autoimmune anti-

Ž .bodies Joos et al., 2000 . I demonstrated an alterna-tive and simple way to simultaneously detect multi-ple protein expression level by combination ofELISA, microspot and ECL. Simplicity, low back-


ground and high sensitivity are among the advantageover other antibody-based array approaches. The pro-tein array system described here, which permits thesimultaneous determination of multiple proteins, willprove to be of particular value in many areas. Oneexample is to study the expression of cytokines,chemokines and growth factors. In this situation, thedifferent capture antibodies are spotted onto mem-brane. Membranes are then incubated with condi-tioned media, sera or other analyte-containing fluids.The non-specific binding is removed and the signalsare revealed by a pool of developing antibodiesagainst different antigens. The nature of collection ofsamples from either conditioned media or patientsera makes the detection of cytokines by a proteinarray system much more attractive than cDNA arraysystem. Another application of protein array technol-ogy described here is detection of antibodies, whichis particularly useful in the study and diagnosis ofinfectious disease. In theory, this is simpler thandetection of cytokine since only one developing anti-body is required. The similarity of different antibod-ies from the same species in mRNA sequence makesit difficult to be detected by RNA-based technology,such as Northern blot, RT-PCR and cDNA microar-ray. The protein-based assay may be the most spe-cific and easiest way to detect antibody.

Other potential applications of protein array tech-nology described here include identification of spe-cific markers for disease, determination of differen-tial protein expression in two different states. Theapproach can also be adopted to study the protein–protein interaction, post-modification and epitopemapping.

The methodology should be easily extended tochip technology. Membrane may be stuck to thesurface of glass, or membrane may be engineeredto 96-well plates. Considering the availability ofmicroarrayer, scanning CCD imager and analysissoftware, such modification will make methodologydescribed here a high throughput approach to simul-taneously detect multiple proteins.

The protein arrays are still in infant state. One ofthe challenging problems we are facing in the devel-opment of antibody-based protein arrays is thecross-reaction among different antibodies. This prob-lem may be overcome by careful selection of anti-bodies, pre-absorption of antibodies with antigens

and application of protein-interacting peptides se-lected by phage library. Nevertheless, the resultspresented here suggest that this is a feasible ap-proach. With the advance of antibody engineeringand thousands of antibodies commercially available,the antibody-based protein arrays will become anincreasingly important tool in studying multiple pro-tein expression.

Acknowledgements

This work was supported by NIHrNCI grantŽ .CA89273 RPH and ACS grant RPG-99-164-01-

Ž .CNE RPH . We would like to express our thanksfor the support by the Helen Dyar King Fund at theArizona Community Foundation for Cancer Re-search. We are grateful to Dr. Sampath Parthasarathyfor providing sera in this study.

References

Anderson, N.L., Matheson, A.D., Steiner, S., 2000. Proteomics:applications in basic and applied biology. Curr. Opin. Biotech-nol. 11, 408–412.

Bradbury, J., 2000. Proteomics: the next step after genomics?Lancet 356, 50.

Bruenner, B.A., Yip, T.T., Hutchens, T.W., 1996. Quantitativeanalysis of oligonucleotides by matrix-assisted laser desorp-tionrionization mass spectrometry. Rapid Commun. MassSpectrom. 10, 1797–1801.

Celis, J.E., Celis, P., Ostergaard, M., Basse, B., Lauridsen, J.B.,Ratz, G., Rasmussen, H.H., Orntoft, T.F., Hein, B., Wolf, H.,Celis, A., 1999. Proteomics and immunohistochemistry definesome of the steps involved in the squamous differentiation ofthe bladder transitional epithelium: a novel strategy for identi-fying metaplastic lesions. Cancer Res. 59, 3003–3009.

de Wildt, R.M., Mundy, C.R., Gorick, B.D., Tomlinson, I.M.,2000. Antibody arrays for high-throughput screening of anti-body–antigen interactions. Nat. Biotechnol. 18, 989–994.

Ekins, R.P., 1998. Ligand assays: from electrophoresis to minia-turized microarrays. Clin. Chem. 44, 2015–2030.

Ekins, R., Chu, F., 1992. Multianalyte microspot immunoassay.The microanalytical ‘compact disk’ of the future. Ann. Biol.

Ž .Clin. Paris 50, 337–353.Ekins, R., Chu, F., 1997. Immunoassay and other ligand assays:

present status and future trends. J. Int. Fed. Clin. Chem. 9,100–109.

Emili, A.Q., Cagney, G., 2000. Large-scale functional analysisusing peptide or protein arrays. Nat. Biotechnol. 18, 393–397.

Emmert-Buck, M.R., Gillespie, J.W., Paweletz, C.P., Ornstein,D.K., Basrur, V., Appella, E., Wang, Q.H., Huang, J., Hu, N.,


Taylor, P., Petricoin, E.F., 2000. An approach to proteomicanalysis of human tumors. Mol. Carcinog. 27, 158–165.

Gygi, S.P., Rochon, Y., Franza, B.R., Aebersold, R., 1999. Corre-lation between protein and mRNA abundance in yeast. Mol.Cell Biol. 19, 1720–1730.

Haynes, P.A., Yates III, J.R., 2000. Proteome profiling—pitfallsand progress. Yeast 17, 81–87.

Huang, R.-P., Fan, Y., deBelle, I., Niemeyer, C., Gottardis, M.M.,Mercola, D., Adamson, E.D., 1997. Decreaed Egr-1 expres-sion in human, mouse and rat mammary cells and tissuecorrelates with tumor formation. Int. J. Cancer 72, 102–109.

Joos, T.O., Schrenk, M., Hopfl, P., Kroger, K., Chowdhury, U.,Stoll, D., Schorner, D., Durr, M., Herick, K., Rupp, S., Sohn,K., Hammerle, H., 2000. A microarray enzyme-linked im-munosorbent assay for autoimmune diagnostics. Electrophore-sis 21, 2641–2650.

Kuwata, H., Yip, T.T., Yip, C.L., Tomita, M., Hutchens, T.W.,1998. Bactericidal domain of lactoferrin: detection, quantita-tion, and characterization of lactoferricin in serum by SELDIaffinity mass spectrometry. Biochem. Biophys. Res. Commun.245, 764–773.

Legrain, P., Jestin, J.L., Schachter, V., 2000. From the analysis ofprotein complexes to proteome-wide linkage maps. Curr. Opin.Biotechnol. 11, 402–407.

MacBeath, G., Schreiber, S.L., 2000. Printing proteins as microar-wrays for high-throughput function determination see com-

xments . Science 289, 1760–1763.Mendoza, L.G., McQuary, P., Mongan, A., Gangadharan, R.,

Brignac, S., Eggers, M., 1999. High-throughput microarray-Ž .based enzyme-linked immunosorbent assay ELISA . Biotech-

Ž .niques 27 778-6 , 788.Page, M.J., Amess, B., Townsend, R.R., Parekh, R., Herath, A.,

Brusten, L., Zvelebil, M.J., Stein, R.C., Waterfield, M.D.,Davies, S.C., O’Hare, M.J., 1999. Proteomic definition ofnormal human luminal and myoepithelial breast cells purifiedfrom reduction mammoplasties. Proc. Natl. Acad. Sci. U. S.A. 96, 12589–12594.

Santanam, N., Jonas, C., Murphy, A.A., Parthasarathy, S., 1998.Vitamin E supplementation decreases antoantibodies to oxi-dized lipid–protein complexes. J. Medicinal Food. 1, 247–251.

Service, R.F., 2000. Biochemistry. Protein arrays step out ofw x w xDNA’s shadow comment news . Science 289, 1673.

Silzel, J.W., Cercek, B., Dodson, C., Tsay, T., Obremski, R.J.,1998. Mass-sensing, multianalyte microarray immunoassaywith imaging detection. Clin. Chem. 44, 2036–2043.

Walter, G., Bussow, K., Cahill, D., Lueking, A., Lehrach, H.,2000. Protein arrays for gene expression and molecular inter-action screening. Curr. Opin. Microbiol. 3, 298–302.

Wang, Y., Wu, T.R., Cai, S., Welte, T., Chin, Y.E., 2000. Stat1 asa component of tumor necrosis factor alpha receptor 1-TRADDsignaling complex to inhibit NF-kappaB activation. Mol. CellBiol. 20, 4505–4512.

Zhu, H., Klemic, J.F., Chang, S., Bertone, P., Casamayor, A.,Klemic, K.G., Smith, D., Gerstein, M., Reed, M.A., Snyder,M., 2000. Analysis of yeast protein kinases using proteinchips. Nat. Genet. 26, 283–289.

Technology

PII: S0022-1759(01)00394-5