ARTICLE pubs.acs.org/jpr
Toward Protein Biomarkers for Allergy: CD4þ T Cell Proteomics in Allergic and Nonallergic Subjects Sampled in and out of Pollen Season Martin Bl€uggel,†,‡ Franc- ois Spertini,†,§ Petra Lutter,†,‡,|| Jacqueline Wassenberg,§ Regine Audran,^ Blaise Corthesy,^ Stefan M€ullner,‡ Stephanie Blum,# Andreas Wattenberg,‡ Annick Mercenier,z Michael Affolter,0 and Martin Kussmann*,0,2,£ ‡
Protagen AG, Otto-Hahn Strasse 15, 44227 Dortmund, Germany Service d'Immunologie et d'Allergie, Centre Hospitalier Universitaire Vaudois (CHUV-BH19), Rue du Bugnon, 1011 Lausanne, Switzerland ^ Division of Immunology and Allergy, H^opital Orthopedique, Avenue Pierre Decker 4, 1005 Lausanne, Switzerland z Allergy Group, Department of Nutrition and Health, Nestle Research Center, Vers-chez-les-Blanc, 1000 Lausanne 26, Switzerland # Immunology Group, Department of Nutrition and Health, Nestle Research Center, Vers-chez-les-Blanc, 1000 Lausanne 26, Switzerland 0 Functional Genomics Group, Department of BioAnalytical Sciences, Nestle Research Center, Vers-chez-les-Blanc, 1000 Lausanne 26, Switzerland 2 Faculty of Science, Aarhus University, Ny Munkegade, 8000 Aarhus C, Denmark §
bS Supporting Information ABSTRACT: Allergy is an immunological disorder of the upper airways, lung, skin, and the gut with a growing prevalence over the last decades in Western countries. Atopy, the genetic predisposition for allergy, is strongly dependent on familial inheritance and environmental factors. These observations call for predictive markers of progression from atopy to allergy, a prerequisite to any active intervention in neonates and children (prophylactic interventions/primary prevention) or in adults (immunomodulatory interventions/secondary prevention). In an attempt to identify early biomarkers of the “atopic march” using minimally invasive sampling, CD4þ T cells from 20 adult volunteers (10 healthy and 10 with respiratory allergies) were isolated and quantitatively analyzed and their proteomes were compared in and out of pollen season (( antigen exposure). The proteome study based on high-resolution 2D gel electrophoresis revealed three candidate protein markers that distinguish the CD4þ T cell proteomes of normal from allergic individuals when sampled out of pollen season, namely Talin 1, Nipsnap homologue 3A, and Glutamate-cysteine ligase regulatory protein. Three proteins were found differentially expressed between the CD4þ T cell proteomes of normal and allergic subjects when sampled during pollen season: carbonyl reductase, glutathione Stransferase ω 1, and 2,4-dienoyl-CoA reductase. The results were partly validated by Western blotting. KEYWORDS: allergy, biomarker, proteomics, CD4þ T cell, PBMC
’ INTRODUCTION Definition of Allergy
Allergy is an immunological disorder, which manifests in the upper airways, lung, skin and the gut and is understood as the exacerbated, IgE-mediated response of an organism toward an allergen.1 Atopy means in more general terms the predisposition for the development of allergic symptoms.2 Allergens can be classified according to their ways of interacting with the host: airborne allergens invade the respiratory system, food allergens are taken up by the gastrointestinal tract (GIT) and contact allergens act through the skin. The way of invasion does not necessarily correspond to the locus of allergy manifestation: some food allergens can, r 2011 American Chemical Society
for example, provoke allergic reactions in the respiratory tract.1 Allergy is mainly governed by T helper 2 (Th2) cells, which express the interleukins (ILs) 4, 5, 9, 10, and 13. T-regulatory cells (Tregs), secreting the anti-inflammatory cytokines IL-10 and TGF-β, control the balance between Th1- and Th2 cells and regulate in this way the specific allergen response to maintain normal immunity.3 Allergy Markers and Mechanisms
Most of the to-date identified twenty or more allergy-associated genetic markers are rather indicators of inflammation than Received: September 15, 2010 Published: February 28, 2011 1558
dx.doi.org/10.1021/pr100939g | J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research of allergy. Important allergy markers currently accepted are IL10,4 TGF-β, TLRs,5 PD-1 and CTLA-4.6 Yet, specific IgE antibody levels are successfully used as indicators for an existing allergic condition. Roughly 150 genes are suspected to be linked with the multiple phenomena of the three allergic diseases atopic dermatitis, hay fever and asthma.7 Today, only a few gene-trait links for allergy susceptibility and predisposition are established, one of which is the identification of a susceptibility locus for asthma-related traits on chromosome seven revealed by a genome-wide scan in a Finnish founder population.8 In contrast to the advanced understanding of allergen structures, the molecular mechanisms leading to a normal versus an allergic phenotype remain partly elusive. Prediction of allergy risk and onset is mainly based on family history data. Their impacts on individual disposition and susceptibility, genetic and environmental influences are difficult to dissect. The environmental imprinting as a counter-player of the genetic determination is most important during the perinatal period until weaning and in early childhood. PBMCs for Allergy Research
Circulating leukocytes (or peripheral blood mononuclear cells, PBMCs) are potential targets for proteomic studies of an individual’s immune status.9 They are available in large amounts from healthy and diseased subjects, can be harvested by minimally invasive means and cultured under defined conditions. Moreover and importantly, PBMCs have a normal active metabolism.10 Differential proteomics of PBMCs require a sufficient number of biological and technical replicates in order to understand the pronounced and meaningful interdonor variability in protein profiles and discern it from the experimental variations.11 A 2D gel database of PBMC proteins has recently been established.12 Jeong et al. assessed the peripheral T-lymphocyte proteome in patients with asthma by 2DE and MALDI-TOF-MS.13 The proteins of CD3þ T-lymphocytes were isolated from whole blood of six steroid-naive asthmatic patients and of six healthy volunteers. Messenger RNA levels of some proteins found differentially expressed were examined by real-time PCR. Thirteen protein spots in the T-lymphocytes of the asthmatic patients were increased and 12 spots were decreased compared to those of the normal subjects; phosphodiesterase 4C and thioredoxin-2 were found increased and glutathione S-transferase M3 was found decreased in the asthmatic patients, all both at protein and mRNA level. The aim of the present study was to identify allergy associated biomarker candidates and to contribute to a better understanding of the mechanism and the onset of allergy. This is to our knowledge the first proteomic study that compares human immune cell proteome patterns before and during pollen season. Besides the between-group comparison, the interseason differences were also addressed by comparing the CD4þ T cell proteomes of the same group of individuals once sampled in winter, that is, out of pollen season and in spring, that is, during grass pollen exposure. To accomplish this undertaking, a total of 120 high resolution and large-format (30 40 cm, pI 211) two-dimensional gels (2 10 subjects 4 conditions 3 technical replicates) were performed and the protein patterns were visualized by silver staining. Approximately 2200 protein spots were detected per gel, quantification results of ∼1200 of them were considered and visually controlled across all biological and technical replicates; 697 protein spots were detected in at least 2/3 of gel replicates and 80% of subjects per condition.
ARTICLE
The analysis of this revealed three candidate protein markers that distinguished the CD4þ T cell proteome signature of allergic from nonallergic subjects on-season (Carbonyl reductase; Glutathione S-transferase ω 1; 2,4-Dienoyl-CoA reductase, the latter not t test confirmed) and another three candidate markers that were differentially expressed between CD4þ T cell proteomes of allergic and nonallergic subjects sampled off-season (Talin 1; Nipsnap homologue 3A; Glutamate-cysteine ligase regulatory protein).
’ EXPERIMENTAL PROCEDURES Clinical Part
Study Protocol and Population. The study was designed as an open, non interventional, two-group trial. Twenty female Caucasians (age 18 to 35), 10 with a history of grass polleninduced rhinitis and for two among these 10 with asthma (allergic group) and 10 without known history of allergy or atopy (nonallergic subjects), were enrolled in this single center study. Only females were recruited to minimize potential deviations in the immune response linked to gender. This was taken as a measure of caution although, to our knowledge, to date no systematic differences between males and females have been reported at the level of the immune response. Allergy diagnosis was based on clinical history of rhinitis and/ or asthma to grass pollen during the last pollen season, as well as on the presence of positive skin prick tests (wheal diameter >3 mm) and/or positive antigrass pollen specific IgE (>0.35 kU/L) as titrated by UniCAP 100 (Pharmacia Diagnostics, Uppsala, Sweden). For the “nonallergic group”, volunteers were included based on the absence of personal and familial history of respiratory or food allergies confirmed by negative skin prick tests and negative IgE for grass and tree pollen (birch, hazel and ash). Two-hundred milliliters of whole blood was collected from each volunteer, the first time out of the pollen season, during the winter 200607 (December and January), and the second time during May 2007 (grass pollen season). Volunteers were excluded if any of the following criteria were present: (1) Anemia at the screening visit, detected by a whole blood count; (2) primary or secondary immunodeficiency or treatment with immunosuppressive drugs; (3) any severe medical condition that could influence the study; (4) participation in another clinical trial at the same time; (5) pregnant or breastfeeding women. The clinical research protocol was approved by the Ethical Review Board of the Faculty of Biology and Medicine, Lausanne. Written informed consent was obtained from each volunteer before the start of the study. CD4þ T Cell Sampling. Heparinized peripheral blood was collected from volunteers out of and during pollen season. PBMCs were separated from blood samples by Ficoll Paque density gradient centrifugation followed by elimination of platelets via centrifugation for 10 min at 230 g and 4 C and lysis of the remaining red blood cells by incubation with 3 volumes of a 0.015 M NH4Cl, 103 M KHCO3 solution per 2 volumes of cell suspension. CD4þ T cells were negatively enriched from 2 108 PBMC by three successive magnetic depletions using the MagCellect Human CD4þ TCell Isolation kit and according to manufacturer’s instructions (R&D Systems, Abingdon, U.K.). CD4þ T cells were washed, counted, tested for viability, pelleted in three identical aliquots, weighted, immediately frozen in liquid nitrogen and stored at 80 C. The cell purity was assessed by 1559
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research FACS analysis using anti-CD4-FITC and anti-CD3-PE staining (BD Biosciences). The resulting cell population was not further divided into CD25þ and CD25 cells. Proteomic Part
All samples submitted to the subsequent proteomic analysis were blinded before shipping. CD4þ T Cell Preparation for Proteome Analysis. The cells of each volunteer were received in triplicates. For each sample vial, the cells were lysed by ultrasonication in the presence of protease inhibitors (mixture of Complete, PMSF, and pepstatin) and CHAPS in phosphate buffer as described in Klose14 and Weingarten et al.15 For the isoelectric focusing (IEF, first separation dimension) of the proteins, urea and thiourea were added to the sample to a final concentration of 7 M and 2M, respectively. To denature the proteins, 65 mM DTT was added. The homogenate was centrifuged for 30 min at 226 000 g at 25 C. The resulting supernatant was used for proteome comparison. Protein Quantification. The protein amount of each sample prior to 2D gel electrophoresis was determined in duplicates using the amido black dye binding method modified according to Popov et al.16 and Schaffner et al.17 Each sample was then supplemented with carrier ampholytes 24 resulting in a final concentration of 2%. 2D Gel Electrophoresis Analytical Gels for Image Analysis. The cells of each volunteer were received in triplicates. Three independent sample preparations were separated on one 2D gel each to ensure the quality and reproducibility of the analysis. Each sample was lysed, individually prepared for 2D PAGE and then separated on one 2D gel, each. IEF was performed according to a method of Klose14 with modifications as described in Weingarten et al.15 using 40 cm rod gels containing 9 M urea, 3.5% acrylamide, 0.3% piperazine diacrylamide and a total of 4% carrier ampholytes pH 211. For analytical gels, 80 μg of protein were applied onto the IEF gels at the anodic side of the tube gels. The proteins were focused under nonequilibrium pH gradient electrophoresis conditions (NEPHGE) for 23 816 Vh. After IEF the tube gels were gently pushed out of the glass tube and equilibrated in 125 mM Tris-Base, 40% (v/v) glycerol, 3% (w/v) SDS, 65 mM DTT, pH 6.8 for 10 min. Afterward, the equilibration buffer was removed and the gel tubes were stored at 80 C. The IEF gels were sliced in halves and each half was applied onto SDS gels of 0.75 250 300 mm3 containing 15% acrylamide and 0.2% bis-acrylamide using the IEF gels as stacking gels. The proteins were separated according to their apparent molecular weight in a continuous buffer system (25 mM Tris, 192 mM glycine, and 0.1% SDS). We used a validated method established also for GMP analytics of protein therapeutics. In-house tests had demonstrated that (most likely batch-to-batch) variability of IPG was significantly higher compared to our in-house developed Ampholine technique under full SOP embedded in the QM system audited by a local authority. Furthermore, the use of three technical replicates allowed compensating for spots affected by gel cutting. IEF rod halves were not cut at exactly the same place allowing for a minimum of two out of three spot replicates at the rod cutting edge to be used for quantitative analysis per sample. Chambers for SDS gels were designed to separate proteins from two SDS gels. This allowed running corresponding gels per IEF rod under exactly the same conditions with negligible variability along the molecular weight axis.
ARTICLE
Preparative Gels for Protein Identification. For preparative gels, 160 μg of protein were applied to the IEF tube gel. For mass spectrometric protein identification, four representative 2D gels were prepared as described above, except for the thickness of the gels: SDS-PAGE for protein identification was performed in 1.5 250 300 mm3 gels. The samples showing the highest abundance of the differentially expressed proteins were selected for protein identification. Protein Staining, Gel Image Analysis and Protein Spot Statistics Analytical Gels for Image Analysis. The 2D separated proteins were stained with silver using a modified method according to Heukeshoven.18 For quantitative comparison the 2D gels were digitized and analyzed with the Proteomweaver 3.0 software (BioRad Laboratories Inc.). The determination of the total spot number, spot quantification and normalization were performed with the software-implemented algorithms. Protein detection and spot matching were controlled visually. Protein spots with a saturated stain were selected manually and compared visually. Spots showing a standard deviation of more than 66% within one subject group and season (gel triplicates), were excluded from further analysis. If visual inspection of such a spot revealed an experimental errorderived issue (e.g., local streaking or spots split during cutting of the IEF gels), only this outlier was excluded with the two remaining spots still included in the analysis. If no experimental error was found for such a deviating spot, the latter was completely excluded from further analysis. Applying these spot selection criteria, a total of 697 well separated protein gel spots were finally selected for quantitative spot comparison. The spot intensities were averaged over the gel triplicates per biological sample. The mean spot intensities and coefficients of variation for both subject groups were calculated and the mean ratios between groups (allergic vs nonallergic) were determined. The high quality of all gels allowed considering spots with an average intensity ratio g1.5 or e0.67 as potentially different between both groups. These significance limits were determined using the “Replicate Quality Test” function of Proteomweaver software with a minimum confidence level of p = 0.05, and assuming a double-sided regulation. The average significant regulation factor was confirmed by a prestudy to assess workflow reproducibility (see below) and therefore applied during gel analysis. However, only statistically confirmed changes were considered. The significance of differential spot intensities was assessed for all protein spots by multiple statistical tests. First, statistics were analyzed by a conventional Student’s t test with and without Bonferroni correction using MS Excel. For the Bonferroni correction, the significance levels were recalculated for the prespecified p values (p = 0.05, p = 0.01, and p = 0.001). Critical t values were calculated with and without Bonferroni correction for f = 18 degrees of freedom for N = 20 samples (Table 1). For samples with less than 18 degrees of freedom, the corresponding t values were used.19 Additionally, the data were evaluated by the software SAM (significance analysis of microarray data),20 student’s t testing and nonparametric Wilcoxon signed-rank test.21 Preparative Gels for Protein Identification. For mass spectrometric protein identification a total of four preparative 2D gels was prepared using representative samples with respect to high intensities of the spots of interest. The separated proteins were stained with silver using a modified method according to Blum et al.22 Otherwise, the preparative gels were processed as the analytical ones. 1560
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research
ARTICLE
Table 1. Critical t Values Used for Student’s t-test critical t values t (p = 0.05; f = 18)
2.101
t (p = 0.01; f = 18)
2.878
t (p = 0.001; f = 18)
3.922
tBonferroni (p = 0.000071; f = 18)
5.122
tBonferroni (p = 0.000014; f = 18)
5.886
tBonferroni (p = 0.0000014; f = 18)
7.045
Western Blotting. Samples as prepared under CD4þ T cell preparation for proteome analysis were supplemented with 4 NuPAGE LDS sample buffer and proteins were denatured for 4 min at 72 C. Afterward samples were separated on NuPAGE Novex 412% Bis-Tris gels using NuPAGE MES SDS running buffer (all Invitrogen). After protein separation, gels were incubated in transfer buffer (0.6% (w/v) Tris, 2.25% (w/v) glycin, 0.25% (w/v) SDS, 10% (v/v) methanol) and proteins transferred by semidry-blotmethod (Multiphor II Nova Blot, GE Healthcare) onto PVDF membrane (Immobilon PSQ, 0.2 μm, Millipor) for 1 h at 2.5 mA/ cm2. Efficiency of the electro-transfer was controlled by silver staining of the gels using a modified method according to Heukeshoven et al.18 The PVDF membrane was incubated in 2% (w/v) BSA in TBS-T (20 mM Tris, 0.5 M NaCl, 0.1% (v/v) Tween 20) for 12 h at room temperature and followed by incubation with the primary antibodies (anti-NIPSNAP3A rabbit IgG (1 μg/mL, Abcam); anti-CBR1 rabbit IgG (2.5 μg/mL, Sigma-Aldrich); antiGCLM mouse IgG, Serotec; anti-GSTΩ, mouse IgG, Serotec; antitalin 1 mouse IgG, Abcam; and antitubulin rabbit IgG, Abcam in 1% BSA in TBS-T) at 5 C overnight. After a washing step, the membrane was incubated with the respective HRP-conjugated secondary antibodies (ECL antirabbit IgG, HRP linked, diluted 1:100 000 in 1% (w/v) BSA in TBS-T) at room temperature for further 12 h. Detection was performed with ECL Plus solution (GE Healthcare) on X-ray films (Hyperfilm ECL, GE Healthcare). Image analysis of the digitized films was performed using ImageQuantTL software (GE Healthcare). Mass Spectrometric Protein Identification. Protein identification was performed manually. The selected protein spots were excised from the preparative gels as duplicates and destained. After drying of the gel pieces, modified porcine trypsin (66 ng, Promega) was added to digest the proteins. The peptides were extracted from the gel, purified on C18 RP material (ZipTip μC18, Millipore) and spotted onto a MALDI sample target (Stainless steel target 384 positions, Bruker Daltonics). Peptide mass fingerprint spectra (PMF, MS) and peptide fragmentation spectra (PFF, MS/MS) were acquired on an Ultraflex III TOF/ TOF mass spectrometer (Bruker Daltonics). Spectra were analyzed on the Proteinscape 1.3 Bioinformatics platform (Bruker Daltonics, Protagen AG). For PMF searches MASCOT and ProFound were utilized. For PFF searches MASCOT, Sequest (SequestMetaScore combined score of Xcorr and deltaCn as implemented into Proteinscape) and Protein Solver (in-house developed algorithm) were utilized. The acquired peptide masses were used for database searches against the NCBI database (mass deviation tolerances: MS 50 ppm, MS/MS data 0.2 Da parent, 1 Da fragment ion; accepted modifications for all ions: variable carboxamidomethyl modification and methionine oxidation, 1 missed cleavage). The sequence database used originally was the NCBInr database version from 2007-11-09; the searches were repeated and confirmed using the NCBInr database version from
2010-10-09. Detailed protein identification data are given in Supplemental Table 1 (Supporting Information). Prestudy to Assess Workflow Reproducibility. To assess the reproducibility of the entire workflow, i.e. isolation of CD4þ T cells, sample shipping, sample preparation and 2D gel electrophoresis, the proteomes of three samples of independently isolated human primary CD4þ T cells were displayed on large format 2D gels in triplicates. As a first step, the reproducibility of sample generation and analysis was controlled. A representative false-color overlay of a pairwise comparison of two independently prepared off-season samples of a nonallergic (orange) and an allergic (blue) subject is displayed in Supplemental Figure 1 (Supporting Information). The high reproducibility and similarity of the gels is demonstrated by the high number of black spots, which are identical in spot intensity and localization on the gels of both overlaid samples. A similar reproducibility was achieved for the on-season samples (data not shown). Spots that appear as blue or orange spots differ in their spot intensities as a result of subject heterogeneity between the two samples shown in Supplemental Figure 1. Approximately 2200 protein spots were detected on each gel. The reproducibility of the entire process was quantitatively analyzed on the basis of the 304 most intensely stained spots. The mean standard deviation of the spot intensities for these 304 most intense protein spots within the gel triplicates equaled 17%. The high quality of the gels allowed considering an average intensity ratio g1.5 or e0.67 as significance limit for differences. Three spots fell outside the defined intensity variability tolerance. Two of these three spots were suspected as hemoglobin, probably introduced into the sample during preparation. For another third spot, the average intensity in Sample 1 was lower than in Sample 2 and 3. The excellent reproducibility of the entire procedure including sample preparation and 2D gel electrophoresis qualified for a differential proteome analysis of human CD4 þ T cells.
’ RESULTS Demographic Data
Twenty female volunteers were enrolled into the study. No one was excluded during the trial. All subjects were Caucasian and 35% were smokers (30% in the allergic group and 40% in the nonallergic group). Subjects were between 19 and 35 years old (mean age 23.3 years). CD4þ T Cell Proteome Comparison between Groups and Seasons
The workflow for the proteome comparison of human CD4þ T cells derived from nonallergic (group A) and allergic adult volunteers (group B) in and out of pollen season is outlined in Figure 1. At each season, three cell samples were taken from each subject and one 2D gel was run per sample, resulting in a total of three gels per subject, 30 gels per group, 60 gels per season and 120 gels for the whole study. Each sample was assigned to one of both groups (A or B) but the affiliation of the groups to allergic or nonallergic subjects was not disclosed during the analysis of differential protein expression. The on-season and off-season samples were analyzed separately. Afterward, the influence of the pollen exposure was analyzed for all subjects together and for each donor individually. Common expression changes over all individuals per group as well as the differentiation of the responses between both groups were analyzed. Trends were considered as significant if they were 1561
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research
ARTICLE
Figure 1. Workflow for the proteome comparison of CD4þ T cells derived from allergic and nonallergic subjects in and out of pollen season. The blood cells collected once from each subject were divided into three aliquots. From each such aliquot one 2D gel was produced, i.e., three gels per subject. The resulting 120 gels from both subject groups and seasons (2 groups 10 subjects 2 conditions 3 replicates) were compared.
found in at least 7 out of 10 subjects per group and confirmed by t testing. For the comparison of the effect of the pollen exposure on protein spot intensities, the ratios off/on season were calculated for each subject individually. The differences between group A and B were analyzed by dependent t testing for matched-paired samples. Spots with t test significance (conventional and SAM) and a mean fold-change of the ratios g ( 1.5-fold were reported. The numbers of individuals with expression ratios exceeding a (2-fold change were counted group-wise and compared. Spots homogenously up- or down-regulated within one group, but unchanged in the other in at least 5 out of 10 subjects per group as well as spots regulated heterogeneously in both directions in at least 7 out of 10 subjects per group were reported. For the quantitative comparison of spot intensities, the protein spots were inspected with regard to outline, saturation, streaking and mismatches. Spots with saturated staining cannot be reliably quantified and were hence excluded from the selection. The quantitative comparison was controlled visually for 1200 spots detected in each gel. Totally, 697 protein spots per gel were detected in at least 66% of the gel replicates and 80% of the subjects per group. These spots were selected for differential proteome analysis. The protein spots of the entire off-season study were distributed into those with consistent protein spot intensities over all subjects and both groups and into spots with minor or major interindividual heterogeneity. For selected protein spots, representatives of these three categories, corresponding histograms are displayed in Supplementary Figures 2ac (Supporting Information). These represent spots with low (Supplemental
Figure 2a), moderate (Supplemental Figure 2b) and high (Supplemental Figure 2c) subject heterogeneity while their mean intensity distribution did not show any difference between the nonallergic and the allergic group. For the CD4þ T cell proteome comparison between groups and within season, the significance of the results was assessed by different statistical methods. In a first step, all matched spot intensities of the technical replicates were averaged for each subject of both groups. The mean spot intensities and the resulting ratios between the two groups were then compared. A scatter plot demonstrated the homogeneity of the CD4þ T cell proteomes over all subjects and both groups based on 697 gel protein spots (data not shown). Figure 2 gives an overview of the distribution of average spot intensity ratios, binned into fold-change intervals: overall, 696 spots, that is, all spots submitted to this analysis except one, are located within the 2-fold change interval and 16 protein spots were found outside the 067/1.5 ratio interval and hence considered subtly but significantly regulated between the two groups. The figure summarizes the high reproducibility of the 2D gel study and few subtle but robust protein changes. The significance of differences between the two groups was first assessed by a conventional Student’s t test. Then, to reduce falsely significant results of this t test, a Bonferroni correction was performed with stricter p values. Table 2 lists the spots with average intensity ratios between 1.5 and 2.0 and between 0.5 and 0.67. The t test significance levels with and without Bonferroni correction are given. All protein spots found with average intensity ratios between 1.5 and 2.0 and between 0.5 and 0.67 and conventional t test significance revealed moderate to high interindividual heterogeneity. 1562
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research
ARTICLE
Figure 2. Distribution of average ratios comparing the mean spot intensities of the allergic group over the nonallergic group. The widths of the foldchange bins are indicated. From a total of 696 protein spots, a total of 16 spots were located outside the 0.67 to 1.5 ratio interval. The mean average ratio was calculated as 1.00.
Table 2. Averaged Protein Spot Intensity Ratios with t-Test Significance Levels With and Without Bonferroni Correctiona SAM significance q value [% FDR]
average ratio spot number
g2
g1.5
e 0.67
t-test significance
t-test significance after Bonferroni
t-test-based
Wilcoxon-based
404520
0.53
0.001
n.s.
0.0
24.69
404641
0.64
0.01
n.s.
n.s.
24.69
0.05
n.s.
n.s.
n.s.
383960
0.48
404412
0.58
0.05
n.s.
n.s.
n.s.
405306
0.58
0.05
n.s.
n.s.
n.s.
405097 404335
0.63 0.54
0.05 n.s.
n.s. n.s.
n.s. n.s.
n.s. n.s.
404574
0.57
n.s.
n.s.
n.s.
n.s.
403860
0.61
n.s.
n.s.
n.s.
n.s.
405387
0.63
n.s.
n.s.
n.s.
n.s.
404481
0.64
n.s.
n.s.
n.s.
n.s.
404678
0.64
n.s.
n.s.
n.s.
n.s.
405036
0.65
n.s.
n.s.
n.s.
n.s.
405632 404391
0.66 0.67
n.s. n.s.
n.s. n.s.
n.s. n.s.
n.s. n.s.
0.67
384022
n.s.
n.s.
n.s.
n.s.
383565
1.51
n.s.
n.s.
n.s.
n.s.
383603
1.51
n.s.
n.s.
n.s.
n.s.
6
0
1
2
Total Spots a
e 0.5
0
2
15
1
SAM q value is the lowest false discovery rate (FDR) (%), at which the spot is considered significant; n.s. = not significant.
For each of those spots, the intragroup variability was comparable for both the nonallergic and the allergic group. The spots with t test significance of p = 0.05 (see Table 3) showed high subject heterogeneity. For a more reliable assessment of these four spots with regard to being different in intensity between allergic and nonallergic individuals, more subjects should be analyzed.
In the second data processing phase, we applied additional statistical tests. Especially for statistical analysis of data sets derived from proteome (or transcriptome) comparisons, the SAM software can be deployed as it assigns a score to each protein spot on the basis of change in spot intensities relative to the standard deviation of repeated measurements of that protein. Proteins with scores greater than a threshold are deemed potentially differentially 1563
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research
ARTICLE
Table 3. Identified Proteins Differentially Expressed between Nonallergic and Allergic Individuals, Either In or Out of Pollen Season, with NCBInr Database Accession Numbers spot name
sample ID
accession
protein
regulation
glutathione S-transferase omega 1 [Homo sapiens]
Group B/A on season: 0.65
gi|4502599
carbonyl reductase 1 [Homo sapiens]
Group B/A on season: 0.66
gi|56554353
A Chain A, Structure Of Human Decr Ternary Complex
gi|4504011
glutamate-cysteine ligase regulatory protein [Homo sapiens]
959140
0511
gi|55957303
958889
05116 0512 05118 05115
(Decr =2,4-dienoyl-CoA reductase) 404412
05117 404520
0514
Group B/A off season: 0.58
not identified gi|22267436
nipsnap homologue 3A [Homo sapiens]
Group B/A off season: 0.53
gi|55859708
talin 1 [Homo sapiens]
Group B/A off season: 0.48
0519 383960 404641
0513 05110 0518
not identified
Group B/A off season: 0.64
not identified
Group B/A off season: 0.58
not identified
Group B/A off season: 0.63
not identified
Ratiooff/on season Group B/A: 1.42
0515 405306
0516 0517
405097
05113 05114
1019872
05111 05112
expressed. The percentage of such proteins identified by chance is the false discovery rate (FDR). Spot 404520 was identified as different between groups in both analyses based on t testing and Wilcoxon rank sum testing. Additionally, the Wilcoxon test confirms spot 404 641 as different between groups. The resulting FDRs are listed in Table 2. Hierarchical single linkage clustering (nearest neighbor technique) and complete linkage clustering (farthest neighbor technique) were recruited to compare the two groups off season as well as the two on season groups. The cluster analyses did not yield any meaningful differentiation between the samples and therefore indicate high similarity of CD4þ T cell proteomes of allergic and nonallergic subjects when sampled out of or during pollen season, respectively. Identification of Proteins Differentially Expressed in CD4þ T Cell Proteomes between Groups and Seasons
A total of nine spots regulated between the CD4þ T cells from nonallergic and allergic adult donors within a given season (either on or off) were analyzed by MALDI tandem mass spectrometry. These nine spots were picked as duplicates from a total of four high-resolution 2D gels generated from different samples. The abundance of most of these spots was very low as indicated by faint staining. For each differentially stained protein spot PMF (peptide mass fingerprint) and, if possible, PFF (peptide fragment fingerprint) spectra were acquired and analyzed. Detailed protein identification data are given in Supplemental Table 1 (Supporting Information). Five different proteins could be identified in four of the nine spots: glutathione S-transferase omega 1, nipsnap homologue 3A, and talin 1 were identified by PMF and PFF in 2 out of 2 replicate gels; carbonyl reductase 1 and glutamate-cysteine ligase regulatory protein were identified by PMF and PFF in 1 out of 2 replicate gels; and Decr Ternary Complex was identified by PMF only, but with 12 matching peptides with a maximum deviation of (32 ppm from the theoretical mass (see Supplemental Table 1). The identified
proteins reported with the NCBInr database accessions are listed in Table 3. Figure 3 shows the proteins found differentially expressed between nonallergic and allergic subjects, either during or out of pollen season, as marked spots on a representative 2D gel (from an allergic subject off season) and analyzed by MALDI mass spectrometry, with the spot numbers and the protein names (see also Tables 3 and 4) directly annotated near the protein spots. Proteins Differentially Expressed between Groups OffSeason. The comparison of CD4þ T cell proteomes between nonallergic (Group A) and allergic subjects (Group B) revealed six spots with spot intensity ratios outside the 1.5-fold or 0.67fold ratio range as assessed by conventional t test significance (Table 3). Differential expression and statistical significance of these six proteins are summarized in Table 4. Among these six spots, three could be identified and corresponded to Nipsnap homologue 3A, Talin 1 and Glutamate-cysteine ligase regulatory protein and these latter three are addressed in more detail in the Discussion of this paper. All six proteins, identified or not, were found at lower levels in the allergic group (B) compared to the nonallergic group (A). The protein spot of Nipsnap homologue 3A was found differentially stained with a t test significance of p = 0.001, the not identified spot 404641 with p = 0.01, and the remaining four spots, including Talin 1 and Glutamate-cystein ligase regulatory protein, were found differentially displayed with p = 0.05. Cluster analyses did not yield any meaningful off-season differentiation of CD4þ T cell proteomes between nonallergic and allergic subjects and therefore indicate high proteome similarity of CD4þ T cells in the absence of allergen challenge (see Discussion). Proteins Differentially Expressed between Groups OnSeason. While all spots of the on-season comparison show intensity changes within the 2-fold change interval, seven spots were found outside the 1.5-fold or 0.67-fold range. The significance of the on-season differences between allergic/nonallergic 1564
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research
ARTICLE
Figure 3. Representative 2D gel of isolated CD4þ T cells from an allergic volunteer sampled off-season. Spots analyzed by MALDI mass spectrometry are annotated with their spot identification numbers and protein name (see also Tables 3 and 4).
Table 4. Protein Spots Differentially Displayed Off- and On-Season between Nonallergic (Group A) and Allergic Subjects (Group B) spot
protein name
average ratio Group B/A
gene locus
t-test significance
Off-season between nonallergic (Group A) and allergic subjects (Group B) 9q31.1
0.001a
0.64 0.48
9p13
0.01a 0.05
Glutamate-cysteine ligase regulatory protein
0.58
1p22.1
405306
Not identified
0.58
0.05
405097
Not identified
0.63
0.05
958889
Carbonyl reductase 1
0.66
21q222.13
0.001a
959140
Glutathione S-transferase omaga 1
0.65
10q25.1
0.05
404520
Nipsnap homologue 3A
0.53
404641 383960
Not identified Talin 1
404412
0.05
On-season between nonallergic (Group A) and allergic subjects (Group B)
a
Confirmed as differential by Wilcoxon test.
individuals was assessed by a conventional Student’s t test. To reduce falsely significant results of the t testing, a Bonferroni correction was performed with stricter p values. Of the seven potentially differentially stained spots, two could be confirmed by Student’s t test: Carbonyl reductase 1 was found differential with a t test significance of 0.001 and Glutathione S-transferase omega 1 with a t test significance of 0.05. Both proteins are found less abundant in allergic individuals (Group B) compared to nonallergic subjects. Differential expression and statistical significance of these two proteins are summarized in Table 4. Proteins Differentially Expressed within Groups between Seasons. The off-season gels were matched to the on-season gels to form the basis for further data evaluation between seasons. Common expression changes over all subjects per group as well as the difference of responses between both groups were analyzed: the spot intensities were averaged over all subjects per one group for onand off-season. The resulting four values (A-on, A-off, B-on and
B-off) were then compared using the intensity ratios built for each subject using t test- and Wilcoxon-based SAM. Neither one of these analyses showed any significant difference in spot intensities upon allergen exposure between the groups (FDR e 25%). One possible cause for the absence of any group-characteristic proteome signature between seasons may lie in the heterogeneity of the proteome pattern of CD4þ T cells (possibly due to differences in cell subpopulation quantities) from different subjects. The variations of spot intensities within one group may obscure the changes between the seasons. To further elucidate the effect of the allergen exposure on CD4þ T cells, a second data analysis strategy was followed: the ratios of the mean spot intensities of each subject were calculated separately comparing on- and off-season. This analysis delivered proteome differences of individual subjects compared between seasons. In a second step, these changes were compared between subjects and between groups. 1565
dx.doi.org/10.1021/pr100939g |J. Proteome Res. 2011, 10, 1558–1570
Journal of Proteome Research
ARTICLE
Table 5. Numbers of Spots with a Fold Change of g (2 between Conditions of the Same Subjecta Group A Subject
01
03
04
06
07
08
09
11
15
18
9 4
16 10
5 11
20 59
23 36
8 10
25 6
14 37
7 7
8 19
Subject
02
05
12
13
14
16
17
19
20
21
g2-fold
9
14
23
49
49
8
2
1
2
4
e0.5 fold
5
11
3
4
65
23
20
3
4
9
g2-fold e0.5 fold
Group B
a
Ratios off/on season of the mean spot intensities for the gel sample triplicates were calculated. Spots with a g ( 2-fold change were considered.
The numbers of spots with a fold change of g ( 2 between conditions of the same subject is given in Table 5. The subjects of both groups showed a wide span of numbers of spots differentially displayed, varying from a total of 4 spots in donor B19 to 114 spots in donor B14. The mean ratios for each individual were around 1.0- to 1.2-fold expression, except for subjects A11 and B14, both with a mean ratio of 1.4. The overall mean ratio over all subjects was calculated as 1.1 ( 0.2. Neither the analysis of common expression changes over all subjects per one group nor the differentiation of the responses between both groups did show any group-significant difference regarding altered protein abundance as induced by allergen exposure (FDR e 25%). For 7 or more of the total of 10 subjects per group, no changes were observed in the intragroup between-season comparison and no common differences were found in differential protein expression upon allergen exposure comparing nonallergic (A) with allergic (B) individuals. This said, protein changes common to less than 7 out of 10 subjects were found. For example, when the limit of consideration was lowered to similar regulation in at least 5 out of 10 subjects, 1 protein spot (1019872, p = 0.05, conventional dependent t testing) was found differently abundant between the groups. This spot was up-regulated in both groups, with the ratio being higher in the allergic group. However, all these latter changes were regarded as rather subjectspecific. Due to the intersubject heterogeneity, no other statistically significant differentially expressed proteins could be found in the comparison of off- and on-season harvested CD4þ T cells. Western-Blot Confirmation of Differently Expressed Proteins
To confirm the differently expressed and identified proteins, Western-Blot analyses were performed using commercially available antibodies. As shown in the proteomics data, the intensities of differentially displayed spots were all faint. Therefore, HRPconjugated secondary antibodies were chosen in combination with ECL Plus substrate and chemiluminescent read-out on X-ray film. Western blot band intensities were compared qualitatively and quantitatively using tubulin as internal loading control. Polyclonal anti-Carbonyl reductase (CBR1) antibody specifically stained CBR1 protein in all samples. The CBR1 band intensities were normalized to tubulin control. Variability within the groups was found much higher than obtained by 2DE. However, the median for Group A = nonallergic (0.30) and Group B = allergic (0.18) confirmed the observations of the 2 DE, namely
that CBR1 is down-regulated in allergic subjects. These results are summarized in Figure 4 that shows Western and box plots. Talin 1 has an approximate molecular weight of 270 kDa and was therefore not detected as intact protein in the 2DE proteome comparison. However, a 30 kDa fragment was found differently expressed. The antibody staining returned multiple signals
