of December 18, 2014.This information is current as

ImmunodeficiencyAbnormalities in Common Variable Unravelling the Complexity of T Cell

Paganelli, Walter Malorni and Fernando AiutiRossella Carello, Antonio Pizzuti, Marco G. Paggi, Roberto Mileo, Marco De Felice, Barbara Mora, Antonella Esposito,Mazzetta, Marco Marziali, Cristina Renzi, Anna Maria Antonello Giovannetti, Marina Pierdominici, Francesca

http://www.jimmunol.org/content/178/6/3932doi: 10.4049/jimmunol.178.6.3932

2007; 178:3932-3943; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/178/6/3932.full#ref-list-1

, 13 of which you can access for free at: cites 49 articlesThis article

Subscriptionshttp://jimmunol.org/subscriptions

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/ji/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/cgi/alerts/etocReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2007 by The American Association of9650 Rockville Pike, Bethesda, MD 20814-3994.The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on D

ecember 18, 2014

http://ww

w.jim

munol.org/

Dow

nloaded from

Unravelling the Complexity of T Cell Abnormalitiesin Common Variable Immunodeficiency1

Antonello Giovannetti,2,3* Marina Pierdominici,2† Francesca Mazzetta,‡ Marco Marziali,‡

Cristina Renzi,§ Anna Maria Mileo,¶ Marco De Felice,† Barbara Mora,� Antonella Esposito,‡

Rossella Carello,# Antonio Pizzuti,� Marco G. Paggi,¶ Roberto Paganelli,** Walter Malorni,††

and Fernando Aiuti‡

We investigated several phenotypic and functional parameters of T cell-mediated immunity in a large series of common variableimmunodeficiency (CVID) patients. We demonstrated that the vast majority of CVID patients presented multiple T cell abnor-malities intimately related among them, the severity of which was reflected in a parallel loss of CD4� naive T cells. A strongcorrelation between the number of CD4� naive T cells and clinical features was observed, supporting the subgrouping of patientsaccording to their number of naive CD4� T lymphocytes. A reduced thymic output and disrupted CD4� and CD8� TCRrepertoires paralleled the contraction of CD4� naive T cell pools. The evaluation of activation markers and cytokine productionindicated a strong T cell activation that was significantly related to the increased levels of T cell turnover and apoptosis. Finally,discrete genetic profiles could be demonstrated in groups of patients showing extremely diverse T cell subset composition andfunction. Naive CD4� T cell levels were significantly associated with the switched memory B cell-based classification, al-though the concordance between the respective subgroups did not exceed 58.8%. In conclusion, our data highlight the keyrole played by the T cell compartment in the pathogenesis of CVID, pointing to the need to consider this aspect forclassification of this disease. The Journal of Immunology, 2007, 178: 3932–3943.

C ommon variable immunodeficiency (CVID)4 represents alargely heterogeneous group of immunologic disorderscharacterized by low serum Ig concentrations, defective

specific Ab production, and increased susceptibility to bacterialinfections (1). CVID patients also frequently develop splenomeg-aly, granulomatous lesions, and autoimmune diseases (2). The as-sociation of CVID with the development of malignancies, mainly

non-Hodgkin’s lymphoma of the gastrointestinal tract, has beenalso documented (3). CVID has a prevalence of �1 in 30,000 andis thus the second most frequent primary immunodeficiency inhumans (4). The 20-year survival rate after the diagnosis of CVIDis 64% for male patients and 67% for female patients comparedwith 92–94% in the general population (5). Initial family studiessuggested the presence of at least two susceptibility loci within theMHC on the short arm of chromosome 6 (6). Recently, it has beendemonstrated that the homozygous loss of ICOS (7, 8) and muta-tions of transmembrane activator and calcium modulator ligandinteractor (9, 10) can result in CVID, although both of the defectsaccount for only 5–10% of the cases.

Clinically overt CVID necessitates long-term Ig replacementand antimicrobial therapy, but treated patients still show a highermortality rate than the general population. This result contrastswith an exclusive role played by the defective Ig production indetermining the course of the disease. A more complete explana-tion would encompass the T cell abnormalities in CVID pathogen-esis. In fact, a number of T cell defects have been demonstrated inthe past in a still undefined proportion of CVID patients, includingdecreased lymphocyte proliferation to mitogens and Ags (2), in-creased T cell apoptosis (11), impaired cytokine production (12,13), absent generation of Ag-primed T cells after prophylactic vac-cination (14, 15), and reduced expression of CD40L on activatedT cells (16, 17). This complex scenario of T cell deficiency prob-ably takes a major part in influencing the clinical outcome ofCVID patients. In this regard, it has been reported that poor T cellfunction at the time of the diagnosis is associated with death at anearlier age (2).

Attempts to classify CVID patients have been based on the as-sessment of Ig synthesis in vitro and phenotypic subsets of periph-eral blood B cells (18–21). To date, a comprehensive evaluation ofthe T cell deficiency is still lacking and CVID patients remain to

*Department of Clinical Medicine, Division of Allergy and Clinical Immunology,“La Sapienza” University, Rome, Italy; †Department of Cell Biology and Neuro-sciences, Istituto Superiore di Sanita, Rome, Italy; ‡Interregional Center for PrimaryImmunodeficiencies, Department of Clinical Medicine, “La Sapienza” University,Rome, Italy; §Epidemiology Unit, Istituto Dermopatico dell’Immacolata, Rome, Italy;¶Department for the Development of Therapeutic Programs, Center for ExperimentalResearch, Regina Elena Cancer Institute, Rome, Italy; �Institute Casa Sollievo dellaSofferenCa-Mendel, Rome, Italy; #Doctoral School of Research in Science of Immu-notherapy, University of Rome “La Sapienza,” Rome, Italy; **Department of Med-icine and Sciences of Aging, “G. d’Annunzio” University, Chieti, Italy; and ††De-partment of Drug Research and Evaluation, Istituto Superiore di Sanita, Rome, Italy

Received for publication March 8, 2006. Accepted for publication December22, 2006.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by grants from University of Rome, Faculty of Medicine2001–2004 and 2004–2006 for Primary Immunodeficiencies (to F.A.); Ministerodell’Universita e della Ricerca Scientifica e Technologica, projects for Primary Im-munodeficiencies, 2002–2003 (to F.A.); Istituto Superiore di Sanita, 2003–2005 (toF.A.); and Istituto Superiore di Sanita, 2003–2004, Grant C3AF (to M.P.).2 A.G. and M.P. contributed equally to this study.3 Address correspondence and reprint requests to Dr. Antonello Giovannetti, Depart-ment of Clinical Medicine, Division of Clinical Immunology and Allergy, “La Sa-pienza” University, Viale dell’Universita 37, 00185 Rome, Italy. E-mail address:antonello.giovannetti@uniroma1.it4 Abbreviations used in this paper: CVID, common variable immunodeficiency; PI,propidium iodide; HCFC1, host cell factor 1; OR, odds ratio; CI, confidence interval;BV, �-chain variable region; DC, dendritic cell.

Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00

The Journal of Immunology

www.jimmunol.org

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

be subdivided into homogeneous subgroups according to their Tcell abnormalities.

Therefore, it has been our interest to perform, in a large numberof CVID patients, a wide panel of immunological investigationsaiming specifically to a quantitative and qualitative evaluation ofthe T cell compartment. Genetic assays have been also performedto detect any possible link between phenotype and discrete geneticprofiles.

Materials and MethodsPatients and controls

This study includes 60 patients admitted and studied from January 1983 toJune 2004 at the Division of Clinical Immunology located at the Universityof Rome “La Sapienza.” All of them have a well-documented CVID di-agnosis according to the criteria established by the European Society forImmunodeficiencies/Pan-American Group for Immunodeficiency group ofexperts (1). The age range was 15–78 years and there were 29 females and31 males. All patients were receiving regular i.v. Ig substitution therapyand were free of any serious infections when tested. As control subjects, werecruited 30 age-matched healthy donors. The study protocol conforms tothe ethical guidelines of the 1975 Declaration of Helsinki. Demographic,clinical, and laboratory characteristics of patients investigated are reportedin Table I. According to the classification described by Warnatz et al. (20),CVID patients were segregated into two groups based on the flow cytometricquantification of class-switched memory B cells (CD19�CD27�IgM�IgD�):CVID group I comprises patients with class-switched memory B cells below0.4% of total PBL, and CVID group II includes all patients with normal num-bers of switched memory B cells (�0.4%). Group I can be subdivided ac-cording to increased (Ia) or normal (Ib) numbers of CD19�CD21� immatureB cells.

Flow cytometry analyses

Surface phenotyping and cytokine intracellular staining of PBL were per-formed with quadruple combinations of mAbs as described before (22). ForB cell surface phenotyping, a mixture of the following Abs at optimalconcentrations was used: CD27 FITC or CD21 FITC, anti-IgD PE, anti-IgM PerCP, and CD19 allophycocyanin (BD Immunocytometry Systems).Apoptosis level was quantified by a double staining flow cytometry methodusing FITC-conjugated annexin V/propidium iodide (PI) apoptosis detec-tion kit (Marine Biological Laboratory) according to the manufacturer’sprotocol. Reported data referred to both early (AV�/PI� cells, still alive)and late (AV�/PI� cells, dead cells) apoptotic cells, and were calculatedwithin the CD4 and CD8 subsets.

For Ki-67 intracellular staining, cryopreserved cells, were thawed, fixed,and stained after permeabilization with Ki-67-PE mAb (BD Immunocy-tometry Systems).

Analyses were done with a FACSCalibur flow cytometer and theCellQuest software (BD Immunocytometry Systems).

Molecular studies

To perform the CDR3 spectratyping, CD4� and CD8� T cells were sep-arated by using CD4 and CD8 MicroBeads and MACS columns accordingto the manufacturer’s protocols (Miltenyi Biotec). Total mRNA was ex-tracted directly from 1 � 106 bead-coated cells using TRIzol-LS Reagent(Invitrogen Life Technologies) and Microcarrier (Molecular Research Center).cDNA samples obtained by retrotranscription of total RNA were amplified byusing a TCRB C1/C2-specific primer and a set of 24 TCRBV-specific primersas described previously (22).

For array hybridization, the Clontech Atlas Human Hematology/Immu-nology Array, composed of 406 human cDNAs, was used. Arrays werescanned using a Bio-Rad PhosphorImager and acquired by the QuantityOne software, version 14 (Bio-Rad). Data analysis, quantification, andcomparison were done using the AtlasImage 2.7 software (BD Clontech).The experiments were performed in duplicate, using the same RNA prep-aration. A mixture of equal amounts of RNA from negatively selected Tlymphocytes purified from 10 different healthy controls or from 10 ran-domly selected CVID patients belonging to group I and III was used. In-tensity values were normalized using the global normalization mode (summethod) allowed by the software. The expression levels of the housekeep-ing genes spotted on the arrays were also used to validate the normalizationprocedure. A 2-fold expression variation (up or down) between patientsand healthy donors was the minimum requirement for a gene to be selected.

Quantitative RT-PCR analysis for host cell factor 1 (HCFC1), CD9 Ag,LFA3, and cyclophilin A were conducted on the LightCycler instrument

using LightCycler RNA amplification kit SYBR Green I (Roche). Theprimer sequences for HCF1 were as follows: forward 5�-GACGAACTGCACGTGTACAAC-3� and reverse 5�-GACGAACTGCACGTGTACAAC-3�; for CD9, forward 5�-GACTCTCAGACCAAGAGCATC-3� and reverse5�-CTTTGATGGCATCAGGACAGG-3�; for LFA-3, forward 5�-CCATGTACCAAGCAATGTGCC-3� and reverse 5�-GTCTTGAATGACCGCTGCTTG-3�; for cyclophilin A, forward 5�-TGGTCAACCCCACCGTGTTC-3� and reverse 5�-GCCATCCAACCACTCAGTC-3�. All of the above-mentioned primer sequences were obtained from the National Center forBiotechnology Information database. RT-PCR was performed in 20 �l ofreaction mixture containing 300 ng of total RNA, 6 mM MgCl2, 2 �l ofeach primer (0.5 �M), 40 �l of LightCycler RT-PCR mix SYBR Green I,0.4 �l of enzyme mix, and ddH2O in a glass capillary. Reverse transcrip-tion was performed at 55°C for 10 min. The cycle program was set at 1cycle of denaturation at 95°C for 30 s; 40 cycles of 95°C for 0 s, annealingat 55°C for 5 s, and extension at 72°C for 18 s; 1 cycle of cooling at 40°Cfor 30 s. Real-time quantitative RT-PCR for human cyclophilin A gene wasperformed on each sample as an internal control for equivalence of tem-plate. PCR amplification dynamics were monitored on-line by fluorescencesignal acquisition. Each RT-PCR amplification was repeated in triplicate.Melting curves and subsequent agarose gel electrophoresis analysis wereperformed to verify the specificity of amplified DNA fragments. Quanti-tative analysis of target genes expression was done using the comparativecrossing point method (23).

Statistical analysis

Groups were compared using the Fisher’s exact test for categorical vari-ables. Regarding continuous variables, the Kruskal-Wallis test was used forcomparing several groups and the two-tailed Mann-Whitney U test forcomparing two groups. Spearman’s rank correlation coefficients were cal-culated for naive CD4� T cell values and overall clinical severity scoresand grade of splenomegaly. Due to the nonnormality of the distribution,naive CD4� T cell values were divided into tertiles. Initially, the univariateassociation between overall clinical severity (severe vs mild) and naiveCD4� T cell level was examined by logistic regression. Subsequently,multivariable logistic regression analysis was performed for examining theassociation between overall clinical severity and naive CD4� T cell level,controlling for variables thought a priori to be potential confounders, suchas age, gender, Ig levels (tertiles), and total number of CD4� T cells (ter-tiles of absolute numbers). Similarly, multiple logistic regression was alsoused for examining the association between splenomegaly and naive CD4�

T cell levels. Odds ratio (OR) and 95% confidence intervals (CIs) wereestimated using the regression coefficients, and their SE was obtained fromthe logistic regression analyses.

Statistical analysis was performed using the computer package STATA-9.0 Statistical Software for Windows (Stata).

ResultsRestricted CD4� and CD8� naive T cell pools in subgroups ofCVID patients

As indicated in Fig. 1, frequency and absolute counts of CD4� Tcells within the PBL population of CVID patients were signifi-cantly reduced in comparison to healthy controls ( p � 0.0001). Agreat interpatient variability was observed with frequencies rang-ing from 18% to 65% and absolute counts from 34 cells/�l to 1632cells/�l (normal ranges, 39–57% and 603-1646 cells/�l, respec-tively). Differently from CD4� T cells, CD8� T cells were sig-nificantly increased in CVID patients but only as a percentage( p � 0.0001). A decreased distribution of CD4� and CD8� naiveT cells was also detected ( p � 0.0001). A high degree of vari-ability was observed for CD4� naive T cells with frequenciesranging between 3% and 72% and absolute counts from 2 cells/�lto 1,175 cells/�l. Table II illustrates the characteristics of the studysample when subclassified by naive CD4� T cell levels. Threehomogeneous subgroups were identified according to this param-eter, each including one-third of cases (tertiles). Patients withCD4� naive T cells falling in the lower, intermediate, and uppertertiles were defined as belonging to group I, II, and III, respec-tively. All of the other variables investigated were then analyzedwith respect to the patients’ subgroups defined by the number ofnaive CD4� T cells. Spearman’s rank correlation coefficients

3933The Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

showed a strong negative correlation for naive CD4� T cell valuesand overall clinical severity scores (r � �0.68) and for naiveCD4� values and levels of splenomegaly (r � �0.76). Overallclinical severity scores showed a weaker correlation with thetotal number of CD4� T cells as percentages (r � �0.31) andas absolute numbers (r � �0.26). Naive CD4� T cell levelswere strongly associated with clinical severity at univariate

analysis: the likelihood of having a severe disease increased7.7-fold for a unit decrease in naive CD4� T cell level, goingfrom high to low to very low levels (OR � 7.65; 95% CI, 2.8to 21.1; p � 0.001).

As shown in Table III, multivariate analysis confirmed the sig-nificant association between naive CD4� T cell levels and clinicalseverity independently of age, gender, total number of CD4� T cells,

Table I. Demographic and clinical characteristics of the study populationa

PatientNos.

CD4�

NaiveT Cells

(%)

CD19�

B Cells(%)

IgM/D�

CD27�

Naive% B

IgM/D�

CD27�

Memory% B

IgM/D�

CD27�

Switched% B

CD21�

% BWarnatz’

Classification SexAge

(year)

Age atOnset(year)

IgG(mg/dl)

IgA(mg/dl)

IgM(mg/dl) Splenomegaly AD

OverallClinicalSeverity

1 3 6 5.2 0.7 0.05 27 Ia F 61 44 286 2 17 Yes (���) Yes 102 4 4 ND ND ND ND ND M 42 32 218 2 11 Yes (���) No 113 4 3 2.8 0.1 0.02 55 Ia F 51 13 310 2 26 Yes (���) Yes 134 4 2 1.8 0.1 0.01 36 Ia F 55 38 114 15 12 Yes (��) No 105 5 2 1.7 0.2 0.01 40 Ia F 44 40 88 12 99 Yes (��) No 106 6 12 ND ND ND ND ND M 43 20 64 85 77 Yes (�) No 97 7 2 1.3 0.1 0.01 23 Ia M 34 19 100 2 2 Yes (���) No 118 7 40 39.6 0.1 0.04 1 Ib M 54 15 200 2 11 Yes (���) No 159 7 5 4.9 0.06 0.01 3 Ib F 46 8 463 42 32 Yes (���) No 1210 8 1.1 1 0.1 0.01 50 Ia M 49 8 188 2 2 Yes (���) No 1411 8 3 2.9 0.06 0.01 43 Ia F 50 37 20 2 67 Yes (���) No 1012 8 2 ND ND ND ND ND M 36 18 100 5 5 Yes (�) Yes 913 9 2 1.1 0.3 0.04 11 Ib F 58 8 215 19 27 Yes (��) No 1014 10 10 7.3 2 0.45 10 Ib F 51 45 220 10 11 Yes (��) Yes 1115 10 2 1.4 0.6 0.01 13 Ib F 54 55 180 12 22 Yes (�) No 516 10 9 ND ND ND ND ND F 61 8 200 2 5 Yes (���) Yes 1217 11 23 12 10.6 0.02 24 Ia M 54 35 564 2 9 Yes (��) No 918 12 14 12 1.7 0.07 4 Ib F 57 15 300 2 28 Yes (���) No 1019 13 6 4 1.7 0.12 12 Ib M 56 22 92 2 2 Yes (���) No 1520 15 16 11.8 4 0.48 7 Ib F 30 8 56 2 16 No No 821 17 22 9.7 12 0.1 11 Ib F 48 8 451 2 16 Yes (��) Yes 922 17 9 4.3 3.7 0.8 6 II M 42 24 601 8 12 Yes (�) No 923 17 12 ND ND ND ND ND F 40 6 133 7 6 Yes (�) No 1024 17 16 12 3.7 0.3 9 Ib F 75 68 75 20 28 Yes (���) No 725 18 11 ND ND ND ND ND F 31 21 194 22 18 Yes (�) Yes 626 19 6 5 0.5 0.1 26 Ia F 42 24 132 2 2 Yes (�) No 827 20 6 ND ND ND ND ND F 75 40 300 2 2 Yes (�) No 728 20 12 10 1.3 0.04 6 Ib M 52 8 289 2 60 Yes (��) Yes 1429 21 19 9 9 0.36 4 Ib F 61 8 400 2 22 No No 830 22 4 3 0.7 0.52 12 II M 76 7 250 8 2 No No 431 24 2 1.7 0.06 0.06 7 Ib F 36 21 65 0 2 Yes (�) No 632 24 10 8.7 1.2 0.1 5 Ib F 66 49 398 87 42 No No 733 25 1.2 1.1 0.02 0.04 38 Ia M 33 10 150 2 2 Yes (�) No 1034 25 6 4.5 0.6 0.8 3 II F 50 8 71 2 49 Yes (�) No 735 26 6 4 1.5 0.3 12 Ib M 55 10 310 11 17 Yes (�) No 636 26 4 3.2 0.4 0.2 3 Ib M 22 8 574 20 50 No No 537 26 16 11 4 0.3 7 Ib M 63 35 300 2 6 Yes (�) No 638 27 9 6 1.6 0.3 5 Ib F 45 8 96 2 2 Yes (�) Yes 739 28 8 4 0.5 0.6 2 II M 15 20 350 10 20 No No 740 28 4 2.6 0.8 0.5 2 II F 78 8 250 2 6 No No 441 30 7 ND ND ND ND ND M 49 8 206 5 11 Yes (�) No 642 31 4 2.7 1 0.2 0.4 Ib M 55 8 319 24 25 Yes (�) Yes 643 31 1.5 0.8 2 0.2 9 Ib M 45 5 354 2 2 Yes (�) No 544 32 6 5 0.7 0.06 15 Ib M 56 46 220 31 91 Yes (�) No 1045 33 15 10 1 3 3 II M 26 7 50 2 2 Yes (�) Yes 1246 35 4 2 0.9 0.9 5 II M 24 7 250 8 2 No No 447 36 15 9 3 2.2 11 II M 40 8 96 2 10 No No 848 36 6 4 1.6 0.2 14 Ib F 24 8 153 20 15 No No 449 37 9 4 3.6 0.9 15 II M 50 40 228 18 2 No No 450 38 2 2 0.1 0.02 12 Ib F 36 8 19 2 2 Yes (�) No 1551 40 11 ND ND ND ND ND M 33 12 143 2 82 No No 452 40 5 1.9 2.5 0.5 0.2 II M 60 60 403 33 35 No No 453 41 17 ND ND ND ND ND F 39 35 411 74 36 No No 654 42 7 6 0.4 0.1 14 Ib M 28 19 59 2 2 No No 555 43 8 ND ND ND ND ND F 41 8 259 2 35 Yes (��) No 856 44 5 3.5 0.7 0.5 14 II M 33 20 218 2 18 No No 457 48 14 ND ND ND ND ND M 34 20 85 2 12 No No 458 52 5 1.7 1.2 1.6 7 II M 45 8 256 2 7 No No 459 57 15 12 2 0.45 4 II M 32 8 29 2 2 No No 460 72 1.3 1.2 0.02 0.03 10 Ib F 42 8 50 2 2 No No 4

a CVID patients are stratified according to the percentage of peripheral CD4� naive (CD45RA�CD62L�) T cells. Serum Ig levels were at time of diagnosis. Splenomegalywas verified by computerised tomography scanner or ultrasonographic examination (�, ��, ���, �12 �15 cm, 15–18 cm, and �18 cm in craniocaudal length,respectively). The overall clinical severity of the disease was quantified on the basis of the following criteria evaluated within the last 5 years: 1) history of severerespiratory tract infections (SRT1 elevated numbers of leukocytes and C-reactive protein, body temperature higher than 38.5°C, and antibiotic therapy); 2) use ofantibiotics; 3) patients’ judgment of severity; 4) physicians’ judgment of severity. A score ranging from 1 (less severe) to 4 (more severe) was assigned to each of theseparameters. For the “history of infections” parameter the scores were as follows: 1, �2 SRT1 per year; 2, �3 and �4 SRTI per year; 3, �5 and �6 SRTI per year;4, �6 SRTI per year. For the “use of antibiotic” parameter, the scores were as follows: 1, �2 times per year; 2, �3 and �4 times per year; 3, �5 and �12 times peryear; 4, �12 times per year. The sum of the scores obtained for each parameter was used to define as overall clinical severity score that was divided into tertiles. AD,Autoimmune diseases; F, female; M, male.

3934 T CELLS IN COMMON VARIABLE IMMUNODEFICIENCY

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

and Ig levels. The only other variable associated with disease severitywas the IgA level.

Moreover, multiple logistic regression analysis was performedexamining the association between splenomegaly and naive CD4�

T cell levels, including gender, age, and Ig levels in the model.Lower naive CD4� T cell levels were significantly associated withan increased likelihood of splenomegaly (OR � 4.78; 95% CI, 1.8to 12.9; p � 0.002 for a unit decrease in naive CD4� T cell levels,going from high to low to very low levels).

Naive CD4� T cell levels were also shown to be significantlyassociated with the Warnatz classification groups (Fisher’s exact test;p � 0.001) (20). In particular, among patients in the lowest naivecategory, i.e., group I, 50% belonged to the Warnatz group Ia and50% to group Ib; among patients in the intermediate naive cate-gory, i.e., group II, the majority of patients (58.8%) belonged toWarnatz group Ib, 29.4% to group II, and 11.7% to group Ia;among patients with values of CD4� naive T cells �30%, i.e.,group III, 53.3% belonged to Warnatz group II and 46.7% togroup Ib.

As indicated in Fig. 2, C and E, the frequency of effector mem-ory cells were significantly increased in all three groups of patientsin both CD4 and CD8 subsets, whereas central memory cells were

increased for CD4� T cells in groups I and II and for CD8� T cellsonly in group I ( p � 0.01). Absolute counts of memory subsets inCD4� T cells showed a decrease of central memory cells in allpatients ( p � 0.01), whereas no significant differences in compar-ison to controls were detected for the effector memory cells (Fig.2D). Differently, in CD8� T cells, a sustained increase in the ab-solute count of central memory cells was detected in group I ( p �0.0124), whereas the effector memory cells were increased in allpatients ( p � 0.02) (Fig. 2F).

Increased T cell activation

Increased frequencies of CD4� and CD8� CD95- and HLA-DR-expressing T cells could be demonstrated in all three groups ofpatients ( p � 0.01) (Fig. 2, G and I). However, in group III, thelevel of T cell activation was less prominent than that observed ingroup I and II of patients (III vs I and II; p � 0.01). Minor dif-ferences were detected in the expression of CCR5 and significantlyincreased in CD4� and CD8� T cells only in group I ( p � 0.0007and p � 0.0010, respectively).

The percentages of CD95-, CCR5-, and HLA-DR-expressing Tcells negatively correlated with the frequency of naive T cells in bothCD4 and CD8 subsets (r � 0.6 and p � 0.0001 for all analyzed

FIGURE 1. Flow cytometric analysis of totalCD4� and CD8� and naive T cell subsets from CVIDpatients and age-matched healthy controls. Data arerepresented as box plots displaying medians, 25th and75th percentiles as boxes, and 10th and 90th percen-tiles as whiskers. The distribution of naive T cells wasevaluated by using mAbs directed toward theCD45RA and CD62L molecules. Differences be-tween patients and controls were compared by theMann-Whitney U test.

3935The Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

populations). The increased frequency of HLA-DR�, CD95�, andCCR5�CD4� T cells was not seen in association with a parallel in-crease in their absolute count, with the exception of CD4�HLA-DR�

cells in group I ( p � 0.0139) (Fig. 2H). A different trend was ob-served in the CD8 subset. Here, high absolute counts were detectedfor both CD95� and HLA-DR� T cells in all CVID patients ( p �0.01 for both markers), whereas no meaningful differences were ob-served for CCR5� T cell count with the exception of group I ( p �0.0102) (Fig. 2L).

Reduced thymic output

Thymic output was evaluated by measuring the expression ofCD31 molecules that allows us to distinguish postthymically ex-panded naive CD4� T cells (centralnaive CD4� T cells) from truerecent thymic emigrants (thymicnaive CD4� T cells) (24). Inhealthy control individuals, the percentage of CD31-expressingCD4� T cells declined with age (Fig. 3). Among CVID patientsthis physiological, age-related decline of CD31-expressing CD4�

T cells was no longer detectable in group I and II. In fact, in thesepatients, the production of new thymicnaive CD4� T cells was al-ready deeply impaired during early adulthood ( p � 0.006, p �0.001, and p � 0.19 for groups I, II, and III vs controls). We alsoevaluated the TCR excision circle content (25) of CD4� T cells inseveral CVID patients studied here, finding a good correlation withthe CD31 data (data not shown).

Disrupted TCRBV repertoires

Anomalies in the relative TCRBV usage were investigated by us-ing a panel of �-chain variable region (BV) subfamily specific

Table II. Characteristics of the study sample by naive CD4� T cell levels

Group I (�15%)(no. of patients � 20)

Naive CD4� T CellLevels

Group II (16–29%)(no. of patients � 20)

Group III (at least 30%)(no. of patients � 20)

p*n (%) n (%) n (%)

GenderMale 8 40.0 8 40.0 15 75.0 0.04Female 12 60.0 12 60.0 5 25.0

Age (year)Mean 49.3 50.3 39.6Median (range) 51 (30–61) 49 (15–78) 39.5 (24–60) 0.02†

Overall clinical severity‡Mild 1 5.3 7 36.8 15 79.0 �0.001Severe 10 52.6 11 57.9 2 10.5Very severe 8 42.1 1 5.3 2 10.5

SplenomegalyNo 1 5.0 6 30.0 13 65.0 �0.001Grade 1 3 15.0 11 55.0 6 30.0Grade 2 5 25.0 2 10.0 1 5.0Grade 3 11 55.0 1 5.0 0 0.0

IgG tertilesLower tertile 7 35.0 5 25.0 8 40.0 0.5Intermediate tertile 8 40.0 5 25.0 7 35.0Upper tertile 5 25.0 10 50.0 5 25.0

IgA tertilesLower tertile 12 60.0 11 55.0 12 60.0 0.9Intermediate tertile 2 10.0 3 15.0 1 5.0Upper tertile 6 30.0 6 30.0 7 35.0

IgM tertilesLower tertile 8 40.0 9 45.0 5 25.0 0.8Intermediate tertile 6 30.0 5 25.0 7 35.0Upper tertile 6 30.0 6 30.0 8 40.0

CD4� T cells (cells/Microliter)Mean 581.4 539 813.3Median (range) 497 (34–1652) 526 (278–970) 659 (404–1635) 0.02†

CD4� T cells (%)Mean 33.4 35.4 39.3Median (range) 30 (18–58) 33.5 (21–65) 39 (19–54) 0.14†

�, Fisher’s exact test.†, Kruskal-Wallis test.‡, Overall clinical severity: mild � lower tertile (score �7); severe � intermediate tertile (score 7–10); very severe � upper tertile (score at least 10).

Table III. Multiple logistic regression ORs for the association betweenoverall clinical severity (severe vs mild disease) and naive CD4� T celllevels, controlling for potential confounders (gender, age, Ig levels, andtotal number of CD4� T cells)a

OR (95% CI) p

Naive CD4� level (fromhigher to lower levels)�

7.46 (2.1 to 26.5) 0.002

GenderMale 1Female 3.13 (0.6 to 15.1) 0.2

Age (year) 1.0 (0.9 to 1.1) 0.9IgG tertiles (from lower to

higher tertiles)†0.69 (0.2 to 2.0) 0.5

IgM tertiles (from lower tohigher tertiles)†

2.31 (0.8 to 7.1) 0.1

IgA tertiles (from lower tohigher tertiles)†

0.35 (0.1 to 0.9) 0.046

CD4� T cells (absolutenumbers in tertiles)

0.79 (0.3 to 2.2) 0.6

a The severe and very severe categories of the overall clinical severity weregrouped together defining them as severe diseases.

�, For unit decrease in naive CD4� T-cell level, going from high to low to verylow levels.

†, For unit increase in Ig levels going from lower to intermediate to upper tertiles.

3936 T CELLS IN COMMON VARIABLE IMMUNODEFICIENCY

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

mAbs covering �60–70% of T cells expressing TCR in healthycontrols (Fig. 4, A and B). Within the CD4 subset significant ex-pansions were observed, among the 22 TCRBV genes investi-gated, mainly in patients belonging to group I. These expansions(n � 96) clustered preferentially to certain TCRBV genes: expan-sions of BV16, BV14, and BV 21.3 were detected in 77, 61, and50%, respectively, of patients belonging to group I. The number ofexpanded CD4 BV families dramatically decreased to 14 in groupII and to 8 in group III, demonstrating a significant associationbetween the level of CD4� naive cells and the restriction of TCRrepertoires ( p � 0.001). Expanded BV families were also ob-

served within the CD8 subset; however, differently from CD4� Tcells, they were not confined to groups I and II, but were alsoobserved in group III (47, 38, and 27 expansions for group I, II,and III, respectively; p � 0.186).

Because the evaluation of TCRBV usage with mAbs does notallow an exclusion of reduced diversity of the TCR repertoire evenin the presence of a normal distribution of individual BV families,we looked for further restrictions of repertoires by performing themore informative CDR3 spectratyping, namely the quantitativeanalysis of the CDR3s with different sizes generated by the randominsertion/deletion of nucleotides during V(D)J rearrangement.

FIGURE 2. Flow cytometric anal-ysis of CD4� (A, C, D, G, and H)and CD8� (B, E, F, I, and L) Tcell subsets of CVID patients andage-matched healthy controls. Dataare represented as mean values SD. The naive subset is defined asCD45RA�CD62L�, whereas the re-maining cells comprise the memorysubsets (CD45RA�CD62L�, centralmemory subset; CD45RA�CD62L�

and CD45RA�CD62L�, effectormemory subsets) (47, 48). �, Statis-tical difference compared withcontrol values (p � 0.05; Mann-Whitney U test).

3937The Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

Nine patients from group I, three from group II, and two fromgroup III were randomly assigned to perform the CDR3spectratyping.

Mathematical analysis of the deviation of patients’ histogramsfrom the normal distribution revealed altered patterns of selectedCD4 BV families in most patients of group I (Fig. 4C). Theseoligoclonal expansions were mainly detected in BV families de-scribed above as preferentially expanded on the basis of cytofluo-rimetric analysis. CDR3 perturbations were negatively correlatedwith the CD4 counts (r � �0.690; p � 0.0131). Low levels ofCDR3 perturbation were observed in the remaining five patientsassigned to group II and III in agreement with their milder clinicalcourses. The sequencing of CDR3 regions, performed in selectedpatients, confirmed the oligoclonal nature of peaks detected byCDR3 spectratyping (data not shown).

By contrast, massive alterations of CDR3 profiles were observedin CD8� T cells of all the 14 patients studied by spectratyping(Fig. 4D). Partially at variance with CD4� T cells, the exception-ally high levels of CD8 CDR3 perturbation were not confined togroup I but were also observed in groups II and III. In CD8�

T cells, disrupted CDR3 profiles were detected in virtually all BVfamily investigated irrespective of the relative CD8 BV usagedemonstrated by cytofluorometry.

FIGURE 3. Frequencies of CD31-expressing CD4� T cells during agingin both CVID patients and age-matched control donors. Thymic output wasevaluated by measuring the expression of CD31 molecules that allows us todistinguish postthymically expanded naive CD4� T cells (centralnaive CD4� Tcells) from true recent thymic emigrants (thymicnaive CD4� T cells) (24). Pe-ripheral blood T cells were stained with anti-CD4, anti-CD45RA, anti-CD62L,and anti-CD31 mAbs and examined by flow cytometry.

FIGURE 4. Diversity of TCRBV repertoires studied by flow cytometric and CDR3 spectratyping analyses in CVID patients. Flow cytometric analyseswere done with BV-specific mAbs in CD4� (A) and CD8� (B) T cells. Values �3 SD above the mean of controls were used to establish skewed use ofBV gene families by CD4� and CD8� T cells. Data are shown as percentages of patients showing expansion of TCRBV families. CDR3 spectratyping wereperformed in CD4� (C) and CD8� (D) T cells from 14 representative CVID patients (9 from group I, 3 from group II, and 2 from group III). To analyzethe level of TCRBV perturbation, the CDR3 spectratypes were first translated into probability distributions as function of the area under the profile for eachCDR3 length. A control profile, representing the nonperturbed repertoire, was determined for each BV by calculating the average distribution of thecorresponding CD4 and CD8 profiles from healthy donors. The extent of perturbation for each CDR3 fragment was then calculated by the differencebetween the sample’s distribution and the control’s distribution. Finally, the TCR repertoire perturbation per BV family was defined as the sum of theabsolute values of the differences between each sample’s CDR3 length and the corresponding control distribution. Values of BV perturbation greater thanthe sum of the SDs calculated in normal blood donors for each CDR3 profile were considered abnormal (f). �, Values within normal range. Patient nos.1–18, group I; patient nos. 22–33, group II; patient nos. 52–59, group III.

3938 T CELLS IN COMMON VARIABLE IMMUNODEFICIENCY

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

Altered cytokine production

Fig. 5 illustrates all three patient groups and controls for the pe-ripheral distribution of CD4� and CD8� IFN-�-, IL-2-, IL-4-, andIL-10-producing T cells. Within the CD4� T cell subset, the pro-duction of IFN-� was significantly increased only in group I ( p �0.0016), whereas IL-4 was within normal range in all CVIDpatients. The percentage of CD4� IFN-�-producing cells was pos-itively related to that of memory/effector cells (r � 0.716; p �0.0001) as well as to their activation state (r � 0.7 and p � �0.0001 vs CD95-, CCR5-, and HLA-DR-expressing CD4� Tcells). Differently, no significant correlation between the level ofIL-4 and the memory/activated CD4� T cell subsets was detected.When CD8� T cells were considered, high levels of IFN-� weredetected in groups I and II ( p � 0.0040 and p � 0.0240, respec-tively), whereas IL-4 production was increased in all three groups( p � 0.0162, p � 0.0140, p � 0.0336, for group I, II, and III,respectively). A positive correlation between IFN-� productionand the memory/effector and activated CD8 subsets was detected

(r � 0.6 and p � 0.001 for all cases). A mild correlation betweenthe IL-4 production and the memory/effector CD8� T cell subsetwas also observed (r � 0.382; p � 0.0450). A significantly re-duced production of IL-10 ( p � 0.05) was detected in all sub-groups of patients in CD4� and CD8� T cells, although no sig-nificant correlation between the composition of the T cell subsetsand the distribution of IL-10-expressing T cells could be observed.No differences between CVID and controls could be observed atthe level of IL-2 production in both CD4� and CD8� T cells.

Increased CD4� and CD8� T cell turnover

Proliferation of CD4� and CD8� T lymphocytes was evaluated bymeasuring the Ki-67 nuclear Ag expression (Fig. 6, A and B, forCD4� and CD8� T cells, respectively). In CVID patients, thepercentage of both Ki-67� CD4� and CD8� T cells was signifi-cantly increased in all three groups as compared with healthy con-trols ( p � 0.05). Significant differences in the percentage of Ki-67expression could also be detected in the naive and CD27� memory

FIGURE 5. Flow cytometric anal-ysis of cytokine production in CVIDpatients and age-matched controls.Cytokine production was studied inPBLs stimulated with ionomycin andPMA in the presence of brefeldin A(49). Data are represented as boxplots displaying medians, 25th and75th percentiles as boxes, and 10thand 90th percentiles as whiskers. Dif-ferences between patients and healthycontrols were compared by the Mann-Whitney U test.

3939The Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

subsets in CVID patients belonging to group I and II for CD4� Tcells and to group I for CD8� T cells, as compared with healthycontrols ( p � 0.05). Naive T cells showed a 7- (CD8) to 15- (CD4)fold elevation in Ki-67� T cell percentage, whereas memory andeffector T cells showed an increase up to 4-fold. In CD4� T cells,the expression of Ki-67 Ag negatively correlated with the percent-age of both naive (r � �0.7; p � 0.0001) and CD27� memory(r � �0.434; p � 0.02) subsets and positively with the percentageof CD27� memory T cells (r � 0.445; p � 0.02). In contrast, inCD8� T cells, a significant correlation could be demonstrated onlybetween the percentage of Ki-67 expression and that of the CD27�

memory T cell subset (r � 0.471; p � 0.0132).

Increased levels of spontaneous apoptosis

The percentage of total lymphocytes expressing annexin V wassignificantly increased in all three groups of CVID patients ( p �0.0040, p � 0.0050, and p � 0.0056 for group I, II, and III, re-spectively) (Fig. 7A). However, the percentage of apoptotic lym-phocytes was massively increased in group I, as compared togroups II and III ( p � 0.01), and negatively correlated with thenumber of CD4� naive cells (r � �0.855; p � 0.0001). A moredetailed analysis of lymphocyte population confirmed, within boththe CD4 and CD8 subsets, significant levels of apoptosis in allthree groups of CVID patients (Fig. 7, B and C). Also, in thisinstance, apoptosis was maximally increased in group I vs groupsII and III (for CD4� T cells, p � 0.01 and for CD8� T cells, p �0.05). The percentage of CD4� apoptotic cells were related to theiractivation state (r � 0.7 and p � 0.0001 for CD95-, CCR5-, HLA-DR-, and IFN-�-expressing CD8� T cells). A positive correlationwas also detected between the percentage of CD4� apoptotic cells

and that of Ki-67�CD4� T cells (r � 0.635; p � 0.0006). Simi-larly to that observed in CD4� T cells, a positive correlation be-tween the rate of CD8� T cell apoptosis and activation/prolifera-tion markers could be demonstrated (r � 0.5; p � 0.001).

Differential gene expression

To find a possible link between the above described T cell defectsand the genotype of CVID patients, we studied the expression of406 immunologic and hematologic potentially relevant genes. Dif-ferences in gene expression (�2-fold up- or down-regulated) forCVID group I and III vs control group are listed in Table IV forgroup I (top) and III (bottom), respectively. Prominently enrichedamong the genes up-regulated in CVID patients of group I weremany genes known to be induced during lymphocyte activationsuch as genes encoding cell adhesion molecules, signal transduc-ers, and transcription activators. Moreover, some of the genesoverexpressed in group I, i.e., LFA-1A, LFA-3, VLA-4, STAT1,and STAT2 were down-regulated in group III, whereas genesdown-regulated in group I such as AF-17 and Ikaros were up-regulated in group III. Adjacent to these sets of differentially ex-pressed genes, another series of genes including ICAM-1, CREB1,

FIGURE 6. Percentage of Ki-67� T lymphocytes within the CD4� andCD8� peripheral blood T cells of CVID patients and age-matched healthycontrols. Ki-67 Ag expression was measured in CD4� (A) and CD8� (B)naive (CD45RO�CD27high), and CD27� memory (CD45RO�/CD27�),CD27� memory (CD45RO�/CD27�) cells, and CD27� effector(CD45RO�/CD27�) T cells (35). Data are represented as mean values SD. An asterisk indicates statistical difference compared with control val-ues (Mann-Whitney U test).

FIGURE 7. Flow cytometric analysis of spontaneous apoptosis inCVID patients and age-matched healthy controls. Spontaneous apoptosiswas evaluated in PBLs (A), CD4� (B), and CD8� (C) T cells after 48-hcultures. Annexin V-positive cells include both cells in the early (FITC-conjugated annexin V single-positive cells) and later stages of apoptosis(FITC-conjugated annexin V and PI double-positive cells). Data are rep-resented as box plots displaying medians, 25th and 75th percentiles asboxes, and 10th and 90th percentiles as whiskers. Differences betweenpatients and controls were compared by the Mann-Whitney U test.

3940 T CELLS IN COMMON VARIABLE IMMUNODEFICIENCY

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

BTG1, NK4, LCP1, and BAT2 could be detected, the expressionpatterns of which were similar in both group I and III.

To validate these macroarray data, we analyzed, by quantitativereal-time RT-PCR, some of the genes whose expression showed

extreme variations among the two groups of patients, i.e., HCF1,LFA-3, and CD9. The results demonstrated a reasonable agree-ment between the data obtained using these two different tech-niques. In particular, in groups I and III, the expression of HCFC1was 12.6 2.3 and 21 13%, the expression of CD9 was 166 3 and 84 4%, and the expression of LFA3 was 116 21 and19 11%, respectively, in comparison to controls normalizedto 100%.

DiscussionDespite intense investigation, the etiology as well as the patho-genesis of CVID, years after its initial recognition, remain largelyunknown. A major obstacle in characterizing the molecular eventsleading to CVID is certainly the great heterogeneity of the disease,which is implied by defining it as a variable. This obstacle has alsohampered the fulfillment of a commonly accepted approach toclassify subgroups of cases with CVID. Achieving classificationsystems has therefore become one of the main goals for peopleinvolved in study of CVID (26), who need to precisely definehomogeneous subpopulations of patients. It is well known thatpatients with CVID may present a vast array of T cell abnormal-ities, but, to date, a comprehensive evaluation of the T cell defi-ciency is lacking. In this study, we demonstrate that CVID con-stantly associates with variable degrees of T cell dysfunction.Multiple T cell abnormalities are simultaneously detected in af-fected individuals intimately related among them, and whose se-verity reflects in a parallel loss of CD4� naive T cells. Due to itsstrong correlation with the severity of the disease, we propose thatthe level of CD4� naive T cell should be considered as a parameteressential to classify CVID and possibly as a valuable prognosticmarker.

Irrespective of the T cell defects affecting the vast majority ofthe patients, the only proposed classifications for CVID are actu-ally based on functional or phenotypic characteristics of B cells.Clustering patients with a diagnosis of CVID according to theirnaive CD4� T cell numbers allows the recognition of discretesubgroups that exhibit several shared abnormalities of T cell-me-diated immunity. On the basis of the results presented in this study,three main groups of CVID patient may be identified. One groupis characterized by marked depletion of CD4� naive T cells, mas-sive T cell activation, proliferation, and apoptosis, disruption ofCD4� and CD8� TCR repertoires, and clinically characterized bysevere immunodeficiency generally associated with splenomegaly;a second group is characterized by less prominent alteration ofCD4� and CD8� T cells and less advanced immunodeficiencyoccasionally presenting moderate splenomegaly; and, finally, athird group of CVID patients is characterized by prevalent CD8�

T cell abnormalities, CD4� naive T cell levels within normalrange, mild clinical symptoms, and slight or absent splenomegaly.Naive CD4� T cell levels were significantly associated with theWarnatz et al. (20) classification, based on phenotypic B cell mat-uration markers, although the three groups did not overlap withthose assigned on the basis of naive CD4� T cells. The correlationhighlighted the presence of more severe memory B cell depletionin those CVID cases that had more severe loss of naive CD4�

lymphocytes, but the resulting assignment to subgroups did notexceed 58.8% concordance between the two systems.

In normal conditions, the size of the peripheral blood T cellcompartment is regulated by the combined effects of thymic outputand peripheral proliferation counterbalanced by apoptosis (27).However, differences exist in the CD4� and CD8� T cell regen-erative pathway that account for the natural tendency of the CD4�

T cell compartment to shrink (28). Differently from reconstitutionof CD8� T cells which is rapid, independent from thymic function

Table IV. Selected genes with differential expression in CVID patientsof group I and III as compared to healthy controls

Group I

Name Accession Nos. Ratio

IncreasedAdhesion/migration

CD9 Ag M38690 11:46ICAM-1 J03132 2:74LFA-1A Y00796 2:43T cell-specific RANTES protein M21121 2:21LFA-3 Y00636 2:16VLA-4 L12002 2:01

Regulation of transcription/signaltransduction/intracellular signalingcascade

CREB1 M34356 5:28Activating transcription factor 4 D90209 3:01STAT2 U18671 2:83STAT1 �� M97935 2:11

Name Accession Nos. Ratio

DecreasedOncogenes and tumor suppressors

Tyrosine-protein kinase lyn M16038 0:38AF-17 protein U07932 0:40B cell translocation gene 1 protein

(BTG1)X61123 0:47

Regulation of transcription/signaltransduction/intracellular signalingcascade

HCFC1 L20010 0:07Zinc finger protein, subfamily 1A,

1 (Ikaros)U40462 0:35

JAK3 U09607 0:40Other immune system proteins

NK cells protein 4 precursor (NK4) M59807 0:14Lymphocyte cytosolic protein 1 (LCP1) M22300 0:43HLA-B-associated transcript 2 (BAT2) M33509 0:48

Group III

Name Accession Nos. Ratio

IncreasedAdhesion/migration

ICAM-1 J03132 3:22Oncogenes and tumor suppressors

AF-17 protein U07932 1:86Regulation of transcription/signal

transduction/intracellular signalingcascade

Ikaros U40462 2:00CREB1 M34356 1:89

Other immune system proteinsFGFR1 X66945 2:90

Name Accession Nos. Ratio

DecreasedAdhesion/migration

LFA-3 Y00636 0:004VLA-4 L12002 0:41LFA-1A Y00796 0:47

Oncogenes and tumor suppressorsBTG1 X61123 0:40

Regulation of transcription/signaltransduction/intracellular signalingcascade

HCFC1 L20010 0:07Nucleophosmin 1 M23613 0:21STAT1 �� M97935 0:28STAT2 U18671 0:29

Other immune system proteinsLCP1 M22300 0:10NK4 M59807 0:17BAT2 M33509 0:17

3941The Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

and occurring by peripheral expansion, reconstitution of CD4� Tcells is slower and depends on thymic output (29, 30). A first,natural obstacle to CD4� T cell recovery is therefore the age-dependent decline of thymopoiesis. The regenerative capability ofthe CD4� T cell compartment is also reduced by the proneness ofCD4� T cells to undergo apoptosis as well as by an intrinsicallylimited expansion of CD4� T cells in response to antigenicstimula-tion (31, 32).

Evidence from mice as well as from HIV-1-infected individualsundergoing highly active antiretroviral therapy treatment, suggeststhat the number of newly formed naive T lymphocytes is deter-mined by the thymic mass (33). According to this result, in normalindividuals the frequency of recent thymic emigrants start to de-cline between 15 and 20 years of age, corresponding to the time ofthymic involution. The reduced frequency of CD31� T cells,which we found in a large part of CVID patients, indicated a re-markable impairment of thymic output, and loss of the physiolog-ical age-dependent decline was observed in controls. Although itwas impossible to establish the effective duration of the disease,our data, showing a deeply reduced number of thymic emigrantseven at 30–35 years of age, contrast with an effect of chronicCVID on T cell homeostasis in the naive compartment. An im-paired supply of bone marrow progenitor cells could play a role inthe premature exhaustion of thymic function, as suggested by thereduced maturation of hemopoietic progenitors and destruction ofstromal elements we recently described in CVID (34).

Enhanced self-replication of T cells was demonstrated in CVIDby increased Ki-67 nuclear Ag expression in both peripheral naiveand memory cells. The mean percentage of dividing naive CD4�

T cells in CVID patients was 10-fold higher than in controls. Sim-ilarly to HIV-1 infection (35), higher increases in percentages ofKi-67�CD4� naive T cells were observed only when the numberof naive cells fell below 20%. As expected, the rate of proliferationsignificantly associated with the levels of T cell apoptosis andactivation. These observations point to a key role exerted in CVIDby T cell activation in inducing an increased cell death first, and anincreased peripheral proliferation of T cells after, to compensatefor the ineffective thymic output. The peripheral homeostatic ex-pansion implies that the diversity of the T cell pool is lower thana comparably sized T cell pool derived from intrathymic matura-tion. A restricted TCR repertoire might result in turn in a decreasedability to counteract infections or cancer.

Only limited information is currently available on the diversityof TCR repertoire of CVID (36, 37). We detected a marked con-traction of the TCRBV repertoire in most CVID patients. The sig-nificant correlation observed between the number of CD4 TCRBVexpansions and the frequency of CD4� naive T cells suggestedthat the impaired influx of new cells played a role in the genesis ofthe disrupted repertoires. Alternatively, expansion of selected BVfamilies could represent the result of chronic antigenic stimulationinducing expansion of specific memory cells. In CVID patients, apreferential BV usage was observed for a number of CD4 BVfamilies, among them BV16, BV14, and BV 21.3. The spectratyp-ing of the VDJ region showed oligoclonal profiles in patients withsevere depletion of CD4� naive T cells (group I, �20% of CD4�

T cells), in striking contrast to Gaussian profiles observed in pa-tients with less or no impairment of T cell homeostasis. Notably,the CDR3 perturbation, exceptionally high in CD4� T cells,reached impressive levels in CD8� T cells, even more than thatobserved in advanced HIV-1 infection, which invariably associateswith disrupted TCR repertoires (38, 39). Interestingly, the oligo-clonality of the CD8 TCRBV repertoire was not confined to clin-

ically more compromised patients but also concerned patients withmilder clinical course, although to a lesser extent.

It has been reported that i.v. Ig treatment can influence the T cellrepertoires (40). However, the TCRBV analysis we performed inCVID patients naive for Ig treatment, resulted in data overlappingthose observed in i.v. Ig-treated patients, allowing us to exclude apossible role of this therapeutic intervention in shaping the oligo-clonal expansions (our unpublished observations).

Our data on cDNA arrays, pointed to discrete genetic profiles forthe different subgroups of CVID patients. In group I of patients, thegene giving the strongest hybridization signal was CD9, a cellsurface molecule belonging to the tetraspanin family. This familyof proteins is involved in an astonishing variety of biologic re-sponses including adhesion, morphology, activation, proliferation,and differentiation of B, T, and other cells (41). Other genes wefound up-regulated in group I were the genes encoding for theadhesion molecules ICAM-1 and LFA-3. This finding was consis-tent with previous reports showing an increased expression ofthese molecules in a subgroup of CVID patients characterized bylow CD4:CD8 ratio, increased T cell activation, and splenomegaly(42–44). Noteworthy was also the differential expression of AF-17and Ikaros genes that we found in patients of group I and III, whichcould represent a useful marker for subgroup prediction andassignment.

Recent evidence also suggest a primary defect of myeloid den-dritic cells (DCs) in CVID (45, 46). DCs have a crucial role in Agpresentation to T lymphocytes and instructively direct their acti-vation and differential functional maturation. It is likely that DCdefects underlie several of the immune abnormalities observed inCVID, and this result should be considered in future studies.

In conclusion, data presented in this study indicate that, far frombeing innocent bystanders, T cells play a key role in the patho-genesis of CVID. Therefore, a proper classification of this diseaseshould include, in addition to functional and phenotypic charac-terization of B cells, the evaluation of T cell functions andnumbers.

DisclosuresThe authors have no financial conflict of interest.

References1. Conley, M. E., L. D. Notarangelo, and A. Etzioni. 1999. Diagnostic criteria for

primary immunodeficiencies: representing PAGID (Pan-American Group for Im-munodeficiency) and ESID (European Society for Immunodeficiencies). Clin.Immunol. 93: 190–197.

2. Cunningham-Rundles, C., and C. Bodian. 1999. Common variable immunodefi-ciency: clinical and immunological features of 248 patients. Clin. Immunol. 92:34–48.

3. Cunningham-Rundles, C., D. L. Cooper, T. P. Duffy, and J. Strauchen. 2002.Lymphomas of mucosal-associated lymphoid tissue in common variable immu-nodeficiency. Am. J. Hematol. 69: 171–178.

4. International Union of Immunological Societies. 1999. Primary immunodefi-ciency diseases: report of an IUIS Scientific Committee. Clin. Exp. Immunol.118(Suppl. 1): 1–28.

5. Ballow, M. 2002. Primary immunodeficiency disorders: antibody deficiency.J. Allergy Clin. Immunol. 109: 581–591.

6. Schroeder, H. W., Jr., H. W. Schroeder, III, and S. M. Sheikh. 2004. The complexgenetics of common variable immunodeficiency. J. Investig. Med. 52: 90–103.

7. Grimbacher, B., A. Hutloff, M. Schlesier, E. Glocker, K. Warnatz, R. Drager,H. Eibel, B. Fischer, A. A. Schaffer, H. W. Mages, et al. 2003. Homozygous lossof ICOS is associated with adult-onset common variable immunodeficiency. Nat.Immunol. 4: 261–268.

8. Salzer, U., A. Maul-Pavicic, C. Cunningham-Rundles, S. Urschel, B. H.Belohradsky, J. Litzman, A. Holm, J. L. Franco, A. Plebani, L. Hammarstrom, et al.2004. ICOS deficiency in patients with common variable immunodeficiency. Clin.Immunol. 113: 234–240.

9. Salzer, U., H. M. Chapel, A. D. Webster, Q. Pan-Hammarstrom, A. Schmitt-Graeff,M. Schlesier, H. H. Peter, J. K. Rockstroh, P. Schneider, A. A. Schaffer, et al. 2005.Mutations in TNFRSF13B encoding TACI are associated with common variableimmunodeficiency in humans. Nat. Genet. 37: 820–828

10. Castigli, E., S. A. Wilson, L. Garibyan, R. Rachid, F. Bonilla, L. Schneider, andR. S. Geha. 2005. TACI is mutant in common variable immunodeficiency andIgA deficiency. Nat. Genet. 37: 829–834.

3942 T CELLS IN COMMON VARIABLE IMMUNODEFICIENCY

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from

11. Di Renzo, M., D. Serrano, Z. Zhou, I. George, K. Becker, and C. Cunningham-Rundles. 2001. Enhanced T cell apoptosis in common variable immunodeficiency:negative role of the fas/fas ligand system and of the Bcl-2 family proteins and pos-sible role of TNF-RS. Clin. Exp. Immunol. 125: 117–122.

12. Sneller, M. C., and W. Strober. 1990. Abnormalities of lymphokine gene expres-sion in patients with common variable immunodeficiency. J. Immunol. 144:3762–3769.

13. Aukrust, P., F. Muller, and S. S. Froland. 1994. Elevated serum levels of inter-leukin-4 and interleukin-6 in patients with common variable immunodeficiency(CVI) are associated with chronic immune activation and low numbers of CD4�

lymphocytes. Clin. Immunol. Immunopathol. 70: 217–224.14. Stagg, A. J., M. Funauchi, S. C. Knight, A. D. Webster, and J. Farrant. 1994.

Failure in antigen responses by T cells from patients with common variableimmunodeficiency (CVID). Clin. Exp. Immunol. 96: 48–53.

15. Kondratenko, I., P. L. Amlot, A. D. Webster, and J. Farrant. 1997. Lack ofspecific antibody response in common variable immunodeficiency (CVID) asso-ciated with failure in production of antigen-specific memory T cells: MRC Im-munodeficiency Group. Clin. Exp. Immunol. 108: 9–13.

16. Farrington, M., L. S. Grosmaire, S. Nonoyama, S. H. Fischer, D. Hollenbaugh,J. A. Ledbetter, R. J. Noelle, A. Aruffo, and H. D. Ochs. 1994. CD40 ligandexpression is defective in a subset of patients with common variable immuno-deficiency. Proc. Natl. Acad. Sci. USA 91: 1099–1103.

17. Oliva, A., E. Scala, I. Quinti, R. Paganelli, I. J. Ansotegui, A. Giovannetti,M. Pierdominici, F. Aiuti, and F. Pandolfi. 1997. IL-10 production and CD40Lexpression in patients with common variable immunodeficiency. Scand. J. Im-munol. 46: 86–90.

18. Bryant, A., N. C. Calver, E. Toubi, A. D. Webster, and J. Farrant. 1990. Clas-sification of patients with common variable immunodeficiency by B cell secretionof IgM and IgG in response to anti-IgM and interleukin-2. Clin. Immunol. Im-munopathol. 56: 239–248.

19. Scott, L. J., A. Bryant, A. D. Webster, and J. Farrant. 1994. Failure in IgAsecretion by surface IgA-positive B cells in common variable immunodeficiency(CVID). Clin. Exp. Immunol. 95: 10–13.

20. Warnatz, K., A. Denz, R. Drager, M. Braun, C. Groth, G. Wolff-Vorbeck,H. Eibel, M. Schlesier, and H. H. Peter. 2002. Severe deficiency of switchedmemory B cells (CD27�IgM�IgD�) in subgroups of patients with common vari-able immunodeficiency: a new approach to classify a heterogeneous disease.Blood 99: 1544–1551.

21. Piqueras, B., C. Lavenu-Bombled, L. Galicier, F. Bergeron-van der Cruyssen,L. Mouthon, S. Chevret, P. Debre, C. Schmitt, and E. Oksenhendler. 2003. Com-mon variable immunodeficiency patient classification based on impaired B cellmemory differentiation correlates with clinical aspects. J. Clin. Immunol. 23:385–400.

22. Giovannetti, A., F. Mazzetta, E. Caprini, A. Aiuti, M. Marziali, M. Pierdominici,A. Cossarizza, L. Chessa, E. Scala, I. Quinti, et al. 2002. Skewed T-cell receptorrepertoire, decreased thymic output, and predominance of terminally differenti-ated T cells in ataxia telangiectasia. Blood 100: 4082–4089.

23. Chen, Y. P., J. A. Higgins, and D. E. Gundersen-Rindal. 2003. Quantitation of aGlyptapanteles indiensis polydnavirus gene expressed in parasitized host, Ly-mantria dispar, by real-time quantitative RT-PCR. J. Virol. Methods 114:125–133.

24. Kimmig, S., G. K. Przybylski, C. A. Schmidt, K. Laurisch, B. Mowes,A. Radbruch, and A. Thiel. 2002. Two subsets of naive T helper cells withdistinct T cell receptor excision circle content in human adult peripheral blood.J. Exp. Med. 195: 789–794.

25. Guazzi, V., F. Aiuti, I. Mezzaroma, F. Mazzetta, G. Andolfi, A. Mortellaro,M. Pierdominici, R. Fantini, M. Marziali, and A. Aiuti. 2002. Assessment ofthymic output in common variable immunodeficiency patients by evaluation of Tcell receptor excision circles. Clin. Exp. Immunol. 129: 346–353.

26. Goldacker, S., and K. Warnatz. 2005. Tackling the heterogeneity of CVID. Curr.Opin. Allergy Clin. Immunol. 5: 504–509.

27. Grossman, Z., B. Min, M. Meier-Schellersheim, and W. E. Paul. 2004. Concom-itant regulation of T-cell activation and homeostasis. Nat. Rev. Immunol. 4:387–395.

28. Douek, D. C., L. J. Picker, and R. A. Koup. 2003. T cell dynamics in HIV-1infection. Annu. Rev. Immunol. 21: 265–304.

29. Atkinson, K., J. A. Hansen, R. Storb, S. Goehle, G. Goldstein, and E. D. Thomas.1982. T-cell subpopulations identified by monoclonal antibodies after humanmarrow transplantation, I: helper-inducer and cytotoxic-suppressor subsets.Blood 59: 1292–1298.

30. Watanabe, N., S. C. De Rosa, A. Cmelak, R. Hoppe, L. A. Herzenberg, andM. Roederer. 1997. Long-term depletion of naive T cells in patients treated forHodgkin’s disease. Blood 90: 3662–3672.

31. Hakim, F. T., R. Cepeda, S. Kaimei, C. L. Mackall, N. McAtee, J. Zujewski,K. Cowan, and R. E. Gress. 1997. Constraints on CD4 recovery postchemo-therapy in adults: thymic insufficiency and apoptotic decline of expanded periph-eral CD4 cells. Blood 90: 3789–3798.

32. Foulds, K. E., L. A. Zenewicz, D. J. Shedlock, J. Jiang, A. E. Troy, and H. Shen.2002. Cutting edge: CD4 and CD8 T cells are intrinsically different in theirproliferative responses. J. Immunol. 168: 1528–1532.

33. Berzins, S. P., D. I. Godfrey, J. F. Miller, and R. L. Boyd. 1999. A central rolefor thymic emigrants in peripheral T cell homeostasis. Proc. Natl. Acad. Sci. USA96: 9787–9791.

34. Isgro, A., M. Marziali, I. Mezzaroma, G. Luzi, A. M. Mazzone, V. Guazzi,G. Andolfi, B. Cassani, A. Aiuti, and F. Aiuti. 2005. Bone marrow clonogeniccapability, cytokine production, and thymic output in patients with common vari-able immunodeficiency. J. Immunol. 174: 5074–5081.

35. Hazenberg, M. D., J. W. Stuart, S. A. Otto, J. C. Borleffs, C. A. Boucher,R. J. de Boer, F. Miedema, and D. Hamann. 2000. T-cell division in humanimmunodeficiency virus (HIV)-1 infection is mainly due to immune activation: alongitudinal analysis in patients before and during highly active antiretroviraltherapy (HAART). Blood 95: 249–255.

36. Duchmann, R., J. Jaffe, R. Ehrhardt, D. W. Alling, and W. Strober. 1996. Dif-ferential usage of T-cell receptor V� gene families by CD4� and CD8� T cellsin patients with CD8hi common variable immunodeficiency: evidence of a post-thymic effect. Immunology 87: 99–107.

37. Serrano, D., K. Becker, C. Cunningham-Rundles, and L. Mayer. 2000. Charac-terization of the T cell receptor repertoire in patients with common variableimmunodeficiency: oligoclonal expansion of CD8� T cells. Clin. Immunol. 97:248–258.

38. Gorochov, G., A. U. Neumann, A. Kereveur, C. Parizot, T. Li, C. Katlama,M. Karmochkine, G. Raguin, B. Autran, and P. Debre. 1998. Perturbation ofCD4� and CD8� T-cell repertoires during progression to AIDS and regulation ofthe CD4� repertoire during antiviral therapy. Nat. Med. 4: 215–221.

39. Giovannetti, A., M. Pierdominici, M. Marziali, F. Mazzetta, E. Caprini, G. Russo,R. Bugarini, M. L. Bernardi, I. Mezzaroma, and F. Aiuti. 2003. Persistentlybiased T-cell receptor repertoires in HIV-1-infected combination antiretroviraltherapy-treated patients despite sustained suppression of viral replication.J. Acquired Immune Defic. Syndr. 34: 140–154.

40. Kazatchkine, M. D., and S. V. Kaveri. 2001. Immunomodulation of autoimmuneand inflammatory diseases with intravenous immune globulin. N. Engl. J. Med.345: 747–755.

41. Boucheix, C., and E. Rubinstein. 2001. Tetraspanins. Cell Mol. Life Sci. 58:1189–1184.

42. Baumert, E., G. Wolff-Vorbeck, M. Schlesier, and H. H. Peter. 1992. Immuno-phenotypical alterations in a subset of patients with common variable immuno-deficiency (CVID). Clin. Exp. Immunol. 90: 25–30.

43. Nordoy, I., F. Muller, P. Aukrust, and S. S. Froland. 1998. Adhesion moleculesin common variable immunodeficiency (CVID): a decrease in L-selectin-positiveT lymphocytes. Clin. Exp. Immunol. 114: 258–263.

44. Holm, A. M., E. A. Sivertsen, S. H. Tunheim, T. Haug, V. Bjerkeli, A. Yndestad,P. Aukrust, and S. S. Froland. 2004. Gene expression analysis of peripheral Tcells in a subgroup of common variable immunodeficiency shows predominanceof CCR7� effector-memory T cells. Clin. Exp. Immunol. 138: 278–289.

45. Bayry, J., S. Lacroix-Desmazes, M. D. Kazatchkine, L. Galicier, Y. Lepelletier,D. Webster, Y. Levy, M. M. Eibl, E. Oksenhendler, O. Hermine, andS. V. Kaveri. 2004. Common variable immunodeficiency is associated with de-fective functions of dendritic cells. Blood 104: 2441–2443.

46. Cunningham-Rundles, C., and L. Radigan. 2005. Deficient IL-12 and dendriticcell function in common variable immune deficiency. Clin. Immunol. 115:147–153.

47. Sallusto, F., D. Lenig, R. Forster, M. Lipp, and A. Lanzavecchia. 1999. Twosubsets of memory T lymphocytes with distinct homing potentials and effectorfunctions. Nature 401: 708–712.

48. Rabin, R. L., M. Roederer, Y. Maldonado, A. Petru, and L. A. Herzenberg. 1995.Altered representation of naive and memory CD8 T cell subsets in HIV-infectedchildren. J. Clin. Invest. 95: 2054–2060.

49. Pierdominici, M., F. Mazzetta, E. Caprini, M. Marziali, M. C. Digilio, B. Marino,A. Aiuti, F. Amati, G. Russo, G. Novelli, et al. 2003. Biased T-cell receptorrepertoires in patients with chromosome 22q11.2 deletion syndrome (DiGeorgesyndrome/velocardiofacial syndrome). Clin. Exp. Immunol. 132: 323–331.

3943The Journal of Immunology

by guest on Decem

ber 18, 2014http://w

ww

.jimm

unol.org/D

ownloaded from