INFECTION AND IMMUNITY, Nov. 2009, p. 4877–4886 Vol. 77, No. 110019-9567/09/$12.00 doi:10.1128/IAI.00698-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Genome-Wide Study of Pseudomonas aeruginosa Outer MembraneProtein Immunogenicity Using Self-Assembling

Protein Microarrays�†Wagner R. Montor,1‡ Jin Huang,2 Yanhui Hu,1 Eugenie Hainsworth,1 Susan Lynch,3

Jeannine-Weiner Kronish,3 Claudia L. Ordonez,4 Tanya Logvinenko,5Stephen Lory,2 and Joshua LaBaer1*

Harvard Institute of Proteomics, Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School,Boston, Massachusetts 021151; Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston,

Massachusetts 021152; Department of Anesthesia and Perioperative Care, University of California—San Francisco,San Francisco, California 941433; Boston Children’s Hospital, Boston, Massachusetts 021154; and Institute for

Clinical Research and Health Policy Studies, Tufts New England Medical Center, Boston, Massachusetts 021105

Received 19 June 2009/Returned for modification 22 July 2009/Accepted 2 September 2009

Pseudomonas aeruginosa is responsible for potentially life-threatening infections in individuals with com-promised defense mechanisms and those with cystic fibrosis. P. aeruginosa infection is notable for the appear-ance of a humoral response to some known antigens, such as flagellin C, elastase, alkaline protease, and others.Although a number of immunogenic proteins are known, no effective vaccine has been approved yet. Here, wereport a comprehensive study of all 262 outer membrane and exported P. aeruginosa PAO1 proteins by amodified protein microarray methodology called the nucleic acid-programmable protein array. From thisstudy, it was possible to identify 12 proteins that trigger an adaptive immune response in cystic fibrosis andacutely infected patients, providing valuable information about which bacterial proteins are actually recog-nized by the immune system in vivo during the natural course of infection. The differential detections of theseproteins in patients and controls proved to be statistically significant (P < 0.01). The study provides a list ofpotential candidates for the improvement of serological diagnostics and the development of vaccines.

Pseudomonas aeruginosa is a gram-negative bacterium rep-resented by different strains. PAO1 is the most representativeone, and it contains one chromosome comprising 5,570 openreading frames (ORFs) (21). Ubiquitous in the environment, itrarely causes respiratory tract infections in healthy individualsbut causes life-threatening lung infections in cystic fibrosis(CF) patients (24, 27), linking it directly to the disease’s poorprognosis (38).

CF is a hereditary disease affecting 30,000 people in theUnited States alone, and it results from the impaired functionof the cystic fibrosis transmembrane conductance regulator-encoded chloride channel. This defect leads to the accumula-tion of nutrient-rich mucus inside the respiratory tract, which isreadily colonized by bacteria and difficult for immune cells andantibiotics to penetrate (13). Although these patients are gen-erally colonized early by other pathogens, such as Staphylococ-cus aureus and Haemophilus influenzae, by the age of 18, 85%have chronic pseudomonal lung infections (23).

P. aeruginosa is also associated with nosocomial infections,especially in the immunocompromised (6, 10). Multidrug-re-sistant strains in intensive care units as well as outside the

hospital setting have been described, making the developmentof alternate treatment strategies a necessity (9, 29). CFpatients and other at-risk individuals would benefit from afunctional vaccine, but none has been approved for use sofar (8, 16).

In the late 1970s, evidence about proteins acting as impor-tant virulence factors in Pseudomonas aeruginosa infectioncalled attention to exotoxin A (protein PA1148), alkaline pro-tease (protein PA1248), and elastase (protein PA3724) (18,43). Currently, diagnostic serology for P. aeruginosa is mostlybased on these three proteins and has a sensitivity of 80% inchronic infections (36). This test is much less sensitive at theonset of infection, when serology alone does not detectpseudomonal infection in more than 43% of patients. Thediscovery of new antigens to complement the current panel ofantigens would increase diagnostic sensitivity in all infectionphases.

Pseudomonas presents different antigens on the surface de-pending on the microenvironment it colonizes, the mucoidphenotype being a hallmark for lung infections (5, 35). Whichproteins are present on the pathogen’s surface in each micro-environment has not yet been fully characterized; however, thisinformation would be invaluable for diagnostics developmentand reverse vaccinology studies.

Early studies executed in the 1980s probed Western blotsmade from gel electrophoresis-fractionated P. aeruginosa ex-tracts with serum to reveal reactive bands. Some of thesereactive bands were identified to be specific outer membraneproteins (OMPs), such as OMPs E, H2, and I, as well as porinF, but due to technical limitations, not all bands could be

* Corresponding author. Present address: Virginia G. Piper Centerfor Personalized Diagnostics, Biodesign Institute, Arizona State Uni-versity, Tempe, AZ 85287. Phone: (480) 965-2805. Fax: (480) 965-3051. E-mail: jlabaer@asu.edu.

‡ Present address: Department of Physiology, Santa Casa de SaoPaulo Medical School, Sao Paulo, Brazil.

† Supplemental material for this article may be found at http://iai.asm.org/.

� Published ahead of print on 8 September 2009.

4877

identified (15, 20). Notably, all of these proteins, as well as thesecreted proteins alkaline protease and elastase from the ear-lier studies, are externally displayed proteins, confirming theimportance of this class of proteins in triggering an immuneresponse, as they are readily seen by the immune system.Therefore, it would be useful to execute a systematic compre-hensive study to analyze all predicted OMPs, both known andhypothetical, in an unbiased screening experiment.

These also represent good vaccine candidates based on datafrom phase I and II clinical trials using two P. aeruginosa OMPsin vaccines (8). A phase III trial of the external surface flagellinprotein also showed promising results with previously unin-fected CF patients (7). However, testing proteins as candidateantigens ordinarily requires antigen expression and purifica-tion, which are notoriously difficult for membrane proteins,making a comprehensive study of all P. aeruginosa OMPs withtraditional methods impossible, particularly because the P.aeruginosa genome is very GC rich, which reduces the yield ofprotein in common expression systems.

Protein microarrays provide one mechanism for executingan unbiased screen. Recent studies demonstrate immune re-sponses from plague patients against a Yersinia pestis proteinmicroarray containing 144 virulence related factors, 14 ofwhich were detected by all the patients tested (25). Similarprotein microarray studies have been done for selected anti-gens from Francisella tularensis (39), Borrelia burgdorferi (44),Coxiella burnetti (2), severe acute respiratory syndrome coro-navirus (45), and others. In some cases, difficulty in producingproteins has been encountered. In the C. burnetti study, �500proteins could not be studied (�25%) due to difficulties inproducing some individual proteins for array spotting. Thehigh-throughput nature of microarrays allows easy executionof comparative studies of the humoral response mountedagainst different pathogens. In a recent study, Keasey andcollaborators demonstrated that by probing a Yersinia pestis-specific protein microarray with serum from animals infectedwith other gram-negative bacteria, it was possible to findshared cross-reactive proteins (common to several pathogens),fingerprint proteins (common to two or more bacteria), andsignature proteins (specific to each pathogen) (17).

However, the traditional protein microarray printing method,which relies upon printing purified proteins, may encounterlimitations. Chief among these is the difficulty in producing andpurifying a broad range of proteins in order to obtain bothadequate and consistent protein yields, resulting in an exces-sively large dynamic range of protein yields displayed on thearray. This is particularly relevant in circumstances in whichthe targets are either hydrophobic or large.

To address these issues, we previously reported the devel-opment of a novel method for displaying proteins on proteinmicroarrays, called the nucleic acid-programmable proteinarray (NAPPA), which is particularly advantageous formembrane proteins. In contrast to the traditional methodfor producing protein microarrays, NAPPA entails printingthe cDNAs that encode the proteins that are “self-assembled”at the time of the experiment by transcribing and translatingproteins in situ on the array surface (33). The cDNAs aresituated in an expression vector such that they produce pro-teins with a carboxyl-terminal fusion tag that both allows pro-tein capture at the surface and confirms full-length translation.

This unique approach obviates the need to express and purifyany protein, avoids protein aggregation by immediate captureon the slide surface, and ensures that the proteins displayedhave been produced freshly at the time of assay.

NAPPAs display a uniform amount of protein in each fea-ture (34) and work well for detecting humoral immune re-sponses in a micro-enzyme-linked immunosorbent assay(micro-ELISA) format (1, 32). NAPPA has been demonstrated toexpress and display membrane proteins with an efficiency thatexceeds 90% (34), perhaps by avoiding cell toxicity and proteinaggregation. Based on the observation that NAPPA-immobi-lized proteins interact with other proteins as expected (33), wereasoned that many relevant epitopes should be displayed andthat NAPPA could be used to test candidate membrane anti-gens in P. aeruginosa.

Using bioinformatics tools based on amino acid composi-tion, homology to known OMPs, and finding transmembraneß-barrels, we identified 266 gene products among all of the5,570 ORFs in PAO1 that were likely to reside in the outermembrane, are part of exposed surface organelles, or are se-creted into the surrounding media. These proteins were rep-resented on the surface of the NAPPAs, and patient sera wereused to screen for their antigenicity.

MATERIALS AND METHODS

Selection of OMPs by using bioinformatics tools. Several different bioinfor-matics tools, such as BOMP (3) and PROFtmb (4), were used for ß-barrelprediction; PROFtmb uses a profile based on a hidden Markov model, whileBOMP uses other sequence-based algorithms to predict bacterial transmem-brane �-barrels in the structure. PSORT (11) and Pence (26) were used for theprediction of subcellular localization. PSORT uses a combined algorithm topredict subcellular location by analyzing signal peptide, transmembrane helices,homology to annotated proteins, amino acid composition, and structural motifs.Pence uses database text annotations from homologs and machine learning tosubstantially improve the prediction of subcellular location. To achieve the mostcomprehensive OMP list, we combined the predicted results from all approachesand compared them to available annotation. A specialist in the field verified theresults to eliminate obvious false positives and to add candidates of interestbased on updated literature that might have been missed by the bioinformaticstools, defining the final set of 266 proteins to be studied.

Transfer of ORFs from the entry vector into the pANT7-GST expressionvector. Pseudomonas aeruginosa entry clones representing full-length ORFs inpDONR221 were previously made in our laboratory (21). The full-length se-quences of all of the ORFs tested in this work were verified. Reactions fortransfer from pDONR221 into pANT7-cGST were done with LR kits fromInvitrogen, following the manufacturer’s instructions and as previously described(28). Transfer reaction products were transformed into a T1 phage-resistantEscherichia coli DH5� host strain and plated onto LB agar-ampicillin 48-wellgrid plates for single-colony selection. After colony growth, a customizedMegapix robot (Genetix) was used to count colonies, capture an image, isolatethe indicated number of single colonies from each sector of the agar dish, andinoculate the colony into 1 ml growth media (LB-antibiotic). Cultures weregrown overnight at 37°C, their growth was verified via measurement of opticaldensities at 600 nm, and aliquots were stored at �80°C as 15% glycerol stocks.

Plasmid DNA isolation. An aliquot of the glycerol stocks was diluted in LBmedium (1:300) and replated on LB agar (100 �g/ml ampicillin), and on thefollowing day, isolated colonies were used to inoculate 1.5-ml Terrific Broth (100�g/ml ampicillin) cultures in the 96-deep-well block format (34). Cultures weregrown for 24 h at 37°C and pelleted by centrifugation at a relative centrifugalforce of 5,000 for 15 min. Plasmid DNA was obtained by regular alkaline lysis,scaled up to a 96-well format, followed by anion exchange purification using thereagents and protocols provided by the resin manufacturer (Macherey-Nagel).Agar spotting and volume transfers from glycerol inoculation to plasmid DNApurification were done using robotics whenever possible (Biomek FX from Beck-man-Coulter, Genmate from Tecan, and WellMate from Matrix).

Sample preparation for microarray printing. In brief, purified DNA wasprecipitated using 10% sodium acetate (pH 5.5) and 60% isopropanol, washed

4878 MONTOR ET AL. INFECT. IMMUN.

with 200 �l of 80% ethanol, and dried. The dried DNA pellet was resuspendedin a printing solution containing 50 ng/�l of capture antibody (Amersham), 3mg/ml of bovine serum albumin (Sigma), and 2 mM bis(sulfosuccinimidyl) su-berate (Pierce). The printing solution was transferred to a 384-well printing plate(Genetix) and arrayed onto aminosilane-coated glass slides. The printing wasperformed using a Genetix Q2 microarrayer with 300-�m solid tungsten pins,and slides were stored dry at room temperature.

NAPPA slide processing. The slides were blocked in Superblock (Pierce) for1 h at room temperature, rinsed with double deionized water, and dried usinghouse air. An incubation chamber (Grace Bio) was applied to the array surface,and the cell-free transcription and translation mix (T7-TNT system; Promega)was prepared following the manufacturer’s instructions and added to the slides.The slides were incubated in a chilling oven (Torrey Pines) at 30°C for 1.5 h andat 15°C for 0.5 h. Following activation with cell-free lysate, the slides wereblocked with blocking buffer (5% milk in phosphate-buffered saline [PBS] sup-plemented with 0.2% Tween 20) for 1 h. For the detection of proteins on thearray, the slides were incubated with 300 �l of primary antibody (10 �g/mlanti-glutathione S-transferase [GST] antibody; Cell Signaling Technologies) byusing a LifterSlip (Erie) for 1 h at room temperature. The slides were rinsed withthe blocking buffer and incubated with secondary antibody (10 �g/ml anti-mouseimmunoglobulin G [IgG]; Amersham) by using the same conditions as for theprimary antibody. The GST moiety is at the carboxyl terminus of each expressedprotein; thus, detection confirms full-length translation. For the detection ofserum signal, slides were placed in a hybridization chamber (Corning) and 2 mlof diluted serum (1:600 in blocking buffer) was added. The slides were incubatedfor 16 h at 4°C on a rotator. Following incubation with serum, slides were washedwith blocking buffer and incubated with secondary antibody against human IgGs(1.6 �g/ml; Jackson ImmunoLabs) by using the same chamber previously de-scribed, for 1 h at room temperature. The slides for both protein signal andserum signal were developed using 600 �l of tyramide signal amplification solu-tion (Perkin Elmer) and scanned using a ScanArray ProScanArray HT scanner(Perkin Elmer). Scanning conditions were 43% photomultiplier tube gain, 79%laser power, and 50-�m resolution for visualization and 33% photomultipliertube gain, 80% laser power, and 20-�m resolution for quantification. False-colorimages were generated using ScanArray Express. DNA binding was measured byincubating the slides in 20 ml 1� PBS containing 34 �l PicoGreen reagent. Thearray images were quantified using MicroVigene software, version 2.501(VigeneTech).

Serum collection. A collection comprising 22 serum samples from patientsattending the Cystic Fibrosis Clinic at the Children’s Hospital in Boston, Mas-sachusetts, was used. The ages of patients ranged from 5 to 54 years. Sputumsamples were collected at the time of the outpatients’ visit or admission, and theywere all positive for P. aeruginosa. At this time, blood, from which the sera for theprotein array were generated, was collected. The lung function of the subjects, asdetermined by spirometry, ranged from normal to 33%. Half of the patientsexperienced exacerbation as determined by the attending physicians.

The 16 non-CF patient samples were collected at various clinics at the Uni-versity of California San Francisco Medical Center. Samples were from intubatedintensive care unit patients with newly acquired P. aeruginosa strains. To confirmthat patients were infected with P. aeruginosa, bacteria were isolated from en-dotracheal aspirates. Isolates from airway secretions with high bacterial counts(mean CFU/ml � 6 � 107) were typically associated with patients diagnosed withventilator-associated pneumonia; 50% of patients in this group died. Culturesfrom airway secretions with lower counts (mean CFU/ml � 3.4 � 105) wereisolated from patients who were subsequently discharged or transferred to acuteor long-term-care facilities. All isolates were cultured on Pseudomonas isolationagar, and the identities were confirmed by MicroScan Combo assay (DadeBehring, CA) according to the manufacturer’s instructions. Sera from bloodsamples obtained from a third group of patients with newly acquired P. aerugi-nosa in their respiratory secretions and a preexisting diagnosis of chronic ob-structive pulmonary disease (COPD) were also used for this study. One of thepatients in this cohort developed clinical signs and symptoms of ventilator-associated pneumonia; this patient subsequently died.

Fifteen healthy, P. aeruginosa-free controls were used to screen NAPPA slidescontaining the 262 Pseudomonas aeruginosa ORFs.

All samples collected for this study were collected with institutional reviewboard approval and handled as deidentified sera. The serum supply for this studywas exhausted during the NAPPA screening and ELISA/Western blot validationstudies.

Image analysis and quantification. After the slides were scanned, spot inten-sities were measured using the MicroVigene software (VigeneTech) for microar-ray analysis. For each serum sample, the average of four spots (two slides perpatient and two spots per antigen on each slide) was calculated for each antigen.

The mean of all spots in each slide was obtained (disregarding controls), and theZ-score for each antigen in relation to the slide mean was generated. Slideimages were also printed for visual confirmation by two independent researchersbased on spot intensity and morphology.

Western blot confirmation. Proteins of interest were produced in solution byusing the in vitro transcription/translation (IVTT) system. Plasmid DNA (500ng) purified by anion exchange and resuspended in diethyl pyrocarbonate-treated water was used to program protein production in IVTT reactions (20 �l).An aliquot of 3 �l of IVTT product was subjected to electrophoresis using 4 to20% Criterion Precast gels (Bio-Rad), and the proteins were then transferred toa polyvinylidene difluoride membrane (GE Healthcare), which was blocked inSuperBlock (Thermo Scientific) overnight and incubated with patient serum(1:600 in 1� PBS containing 5% milk and 0.05% Tween 20) for 1 h at roomtemperature. Signal detection was performed using 1:1,000 anti-human IgG(Jackson ImmunoResearch) and SuperSignal West Femto (Thermo Scientific).Images were obtained using Chemi Doc (Bio-Rad).

ELISA for confirmation. Proteins of interest were produced as described forWestern blotting and bound to a preblocked anti-GST-coated plate (GE Health-care). A standard ELISA protocol was followed using patient or control serumdiluted at 1:300 in 1� PBS containing 5% milk and 0.05% Tween 20. Anti-human IgG (Jackson ImmunoResearch) and Super Signal ELISA Pico (ThermoScientific) were used for detection. Luminescence was read on the SpectraFluorPlus reader (Tecan), and images were obtained using Chemi Doc (Bio-Rad).

Statistical analysis. To examine reproducibility of the signals obtained inELISAs, duplicate features of selected antigens were probed with serum frommultiple patients. For each of the patients, we computed the coefficients ofvariation ([standard deviation/mean] � 100%) of the log-transformed signals ofthe duplicate spots. The coefficients of variation ranged from 0.03% to 1.28%.The ratio of the positive samples and controls for each antigen was computed asthe average ratio of the antigen’s raw signal to the average raw signal of negativeinternal controls across the assays. The signal intensities were log transformed,and positive samples were compared to the internal negative controls within eachassay by using an analysis of variance adjusted for assay effect. The normalityassumptions for each antigen were verified using the Shapiro-Wilks test. Allratios and P values adjusted for multiple comparisons for the tested antigens arelisted in Table 1 and Table S2 in the supplemental material. The P values wereadjusted for multiple comparisons by using the Benjamini-Hochberg procedure.

For NAPPA, the reproducibility between duplicate NAPPA spots (serum signal)was determined by the coefficient of variation ([standard deviation/mean] � 100%)for the within-slide replicates. More than 70% of the antigens had coefficients ofvariation below 25%. For comparison among patients and controls, serum signalintensities for each particular antigen were computed as the average of four spots(two replicate spots from two duplicate slides), and a Z-score was computed foreach antigen based on the average signal generated by that particular patient.The cutoff for positivity was defined as 1.6. A one-sided Wilcoxon test was usedto compare the signals between controls and positive patients. The P values wereadjusted for multiple comparisons by using the Benjamini-Hochberg procedure.

RESULTS

Selection of the proteins to be represented on the arrays.Using the different tools described, it was possible to predictthe proteins of interest as being those externally exposed by thepathogen. Table 2 was generated with the results from thesedifferent complementary tools plus the final hand curationfrom a specialist in the field (Stephen Lory). Many of thesetools pointed to the same proteins, so a final nonredundant listof 266 proteins was assembled (see Table S1 in the supplemen-tal material). This final list includes all the proteins mentionedin the introduction, which had been previously described asimmunogenic, such as flagellin, exotoxin A, elastase, alkalineprotease, OprI, OprH2, OprE, and OprF.

To be sure that the genes we tested were not unique to thePAO1 strain, we checked the Microbial Genome Database forComparative Analysis (http://mbgd.genome.ad.jp/) to analyzethe four P. aeruginosa strains whose genomic sequences werepublicly available (PAO1, PA7, LESB58, and UCBPP-PA14)and found that more than 90% of the ORFs present in the

VOL. 77, 2009 P. AERUGINOSA ANTIGENS DETECTED BY MICROARRAY 4879

PAO1 strain have nearly identical orthologs in at least one ofthe other strains. Specifically, for the 266 OMPs selected fromPAO1, 230 have orthologs in all three of the other strains, 30have orthologs in two other strains, and 3 have orthologs in oneother strain. Thus, only two OMPs are PAO1 specific.

Generation of the slides containing P. aeruginosa proteins ofinterest. After PCR-based cloning of the full-length ORFs andfull-length sequence verification (40) in the Gateway system(21), 262 of these 266 ORFs were successfully transferred tothe in vitro expression vector pANT7-cGST and printed ontoNAPPA slides in duplicate. NAPPA is a modified proteinmicroarray in which expression clones (cDNAs) are initiallyprinted and then transcribed/translated in situ to produce pro-tein at the time of the experiment. As a first quality controlstep, we assessed binding of the plasmid DNA on the slides bystaining with PicoGreen, which fluoresces when bound toDNA (Fig. 1a). The protein-encoding sequences are arrangedsuch that they encode carboxyl-terminal GST tags for all pro-teins. Thus, to confirm protein expression and capture after theaddition of the rabbit reticulocyte-based in vitro expressionsystem, anti-GST antibody was used to detect the freshly pro-duced proteins (Fig. 1a).

The false-color image in Fig. 1a representing the proteinarray was adjusted to avoid saturating the bright pixels, andbecause of the narrow dynamic range of the print image, somespots appear to have no protein. However, as seen in theplotted quantitative values for the signal intensity in Fig. 1b, allof the array features where we printed an ORF had signals thatwere more than three standard deviations higher than thebackground signal from the control features (features with thesame chemistry but which do not express protein).

To demonstrate that these arrays could be used to detect ahuman serum response and that the epitopes on the displayedproteins are accessible, a serum sample from a CF patient wasused as a primary antibody to bind to the arrays (Fig. 1a). Afterwe washed away the patient serum, a horseradish peroxidase-labeled secondary anti-human IgG was added and visualizedwith the tyramide signal amplification system, which generatesfluorescent free radicals that attach to the slide surface at thesite of antibody binding.

The resulting images were entered into MicroVigene mi-croarray quantification software and printed for visual analysis.Quantification showed that all the clones of interest (Fig. 1b)have sufficient immobilized DNA and display protein above

TA

BL

E1.

NA

PPA

Z-s

core

resu

lts,v

isua

lcon

firm

atio

n,E

LIS

Ara

tios,

and

stat

istic

alan

alys

isre

sults

for

the

top

12an

tigen

sa

Prot

ein

CF

patie

nts

(n�

22)

NC

Fpa

tient

s(n

�16

)C

ontr

ols

(n�

15)

Z-s

core

hit

no.o

fpa

tient

s

Vis

ualh

itno

.of

patie

nts

Max

Zin

hits

NA

PPA

Pva

lueb

Hit/

cont

rola

ntig

enE

LIS

Ara

tio(P

valu

e)

Z-s

core

hit

no.o

fpa

tient

s

Vis

ualh

itno

.of

patie

nts

Max

Zin

hits

NA

PPA

Pva

lueb

Hit/

cont

rola

ntig

enE

LIS

Ara

tio(P

valu

e)

Z-s

core

hit

no.o

fco

ntro

ls

Vis

ualh

itno

.of

cont

rols

Max

Zin

hits

PA00

4421

2014

.40

0.00

5281

725.

6(�

0.01

)7

89.

030.

7683

4409

33.

5(�

0.01

)5

410

.93

FliC

_A20

2012

.75

5.93

484E

�06

10.7

(�0.

001)

12

3.34

0.39

2514

931

2.4

(�0.

01)

11

2.88

PA53

6913

206.

882.

1067

1E�

052.

2(�

0.01

)0

12.

0(0

.44)

00

PA10

8010

144.

790.

0219

9524

211

.2(�

0.00

1)1

11.

721

4.2

(�0.

01)

00

PA38

4110

139.

277.

7305

1E�

067.

9(�

0.00

1)2

41.

950.

0077

4609

45.

0(�

0.00

1)0

0PA

0973

1213

5.33

1.61

306E

�05

8.2

(0.0

01)

11

5.80

0.05

5059

338

23.5

(�0.

05)

00

PA49

2212

129.

690.

0183

0299

28.

6(�

0.00

1)1

12.

690.

0084

1139

56.

00

0PA

0807

1312

8.43

0.00

2835

889

8.5

(0.0

7)0

00

0PA

3931

812

5.95

0.00

0122

245

5.5

00

00

PA34

0710

103.

710.

0780

6001

913

.6(�

0.00

1)15

107.

771.

8017

6E�

0515

.1(�

0.01

)2

01.

83PA

4110

910

10.2

81.

2097

9E�

0513

.0(�

0.00

1)1

23.

840.

0018

1819

39.

4(�

0.01

)0

0PA

2300

129

6.99

2.24

864E

�05

6.0

(�0.

001)

01

2.6

(0.0

1)0

0

aT

here

sults

for

the

12to

pan

tigen

sfo

und

are

here

show

n,in

clud

ing

the

num

ber

ofpa

tient

sin

whi

chth

eir

Z-s

core

was

posi

tive,

the

num

ber

ofpa

tient

sin

whi

chth

eyw

ere

avi

sual

hit,

the

max

imum

(Max

)Z

-sco

refo

rth

atsp

ecifi

can

tigen

amon

ghi

ts,t

heP

valu

efo

rth

ean

alys

isof

the

inte

nsity

ofth

esp

ecifi

can

tigen

inpa

tient

san

dco

ntro

lgro

ups,

and

the

EL

ISA

resu

lts.T

able

S2in

the

supp

lem

enta

lmat

eria

lsho

ws

the

exte

nded

anal

ysis

,inc

ludi

ngre

sults

for

allt

he26

2an

tigen

ste

sted

.b

Pva

lues

are

for

com

pari

son

betw

een

the

patie

ntgr

oup

and

the

cont

rolg

roup

.

TABLE 2. Tools used to predict OMPsa

Source

No. of OMPs

Predicted Confirmed byannotation

Passed handcuration

BOMP 120 61 112PROFtmb 62 55 62PSORT 165 108 154Pence 142 112 137Genome annotation 129 NA 123Experiment 53 NA 53

Total 293 119 266

a The chart displays the number of OMPs selected by each tool used, as wellas the proportion of those that were confirmed by annotation and hand curation.

4880 MONTOR ET AL. INFECT. IMMUN.

the detection threshold (red lines) set three standard devia-tions higher than the no-DNA controls (blue spots). Moreover,the pilot serum screening revealed that even weak proteinspots were detected by antibodies in patients’ sera (Fig. 1c).

In addition to the quantitative analysis, the slides were ex-amined for responses that were visually obvious (Fig. 1c). Anexample of a pair of duplicate features is indicated in thefigure. Based on the range of proteins that were detected, all ofthe protein display levels in this experiment were adequate forserum detection.

Screening patients and controls. Replicate slides containingduplicate features were screened with independently preparedserum samples from each of 38 patients with documented P.aeruginosa infections (22 CF and 16 non-CF patients with var-ious acute P. aeruginosa infections), as well as 15 healthy con-trols (Fig. 2a). Responses to the P. aeruginosa antigens weregenerally stronger among the CF patients than among non-CFpatients in antigen number, in signal intensity, and in the frac-tion of responders. The mean signal intensities for the fourfeatures for each antigen (duplicate spots from replicate slides)were calculated for each patient, and a Z-score was computedfor each antigen in each patient relative to the mean signalintensity of the slide [Z � (signal � mean)/standard deviation].An initial cutoff for positives was set at a Z-score of �1.6,

based on internal controls. The reproducibility for this quan-tification method is represented by the scatter plots of thesame serum samples tested on different arrays seen in Fig.2b and c. Samples are tightly grouped around the 45°-angleline that represents excellent reproducibility between repli-cate slides.

After this objective quantitative evaluation, slides were sub-mitted to visual review and a list of visual hits was obtained. Aheat map based on the Z-scores for this overlapping subset(Fig. 3a) shows the more intense response of CF patients. Asseen on the map and in Table S2 in the supplemental material,responses in healthy individuals were rare and weak. A map ofvisually confirmed hits is shown in Fig. 3b.

A total of 48 antigens were found to be positive in at leastone patient. Of these, 12 antigens were detected in 10 or morepatients, comprising a group of more promising candidates forfurther studies.

Confirmation of microarray results. All antigens that werepositive by both Z-score and visual review as well as subsets ofantigens positive by visual review only and Z-score only weretested by ELISA for confirmation, using sera from at least twodifferent patients per antigen. A large percentage of the hitsfor antigens positive by both Z-score and visual review (70%)was also positive by ELISA, whereas none of the antigens

FIG. 1. NAPPA quality control. (a) Expression clones encoding the target proteins fused to a C-terminal GST tag were printed along with apolyclonal anti-GST antibody in duplicate on the array surface. DNA capture was confirmed by PicoGreen (PG) staining (DNA), in situ proteinexpression and capture were assessed by GST detection using a monoclonal antibody (protein), and after incubation with serum, the response wasmeasured with a horseradish peroxidase-linked anti-human IgG secondary antibody (serum). (b) DNA-versus-protein scatter plot showingbackground negative control spots (blue) and expression clone spots (brown) on a logarithmic plot. All clone spots showed protein and DNA levelsat least three standard deviations higher than the mean background level (red lines). (c) Scatter plot comparing protein signal intensities (anti-GSTantibody) and serum signal intensities (anti-human IgG antibody) for data from panel a. Red features correspond to visually confirmed hits.Duplicate signals for one example of the hits are indicated by the black arrows on both the serum panel and the graph.

VOL. 77, 2009 P. AERUGINOSA ANTIGENS DETECTED BY MICROARRAY 4881

positive by Z-score only or antigens negative by both methodsdemonstrated a positive ELISA response (Fig. S1 in the sup-plemental material). Interestingly, 87% of the hits correspond-ing to antigens positive only by visual review were also positiveby ELISA, emphasizing the value of the visual review.

We also tested some of the antigens for response by immu-noblotting. As expected, the proteins that were negative byNAPPA were also negative by immunoblotting. Moreover, sixof the nine NAPPA-positive antigens tested were also positiveby immunoblotting. The remaining three proteins were negativeby immunoblotting but were unmistakably positive by NAPPA(Fig. 4), two of them being confirmed as positive by ELISA(PA1082 and PA4837). PA1302 was detected only by NAPPA. Itis possible that their microarray signal arises from conformationalepitopes displayed on the array but absent from the denaturedimmunoblots.

In silico analysis of the hits. P. aeruginosa genome annota-tion was used to determine if the 48 positive hits showed anytendency to cluster based on their structure, presence of spe-cific domains, protein function, or pathway involvement. Aftera thorough evaluation, it was not possible to find any commonfeature that drove subclustering of the hits (data not shown).Of course, these were all proteins preselected on the basis ofcommon features, which might make it difficult to subclusterthem.

Antigen prediction tools were also used to determine if thelist of hits would have been predicted in silico. Without struc-

tural data on the OMPs, we were limited to using tools thatpredicted linear epitopes from the primary structure. “Anti-genic” is one such tool in EMBOSS, predicting potentiallyantigenic regions of a protein sequence by using the method ofKolaskar and Tongaonkar (19, 30, 37). “BepiPred” is anothertool using a hidden Markov model and propensity scalemethod (22). Both of them can analyze protein sequences inbatch mode. The Antigenic tool predicted that all the OMPswould be antigenic, whereas BepiPred predicted 27 OMPs tobe immunogenic, only 2 of which were actually detected inthese experiments. A list of 117 annotated B-cell epitope se-quences for P. aeruginosa from the Immune Epitope Databasewas also assembled (31). The protein sequences of the 262OMPs were scanned, and 65 OMPs containing one or more ofthese epitopes were found; two of these OMPs overlapped withNAPPA hits. In conclusion, current antigen prediction toolscould not predict the observed NAPPA results.

DISCUSSION

The number of CF patients and the growing number ofimmunodeficient patients, especially due to human immuno-deficiency virus infection, draw attention to the importance ofopportunistic pathogens such as Pseudomonas aeruginosa. Be-cause the abundance of P. aeruginosa proteins changes accord-ing to the microenvironment of the infection (5, 35), empiricalstudies that provide a better understanding of the human hu-

FIG. 2. Serum screening. (a) A representative group of serum slides for controls and CF and non-CF (NCF) patients. (b and c) Reproducibilityscatter plots of the serum responses for independently processed samples (samples A and B) from patients CF-1 and NCF-1, as seen in panel a.All detected spots are represented in green, and the visually confirmed hits are highlighted with red circles. The R2 values for CF-1 and NCF-1were 0.85 and 0.80, respectively.

4882 MONTOR ET AL. INFECT. IMMUN.

moral response against this pathogen can contribute to im-proved diagnostics and specific vaccine development by high-lighting the P. aeruginosa proteins that are both expressed andexposed during infection in vivo.

Our objective was to select the proteins predicted to resideon the outer surface of the pathogen by using bioinformaticstools. A list of 266 candidate proteins was defined, 262 of whichwere successfully subcloned in the in vitro expression vector ofinterest, for NAPPA microarray expression. In addition tooffering a tool for unbiased representation for parallel study,the use of NAPPA circumvented some of the common prob-lems associated with the study of membrane proteins, namely,the difficulty in producing and purifying them, as well as theirinstability. As soon as the protein chains are synthesized, they

are captured and immobilized by the capture antibody, whichmight avoid aggregation. Moreover, the cell-free expressionsystem also avoids the host toxicity problems associated withoverloading the protein transportation machinery (14, 41). Al-together, this approach generated a uniform display of mem-brane proteins on the slides.

By displaying the proteins of interest in duplicate on a singleslide, experimental artifacts are minimized, as all features areexposed to the same serum sample at the same time.

Among the 262 OMPs and the controls tested, 48 antigenswere detected in at least one patient. Importantly, 12 anti-gens were positive in 10 or more patients (PA0044, PA5369,PA3407, PA3841, PA1080, PA0973, PA4922, PA4110,PA0807, PA3931, PA2300, and FliC_A), three of which were

FIG. 3. Microarray analysis. (a) Z-scores were computed from the averages of all four features for each antigen (duplicate features fromduplicate slides). Here, the Z-scores of the 48 visually confirmed positives were used to create a heat map. (b) A map displaying the visuallyconfirmed hits from panel a. The order of antigens in the heat map and the order in the chart are the same.

FIG. 4. Western blot confirmation. Proteins of interest were generated by IVTT and analyzed by Western blotting using patient serum as theprimary antibody. The numbers above the lanes correspond to antigens tested, and their NAPPA results for that specific patient are representedby a plus or minus sign. NAPPA images for some hits that were negative by Western blotting are shown. Proteins tested are numbered as follows:1, PA1080; 2, PA0044; 3, PA2373; 4, PA2130; 5, PA2760; 6, no DNA control; 7, PA3841; 8, PA0973; 9, PA0807; 10, PA2685; 11, PA4179; 12,PA1082; 13, PA1302; 14, PA1094; 15, PA4837. Molecular masses are given to the left of the gels.

VOL. 77, 2009 P. AERUGINOSA ANTIGENS DETECTED BY MICROARRAY 4883

frequently observed for both CF and non-CF patient groups(PA0044, PA3407, and PA3841). We found several proteinclasses known to be immunogenic, such as exoenzymes andflagellum-related proteins. We also found three hypotheticalproteins (PA5369, PA0807, and PA3931) that were among thetop 12 of our list. The strong differential responses for theseantigens in patients and controls support the notion that theprotein microarray can be used to detect immune responses tomembrane proteins, as can be seen by the significant P values(P � 0.01) in Table 1, representing the significance of NAPPAdifferential intensity detection between patients and controlsfor the top 12 hits. Moreover, the serum responses demon-strate that at least 48 of these antigens are clearly presented bythe pathogen during infection. These antigens can be used forfurther studies targeting vaccine development and diagnostics,particularly the promising top 12, which were detected by alarge number of patients.

There was a clear difference in the response patterns of CFand non-CF patients, considering both the number of antigensand the intensity of response. We attribute the stronger re-sponse of CF patients to the duration of infection, as thesepatients were chronically colonized since adolescence, whereasnon-CF patients presented more-recent infections. This idea iscorroborated by the observation that among the 11 non-CFpatients who responded to at least one of these antigens, 7either had COPD or were hospitalized in a long-term-carefacility (see Table S3 in the supplemental material). Thechronic nature of COPD can make the lung environment sim-ilar to that of a CF patient; however, a larger number of COPDpatients would be needed to confirm this apparent tendency.Both children and adults mounted an immune response to thepathogen, and the evaluation of data on disease severity(forced expiratory volume in 1 s, disease outcome, and bacte-rial count), when available, showed no obvious correlation withthe humoral response pattern. It is interesting to notice thatsome patients from both the CF group and the non-CF groupshowed no response to any of the antigens on the array. It isnot clear if the absence of responses is due to host or pathogenfactors, and the limited clinical data available for these dei-dentified samples make it impossible to address this here.Comparisons among the clinical and molecular data showed nocorrelation connecting specific antigens to prognosis.

Using NAPPA, we observed responses to most of the pre-viously known antigens, such as exotoxin A, flagellin, OprH2,and porin F, as well as the more recently described OprL(PA0973) and ExoS (PA3841). We also observed responses toa considerable list of new P. aeruginosa antigens, includinghypothetical proteins; however, we did not observe responsesto alkaline protease or elastase. Although CF patients aregenerally described as chronically infected, they undergophases of relative clearance, promoted by antibiotic treatment,during which the antibody titers to some pseudomonas anti-gens decrease significantly. This could explain some of thevariation in responses among different patients and why wemay have missed responses to some proteins. Also, it is knownthat responses to elastase and alkaline protease are weakerearly in infection, ranging from 20% prevalence to 38% prev-alence. Thus, given that these were all closely monitored pa-tients, the samples may have been obtained prior to the devel-opment of strong humoral response to these late-responding

proteins. Finally, as with any immunological method, NAPPAmay not be the ideal presentation method for all antigens.Nevertheless, it successfully found antigens that had not beendetected by other methods.

Of the three OMPs initially described to be immunogenic inthe 1980s (OprI, OprH2, and OprE), only responses to OprH2were found among our patients, but it is interesting to noticethat no other recent studies report immunogenicity for OprI orOprE after natural Pseudomonas infection. It is possible thatthe use of current refined-protein identification techniqueswould identify what was called OprE and OprI as differentOMPs. More-recent studies use OprI as an innate immunitytrigger for its ability to act on Toll-like receptors (12).

The success of a protein-based vaccine for P. aeruginosa willrely on demonstrating protective immunity from the selectedantigens. We regard this work as a first step toward identifyingantigens that may prove useful in vaccines, that is, the identi-fication of antigens that induce a humoral response in multiplepatients. In these deidentified samples, we do not knowwhether these humoral responses correlated with bacterialclearance or not. Both sets of patients were monitored closelyand were likely to have received antibiotics at the earliest signof infection. The approaches by which vaccines have proveduseful depend upon, among other factors, how the antigens areprepared, whether the antigens are engineered to enhanceresponses, which adjuvants are used, what vaccination sched-ule is used, and when the patients are vaccinated relative totheir encounter with the pathogen. The hope here is that avaccine would prepare individuals to mount a much moreprompt response during early colonization, when the bacterialload is still controllable.

Given the multiple strains of P. aeruginosa and the variationin antigen presentation by the pathogen in different infectiontypes, a multivalent vaccine seems a likely solution. The 12most frequently detected antigens here are good candidates forfurther evaluation, although we would not necessarily discardthe other hits. Some of the other hits may have correspondedto early responses that we missed but which nevertheless con-tribute to a protective effect.

All the top 12 antigens plus several of the others were con-firmed visually and by ELISA. Subsets of these may proveuseful for diagnostic applications. A test analyzing the re-sponses to PA5369, PA1080, PA3841, PA3407, and PA4837displays a sensitivity of 87%, specificity of 100%, positive pre-dictive value of 100%, and negative predictive value of 72% inthis limited group of patients. These values improve when CFis established as a prior condition, suggesting possible applica-tions for the early detection of P. aeruginosa infection in indi-viduals with CF (42). Although further validation studies withindependent samples would be needed to confirm these find-ings, these early results are promising.

Bioinformatics analysis of all positive antigens revealed nostructural or functional commonalities among the proteins thatacted as strong immunogens, and the antigen prediction toolsdid not predict which ones would show a strong humoral re-sponse. Thus, at the present, the experimental determinationof the humoral response by using high-throughput tools likeNAPPA may remain the only reliable method for assessing thehumoral response to a large repertoire of proteins produced bya bacterial pathogen.

4884 MONTOR ET AL. INFECT. IMMUN.

These promising antigens should now be submitted to ani-mal studies to evaluate protective properties, both individuallyand in groups, and also tested for diagnostic purposes in alarger population of patients and controls.

Here, we have presented a list of 48 antigens from Pseudo-monas aeruginosa which are clearly expressed during pulmo-nary infections and are potential candidates for diagnostics andvaccine development. As well, we demonstrated the appli-cability of a novel method for rapidly and comprehensivelytesting full-length bacterial OMPs for their ability to inducea humoral immune response in infected individuals. Thismethod exploits the ability of NAPPAs to efficiently produceand display membrane proteins. Because of the ease of dis-playing proteins in the NAPPA format and the minute volumesof serum needed (�10 �l), this method can be readily appliedto many other pathogens. It clearly demonstrates the useful-ness of the methodology for large-scale epidemiological sur-veys of human humoral responses, correlating clinical condi-tions of patients with the presence or lack of antibodies tospecific bacterial products, thus identifying potentially protec-tive antigens.

ACKNOWLEDGMENTS

We thank E. A. Mendoza and J. Raphael (Harvard Institute ofProteomics) for excellent technical support.

This work was supported by NIAID contract DHHSN266200400053C and by a grant from the Cystic Fibrosis Foundation.

REFERENCES

1. Anderson, K. S., N. Ramachandran, J. Wong, J. V. Raphael, E. Hainsworth,G. Demirkan, D. Cramer, D. Aronzon, F. S. Hodi, L. Harris, T. Logvinenko,and J. Labaer. 2008. Application of protein microarrays for multiplexeddetection of antibodies to tumor antigens in breast cancer. J. Proteome Res.7:1490–1499.

2. Beare, P. A., C. Chen, T. Bouman, J. Pablo, B. Unal, D. C. Cockrell, W. C.Brown, K. D. Barbian, S. F. Porcella, J. E. Samuel, P. L. Felgner, and R. A.Heinzen. 2008. Candidate antigens for Q fever serodiagnosis revealed byimmunoscreening of a Coxiella burnetii protein microarray. Clin. VaccineImmunol. 15:1771–1779.

3. Berven, F. S., K. Flikka, H. B. Jensen, and I. Eidhammer. 2004. BOMP: aprogram to predict integral beta-barrel outer membrane proteins encodedwithin genomes of Gram-negative bacteria. Nucleic Acids Res. 32:W394–W399.

4. Bigelow, H. R., D. S. Petrey, J. Liu, D. Przybylski, and B. Rost. 2004.Predicting transmembrane beta-barrels in proteomes. Nucleic Acids Res.32:2566–2577.

5. Bragonzi, A., D. Worlitzsch, G. B. Pier, P. Timpert, M. Ulrich, M. Hentzer,J. B. Andersen, M. Givskov, M. Conese, and G. Doring. 2005. NonmucoidPseudomonas aeruginosa expresses alginate in the lungs of patients withcystic fibrosis and in a mouse model. J. Infect. Dis. 192:410–419.

6. Diaz, E., E. Munoz, K. Agbaht, and J. Rello. 2007. Management of ventila-tor-associated pneumonia caused by multiresistant bacteria. Curr. Opin.Crit. Care 13:45–50.

7. Doring, G., C. Meisner, and M. Stern. 2007. A double-blind randomizedplacebo-controlled phase III study of a Pseudomonas aeruginosa flagellavaccine in cystic fibrosis patients. Proc. Natl. Acad. Sci. USA 104:11020–11025.

8. Doring, G., and G. B. Pier. 2008. Vaccines and immunotherapy againstPseudomonas aeruginosa. Vaccine 26:1011–1024.

9. Driscoll, J. A., S. L. Brody, and M. H. Kollef. 2007. The epidemiology,pathogenesis and treatment of Pseudomonas aeruginosa infections. Drugs67:351–368.

10. Elting, L. S., E. B. Rubenstein, K. V. Rolston, and G. P. Bodey. 1997.Outcomes of bacteremia in patients with cancer and neutropenia: observa-tions from two decades of epidemiological and clinical trials. Clin. Infect.Dis. 25:247–259.

11. Gardy, J. L., M. R. Laird, F. Chen, S. Rey, C. J. Walsh, M. Ester, and F. S.Brinkman. 2005. PSORTb v. 2.0: expanded prediction of bacterial proteinsubcellular localization and insights gained from comparative proteomeanalysis. Bioinformatics 21:617–623.

12. Gartner, T., M. Baeten, S. Otieno, H. Revets, P. De Baetselier, and K.Huygen. 2007. Mucosal prime-boost vaccination for tuberculosis based on

TLR triggering OprI lipoprotein from Pseudomonas aeruginosa fused tomycolyl-transferase Ag85A. Immunol. Lett. 111:26–35.

13. Gibson, R. L., J. L. Burns, and B. W. Ramsey. 2003. Pathophysiology andmanagement of pulmonary infections in cystic fibrosis. Am. J. Respir. Crit.Care Med. 168:918–951.

14. Grisshammer, R. 2006. Understanding recombinant expression of mem-brane proteins. Curr. Opin. Biotechnol. 17:337–340.

15. Hancock, R. E., E. C. Mouat, and D. P. Speert. 1984. Quantitation andidentification of antibodies to outer-membrane proteins of Pseudomonasaeruginosa in sera of patients with cystic fibrosis. J. Infect. Dis. 149:220–226.

16. Holder, I. A. 2004. Pseudomonas immunotherapy: a historical overview.Vaccine 22:831–839.

17. Keasey, S. L., K. E. Schmid, M. S. Lee, J. Meegan, P. Tomas, M. Minto, A. P.Tikhonov, B. Schweitzer, and R. G. Ulrich. 2009. Extensive antibody cross-reactivity among infectious gram-negative bacteria revealed by proteomemicroarray analysis. Mol. Cell. Proteomics 8:924–935.

18. Klinger, J. D., D. C. Straus, C. B. Hilton, and J. A. Bass. 1978. Antibodiesto proteases and exotoxin A of Pseudomonas aeruginosa in patients withcystic fibrosis: demonstration by radioimmunoassay. J. Infect. Dis. 138:49–58.

19. Kolaskar, A. S., and P. C. Tongaonkar. 1990. A semi-empirical method forprediction of antigenic determinants on protein antigens. FEBS Lett. 276:172–174.

20. Kubesch, P., B. U. von Specht, and B. Tummler. 1988. Immune response incystic fibrosis to outer membrane proteins of Pseudomonas aeruginosa.Zentralbl. Bakteriol. Mikrobiol. Hyg. A 269:395–410.

21. Labaer, J., Q. Qiu, A. Anumanthan, W. Mar, D. Zuo, T. V. Murthy, H.Taycher, A. Halleck, E. Hainsworth, S. Lory, and L. Brizuela. 2004. ThePseudomonas aeruginosa PA01 gene collection. Genome Res. 14:2190–2200.

22. Larsen, J. E., O. Lund, and M. Nielsen. 2006. Improved method for pre-dicting linear B-cell epitopes. Immunome Res. 2:2.

23. Laurans, M., A. Arion, M. Fines-Guyon, A. Regeasse, J. Brouard, R.Leclercq, and J. F. Duhamel. 2006. Pseudomonas aeruginosa and cysticfibrosis: first colonization to chronic infection. Arch. Pediatr. 13(Suppl. 1):S22–S29. (In French.)

24. Leonard, A., T. Leal, and P. Lebecque. 2007. Cystic fibrosis and pseudomonasaeruginosa current and future strategies. J. Pharm. Belg. 62:25–28. (In French.)

25. Li, B., D. Zhou, Z. Wang, Z. Song, H. Wang, M. Li, X. Dong, M. Wu, Z. Guo,and R. Yang. 2008. Antibody profiling in plague patients by protein microar-ray. Microbes Infect. 10:45–51.

26. Lu, Z., D. Szafron, R. Greiner, P. Lu, D. S. Wishart, B. Poulin, J. Anvik, C.Macdonell, and R. Eisner. 2004. Predicting subcellular localization of pro-teins using machine-learned classifiers. Bioinformatics 20:547–556.

27. Lyczak, J. B., C. L. Cannon, and G. B. Pier. 2002. Lung infections associatedwith cystic fibrosis. Clin. Microbiol. Rev. 15:194–222.

28. Murthy, T., A. Rolfs, Y. Hu, Z. Shi, J. Raphael, D. Moreira, F. Kelley, S.McCarron, D. Jepson, E. Taycher, D. Zuo, S. E. Mohr, M. Fernandez, L.Brizuela, and J. Labaer. 2007. A full-genomic sequence-verified protein-coding gene collection for Francisella tularensis. PLoS ONE 2:e577.

29. Obritsch, M. D., D. N. Fish, R. MacLaren, and R. Jung. 2005. Nosocomialinfections due to multidrug-resistant Pseudomonas aeruginosa: epidemiol-ogy and treatment options. Pharmacotherapy 25:1353–1364.

30. Parker, J. M., D. Guo, and R. S. Hodges. 1986. New hydrophilicity scalederived from high-performance liquid chromatography peptide retentiondata: correlation of predicted surface residues with antigenicity and X-ray-derived accessible sites. Biochemistry 25:5425–5432.

31. Peters, B., J. Sidney, P. Bourne, H. H. Bui, S. Buus, G. Doh, W. Fleri, M.Kronenberg, R. Kubo, O. Lund, D. Nemazee, J. V. Ponomarenko, M. Sathia-murthy, S. Schoenberger, S. Stewart, P. Surko, S. Way, S. Wilson, and A.Sette. 2005. The immune epitope database and analysis resource: from visionto blueprint. PLoS Biol. 3:e91.

32. Ramachandran, N., K. S. Anderson, J. Raphael, E. Hainsworth, S. Sibani,W. R. Montor, M. Pacek, J. Wong, M. Eljanne, M. G. Sanda, Y. Hu, T.Logvinenko, and J. Labaer. 2008. Tracking humoral responses using selfassembling protein microarrays. Proteomics Clin. Appl. 2:1518–1527.

33. Ramachandran, N., E. Hainsworth, B. Bhullar, S. Eisenstein, B. Rosen, A. Y.Lau, J. C. Walter, and J. LaBaer. 2004. Self-assembling protein microarrays.Science 305:86–90.

34. Ramachandran, N., J. V. Raphael, E. Hainsworth, G. Demirkan, M. G.Fuentes, A. Rolfs, Y. Hu, and J. Labaer. 2008. Next-generation high-densityself-assembling functional protein arrays. Nat. Methods 5:535–538.

35. Ramsey, D. M., and D. J. Wozniak. 2005. Understanding the control ofPseudomonas aeruginosa alginate synthesis and the prospects for manage-ment of chronic infections in cystic fibrosis. Mol. Microbiol. 56:309–322.

36. Ratjen, F., H. Walter, M. Haug, C. Meisner, H. Grasemann, and G. Doring.2007. Diagnostic value of serum antibodies in early Pseudomonas aeruginosainfection in cystic fibrosis patients. Pediatr. Pulmonol. 42:249–255.

37. Rice, P., I. Longden, and A. Bleasby. 2000. EMBOSS: the European Molec-ular Biology Open Software Suite. Trends Genet. 16:276–277.

38. Saiman, L. 2004. Microbiology of early CF lung disease. Paediatr. Respir.Rev. 5(Suppl. A):S367–S369.

39. Sundaresh, S., A. Randall, B. Unal, J. M. Petersen, J. T. Belisle, M. G.Hartley, M. Duffield, R. W. Titball, D. H. Davies, P. L. Felgner, and P. Baldi.

VOL. 77, 2009 P. AERUGINOSA ANTIGENS DETECTED BY MICROARRAY 4885

2007. From protein microarrays to diagnostic antigen discovery: a study ofthe pathogen Francisella tularensis. Bioinformatics 23:i508–i518.

40. Taycher, E., A. Rolfs, Y. Hu, D. Zuo, S. E. Mohr, J. Williamson, andJ. Labaer. 2007. A novel approach to sequence validating protein expressionclones with automated decision making. BMC Bioinformatics 8:198.

41. Wagner, S., M. L. Bader, D. Drew, and J. W. de Gier. 2006. Rationalizingmembrane protein overexpression. Trends Biotechnol. 24:364–371.

42. West, S. E., L. Zeng, B. L. Lee, M. R. Kosorok, A. Laxova, M. J. Rock, M. J.Splaingard, and P. M. Farrell. 2002. Respiratory infections with Pseudomo-nas aeruginosa in children with cystic fibrosis: early detection by serology andassessment of risk factors. JAMA 287:2958–2967.

43. Woods, D. E., S. J. Cryz, R. L. Friedman, and B. H. Iglewski. 1982. Contri-bution of toxin A and elastase to virulence of Pseudomonas aeruginosa inchronic lung infections of rats. Infect. Immun. 36:1223–1228.

44. Xu, Y., J. F. Bruno, and B. J. Luft. 2008. Profiling the humoral immuneresponse to Borrelia burgdorferi infection with protein microarrays. Microb.Pathog. 45:403–407.

45. Zhu, H., S. Hu, G. Jona, X. Zhu, N. Kreiswirth, B. M. Willey, T. Mazzulli, G.Liu, Q. Song, P. Chen, M. Cameron, A. Tyler, J. Wang, J. Wen, W. Chen, S.Compton, and M. Snyder. 2006. Severe acute respiratory syndrome diagnos-tics using a coronavirus protein microarray. Proc. Natl. Acad. Sci. USA103:4011–4016.

Editor: J. N. Weiser

4886 MONTOR ET AL. INFECT. IMMUN.